{
  "best_global_step": 240500,
  "best_metric": 0.9813696191263432,
  "best_model_checkpoint": "/workspace/output/lora_r4/checkpoint-240500",
  "epoch": 68.69145614533069,
  "eval_steps": 500,
  "global_step": 242000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002838489923360772,
      "grad_norm": 7.826807975769043,
      "learning_rate": 9.999744535906897e-05,
      "loss": 7.1339569091796875,
      "step": 10
    },
    {
      "epoch": 0.005676979846721544,
      "grad_norm": 7.7576518058776855,
      "learning_rate": 9.999460686914563e-05,
      "loss": 6.710009765625,
      "step": 20
    },
    {
      "epoch": 0.008515469770082317,
      "grad_norm": 7.841175556182861,
      "learning_rate": 9.999176837922227e-05,
      "loss": 6.468508911132813,
      "step": 30
    },
    {
      "epoch": 0.011353959693443088,
      "grad_norm": 7.927770137786865,
      "learning_rate": 9.99889298892989e-05,
      "loss": 6.272714233398437,
      "step": 40
    },
    {
      "epoch": 0.01419244961680386,
      "grad_norm": 8.131056785583496,
      "learning_rate": 9.998609139937554e-05,
      "loss": 6.034750366210938,
      "step": 50
    },
    {
      "epoch": 0.017030939540164634,
      "grad_norm": 7.635435104370117,
      "learning_rate": 9.998325290945218e-05,
      "loss": 5.990374755859375,
      "step": 60
    },
    {
      "epoch": 0.019869429463525403,
      "grad_norm": 8.140981674194336,
      "learning_rate": 9.99804144195288e-05,
      "loss": 5.84742431640625,
      "step": 70
    },
    {
      "epoch": 0.022707919386886176,
      "grad_norm": 8.634374618530273,
      "learning_rate": 9.997757592960545e-05,
      "loss": 5.69959716796875,
      "step": 80
    },
    {
      "epoch": 0.02554640931024695,
      "grad_norm": 8.20428466796875,
      "learning_rate": 9.99747374396821e-05,
      "loss": 5.6543731689453125,
      "step": 90
    },
    {
      "epoch": 0.02838489923360772,
      "grad_norm": 8.128388404846191,
      "learning_rate": 9.997189894975873e-05,
      "loss": 5.422869873046875,
      "step": 100
    },
    {
      "epoch": 0.031223389156968494,
      "grad_norm": 8.217612266540527,
      "learning_rate": 9.996906045983537e-05,
      "loss": 5.347100830078125,
      "step": 110
    },
    {
      "epoch": 0.03406187908032927,
      "grad_norm": 8.654301643371582,
      "learning_rate": 9.996622196991201e-05,
      "loss": 5.253237915039063,
      "step": 120
    },
    {
      "epoch": 0.03690036900369004,
      "grad_norm": 8.540105819702148,
      "learning_rate": 9.996338347998865e-05,
      "loss": 5.096551513671875,
      "step": 130
    },
    {
      "epoch": 0.039738858927050806,
      "grad_norm": 8.838942527770996,
      "learning_rate": 9.996054499006528e-05,
      "loss": 4.921728515625,
      "step": 140
    },
    {
      "epoch": 0.04257734885041158,
      "grad_norm": 8.903046607971191,
      "learning_rate": 9.995770650014194e-05,
      "loss": 4.8088134765625,
      "step": 150
    },
    {
      "epoch": 0.04541583877377235,
      "grad_norm": 9.810368537902832,
      "learning_rate": 9.995486801021858e-05,
      "loss": 4.75050048828125,
      "step": 160
    },
    {
      "epoch": 0.04825432869713313,
      "grad_norm": 10.207559585571289,
      "learning_rate": 9.99520295202952e-05,
      "loss": 4.630392456054688,
      "step": 170
    },
    {
      "epoch": 0.0510928186204939,
      "grad_norm": 10.042340278625488,
      "learning_rate": 9.994919103037185e-05,
      "loss": 4.4931793212890625,
      "step": 180
    },
    {
      "epoch": 0.05393130854385467,
      "grad_norm": 10.114358901977539,
      "learning_rate": 9.994635254044849e-05,
      "loss": 4.399362182617187,
      "step": 190
    },
    {
      "epoch": 0.05676979846721544,
      "grad_norm": 10.164721488952637,
      "learning_rate": 9.994351405052512e-05,
      "loss": 4.324098205566406,
      "step": 200
    },
    {
      "epoch": 0.05960828839057621,
      "grad_norm": 11.823846817016602,
      "learning_rate": 9.994067556060176e-05,
      "loss": 4.284576416015625,
      "step": 210
    },
    {
      "epoch": 0.06244677831393699,
      "grad_norm": 11.347098350524902,
      "learning_rate": 9.993783707067841e-05,
      "loss": 4.103669738769531,
      "step": 220
    },
    {
      "epoch": 0.06528526823729776,
      "grad_norm": 11.463887214660645,
      "learning_rate": 9.993499858075504e-05,
      "loss": 4.066629028320312,
      "step": 230
    },
    {
      "epoch": 0.06812375816065853,
      "grad_norm": 12.289628982543945,
      "learning_rate": 9.993216009083168e-05,
      "loss": 4.048051452636718,
      "step": 240
    },
    {
      "epoch": 0.0709622480840193,
      "grad_norm": 12.22239875793457,
      "learning_rate": 9.992932160090832e-05,
      "loss": 3.9947723388671874,
      "step": 250
    },
    {
      "epoch": 0.07380073800738007,
      "grad_norm": 12.375640869140625,
      "learning_rate": 9.992648311098496e-05,
      "loss": 3.9365478515625,
      "step": 260
    },
    {
      "epoch": 0.07663922793074085,
      "grad_norm": 12.353590965270996,
      "learning_rate": 9.992364462106159e-05,
      "loss": 3.8360382080078126,
      "step": 270
    },
    {
      "epoch": 0.07947771785410161,
      "grad_norm": 11.50862979888916,
      "learning_rate": 9.992080613113823e-05,
      "loss": 3.845257568359375,
      "step": 280
    },
    {
      "epoch": 0.08231620777746239,
      "grad_norm": 12.36604118347168,
      "learning_rate": 9.991796764121489e-05,
      "loss": 3.7559188842773437,
      "step": 290
    },
    {
      "epoch": 0.08515469770082316,
      "grad_norm": 13.271099090576172,
      "learning_rate": 9.991512915129152e-05,
      "loss": 3.75750732421875,
      "step": 300
    },
    {
      "epoch": 0.08799318762418393,
      "grad_norm": 12.769731521606445,
      "learning_rate": 9.991229066136816e-05,
      "loss": 3.6090957641601564,
      "step": 310
    },
    {
      "epoch": 0.0908316775475447,
      "grad_norm": 12.775890350341797,
      "learning_rate": 9.99094521714448e-05,
      "loss": 3.6455108642578127,
      "step": 320
    },
    {
      "epoch": 0.09367016747090548,
      "grad_norm": 12.35387897491455,
      "learning_rate": 9.990661368152143e-05,
      "loss": 3.5637794494628907,
      "step": 330
    },
    {
      "epoch": 0.09650865739426626,
      "grad_norm": 13.14075756072998,
      "learning_rate": 9.990377519159807e-05,
      "loss": 3.536479187011719,
      "step": 340
    },
    {
      "epoch": 0.09934714731762702,
      "grad_norm": 13.470285415649414,
      "learning_rate": 9.990093670167472e-05,
      "loss": 3.44134521484375,
      "step": 350
    },
    {
      "epoch": 0.1021856372409878,
      "grad_norm": 13.043403625488281,
      "learning_rate": 9.989809821175135e-05,
      "loss": 3.4576507568359376,
      "step": 360
    },
    {
      "epoch": 0.10502412716434857,
      "grad_norm": 13.769856452941895,
      "learning_rate": 9.989525972182799e-05,
      "loss": 3.3593673706054688,
      "step": 370
    },
    {
      "epoch": 0.10786261708770933,
      "grad_norm": 12.712862014770508,
      "learning_rate": 9.989242123190463e-05,
      "loss": 3.2312248229980467,
      "step": 380
    },
    {
      "epoch": 0.11070110701107011,
      "grad_norm": 13.926446914672852,
      "learning_rate": 9.988958274198128e-05,
      "loss": 3.353959655761719,
      "step": 390
    },
    {
      "epoch": 0.11353959693443089,
      "grad_norm": 13.616968154907227,
      "learning_rate": 9.98867442520579e-05,
      "loss": 3.261731719970703,
      "step": 400
    },
    {
      "epoch": 0.11637808685779165,
      "grad_norm": 14.255325317382812,
      "learning_rate": 9.988390576213454e-05,
      "loss": 3.2977920532226563,
      "step": 410
    },
    {
      "epoch": 0.11921657678115243,
      "grad_norm": 14.241050720214844,
      "learning_rate": 9.988106727221119e-05,
      "loss": 3.2018287658691404,
      "step": 420
    },
    {
      "epoch": 0.1220550667045132,
      "grad_norm": 16.304302215576172,
      "learning_rate": 9.987822878228783e-05,
      "loss": 3.2613784790039064,
      "step": 430
    },
    {
      "epoch": 0.12489355662787398,
      "grad_norm": 13.858403205871582,
      "learning_rate": 9.987539029236447e-05,
      "loss": 3.149138641357422,
      "step": 440
    },
    {
      "epoch": 0.12773204655123474,
      "grad_norm": 13.387377738952637,
      "learning_rate": 9.987283565143344e-05,
      "loss": 3.088349151611328,
      "step": 450
    },
    {
      "epoch": 0.13057053647459552,
      "grad_norm": 13.450538635253906,
      "learning_rate": 9.986999716151008e-05,
      "loss": 3.0058448791503904,
      "step": 460
    },
    {
      "epoch": 0.1334090263979563,
      "grad_norm": 13.723711013793945,
      "learning_rate": 9.986715867158672e-05,
      "loss": 2.9626235961914062,
      "step": 470
    },
    {
      "epoch": 0.13624751632131707,
      "grad_norm": 14.08718204498291,
      "learning_rate": 9.986432018166336e-05,
      "loss": 3.0352642059326174,
      "step": 480
    },
    {
      "epoch": 0.13908600624467782,
      "grad_norm": 13.817708969116211,
      "learning_rate": 9.986148169174e-05,
      "loss": 2.928153228759766,
      "step": 490
    },
    {
      "epoch": 0.1419244961680386,
      "grad_norm": 14.170774459838867,
      "learning_rate": 9.985864320181664e-05,
      "loss": 2.9139144897460936,
      "step": 500
    },
    {
      "epoch": 0.1419244961680386,
      "eval_accuracy": 0.18929229986647167,
      "eval_loss": 3.110288143157959,
      "eval_runtime": 31.3673,
      "eval_samples_per_second": 501.382,
      "eval_steps_per_second": 7.843,
      "step": 500
    },
    {
      "epoch": 0.14476298609139937,
      "grad_norm": 15.852224349975586,
      "learning_rate": 9.985580471189327e-05,
      "loss": 2.839629364013672,
      "step": 510
    },
    {
      "epoch": 0.14760147601476015,
      "grad_norm": 13.73141860961914,
      "learning_rate": 9.985296622196991e-05,
      "loss": 2.8279544830322267,
      "step": 520
    },
    {
      "epoch": 0.15043996593812092,
      "grad_norm": 16.905445098876953,
      "learning_rate": 9.985012773204657e-05,
      "loss": 2.8459964752197267,
      "step": 530
    },
    {
      "epoch": 0.1532784558614817,
      "grad_norm": 15.00737190246582,
      "learning_rate": 9.98472892421232e-05,
      "loss": 2.8309253692626952,
      "step": 540
    },
    {
      "epoch": 0.15611694578484248,
      "grad_norm": 15.849299430847168,
      "learning_rate": 9.984445075219984e-05,
      "loss": 2.8491188049316407,
      "step": 550
    },
    {
      "epoch": 0.15895543570820322,
      "grad_norm": 13.277036666870117,
      "learning_rate": 9.984161226227648e-05,
      "loss": 2.750569152832031,
      "step": 560
    },
    {
      "epoch": 0.161793925631564,
      "grad_norm": 19.458072662353516,
      "learning_rate": 9.98387737723531e-05,
      "loss": 2.669076156616211,
      "step": 570
    },
    {
      "epoch": 0.16463241555492478,
      "grad_norm": 15.88986587524414,
      "learning_rate": 9.983593528242975e-05,
      "loss": 2.7419994354248045,
      "step": 580
    },
    {
      "epoch": 0.16747090547828555,
      "grad_norm": 17.070947647094727,
      "learning_rate": 9.983309679250639e-05,
      "loss": 2.667735481262207,
      "step": 590
    },
    {
      "epoch": 0.17030939540164633,
      "grad_norm": 17.319679260253906,
      "learning_rate": 9.983025830258303e-05,
      "loss": 2.7211326599121093,
      "step": 600
    },
    {
      "epoch": 0.1731478853250071,
      "grad_norm": 16.1612606048584,
      "learning_rate": 9.982741981265967e-05,
      "loss": 2.620807075500488,
      "step": 610
    },
    {
      "epoch": 0.17598637524836785,
      "grad_norm": 17.112060546875,
      "learning_rate": 9.982458132273631e-05,
      "loss": 2.6665264129638673,
      "step": 620
    },
    {
      "epoch": 0.17882486517172863,
      "grad_norm": 14.721344947814941,
      "learning_rate": 9.982174283281295e-05,
      "loss": 2.5853302001953127,
      "step": 630
    },
    {
      "epoch": 0.1816633550950894,
      "grad_norm": 14.741107940673828,
      "learning_rate": 9.981890434288958e-05,
      "loss": 2.572748565673828,
      "step": 640
    },
    {
      "epoch": 0.18450184501845018,
      "grad_norm": 15.590767860412598,
      "learning_rate": 9.981606585296622e-05,
      "loss": 2.5285110473632812,
      "step": 650
    },
    {
      "epoch": 0.18734033494181096,
      "grad_norm": 18.87065315246582,
      "learning_rate": 9.981322736304286e-05,
      "loss": 2.5050926208496094,
      "step": 660
    },
    {
      "epoch": 0.19017882486517174,
      "grad_norm": 16.421432495117188,
      "learning_rate": 9.98103888731195e-05,
      "loss": 2.427301597595215,
      "step": 670
    },
    {
      "epoch": 0.1930173147885325,
      "grad_norm": 18.1635684967041,
      "learning_rate": 9.980755038319615e-05,
      "loss": 2.3481977462768553,
      "step": 680
    },
    {
      "epoch": 0.19585580471189326,
      "grad_norm": 18.698192596435547,
      "learning_rate": 9.980471189327279e-05,
      "loss": 2.39544734954834,
      "step": 690
    },
    {
      "epoch": 0.19869429463525404,
      "grad_norm": 16.736194610595703,
      "learning_rate": 9.980187340334942e-05,
      "loss": 2.4873001098632814,
      "step": 700
    },
    {
      "epoch": 0.2015327845586148,
      "grad_norm": 19.857229232788086,
      "learning_rate": 9.979903491342606e-05,
      "loss": 2.3658916473388674,
      "step": 710
    },
    {
      "epoch": 0.2043712744819756,
      "grad_norm": 17.635812759399414,
      "learning_rate": 9.97961964235027e-05,
      "loss": 2.336787796020508,
      "step": 720
    },
    {
      "epoch": 0.20720976440533637,
      "grad_norm": 16.26897430419922,
      "learning_rate": 9.979335793357934e-05,
      "loss": 2.271711540222168,
      "step": 730
    },
    {
      "epoch": 0.21004825432869714,
      "grad_norm": 16.782176971435547,
      "learning_rate": 9.979051944365598e-05,
      "loss": 2.321268653869629,
      "step": 740
    },
    {
      "epoch": 0.21288674425205792,
      "grad_norm": 23.24675178527832,
      "learning_rate": 9.978768095373262e-05,
      "loss": 2.2826637268066405,
      "step": 750
    },
    {
      "epoch": 0.21572523417541867,
      "grad_norm": 16.791034698486328,
      "learning_rate": 9.978484246380926e-05,
      "loss": 2.306294631958008,
      "step": 760
    },
    {
      "epoch": 0.21856372409877944,
      "grad_norm": 26.815982818603516,
      "learning_rate": 9.978200397388589e-05,
      "loss": 2.3874790191650392,
      "step": 770
    },
    {
      "epoch": 0.22140221402214022,
      "grad_norm": 14.967060089111328,
      "learning_rate": 9.977916548396253e-05,
      "loss": 2.256601905822754,
      "step": 780
    },
    {
      "epoch": 0.224240703945501,
      "grad_norm": 15.23949146270752,
      "learning_rate": 9.977632699403917e-05,
      "loss": 2.1650556564331054,
      "step": 790
    },
    {
      "epoch": 0.22707919386886177,
      "grad_norm": 17.264827728271484,
      "learning_rate": 9.977348850411582e-05,
      "loss": 2.195321273803711,
      "step": 800
    },
    {
      "epoch": 0.22991768379222255,
      "grad_norm": 20.194473266601562,
      "learning_rate": 9.977065001419246e-05,
      "loss": 2.245586967468262,
      "step": 810
    },
    {
      "epoch": 0.2327561737155833,
      "grad_norm": 17.323143005371094,
      "learning_rate": 9.97678115242691e-05,
      "loss": 2.259768486022949,
      "step": 820
    },
    {
      "epoch": 0.23559466363894407,
      "grad_norm": 15.63845443725586,
      "learning_rate": 9.976497303434573e-05,
      "loss": 2.2193319320678713,
      "step": 830
    },
    {
      "epoch": 0.23843315356230485,
      "grad_norm": 17.313743591308594,
      "learning_rate": 9.976213454442237e-05,
      "loss": 2.1672977447509765,
      "step": 840
    },
    {
      "epoch": 0.24127164348566563,
      "grad_norm": 17.40256690979004,
      "learning_rate": 9.975929605449901e-05,
      "loss": 2.0945472717285156,
      "step": 850
    },
    {
      "epoch": 0.2441101334090264,
      "grad_norm": 21.776893615722656,
      "learning_rate": 9.975645756457565e-05,
      "loss": 2.078630256652832,
      "step": 860
    },
    {
      "epoch": 0.24694862333238718,
      "grad_norm": 19.48147964477539,
      "learning_rate": 9.975361907465229e-05,
      "loss": 2.1304529190063475,
      "step": 870
    },
    {
      "epoch": 0.24978711325574796,
      "grad_norm": 16.93062973022461,
      "learning_rate": 9.975078058472893e-05,
      "loss": 2.069942092895508,
      "step": 880
    },
    {
      "epoch": 0.2526256031791087,
      "grad_norm": 17.80629539489746,
      "learning_rate": 9.974794209480557e-05,
      "loss": 2.0870670318603515,
      "step": 890
    },
    {
      "epoch": 0.2554640931024695,
      "grad_norm": 20.66566276550293,
      "learning_rate": 9.97451036048822e-05,
      "loss": 2.1093101501464844,
      "step": 900
    },
    {
      "epoch": 0.25830258302583026,
      "grad_norm": 16.249364852905273,
      "learning_rate": 9.974226511495884e-05,
      "loss": 2.1144399642944336,
      "step": 910
    },
    {
      "epoch": 0.26114107294919103,
      "grad_norm": 18.46486473083496,
      "learning_rate": 9.973942662503549e-05,
      "loss": 2.0278759002685547,
      "step": 920
    },
    {
      "epoch": 0.2639795628725518,
      "grad_norm": 17.295791625976562,
      "learning_rate": 9.973658813511213e-05,
      "loss": 1.9899866104125976,
      "step": 930
    },
    {
      "epoch": 0.2668180527959126,
      "grad_norm": 16.32284164428711,
      "learning_rate": 9.973374964518877e-05,
      "loss": 1.9527853012084961,
      "step": 940
    },
    {
      "epoch": 0.26965654271927336,
      "grad_norm": 18.139652252197266,
      "learning_rate": 9.973091115526541e-05,
      "loss": 1.9534086227416991,
      "step": 950
    },
    {
      "epoch": 0.27249503264263414,
      "grad_norm": 16.14771842956543,
      "learning_rate": 9.972807266534204e-05,
      "loss": 1.9304651260375976,
      "step": 960
    },
    {
      "epoch": 0.2753335225659949,
      "grad_norm": 18.864286422729492,
      "learning_rate": 9.972523417541868e-05,
      "loss": 2.0156524658203123,
      "step": 970
    },
    {
      "epoch": 0.27817201248935564,
      "grad_norm": 15.858384132385254,
      "learning_rate": 9.972239568549532e-05,
      "loss": 1.974644088745117,
      "step": 980
    },
    {
      "epoch": 0.2810105024127164,
      "grad_norm": 28.33545684814453,
      "learning_rate": 9.971955719557196e-05,
      "loss": 1.913672637939453,
      "step": 990
    },
    {
      "epoch": 0.2838489923360772,
      "grad_norm": 20.52104377746582,
      "learning_rate": 9.971700255464094e-05,
      "loss": 1.8821027755737305,
      "step": 1000
    },
    {
      "epoch": 0.2838489923360772,
      "eval_accuracy": 0.3853246010046417,
      "eval_loss": 2.1993799209594727,
      "eval_runtime": 30.911,
      "eval_samples_per_second": 508.784,
      "eval_steps_per_second": 7.958,
      "step": 1000
    },
    {
      "epoch": 0.28668748225943796,
      "grad_norm": 17.233631134033203,
      "learning_rate": 9.971416406471757e-05,
      "loss": 1.888556671142578,
      "step": 1010
    },
    {
      "epoch": 0.28952597218279874,
      "grad_norm": 18.930858612060547,
      "learning_rate": 9.971132557479421e-05,
      "loss": 1.9341995239257812,
      "step": 1020
    },
    {
      "epoch": 0.2923644621061595,
      "grad_norm": 23.888870239257812,
      "learning_rate": 9.970848708487085e-05,
      "loss": 1.8232383728027344,
      "step": 1030
    },
    {
      "epoch": 0.2952029520295203,
      "grad_norm": 21.316509246826172,
      "learning_rate": 9.970564859494749e-05,
      "loss": 1.9536394119262694,
      "step": 1040
    },
    {
      "epoch": 0.29804144195288107,
      "grad_norm": 21.5119571685791,
      "learning_rate": 9.970281010502413e-05,
      "loss": 1.856839370727539,
      "step": 1050
    },
    {
      "epoch": 0.30087993187624185,
      "grad_norm": 18.16112518310547,
      "learning_rate": 9.969997161510078e-05,
      "loss": 1.801273727416992,
      "step": 1060
    },
    {
      "epoch": 0.3037184217996026,
      "grad_norm": 14.466642379760742,
      "learning_rate": 9.969713312517742e-05,
      "loss": 1.7231252670288086,
      "step": 1070
    },
    {
      "epoch": 0.3065569117229634,
      "grad_norm": 31.817644119262695,
      "learning_rate": 9.969429463525405e-05,
      "loss": 1.9028480529785157,
      "step": 1080
    },
    {
      "epoch": 0.3093954016463242,
      "grad_norm": 20.963695526123047,
      "learning_rate": 9.969145614533069e-05,
      "loss": 1.851097869873047,
      "step": 1090
    },
    {
      "epoch": 0.31223389156968495,
      "grad_norm": 27.92727279663086,
      "learning_rate": 9.968861765540733e-05,
      "loss": 1.7934017181396484,
      "step": 1100
    },
    {
      "epoch": 0.3150723814930457,
      "grad_norm": 22.669452667236328,
      "learning_rate": 9.968577916548397e-05,
      "loss": 1.7833684921264648,
      "step": 1110
    },
    {
      "epoch": 0.31791087141640645,
      "grad_norm": 20.252046585083008,
      "learning_rate": 9.968294067556061e-05,
      "loss": 1.7792678833007813,
      "step": 1120
    },
    {
      "epoch": 0.3207493613397672,
      "grad_norm": 18.623905181884766,
      "learning_rate": 9.968010218563725e-05,
      "loss": 1.8188812255859375,
      "step": 1130
    },
    {
      "epoch": 0.323587851263128,
      "grad_norm": 19.032197952270508,
      "learning_rate": 9.967726369571388e-05,
      "loss": 1.6491582870483399,
      "step": 1140
    },
    {
      "epoch": 0.3264263411864888,
      "grad_norm": 22.769596099853516,
      "learning_rate": 9.967442520579052e-05,
      "loss": 1.7024646759033204,
      "step": 1150
    },
    {
      "epoch": 0.32926483110984955,
      "grad_norm": 17.219308853149414,
      "learning_rate": 9.967158671586716e-05,
      "loss": 1.701007080078125,
      "step": 1160
    },
    {
      "epoch": 0.33210332103321033,
      "grad_norm": 18.769567489624023,
      "learning_rate": 9.96687482259438e-05,
      "loss": 1.747934341430664,
      "step": 1170
    },
    {
      "epoch": 0.3349418109565711,
      "grad_norm": 16.897687911987305,
      "learning_rate": 9.966590973602045e-05,
      "loss": 1.6520572662353517,
      "step": 1180
    },
    {
      "epoch": 0.3377803008799319,
      "grad_norm": 18.1732234954834,
      "learning_rate": 9.966307124609709e-05,
      "loss": 1.717509651184082,
      "step": 1190
    },
    {
      "epoch": 0.34061879080329266,
      "grad_norm": 18.14131736755371,
      "learning_rate": 9.966023275617373e-05,
      "loss": 1.6966127395629882,
      "step": 1200
    },
    {
      "epoch": 0.34345728072665344,
      "grad_norm": 24.596132278442383,
      "learning_rate": 9.965739426625036e-05,
      "loss": 1.6875707626342773,
      "step": 1210
    },
    {
      "epoch": 0.3462957706500142,
      "grad_norm": 20.04149055480957,
      "learning_rate": 9.9654555776327e-05,
      "loss": 1.6078899383544922,
      "step": 1220
    },
    {
      "epoch": 0.349134260573375,
      "grad_norm": 18.574588775634766,
      "learning_rate": 9.965171728640364e-05,
      "loss": 1.682082748413086,
      "step": 1230
    },
    {
      "epoch": 0.3519727504967357,
      "grad_norm": 20.25678253173828,
      "learning_rate": 9.964887879648028e-05,
      "loss": 1.7114227294921875,
      "step": 1240
    },
    {
      "epoch": 0.3548112404200965,
      "grad_norm": 20.133079528808594,
      "learning_rate": 9.964604030655692e-05,
      "loss": 1.6490610122680665,
      "step": 1250
    },
    {
      "epoch": 0.35764973034345726,
      "grad_norm": 22.176372528076172,
      "learning_rate": 9.964320181663356e-05,
      "loss": 1.676508903503418,
      "step": 1260
    },
    {
      "epoch": 0.36048822026681804,
      "grad_norm": 19.151033401489258,
      "learning_rate": 9.964036332671019e-05,
      "loss": 1.6343936920166016,
      "step": 1270
    },
    {
      "epoch": 0.3633267101901788,
      "grad_norm": 18.87334442138672,
      "learning_rate": 9.963752483678683e-05,
      "loss": 1.5801480293273926,
      "step": 1280
    },
    {
      "epoch": 0.3661652001135396,
      "grad_norm": 18.33944320678711,
      "learning_rate": 9.963468634686347e-05,
      "loss": 1.6038869857788085,
      "step": 1290
    },
    {
      "epoch": 0.36900369003690037,
      "grad_norm": 26.26329231262207,
      "learning_rate": 9.963184785694011e-05,
      "loss": 1.5226101875305176,
      "step": 1300
    },
    {
      "epoch": 0.37184217996026114,
      "grad_norm": 23.27513885498047,
      "learning_rate": 9.962900936701676e-05,
      "loss": 1.6262874603271484,
      "step": 1310
    },
    {
      "epoch": 0.3746806698836219,
      "grad_norm": 20.425983428955078,
      "learning_rate": 9.96261708770934e-05,
      "loss": 1.5267268180847169,
      "step": 1320
    },
    {
      "epoch": 0.3775191598069827,
      "grad_norm": 31.24871063232422,
      "learning_rate": 9.962333238717003e-05,
      "loss": 1.5792694091796875,
      "step": 1330
    },
    {
      "epoch": 0.3803576497303435,
      "grad_norm": 19.59308433532715,
      "learning_rate": 9.962049389724667e-05,
      "loss": 1.5731708526611328,
      "step": 1340
    },
    {
      "epoch": 0.38319613965370425,
      "grad_norm": 23.49835205078125,
      "learning_rate": 9.961765540732331e-05,
      "loss": 1.5430625915527343,
      "step": 1350
    },
    {
      "epoch": 0.386034629577065,
      "grad_norm": 18.125337600708008,
      "learning_rate": 9.961481691739995e-05,
      "loss": 1.6195716857910156,
      "step": 1360
    },
    {
      "epoch": 0.3888731195004258,
      "grad_norm": 20.855470657348633,
      "learning_rate": 9.961197842747658e-05,
      "loss": 1.5137943267822265,
      "step": 1370
    },
    {
      "epoch": 0.3917116094237865,
      "grad_norm": 19.750150680541992,
      "learning_rate": 9.960913993755323e-05,
      "loss": 1.5584377288818358,
      "step": 1380
    },
    {
      "epoch": 0.3945500993471473,
      "grad_norm": 17.54871940612793,
      "learning_rate": 9.960630144762987e-05,
      "loss": 1.509945011138916,
      "step": 1390
    },
    {
      "epoch": 0.3973885892705081,
      "grad_norm": 19.041837692260742,
      "learning_rate": 9.96034629577065e-05,
      "loss": 1.4547015190124513,
      "step": 1400
    },
    {
      "epoch": 0.40022707919386885,
      "grad_norm": 23.58037757873535,
      "learning_rate": 9.960062446778314e-05,
      "loss": 1.4935945510864257,
      "step": 1410
    },
    {
      "epoch": 0.4030655691172296,
      "grad_norm": 21.330665588378906,
      "learning_rate": 9.959778597785978e-05,
      "loss": 1.5234458923339844,
      "step": 1420
    },
    {
      "epoch": 0.4059040590405904,
      "grad_norm": 27.210187911987305,
      "learning_rate": 9.959494748793641e-05,
      "loss": 1.4157642364501952,
      "step": 1430
    },
    {
      "epoch": 0.4087425489639512,
      "grad_norm": 24.996427536010742,
      "learning_rate": 9.959210899801307e-05,
      "loss": 1.4804776191711426,
      "step": 1440
    },
    {
      "epoch": 0.41158103888731196,
      "grad_norm": 20.157520294189453,
      "learning_rate": 9.958927050808971e-05,
      "loss": 1.5209645271301269,
      "step": 1450
    },
    {
      "epoch": 0.41441952881067273,
      "grad_norm": 24.016756057739258,
      "learning_rate": 9.958643201816634e-05,
      "loss": 1.5178532600402832,
      "step": 1460
    },
    {
      "epoch": 0.4172580187340335,
      "grad_norm": 24.519412994384766,
      "learning_rate": 9.958359352824298e-05,
      "loss": 1.512025737762451,
      "step": 1470
    },
    {
      "epoch": 0.4200965086573943,
      "grad_norm": 26.331186294555664,
      "learning_rate": 9.958075503831962e-05,
      "loss": 1.4657453536987304,
      "step": 1480
    },
    {
      "epoch": 0.42293499858075506,
      "grad_norm": 22.380062103271484,
      "learning_rate": 9.957791654839626e-05,
      "loss": 1.468503761291504,
      "step": 1490
    },
    {
      "epoch": 0.42577348850411584,
      "grad_norm": 20.46269989013672,
      "learning_rate": 9.957507805847289e-05,
      "loss": 1.493769645690918,
      "step": 1500
    },
    {
      "epoch": 0.42577348850411584,
      "eval_accuracy": 0.4980606600114453,
      "eval_loss": 1.7268073558807373,
      "eval_runtime": 31.0286,
      "eval_samples_per_second": 506.854,
      "eval_steps_per_second": 7.928,
      "step": 1500
    },
    {
      "epoch": 0.42861197842747656,
      "grad_norm": 31.702930450439453,
      "learning_rate": 9.957223956854954e-05,
      "loss": 1.5163578033447265,
      "step": 1510
    },
    {
      "epoch": 0.43145046835083734,
      "grad_norm": 18.17645263671875,
      "learning_rate": 9.956940107862618e-05,
      "loss": 1.4007353782653809,
      "step": 1520
    },
    {
      "epoch": 0.4342889582741981,
      "grad_norm": 18.51492691040039,
      "learning_rate": 9.956656258870281e-05,
      "loss": 1.3827913284301758,
      "step": 1530
    },
    {
      "epoch": 0.4371274481975589,
      "grad_norm": 19.405197143554688,
      "learning_rate": 9.956372409877945e-05,
      "loss": 1.3922388076782226,
      "step": 1540
    },
    {
      "epoch": 0.43996593812091966,
      "grad_norm": 31.843347549438477,
      "learning_rate": 9.95608856088561e-05,
      "loss": 1.4234270095825194,
      "step": 1550
    },
    {
      "epoch": 0.44280442804428044,
      "grad_norm": 35.88179016113281,
      "learning_rate": 9.955804711893272e-05,
      "loss": 1.4969772338867187,
      "step": 1560
    },
    {
      "epoch": 0.4456429179676412,
      "grad_norm": 20.38347816467285,
      "learning_rate": 9.955520862900936e-05,
      "loss": 1.4426648139953613,
      "step": 1570
    },
    {
      "epoch": 0.448481407891002,
      "grad_norm": 19.19594955444336,
      "learning_rate": 9.955237013908602e-05,
      "loss": 1.3540478706359864,
      "step": 1580
    },
    {
      "epoch": 0.45131989781436277,
      "grad_norm": 20.872447967529297,
      "learning_rate": 9.954953164916265e-05,
      "loss": 1.392230224609375,
      "step": 1590
    },
    {
      "epoch": 0.45415838773772355,
      "grad_norm": 25.477401733398438,
      "learning_rate": 9.954669315923929e-05,
      "loss": 1.3302290916442872,
      "step": 1600
    },
    {
      "epoch": 0.4569968776610843,
      "grad_norm": 19.187055587768555,
      "learning_rate": 9.954385466931593e-05,
      "loss": 1.3069095611572266,
      "step": 1610
    },
    {
      "epoch": 0.4598353675844451,
      "grad_norm": 25.863494873046875,
      "learning_rate": 9.954101617939257e-05,
      "loss": 1.321756935119629,
      "step": 1620
    },
    {
      "epoch": 0.4626738575078059,
      "grad_norm": 18.916994094848633,
      "learning_rate": 9.95381776894692e-05,
      "loss": 1.322645378112793,
      "step": 1630
    },
    {
      "epoch": 0.4655123474311666,
      "grad_norm": 21.816265106201172,
      "learning_rate": 9.953533919954585e-05,
      "loss": 1.330703067779541,
      "step": 1640
    },
    {
      "epoch": 0.46835083735452737,
      "grad_norm": 20.472036361694336,
      "learning_rate": 9.95325007096225e-05,
      "loss": 1.334905242919922,
      "step": 1650
    },
    {
      "epoch": 0.47118932727788815,
      "grad_norm": 24.386188507080078,
      "learning_rate": 9.952966221969912e-05,
      "loss": 1.3683245658874512,
      "step": 1660
    },
    {
      "epoch": 0.4740278172012489,
      "grad_norm": 28.326887130737305,
      "learning_rate": 9.952682372977576e-05,
      "loss": 1.2968414306640625,
      "step": 1670
    },
    {
      "epoch": 0.4768663071246097,
      "grad_norm": 17.494138717651367,
      "learning_rate": 9.95239852398524e-05,
      "loss": 1.3585668563842774,
      "step": 1680
    },
    {
      "epoch": 0.4797047970479705,
      "grad_norm": 19.921358108520508,
      "learning_rate": 9.952114674992903e-05,
      "loss": 1.3288690567016601,
      "step": 1690
    },
    {
      "epoch": 0.48254328697133125,
      "grad_norm": 20.302352905273438,
      "learning_rate": 9.951830826000567e-05,
      "loss": 1.3096282958984375,
      "step": 1700
    },
    {
      "epoch": 0.48538177689469203,
      "grad_norm": 28.344350814819336,
      "learning_rate": 9.951546977008233e-05,
      "loss": 1.3355764389038085,
      "step": 1710
    },
    {
      "epoch": 0.4882202668180528,
      "grad_norm": 18.56201171875,
      "learning_rate": 9.951263128015896e-05,
      "loss": 1.3571157455444336,
      "step": 1720
    },
    {
      "epoch": 0.4910587567414136,
      "grad_norm": 20.75018310546875,
      "learning_rate": 9.95097927902356e-05,
      "loss": 1.295008087158203,
      "step": 1730
    },
    {
      "epoch": 0.49389724666477436,
      "grad_norm": 21.700300216674805,
      "learning_rate": 9.950695430031224e-05,
      "loss": 1.3051130294799804,
      "step": 1740
    },
    {
      "epoch": 0.49673573658813513,
      "grad_norm": 24.367721557617188,
      "learning_rate": 9.950411581038888e-05,
      "loss": 1.2838464736938477,
      "step": 1750
    },
    {
      "epoch": 0.4995742265114959,
      "grad_norm": 18.05948257446289,
      "learning_rate": 9.950127732046551e-05,
      "loss": 1.2749036788940429,
      "step": 1760
    },
    {
      "epoch": 0.5024127164348566,
      "grad_norm": 26.424760818481445,
      "learning_rate": 9.949843883054215e-05,
      "loss": 1.1802459716796876,
      "step": 1770
    },
    {
      "epoch": 0.5052512063582174,
      "grad_norm": 23.55957794189453,
      "learning_rate": 9.94956003406188e-05,
      "loss": 1.279608917236328,
      "step": 1780
    },
    {
      "epoch": 0.5080896962815782,
      "grad_norm": 24.43779754638672,
      "learning_rate": 9.949276185069543e-05,
      "loss": 1.3498580932617188,
      "step": 1790
    },
    {
      "epoch": 0.510928186204939,
      "grad_norm": 26.3443660736084,
      "learning_rate": 9.948992336077207e-05,
      "loss": 1.3418119430541993,
      "step": 1800
    },
    {
      "epoch": 0.5137666761282997,
      "grad_norm": 22.686662673950195,
      "learning_rate": 9.948708487084872e-05,
      "loss": 1.2427737236022949,
      "step": 1810
    },
    {
      "epoch": 0.5166051660516605,
      "grad_norm": 27.369094848632812,
      "learning_rate": 9.948424638092534e-05,
      "loss": 1.3421775817871093,
      "step": 1820
    },
    {
      "epoch": 0.5194436559750213,
      "grad_norm": 20.854188919067383,
      "learning_rate": 9.948140789100199e-05,
      "loss": 1.2650754928588868,
      "step": 1830
    },
    {
      "epoch": 0.5222821458983821,
      "grad_norm": 19.906824111938477,
      "learning_rate": 9.947856940107864e-05,
      "loss": 1.1981738090515137,
      "step": 1840
    },
    {
      "epoch": 0.5251206358217428,
      "grad_norm": 18.435914993286133,
      "learning_rate": 9.947573091115527e-05,
      "loss": 1.2321277618408204,
      "step": 1850
    },
    {
      "epoch": 0.5279591257451036,
      "grad_norm": 23.23478126525879,
      "learning_rate": 9.947289242123191e-05,
      "loss": 1.2562955856323241,
      "step": 1860
    },
    {
      "epoch": 0.5307976156684644,
      "grad_norm": 21.96221160888672,
      "learning_rate": 9.947005393130855e-05,
      "loss": 1.2214305877685547,
      "step": 1870
    },
    {
      "epoch": 0.5336361055918252,
      "grad_norm": 22.680828094482422,
      "learning_rate": 9.946721544138519e-05,
      "loss": 1.2809284210205079,
      "step": 1880
    },
    {
      "epoch": 0.536474595515186,
      "grad_norm": 19.553720474243164,
      "learning_rate": 9.946437695146182e-05,
      "loss": 1.15775146484375,
      "step": 1890
    },
    {
      "epoch": 0.5393130854385467,
      "grad_norm": 21.628679275512695,
      "learning_rate": 9.946153846153846e-05,
      "loss": 1.1703840255737306,
      "step": 1900
    },
    {
      "epoch": 0.5421515753619075,
      "grad_norm": 27.79534912109375,
      "learning_rate": 9.945869997161512e-05,
      "loss": 1.1535178184509278,
      "step": 1910
    },
    {
      "epoch": 0.5449900652852683,
      "grad_norm": 23.678956985473633,
      "learning_rate": 9.945586148169174e-05,
      "loss": 1.2113995552062988,
      "step": 1920
    },
    {
      "epoch": 0.547828555208629,
      "grad_norm": 19.417715072631836,
      "learning_rate": 9.945302299176839e-05,
      "loss": 1.2194194793701172,
      "step": 1930
    },
    {
      "epoch": 0.5506670451319898,
      "grad_norm": 17.17325210571289,
      "learning_rate": 9.945018450184503e-05,
      "loss": 1.1678752899169922,
      "step": 1940
    },
    {
      "epoch": 0.5535055350553506,
      "grad_norm": 20.972352981567383,
      "learning_rate": 9.944734601192165e-05,
      "loss": 1.1621131896972656,
      "step": 1950
    },
    {
      "epoch": 0.5563440249787113,
      "grad_norm": 19.941749572753906,
      "learning_rate": 9.94445075219983e-05,
      "loss": 1.1394742965698241,
      "step": 1960
    },
    {
      "epoch": 0.559182514902072,
      "grad_norm": 21.390018463134766,
      "learning_rate": 9.944166903207494e-05,
      "loss": 1.1680900573730468,
      "step": 1970
    },
    {
      "epoch": 0.5620210048254328,
      "grad_norm": 24.322654724121094,
      "learning_rate": 9.943883054215158e-05,
      "loss": 1.0931670188903808,
      "step": 1980
    },
    {
      "epoch": 0.5648594947487936,
      "grad_norm": 18.85932731628418,
      "learning_rate": 9.943599205222822e-05,
      "loss": 1.1821538925170898,
      "step": 1990
    },
    {
      "epoch": 0.5676979846721544,
      "grad_norm": 21.754899978637695,
      "learning_rate": 9.943315356230486e-05,
      "loss": 1.163304901123047,
      "step": 2000
    },
    {
      "epoch": 0.5676979846721544,
      "eval_accuracy": 0.5731544477649901,
      "eval_loss": 1.4467488527297974,
      "eval_runtime": 31.4133,
      "eval_samples_per_second": 500.647,
      "eval_steps_per_second": 7.831,
      "step": 2000
    },
    {
      "epoch": 0.5705364745955152,
      "grad_norm": 22.230655670166016,
      "learning_rate": 9.94303150723815e-05,
      "loss": 1.1267522811889648,
      "step": 2010
    },
    {
      "epoch": 0.5733749645188759,
      "grad_norm": 27.563358306884766,
      "learning_rate": 9.942747658245813e-05,
      "loss": 1.1471932411193848,
      "step": 2020
    },
    {
      "epoch": 0.5762134544422367,
      "grad_norm": 23.25899887084961,
      "learning_rate": 9.942463809253477e-05,
      "loss": 1.12424898147583,
      "step": 2030
    },
    {
      "epoch": 0.5790519443655975,
      "grad_norm": 24.0449161529541,
      "learning_rate": 9.942179960261143e-05,
      "loss": 1.0815184593200684,
      "step": 2040
    },
    {
      "epoch": 0.5818904342889583,
      "grad_norm": 33.145957946777344,
      "learning_rate": 9.941896111268806e-05,
      "loss": 1.100113582611084,
      "step": 2050
    },
    {
      "epoch": 0.584728924212319,
      "grad_norm": 18.245145797729492,
      "learning_rate": 9.94161226227647e-05,
      "loss": 1.14227352142334,
      "step": 2060
    },
    {
      "epoch": 0.5875674141356798,
      "grad_norm": 24.452871322631836,
      "learning_rate": 9.941328413284134e-05,
      "loss": 1.0578697204589844,
      "step": 2070
    },
    {
      "epoch": 0.5904059040590406,
      "grad_norm": 21.05449867248535,
      "learning_rate": 9.941044564291797e-05,
      "loss": 1.0928358078002929,
      "step": 2080
    },
    {
      "epoch": 0.5932443939824014,
      "grad_norm": 19.354047775268555,
      "learning_rate": 9.940760715299461e-05,
      "loss": 1.088101291656494,
      "step": 2090
    },
    {
      "epoch": 0.5960828839057621,
      "grad_norm": 23.127643585205078,
      "learning_rate": 9.940476866307125e-05,
      "loss": 1.1032953262329102,
      "step": 2100
    },
    {
      "epoch": 0.5989213738291229,
      "grad_norm": 23.332597732543945,
      "learning_rate": 9.940193017314789e-05,
      "loss": 1.1219491958618164,
      "step": 2110
    },
    {
      "epoch": 0.6017598637524837,
      "grad_norm": 27.56525421142578,
      "learning_rate": 9.939909168322453e-05,
      "loss": 1.1679291725158691,
      "step": 2120
    },
    {
      "epoch": 0.6045983536758445,
      "grad_norm": 25.26095962524414,
      "learning_rate": 9.939625319330117e-05,
      "loss": 1.0028508186340332,
      "step": 2130
    },
    {
      "epoch": 0.6074368435992052,
      "grad_norm": 22.253829956054688,
      "learning_rate": 9.939341470337781e-05,
      "loss": 1.0808586120605468,
      "step": 2140
    },
    {
      "epoch": 0.610275333522566,
      "grad_norm": 24.0926456451416,
      "learning_rate": 9.939057621345444e-05,
      "loss": 1.0530101776123046,
      "step": 2150
    },
    {
      "epoch": 0.6131138234459268,
      "grad_norm": 24.591312408447266,
      "learning_rate": 9.938773772353108e-05,
      "loss": 1.057276725769043,
      "step": 2160
    },
    {
      "epoch": 0.6159523133692876,
      "grad_norm": 32.76468276977539,
      "learning_rate": 9.938489923360772e-05,
      "loss": 1.0916509628295898,
      "step": 2170
    },
    {
      "epoch": 0.6187908032926484,
      "grad_norm": 37.99615478515625,
      "learning_rate": 9.938206074368437e-05,
      "loss": 1.0856529235839845,
      "step": 2180
    },
    {
      "epoch": 0.6216292932160091,
      "grad_norm": 29.176586151123047,
      "learning_rate": 9.937922225376101e-05,
      "loss": 1.0997905731201172,
      "step": 2190
    },
    {
      "epoch": 0.6244677831393699,
      "grad_norm": 28.28167152404785,
      "learning_rate": 9.937638376383765e-05,
      "loss": 1.0920446395874024,
      "step": 2200
    },
    {
      "epoch": 0.6273062730627307,
      "grad_norm": 19.030431747436523,
      "learning_rate": 9.937354527391428e-05,
      "loss": 1.0603958129882813,
      "step": 2210
    },
    {
      "epoch": 0.6301447629860913,
      "grad_norm": 18.90097427368164,
      "learning_rate": 9.937070678399092e-05,
      "loss": 1.085594654083252,
      "step": 2220
    },
    {
      "epoch": 0.6329832529094521,
      "grad_norm": 41.02661895751953,
      "learning_rate": 9.936786829406756e-05,
      "loss": 1.0644072532653808,
      "step": 2230
    },
    {
      "epoch": 0.6358217428328129,
      "grad_norm": 28.411516189575195,
      "learning_rate": 9.93650298041442e-05,
      "loss": 1.0713820457458496,
      "step": 2240
    },
    {
      "epoch": 0.6386602327561737,
      "grad_norm": 22.9285945892334,
      "learning_rate": 9.936219131422084e-05,
      "loss": 1.0817390441894532,
      "step": 2250
    },
    {
      "epoch": 0.6414987226795344,
      "grad_norm": 28.542499542236328,
      "learning_rate": 9.935935282429748e-05,
      "loss": 1.1385957717895507,
      "step": 2260
    },
    {
      "epoch": 0.6443372126028952,
      "grad_norm": 21.064485549926758,
      "learning_rate": 9.935651433437411e-05,
      "loss": 1.0123428344726562,
      "step": 2270
    },
    {
      "epoch": 0.647175702526256,
      "grad_norm": 22.37677001953125,
      "learning_rate": 9.935367584445075e-05,
      "loss": 1.0572824478149414,
      "step": 2280
    },
    {
      "epoch": 0.6500141924496168,
      "grad_norm": 20.43739128112793,
      "learning_rate": 9.93508373545274e-05,
      "loss": 1.0530651092529297,
      "step": 2290
    },
    {
      "epoch": 0.6528526823729776,
      "grad_norm": 30.628801345825195,
      "learning_rate": 9.934799886460404e-05,
      "loss": 1.0473257064819337,
      "step": 2300
    },
    {
      "epoch": 0.6556911722963383,
      "grad_norm": 31.006044387817383,
      "learning_rate": 9.934516037468068e-05,
      "loss": 1.0243597030639648,
      "step": 2310
    },
    {
      "epoch": 0.6585296622196991,
      "grad_norm": 24.01661491394043,
      "learning_rate": 9.934232188475732e-05,
      "loss": 1.0115116119384766,
      "step": 2320
    },
    {
      "epoch": 0.6613681521430599,
      "grad_norm": 20.445096969604492,
      "learning_rate": 9.933948339483396e-05,
      "loss": 1.0294639587402343,
      "step": 2330
    },
    {
      "epoch": 0.6642066420664207,
      "grad_norm": 20.086233139038086,
      "learning_rate": 9.933664490491059e-05,
      "loss": 1.017643356323242,
      "step": 2340
    },
    {
      "epoch": 0.6670451319897814,
      "grad_norm": 23.536048889160156,
      "learning_rate": 9.933380641498723e-05,
      "loss": 1.0868244171142578,
      "step": 2350
    },
    {
      "epoch": 0.6698836219131422,
      "grad_norm": 26.774625778198242,
      "learning_rate": 9.933096792506387e-05,
      "loss": 1.0430272102355957,
      "step": 2360
    },
    {
      "epoch": 0.672722111836503,
      "grad_norm": 32.199214935302734,
      "learning_rate": 9.93281294351405e-05,
      "loss": 0.9647135734558105,
      "step": 2370
    },
    {
      "epoch": 0.6755606017598638,
      "grad_norm": 28.380857467651367,
      "learning_rate": 9.932529094521715e-05,
      "loss": 0.9824605941772461,
      "step": 2380
    },
    {
      "epoch": 0.6783990916832245,
      "grad_norm": 26.344202041625977,
      "learning_rate": 9.93224524552938e-05,
      "loss": 0.9529353141784668,
      "step": 2390
    },
    {
      "epoch": 0.6812375816065853,
      "grad_norm": 19.980390548706055,
      "learning_rate": 9.931961396537042e-05,
      "loss": 1.0057503700256347,
      "step": 2400
    },
    {
      "epoch": 0.6840760715299461,
      "grad_norm": 24.75773811340332,
      "learning_rate": 9.931677547544706e-05,
      "loss": 1.0897193908691407,
      "step": 2410
    },
    {
      "epoch": 0.6869145614533069,
      "grad_norm": 31.166263580322266,
      "learning_rate": 9.93139369855237e-05,
      "loss": 0.9823177337646485,
      "step": 2420
    },
    {
      "epoch": 0.6897530513766676,
      "grad_norm": 14.208596229553223,
      "learning_rate": 9.931109849560035e-05,
      "loss": 0.998142147064209,
      "step": 2430
    },
    {
      "epoch": 0.6925915413000284,
      "grad_norm": 22.746000289916992,
      "learning_rate": 9.930826000567699e-05,
      "loss": 1.0114707946777344,
      "step": 2440
    },
    {
      "epoch": 0.6954300312233892,
      "grad_norm": 25.27461051940918,
      "learning_rate": 9.930542151575363e-05,
      "loss": 0.9773696899414063,
      "step": 2450
    },
    {
      "epoch": 0.69826852114675,
      "grad_norm": 22.146265029907227,
      "learning_rate": 9.930258302583027e-05,
      "loss": 1.0271379470825195,
      "step": 2460
    },
    {
      "epoch": 0.7011070110701108,
      "grad_norm": 19.95235824584961,
      "learning_rate": 9.92997445359069e-05,
      "loss": 0.9382213592529297,
      "step": 2470
    },
    {
      "epoch": 0.7039455009934714,
      "grad_norm": 16.78559684753418,
      "learning_rate": 9.929690604598354e-05,
      "loss": 0.9645964622497558,
      "step": 2480
    },
    {
      "epoch": 0.7067839909168322,
      "grad_norm": 18.311519622802734,
      "learning_rate": 9.929406755606018e-05,
      "loss": 0.9811866760253907,
      "step": 2490
    },
    {
      "epoch": 0.709622480840193,
      "grad_norm": 34.024078369140625,
      "learning_rate": 9.929122906613681e-05,
      "loss": 0.9456801414489746,
      "step": 2500
    },
    {
      "epoch": 0.709622480840193,
      "eval_accuracy": 0.6204616265021937,
      "eval_loss": 1.2559458017349243,
      "eval_runtime": 30.558,
      "eval_samples_per_second": 514.661,
      "eval_steps_per_second": 8.05,
      "step": 2500
    },
    {
      "epoch": 0.7124609707635537,
      "grad_norm": 24.287723541259766,
      "learning_rate": 9.928839057621346e-05,
      "loss": 0.9708761215209961,
      "step": 2510
    },
    {
      "epoch": 0.7152994606869145,
      "grad_norm": 22.299070358276367,
      "learning_rate": 9.92855520862901e-05,
      "loss": 1.0146733283996583,
      "step": 2520
    },
    {
      "epoch": 0.7181379506102753,
      "grad_norm": 23.246183395385742,
      "learning_rate": 9.928271359636673e-05,
      "loss": 1.0145591735839843,
      "step": 2530
    },
    {
      "epoch": 0.7209764405336361,
      "grad_norm": 30.1379337310791,
      "learning_rate": 9.927987510644337e-05,
      "loss": 0.9871024131774903,
      "step": 2540
    },
    {
      "epoch": 0.7238149304569969,
      "grad_norm": 28.686357498168945,
      "learning_rate": 9.927703661652002e-05,
      "loss": 0.962646770477295,
      "step": 2550
    },
    {
      "epoch": 0.7266534203803576,
      "grad_norm": 22.367624282836914,
      "learning_rate": 9.927419812659666e-05,
      "loss": 0.9451247215270996,
      "step": 2560
    },
    {
      "epoch": 0.7294919103037184,
      "grad_norm": 23.129642486572266,
      "learning_rate": 9.92713596366733e-05,
      "loss": 0.9488322257995605,
      "step": 2570
    },
    {
      "epoch": 0.7323304002270792,
      "grad_norm": 19.723764419555664,
      "learning_rate": 9.926852114674994e-05,
      "loss": 0.9426379203796387,
      "step": 2580
    },
    {
      "epoch": 0.73516889015044,
      "grad_norm": 24.903884887695312,
      "learning_rate": 9.926568265682658e-05,
      "loss": 0.9138444900512696,
      "step": 2590
    },
    {
      "epoch": 0.7380073800738007,
      "grad_norm": 34.61288070678711,
      "learning_rate": 9.926284416690321e-05,
      "loss": 0.95067138671875,
      "step": 2600
    },
    {
      "epoch": 0.7408458699971615,
      "grad_norm": 23.43537139892578,
      "learning_rate": 9.926000567697985e-05,
      "loss": 0.9514213562011719,
      "step": 2610
    },
    {
      "epoch": 0.7436843599205223,
      "grad_norm": 20.223928451538086,
      "learning_rate": 9.925716718705649e-05,
      "loss": 0.9066153526306152,
      "step": 2620
    },
    {
      "epoch": 0.7465228498438831,
      "grad_norm": 30.943580627441406,
      "learning_rate": 9.925432869713312e-05,
      "loss": 0.9603883743286132,
      "step": 2630
    },
    {
      "epoch": 0.7493613397672438,
      "grad_norm": 26.97696876525879,
      "learning_rate": 9.925149020720977e-05,
      "loss": 0.8883302688598633,
      "step": 2640
    },
    {
      "epoch": 0.7521998296906046,
      "grad_norm": 28.247447967529297,
      "learning_rate": 9.924865171728642e-05,
      "loss": 0.9468140602111816,
      "step": 2650
    },
    {
      "epoch": 0.7550383196139654,
      "grad_norm": 33.24482727050781,
      "learning_rate": 9.924581322736304e-05,
      "loss": 0.8988624572753906,
      "step": 2660
    },
    {
      "epoch": 0.7578768095373262,
      "grad_norm": 21.37324333190918,
      "learning_rate": 9.924297473743968e-05,
      "loss": 0.9730520248413086,
      "step": 2670
    },
    {
      "epoch": 0.760715299460687,
      "grad_norm": 21.114492416381836,
      "learning_rate": 9.924013624751633e-05,
      "loss": 0.9491323471069336,
      "step": 2680
    },
    {
      "epoch": 0.7635537893840477,
      "grad_norm": 16.801727294921875,
      "learning_rate": 9.923729775759297e-05,
      "loss": 0.8683090209960938,
      "step": 2690
    },
    {
      "epoch": 0.7663922793074085,
      "grad_norm": 29.216243743896484,
      "learning_rate": 9.92344592676696e-05,
      "loss": 0.9094600677490234,
      "step": 2700
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 35.45439910888672,
      "learning_rate": 9.923162077774625e-05,
      "loss": 0.9245884895324707,
      "step": 2710
    },
    {
      "epoch": 0.77206925915413,
      "grad_norm": 23.179960250854492,
      "learning_rate": 9.922878228782289e-05,
      "loss": 0.9042709350585938,
      "step": 2720
    },
    {
      "epoch": 0.7749077490774908,
      "grad_norm": 16.718400955200195,
      "learning_rate": 9.922594379789952e-05,
      "loss": 0.8826107978820801,
      "step": 2730
    },
    {
      "epoch": 0.7777462390008516,
      "grad_norm": 24.61229705810547,
      "learning_rate": 9.922310530797616e-05,
      "loss": 0.9001324653625489,
      "step": 2740
    },
    {
      "epoch": 0.7805847289242123,
      "grad_norm": 30.58627700805664,
      "learning_rate": 9.92202668180528e-05,
      "loss": 0.8385726928710937,
      "step": 2750
    },
    {
      "epoch": 0.783423218847573,
      "grad_norm": 16.988853454589844,
      "learning_rate": 9.921742832812943e-05,
      "loss": 0.8284055709838867,
      "step": 2760
    },
    {
      "epoch": 0.7862617087709338,
      "grad_norm": 28.70233917236328,
      "learning_rate": 9.921458983820609e-05,
      "loss": 0.8698087692260742,
      "step": 2770
    },
    {
      "epoch": 0.7891001986942946,
      "grad_norm": 19.30978012084961,
      "learning_rate": 9.921175134828273e-05,
      "loss": 0.8799326896667481,
      "step": 2780
    },
    {
      "epoch": 0.7919386886176554,
      "grad_norm": 20.468929290771484,
      "learning_rate": 9.920891285835935e-05,
      "loss": 0.8119211196899414,
      "step": 2790
    },
    {
      "epoch": 0.7947771785410161,
      "grad_norm": 33.029483795166016,
      "learning_rate": 9.9206074368436e-05,
      "loss": 0.8412540435791016,
      "step": 2800
    },
    {
      "epoch": 0.7976156684643769,
      "grad_norm": 24.97670555114746,
      "learning_rate": 9.920323587851264e-05,
      "loss": 0.8865632057189942,
      "step": 2810
    },
    {
      "epoch": 0.8004541583877377,
      "grad_norm": 28.665922164916992,
      "learning_rate": 9.920039738858928e-05,
      "loss": 0.9044490814208984,
      "step": 2820
    },
    {
      "epoch": 0.8032926483110985,
      "grad_norm": 21.4758358001709,
      "learning_rate": 9.91975588986659e-05,
      "loss": 0.8875882148742675,
      "step": 2830
    },
    {
      "epoch": 0.8061311382344593,
      "grad_norm": 26.156322479248047,
      "learning_rate": 9.919472040874256e-05,
      "loss": 0.8137760162353516,
      "step": 2840
    },
    {
      "epoch": 0.80896962815782,
      "grad_norm": 17.85251808166504,
      "learning_rate": 9.91918819188192e-05,
      "loss": 0.9045653343200684,
      "step": 2850
    },
    {
      "epoch": 0.8118081180811808,
      "grad_norm": 24.476266860961914,
      "learning_rate": 9.918904342889583e-05,
      "loss": 0.8285739898681641,
      "step": 2860
    },
    {
      "epoch": 0.8146466080045416,
      "grad_norm": 19.197765350341797,
      "learning_rate": 9.918620493897247e-05,
      "loss": 0.9241982460021972,
      "step": 2870
    },
    {
      "epoch": 0.8174850979279024,
      "grad_norm": 22.9984188079834,
      "learning_rate": 9.918336644904911e-05,
      "loss": 0.8890705108642578,
      "step": 2880
    },
    {
      "epoch": 0.8203235878512631,
      "grad_norm": 23.804393768310547,
      "learning_rate": 9.918052795912574e-05,
      "loss": 0.8551264762878418,
      "step": 2890
    },
    {
      "epoch": 0.8231620777746239,
      "grad_norm": 27.835195541381836,
      "learning_rate": 9.917768946920238e-05,
      "loss": 0.8502161026000976,
      "step": 2900
    },
    {
      "epoch": 0.8260005676979847,
      "grad_norm": 21.51314353942871,
      "learning_rate": 9.917485097927904e-05,
      "loss": 0.7782979011535645,
      "step": 2910
    },
    {
      "epoch": 0.8288390576213455,
      "grad_norm": 27.626466751098633,
      "learning_rate": 9.917201248935566e-05,
      "loss": 0.8109992027282715,
      "step": 2920
    },
    {
      "epoch": 0.8316775475447062,
      "grad_norm": 21.60063362121582,
      "learning_rate": 9.91691739994323e-05,
      "loss": 0.8673000335693359,
      "step": 2930
    },
    {
      "epoch": 0.834516037468067,
      "grad_norm": 19.806446075439453,
      "learning_rate": 9.916633550950895e-05,
      "loss": 0.8503800392150879,
      "step": 2940
    },
    {
      "epoch": 0.8373545273914278,
      "grad_norm": 19.74971580505371,
      "learning_rate": 9.916349701958559e-05,
      "loss": 0.8043561935424804,
      "step": 2950
    },
    {
      "epoch": 0.8401930173147886,
      "grad_norm": 29.61273956298828,
      "learning_rate": 9.916065852966222e-05,
      "loss": 0.7918411731719971,
      "step": 2960
    },
    {
      "epoch": 0.8430315072381493,
      "grad_norm": 16.50292205810547,
      "learning_rate": 9.915782003973887e-05,
      "loss": 0.7851666450500489,
      "step": 2970
    },
    {
      "epoch": 0.8458699971615101,
      "grad_norm": 21.183700561523438,
      "learning_rate": 9.915498154981551e-05,
      "loss": 0.810921573638916,
      "step": 2980
    },
    {
      "epoch": 0.8487084870848709,
      "grad_norm": 26.505327224731445,
      "learning_rate": 9.915214305989214e-05,
      "loss": 0.791579294204712,
      "step": 2990
    },
    {
      "epoch": 0.8515469770082317,
      "grad_norm": Infinity,
      "learning_rate": 9.914930456996878e-05,
      "loss": 0.7944314002990722,
      "step": 3000
    },
    {
      "epoch": 0.8515469770082317,
      "eval_accuracy": 0.6640172950976029,
      "eval_loss": 1.1031973361968994,
      "eval_runtime": 30.454,
      "eval_samples_per_second": 516.418,
      "eval_steps_per_second": 8.078,
      "step": 3000
    },
    {
      "epoch": 0.8543854669315923,
      "grad_norm": 20.32843017578125,
      "learning_rate": 9.914674992903775e-05,
      "loss": 0.7646556377410889,
      "step": 3010
    },
    {
      "epoch": 0.8572239568549531,
      "grad_norm": 22.286470413208008,
      "learning_rate": 9.91439114391144e-05,
      "loss": 0.792286729812622,
      "step": 3020
    },
    {
      "epoch": 0.8600624467783139,
      "grad_norm": 19.36547088623047,
      "learning_rate": 9.914107294919105e-05,
      "loss": 0.8225823402404785,
      "step": 3030
    },
    {
      "epoch": 0.8629009367016747,
      "grad_norm": 21.49129867553711,
      "learning_rate": 9.913823445926767e-05,
      "loss": 0.8195921897888183,
      "step": 3040
    },
    {
      "epoch": 0.8657394266250354,
      "grad_norm": 25.84406852722168,
      "learning_rate": 9.913539596934431e-05,
      "loss": 0.8258531570434571,
      "step": 3050
    },
    {
      "epoch": 0.8685779165483962,
      "grad_norm": 21.42317771911621,
      "learning_rate": 9.913255747942096e-05,
      "loss": 0.7712396621704102,
      "step": 3060
    },
    {
      "epoch": 0.871416406471757,
      "grad_norm": 20.756650924682617,
      "learning_rate": 9.912971898949758e-05,
      "loss": 0.8155352592468261,
      "step": 3070
    },
    {
      "epoch": 0.8742548963951178,
      "grad_norm": 24.09754753112793,
      "learning_rate": 9.912688049957422e-05,
      "loss": 0.7634074211120605,
      "step": 3080
    },
    {
      "epoch": 0.8770933863184786,
      "grad_norm": 28.720779418945312,
      "learning_rate": 9.912404200965088e-05,
      "loss": 0.7250977039337159,
      "step": 3090
    },
    {
      "epoch": 0.8799318762418393,
      "grad_norm": 22.041954040527344,
      "learning_rate": 9.912120351972751e-05,
      "loss": 0.7625031471252441,
      "step": 3100
    },
    {
      "epoch": 0.8827703661652001,
      "grad_norm": 21.66750144958496,
      "learning_rate": 9.911836502980415e-05,
      "loss": 0.7681513786315918,
      "step": 3110
    },
    {
      "epoch": 0.8856088560885609,
      "grad_norm": 23.560832977294922,
      "learning_rate": 9.911552653988079e-05,
      "loss": 0.7657046318054199,
      "step": 3120
    },
    {
      "epoch": 0.8884473460119217,
      "grad_norm": 28.501028060913086,
      "learning_rate": 9.911268804995743e-05,
      "loss": 0.782001543045044,
      "step": 3130
    },
    {
      "epoch": 0.8912858359352824,
      "grad_norm": 19.86785888671875,
      "learning_rate": 9.910984956003406e-05,
      "loss": 0.8034740447998047,
      "step": 3140
    },
    {
      "epoch": 0.8941243258586432,
      "grad_norm": 23.600494384765625,
      "learning_rate": 9.910701107011071e-05,
      "loss": 0.8363421440124512,
      "step": 3150
    },
    {
      "epoch": 0.896962815782004,
      "grad_norm": 18.663354873657227,
      "learning_rate": 9.910417258018734e-05,
      "loss": 0.7630080223083496,
      "step": 3160
    },
    {
      "epoch": 0.8998013057053648,
      "grad_norm": 29.64305305480957,
      "learning_rate": 9.910133409026398e-05,
      "loss": 0.7915078639984131,
      "step": 3170
    },
    {
      "epoch": 0.9026397956287255,
      "grad_norm": 32.460205078125,
      "learning_rate": 9.909849560034063e-05,
      "loss": 0.8089962959289551,
      "step": 3180
    },
    {
      "epoch": 0.9054782855520863,
      "grad_norm": 20.414539337158203,
      "learning_rate": 9.909565711041727e-05,
      "loss": 0.7740297794342041,
      "step": 3190
    },
    {
      "epoch": 0.9083167754754471,
      "grad_norm": 33.839813232421875,
      "learning_rate": 9.90928186204939e-05,
      "loss": 0.752277946472168,
      "step": 3200
    },
    {
      "epoch": 0.9111552653988079,
      "grad_norm": 18.867141723632812,
      "learning_rate": 9.909026397956287e-05,
      "loss": 0.7238951683044433,
      "step": 3210
    },
    {
      "epoch": 0.9139937553221686,
      "grad_norm": 22.304485321044922,
      "learning_rate": 9.908742548963952e-05,
      "loss": 0.7545382976531982,
      "step": 3220
    },
    {
      "epoch": 0.9168322452455294,
      "grad_norm": 33.62503433227539,
      "learning_rate": 9.908458699971616e-05,
      "loss": 0.8270857810974122,
      "step": 3230
    },
    {
      "epoch": 0.9196707351688902,
      "grad_norm": 18.09872817993164,
      "learning_rate": 9.90817485097928e-05,
      "loss": 0.7963118553161621,
      "step": 3240
    },
    {
      "epoch": 0.922509225092251,
      "grad_norm": 22.346670150756836,
      "learning_rate": 9.907891001986943e-05,
      "loss": 0.7598023414611816,
      "step": 3250
    },
    {
      "epoch": 0.9253477150156117,
      "grad_norm": 24.197389602661133,
      "learning_rate": 9.907607152994607e-05,
      "loss": 0.7657143115997315,
      "step": 3260
    },
    {
      "epoch": 0.9281862049389724,
      "grad_norm": 26.21255111694336,
      "learning_rate": 9.907323304002272e-05,
      "loss": 0.7388171195983887,
      "step": 3270
    },
    {
      "epoch": 0.9310246948623332,
      "grad_norm": 28.20396614074707,
      "learning_rate": 9.907039455009935e-05,
      "loss": 0.7233112812042236,
      "step": 3280
    },
    {
      "epoch": 0.933863184785694,
      "grad_norm": 23.623817443847656,
      "learning_rate": 9.906755606017599e-05,
      "loss": 0.7306665897369384,
      "step": 3290
    },
    {
      "epoch": 0.9367016747090547,
      "grad_norm": 23.831466674804688,
      "learning_rate": 9.906471757025263e-05,
      "loss": 0.8213624000549317,
      "step": 3300
    },
    {
      "epoch": 0.9395401646324155,
      "grad_norm": 22.057376861572266,
      "learning_rate": 9.906187908032926e-05,
      "loss": 0.7966175556182862,
      "step": 3310
    },
    {
      "epoch": 0.9423786545557763,
      "grad_norm": 20.871597290039062,
      "learning_rate": 9.90590405904059e-05,
      "loss": 0.7152708053588868,
      "step": 3320
    },
    {
      "epoch": 0.9452171444791371,
      "grad_norm": 18.533130645751953,
      "learning_rate": 9.905620210048256e-05,
      "loss": 0.7805844783782959,
      "step": 3330
    },
    {
      "epoch": 0.9480556344024978,
      "grad_norm": 26.149999618530273,
      "learning_rate": 9.905336361055918e-05,
      "loss": 0.7450548648834229,
      "step": 3340
    },
    {
      "epoch": 0.9508941243258586,
      "grad_norm": 24.809926986694336,
      "learning_rate": 9.905052512063583e-05,
      "loss": 0.7699018955230713,
      "step": 3350
    },
    {
      "epoch": 0.9537326142492194,
      "grad_norm": 34.5396728515625,
      "learning_rate": 9.904768663071247e-05,
      "loss": 0.779564380645752,
      "step": 3360
    },
    {
      "epoch": 0.9565711041725802,
      "grad_norm": 23.306936264038086,
      "learning_rate": 9.904484814078911e-05,
      "loss": 0.7437909126281739,
      "step": 3370
    },
    {
      "epoch": 0.959409594095941,
      "grad_norm": 21.1256103515625,
      "learning_rate": 9.904200965086574e-05,
      "loss": 0.7724534034729004,
      "step": 3380
    },
    {
      "epoch": 0.9622480840193017,
      "grad_norm": 34.26607894897461,
      "learning_rate": 9.903917116094238e-05,
      "loss": 0.7829182147979736,
      "step": 3390
    },
    {
      "epoch": 0.9650865739426625,
      "grad_norm": 25.866369247436523,
      "learning_rate": 9.903633267101903e-05,
      "loss": 0.7695268630981446,
      "step": 3400
    },
    {
      "epoch": 0.9679250638660233,
      "grad_norm": 23.50071907043457,
      "learning_rate": 9.903349418109566e-05,
      "loss": 0.7211739540100097,
      "step": 3410
    },
    {
      "epoch": 0.9707635537893841,
      "grad_norm": 27.50710105895996,
      "learning_rate": 9.90306556911723e-05,
      "loss": 0.7612006187438964,
      "step": 3420
    },
    {
      "epoch": 0.9736020437127448,
      "grad_norm": 27.45734405517578,
      "learning_rate": 9.902781720124894e-05,
      "loss": 0.6757657527923584,
      "step": 3430
    },
    {
      "epoch": 0.9764405336361056,
      "grad_norm": 29.132966995239258,
      "learning_rate": 9.902497871132557e-05,
      "loss": 0.7574927806854248,
      "step": 3440
    },
    {
      "epoch": 0.9792790235594664,
      "grad_norm": 29.643259048461914,
      "learning_rate": 9.902214022140221e-05,
      "loss": 0.7482147693634034,
      "step": 3450
    },
    {
      "epoch": 0.9821175134828272,
      "grad_norm": 21.631839752197266,
      "learning_rate": 9.901930173147885e-05,
      "loss": 0.7287088871002197,
      "step": 3460
    },
    {
      "epoch": 0.9849560034061879,
      "grad_norm": 19.11096954345703,
      "learning_rate": 9.90164632415555e-05,
      "loss": 0.6649835586547852,
      "step": 3470
    },
    {
      "epoch": 0.9877944933295487,
      "grad_norm": 22.88902473449707,
      "learning_rate": 9.901362475163214e-05,
      "loss": 0.7077507972717285,
      "step": 3480
    },
    {
      "epoch": 0.9906329832529095,
      "grad_norm": 21.8216609954834,
      "learning_rate": 9.901078626170878e-05,
      "loss": 0.7006841182708741,
      "step": 3490
    },
    {
      "epoch": 0.9934714731762703,
      "grad_norm": 23.871532440185547,
      "learning_rate": 9.900794777178542e-05,
      "loss": 0.705707597732544,
      "step": 3500
    },
    {
      "epoch": 0.9934714731762703,
      "eval_accuracy": 0.6953010745851084,
      "eval_loss": 0.980442464351654,
      "eval_runtime": 30.9652,
      "eval_samples_per_second": 507.892,
      "eval_steps_per_second": 7.944,
      "step": 3500
    },
    {
      "epoch": 0.996309963099631,
      "grad_norm": 19.31228256225586,
      "learning_rate": 9.900510928186205e-05,
      "loss": 0.6550156593322753,
      "step": 3510
    },
    {
      "epoch": 0.9991484530229918,
      "grad_norm": 19.09356689453125,
      "learning_rate": 9.900227079193869e-05,
      "loss": 0.6604233264923096,
      "step": 3520
    },
    {
      "epoch": 1.0019869429463526,
      "grad_norm": 28.049551010131836,
      "learning_rate": 9.899943230201534e-05,
      "loss": 0.6727694511413574,
      "step": 3530
    },
    {
      "epoch": 1.0048254328697133,
      "grad_norm": 26.254486083984375,
      "learning_rate": 9.899659381209197e-05,
      "loss": 0.6348094940185547,
      "step": 3540
    },
    {
      "epoch": 1.0076639227930742,
      "grad_norm": 25.290700912475586,
      "learning_rate": 9.899375532216861e-05,
      "loss": 0.651258897781372,
      "step": 3550
    },
    {
      "epoch": 1.0105024127164348,
      "grad_norm": 20.58706283569336,
      "learning_rate": 9.899091683224525e-05,
      "loss": 0.6630977630615235,
      "step": 3560
    },
    {
      "epoch": 1.0133409026397957,
      "grad_norm": 33.84420394897461,
      "learning_rate": 9.898807834232188e-05,
      "loss": 0.6907260417938232,
      "step": 3570
    },
    {
      "epoch": 1.0161793925631564,
      "grad_norm": 22.32529640197754,
      "learning_rate": 9.898523985239852e-05,
      "loss": 0.6630112648010253,
      "step": 3580
    },
    {
      "epoch": 1.0190178824865173,
      "grad_norm": 31.985055923461914,
      "learning_rate": 9.898240136247517e-05,
      "loss": 0.7523481369018554,
      "step": 3590
    },
    {
      "epoch": 1.021856372409878,
      "grad_norm": 26.934823989868164,
      "learning_rate": 9.89795628725518e-05,
      "loss": 0.693488073348999,
      "step": 3600
    },
    {
      "epoch": 1.0246948623332388,
      "grad_norm": 30.164888381958008,
      "learning_rate": 9.897672438262845e-05,
      "loss": 0.6415844917297363,
      "step": 3610
    },
    {
      "epoch": 1.0275333522565995,
      "grad_norm": 18.834300994873047,
      "learning_rate": 9.897388589270509e-05,
      "loss": 0.6850387573242187,
      "step": 3620
    },
    {
      "epoch": 1.0303718421799604,
      "grad_norm": 21.231327056884766,
      "learning_rate": 9.897104740278173e-05,
      "loss": 0.6446299076080322,
      "step": 3630
    },
    {
      "epoch": 1.033210332103321,
      "grad_norm": 23.240516662597656,
      "learning_rate": 9.896820891285836e-05,
      "loss": 0.6083054065704345,
      "step": 3640
    },
    {
      "epoch": 1.036048822026682,
      "grad_norm": 17.034574508666992,
      "learning_rate": 9.8965370422935e-05,
      "loss": 0.6667216300964356,
      "step": 3650
    },
    {
      "epoch": 1.0388873119500426,
      "grad_norm": 21.844274520874023,
      "learning_rate": 9.896253193301164e-05,
      "loss": 0.663185977935791,
      "step": 3660
    },
    {
      "epoch": 1.0417258018734032,
      "grad_norm": 27.365386962890625,
      "learning_rate": 9.895969344308828e-05,
      "loss": 0.6416357517242431,
      "step": 3670
    },
    {
      "epoch": 1.0445642917967641,
      "grad_norm": 19.69205093383789,
      "learning_rate": 9.895685495316492e-05,
      "loss": 0.6922329425811767,
      "step": 3680
    },
    {
      "epoch": 1.0474027817201248,
      "grad_norm": 18.737306594848633,
      "learning_rate": 9.895401646324157e-05,
      "loss": 0.5935844898223877,
      "step": 3690
    },
    {
      "epoch": 1.0502412716434857,
      "grad_norm": 26.93574333190918,
      "learning_rate": 9.895117797331819e-05,
      "loss": 0.713580322265625,
      "step": 3700
    },
    {
      "epoch": 1.0530797615668464,
      "grad_norm": 21.484880447387695,
      "learning_rate": 9.894833948339483e-05,
      "loss": 0.6020592212677002,
      "step": 3710
    },
    {
      "epoch": 1.0559182514902072,
      "grad_norm": 19.4040584564209,
      "learning_rate": 9.894550099347148e-05,
      "loss": 0.6169495582580566,
      "step": 3720
    },
    {
      "epoch": 1.058756741413568,
      "grad_norm": 23.88226318359375,
      "learning_rate": 9.894266250354812e-05,
      "loss": 0.6956416606903076,
      "step": 3730
    },
    {
      "epoch": 1.0615952313369288,
      "grad_norm": 16.01617431640625,
      "learning_rate": 9.893982401362476e-05,
      "loss": 0.6439888954162598,
      "step": 3740
    },
    {
      "epoch": 1.0644337212602895,
      "grad_norm": 19.912168502807617,
      "learning_rate": 9.89369855237014e-05,
      "loss": 0.6417697429656982,
      "step": 3750
    },
    {
      "epoch": 1.0672722111836503,
      "grad_norm": 19.506757736206055,
      "learning_rate": 9.893414703377804e-05,
      "loss": 0.6494669914245605,
      "step": 3760
    },
    {
      "epoch": 1.070110701107011,
      "grad_norm": 24.823179244995117,
      "learning_rate": 9.893130854385467e-05,
      "loss": 0.6445348739624024,
      "step": 3770
    },
    {
      "epoch": 1.072949191030372,
      "grad_norm": 15.19336986541748,
      "learning_rate": 9.892847005393131e-05,
      "loss": 0.6461805820465087,
      "step": 3780
    },
    {
      "epoch": 1.0757876809537326,
      "grad_norm": 27.75702476501465,
      "learning_rate": 9.892563156400795e-05,
      "loss": 0.5833687305450439,
      "step": 3790
    },
    {
      "epoch": 1.0786261708770934,
      "grad_norm": 24.98112678527832,
      "learning_rate": 9.89227930740846e-05,
      "loss": 0.5608679294586182,
      "step": 3800
    },
    {
      "epoch": 1.0814646608004541,
      "grad_norm": 27.189119338989258,
      "learning_rate": 9.891995458416123e-05,
      "loss": 0.6453812122344971,
      "step": 3810
    },
    {
      "epoch": 1.084303150723815,
      "grad_norm": 21.3582706451416,
      "learning_rate": 9.891711609423788e-05,
      "loss": 0.613520622253418,
      "step": 3820
    },
    {
      "epoch": 1.0871416406471757,
      "grad_norm": 21.198776245117188,
      "learning_rate": 9.89142776043145e-05,
      "loss": 0.6817501068115235,
      "step": 3830
    },
    {
      "epoch": 1.0899801305705366,
      "grad_norm": 23.9219970703125,
      "learning_rate": 9.891143911439115e-05,
      "loss": 0.6310290813446044,
      "step": 3840
    },
    {
      "epoch": 1.0928186204938972,
      "grad_norm": 15.712692260742188,
      "learning_rate": 9.890860062446779e-05,
      "loss": 0.6560925960540771,
      "step": 3850
    },
    {
      "epoch": 1.095657110417258,
      "grad_norm": 27.00271224975586,
      "learning_rate": 9.890576213454443e-05,
      "loss": 0.5857605457305908,
      "step": 3860
    },
    {
      "epoch": 1.0984956003406188,
      "grad_norm": 22.93274688720703,
      "learning_rate": 9.890292364462107e-05,
      "loss": 0.609201717376709,
      "step": 3870
    },
    {
      "epoch": 1.1013340902639797,
      "grad_norm": 17.351171493530273,
      "learning_rate": 9.890008515469771e-05,
      "loss": 0.6610034465789795,
      "step": 3880
    },
    {
      "epoch": 1.1041725801873403,
      "grad_norm": 29.707874298095703,
      "learning_rate": 9.889724666477435e-05,
      "loss": 0.666200828552246,
      "step": 3890
    },
    {
      "epoch": 1.1070110701107012,
      "grad_norm": 20.44675636291504,
      "learning_rate": 9.889440817485098e-05,
      "loss": 0.6511158466339111,
      "step": 3900
    },
    {
      "epoch": 1.1098495600340619,
      "grad_norm": 18.375123977661133,
      "learning_rate": 9.889156968492762e-05,
      "loss": 0.5879981994628907,
      "step": 3910
    },
    {
      "epoch": 1.1126880499574225,
      "grad_norm": 38.74342346191406,
      "learning_rate": 9.888873119500426e-05,
      "loss": 0.6463605403900147,
      "step": 3920
    },
    {
      "epoch": 1.1155265398807834,
      "grad_norm": 35.702579498291016,
      "learning_rate": 9.88858927050809e-05,
      "loss": 0.7014100551605225,
      "step": 3930
    },
    {
      "epoch": 1.118365029804144,
      "grad_norm": 20.338956832885742,
      "learning_rate": 9.888305421515755e-05,
      "loss": 0.6385445117950439,
      "step": 3940
    },
    {
      "epoch": 1.121203519727505,
      "grad_norm": 30.120508193969727,
      "learning_rate": 9.888021572523419e-05,
      "loss": 0.6235272884368896,
      "step": 3950
    },
    {
      "epoch": 1.1240420096508656,
      "grad_norm": 26.171878814697266,
      "learning_rate": 9.887737723531081e-05,
      "loss": 0.6421665191650391,
      "step": 3960
    },
    {
      "epoch": 1.1268804995742265,
      "grad_norm": 21.68423080444336,
      "learning_rate": 9.887453874538746e-05,
      "loss": 0.6641988277435302,
      "step": 3970
    },
    {
      "epoch": 1.1297189894975872,
      "grad_norm": 28.949514389038086,
      "learning_rate": 9.88717002554641e-05,
      "loss": 0.5635973930358886,
      "step": 3980
    },
    {
      "epoch": 1.132557479420948,
      "grad_norm": 19.748165130615234,
      "learning_rate": 9.886886176554074e-05,
      "loss": 0.5637505054473877,
      "step": 3990
    },
    {
      "epoch": 1.1353959693443088,
      "grad_norm": 17.690265655517578,
      "learning_rate": 9.886602327561738e-05,
      "loss": 0.5907973766326904,
      "step": 4000
    },
    {
      "epoch": 1.1353959693443088,
      "eval_accuracy": 0.7279201373434221,
      "eval_loss": 0.8885079622268677,
      "eval_runtime": 31.2115,
      "eval_samples_per_second": 503.885,
      "eval_steps_per_second": 7.882,
      "step": 4000
    },
    {
      "epoch": 1.1382344592676696,
      "grad_norm": 34.16864776611328,
      "learning_rate": 9.886318478569402e-05,
      "loss": 0.6118478298187255,
      "step": 4010
    },
    {
      "epoch": 1.1410729491910303,
      "grad_norm": 19.992345809936523,
      "learning_rate": 9.886034629577065e-05,
      "loss": 0.707521390914917,
      "step": 4020
    },
    {
      "epoch": 1.1439114391143912,
      "grad_norm": 27.760913848876953,
      "learning_rate": 9.885750780584729e-05,
      "loss": 0.5689629554748535,
      "step": 4030
    },
    {
      "epoch": 1.1467499290377519,
      "grad_norm": 30.910898208618164,
      "learning_rate": 9.885466931592393e-05,
      "loss": 0.6241819381713867,
      "step": 4040
    },
    {
      "epoch": 1.1495884189611127,
      "grad_norm": 21.5301513671875,
      "learning_rate": 9.885183082600057e-05,
      "loss": 0.5930325508117675,
      "step": 4050
    },
    {
      "epoch": 1.1524269088844734,
      "grad_norm": 37.531471252441406,
      "learning_rate": 9.88489923360772e-05,
      "loss": 0.6504276275634766,
      "step": 4060
    },
    {
      "epoch": 1.1552653988078343,
      "grad_norm": 16.46526336669922,
      "learning_rate": 9.884615384615386e-05,
      "loss": 0.6098124980926514,
      "step": 4070
    },
    {
      "epoch": 1.158103888731195,
      "grad_norm": 30.237173080444336,
      "learning_rate": 9.88433153562305e-05,
      "loss": 0.641084861755371,
      "step": 4080
    },
    {
      "epoch": 1.1609423786545559,
      "grad_norm": 26.651351928710938,
      "learning_rate": 9.884047686630713e-05,
      "loss": 0.6080304622650147,
      "step": 4090
    },
    {
      "epoch": 1.1637808685779165,
      "grad_norm": 25.55205726623535,
      "learning_rate": 9.883763837638377e-05,
      "loss": 0.5625734806060791,
      "step": 4100
    },
    {
      "epoch": 1.1666193585012774,
      "grad_norm": 26.443571090698242,
      "learning_rate": 9.883479988646041e-05,
      "loss": 0.5978858470916748,
      "step": 4110
    },
    {
      "epoch": 1.169457848424638,
      "grad_norm": 20.061460494995117,
      "learning_rate": 9.883196139653704e-05,
      "loss": 0.5961314678192139,
      "step": 4120
    },
    {
      "epoch": 1.172296338347999,
      "grad_norm": 17.195859909057617,
      "learning_rate": 9.882912290661369e-05,
      "loss": 0.5361915588378906,
      "step": 4130
    },
    {
      "epoch": 1.1751348282713596,
      "grad_norm": 32.27775955200195,
      "learning_rate": 9.882628441669033e-05,
      "loss": 0.5996068000793457,
      "step": 4140
    },
    {
      "epoch": 1.1779733181947205,
      "grad_norm": 24.170778274536133,
      "learning_rate": 9.882344592676696e-05,
      "loss": 0.5598326206207276,
      "step": 4150
    },
    {
      "epoch": 1.1808118081180812,
      "grad_norm": 20.161243438720703,
      "learning_rate": 9.88206074368436e-05,
      "loss": 0.5899753093719482,
      "step": 4160
    },
    {
      "epoch": 1.1836502980414418,
      "grad_norm": 25.8084659576416,
      "learning_rate": 9.881776894692024e-05,
      "loss": 0.648932933807373,
      "step": 4170
    },
    {
      "epoch": 1.1864887879648027,
      "grad_norm": 21.893795013427734,
      "learning_rate": 9.881493045699688e-05,
      "loss": 0.582167673110962,
      "step": 4180
    },
    {
      "epoch": 1.1893272778881636,
      "grad_norm": 24.662321090698242,
      "learning_rate": 9.881209196707351e-05,
      "loss": 0.6225980281829834,
      "step": 4190
    },
    {
      "epoch": 1.1921657678115243,
      "grad_norm": 27.636730194091797,
      "learning_rate": 9.880925347715017e-05,
      "loss": 0.6037321567535401,
      "step": 4200
    },
    {
      "epoch": 1.195004257734885,
      "grad_norm": 22.676822662353516,
      "learning_rate": 9.880641498722681e-05,
      "loss": 0.6260679721832275,
      "step": 4210
    },
    {
      "epoch": 1.1978427476582458,
      "grad_norm": 29.929536819458008,
      "learning_rate": 9.880357649730344e-05,
      "loss": 0.5525911331176758,
      "step": 4220
    },
    {
      "epoch": 1.2006812375816065,
      "grad_norm": 28.129732131958008,
      "learning_rate": 9.880073800738008e-05,
      "loss": 0.6073970317840576,
      "step": 4230
    },
    {
      "epoch": 1.2035197275049674,
      "grad_norm": 19.525175094604492,
      "learning_rate": 9.879789951745672e-05,
      "loss": 0.5686793327331543,
      "step": 4240
    },
    {
      "epoch": 1.206358217428328,
      "grad_norm": 17.232412338256836,
      "learning_rate": 9.879506102753335e-05,
      "loss": 0.5829248905181885,
      "step": 4250
    },
    {
      "epoch": 1.209196707351689,
      "grad_norm": 27.632844924926758,
      "learning_rate": 9.879222253760999e-05,
      "loss": 0.5319347858428956,
      "step": 4260
    },
    {
      "epoch": 1.2120351972750496,
      "grad_norm": 24.55426025390625,
      "learning_rate": 9.878938404768664e-05,
      "loss": 0.5863132476806641,
      "step": 4270
    },
    {
      "epoch": 1.2148736871984105,
      "grad_norm": 23.3425350189209,
      "learning_rate": 9.878654555776327e-05,
      "loss": 0.6243361949920654,
      "step": 4280
    },
    {
      "epoch": 1.2177121771217712,
      "grad_norm": 30.333187103271484,
      "learning_rate": 9.878370706783991e-05,
      "loss": 0.5781556129455566,
      "step": 4290
    },
    {
      "epoch": 1.220550667045132,
      "grad_norm": 27.743261337280273,
      "learning_rate": 9.878086857791655e-05,
      "loss": 0.6214333057403565,
      "step": 4300
    },
    {
      "epoch": 1.2233891569684927,
      "grad_norm": 21.51885223388672,
      "learning_rate": 9.87780300879932e-05,
      "loss": 0.623208475112915,
      "step": 4310
    },
    {
      "epoch": 1.2262276468918536,
      "grad_norm": 19.256988525390625,
      "learning_rate": 9.877519159806982e-05,
      "loss": 0.5246468544006347,
      "step": 4320
    },
    {
      "epoch": 1.2290661368152143,
      "grad_norm": 29.382707595825195,
      "learning_rate": 9.877235310814648e-05,
      "loss": 0.622248363494873,
      "step": 4330
    },
    {
      "epoch": 1.2319046267385751,
      "grad_norm": 25.908470153808594,
      "learning_rate": 9.876951461822312e-05,
      "loss": 0.5938609600067138,
      "step": 4340
    },
    {
      "epoch": 1.2347431166619358,
      "grad_norm": 19.781829833984375,
      "learning_rate": 9.876667612829975e-05,
      "loss": 0.5751468658447265,
      "step": 4350
    },
    {
      "epoch": 1.2375816065852967,
      "grad_norm": 22.52198028564453,
      "learning_rate": 9.876383763837639e-05,
      "loss": 0.6422913074493408,
      "step": 4360
    },
    {
      "epoch": 1.2404200965086574,
      "grad_norm": 23.624773025512695,
      "learning_rate": 9.876099914845303e-05,
      "loss": 0.5497620105743408,
      "step": 4370
    },
    {
      "epoch": 1.2432585864320183,
      "grad_norm": 24.279569625854492,
      "learning_rate": 9.875816065852966e-05,
      "loss": 0.5888115406036377,
      "step": 4380
    },
    {
      "epoch": 1.246097076355379,
      "grad_norm": 20.661331176757812,
      "learning_rate": 9.87553221686063e-05,
      "loss": 0.5441700935363769,
      "step": 4390
    },
    {
      "epoch": 1.2489355662787398,
      "grad_norm": 23.45940589904785,
      "learning_rate": 9.875248367868295e-05,
      "loss": 0.5755167007446289,
      "step": 4400
    },
    {
      "epoch": 1.2517740562021005,
      "grad_norm": 36.26948928833008,
      "learning_rate": 9.874964518875958e-05,
      "loss": 0.6581007480621338,
      "step": 4410
    },
    {
      "epoch": 1.2546125461254611,
      "grad_norm": 28.99075698852539,
      "learning_rate": 9.874680669883622e-05,
      "loss": 0.5576618194580079,
      "step": 4420
    },
    {
      "epoch": 1.257451036048822,
      "grad_norm": 15.188714981079102,
      "learning_rate": 9.874396820891286e-05,
      "loss": 0.5354149818420411,
      "step": 4430
    },
    {
      "epoch": 1.260289525972183,
      "grad_norm": 33.135074615478516,
      "learning_rate": 9.87411297189895e-05,
      "loss": 0.6364504814147949,
      "step": 4440
    },
    {
      "epoch": 1.2631280158955436,
      "grad_norm": 21.79518699645996,
      "learning_rate": 9.873829122906613e-05,
      "loss": 0.6258384227752686,
      "step": 4450
    },
    {
      "epoch": 1.2659665058189042,
      "grad_norm": 23.883333206176758,
      "learning_rate": 9.873545273914278e-05,
      "loss": 0.549041748046875,
      "step": 4460
    },
    {
      "epoch": 1.2688049957422651,
      "grad_norm": 17.414018630981445,
      "learning_rate": 9.873261424921943e-05,
      "loss": 0.5412386417388916,
      "step": 4470
    },
    {
      "epoch": 1.271643485665626,
      "grad_norm": 37.17933654785156,
      "learning_rate": 9.872977575929606e-05,
      "loss": 0.5415035247802734,
      "step": 4480
    },
    {
      "epoch": 1.2744819755889867,
      "grad_norm": 20.74306869506836,
      "learning_rate": 9.87269372693727e-05,
      "loss": 0.5566246986389161,
      "step": 4490
    },
    {
      "epoch": 1.2773204655123473,
      "grad_norm": 18.906362533569336,
      "learning_rate": 9.872409877944934e-05,
      "loss": 0.5103561401367187,
      "step": 4500
    },
    {
      "epoch": 1.2773204655123473,
      "eval_accuracy": 0.7405735359572709,
      "eval_loss": 0.8171529769897461,
      "eval_runtime": 31.2672,
      "eval_samples_per_second": 502.988,
      "eval_steps_per_second": 7.868,
      "step": 4500
    },
    {
      "epoch": 1.2801589554357082,
      "grad_norm": 20.318050384521484,
      "learning_rate": 9.872126028952597e-05,
      "loss": 0.56712646484375,
      "step": 4510
    },
    {
      "epoch": 1.282997445359069,
      "grad_norm": 22.702953338623047,
      "learning_rate": 9.871842179960261e-05,
      "loss": 0.5166520595550537,
      "step": 4520
    },
    {
      "epoch": 1.2858359352824298,
      "grad_norm": 17.20403289794922,
      "learning_rate": 9.871558330967926e-05,
      "loss": 0.5455490112304687,
      "step": 4530
    },
    {
      "epoch": 1.2886744252057905,
      "grad_norm": 23.980810165405273,
      "learning_rate": 9.871274481975589e-05,
      "loss": 0.545859432220459,
      "step": 4540
    },
    {
      "epoch": 1.2915129151291513,
      "grad_norm": 22.90290069580078,
      "learning_rate": 9.870990632983253e-05,
      "loss": 0.5312858581542969,
      "step": 4550
    },
    {
      "epoch": 1.294351405052512,
      "grad_norm": 15.267108917236328,
      "learning_rate": 9.870706783990918e-05,
      "loss": 0.4967654705047607,
      "step": 4560
    },
    {
      "epoch": 1.297189894975873,
      "grad_norm": 23.086891174316406,
      "learning_rate": 9.870422934998582e-05,
      "loss": 0.4933491230010986,
      "step": 4570
    },
    {
      "epoch": 1.3000283848992336,
      "grad_norm": 30.994117736816406,
      "learning_rate": 9.870139086006244e-05,
      "loss": 0.5735067844390869,
      "step": 4580
    },
    {
      "epoch": 1.3028668748225944,
      "grad_norm": 21.239145278930664,
      "learning_rate": 9.869855237013909e-05,
      "loss": 0.540119743347168,
      "step": 4590
    },
    {
      "epoch": 1.305705364745955,
      "grad_norm": 22.86792755126953,
      "learning_rate": 9.869571388021574e-05,
      "loss": 0.5780026912689209,
      "step": 4600
    },
    {
      "epoch": 1.308543854669316,
      "grad_norm": 30.0904598236084,
      "learning_rate": 9.869287539029237e-05,
      "loss": 0.5557725429534912,
      "step": 4610
    },
    {
      "epoch": 1.3113823445926767,
      "grad_norm": 19.40277671813965,
      "learning_rate": 9.869003690036901e-05,
      "loss": 0.5452414989471436,
      "step": 4620
    },
    {
      "epoch": 1.3142208345160376,
      "grad_norm": 20.77096176147461,
      "learning_rate": 9.868719841044565e-05,
      "loss": 0.610050630569458,
      "step": 4630
    },
    {
      "epoch": 1.3170593244393982,
      "grad_norm": 22.80849266052246,
      "learning_rate": 9.868435992052228e-05,
      "loss": 0.5045920848846436,
      "step": 4640
    },
    {
      "epoch": 1.319897814362759,
      "grad_norm": 23.759414672851562,
      "learning_rate": 9.868152143059892e-05,
      "loss": 0.5491971492767334,
      "step": 4650
    },
    {
      "epoch": 1.3227363042861198,
      "grad_norm": 24.919057846069336,
      "learning_rate": 9.867868294067558e-05,
      "loss": 0.5136069297790528,
      "step": 4660
    },
    {
      "epoch": 1.3255747942094804,
      "grad_norm": 23.633241653442383,
      "learning_rate": 9.86758444507522e-05,
      "loss": 0.5374538421630859,
      "step": 4670
    },
    {
      "epoch": 1.3284132841328413,
      "grad_norm": 19.171987533569336,
      "learning_rate": 9.867300596082884e-05,
      "loss": 0.523595142364502,
      "step": 4680
    },
    {
      "epoch": 1.3312517740562022,
      "grad_norm": 39.36833572387695,
      "learning_rate": 9.867016747090549e-05,
      "loss": 0.6079312324523926,
      "step": 4690
    },
    {
      "epoch": 1.3340902639795629,
      "grad_norm": 22.341962814331055,
      "learning_rate": 9.866732898098213e-05,
      "loss": 0.5197252750396728,
      "step": 4700
    },
    {
      "epoch": 1.3369287539029235,
      "grad_norm": 18.308134078979492,
      "learning_rate": 9.866449049105876e-05,
      "loss": 0.5464882850646973,
      "step": 4710
    },
    {
      "epoch": 1.3397672438262844,
      "grad_norm": 17.265478134155273,
      "learning_rate": 9.86616520011354e-05,
      "loss": 0.5506086349487305,
      "step": 4720
    },
    {
      "epoch": 1.3426057337496453,
      "grad_norm": 23.074758529663086,
      "learning_rate": 9.865881351121205e-05,
      "loss": 0.5192683696746826,
      "step": 4730
    },
    {
      "epoch": 1.345444223673006,
      "grad_norm": 20.067684173583984,
      "learning_rate": 9.865597502128868e-05,
      "loss": 0.5306649208068848,
      "step": 4740
    },
    {
      "epoch": 1.3482827135963666,
      "grad_norm": 20.553401947021484,
      "learning_rate": 9.865313653136532e-05,
      "loss": 0.49111275672912597,
      "step": 4750
    },
    {
      "epoch": 1.3511212035197275,
      "grad_norm": 31.436376571655273,
      "learning_rate": 9.865029804144196e-05,
      "loss": 0.5791452884674072,
      "step": 4760
    },
    {
      "epoch": 1.3539596934430882,
      "grad_norm": 36.756351470947266,
      "learning_rate": 9.864745955151859e-05,
      "loss": 0.5307989597320557,
      "step": 4770
    },
    {
      "epoch": 1.356798183366449,
      "grad_norm": 26.145814895629883,
      "learning_rate": 9.864462106159523e-05,
      "loss": 0.48682713508605957,
      "step": 4780
    },
    {
      "epoch": 1.3596366732898097,
      "grad_norm": 22.17993927001953,
      "learning_rate": 9.864178257167187e-05,
      "loss": 0.5243121147155761,
      "step": 4790
    },
    {
      "epoch": 1.3624751632131706,
      "grad_norm": 23.06700897216797,
      "learning_rate": 9.863894408174851e-05,
      "loss": 0.594422721862793,
      "step": 4800
    },
    {
      "epoch": 1.3653136531365313,
      "grad_norm": 17.529457092285156,
      "learning_rate": 9.863610559182516e-05,
      "loss": 0.5081527709960938,
      "step": 4810
    },
    {
      "epoch": 1.3681521430598922,
      "grad_norm": 19.19485092163086,
      "learning_rate": 9.86332671019018e-05,
      "loss": 0.5057069301605225,
      "step": 4820
    },
    {
      "epoch": 1.3709906329832529,
      "grad_norm": 24.305240631103516,
      "learning_rate": 9.863042861197844e-05,
      "loss": 0.5302314758300781,
      "step": 4830
    },
    {
      "epoch": 1.3738291229066137,
      "grad_norm": 32.79861831665039,
      "learning_rate": 9.862759012205507e-05,
      "loss": 0.5537318706512451,
      "step": 4840
    },
    {
      "epoch": 1.3766676128299744,
      "grad_norm": 21.89700698852539,
      "learning_rate": 9.862475163213171e-05,
      "loss": 0.5480470657348633,
      "step": 4850
    },
    {
      "epoch": 1.3795061027533353,
      "grad_norm": 20.300912857055664,
      "learning_rate": 9.862191314220835e-05,
      "loss": 0.505549144744873,
      "step": 4860
    },
    {
      "epoch": 1.382344592676696,
      "grad_norm": 20.352535247802734,
      "learning_rate": 9.861907465228499e-05,
      "loss": 0.5526429176330566,
      "step": 4870
    },
    {
      "epoch": 1.3851830826000568,
      "grad_norm": 23.988351821899414,
      "learning_rate": 9.861623616236163e-05,
      "loss": 0.5067482948303222,
      "step": 4880
    },
    {
      "epoch": 1.3880215725234175,
      "grad_norm": 30.3094425201416,
      "learning_rate": 9.861339767243827e-05,
      "loss": 0.5444604396820069,
      "step": 4890
    },
    {
      "epoch": 1.3908600624467784,
      "grad_norm": 15.551344871520996,
      "learning_rate": 9.86105591825149e-05,
      "loss": 0.5170631885528565,
      "step": 4900
    },
    {
      "epoch": 1.393698552370139,
      "grad_norm": 22.54707145690918,
      "learning_rate": 9.860772069259154e-05,
      "loss": 0.4765817165374756,
      "step": 4910
    },
    {
      "epoch": 1.3965370422934997,
      "grad_norm": 21.261871337890625,
      "learning_rate": 9.860488220266818e-05,
      "loss": 0.4708756446838379,
      "step": 4920
    },
    {
      "epoch": 1.3993755322168606,
      "grad_norm": 22.6719913482666,
      "learning_rate": 9.860204371274482e-05,
      "loss": 0.5953768730163574,
      "step": 4930
    },
    {
      "epoch": 1.4022140221402215,
      "grad_norm": 25.462444305419922,
      "learning_rate": 9.859920522282147e-05,
      "loss": 0.5883163452148438,
      "step": 4940
    },
    {
      "epoch": 1.4050525120635822,
      "grad_norm": 14.67077350616455,
      "learning_rate": 9.859636673289811e-05,
      "loss": 0.487822437286377,
      "step": 4950
    },
    {
      "epoch": 1.4078910019869428,
      "grad_norm": 23.763290405273438,
      "learning_rate": 9.859352824297474e-05,
      "loss": 0.5172186851501465,
      "step": 4960
    },
    {
      "epoch": 1.4107294919103037,
      "grad_norm": 20.53472900390625,
      "learning_rate": 9.859068975305138e-05,
      "loss": 0.5040953159332275,
      "step": 4970
    },
    {
      "epoch": 1.4135679818336646,
      "grad_norm": 29.127592086791992,
      "learning_rate": 9.858785126312802e-05,
      "loss": 0.5164206504821778,
      "step": 4980
    },
    {
      "epoch": 1.4164064717570253,
      "grad_norm": 21.441164016723633,
      "learning_rate": 9.858501277320466e-05,
      "loss": 0.5650917053222656,
      "step": 4990
    },
    {
      "epoch": 1.419244961680386,
      "grad_norm": 24.15547752380371,
      "learning_rate": 9.85821742832813e-05,
      "loss": 0.5443023204803467,
      "step": 5000
    },
    {
      "epoch": 1.419244961680386,
      "eval_accuracy": 0.7549437273478731,
      "eval_loss": 0.7658727169036865,
      "eval_runtime": 30.6986,
      "eval_samples_per_second": 512.304,
      "eval_steps_per_second": 8.013,
      "step": 5000
    },
    {
      "epoch": 1.4220834516037468,
      "grad_norm": 22.526084899902344,
      "learning_rate": 9.857933579335794e-05,
      "loss": 0.4786521434783936,
      "step": 5010
    },
    {
      "epoch": 1.4249219415271077,
      "grad_norm": 15.852435111999512,
      "learning_rate": 9.857649730343458e-05,
      "loss": 0.5006365299224853,
      "step": 5020
    },
    {
      "epoch": 1.4277604314504684,
      "grad_norm": 15.36372184753418,
      "learning_rate": 9.857365881351121e-05,
      "loss": 0.44577822685241697,
      "step": 5030
    },
    {
      "epoch": 1.430598921373829,
      "grad_norm": 15.919493675231934,
      "learning_rate": 9.857082032358785e-05,
      "loss": 0.5354157447814941,
      "step": 5040
    },
    {
      "epoch": 1.43343741129719,
      "grad_norm": 18.767841339111328,
      "learning_rate": 9.85679818336645e-05,
      "loss": 0.5587405204772949,
      "step": 5050
    },
    {
      "epoch": 1.4362759012205506,
      "grad_norm": 25.292882919311523,
      "learning_rate": 9.856514334374114e-05,
      "loss": 0.5098119735717773,
      "step": 5060
    },
    {
      "epoch": 1.4391143911439115,
      "grad_norm": 22.182270050048828,
      "learning_rate": 9.856230485381778e-05,
      "loss": 0.5585176467895507,
      "step": 5070
    },
    {
      "epoch": 1.4419528810672722,
      "grad_norm": 28.664527893066406,
      "learning_rate": 9.855946636389442e-05,
      "loss": 0.5147800922393799,
      "step": 5080
    },
    {
      "epoch": 1.444791370990633,
      "grad_norm": 18.47371482849121,
      "learning_rate": 9.855662787397105e-05,
      "loss": 0.536253547668457,
      "step": 5090
    },
    {
      "epoch": 1.4476298609139937,
      "grad_norm": 25.276735305786133,
      "learning_rate": 9.855378938404769e-05,
      "loss": 0.534224271774292,
      "step": 5100
    },
    {
      "epoch": 1.4504683508373546,
      "grad_norm": 19.44446563720703,
      "learning_rate": 9.855095089412433e-05,
      "loss": 0.5034564971923828,
      "step": 5110
    },
    {
      "epoch": 1.4533068407607153,
      "grad_norm": 23.57341957092285,
      "learning_rate": 9.854811240420097e-05,
      "loss": 0.5047745704650879,
      "step": 5120
    },
    {
      "epoch": 1.4561453306840761,
      "grad_norm": 30.271663665771484,
      "learning_rate": 9.854527391427761e-05,
      "loss": 0.5482126235961914,
      "step": 5130
    },
    {
      "epoch": 1.4589838206074368,
      "grad_norm": 28.177982330322266,
      "learning_rate": 9.854243542435425e-05,
      "loss": 0.5251502513885498,
      "step": 5140
    },
    {
      "epoch": 1.4618223105307977,
      "grad_norm": 33.12307357788086,
      "learning_rate": 9.85395969344309e-05,
      "loss": 0.49888076782226565,
      "step": 5150
    },
    {
      "epoch": 1.4646608004541584,
      "grad_norm": 25.201351165771484,
      "learning_rate": 9.853675844450752e-05,
      "loss": 0.5624466896057129,
      "step": 5160
    },
    {
      "epoch": 1.467499290377519,
      "grad_norm": 16.79703140258789,
      "learning_rate": 9.853391995458416e-05,
      "loss": 0.5105113029479981,
      "step": 5170
    },
    {
      "epoch": 1.47033778030088,
      "grad_norm": 20.61594009399414,
      "learning_rate": 9.85310814646608e-05,
      "loss": 0.5088314533233642,
      "step": 5180
    },
    {
      "epoch": 1.4731762702242408,
      "grad_norm": 22.573232650756836,
      "learning_rate": 9.852824297473743e-05,
      "loss": 0.47755999565124513,
      "step": 5190
    },
    {
      "epoch": 1.4760147601476015,
      "grad_norm": 18.0330753326416,
      "learning_rate": 9.852540448481409e-05,
      "loss": 0.46817960739135744,
      "step": 5200
    },
    {
      "epoch": 1.4788532500709621,
      "grad_norm": 26.57672119140625,
      "learning_rate": 9.852256599489073e-05,
      "loss": 0.44509220123291016,
      "step": 5210
    },
    {
      "epoch": 1.481691739994323,
      "grad_norm": 27.457799911499023,
      "learning_rate": 9.851972750496736e-05,
      "loss": 0.49938220977783204,
      "step": 5220
    },
    {
      "epoch": 1.484530229917684,
      "grad_norm": 29.482223510742188,
      "learning_rate": 9.8516889015044e-05,
      "loss": 0.48114356994628904,
      "step": 5230
    },
    {
      "epoch": 1.4873687198410446,
      "grad_norm": 19.489933013916016,
      "learning_rate": 9.851405052512064e-05,
      "loss": 0.5131722450256347,
      "step": 5240
    },
    {
      "epoch": 1.4902072097644052,
      "grad_norm": 25.34461784362793,
      "learning_rate": 9.851149588418962e-05,
      "loss": 0.47336368560791015,
      "step": 5250
    },
    {
      "epoch": 1.4930456996877661,
      "grad_norm": 22.344881057739258,
      "learning_rate": 9.850865739426626e-05,
      "loss": 0.423797082901001,
      "step": 5260
    },
    {
      "epoch": 1.495884189611127,
      "grad_norm": 21.193986892700195,
      "learning_rate": 9.850581890434289e-05,
      "loss": 0.509946632385254,
      "step": 5270
    },
    {
      "epoch": 1.4987226795344877,
      "grad_norm": 25.462278366088867,
      "learning_rate": 9.850298041441953e-05,
      "loss": 0.5746853828430176,
      "step": 5280
    },
    {
      "epoch": 1.5015611694578483,
      "grad_norm": 21.391799926757812,
      "learning_rate": 9.850014192449617e-05,
      "loss": 0.49782500267028806,
      "step": 5290
    },
    {
      "epoch": 1.5043996593812092,
      "grad_norm": 30.367361068725586,
      "learning_rate": 9.849730343457281e-05,
      "loss": 0.4601006031036377,
      "step": 5300
    },
    {
      "epoch": 1.5072381493045701,
      "grad_norm": 18.497825622558594,
      "learning_rate": 9.849446494464945e-05,
      "loss": 0.47089128494262694,
      "step": 5310
    },
    {
      "epoch": 1.5100766392279308,
      "grad_norm": 24.439403533935547,
      "learning_rate": 9.84916264547261e-05,
      "loss": 0.4955033779144287,
      "step": 5320
    },
    {
      "epoch": 1.5129151291512914,
      "grad_norm": 18.78842544555664,
      "learning_rate": 9.848878796480274e-05,
      "loss": 0.4948451042175293,
      "step": 5330
    },
    {
      "epoch": 1.5157536190746523,
      "grad_norm": 21.88657569885254,
      "learning_rate": 9.848594947487936e-05,
      "loss": 0.4650012969970703,
      "step": 5340
    },
    {
      "epoch": 1.518592108998013,
      "grad_norm": 21.758548736572266,
      "learning_rate": 9.8483110984956e-05,
      "loss": 0.4517825603485107,
      "step": 5350
    },
    {
      "epoch": 1.521430598921374,
      "grad_norm": 22.636329650878906,
      "learning_rate": 9.848027249503265e-05,
      "loss": 0.5213921546936036,
      "step": 5360
    },
    {
      "epoch": 1.5242690888447346,
      "grad_norm": 32.18223571777344,
      "learning_rate": 9.847743400510928e-05,
      "loss": 0.5041645526885986,
      "step": 5370
    },
    {
      "epoch": 1.5271075787680952,
      "grad_norm": 18.59054183959961,
      "learning_rate": 9.847459551518593e-05,
      "loss": 0.5283419609069824,
      "step": 5380
    },
    {
      "epoch": 1.529946068691456,
      "grad_norm": 26.19089698791504,
      "learning_rate": 9.847175702526257e-05,
      "loss": 0.49493966102600095,
      "step": 5390
    },
    {
      "epoch": 1.532784558614817,
      "grad_norm": 19.25995635986328,
      "learning_rate": 9.84689185353392e-05,
      "loss": 0.46999635696411135,
      "step": 5400
    },
    {
      "epoch": 1.5356230485381777,
      "grad_norm": 21.66840171813965,
      "learning_rate": 9.846608004541584e-05,
      "loss": 0.46753621101379395,
      "step": 5410
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 21.555118560791016,
      "learning_rate": 9.846324155549248e-05,
      "loss": 0.4348593235015869,
      "step": 5420
    },
    {
      "epoch": 1.5413000283848992,
      "grad_norm": 20.662364959716797,
      "learning_rate": 9.846040306556912e-05,
      "loss": 0.47795305252075193,
      "step": 5430
    },
    {
      "epoch": 1.54413851830826,
      "grad_norm": 26.63555908203125,
      "learning_rate": 9.845756457564577e-05,
      "loss": 0.5018423080444336,
      "step": 5440
    },
    {
      "epoch": 1.5469770082316208,
      "grad_norm": 17.335065841674805,
      "learning_rate": 9.84547260857224e-05,
      "loss": 0.45438499450683595,
      "step": 5450
    },
    {
      "epoch": 1.5498154981549814,
      "grad_norm": 27.396347045898438,
      "learning_rate": 9.845188759579905e-05,
      "loss": 0.5069549560546875,
      "step": 5460
    },
    {
      "epoch": 1.5526539880783423,
      "grad_norm": 28.821380615234375,
      "learning_rate": 9.844904910587568e-05,
      "loss": 0.4687957763671875,
      "step": 5470
    },
    {
      "epoch": 1.5554924780017032,
      "grad_norm": 32.41615676879883,
      "learning_rate": 9.844621061595232e-05,
      "loss": 0.4600421905517578,
      "step": 5480
    },
    {
      "epoch": 1.5583309679250639,
      "grad_norm": 17.59092903137207,
      "learning_rate": 9.844337212602896e-05,
      "loss": 0.4343866348266602,
      "step": 5490
    },
    {
      "epoch": 1.5611694578484245,
      "grad_norm": 25.264320373535156,
      "learning_rate": 9.844053363610559e-05,
      "loss": 0.47960448265075684,
      "step": 5500
    },
    {
      "epoch": 1.5611694578484245,
      "eval_accuracy": 0.7705220321739683,
      "eval_loss": 0.7081770300865173,
      "eval_runtime": 30.734,
      "eval_samples_per_second": 511.714,
      "eval_steps_per_second": 8.004,
      "step": 5500
    },
    {
      "epoch": 1.5640079477717854,
      "grad_norm": 18.138931274414062,
      "learning_rate": 9.843769514618224e-05,
      "loss": 0.43840909004211426,
      "step": 5510
    },
    {
      "epoch": 1.5668464376951463,
      "grad_norm": 31.31865119934082,
      "learning_rate": 9.843485665625888e-05,
      "loss": 0.5786248207092285,
      "step": 5520
    },
    {
      "epoch": 1.569684927618507,
      "grad_norm": 17.037433624267578,
      "learning_rate": 9.843201816633551e-05,
      "loss": 0.4403118133544922,
      "step": 5530
    },
    {
      "epoch": 1.5725234175418676,
      "grad_norm": 21.156137466430664,
      "learning_rate": 9.842917967641215e-05,
      "loss": 0.45325593948364257,
      "step": 5540
    },
    {
      "epoch": 1.5753619074652285,
      "grad_norm": 20.449691772460938,
      "learning_rate": 9.842634118648879e-05,
      "loss": 0.5135906219482422,
      "step": 5550
    },
    {
      "epoch": 1.5782003973885894,
      "grad_norm": 25.842815399169922,
      "learning_rate": 9.842350269656543e-05,
      "loss": 0.44756927490234377,
      "step": 5560
    },
    {
      "epoch": 1.58103888731195,
      "grad_norm": 24.721166610717773,
      "learning_rate": 9.842066420664206e-05,
      "loss": 0.4527876853942871,
      "step": 5570
    },
    {
      "epoch": 1.5838773772353107,
      "grad_norm": 29.005586624145508,
      "learning_rate": 9.841782571671872e-05,
      "loss": 0.4674509048461914,
      "step": 5580
    },
    {
      "epoch": 1.5867158671586716,
      "grad_norm": 26.988758087158203,
      "learning_rate": 9.841498722679536e-05,
      "loss": 0.4187464237213135,
      "step": 5590
    },
    {
      "epoch": 1.5895543570820325,
      "grad_norm": 21.200258255004883,
      "learning_rate": 9.841214873687199e-05,
      "loss": 0.4542283058166504,
      "step": 5600
    },
    {
      "epoch": 1.5923928470053932,
      "grad_norm": 21.527809143066406,
      "learning_rate": 9.840931024694863e-05,
      "loss": 0.45181856155395506,
      "step": 5610
    },
    {
      "epoch": 1.5952313369287539,
      "grad_norm": 28.10869789123535,
      "learning_rate": 9.840647175702527e-05,
      "loss": 0.5212711334228516,
      "step": 5620
    },
    {
      "epoch": 1.5980698268521145,
      "grad_norm": 22.373567581176758,
      "learning_rate": 9.84036332671019e-05,
      "loss": 0.441100549697876,
      "step": 5630
    },
    {
      "epoch": 1.6009083167754754,
      "grad_norm": 20.653213500976562,
      "learning_rate": 9.840079477717855e-05,
      "loss": 0.4398199558258057,
      "step": 5640
    },
    {
      "epoch": 1.6037468066988363,
      "grad_norm": 18.533262252807617,
      "learning_rate": 9.83979562872552e-05,
      "loss": 0.40874290466308594,
      "step": 5650
    },
    {
      "epoch": 1.606585296622197,
      "grad_norm": 17.033687591552734,
      "learning_rate": 9.839511779733182e-05,
      "loss": 0.43529257774353025,
      "step": 5660
    },
    {
      "epoch": 1.6094237865455576,
      "grad_norm": 25.372634887695312,
      "learning_rate": 9.839227930740846e-05,
      "loss": 0.5042860984802247,
      "step": 5670
    },
    {
      "epoch": 1.6122622764689185,
      "grad_norm": 26.601974487304688,
      "learning_rate": 9.83894408174851e-05,
      "loss": 0.4922661781311035,
      "step": 5680
    },
    {
      "epoch": 1.6151007663922794,
      "grad_norm": 20.631139755249023,
      "learning_rate": 9.838660232756175e-05,
      "loss": 0.4670083999633789,
      "step": 5690
    },
    {
      "epoch": 1.61793925631564,
      "grad_norm": 15.214738845825195,
      "learning_rate": 9.838376383763837e-05,
      "loss": 0.5103687286376953,
      "step": 5700
    },
    {
      "epoch": 1.6207777462390007,
      "grad_norm": 22.460304260253906,
      "learning_rate": 9.838092534771503e-05,
      "loss": 0.4493797779083252,
      "step": 5710
    },
    {
      "epoch": 1.6236162361623616,
      "grad_norm": 20.383995056152344,
      "learning_rate": 9.837808685779167e-05,
      "loss": 0.5087112426757813,
      "step": 5720
    },
    {
      "epoch": 1.6264547260857225,
      "grad_norm": 31.7694091796875,
      "learning_rate": 9.83752483678683e-05,
      "loss": 0.48621907234191897,
      "step": 5730
    },
    {
      "epoch": 1.6292932160090832,
      "grad_norm": 24.267576217651367,
      "learning_rate": 9.837240987794494e-05,
      "loss": 0.43189225196838377,
      "step": 5740
    },
    {
      "epoch": 1.6321317059324438,
      "grad_norm": 17.485614776611328,
      "learning_rate": 9.836957138802158e-05,
      "loss": 0.4363883018493652,
      "step": 5750
    },
    {
      "epoch": 1.6349701958558047,
      "grad_norm": 17.831676483154297,
      "learning_rate": 9.836673289809821e-05,
      "loss": 0.36375472545623777,
      "step": 5760
    },
    {
      "epoch": 1.6378086857791656,
      "grad_norm": 38.249237060546875,
      "learning_rate": 9.836389440817485e-05,
      "loss": 0.487896203994751,
      "step": 5770
    },
    {
      "epoch": 1.6406471757025263,
      "grad_norm": 21.259401321411133,
      "learning_rate": 9.83610559182515e-05,
      "loss": 0.4958378314971924,
      "step": 5780
    },
    {
      "epoch": 1.643485665625887,
      "grad_norm": 18.802688598632812,
      "learning_rate": 9.835821742832813e-05,
      "loss": 0.4471120834350586,
      "step": 5790
    },
    {
      "epoch": 1.6463241555492478,
      "grad_norm": 18.47874641418457,
      "learning_rate": 9.835537893840477e-05,
      "loss": 0.46596393585205076,
      "step": 5800
    },
    {
      "epoch": 1.6491626454726087,
      "grad_norm": 22.69192123413086,
      "learning_rate": 9.835254044848141e-05,
      "loss": 0.5249762535095215,
      "step": 5810
    },
    {
      "epoch": 1.6520011353959694,
      "grad_norm": 23.77510643005371,
      "learning_rate": 9.834970195855806e-05,
      "loss": 0.49897050857543945,
      "step": 5820
    },
    {
      "epoch": 1.65483962531933,
      "grad_norm": 20.74818992614746,
      "learning_rate": 9.834686346863468e-05,
      "loss": 0.43282279968261717,
      "step": 5830
    },
    {
      "epoch": 1.657678115242691,
      "grad_norm": 27.07232093811035,
      "learning_rate": 9.834402497871134e-05,
      "loss": 0.452546501159668,
      "step": 5840
    },
    {
      "epoch": 1.6605166051660518,
      "grad_norm": 18.120189666748047,
      "learning_rate": 9.834118648878797e-05,
      "loss": 0.45517935752868655,
      "step": 5850
    },
    {
      "epoch": 1.6633550950894125,
      "grad_norm": 15.141501426696777,
      "learning_rate": 9.833834799886461e-05,
      "loss": 0.4280442237854004,
      "step": 5860
    },
    {
      "epoch": 1.6661935850127731,
      "grad_norm": 29.13498306274414,
      "learning_rate": 9.833550950894125e-05,
      "loss": 0.45500760078430175,
      "step": 5870
    },
    {
      "epoch": 1.6690320749361338,
      "grad_norm": 17.46603012084961,
      "learning_rate": 9.833267101901789e-05,
      "loss": 0.5003047943115234,
      "step": 5880
    },
    {
      "epoch": 1.6718705648594947,
      "grad_norm": 47.59585952758789,
      "learning_rate": 9.832983252909452e-05,
      "loss": 0.4896832466125488,
      "step": 5890
    },
    {
      "epoch": 1.6747090547828556,
      "grad_norm": 24.156326293945312,
      "learning_rate": 9.832699403917116e-05,
      "loss": 0.48364977836608886,
      "step": 5900
    },
    {
      "epoch": 1.6775475447062163,
      "grad_norm": 27.248226165771484,
      "learning_rate": 9.832415554924781e-05,
      "loss": 0.45184640884399413,
      "step": 5910
    },
    {
      "epoch": 1.680386034629577,
      "grad_norm": 21.402156829833984,
      "learning_rate": 9.832131705932444e-05,
      "loss": 0.4230981349945068,
      "step": 5920
    },
    {
      "epoch": 1.6832245245529378,
      "grad_norm": 21.467369079589844,
      "learning_rate": 9.831847856940108e-05,
      "loss": 0.39625232219696044,
      "step": 5930
    },
    {
      "epoch": 1.6860630144762987,
      "grad_norm": 17.363279342651367,
      "learning_rate": 9.831564007947773e-05,
      "loss": 0.4277812957763672,
      "step": 5940
    },
    {
      "epoch": 1.6889015043996594,
      "grad_norm": 22.998035430908203,
      "learning_rate": 9.831280158955435e-05,
      "loss": 0.3986526012420654,
      "step": 5950
    },
    {
      "epoch": 1.69173999432302,
      "grad_norm": 26.156076431274414,
      "learning_rate": 9.8309963099631e-05,
      "loss": 0.4157834053039551,
      "step": 5960
    },
    {
      "epoch": 1.694578484246381,
      "grad_norm": 27.324661254882812,
      "learning_rate": 9.830712460970764e-05,
      "loss": 0.40967526435852053,
      "step": 5970
    },
    {
      "epoch": 1.6974169741697418,
      "grad_norm": 31.61458396911621,
      "learning_rate": 9.830428611978428e-05,
      "loss": 0.49207186698913574,
      "step": 5980
    },
    {
      "epoch": 1.7002554640931025,
      "grad_norm": 15.488746643066406,
      "learning_rate": 9.830144762986092e-05,
      "loss": 0.4254718780517578,
      "step": 5990
    },
    {
      "epoch": 1.7030939540164631,
      "grad_norm": 19.080184936523438,
      "learning_rate": 9.829860913993756e-05,
      "loss": 0.3942666530609131,
      "step": 6000
    },
    {
      "epoch": 1.7030939540164631,
      "eval_accuracy": 0.7946843008838304,
      "eval_loss": 0.639552891254425,
      "eval_runtime": 31.3826,
      "eval_samples_per_second": 501.138,
      "eval_steps_per_second": 7.839,
      "step": 6000
    },
    {
      "epoch": 1.705932443939824,
      "grad_norm": 17.96277618408203,
      "learning_rate": 9.82957706500142e-05,
      "loss": 0.45555100440979,
      "step": 6010
    },
    {
      "epoch": 1.708770933863185,
      "grad_norm": 25.49015998840332,
      "learning_rate": 9.829293216009083e-05,
      "loss": 0.4165348052978516,
      "step": 6020
    },
    {
      "epoch": 1.7116094237865456,
      "grad_norm": 18.687419891357422,
      "learning_rate": 9.829009367016747e-05,
      "loss": 0.42314891815185546,
      "step": 6030
    },
    {
      "epoch": 1.7144479137099062,
      "grad_norm": 28.25301170349121,
      "learning_rate": 9.828725518024413e-05,
      "loss": 0.46349000930786133,
      "step": 6040
    },
    {
      "epoch": 1.7172864036332671,
      "grad_norm": 18.832815170288086,
      "learning_rate": 9.828441669032075e-05,
      "loss": 0.3894632816314697,
      "step": 6050
    },
    {
      "epoch": 1.720124893556628,
      "grad_norm": 18.941354751586914,
      "learning_rate": 9.82815782003974e-05,
      "loss": 0.424459171295166,
      "step": 6060
    },
    {
      "epoch": 1.7229633834799887,
      "grad_norm": 21.745758056640625,
      "learning_rate": 9.827873971047404e-05,
      "loss": 0.4320863723754883,
      "step": 6070
    },
    {
      "epoch": 1.7258018734033493,
      "grad_norm": 26.7862491607666,
      "learning_rate": 9.827590122055066e-05,
      "loss": 0.4254425525665283,
      "step": 6080
    },
    {
      "epoch": 1.7286403633267102,
      "grad_norm": 29.282751083374023,
      "learning_rate": 9.82730627306273e-05,
      "loss": 0.4424506664276123,
      "step": 6090
    },
    {
      "epoch": 1.7314788532500711,
      "grad_norm": 22.056499481201172,
      "learning_rate": 9.827022424070395e-05,
      "loss": 0.40218181610107423,
      "step": 6100
    },
    {
      "epoch": 1.7343173431734318,
      "grad_norm": 18.969566345214844,
      "learning_rate": 9.826738575078059e-05,
      "loss": 0.4059330463409424,
      "step": 6110
    },
    {
      "epoch": 1.7371558330967924,
      "grad_norm": 26.770959854125977,
      "learning_rate": 9.826454726085723e-05,
      "loss": 0.4434517383575439,
      "step": 6120
    },
    {
      "epoch": 1.7399943230201533,
      "grad_norm": 16.149120330810547,
      "learning_rate": 9.826170877093387e-05,
      "loss": 0.41785407066345215,
      "step": 6130
    },
    {
      "epoch": 1.742832812943514,
      "grad_norm": 17.30535316467285,
      "learning_rate": 9.825887028101051e-05,
      "loss": 0.4103550434112549,
      "step": 6140
    },
    {
      "epoch": 1.7456713028668749,
      "grad_norm": 18.982412338256836,
      "learning_rate": 9.825603179108714e-05,
      "loss": 0.41153688430786134,
      "step": 6150
    },
    {
      "epoch": 1.7485097927902356,
      "grad_norm": 35.373897552490234,
      "learning_rate": 9.825319330116378e-05,
      "loss": 0.44315595626831056,
      "step": 6160
    },
    {
      "epoch": 1.7513482827135962,
      "grad_norm": 19.156211853027344,
      "learning_rate": 9.825035481124042e-05,
      "loss": 0.404082727432251,
      "step": 6170
    },
    {
      "epoch": 1.754186772636957,
      "grad_norm": 26.301593780517578,
      "learning_rate": 9.824751632131706e-05,
      "loss": 0.4820054054260254,
      "step": 6180
    },
    {
      "epoch": 1.757025262560318,
      "grad_norm": 23.461977005004883,
      "learning_rate": 9.82446778313937e-05,
      "loss": 0.41650118827819826,
      "step": 6190
    },
    {
      "epoch": 1.7598637524836787,
      "grad_norm": 18.080190658569336,
      "learning_rate": 9.824183934147035e-05,
      "loss": 0.4413625240325928,
      "step": 6200
    },
    {
      "epoch": 1.7627022424070393,
      "grad_norm": 29.39219093322754,
      "learning_rate": 9.823900085154697e-05,
      "loss": 0.4654662609100342,
      "step": 6210
    },
    {
      "epoch": 1.7655407323304002,
      "grad_norm": 24.986982345581055,
      "learning_rate": 9.823616236162362e-05,
      "loss": 0.3866588592529297,
      "step": 6220
    },
    {
      "epoch": 1.768379222253761,
      "grad_norm": 22.976842880249023,
      "learning_rate": 9.823332387170026e-05,
      "loss": 0.39845123291015627,
      "step": 6230
    },
    {
      "epoch": 1.7712177121771218,
      "grad_norm": 22.65195083618164,
      "learning_rate": 9.82304853817769e-05,
      "loss": 0.35651321411132814,
      "step": 6240
    },
    {
      "epoch": 1.7740562021004824,
      "grad_norm": 26.74050521850586,
      "learning_rate": 9.822764689185354e-05,
      "loss": 0.34719023704528806,
      "step": 6250
    },
    {
      "epoch": 1.7768946920238433,
      "grad_norm": 17.39774513244629,
      "learning_rate": 9.822480840193018e-05,
      "loss": 0.4214783668518066,
      "step": 6260
    },
    {
      "epoch": 1.7797331819472042,
      "grad_norm": 16.551368713378906,
      "learning_rate": 9.822196991200682e-05,
      "loss": 0.436419153213501,
      "step": 6270
    },
    {
      "epoch": 1.7825716718705649,
      "grad_norm": 29.02804946899414,
      "learning_rate": 9.821913142208345e-05,
      "loss": 0.4335481643676758,
      "step": 6280
    },
    {
      "epoch": 1.7854101617939255,
      "grad_norm": 22.586355209350586,
      "learning_rate": 9.821629293216009e-05,
      "loss": 0.35754120349884033,
      "step": 6290
    },
    {
      "epoch": 1.7882486517172864,
      "grad_norm": 21.191234588623047,
      "learning_rate": 9.821345444223673e-05,
      "loss": 0.4479619026184082,
      "step": 6300
    },
    {
      "epoch": 1.7910871416406473,
      "grad_norm": 29.112369537353516,
      "learning_rate": 9.821061595231337e-05,
      "loss": 0.4265907287597656,
      "step": 6310
    },
    {
      "epoch": 1.793925631564008,
      "grad_norm": 20.455699920654297,
      "learning_rate": 9.820777746239002e-05,
      "loss": 0.3897282600402832,
      "step": 6320
    },
    {
      "epoch": 1.7967641214873686,
      "grad_norm": 18.384050369262695,
      "learning_rate": 9.820493897246666e-05,
      "loss": 0.37078032493591306,
      "step": 6330
    },
    {
      "epoch": 1.7996026114107295,
      "grad_norm": 17.571855545043945,
      "learning_rate": 9.820210048254329e-05,
      "loss": 0.43453564643859866,
      "step": 6340
    },
    {
      "epoch": 1.8024411013340904,
      "grad_norm": 19.029672622680664,
      "learning_rate": 9.819926199261993e-05,
      "loss": 0.4427356719970703,
      "step": 6350
    },
    {
      "epoch": 1.805279591257451,
      "grad_norm": 17.716753005981445,
      "learning_rate": 9.819642350269657e-05,
      "loss": 0.4016223430633545,
      "step": 6360
    },
    {
      "epoch": 1.8081180811808117,
      "grad_norm": 21.720796585083008,
      "learning_rate": 9.819358501277321e-05,
      "loss": 0.41228551864624025,
      "step": 6370
    },
    {
      "epoch": 1.8109565711041726,
      "grad_norm": 17.8159122467041,
      "learning_rate": 9.819074652284985e-05,
      "loss": 0.41730575561523436,
      "step": 6380
    },
    {
      "epoch": 1.8137950610275335,
      "grad_norm": 28.501855850219727,
      "learning_rate": 9.818790803292649e-05,
      "loss": 0.43465280532836914,
      "step": 6390
    },
    {
      "epoch": 1.8166335509508942,
      "grad_norm": 17.53424644470215,
      "learning_rate": 9.818506954300313e-05,
      "loss": 0.43177337646484376,
      "step": 6400
    },
    {
      "epoch": 1.8194720408742548,
      "grad_norm": 17.43556785583496,
      "learning_rate": 9.818223105307976e-05,
      "loss": 0.3786790370941162,
      "step": 6410
    },
    {
      "epoch": 1.8223105307976155,
      "grad_norm": 24.341060638427734,
      "learning_rate": 9.81793925631564e-05,
      "loss": 0.4329082489013672,
      "step": 6420
    },
    {
      "epoch": 1.8251490207209764,
      "grad_norm": 18.160860061645508,
      "learning_rate": 9.817655407323304e-05,
      "loss": 0.33026275634765623,
      "step": 6430
    },
    {
      "epoch": 1.8279875106443373,
      "grad_norm": 29.488039016723633,
      "learning_rate": 9.817371558330969e-05,
      "loss": 0.36414299011230467,
      "step": 6440
    },
    {
      "epoch": 1.830826000567698,
      "grad_norm": 24.25476837158203,
      "learning_rate": 9.817087709338633e-05,
      "loss": 0.36670989990234376,
      "step": 6450
    },
    {
      "epoch": 1.8336644904910586,
      "grad_norm": 22.606950759887695,
      "learning_rate": 9.816803860346297e-05,
      "loss": 0.4424624443054199,
      "step": 6460
    },
    {
      "epoch": 1.8365029804144195,
      "grad_norm": 21.106725692749023,
      "learning_rate": 9.81652001135396e-05,
      "loss": 0.42706780433654784,
      "step": 6470
    },
    {
      "epoch": 1.8393414703377804,
      "grad_norm": 22.923452377319336,
      "learning_rate": 9.816236162361624e-05,
      "loss": 0.3846797704696655,
      "step": 6480
    },
    {
      "epoch": 1.842179960261141,
      "grad_norm": 16.407604217529297,
      "learning_rate": 9.815952313369288e-05,
      "loss": 0.4001880168914795,
      "step": 6490
    },
    {
      "epoch": 1.8450184501845017,
      "grad_norm": 22.22679901123047,
      "learning_rate": 9.815668464376952e-05,
      "loss": 0.381205153465271,
      "step": 6500
    },
    {
      "epoch": 1.8450184501845017,
      "eval_accuracy": 0.7997075093787753,
      "eval_loss": 0.6205105781555176,
      "eval_runtime": 30.6759,
      "eval_samples_per_second": 512.683,
      "eval_steps_per_second": 8.019,
      "step": 6500
    },
    {
      "epoch": 1.8478569401078626,
      "grad_norm": 23.107955932617188,
      "learning_rate": 9.815384615384616e-05,
      "loss": 0.3948862314224243,
      "step": 6510
    },
    {
      "epoch": 1.8506954300312235,
      "grad_norm": 15.464545249938965,
      "learning_rate": 9.81510076639228e-05,
      "loss": 0.3841205835342407,
      "step": 6520
    },
    {
      "epoch": 1.8535339199545842,
      "grad_norm": 24.38384437561035,
      "learning_rate": 9.814816917399944e-05,
      "loss": 0.34784417152404784,
      "step": 6530
    },
    {
      "epoch": 1.8563724098779448,
      "grad_norm": 32.742393493652344,
      "learning_rate": 9.814533068407607e-05,
      "loss": 0.40121960639953613,
      "step": 6540
    },
    {
      "epoch": 1.8592108998013057,
      "grad_norm": 30.2054386138916,
      "learning_rate": 9.814249219415271e-05,
      "loss": 0.46240997314453125,
      "step": 6550
    },
    {
      "epoch": 1.8620493897246666,
      "grad_norm": 25.12553596496582,
      "learning_rate": 9.813965370422936e-05,
      "loss": 0.4601171016693115,
      "step": 6560
    },
    {
      "epoch": 1.8648878796480273,
      "grad_norm": 23.124380111694336,
      "learning_rate": 9.8136815214306e-05,
      "loss": 0.38168768882751464,
      "step": 6570
    },
    {
      "epoch": 1.867726369571388,
      "grad_norm": 26.014198303222656,
      "learning_rate": 9.813397672438264e-05,
      "loss": 0.4082041263580322,
      "step": 6580
    },
    {
      "epoch": 1.8705648594947488,
      "grad_norm": 23.301454544067383,
      "learning_rate": 9.813113823445928e-05,
      "loss": 0.45189976692199707,
      "step": 6590
    },
    {
      "epoch": 1.8734033494181097,
      "grad_norm": 26.73699188232422,
      "learning_rate": 9.812829974453591e-05,
      "loss": 0.41971330642700194,
      "step": 6600
    },
    {
      "epoch": 1.8762418393414704,
      "grad_norm": 19.089204788208008,
      "learning_rate": 9.812546125461255e-05,
      "loss": 0.4108613967895508,
      "step": 6610
    },
    {
      "epoch": 1.879080329264831,
      "grad_norm": 30.80247688293457,
      "learning_rate": 9.812262276468919e-05,
      "loss": 0.3896260499954224,
      "step": 6620
    },
    {
      "epoch": 1.881918819188192,
      "grad_norm": 16.476621627807617,
      "learning_rate": 9.811978427476583e-05,
      "loss": 0.4086155891418457,
      "step": 6630
    },
    {
      "epoch": 1.8847573091115528,
      "grad_norm": 30.7562198638916,
      "learning_rate": 9.811694578484247e-05,
      "loss": 0.3979541301727295,
      "step": 6640
    },
    {
      "epoch": 1.8875957990349135,
      "grad_norm": 24.587360382080078,
      "learning_rate": 9.811410729491911e-05,
      "loss": 0.4108292102813721,
      "step": 6650
    },
    {
      "epoch": 1.8904342889582741,
      "grad_norm": 29.429893493652344,
      "learning_rate": 9.811126880499576e-05,
      "loss": 0.4692255973815918,
      "step": 6660
    },
    {
      "epoch": 1.8932727788816348,
      "grad_norm": 26.78128433227539,
      "learning_rate": 9.810843031507238e-05,
      "loss": 0.4545595645904541,
      "step": 6670
    },
    {
      "epoch": 1.8961112688049957,
      "grad_norm": 32.98216247558594,
      "learning_rate": 9.810559182514902e-05,
      "loss": 0.39421920776367186,
      "step": 6680
    },
    {
      "epoch": 1.8989497587283566,
      "grad_norm": 20.42596435546875,
      "learning_rate": 9.810275333522567e-05,
      "loss": 0.3842628479003906,
      "step": 6690
    },
    {
      "epoch": 1.9017882486517173,
      "grad_norm": 15.698198318481445,
      "learning_rate": 9.80999148453023e-05,
      "loss": 0.3461360692977905,
      "step": 6700
    },
    {
      "epoch": 1.904626738575078,
      "grad_norm": 29.326900482177734,
      "learning_rate": 9.809707635537895e-05,
      "loss": 0.3522496223449707,
      "step": 6710
    },
    {
      "epoch": 1.9074652284984388,
      "grad_norm": 14.160161972045898,
      "learning_rate": 9.809423786545559e-05,
      "loss": 0.36911160945892335,
      "step": 6720
    },
    {
      "epoch": 1.9103037184217997,
      "grad_norm": 25.620317459106445,
      "learning_rate": 9.809139937553222e-05,
      "loss": 0.4214602470397949,
      "step": 6730
    },
    {
      "epoch": 1.9131422083451604,
      "grad_norm": 18.86240005493164,
      "learning_rate": 9.808856088560886e-05,
      "loss": 0.3971108913421631,
      "step": 6740
    },
    {
      "epoch": 1.915980698268521,
      "grad_norm": 26.27129364013672,
      "learning_rate": 9.80857223956855e-05,
      "loss": 0.4118006229400635,
      "step": 6750
    },
    {
      "epoch": 1.918819188191882,
      "grad_norm": 23.562463760375977,
      "learning_rate": 9.808288390576214e-05,
      "loss": 0.4610177516937256,
      "step": 6760
    },
    {
      "epoch": 1.9216576781152428,
      "grad_norm": 16.37548065185547,
      "learning_rate": 9.808004541583878e-05,
      "loss": 0.3787146806716919,
      "step": 6770
    },
    {
      "epoch": 1.9244961680386035,
      "grad_norm": 16.402164459228516,
      "learning_rate": 9.807720692591542e-05,
      "loss": 0.314204478263855,
      "step": 6780
    },
    {
      "epoch": 1.9273346579619641,
      "grad_norm": 23.63580322265625,
      "learning_rate": 9.807436843599205e-05,
      "loss": 0.4527307033538818,
      "step": 6790
    },
    {
      "epoch": 1.930173147885325,
      "grad_norm": 16.004045486450195,
      "learning_rate": 9.80715299460687e-05,
      "loss": 0.39499173164367674,
      "step": 6800
    },
    {
      "epoch": 1.933011637808686,
      "grad_norm": 25.094459533691406,
      "learning_rate": 9.806869145614534e-05,
      "loss": 0.37181546688079836,
      "step": 6810
    },
    {
      "epoch": 1.9358501277320466,
      "grad_norm": 19.275381088256836,
      "learning_rate": 9.806585296622198e-05,
      "loss": 0.3720922708511353,
      "step": 6820
    },
    {
      "epoch": 1.9386886176554072,
      "grad_norm": 24.782304763793945,
      "learning_rate": 9.80630144762986e-05,
      "loss": 0.4120591640472412,
      "step": 6830
    },
    {
      "epoch": 1.9415271075787681,
      "grad_norm": 22.285507202148438,
      "learning_rate": 9.806017598637526e-05,
      "loss": 0.32892069816589353,
      "step": 6840
    },
    {
      "epoch": 1.944365597502129,
      "grad_norm": 17.815025329589844,
      "learning_rate": 9.80573374964519e-05,
      "loss": 0.37518939971923826,
      "step": 6850
    },
    {
      "epoch": 1.9472040874254897,
      "grad_norm": 18.14682388305664,
      "learning_rate": 9.805449900652853e-05,
      "loss": 0.3778093338012695,
      "step": 6860
    },
    {
      "epoch": 1.9500425773488503,
      "grad_norm": 21.517192840576172,
      "learning_rate": 9.805166051660517e-05,
      "loss": 0.3278337001800537,
      "step": 6870
    },
    {
      "epoch": 1.9528810672722112,
      "grad_norm": 13.659611701965332,
      "learning_rate": 9.804882202668181e-05,
      "loss": 0.3915505647659302,
      "step": 6880
    },
    {
      "epoch": 1.9557195571955721,
      "grad_norm": 14.353165626525879,
      "learning_rate": 9.804598353675844e-05,
      "loss": 0.37442266941070557,
      "step": 6890
    },
    {
      "epoch": 1.9585580471189328,
      "grad_norm": 22.526288986206055,
      "learning_rate": 9.804314504683508e-05,
      "loss": 0.3907177209854126,
      "step": 6900
    },
    {
      "epoch": 1.9613965370422934,
      "grad_norm": 21.90485191345215,
      "learning_rate": 9.804030655691174e-05,
      "loss": 0.3444492816925049,
      "step": 6910
    },
    {
      "epoch": 1.964235026965654,
      "grad_norm": 14.914915084838867,
      "learning_rate": 9.803746806698836e-05,
      "loss": 0.43480501174926756,
      "step": 6920
    },
    {
      "epoch": 1.967073516889015,
      "grad_norm": 23.14503288269043,
      "learning_rate": 9.8034629577065e-05,
      "loss": 0.4471869945526123,
      "step": 6930
    },
    {
      "epoch": 1.9699120068123759,
      "grad_norm": 23.77007484436035,
      "learning_rate": 9.803179108714165e-05,
      "loss": 0.3919124364852905,
      "step": 6940
    },
    {
      "epoch": 1.9727504967357365,
      "grad_norm": 26.774288177490234,
      "learning_rate": 9.802895259721829e-05,
      "loss": 0.39920334815979003,
      "step": 6950
    },
    {
      "epoch": 1.9755889866590972,
      "grad_norm": 22.82596778869629,
      "learning_rate": 9.802611410729492e-05,
      "loss": 0.3753786087036133,
      "step": 6960
    },
    {
      "epoch": 1.978427476582458,
      "grad_norm": 17.98094940185547,
      "learning_rate": 9.802327561737157e-05,
      "loss": 0.3512239933013916,
      "step": 6970
    },
    {
      "epoch": 1.981265966505819,
      "grad_norm": 18.104963302612305,
      "learning_rate": 9.802043712744821e-05,
      "loss": 0.3635380268096924,
      "step": 6980
    },
    {
      "epoch": 1.9841044564291797,
      "grad_norm": 22.595094680786133,
      "learning_rate": 9.801759863752484e-05,
      "loss": 0.3666935920715332,
      "step": 6990
    },
    {
      "epoch": 1.9869429463525403,
      "grad_norm": 29.48496437072754,
      "learning_rate": 9.801476014760148e-05,
      "loss": 0.45305633544921875,
      "step": 7000
    },
    {
      "epoch": 1.9869429463525403,
      "eval_accuracy": 0.789979016977173,
      "eval_loss": 0.6328722834587097,
      "eval_runtime": 30.9315,
      "eval_samples_per_second": 508.445,
      "eval_steps_per_second": 7.953,
      "step": 7000
    },
    {
      "epoch": 1.9897814362759012,
      "grad_norm": 22.97522735595703,
      "learning_rate": 9.801192165767812e-05,
      "loss": 0.36184446811676024,
      "step": 7010
    },
    {
      "epoch": 1.992619926199262,
      "grad_norm": 27.06800651550293,
      "learning_rate": 9.800908316775475e-05,
      "loss": 0.37371642589569093,
      "step": 7020
    },
    {
      "epoch": 1.9954584161226228,
      "grad_norm": 22.68690299987793,
      "learning_rate": 9.800624467783139e-05,
      "loss": 0.4028749942779541,
      "step": 7030
    },
    {
      "epoch": 1.9982969060459834,
      "grad_norm": 28.094987869262695,
      "learning_rate": 9.800340618790805e-05,
      "loss": 0.4097541332244873,
      "step": 7040
    },
    {
      "epoch": 2.001135395969344,
      "grad_norm": 25.623306274414062,
      "learning_rate": 9.800056769798467e-05,
      "loss": 0.3332174777984619,
      "step": 7050
    },
    {
      "epoch": 2.003973885892705,
      "grad_norm": 26.121572494506836,
      "learning_rate": 9.799772920806132e-05,
      "loss": 0.3804765224456787,
      "step": 7060
    },
    {
      "epoch": 2.006812375816066,
      "grad_norm": 18.207355499267578,
      "learning_rate": 9.799489071813796e-05,
      "loss": 0.36211156845092773,
      "step": 7070
    },
    {
      "epoch": 2.0096508657394265,
      "grad_norm": 26.486698150634766,
      "learning_rate": 9.79920522282146e-05,
      "loss": 0.37827110290527344,
      "step": 7080
    },
    {
      "epoch": 2.012489355662787,
      "grad_norm": 27.40904998779297,
      "learning_rate": 9.798921373829123e-05,
      "loss": 0.3477623701095581,
      "step": 7090
    },
    {
      "epoch": 2.0153278455861483,
      "grad_norm": 26.22162628173828,
      "learning_rate": 9.798637524836787e-05,
      "loss": 0.4224153995513916,
      "step": 7100
    },
    {
      "epoch": 2.018166335509509,
      "grad_norm": 11.713774681091309,
      "learning_rate": 9.798353675844452e-05,
      "loss": 0.38394105434417725,
      "step": 7110
    },
    {
      "epoch": 2.0210048254328696,
      "grad_norm": 16.71647071838379,
      "learning_rate": 9.798069826852115e-05,
      "loss": 0.3284349203109741,
      "step": 7120
    },
    {
      "epoch": 2.0238433153562303,
      "grad_norm": 20.081064224243164,
      "learning_rate": 9.797785977859779e-05,
      "loss": 0.35011241436004636,
      "step": 7130
    },
    {
      "epoch": 2.0266818052795914,
      "grad_norm": 33.82632064819336,
      "learning_rate": 9.797502128867443e-05,
      "loss": 0.3723596096038818,
      "step": 7140
    },
    {
      "epoch": 2.029520295202952,
      "grad_norm": 24.423551559448242,
      "learning_rate": 9.797218279875106e-05,
      "loss": 0.34326722621917727,
      "step": 7150
    },
    {
      "epoch": 2.0323587851263127,
      "grad_norm": 15.357254981994629,
      "learning_rate": 9.79693443088277e-05,
      "loss": 0.3487834453582764,
      "step": 7160
    },
    {
      "epoch": 2.0351972750496734,
      "grad_norm": 18.3219051361084,
      "learning_rate": 9.796650581890436e-05,
      "loss": 0.3497222900390625,
      "step": 7170
    },
    {
      "epoch": 2.0380357649730345,
      "grad_norm": 17.77417755126953,
      "learning_rate": 9.796366732898098e-05,
      "loss": 0.3414299488067627,
      "step": 7180
    },
    {
      "epoch": 2.040874254896395,
      "grad_norm": 19.74038314819336,
      "learning_rate": 9.796082883905763e-05,
      "loss": 0.34571359157562254,
      "step": 7190
    },
    {
      "epoch": 2.043712744819756,
      "grad_norm": 29.782472610473633,
      "learning_rate": 9.795799034913427e-05,
      "loss": 0.36627726554870604,
      "step": 7200
    },
    {
      "epoch": 2.0465512347431165,
      "grad_norm": 18.892847061157227,
      "learning_rate": 9.795515185921091e-05,
      "loss": 0.4110989570617676,
      "step": 7210
    },
    {
      "epoch": 2.0493897246664776,
      "grad_norm": 18.120851516723633,
      "learning_rate": 9.795231336928754e-05,
      "loss": 0.3553095817565918,
      "step": 7220
    },
    {
      "epoch": 2.0522282145898383,
      "grad_norm": 18.801067352294922,
      "learning_rate": 9.794947487936418e-05,
      "loss": 0.331646466255188,
      "step": 7230
    },
    {
      "epoch": 2.055066704513199,
      "grad_norm": 21.328035354614258,
      "learning_rate": 9.794663638944083e-05,
      "loss": 0.3528684854507446,
      "step": 7240
    },
    {
      "epoch": 2.0579051944365596,
      "grad_norm": 22.32634735107422,
      "learning_rate": 9.794379789951746e-05,
      "loss": 0.2940554618835449,
      "step": 7250
    },
    {
      "epoch": 2.0607436843599207,
      "grad_norm": 22.976261138916016,
      "learning_rate": 9.79409594095941e-05,
      "loss": 0.36672842502593994,
      "step": 7260
    },
    {
      "epoch": 2.0635821742832814,
      "grad_norm": 18.52858543395996,
      "learning_rate": 9.793812091967074e-05,
      "loss": 0.32463598251342773,
      "step": 7270
    },
    {
      "epoch": 2.066420664206642,
      "grad_norm": 28.223352432250977,
      "learning_rate": 9.793528242974737e-05,
      "loss": 0.34141054153442385,
      "step": 7280
    },
    {
      "epoch": 2.0692591541300027,
      "grad_norm": 18.35302734375,
      "learning_rate": 9.793244393982401e-05,
      "loss": 0.3978186845779419,
      "step": 7290
    },
    {
      "epoch": 2.072097644053364,
      "grad_norm": 24.369665145874023,
      "learning_rate": 9.792960544990065e-05,
      "loss": 0.3742377519607544,
      "step": 7300
    },
    {
      "epoch": 2.0749361339767245,
      "grad_norm": 24.92734146118164,
      "learning_rate": 9.79267669599773e-05,
      "loss": 0.3147447109222412,
      "step": 7310
    },
    {
      "epoch": 2.077774623900085,
      "grad_norm": 26.140825271606445,
      "learning_rate": 9.792392847005394e-05,
      "loss": 0.3315492868423462,
      "step": 7320
    },
    {
      "epoch": 2.080613113823446,
      "grad_norm": 20.15946388244629,
      "learning_rate": 9.792108998013058e-05,
      "loss": 0.37865581512451174,
      "step": 7330
    },
    {
      "epoch": 2.0834516037468065,
      "grad_norm": 29.028505325317383,
      "learning_rate": 9.791825149020722e-05,
      "loss": 0.3692589998245239,
      "step": 7340
    },
    {
      "epoch": 2.0862900936701676,
      "grad_norm": 18.025115966796875,
      "learning_rate": 9.791541300028385e-05,
      "loss": 0.34498021602630613,
      "step": 7350
    },
    {
      "epoch": 2.0891285835935283,
      "grad_norm": 19.393327713012695,
      "learning_rate": 9.791257451036049e-05,
      "loss": 0.3126228332519531,
      "step": 7360
    },
    {
      "epoch": 2.091967073516889,
      "grad_norm": 25.120500564575195,
      "learning_rate": 9.790973602043714e-05,
      "loss": 0.299014687538147,
      "step": 7370
    },
    {
      "epoch": 2.0948055634402496,
      "grad_norm": 16.622690200805664,
      "learning_rate": 9.790689753051377e-05,
      "loss": 0.33715310096740725,
      "step": 7380
    },
    {
      "epoch": 2.0976440533636107,
      "grad_norm": 19.6770076751709,
      "learning_rate": 9.790405904059041e-05,
      "loss": 0.3389828443527222,
      "step": 7390
    },
    {
      "epoch": 2.1004825432869714,
      "grad_norm": 14.121631622314453,
      "learning_rate": 9.790122055066705e-05,
      "loss": 0.3276146173477173,
      "step": 7400
    },
    {
      "epoch": 2.103321033210332,
      "grad_norm": 19.206398010253906,
      "learning_rate": 9.789838206074368e-05,
      "loss": 0.33128836154937746,
      "step": 7410
    },
    {
      "epoch": 2.1061595231336927,
      "grad_norm": 24.142192840576172,
      "learning_rate": 9.789554357082032e-05,
      "loss": 0.3608201503753662,
      "step": 7420
    },
    {
      "epoch": 2.108998013057054,
      "grad_norm": 19.778400421142578,
      "learning_rate": 9.789270508089696e-05,
      "loss": 0.3426323890686035,
      "step": 7430
    },
    {
      "epoch": 2.1118365029804145,
      "grad_norm": 24.99077033996582,
      "learning_rate": 9.78898665909736e-05,
      "loss": 0.3042409420013428,
      "step": 7440
    },
    {
      "epoch": 2.114674992903775,
      "grad_norm": 22.850847244262695,
      "learning_rate": 9.788702810105025e-05,
      "loss": 0.3346649885177612,
      "step": 7450
    },
    {
      "epoch": 2.117513482827136,
      "grad_norm": 17.953664779663086,
      "learning_rate": 9.788418961112689e-05,
      "loss": 0.35388185977935793,
      "step": 7460
    },
    {
      "epoch": 2.120351972750497,
      "grad_norm": 25.097293853759766,
      "learning_rate": 9.788163497019586e-05,
      "loss": 0.3761943578720093,
      "step": 7470
    },
    {
      "epoch": 2.1231904626738576,
      "grad_norm": 17.763038635253906,
      "learning_rate": 9.78787964802725e-05,
      "loss": 0.32521159648895265,
      "step": 7480
    },
    {
      "epoch": 2.1260289525972182,
      "grad_norm": 14.84680461883545,
      "learning_rate": 9.787595799034914e-05,
      "loss": 0.29300243854522706,
      "step": 7490
    },
    {
      "epoch": 2.128867442520579,
      "grad_norm": 22.55581283569336,
      "learning_rate": 9.787311950042578e-05,
      "loss": 0.2999121189117432,
      "step": 7500
    },
    {
      "epoch": 2.128867442520579,
      "eval_accuracy": 0.8193552489349526,
      "eval_loss": 0.5586711764335632,
      "eval_runtime": 30.6876,
      "eval_samples_per_second": 512.486,
      "eval_steps_per_second": 8.016,
      "step": 7500
    },
    {
      "epoch": 2.13170593244394,
      "grad_norm": 22.64993667602539,
      "learning_rate": 9.787028101050242e-05,
      "loss": 0.33211965560913087,
      "step": 7510
    },
    {
      "epoch": 2.1345444223673007,
      "grad_norm": 29.761520385742188,
      "learning_rate": 9.786744252057906e-05,
      "loss": 0.3662957429885864,
      "step": 7520
    },
    {
      "epoch": 2.1373829122906614,
      "grad_norm": 21.352767944335938,
      "learning_rate": 9.786460403065569e-05,
      "loss": 0.34453125,
      "step": 7530
    },
    {
      "epoch": 2.140221402214022,
      "grad_norm": 18.434667587280273,
      "learning_rate": 9.786176554073233e-05,
      "loss": 0.3489013671875,
      "step": 7540
    },
    {
      "epoch": 2.1430598921373827,
      "grad_norm": 12.389979362487793,
      "learning_rate": 9.785892705080897e-05,
      "loss": 0.3259838581085205,
      "step": 7550
    },
    {
      "epoch": 2.145898382060744,
      "grad_norm": 26.911739349365234,
      "learning_rate": 9.785608856088561e-05,
      "loss": 0.3636974096298218,
      "step": 7560
    },
    {
      "epoch": 2.1487368719841045,
      "grad_norm": 17.13495445251465,
      "learning_rate": 9.785325007096226e-05,
      "loss": 0.3096443176269531,
      "step": 7570
    },
    {
      "epoch": 2.151575361907465,
      "grad_norm": 13.735611915588379,
      "learning_rate": 9.78504115810389e-05,
      "loss": 0.28812298774719236,
      "step": 7580
    },
    {
      "epoch": 2.1544138518308262,
      "grad_norm": 13.61890983581543,
      "learning_rate": 9.784757309111552e-05,
      "loss": 0.27205934524536135,
      "step": 7590
    },
    {
      "epoch": 2.157252341754187,
      "grad_norm": 17.72872543334961,
      "learning_rate": 9.784473460119217e-05,
      "loss": 0.36398863792419434,
      "step": 7600
    },
    {
      "epoch": 2.1600908316775476,
      "grad_norm": 18.36585235595703,
      "learning_rate": 9.784189611126881e-05,
      "loss": 0.32454309463500974,
      "step": 7610
    },
    {
      "epoch": 2.1629293216009082,
      "grad_norm": 28.443626403808594,
      "learning_rate": 9.783905762134545e-05,
      "loss": 0.30087525844573976,
      "step": 7620
    },
    {
      "epoch": 2.165767811524269,
      "grad_norm": 16.3806209564209,
      "learning_rate": 9.783621913142209e-05,
      "loss": 0.33084516525268554,
      "step": 7630
    },
    {
      "epoch": 2.16860630144763,
      "grad_norm": 20.08734130859375,
      "learning_rate": 9.783338064149873e-05,
      "loss": 0.35816125869750975,
      "step": 7640
    },
    {
      "epoch": 2.1714447913709907,
      "grad_norm": 16.4975643157959,
      "learning_rate": 9.783054215157536e-05,
      "loss": 0.29642434120178224,
      "step": 7650
    },
    {
      "epoch": 2.1742832812943513,
      "grad_norm": 25.36775016784668,
      "learning_rate": 9.7827703661652e-05,
      "loss": 0.31282219886779783,
      "step": 7660
    },
    {
      "epoch": 2.177121771217712,
      "grad_norm": 11.673345565795898,
      "learning_rate": 9.782486517172864e-05,
      "loss": 0.32711076736450195,
      "step": 7670
    },
    {
      "epoch": 2.179960261141073,
      "grad_norm": 23.321889877319336,
      "learning_rate": 9.782202668180528e-05,
      "loss": 0.3195405721664429,
      "step": 7680
    },
    {
      "epoch": 2.1827987510644338,
      "grad_norm": 29.165145874023438,
      "learning_rate": 9.781918819188193e-05,
      "loss": 0.3221855163574219,
      "step": 7690
    },
    {
      "epoch": 2.1856372409877944,
      "grad_norm": 19.81364631652832,
      "learning_rate": 9.781634970195857e-05,
      "loss": 0.3239097833633423,
      "step": 7700
    },
    {
      "epoch": 2.188475730911155,
      "grad_norm": 15.657233238220215,
      "learning_rate": 9.781351121203521e-05,
      "loss": 0.29824137687683105,
      "step": 7710
    },
    {
      "epoch": 2.191314220834516,
      "grad_norm": 17.727561950683594,
      "learning_rate": 9.781067272211184e-05,
      "loss": 0.2912928581237793,
      "step": 7720
    },
    {
      "epoch": 2.194152710757877,
      "grad_norm": 16.810564041137695,
      "learning_rate": 9.780783423218848e-05,
      "loss": 0.3414383172988892,
      "step": 7730
    },
    {
      "epoch": 2.1969912006812375,
      "grad_norm": 15.82119083404541,
      "learning_rate": 9.780499574226512e-05,
      "loss": 0.3083316802978516,
      "step": 7740
    },
    {
      "epoch": 2.199829690604598,
      "grad_norm": 15.556926727294922,
      "learning_rate": 9.780215725234176e-05,
      "loss": 0.3164150476455688,
      "step": 7750
    },
    {
      "epoch": 2.2026681805279593,
      "grad_norm": 21.46554946899414,
      "learning_rate": 9.77993187624184e-05,
      "loss": 0.3076106309890747,
      "step": 7760
    },
    {
      "epoch": 2.20550667045132,
      "grad_norm": 13.634368896484375,
      "learning_rate": 9.779648027249504e-05,
      "loss": 0.319075345993042,
      "step": 7770
    },
    {
      "epoch": 2.2083451603746806,
      "grad_norm": 15.295550346374512,
      "learning_rate": 9.779364178257167e-05,
      "loss": 0.29931752681732177,
      "step": 7780
    },
    {
      "epoch": 2.2111836502980413,
      "grad_norm": 24.052366256713867,
      "learning_rate": 9.779080329264831e-05,
      "loss": 0.3561910390853882,
      "step": 7790
    },
    {
      "epoch": 2.2140221402214024,
      "grad_norm": 17.625972747802734,
      "learning_rate": 9.778796480272495e-05,
      "loss": 0.330292272567749,
      "step": 7800
    },
    {
      "epoch": 2.216860630144763,
      "grad_norm": 22.240184783935547,
      "learning_rate": 9.77851263128016e-05,
      "loss": 0.3710068941116333,
      "step": 7810
    },
    {
      "epoch": 2.2196991200681238,
      "grad_norm": 16.99149513244629,
      "learning_rate": 9.778228782287824e-05,
      "loss": 0.3270531892776489,
      "step": 7820
    },
    {
      "epoch": 2.2225376099914844,
      "grad_norm": 26.887069702148438,
      "learning_rate": 9.777944933295488e-05,
      "loss": 0.28733437061309813,
      "step": 7830
    },
    {
      "epoch": 2.225376099914845,
      "grad_norm": 23.501020431518555,
      "learning_rate": 9.777661084303152e-05,
      "loss": 0.3077794075012207,
      "step": 7840
    },
    {
      "epoch": 2.228214589838206,
      "grad_norm": 14.051553726196289,
      "learning_rate": 9.777377235310815e-05,
      "loss": 0.3143048048019409,
      "step": 7850
    },
    {
      "epoch": 2.231053079761567,
      "grad_norm": 16.144062042236328,
      "learning_rate": 9.777093386318479e-05,
      "loss": 0.3136625051498413,
      "step": 7860
    },
    {
      "epoch": 2.2338915696849275,
      "grad_norm": 27.20977783203125,
      "learning_rate": 9.776809537326143e-05,
      "loss": 0.34062178134918214,
      "step": 7870
    },
    {
      "epoch": 2.236730059608288,
      "grad_norm": 12.354023933410645,
      "learning_rate": 9.776525688333806e-05,
      "loss": 0.3248645544052124,
      "step": 7880
    },
    {
      "epoch": 2.2395685495316493,
      "grad_norm": 21.091035842895508,
      "learning_rate": 9.776241839341471e-05,
      "loss": 0.32216098308563235,
      "step": 7890
    },
    {
      "epoch": 2.24240703945501,
      "grad_norm": 17.950258255004883,
      "learning_rate": 9.775957990349135e-05,
      "loss": 0.3246042013168335,
      "step": 7900
    },
    {
      "epoch": 2.2452455293783706,
      "grad_norm": 28.928653717041016,
      "learning_rate": 9.775674141356798e-05,
      "loss": 0.3273427724838257,
      "step": 7910
    },
    {
      "epoch": 2.2480840193017313,
      "grad_norm": 22.439340591430664,
      "learning_rate": 9.775390292364462e-05,
      "loss": 0.3079803228378296,
      "step": 7920
    },
    {
      "epoch": 2.2509225092250924,
      "grad_norm": 15.47249698638916,
      "learning_rate": 9.775106443372126e-05,
      "loss": 0.29904541969299314,
      "step": 7930
    },
    {
      "epoch": 2.253760999148453,
      "grad_norm": 18.008270263671875,
      "learning_rate": 9.77482259437979e-05,
      "loss": 0.3077368259429932,
      "step": 7940
    },
    {
      "epoch": 2.2565994890718137,
      "grad_norm": 19.825355529785156,
      "learning_rate": 9.774538745387455e-05,
      "loss": 0.32225110530853274,
      "step": 7950
    },
    {
      "epoch": 2.2594379789951744,
      "grad_norm": 14.30937671661377,
      "learning_rate": 9.774254896395119e-05,
      "loss": 0.31217069625854493,
      "step": 7960
    },
    {
      "epoch": 2.2622764689185355,
      "grad_norm": 23.95633888244629,
      "learning_rate": 9.773971047402783e-05,
      "loss": 0.2829510450363159,
      "step": 7970
    },
    {
      "epoch": 2.265114958841896,
      "grad_norm": 21.46575927734375,
      "learning_rate": 9.773687198410446e-05,
      "loss": 0.30298519134521484,
      "step": 7980
    },
    {
      "epoch": 2.267953448765257,
      "grad_norm": 16.935693740844727,
      "learning_rate": 9.77340334941811e-05,
      "loss": 0.2906353235244751,
      "step": 7990
    },
    {
      "epoch": 2.2707919386886175,
      "grad_norm": 14.386497497558594,
      "learning_rate": 9.773119500425774e-05,
      "loss": 0.24933280944824218,
      "step": 8000
    },
    {
      "epoch": 2.2707919386886175,
      "eval_accuracy": 0.8336618554078973,
      "eval_loss": 0.510450005531311,
      "eval_runtime": 30.7075,
      "eval_samples_per_second": 512.154,
      "eval_steps_per_second": 8.011,
      "step": 8000
    },
    {
      "epoch": 2.2736304286119786,
      "grad_norm": 15.910746574401855,
      "learning_rate": 9.772835651433437e-05,
      "loss": 0.33203461170196535,
      "step": 8010
    },
    {
      "epoch": 2.2764689185353393,
      "grad_norm": 20.63860511779785,
      "learning_rate": 9.772551802441102e-05,
      "loss": 0.3112318992614746,
      "step": 8020
    },
    {
      "epoch": 2.2793074084587,
      "grad_norm": 12.859118461608887,
      "learning_rate": 9.772267953448766e-05,
      "loss": 0.30755631923675536,
      "step": 8030
    },
    {
      "epoch": 2.2821458983820606,
      "grad_norm": 22.16994285583496,
      "learning_rate": 9.771984104456429e-05,
      "loss": 0.32435686588287355,
      "step": 8040
    },
    {
      "epoch": 2.2849843883054213,
      "grad_norm": 24.568498611450195,
      "learning_rate": 9.771700255464093e-05,
      "loss": 0.28578126430511475,
      "step": 8050
    },
    {
      "epoch": 2.2878228782287824,
      "grad_norm": 16.076248168945312,
      "learning_rate": 9.771416406471757e-05,
      "loss": 0.29748618602752686,
      "step": 8060
    },
    {
      "epoch": 2.290661368152143,
      "grad_norm": 28.567386627197266,
      "learning_rate": 9.771132557479422e-05,
      "loss": 0.2868338584899902,
      "step": 8070
    },
    {
      "epoch": 2.2934998580755037,
      "grad_norm": 25.460519790649414,
      "learning_rate": 9.770848708487086e-05,
      "loss": 0.30639047622680665,
      "step": 8080
    },
    {
      "epoch": 2.296338347998865,
      "grad_norm": 12.130402565002441,
      "learning_rate": 9.77056485949475e-05,
      "loss": 0.34181952476501465,
      "step": 8090
    },
    {
      "epoch": 2.2991768379222255,
      "grad_norm": 17.020689010620117,
      "learning_rate": 9.770281010502414e-05,
      "loss": 0.3766122579574585,
      "step": 8100
    },
    {
      "epoch": 2.302015327845586,
      "grad_norm": 19.27068519592285,
      "learning_rate": 9.769997161510077e-05,
      "loss": 0.32407164573669434,
      "step": 8110
    },
    {
      "epoch": 2.304853817768947,
      "grad_norm": 22.321697235107422,
      "learning_rate": 9.769713312517741e-05,
      "loss": 0.30633792877197263,
      "step": 8120
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 27.439123153686523,
      "learning_rate": 9.769429463525405e-05,
      "loss": 0.3041942358016968,
      "step": 8130
    },
    {
      "epoch": 2.3105307976156686,
      "grad_norm": 18.074853897094727,
      "learning_rate": 9.769145614533068e-05,
      "loss": 0.29930384159088136,
      "step": 8140
    },
    {
      "epoch": 2.3133692875390293,
      "grad_norm": 17.662498474121094,
      "learning_rate": 9.768861765540733e-05,
      "loss": 0.31606063842773435,
      "step": 8150
    },
    {
      "epoch": 2.31620777746239,
      "grad_norm": 21.386085510253906,
      "learning_rate": 9.768577916548397e-05,
      "loss": 0.33449127674102785,
      "step": 8160
    },
    {
      "epoch": 2.3190462673857506,
      "grad_norm": 23.739849090576172,
      "learning_rate": 9.76829406755606e-05,
      "loss": 0.4126762866973877,
      "step": 8170
    },
    {
      "epoch": 2.3218847573091117,
      "grad_norm": 23.911813735961914,
      "learning_rate": 9.768010218563724e-05,
      "loss": 0.3007689952850342,
      "step": 8180
    },
    {
      "epoch": 2.3247232472324724,
      "grad_norm": 18.675647735595703,
      "learning_rate": 9.767726369571389e-05,
      "loss": 0.3121593475341797,
      "step": 8190
    },
    {
      "epoch": 2.327561737155833,
      "grad_norm": 25.354106903076172,
      "learning_rate": 9.767442520579053e-05,
      "loss": 0.36575496196746826,
      "step": 8200
    },
    {
      "epoch": 2.3304002270791937,
      "grad_norm": 30.7576904296875,
      "learning_rate": 9.767158671586715e-05,
      "loss": 0.32697985172271726,
      "step": 8210
    },
    {
      "epoch": 2.333238717002555,
      "grad_norm": 20.862083435058594,
      "learning_rate": 9.766874822594381e-05,
      "loss": 0.3296887636184692,
      "step": 8220
    },
    {
      "epoch": 2.3360772069259155,
      "grad_norm": 12.933531761169434,
      "learning_rate": 9.766590973602045e-05,
      "loss": 0.2782109260559082,
      "step": 8230
    },
    {
      "epoch": 2.338915696849276,
      "grad_norm": 19.33204460144043,
      "learning_rate": 9.766307124609708e-05,
      "loss": 0.36731810569763185,
      "step": 8240
    },
    {
      "epoch": 2.341754186772637,
      "grad_norm": 33.66826629638672,
      "learning_rate": 9.766023275617372e-05,
      "loss": 0.31624741554260255,
      "step": 8250
    },
    {
      "epoch": 2.344592676695998,
      "grad_norm": 29.914854049682617,
      "learning_rate": 9.765739426625036e-05,
      "loss": 0.3358884334564209,
      "step": 8260
    },
    {
      "epoch": 2.3474311666193586,
      "grad_norm": 19.980573654174805,
      "learning_rate": 9.765455577632699e-05,
      "loss": 0.3604869365692139,
      "step": 8270
    },
    {
      "epoch": 2.3502696565427192,
      "grad_norm": 15.825258255004883,
      "learning_rate": 9.765171728640364e-05,
      "loss": 0.29133899211883546,
      "step": 8280
    },
    {
      "epoch": 2.35310814646608,
      "grad_norm": 20.521116256713867,
      "learning_rate": 9.764887879648029e-05,
      "loss": 0.3142831325531006,
      "step": 8290
    },
    {
      "epoch": 2.355946636389441,
      "grad_norm": 16.169811248779297,
      "learning_rate": 9.764604030655691e-05,
      "loss": 0.2787348985671997,
      "step": 8300
    },
    {
      "epoch": 2.3587851263128017,
      "grad_norm": 31.150718688964844,
      "learning_rate": 9.764320181663355e-05,
      "loss": 0.37035021781921384,
      "step": 8310
    },
    {
      "epoch": 2.3616236162361623,
      "grad_norm": 17.649940490722656,
      "learning_rate": 9.76403633267102e-05,
      "loss": 0.3348374843597412,
      "step": 8320
    },
    {
      "epoch": 2.364462106159523,
      "grad_norm": 13.433549880981445,
      "learning_rate": 9.763752483678684e-05,
      "loss": 0.2467961311340332,
      "step": 8330
    },
    {
      "epoch": 2.3673005960828837,
      "grad_norm": 21.340105056762695,
      "learning_rate": 9.763468634686347e-05,
      "loss": 0.31401829719543456,
      "step": 8340
    },
    {
      "epoch": 2.370139086006245,
      "grad_norm": 17.576339721679688,
      "learning_rate": 9.763184785694012e-05,
      "loss": 0.28417556285858153,
      "step": 8350
    },
    {
      "epoch": 2.3729775759296055,
      "grad_norm": 24.587268829345703,
      "learning_rate": 9.762900936701676e-05,
      "loss": 0.32737605571746825,
      "step": 8360
    },
    {
      "epoch": 2.375816065852966,
      "grad_norm": 19.446542739868164,
      "learning_rate": 9.762617087709339e-05,
      "loss": 0.29132063388824464,
      "step": 8370
    },
    {
      "epoch": 2.3786545557763272,
      "grad_norm": 11.685669898986816,
      "learning_rate": 9.762333238717003e-05,
      "loss": 0.3012071132659912,
      "step": 8380
    },
    {
      "epoch": 2.381493045699688,
      "grad_norm": 15.930485725402832,
      "learning_rate": 9.762049389724667e-05,
      "loss": 0.27760627269744875,
      "step": 8390
    },
    {
      "epoch": 2.3843315356230486,
      "grad_norm": 25.309608459472656,
      "learning_rate": 9.76176554073233e-05,
      "loss": 0.33383493423461913,
      "step": 8400
    },
    {
      "epoch": 2.3871700255464092,
      "grad_norm": 10.296014785766602,
      "learning_rate": 9.761481691739994e-05,
      "loss": 0.3022672176361084,
      "step": 8410
    },
    {
      "epoch": 2.39000851546977,
      "grad_norm": 19.012435913085938,
      "learning_rate": 9.76119784274766e-05,
      "loss": 0.3082521200180054,
      "step": 8420
    },
    {
      "epoch": 2.392847005393131,
      "grad_norm": 14.833815574645996,
      "learning_rate": 9.760913993755322e-05,
      "loss": 0.3190229177474976,
      "step": 8430
    },
    {
      "epoch": 2.3956854953164917,
      "grad_norm": 20.57181167602539,
      "learning_rate": 9.760630144762987e-05,
      "loss": 0.30741300582885744,
      "step": 8440
    },
    {
      "epoch": 2.3985239852398523,
      "grad_norm": 17.030780792236328,
      "learning_rate": 9.760346295770651e-05,
      "loss": 0.3428426027297974,
      "step": 8450
    },
    {
      "epoch": 2.401362475163213,
      "grad_norm": 22.705467224121094,
      "learning_rate": 9.760062446778315e-05,
      "loss": 0.30942912101745607,
      "step": 8460
    },
    {
      "epoch": 2.404200965086574,
      "grad_norm": 20.66131019592285,
      "learning_rate": 9.759778597785978e-05,
      "loss": 0.3165341854095459,
      "step": 8470
    },
    {
      "epoch": 2.4070394550099348,
      "grad_norm": 21.866134643554688,
      "learning_rate": 9.759494748793643e-05,
      "loss": 0.31316912174224854,
      "step": 8480
    },
    {
      "epoch": 2.4098779449332954,
      "grad_norm": 20.54009246826172,
      "learning_rate": 9.759210899801306e-05,
      "loss": 0.3653756618499756,
      "step": 8490
    },
    {
      "epoch": 2.412716434856656,
      "grad_norm": 18.931283950805664,
      "learning_rate": 9.75892705080897e-05,
      "loss": 0.3053940773010254,
      "step": 8500
    },
    {
      "epoch": 2.412716434856656,
      "eval_accuracy": 0.839130158326445,
      "eval_loss": 0.48315316438674927,
      "eval_runtime": 31.3929,
      "eval_samples_per_second": 500.973,
      "eval_steps_per_second": 7.836,
      "step": 8500
    },
    {
      "epoch": 2.415554924780017,
      "grad_norm": 32.2748908996582,
      "learning_rate": 9.758643201816634e-05,
      "loss": 0.3398677587509155,
      "step": 8510
    },
    {
      "epoch": 2.418393414703378,
      "grad_norm": 15.286317825317383,
      "learning_rate": 9.758359352824298e-05,
      "loss": 0.3296861410140991,
      "step": 8520
    },
    {
      "epoch": 2.4212319046267385,
      "grad_norm": 19.433284759521484,
      "learning_rate": 9.758075503831961e-05,
      "loss": 0.35013091564178467,
      "step": 8530
    },
    {
      "epoch": 2.424070394550099,
      "grad_norm": 21.255708694458008,
      "learning_rate": 9.757791654839625e-05,
      "loss": 0.3604008436203003,
      "step": 8540
    },
    {
      "epoch": 2.42690888447346,
      "grad_norm": 28.56644058227539,
      "learning_rate": 9.757507805847291e-05,
      "loss": 0.32774953842163085,
      "step": 8550
    },
    {
      "epoch": 2.429747374396821,
      "grad_norm": 20.850976943969727,
      "learning_rate": 9.757223956854953e-05,
      "loss": 0.2773128986358643,
      "step": 8560
    },
    {
      "epoch": 2.4325858643201816,
      "grad_norm": 18.52234649658203,
      "learning_rate": 9.756940107862618e-05,
      "loss": 0.3332803249359131,
      "step": 8570
    },
    {
      "epoch": 2.4354243542435423,
      "grad_norm": 19.95413589477539,
      "learning_rate": 9.756656258870282e-05,
      "loss": 0.3028581619262695,
      "step": 8580
    },
    {
      "epoch": 2.4382628441669034,
      "grad_norm": 41.320655822753906,
      "learning_rate": 9.756372409877945e-05,
      "loss": 0.3280078172683716,
      "step": 8590
    },
    {
      "epoch": 2.441101334090264,
      "grad_norm": 13.023110389709473,
      "learning_rate": 9.756088560885609e-05,
      "loss": 0.29077088832855225,
      "step": 8600
    },
    {
      "epoch": 2.4439398240136248,
      "grad_norm": 18.225536346435547,
      "learning_rate": 9.755804711893273e-05,
      "loss": 0.30920071601867677,
      "step": 8610
    },
    {
      "epoch": 2.4467783139369854,
      "grad_norm": 24.94522476196289,
      "learning_rate": 9.755520862900937e-05,
      "loss": 0.2795328378677368,
      "step": 8620
    },
    {
      "epoch": 2.449616803860346,
      "grad_norm": 13.974796295166016,
      "learning_rate": 9.755237013908601e-05,
      "loss": 0.316956615447998,
      "step": 8630
    },
    {
      "epoch": 2.452455293783707,
      "grad_norm": 17.750377655029297,
      "learning_rate": 9.754953164916265e-05,
      "loss": 0.30039684772491454,
      "step": 8640
    },
    {
      "epoch": 2.455293783707068,
      "grad_norm": 16.818546295166016,
      "learning_rate": 9.75466931592393e-05,
      "loss": 0.3117759943008423,
      "step": 8650
    },
    {
      "epoch": 2.4581322736304285,
      "grad_norm": 19.495594024658203,
      "learning_rate": 9.754385466931592e-05,
      "loss": 0.25922765731811526,
      "step": 8660
    },
    {
      "epoch": 2.4609707635537896,
      "grad_norm": 25.982568740844727,
      "learning_rate": 9.754101617939256e-05,
      "loss": 0.299273681640625,
      "step": 8670
    },
    {
      "epoch": 2.4638092534771503,
      "grad_norm": 22.021488189697266,
      "learning_rate": 9.753817768946922e-05,
      "loss": 0.3233955383300781,
      "step": 8680
    },
    {
      "epoch": 2.466647743400511,
      "grad_norm": 18.47892189025879,
      "learning_rate": 9.753533919954585e-05,
      "loss": 0.2856843709945679,
      "step": 8690
    },
    {
      "epoch": 2.4694862333238716,
      "grad_norm": 14.465981483459473,
      "learning_rate": 9.753250070962249e-05,
      "loss": 0.29565207958221434,
      "step": 8700
    },
    {
      "epoch": 2.4723247232472323,
      "grad_norm": 20.1898136138916,
      "learning_rate": 9.752966221969913e-05,
      "loss": 0.3390942096710205,
      "step": 8710
    },
    {
      "epoch": 2.4751632131705934,
      "grad_norm": 17.767154693603516,
      "learning_rate": 9.752682372977576e-05,
      "loss": 0.3217710733413696,
      "step": 8720
    },
    {
      "epoch": 2.478001703093954,
      "grad_norm": 30.87446403503418,
      "learning_rate": 9.75239852398524e-05,
      "loss": 0.2890691041946411,
      "step": 8730
    },
    {
      "epoch": 2.4808401930173147,
      "grad_norm": 20.463594436645508,
      "learning_rate": 9.752114674992904e-05,
      "loss": 0.30511558055877686,
      "step": 8740
    },
    {
      "epoch": 2.4836786829406754,
      "grad_norm": 23.91307258605957,
      "learning_rate": 9.751830826000568e-05,
      "loss": 0.3336294174194336,
      "step": 8750
    },
    {
      "epoch": 2.4865171728640365,
      "grad_norm": 16.377883911132812,
      "learning_rate": 9.751546977008232e-05,
      "loss": 0.28200268745422363,
      "step": 8760
    },
    {
      "epoch": 2.489355662787397,
      "grad_norm": 16.278593063354492,
      "learning_rate": 9.751263128015896e-05,
      "loss": 0.2989437818527222,
      "step": 8770
    },
    {
      "epoch": 2.492194152710758,
      "grad_norm": 22.372589111328125,
      "learning_rate": 9.75097927902356e-05,
      "loss": 0.298288631439209,
      "step": 8780
    },
    {
      "epoch": 2.4950326426341185,
      "grad_norm": 15.453804016113281,
      "learning_rate": 9.750695430031223e-05,
      "loss": 0.26072652339935304,
      "step": 8790
    },
    {
      "epoch": 2.4978711325574796,
      "grad_norm": 19.527374267578125,
      "learning_rate": 9.750411581038887e-05,
      "loss": 0.26543455123901366,
      "step": 8800
    },
    {
      "epoch": 2.5007096224808403,
      "grad_norm": 19.62587547302246,
      "learning_rate": 9.750127732046552e-05,
      "loss": 0.37260947227478025,
      "step": 8810
    },
    {
      "epoch": 2.503548112404201,
      "grad_norm": 19.925113677978516,
      "learning_rate": 9.749843883054216e-05,
      "loss": 0.33139777183532715,
      "step": 8820
    },
    {
      "epoch": 2.5063866023275616,
      "grad_norm": 17.96927261352539,
      "learning_rate": 9.74956003406188e-05,
      "loss": 0.27111947536468506,
      "step": 8830
    },
    {
      "epoch": 2.5092250922509223,
      "grad_norm": 13.09363079071045,
      "learning_rate": 9.749276185069544e-05,
      "loss": 0.2507824420928955,
      "step": 8840
    },
    {
      "epoch": 2.5120635821742834,
      "grad_norm": 10.441879272460938,
      "learning_rate": 9.748992336077207e-05,
      "loss": 0.2787137985229492,
      "step": 8850
    },
    {
      "epoch": 2.514902072097644,
      "grad_norm": 18.93663215637207,
      "learning_rate": 9.748708487084871e-05,
      "loss": 0.28047478199005127,
      "step": 8860
    },
    {
      "epoch": 2.5177405620210047,
      "grad_norm": 17.196422576904297,
      "learning_rate": 9.748424638092535e-05,
      "loss": 0.2845618486404419,
      "step": 8870
    },
    {
      "epoch": 2.520579051944366,
      "grad_norm": 22.02267837524414,
      "learning_rate": 9.748140789100199e-05,
      "loss": 0.25998194217681886,
      "step": 8880
    },
    {
      "epoch": 2.5234175418677265,
      "grad_norm": 12.58385181427002,
      "learning_rate": 9.747856940107863e-05,
      "loss": 0.24599096775054932,
      "step": 8890
    },
    {
      "epoch": 2.526256031791087,
      "grad_norm": 18.140384674072266,
      "learning_rate": 9.747573091115527e-05,
      "loss": 0.2813450336456299,
      "step": 8900
    },
    {
      "epoch": 2.529094521714448,
      "grad_norm": 19.379297256469727,
      "learning_rate": 9.747289242123192e-05,
      "loss": 0.2888960361480713,
      "step": 8910
    },
    {
      "epoch": 2.5319330116378085,
      "grad_norm": 19.3806209564209,
      "learning_rate": 9.747005393130854e-05,
      "loss": 0.31046576499938966,
      "step": 8920
    },
    {
      "epoch": 2.5347715015611696,
      "grad_norm": 19.743083953857422,
      "learning_rate": 9.746721544138518e-05,
      "loss": 0.3509793519973755,
      "step": 8930
    },
    {
      "epoch": 2.5376099914845303,
      "grad_norm": 25.64822769165039,
      "learning_rate": 9.746437695146183e-05,
      "loss": 0.32240374088287355,
      "step": 8940
    },
    {
      "epoch": 2.540448481407891,
      "grad_norm": 19.924394607543945,
      "learning_rate": 9.746153846153847e-05,
      "loss": 0.3332366466522217,
      "step": 8950
    },
    {
      "epoch": 2.543286971331252,
      "grad_norm": 14.92116641998291,
      "learning_rate": 9.745869997161511e-05,
      "loss": 0.2333456516265869,
      "step": 8960
    },
    {
      "epoch": 2.5461254612546127,
      "grad_norm": 12.108552932739258,
      "learning_rate": 9.745586148169175e-05,
      "loss": 0.286391282081604,
      "step": 8970
    },
    {
      "epoch": 2.5489639511779734,
      "grad_norm": 29.60675811767578,
      "learning_rate": 9.745302299176838e-05,
      "loss": 0.30680859088897705,
      "step": 8980
    },
    {
      "epoch": 2.551802441101334,
      "grad_norm": 22.1567440032959,
      "learning_rate": 9.745018450184502e-05,
      "loss": 0.2720050573348999,
      "step": 8990
    },
    {
      "epoch": 2.5546409310246947,
      "grad_norm": 14.802998542785645,
      "learning_rate": 9.744734601192166e-05,
      "loss": 0.271009635925293,
      "step": 9000
    },
    {
      "epoch": 2.5546409310246947,
      "eval_accuracy": 0.8457429897628282,
      "eval_loss": 0.47256723046302795,
      "eval_runtime": 31.5425,
      "eval_samples_per_second": 498.597,
      "eval_steps_per_second": 7.799,
      "step": 9000
    },
    {
      "epoch": 2.557479420948056,
      "grad_norm": 14.7639799118042,
      "learning_rate": 9.74445075219983e-05,
      "loss": 0.257672119140625,
      "step": 9010
    },
    {
      "epoch": 2.5603179108714165,
      "grad_norm": 18.669347763061523,
      "learning_rate": 9.744166903207494e-05,
      "loss": 0.28345017433166503,
      "step": 9020
    },
    {
      "epoch": 2.563156400794777,
      "grad_norm": 19.739337921142578,
      "learning_rate": 9.743883054215158e-05,
      "loss": 0.2709732294082642,
      "step": 9030
    },
    {
      "epoch": 2.565994890718138,
      "grad_norm": 16.077856063842773,
      "learning_rate": 9.743599205222823e-05,
      "loss": 0.2761526584625244,
      "step": 9040
    },
    {
      "epoch": 2.5688333806414985,
      "grad_norm": 26.328725814819336,
      "learning_rate": 9.743315356230485e-05,
      "loss": 0.2954942941665649,
      "step": 9050
    },
    {
      "epoch": 2.5716718705648596,
      "grad_norm": 17.859054565429688,
      "learning_rate": 9.74303150723815e-05,
      "loss": 0.28468611240386965,
      "step": 9060
    },
    {
      "epoch": 2.5745103604882202,
      "grad_norm": 12.289284706115723,
      "learning_rate": 9.742747658245814e-05,
      "loss": 0.24026684761047362,
      "step": 9070
    },
    {
      "epoch": 2.577348850411581,
      "grad_norm": 17.859779357910156,
      "learning_rate": 9.742463809253478e-05,
      "loss": 0.2966356992721558,
      "step": 9080
    },
    {
      "epoch": 2.580187340334942,
      "grad_norm": 17.773258209228516,
      "learning_rate": 9.742179960261142e-05,
      "loss": 0.2913994312286377,
      "step": 9090
    },
    {
      "epoch": 2.5830258302583027,
      "grad_norm": 25.365699768066406,
      "learning_rate": 9.741896111268806e-05,
      "loss": 0.2949223518371582,
      "step": 9100
    },
    {
      "epoch": 2.5858643201816633,
      "grad_norm": 17.04157066345215,
      "learning_rate": 9.741612262276469e-05,
      "loss": 0.28518497943878174,
      "step": 9110
    },
    {
      "epoch": 2.588702810105024,
      "grad_norm": 14.898686408996582,
      "learning_rate": 9.741328413284133e-05,
      "loss": 0.24409422874450684,
      "step": 9120
    },
    {
      "epoch": 2.5915413000283847,
      "grad_norm": 19.66142463684082,
      "learning_rate": 9.741044564291797e-05,
      "loss": 0.280489444732666,
      "step": 9130
    },
    {
      "epoch": 2.594379789951746,
      "grad_norm": 13.848917007446289,
      "learning_rate": 9.740760715299461e-05,
      "loss": 0.25444056987762453,
      "step": 9140
    },
    {
      "epoch": 2.5972182798751065,
      "grad_norm": 12.390437126159668,
      "learning_rate": 9.740476866307125e-05,
      "loss": 0.3165001392364502,
      "step": 9150
    },
    {
      "epoch": 2.600056769798467,
      "grad_norm": 9.805212020874023,
      "learning_rate": 9.74019301731479e-05,
      "loss": 0.2544929027557373,
      "step": 9160
    },
    {
      "epoch": 2.6028952597218282,
      "grad_norm": 21.45377540588379,
      "learning_rate": 9.739909168322454e-05,
      "loss": 0.2922818183898926,
      "step": 9170
    },
    {
      "epoch": 2.605733749645189,
      "grad_norm": 12.3560209274292,
      "learning_rate": 9.739625319330116e-05,
      "loss": 0.2575647830963135,
      "step": 9180
    },
    {
      "epoch": 2.6085722395685496,
      "grad_norm": 19.43159294128418,
      "learning_rate": 9.73934147033778e-05,
      "loss": 0.3042638301849365,
      "step": 9190
    },
    {
      "epoch": 2.61141072949191,
      "grad_norm": 19.24188995361328,
      "learning_rate": 9.739057621345445e-05,
      "loss": 0.2710337400436401,
      "step": 9200
    },
    {
      "epoch": 2.614249219415271,
      "grad_norm": 17.602937698364258,
      "learning_rate": 9.738773772353108e-05,
      "loss": 0.35931103229522704,
      "step": 9210
    },
    {
      "epoch": 2.617087709338632,
      "grad_norm": 25.234039306640625,
      "learning_rate": 9.738489923360773e-05,
      "loss": 0.3196626424789429,
      "step": 9220
    },
    {
      "epoch": 2.6199261992619927,
      "grad_norm": 21.392139434814453,
      "learning_rate": 9.738206074368437e-05,
      "loss": 0.2411559820175171,
      "step": 9230
    },
    {
      "epoch": 2.6227646891853533,
      "grad_norm": 14.251041412353516,
      "learning_rate": 9.7379222253761e-05,
      "loss": 0.27556719779968264,
      "step": 9240
    },
    {
      "epoch": 2.6256031791087144,
      "grad_norm": 16.886171340942383,
      "learning_rate": 9.737638376383764e-05,
      "loss": 0.26039795875549315,
      "step": 9250
    },
    {
      "epoch": 2.628441669032075,
      "grad_norm": 15.684998512268066,
      "learning_rate": 9.737354527391428e-05,
      "loss": 0.2410115957260132,
      "step": 9260
    },
    {
      "epoch": 2.6312801589554358,
      "grad_norm": 13.421048164367676,
      "learning_rate": 9.737070678399092e-05,
      "loss": 0.2873006105422974,
      "step": 9270
    },
    {
      "epoch": 2.6341186488787964,
      "grad_norm": 23.026832580566406,
      "learning_rate": 9.736786829406756e-05,
      "loss": 0.30839715003967283,
      "step": 9280
    },
    {
      "epoch": 2.636957138802157,
      "grad_norm": 13.106962203979492,
      "learning_rate": 9.73650298041442e-05,
      "loss": 0.31653358936309817,
      "step": 9290
    },
    {
      "epoch": 2.639795628725518,
      "grad_norm": 17.663724899291992,
      "learning_rate": 9.736219131422085e-05,
      "loss": 0.3148358821868896,
      "step": 9300
    },
    {
      "epoch": 2.642634118648879,
      "grad_norm": 13.780986785888672,
      "learning_rate": 9.735935282429748e-05,
      "loss": 0.2812326669692993,
      "step": 9310
    },
    {
      "epoch": 2.6454726085722395,
      "grad_norm": 15.959388732910156,
      "learning_rate": 9.735651433437412e-05,
      "loss": 0.27190275192260743,
      "step": 9320
    },
    {
      "epoch": 2.6483110984956,
      "grad_norm": 19.790674209594727,
      "learning_rate": 9.735367584445076e-05,
      "loss": 0.28694896697998046,
      "step": 9330
    },
    {
      "epoch": 2.651149588418961,
      "grad_norm": 24.940673828125,
      "learning_rate": 9.735083735452739e-05,
      "loss": 0.2751932621002197,
      "step": 9340
    },
    {
      "epoch": 2.653988078342322,
      "grad_norm": 14.492531776428223,
      "learning_rate": 9.734799886460404e-05,
      "loss": 0.2863935947418213,
      "step": 9350
    },
    {
      "epoch": 2.6568265682656826,
      "grad_norm": 20.169912338256836,
      "learning_rate": 9.734516037468068e-05,
      "loss": 0.3258250951766968,
      "step": 9360
    },
    {
      "epoch": 2.6596650581890433,
      "grad_norm": 25.907611846923828,
      "learning_rate": 9.734232188475731e-05,
      "loss": 0.313360857963562,
      "step": 9370
    },
    {
      "epoch": 2.6625035481124044,
      "grad_norm": 17.57925796508789,
      "learning_rate": 9.733948339483395e-05,
      "loss": 0.2980458974838257,
      "step": 9380
    },
    {
      "epoch": 2.665342038035765,
      "grad_norm": 16.746706008911133,
      "learning_rate": 9.733664490491059e-05,
      "loss": 0.28972275257110597,
      "step": 9390
    },
    {
      "epoch": 2.6681805279591257,
      "grad_norm": 30.41115951538086,
      "learning_rate": 9.733380641498723e-05,
      "loss": 0.2900726079940796,
      "step": 9400
    },
    {
      "epoch": 2.6710190178824864,
      "grad_norm": 13.418630599975586,
      "learning_rate": 9.733096792506388e-05,
      "loss": 0.2452016830444336,
      "step": 9410
    },
    {
      "epoch": 2.673857507805847,
      "grad_norm": 20.504331588745117,
      "learning_rate": 9.732812943514052e-05,
      "loss": 0.24994661808013915,
      "step": 9420
    },
    {
      "epoch": 2.676695997729208,
      "grad_norm": 29.031526565551758,
      "learning_rate": 9.732529094521714e-05,
      "loss": 0.31679112911224366,
      "step": 9430
    },
    {
      "epoch": 2.679534487652569,
      "grad_norm": 23.531126022338867,
      "learning_rate": 9.732245245529379e-05,
      "loss": 0.2530877351760864,
      "step": 9440
    },
    {
      "epoch": 2.6823729775759295,
      "grad_norm": 11.991886138916016,
      "learning_rate": 9.731961396537043e-05,
      "loss": 0.24304654598236083,
      "step": 9450
    },
    {
      "epoch": 2.6852114674992906,
      "grad_norm": 12.916497230529785,
      "learning_rate": 9.731677547544707e-05,
      "loss": 0.2319650650024414,
      "step": 9460
    },
    {
      "epoch": 2.6880499574226513,
      "grad_norm": 17.324609756469727,
      "learning_rate": 9.73139369855237e-05,
      "loss": 0.23163559436798095,
      "step": 9470
    },
    {
      "epoch": 2.690888447346012,
      "grad_norm": 19.111806869506836,
      "learning_rate": 9.731109849560035e-05,
      "loss": 0.2528048515319824,
      "step": 9480
    },
    {
      "epoch": 2.6937269372693726,
      "grad_norm": 17.495214462280273,
      "learning_rate": 9.730826000567699e-05,
      "loss": 0.30721681118011473,
      "step": 9490
    },
    {
      "epoch": 2.6965654271927333,
      "grad_norm": 21.38582992553711,
      "learning_rate": 9.730542151575362e-05,
      "loss": 0.26047730445861816,
      "step": 9500
    },
    {
      "epoch": 2.6965654271927333,
      "eval_accuracy": 0.8493673300693075,
      "eval_loss": 0.44921496510505676,
      "eval_runtime": 31.0649,
      "eval_samples_per_second": 506.263,
      "eval_steps_per_second": 7.919,
      "step": 9500
    },
    {
      "epoch": 2.6994039171160944,
      "grad_norm": 20.519363403320312,
      "learning_rate": 9.730258302583026e-05,
      "loss": 0.29499478340148927,
      "step": 9510
    },
    {
      "epoch": 2.702242407039455,
      "grad_norm": 19.01187515258789,
      "learning_rate": 9.72997445359069e-05,
      "loss": 0.2947841644287109,
      "step": 9520
    },
    {
      "epoch": 2.7050808969628157,
      "grad_norm": 21.62796401977539,
      "learning_rate": 9.729690604598353e-05,
      "loss": 0.3381195545196533,
      "step": 9530
    },
    {
      "epoch": 2.7079193868861764,
      "grad_norm": 18.938907623291016,
      "learning_rate": 9.729406755606017e-05,
      "loss": 0.2477346420288086,
      "step": 9540
    },
    {
      "epoch": 2.710757876809537,
      "grad_norm": 22.521770477294922,
      "learning_rate": 9.729122906613683e-05,
      "loss": 0.2575953960418701,
      "step": 9550
    },
    {
      "epoch": 2.713596366732898,
      "grad_norm": 22.870479583740234,
      "learning_rate": 9.728839057621346e-05,
      "loss": 0.2541611909866333,
      "step": 9560
    },
    {
      "epoch": 2.716434856656259,
      "grad_norm": 21.431659698486328,
      "learning_rate": 9.72855520862901e-05,
      "loss": 0.2780029773712158,
      "step": 9570
    },
    {
      "epoch": 2.7192733465796195,
      "grad_norm": 24.88163948059082,
      "learning_rate": 9.728271359636674e-05,
      "loss": 0.29173128604888915,
      "step": 9580
    },
    {
      "epoch": 2.7221118365029806,
      "grad_norm": 25.26793098449707,
      "learning_rate": 9.727987510644338e-05,
      "loss": 0.31362454891204833,
      "step": 9590
    },
    {
      "epoch": 2.7249503264263413,
      "grad_norm": 18.78021812438965,
      "learning_rate": 9.727703661652001e-05,
      "loss": 0.2751483917236328,
      "step": 9600
    },
    {
      "epoch": 2.727788816349702,
      "grad_norm": 29.22690200805664,
      "learning_rate": 9.727419812659666e-05,
      "loss": 0.29752316474914553,
      "step": 9610
    },
    {
      "epoch": 2.7306273062730626,
      "grad_norm": 23.88349151611328,
      "learning_rate": 9.72713596366733e-05,
      "loss": 0.25617666244506837,
      "step": 9620
    },
    {
      "epoch": 2.7334657961964233,
      "grad_norm": 12.977919578552246,
      "learning_rate": 9.726852114674993e-05,
      "loss": 0.23914566040039062,
      "step": 9630
    },
    {
      "epoch": 2.7363042861197844,
      "grad_norm": 18.041250228881836,
      "learning_rate": 9.726568265682657e-05,
      "loss": 0.26929564476013185,
      "step": 9640
    },
    {
      "epoch": 2.739142776043145,
      "grad_norm": 13.003925323486328,
      "learning_rate": 9.726284416690321e-05,
      "loss": 0.28539423942565917,
      "step": 9650
    },
    {
      "epoch": 2.7419812659665057,
      "grad_norm": 17.521793365478516,
      "learning_rate": 9.726000567697984e-05,
      "loss": 0.24657120704650878,
      "step": 9660
    },
    {
      "epoch": 2.744819755889867,
      "grad_norm": 13.071157455444336,
      "learning_rate": 9.725716718705648e-05,
      "loss": 0.2956104040145874,
      "step": 9670
    },
    {
      "epoch": 2.7476582458132275,
      "grad_norm": 15.85848331451416,
      "learning_rate": 9.725432869713314e-05,
      "loss": 0.28966264724731444,
      "step": 9680
    },
    {
      "epoch": 2.750496735736588,
      "grad_norm": 23.094085693359375,
      "learning_rate": 9.725149020720977e-05,
      "loss": 0.33190135955810546,
      "step": 9690
    },
    {
      "epoch": 2.753335225659949,
      "grad_norm": 13.309473037719727,
      "learning_rate": 9.724865171728641e-05,
      "loss": 0.2656923770904541,
      "step": 9700
    },
    {
      "epoch": 2.7561737155833095,
      "grad_norm": 15.193010330200195,
      "learning_rate": 9.724581322736305e-05,
      "loss": 0.3423954725265503,
      "step": 9710
    },
    {
      "epoch": 2.7590122055066706,
      "grad_norm": 13.571406364440918,
      "learning_rate": 9.724297473743969e-05,
      "loss": 0.2532841920852661,
      "step": 9720
    },
    {
      "epoch": 2.7618506954300313,
      "grad_norm": 11.178678512573242,
      "learning_rate": 9.724013624751632e-05,
      "loss": 0.25420119762420657,
      "step": 9730
    },
    {
      "epoch": 2.764689185353392,
      "grad_norm": 19.80664825439453,
      "learning_rate": 9.723729775759296e-05,
      "loss": 0.2993446350097656,
      "step": 9740
    },
    {
      "epoch": 2.767527675276753,
      "grad_norm": 17.400503158569336,
      "learning_rate": 9.723445926766961e-05,
      "loss": 0.26490628719329834,
      "step": 9750
    },
    {
      "epoch": 2.7703661652001137,
      "grad_norm": 17.312856674194336,
      "learning_rate": 9.723162077774624e-05,
      "loss": 0.25854156017303465,
      "step": 9760
    },
    {
      "epoch": 2.7732046551234744,
      "grad_norm": 15.450798988342285,
      "learning_rate": 9.722878228782288e-05,
      "loss": 0.222100830078125,
      "step": 9770
    },
    {
      "epoch": 2.776043145046835,
      "grad_norm": 27.243274688720703,
      "learning_rate": 9.722594379789953e-05,
      "loss": 0.26625804901123046,
      "step": 9780
    },
    {
      "epoch": 2.7788816349701957,
      "grad_norm": 14.516337394714355,
      "learning_rate": 9.722310530797615e-05,
      "loss": 0.2396000862121582,
      "step": 9790
    },
    {
      "epoch": 2.781720124893557,
      "grad_norm": 22.565673828125,
      "learning_rate": 9.72202668180528e-05,
      "loss": 0.2102586030960083,
      "step": 9800
    },
    {
      "epoch": 2.7845586148169175,
      "grad_norm": 16.52597999572754,
      "learning_rate": 9.721742832812945e-05,
      "loss": 0.263578200340271,
      "step": 9810
    },
    {
      "epoch": 2.787397104740278,
      "grad_norm": 7.076500415802002,
      "learning_rate": 9.721458983820608e-05,
      "loss": 0.2602548360824585,
      "step": 9820
    },
    {
      "epoch": 2.790235594663639,
      "grad_norm": 11.145809173583984,
      "learning_rate": 9.721175134828272e-05,
      "loss": 0.2652809858322144,
      "step": 9830
    },
    {
      "epoch": 2.7930740845869995,
      "grad_norm": 20.258697509765625,
      "learning_rate": 9.720891285835936e-05,
      "loss": 0.2715002536773682,
      "step": 9840
    },
    {
      "epoch": 2.7959125745103606,
      "grad_norm": 16.852272033691406,
      "learning_rate": 9.7206074368436e-05,
      "loss": 0.22555181980133057,
      "step": 9850
    },
    {
      "epoch": 2.7987510644337212,
      "grad_norm": 17.305248260498047,
      "learning_rate": 9.720323587851263e-05,
      "loss": 0.21825485229492186,
      "step": 9860
    },
    {
      "epoch": 2.801589554357082,
      "grad_norm": 13.663861274719238,
      "learning_rate": 9.720039738858927e-05,
      "loss": 0.24384195804595948,
      "step": 9870
    },
    {
      "epoch": 2.804428044280443,
      "grad_norm": 21.773473739624023,
      "learning_rate": 9.719755889866593e-05,
      "loss": 0.31784915924072266,
      "step": 9880
    },
    {
      "epoch": 2.8072665342038037,
      "grad_norm": 23.478778839111328,
      "learning_rate": 9.719472040874255e-05,
      "loss": 0.2890221357345581,
      "step": 9890
    },
    {
      "epoch": 2.8101050241271643,
      "grad_norm": 15.169078826904297,
      "learning_rate": 9.71918819188192e-05,
      "loss": 0.27158963680267334,
      "step": 9900
    },
    {
      "epoch": 2.812943514050525,
      "grad_norm": 21.22194480895996,
      "learning_rate": 9.718904342889584e-05,
      "loss": 0.2151625633239746,
      "step": 9910
    },
    {
      "epoch": 2.8157820039738857,
      "grad_norm": 23.666257858276367,
      "learning_rate": 9.718620493897246e-05,
      "loss": 0.2504837989807129,
      "step": 9920
    },
    {
      "epoch": 2.818620493897247,
      "grad_norm": 19.463533401489258,
      "learning_rate": 9.71833664490491e-05,
      "loss": 0.2768238544464111,
      "step": 9930
    },
    {
      "epoch": 2.8214589838206074,
      "grad_norm": 13.129846572875977,
      "learning_rate": 9.718052795912575e-05,
      "loss": 0.3096976041793823,
      "step": 9940
    },
    {
      "epoch": 2.824297473743968,
      "grad_norm": 16.599721908569336,
      "learning_rate": 9.717768946920239e-05,
      "loss": 0.27189714908599855,
      "step": 9950
    },
    {
      "epoch": 2.827135963667329,
      "grad_norm": 21.78156089782715,
      "learning_rate": 9.717485097927903e-05,
      "loss": 0.25347445011138914,
      "step": 9960
    },
    {
      "epoch": 2.82997445359069,
      "grad_norm": 14.555994987487793,
      "learning_rate": 9.717201248935567e-05,
      "loss": 0.3180224895477295,
      "step": 9970
    },
    {
      "epoch": 2.8328129435140506,
      "grad_norm": 17.541913986206055,
      "learning_rate": 9.716917399943231e-05,
      "loss": 0.2798717498779297,
      "step": 9980
    },
    {
      "epoch": 2.835651433437411,
      "grad_norm": 15.366768836975098,
      "learning_rate": 9.716633550950894e-05,
      "loss": 0.23516781330108644,
      "step": 9990
    },
    {
      "epoch": 2.838489923360772,
      "grad_norm": 18.413864135742188,
      "learning_rate": 9.716349701958558e-05,
      "loss": 0.2668823003768921,
      "step": 10000
    },
    {
      "epoch": 2.838489923360772,
      "eval_accuracy": 0.8517835569402937,
      "eval_loss": 0.4507799446582794,
      "eval_runtime": 30.7608,
      "eval_samples_per_second": 511.268,
      "eval_steps_per_second": 7.997,
      "step": 10000
    },
    {
      "epoch": 2.841328413284133,
      "grad_norm": 21.122041702270508,
      "learning_rate": 9.716065852966224e-05,
      "loss": 0.2700818061828613,
      "step": 10010
    },
    {
      "epoch": 2.8441669032074937,
      "grad_norm": 18.726993560791016,
      "learning_rate": 9.715782003973886e-05,
      "loss": 0.2364410638809204,
      "step": 10020
    },
    {
      "epoch": 2.8470053931308543,
      "grad_norm": 16.451772689819336,
      "learning_rate": 9.71549815498155e-05,
      "loss": 0.2411497116088867,
      "step": 10030
    },
    {
      "epoch": 2.8498438830542154,
      "grad_norm": 24.123859405517578,
      "learning_rate": 9.715214305989215e-05,
      "loss": 0.2889240264892578,
      "step": 10040
    },
    {
      "epoch": 2.8526823729775757,
      "grad_norm": 13.031628608703613,
      "learning_rate": 9.714930456996877e-05,
      "loss": 0.2588303804397583,
      "step": 10050
    },
    {
      "epoch": 2.8555208629009368,
      "grad_norm": 17.849319458007812,
      "learning_rate": 9.714646608004542e-05,
      "loss": 0.25627334117889405,
      "step": 10060
    },
    {
      "epoch": 2.8583593528242974,
      "grad_norm": 19.63104820251465,
      "learning_rate": 9.714362759012206e-05,
      "loss": 0.2885117530822754,
      "step": 10070
    },
    {
      "epoch": 2.861197842747658,
      "grad_norm": 13.781159400939941,
      "learning_rate": 9.71407891001987e-05,
      "loss": 0.2903496265411377,
      "step": 10080
    },
    {
      "epoch": 2.864036332671019,
      "grad_norm": 12.115455627441406,
      "learning_rate": 9.713795061027534e-05,
      "loss": 0.2852003812789917,
      "step": 10090
    },
    {
      "epoch": 2.86687482259438,
      "grad_norm": 17.76203155517578,
      "learning_rate": 9.713511212035198e-05,
      "loss": 0.2570801258087158,
      "step": 10100
    },
    {
      "epoch": 2.8697133125177405,
      "grad_norm": 18.625986099243164,
      "learning_rate": 9.713227363042862e-05,
      "loss": 0.2592076301574707,
      "step": 10110
    },
    {
      "epoch": 2.872551802441101,
      "grad_norm": 13.73946475982666,
      "learning_rate": 9.712943514050525e-05,
      "loss": 0.22207555770874024,
      "step": 10120
    },
    {
      "epoch": 2.875390292364462,
      "grad_norm": 12.634784698486328,
      "learning_rate": 9.712659665058189e-05,
      "loss": 0.24926607608795165,
      "step": 10130
    },
    {
      "epoch": 2.878228782287823,
      "grad_norm": 19.480297088623047,
      "learning_rate": 9.712375816065853e-05,
      "loss": 0.26073637008666994,
      "step": 10140
    },
    {
      "epoch": 2.8810672722111836,
      "grad_norm": 14.653205871582031,
      "learning_rate": 9.712091967073517e-05,
      "loss": 0.23487539291381837,
      "step": 10150
    },
    {
      "epoch": 2.8839057621345443,
      "grad_norm": 15.37935733795166,
      "learning_rate": 9.711808118081182e-05,
      "loss": 0.24443528652191163,
      "step": 10160
    },
    {
      "epoch": 2.8867442520579054,
      "grad_norm": 17.303863525390625,
      "learning_rate": 9.711524269088846e-05,
      "loss": 0.2529600143432617,
      "step": 10170
    },
    {
      "epoch": 2.889582741981266,
      "grad_norm": 33.37606430053711,
      "learning_rate": 9.711240420096509e-05,
      "loss": 0.2778038263320923,
      "step": 10180
    },
    {
      "epoch": 2.8924212319046267,
      "grad_norm": 19.504133224487305,
      "learning_rate": 9.710956571104173e-05,
      "loss": 0.29585649967193606,
      "step": 10190
    },
    {
      "epoch": 2.8952597218279874,
      "grad_norm": 17.723546981811523,
      "learning_rate": 9.710672722111837e-05,
      "loss": 0.24174587726593016,
      "step": 10200
    },
    {
      "epoch": 2.898098211751348,
      "grad_norm": 15.686098098754883,
      "learning_rate": 9.710388873119501e-05,
      "loss": 0.27827978134155273,
      "step": 10210
    },
    {
      "epoch": 2.900936701674709,
      "grad_norm": 14.3742094039917,
      "learning_rate": 9.710105024127165e-05,
      "loss": 0.21491069793701173,
      "step": 10220
    },
    {
      "epoch": 2.90377519159807,
      "grad_norm": 18.604324340820312,
      "learning_rate": 9.709821175134829e-05,
      "loss": 0.23633861541748047,
      "step": 10230
    },
    {
      "epoch": 2.9066136815214305,
      "grad_norm": 24.143869400024414,
      "learning_rate": 9.709537326142493e-05,
      "loss": 0.2693354368209839,
      "step": 10240
    },
    {
      "epoch": 2.9094521714447916,
      "grad_norm": 21.800477981567383,
      "learning_rate": 9.709253477150156e-05,
      "loss": 0.22950847148895265,
      "step": 10250
    },
    {
      "epoch": 2.9122906613681523,
      "grad_norm": 20.529905319213867,
      "learning_rate": 9.70896962815782e-05,
      "loss": 0.2422048568725586,
      "step": 10260
    },
    {
      "epoch": 2.915129151291513,
      "grad_norm": 20.537445068359375,
      "learning_rate": 9.708685779165484e-05,
      "loss": 0.2803179264068604,
      "step": 10270
    },
    {
      "epoch": 2.9179676412148736,
      "grad_norm": 16.101655960083008,
      "learning_rate": 9.708401930173149e-05,
      "loss": 0.24713599681854248,
      "step": 10280
    },
    {
      "epoch": 2.9208061311382343,
      "grad_norm": 10.84926986694336,
      "learning_rate": 9.708118081180813e-05,
      "loss": 0.2222980499267578,
      "step": 10290
    },
    {
      "epoch": 2.9236446210615954,
      "grad_norm": 12.458867073059082,
      "learning_rate": 9.707834232188477e-05,
      "loss": 0.24086337089538573,
      "step": 10300
    },
    {
      "epoch": 2.926483110984956,
      "grad_norm": 17.04847526550293,
      "learning_rate": 9.70755038319614e-05,
      "loss": 0.2354985237121582,
      "step": 10310
    },
    {
      "epoch": 2.9293216009083167,
      "grad_norm": 25.308412551879883,
      "learning_rate": 9.707266534203804e-05,
      "loss": 0.27909610271453855,
      "step": 10320
    },
    {
      "epoch": 2.9321600908316774,
      "grad_norm": 17.994943618774414,
      "learning_rate": 9.706982685211468e-05,
      "loss": 0.2386638879776001,
      "step": 10330
    },
    {
      "epoch": 2.934998580755038,
      "grad_norm": 14.818497657775879,
      "learning_rate": 9.706698836219132e-05,
      "loss": 0.240454363822937,
      "step": 10340
    },
    {
      "epoch": 2.937837070678399,
      "grad_norm": 13.345973014831543,
      "learning_rate": 9.706414987226796e-05,
      "loss": 0.24808545112609864,
      "step": 10350
    },
    {
      "epoch": 2.94067556060176,
      "grad_norm": 21.07094383239746,
      "learning_rate": 9.70613113823446e-05,
      "loss": 0.2733091592788696,
      "step": 10360
    },
    {
      "epoch": 2.9435140505251205,
      "grad_norm": 18.607572555541992,
      "learning_rate": 9.705847289242123e-05,
      "loss": 0.25334250926971436,
      "step": 10370
    },
    {
      "epoch": 2.9463525404484816,
      "grad_norm": 18.747655868530273,
      "learning_rate": 9.705563440249787e-05,
      "loss": 0.2256793737411499,
      "step": 10380
    },
    {
      "epoch": 2.9491910303718423,
      "grad_norm": 14.889187812805176,
      "learning_rate": 9.705279591257451e-05,
      "loss": 0.21279048919677734,
      "step": 10390
    },
    {
      "epoch": 2.952029520295203,
      "grad_norm": 14.380620002746582,
      "learning_rate": 9.704995742265115e-05,
      "loss": 0.24076695442199708,
      "step": 10400
    },
    {
      "epoch": 2.9548680102185636,
      "grad_norm": 30.139314651489258,
      "learning_rate": 9.70471189327278e-05,
      "loss": 0.26405456066131594,
      "step": 10410
    },
    {
      "epoch": 2.9577065001419243,
      "grad_norm": 22.04210090637207,
      "learning_rate": 9.704428044280444e-05,
      "loss": 0.27288987636566164,
      "step": 10420
    },
    {
      "epoch": 2.9605449900652854,
      "grad_norm": 10.629016876220703,
      "learning_rate": 9.704144195288108e-05,
      "loss": 0.19840748310089112,
      "step": 10430
    },
    {
      "epoch": 2.963383479988646,
      "grad_norm": 14.220978736877441,
      "learning_rate": 9.70386034629577e-05,
      "loss": 0.2405843496322632,
      "step": 10440
    },
    {
      "epoch": 2.9662219699120067,
      "grad_norm": 28.67279815673828,
      "learning_rate": 9.703576497303435e-05,
      "loss": 0.23020591735839843,
      "step": 10450
    },
    {
      "epoch": 2.969060459835368,
      "grad_norm": 17.291454315185547,
      "learning_rate": 9.703292648311099e-05,
      "loss": 0.2988447666168213,
      "step": 10460
    },
    {
      "epoch": 2.9718989497587285,
      "grad_norm": 13.446309089660645,
      "learning_rate": 9.703008799318762e-05,
      "loss": 0.268503737449646,
      "step": 10470
    },
    {
      "epoch": 2.974737439682089,
      "grad_norm": 12.08239459991455,
      "learning_rate": 9.702724950326427e-05,
      "loss": 0.2384115695953369,
      "step": 10480
    },
    {
      "epoch": 2.97757592960545,
      "grad_norm": 12.028255462646484,
      "learning_rate": 9.702441101334091e-05,
      "loss": 0.24648849964141845,
      "step": 10490
    },
    {
      "epoch": 2.9804144195288105,
      "grad_norm": 14.870538711547852,
      "learning_rate": 9.702157252341754e-05,
      "loss": 0.26308369636535645,
      "step": 10500
    },
    {
      "epoch": 2.9804144195288105,
      "eval_accuracy": 0.864373370636485,
      "eval_loss": 0.41203802824020386,
      "eval_runtime": 31.377,
      "eval_samples_per_second": 501.228,
      "eval_steps_per_second": 7.84,
      "step": 10500
    },
    {
      "epoch": 2.9832529094521716,
      "grad_norm": 14.513052940368652,
      "learning_rate": 9.701873403349418e-05,
      "loss": 0.2901655912399292,
      "step": 10510
    },
    {
      "epoch": 2.9860913993755323,
      "grad_norm": 24.498777389526367,
      "learning_rate": 9.701589554357082e-05,
      "loss": 0.2695554256439209,
      "step": 10520
    },
    {
      "epoch": 2.988929889298893,
      "grad_norm": 28.75697135925293,
      "learning_rate": 9.701305705364747e-05,
      "loss": 0.22772583961486817,
      "step": 10530
    },
    {
      "epoch": 2.991768379222254,
      "grad_norm": 9.938371658325195,
      "learning_rate": 9.70102185637241e-05,
      "loss": 0.2511621952056885,
      "step": 10540
    },
    {
      "epoch": 2.9946068691456147,
      "grad_norm": 20.586977005004883,
      "learning_rate": 9.700738007380075e-05,
      "loss": 0.24575157165527345,
      "step": 10550
    },
    {
      "epoch": 2.9974453590689754,
      "grad_norm": 12.629380226135254,
      "learning_rate": 9.700454158387739e-05,
      "loss": 0.2179649591445923,
      "step": 10560
    },
    {
      "epoch": 3.000283848992336,
      "grad_norm": 12.893305778503418,
      "learning_rate": 9.700170309395402e-05,
      "loss": 0.22755892276763917,
      "step": 10570
    },
    {
      "epoch": 3.0031223389156967,
      "grad_norm": 14.128294944763184,
      "learning_rate": 9.699886460403066e-05,
      "loss": 0.21660025119781495,
      "step": 10580
    },
    {
      "epoch": 3.005960828839058,
      "grad_norm": 25.21160888671875,
      "learning_rate": 9.69960261141073e-05,
      "loss": 0.22867438793182374,
      "step": 10590
    },
    {
      "epoch": 3.0087993187624185,
      "grad_norm": 13.244644165039062,
      "learning_rate": 9.699318762418393e-05,
      "loss": 0.24277143478393554,
      "step": 10600
    },
    {
      "epoch": 3.011637808685779,
      "grad_norm": 18.281429290771484,
      "learning_rate": 9.699034913426058e-05,
      "loss": 0.2278064727783203,
      "step": 10610
    },
    {
      "epoch": 3.01447629860914,
      "grad_norm": 16.516063690185547,
      "learning_rate": 9.698751064433722e-05,
      "loss": 0.21306154727935792,
      "step": 10620
    },
    {
      "epoch": 3.017314788532501,
      "grad_norm": 19.363893508911133,
      "learning_rate": 9.698467215441385e-05,
      "loss": 0.2018261432647705,
      "step": 10630
    },
    {
      "epoch": 3.0201532784558616,
      "grad_norm": 8.60107707977295,
      "learning_rate": 9.69818336644905e-05,
      "loss": 0.1982879161834717,
      "step": 10640
    },
    {
      "epoch": 3.0229917683792222,
      "grad_norm": 22.62881851196289,
      "learning_rate": 9.697899517456713e-05,
      "loss": 0.2999326229095459,
      "step": 10650
    },
    {
      "epoch": 3.025830258302583,
      "grad_norm": 13.846975326538086,
      "learning_rate": 9.697615668464378e-05,
      "loss": 0.23673901557922364,
      "step": 10660
    },
    {
      "epoch": 3.028668748225944,
      "grad_norm": 18.202268600463867,
      "learning_rate": 9.69733181947204e-05,
      "loss": 0.2435927391052246,
      "step": 10670
    },
    {
      "epoch": 3.0315072381493047,
      "grad_norm": 11.563315391540527,
      "learning_rate": 9.697047970479706e-05,
      "loss": 0.2007399320602417,
      "step": 10680
    },
    {
      "epoch": 3.0343457280726653,
      "grad_norm": 19.236194610595703,
      "learning_rate": 9.69676412148737e-05,
      "loss": 0.25865540504455564,
      "step": 10690
    },
    {
      "epoch": 3.037184217996026,
      "grad_norm": 16.745946884155273,
      "learning_rate": 9.696480272495033e-05,
      "loss": 0.20650393962860109,
      "step": 10700
    },
    {
      "epoch": 3.0400227079193867,
      "grad_norm": 18.521982192993164,
      "learning_rate": 9.696196423502697e-05,
      "loss": 0.20056042671203614,
      "step": 10710
    },
    {
      "epoch": 3.042861197842748,
      "grad_norm": 19.786373138427734,
      "learning_rate": 9.695912574510361e-05,
      "loss": 0.23505036830902098,
      "step": 10720
    },
    {
      "epoch": 3.0456996877661084,
      "grad_norm": 10.31456470489502,
      "learning_rate": 9.695628725518024e-05,
      "loss": 0.21039993762969972,
      "step": 10730
    },
    {
      "epoch": 3.048538177689469,
      "grad_norm": 10.725625991821289,
      "learning_rate": 9.69534487652569e-05,
      "loss": 0.255189847946167,
      "step": 10740
    },
    {
      "epoch": 3.0513766676128298,
      "grad_norm": 16.50446891784668,
      "learning_rate": 9.695061027533354e-05,
      "loss": 0.2216968297958374,
      "step": 10750
    },
    {
      "epoch": 3.054215157536191,
      "grad_norm": 16.032379150390625,
      "learning_rate": 9.694777178541016e-05,
      "loss": 0.20543622970581055,
      "step": 10760
    },
    {
      "epoch": 3.0570536474595515,
      "grad_norm": 12.301016807556152,
      "learning_rate": 9.69449332954868e-05,
      "loss": 0.1975337028503418,
      "step": 10770
    },
    {
      "epoch": 3.059892137382912,
      "grad_norm": 16.790771484375,
      "learning_rate": 9.694209480556345e-05,
      "loss": 0.25299646854400637,
      "step": 10780
    },
    {
      "epoch": 3.062730627306273,
      "grad_norm": 25.425148010253906,
      "learning_rate": 9.693925631564009e-05,
      "loss": 0.25143113136291506,
      "step": 10790
    },
    {
      "epoch": 3.065569117229634,
      "grad_norm": 17.002199172973633,
      "learning_rate": 9.693641782571671e-05,
      "loss": 0.23155615329742432,
      "step": 10800
    },
    {
      "epoch": 3.0684076071529947,
      "grad_norm": 12.678319931030273,
      "learning_rate": 9.693357933579337e-05,
      "loss": 0.1883660674095154,
      "step": 10810
    },
    {
      "epoch": 3.0712460970763553,
      "grad_norm": 17.38966941833496,
      "learning_rate": 9.693074084587001e-05,
      "loss": 0.24447345733642578,
      "step": 10820
    },
    {
      "epoch": 3.074084586999716,
      "grad_norm": 14.343957901000977,
      "learning_rate": 9.692790235594664e-05,
      "loss": 0.24958877563476561,
      "step": 10830
    },
    {
      "epoch": 3.076923076923077,
      "grad_norm": 12.602128028869629,
      "learning_rate": 9.692506386602328e-05,
      "loss": 0.18532336950302125,
      "step": 10840
    },
    {
      "epoch": 3.0797615668464378,
      "grad_norm": 12.524943351745605,
      "learning_rate": 9.692222537609992e-05,
      "loss": 0.24769043922424316,
      "step": 10850
    },
    {
      "epoch": 3.0826000567697984,
      "grad_norm": 13.8504638671875,
      "learning_rate": 9.691938688617655e-05,
      "loss": 0.18376538753509522,
      "step": 10860
    },
    {
      "epoch": 3.085438546693159,
      "grad_norm": 22.01972198486328,
      "learning_rate": 9.691654839625319e-05,
      "loss": 0.2254868984222412,
      "step": 10870
    },
    {
      "epoch": 3.08827703661652,
      "grad_norm": 16.021013259887695,
      "learning_rate": 9.691370990632985e-05,
      "loss": 0.21665406227111816,
      "step": 10880
    },
    {
      "epoch": 3.091115526539881,
      "grad_norm": 23.66881561279297,
      "learning_rate": 9.691087141640647e-05,
      "loss": 0.19602859020233154,
      "step": 10890
    },
    {
      "epoch": 3.0939540164632415,
      "grad_norm": 19.660036087036133,
      "learning_rate": 9.690803292648312e-05,
      "loss": 0.2671362400054932,
      "step": 10900
    },
    {
      "epoch": 3.096792506386602,
      "grad_norm": 19.80447769165039,
      "learning_rate": 9.690519443655976e-05,
      "loss": 0.20511860847473146,
      "step": 10910
    },
    {
      "epoch": 3.0996309963099633,
      "grad_norm": 12.218652725219727,
      "learning_rate": 9.69023559466364e-05,
      "loss": 0.22975389957427977,
      "step": 10920
    },
    {
      "epoch": 3.102469486233324,
      "grad_norm": 15.66241455078125,
      "learning_rate": 9.689951745671303e-05,
      "loss": 0.20163025856018066,
      "step": 10930
    },
    {
      "epoch": 3.1053079761566846,
      "grad_norm": 11.016373634338379,
      "learning_rate": 9.689667896678968e-05,
      "loss": 0.1971202850341797,
      "step": 10940
    },
    {
      "epoch": 3.1081464660800453,
      "grad_norm": 10.28996753692627,
      "learning_rate": 9.689384047686632e-05,
      "loss": 0.18606898784637452,
      "step": 10950
    },
    {
      "epoch": 3.1109849560034064,
      "grad_norm": 13.315267562866211,
      "learning_rate": 9.689100198694295e-05,
      "loss": 0.21283600330352784,
      "step": 10960
    },
    {
      "epoch": 3.113823445926767,
      "grad_norm": 15.513428688049316,
      "learning_rate": 9.688816349701959e-05,
      "loss": 0.21595463752746583,
      "step": 10970
    },
    {
      "epoch": 3.1166619358501277,
      "grad_norm": 14.546529769897461,
      "learning_rate": 9.688532500709623e-05,
      "loss": 0.18798352479934693,
      "step": 10980
    },
    {
      "epoch": 3.1195004257734884,
      "grad_norm": 17.30731964111328,
      "learning_rate": 9.688248651717286e-05,
      "loss": 0.21172025203704833,
      "step": 10990
    },
    {
      "epoch": 3.122338915696849,
      "grad_norm": 17.227052688598633,
      "learning_rate": 9.68796480272495e-05,
      "loss": 0.19874925613403321,
      "step": 11000
    },
    {
      "epoch": 3.122338915696849,
      "eval_accuracy": 0.8715584663317861,
      "eval_loss": 0.39016368985176086,
      "eval_runtime": 31.3018,
      "eval_samples_per_second": 502.432,
      "eval_steps_per_second": 7.859,
      "step": 11000
    },
    {
      "epoch": 3.12517740562021,
      "grad_norm": 29.575613021850586,
      "learning_rate": 9.687680953732616e-05,
      "loss": 0.2470324993133545,
      "step": 11010
    },
    {
      "epoch": 3.128015895543571,
      "grad_norm": 11.799863815307617,
      "learning_rate": 9.687397104740278e-05,
      "loss": 0.19145163297653198,
      "step": 11020
    },
    {
      "epoch": 3.1308543854669315,
      "grad_norm": 16.00745391845703,
      "learning_rate": 9.687113255747943e-05,
      "loss": 0.21735126972198487,
      "step": 11030
    },
    {
      "epoch": 3.133692875390292,
      "grad_norm": 16.77241325378418,
      "learning_rate": 9.686829406755607e-05,
      "loss": 0.21074838638305665,
      "step": 11040
    },
    {
      "epoch": 3.1365313653136533,
      "grad_norm": 16.993755340576172,
      "learning_rate": 9.686545557763271e-05,
      "loss": 0.23533854484558106,
      "step": 11050
    },
    {
      "epoch": 3.139369855237014,
      "grad_norm": 13.329765319824219,
      "learning_rate": 9.686261708770934e-05,
      "loss": 0.23303022384643554,
      "step": 11060
    },
    {
      "epoch": 3.1422083451603746,
      "grad_norm": 16.305633544921875,
      "learning_rate": 9.685977859778598e-05,
      "loss": 0.22058229446411132,
      "step": 11070
    },
    {
      "epoch": 3.1450468350837353,
      "grad_norm": 18.921321868896484,
      "learning_rate": 9.685694010786262e-05,
      "loss": 0.18921666145324706,
      "step": 11080
    },
    {
      "epoch": 3.1478853250070964,
      "grad_norm": 13.634056091308594,
      "learning_rate": 9.685410161793926e-05,
      "loss": 0.1699698805809021,
      "step": 11090
    },
    {
      "epoch": 3.150723814930457,
      "grad_norm": 16.445297241210938,
      "learning_rate": 9.68512631280159e-05,
      "loss": 0.2500131845474243,
      "step": 11100
    },
    {
      "epoch": 3.1535623048538177,
      "grad_norm": 14.930386543273926,
      "learning_rate": 9.684842463809254e-05,
      "loss": 0.23129620552062988,
      "step": 11110
    },
    {
      "epoch": 3.1564007947771784,
      "grad_norm": 21.532629013061523,
      "learning_rate": 9.684558614816917e-05,
      "loss": 0.24727993011474608,
      "step": 11120
    },
    {
      "epoch": 3.1592392847005395,
      "grad_norm": 18.688865661621094,
      "learning_rate": 9.684274765824581e-05,
      "loss": 0.19990921020507812,
      "step": 11130
    },
    {
      "epoch": 3.1620777746239,
      "grad_norm": 21.205503463745117,
      "learning_rate": 9.683990916832247e-05,
      "loss": 0.23148727416992188,
      "step": 11140
    },
    {
      "epoch": 3.164916264547261,
      "grad_norm": 17.55402183532715,
      "learning_rate": 9.68370706783991e-05,
      "loss": 0.23841476440429688,
      "step": 11150
    },
    {
      "epoch": 3.1677547544706215,
      "grad_norm": 19.211013793945312,
      "learning_rate": 9.683423218847574e-05,
      "loss": 0.20569534301757814,
      "step": 11160
    },
    {
      "epoch": 3.1705932443939826,
      "grad_norm": 23.154470443725586,
      "learning_rate": 9.683139369855238e-05,
      "loss": 0.19128456115722656,
      "step": 11170
    },
    {
      "epoch": 3.1734317343173433,
      "grad_norm": 20.153133392333984,
      "learning_rate": 9.682855520862902e-05,
      "loss": 0.2156306028366089,
      "step": 11180
    },
    {
      "epoch": 3.176270224240704,
      "grad_norm": 17.49920654296875,
      "learning_rate": 9.682571671870565e-05,
      "loss": 0.17330524921417237,
      "step": 11190
    },
    {
      "epoch": 3.1791087141640646,
      "grad_norm": 11.474637985229492,
      "learning_rate": 9.682287822878229e-05,
      "loss": 0.19887797832489013,
      "step": 11200
    },
    {
      "epoch": 3.1819472040874253,
      "grad_norm": 15.783727645874023,
      "learning_rate": 9.682003973885893e-05,
      "loss": 0.220632004737854,
      "step": 11210
    },
    {
      "epoch": 3.1847856940107864,
      "grad_norm": 27.546283721923828,
      "learning_rate": 9.681720124893557e-05,
      "loss": 0.25777783393859866,
      "step": 11220
    },
    {
      "epoch": 3.187624183934147,
      "grad_norm": 12.006989479064941,
      "learning_rate": 9.681436275901221e-05,
      "loss": 0.2657755374908447,
      "step": 11230
    },
    {
      "epoch": 3.1904626738575077,
      "grad_norm": 21.779420852661133,
      "learning_rate": 9.681152426908885e-05,
      "loss": 0.25869054794311525,
      "step": 11240
    },
    {
      "epoch": 3.193301163780869,
      "grad_norm": 15.037901878356934,
      "learning_rate": 9.680868577916548e-05,
      "loss": 0.19070355892181395,
      "step": 11250
    },
    {
      "epoch": 3.1961396537042295,
      "grad_norm": 13.57245922088623,
      "learning_rate": 9.680584728924212e-05,
      "loss": 0.19707906246185303,
      "step": 11260
    },
    {
      "epoch": 3.19897814362759,
      "grad_norm": 17.515422821044922,
      "learning_rate": 9.680300879931876e-05,
      "loss": 0.21522417068481445,
      "step": 11270
    },
    {
      "epoch": 3.201816633550951,
      "grad_norm": 17.102718353271484,
      "learning_rate": 9.68001703093954e-05,
      "loss": 0.2115429401397705,
      "step": 11280
    },
    {
      "epoch": 3.2046551234743115,
      "grad_norm": 22.954265594482422,
      "learning_rate": 9.679733181947205e-05,
      "loss": 0.24711823463439941,
      "step": 11290
    },
    {
      "epoch": 3.2074936133976726,
      "grad_norm": 12.616451263427734,
      "learning_rate": 9.679449332954869e-05,
      "loss": 0.2089317560195923,
      "step": 11300
    },
    {
      "epoch": 3.2103321033210332,
      "grad_norm": 10.590105056762695,
      "learning_rate": 9.679165483962532e-05,
      "loss": 0.219266939163208,
      "step": 11310
    },
    {
      "epoch": 3.213170593244394,
      "grad_norm": 21.14474105834961,
      "learning_rate": 9.678881634970196e-05,
      "loss": 0.22574622631073,
      "step": 11320
    },
    {
      "epoch": 3.2160090831677546,
      "grad_norm": 29.06281852722168,
      "learning_rate": 9.67859778597786e-05,
      "loss": 0.22632551193237305,
      "step": 11330
    },
    {
      "epoch": 3.2188475730911157,
      "grad_norm": 13.382645606994629,
      "learning_rate": 9.678313936985524e-05,
      "loss": 0.2735266208648682,
      "step": 11340
    },
    {
      "epoch": 3.2216860630144764,
      "grad_norm": 14.074967384338379,
      "learning_rate": 9.678030087993188e-05,
      "loss": 0.1919063687324524,
      "step": 11350
    },
    {
      "epoch": 3.224524552937837,
      "grad_norm": 15.914581298828125,
      "learning_rate": 9.677746239000852e-05,
      "loss": 0.2603232145309448,
      "step": 11360
    },
    {
      "epoch": 3.2273630428611977,
      "grad_norm": 18.473819732666016,
      "learning_rate": 9.677462390008516e-05,
      "loss": 0.2510881185531616,
      "step": 11370
    },
    {
      "epoch": 3.230201532784559,
      "grad_norm": 22.8264217376709,
      "learning_rate": 9.677178541016179e-05,
      "loss": 0.2662872076034546,
      "step": 11380
    },
    {
      "epoch": 3.2330400227079195,
      "grad_norm": 16.1934871673584,
      "learning_rate": 9.676894692023843e-05,
      "loss": 0.2068319797515869,
      "step": 11390
    },
    {
      "epoch": 3.23587851263128,
      "grad_norm": 11.467836380004883,
      "learning_rate": 9.676610843031508e-05,
      "loss": 0.20871257781982422,
      "step": 11400
    },
    {
      "epoch": 3.238717002554641,
      "grad_norm": 15.661177635192871,
      "learning_rate": 9.676326994039172e-05,
      "loss": 0.19979265928268433,
      "step": 11410
    },
    {
      "epoch": 3.241555492478002,
      "grad_norm": 12.25120735168457,
      "learning_rate": 9.676043145046836e-05,
      "loss": 0.2359541177749634,
      "step": 11420
    },
    {
      "epoch": 3.2443939824013626,
      "grad_norm": 15.13478946685791,
      "learning_rate": 9.6757592960545e-05,
      "loss": 0.17350263595581056,
      "step": 11430
    },
    {
      "epoch": 3.2472324723247232,
      "grad_norm": 17.476802825927734,
      "learning_rate": 9.675475447062163e-05,
      "loss": 0.224137282371521,
      "step": 11440
    },
    {
      "epoch": 3.250070962248084,
      "grad_norm": 24.93374252319336,
      "learning_rate": 9.675191598069827e-05,
      "loss": 0.20631773471832277,
      "step": 11450
    },
    {
      "epoch": 3.252909452171445,
      "grad_norm": 26.18758773803711,
      "learning_rate": 9.674907749077491e-05,
      "loss": 0.258397650718689,
      "step": 11460
    },
    {
      "epoch": 3.2557479420948057,
      "grad_norm": 13.750945091247559,
      "learning_rate": 9.674652284984389e-05,
      "loss": 0.21751623153686522,
      "step": 11470
    },
    {
      "epoch": 3.2585864320181663,
      "grad_norm": 17.093978881835938,
      "learning_rate": 9.674368435992053e-05,
      "loss": 0.18042036294937133,
      "step": 11480
    },
    {
      "epoch": 3.261424921941527,
      "grad_norm": 16.70892333984375,
      "learning_rate": 9.674084586999716e-05,
      "loss": 0.19773545265197753,
      "step": 11490
    },
    {
      "epoch": 3.2642634118648877,
      "grad_norm": 15.371891975402832,
      "learning_rate": 9.67380073800738e-05,
      "loss": 0.21635334491729735,
      "step": 11500
    },
    {
      "epoch": 3.2642634118648877,
      "eval_accuracy": 0.8788071469447447,
      "eval_loss": 0.3605816960334778,
      "eval_runtime": 30.4556,
      "eval_samples_per_second": 516.392,
      "eval_steps_per_second": 8.077,
      "step": 11500
    },
    {
      "epoch": 3.2671019017882488,
      "grad_norm": 12.340066909790039,
      "learning_rate": 9.673516889015044e-05,
      "loss": 0.18395260572433472,
      "step": 11510
    },
    {
      "epoch": 3.2699403917116094,
      "grad_norm": 15.827431678771973,
      "learning_rate": 9.673233040022708e-05,
      "loss": 0.17612080574035643,
      "step": 11520
    },
    {
      "epoch": 3.27277888163497,
      "grad_norm": 14.852681159973145,
      "learning_rate": 9.672949191030372e-05,
      "loss": 0.18402761220932007,
      "step": 11530
    },
    {
      "epoch": 3.275617371558331,
      "grad_norm": 17.557153701782227,
      "learning_rate": 9.672665342038037e-05,
      "loss": 0.20240285396575927,
      "step": 11540
    },
    {
      "epoch": 3.278455861481692,
      "grad_norm": 18.5665283203125,
      "learning_rate": 9.672381493045701e-05,
      "loss": 0.20262906551361085,
      "step": 11550
    },
    {
      "epoch": 3.2812943514050525,
      "grad_norm": 12.783393859863281,
      "learning_rate": 9.672097644053364e-05,
      "loss": 0.2017357349395752,
      "step": 11560
    },
    {
      "epoch": 3.284132841328413,
      "grad_norm": 18.27359390258789,
      "learning_rate": 9.671813795061028e-05,
      "loss": 0.21872055530548096,
      "step": 11570
    },
    {
      "epoch": 3.286971331251774,
      "grad_norm": 18.7623348236084,
      "learning_rate": 9.671529946068692e-05,
      "loss": 0.20855932235717772,
      "step": 11580
    },
    {
      "epoch": 3.289809821175135,
      "grad_norm": 10.302472114562988,
      "learning_rate": 9.671246097076356e-05,
      "loss": 0.18217077255249023,
      "step": 11590
    },
    {
      "epoch": 3.2926483110984957,
      "grad_norm": 20.435976028442383,
      "learning_rate": 9.67096224808402e-05,
      "loss": 0.2270435571670532,
      "step": 11600
    },
    {
      "epoch": 3.2954868010218563,
      "grad_norm": 10.924196243286133,
      "learning_rate": 9.670678399091684e-05,
      "loss": 0.23111917972564697,
      "step": 11610
    },
    {
      "epoch": 3.298325290945217,
      "grad_norm": 16.25908851623535,
      "learning_rate": 9.670394550099347e-05,
      "loss": 0.24706826210021973,
      "step": 11620
    },
    {
      "epoch": 3.301163780868578,
      "grad_norm": 12.188825607299805,
      "learning_rate": 9.670110701107011e-05,
      "loss": 0.20542511940002442,
      "step": 11630
    },
    {
      "epoch": 3.3040022707919388,
      "grad_norm": 17.599225997924805,
      "learning_rate": 9.669826852114675e-05,
      "loss": 0.2173429012298584,
      "step": 11640
    },
    {
      "epoch": 3.3068407607152994,
      "grad_norm": 14.959057807922363,
      "learning_rate": 9.66954300312234e-05,
      "loss": 0.19175952672958374,
      "step": 11650
    },
    {
      "epoch": 3.30967925063866,
      "grad_norm": 11.093716621398926,
      "learning_rate": 9.669259154130004e-05,
      "loss": 0.1973414897918701,
      "step": 11660
    },
    {
      "epoch": 3.312517740562021,
      "grad_norm": 20.466720581054688,
      "learning_rate": 9.668975305137668e-05,
      "loss": 0.24486141204833983,
      "step": 11670
    },
    {
      "epoch": 3.315356230485382,
      "grad_norm": 13.740006446838379,
      "learning_rate": 9.668691456145332e-05,
      "loss": 0.21142239570617677,
      "step": 11680
    },
    {
      "epoch": 3.3181947204087425,
      "grad_norm": 15.002001762390137,
      "learning_rate": 9.668407607152995e-05,
      "loss": 0.20313360691070556,
      "step": 11690
    },
    {
      "epoch": 3.321033210332103,
      "grad_norm": 8.367109298706055,
      "learning_rate": 9.668123758160659e-05,
      "loss": 0.22884137630462648,
      "step": 11700
    },
    {
      "epoch": 3.323871700255464,
      "grad_norm": 20.374404907226562,
      "learning_rate": 9.667839909168323e-05,
      "loss": 0.19571526050567628,
      "step": 11710
    },
    {
      "epoch": 3.326710190178825,
      "grad_norm": 14.498163223266602,
      "learning_rate": 9.667556060175987e-05,
      "loss": 0.20029008388519287,
      "step": 11720
    },
    {
      "epoch": 3.3295486801021856,
      "grad_norm": 8.970108985900879,
      "learning_rate": 9.667272211183651e-05,
      "loss": 0.18074514865875244,
      "step": 11730
    },
    {
      "epoch": 3.3323871700255463,
      "grad_norm": 18.17815589904785,
      "learning_rate": 9.666988362191315e-05,
      "loss": 0.20803744792938234,
      "step": 11740
    },
    {
      "epoch": 3.3352256599489074,
      "grad_norm": 12.169970512390137,
      "learning_rate": 9.666704513198978e-05,
      "loss": 0.21330771446228028,
      "step": 11750
    },
    {
      "epoch": 3.338064149872268,
      "grad_norm": 17.670808792114258,
      "learning_rate": 9.666420664206642e-05,
      "loss": 0.21243536472320557,
      "step": 11760
    },
    {
      "epoch": 3.3409026397956287,
      "grad_norm": 10.695698738098145,
      "learning_rate": 9.666136815214306e-05,
      "loss": 0.20828895568847655,
      "step": 11770
    },
    {
      "epoch": 3.3437411297189894,
      "grad_norm": 11.600749969482422,
      "learning_rate": 9.66585296622197e-05,
      "loss": 0.2504126787185669,
      "step": 11780
    },
    {
      "epoch": 3.34657961964235,
      "grad_norm": 24.69825553894043,
      "learning_rate": 9.665569117229635e-05,
      "loss": 0.2206576108932495,
      "step": 11790
    },
    {
      "epoch": 3.349418109565711,
      "grad_norm": 18.680410385131836,
      "learning_rate": 9.665285268237299e-05,
      "loss": 0.21109144687652587,
      "step": 11800
    },
    {
      "epoch": 3.352256599489072,
      "grad_norm": 18.611135482788086,
      "learning_rate": 9.665001419244963e-05,
      "loss": 0.22174811363220215,
      "step": 11810
    },
    {
      "epoch": 3.3550950894124325,
      "grad_norm": 17.504030227661133,
      "learning_rate": 9.664717570252626e-05,
      "loss": 0.21745755672454833,
      "step": 11820
    },
    {
      "epoch": 3.357933579335793,
      "grad_norm": 12.793354988098145,
      "learning_rate": 9.66443372126029e-05,
      "loss": 0.2557119369506836,
      "step": 11830
    },
    {
      "epoch": 3.3607720692591543,
      "grad_norm": 13.081313133239746,
      "learning_rate": 9.664149872267954e-05,
      "loss": 0.198041570186615,
      "step": 11840
    },
    {
      "epoch": 3.363610559182515,
      "grad_norm": 31.202713012695312,
      "learning_rate": 9.663866023275617e-05,
      "loss": 0.2324148654937744,
      "step": 11850
    },
    {
      "epoch": 3.3664490491058756,
      "grad_norm": 12.557838439941406,
      "learning_rate": 9.663582174283282e-05,
      "loss": 0.1963936448097229,
      "step": 11860
    },
    {
      "epoch": 3.3692875390292363,
      "grad_norm": 9.48736572265625,
      "learning_rate": 9.663298325290946e-05,
      "loss": 0.21634628772735595,
      "step": 11870
    },
    {
      "epoch": 3.3721260289525974,
      "grad_norm": 16.349489212036133,
      "learning_rate": 9.663014476298609e-05,
      "loss": 0.2506218910217285,
      "step": 11880
    },
    {
      "epoch": 3.374964518875958,
      "grad_norm": 10.301873207092285,
      "learning_rate": 9.662730627306273e-05,
      "loss": 0.20182456970214843,
      "step": 11890
    },
    {
      "epoch": 3.3778030087993187,
      "grad_norm": 24.21087646484375,
      "learning_rate": 9.662446778313937e-05,
      "loss": 0.21132652759552,
      "step": 11900
    },
    {
      "epoch": 3.3806414987226794,
      "grad_norm": 13.918732643127441,
      "learning_rate": 9.662162929321602e-05,
      "loss": 0.21867427825927735,
      "step": 11910
    },
    {
      "epoch": 3.3834799886460405,
      "grad_norm": 13.517630577087402,
      "learning_rate": 9.661879080329266e-05,
      "loss": 0.2162034273147583,
      "step": 11920
    },
    {
      "epoch": 3.386318478569401,
      "grad_norm": 18.84538459777832,
      "learning_rate": 9.66159523133693e-05,
      "loss": 0.21188414096832275,
      "step": 11930
    },
    {
      "epoch": 3.389156968492762,
      "grad_norm": 24.080352783203125,
      "learning_rate": 9.661311382344594e-05,
      "loss": 0.21764988899230958,
      "step": 11940
    },
    {
      "epoch": 3.3919954584161225,
      "grad_norm": 15.010066032409668,
      "learning_rate": 9.661027533352257e-05,
      "loss": 0.23972277641296386,
      "step": 11950
    },
    {
      "epoch": 3.3948339483394836,
      "grad_norm": 11.731785774230957,
      "learning_rate": 9.660743684359921e-05,
      "loss": 0.21554033756256102,
      "step": 11960
    },
    {
      "epoch": 3.3976724382628443,
      "grad_norm": 17.53189468383789,
      "learning_rate": 9.660459835367585e-05,
      "loss": 0.1813856840133667,
      "step": 11970
    },
    {
      "epoch": 3.400510928186205,
      "grad_norm": 26.230566024780273,
      "learning_rate": 9.660175986375248e-05,
      "loss": 0.1830596923828125,
      "step": 11980
    },
    {
      "epoch": 3.4033494181095656,
      "grad_norm": 9.721470832824707,
      "learning_rate": 9.659892137382913e-05,
      "loss": 0.22232580184936523,
      "step": 11990
    },
    {
      "epoch": 3.4061879080329263,
      "grad_norm": 13.806138038635254,
      "learning_rate": 9.659608288390577e-05,
      "loss": 0.20747621059417726,
      "step": 12000
    },
    {
      "epoch": 3.4061879080329263,
      "eval_accuracy": 0.8741018630380873,
      "eval_loss": 0.38113686442375183,
      "eval_runtime": 31.2293,
      "eval_samples_per_second": 503.598,
      "eval_steps_per_second": 7.877,
      "step": 12000
    },
    {
      "epoch": 3.4090263979562874,
      "grad_norm": 10.555680274963379,
      "learning_rate": 9.65932443939824e-05,
      "loss": 0.2150982141494751,
      "step": 12010
    },
    {
      "epoch": 3.411864887879648,
      "grad_norm": 13.711437225341797,
      "learning_rate": 9.659040590405904e-05,
      "loss": 0.2441176652908325,
      "step": 12020
    },
    {
      "epoch": 3.4147033778030087,
      "grad_norm": 14.219337463378906,
      "learning_rate": 9.658756741413569e-05,
      "loss": 0.20845234394073486,
      "step": 12030
    },
    {
      "epoch": 3.41754186772637,
      "grad_norm": 19.067289352416992,
      "learning_rate": 9.658472892421233e-05,
      "loss": 0.19756081104278564,
      "step": 12040
    },
    {
      "epoch": 3.4203803576497305,
      "grad_norm": 13.542132377624512,
      "learning_rate": 9.658189043428895e-05,
      "loss": 0.220689058303833,
      "step": 12050
    },
    {
      "epoch": 3.423218847573091,
      "grad_norm": 18.156505584716797,
      "learning_rate": 9.657905194436561e-05,
      "loss": 0.18415558338165283,
      "step": 12060
    },
    {
      "epoch": 3.426057337496452,
      "grad_norm": 17.014469146728516,
      "learning_rate": 9.657621345444224e-05,
      "loss": 0.22088563442230225,
      "step": 12070
    },
    {
      "epoch": 3.4288958274198125,
      "grad_norm": 14.176739692687988,
      "learning_rate": 9.657337496451888e-05,
      "loss": 0.2342505931854248,
      "step": 12080
    },
    {
      "epoch": 3.4317343173431736,
      "grad_norm": 18.460636138916016,
      "learning_rate": 9.657053647459552e-05,
      "loss": 0.19163500070571898,
      "step": 12090
    },
    {
      "epoch": 3.4345728072665342,
      "grad_norm": 21.631229400634766,
      "learning_rate": 9.656769798467216e-05,
      "loss": 0.17132577896118165,
      "step": 12100
    },
    {
      "epoch": 3.437411297189895,
      "grad_norm": 16.773332595825195,
      "learning_rate": 9.656485949474879e-05,
      "loss": 0.19617114067077637,
      "step": 12110
    },
    {
      "epoch": 3.4402497871132556,
      "grad_norm": 16.87520980834961,
      "learning_rate": 9.656202100482544e-05,
      "loss": 0.18289507627487184,
      "step": 12120
    },
    {
      "epoch": 3.4430882770366167,
      "grad_norm": 23.625057220458984,
      "learning_rate": 9.655918251490209e-05,
      "loss": 0.18987154960632324,
      "step": 12130
    },
    {
      "epoch": 3.4459267669599773,
      "grad_norm": 12.312173843383789,
      "learning_rate": 9.655634402497871e-05,
      "loss": 0.21520578861236572,
      "step": 12140
    },
    {
      "epoch": 3.448765256883338,
      "grad_norm": 11.222692489624023,
      "learning_rate": 9.655350553505535e-05,
      "loss": 0.19222476482391357,
      "step": 12150
    },
    {
      "epoch": 3.4516037468066987,
      "grad_norm": 18.204130172729492,
      "learning_rate": 9.6550667045132e-05,
      "loss": 0.16466399431228637,
      "step": 12160
    },
    {
      "epoch": 3.45444223673006,
      "grad_norm": 15.86926555633545,
      "learning_rate": 9.654782855520862e-05,
      "loss": 0.1836155414581299,
      "step": 12170
    },
    {
      "epoch": 3.4572807266534205,
      "grad_norm": 18.80999755859375,
      "learning_rate": 9.654499006528527e-05,
      "loss": 0.18175710439682008,
      "step": 12180
    },
    {
      "epoch": 3.460119216576781,
      "grad_norm": 12.337085723876953,
      "learning_rate": 9.654215157536192e-05,
      "loss": 0.22268054485321045,
      "step": 12190
    },
    {
      "epoch": 3.462957706500142,
      "grad_norm": 17.18171501159668,
      "learning_rate": 9.653931308543855e-05,
      "loss": 0.2185718059539795,
      "step": 12200
    },
    {
      "epoch": 3.4657961964235025,
      "grad_norm": 16.527612686157227,
      "learning_rate": 9.653647459551519e-05,
      "loss": 0.17128726243972778,
      "step": 12210
    },
    {
      "epoch": 3.4686346863468636,
      "grad_norm": 17.636186599731445,
      "learning_rate": 9.653363610559183e-05,
      "loss": 0.2028270721435547,
      "step": 12220
    },
    {
      "epoch": 3.4714731762702242,
      "grad_norm": 14.152024269104004,
      "learning_rate": 9.653079761566847e-05,
      "loss": 0.21990017890930175,
      "step": 12230
    },
    {
      "epoch": 3.474311666193585,
      "grad_norm": 29.320327758789062,
      "learning_rate": 9.65279591257451e-05,
      "loss": 0.18856503963470458,
      "step": 12240
    },
    {
      "epoch": 3.477150156116946,
      "grad_norm": 18.46135902404785,
      "learning_rate": 9.652512063582174e-05,
      "loss": 0.17718451023101806,
      "step": 12250
    },
    {
      "epoch": 3.4799886460403067,
      "grad_norm": 18.069143295288086,
      "learning_rate": 9.65222821458984e-05,
      "loss": 0.24148759841918946,
      "step": 12260
    },
    {
      "epoch": 3.4828271359636673,
      "grad_norm": 19.718914031982422,
      "learning_rate": 9.651944365597502e-05,
      "loss": 0.1954998254776001,
      "step": 12270
    },
    {
      "epoch": 3.485665625887028,
      "grad_norm": 14.599784851074219,
      "learning_rate": 9.651660516605167e-05,
      "loss": 0.24097776412963867,
      "step": 12280
    },
    {
      "epoch": 3.4885041158103887,
      "grad_norm": 21.432050704956055,
      "learning_rate": 9.65137666761283e-05,
      "loss": 0.20681569576263428,
      "step": 12290
    },
    {
      "epoch": 3.4913426057337498,
      "grad_norm": 7.485457420349121,
      "learning_rate": 9.651092818620493e-05,
      "loss": 0.18294481039047242,
      "step": 12300
    },
    {
      "epoch": 3.4941810956571104,
      "grad_norm": 10.234678268432617,
      "learning_rate": 9.650808969628158e-05,
      "loss": 0.2259777545928955,
      "step": 12310
    },
    {
      "epoch": 3.497019585580471,
      "grad_norm": 17.758129119873047,
      "learning_rate": 9.650525120635823e-05,
      "loss": 0.22350800037384033,
      "step": 12320
    },
    {
      "epoch": 3.499858075503832,
      "grad_norm": 16.66558074951172,
      "learning_rate": 9.650241271643486e-05,
      "loss": 0.19122059345245362,
      "step": 12330
    },
    {
      "epoch": 3.502696565427193,
      "grad_norm": 15.982666969299316,
      "learning_rate": 9.64995742265115e-05,
      "loss": 0.23698322772979735,
      "step": 12340
    },
    {
      "epoch": 3.5055350553505535,
      "grad_norm": 21.78820037841797,
      "learning_rate": 9.649673573658814e-05,
      "loss": 0.2171928644180298,
      "step": 12350
    },
    {
      "epoch": 3.508373545273914,
      "grad_norm": 24.82773208618164,
      "learning_rate": 9.649389724666478e-05,
      "loss": 0.23034744262695311,
      "step": 12360
    },
    {
      "epoch": 3.511212035197275,
      "grad_norm": 24.12273597717285,
      "learning_rate": 9.649105875674141e-05,
      "loss": 0.24887030124664306,
      "step": 12370
    },
    {
      "epoch": 3.514050525120636,
      "grad_norm": 14.947895050048828,
      "learning_rate": 9.648822026681805e-05,
      "loss": 0.16475038528442382,
      "step": 12380
    },
    {
      "epoch": 3.5168890150439966,
      "grad_norm": 15.203449249267578,
      "learning_rate": 9.648538177689471e-05,
      "loss": 0.22003324031829835,
      "step": 12390
    },
    {
      "epoch": 3.5197275049673573,
      "grad_norm": 21.137218475341797,
      "learning_rate": 9.648254328697133e-05,
      "loss": 0.19501237869262694,
      "step": 12400
    },
    {
      "epoch": 3.522565994890718,
      "grad_norm": 20.264305114746094,
      "learning_rate": 9.647970479704798e-05,
      "loss": 0.21154956817626952,
      "step": 12410
    },
    {
      "epoch": 3.5254044848140786,
      "grad_norm": 10.199299812316895,
      "learning_rate": 9.647686630712462e-05,
      "loss": 0.21863808631896972,
      "step": 12420
    },
    {
      "epoch": 3.5282429747374398,
      "grad_norm": 14.300342559814453,
      "learning_rate": 9.647402781720125e-05,
      "loss": 0.18522413969039916,
      "step": 12430
    },
    {
      "epoch": 3.5310814646608004,
      "grad_norm": 23.800457000732422,
      "learning_rate": 9.647118932727789e-05,
      "loss": 0.20307679176330568,
      "step": 12440
    },
    {
      "epoch": 3.533919954584161,
      "grad_norm": 14.285547256469727,
      "learning_rate": 9.646835083735454e-05,
      "loss": 0.17858246564865113,
      "step": 12450
    },
    {
      "epoch": 3.536758444507522,
      "grad_norm": 18.144193649291992,
      "learning_rate": 9.646551234743117e-05,
      "loss": 0.16114590167999268,
      "step": 12460
    },
    {
      "epoch": 3.539596934430883,
      "grad_norm": 13.867683410644531,
      "learning_rate": 9.646267385750781e-05,
      "loss": 0.19620091915130616,
      "step": 12470
    },
    {
      "epoch": 3.5424354243542435,
      "grad_norm": 17.450275421142578,
      "learning_rate": 9.645983536758445e-05,
      "loss": 0.20018246173858642,
      "step": 12480
    },
    {
      "epoch": 3.545273914277604,
      "grad_norm": 12.375514030456543,
      "learning_rate": 9.64569968776611e-05,
      "loss": 0.18790247440338134,
      "step": 12490
    },
    {
      "epoch": 3.548112404200965,
      "grad_norm": 14.685271263122559,
      "learning_rate": 9.645415838773772e-05,
      "loss": 0.1831412434577942,
      "step": 12500
    },
    {
      "epoch": 3.548112404200965,
      "eval_accuracy": 0.8881541298404019,
      "eval_loss": 0.3331910967826843,
      "eval_runtime": 31.2921,
      "eval_samples_per_second": 502.587,
      "eval_steps_per_second": 7.861,
      "step": 12500
    },
    {
      "epoch": 3.550950894124326,
      "grad_norm": 18.61245346069336,
      "learning_rate": 9.645131989781436e-05,
      "loss": 0.19272198677062988,
      "step": 12510
    },
    {
      "epoch": 3.5537893840476866,
      "grad_norm": 7.7992401123046875,
      "learning_rate": 9.644848140789102e-05,
      "loss": 0.18955307006835936,
      "step": 12520
    },
    {
      "epoch": 3.5566278739710473,
      "grad_norm": 18.963354110717773,
      "learning_rate": 9.644564291796765e-05,
      "loss": 0.20854313373565675,
      "step": 12530
    },
    {
      "epoch": 3.5594663638944084,
      "grad_norm": 21.342796325683594,
      "learning_rate": 9.644280442804429e-05,
      "loss": 0.2071152448654175,
      "step": 12540
    },
    {
      "epoch": 3.562304853817769,
      "grad_norm": 18.822126388549805,
      "learning_rate": 9.643996593812093e-05,
      "loss": 0.19297002553939818,
      "step": 12550
    },
    {
      "epoch": 3.5651433437411297,
      "grad_norm": 9.989988327026367,
      "learning_rate": 9.643712744819756e-05,
      "loss": 0.203399395942688,
      "step": 12560
    },
    {
      "epoch": 3.5679818336644904,
      "grad_norm": 7.88776159286499,
      "learning_rate": 9.64342889582742e-05,
      "loss": 0.14284805059432984,
      "step": 12570
    },
    {
      "epoch": 3.570820323587851,
      "grad_norm": 19.929224014282227,
      "learning_rate": 9.643145046835084e-05,
      "loss": 0.20520339012145997,
      "step": 12580
    },
    {
      "epoch": 3.573658813511212,
      "grad_norm": 13.769087791442871,
      "learning_rate": 9.642861197842748e-05,
      "loss": 0.20943028926849366,
      "step": 12590
    },
    {
      "epoch": 3.576497303434573,
      "grad_norm": 16.053749084472656,
      "learning_rate": 9.642577348850412e-05,
      "loss": 0.24009351730346679,
      "step": 12600
    },
    {
      "epoch": 3.5793357933579335,
      "grad_norm": 18.332489013671875,
      "learning_rate": 9.642293499858076e-05,
      "loss": 0.1855130076408386,
      "step": 12610
    },
    {
      "epoch": 3.5821742832812946,
      "grad_norm": 14.052803039550781,
      "learning_rate": 9.64200965086574e-05,
      "loss": 0.16475338935852052,
      "step": 12620
    },
    {
      "epoch": 3.5850127732046553,
      "grad_norm": 13.83219051361084,
      "learning_rate": 9.641725801873403e-05,
      "loss": 0.1916528820991516,
      "step": 12630
    },
    {
      "epoch": 3.587851263128016,
      "grad_norm": 14.276023864746094,
      "learning_rate": 9.641441952881067e-05,
      "loss": 0.18999066352844238,
      "step": 12640
    },
    {
      "epoch": 3.5906897530513766,
      "grad_norm": 21.867109298706055,
      "learning_rate": 9.641158103888733e-05,
      "loss": 0.23952841758728027,
      "step": 12650
    },
    {
      "epoch": 3.5935282429747373,
      "grad_norm": 14.864910125732422,
      "learning_rate": 9.640874254896396e-05,
      "loss": 0.15287126302719117,
      "step": 12660
    },
    {
      "epoch": 3.5963667328980984,
      "grad_norm": 10.48604679107666,
      "learning_rate": 9.64059040590406e-05,
      "loss": 0.15772616863250732,
      "step": 12670
    },
    {
      "epoch": 3.599205222821459,
      "grad_norm": 14.08375358581543,
      "learning_rate": 9.640306556911724e-05,
      "loss": 0.17585846185684204,
      "step": 12680
    },
    {
      "epoch": 3.6020437127448197,
      "grad_norm": 9.9944429397583,
      "learning_rate": 9.640022707919387e-05,
      "loss": 0.17443268299102782,
      "step": 12690
    },
    {
      "epoch": 3.6048822026681804,
      "grad_norm": 18.99468994140625,
      "learning_rate": 9.639738858927051e-05,
      "loss": 0.23786187171936035,
      "step": 12700
    },
    {
      "epoch": 3.607720692591541,
      "grad_norm": 16.31536102294922,
      "learning_rate": 9.639455009934715e-05,
      "loss": 0.20249197483062745,
      "step": 12710
    },
    {
      "epoch": 3.610559182514902,
      "grad_norm": 18.80646514892578,
      "learning_rate": 9.639171160942379e-05,
      "loss": 0.1999928832054138,
      "step": 12720
    },
    {
      "epoch": 3.613397672438263,
      "grad_norm": 21.134939193725586,
      "learning_rate": 9.638887311950043e-05,
      "loss": 0.16092418432235717,
      "step": 12730
    },
    {
      "epoch": 3.6162361623616235,
      "grad_norm": 18.907377243041992,
      "learning_rate": 9.638603462957707e-05,
      "loss": 0.1759500741958618,
      "step": 12740
    },
    {
      "epoch": 3.6190746522849846,
      "grad_norm": 16.169979095458984,
      "learning_rate": 9.638319613965371e-05,
      "loss": 0.1590247392654419,
      "step": 12750
    },
    {
      "epoch": 3.6219131422083453,
      "grad_norm": 17.707767486572266,
      "learning_rate": 9.638035764973034e-05,
      "loss": 0.19387505054473878,
      "step": 12760
    },
    {
      "epoch": 3.624751632131706,
      "grad_norm": 14.954610824584961,
      "learning_rate": 9.637751915980698e-05,
      "loss": 0.22697823047637938,
      "step": 12770
    },
    {
      "epoch": 3.6275901220550666,
      "grad_norm": 16.924896240234375,
      "learning_rate": 9.637468066988363e-05,
      "loss": 0.1832563042640686,
      "step": 12780
    },
    {
      "epoch": 3.6304286119784273,
      "grad_norm": 9.252005577087402,
      "learning_rate": 9.637184217996027e-05,
      "loss": 0.18793096542358398,
      "step": 12790
    },
    {
      "epoch": 3.6332671019017884,
      "grad_norm": 15.584880828857422,
      "learning_rate": 9.636900369003691e-05,
      "loss": 0.16042823791503907,
      "step": 12800
    },
    {
      "epoch": 3.636105591825149,
      "grad_norm": 21.544593811035156,
      "learning_rate": 9.636616520011355e-05,
      "loss": 0.20613737106323243,
      "step": 12810
    },
    {
      "epoch": 3.6389440817485097,
      "grad_norm": 28.40416145324707,
      "learning_rate": 9.636332671019018e-05,
      "loss": 0.20446252822875977,
      "step": 12820
    },
    {
      "epoch": 3.641782571671871,
      "grad_norm": 21.0313777923584,
      "learning_rate": 9.636048822026682e-05,
      "loss": 0.232306170463562,
      "step": 12830
    },
    {
      "epoch": 3.6446210615952315,
      "grad_norm": 19.987058639526367,
      "learning_rate": 9.635764973034346e-05,
      "loss": 0.18236492872238158,
      "step": 12840
    },
    {
      "epoch": 3.647459551518592,
      "grad_norm": 14.75471305847168,
      "learning_rate": 9.63548112404201e-05,
      "loss": 0.2041480302810669,
      "step": 12850
    },
    {
      "epoch": 3.650298041441953,
      "grad_norm": 10.284089088439941,
      "learning_rate": 9.635197275049674e-05,
      "loss": 0.1662386417388916,
      "step": 12860
    },
    {
      "epoch": 3.6531365313653135,
      "grad_norm": 22.920259475708008,
      "learning_rate": 9.634913426057338e-05,
      "loss": 0.20379331111907958,
      "step": 12870
    },
    {
      "epoch": 3.6559750212886746,
      "grad_norm": 17.91655731201172,
      "learning_rate": 9.634629577065003e-05,
      "loss": 0.19459644556045533,
      "step": 12880
    },
    {
      "epoch": 3.6588135112120352,
      "grad_norm": 14.011137008666992,
      "learning_rate": 9.634345728072665e-05,
      "loss": 0.1959167242050171,
      "step": 12890
    },
    {
      "epoch": 3.661652001135396,
      "grad_norm": 9.336225509643555,
      "learning_rate": 9.63406187908033e-05,
      "loss": 0.20528478622436525,
      "step": 12900
    },
    {
      "epoch": 3.664490491058757,
      "grad_norm": 14.320165634155273,
      "learning_rate": 9.633778030087994e-05,
      "loss": 0.19499990940093995,
      "step": 12910
    },
    {
      "epoch": 3.6673289809821172,
      "grad_norm": 12.04212760925293,
      "learning_rate": 9.633494181095658e-05,
      "loss": 0.21102490425109863,
      "step": 12920
    },
    {
      "epoch": 3.6701674709054783,
      "grad_norm": 17.68199348449707,
      "learning_rate": 9.633210332103322e-05,
      "loss": 0.1669550657272339,
      "step": 12930
    },
    {
      "epoch": 3.673005960828839,
      "grad_norm": 13.608904838562012,
      "learning_rate": 9.632926483110986e-05,
      "loss": 0.2338263511657715,
      "step": 12940
    },
    {
      "epoch": 3.6758444507521997,
      "grad_norm": 20.929216384887695,
      "learning_rate": 9.632642634118649e-05,
      "loss": 0.20095372200012207,
      "step": 12950
    },
    {
      "epoch": 3.678682940675561,
      "grad_norm": 10.120415687561035,
      "learning_rate": 9.632358785126313e-05,
      "loss": 0.19782148599624633,
      "step": 12960
    },
    {
      "epoch": 3.6815214305989215,
      "grad_norm": 17.33092498779297,
      "learning_rate": 9.632074936133977e-05,
      "loss": 0.19996261596679688,
      "step": 12970
    },
    {
      "epoch": 3.684359920522282,
      "grad_norm": 15.53795337677002,
      "learning_rate": 9.631791087141641e-05,
      "loss": 0.19260650873184204,
      "step": 12980
    },
    {
      "epoch": 3.687198410445643,
      "grad_norm": 17.326200485229492,
      "learning_rate": 9.631507238149305e-05,
      "loss": 0.1979435443878174,
      "step": 12990
    },
    {
      "epoch": 3.6900369003690034,
      "grad_norm": 12.190852165222168,
      "learning_rate": 9.63122338915697e-05,
      "loss": 0.2052302837371826,
      "step": 13000
    },
    {
      "epoch": 3.6900369003690034,
      "eval_accuracy": 0.8860558275577033,
      "eval_loss": 0.3456185460090637,
      "eval_runtime": 31.2758,
      "eval_samples_per_second": 502.849,
      "eval_steps_per_second": 7.866,
      "step": 13000
    },
    {
      "epoch": 3.6928753902923646,
      "grad_norm": 11.41498851776123,
      "learning_rate": 9.630939540164632e-05,
      "loss": 0.16218639612197877,
      "step": 13010
    },
    {
      "epoch": 3.6957138802157252,
      "grad_norm": 21.549772262573242,
      "learning_rate": 9.630655691172296e-05,
      "loss": 0.18964788913726807,
      "step": 13020
    },
    {
      "epoch": 3.698552370139086,
      "grad_norm": 15.30506706237793,
      "learning_rate": 9.63037184217996e-05,
      "loss": 0.17965257167816162,
      "step": 13030
    },
    {
      "epoch": 3.701390860062447,
      "grad_norm": 18.242279052734375,
      "learning_rate": 9.630087993187625e-05,
      "loss": 0.22642974853515624,
      "step": 13040
    },
    {
      "epoch": 3.7042293499858077,
      "grad_norm": 13.957133293151855,
      "learning_rate": 9.629804144195289e-05,
      "loss": 0.2009499788284302,
      "step": 13050
    },
    {
      "epoch": 3.7070678399091683,
      "grad_norm": 19.624011993408203,
      "learning_rate": 9.629520295202953e-05,
      "loss": 0.21221296787261962,
      "step": 13060
    },
    {
      "epoch": 3.709906329832529,
      "grad_norm": 19.296112060546875,
      "learning_rate": 9.629236446210617e-05,
      "loss": 0.1756616234779358,
      "step": 13070
    },
    {
      "epoch": 3.7127448197558897,
      "grad_norm": 3.713320255279541,
      "learning_rate": 9.62895259721828e-05,
      "loss": 0.1612323999404907,
      "step": 13080
    },
    {
      "epoch": 3.7155833096792508,
      "grad_norm": 12.882383346557617,
      "learning_rate": 9.628668748225944e-05,
      "loss": 0.20692853927612304,
      "step": 13090
    },
    {
      "epoch": 3.7184217996026114,
      "grad_norm": 15.465047836303711,
      "learning_rate": 9.628384899233608e-05,
      "loss": 0.23029165267944335,
      "step": 13100
    },
    {
      "epoch": 3.721260289525972,
      "grad_norm": 20.43402099609375,
      "learning_rate": 9.628101050241271e-05,
      "loss": 0.20411965847015381,
      "step": 13110
    },
    {
      "epoch": 3.724098779449333,
      "grad_norm": 18.477548599243164,
      "learning_rate": 9.627817201248936e-05,
      "loss": 0.22115280628204345,
      "step": 13120
    },
    {
      "epoch": 3.726937269372694,
      "grad_norm": 14.626617431640625,
      "learning_rate": 9.6275333522566e-05,
      "loss": 0.20535669326782227,
      "step": 13130
    },
    {
      "epoch": 3.7297757592960545,
      "grad_norm": 9.742225646972656,
      "learning_rate": 9.627249503264263e-05,
      "loss": 0.19262950420379638,
      "step": 13140
    },
    {
      "epoch": 3.732614249219415,
      "grad_norm": 13.3360013961792,
      "learning_rate": 9.626965654271928e-05,
      "loss": 0.20899879932403564,
      "step": 13150
    },
    {
      "epoch": 3.735452739142776,
      "grad_norm": 21.276222229003906,
      "learning_rate": 9.626681805279592e-05,
      "loss": 0.18748929500579833,
      "step": 13160
    },
    {
      "epoch": 3.738291229066137,
      "grad_norm": 17.293577194213867,
      "learning_rate": 9.626397956287256e-05,
      "loss": 0.16237835884094237,
      "step": 13170
    },
    {
      "epoch": 3.7411297189894976,
      "grad_norm": 21.195192337036133,
      "learning_rate": 9.626114107294919e-05,
      "loss": 0.20309970378875733,
      "step": 13180
    },
    {
      "epoch": 3.7439682089128583,
      "grad_norm": 14.44090747833252,
      "learning_rate": 9.625830258302584e-05,
      "loss": 0.235282564163208,
      "step": 13190
    },
    {
      "epoch": 3.746806698836219,
      "grad_norm": 13.884706497192383,
      "learning_rate": 9.625546409310248e-05,
      "loss": 0.18537503480911255,
      "step": 13200
    },
    {
      "epoch": 3.7496451887595796,
      "grad_norm": 14.233735084533691,
      "learning_rate": 9.625262560317911e-05,
      "loss": 0.2119133710861206,
      "step": 13210
    },
    {
      "epoch": 3.7524836786829407,
      "grad_norm": 16.935958862304688,
      "learning_rate": 9.624978711325575e-05,
      "loss": 0.20455503463745117,
      "step": 13220
    },
    {
      "epoch": 3.7553221686063014,
      "grad_norm": 11.480093002319336,
      "learning_rate": 9.624694862333239e-05,
      "loss": 0.2135927677154541,
      "step": 13230
    },
    {
      "epoch": 3.758160658529662,
      "grad_norm": 16.112228393554688,
      "learning_rate": 9.624411013340902e-05,
      "loss": 0.19972648620605468,
      "step": 13240
    },
    {
      "epoch": 3.760999148453023,
      "grad_norm": 12.924802780151367,
      "learning_rate": 9.624127164348568e-05,
      "loss": 0.20564627647399902,
      "step": 13250
    },
    {
      "epoch": 3.763837638376384,
      "grad_norm": 18.588260650634766,
      "learning_rate": 9.623843315356232e-05,
      "loss": 0.20453810691833496,
      "step": 13260
    },
    {
      "epoch": 3.7666761282997445,
      "grad_norm": 18.94190788269043,
      "learning_rate": 9.623559466363894e-05,
      "loss": 0.19725641012191772,
      "step": 13270
    },
    {
      "epoch": 3.769514618223105,
      "grad_norm": 27.818538665771484,
      "learning_rate": 9.623275617371559e-05,
      "loss": 0.17562634944915773,
      "step": 13280
    },
    {
      "epoch": 3.772353108146466,
      "grad_norm": 16.601388931274414,
      "learning_rate": 9.622991768379223e-05,
      "loss": 0.16375694274902344,
      "step": 13290
    },
    {
      "epoch": 3.775191598069827,
      "grad_norm": 14.693408966064453,
      "learning_rate": 9.622707919386887e-05,
      "loss": 0.18652769327163696,
      "step": 13300
    },
    {
      "epoch": 3.7780300879931876,
      "grad_norm": 25.490880966186523,
      "learning_rate": 9.62242407039455e-05,
      "loss": 0.23232064247131348,
      "step": 13310
    },
    {
      "epoch": 3.7808685779165483,
      "grad_norm": 18.175931930541992,
      "learning_rate": 9.622140221402215e-05,
      "loss": 0.17263089418411254,
      "step": 13320
    },
    {
      "epoch": 3.7837070678399094,
      "grad_norm": 12.457060813903809,
      "learning_rate": 9.621856372409879e-05,
      "loss": 0.17455865144729615,
      "step": 13330
    },
    {
      "epoch": 3.78654555776327,
      "grad_norm": 18.688642501831055,
      "learning_rate": 9.621572523417542e-05,
      "loss": 0.20567383766174316,
      "step": 13340
    },
    {
      "epoch": 3.7893840476866307,
      "grad_norm": 16.189008712768555,
      "learning_rate": 9.621288674425206e-05,
      "loss": 0.20075876712799073,
      "step": 13350
    },
    {
      "epoch": 3.7922225376099914,
      "grad_norm": 13.119763374328613,
      "learning_rate": 9.62100482543287e-05,
      "loss": 0.2060337781906128,
      "step": 13360
    },
    {
      "epoch": 3.795061027533352,
      "grad_norm": 16.44010353088379,
      "learning_rate": 9.620720976440533e-05,
      "loss": 0.21363377571105957,
      "step": 13370
    },
    {
      "epoch": 3.797899517456713,
      "grad_norm": 10.756022453308105,
      "learning_rate": 9.620437127448197e-05,
      "loss": 0.17126485109329223,
      "step": 13380
    },
    {
      "epoch": 3.800738007380074,
      "grad_norm": 21.037017822265625,
      "learning_rate": 9.620153278455863e-05,
      "loss": 0.20660436153411865,
      "step": 13390
    },
    {
      "epoch": 3.8035764973034345,
      "grad_norm": 11.497673988342285,
      "learning_rate": 9.619869429463526e-05,
      "loss": 0.1891492486000061,
      "step": 13400
    },
    {
      "epoch": 3.8064149872267956,
      "grad_norm": 14.413599014282227,
      "learning_rate": 9.61958558047119e-05,
      "loss": 0.1766159415245056,
      "step": 13410
    },
    {
      "epoch": 3.8092534771501563,
      "grad_norm": 17.342639923095703,
      "learning_rate": 9.619301731478854e-05,
      "loss": 0.16026729345321655,
      "step": 13420
    },
    {
      "epoch": 3.812091967073517,
      "grad_norm": 14.935521125793457,
      "learning_rate": 9.619017882486518e-05,
      "loss": 0.1667719602584839,
      "step": 13430
    },
    {
      "epoch": 3.8149304569968776,
      "grad_norm": 9.354028701782227,
      "learning_rate": 9.618734033494181e-05,
      "loss": 0.23012104034423828,
      "step": 13440
    },
    {
      "epoch": 3.8177689469202383,
      "grad_norm": 22.844907760620117,
      "learning_rate": 9.618450184501846e-05,
      "loss": 0.23156285285949707,
      "step": 13450
    },
    {
      "epoch": 3.8206074368435994,
      "grad_norm": 10.652470588684082,
      "learning_rate": 9.61816633550951e-05,
      "loss": 0.19911113977432252,
      "step": 13460
    },
    {
      "epoch": 3.82344592676696,
      "grad_norm": 11.36054801940918,
      "learning_rate": 9.617882486517173e-05,
      "loss": 0.21364669799804686,
      "step": 13470
    },
    {
      "epoch": 3.8262844166903207,
      "grad_norm": 12.714508056640625,
      "learning_rate": 9.617627022424071e-05,
      "loss": 0.2182408094406128,
      "step": 13480
    },
    {
      "epoch": 3.8291229066136814,
      "grad_norm": 16.124755859375,
      "learning_rate": 9.617343173431734e-05,
      "loss": 0.1787625789642334,
      "step": 13490
    },
    {
      "epoch": 3.831961396537042,
      "grad_norm": 17.289695739746094,
      "learning_rate": 9.6170593244394e-05,
      "loss": 0.21091160774230958,
      "step": 13500
    },
    {
      "epoch": 3.831961396537042,
      "eval_accuracy": 0.8963565842182234,
      "eval_loss": 0.3161424696445465,
      "eval_runtime": 31.0853,
      "eval_samples_per_second": 505.93,
      "eval_steps_per_second": 7.914,
      "step": 13500
    },
    {
      "epoch": 3.834799886460403,
      "grad_norm": 14.816575050354004,
      "learning_rate": 9.616775475447064e-05,
      "loss": 0.18416981697082518,
      "step": 13510
    },
    {
      "epoch": 3.837638376383764,
      "grad_norm": 20.717742919921875,
      "learning_rate": 9.616491626454726e-05,
      "loss": 0.19217147827148437,
      "step": 13520
    },
    {
      "epoch": 3.8404768663071245,
      "grad_norm": 12.364614486694336,
      "learning_rate": 9.61620777746239e-05,
      "loss": 0.1885585069656372,
      "step": 13530
    },
    {
      "epoch": 3.8433153562304856,
      "grad_norm": 11.606247901916504,
      "learning_rate": 9.615923928470055e-05,
      "loss": 0.18612173795700074,
      "step": 13540
    },
    {
      "epoch": 3.8461538461538463,
      "grad_norm": 17.668453216552734,
      "learning_rate": 9.615640079477717e-05,
      "loss": 0.16202316284179688,
      "step": 13550
    },
    {
      "epoch": 3.848992336077207,
      "grad_norm": 17.570472717285156,
      "learning_rate": 9.615356230485382e-05,
      "loss": 0.22200171947479247,
      "step": 13560
    },
    {
      "epoch": 3.8518308260005676,
      "grad_norm": 17.335346221923828,
      "learning_rate": 9.615072381493047e-05,
      "loss": 0.15838404893875122,
      "step": 13570
    },
    {
      "epoch": 3.8546693159239283,
      "grad_norm": 10.258763313293457,
      "learning_rate": 9.61478853250071e-05,
      "loss": 0.14664362668991088,
      "step": 13580
    },
    {
      "epoch": 3.8575078058472894,
      "grad_norm": 12.704676628112793,
      "learning_rate": 9.614504683508374e-05,
      "loss": 0.1908525347709656,
      "step": 13590
    },
    {
      "epoch": 3.86034629577065,
      "grad_norm": 14.854520797729492,
      "learning_rate": 9.614220834516038e-05,
      "loss": 0.18227908611297608,
      "step": 13600
    },
    {
      "epoch": 3.8631847856940107,
      "grad_norm": 13.429558753967285,
      "learning_rate": 9.613936985523702e-05,
      "loss": 0.2027650833129883,
      "step": 13610
    },
    {
      "epoch": 3.866023275617372,
      "grad_norm": 12.316184043884277,
      "learning_rate": 9.613653136531365e-05,
      "loss": 0.17951679229736328,
      "step": 13620
    },
    {
      "epoch": 3.8688617655407325,
      "grad_norm": 11.702552795410156,
      "learning_rate": 9.61336928753903e-05,
      "loss": 0.17083731889724732,
      "step": 13630
    },
    {
      "epoch": 3.871700255464093,
      "grad_norm": 10.787964820861816,
      "learning_rate": 9.613085438546695e-05,
      "loss": 0.19240431785583495,
      "step": 13640
    },
    {
      "epoch": 3.874538745387454,
      "grad_norm": 22.3664608001709,
      "learning_rate": 9.612801589554357e-05,
      "loss": 0.2139369487762451,
      "step": 13650
    },
    {
      "epoch": 3.8773772353108145,
      "grad_norm": 17.102693557739258,
      "learning_rate": 9.612517740562022e-05,
      "loss": 0.1737605571746826,
      "step": 13660
    },
    {
      "epoch": 3.8802157252341756,
      "grad_norm": 17.290803909301758,
      "learning_rate": 9.612233891569686e-05,
      "loss": 0.19548892974853516,
      "step": 13670
    },
    {
      "epoch": 3.8830542151575362,
      "grad_norm": 14.182625770568848,
      "learning_rate": 9.611950042577348e-05,
      "loss": 0.18373520374298097,
      "step": 13680
    },
    {
      "epoch": 3.885892705080897,
      "grad_norm": 13.956124305725098,
      "learning_rate": 9.611666193585013e-05,
      "loss": 0.23870363235473632,
      "step": 13690
    },
    {
      "epoch": 3.888731195004258,
      "grad_norm": 8.751736640930176,
      "learning_rate": 9.611382344592678e-05,
      "loss": 0.14544075727462769,
      "step": 13700
    },
    {
      "epoch": 3.8915696849276182,
      "grad_norm": 8.248520851135254,
      "learning_rate": 9.611098495600341e-05,
      "loss": 0.15393333435058593,
      "step": 13710
    },
    {
      "epoch": 3.8944081748509793,
      "grad_norm": 18.234996795654297,
      "learning_rate": 9.610814646608005e-05,
      "loss": 0.18511407375335692,
      "step": 13720
    },
    {
      "epoch": 3.89724666477434,
      "grad_norm": 12.562544822692871,
      "learning_rate": 9.610530797615669e-05,
      "loss": 0.15003457069396972,
      "step": 13730
    },
    {
      "epoch": 3.9000851546977007,
      "grad_norm": 16.743684768676758,
      "learning_rate": 9.610246948623333e-05,
      "loss": 0.21348919868469238,
      "step": 13740
    },
    {
      "epoch": 3.902923644621062,
      "grad_norm": 12.473259925842285,
      "learning_rate": 9.609963099630996e-05,
      "loss": 0.15301227569580078,
      "step": 13750
    },
    {
      "epoch": 3.9057621345444224,
      "grad_norm": 14.41502571105957,
      "learning_rate": 9.60967925063866e-05,
      "loss": 0.1809757947921753,
      "step": 13760
    },
    {
      "epoch": 3.908600624467783,
      "grad_norm": 9.003304481506348,
      "learning_rate": 9.609395401646324e-05,
      "loss": 0.14672800302505493,
      "step": 13770
    },
    {
      "epoch": 3.911439114391144,
      "grad_norm": 9.885737419128418,
      "learning_rate": 9.609111552653988e-05,
      "loss": 0.12219957113265992,
      "step": 13780
    },
    {
      "epoch": 3.9142776043145044,
      "grad_norm": 14.666175842285156,
      "learning_rate": 9.608827703661653e-05,
      "loss": 0.2119225263595581,
      "step": 13790
    },
    {
      "epoch": 3.9171160942378656,
      "grad_norm": 17.990489959716797,
      "learning_rate": 9.608543854669317e-05,
      "loss": 0.23682405948638915,
      "step": 13800
    },
    {
      "epoch": 3.919954584161226,
      "grad_norm": 15.3749361038208,
      "learning_rate": 9.60826000567698e-05,
      "loss": 0.1940447211265564,
      "step": 13810
    },
    {
      "epoch": 3.922793074084587,
      "grad_norm": 13.172243118286133,
      "learning_rate": 9.607976156684644e-05,
      "loss": 0.14192864894866944,
      "step": 13820
    },
    {
      "epoch": 3.925631564007948,
      "grad_norm": 13.275805473327637,
      "learning_rate": 9.607692307692309e-05,
      "loss": 0.19739904403686523,
      "step": 13830
    },
    {
      "epoch": 3.9284700539313087,
      "grad_norm": 14.367815017700195,
      "learning_rate": 9.607408458699972e-05,
      "loss": 0.19411333799362182,
      "step": 13840
    },
    {
      "epoch": 3.9313085438546693,
      "grad_norm": 13.982487678527832,
      "learning_rate": 9.607124609707636e-05,
      "loss": 0.17054067850112914,
      "step": 13850
    },
    {
      "epoch": 3.93414703377803,
      "grad_norm": 11.844534873962402,
      "learning_rate": 9.6068407607153e-05,
      "loss": 0.24255647659301757,
      "step": 13860
    },
    {
      "epoch": 3.9369855237013907,
      "grad_norm": 13.950010299682617,
      "learning_rate": 9.606556911722963e-05,
      "loss": 0.1747615694999695,
      "step": 13870
    },
    {
      "epoch": 3.9398240136247518,
      "grad_norm": 14.970807075500488,
      "learning_rate": 9.606273062730627e-05,
      "loss": 0.16250706911087037,
      "step": 13880
    },
    {
      "epoch": 3.9426625035481124,
      "grad_norm": 17.797439575195312,
      "learning_rate": 9.605989213738291e-05,
      "loss": 0.18647220134735107,
      "step": 13890
    },
    {
      "epoch": 3.945500993471473,
      "grad_norm": 8.312163352966309,
      "learning_rate": 9.605705364745955e-05,
      "loss": 0.1735267996788025,
      "step": 13900
    },
    {
      "epoch": 3.948339483394834,
      "grad_norm": 9.662410736083984,
      "learning_rate": 9.60542151575362e-05,
      "loss": 0.17853622436523436,
      "step": 13910
    },
    {
      "epoch": 3.951177973318195,
      "grad_norm": 10.913804054260254,
      "learning_rate": 9.605137666761284e-05,
      "loss": 0.17758163213729858,
      "step": 13920
    },
    {
      "epoch": 3.9540164632415555,
      "grad_norm": 13.551321983337402,
      "learning_rate": 9.604853817768948e-05,
      "loss": 0.19117436408996583,
      "step": 13930
    },
    {
      "epoch": 3.956854953164916,
      "grad_norm": 8.479565620422363,
      "learning_rate": 9.60456996877661e-05,
      "loss": 0.18471699953079224,
      "step": 13940
    },
    {
      "epoch": 3.959693443088277,
      "grad_norm": 18.360044479370117,
      "learning_rate": 9.604286119784275e-05,
      "loss": 0.2238318681716919,
      "step": 13950
    },
    {
      "epoch": 3.962531933011638,
      "grad_norm": 10.409412384033203,
      "learning_rate": 9.604002270791939e-05,
      "loss": 0.19304097890853883,
      "step": 13960
    },
    {
      "epoch": 3.9653704229349986,
      "grad_norm": 8.591785430908203,
      "learning_rate": 9.603718421799603e-05,
      "loss": 0.17751789093017578,
      "step": 13970
    },
    {
      "epoch": 3.9682089128583593,
      "grad_norm": 13.945221900939941,
      "learning_rate": 9.603434572807267e-05,
      "loss": 0.17351099252700805,
      "step": 13980
    },
    {
      "epoch": 3.97104740278172,
      "grad_norm": 7.1689229011535645,
      "learning_rate": 9.603150723814931e-05,
      "loss": 0.17921603918075563,
      "step": 13990
    },
    {
      "epoch": 3.9738858927050806,
      "grad_norm": 17.593414306640625,
      "learning_rate": 9.602866874822594e-05,
      "loss": 0.24585111141204835,
      "step": 14000
    },
    {
      "epoch": 3.9738858927050806,
      "eval_accuracy": 0.902651491066319,
      "eval_loss": 0.2931208908557892,
      "eval_runtime": 31.3853,
      "eval_samples_per_second": 501.095,
      "eval_steps_per_second": 7.838,
      "step": 14000
    },
    {
      "epoch": 3.9767243826284417,
      "grad_norm": 14.007991790771484,
      "learning_rate": 9.602583025830258e-05,
      "loss": 0.20412559509277345,
      "step": 14010
    },
    {
      "epoch": 3.9795628725518024,
      "grad_norm": 9.47784423828125,
      "learning_rate": 9.602299176837922e-05,
      "loss": 0.1461554765701294,
      "step": 14020
    },
    {
      "epoch": 3.982401362475163,
      "grad_norm": 19.436687469482422,
      "learning_rate": 9.602015327845586e-05,
      "loss": 0.19939231872558594,
      "step": 14030
    },
    {
      "epoch": 3.985239852398524,
      "grad_norm": 16.153579711914062,
      "learning_rate": 9.60173147885325e-05,
      "loss": 0.1849508047103882,
      "step": 14040
    },
    {
      "epoch": 3.988078342321885,
      "grad_norm": 10.739173889160156,
      "learning_rate": 9.601447629860915e-05,
      "loss": 0.17080507278442383,
      "step": 14050
    },
    {
      "epoch": 3.9909168322452455,
      "grad_norm": 18.857379913330078,
      "learning_rate": 9.601163780868579e-05,
      "loss": 0.17809062004089354,
      "step": 14060
    },
    {
      "epoch": 3.993755322168606,
      "grad_norm": 14.858738899230957,
      "learning_rate": 9.600879931876242e-05,
      "loss": 0.2061694622039795,
      "step": 14070
    },
    {
      "epoch": 3.996593812091967,
      "grad_norm": 9.384162902832031,
      "learning_rate": 9.600596082883906e-05,
      "loss": 0.14361640214920043,
      "step": 14080
    },
    {
      "epoch": 3.999432302015328,
      "grad_norm": 15.557256698608398,
      "learning_rate": 9.60031223389157e-05,
      "loss": 0.13269511461257935,
      "step": 14090
    },
    {
      "epoch": 4.002270791938688,
      "grad_norm": 15.339125633239746,
      "learning_rate": 9.600028384899234e-05,
      "loss": 0.15242763757705688,
      "step": 14100
    },
    {
      "epoch": 4.005109281862049,
      "grad_norm": 11.810029029846191,
      "learning_rate": 9.599744535906898e-05,
      "loss": 0.14725091457366943,
      "step": 14110
    },
    {
      "epoch": 4.00794777178541,
      "grad_norm": 15.168164253234863,
      "learning_rate": 9.599460686914562e-05,
      "loss": 0.13273046016693116,
      "step": 14120
    },
    {
      "epoch": 4.010786261708771,
      "grad_norm": 15.249293327331543,
      "learning_rate": 9.599176837922225e-05,
      "loss": 0.17031476497650147,
      "step": 14130
    },
    {
      "epoch": 4.013624751632132,
      "grad_norm": 10.77449893951416,
      "learning_rate": 9.598892988929889e-05,
      "loss": 0.15205686092376708,
      "step": 14140
    },
    {
      "epoch": 4.016463241555493,
      "grad_norm": 13.917808532714844,
      "learning_rate": 9.598609139937553e-05,
      "loss": 0.18895701169967652,
      "step": 14150
    },
    {
      "epoch": 4.019301731478853,
      "grad_norm": 17.4494686126709,
      "learning_rate": 9.598325290945218e-05,
      "loss": 0.12264747619628906,
      "step": 14160
    },
    {
      "epoch": 4.022140221402214,
      "grad_norm": 13.52995491027832,
      "learning_rate": 9.598041441952882e-05,
      "loss": 0.1630455732345581,
      "step": 14170
    },
    {
      "epoch": 4.024978711325574,
      "grad_norm": 8.253148078918457,
      "learning_rate": 9.597757592960546e-05,
      "loss": 0.1761770248413086,
      "step": 14180
    },
    {
      "epoch": 4.0278172012489355,
      "grad_norm": 16.79362678527832,
      "learning_rate": 9.59747374396821e-05,
      "loss": 0.16914713382720947,
      "step": 14190
    },
    {
      "epoch": 4.030655691172297,
      "grad_norm": 10.3887357711792,
      "learning_rate": 9.597189894975873e-05,
      "loss": 0.15404207706451417,
      "step": 14200
    },
    {
      "epoch": 4.033494181095657,
      "grad_norm": 17.84835433959961,
      "learning_rate": 9.596906045983537e-05,
      "loss": 0.15989917516708374,
      "step": 14210
    },
    {
      "epoch": 4.036332671019018,
      "grad_norm": 10.473235130310059,
      "learning_rate": 9.596622196991201e-05,
      "loss": 0.1424857497215271,
      "step": 14220
    },
    {
      "epoch": 4.039171160942379,
      "grad_norm": 8.463979721069336,
      "learning_rate": 9.596338347998865e-05,
      "loss": 0.12286888360977173,
      "step": 14230
    },
    {
      "epoch": 4.042009650865739,
      "grad_norm": 11.434859275817871,
      "learning_rate": 9.596054499006529e-05,
      "loss": 0.13087767362594604,
      "step": 14240
    },
    {
      "epoch": 4.0448481407891,
      "grad_norm": 14.250101089477539,
      "learning_rate": 9.595770650014193e-05,
      "loss": 0.1534764885902405,
      "step": 14250
    },
    {
      "epoch": 4.047686630712461,
      "grad_norm": 14.397462844848633,
      "learning_rate": 9.595486801021856e-05,
      "loss": 0.134521222114563,
      "step": 14260
    },
    {
      "epoch": 4.050525120635822,
      "grad_norm": 19.012601852416992,
      "learning_rate": 9.59520295202952e-05,
      "loss": 0.14702656269073486,
      "step": 14270
    },
    {
      "epoch": 4.053363610559183,
      "grad_norm": 21.502548217773438,
      "learning_rate": 9.594919103037185e-05,
      "loss": 0.16444971561431884,
      "step": 14280
    },
    {
      "epoch": 4.056202100482543,
      "grad_norm": 9.994865417480469,
      "learning_rate": 9.594635254044849e-05,
      "loss": 0.17271435260772705,
      "step": 14290
    },
    {
      "epoch": 4.059040590405904,
      "grad_norm": 13.898537635803223,
      "learning_rate": 9.594351405052513e-05,
      "loss": 0.14536019563674926,
      "step": 14300
    },
    {
      "epoch": 4.061879080329265,
      "grad_norm": 18.628498077392578,
      "learning_rate": 9.594067556060177e-05,
      "loss": 0.15034260749816894,
      "step": 14310
    },
    {
      "epoch": 4.0647175702526255,
      "grad_norm": 12.507842063903809,
      "learning_rate": 9.593783707067841e-05,
      "loss": 0.15725717544555665,
      "step": 14320
    },
    {
      "epoch": 4.067556060175987,
      "grad_norm": 9.733858108520508,
      "learning_rate": 9.593499858075504e-05,
      "loss": 0.1314501404762268,
      "step": 14330
    },
    {
      "epoch": 4.070394550099347,
      "grad_norm": 11.284711837768555,
      "learning_rate": 9.593216009083168e-05,
      "loss": 0.160497784614563,
      "step": 14340
    },
    {
      "epoch": 4.073233040022708,
      "grad_norm": 10.518166542053223,
      "learning_rate": 9.592932160090832e-05,
      "loss": 0.11356463432312011,
      "step": 14350
    },
    {
      "epoch": 4.076071529946069,
      "grad_norm": 24.16094970703125,
      "learning_rate": 9.592648311098496e-05,
      "loss": 0.15804142951965333,
      "step": 14360
    },
    {
      "epoch": 4.078910019869429,
      "grad_norm": 12.228034019470215,
      "learning_rate": 9.592392847005394e-05,
      "loss": 0.15015305280685426,
      "step": 14370
    },
    {
      "epoch": 4.08174850979279,
      "grad_norm": 12.908282279968262,
      "learning_rate": 9.592108998013057e-05,
      "loss": 0.12004361152648926,
      "step": 14380
    },
    {
      "epoch": 4.084586999716151,
      "grad_norm": 25.25391387939453,
      "learning_rate": 9.591825149020721e-05,
      "loss": 0.1858546495437622,
      "step": 14390
    },
    {
      "epoch": 4.087425489639512,
      "grad_norm": 15.679829597473145,
      "learning_rate": 9.591541300028385e-05,
      "loss": 0.18722642660140992,
      "step": 14400
    },
    {
      "epoch": 4.090263979562873,
      "grad_norm": 8.558819770812988,
      "learning_rate": 9.59125745103605e-05,
      "loss": 0.14437999725341796,
      "step": 14410
    },
    {
      "epoch": 4.093102469486233,
      "grad_norm": 19.275169372558594,
      "learning_rate": 9.590973602043714e-05,
      "loss": 0.14627193212509154,
      "step": 14420
    },
    {
      "epoch": 4.095940959409594,
      "grad_norm": 7.874567031860352,
      "learning_rate": 9.590689753051378e-05,
      "loss": 0.12814124822616577,
      "step": 14430
    },
    {
      "epoch": 4.098779449332955,
      "grad_norm": 11.124398231506348,
      "learning_rate": 9.59040590405904e-05,
      "loss": 0.16742610931396484,
      "step": 14440
    },
    {
      "epoch": 4.1016179392563155,
      "grad_norm": 17.87415313720703,
      "learning_rate": 9.590122055066705e-05,
      "loss": 0.15107352733612062,
      "step": 14450
    },
    {
      "epoch": 4.104456429179677,
      "grad_norm": 11.556680679321289,
      "learning_rate": 9.589838206074369e-05,
      "loss": 0.13166494369506837,
      "step": 14460
    },
    {
      "epoch": 4.107294919103037,
      "grad_norm": 17.838333129882812,
      "learning_rate": 9.589554357082033e-05,
      "loss": 0.1744963765144348,
      "step": 14470
    },
    {
      "epoch": 4.110133409026398,
      "grad_norm": 15.847147941589355,
      "learning_rate": 9.589270508089697e-05,
      "loss": 0.1760677695274353,
      "step": 14480
    },
    {
      "epoch": 4.112971898949759,
      "grad_norm": 16.368736267089844,
      "learning_rate": 9.588986659097361e-05,
      "loss": 0.16585955619812012,
      "step": 14490
    },
    {
      "epoch": 4.115810388873119,
      "grad_norm": 18.406042098999023,
      "learning_rate": 9.588702810105025e-05,
      "loss": 0.16642417907714843,
      "step": 14500
    },
    {
      "epoch": 4.115810388873119,
      "eval_accuracy": 0.8910790360526483,
      "eval_loss": 0.3261353671550751,
      "eval_runtime": 31.5065,
      "eval_samples_per_second": 499.166,
      "eval_steps_per_second": 7.808,
      "step": 14500
    },
    {
      "epoch": 4.11864887879648,
      "grad_norm": 15.058733940124512,
      "learning_rate": 9.588418961112688e-05,
      "loss": 0.17857513427734376,
      "step": 14510
    },
    {
      "epoch": 4.1214873687198414,
      "grad_norm": 10.961214065551758,
      "learning_rate": 9.588135112120352e-05,
      "loss": 0.14090417623519896,
      "step": 14520
    },
    {
      "epoch": 4.124325858643202,
      "grad_norm": 13.186880111694336,
      "learning_rate": 9.587851263128016e-05,
      "loss": 0.15759789943695068,
      "step": 14530
    },
    {
      "epoch": 4.127164348566563,
      "grad_norm": 14.095049858093262,
      "learning_rate": 9.58756741413568e-05,
      "loss": 0.11720154285430909,
      "step": 14540
    },
    {
      "epoch": 4.130002838489923,
      "grad_norm": 10.82900333404541,
      "learning_rate": 9.587283565143345e-05,
      "loss": 0.13271580934524535,
      "step": 14550
    },
    {
      "epoch": 4.132841328413284,
      "grad_norm": 12.094626426696777,
      "learning_rate": 9.586999716151009e-05,
      "loss": 0.17368659973144532,
      "step": 14560
    },
    {
      "epoch": 4.135679818336645,
      "grad_norm": 15.703944206237793,
      "learning_rate": 9.586715867158672e-05,
      "loss": 0.15307741165161132,
      "step": 14570
    },
    {
      "epoch": 4.138518308260005,
      "grad_norm": 9.091362953186035,
      "learning_rate": 9.586432018166336e-05,
      "loss": 0.12892868518829345,
      "step": 14580
    },
    {
      "epoch": 4.1413567981833665,
      "grad_norm": 13.104792594909668,
      "learning_rate": 9.586148169174e-05,
      "loss": 0.15084669589996338,
      "step": 14590
    },
    {
      "epoch": 4.144195288106728,
      "grad_norm": 19.178945541381836,
      "learning_rate": 9.585864320181664e-05,
      "loss": 0.15708644390106202,
      "step": 14600
    },
    {
      "epoch": 4.147033778030088,
      "grad_norm": 8.34633731842041,
      "learning_rate": 9.585580471189328e-05,
      "loss": 0.16216639280319214,
      "step": 14610
    },
    {
      "epoch": 4.149872267953449,
      "grad_norm": 17.990982055664062,
      "learning_rate": 9.585296622196992e-05,
      "loss": 0.17228389978408815,
      "step": 14620
    },
    {
      "epoch": 4.152710757876809,
      "grad_norm": 15.340961456298828,
      "learning_rate": 9.585012773204656e-05,
      "loss": 0.15551118850708007,
      "step": 14630
    },
    {
      "epoch": 4.15554924780017,
      "grad_norm": 18.991130828857422,
      "learning_rate": 9.584728924212319e-05,
      "loss": 0.12890100479125977,
      "step": 14640
    },
    {
      "epoch": 4.158387737723531,
      "grad_norm": 21.018190383911133,
      "learning_rate": 9.584445075219983e-05,
      "loss": 0.18114593029022216,
      "step": 14650
    },
    {
      "epoch": 4.161226227646892,
      "grad_norm": 6.306010723114014,
      "learning_rate": 9.584161226227647e-05,
      "loss": 0.11897997856140137,
      "step": 14660
    },
    {
      "epoch": 4.164064717570253,
      "grad_norm": 11.076420783996582,
      "learning_rate": 9.58387737723531e-05,
      "loss": 0.1446991205215454,
      "step": 14670
    },
    {
      "epoch": 4.166903207493613,
      "grad_norm": 12.878436088562012,
      "learning_rate": 9.583593528242976e-05,
      "loss": 0.15413979291915894,
      "step": 14680
    },
    {
      "epoch": 4.169741697416974,
      "grad_norm": 15.184830665588379,
      "learning_rate": 9.58330967925064e-05,
      "loss": 0.1326911687850952,
      "step": 14690
    },
    {
      "epoch": 4.172580187340335,
      "grad_norm": 17.91576385498047,
      "learning_rate": 9.583025830258303e-05,
      "loss": 0.17909533977508546,
      "step": 14700
    },
    {
      "epoch": 4.175418677263695,
      "grad_norm": 14.837722778320312,
      "learning_rate": 9.582741981265967e-05,
      "loss": 0.1566817045211792,
      "step": 14710
    },
    {
      "epoch": 4.1782571671870565,
      "grad_norm": 12.023417472839355,
      "learning_rate": 9.582458132273631e-05,
      "loss": 0.1333454966545105,
      "step": 14720
    },
    {
      "epoch": 4.181095657110418,
      "grad_norm": 7.271909713745117,
      "learning_rate": 9.582174283281295e-05,
      "loss": 0.14408345222473146,
      "step": 14730
    },
    {
      "epoch": 4.183934147033778,
      "grad_norm": 11.925993919372559,
      "learning_rate": 9.581890434288959e-05,
      "loss": 0.1441882371902466,
      "step": 14740
    },
    {
      "epoch": 4.186772636957139,
      "grad_norm": 12.516487121582031,
      "learning_rate": 9.581606585296623e-05,
      "loss": 0.13351558446884154,
      "step": 14750
    },
    {
      "epoch": 4.189611126880499,
      "grad_norm": 13.141271591186523,
      "learning_rate": 9.581322736304286e-05,
      "loss": 0.14427276849746704,
      "step": 14760
    },
    {
      "epoch": 4.19244961680386,
      "grad_norm": 12.104243278503418,
      "learning_rate": 9.58103888731195e-05,
      "loss": 0.17315857410430907,
      "step": 14770
    },
    {
      "epoch": 4.195288106727221,
      "grad_norm": 13.825701713562012,
      "learning_rate": 9.580755038319614e-05,
      "loss": 0.19491254091262816,
      "step": 14780
    },
    {
      "epoch": 4.198126596650582,
      "grad_norm": 8.500093460083008,
      "learning_rate": 9.580471189327279e-05,
      "loss": 0.1493491530418396,
      "step": 14790
    },
    {
      "epoch": 4.200965086573943,
      "grad_norm": 15.029679298400879,
      "learning_rate": 9.580187340334941e-05,
      "loss": 0.1404767155647278,
      "step": 14800
    },
    {
      "epoch": 4.203803576497304,
      "grad_norm": 14.780632972717285,
      "learning_rate": 9.579903491342607e-05,
      "loss": 0.15077284574508668,
      "step": 14810
    },
    {
      "epoch": 4.206642066420664,
      "grad_norm": 16.46811866760254,
      "learning_rate": 9.579619642350271e-05,
      "loss": 0.17287200689315796,
      "step": 14820
    },
    {
      "epoch": 4.209480556344025,
      "grad_norm": 15.16257095336914,
      "learning_rate": 9.579335793357934e-05,
      "loss": 0.16240355968475342,
      "step": 14830
    },
    {
      "epoch": 4.212319046267385,
      "grad_norm": 16.409542083740234,
      "learning_rate": 9.579051944365598e-05,
      "loss": 0.1358044147491455,
      "step": 14840
    },
    {
      "epoch": 4.2151575361907465,
      "grad_norm": 9.065581321716309,
      "learning_rate": 9.578768095373262e-05,
      "loss": 0.11479216814041138,
      "step": 14850
    },
    {
      "epoch": 4.217996026114108,
      "grad_norm": 16.2310791015625,
      "learning_rate": 9.578484246380925e-05,
      "loss": 0.16164865493774414,
      "step": 14860
    },
    {
      "epoch": 4.220834516037468,
      "grad_norm": 10.68352222442627,
      "learning_rate": 9.578200397388589e-05,
      "loss": 0.12752039432525636,
      "step": 14870
    },
    {
      "epoch": 4.223673005960829,
      "grad_norm": 13.836865425109863,
      "learning_rate": 9.577916548396254e-05,
      "loss": 0.20279693603515625,
      "step": 14880
    },
    {
      "epoch": 4.226511495884189,
      "grad_norm": 10.64200210571289,
      "learning_rate": 9.577632699403917e-05,
      "loss": 0.1932573914527893,
      "step": 14890
    },
    {
      "epoch": 4.22934998580755,
      "grad_norm": 11.400484085083008,
      "learning_rate": 9.577348850411581e-05,
      "loss": 0.16212347745895386,
      "step": 14900
    },
    {
      "epoch": 4.232188475730911,
      "grad_norm": 15.307188987731934,
      "learning_rate": 9.577065001419245e-05,
      "loss": 0.20311076641082765,
      "step": 14910
    },
    {
      "epoch": 4.235026965654272,
      "grad_norm": 15.208539962768555,
      "learning_rate": 9.57678115242691e-05,
      "loss": 0.15278089046478271,
      "step": 14920
    },
    {
      "epoch": 4.237865455577633,
      "grad_norm": 12.419407844543457,
      "learning_rate": 9.576497303434572e-05,
      "loss": 0.1347472071647644,
      "step": 14930
    },
    {
      "epoch": 4.240703945500994,
      "grad_norm": 21.679683685302734,
      "learning_rate": 9.576213454442238e-05,
      "loss": 0.16810553073883056,
      "step": 14940
    },
    {
      "epoch": 4.243542435424354,
      "grad_norm": 14.06842041015625,
      "learning_rate": 9.575929605449902e-05,
      "loss": 0.15204988718032836,
      "step": 14950
    },
    {
      "epoch": 4.246380925347715,
      "grad_norm": 8.801273345947266,
      "learning_rate": 9.575645756457565e-05,
      "loss": 0.15220837593078612,
      "step": 14960
    },
    {
      "epoch": 4.249219415271075,
      "grad_norm": 19.13870620727539,
      "learning_rate": 9.575361907465229e-05,
      "loss": 0.19406039714813234,
      "step": 14970
    },
    {
      "epoch": 4.2520579051944365,
      "grad_norm": 17.72483253479004,
      "learning_rate": 9.575078058472893e-05,
      "loss": 0.15963964462280272,
      "step": 14980
    },
    {
      "epoch": 4.254896395117798,
      "grad_norm": 19.06657600402832,
      "learning_rate": 9.574794209480556e-05,
      "loss": 0.15715559720993041,
      "step": 14990
    },
    {
      "epoch": 4.257734885041158,
      "grad_norm": 10.211169242858887,
      "learning_rate": 9.57451036048822e-05,
      "loss": 0.1485607624053955,
      "step": 15000
    },
    {
      "epoch": 4.257734885041158,
      "eval_accuracy": 0.8908246963820182,
      "eval_loss": 0.3205018937587738,
      "eval_runtime": 31.4678,
      "eval_samples_per_second": 499.781,
      "eval_steps_per_second": 7.818,
      "step": 15000
    },
    {
      "epoch": 4.260573374964519,
      "grad_norm": 18.0892276763916,
      "learning_rate": 9.574226511495885e-05,
      "loss": 0.1473289966583252,
      "step": 15010
    },
    {
      "epoch": 4.26341186488788,
      "grad_norm": 16.001670837402344,
      "learning_rate": 9.573942662503548e-05,
      "loss": 0.1674429416656494,
      "step": 15020
    },
    {
      "epoch": 4.26625035481124,
      "grad_norm": 18.877471923828125,
      "learning_rate": 9.573658813511212e-05,
      "loss": 0.16755801439285278,
      "step": 15030
    },
    {
      "epoch": 4.269088844734601,
      "grad_norm": 7.050081253051758,
      "learning_rate": 9.573374964518877e-05,
      "loss": 0.15270453691482544,
      "step": 15040
    },
    {
      "epoch": 4.271927334657962,
      "grad_norm": 14.125801086425781,
      "learning_rate": 9.573091115526541e-05,
      "loss": 0.15930687189102172,
      "step": 15050
    },
    {
      "epoch": 4.274765824581323,
      "grad_norm": 14.005635261535645,
      "learning_rate": 9.572807266534203e-05,
      "loss": 0.1637762665748596,
      "step": 15060
    },
    {
      "epoch": 4.277604314504684,
      "grad_norm": 13.851651191711426,
      "learning_rate": 9.572523417541868e-05,
      "loss": 0.1451090931892395,
      "step": 15070
    },
    {
      "epoch": 4.280442804428044,
      "grad_norm": 11.01707649230957,
      "learning_rate": 9.572239568549533e-05,
      "loss": 0.13447017669677735,
      "step": 15080
    },
    {
      "epoch": 4.283281294351405,
      "grad_norm": 16.21165657043457,
      "learning_rate": 9.571955719557196e-05,
      "loss": 0.1568738341331482,
      "step": 15090
    },
    {
      "epoch": 4.286119784274765,
      "grad_norm": 13.440966606140137,
      "learning_rate": 9.57167187056486e-05,
      "loss": 0.15765141248703002,
      "step": 15100
    },
    {
      "epoch": 4.2889582741981265,
      "grad_norm": 8.969966888427734,
      "learning_rate": 9.571388021572524e-05,
      "loss": 0.11767923831939697,
      "step": 15110
    },
    {
      "epoch": 4.291796764121488,
      "grad_norm": 18.897871017456055,
      "learning_rate": 9.571104172580187e-05,
      "loss": 0.1935199975967407,
      "step": 15120
    },
    {
      "epoch": 4.294635254044848,
      "grad_norm": 7.486785888671875,
      "learning_rate": 9.570820323587851e-05,
      "loss": 0.15177110433578492,
      "step": 15130
    },
    {
      "epoch": 4.297473743968209,
      "grad_norm": 14.913260459899902,
      "learning_rate": 9.570536474595517e-05,
      "loss": 0.14765549898147584,
      "step": 15140
    },
    {
      "epoch": 4.30031223389157,
      "grad_norm": 16.408605575561523,
      "learning_rate": 9.57025262560318e-05,
      "loss": 0.14995042085647584,
      "step": 15150
    },
    {
      "epoch": 4.30315072381493,
      "grad_norm": 19.2226505279541,
      "learning_rate": 9.569968776610843e-05,
      "loss": 0.13153890371322632,
      "step": 15160
    },
    {
      "epoch": 4.305989213738291,
      "grad_norm": 13.076022148132324,
      "learning_rate": 9.569684927618508e-05,
      "loss": 0.1534234404563904,
      "step": 15170
    },
    {
      "epoch": 4.3088277036616525,
      "grad_norm": 12.501521110534668,
      "learning_rate": 9.569401078626172e-05,
      "loss": 0.11859639883041381,
      "step": 15180
    },
    {
      "epoch": 4.311666193585013,
      "grad_norm": 25.320295333862305,
      "learning_rate": 9.569117229633835e-05,
      "loss": 0.14102661609649658,
      "step": 15190
    },
    {
      "epoch": 4.314504683508374,
      "grad_norm": 11.505056381225586,
      "learning_rate": 9.568833380641499e-05,
      "loss": 0.1731319546699524,
      "step": 15200
    },
    {
      "epoch": 4.317343173431734,
      "grad_norm": 9.780108451843262,
      "learning_rate": 9.568549531649164e-05,
      "loss": 0.17432491779327391,
      "step": 15210
    },
    {
      "epoch": 4.320181663355095,
      "grad_norm": 5.729079723358154,
      "learning_rate": 9.568265682656827e-05,
      "loss": 0.12856547832489013,
      "step": 15220
    },
    {
      "epoch": 4.323020153278456,
      "grad_norm": 17.16070556640625,
      "learning_rate": 9.567981833664491e-05,
      "loss": 0.1171339750289917,
      "step": 15230
    },
    {
      "epoch": 4.3258586432018165,
      "grad_norm": 19.715206146240234,
      "learning_rate": 9.567697984672155e-05,
      "loss": 0.18454134464263916,
      "step": 15240
    },
    {
      "epoch": 4.328697133125178,
      "grad_norm": 14.777257919311523,
      "learning_rate": 9.567414135679818e-05,
      "loss": 0.1561437129974365,
      "step": 15250
    },
    {
      "epoch": 4.331535623048538,
      "grad_norm": 12.500276565551758,
      "learning_rate": 9.567130286687482e-05,
      "loss": 0.13659229278564453,
      "step": 15260
    },
    {
      "epoch": 4.334374112971899,
      "grad_norm": 17.76792335510254,
      "learning_rate": 9.566846437695146e-05,
      "loss": 0.17823370695114135,
      "step": 15270
    },
    {
      "epoch": 4.33721260289526,
      "grad_norm": 10.953511238098145,
      "learning_rate": 9.56656258870281e-05,
      "loss": 0.1756179451942444,
      "step": 15280
    },
    {
      "epoch": 4.34005109281862,
      "grad_norm": 9.660861015319824,
      "learning_rate": 9.566278739710475e-05,
      "loss": 0.15249955654144287,
      "step": 15290
    },
    {
      "epoch": 4.342889582741981,
      "grad_norm": 14.300312042236328,
      "learning_rate": 9.565994890718139e-05,
      "loss": 0.16629996299743652,
      "step": 15300
    },
    {
      "epoch": 4.3457280726653424,
      "grad_norm": 14.198461532592773,
      "learning_rate": 9.565711041725803e-05,
      "loss": 0.15525127649307252,
      "step": 15310
    },
    {
      "epoch": 4.348566562588703,
      "grad_norm": 17.374862670898438,
      "learning_rate": 9.565427192733466e-05,
      "loss": 0.1522218942642212,
      "step": 15320
    },
    {
      "epoch": 4.351405052512064,
      "grad_norm": 11.194106101989746,
      "learning_rate": 9.56514334374113e-05,
      "loss": 0.14212888479232788,
      "step": 15330
    },
    {
      "epoch": 4.354243542435424,
      "grad_norm": 9.271810531616211,
      "learning_rate": 9.564859494748795e-05,
      "loss": 0.13489923477172852,
      "step": 15340
    },
    {
      "epoch": 4.357082032358785,
      "grad_norm": 14.644876480102539,
      "learning_rate": 9.564575645756458e-05,
      "loss": 0.11552230119705201,
      "step": 15350
    },
    {
      "epoch": 4.359920522282146,
      "grad_norm": 13.42244815826416,
      "learning_rate": 9.564291796764122e-05,
      "loss": 0.14517542123794555,
      "step": 15360
    },
    {
      "epoch": 4.362759012205506,
      "grad_norm": 17.174306869506836,
      "learning_rate": 9.564007947771786e-05,
      "loss": 0.14905043840408325,
      "step": 15370
    },
    {
      "epoch": 4.3655975021288675,
      "grad_norm": 14.001357078552246,
      "learning_rate": 9.563724098779449e-05,
      "loss": 0.1530600070953369,
      "step": 15380
    },
    {
      "epoch": 4.368435992052229,
      "grad_norm": 20.78911018371582,
      "learning_rate": 9.563440249787113e-05,
      "loss": 0.16638636589050293,
      "step": 15390
    },
    {
      "epoch": 4.371274481975589,
      "grad_norm": 14.357483863830566,
      "learning_rate": 9.563156400794777e-05,
      "loss": 0.14346433877944947,
      "step": 15400
    },
    {
      "epoch": 4.37411297189895,
      "grad_norm": 12.634615898132324,
      "learning_rate": 9.562872551802442e-05,
      "loss": 0.15297379493713378,
      "step": 15410
    },
    {
      "epoch": 4.37695146182231,
      "grad_norm": 12.296581268310547,
      "learning_rate": 9.562588702810106e-05,
      "loss": 0.13567272424697877,
      "step": 15420
    },
    {
      "epoch": 4.379789951745671,
      "grad_norm": 14.986875534057617,
      "learning_rate": 9.56230485381777e-05,
      "loss": 0.1665493965148926,
      "step": 15430
    },
    {
      "epoch": 4.382628441669032,
      "grad_norm": 6.609609127044678,
      "learning_rate": 9.562021004825434e-05,
      "loss": 0.12005674839019775,
      "step": 15440
    },
    {
      "epoch": 4.385466931592393,
      "grad_norm": 11.34575366973877,
      "learning_rate": 9.561737155833097e-05,
      "loss": 0.1485649585723877,
      "step": 15450
    },
    {
      "epoch": 4.388305421515754,
      "grad_norm": 14.887136459350586,
      "learning_rate": 9.561453306840761e-05,
      "loss": 0.1593350887298584,
      "step": 15460
    },
    {
      "epoch": 4.391143911439114,
      "grad_norm": 17.065351486206055,
      "learning_rate": 9.561169457848425e-05,
      "loss": 0.1566593885421753,
      "step": 15470
    },
    {
      "epoch": 4.393982401362475,
      "grad_norm": 12.026158332824707,
      "learning_rate": 9.560885608856089e-05,
      "loss": 0.13626867532730103,
      "step": 15480
    },
    {
      "epoch": 4.396820891285836,
      "grad_norm": 13.356855392456055,
      "learning_rate": 9.560601759863753e-05,
      "loss": 0.11659452915191651,
      "step": 15490
    },
    {
      "epoch": 4.399659381209196,
      "grad_norm": 19.59092140197754,
      "learning_rate": 9.560317910871417e-05,
      "loss": 0.15149736404418945,
      "step": 15500
    },
    {
      "epoch": 4.399659381209196,
      "eval_accuracy": 0.900489603865963,
      "eval_loss": 0.2935667335987091,
      "eval_runtime": 31.6322,
      "eval_samples_per_second": 497.184,
      "eval_steps_per_second": 7.777,
      "step": 15500
    },
    {
      "epoch": 4.4024978711325575,
      "grad_norm": 11.592808723449707,
      "learning_rate": 9.56003406187908e-05,
      "loss": 0.1285654067993164,
      "step": 15510
    },
    {
      "epoch": 4.405336361055919,
      "grad_norm": 9.41304874420166,
      "learning_rate": 9.559750212886744e-05,
      "loss": 0.13084110021591186,
      "step": 15520
    },
    {
      "epoch": 4.408174850979279,
      "grad_norm": 11.23768138885498,
      "learning_rate": 9.559466363894408e-05,
      "loss": 0.1178362488746643,
      "step": 15530
    },
    {
      "epoch": 4.41101334090264,
      "grad_norm": 10.919793128967285,
      "learning_rate": 9.559182514902073e-05,
      "loss": 0.16446350812911986,
      "step": 15540
    },
    {
      "epoch": 4.413851830826,
      "grad_norm": 17.316804885864258,
      "learning_rate": 9.558898665909737e-05,
      "loss": 0.1307805061340332,
      "step": 15550
    },
    {
      "epoch": 4.416690320749361,
      "grad_norm": 15.07625675201416,
      "learning_rate": 9.558614816917401e-05,
      "loss": 0.15423996448516847,
      "step": 15560
    },
    {
      "epoch": 4.419528810672722,
      "grad_norm": 18.693843841552734,
      "learning_rate": 9.558330967925065e-05,
      "loss": 0.18187485933303832,
      "step": 15570
    },
    {
      "epoch": 4.422367300596083,
      "grad_norm": 9.033263206481934,
      "learning_rate": 9.558047118932728e-05,
      "loss": 0.17740638256073,
      "step": 15580
    },
    {
      "epoch": 4.425205790519444,
      "grad_norm": 12.151775360107422,
      "learning_rate": 9.557763269940392e-05,
      "loss": 0.13484522104263305,
      "step": 15590
    },
    {
      "epoch": 4.428044280442805,
      "grad_norm": 17.459026336669922,
      "learning_rate": 9.557479420948056e-05,
      "loss": 0.14064306020736694,
      "step": 15600
    },
    {
      "epoch": 4.430882770366165,
      "grad_norm": 17.531124114990234,
      "learning_rate": 9.55719557195572e-05,
      "loss": 0.1496012330055237,
      "step": 15610
    },
    {
      "epoch": 4.433721260289526,
      "grad_norm": 18.614593505859375,
      "learning_rate": 9.556911722963384e-05,
      "loss": 0.1877504825592041,
      "step": 15620
    },
    {
      "epoch": 4.436559750212886,
      "grad_norm": 15.525225639343262,
      "learning_rate": 9.556627873971048e-05,
      "loss": 0.16716893911361694,
      "step": 15630
    },
    {
      "epoch": 4.4393982401362475,
      "grad_norm": 11.920730590820312,
      "learning_rate": 9.556344024978711e-05,
      "loss": 0.17789096832275392,
      "step": 15640
    },
    {
      "epoch": 4.442236730059609,
      "grad_norm": 14.344074249267578,
      "learning_rate": 9.556060175986375e-05,
      "loss": 0.1514929175376892,
      "step": 15650
    },
    {
      "epoch": 4.445075219982969,
      "grad_norm": 14.779162406921387,
      "learning_rate": 9.55577632699404e-05,
      "loss": 0.12468938827514649,
      "step": 15660
    },
    {
      "epoch": 4.44791370990633,
      "grad_norm": 10.21207332611084,
      "learning_rate": 9.555492478001704e-05,
      "loss": 0.13630890846252441,
      "step": 15670
    },
    {
      "epoch": 4.45075219982969,
      "grad_norm": 13.229500770568848,
      "learning_rate": 9.555208629009368e-05,
      "loss": 0.1446927785873413,
      "step": 15680
    },
    {
      "epoch": 4.453590689753051,
      "grad_norm": 8.497537612915039,
      "learning_rate": 9.554924780017032e-05,
      "loss": 0.1572495937347412,
      "step": 15690
    },
    {
      "epoch": 4.456429179676412,
      "grad_norm": 6.93145751953125,
      "learning_rate": 9.554640931024695e-05,
      "loss": 0.13637943267822267,
      "step": 15700
    },
    {
      "epoch": 4.459267669599773,
      "grad_norm": 8.381741523742676,
      "learning_rate": 9.554357082032359e-05,
      "loss": 0.15186114311218263,
      "step": 15710
    },
    {
      "epoch": 4.462106159523134,
      "grad_norm": 15.00367546081543,
      "learning_rate": 9.554073233040023e-05,
      "loss": 0.15030059814453126,
      "step": 15720
    },
    {
      "epoch": 4.464944649446495,
      "grad_norm": 17.346628189086914,
      "learning_rate": 9.553789384047687e-05,
      "loss": 0.1418553590774536,
      "step": 15730
    },
    {
      "epoch": 4.467783139369855,
      "grad_norm": 13.140084266662598,
      "learning_rate": 9.553505535055351e-05,
      "loss": 0.12391524314880371,
      "step": 15740
    },
    {
      "epoch": 4.470621629293216,
      "grad_norm": 9.749346733093262,
      "learning_rate": 9.553221686063015e-05,
      "loss": 0.12974709272384644,
      "step": 15750
    },
    {
      "epoch": 4.473460119216576,
      "grad_norm": 10.463181495666504,
      "learning_rate": 9.55293783707068e-05,
      "loss": 0.17821048498153685,
      "step": 15760
    },
    {
      "epoch": 4.4762986091399375,
      "grad_norm": 7.689849376678467,
      "learning_rate": 9.552653988078342e-05,
      "loss": 0.1456715941429138,
      "step": 15770
    },
    {
      "epoch": 4.479137099063299,
      "grad_norm": 16.416494369506836,
      "learning_rate": 9.552370139086006e-05,
      "loss": 0.15434129238128663,
      "step": 15780
    },
    {
      "epoch": 4.481975588986659,
      "grad_norm": 12.501565933227539,
      "learning_rate": 9.55208629009367e-05,
      "loss": 0.15272268056869506,
      "step": 15790
    },
    {
      "epoch": 4.48481407891002,
      "grad_norm": 16.444988250732422,
      "learning_rate": 9.551802441101333e-05,
      "loss": 0.144976806640625,
      "step": 15800
    },
    {
      "epoch": 4.487652568833381,
      "grad_norm": 12.045256614685059,
      "learning_rate": 9.551518592108999e-05,
      "loss": 0.13219764232635497,
      "step": 15810
    },
    {
      "epoch": 4.490491058756741,
      "grad_norm": 13.199865341186523,
      "learning_rate": 9.551234743116663e-05,
      "loss": 0.17169543504714965,
      "step": 15820
    },
    {
      "epoch": 4.493329548680102,
      "grad_norm": 18.176637649536133,
      "learning_rate": 9.550950894124326e-05,
      "loss": 0.16755325794219972,
      "step": 15830
    },
    {
      "epoch": 4.496168038603463,
      "grad_norm": 11.754714965820312,
      "learning_rate": 9.55066704513199e-05,
      "loss": 0.16987472772598267,
      "step": 15840
    },
    {
      "epoch": 4.499006528526824,
      "grad_norm": 17.328136444091797,
      "learning_rate": 9.550383196139654e-05,
      "loss": 0.1631556987762451,
      "step": 15850
    },
    {
      "epoch": 4.501845018450185,
      "grad_norm": 10.793828010559082,
      "learning_rate": 9.550099347147318e-05,
      "loss": 0.11724253892898559,
      "step": 15860
    },
    {
      "epoch": 4.504683508373545,
      "grad_norm": 18.185853958129883,
      "learning_rate": 9.549815498154982e-05,
      "loss": 0.16927456855773926,
      "step": 15870
    },
    {
      "epoch": 4.507521998296906,
      "grad_norm": 18.749469757080078,
      "learning_rate": 9.549531649162646e-05,
      "loss": 0.1657538652420044,
      "step": 15880
    },
    {
      "epoch": 4.510360488220266,
      "grad_norm": 14.60150146484375,
      "learning_rate": 9.54924780017031e-05,
      "loss": 0.13404953479766846,
      "step": 15890
    },
    {
      "epoch": 4.5131989781436275,
      "grad_norm": 11.472158432006836,
      "learning_rate": 9.548963951177973e-05,
      "loss": 0.15821380615234376,
      "step": 15900
    },
    {
      "epoch": 4.516037468066989,
      "grad_norm": 11.78376293182373,
      "learning_rate": 9.548680102185638e-05,
      "loss": 0.16594191789627075,
      "step": 15910
    },
    {
      "epoch": 4.518875957990349,
      "grad_norm": 15.846101760864258,
      "learning_rate": 9.548396253193302e-05,
      "loss": 0.12899974584579468,
      "step": 15920
    },
    {
      "epoch": 4.52171444791371,
      "grad_norm": 11.677336692810059,
      "learning_rate": 9.548112404200964e-05,
      "loss": 0.13508477210998535,
      "step": 15930
    },
    {
      "epoch": 4.524552937837071,
      "grad_norm": 11.68274974822998,
      "learning_rate": 9.54782855520863e-05,
      "loss": 0.10200138092041015,
      "step": 15940
    },
    {
      "epoch": 4.527391427760431,
      "grad_norm": 14.795760154724121,
      "learning_rate": 9.547544706216294e-05,
      "loss": 0.14780919551849364,
      "step": 15950
    },
    {
      "epoch": 4.530229917683792,
      "grad_norm": 10.435012817382812,
      "learning_rate": 9.547260857223957e-05,
      "loss": 0.1737540364265442,
      "step": 15960
    },
    {
      "epoch": 4.5330684076071535,
      "grad_norm": 10.23220157623291,
      "learning_rate": 9.546977008231621e-05,
      "loss": 0.12276431322097778,
      "step": 15970
    },
    {
      "epoch": 4.535906897530514,
      "grad_norm": 8.545750617980957,
      "learning_rate": 9.546693159239285e-05,
      "loss": 0.1404184579849243,
      "step": 15980
    },
    {
      "epoch": 4.538745387453875,
      "grad_norm": 17.83550262451172,
      "learning_rate": 9.546409310246949e-05,
      "loss": 0.1576627016067505,
      "step": 15990
    },
    {
      "epoch": 4.541583877377235,
      "grad_norm": 19.00641632080078,
      "learning_rate": 9.546125461254612e-05,
      "loss": 0.14220670461654664,
      "step": 16000
    },
    {
      "epoch": 4.541583877377235,
      "eval_accuracy": 0.9019520569720862,
      "eval_loss": 0.2884344756603241,
      "eval_runtime": 31.391,
      "eval_samples_per_second": 501.004,
      "eval_steps_per_second": 7.837,
      "step": 16000
    },
    {
      "epoch": 4.544422367300596,
      "grad_norm": 13.240455627441406,
      "learning_rate": 9.545841612262278e-05,
      "loss": 0.1689929962158203,
      "step": 16010
    },
    {
      "epoch": 4.547260857223957,
      "grad_norm": 19.55441665649414,
      "learning_rate": 9.545557763269942e-05,
      "loss": 0.14297375679016114,
      "step": 16020
    },
    {
      "epoch": 4.5500993471473175,
      "grad_norm": 10.37378978729248,
      "learning_rate": 9.545273914277604e-05,
      "loss": 0.14217883348464966,
      "step": 16030
    },
    {
      "epoch": 4.552937837070679,
      "grad_norm": 22.87603759765625,
      "learning_rate": 9.544990065285269e-05,
      "loss": 0.17013124227523804,
      "step": 16040
    },
    {
      "epoch": 4.555776326994039,
      "grad_norm": 16.714683532714844,
      "learning_rate": 9.544706216292933e-05,
      "loss": 0.18330156803131104,
      "step": 16050
    },
    {
      "epoch": 4.5586148169174,
      "grad_norm": 12.552401542663574,
      "learning_rate": 9.544422367300596e-05,
      "loss": 0.13127771615982056,
      "step": 16060
    },
    {
      "epoch": 4.561453306840761,
      "grad_norm": 20.79747772216797,
      "learning_rate": 9.544138518308261e-05,
      "loss": 0.1300564408302307,
      "step": 16070
    },
    {
      "epoch": 4.564291796764121,
      "grad_norm": 18.15772819519043,
      "learning_rate": 9.543854669315925e-05,
      "loss": 0.15373085737228392,
      "step": 16080
    },
    {
      "epoch": 4.567130286687482,
      "grad_norm": 12.785666465759277,
      "learning_rate": 9.543570820323588e-05,
      "loss": 0.16391913890838622,
      "step": 16090
    },
    {
      "epoch": 4.5699687766108426,
      "grad_norm": 7.867143630981445,
      "learning_rate": 9.543286971331252e-05,
      "loss": 0.14312713146209716,
      "step": 16100
    },
    {
      "epoch": 4.572807266534204,
      "grad_norm": 12.859415054321289,
      "learning_rate": 9.543003122338916e-05,
      "loss": 0.13727431297302245,
      "step": 16110
    },
    {
      "epoch": 4.575645756457565,
      "grad_norm": 16.947664260864258,
      "learning_rate": 9.54271927334658e-05,
      "loss": 0.14375555515289307,
      "step": 16120
    },
    {
      "epoch": 4.578484246380925,
      "grad_norm": 10.144893646240234,
      "learning_rate": 9.542435424354243e-05,
      "loss": 0.13479676246643066,
      "step": 16130
    },
    {
      "epoch": 4.581322736304286,
      "grad_norm": 15.607450485229492,
      "learning_rate": 9.542151575361909e-05,
      "loss": 0.15900163650512694,
      "step": 16140
    },
    {
      "epoch": 4.584161226227647,
      "grad_norm": 10.937230110168457,
      "learning_rate": 9.541867726369573e-05,
      "loss": 0.15017277002334595,
      "step": 16150
    },
    {
      "epoch": 4.586999716151007,
      "grad_norm": 13.372411727905273,
      "learning_rate": 9.541583877377236e-05,
      "loss": 0.16198084354400635,
      "step": 16160
    },
    {
      "epoch": 4.5898382060743685,
      "grad_norm": 16.571212768554688,
      "learning_rate": 9.5413000283849e-05,
      "loss": 0.13171954154968263,
      "step": 16170
    },
    {
      "epoch": 4.59267669599773,
      "grad_norm": 13.738970756530762,
      "learning_rate": 9.541016179392564e-05,
      "loss": 0.15878405570983886,
      "step": 16180
    },
    {
      "epoch": 4.59551518592109,
      "grad_norm": 18.799890518188477,
      "learning_rate": 9.540732330400227e-05,
      "loss": 0.15071109533309937,
      "step": 16190
    },
    {
      "epoch": 4.598353675844451,
      "grad_norm": 24.42254638671875,
      "learning_rate": 9.540448481407891e-05,
      "loss": 0.17055392265319824,
      "step": 16200
    },
    {
      "epoch": 4.601192165767811,
      "grad_norm": 13.94593334197998,
      "learning_rate": 9.540164632415556e-05,
      "loss": 0.12507585287094117,
      "step": 16210
    },
    {
      "epoch": 4.604030655691172,
      "grad_norm": 12.092589378356934,
      "learning_rate": 9.539880783423219e-05,
      "loss": 0.12754325866699218,
      "step": 16220
    },
    {
      "epoch": 4.606869145614533,
      "grad_norm": 9.539417266845703,
      "learning_rate": 9.539596934430883e-05,
      "loss": 0.15025877952575684,
      "step": 16230
    },
    {
      "epoch": 4.609707635537894,
      "grad_norm": 15.437283515930176,
      "learning_rate": 9.539313085438547e-05,
      "loss": 0.1816194772720337,
      "step": 16240
    },
    {
      "epoch": 4.612546125461255,
      "grad_norm": 17.008625030517578,
      "learning_rate": 9.539029236446211e-05,
      "loss": 0.18534044027328492,
      "step": 16250
    },
    {
      "epoch": 4.615384615384615,
      "grad_norm": 16.05769157409668,
      "learning_rate": 9.538745387453874e-05,
      "loss": 0.12883687019348145,
      "step": 16260
    },
    {
      "epoch": 4.618223105307976,
      "grad_norm": 13.185182571411133,
      "learning_rate": 9.53846153846154e-05,
      "loss": 0.12575019598007203,
      "step": 16270
    },
    {
      "epoch": 4.621061595231337,
      "grad_norm": 7.532161235809326,
      "learning_rate": 9.538177689469204e-05,
      "loss": 0.11277498006820678,
      "step": 16280
    },
    {
      "epoch": 4.623900085154697,
      "grad_norm": 12.064852714538574,
      "learning_rate": 9.537893840476867e-05,
      "loss": 0.1498566150665283,
      "step": 16290
    },
    {
      "epoch": 4.6267385750780585,
      "grad_norm": 15.776162147521973,
      "learning_rate": 9.537609991484531e-05,
      "loss": 0.2174987554550171,
      "step": 16300
    },
    {
      "epoch": 4.62957706500142,
      "grad_norm": 6.656815528869629,
      "learning_rate": 9.537326142492195e-05,
      "loss": 0.11988822221755982,
      "step": 16310
    },
    {
      "epoch": 4.63241555492478,
      "grad_norm": 10.42570686340332,
      "learning_rate": 9.537042293499858e-05,
      "loss": 0.153936505317688,
      "step": 16320
    },
    {
      "epoch": 4.635254044848141,
      "grad_norm": 10.324726104736328,
      "learning_rate": 9.536758444507522e-05,
      "loss": 0.13733071088790894,
      "step": 16330
    },
    {
      "epoch": 4.638092534771501,
      "grad_norm": 16.021486282348633,
      "learning_rate": 9.536474595515187e-05,
      "loss": 0.1552325129508972,
      "step": 16340
    },
    {
      "epoch": 4.640931024694862,
      "grad_norm": 10.999706268310547,
      "learning_rate": 9.53619074652285e-05,
      "loss": 0.1193424105644226,
      "step": 16350
    },
    {
      "epoch": 4.643769514618223,
      "grad_norm": 7.4361138343811035,
      "learning_rate": 9.535906897530514e-05,
      "loss": 0.12221949100494385,
      "step": 16360
    },
    {
      "epoch": 4.646608004541584,
      "grad_norm": 23.843963623046875,
      "learning_rate": 9.535623048538178e-05,
      "loss": 0.18717553615570068,
      "step": 16370
    },
    {
      "epoch": 4.649446494464945,
      "grad_norm": 9.559157371520996,
      "learning_rate": 9.535339199545843e-05,
      "loss": 0.11549685001373292,
      "step": 16380
    },
    {
      "epoch": 4.652284984388306,
      "grad_norm": 9.604438781738281,
      "learning_rate": 9.535055350553505e-05,
      "loss": 0.1513538956642151,
      "step": 16390
    },
    {
      "epoch": 4.655123474311666,
      "grad_norm": 12.756381034851074,
      "learning_rate": 9.53477150156117e-05,
      "loss": 0.14272687435150147,
      "step": 16400
    },
    {
      "epoch": 4.657961964235027,
      "grad_norm": 13.752169609069824,
      "learning_rate": 9.534487652568835e-05,
      "loss": 0.13836480379104615,
      "step": 16410
    },
    {
      "epoch": 4.660800454158387,
      "grad_norm": 10.84335708618164,
      "learning_rate": 9.534203803576498e-05,
      "loss": 0.16139088869094848,
      "step": 16420
    },
    {
      "epoch": 4.6636389440817485,
      "grad_norm": 18.47174072265625,
      "learning_rate": 9.533919954584162e-05,
      "loss": 0.15852797031402588,
      "step": 16430
    },
    {
      "epoch": 4.66647743400511,
      "grad_norm": 23.645231246948242,
      "learning_rate": 9.533636105591826e-05,
      "loss": 0.14505040645599365,
      "step": 16440
    },
    {
      "epoch": 4.66931592392847,
      "grad_norm": 8.398719787597656,
      "learning_rate": 9.533352256599489e-05,
      "loss": 0.15253835916519165,
      "step": 16450
    },
    {
      "epoch": 4.672154413851831,
      "grad_norm": 14.186738967895508,
      "learning_rate": 9.533068407607153e-05,
      "loss": 0.1989857077598572,
      "step": 16460
    },
    {
      "epoch": 4.674992903775191,
      "grad_norm": 10.147993087768555,
      "learning_rate": 9.532784558614818e-05,
      "loss": 0.1429014563560486,
      "step": 16470
    },
    {
      "epoch": 4.677831393698552,
      "grad_norm": 16.096384048461914,
      "learning_rate": 9.532500709622481e-05,
      "loss": 0.14322514533996583,
      "step": 16480
    },
    {
      "epoch": 4.680669883621913,
      "grad_norm": 18.274959564208984,
      "learning_rate": 9.532216860630145e-05,
      "loss": 0.1792846441268921,
      "step": 16490
    },
    {
      "epoch": 4.683508373545274,
      "grad_norm": 12.338505744934082,
      "learning_rate": 9.53193301163781e-05,
      "loss": 0.18460575342178345,
      "step": 16500
    },
    {
      "epoch": 4.683508373545274,
      "eval_accuracy": 0.9016341323837985,
      "eval_loss": 0.28733178973197937,
      "eval_runtime": 30.4821,
      "eval_samples_per_second": 515.942,
      "eval_steps_per_second": 8.07,
      "step": 16500
    },
    {
      "epoch": 4.686346863468635,
      "grad_norm": 12.89003849029541,
      "learning_rate": 9.531649162645474e-05,
      "loss": 0.1602902889251709,
      "step": 16510
    },
    {
      "epoch": 4.689185353391996,
      "grad_norm": 19.42878532409668,
      "learning_rate": 9.531365313653136e-05,
      "loss": 0.16873326301574706,
      "step": 16520
    },
    {
      "epoch": 4.692023843315356,
      "grad_norm": 16.303058624267578,
      "learning_rate": 9.5310814646608e-05,
      "loss": 0.1738981008529663,
      "step": 16530
    },
    {
      "epoch": 4.694862333238717,
      "grad_norm": 15.617766380310059,
      "learning_rate": 9.530797615668465e-05,
      "loss": 0.19199310541152953,
      "step": 16540
    },
    {
      "epoch": 4.697700823162078,
      "grad_norm": 19.003225326538086,
      "learning_rate": 9.530513766676129e-05,
      "loss": 0.12804521322250367,
      "step": 16550
    },
    {
      "epoch": 4.7005393130854385,
      "grad_norm": 11.42994499206543,
      "learning_rate": 9.530229917683793e-05,
      "loss": 0.17322299480438233,
      "step": 16560
    },
    {
      "epoch": 4.7033778030088,
      "grad_norm": 14.426518440246582,
      "learning_rate": 9.529946068691457e-05,
      "loss": 0.16375750303268433,
      "step": 16570
    },
    {
      "epoch": 4.70621629293216,
      "grad_norm": 18.268091201782227,
      "learning_rate": 9.52966221969912e-05,
      "loss": 0.13991639614105225,
      "step": 16580
    },
    {
      "epoch": 4.709054782855521,
      "grad_norm": 15.145362854003906,
      "learning_rate": 9.529378370706784e-05,
      "loss": 0.14734928607940673,
      "step": 16590
    },
    {
      "epoch": 4.711893272778882,
      "grad_norm": 12.734895706176758,
      "learning_rate": 9.529094521714448e-05,
      "loss": 0.18087154626846313,
      "step": 16600
    },
    {
      "epoch": 4.714731762702242,
      "grad_norm": 13.470159530639648,
      "learning_rate": 9.528810672722112e-05,
      "loss": 0.1604411482810974,
      "step": 16610
    },
    {
      "epoch": 4.717570252625603,
      "grad_norm": 14.53005599975586,
      "learning_rate": 9.528526823729776e-05,
      "loss": 0.13916300535202025,
      "step": 16620
    },
    {
      "epoch": 4.720408742548964,
      "grad_norm": 19.849550247192383,
      "learning_rate": 9.52824297473744e-05,
      "loss": 0.1767083764076233,
      "step": 16630
    },
    {
      "epoch": 4.723247232472325,
      "grad_norm": 8.63602066040039,
      "learning_rate": 9.527959125745103e-05,
      "loss": 0.1522284507751465,
      "step": 16640
    },
    {
      "epoch": 4.726085722395686,
      "grad_norm": 10.36530590057373,
      "learning_rate": 9.527675276752767e-05,
      "loss": 0.14065526723861693,
      "step": 16650
    },
    {
      "epoch": 4.728924212319046,
      "grad_norm": 9.742255210876465,
      "learning_rate": 9.527391427760432e-05,
      "loss": 0.14950262308120726,
      "step": 16660
    },
    {
      "epoch": 4.731762702242407,
      "grad_norm": 10.825611114501953,
      "learning_rate": 9.527107578768096e-05,
      "loss": 0.15421334505081177,
      "step": 16670
    },
    {
      "epoch": 4.734601192165767,
      "grad_norm": 6.3446946144104,
      "learning_rate": 9.52682372977576e-05,
      "loss": 0.11357299089431763,
      "step": 16680
    },
    {
      "epoch": 4.7374396820891285,
      "grad_norm": 12.798687934875488,
      "learning_rate": 9.526539880783424e-05,
      "loss": 0.13540432453155518,
      "step": 16690
    },
    {
      "epoch": 4.74027817201249,
      "grad_norm": 25.06239891052246,
      "learning_rate": 9.526256031791088e-05,
      "loss": 0.16007757186889648,
      "step": 16700
    },
    {
      "epoch": 4.74311666193585,
      "grad_norm": 16.52006721496582,
      "learning_rate": 9.525972182798751e-05,
      "loss": 0.15668725967407227,
      "step": 16710
    },
    {
      "epoch": 4.745955151859211,
      "grad_norm": 22.51520538330078,
      "learning_rate": 9.525688333806415e-05,
      "loss": 0.1538001298904419,
      "step": 16720
    },
    {
      "epoch": 4.748793641782572,
      "grad_norm": 27.855426788330078,
      "learning_rate": 9.525404484814079e-05,
      "loss": 0.1712619423866272,
      "step": 16730
    },
    {
      "epoch": 4.751632131705932,
      "grad_norm": 20.748098373413086,
      "learning_rate": 9.525120635821743e-05,
      "loss": 0.12831650972366332,
      "step": 16740
    },
    {
      "epoch": 4.754470621629293,
      "grad_norm": 12.2571439743042,
      "learning_rate": 9.524836786829407e-05,
      "loss": 0.1558891773223877,
      "step": 16750
    },
    {
      "epoch": 4.7573091115526545,
      "grad_norm": 8.167339324951172,
      "learning_rate": 9.524552937837072e-05,
      "loss": 0.11506112813949584,
      "step": 16760
    },
    {
      "epoch": 4.760147601476015,
      "grad_norm": 15.929369926452637,
      "learning_rate": 9.524269088844734e-05,
      "loss": 0.12471755743026733,
      "step": 16770
    },
    {
      "epoch": 4.762986091399376,
      "grad_norm": 30.671676635742188,
      "learning_rate": 9.523985239852399e-05,
      "loss": 0.174582839012146,
      "step": 16780
    },
    {
      "epoch": 4.765824581322736,
      "grad_norm": 14.295692443847656,
      "learning_rate": 9.523701390860063e-05,
      "loss": 0.15542949438095094,
      "step": 16790
    },
    {
      "epoch": 4.768663071246097,
      "grad_norm": 12.083245277404785,
      "learning_rate": 9.523417541867727e-05,
      "loss": 0.15535662174224854,
      "step": 16800
    },
    {
      "epoch": 4.771501561169458,
      "grad_norm": 6.681707859039307,
      "learning_rate": 9.523133692875391e-05,
      "loss": 0.140674364566803,
      "step": 16810
    },
    {
      "epoch": 4.7743400510928184,
      "grad_norm": 22.708444595336914,
      "learning_rate": 9.522849843883055e-05,
      "loss": 0.11735382080078124,
      "step": 16820
    },
    {
      "epoch": 4.77717854101618,
      "grad_norm": 14.901098251342773,
      "learning_rate": 9.522565994890719e-05,
      "loss": 0.1418764591217041,
      "step": 16830
    },
    {
      "epoch": 4.78001703093954,
      "grad_norm": 10.739974021911621,
      "learning_rate": 9.522282145898382e-05,
      "loss": 0.15952656269073487,
      "step": 16840
    },
    {
      "epoch": 4.782855520862901,
      "grad_norm": 10.795823097229004,
      "learning_rate": 9.521998296906046e-05,
      "loss": 0.1389915704727173,
      "step": 16850
    },
    {
      "epoch": 4.785694010786262,
      "grad_norm": 20.781103134155273,
      "learning_rate": 9.52171444791371e-05,
      "loss": 0.1436087965965271,
      "step": 16860
    },
    {
      "epoch": 4.788532500709622,
      "grad_norm": 13.771445274353027,
      "learning_rate": 9.521430598921374e-05,
      "loss": 0.13742592334747314,
      "step": 16870
    },
    {
      "epoch": 4.791370990632983,
      "grad_norm": 7.738908767700195,
      "learning_rate": 9.521146749929039e-05,
      "loss": 0.16615771055221557,
      "step": 16880
    },
    {
      "epoch": 4.7942094805563436,
      "grad_norm": 8.936503410339355,
      "learning_rate": 9.520862900936703e-05,
      "loss": 0.13377295732498168,
      "step": 16890
    },
    {
      "epoch": 4.797047970479705,
      "grad_norm": 11.112290382385254,
      "learning_rate": 9.520579051944365e-05,
      "loss": 0.1536064028739929,
      "step": 16900
    },
    {
      "epoch": 4.799886460403066,
      "grad_norm": 14.887866973876953,
      "learning_rate": 9.52029520295203e-05,
      "loss": 0.20455670356750488,
      "step": 16910
    },
    {
      "epoch": 4.802724950326426,
      "grad_norm": 11.61521053314209,
      "learning_rate": 9.520011353959694e-05,
      "loss": 0.1540137529373169,
      "step": 16920
    },
    {
      "epoch": 4.805563440249787,
      "grad_norm": 8.329435348510742,
      "learning_rate": 9.519727504967358e-05,
      "loss": 0.13170146942138672,
      "step": 16930
    },
    {
      "epoch": 4.808401930173148,
      "grad_norm": 15.267801284790039,
      "learning_rate": 9.519443655975022e-05,
      "loss": 0.12724708318710326,
      "step": 16940
    },
    {
      "epoch": 4.811240420096508,
      "grad_norm": 14.530689239501953,
      "learning_rate": 9.519159806982686e-05,
      "loss": 0.12465198040008545,
      "step": 16950
    },
    {
      "epoch": 4.8140789100198695,
      "grad_norm": 13.377323150634766,
      "learning_rate": 9.51887595799035e-05,
      "loss": 0.12636666297912597,
      "step": 16960
    },
    {
      "epoch": 4.816917399943231,
      "grad_norm": 16.755334854125977,
      "learning_rate": 9.518592108998013e-05,
      "loss": 0.11700376272201538,
      "step": 16970
    },
    {
      "epoch": 4.819755889866591,
      "grad_norm": 11.901659965515137,
      "learning_rate": 9.518308260005677e-05,
      "loss": 0.15317986011505128,
      "step": 16980
    },
    {
      "epoch": 4.822594379789952,
      "grad_norm": 15.167632102966309,
      "learning_rate": 9.518024411013341e-05,
      "loss": 0.15732300281524658,
      "step": 16990
    },
    {
      "epoch": 4.825432869713312,
      "grad_norm": 10.112268447875977,
      "learning_rate": 9.517740562021005e-05,
      "loss": 0.15507109165191652,
      "step": 17000
    },
    {
      "epoch": 4.825432869713312,
      "eval_accuracy": 0.9168309277039486,
      "eval_loss": 0.2476796805858612,
      "eval_runtime": 30.1905,
      "eval_samples_per_second": 520.925,
      "eval_steps_per_second": 8.148,
      "step": 17000
    },
    {
      "epoch": 4.828271359636673,
      "grad_norm": 15.758947372436523,
      "learning_rate": 9.51745671302867e-05,
      "loss": 0.1439222812652588,
      "step": 17010
    },
    {
      "epoch": 4.831109849560034,
      "grad_norm": 12.99755573272705,
      "learning_rate": 9.517172864036334e-05,
      "loss": 0.12416312694549561,
      "step": 17020
    },
    {
      "epoch": 4.833948339483395,
      "grad_norm": 11.258211135864258,
      "learning_rate": 9.516889015043997e-05,
      "loss": 0.14969966411590577,
      "step": 17030
    },
    {
      "epoch": 4.836786829406756,
      "grad_norm": 8.632542610168457,
      "learning_rate": 9.516605166051661e-05,
      "loss": 0.14188437461853026,
      "step": 17040
    },
    {
      "epoch": 4.839625319330116,
      "grad_norm": 12.363591194152832,
      "learning_rate": 9.516321317059325e-05,
      "loss": 0.12322738170623779,
      "step": 17050
    },
    {
      "epoch": 4.842463809253477,
      "grad_norm": 14.226353645324707,
      "learning_rate": 9.516037468066989e-05,
      "loss": 0.11890636682510376,
      "step": 17060
    },
    {
      "epoch": 4.845302299176838,
      "grad_norm": 7.976502418518066,
      "learning_rate": 9.515753619074653e-05,
      "loss": 0.13820751905441284,
      "step": 17070
    },
    {
      "epoch": 4.848140789100198,
      "grad_norm": 9.916215896606445,
      "learning_rate": 9.515469770082317e-05,
      "loss": 0.12979166507720946,
      "step": 17080
    },
    {
      "epoch": 4.8509792790235595,
      "grad_norm": 14.709095001220703,
      "learning_rate": 9.515185921089981e-05,
      "loss": 0.1539337158203125,
      "step": 17090
    },
    {
      "epoch": 4.85381776894692,
      "grad_norm": 16.777894973754883,
      "learning_rate": 9.514902072097644e-05,
      "loss": 0.13372565507888795,
      "step": 17100
    },
    {
      "epoch": 4.856656258870281,
      "grad_norm": 16.204736709594727,
      "learning_rate": 9.514618223105308e-05,
      "loss": 0.18906499147415162,
      "step": 17110
    },
    {
      "epoch": 4.859494748793642,
      "grad_norm": 15.491095542907715,
      "learning_rate": 9.514334374112972e-05,
      "loss": 0.1642136812210083,
      "step": 17120
    },
    {
      "epoch": 4.862333238717002,
      "grad_norm": 16.478290557861328,
      "learning_rate": 9.514050525120635e-05,
      "loss": 0.16236262321472167,
      "step": 17130
    },
    {
      "epoch": 4.865171728640363,
      "grad_norm": 6.901222229003906,
      "learning_rate": 9.513766676128301e-05,
      "loss": 0.2116683006286621,
      "step": 17140
    },
    {
      "epoch": 4.868010218563724,
      "grad_norm": 8.407983779907227,
      "learning_rate": 9.513482827135965e-05,
      "loss": 0.12644619941711427,
      "step": 17150
    },
    {
      "epoch": 4.870848708487085,
      "grad_norm": 12.428938865661621,
      "learning_rate": 9.513198978143628e-05,
      "loss": 0.14198888540267945,
      "step": 17160
    },
    {
      "epoch": 4.873687198410446,
      "grad_norm": 11.895127296447754,
      "learning_rate": 9.512915129151292e-05,
      "loss": 0.17176514863967896,
      "step": 17170
    },
    {
      "epoch": 4.876525688333807,
      "grad_norm": 12.704463958740234,
      "learning_rate": 9.512631280158956e-05,
      "loss": 0.15700401067733766,
      "step": 17180
    },
    {
      "epoch": 4.879364178257167,
      "grad_norm": 9.89565372467041,
      "learning_rate": 9.51234743116662e-05,
      "loss": 0.13275704383850098,
      "step": 17190
    },
    {
      "epoch": 4.882202668180528,
      "grad_norm": 10.422303199768066,
      "learning_rate": 9.512063582174284e-05,
      "loss": 0.13211022615432738,
      "step": 17200
    },
    {
      "epoch": 4.885041158103888,
      "grad_norm": 22.244260787963867,
      "learning_rate": 9.511779733181948e-05,
      "loss": 0.17186893224716188,
      "step": 17210
    },
    {
      "epoch": 4.8878796480272495,
      "grad_norm": 12.746564865112305,
      "learning_rate": 9.511495884189612e-05,
      "loss": 0.10704275369644164,
      "step": 17220
    },
    {
      "epoch": 4.890718137950611,
      "grad_norm": 11.201715469360352,
      "learning_rate": 9.511212035197275e-05,
      "loss": 0.13733270168304443,
      "step": 17230
    },
    {
      "epoch": 4.893556627873971,
      "grad_norm": 11.69204330444336,
      "learning_rate": 9.51092818620494e-05,
      "loss": 0.13713716268539428,
      "step": 17240
    },
    {
      "epoch": 4.896395117797332,
      "grad_norm": 8.279518127441406,
      "learning_rate": 9.510644337212603e-05,
      "loss": 0.08992868661880493,
      "step": 17250
    },
    {
      "epoch": 4.899233607720692,
      "grad_norm": 16.476802825927734,
      "learning_rate": 9.510360488220266e-05,
      "loss": 0.12493051290512085,
      "step": 17260
    },
    {
      "epoch": 4.902072097644053,
      "grad_norm": 15.440129280090332,
      "learning_rate": 9.510076639227932e-05,
      "loss": 0.14535372257232665,
      "step": 17270
    },
    {
      "epoch": 4.904910587567414,
      "grad_norm": 15.233850479125977,
      "learning_rate": 9.509792790235596e-05,
      "loss": 0.1533084034919739,
      "step": 17280
    },
    {
      "epoch": 4.907749077490775,
      "grad_norm": 7.581547260284424,
      "learning_rate": 9.509508941243259e-05,
      "loss": 0.1507856249809265,
      "step": 17290
    },
    {
      "epoch": 4.910587567414136,
      "grad_norm": 7.301828384399414,
      "learning_rate": 9.509225092250923e-05,
      "loss": 0.11710257530212402,
      "step": 17300
    },
    {
      "epoch": 4.913426057337497,
      "grad_norm": 16.295379638671875,
      "learning_rate": 9.508941243258587e-05,
      "loss": 0.14207034111022948,
      "step": 17310
    },
    {
      "epoch": 4.916264547260857,
      "grad_norm": 15.208168029785156,
      "learning_rate": 9.508657394266251e-05,
      "loss": 0.2113130807876587,
      "step": 17320
    },
    {
      "epoch": 4.919103037184218,
      "grad_norm": 17.72686004638672,
      "learning_rate": 9.508373545273914e-05,
      "loss": 0.1869381308555603,
      "step": 17330
    },
    {
      "epoch": 4.921941527107579,
      "grad_norm": 6.089931964874268,
      "learning_rate": 9.50808969628158e-05,
      "loss": 0.14198582172393798,
      "step": 17340
    },
    {
      "epoch": 4.9247800170309395,
      "grad_norm": 18.309528350830078,
      "learning_rate": 9.507805847289242e-05,
      "loss": 0.13088276386260986,
      "step": 17350
    },
    {
      "epoch": 4.927618506954301,
      "grad_norm": 10.71265697479248,
      "learning_rate": 9.507521998296906e-05,
      "loss": 0.14974571466445924,
      "step": 17360
    },
    {
      "epoch": 4.930456996877661,
      "grad_norm": 17.39009666442871,
      "learning_rate": 9.50723814930457e-05,
      "loss": 0.1483999013900757,
      "step": 17370
    },
    {
      "epoch": 4.933295486801022,
      "grad_norm": 14.069708824157715,
      "learning_rate": 9.506954300312235e-05,
      "loss": 0.1300961494445801,
      "step": 17380
    },
    {
      "epoch": 4.936133976724383,
      "grad_norm": 16.622167587280273,
      "learning_rate": 9.506670451319897e-05,
      "loss": 0.1281723976135254,
      "step": 17390
    },
    {
      "epoch": 4.938972466647743,
      "grad_norm": 11.97647476196289,
      "learning_rate": 9.506386602327563e-05,
      "loss": 0.12952659130096436,
      "step": 17400
    },
    {
      "epoch": 4.941810956571104,
      "grad_norm": 16.319665908813477,
      "learning_rate": 9.506102753335227e-05,
      "loss": 0.10434367656707763,
      "step": 17410
    },
    {
      "epoch": 4.944649446494465,
      "grad_norm": 15.70507526397705,
      "learning_rate": 9.50581890434289e-05,
      "loss": 0.12239481210708618,
      "step": 17420
    },
    {
      "epoch": 4.947487936417826,
      "grad_norm": 15.570923805236816,
      "learning_rate": 9.505535055350554e-05,
      "loss": 0.1408989191055298,
      "step": 17430
    },
    {
      "epoch": 4.950326426341187,
      "grad_norm": 9.592642784118652,
      "learning_rate": 9.505251206358218e-05,
      "loss": 0.15614187717437744,
      "step": 17440
    },
    {
      "epoch": 4.953164916264547,
      "grad_norm": 13.035661697387695,
      "learning_rate": 9.504967357365882e-05,
      "loss": 0.1270779013633728,
      "step": 17450
    },
    {
      "epoch": 4.956003406187908,
      "grad_norm": 9.722272872924805,
      "learning_rate": 9.504683508373545e-05,
      "loss": 0.11495968103408813,
      "step": 17460
    },
    {
      "epoch": 4.958841896111268,
      "grad_norm": 14.22462272644043,
      "learning_rate": 9.50439965938121e-05,
      "loss": 0.156838059425354,
      "step": 17470
    },
    {
      "epoch": 4.9616803860346295,
      "grad_norm": 12.47933578491211,
      "learning_rate": 9.504115810388873e-05,
      "loss": 0.14552901983261107,
      "step": 17480
    },
    {
      "epoch": 4.964518875957991,
      "grad_norm": 12.599534034729004,
      "learning_rate": 9.503831961396537e-05,
      "loss": 0.1571561336517334,
      "step": 17490
    },
    {
      "epoch": 4.967357365881351,
      "grad_norm": 9.423718452453613,
      "learning_rate": 9.503548112404202e-05,
      "loss": 0.15664663314819335,
      "step": 17500
    },
    {
      "epoch": 4.967357365881351,
      "eval_accuracy": 0.9085648884084695,
      "eval_loss": 0.2735145092010498,
      "eval_runtime": 30.6411,
      "eval_samples_per_second": 513.265,
      "eval_steps_per_second": 8.028,
      "step": 17500
    },
    {
      "epoch": 4.970195855804712,
      "grad_norm": 8.23122787475586,
      "learning_rate": 9.503264263411866e-05,
      "loss": 0.15310896635055543,
      "step": 17510
    },
    {
      "epoch": 4.973034345728073,
      "grad_norm": 11.388163566589355,
      "learning_rate": 9.502980414419528e-05,
      "loss": 0.12541829347610473,
      "step": 17520
    },
    {
      "epoch": 4.975872835651433,
      "grad_norm": 8.645771980285645,
      "learning_rate": 9.502696565427193e-05,
      "loss": 0.11650108098983765,
      "step": 17530
    },
    {
      "epoch": 4.978711325574794,
      "grad_norm": 8.968881607055664,
      "learning_rate": 9.502412716434858e-05,
      "loss": 0.160172176361084,
      "step": 17540
    },
    {
      "epoch": 4.9815498154981555,
      "grad_norm": 17.769880294799805,
      "learning_rate": 9.502128867442521e-05,
      "loss": 0.16284531354904175,
      "step": 17550
    },
    {
      "epoch": 4.984388305421516,
      "grad_norm": 15.417827606201172,
      "learning_rate": 9.501845018450185e-05,
      "loss": 0.14024698734283447,
      "step": 17560
    },
    {
      "epoch": 4.987226795344877,
      "grad_norm": 18.253232955932617,
      "learning_rate": 9.501561169457849e-05,
      "loss": 0.16308448314666749,
      "step": 17570
    },
    {
      "epoch": 4.990065285268237,
      "grad_norm": 8.247485160827637,
      "learning_rate": 9.501277320465512e-05,
      "loss": 0.1292736291885376,
      "step": 17580
    },
    {
      "epoch": 4.992903775191598,
      "grad_norm": 17.825857162475586,
      "learning_rate": 9.500993471473176e-05,
      "loss": 0.14941771030426027,
      "step": 17590
    },
    {
      "epoch": 4.995742265114959,
      "grad_norm": 17.01661491394043,
      "learning_rate": 9.500709622480842e-05,
      "loss": 0.14126372337341309,
      "step": 17600
    },
    {
      "epoch": 4.9985807550383194,
      "grad_norm": 11.892715454101562,
      "learning_rate": 9.500425773488504e-05,
      "loss": 0.1165812611579895,
      "step": 17610
    },
    {
      "epoch": 5.0014192449616806,
      "grad_norm": 13.998254776000977,
      "learning_rate": 9.500141924496168e-05,
      "loss": 0.1303948163986206,
      "step": 17620
    },
    {
      "epoch": 5.004257734885041,
      "grad_norm": 13.037029266357422,
      "learning_rate": 9.499858075503833e-05,
      "loss": 0.10122346878051758,
      "step": 17630
    },
    {
      "epoch": 5.007096224808402,
      "grad_norm": 13.354364395141602,
      "learning_rate": 9.499574226511497e-05,
      "loss": 0.12721903324127198,
      "step": 17640
    },
    {
      "epoch": 5.009934714731763,
      "grad_norm": 14.000283241271973,
      "learning_rate": 9.49929037751916e-05,
      "loss": 0.11292828321456909,
      "step": 17650
    },
    {
      "epoch": 5.012773204655123,
      "grad_norm": 13.819226264953613,
      "learning_rate": 9.499006528526824e-05,
      "loss": 0.12236974239349366,
      "step": 17660
    },
    {
      "epoch": 5.015611694578484,
      "grad_norm": 13.245518684387207,
      "learning_rate": 9.498722679534489e-05,
      "loss": 0.14156126976013184,
      "step": 17670
    },
    {
      "epoch": 5.018450184501845,
      "grad_norm": 15.717375755310059,
      "learning_rate": 9.498438830542152e-05,
      "loss": 0.0959846019744873,
      "step": 17680
    },
    {
      "epoch": 5.021288674425206,
      "grad_norm": 15.254039764404297,
      "learning_rate": 9.498154981549816e-05,
      "loss": 0.10405156612396241,
      "step": 17690
    },
    {
      "epoch": 5.024127164348567,
      "grad_norm": 9.055869102478027,
      "learning_rate": 9.49787113255748e-05,
      "loss": 0.10309066772460937,
      "step": 17700
    },
    {
      "epoch": 5.026965654271927,
      "grad_norm": 8.052003860473633,
      "learning_rate": 9.497587283565143e-05,
      "loss": 0.10911027193069459,
      "step": 17710
    },
    {
      "epoch": 5.029804144195288,
      "grad_norm": 7.9866228103637695,
      "learning_rate": 9.497303434572807e-05,
      "loss": 0.13897613286972046,
      "step": 17720
    },
    {
      "epoch": 5.032642634118649,
      "grad_norm": 15.06964111328125,
      "learning_rate": 9.497019585580471e-05,
      "loss": 0.11954238414764404,
      "step": 17730
    },
    {
      "epoch": 5.035481124042009,
      "grad_norm": 10.784893035888672,
      "learning_rate": 9.496735736588135e-05,
      "loss": 0.13433692455291749,
      "step": 17740
    },
    {
      "epoch": 5.0383196139653705,
      "grad_norm": 14.187119483947754,
      "learning_rate": 9.4964518875958e-05,
      "loss": 0.10665247440338135,
      "step": 17750
    },
    {
      "epoch": 5.041158103888731,
      "grad_norm": 12.727693557739258,
      "learning_rate": 9.496168038603464e-05,
      "loss": 0.1077520489692688,
      "step": 17760
    },
    {
      "epoch": 5.043996593812092,
      "grad_norm": 11.795356750488281,
      "learning_rate": 9.495884189611128e-05,
      "loss": 0.12967787981033324,
      "step": 17770
    },
    {
      "epoch": 5.046835083735453,
      "grad_norm": 16.47637367248535,
      "learning_rate": 9.49560034061879e-05,
      "loss": 0.11486667394638062,
      "step": 17780
    },
    {
      "epoch": 5.049673573658813,
      "grad_norm": 14.412755966186523,
      "learning_rate": 9.495316491626455e-05,
      "loss": 0.12002516984939575,
      "step": 17790
    },
    {
      "epoch": 5.052512063582174,
      "grad_norm": 19.657102584838867,
      "learning_rate": 9.49503264263412e-05,
      "loss": 0.12900152206420898,
      "step": 17800
    },
    {
      "epoch": 5.055350553505535,
      "grad_norm": 10.46070384979248,
      "learning_rate": 9.494748793641783e-05,
      "loss": 0.12214690446853638,
      "step": 17810
    },
    {
      "epoch": 5.058189043428896,
      "grad_norm": 10.935361862182617,
      "learning_rate": 9.494464944649447e-05,
      "loss": 0.11936023235321044,
      "step": 17820
    },
    {
      "epoch": 5.061027533352257,
      "grad_norm": 14.286049842834473,
      "learning_rate": 9.494181095657111e-05,
      "loss": 0.12046651840209961,
      "step": 17830
    },
    {
      "epoch": 5.063866023275617,
      "grad_norm": 15.395772933959961,
      "learning_rate": 9.493897246664774e-05,
      "loss": 0.15066107511520385,
      "step": 17840
    },
    {
      "epoch": 5.066704513198978,
      "grad_norm": 14.674049377441406,
      "learning_rate": 9.493613397672438e-05,
      "loss": 0.09819283485412597,
      "step": 17850
    },
    {
      "epoch": 5.069543003122339,
      "grad_norm": 18.47867774963379,
      "learning_rate": 9.493329548680102e-05,
      "loss": 0.15087668895721434,
      "step": 17860
    },
    {
      "epoch": 5.072381493045699,
      "grad_norm": 7.503854751586914,
      "learning_rate": 9.493045699687766e-05,
      "loss": 0.1093890905380249,
      "step": 17870
    },
    {
      "epoch": 5.0752199829690605,
      "grad_norm": 12.142801284790039,
      "learning_rate": 9.49276185069543e-05,
      "loss": 0.1050212025642395,
      "step": 17880
    },
    {
      "epoch": 5.078058472892422,
      "grad_norm": 12.580880165100098,
      "learning_rate": 9.492478001703095e-05,
      "loss": 0.146125328540802,
      "step": 17890
    },
    {
      "epoch": 5.080896962815782,
      "grad_norm": 13.052600860595703,
      "learning_rate": 9.492194152710759e-05,
      "loss": 0.10773894786834717,
      "step": 17900
    },
    {
      "epoch": 5.083735452739143,
      "grad_norm": 17.98420524597168,
      "learning_rate": 9.491910303718422e-05,
      "loss": 0.1375831127166748,
      "step": 17910
    },
    {
      "epoch": 5.086573942662503,
      "grad_norm": 11.845067024230957,
      "learning_rate": 9.491626454726086e-05,
      "loss": 0.10428259372711182,
      "step": 17920
    },
    {
      "epoch": 5.089412432585864,
      "grad_norm": 9.506842613220215,
      "learning_rate": 9.49134260573375e-05,
      "loss": 0.113460111618042,
      "step": 17930
    },
    {
      "epoch": 5.092250922509225,
      "grad_norm": 10.9423246383667,
      "learning_rate": 9.491058756741414e-05,
      "loss": 0.11058430671691895,
      "step": 17940
    },
    {
      "epoch": 5.095089412432586,
      "grad_norm": 11.014501571655273,
      "learning_rate": 9.490774907749078e-05,
      "loss": 0.11267154216766358,
      "step": 17950
    },
    {
      "epoch": 5.097927902355947,
      "grad_norm": 7.578512668609619,
      "learning_rate": 9.490491058756742e-05,
      "loss": 0.1101533055305481,
      "step": 17960
    },
    {
      "epoch": 5.100766392279308,
      "grad_norm": 15.54187297821045,
      "learning_rate": 9.490207209764405e-05,
      "loss": 0.12509812116622926,
      "step": 17970
    },
    {
      "epoch": 5.103604882202668,
      "grad_norm": 12.168378829956055,
      "learning_rate": 9.489923360772069e-05,
      "loss": 0.12506475448608398,
      "step": 17980
    },
    {
      "epoch": 5.106443372126029,
      "grad_norm": 16.660964965820312,
      "learning_rate": 9.489639511779733e-05,
      "loss": 0.11895430088043213,
      "step": 17990
    },
    {
      "epoch": 5.109281862049389,
      "grad_norm": 10.629940032958984,
      "learning_rate": 9.489355662787398e-05,
      "loss": 0.08650976419448853,
      "step": 18000
    },
    {
      "epoch": 5.109281862049389,
      "eval_accuracy": 0.9173396070452089,
      "eval_loss": 0.24206331372261047,
      "eval_runtime": 31.1193,
      "eval_samples_per_second": 505.378,
      "eval_steps_per_second": 7.905,
      "step": 18000
    },
    {
      "epoch": 5.1121203519727505,
      "grad_norm": 11.882974624633789,
      "learning_rate": 9.489071813795062e-05,
      "loss": 0.13628501892089845,
      "step": 18010
    },
    {
      "epoch": 5.114958841896112,
      "grad_norm": 11.58086109161377,
      "learning_rate": 9.488787964802726e-05,
      "loss": 0.11117228269577026,
      "step": 18020
    },
    {
      "epoch": 5.117797331819472,
      "grad_norm": 5.872004985809326,
      "learning_rate": 9.48850411581039e-05,
      "loss": 0.09934443831443787,
      "step": 18030
    },
    {
      "epoch": 5.120635821742833,
      "grad_norm": 27.13557243347168,
      "learning_rate": 9.488220266818053e-05,
      "loss": 0.12184182405471802,
      "step": 18040
    },
    {
      "epoch": 5.123474311666193,
      "grad_norm": 7.560076713562012,
      "learning_rate": 9.487936417825717e-05,
      "loss": 0.11554148197174072,
      "step": 18050
    },
    {
      "epoch": 5.126312801589554,
      "grad_norm": 8.092820167541504,
      "learning_rate": 9.487652568833381e-05,
      "loss": 0.083376944065094,
      "step": 18060
    },
    {
      "epoch": 5.129151291512915,
      "grad_norm": 14.400120735168457,
      "learning_rate": 9.487368719841045e-05,
      "loss": 0.12058475017547607,
      "step": 18070
    },
    {
      "epoch": 5.131989781436276,
      "grad_norm": 7.0960469245910645,
      "learning_rate": 9.487084870848709e-05,
      "loss": 0.08904277682304382,
      "step": 18080
    },
    {
      "epoch": 5.134828271359637,
      "grad_norm": 11.378924369812012,
      "learning_rate": 9.486801021856373e-05,
      "loss": 0.11610628366470337,
      "step": 18090
    },
    {
      "epoch": 5.137666761282998,
      "grad_norm": 14.249679565429688,
      "learning_rate": 9.486517172864036e-05,
      "loss": 0.1266200304031372,
      "step": 18100
    },
    {
      "epoch": 5.140505251206358,
      "grad_norm": 13.318857192993164,
      "learning_rate": 9.4862333238717e-05,
      "loss": 0.13533298969268798,
      "step": 18110
    },
    {
      "epoch": 5.143343741129719,
      "grad_norm": 7.622172832489014,
      "learning_rate": 9.485949474879364e-05,
      "loss": 0.13788855075836182,
      "step": 18120
    },
    {
      "epoch": 5.146182231053079,
      "grad_norm": 10.998847007751465,
      "learning_rate": 9.485665625887029e-05,
      "loss": 0.11215749979019166,
      "step": 18130
    },
    {
      "epoch": 5.1490207209764405,
      "grad_norm": 7.949224472045898,
      "learning_rate": 9.485381776894693e-05,
      "loss": 0.08397881388664245,
      "step": 18140
    },
    {
      "epoch": 5.151859210899802,
      "grad_norm": 7.0520195960998535,
      "learning_rate": 9.485097927902357e-05,
      "loss": 0.08436256647109985,
      "step": 18150
    },
    {
      "epoch": 5.154697700823162,
      "grad_norm": 7.95007848739624,
      "learning_rate": 9.484814078910021e-05,
      "loss": 0.08683756589889527,
      "step": 18160
    },
    {
      "epoch": 5.157536190746523,
      "grad_norm": 16.676387786865234,
      "learning_rate": 9.484530229917684e-05,
      "loss": 0.09507114887237549,
      "step": 18170
    },
    {
      "epoch": 5.160374680669884,
      "grad_norm": 9.202858924865723,
      "learning_rate": 9.484246380925348e-05,
      "loss": 0.09739378094673157,
      "step": 18180
    },
    {
      "epoch": 5.163213170593244,
      "grad_norm": 13.040538787841797,
      "learning_rate": 9.483962531933012e-05,
      "loss": 0.136370849609375,
      "step": 18190
    },
    {
      "epoch": 5.166051660516605,
      "grad_norm": 12.115922927856445,
      "learning_rate": 9.483678682940676e-05,
      "loss": 0.11529729366302491,
      "step": 18200
    },
    {
      "epoch": 5.168890150439966,
      "grad_norm": 10.48679256439209,
      "learning_rate": 9.48339483394834e-05,
      "loss": 0.13028812408447266,
      "step": 18210
    },
    {
      "epoch": 5.171728640363327,
      "grad_norm": 14.922074317932129,
      "learning_rate": 9.483110984956005e-05,
      "loss": 0.11457020044326782,
      "step": 18220
    },
    {
      "epoch": 5.174567130286688,
      "grad_norm": 8.332785606384277,
      "learning_rate": 9.482827135963667e-05,
      "loss": 0.10092849731445312,
      "step": 18230
    },
    {
      "epoch": 5.177405620210048,
      "grad_norm": 14.685294151306152,
      "learning_rate": 9.482543286971331e-05,
      "loss": 0.118309485912323,
      "step": 18240
    },
    {
      "epoch": 5.180244110133409,
      "grad_norm": 3.95269513130188,
      "learning_rate": 9.482259437978996e-05,
      "loss": 0.09387722015380859,
      "step": 18250
    },
    {
      "epoch": 5.183082600056769,
      "grad_norm": 7.29939079284668,
      "learning_rate": 9.48197558898666e-05,
      "loss": 0.10037614107131958,
      "step": 18260
    },
    {
      "epoch": 5.1859210899801305,
      "grad_norm": 9.66841983795166,
      "learning_rate": 9.481691739994324e-05,
      "loss": 0.12975598573684693,
      "step": 18270
    },
    {
      "epoch": 5.188759579903492,
      "grad_norm": 25.23455238342285,
      "learning_rate": 9.481407891001988e-05,
      "loss": 0.18081820011138916,
      "step": 18280
    },
    {
      "epoch": 5.191598069826852,
      "grad_norm": 14.022313117980957,
      "learning_rate": 9.481124042009651e-05,
      "loss": 0.11226980686187744,
      "step": 18290
    },
    {
      "epoch": 5.194436559750213,
      "grad_norm": 4.847559928894043,
      "learning_rate": 9.480840193017315e-05,
      "loss": 0.09739605188369752,
      "step": 18300
    },
    {
      "epoch": 5.197275049673574,
      "grad_norm": 14.528687477111816,
      "learning_rate": 9.480556344024979e-05,
      "loss": 0.09656654000282287,
      "step": 18310
    },
    {
      "epoch": 5.200113539596934,
      "grad_norm": 15.67531967163086,
      "learning_rate": 9.480272495032643e-05,
      "loss": 0.11359162330627441,
      "step": 18320
    },
    {
      "epoch": 5.202952029520295,
      "grad_norm": 11.566393852233887,
      "learning_rate": 9.479988646040307e-05,
      "loss": 0.11102167367935181,
      "step": 18330
    },
    {
      "epoch": 5.205790519443656,
      "grad_norm": 18.80540657043457,
      "learning_rate": 9.479704797047971e-05,
      "loss": 0.14986209869384765,
      "step": 18340
    },
    {
      "epoch": 5.208629009367017,
      "grad_norm": 14.605559349060059,
      "learning_rate": 9.479420948055636e-05,
      "loss": 0.09588061571121216,
      "step": 18350
    },
    {
      "epoch": 5.211467499290378,
      "grad_norm": 24.594322204589844,
      "learning_rate": 9.479137099063298e-05,
      "loss": 0.16140575408935548,
      "step": 18360
    },
    {
      "epoch": 5.214305989213738,
      "grad_norm": 9.228494644165039,
      "learning_rate": 9.478853250070963e-05,
      "loss": 0.12389692068099975,
      "step": 18370
    },
    {
      "epoch": 5.217144479137099,
      "grad_norm": 11.408028602600098,
      "learning_rate": 9.478569401078627e-05,
      "loss": 0.11617081165313721,
      "step": 18380
    },
    {
      "epoch": 5.21998296906046,
      "grad_norm": 13.211589813232422,
      "learning_rate": 9.47828555208629e-05,
      "loss": 0.1088978886604309,
      "step": 18390
    },
    {
      "epoch": 5.22282145898382,
      "grad_norm": 9.789724349975586,
      "learning_rate": 9.478030087993189e-05,
      "loss": 0.10675936937332153,
      "step": 18400
    },
    {
      "epoch": 5.2256599489071816,
      "grad_norm": 5.0881266593933105,
      "learning_rate": 9.477746239000852e-05,
      "loss": 0.11985863447189331,
      "step": 18410
    },
    {
      "epoch": 5.228498438830542,
      "grad_norm": 22.814834594726562,
      "learning_rate": 9.477462390008516e-05,
      "loss": 0.13029314279556276,
      "step": 18420
    },
    {
      "epoch": 5.231336928753903,
      "grad_norm": 11.96026611328125,
      "learning_rate": 9.47717854101618e-05,
      "loss": 0.1350261926651001,
      "step": 18430
    },
    {
      "epoch": 5.234175418677264,
      "grad_norm": 9.897076606750488,
      "learning_rate": 9.476894692023843e-05,
      "loss": 0.1350799560546875,
      "step": 18440
    },
    {
      "epoch": 5.237013908600624,
      "grad_norm": 12.540698051452637,
      "learning_rate": 9.476610843031508e-05,
      "loss": 0.11492029428482056,
      "step": 18450
    },
    {
      "epoch": 5.239852398523985,
      "grad_norm": 13.65211296081543,
      "learning_rate": 9.476326994039172e-05,
      "loss": 0.12582682371139525,
      "step": 18460
    },
    {
      "epoch": 5.242690888447346,
      "grad_norm": 13.934420585632324,
      "learning_rate": 9.476043145046835e-05,
      "loss": 0.11009331941604614,
      "step": 18470
    },
    {
      "epoch": 5.245529378370707,
      "grad_norm": 13.780197143554688,
      "learning_rate": 9.475759296054499e-05,
      "loss": 0.1336338996887207,
      "step": 18480
    },
    {
      "epoch": 5.248367868294068,
      "grad_norm": 14.090015411376953,
      "learning_rate": 9.475475447062163e-05,
      "loss": 0.13067694902420043,
      "step": 18490
    },
    {
      "epoch": 5.251206358217428,
      "grad_norm": 15.602925300598145,
      "learning_rate": 9.475191598069827e-05,
      "loss": 0.12691853046417237,
      "step": 18500
    },
    {
      "epoch": 5.251206358217428,
      "eval_accuracy": 0.9096458320086476,
      "eval_loss": 0.270553857088089,
      "eval_runtime": 31.0448,
      "eval_samples_per_second": 506.59,
      "eval_steps_per_second": 7.924,
      "step": 18500
    },
    {
      "epoch": 5.254044848140789,
      "grad_norm": 14.531049728393555,
      "learning_rate": 9.47490774907749e-05,
      "loss": 0.10472745895385742,
      "step": 18510
    },
    {
      "epoch": 5.25688333806415,
      "grad_norm": 6.957664966583252,
      "learning_rate": 9.474623900085156e-05,
      "loss": 0.10247851610183716,
      "step": 18520
    },
    {
      "epoch": 5.25972182798751,
      "grad_norm": 13.993846893310547,
      "learning_rate": 9.47434005109282e-05,
      "loss": 0.14197319746017456,
      "step": 18530
    },
    {
      "epoch": 5.2625603179108715,
      "grad_norm": 21.744396209716797,
      "learning_rate": 9.474056202100483e-05,
      "loss": 0.1284420371055603,
      "step": 18540
    },
    {
      "epoch": 5.265398807834233,
      "grad_norm": 10.779120445251465,
      "learning_rate": 9.473772353108147e-05,
      "loss": 0.10145989656448365,
      "step": 18550
    },
    {
      "epoch": 5.268237297757593,
      "grad_norm": 13.402883529663086,
      "learning_rate": 9.473488504115811e-05,
      "loss": 0.13248554468154908,
      "step": 18560
    },
    {
      "epoch": 5.271075787680954,
      "grad_norm": 13.94957160949707,
      "learning_rate": 9.473204655123474e-05,
      "loss": 0.12498775720596314,
      "step": 18570
    },
    {
      "epoch": 5.273914277604314,
      "grad_norm": 8.157875061035156,
      "learning_rate": 9.472920806131139e-05,
      "loss": 0.12708642482757568,
      "step": 18580
    },
    {
      "epoch": 5.276752767527675,
      "grad_norm": 20.33916664123535,
      "learning_rate": 9.472636957138803e-05,
      "loss": 0.09488003253936768,
      "step": 18590
    },
    {
      "epoch": 5.279591257451036,
      "grad_norm": 11.896894454956055,
      "learning_rate": 9.472353108146466e-05,
      "loss": 0.1267958402633667,
      "step": 18600
    },
    {
      "epoch": 5.282429747374397,
      "grad_norm": 13.123417854309082,
      "learning_rate": 9.47206925915413e-05,
      "loss": 0.10330753326416016,
      "step": 18610
    },
    {
      "epoch": 5.285268237297758,
      "grad_norm": 10.870304107666016,
      "learning_rate": 9.471785410161794e-05,
      "loss": 0.12085975408554077,
      "step": 18620
    },
    {
      "epoch": 5.288106727221118,
      "grad_norm": 17.0112361907959,
      "learning_rate": 9.471501561169459e-05,
      "loss": 0.13956127166748047,
      "step": 18630
    },
    {
      "epoch": 5.290945217144479,
      "grad_norm": 10.220715522766113,
      "learning_rate": 9.471217712177121e-05,
      "loss": 0.12341954708099365,
      "step": 18640
    },
    {
      "epoch": 5.29378370706784,
      "grad_norm": 14.950878143310547,
      "learning_rate": 9.470933863184787e-05,
      "loss": 0.14755876064300538,
      "step": 18650
    },
    {
      "epoch": 5.2966221969912,
      "grad_norm": 17.65648651123047,
      "learning_rate": 9.470650014192451e-05,
      "loss": 0.139887273311615,
      "step": 18660
    },
    {
      "epoch": 5.2994606869145615,
      "grad_norm": 10.992133140563965,
      "learning_rate": 9.470366165200114e-05,
      "loss": 0.10425843000411987,
      "step": 18670
    },
    {
      "epoch": 5.302299176837923,
      "grad_norm": 8.081459999084473,
      "learning_rate": 9.470082316207778e-05,
      "loss": 0.14321550130844116,
      "step": 18680
    },
    {
      "epoch": 5.305137666761283,
      "grad_norm": 19.88718032836914,
      "learning_rate": 9.469798467215442e-05,
      "loss": 0.1343953251838684,
      "step": 18690
    },
    {
      "epoch": 5.307976156684644,
      "grad_norm": 11.913741111755371,
      "learning_rate": 9.469514618223105e-05,
      "loss": 0.11695908308029175,
      "step": 18700
    },
    {
      "epoch": 5.310814646608004,
      "grad_norm": 16.517282485961914,
      "learning_rate": 9.46923076923077e-05,
      "loss": 0.12505130767822265,
      "step": 18710
    },
    {
      "epoch": 5.313653136531365,
      "grad_norm": 8.039443016052246,
      "learning_rate": 9.468946920238434e-05,
      "loss": 0.10022202730178834,
      "step": 18720
    },
    {
      "epoch": 5.316491626454726,
      "grad_norm": 16.427576065063477,
      "learning_rate": 9.468663071246097e-05,
      "loss": 0.13602070808410643,
      "step": 18730
    },
    {
      "epoch": 5.319330116378087,
      "grad_norm": 12.405729293823242,
      "learning_rate": 9.468379222253761e-05,
      "loss": 0.09898651242256165,
      "step": 18740
    },
    {
      "epoch": 5.322168606301448,
      "grad_norm": 5.045668601989746,
      "learning_rate": 9.468095373261425e-05,
      "loss": 0.11487486362457275,
      "step": 18750
    },
    {
      "epoch": 5.325007096224809,
      "grad_norm": 8.068799018859863,
      "learning_rate": 9.46781152426909e-05,
      "loss": 0.074915611743927,
      "step": 18760
    },
    {
      "epoch": 5.327845586148169,
      "grad_norm": 9.10688304901123,
      "learning_rate": 9.467527675276752e-05,
      "loss": 0.1302726984024048,
      "step": 18770
    },
    {
      "epoch": 5.33068407607153,
      "grad_norm": 18.554283142089844,
      "learning_rate": 9.467243826284418e-05,
      "loss": 0.12635024785995483,
      "step": 18780
    },
    {
      "epoch": 5.33352256599489,
      "grad_norm": 9.430561065673828,
      "learning_rate": 9.466959977292082e-05,
      "loss": 0.13685967922210693,
      "step": 18790
    },
    {
      "epoch": 5.3363610559182515,
      "grad_norm": 6.459030628204346,
      "learning_rate": 9.466676128299745e-05,
      "loss": 0.13198906183242798,
      "step": 18800
    },
    {
      "epoch": 5.339199545841613,
      "grad_norm": 6.607358455657959,
      "learning_rate": 9.466392279307409e-05,
      "loss": 0.12675567865371704,
      "step": 18810
    },
    {
      "epoch": 5.342038035764973,
      "grad_norm": 14.684468269348145,
      "learning_rate": 9.466108430315073e-05,
      "loss": 0.13726824522018433,
      "step": 18820
    },
    {
      "epoch": 5.344876525688334,
      "grad_norm": 12.43384838104248,
      "learning_rate": 9.465824581322736e-05,
      "loss": 0.11709840297698974,
      "step": 18830
    },
    {
      "epoch": 5.347715015611694,
      "grad_norm": 7.708503246307373,
      "learning_rate": 9.4655407323304e-05,
      "loss": 0.11707191467285157,
      "step": 18840
    },
    {
      "epoch": 5.350553505535055,
      "grad_norm": 12.617171287536621,
      "learning_rate": 9.465256883338065e-05,
      "loss": 0.1273653030395508,
      "step": 18850
    },
    {
      "epoch": 5.353391995458416,
      "grad_norm": 11.896539688110352,
      "learning_rate": 9.464973034345728e-05,
      "loss": 0.11616818904876709,
      "step": 18860
    },
    {
      "epoch": 5.356230485381777,
      "grad_norm": 10.566817283630371,
      "learning_rate": 9.464689185353392e-05,
      "loss": 0.12600181102752686,
      "step": 18870
    },
    {
      "epoch": 5.359068975305138,
      "grad_norm": 2.9978444576263428,
      "learning_rate": 9.464405336361057e-05,
      "loss": 0.09637483358383178,
      "step": 18880
    },
    {
      "epoch": 5.361907465228499,
      "grad_norm": 6.65971040725708,
      "learning_rate": 9.46412148736872e-05,
      "loss": 0.13089643716812133,
      "step": 18890
    },
    {
      "epoch": 5.364745955151859,
      "grad_norm": 18.470901489257812,
      "learning_rate": 9.463837638376383e-05,
      "loss": 0.09059785604476929,
      "step": 18900
    },
    {
      "epoch": 5.36758444507522,
      "grad_norm": 13.260082244873047,
      "learning_rate": 9.463553789384049e-05,
      "loss": 0.11366134881973267,
      "step": 18910
    },
    {
      "epoch": 5.37042293499858,
      "grad_norm": 14.940563201904297,
      "learning_rate": 9.463269940391713e-05,
      "loss": 0.09691439867019654,
      "step": 18920
    },
    {
      "epoch": 5.3732614249219415,
      "grad_norm": 10.499874114990234,
      "learning_rate": 9.462986091399376e-05,
      "loss": 0.10260868072509766,
      "step": 18930
    },
    {
      "epoch": 5.376099914845303,
      "grad_norm": 10.452183723449707,
      "learning_rate": 9.46270224240704e-05,
      "loss": 0.09258331656455994,
      "step": 18940
    },
    {
      "epoch": 5.378938404768663,
      "grad_norm": 20.69902801513672,
      "learning_rate": 9.462418393414704e-05,
      "loss": 0.1747816801071167,
      "step": 18950
    },
    {
      "epoch": 5.381776894692024,
      "grad_norm": 14.414055824279785,
      "learning_rate": 9.462134544422367e-05,
      "loss": 0.12152936458587646,
      "step": 18960
    },
    {
      "epoch": 5.384615384615385,
      "grad_norm": 6.830984115600586,
      "learning_rate": 9.461850695430031e-05,
      "loss": 0.10256987810134888,
      "step": 18970
    },
    {
      "epoch": 5.387453874538745,
      "grad_norm": 16.298723220825195,
      "learning_rate": 9.461566846437697e-05,
      "loss": 0.13600679636001586,
      "step": 18980
    },
    {
      "epoch": 5.390292364462106,
      "grad_norm": 14.064005851745605,
      "learning_rate": 9.46128299744536e-05,
      "loss": 0.10061171054840087,
      "step": 18990
    },
    {
      "epoch": 5.393130854385467,
      "grad_norm": 12.299553871154785,
      "learning_rate": 9.460999148453023e-05,
      "loss": 0.11337920427322387,
      "step": 19000
    },
    {
      "epoch": 5.393130854385467,
      "eval_accuracy": 0.9239524384815921,
      "eval_loss": 0.23114651441574097,
      "eval_runtime": 31.3393,
      "eval_samples_per_second": 501.83,
      "eval_steps_per_second": 7.85,
      "step": 19000
    },
    {
      "epoch": 5.395969344308828,
      "grad_norm": 12.65729808807373,
      "learning_rate": 9.460715299460688e-05,
      "loss": 0.10458855628967285,
      "step": 19010
    },
    {
      "epoch": 5.398807834232189,
      "grad_norm": 11.864614486694336,
      "learning_rate": 9.460431450468352e-05,
      "loss": 0.12202335596084594,
      "step": 19020
    },
    {
      "epoch": 5.401646324155549,
      "grad_norm": 10.886442184448242,
      "learning_rate": 9.460147601476015e-05,
      "loss": 0.12154321670532227,
      "step": 19030
    },
    {
      "epoch": 5.40448481407891,
      "grad_norm": 13.320350646972656,
      "learning_rate": 9.459863752483679e-05,
      "loss": 0.10878154039382934,
      "step": 19040
    },
    {
      "epoch": 5.40732330400227,
      "grad_norm": 8.635933876037598,
      "learning_rate": 9.459579903491344e-05,
      "loss": 0.12345335483551026,
      "step": 19050
    },
    {
      "epoch": 5.4101617939256315,
      "grad_norm": 16.82415771484375,
      "learning_rate": 9.459296054499007e-05,
      "loss": 0.1340232253074646,
      "step": 19060
    },
    {
      "epoch": 5.413000283848993,
      "grad_norm": 5.474141597747803,
      "learning_rate": 9.459012205506671e-05,
      "loss": 0.0995680570602417,
      "step": 19070
    },
    {
      "epoch": 5.415838773772353,
      "grad_norm": 8.26756763458252,
      "learning_rate": 9.458728356514335e-05,
      "loss": 0.10731569528579712,
      "step": 19080
    },
    {
      "epoch": 5.418677263695714,
      "grad_norm": 12.569828987121582,
      "learning_rate": 9.458444507521998e-05,
      "loss": 0.10253074169158935,
      "step": 19090
    },
    {
      "epoch": 5.421515753619075,
      "grad_norm": 23.679319381713867,
      "learning_rate": 9.458160658529662e-05,
      "loss": 0.13541394472122192,
      "step": 19100
    },
    {
      "epoch": 5.424354243542435,
      "grad_norm": 10.604979515075684,
      "learning_rate": 9.457876809537328e-05,
      "loss": 0.1214879035949707,
      "step": 19110
    },
    {
      "epoch": 5.427192733465796,
      "grad_norm": 13.791625022888184,
      "learning_rate": 9.45759296054499e-05,
      "loss": 0.13752079010009766,
      "step": 19120
    },
    {
      "epoch": 5.430031223389157,
      "grad_norm": 20.201370239257812,
      "learning_rate": 9.457309111552655e-05,
      "loss": 0.12793558835983276,
      "step": 19130
    },
    {
      "epoch": 5.432869713312518,
      "grad_norm": 14.36790657043457,
      "learning_rate": 9.457025262560319e-05,
      "loss": 0.1193717360496521,
      "step": 19140
    },
    {
      "epoch": 5.435708203235879,
      "grad_norm": 9.850201606750488,
      "learning_rate": 9.456741413567983e-05,
      "loss": 0.12861897945404052,
      "step": 19150
    },
    {
      "epoch": 5.438546693159239,
      "grad_norm": 6.269471645355225,
      "learning_rate": 9.456457564575646e-05,
      "loss": 0.0933423638343811,
      "step": 19160
    },
    {
      "epoch": 5.4413851830826,
      "grad_norm": 19.406147003173828,
      "learning_rate": 9.45617371558331e-05,
      "loss": 0.12077394723892212,
      "step": 19170
    },
    {
      "epoch": 5.444223673005961,
      "grad_norm": 12.03589153289795,
      "learning_rate": 9.455889866590974e-05,
      "loss": 0.1468037486076355,
      "step": 19180
    },
    {
      "epoch": 5.447062162929321,
      "grad_norm": 12.559707641601562,
      "learning_rate": 9.455606017598638e-05,
      "loss": 0.12684504985809325,
      "step": 19190
    },
    {
      "epoch": 5.4499006528526825,
      "grad_norm": 5.962762355804443,
      "learning_rate": 9.455322168606302e-05,
      "loss": 0.10693166255950928,
      "step": 19200
    },
    {
      "epoch": 5.452739142776043,
      "grad_norm": 11.452478408813477,
      "learning_rate": 9.455038319613966e-05,
      "loss": 0.10182032585144044,
      "step": 19210
    },
    {
      "epoch": 5.455577632699404,
      "grad_norm": 15.097282409667969,
      "learning_rate": 9.454754470621629e-05,
      "loss": 0.11651647090911865,
      "step": 19220
    },
    {
      "epoch": 5.458416122622765,
      "grad_norm": 8.956564903259277,
      "learning_rate": 9.454470621629293e-05,
      "loss": 0.13075654506683348,
      "step": 19230
    },
    {
      "epoch": 5.461254612546125,
      "grad_norm": 15.3197603225708,
      "learning_rate": 9.454186772636957e-05,
      "loss": 0.11533890962600708,
      "step": 19240
    },
    {
      "epoch": 5.464093102469486,
      "grad_norm": 11.846577644348145,
      "learning_rate": 9.453902923644621e-05,
      "loss": 0.12017837762832642,
      "step": 19250
    },
    {
      "epoch": 5.4669315923928465,
      "grad_norm": 4.201696872711182,
      "learning_rate": 9.453619074652286e-05,
      "loss": 0.12061038017272949,
      "step": 19260
    },
    {
      "epoch": 5.469770082316208,
      "grad_norm": 11.541903495788574,
      "learning_rate": 9.45333522565995e-05,
      "loss": 0.11912944316864013,
      "step": 19270
    },
    {
      "epoch": 5.472608572239569,
      "grad_norm": 12.73418140411377,
      "learning_rate": 9.453051376667613e-05,
      "loss": 0.1129902720451355,
      "step": 19280
    },
    {
      "epoch": 5.475447062162929,
      "grad_norm": 5.762692451477051,
      "learning_rate": 9.452767527675277e-05,
      "loss": 0.07790331244468689,
      "step": 19290
    },
    {
      "epoch": 5.47828555208629,
      "grad_norm": 13.306546211242676,
      "learning_rate": 9.452483678682941e-05,
      "loss": 0.13790605068206788,
      "step": 19300
    },
    {
      "epoch": 5.481124042009651,
      "grad_norm": 13.98701000213623,
      "learning_rate": 9.452199829690605e-05,
      "loss": 0.12352221012115479,
      "step": 19310
    },
    {
      "epoch": 5.483962531933011,
      "grad_norm": 9.166779518127441,
      "learning_rate": 9.451915980698269e-05,
      "loss": 0.09548219442367553,
      "step": 19320
    },
    {
      "epoch": 5.4868010218563725,
      "grad_norm": 16.246734619140625,
      "learning_rate": 9.451632131705933e-05,
      "loss": 0.1522259831428528,
      "step": 19330
    },
    {
      "epoch": 5.489639511779734,
      "grad_norm": 9.676566123962402,
      "learning_rate": 9.451348282713597e-05,
      "loss": 0.11040685176849366,
      "step": 19340
    },
    {
      "epoch": 5.492478001703094,
      "grad_norm": 12.970476150512695,
      "learning_rate": 9.45106443372126e-05,
      "loss": 0.14873800277709961,
      "step": 19350
    },
    {
      "epoch": 5.495316491626455,
      "grad_norm": 3.8955798149108887,
      "learning_rate": 9.450780584728924e-05,
      "loss": 0.12185837030410766,
      "step": 19360
    },
    {
      "epoch": 5.498154981549815,
      "grad_norm": 12.275150299072266,
      "learning_rate": 9.450496735736588e-05,
      "loss": 0.1141619086265564,
      "step": 19370
    },
    {
      "epoch": 5.500993471473176,
      "grad_norm": 7.498418807983398,
      "learning_rate": 9.450212886744253e-05,
      "loss": 0.09539836645126343,
      "step": 19380
    },
    {
      "epoch": 5.503831961396537,
      "grad_norm": 9.892207145690918,
      "learning_rate": 9.449929037751917e-05,
      "loss": 0.09321224093437194,
      "step": 19390
    },
    {
      "epoch": 5.506670451319898,
      "grad_norm": 9.417450904846191,
      "learning_rate": 9.449645188759581e-05,
      "loss": 0.13593860864639282,
      "step": 19400
    },
    {
      "epoch": 5.509508941243259,
      "grad_norm": 12.9541597366333,
      "learning_rate": 9.449361339767244e-05,
      "loss": 0.12845546007156372,
      "step": 19410
    },
    {
      "epoch": 5.512347431166619,
      "grad_norm": 12.885693550109863,
      "learning_rate": 9.449077490774908e-05,
      "loss": 0.10136137008666993,
      "step": 19420
    },
    {
      "epoch": 5.51518592108998,
      "grad_norm": 11.74797248840332,
      "learning_rate": 9.448793641782572e-05,
      "loss": 0.14557892084121704,
      "step": 19430
    },
    {
      "epoch": 5.518024411013341,
      "grad_norm": 11.357150077819824,
      "learning_rate": 9.448509792790236e-05,
      "loss": 0.09730833172798156,
      "step": 19440
    },
    {
      "epoch": 5.520862900936701,
      "grad_norm": 11.29161262512207,
      "learning_rate": 9.4482259437979e-05,
      "loss": 0.10603079795837403,
      "step": 19450
    },
    {
      "epoch": 5.5237013908600625,
      "grad_norm": 28.32952308654785,
      "learning_rate": 9.447942094805564e-05,
      "loss": 0.1581097960472107,
      "step": 19460
    },
    {
      "epoch": 5.526539880783423,
      "grad_norm": 4.153390407562256,
      "learning_rate": 9.447658245813228e-05,
      "loss": 0.11754984855651855,
      "step": 19470
    },
    {
      "epoch": 5.529378370706784,
      "grad_norm": 10.587778091430664,
      "learning_rate": 9.447374396820891e-05,
      "loss": 0.15817960500717163,
      "step": 19480
    },
    {
      "epoch": 5.532216860630145,
      "grad_norm": 9.358624458312988,
      "learning_rate": 9.447090547828555e-05,
      "loss": 0.12570657730102539,
      "step": 19490
    },
    {
      "epoch": 5.535055350553505,
      "grad_norm": 10.29257583618164,
      "learning_rate": 9.44680669883622e-05,
      "loss": 0.12990448474884034,
      "step": 19500
    },
    {
      "epoch": 5.535055350553505,
      "eval_accuracy": 0.9264958351878935,
      "eval_loss": 0.22490718960762024,
      "eval_runtime": 32.3009,
      "eval_samples_per_second": 486.891,
      "eval_steps_per_second": 7.616,
      "step": 19500
    },
    {
      "epoch": 5.537893840476866,
      "grad_norm": 25.854448318481445,
      "learning_rate": 9.446522849843884e-05,
      "loss": 0.136379873752594,
      "step": 19510
    },
    {
      "epoch": 5.540732330400227,
      "grad_norm": 7.664098739624023,
      "learning_rate": 9.446239000851548e-05,
      "loss": 0.14082088470458984,
      "step": 19520
    },
    {
      "epoch": 5.543570820323588,
      "grad_norm": 16.849760055541992,
      "learning_rate": 9.445955151859212e-05,
      "loss": 0.11256475448608398,
      "step": 19530
    },
    {
      "epoch": 5.546409310246949,
      "grad_norm": 7.661827564239502,
      "learning_rate": 9.445671302866875e-05,
      "loss": 0.12128474712371826,
      "step": 19540
    },
    {
      "epoch": 5.54924780017031,
      "grad_norm": 13.366745948791504,
      "learning_rate": 9.445387453874539e-05,
      "loss": 0.12486085891723633,
      "step": 19550
    },
    {
      "epoch": 5.55208629009367,
      "grad_norm": 19.34494972229004,
      "learning_rate": 9.445103604882203e-05,
      "loss": 0.08632761240005493,
      "step": 19560
    },
    {
      "epoch": 5.554924780017031,
      "grad_norm": 16.57925796508789,
      "learning_rate": 9.444819755889867e-05,
      "loss": 0.1167499303817749,
      "step": 19570
    },
    {
      "epoch": 5.557763269940391,
      "grad_norm": 7.807219982147217,
      "learning_rate": 9.444535906897531e-05,
      "loss": 0.10497434139251709,
      "step": 19580
    },
    {
      "epoch": 5.5606017598637525,
      "grad_norm": 8.378154754638672,
      "learning_rate": 9.444252057905195e-05,
      "loss": 0.13328760862350464,
      "step": 19590
    },
    {
      "epoch": 5.563440249787114,
      "grad_norm": 17.223222732543945,
      "learning_rate": 9.44396820891286e-05,
      "loss": 0.12755695581436158,
      "step": 19600
    },
    {
      "epoch": 5.566278739710474,
      "grad_norm": 11.267366409301758,
      "learning_rate": 9.443684359920522e-05,
      "loss": 0.1144278883934021,
      "step": 19610
    },
    {
      "epoch": 5.569117229633835,
      "grad_norm": 12.408498764038086,
      "learning_rate": 9.443400510928186e-05,
      "loss": 0.09208456873893738,
      "step": 19620
    },
    {
      "epoch": 5.571955719557195,
      "grad_norm": 12.417707443237305,
      "learning_rate": 9.44311666193585e-05,
      "loss": 0.09806557297706604,
      "step": 19630
    },
    {
      "epoch": 5.574794209480556,
      "grad_norm": 10.28376293182373,
      "learning_rate": 9.442832812943513e-05,
      "loss": 0.1239471435546875,
      "step": 19640
    },
    {
      "epoch": 5.577632699403917,
      "grad_norm": 13.795941352844238,
      "learning_rate": 9.442548963951179e-05,
      "loss": 0.10845887660980225,
      "step": 19650
    },
    {
      "epoch": 5.580471189327278,
      "grad_norm": 19.030147552490234,
      "learning_rate": 9.442265114958843e-05,
      "loss": 0.11543688774108887,
      "step": 19660
    },
    {
      "epoch": 5.583309679250639,
      "grad_norm": 9.811711311340332,
      "learning_rate": 9.441981265966506e-05,
      "loss": 0.11494889259338378,
      "step": 19670
    },
    {
      "epoch": 5.586148169174,
      "grad_norm": 12.787882804870605,
      "learning_rate": 9.44169741697417e-05,
      "loss": 0.0939164400100708,
      "step": 19680
    },
    {
      "epoch": 5.58898665909736,
      "grad_norm": 12.61274528503418,
      "learning_rate": 9.441413567981834e-05,
      "loss": 0.09798465967178345,
      "step": 19690
    },
    {
      "epoch": 5.591825149020721,
      "grad_norm": 16.24980354309082,
      "learning_rate": 9.441129718989498e-05,
      "loss": 0.14927272796630858,
      "step": 19700
    },
    {
      "epoch": 5.594663638944081,
      "grad_norm": 11.760284423828125,
      "learning_rate": 9.440845869997162e-05,
      "loss": 0.12690526247024536,
      "step": 19710
    },
    {
      "epoch": 5.5975021288674425,
      "grad_norm": 20.020917892456055,
      "learning_rate": 9.440562021004826e-05,
      "loss": 0.1077381730079651,
      "step": 19720
    },
    {
      "epoch": 5.600340618790804,
      "grad_norm": 10.436101913452148,
      "learning_rate": 9.44027817201249e-05,
      "loss": 0.1191368818283081,
      "step": 19730
    },
    {
      "epoch": 5.603179108714164,
      "grad_norm": 9.501431465148926,
      "learning_rate": 9.439994323020153e-05,
      "loss": 0.10124939680099487,
      "step": 19740
    },
    {
      "epoch": 5.606017598637525,
      "grad_norm": 13.978364944458008,
      "learning_rate": 9.439710474027818e-05,
      "loss": 0.11254279613494873,
      "step": 19750
    },
    {
      "epoch": 5.608856088560886,
      "grad_norm": 13.982394218444824,
      "learning_rate": 9.439426625035482e-05,
      "loss": 0.10022561550140381,
      "step": 19760
    },
    {
      "epoch": 5.611694578484246,
      "grad_norm": 9.231678009033203,
      "learning_rate": 9.439142776043144e-05,
      "loss": 0.07619082927703857,
      "step": 19770
    },
    {
      "epoch": 5.614533068407607,
      "grad_norm": 10.90194320678711,
      "learning_rate": 9.43885892705081e-05,
      "loss": 0.11572556495666504,
      "step": 19780
    },
    {
      "epoch": 5.617371558330968,
      "grad_norm": 4.109992980957031,
      "learning_rate": 9.438575078058474e-05,
      "loss": 0.09055798649787902,
      "step": 19790
    },
    {
      "epoch": 5.620210048254329,
      "grad_norm": 6.505393028259277,
      "learning_rate": 9.438291229066137e-05,
      "loss": 0.08564833402633668,
      "step": 19800
    },
    {
      "epoch": 5.62304853817769,
      "grad_norm": 11.80860424041748,
      "learning_rate": 9.438007380073801e-05,
      "loss": 0.1683119058609009,
      "step": 19810
    },
    {
      "epoch": 5.62588702810105,
      "grad_norm": 19.342016220092773,
      "learning_rate": 9.437723531081465e-05,
      "loss": 0.13672653436660767,
      "step": 19820
    },
    {
      "epoch": 5.628725518024411,
      "grad_norm": 28.116464614868164,
      "learning_rate": 9.437439682089129e-05,
      "loss": 0.09803901314735412,
      "step": 19830
    },
    {
      "epoch": 5.631564007947771,
      "grad_norm": 9.503965377807617,
      "learning_rate": 9.437155833096792e-05,
      "loss": 0.10152112245559693,
      "step": 19840
    },
    {
      "epoch": 5.6344024978711325,
      "grad_norm": 16.11463165283203,
      "learning_rate": 9.436871984104458e-05,
      "loss": 0.11753343343734741,
      "step": 19850
    },
    {
      "epoch": 5.637240987794494,
      "grad_norm": 10.268158912658691,
      "learning_rate": 9.436588135112122e-05,
      "loss": 0.1271757960319519,
      "step": 19860
    },
    {
      "epoch": 5.640079477717854,
      "grad_norm": 23.734561920166016,
      "learning_rate": 9.436304286119784e-05,
      "loss": 0.12662642002105712,
      "step": 19870
    },
    {
      "epoch": 5.642917967641215,
      "grad_norm": 12.347908973693848,
      "learning_rate": 9.436020437127449e-05,
      "loss": 0.12965919971466064,
      "step": 19880
    },
    {
      "epoch": 5.645756457564576,
      "grad_norm": 22.378528594970703,
      "learning_rate": 9.435736588135113e-05,
      "loss": 0.1164097785949707,
      "step": 19890
    },
    {
      "epoch": 5.648594947487936,
      "grad_norm": 7.571527004241943,
      "learning_rate": 9.435452739142776e-05,
      "loss": 0.09827263355255127,
      "step": 19900
    },
    {
      "epoch": 5.651433437411297,
      "grad_norm": 13.348270416259766,
      "learning_rate": 9.435168890150441e-05,
      "loss": 0.1280358076095581,
      "step": 19910
    },
    {
      "epoch": 5.654271927334658,
      "grad_norm": 10.951181411743164,
      "learning_rate": 9.434885041158105e-05,
      "loss": 0.13780926465988158,
      "step": 19920
    },
    {
      "epoch": 5.657110417258019,
      "grad_norm": 3.0407769680023193,
      "learning_rate": 9.434601192165768e-05,
      "loss": 0.11337422132492066,
      "step": 19930
    },
    {
      "epoch": 5.65994890718138,
      "grad_norm": 10.648685455322266,
      "learning_rate": 9.434317343173432e-05,
      "loss": 0.11053813695907592,
      "step": 19940
    },
    {
      "epoch": 5.66278739710474,
      "grad_norm": 10.185513496398926,
      "learning_rate": 9.434033494181096e-05,
      "loss": 0.1217140793800354,
      "step": 19950
    },
    {
      "epoch": 5.665625887028101,
      "grad_norm": 9.673163414001465,
      "learning_rate": 9.43374964518876e-05,
      "loss": 0.1199615478515625,
      "step": 19960
    },
    {
      "epoch": 5.668464376951462,
      "grad_norm": 17.210432052612305,
      "learning_rate": 9.433465796196423e-05,
      "loss": 0.10232281684875488,
      "step": 19970
    },
    {
      "epoch": 5.671302866874822,
      "grad_norm": 7.972498416900635,
      "learning_rate": 9.433181947204089e-05,
      "loss": 0.1017485499382019,
      "step": 19980
    },
    {
      "epoch": 5.6741413567981835,
      "grad_norm": 25.447080612182617,
      "learning_rate": 9.432898098211753e-05,
      "loss": 0.13863906860351563,
      "step": 19990
    },
    {
      "epoch": 5.676979846721544,
      "grad_norm": 14.742323875427246,
      "learning_rate": 9.432614249219416e-05,
      "loss": 0.1260089874267578,
      "step": 20000
    },
    {
      "epoch": 5.676979846721544,
      "eval_accuracy": 0.9208367775163732,
      "eval_loss": 0.23497049510478973,
      "eval_runtime": 31.2279,
      "eval_samples_per_second": 503.62,
      "eval_steps_per_second": 7.878,
      "step": 20000
    },
    {
      "epoch": 5.679818336644905,
      "grad_norm": 22.31974220275879,
      "learning_rate": 9.43233040022708e-05,
      "loss": 0.09978458285331726,
      "step": 20010
    },
    {
      "epoch": 5.682656826568266,
      "grad_norm": 4.697038650512695,
      "learning_rate": 9.432046551234744e-05,
      "loss": 0.1066719651222229,
      "step": 20020
    },
    {
      "epoch": 5.685495316491626,
      "grad_norm": 7.4572529792785645,
      "learning_rate": 9.431762702242407e-05,
      "loss": 0.10995265245437622,
      "step": 20030
    },
    {
      "epoch": 5.688333806414987,
      "grad_norm": 10.074922561645508,
      "learning_rate": 9.431478853250072e-05,
      "loss": 0.11781272888183594,
      "step": 20040
    },
    {
      "epoch": 5.6911722963383475,
      "grad_norm": 12.172504425048828,
      "learning_rate": 9.431195004257736e-05,
      "loss": 0.12341436147689819,
      "step": 20050
    },
    {
      "epoch": 5.694010786261709,
      "grad_norm": 13.814973831176758,
      "learning_rate": 9.430911155265399e-05,
      "loss": 0.08532713651657105,
      "step": 20060
    },
    {
      "epoch": 5.69684927618507,
      "grad_norm": 8.905369758605957,
      "learning_rate": 9.430627306273063e-05,
      "loss": 0.09130767583847046,
      "step": 20070
    },
    {
      "epoch": 5.69968776610843,
      "grad_norm": 10.224356651306152,
      "learning_rate": 9.430343457280727e-05,
      "loss": 0.1232654333114624,
      "step": 20080
    },
    {
      "epoch": 5.702526256031791,
      "grad_norm": 9.009565353393555,
      "learning_rate": 9.430059608288391e-05,
      "loss": 0.11865953207015992,
      "step": 20090
    },
    {
      "epoch": 5.705364745955152,
      "grad_norm": 8.753589630126953,
      "learning_rate": 9.429775759296054e-05,
      "loss": 0.11695575714111328,
      "step": 20100
    },
    {
      "epoch": 5.708203235878512,
      "grad_norm": 5.8907880783081055,
      "learning_rate": 9.42949191030372e-05,
      "loss": 0.09317396879196167,
      "step": 20110
    },
    {
      "epoch": 5.7110417258018735,
      "grad_norm": 12.783473014831543,
      "learning_rate": 9.429208061311382e-05,
      "loss": 0.14268021583557128,
      "step": 20120
    },
    {
      "epoch": 5.713880215725235,
      "grad_norm": 13.65078067779541,
      "learning_rate": 9.428924212319047e-05,
      "loss": 0.17956950664520263,
      "step": 20130
    },
    {
      "epoch": 5.716718705648595,
      "grad_norm": 13.46239948272705,
      "learning_rate": 9.428640363326711e-05,
      "loss": 0.10398170948028565,
      "step": 20140
    },
    {
      "epoch": 5.719557195571956,
      "grad_norm": 13.398715019226074,
      "learning_rate": 9.428356514334375e-05,
      "loss": 0.08553057909011841,
      "step": 20150
    },
    {
      "epoch": 5.722395685495316,
      "grad_norm": 13.870940208435059,
      "learning_rate": 9.428072665342038e-05,
      "loss": 0.12373956441879272,
      "step": 20160
    },
    {
      "epoch": 5.725234175418677,
      "grad_norm": 7.184823036193848,
      "learning_rate": 9.427788816349702e-05,
      "loss": 0.11054706573486328,
      "step": 20170
    },
    {
      "epoch": 5.728072665342038,
      "grad_norm": 16.883848190307617,
      "learning_rate": 9.427504967357367e-05,
      "loss": 0.11671416759490967,
      "step": 20180
    },
    {
      "epoch": 5.730911155265399,
      "grad_norm": 7.285149574279785,
      "learning_rate": 9.42722111836503e-05,
      "loss": 0.10194882154464721,
      "step": 20190
    },
    {
      "epoch": 5.73374964518876,
      "grad_norm": 17.83062744140625,
      "learning_rate": 9.426937269372694e-05,
      "loss": 0.12979027032852172,
      "step": 20200
    },
    {
      "epoch": 5.73658813511212,
      "grad_norm": 10.121158599853516,
      "learning_rate": 9.426653420380358e-05,
      "loss": 0.11723473072052001,
      "step": 20210
    },
    {
      "epoch": 5.739426625035481,
      "grad_norm": 9.345041275024414,
      "learning_rate": 9.426369571388021e-05,
      "loss": 0.09484395980834961,
      "step": 20220
    },
    {
      "epoch": 5.742265114958842,
      "grad_norm": 16.191144943237305,
      "learning_rate": 9.426085722395685e-05,
      "loss": 0.13823035955429078,
      "step": 20230
    },
    {
      "epoch": 5.745103604882202,
      "grad_norm": 8.47496509552002,
      "learning_rate": 9.425801873403351e-05,
      "loss": 0.09822333455085755,
      "step": 20240
    },
    {
      "epoch": 5.7479420948055635,
      "grad_norm": 9.439685821533203,
      "learning_rate": 9.425518024411014e-05,
      "loss": 0.1134137749671936,
      "step": 20250
    },
    {
      "epoch": 5.750780584728924,
      "grad_norm": 17.277421951293945,
      "learning_rate": 9.425234175418678e-05,
      "loss": 0.10476200580596924,
      "step": 20260
    },
    {
      "epoch": 5.753619074652285,
      "grad_norm": 6.557984828948975,
      "learning_rate": 9.424950326426342e-05,
      "loss": 0.09636909365653992,
      "step": 20270
    },
    {
      "epoch": 5.756457564575646,
      "grad_norm": 7.983254909515381,
      "learning_rate": 9.424666477434006e-05,
      "loss": 0.08869478702545167,
      "step": 20280
    },
    {
      "epoch": 5.759296054499006,
      "grad_norm": 12.20524787902832,
      "learning_rate": 9.424382628441669e-05,
      "loss": 0.1241506814956665,
      "step": 20290
    },
    {
      "epoch": 5.762134544422367,
      "grad_norm": 11.937776565551758,
      "learning_rate": 9.424098779449333e-05,
      "loss": 0.10028040409088135,
      "step": 20300
    },
    {
      "epoch": 5.764973034345728,
      "grad_norm": 11.134846687316895,
      "learning_rate": 9.423814930456998e-05,
      "loss": 0.10375795364379883,
      "step": 20310
    },
    {
      "epoch": 5.767811524269089,
      "grad_norm": 5.081900596618652,
      "learning_rate": 9.423531081464661e-05,
      "loss": 0.08871475458145142,
      "step": 20320
    },
    {
      "epoch": 5.77065001419245,
      "grad_norm": 13.92850399017334,
      "learning_rate": 9.423247232472325e-05,
      "loss": 0.11963483095169067,
      "step": 20330
    },
    {
      "epoch": 5.773488504115811,
      "grad_norm": 19.87070083618164,
      "learning_rate": 9.42296338347999e-05,
      "loss": 0.13282688856124877,
      "step": 20340
    },
    {
      "epoch": 5.776326994039171,
      "grad_norm": 10.71762466430664,
      "learning_rate": 9.422679534487652e-05,
      "loss": 0.11580984592437744,
      "step": 20350
    },
    {
      "epoch": 5.779165483962532,
      "grad_norm": 17.5373592376709,
      "learning_rate": 9.422395685495316e-05,
      "loss": 0.1278702974319458,
      "step": 20360
    },
    {
      "epoch": 5.782003973885892,
      "grad_norm": 8.061185836791992,
      "learning_rate": 9.42211183650298e-05,
      "loss": 0.11274677515029907,
      "step": 20370
    },
    {
      "epoch": 5.7848424638092535,
      "grad_norm": 7.343835830688477,
      "learning_rate": 9.421827987510645e-05,
      "loss": 0.10580188035964966,
      "step": 20380
    },
    {
      "epoch": 5.787680953732615,
      "grad_norm": 7.6651177406311035,
      "learning_rate": 9.421544138518309e-05,
      "loss": 0.11666873693466187,
      "step": 20390
    },
    {
      "epoch": 5.790519443655975,
      "grad_norm": 14.452144622802734,
      "learning_rate": 9.421260289525973e-05,
      "loss": 0.12069786787033081,
      "step": 20400
    },
    {
      "epoch": 5.793357933579336,
      "grad_norm": 16.657556533813477,
      "learning_rate": 9.420976440533637e-05,
      "loss": 0.10045976638793945,
      "step": 20410
    },
    {
      "epoch": 5.796196423502696,
      "grad_norm": 9.178298950195312,
      "learning_rate": 9.4206925915413e-05,
      "loss": 0.12472000122070312,
      "step": 20420
    },
    {
      "epoch": 5.799034913426057,
      "grad_norm": 11.092296600341797,
      "learning_rate": 9.420408742548964e-05,
      "loss": 0.12591469287872314,
      "step": 20430
    },
    {
      "epoch": 5.801873403349418,
      "grad_norm": 7.71895170211792,
      "learning_rate": 9.42012489355663e-05,
      "loss": 0.11758935451507568,
      "step": 20440
    },
    {
      "epoch": 5.804711893272779,
      "grad_norm": 11.795618057250977,
      "learning_rate": 9.419841044564292e-05,
      "loss": 0.10448135137557983,
      "step": 20450
    },
    {
      "epoch": 5.80755038319614,
      "grad_norm": 9.91723918914795,
      "learning_rate": 9.419557195571956e-05,
      "loss": 0.10342755317687988,
      "step": 20460
    },
    {
      "epoch": 5.810388873119501,
      "grad_norm": 22.000638961791992,
      "learning_rate": 9.41927334657962e-05,
      "loss": 0.14892503023147582,
      "step": 20470
    },
    {
      "epoch": 5.813227363042861,
      "grad_norm": 7.675416469573975,
      "learning_rate": 9.418989497587283e-05,
      "loss": 0.09208612442016602,
      "step": 20480
    },
    {
      "epoch": 5.816065852966222,
      "grad_norm": 16.408016204833984,
      "learning_rate": 9.418705648594947e-05,
      "loss": 0.08327261209487916,
      "step": 20490
    },
    {
      "epoch": 5.818904342889582,
      "grad_norm": 7.683635711669922,
      "learning_rate": 9.418421799602612e-05,
      "loss": 0.1132471203804016,
      "step": 20500
    },
    {
      "epoch": 5.818904342889582,
      "eval_accuracy": 0.9238888535639347,
      "eval_loss": 0.22741714119911194,
      "eval_runtime": 31.0664,
      "eval_samples_per_second": 506.237,
      "eval_steps_per_second": 7.919,
      "step": 20500
    },
    {
      "epoch": 5.8217428328129435,
      "grad_norm": 8.033111572265625,
      "learning_rate": 9.41816633550951e-05,
      "loss": 0.10498391389846802,
      "step": 20510
    },
    {
      "epoch": 5.824581322736305,
      "grad_norm": 16.93895149230957,
      "learning_rate": 9.417882486517174e-05,
      "loss": 0.13192824125289918,
      "step": 20520
    },
    {
      "epoch": 5.827419812659665,
      "grad_norm": 17.369476318359375,
      "learning_rate": 9.417598637524836e-05,
      "loss": 0.15906789302825927,
      "step": 20530
    },
    {
      "epoch": 5.830258302583026,
      "grad_norm": 25.607196807861328,
      "learning_rate": 9.4173147885325e-05,
      "loss": 0.08330153226852417,
      "step": 20540
    },
    {
      "epoch": 5.833096792506387,
      "grad_norm": 10.629730224609375,
      "learning_rate": 9.417030939540165e-05,
      "loss": 0.13406455516815186,
      "step": 20550
    },
    {
      "epoch": 5.835935282429747,
      "grad_norm": 8.932452201843262,
      "learning_rate": 9.416747090547829e-05,
      "loss": 0.11395784616470336,
      "step": 20560
    },
    {
      "epoch": 5.838773772353108,
      "grad_norm": 14.644490242004395,
      "learning_rate": 9.416463241555493e-05,
      "loss": 0.13847874402999877,
      "step": 20570
    },
    {
      "epoch": 5.841612262276469,
      "grad_norm": 16.633623123168945,
      "learning_rate": 9.416179392563157e-05,
      "loss": 0.1258102297782898,
      "step": 20580
    },
    {
      "epoch": 5.84445075219983,
      "grad_norm": 14.985760688781738,
      "learning_rate": 9.415895543570821e-05,
      "loss": 0.12671819925308228,
      "step": 20590
    },
    {
      "epoch": 5.847289242123191,
      "grad_norm": 7.696166515350342,
      "learning_rate": 9.415611694578484e-05,
      "loss": 0.14628607034683228,
      "step": 20600
    },
    {
      "epoch": 5.850127732046551,
      "grad_norm": 15.762514114379883,
      "learning_rate": 9.415327845586148e-05,
      "loss": 0.11489802598953247,
      "step": 20610
    },
    {
      "epoch": 5.852966221969912,
      "grad_norm": 12.863702774047852,
      "learning_rate": 9.415043996593814e-05,
      "loss": 0.09319427013397216,
      "step": 20620
    },
    {
      "epoch": 5.855804711893272,
      "grad_norm": 13.40141773223877,
      "learning_rate": 9.414760147601476e-05,
      "loss": 0.13039968013763428,
      "step": 20630
    },
    {
      "epoch": 5.8586432018166335,
      "grad_norm": 6.219179153442383,
      "learning_rate": 9.41447629860914e-05,
      "loss": 0.08061777353286743,
      "step": 20640
    },
    {
      "epoch": 5.861481691739995,
      "grad_norm": 23.412900924682617,
      "learning_rate": 9.414192449616805e-05,
      "loss": 0.11036337614059448,
      "step": 20650
    },
    {
      "epoch": 5.864320181663355,
      "grad_norm": 11.657022476196289,
      "learning_rate": 9.413908600624468e-05,
      "loss": 0.09171515107154846,
      "step": 20660
    },
    {
      "epoch": 5.867158671586716,
      "grad_norm": 10.708574295043945,
      "learning_rate": 9.413624751632132e-05,
      "loss": 0.1260165572166443,
      "step": 20670
    },
    {
      "epoch": 5.869997161510077,
      "grad_norm": 4.3276777267456055,
      "learning_rate": 9.413340902639796e-05,
      "loss": 0.10439838171005249,
      "step": 20680
    },
    {
      "epoch": 5.872835651433437,
      "grad_norm": 13.661375045776367,
      "learning_rate": 9.41305705364746e-05,
      "loss": 0.11782879829406738,
      "step": 20690
    },
    {
      "epoch": 5.875674141356798,
      "grad_norm": 10.946717262268066,
      "learning_rate": 9.412773204655124e-05,
      "loss": 0.11104187965393067,
      "step": 20700
    },
    {
      "epoch": 5.878512631280159,
      "grad_norm": 14.004988670349121,
      "learning_rate": 9.412489355662788e-05,
      "loss": 0.0921678900718689,
      "step": 20710
    },
    {
      "epoch": 5.88135112120352,
      "grad_norm": 7.5447773933410645,
      "learning_rate": 9.412205506670452e-05,
      "loss": 0.13092324733734131,
      "step": 20720
    },
    {
      "epoch": 5.884189611126881,
      "grad_norm": 8.072649955749512,
      "learning_rate": 9.411921657678115e-05,
      "loss": 0.1444801926612854,
      "step": 20730
    },
    {
      "epoch": 5.887028101050241,
      "grad_norm": 16.72834587097168,
      "learning_rate": 9.411637808685779e-05,
      "loss": 0.13325690031051635,
      "step": 20740
    },
    {
      "epoch": 5.889866590973602,
      "grad_norm": 21.628686904907227,
      "learning_rate": 9.411353959693443e-05,
      "loss": 0.12417876720428467,
      "step": 20750
    },
    {
      "epoch": 5.892705080896963,
      "grad_norm": 10.818861961364746,
      "learning_rate": 9.411070110701108e-05,
      "loss": 0.12724164724349976,
      "step": 20760
    },
    {
      "epoch": 5.895543570820323,
      "grad_norm": 8.504688262939453,
      "learning_rate": 9.410786261708772e-05,
      "loss": 0.10596181154251098,
      "step": 20770
    },
    {
      "epoch": 5.8983820607436845,
      "grad_norm": 14.816200256347656,
      "learning_rate": 9.410502412716436e-05,
      "loss": 0.094748854637146,
      "step": 20780
    },
    {
      "epoch": 5.901220550667045,
      "grad_norm": 18.645090103149414,
      "learning_rate": 9.410218563724099e-05,
      "loss": 0.15787179470062257,
      "step": 20790
    },
    {
      "epoch": 5.904059040590406,
      "grad_norm": 7.746403694152832,
      "learning_rate": 9.409934714731763e-05,
      "loss": 0.11731476783752441,
      "step": 20800
    },
    {
      "epoch": 5.906897530513767,
      "grad_norm": 13.322840690612793,
      "learning_rate": 9.409650865739427e-05,
      "loss": 0.0997210144996643,
      "step": 20810
    },
    {
      "epoch": 5.909736020437127,
      "grad_norm": 9.661260604858398,
      "learning_rate": 9.409367016747091e-05,
      "loss": 0.08642529845237731,
      "step": 20820
    },
    {
      "epoch": 5.912574510360488,
      "grad_norm": 10.458847045898438,
      "learning_rate": 9.409083167754755e-05,
      "loss": 0.12587547302246094,
      "step": 20830
    },
    {
      "epoch": 5.9154130002838485,
      "grad_norm": 14.758914947509766,
      "learning_rate": 9.408799318762419e-05,
      "loss": 0.07983130216598511,
      "step": 20840
    },
    {
      "epoch": 5.91825149020721,
      "grad_norm": 13.798670768737793,
      "learning_rate": 9.408515469770083e-05,
      "loss": 0.10943112373352051,
      "step": 20850
    },
    {
      "epoch": 5.921089980130571,
      "grad_norm": 9.510903358459473,
      "learning_rate": 9.408231620777746e-05,
      "loss": 0.12384045124053955,
      "step": 20860
    },
    {
      "epoch": 5.923928470053931,
      "grad_norm": 24.714670181274414,
      "learning_rate": 9.40794777178541e-05,
      "loss": 0.11439346075057984,
      "step": 20870
    },
    {
      "epoch": 5.926766959977292,
      "grad_norm": 11.630462646484375,
      "learning_rate": 9.407663922793075e-05,
      "loss": 0.10922585725784302,
      "step": 20880
    },
    {
      "epoch": 5.929605449900653,
      "grad_norm": 6.719127178192139,
      "learning_rate": 9.407380073800739e-05,
      "loss": 0.10444743633270263,
      "step": 20890
    },
    {
      "epoch": 5.932443939824013,
      "grad_norm": 12.36545467376709,
      "learning_rate": 9.407096224808403e-05,
      "loss": 0.12218999862670898,
      "step": 20900
    },
    {
      "epoch": 5.9352824297473745,
      "grad_norm": 11.014594078063965,
      "learning_rate": 9.406812375816067e-05,
      "loss": 0.08795750737190247,
      "step": 20910
    },
    {
      "epoch": 5.938120919670736,
      "grad_norm": 5.8085174560546875,
      "learning_rate": 9.40652852682373e-05,
      "loss": 0.09594220519065857,
      "step": 20920
    },
    {
      "epoch": 5.940959409594096,
      "grad_norm": 17.007211685180664,
      "learning_rate": 9.406244677831394e-05,
      "loss": 0.08541787266731263,
      "step": 20930
    },
    {
      "epoch": 5.943797899517457,
      "grad_norm": 12.685742378234863,
      "learning_rate": 9.405960828839058e-05,
      "loss": 0.10849156379699706,
      "step": 20940
    },
    {
      "epoch": 5.946636389440817,
      "grad_norm": 6.9238715171813965,
      "learning_rate": 9.405676979846722e-05,
      "loss": 0.12375425100326538,
      "step": 20950
    },
    {
      "epoch": 5.949474879364178,
      "grad_norm": 4.230216026306152,
      "learning_rate": 9.405393130854386e-05,
      "loss": 0.08689563870429992,
      "step": 20960
    },
    {
      "epoch": 5.952313369287539,
      "grad_norm": 10.104366302490234,
      "learning_rate": 9.40510928186205e-05,
      "loss": 0.13967742919921874,
      "step": 20970
    },
    {
      "epoch": 5.9551518592109,
      "grad_norm": 11.409494400024414,
      "learning_rate": 9.404825432869713e-05,
      "loss": 0.1268508553504944,
      "step": 20980
    },
    {
      "epoch": 5.957990349134261,
      "grad_norm": 8.809844017028809,
      "learning_rate": 9.404541583877377e-05,
      "loss": 0.11903468370437623,
      "step": 20990
    },
    {
      "epoch": 5.960828839057621,
      "grad_norm": 8.836562156677246,
      "learning_rate": 9.404257734885041e-05,
      "loss": 0.09153556823730469,
      "step": 21000
    },
    {
      "epoch": 5.960828839057621,
      "eval_accuracy": 0.9313282889298659,
      "eval_loss": 0.2052992433309555,
      "eval_runtime": 30.996,
      "eval_samples_per_second": 507.389,
      "eval_steps_per_second": 7.937,
      "step": 21000
    },
    {
      "epoch": 5.963667328980982,
      "grad_norm": 12.423620223999023,
      "learning_rate": 9.403973885892706e-05,
      "loss": 0.11621305942535401,
      "step": 21010
    },
    {
      "epoch": 5.966505818904343,
      "grad_norm": 1.8047871589660645,
      "learning_rate": 9.40369003690037e-05,
      "loss": 0.09838481545448304,
      "step": 21020
    },
    {
      "epoch": 5.969344308827703,
      "grad_norm": 13.596325874328613,
      "learning_rate": 9.403406187908034e-05,
      "loss": 0.12000309228897095,
      "step": 21030
    },
    {
      "epoch": 5.9721827987510645,
      "grad_norm": 15.063260078430176,
      "learning_rate": 9.403122338915698e-05,
      "loss": 0.16305012702941896,
      "step": 21040
    },
    {
      "epoch": 5.975021288674425,
      "grad_norm": 9.36721420288086,
      "learning_rate": 9.402838489923361e-05,
      "loss": 0.1107431173324585,
      "step": 21050
    },
    {
      "epoch": 5.977859778597786,
      "grad_norm": 16.601165771484375,
      "learning_rate": 9.402554640931025e-05,
      "loss": 0.13110706806182862,
      "step": 21060
    },
    {
      "epoch": 5.980698268521147,
      "grad_norm": 16.701623916625977,
      "learning_rate": 9.402270791938689e-05,
      "loss": 0.14415010213851928,
      "step": 21070
    },
    {
      "epoch": 5.983536758444507,
      "grad_norm": 15.612679481506348,
      "learning_rate": 9.401986942946352e-05,
      "loss": 0.12169110774993896,
      "step": 21080
    },
    {
      "epoch": 5.986375248367868,
      "grad_norm": 12.671293258666992,
      "learning_rate": 9.401703093954017e-05,
      "loss": 0.14987602233886718,
      "step": 21090
    },
    {
      "epoch": 5.989213738291229,
      "grad_norm": 17.24456214904785,
      "learning_rate": 9.401419244961681e-05,
      "loss": 0.15189071893692016,
      "step": 21100
    },
    {
      "epoch": 5.99205222821459,
      "grad_norm": 21.735652923583984,
      "learning_rate": 9.401135395969344e-05,
      "loss": 0.14007269144058226,
      "step": 21110
    },
    {
      "epoch": 5.994890718137951,
      "grad_norm": 20.80341339111328,
      "learning_rate": 9.400851546977008e-05,
      "loss": 0.10591450929641724,
      "step": 21120
    },
    {
      "epoch": 5.997729208061312,
      "grad_norm": 14.216853141784668,
      "learning_rate": 9.400567697984673e-05,
      "loss": 0.09699013829231262,
      "step": 21130
    },
    {
      "epoch": 6.000567697984672,
      "grad_norm": 10.743067741394043,
      "learning_rate": 9.400283848992337e-05,
      "loss": 0.08979021310806275,
      "step": 21140
    },
    {
      "epoch": 6.003406187908033,
      "grad_norm": 3.1110355854034424,
      "learning_rate": 9.4e-05,
      "loss": 0.07359349727630615,
      "step": 21150
    },
    {
      "epoch": 6.006244677831393,
      "grad_norm": 11.48859977722168,
      "learning_rate": 9.399716151007665e-05,
      "loss": 0.08522689342498779,
      "step": 21160
    },
    {
      "epoch": 6.0090831677547545,
      "grad_norm": 6.143226623535156,
      "learning_rate": 9.399432302015329e-05,
      "loss": 0.09884411692619324,
      "step": 21170
    },
    {
      "epoch": 6.011921657678116,
      "grad_norm": 11.420652389526367,
      "learning_rate": 9.399148453022992e-05,
      "loss": 0.09911879301071166,
      "step": 21180
    },
    {
      "epoch": 6.014760147601476,
      "grad_norm": 7.774734973907471,
      "learning_rate": 9.398864604030656e-05,
      "loss": 0.10043747425079345,
      "step": 21190
    },
    {
      "epoch": 6.017598637524837,
      "grad_norm": 4.420595169067383,
      "learning_rate": 9.39858075503832e-05,
      "loss": 0.07095736861228943,
      "step": 21200
    },
    {
      "epoch": 6.020437127448197,
      "grad_norm": 11.891218185424805,
      "learning_rate": 9.398296906045983e-05,
      "loss": 0.07457776069641113,
      "step": 21210
    },
    {
      "epoch": 6.023275617371558,
      "grad_norm": 11.142216682434082,
      "learning_rate": 9.398013057053648e-05,
      "loss": 0.072478985786438,
      "step": 21220
    },
    {
      "epoch": 6.026114107294919,
      "grad_norm": 10.591689109802246,
      "learning_rate": 9.397729208061313e-05,
      "loss": 0.10340570211410523,
      "step": 21230
    },
    {
      "epoch": 6.02895259721828,
      "grad_norm": 15.661985397338867,
      "learning_rate": 9.397445359068975e-05,
      "loss": 0.10229082107543945,
      "step": 21240
    },
    {
      "epoch": 6.031791087141641,
      "grad_norm": 6.8917317390441895,
      "learning_rate": 9.39716151007664e-05,
      "loss": 0.11099512577056884,
      "step": 21250
    },
    {
      "epoch": 6.034629577065002,
      "grad_norm": 7.677479267120361,
      "learning_rate": 9.396877661084304e-05,
      "loss": 0.07190749645233155,
      "step": 21260
    },
    {
      "epoch": 6.037468066988362,
      "grad_norm": 10.08597469329834,
      "learning_rate": 9.396593812091968e-05,
      "loss": 0.08615765571594239,
      "step": 21270
    },
    {
      "epoch": 6.040306556911723,
      "grad_norm": 14.094669342041016,
      "learning_rate": 9.39630996309963e-05,
      "loss": 0.10557165145874023,
      "step": 21280
    },
    {
      "epoch": 6.043145046835083,
      "grad_norm": 11.102452278137207,
      "learning_rate": 9.396026114107296e-05,
      "loss": 0.08731478452682495,
      "step": 21290
    },
    {
      "epoch": 6.0459835367584445,
      "grad_norm": 18.687788009643555,
      "learning_rate": 9.39574226511496e-05,
      "loss": 0.11136163473129272,
      "step": 21300
    },
    {
      "epoch": 6.048822026681806,
      "grad_norm": 12.847024917602539,
      "learning_rate": 9.395458416122623e-05,
      "loss": 0.09557934403419495,
      "step": 21310
    },
    {
      "epoch": 6.051660516605166,
      "grad_norm": 8.790975570678711,
      "learning_rate": 9.395174567130287e-05,
      "loss": 0.10499650239944458,
      "step": 21320
    },
    {
      "epoch": 6.054499006528527,
      "grad_norm": 11.057319641113281,
      "learning_rate": 9.394890718137951e-05,
      "loss": 0.09857409000396729,
      "step": 21330
    },
    {
      "epoch": 6.057337496451888,
      "grad_norm": 18.362462997436523,
      "learning_rate": 9.394606869145614e-05,
      "loss": 0.09969000816345215,
      "step": 21340
    },
    {
      "epoch": 6.060175986375248,
      "grad_norm": 13.130828857421875,
      "learning_rate": 9.394323020153278e-05,
      "loss": 0.10471675395965577,
      "step": 21350
    },
    {
      "epoch": 6.063014476298609,
      "grad_norm": 9.789393424987793,
      "learning_rate": 9.394039171160944e-05,
      "loss": 0.08259748220443726,
      "step": 21360
    },
    {
      "epoch": 6.06585296622197,
      "grad_norm": 11.308144569396973,
      "learning_rate": 9.393755322168606e-05,
      "loss": 0.11735329627990723,
      "step": 21370
    },
    {
      "epoch": 6.068691456145331,
      "grad_norm": 5.1346025466918945,
      "learning_rate": 9.39347147317627e-05,
      "loss": 0.08542198538780213,
      "step": 21380
    },
    {
      "epoch": 6.071529946068692,
      "grad_norm": 9.559159278869629,
      "learning_rate": 9.393187624183935e-05,
      "loss": 0.0788351058959961,
      "step": 21390
    },
    {
      "epoch": 6.074368435992052,
      "grad_norm": 9.88121509552002,
      "learning_rate": 9.392903775191599e-05,
      "loss": 0.0921436071395874,
      "step": 21400
    },
    {
      "epoch": 6.077206925915413,
      "grad_norm": 18.608478546142578,
      "learning_rate": 9.392619926199262e-05,
      "loss": 0.0845526933670044,
      "step": 21410
    },
    {
      "epoch": 6.080045415838773,
      "grad_norm": 4.436546325683594,
      "learning_rate": 9.392336077206927e-05,
      "loss": 0.07260103225708008,
      "step": 21420
    },
    {
      "epoch": 6.0828839057621344,
      "grad_norm": 9.707710266113281,
      "learning_rate": 9.392052228214591e-05,
      "loss": 0.08143643736839294,
      "step": 21430
    },
    {
      "epoch": 6.085722395685496,
      "grad_norm": 7.061699390411377,
      "learning_rate": 9.391768379222254e-05,
      "loss": 0.09043810963630676,
      "step": 21440
    },
    {
      "epoch": 6.088560885608856,
      "grad_norm": 13.426015853881836,
      "learning_rate": 9.391484530229918e-05,
      "loss": 0.060395896434783936,
      "step": 21450
    },
    {
      "epoch": 6.091399375532217,
      "grad_norm": 15.57006549835205,
      "learning_rate": 9.391200681237582e-05,
      "loss": 0.11533644199371337,
      "step": 21460
    },
    {
      "epoch": 6.094237865455578,
      "grad_norm": 5.301523208618164,
      "learning_rate": 9.390916832245245e-05,
      "loss": 0.09352814555168151,
      "step": 21470
    },
    {
      "epoch": 6.097076355378938,
      "grad_norm": 14.802569389343262,
      "learning_rate": 9.390632983252909e-05,
      "loss": 0.11143943071365356,
      "step": 21480
    },
    {
      "epoch": 6.099914845302299,
      "grad_norm": 10.134464263916016,
      "learning_rate": 9.390349134260575e-05,
      "loss": 0.09743857979774476,
      "step": 21490
    },
    {
      "epoch": 6.1027533352256595,
      "grad_norm": 10.993227005004883,
      "learning_rate": 9.390065285268237e-05,
      "loss": 0.10054000616073608,
      "step": 21500
    },
    {
      "epoch": 6.1027533352256595,
      "eval_accuracy": 0.9324728174477014,
      "eval_loss": 0.20066311955451965,
      "eval_runtime": 31.6206,
      "eval_samples_per_second": 497.366,
      "eval_steps_per_second": 7.78,
      "step": 21500
    },
    {
      "epoch": 6.105591825149021,
      "grad_norm": 13.049242973327637,
      "learning_rate": 9.389781436275902e-05,
      "loss": 0.09531475901603699,
      "step": 21510
    },
    {
      "epoch": 6.108430315072382,
      "grad_norm": 10.597549438476562,
      "learning_rate": 9.389497587283566e-05,
      "loss": 0.08681356906890869,
      "step": 21520
    },
    {
      "epoch": 6.111268804995742,
      "grad_norm": 13.929612159729004,
      "learning_rate": 9.38921373829123e-05,
      "loss": 0.09679096937179565,
      "step": 21530
    },
    {
      "epoch": 6.114107294919103,
      "grad_norm": 9.482136726379395,
      "learning_rate": 9.388929889298893e-05,
      "loss": 0.08291661143302917,
      "step": 21540
    },
    {
      "epoch": 6.116945784842464,
      "grad_norm": 10.108871459960938,
      "learning_rate": 9.388646040306557e-05,
      "loss": 0.12714382410049438,
      "step": 21550
    },
    {
      "epoch": 6.119784274765824,
      "grad_norm": 15.778446197509766,
      "learning_rate": 9.388362191314222e-05,
      "loss": 0.10163508653640747,
      "step": 21560
    },
    {
      "epoch": 6.1226227646891855,
      "grad_norm": 18.318805694580078,
      "learning_rate": 9.388078342321885e-05,
      "loss": 0.09296117424964905,
      "step": 21570
    },
    {
      "epoch": 6.125461254612546,
      "grad_norm": 5.008502960205078,
      "learning_rate": 9.387794493329549e-05,
      "loss": 0.06724562048912049,
      "step": 21580
    },
    {
      "epoch": 6.128299744535907,
      "grad_norm": 10.594789505004883,
      "learning_rate": 9.387510644337213e-05,
      "loss": 0.084965181350708,
      "step": 21590
    },
    {
      "epoch": 6.131138234459268,
      "grad_norm": 8.914656639099121,
      "learning_rate": 9.387226795344876e-05,
      "loss": 0.06451302170753478,
      "step": 21600
    },
    {
      "epoch": 6.133976724382628,
      "grad_norm": 5.62416934967041,
      "learning_rate": 9.38694294635254e-05,
      "loss": 0.09203835725784301,
      "step": 21610
    },
    {
      "epoch": 6.136815214305989,
      "grad_norm": 13.042671203613281,
      "learning_rate": 9.386659097360206e-05,
      "loss": 0.0830959439277649,
      "step": 21620
    },
    {
      "epoch": 6.1396537042293495,
      "grad_norm": 15.792468070983887,
      "learning_rate": 9.386375248367869e-05,
      "loss": 0.1132962703704834,
      "step": 21630
    },
    {
      "epoch": 6.142492194152711,
      "grad_norm": 12.245312690734863,
      "learning_rate": 9.386091399375533e-05,
      "loss": 0.11983513832092285,
      "step": 21640
    },
    {
      "epoch": 6.145330684076072,
      "grad_norm": 11.293359756469727,
      "learning_rate": 9.385807550383197e-05,
      "loss": 0.09297634959220887,
      "step": 21650
    },
    {
      "epoch": 6.148169173999432,
      "grad_norm": 7.058413028717041,
      "learning_rate": 9.385523701390861e-05,
      "loss": 0.0682492196559906,
      "step": 21660
    },
    {
      "epoch": 6.151007663922793,
      "grad_norm": 21.44993019104004,
      "learning_rate": 9.385239852398524e-05,
      "loss": 0.1442980408668518,
      "step": 21670
    },
    {
      "epoch": 6.153846153846154,
      "grad_norm": 7.716861724853516,
      "learning_rate": 9.384956003406188e-05,
      "loss": 0.09743369221687317,
      "step": 21680
    },
    {
      "epoch": 6.156684643769514,
      "grad_norm": 8.692341804504395,
      "learning_rate": 9.384672154413853e-05,
      "loss": 0.08535637259483338,
      "step": 21690
    },
    {
      "epoch": 6.1595231336928755,
      "grad_norm": 24.293867111206055,
      "learning_rate": 9.384388305421516e-05,
      "loss": 0.116846764087677,
      "step": 21700
    },
    {
      "epoch": 6.162361623616236,
      "grad_norm": 7.000239849090576,
      "learning_rate": 9.38410445642918e-05,
      "loss": 0.10343834161758422,
      "step": 21710
    },
    {
      "epoch": 6.165200113539597,
      "grad_norm": 14.319605827331543,
      "learning_rate": 9.383820607436844e-05,
      "loss": 0.09635480642318725,
      "step": 21720
    },
    {
      "epoch": 6.168038603462958,
      "grad_norm": 12.32835865020752,
      "learning_rate": 9.383536758444507e-05,
      "loss": 0.10151472091674804,
      "step": 21730
    },
    {
      "epoch": 6.170877093386318,
      "grad_norm": 8.093631744384766,
      "learning_rate": 9.383252909452171e-05,
      "loss": 0.08099229335784912,
      "step": 21740
    },
    {
      "epoch": 6.173715583309679,
      "grad_norm": 3.1243510246276855,
      "learning_rate": 9.382969060459837e-05,
      "loss": 0.07208704352378845,
      "step": 21750
    },
    {
      "epoch": 6.17655407323304,
      "grad_norm": 12.66645336151123,
      "learning_rate": 9.3826852114675e-05,
      "loss": 0.0819431185722351,
      "step": 21760
    },
    {
      "epoch": 6.179392563156401,
      "grad_norm": 20.28594970703125,
      "learning_rate": 9.382401362475164e-05,
      "loss": 0.08638877868652343,
      "step": 21770
    },
    {
      "epoch": 6.182231053079762,
      "grad_norm": 11.864381790161133,
      "learning_rate": 9.382117513482828e-05,
      "loss": 0.12183315753936767,
      "step": 21780
    },
    {
      "epoch": 6.185069543003122,
      "grad_norm": 10.574756622314453,
      "learning_rate": 9.381833664490492e-05,
      "loss": 0.11219758987426758,
      "step": 21790
    },
    {
      "epoch": 6.187908032926483,
      "grad_norm": 9.806113243103027,
      "learning_rate": 9.381549815498155e-05,
      "loss": 0.07526425123214722,
      "step": 21800
    },
    {
      "epoch": 6.190746522849844,
      "grad_norm": 12.798103332519531,
      "learning_rate": 9.381265966505819e-05,
      "loss": 0.07418632507324219,
      "step": 21810
    },
    {
      "epoch": 6.193585012773204,
      "grad_norm": 12.292802810668945,
      "learning_rate": 9.380982117513483e-05,
      "loss": 0.08453023433685303,
      "step": 21820
    },
    {
      "epoch": 6.1964235026965655,
      "grad_norm": 17.0206356048584,
      "learning_rate": 9.380698268521147e-05,
      "loss": 0.09070110321044922,
      "step": 21830
    },
    {
      "epoch": 6.199261992619927,
      "grad_norm": 6.9051079750061035,
      "learning_rate": 9.380414419528811e-05,
      "loss": 0.09951165318489075,
      "step": 21840
    },
    {
      "epoch": 6.202100482543287,
      "grad_norm": 8.99085521697998,
      "learning_rate": 9.380130570536476e-05,
      "loss": 0.11273281574249268,
      "step": 21850
    },
    {
      "epoch": 6.204938972466648,
      "grad_norm": 10.897320747375488,
      "learning_rate": 9.379846721544138e-05,
      "loss": 0.08586341142654419,
      "step": 21860
    },
    {
      "epoch": 6.207777462390008,
      "grad_norm": 7.794835090637207,
      "learning_rate": 9.379562872551802e-05,
      "loss": 0.07840330600738525,
      "step": 21870
    },
    {
      "epoch": 6.210615952313369,
      "grad_norm": 9.387540817260742,
      "learning_rate": 9.379279023559467e-05,
      "loss": 0.09344335794448852,
      "step": 21880
    },
    {
      "epoch": 6.21345444223673,
      "grad_norm": 9.73838996887207,
      "learning_rate": 9.378995174567131e-05,
      "loss": 0.08935476541519165,
      "step": 21890
    },
    {
      "epoch": 6.216292932160091,
      "grad_norm": 12.763219833374023,
      "learning_rate": 9.378711325574795e-05,
      "loss": 0.0860913872718811,
      "step": 21900
    },
    {
      "epoch": 6.219131422083452,
      "grad_norm": 6.4986443519592285,
      "learning_rate": 9.378427476582459e-05,
      "loss": 0.11419270038604737,
      "step": 21910
    },
    {
      "epoch": 6.221969912006813,
      "grad_norm": 9.286530494689941,
      "learning_rate": 9.378143627590122e-05,
      "loss": 0.10870224237442017,
      "step": 21920
    },
    {
      "epoch": 6.224808401930173,
      "grad_norm": 5.983080863952637,
      "learning_rate": 9.377859778597786e-05,
      "loss": 0.08914061188697815,
      "step": 21930
    },
    {
      "epoch": 6.227646891853534,
      "grad_norm": 11.714787483215332,
      "learning_rate": 9.37757592960545e-05,
      "loss": 0.08509978652000427,
      "step": 21940
    },
    {
      "epoch": 6.230485381776894,
      "grad_norm": 20.62018585205078,
      "learning_rate": 9.377292080613114e-05,
      "loss": 0.11447678804397583,
      "step": 21950
    },
    {
      "epoch": 6.2333238717002555,
      "grad_norm": 12.139809608459473,
      "learning_rate": 9.377008231620778e-05,
      "loss": 0.10462168455123902,
      "step": 21960
    },
    {
      "epoch": 6.236162361623617,
      "grad_norm": 12.046883583068848,
      "learning_rate": 9.376724382628442e-05,
      "loss": 0.07254236936569214,
      "step": 21970
    },
    {
      "epoch": 6.239000851546977,
      "grad_norm": 25.686941146850586,
      "learning_rate": 9.376440533636107e-05,
      "loss": 0.08625777959823608,
      "step": 21980
    },
    {
      "epoch": 6.241839341470338,
      "grad_norm": 19.759035110473633,
      "learning_rate": 9.37615668464377e-05,
      "loss": 0.13934781551361083,
      "step": 21990
    },
    {
      "epoch": 6.244677831393698,
      "grad_norm": 8.881654739379883,
      "learning_rate": 9.375872835651434e-05,
      "loss": 0.0812686800956726,
      "step": 22000
    },
    {
      "epoch": 6.244677831393698,
      "eval_accuracy": 0.9270045145291537,
      "eval_loss": 0.21792440116405487,
      "eval_runtime": 31.0504,
      "eval_samples_per_second": 506.499,
      "eval_steps_per_second": 7.923,
      "step": 22000
    },
    {
      "epoch": 6.247516321317059,
      "grad_norm": 11.264602661132812,
      "learning_rate": 9.375588986659098e-05,
      "loss": 0.10338951349258423,
      "step": 22010
    },
    {
      "epoch": 6.25035481124042,
      "grad_norm": 15.521607398986816,
      "learning_rate": 9.375305137666762e-05,
      "loss": 0.09795397520065308,
      "step": 22020
    },
    {
      "epoch": 6.253193301163781,
      "grad_norm": 11.035134315490723,
      "learning_rate": 9.375021288674426e-05,
      "loss": 0.10320013761520386,
      "step": 22030
    },
    {
      "epoch": 6.256031791087142,
      "grad_norm": 16.9571590423584,
      "learning_rate": 9.37473743968209e-05,
      "loss": 0.10504947900772095,
      "step": 22040
    },
    {
      "epoch": 6.258870281010503,
      "grad_norm": 13.23913860321045,
      "learning_rate": 9.374453590689753e-05,
      "loss": 0.11344741582870484,
      "step": 22050
    },
    {
      "epoch": 6.261708770933863,
      "grad_norm": 9.76651668548584,
      "learning_rate": 9.374169741697417e-05,
      "loss": 0.08526365160942077,
      "step": 22060
    },
    {
      "epoch": 6.264547260857224,
      "grad_norm": 17.147823333740234,
      "learning_rate": 9.373885892705081e-05,
      "loss": 0.0857745885848999,
      "step": 22070
    },
    {
      "epoch": 6.267385750780584,
      "grad_norm": 11.726737976074219,
      "learning_rate": 9.373602043712745e-05,
      "loss": 0.07491317987442017,
      "step": 22080
    },
    {
      "epoch": 6.2702242407039455,
      "grad_norm": 5.284913063049316,
      "learning_rate": 9.37331819472041e-05,
      "loss": 0.05360146760940552,
      "step": 22090
    },
    {
      "epoch": 6.273062730627307,
      "grad_norm": 9.205575942993164,
      "learning_rate": 9.373034345728074e-05,
      "loss": 0.0956966757774353,
      "step": 22100
    },
    {
      "epoch": 6.275901220550667,
      "grad_norm": 18.47577476501465,
      "learning_rate": 9.372750496735738e-05,
      "loss": 0.0897540032863617,
      "step": 22110
    },
    {
      "epoch": 6.278739710474028,
      "grad_norm": 13.804877281188965,
      "learning_rate": 9.3724666477434e-05,
      "loss": 0.06971631646156311,
      "step": 22120
    },
    {
      "epoch": 6.281578200397389,
      "grad_norm": 13.640564918518066,
      "learning_rate": 9.372182798751065e-05,
      "loss": 0.09559376239776611,
      "step": 22130
    },
    {
      "epoch": 6.284416690320749,
      "grad_norm": 7.796805381774902,
      "learning_rate": 9.371898949758729e-05,
      "loss": 0.087092125415802,
      "step": 22140
    },
    {
      "epoch": 6.28725518024411,
      "grad_norm": 17.170866012573242,
      "learning_rate": 9.371615100766393e-05,
      "loss": 0.1096609354019165,
      "step": 22150
    },
    {
      "epoch": 6.290093670167471,
      "grad_norm": 6.7061309814453125,
      "learning_rate": 9.371331251774057e-05,
      "loss": 0.0764557421207428,
      "step": 22160
    },
    {
      "epoch": 6.292932160090832,
      "grad_norm": 8.954500198364258,
      "learning_rate": 9.371047402781721e-05,
      "loss": 0.0790553092956543,
      "step": 22170
    },
    {
      "epoch": 6.295770650014193,
      "grad_norm": 10.993725776672363,
      "learning_rate": 9.370763553789384e-05,
      "loss": 0.09968703389167785,
      "step": 22180
    },
    {
      "epoch": 6.298609139937553,
      "grad_norm": 2.6468212604522705,
      "learning_rate": 9.370479704797048e-05,
      "loss": 0.08798613548278808,
      "step": 22190
    },
    {
      "epoch": 6.301447629860914,
      "grad_norm": 12.747579574584961,
      "learning_rate": 9.370195855804712e-05,
      "loss": 0.09578362703323365,
      "step": 22200
    },
    {
      "epoch": 6.304286119784274,
      "grad_norm": 15.204248428344727,
      "learning_rate": 9.369912006812376e-05,
      "loss": 0.1036005973815918,
      "step": 22210
    },
    {
      "epoch": 6.307124609707635,
      "grad_norm": 18.201011657714844,
      "learning_rate": 9.36962815782004e-05,
      "loss": 0.11049052476882934,
      "step": 22220
    },
    {
      "epoch": 6.3099630996309966,
      "grad_norm": 6.651350975036621,
      "learning_rate": 9.369344308827705e-05,
      "loss": 0.058932000398635866,
      "step": 22230
    },
    {
      "epoch": 6.312801589554357,
      "grad_norm": 13.46920108795166,
      "learning_rate": 9.369060459835369e-05,
      "loss": 0.09388837814331055,
      "step": 22240
    },
    {
      "epoch": 6.315640079477718,
      "grad_norm": 6.388342380523682,
      "learning_rate": 9.368776610843032e-05,
      "loss": 0.08493118286132813,
      "step": 22250
    },
    {
      "epoch": 6.318478569401079,
      "grad_norm": 10.680513381958008,
      "learning_rate": 9.368492761850696e-05,
      "loss": 0.11544784307479858,
      "step": 22260
    },
    {
      "epoch": 6.321317059324439,
      "grad_norm": 12.826972961425781,
      "learning_rate": 9.36820891285836e-05,
      "loss": 0.07117531895637512,
      "step": 22270
    },
    {
      "epoch": 6.3241555492478,
      "grad_norm": 11.348532676696777,
      "learning_rate": 9.367925063866023e-05,
      "loss": 0.12110992670059204,
      "step": 22280
    },
    {
      "epoch": 6.3269940391711605,
      "grad_norm": 16.930397033691406,
      "learning_rate": 9.367641214873688e-05,
      "loss": 0.16743370294570922,
      "step": 22290
    },
    {
      "epoch": 6.329832529094522,
      "grad_norm": 9.169883728027344,
      "learning_rate": 9.367357365881352e-05,
      "loss": 0.07741034030914307,
      "step": 22300
    },
    {
      "epoch": 6.332671019017883,
      "grad_norm": 7.381467342376709,
      "learning_rate": 9.367073516889015e-05,
      "loss": 0.11755729913711548,
      "step": 22310
    },
    {
      "epoch": 6.335509508941243,
      "grad_norm": 10.883224487304688,
      "learning_rate": 9.366789667896679e-05,
      "loss": 0.1140102505683899,
      "step": 22320
    },
    {
      "epoch": 6.338347998864604,
      "grad_norm": 15.579988479614258,
      "learning_rate": 9.366505818904343e-05,
      "loss": 0.10063846111297607,
      "step": 22330
    },
    {
      "epoch": 6.341186488787965,
      "grad_norm": 10.76476001739502,
      "learning_rate": 9.366221969912007e-05,
      "loss": 0.13289889097213745,
      "step": 22340
    },
    {
      "epoch": 6.344024978711325,
      "grad_norm": 6.35222053527832,
      "learning_rate": 9.365938120919672e-05,
      "loss": 0.09211413860321045,
      "step": 22350
    },
    {
      "epoch": 6.3468634686346865,
      "grad_norm": 10.59709358215332,
      "learning_rate": 9.365654271927336e-05,
      "loss": 0.09625858068466187,
      "step": 22360
    },
    {
      "epoch": 6.349701958558047,
      "grad_norm": 13.786771774291992,
      "learning_rate": 9.365370422935e-05,
      "loss": 0.07938930988311768,
      "step": 22370
    },
    {
      "epoch": 6.352540448481408,
      "grad_norm": 11.998483657836914,
      "learning_rate": 9.365086573942663e-05,
      "loss": 0.14438967704772948,
      "step": 22380
    },
    {
      "epoch": 6.355378938404769,
      "grad_norm": 8.744522094726562,
      "learning_rate": 9.364802724950327e-05,
      "loss": 0.10890406370162964,
      "step": 22390
    },
    {
      "epoch": 6.358217428328129,
      "grad_norm": 4.687183380126953,
      "learning_rate": 9.364518875957991e-05,
      "loss": 0.07402298450469971,
      "step": 22400
    },
    {
      "epoch": 6.36105591825149,
      "grad_norm": 17.525976181030273,
      "learning_rate": 9.364235026965654e-05,
      "loss": 0.118218994140625,
      "step": 22410
    },
    {
      "epoch": 6.3638944081748505,
      "grad_norm": 6.600802421569824,
      "learning_rate": 9.363951177973319e-05,
      "loss": 0.11827734708786011,
      "step": 22420
    },
    {
      "epoch": 6.366732898098212,
      "grad_norm": 8.856054306030273,
      "learning_rate": 9.363667328980983e-05,
      "loss": 0.06804376244544982,
      "step": 22430
    },
    {
      "epoch": 6.369571388021573,
      "grad_norm": 29.660568237304688,
      "learning_rate": 9.363383479988646e-05,
      "loss": 0.09639835357666016,
      "step": 22440
    },
    {
      "epoch": 6.372409877944933,
      "grad_norm": 16.50444984436035,
      "learning_rate": 9.36309963099631e-05,
      "loss": 0.10021114349365234,
      "step": 22450
    },
    {
      "epoch": 6.375248367868294,
      "grad_norm": 20.486492156982422,
      "learning_rate": 9.362815782003974e-05,
      "loss": 0.10686763525009155,
      "step": 22460
    },
    {
      "epoch": 6.378086857791655,
      "grad_norm": 10.710816383361816,
      "learning_rate": 9.362531933011638e-05,
      "loss": 0.08526849746704102,
      "step": 22470
    },
    {
      "epoch": 6.380925347715015,
      "grad_norm": 13.139506340026855,
      "learning_rate": 9.362248084019301e-05,
      "loss": 0.09127248525619507,
      "step": 22480
    },
    {
      "epoch": 6.3837638376383765,
      "grad_norm": 9.385111808776855,
      "learning_rate": 9.361964235026967e-05,
      "loss": 0.0844616413116455,
      "step": 22490
    },
    {
      "epoch": 6.386602327561738,
      "grad_norm": 5.706073760986328,
      "learning_rate": 9.361680386034631e-05,
      "loss": 0.09170372486114502,
      "step": 22500
    },
    {
      "epoch": 6.386602327561738,
      "eval_accuracy": 0.9301837604120302,
      "eval_loss": 0.2105872929096222,
      "eval_runtime": 31.9278,
      "eval_samples_per_second": 492.58,
      "eval_steps_per_second": 7.705,
      "step": 22500
    },
    {
      "epoch": 6.389440817485098,
      "grad_norm": 9.788163185119629,
      "learning_rate": 9.361396537042294e-05,
      "loss": 0.08152780532836915,
      "step": 22510
    },
    {
      "epoch": 6.392279307408459,
      "grad_norm": 15.817776679992676,
      "learning_rate": 9.361112688049958e-05,
      "loss": 0.09825285077095032,
      "step": 22520
    },
    {
      "epoch": 6.395117797331819,
      "grad_norm": 11.550987243652344,
      "learning_rate": 9.360828839057622e-05,
      "loss": 0.1070898175239563,
      "step": 22530
    },
    {
      "epoch": 6.39795628725518,
      "grad_norm": 14.608263969421387,
      "learning_rate": 9.360544990065285e-05,
      "loss": 0.0923076331615448,
      "step": 22540
    },
    {
      "epoch": 6.400794777178541,
      "grad_norm": 12.77895736694336,
      "learning_rate": 9.36026114107295e-05,
      "loss": 0.10174696445465088,
      "step": 22550
    },
    {
      "epoch": 6.403633267101902,
      "grad_norm": 7.593853950500488,
      "learning_rate": 9.359977292080614e-05,
      "loss": 0.0696295976638794,
      "step": 22560
    },
    {
      "epoch": 6.406471757025263,
      "grad_norm": 11.700136184692383,
      "learning_rate": 9.359693443088277e-05,
      "loss": 0.06687184572219848,
      "step": 22570
    },
    {
      "epoch": 6.409310246948623,
      "grad_norm": 7.089961528778076,
      "learning_rate": 9.359409594095941e-05,
      "loss": 0.0826976478099823,
      "step": 22580
    },
    {
      "epoch": 6.412148736871984,
      "grad_norm": 24.44202423095703,
      "learning_rate": 9.359154130002838e-05,
      "loss": 0.13295964002609253,
      "step": 22590
    },
    {
      "epoch": 6.414987226795345,
      "grad_norm": 6.154026985168457,
      "learning_rate": 9.358870281010503e-05,
      "loss": 0.09617015719413757,
      "step": 22600
    },
    {
      "epoch": 6.417825716718705,
      "grad_norm": 14.29837703704834,
      "learning_rate": 9.358586432018168e-05,
      "loss": 0.1076723575592041,
      "step": 22610
    },
    {
      "epoch": 6.4206642066420665,
      "grad_norm": 6.377991199493408,
      "learning_rate": 9.35830258302583e-05,
      "loss": 0.10216890573501587,
      "step": 22620
    },
    {
      "epoch": 6.423502696565428,
      "grad_norm": 1.7779245376586914,
      "learning_rate": 9.358018734033494e-05,
      "loss": 0.07874549627304077,
      "step": 22630
    },
    {
      "epoch": 6.426341186488788,
      "grad_norm": 17.45425033569336,
      "learning_rate": 9.357734885041159e-05,
      "loss": 0.09025533795356751,
      "step": 22640
    },
    {
      "epoch": 6.429179676412149,
      "grad_norm": 14.70903491973877,
      "learning_rate": 9.357451036048823e-05,
      "loss": 0.11120609045028687,
      "step": 22650
    },
    {
      "epoch": 6.432018166335509,
      "grad_norm": 16.77193260192871,
      "learning_rate": 9.357167187056486e-05,
      "loss": 0.10498535633087158,
      "step": 22660
    },
    {
      "epoch": 6.43485665625887,
      "grad_norm": 4.400278568267822,
      "learning_rate": 9.356883338064151e-05,
      "loss": 0.10595424175262451,
      "step": 22670
    },
    {
      "epoch": 6.437695146182231,
      "grad_norm": 11.257668495178223,
      "learning_rate": 9.356599489071815e-05,
      "loss": 0.07874829769134521,
      "step": 22680
    },
    {
      "epoch": 6.440533636105592,
      "grad_norm": 12.653393745422363,
      "learning_rate": 9.356315640079478e-05,
      "loss": 0.09990564584732056,
      "step": 22690
    },
    {
      "epoch": 6.443372126028953,
      "grad_norm": 10.325603485107422,
      "learning_rate": 9.356031791087142e-05,
      "loss": 0.07696129083633423,
      "step": 22700
    },
    {
      "epoch": 6.446210615952314,
      "grad_norm": 10.782560348510742,
      "learning_rate": 9.355747942094806e-05,
      "loss": 0.08075250387191772,
      "step": 22710
    },
    {
      "epoch": 6.449049105875674,
      "grad_norm": 4.658320426940918,
      "learning_rate": 9.355464093102469e-05,
      "loss": 0.06197248101234436,
      "step": 22720
    },
    {
      "epoch": 6.451887595799035,
      "grad_norm": 14.929058074951172,
      "learning_rate": 9.355180244110134e-05,
      "loss": 0.11998610496520996,
      "step": 22730
    },
    {
      "epoch": 6.454726085722395,
      "grad_norm": 7.436862468719482,
      "learning_rate": 9.354896395117799e-05,
      "loss": 0.0861608326435089,
      "step": 22740
    },
    {
      "epoch": 6.4575645756457565,
      "grad_norm": 5.92672061920166,
      "learning_rate": 9.354612546125461e-05,
      "loss": 0.07573016881942748,
      "step": 22750
    },
    {
      "epoch": 6.460403065569118,
      "grad_norm": 11.298891067504883,
      "learning_rate": 9.354328697133126e-05,
      "loss": 0.10512068271636962,
      "step": 22760
    },
    {
      "epoch": 6.463241555492478,
      "grad_norm": 3.5694100856781006,
      "learning_rate": 9.35404484814079e-05,
      "loss": 0.08416117429733276,
      "step": 22770
    },
    {
      "epoch": 6.466080045415839,
      "grad_norm": 14.55178165435791,
      "learning_rate": 9.353760999148454e-05,
      "loss": 0.10444525480270386,
      "step": 22780
    },
    {
      "epoch": 6.468918535339199,
      "grad_norm": 11.943623542785645,
      "learning_rate": 9.353477150156117e-05,
      "loss": 0.11375449895858765,
      "step": 22790
    },
    {
      "epoch": 6.47175702526256,
      "grad_norm": 14.246231079101562,
      "learning_rate": 9.353193301163782e-05,
      "loss": 0.07867357134819031,
      "step": 22800
    },
    {
      "epoch": 6.474595515185921,
      "grad_norm": 12.942447662353516,
      "learning_rate": 9.352909452171445e-05,
      "loss": 0.08727706670761108,
      "step": 22810
    },
    {
      "epoch": 6.477434005109282,
      "grad_norm": 15.228692054748535,
      "learning_rate": 9.352625603179109e-05,
      "loss": 0.11858946084976196,
      "step": 22820
    },
    {
      "epoch": 6.480272495032643,
      "grad_norm": 9.932172775268555,
      "learning_rate": 9.352341754186773e-05,
      "loss": 0.07700104713439941,
      "step": 22830
    },
    {
      "epoch": 6.483110984956004,
      "grad_norm": 18.2576904296875,
      "learning_rate": 9.352057905194437e-05,
      "loss": 0.09337477684020996,
      "step": 22840
    },
    {
      "epoch": 6.485949474879364,
      "grad_norm": 9.911008834838867,
      "learning_rate": 9.3517740562021e-05,
      "loss": 0.08646316528320312,
      "step": 22850
    },
    {
      "epoch": 6.488787964802725,
      "grad_norm": 14.605977058410645,
      "learning_rate": 9.351490207209764e-05,
      "loss": 0.09186390638351441,
      "step": 22860
    },
    {
      "epoch": 6.491626454726085,
      "grad_norm": 8.759679794311523,
      "learning_rate": 9.35120635821743e-05,
      "loss": 0.08041915893554688,
      "step": 22870
    },
    {
      "epoch": 6.4944649446494465,
      "grad_norm": 8.287078857421875,
      "learning_rate": 9.350922509225092e-05,
      "loss": 0.09436277747154236,
      "step": 22880
    },
    {
      "epoch": 6.497303434572808,
      "grad_norm": 12.57153034210205,
      "learning_rate": 9.350638660232757e-05,
      "loss": 0.09365052580833436,
      "step": 22890
    },
    {
      "epoch": 6.500141924496168,
      "grad_norm": 11.6986083984375,
      "learning_rate": 9.350354811240421e-05,
      "loss": 0.08615879416465759,
      "step": 22900
    },
    {
      "epoch": 6.502980414419529,
      "grad_norm": 8.28325080871582,
      "learning_rate": 9.350070962248084e-05,
      "loss": 0.14441734552383423,
      "step": 22910
    },
    {
      "epoch": 6.50581890434289,
      "grad_norm": 2.1102166175842285,
      "learning_rate": 9.349787113255748e-05,
      "loss": 0.0824457049369812,
      "step": 22920
    },
    {
      "epoch": 6.50865739426625,
      "grad_norm": 9.971158981323242,
      "learning_rate": 9.349503264263413e-05,
      "loss": 0.10625452995300293,
      "step": 22930
    },
    {
      "epoch": 6.511495884189611,
      "grad_norm": 4.650021076202393,
      "learning_rate": 9.349219415271076e-05,
      "loss": 0.07723555564880372,
      "step": 22940
    },
    {
      "epoch": 6.514334374112972,
      "grad_norm": 8.099427223205566,
      "learning_rate": 9.34893556627874e-05,
      "loss": 0.12295387983322144,
      "step": 22950
    },
    {
      "epoch": 6.517172864036333,
      "grad_norm": 13.784622192382812,
      "learning_rate": 9.348651717286404e-05,
      "loss": 0.0961941123008728,
      "step": 22960
    },
    {
      "epoch": 6.520011353959694,
      "grad_norm": 2.625196695327759,
      "learning_rate": 9.348367868294068e-05,
      "loss": 0.11788196563720703,
      "step": 22970
    },
    {
      "epoch": 6.522849843883054,
      "grad_norm": 3.4829862117767334,
      "learning_rate": 9.348084019301731e-05,
      "loss": 0.06416496038436889,
      "step": 22980
    },
    {
      "epoch": 6.525688333806415,
      "grad_norm": 11.49492073059082,
      "learning_rate": 9.347800170309395e-05,
      "loss": 0.1067654013633728,
      "step": 22990
    },
    {
      "epoch": 6.528526823729775,
      "grad_norm": 13.783095359802246,
      "learning_rate": 9.347516321317061e-05,
      "loss": 0.08907094001770019,
      "step": 23000
    },
    {
      "epoch": 6.528526823729775,
      "eval_accuracy": 0.931582628600496,
      "eval_loss": 0.2043585479259491,
      "eval_runtime": 32.1399,
      "eval_samples_per_second": 489.33,
      "eval_steps_per_second": 7.654,
      "step": 23000
    },
    {
      "epoch": 6.531365313653136,
      "grad_norm": 12.19428539276123,
      "learning_rate": 9.347232472324724e-05,
      "loss": 0.11192913055419922,
      "step": 23010
    },
    {
      "epoch": 6.5342038035764975,
      "grad_norm": 9.6806058883667,
      "learning_rate": 9.346948623332388e-05,
      "loss": 0.11081634759902954,
      "step": 23020
    },
    {
      "epoch": 6.537042293499858,
      "grad_norm": 9.315773963928223,
      "learning_rate": 9.346664774340052e-05,
      "loss": 0.09858989119529724,
      "step": 23030
    },
    {
      "epoch": 6.539880783423219,
      "grad_norm": 7.326742649078369,
      "learning_rate": 9.346380925347715e-05,
      "loss": 0.0827315092086792,
      "step": 23040
    },
    {
      "epoch": 6.54271927334658,
      "grad_norm": 14.56912612915039,
      "learning_rate": 9.346097076355379e-05,
      "loss": 0.10821828842163086,
      "step": 23050
    },
    {
      "epoch": 6.54555776326994,
      "grad_norm": 4.597448825836182,
      "learning_rate": 9.345813227363043e-05,
      "loss": 0.09526298642158508,
      "step": 23060
    },
    {
      "epoch": 6.548396253193301,
      "grad_norm": 11.362309455871582,
      "learning_rate": 9.345529378370707e-05,
      "loss": 0.09052236676216126,
      "step": 23070
    },
    {
      "epoch": 6.551234743116662,
      "grad_norm": 5.615175247192383,
      "learning_rate": 9.345245529378371e-05,
      "loss": 0.07679804563522338,
      "step": 23080
    },
    {
      "epoch": 6.554073233040023,
      "grad_norm": 9.740547180175781,
      "learning_rate": 9.344961680386035e-05,
      "loss": 0.08708915710449219,
      "step": 23090
    },
    {
      "epoch": 6.556911722963384,
      "grad_norm": 4.926576614379883,
      "learning_rate": 9.3446778313937e-05,
      "loss": 0.0837101936340332,
      "step": 23100
    },
    {
      "epoch": 6.559750212886744,
      "grad_norm": 7.401463031768799,
      "learning_rate": 9.344393982401362e-05,
      "loss": 0.09549596309661865,
      "step": 23110
    },
    {
      "epoch": 6.562588702810105,
      "grad_norm": 19.454238891601562,
      "learning_rate": 9.344110133409026e-05,
      "loss": 0.06996254324913025,
      "step": 23120
    },
    {
      "epoch": 6.565427192733466,
      "grad_norm": 12.590356826782227,
      "learning_rate": 9.343826284416692e-05,
      "loss": 0.11268901824951172,
      "step": 23130
    },
    {
      "epoch": 6.568265682656826,
      "grad_norm": 15.136454582214355,
      "learning_rate": 9.343542435424355e-05,
      "loss": 0.11089028120040893,
      "step": 23140
    },
    {
      "epoch": 6.5711041725801875,
      "grad_norm": 20.47833251953125,
      "learning_rate": 9.343258586432019e-05,
      "loss": 0.16502034664154053,
      "step": 23150
    },
    {
      "epoch": 6.573942662503548,
      "grad_norm": 18.76143455505371,
      "learning_rate": 9.342974737439683e-05,
      "loss": 0.1134371042251587,
      "step": 23160
    },
    {
      "epoch": 6.576781152426909,
      "grad_norm": 5.621868133544922,
      "learning_rate": 9.342690888447346e-05,
      "loss": 0.07961916327476501,
      "step": 23170
    },
    {
      "epoch": 6.57961964235027,
      "grad_norm": 4.496213912963867,
      "learning_rate": 9.34240703945501e-05,
      "loss": 0.06913825273513793,
      "step": 23180
    },
    {
      "epoch": 6.58245813227363,
      "grad_norm": 5.536193370819092,
      "learning_rate": 9.342123190462674e-05,
      "loss": 0.0882467806339264,
      "step": 23190
    },
    {
      "epoch": 6.585296622196991,
      "grad_norm": 7.135318756103516,
      "learning_rate": 9.341839341470338e-05,
      "loss": 0.08491207361221313,
      "step": 23200
    },
    {
      "epoch": 6.5881351121203515,
      "grad_norm": 10.8344087600708,
      "learning_rate": 9.341555492478002e-05,
      "loss": 0.0774510383605957,
      "step": 23210
    },
    {
      "epoch": 6.590973602043713,
      "grad_norm": 13.20254898071289,
      "learning_rate": 9.341271643485666e-05,
      "loss": 0.08600361347198486,
      "step": 23220
    },
    {
      "epoch": 6.593812091967074,
      "grad_norm": 10.479114532470703,
      "learning_rate": 9.34098779449333e-05,
      "loss": 0.11895914077758789,
      "step": 23230
    },
    {
      "epoch": 6.596650581890434,
      "grad_norm": 15.753108024597168,
      "learning_rate": 9.340703945500993e-05,
      "loss": 0.09777660369873047,
      "step": 23240
    },
    {
      "epoch": 6.599489071813795,
      "grad_norm": 6.3389081954956055,
      "learning_rate": 9.340420096508657e-05,
      "loss": 0.08923123478889465,
      "step": 23250
    },
    {
      "epoch": 6.602327561737156,
      "grad_norm": 7.197924613952637,
      "learning_rate": 9.340136247516322e-05,
      "loss": 0.08538019061088561,
      "step": 23260
    },
    {
      "epoch": 6.605166051660516,
      "grad_norm": 12.7660493850708,
      "learning_rate": 9.339852398523986e-05,
      "loss": 0.11998507976531983,
      "step": 23270
    },
    {
      "epoch": 6.6080045415838775,
      "grad_norm": 17.26128578186035,
      "learning_rate": 9.33956854953165e-05,
      "loss": 0.08162992000579834,
      "step": 23280
    },
    {
      "epoch": 6.610843031507239,
      "grad_norm": 23.859159469604492,
      "learning_rate": 9.339284700539314e-05,
      "loss": 0.10878651142120362,
      "step": 23290
    },
    {
      "epoch": 6.613681521430599,
      "grad_norm": 5.97392463684082,
      "learning_rate": 9.339000851546977e-05,
      "loss": 0.08939759135246277,
      "step": 23300
    },
    {
      "epoch": 6.61652001135396,
      "grad_norm": 6.42591667175293,
      "learning_rate": 9.338717002554641e-05,
      "loss": 0.0937576413154602,
      "step": 23310
    },
    {
      "epoch": 6.61935850127732,
      "grad_norm": 12.023765563964844,
      "learning_rate": 9.338433153562305e-05,
      "loss": 0.0607508659362793,
      "step": 23320
    },
    {
      "epoch": 6.622196991200681,
      "grad_norm": 11.77810001373291,
      "learning_rate": 9.338149304569969e-05,
      "loss": 0.10685676336288452,
      "step": 23330
    },
    {
      "epoch": 6.625035481124042,
      "grad_norm": 7.256436824798584,
      "learning_rate": 9.337865455577633e-05,
      "loss": 0.10113931894302368,
      "step": 23340
    },
    {
      "epoch": 6.627873971047403,
      "grad_norm": 11.541536331176758,
      "learning_rate": 9.337581606585297e-05,
      "loss": 0.060240209102630615,
      "step": 23350
    },
    {
      "epoch": 6.630712460970764,
      "grad_norm": 9.527950286865234,
      "learning_rate": 9.337297757592962e-05,
      "loss": 0.09862492084503174,
      "step": 23360
    },
    {
      "epoch": 6.633550950894124,
      "grad_norm": 5.018553256988525,
      "learning_rate": 9.337013908600624e-05,
      "loss": 0.0929749608039856,
      "step": 23370
    },
    {
      "epoch": 6.636389440817485,
      "grad_norm": 15.911595344543457,
      "learning_rate": 9.336730059608289e-05,
      "loss": 0.07330633997917176,
      "step": 23380
    },
    {
      "epoch": 6.639227930740846,
      "grad_norm": 8.000524520874023,
      "learning_rate": 9.336446210615953e-05,
      "loss": 0.09455389976501465,
      "step": 23390
    },
    {
      "epoch": 6.642066420664206,
      "grad_norm": 8.035616874694824,
      "learning_rate": 9.336162361623617e-05,
      "loss": 0.08320001363754273,
      "step": 23400
    },
    {
      "epoch": 6.6449049105875675,
      "grad_norm": 7.5174760818481445,
      "learning_rate": 9.335878512631281e-05,
      "loss": 0.06982924342155457,
      "step": 23410
    },
    {
      "epoch": 6.647743400510928,
      "grad_norm": 4.479267120361328,
      "learning_rate": 9.335594663638945e-05,
      "loss": 0.07252361178398133,
      "step": 23420
    },
    {
      "epoch": 6.650581890434289,
      "grad_norm": 7.234189510345459,
      "learning_rate": 9.335310814646608e-05,
      "loss": 0.07169376611709595,
      "step": 23430
    },
    {
      "epoch": 6.65342038035765,
      "grad_norm": 18.071443557739258,
      "learning_rate": 9.335026965654272e-05,
      "loss": 0.09714398384094239,
      "step": 23440
    },
    {
      "epoch": 6.65625887028101,
      "grad_norm": 8.038203239440918,
      "learning_rate": 9.334743116661936e-05,
      "loss": 0.07264513969421386,
      "step": 23450
    },
    {
      "epoch": 6.659097360204371,
      "grad_norm": 9.719202995300293,
      "learning_rate": 9.3344592676696e-05,
      "loss": 0.10534716844558716,
      "step": 23460
    },
    {
      "epoch": 6.661935850127732,
      "grad_norm": 5.747554302215576,
      "learning_rate": 9.334175418677264e-05,
      "loss": 0.11357256174087524,
      "step": 23470
    },
    {
      "epoch": 6.664774340051093,
      "grad_norm": 7.582512855529785,
      "learning_rate": 9.333891569684929e-05,
      "loss": 0.10980550050735474,
      "step": 23480
    },
    {
      "epoch": 6.667612829974454,
      "grad_norm": 11.435025215148926,
      "learning_rate": 9.333607720692593e-05,
      "loss": 0.08182085752487182,
      "step": 23490
    },
    {
      "epoch": 6.670451319897815,
      "grad_norm": 4.0055623054504395,
      "learning_rate": 9.333323871700255e-05,
      "loss": 0.10198774337768554,
      "step": 23500
    },
    {
      "epoch": 6.670451319897815,
      "eval_accuracy": 0.9365422521777834,
      "eval_loss": 0.19689252972602844,
      "eval_runtime": 31.32,
      "eval_samples_per_second": 502.14,
      "eval_steps_per_second": 7.854,
      "step": 23500
    },
    {
      "epoch": 6.673289809821175,
      "grad_norm": 5.353410720825195,
      "learning_rate": 9.33304002270792e-05,
      "loss": 0.07309556007385254,
      "step": 23510
    },
    {
      "epoch": 6.676128299744536,
      "grad_norm": 18.983692169189453,
      "learning_rate": 9.332756173715584e-05,
      "loss": 0.10379167795181274,
      "step": 23520
    },
    {
      "epoch": 6.678966789667896,
      "grad_norm": 12.102860450744629,
      "learning_rate": 9.332472324723248e-05,
      "loss": 0.14615061283111572,
      "step": 23530
    },
    {
      "epoch": 6.6818052795912575,
      "grad_norm": 16.388975143432617,
      "learning_rate": 9.332188475730912e-05,
      "loss": 0.0960249125957489,
      "step": 23540
    },
    {
      "epoch": 6.684643769514619,
      "grad_norm": 9.716329574584961,
      "learning_rate": 9.331904626738576e-05,
      "loss": 0.10273417234420776,
      "step": 23550
    },
    {
      "epoch": 6.687482259437979,
      "grad_norm": 7.821311950683594,
      "learning_rate": 9.331620777746239e-05,
      "loss": 0.11158208847045899,
      "step": 23560
    },
    {
      "epoch": 6.69032074936134,
      "grad_norm": 15.425642013549805,
      "learning_rate": 9.331336928753903e-05,
      "loss": 0.11695325374603271,
      "step": 23570
    },
    {
      "epoch": 6.6931592392847,
      "grad_norm": 8.43350601196289,
      "learning_rate": 9.331053079761567e-05,
      "loss": 0.12026833295822144,
      "step": 23580
    },
    {
      "epoch": 6.695997729208061,
      "grad_norm": 12.40167236328125,
      "learning_rate": 9.330769230769231e-05,
      "loss": 0.10209051370620728,
      "step": 23590
    },
    {
      "epoch": 6.698836219131422,
      "grad_norm": 9.377254486083984,
      "learning_rate": 9.330485381776895e-05,
      "loss": 0.10837109088897705,
      "step": 23600
    },
    {
      "epoch": 6.701674709054783,
      "grad_norm": 6.663228511810303,
      "learning_rate": 9.33020153278456e-05,
      "loss": 0.11400752067565918,
      "step": 23610
    },
    {
      "epoch": 6.704513198978144,
      "grad_norm": 9.539036750793457,
      "learning_rate": 9.329917683792224e-05,
      "loss": 0.07644888758659363,
      "step": 23620
    },
    {
      "epoch": 6.707351688901504,
      "grad_norm": 12.6395845413208,
      "learning_rate": 9.329633834799887e-05,
      "loss": 0.11350473165512084,
      "step": 23630
    },
    {
      "epoch": 6.710190178824865,
      "grad_norm": 9.589407920837402,
      "learning_rate": 9.329349985807551e-05,
      "loss": 0.08152202367782593,
      "step": 23640
    },
    {
      "epoch": 6.713028668748226,
      "grad_norm": 15.663517951965332,
      "learning_rate": 9.329066136815215e-05,
      "loss": 0.11970263719558716,
      "step": 23650
    },
    {
      "epoch": 6.715867158671586,
      "grad_norm": 12.597955703735352,
      "learning_rate": 9.328782287822879e-05,
      "loss": 0.09735053777694702,
      "step": 23660
    },
    {
      "epoch": 6.7187056485949475,
      "grad_norm": 18.34467124938965,
      "learning_rate": 9.328498438830543e-05,
      "loss": 0.10283708572387695,
      "step": 23670
    },
    {
      "epoch": 6.721544138518309,
      "grad_norm": 12.121549606323242,
      "learning_rate": 9.328214589838207e-05,
      "loss": 0.11331470012664795,
      "step": 23680
    },
    {
      "epoch": 6.724382628441669,
      "grad_norm": 13.72669792175293,
      "learning_rate": 9.32793074084587e-05,
      "loss": 0.09874246716499328,
      "step": 23690
    },
    {
      "epoch": 6.72722111836503,
      "grad_norm": 11.512922286987305,
      "learning_rate": 9.327646891853534e-05,
      "loss": 0.08080302476882935,
      "step": 23700
    },
    {
      "epoch": 6.730059608288391,
      "grad_norm": 10.64036750793457,
      "learning_rate": 9.327363042861198e-05,
      "loss": 0.1081117868423462,
      "step": 23710
    },
    {
      "epoch": 6.732898098211751,
      "grad_norm": 19.117401123046875,
      "learning_rate": 9.327079193868862e-05,
      "loss": 0.09446322917938232,
      "step": 23720
    },
    {
      "epoch": 6.735736588135112,
      "grad_norm": 5.025933742523193,
      "learning_rate": 9.326795344876527e-05,
      "loss": 0.11468440294265747,
      "step": 23730
    },
    {
      "epoch": 6.738575078058473,
      "grad_norm": 6.562473773956299,
      "learning_rate": 9.326511495884191e-05,
      "loss": 0.08148735761642456,
      "step": 23740
    },
    {
      "epoch": 6.741413567981834,
      "grad_norm": 10.234820365905762,
      "learning_rate": 9.326227646891853e-05,
      "loss": 0.10205699205398559,
      "step": 23750
    },
    {
      "epoch": 6.744252057905195,
      "grad_norm": 8.898704528808594,
      "learning_rate": 9.325943797899518e-05,
      "loss": 0.06831822991371155,
      "step": 23760
    },
    {
      "epoch": 6.747090547828555,
      "grad_norm": 17.5389347076416,
      "learning_rate": 9.325659948907182e-05,
      "loss": 0.1090964913368225,
      "step": 23770
    },
    {
      "epoch": 6.749929037751916,
      "grad_norm": 9.973577499389648,
      "learning_rate": 9.325376099914846e-05,
      "loss": 0.08267690539360047,
      "step": 23780
    },
    {
      "epoch": 6.752767527675276,
      "grad_norm": 9.735045433044434,
      "learning_rate": 9.325092250922509e-05,
      "loss": 0.09638917446136475,
      "step": 23790
    },
    {
      "epoch": 6.755606017598637,
      "grad_norm": 14.622360229492188,
      "learning_rate": 9.324808401930174e-05,
      "loss": 0.1093931794166565,
      "step": 23800
    },
    {
      "epoch": 6.7584445075219985,
      "grad_norm": 10.59725284576416,
      "learning_rate": 9.324524552937838e-05,
      "loss": 0.11809722185134888,
      "step": 23810
    },
    {
      "epoch": 6.761282997445359,
      "grad_norm": 10.137466430664062,
      "learning_rate": 9.324240703945501e-05,
      "loss": 0.11568174362182618,
      "step": 23820
    },
    {
      "epoch": 6.76412148736872,
      "grad_norm": 14.032526016235352,
      "learning_rate": 9.323956854953165e-05,
      "loss": 0.09648122787475585,
      "step": 23830
    },
    {
      "epoch": 6.766959977292081,
      "grad_norm": 9.677608489990234,
      "learning_rate": 9.32367300596083e-05,
      "loss": 0.06552197337150574,
      "step": 23840
    },
    {
      "epoch": 6.769798467215441,
      "grad_norm": 12.235411643981934,
      "learning_rate": 9.323389156968492e-05,
      "loss": 0.10965950489044189,
      "step": 23850
    },
    {
      "epoch": 6.772636957138802,
      "grad_norm": 9.106744766235352,
      "learning_rate": 9.323105307976158e-05,
      "loss": 0.08939334750175476,
      "step": 23860
    },
    {
      "epoch": 6.775475447062163,
      "grad_norm": 6.746470928192139,
      "learning_rate": 9.322821458983822e-05,
      "loss": 0.07405282855033875,
      "step": 23870
    },
    {
      "epoch": 6.778313936985524,
      "grad_norm": 12.039777755737305,
      "learning_rate": 9.322537609991485e-05,
      "loss": 0.09850831627845764,
      "step": 23880
    },
    {
      "epoch": 6.781152426908885,
      "grad_norm": 15.966470718383789,
      "learning_rate": 9.322253760999149e-05,
      "loss": 0.08547778129577636,
      "step": 23890
    },
    {
      "epoch": 6.783990916832245,
      "grad_norm": 18.932947158813477,
      "learning_rate": 9.321969912006813e-05,
      "loss": 0.09407666325569153,
      "step": 23900
    },
    {
      "epoch": 6.786829406755606,
      "grad_norm": 16.5733585357666,
      "learning_rate": 9.321686063014477e-05,
      "loss": 0.12048181295394897,
      "step": 23910
    },
    {
      "epoch": 6.789667896678967,
      "grad_norm": 10.802082061767578,
      "learning_rate": 9.32140221402214e-05,
      "loss": 0.13048343658447265,
      "step": 23920
    },
    {
      "epoch": 6.792506386602327,
      "grad_norm": 8.369768142700195,
      "learning_rate": 9.321118365029805e-05,
      "loss": 0.08966717720031739,
      "step": 23930
    },
    {
      "epoch": 6.7953448765256885,
      "grad_norm": 10.59682559967041,
      "learning_rate": 9.32083451603747e-05,
      "loss": 0.053786516189575195,
      "step": 23940
    },
    {
      "epoch": 6.798183366449049,
      "grad_norm": 10.559469223022461,
      "learning_rate": 9.320550667045132e-05,
      "loss": 0.08415335416793823,
      "step": 23950
    },
    {
      "epoch": 6.80102185637241,
      "grad_norm": 11.114952087402344,
      "learning_rate": 9.320266818052796e-05,
      "loss": 0.12069662809371948,
      "step": 23960
    },
    {
      "epoch": 6.803860346295771,
      "grad_norm": 10.029012680053711,
      "learning_rate": 9.31998296906046e-05,
      "loss": 0.09029229879379272,
      "step": 23970
    },
    {
      "epoch": 6.806698836219131,
      "grad_norm": 14.684560775756836,
      "learning_rate": 9.319699120068123e-05,
      "loss": 0.07971057891845704,
      "step": 23980
    },
    {
      "epoch": 6.809537326142492,
      "grad_norm": 10.083139419555664,
      "learning_rate": 9.319415271075787e-05,
      "loss": 0.08903594613075257,
      "step": 23990
    },
    {
      "epoch": 6.8123758160658525,
      "grad_norm": 17.14006233215332,
      "learning_rate": 9.319131422083453e-05,
      "loss": 0.10128430128097535,
      "step": 24000
    },
    {
      "epoch": 6.8123758160658525,
      "eval_accuracy": 0.9203280981751129,
      "eval_loss": 0.23221370577812195,
      "eval_runtime": 31.6746,
      "eval_samples_per_second": 496.518,
      "eval_steps_per_second": 7.766,
      "step": 24000
    },
    {
      "epoch": 6.815214305989214,
      "grad_norm": 5.353432655334473,
      "learning_rate": 9.318847573091116e-05,
      "loss": 0.09382016658782959,
      "step": 24010
    },
    {
      "epoch": 6.818052795912575,
      "grad_norm": 9.091730117797852,
      "learning_rate": 9.31856372409878e-05,
      "loss": 0.08397729992866516,
      "step": 24020
    },
    {
      "epoch": 6.820891285835935,
      "grad_norm": 8.414477348327637,
      "learning_rate": 9.318279875106444e-05,
      "loss": 0.10697766542434692,
      "step": 24030
    },
    {
      "epoch": 6.823729775759296,
      "grad_norm": 8.535869598388672,
      "learning_rate": 9.317996026114108e-05,
      "loss": 0.07649033665657043,
      "step": 24040
    },
    {
      "epoch": 6.826568265682657,
      "grad_norm": 23.232709884643555,
      "learning_rate": 9.317712177121771e-05,
      "loss": 0.09270018935203553,
      "step": 24050
    },
    {
      "epoch": 6.829406755606017,
      "grad_norm": 12.610830307006836,
      "learning_rate": 9.317428328129436e-05,
      "loss": 0.07556609511375427,
      "step": 24060
    },
    {
      "epoch": 6.8322452455293785,
      "grad_norm": 6.084565162658691,
      "learning_rate": 9.3171444791371e-05,
      "loss": 0.10252330303192139,
      "step": 24070
    },
    {
      "epoch": 6.83508373545274,
      "grad_norm": 4.155466079711914,
      "learning_rate": 9.316860630144763e-05,
      "loss": 0.06510661244392395,
      "step": 24080
    },
    {
      "epoch": 6.8379222253761,
      "grad_norm": 16.93148422241211,
      "learning_rate": 9.316576781152427e-05,
      "loss": 0.10527651309967041,
      "step": 24090
    },
    {
      "epoch": 6.840760715299461,
      "grad_norm": 20.688505172729492,
      "learning_rate": 9.316292932160092e-05,
      "loss": 0.09468711614608764,
      "step": 24100
    },
    {
      "epoch": 6.843599205222821,
      "grad_norm": 18.16554832458496,
      "learning_rate": 9.316009083167754e-05,
      "loss": 0.11274542808532714,
      "step": 24110
    },
    {
      "epoch": 6.846437695146182,
      "grad_norm": 9.878766059875488,
      "learning_rate": 9.315725234175418e-05,
      "loss": 0.10573322772979736,
      "step": 24120
    },
    {
      "epoch": 6.849276185069543,
      "grad_norm": 17.459016799926758,
      "learning_rate": 9.315441385183084e-05,
      "loss": 0.11389918327331543,
      "step": 24130
    },
    {
      "epoch": 6.852114674992904,
      "grad_norm": 11.15742015838623,
      "learning_rate": 9.315157536190747e-05,
      "loss": 0.11135003566741944,
      "step": 24140
    },
    {
      "epoch": 6.854953164916265,
      "grad_norm": 18.712249755859375,
      "learning_rate": 9.314873687198411e-05,
      "loss": 0.11905930042266846,
      "step": 24150
    },
    {
      "epoch": 6.857791654839625,
      "grad_norm": 11.562289237976074,
      "learning_rate": 9.314589838206075e-05,
      "loss": 0.08661167025566101,
      "step": 24160
    },
    {
      "epoch": 6.860630144762986,
      "grad_norm": 13.272250175476074,
      "learning_rate": 9.314305989213739e-05,
      "loss": 0.10623058080673217,
      "step": 24170
    },
    {
      "epoch": 6.863468634686347,
      "grad_norm": 9.34141731262207,
      "learning_rate": 9.314022140221402e-05,
      "loss": 0.09637885093688965,
      "step": 24180
    },
    {
      "epoch": 6.866307124609707,
      "grad_norm": 17.596288681030273,
      "learning_rate": 9.313738291229066e-05,
      "loss": 0.09468052983283996,
      "step": 24190
    },
    {
      "epoch": 6.8691456145330685,
      "grad_norm": 19.334457397460938,
      "learning_rate": 9.313454442236732e-05,
      "loss": 0.09870043992996216,
      "step": 24200
    },
    {
      "epoch": 6.871984104456429,
      "grad_norm": 7.195592403411865,
      "learning_rate": 9.313170593244394e-05,
      "loss": 0.09635913968086243,
      "step": 24210
    },
    {
      "epoch": 6.87482259437979,
      "grad_norm": 4.203941822052002,
      "learning_rate": 9.312886744252058e-05,
      "loss": 0.09320802688598633,
      "step": 24220
    },
    {
      "epoch": 6.877661084303151,
      "grad_norm": 5.168039798736572,
      "learning_rate": 9.312602895259723e-05,
      "loss": 0.08890237808227539,
      "step": 24230
    },
    {
      "epoch": 6.880499574226511,
      "grad_norm": 11.354249000549316,
      "learning_rate": 9.312319046267385e-05,
      "loss": 0.1269654154777527,
      "step": 24240
    },
    {
      "epoch": 6.883338064149872,
      "grad_norm": 11.18812084197998,
      "learning_rate": 9.31203519727505e-05,
      "loss": 0.09434882998466491,
      "step": 24250
    },
    {
      "epoch": 6.886176554073233,
      "grad_norm": 9.15461540222168,
      "learning_rate": 9.311751348282715e-05,
      "loss": 0.09702085852622985,
      "step": 24260
    },
    {
      "epoch": 6.889015043996594,
      "grad_norm": 13.384735107421875,
      "learning_rate": 9.311467499290378e-05,
      "loss": 0.05970205664634705,
      "step": 24270
    },
    {
      "epoch": 6.891853533919955,
      "grad_norm": 9.660673141479492,
      "learning_rate": 9.311183650298042e-05,
      "loss": 0.07755666375160217,
      "step": 24280
    },
    {
      "epoch": 6.894692023843316,
      "grad_norm": 13.448535919189453,
      "learning_rate": 9.310899801305706e-05,
      "loss": 0.09545423984527587,
      "step": 24290
    },
    {
      "epoch": 6.897530513766676,
      "grad_norm": 15.696562767028809,
      "learning_rate": 9.31061595231337e-05,
      "loss": 0.12689557075500488,
      "step": 24300
    },
    {
      "epoch": 6.900369003690037,
      "grad_norm": 9.575885772705078,
      "learning_rate": 9.310332103321033e-05,
      "loss": 0.08498870134353638,
      "step": 24310
    },
    {
      "epoch": 6.903207493613397,
      "grad_norm": 2.8332626819610596,
      "learning_rate": 9.310048254328697e-05,
      "loss": 0.08259831666946411,
      "step": 24320
    },
    {
      "epoch": 6.9060459835367585,
      "grad_norm": 12.012787818908691,
      "learning_rate": 9.309764405336363e-05,
      "loss": 0.12632443904876708,
      "step": 24330
    },
    {
      "epoch": 6.90888447346012,
      "grad_norm": 4.253190040588379,
      "learning_rate": 9.309480556344025e-05,
      "loss": 0.0795013964176178,
      "step": 24340
    },
    {
      "epoch": 6.91172296338348,
      "grad_norm": 8.718367576599121,
      "learning_rate": 9.30919670735169e-05,
      "loss": 0.11982712745666504,
      "step": 24350
    },
    {
      "epoch": 6.914561453306841,
      "grad_norm": 10.600751876831055,
      "learning_rate": 9.308912858359354e-05,
      "loss": 0.10106738805770873,
      "step": 24360
    },
    {
      "epoch": 6.917399943230201,
      "grad_norm": 12.918252944946289,
      "learning_rate": 9.308629009367016e-05,
      "loss": 0.08397529125213624,
      "step": 24370
    },
    {
      "epoch": 6.920238433153562,
      "grad_norm": 7.796765327453613,
      "learning_rate": 9.30834516037468e-05,
      "loss": 0.062130266427993776,
      "step": 24380
    },
    {
      "epoch": 6.923076923076923,
      "grad_norm": 11.75627613067627,
      "learning_rate": 9.308061311382345e-05,
      "loss": 0.0711436927318573,
      "step": 24390
    },
    {
      "epoch": 6.925915413000284,
      "grad_norm": 14.818982124328613,
      "learning_rate": 9.307777462390009e-05,
      "loss": 0.09907492399215698,
      "step": 24400
    },
    {
      "epoch": 6.928753902923645,
      "grad_norm": 8.472296714782715,
      "learning_rate": 9.307493613397673e-05,
      "loss": 0.10106189250946045,
      "step": 24410
    },
    {
      "epoch": 6.931592392847005,
      "grad_norm": 10.537667274475098,
      "learning_rate": 9.307209764405337e-05,
      "loss": 0.07931021451950074,
      "step": 24420
    },
    {
      "epoch": 6.934430882770366,
      "grad_norm": 18.851049423217773,
      "learning_rate": 9.306925915413001e-05,
      "loss": 0.11357226371765136,
      "step": 24430
    },
    {
      "epoch": 6.937269372693727,
      "grad_norm": 21.69588279724121,
      "learning_rate": 9.306642066420664e-05,
      "loss": 0.13082100152969361,
      "step": 24440
    },
    {
      "epoch": 6.940107862617087,
      "grad_norm": 9.200159072875977,
      "learning_rate": 9.306358217428328e-05,
      "loss": 0.08630244731903076,
      "step": 24450
    },
    {
      "epoch": 6.9429463525404485,
      "grad_norm": 8.029813766479492,
      "learning_rate": 9.306074368435992e-05,
      "loss": 0.08214585185050964,
      "step": 24460
    },
    {
      "epoch": 6.94578484246381,
      "grad_norm": 10.202174186706543,
      "learning_rate": 9.305790519443656e-05,
      "loss": 0.1129606008529663,
      "step": 24470
    },
    {
      "epoch": 6.94862333238717,
      "grad_norm": 11.96861457824707,
      "learning_rate": 9.30550667045132e-05,
      "loss": 0.15480034351348876,
      "step": 24480
    },
    {
      "epoch": 6.951461822310531,
      "grad_norm": 11.820459365844727,
      "learning_rate": 9.305222821458985e-05,
      "loss": 0.07899882793426513,
      "step": 24490
    },
    {
      "epoch": 6.954300312233892,
      "grad_norm": 10.035128593444824,
      "learning_rate": 9.304938972466648e-05,
      "loss": 0.10380600690841675,
      "step": 24500
    },
    {
      "epoch": 6.954300312233892,
      "eval_accuracy": 0.9291028168118523,
      "eval_loss": 0.2089700549840927,
      "eval_runtime": 31.3539,
      "eval_samples_per_second": 501.597,
      "eval_steps_per_second": 7.846,
      "step": 24500
    },
    {
      "epoch": 6.957138802157252,
      "grad_norm": 5.837090015411377,
      "learning_rate": 9.304655123474312e-05,
      "loss": 0.09950883984565735,
      "step": 24510
    },
    {
      "epoch": 6.959977292080613,
      "grad_norm": 11.904397964477539,
      "learning_rate": 9.304371274481976e-05,
      "loss": 0.08639371395111084,
      "step": 24520
    },
    {
      "epoch": 6.9628157820039736,
      "grad_norm": 5.764144420623779,
      "learning_rate": 9.30408742548964e-05,
      "loss": 0.09941230416297912,
      "step": 24530
    },
    {
      "epoch": 6.965654271927335,
      "grad_norm": 16.358959197998047,
      "learning_rate": 9.303803576497304e-05,
      "loss": 0.09886674880981446,
      "step": 24540
    },
    {
      "epoch": 6.968492761850696,
      "grad_norm": 6.199477195739746,
      "learning_rate": 9.303519727504968e-05,
      "loss": 0.09478334188461304,
      "step": 24550
    },
    {
      "epoch": 6.971331251774056,
      "grad_norm": 19.868894577026367,
      "learning_rate": 9.303235878512631e-05,
      "loss": 0.10347808599472046,
      "step": 24560
    },
    {
      "epoch": 6.974169741697417,
      "grad_norm": 11.553400993347168,
      "learning_rate": 9.302952029520295e-05,
      "loss": 0.10487091541290283,
      "step": 24570
    },
    {
      "epoch": 6.977008231620777,
      "grad_norm": 6.1727447509765625,
      "learning_rate": 9.302668180527959e-05,
      "loss": 0.1479036331176758,
      "step": 24580
    },
    {
      "epoch": 6.979846721544138,
      "grad_norm": 6.307447910308838,
      "learning_rate": 9.302384331535623e-05,
      "loss": 0.09229648709297181,
      "step": 24590
    },
    {
      "epoch": 6.9826852114674995,
      "grad_norm": 9.719098091125488,
      "learning_rate": 9.302100482543288e-05,
      "loss": 0.0741078019142151,
      "step": 24600
    },
    {
      "epoch": 6.98552370139086,
      "grad_norm": 10.375625610351562,
      "learning_rate": 9.301816633550952e-05,
      "loss": 0.08486036658287048,
      "step": 24610
    },
    {
      "epoch": 6.988362191314221,
      "grad_norm": 10.996009826660156,
      "learning_rate": 9.301532784558616e-05,
      "loss": 0.11396535634994506,
      "step": 24620
    },
    {
      "epoch": 6.991200681237582,
      "grad_norm": 5.173250675201416,
      "learning_rate": 9.301248935566279e-05,
      "loss": 0.08213579654693604,
      "step": 24630
    },
    {
      "epoch": 6.994039171160942,
      "grad_norm": 12.531560897827148,
      "learning_rate": 9.300965086573943e-05,
      "loss": 0.11228445768356324,
      "step": 24640
    },
    {
      "epoch": 6.996877661084303,
      "grad_norm": 12.625089645385742,
      "learning_rate": 9.300681237581607e-05,
      "loss": 0.10983151197433472,
      "step": 24650
    },
    {
      "epoch": 6.999716151007664,
      "grad_norm": 12.65639591217041,
      "learning_rate": 9.300397388589271e-05,
      "loss": 0.07384233474731446,
      "step": 24660
    },
    {
      "epoch": 7.002554640931025,
      "grad_norm": 8.278828620910645,
      "learning_rate": 9.300141924496169e-05,
      "loss": 0.08259215950965881,
      "step": 24670
    },
    {
      "epoch": 7.005393130854386,
      "grad_norm": 3.697892665863037,
      "learning_rate": 9.299858075503832e-05,
      "loss": 0.06144115328788757,
      "step": 24680
    },
    {
      "epoch": 7.008231620777746,
      "grad_norm": 13.145040512084961,
      "learning_rate": 9.299574226511496e-05,
      "loss": 0.06724100708961486,
      "step": 24690
    },
    {
      "epoch": 7.011070110701107,
      "grad_norm": 6.598389148712158,
      "learning_rate": 9.29929037751916e-05,
      "loss": 0.06451724171638488,
      "step": 24700
    },
    {
      "epoch": 7.013908600624468,
      "grad_norm": 12.674330711364746,
      "learning_rate": 9.299006528526824e-05,
      "loss": 0.05395633578300476,
      "step": 24710
    },
    {
      "epoch": 7.016747090547828,
      "grad_norm": 10.196317672729492,
      "learning_rate": 9.298722679534488e-05,
      "loss": 0.07047267556190491,
      "step": 24720
    },
    {
      "epoch": 7.0195855804711895,
      "grad_norm": 10.404425621032715,
      "learning_rate": 9.298438830542152e-05,
      "loss": 0.09644900560379029,
      "step": 24730
    },
    {
      "epoch": 7.02242407039455,
      "grad_norm": 3.890166759490967,
      "learning_rate": 9.298154981549815e-05,
      "loss": 0.07013571858406067,
      "step": 24740
    },
    {
      "epoch": 7.025262560317911,
      "grad_norm": 6.217899799346924,
      "learning_rate": 9.29787113255748e-05,
      "loss": 0.06245779395103455,
      "step": 24750
    },
    {
      "epoch": 7.028101050241272,
      "grad_norm": 6.580312252044678,
      "learning_rate": 9.297587283565144e-05,
      "loss": 0.06836444139480591,
      "step": 24760
    },
    {
      "epoch": 7.030939540164632,
      "grad_norm": 6.5935516357421875,
      "learning_rate": 9.297303434572808e-05,
      "loss": 0.07499512434005737,
      "step": 24770
    },
    {
      "epoch": 7.033778030087993,
      "grad_norm": 5.788921356201172,
      "learning_rate": 9.297019585580472e-05,
      "loss": 0.09312869906425476,
      "step": 24780
    },
    {
      "epoch": 7.0366165200113535,
      "grad_norm": 2.3077564239501953,
      "learning_rate": 9.296735736588136e-05,
      "loss": 0.0823489785194397,
      "step": 24790
    },
    {
      "epoch": 7.039455009934715,
      "grad_norm": 15.5554838180542,
      "learning_rate": 9.2964518875958e-05,
      "loss": 0.08530140519142151,
      "step": 24800
    },
    {
      "epoch": 7.042293499858076,
      "grad_norm": 6.212428569793701,
      "learning_rate": 9.296168038603463e-05,
      "loss": 0.05804241299629211,
      "step": 24810
    },
    {
      "epoch": 7.045131989781436,
      "grad_norm": 22.636356353759766,
      "learning_rate": 9.295884189611127e-05,
      "loss": 0.09581531882286072,
      "step": 24820
    },
    {
      "epoch": 7.047970479704797,
      "grad_norm": 16.66684913635254,
      "learning_rate": 9.295600340618791e-05,
      "loss": 0.09299579858779908,
      "step": 24830
    },
    {
      "epoch": 7.050808969628158,
      "grad_norm": 15.202995300292969,
      "learning_rate": 9.295316491626455e-05,
      "loss": 0.0945294976234436,
      "step": 24840
    },
    {
      "epoch": 7.053647459551518,
      "grad_norm": 9.286093711853027,
      "learning_rate": 9.29503264263412e-05,
      "loss": 0.08990670442581176,
      "step": 24850
    },
    {
      "epoch": 7.0564859494748795,
      "grad_norm": 10.716316223144531,
      "learning_rate": 9.294748793641784e-05,
      "loss": 0.0868717074394226,
      "step": 24860
    },
    {
      "epoch": 7.05932443939824,
      "grad_norm": 14.390135765075684,
      "learning_rate": 9.294464944649446e-05,
      "loss": 0.06352605223655701,
      "step": 24870
    },
    {
      "epoch": 7.062162929321601,
      "grad_norm": 9.363687515258789,
      "learning_rate": 9.29418109565711e-05,
      "loss": 0.0662489116191864,
      "step": 24880
    },
    {
      "epoch": 7.065001419244962,
      "grad_norm": 16.993261337280273,
      "learning_rate": 9.293897246664775e-05,
      "loss": 0.0787426471710205,
      "step": 24890
    },
    {
      "epoch": 7.067839909168322,
      "grad_norm": 13.533523559570312,
      "learning_rate": 9.293613397672439e-05,
      "loss": 0.08902983665466309,
      "step": 24900
    },
    {
      "epoch": 7.070678399091683,
      "grad_norm": 13.833831787109375,
      "learning_rate": 9.293329548680103e-05,
      "loss": 0.09883632063865662,
      "step": 24910
    },
    {
      "epoch": 7.073516889015044,
      "grad_norm": 4.1236491203308105,
      "learning_rate": 9.293045699687767e-05,
      "loss": 0.06012184619903564,
      "step": 24920
    },
    {
      "epoch": 7.076355378938405,
      "grad_norm": 8.891768455505371,
      "learning_rate": 9.292761850695431e-05,
      "loss": 0.0735208511352539,
      "step": 24930
    },
    {
      "epoch": 7.079193868861766,
      "grad_norm": 15.003506660461426,
      "learning_rate": 9.292478001703094e-05,
      "loss": 0.07380845546722412,
      "step": 24940
    },
    {
      "epoch": 7.082032358785126,
      "grad_norm": 9.788448333740234,
      "learning_rate": 9.292194152710758e-05,
      "loss": 0.0688049852848053,
      "step": 24950
    },
    {
      "epoch": 7.084870848708487,
      "grad_norm": 5.41404914855957,
      "learning_rate": 9.291910303718422e-05,
      "loss": 0.08030549287796021,
      "step": 24960
    },
    {
      "epoch": 7.087709338631848,
      "grad_norm": 12.876459121704102,
      "learning_rate": 9.291626454726085e-05,
      "loss": 0.07650284171104431,
      "step": 24970
    },
    {
      "epoch": 7.090547828555208,
      "grad_norm": 9.432177543640137,
      "learning_rate": 9.29134260573375e-05,
      "loss": 0.06650417447090148,
      "step": 24980
    },
    {
      "epoch": 7.0933863184785695,
      "grad_norm": 8.356846809387207,
      "learning_rate": 9.291058756741415e-05,
      "loss": 0.07502039670944213,
      "step": 24990
    },
    {
      "epoch": 7.096224808401931,
      "grad_norm": 13.065507888793945,
      "learning_rate": 9.290774907749077e-05,
      "loss": 0.07929702997207641,
      "step": 25000
    },
    {
      "epoch": 7.096224808401931,
      "eval_accuracy": 0.9442360272143447,
      "eval_loss": 0.16748574376106262,
      "eval_runtime": 31.0544,
      "eval_samples_per_second": 506.433,
      "eval_steps_per_second": 7.922,
      "step": 25000
    },
    {
      "epoch": 7.099063298325291,
      "grad_norm": 16.9394588470459,
      "learning_rate": 9.290491058756742e-05,
      "loss": 0.06433024406433105,
      "step": 25010
    },
    {
      "epoch": 7.101901788248652,
      "grad_norm": 12.390459060668945,
      "learning_rate": 9.290207209764406e-05,
      "loss": 0.07576498985290528,
      "step": 25020
    },
    {
      "epoch": 7.104740278172012,
      "grad_norm": 12.055755615234375,
      "learning_rate": 9.28992336077207e-05,
      "loss": 0.1016196370124817,
      "step": 25030
    },
    {
      "epoch": 7.107578768095373,
      "grad_norm": 10.71986198425293,
      "learning_rate": 9.289639511779734e-05,
      "loss": 0.1038313627243042,
      "step": 25040
    },
    {
      "epoch": 7.110417258018734,
      "grad_norm": 10.531457901000977,
      "learning_rate": 9.289355662787398e-05,
      "loss": 0.07117497324943542,
      "step": 25050
    },
    {
      "epoch": 7.113255747942095,
      "grad_norm": 16.930017471313477,
      "learning_rate": 9.289071813795062e-05,
      "loss": 0.0703173041343689,
      "step": 25060
    },
    {
      "epoch": 7.116094237865456,
      "grad_norm": 21.771995544433594,
      "learning_rate": 9.288787964802725e-05,
      "loss": 0.09848902225494385,
      "step": 25070
    },
    {
      "epoch": 7.118932727788816,
      "grad_norm": 13.917851448059082,
      "learning_rate": 9.288504115810389e-05,
      "loss": 0.0688615083694458,
      "step": 25080
    },
    {
      "epoch": 7.121771217712177,
      "grad_norm": 18.24556541442871,
      "learning_rate": 9.288220266818053e-05,
      "loss": 0.10535664558410644,
      "step": 25090
    },
    {
      "epoch": 7.124609707635538,
      "grad_norm": 5.248446464538574,
      "learning_rate": 9.287936417825716e-05,
      "loss": 0.0779831051826477,
      "step": 25100
    },
    {
      "epoch": 7.127448197558898,
      "grad_norm": 9.42318058013916,
      "learning_rate": 9.287652568833382e-05,
      "loss": 0.08774685859680176,
      "step": 25110
    },
    {
      "epoch": 7.1302866874822595,
      "grad_norm": 13.35573673248291,
      "learning_rate": 9.287368719841046e-05,
      "loss": 0.0786705732345581,
      "step": 25120
    },
    {
      "epoch": 7.133125177405621,
      "grad_norm": 9.384979248046875,
      "learning_rate": 9.287084870848708e-05,
      "loss": 0.06326441168785095,
      "step": 25130
    },
    {
      "epoch": 7.135963667328981,
      "grad_norm": 9.870256423950195,
      "learning_rate": 9.286801021856373e-05,
      "loss": 0.08181899189949035,
      "step": 25140
    },
    {
      "epoch": 7.138802157252342,
      "grad_norm": 3.5451319217681885,
      "learning_rate": 9.286517172864037e-05,
      "loss": 0.060170650482177734,
      "step": 25150
    },
    {
      "epoch": 7.141640647175702,
      "grad_norm": 5.058857440948486,
      "learning_rate": 9.286233323871701e-05,
      "loss": 0.0686713457107544,
      "step": 25160
    },
    {
      "epoch": 7.144479137099063,
      "grad_norm": 14.10214900970459,
      "learning_rate": 9.285949474879365e-05,
      "loss": 0.08296550512313842,
      "step": 25170
    },
    {
      "epoch": 7.147317627022424,
      "grad_norm": 9.43695068359375,
      "learning_rate": 9.285665625887029e-05,
      "loss": 0.07093438506126404,
      "step": 25180
    },
    {
      "epoch": 7.150156116945785,
      "grad_norm": 9.93986988067627,
      "learning_rate": 9.285381776894693e-05,
      "loss": 0.08041448593139648,
      "step": 25190
    },
    {
      "epoch": 7.152994606869146,
      "grad_norm": 8.851746559143066,
      "learning_rate": 9.285097927902356e-05,
      "loss": 0.07922419905662537,
      "step": 25200
    },
    {
      "epoch": 7.155833096792507,
      "grad_norm": 6.401785850524902,
      "learning_rate": 9.28481407891002e-05,
      "loss": 0.06062203049659729,
      "step": 25210
    },
    {
      "epoch": 7.158671586715867,
      "grad_norm": 18.311893463134766,
      "learning_rate": 9.284530229917684e-05,
      "loss": 0.08666465282440186,
      "step": 25220
    },
    {
      "epoch": 7.161510076639228,
      "grad_norm": 11.923754692077637,
      "learning_rate": 9.284246380925347e-05,
      "loss": 0.06378366947174072,
      "step": 25230
    },
    {
      "epoch": 7.164348566562588,
      "grad_norm": 5.056766033172607,
      "learning_rate": 9.283962531933013e-05,
      "loss": 0.05328802466392517,
      "step": 25240
    },
    {
      "epoch": 7.1671870564859494,
      "grad_norm": 9.68974781036377,
      "learning_rate": 9.283678682940677e-05,
      "loss": 0.09437194466590881,
      "step": 25250
    },
    {
      "epoch": 7.170025546409311,
      "grad_norm": 7.316320896148682,
      "learning_rate": 9.28339483394834e-05,
      "loss": 0.06968687176704406,
      "step": 25260
    },
    {
      "epoch": 7.172864036332671,
      "grad_norm": 11.415355682373047,
      "learning_rate": 9.283110984956004e-05,
      "loss": 0.05861054658889771,
      "step": 25270
    },
    {
      "epoch": 7.175702526256032,
      "grad_norm": 7.921270847320557,
      "learning_rate": 9.282827135963668e-05,
      "loss": 0.07577463984489441,
      "step": 25280
    },
    {
      "epoch": 7.178541016179393,
      "grad_norm": 12.708183288574219,
      "learning_rate": 9.282543286971332e-05,
      "loss": 0.0669076383113861,
      "step": 25290
    },
    {
      "epoch": 7.181379506102753,
      "grad_norm": 11.058527946472168,
      "learning_rate": 9.282259437978995e-05,
      "loss": 0.07173556685447693,
      "step": 25300
    },
    {
      "epoch": 7.184217996026114,
      "grad_norm": 4.351419925689697,
      "learning_rate": 9.28197558898666e-05,
      "loss": 0.08426600694656372,
      "step": 25310
    },
    {
      "epoch": 7.1870564859494745,
      "grad_norm": 11.54439926147461,
      "learning_rate": 9.281691739994324e-05,
      "loss": 0.066134774684906,
      "step": 25320
    },
    {
      "epoch": 7.189894975872836,
      "grad_norm": 16.16973114013672,
      "learning_rate": 9.281407891001987e-05,
      "loss": 0.08736181855201722,
      "step": 25330
    },
    {
      "epoch": 7.192733465796197,
      "grad_norm": 8.346450805664062,
      "learning_rate": 9.281124042009651e-05,
      "loss": 0.06326338052749633,
      "step": 25340
    },
    {
      "epoch": 7.195571955719557,
      "grad_norm": 3.7849199771881104,
      "learning_rate": 9.280840193017315e-05,
      "loss": 0.08217431902885437,
      "step": 25350
    },
    {
      "epoch": 7.198410445642918,
      "grad_norm": 9.105610847473145,
      "learning_rate": 9.280556344024978e-05,
      "loss": 0.09630203247070312,
      "step": 25360
    },
    {
      "epoch": 7.201248935566278,
      "grad_norm": 20.390913009643555,
      "learning_rate": 9.280272495032644e-05,
      "loss": 0.10159130096435547,
      "step": 25370
    },
    {
      "epoch": 7.204087425489639,
      "grad_norm": 10.270111083984375,
      "learning_rate": 9.279988646040308e-05,
      "loss": 0.073513263463974,
      "step": 25380
    },
    {
      "epoch": 7.2069259154130005,
      "grad_norm": 12.678959846496582,
      "learning_rate": 9.27970479704797e-05,
      "loss": 0.059249311685562134,
      "step": 25390
    },
    {
      "epoch": 7.209764405336361,
      "grad_norm": 22.306011199951172,
      "learning_rate": 9.279420948055635e-05,
      "loss": 0.08024435043334961,
      "step": 25400
    },
    {
      "epoch": 7.212602895259722,
      "grad_norm": 4.443155288696289,
      "learning_rate": 9.279137099063299e-05,
      "loss": 0.07389181852340698,
      "step": 25410
    },
    {
      "epoch": 7.215441385183083,
      "grad_norm": 5.50084114074707,
      "learning_rate": 9.278853250070963e-05,
      "loss": 0.07785683870315552,
      "step": 25420
    },
    {
      "epoch": 7.218279875106443,
      "grad_norm": 8.419054985046387,
      "learning_rate": 9.278569401078626e-05,
      "loss": 0.08136615753173829,
      "step": 25430
    },
    {
      "epoch": 7.221118365029804,
      "grad_norm": 8.237689018249512,
      "learning_rate": 9.278285552086291e-05,
      "loss": 0.05619444847106934,
      "step": 25440
    },
    {
      "epoch": 7.2239568549531645,
      "grad_norm": 8.519165992736816,
      "learning_rate": 9.278001703093954e-05,
      "loss": 0.05788261890411377,
      "step": 25450
    },
    {
      "epoch": 7.226795344876526,
      "grad_norm": 11.493070602416992,
      "learning_rate": 9.277717854101618e-05,
      "loss": 0.08106710910797119,
      "step": 25460
    },
    {
      "epoch": 7.229633834799887,
      "grad_norm": 10.802931785583496,
      "learning_rate": 9.277434005109282e-05,
      "loss": 0.05255587697029114,
      "step": 25470
    },
    {
      "epoch": 7.232472324723247,
      "grad_norm": 10.36487865447998,
      "learning_rate": 9.277150156116947e-05,
      "loss": 0.10177537202835082,
      "step": 25480
    },
    {
      "epoch": 7.235310814646608,
      "grad_norm": 6.762228965759277,
      "learning_rate": 9.276866307124609e-05,
      "loss": 0.0843475341796875,
      "step": 25490
    },
    {
      "epoch": 7.238149304569969,
      "grad_norm": 12.662060737609863,
      "learning_rate": 9.276582458132273e-05,
      "loss": 0.0893014907836914,
      "step": 25500
    },
    {
      "epoch": 7.238149304569969,
      "eval_accuracy": 0.9397850829783175,
      "eval_loss": 0.1832467019557953,
      "eval_runtime": 31.3095,
      "eval_samples_per_second": 502.308,
      "eval_steps_per_second": 7.857,
      "step": 25500
    },
    {
      "epoch": 7.240987794493329,
      "grad_norm": 8.080511093139648,
      "learning_rate": 9.276298609139939e-05,
      "loss": 0.05834176540374756,
      "step": 25510
    },
    {
      "epoch": 7.2438262844166905,
      "grad_norm": 6.830284118652344,
      "learning_rate": 9.276014760147602e-05,
      "loss": 0.06360492706298829,
      "step": 25520
    },
    {
      "epoch": 7.246664774340051,
      "grad_norm": 17.548480987548828,
      "learning_rate": 9.275730911155266e-05,
      "loss": 0.0802126407623291,
      "step": 25530
    },
    {
      "epoch": 7.249503264263412,
      "grad_norm": 9.045568466186523,
      "learning_rate": 9.27544706216293e-05,
      "loss": 0.07861078977584839,
      "step": 25540
    },
    {
      "epoch": 7.252341754186773,
      "grad_norm": 7.015875816345215,
      "learning_rate": 9.275163213170593e-05,
      "loss": 0.09540387988090515,
      "step": 25550
    },
    {
      "epoch": 7.255180244110133,
      "grad_norm": 3.82190203666687,
      "learning_rate": 9.274879364178257e-05,
      "loss": 0.06535564661026001,
      "step": 25560
    },
    {
      "epoch": 7.258018734033494,
      "grad_norm": 9.843457221984863,
      "learning_rate": 9.274595515185922e-05,
      "loss": 0.0892251968383789,
      "step": 25570
    },
    {
      "epoch": 7.2608572239568545,
      "grad_norm": 9.680808067321777,
      "learning_rate": 9.274311666193585e-05,
      "loss": 0.0860748529434204,
      "step": 25580
    },
    {
      "epoch": 7.263695713880216,
      "grad_norm": 2.475742816925049,
      "learning_rate": 9.27402781720125e-05,
      "loss": 0.08024742007255554,
      "step": 25590
    },
    {
      "epoch": 7.266534203803577,
      "grad_norm": 6.437875747680664,
      "learning_rate": 9.273743968208913e-05,
      "loss": 0.09326879382133484,
      "step": 25600
    },
    {
      "epoch": 7.269372693726937,
      "grad_norm": 7.342555999755859,
      "learning_rate": 9.273460119216578e-05,
      "loss": 0.10031281709671021,
      "step": 25610
    },
    {
      "epoch": 7.272211183650298,
      "grad_norm": 6.315887451171875,
      "learning_rate": 9.27317627022424e-05,
      "loss": 0.09980835318565369,
      "step": 25620
    },
    {
      "epoch": 7.275049673573659,
      "grad_norm": 12.160181045532227,
      "learning_rate": 9.272892421231905e-05,
      "loss": 0.07057795524597169,
      "step": 25630
    },
    {
      "epoch": 7.277888163497019,
      "grad_norm": 10.519721984863281,
      "learning_rate": 9.27260857223957e-05,
      "loss": 0.09350953102111817,
      "step": 25640
    },
    {
      "epoch": 7.2807266534203805,
      "grad_norm": 8.615524291992188,
      "learning_rate": 9.272324723247233e-05,
      "loss": 0.08921960592269898,
      "step": 25650
    },
    {
      "epoch": 7.283565143343741,
      "grad_norm": 14.594928741455078,
      "learning_rate": 9.272040874254897e-05,
      "loss": 0.07720339298248291,
      "step": 25660
    },
    {
      "epoch": 7.286403633267102,
      "grad_norm": 2.612436532974243,
      "learning_rate": 9.271757025262561e-05,
      "loss": 0.059430062770843506,
      "step": 25670
    },
    {
      "epoch": 7.289242123190463,
      "grad_norm": 12.778146743774414,
      "learning_rate": 9.271473176270224e-05,
      "loss": 0.054558998346328734,
      "step": 25680
    },
    {
      "epoch": 7.292080613113823,
      "grad_norm": 10.291651725769043,
      "learning_rate": 9.271189327277888e-05,
      "loss": 0.08524532318115234,
      "step": 25690
    },
    {
      "epoch": 7.294919103037184,
      "grad_norm": 3.862149715423584,
      "learning_rate": 9.270905478285552e-05,
      "loss": 0.08555809259414673,
      "step": 25700
    },
    {
      "epoch": 7.297757592960545,
      "grad_norm": 18.30660057067871,
      "learning_rate": 9.270621629293216e-05,
      "loss": 0.07146743535995484,
      "step": 25710
    },
    {
      "epoch": 7.300596082883906,
      "grad_norm": 19.888212203979492,
      "learning_rate": 9.27033778030088e-05,
      "loss": 0.08293269872665406,
      "step": 25720
    },
    {
      "epoch": 7.303434572807267,
      "grad_norm": 17.063730239868164,
      "learning_rate": 9.270053931308545e-05,
      "loss": 0.07968592047691345,
      "step": 25730
    },
    {
      "epoch": 7.306273062730627,
      "grad_norm": 2.37996244430542,
      "learning_rate": 9.269770082316209e-05,
      "loss": 0.06406975388526917,
      "step": 25740
    },
    {
      "epoch": 7.309111552653988,
      "grad_norm": 4.29535436630249,
      "learning_rate": 9.269486233323871e-05,
      "loss": 0.05949164628982544,
      "step": 25750
    },
    {
      "epoch": 7.311950042577349,
      "grad_norm": 7.4490251541137695,
      "learning_rate": 9.269202384331536e-05,
      "loss": 0.04965499639511108,
      "step": 25760
    },
    {
      "epoch": 7.314788532500709,
      "grad_norm": 8.656577110290527,
      "learning_rate": 9.268918535339201e-05,
      "loss": 0.10513595342636109,
      "step": 25770
    },
    {
      "epoch": 7.3176270224240705,
      "grad_norm": 4.1642374992370605,
      "learning_rate": 9.268634686346864e-05,
      "loss": 0.06156970858573914,
      "step": 25780
    },
    {
      "epoch": 7.320465512347431,
      "grad_norm": 12.691108703613281,
      "learning_rate": 9.268350837354528e-05,
      "loss": 0.08913788199424744,
      "step": 25790
    },
    {
      "epoch": 7.323304002270792,
      "grad_norm": 14.294926643371582,
      "learning_rate": 9.268066988362192e-05,
      "loss": 0.11012557744979859,
      "step": 25800
    },
    {
      "epoch": 7.326142492194153,
      "grad_norm": 8.799478530883789,
      "learning_rate": 9.267783139369855e-05,
      "loss": 0.07678711414337158,
      "step": 25810
    },
    {
      "epoch": 7.328980982117513,
      "grad_norm": 6.116623878479004,
      "learning_rate": 9.267499290377519e-05,
      "loss": 0.06817195415496827,
      "step": 25820
    },
    {
      "epoch": 7.331819472040874,
      "grad_norm": 16.788000106811523,
      "learning_rate": 9.267215441385183e-05,
      "loss": 0.07637510299682618,
      "step": 25830
    },
    {
      "epoch": 7.334657961964235,
      "grad_norm": 12.286406517028809,
      "learning_rate": 9.266931592392847e-05,
      "loss": 0.08467528223991394,
      "step": 25840
    },
    {
      "epoch": 7.337496451887596,
      "grad_norm": 10.662586212158203,
      "learning_rate": 9.266647743400511e-05,
      "loss": 0.10300173759460449,
      "step": 25850
    },
    {
      "epoch": 7.340334941810957,
      "grad_norm": 7.952162265777588,
      "learning_rate": 9.266363894408176e-05,
      "loss": 0.0618781566619873,
      "step": 25860
    },
    {
      "epoch": 7.343173431734318,
      "grad_norm": 12.708575248718262,
      "learning_rate": 9.26608004541584e-05,
      "loss": 0.07204200029373169,
      "step": 25870
    },
    {
      "epoch": 7.346011921657678,
      "grad_norm": 3.645596504211426,
      "learning_rate": 9.265796196423503e-05,
      "loss": 0.0725257933139801,
      "step": 25880
    },
    {
      "epoch": 7.348850411581039,
      "grad_norm": 7.456458568572998,
      "learning_rate": 9.265512347431167e-05,
      "loss": 0.0889801800251007,
      "step": 25890
    },
    {
      "epoch": 7.351688901504399,
      "grad_norm": 9.785014152526855,
      "learning_rate": 9.265228498438831e-05,
      "loss": 0.06083427667617798,
      "step": 25900
    },
    {
      "epoch": 7.3545273914277605,
      "grad_norm": 7.05605936050415,
      "learning_rate": 9.264944649446495e-05,
      "loss": 0.07014908790588378,
      "step": 25910
    },
    {
      "epoch": 7.357365881351122,
      "grad_norm": 6.651464939117432,
      "learning_rate": 9.264660800454159e-05,
      "loss": 0.11011708974838257,
      "step": 25920
    },
    {
      "epoch": 7.360204371274482,
      "grad_norm": 10.310638427734375,
      "learning_rate": 9.264376951461823e-05,
      "loss": 0.06599188446998597,
      "step": 25930
    },
    {
      "epoch": 7.363042861197843,
      "grad_norm": 15.391011238098145,
      "learning_rate": 9.264093102469486e-05,
      "loss": 0.08896167874336243,
      "step": 25940
    },
    {
      "epoch": 7.365881351121203,
      "grad_norm": 9.73863410949707,
      "learning_rate": 9.26380925347715e-05,
      "loss": 0.07937895059585572,
      "step": 25950
    },
    {
      "epoch": 7.368719841044564,
      "grad_norm": 9.990856170654297,
      "learning_rate": 9.263525404484814e-05,
      "loss": 0.09625785946846008,
      "step": 25960
    },
    {
      "epoch": 7.371558330967925,
      "grad_norm": 10.198019981384277,
      "learning_rate": 9.263241555492478e-05,
      "loss": 0.07414093613624573,
      "step": 25970
    },
    {
      "epoch": 7.374396820891286,
      "grad_norm": 4.457877159118652,
      "learning_rate": 9.262957706500143e-05,
      "loss": 0.06367641687393188,
      "step": 25980
    },
    {
      "epoch": 7.377235310814647,
      "grad_norm": 16.091041564941406,
      "learning_rate": 9.262673857507807e-05,
      "loss": 0.08975282311439514,
      "step": 25990
    },
    {
      "epoch": 7.380073800738008,
      "grad_norm": 9.023719787597656,
      "learning_rate": 9.262390008515471e-05,
      "loss": 0.08124292492866517,
      "step": 26000
    },
    {
      "epoch": 7.380073800738008,
      "eval_accuracy": 0.9362879125071533,
      "eval_loss": 0.1925540417432785,
      "eval_runtime": 31.0429,
      "eval_samples_per_second": 506.621,
      "eval_steps_per_second": 7.925,
      "step": 26000
    },
    {
      "epoch": 7.382912290661368,
      "grad_norm": 7.7657036781311035,
      "learning_rate": 9.262106159523134e-05,
      "loss": 0.0755397617816925,
      "step": 26010
    },
    {
      "epoch": 7.385750780584729,
      "grad_norm": 12.042271614074707,
      "learning_rate": 9.261822310530798e-05,
      "loss": 0.08097591400146484,
      "step": 26020
    },
    {
      "epoch": 7.388589270508089,
      "grad_norm": 2.8813228607177734,
      "learning_rate": 9.261538461538462e-05,
      "loss": 0.0630192518234253,
      "step": 26030
    },
    {
      "epoch": 7.3914277604314504,
      "grad_norm": 16.896039962768555,
      "learning_rate": 9.261254612546126e-05,
      "loss": 0.0758962631225586,
      "step": 26040
    },
    {
      "epoch": 7.3942662503548116,
      "grad_norm": 11.112716674804688,
      "learning_rate": 9.26097076355379e-05,
      "loss": 0.09962148070335389,
      "step": 26050
    },
    {
      "epoch": 7.397104740278172,
      "grad_norm": 9.072155952453613,
      "learning_rate": 9.260686914561454e-05,
      "loss": 0.08826958537101745,
      "step": 26060
    },
    {
      "epoch": 7.399943230201533,
      "grad_norm": 11.082853317260742,
      "learning_rate": 9.260403065569117e-05,
      "loss": 0.09242720603942871,
      "step": 26070
    },
    {
      "epoch": 7.402781720124894,
      "grad_norm": 9.824169158935547,
      "learning_rate": 9.260119216576781e-05,
      "loss": 0.08743641376495362,
      "step": 26080
    },
    {
      "epoch": 7.405620210048254,
      "grad_norm": 4.256606578826904,
      "learning_rate": 9.259835367584445e-05,
      "loss": 0.06762970089912415,
      "step": 26090
    },
    {
      "epoch": 7.408458699971615,
      "grad_norm": 8.099950790405273,
      "learning_rate": 9.25955151859211e-05,
      "loss": 0.09003919959068299,
      "step": 26100
    },
    {
      "epoch": 7.4112971898949755,
      "grad_norm": 5.497006893157959,
      "learning_rate": 9.259267669599774e-05,
      "loss": 0.06235863566398621,
      "step": 26110
    },
    {
      "epoch": 7.414135679818337,
      "grad_norm": 8.308137893676758,
      "learning_rate": 9.258983820607438e-05,
      "loss": 0.05843211412429809,
      "step": 26120
    },
    {
      "epoch": 7.416974169741698,
      "grad_norm": 8.485076904296875,
      "learning_rate": 9.258699971615102e-05,
      "loss": 0.08134673237800598,
      "step": 26130
    },
    {
      "epoch": 7.419812659665058,
      "grad_norm": 12.66151237487793,
      "learning_rate": 9.258416122622765e-05,
      "loss": 0.07153877019882202,
      "step": 26140
    },
    {
      "epoch": 7.422651149588419,
      "grad_norm": 9.781754493713379,
      "learning_rate": 9.258132273630429e-05,
      "loss": 0.08094428777694702,
      "step": 26150
    },
    {
      "epoch": 7.425489639511779,
      "grad_norm": 11.617583274841309,
      "learning_rate": 9.257848424638093e-05,
      "loss": 0.08042703866958618,
      "step": 26160
    },
    {
      "epoch": 7.42832812943514,
      "grad_norm": 4.851502895355225,
      "learning_rate": 9.257564575645757e-05,
      "loss": 0.07313272953033448,
      "step": 26170
    },
    {
      "epoch": 7.4311666193585015,
      "grad_norm": 12.004612922668457,
      "learning_rate": 9.257280726653421e-05,
      "loss": 0.06990544199943542,
      "step": 26180
    },
    {
      "epoch": 7.434005109281862,
      "grad_norm": 8.36447525024414,
      "learning_rate": 9.256996877661085e-05,
      "loss": 0.06553784012794495,
      "step": 26190
    },
    {
      "epoch": 7.436843599205223,
      "grad_norm": 11.0001802444458,
      "learning_rate": 9.256713028668748e-05,
      "loss": 0.10173889398574829,
      "step": 26200
    },
    {
      "epoch": 7.439682089128584,
      "grad_norm": 10.433905601501465,
      "learning_rate": 9.256429179676412e-05,
      "loss": 0.11588386297225953,
      "step": 26210
    },
    {
      "epoch": 7.442520579051944,
      "grad_norm": 8.971481323242188,
      "learning_rate": 9.256145330684076e-05,
      "loss": 0.06656346321105958,
      "step": 26220
    },
    {
      "epoch": 7.445359068975305,
      "grad_norm": 10.237892150878906,
      "learning_rate": 9.25586148169174e-05,
      "loss": 0.09673588871955871,
      "step": 26230
    },
    {
      "epoch": 7.4481975588986655,
      "grad_norm": 7.237141132354736,
      "learning_rate": 9.255577632699405e-05,
      "loss": 0.11670624017715454,
      "step": 26240
    },
    {
      "epoch": 7.451036048822027,
      "grad_norm": 8.963961601257324,
      "learning_rate": 9.255293783707069e-05,
      "loss": 0.0728587567806244,
      "step": 26250
    },
    {
      "epoch": 7.453874538745388,
      "grad_norm": 9.15424633026123,
      "learning_rate": 9.255009934714733e-05,
      "loss": 0.07129738330841065,
      "step": 26260
    },
    {
      "epoch": 7.456713028668748,
      "grad_norm": 23.21555519104004,
      "learning_rate": 9.254726085722396e-05,
      "loss": 0.11262639760971069,
      "step": 26270
    },
    {
      "epoch": 7.459551518592109,
      "grad_norm": 8.273276329040527,
      "learning_rate": 9.25444223673006e-05,
      "loss": 0.11315351724624634,
      "step": 26280
    },
    {
      "epoch": 7.46239000851547,
      "grad_norm": 10.806933403015137,
      "learning_rate": 9.254158387737724e-05,
      "loss": 0.08202406167984008,
      "step": 26290
    },
    {
      "epoch": 7.46522849843883,
      "grad_norm": 13.437762260437012,
      "learning_rate": 9.253874538745387e-05,
      "loss": 0.07173805236816407,
      "step": 26300
    },
    {
      "epoch": 7.4680669883621915,
      "grad_norm": 8.283321380615234,
      "learning_rate": 9.253590689753052e-05,
      "loss": 0.06738618612289429,
      "step": 26310
    },
    {
      "epoch": 7.470905478285552,
      "grad_norm": 6.9814300537109375,
      "learning_rate": 9.253306840760716e-05,
      "loss": 0.07975713014602662,
      "step": 26320
    },
    {
      "epoch": 7.473743968208913,
      "grad_norm": 11.191336631774902,
      "learning_rate": 9.253022991768379e-05,
      "loss": 0.08927326202392578,
      "step": 26330
    },
    {
      "epoch": 7.476582458132274,
      "grad_norm": 8.341598510742188,
      "learning_rate": 9.252739142776043e-05,
      "loss": 0.0828757643699646,
      "step": 26340
    },
    {
      "epoch": 7.479420948055634,
      "grad_norm": 11.498359680175781,
      "learning_rate": 9.252455293783708e-05,
      "loss": 0.08253850936889648,
      "step": 26350
    },
    {
      "epoch": 7.482259437978995,
      "grad_norm": 18.451519012451172,
      "learning_rate": 9.252171444791372e-05,
      "loss": 0.07351793646812439,
      "step": 26360
    },
    {
      "epoch": 7.4850979279023555,
      "grad_norm": 6.21812105178833,
      "learning_rate": 9.251887595799036e-05,
      "loss": 0.05628350377082825,
      "step": 26370
    },
    {
      "epoch": 7.487936417825717,
      "grad_norm": 7.223512172698975,
      "learning_rate": 9.2516037468067e-05,
      "loss": 0.07334859371185302,
      "step": 26380
    },
    {
      "epoch": 7.490774907749078,
      "grad_norm": 10.63648509979248,
      "learning_rate": 9.251319897814363e-05,
      "loss": 0.07936392426490783,
      "step": 26390
    },
    {
      "epoch": 7.493613397672438,
      "grad_norm": 2.5491249561309814,
      "learning_rate": 9.251036048822027e-05,
      "loss": 0.07642608284950256,
      "step": 26400
    },
    {
      "epoch": 7.496451887595799,
      "grad_norm": 9.710280418395996,
      "learning_rate": 9.250752199829691e-05,
      "loss": 0.09321855306625366,
      "step": 26410
    },
    {
      "epoch": 7.49929037751916,
      "grad_norm": 5.702704429626465,
      "learning_rate": 9.250468350837355e-05,
      "loss": 0.06718577146530151,
      "step": 26420
    },
    {
      "epoch": 7.50212886744252,
      "grad_norm": 7.61082649230957,
      "learning_rate": 9.250184501845018e-05,
      "loss": 0.07724987268447876,
      "step": 26430
    },
    {
      "epoch": 7.5049673573658815,
      "grad_norm": 3.6927883625030518,
      "learning_rate": 9.249900652852683e-05,
      "loss": 0.0692366361618042,
      "step": 26440
    },
    {
      "epoch": 7.507805847289243,
      "grad_norm": 6.430363178253174,
      "learning_rate": 9.249616803860348e-05,
      "loss": 0.07843012213706971,
      "step": 26450
    },
    {
      "epoch": 7.510644337212603,
      "grad_norm": 11.592170715332031,
      "learning_rate": 9.24933295486801e-05,
      "loss": 0.08535722494125367,
      "step": 26460
    },
    {
      "epoch": 7.513482827135964,
      "grad_norm": 12.934331893920898,
      "learning_rate": 9.249049105875674e-05,
      "loss": 0.0967501938343048,
      "step": 26470
    },
    {
      "epoch": 7.516321317059324,
      "grad_norm": 8.168246269226074,
      "learning_rate": 9.248765256883339e-05,
      "loss": 0.057279002666473386,
      "step": 26480
    },
    {
      "epoch": 7.519159806982685,
      "grad_norm": 6.954280376434326,
      "learning_rate": 9.248481407891001e-05,
      "loss": 0.06227178573608398,
      "step": 26490
    },
    {
      "epoch": 7.521998296906046,
      "grad_norm": 9.486580848693848,
      "learning_rate": 9.248197558898667e-05,
      "loss": 0.09544625282287597,
      "step": 26500
    },
    {
      "epoch": 7.521998296906046,
      "eval_accuracy": 0.9319641381064412,
      "eval_loss": 0.19461214542388916,
      "eval_runtime": 31.5747,
      "eval_samples_per_second": 498.088,
      "eval_steps_per_second": 7.791,
      "step": 26500
    },
    {
      "epoch": 7.524836786829407,
      "grad_norm": 12.489873886108398,
      "learning_rate": 9.247913709906331e-05,
      "loss": 0.09371354579925537,
      "step": 26510
    },
    {
      "epoch": 7.527675276752768,
      "grad_norm": 7.647334575653076,
      "learning_rate": 9.247629860913994e-05,
      "loss": 0.07879270315170288,
      "step": 26520
    },
    {
      "epoch": 7.530513766676128,
      "grad_norm": 9.045136451721191,
      "learning_rate": 9.247346011921658e-05,
      "loss": 0.08057585954666138,
      "step": 26530
    },
    {
      "epoch": 7.533352256599489,
      "grad_norm": 12.849018096923828,
      "learning_rate": 9.247062162929322e-05,
      "loss": 0.08601323962211609,
      "step": 26540
    },
    {
      "epoch": 7.53619074652285,
      "grad_norm": 9.870349884033203,
      "learning_rate": 9.246778313936986e-05,
      "loss": 0.08743392825126647,
      "step": 26550
    },
    {
      "epoch": 7.53902923644621,
      "grad_norm": 3.759732723236084,
      "learning_rate": 9.246494464944649e-05,
      "loss": 0.08128655552864075,
      "step": 26560
    },
    {
      "epoch": 7.5418677263695715,
      "grad_norm": 8.3211088180542,
      "learning_rate": 9.246210615952314e-05,
      "loss": 0.056488925218582155,
      "step": 26570
    },
    {
      "epoch": 7.544706216292932,
      "grad_norm": 16.501707077026367,
      "learning_rate": 9.245926766959979e-05,
      "loss": 0.05979335308074951,
      "step": 26580
    },
    {
      "epoch": 7.547544706216293,
      "grad_norm": 16.16169548034668,
      "learning_rate": 9.245642917967641e-05,
      "loss": 0.07142907381057739,
      "step": 26590
    },
    {
      "epoch": 7.550383196139654,
      "grad_norm": 16.38370704650879,
      "learning_rate": 9.245359068975306e-05,
      "loss": 0.09806442260742188,
      "step": 26600
    },
    {
      "epoch": 7.553221686063014,
      "grad_norm": 13.115726470947266,
      "learning_rate": 9.24507521998297e-05,
      "loss": 0.09101833701133728,
      "step": 26610
    },
    {
      "epoch": 7.556060175986375,
      "grad_norm": 11.214066505432129,
      "learning_rate": 9.244791370990632e-05,
      "loss": 0.09396579265594482,
      "step": 26620
    },
    {
      "epoch": 7.558898665909736,
      "grad_norm": 9.216989517211914,
      "learning_rate": 9.244507521998297e-05,
      "loss": 0.06569300889968872,
      "step": 26630
    },
    {
      "epoch": 7.561737155833097,
      "grad_norm": 10.804996490478516,
      "learning_rate": 9.244223673005962e-05,
      "loss": 0.09532834887504578,
      "step": 26640
    },
    {
      "epoch": 7.564575645756458,
      "grad_norm": 17.074663162231445,
      "learning_rate": 9.243939824013625e-05,
      "loss": 0.05799614787101746,
      "step": 26650
    },
    {
      "epoch": 7.567414135679819,
      "grad_norm": 5.555571556091309,
      "learning_rate": 9.243655975021289e-05,
      "loss": 0.055042868852615355,
      "step": 26660
    },
    {
      "epoch": 7.570252625603179,
      "grad_norm": 17.987028121948242,
      "learning_rate": 9.243372126028953e-05,
      "loss": 0.09240897893905639,
      "step": 26670
    },
    {
      "epoch": 7.57309111552654,
      "grad_norm": 2.771324634552002,
      "learning_rate": 9.243088277036617e-05,
      "loss": 0.07774472236633301,
      "step": 26680
    },
    {
      "epoch": 7.5759296054499,
      "grad_norm": 15.184845924377441,
      "learning_rate": 9.24280442804428e-05,
      "loss": 0.0839342713356018,
      "step": 26690
    },
    {
      "epoch": 7.5787680953732615,
      "grad_norm": 11.460173606872559,
      "learning_rate": 9.242520579051946e-05,
      "loss": 0.07380360960960389,
      "step": 26700
    },
    {
      "epoch": 7.581606585296623,
      "grad_norm": 1.1493022441864014,
      "learning_rate": 9.24223673005961e-05,
      "loss": 0.07841793298721314,
      "step": 26710
    },
    {
      "epoch": 7.584445075219983,
      "grad_norm": 4.56388521194458,
      "learning_rate": 9.241952881067272e-05,
      "loss": 0.06686785221099853,
      "step": 26720
    },
    {
      "epoch": 7.587283565143344,
      "grad_norm": 4.601447582244873,
      "learning_rate": 9.241669032074937e-05,
      "loss": 0.06081027388572693,
      "step": 26730
    },
    {
      "epoch": 7.590122055066704,
      "grad_norm": 6.164660930633545,
      "learning_rate": 9.241385183082601e-05,
      "loss": 0.06688077449798584,
      "step": 26740
    },
    {
      "epoch": 7.592960544990065,
      "grad_norm": 11.12303638458252,
      "learning_rate": 9.241101334090264e-05,
      "loss": 0.08214783072471618,
      "step": 26750
    },
    {
      "epoch": 7.595799034913426,
      "grad_norm": 16.310251235961914,
      "learning_rate": 9.240817485097928e-05,
      "loss": 0.13177411556243895,
      "step": 26760
    },
    {
      "epoch": 7.598637524836787,
      "grad_norm": 9.675166130065918,
      "learning_rate": 9.240533636105593e-05,
      "loss": 0.055321496725082395,
      "step": 26770
    },
    {
      "epoch": 7.601476014760148,
      "grad_norm": 6.835315704345703,
      "learning_rate": 9.24027817201249e-05,
      "loss": 0.09739679098129272,
      "step": 26780
    },
    {
      "epoch": 7.604314504683508,
      "grad_norm": 5.439204216003418,
      "learning_rate": 9.239994323020154e-05,
      "loss": 0.06536415219306946,
      "step": 26790
    },
    {
      "epoch": 7.607152994606869,
      "grad_norm": 9.276590347290039,
      "learning_rate": 9.239710474027817e-05,
      "loss": 0.07503811717033386,
      "step": 26800
    },
    {
      "epoch": 7.60999148453023,
      "grad_norm": 9.135767936706543,
      "learning_rate": 9.239426625035481e-05,
      "loss": 0.0628929615020752,
      "step": 26810
    },
    {
      "epoch": 7.61282997445359,
      "grad_norm": 1.528260588645935,
      "learning_rate": 9.239142776043146e-05,
      "loss": 0.06634442210197448,
      "step": 26820
    },
    {
      "epoch": 7.615668464376951,
      "grad_norm": 6.177563190460205,
      "learning_rate": 9.238858927050809e-05,
      "loss": 0.07088123559951783,
      "step": 26830
    },
    {
      "epoch": 7.6185069543003126,
      "grad_norm": 1.8341660499572754,
      "learning_rate": 9.238575078058473e-05,
      "loss": 0.09534658789634705,
      "step": 26840
    },
    {
      "epoch": 7.621345444223673,
      "grad_norm": 13.950006484985352,
      "learning_rate": 9.238291229066137e-05,
      "loss": 0.10807677507400512,
      "step": 26850
    },
    {
      "epoch": 7.624183934147034,
      "grad_norm": 12.056605339050293,
      "learning_rate": 9.238007380073802e-05,
      "loss": 0.0799710988998413,
      "step": 26860
    },
    {
      "epoch": 7.627022424070395,
      "grad_norm": 5.1240153312683105,
      "learning_rate": 9.237723531081464e-05,
      "loss": 0.08616395592689514,
      "step": 26870
    },
    {
      "epoch": 7.629860913993755,
      "grad_norm": 8.71175479888916,
      "learning_rate": 9.23743968208913e-05,
      "loss": 0.07502010464668274,
      "step": 26880
    },
    {
      "epoch": 7.632699403917116,
      "grad_norm": 12.677748680114746,
      "learning_rate": 9.237155833096794e-05,
      "loss": 0.0809528648853302,
      "step": 26890
    },
    {
      "epoch": 7.6355378938404765,
      "grad_norm": 8.12935733795166,
      "learning_rate": 9.236871984104457e-05,
      "loss": 0.11166051626205445,
      "step": 26900
    },
    {
      "epoch": 7.638376383763838,
      "grad_norm": 8.465171813964844,
      "learning_rate": 9.236588135112121e-05,
      "loss": 0.07791109681129456,
      "step": 26910
    },
    {
      "epoch": 7.641214873687199,
      "grad_norm": 5.829082489013672,
      "learning_rate": 9.236304286119785e-05,
      "loss": 0.05595495700836182,
      "step": 26920
    },
    {
      "epoch": 7.644053363610559,
      "grad_norm": 8.654165267944336,
      "learning_rate": 9.236020437127448e-05,
      "loss": 0.08592514991760254,
      "step": 26930
    },
    {
      "epoch": 7.64689185353392,
      "grad_norm": 7.819706916809082,
      "learning_rate": 9.235736588135112e-05,
      "loss": 0.05950862169265747,
      "step": 26940
    },
    {
      "epoch": 7.64973034345728,
      "grad_norm": 11.006224632263184,
      "learning_rate": 9.235452739142777e-05,
      "loss": 0.06624175906181336,
      "step": 26950
    },
    {
      "epoch": 7.652568833380641,
      "grad_norm": 13.52245807647705,
      "learning_rate": 9.23516889015044e-05,
      "loss": 0.0643926978111267,
      "step": 26960
    },
    {
      "epoch": 7.6554073233040025,
      "grad_norm": 8.29045295715332,
      "learning_rate": 9.234885041158104e-05,
      "loss": 0.09127501845359802,
      "step": 26970
    },
    {
      "epoch": 7.658245813227363,
      "grad_norm": 6.466549873352051,
      "learning_rate": 9.234601192165768e-05,
      "loss": 0.08462011814117432,
      "step": 26980
    },
    {
      "epoch": 7.661084303150724,
      "grad_norm": 10.676262855529785,
      "learning_rate": 9.234317343173433e-05,
      "loss": 0.08582854270935059,
      "step": 26990
    },
    {
      "epoch": 7.663922793074085,
      "grad_norm": 10.218280792236328,
      "learning_rate": 9.234033494181095e-05,
      "loss": 0.06660287976264953,
      "step": 27000
    },
    {
      "epoch": 7.663922793074085,
      "eval_accuracy": 0.9253513066700578,
      "eval_loss": 0.21773947775363922,
      "eval_runtime": 31.841,
      "eval_samples_per_second": 493.923,
      "eval_steps_per_second": 7.726,
      "step": 27000
    },
    {
      "epoch": 7.666761282997445,
      "grad_norm": 10.956585884094238,
      "learning_rate": 9.23374964518876e-05,
      "loss": 0.07425179481506347,
      "step": 27010
    },
    {
      "epoch": 7.669599772920806,
      "grad_norm": 17.00250244140625,
      "learning_rate": 9.233465796196425e-05,
      "loss": 0.07376413345336914,
      "step": 27020
    },
    {
      "epoch": 7.6724382628441665,
      "grad_norm": 21.057472229003906,
      "learning_rate": 9.233181947204088e-05,
      "loss": 0.09075461030006408,
      "step": 27030
    },
    {
      "epoch": 7.675276752767528,
      "grad_norm": 10.701294898986816,
      "learning_rate": 9.232898098211752e-05,
      "loss": 0.07240979075431823,
      "step": 27040
    },
    {
      "epoch": 7.678115242690889,
      "grad_norm": 6.992650032043457,
      "learning_rate": 9.232614249219416e-05,
      "loss": 0.06210773587226868,
      "step": 27050
    },
    {
      "epoch": 7.680953732614249,
      "grad_norm": 9.65118408203125,
      "learning_rate": 9.232330400227079e-05,
      "loss": 0.1013252854347229,
      "step": 27060
    },
    {
      "epoch": 7.68379222253761,
      "grad_norm": 18.297809600830078,
      "learning_rate": 9.232046551234743e-05,
      "loss": 0.08987092971801758,
      "step": 27070
    },
    {
      "epoch": 7.686630712460971,
      "grad_norm": 8.46811580657959,
      "learning_rate": 9.231762702242409e-05,
      "loss": 0.08325701951980591,
      "step": 27080
    },
    {
      "epoch": 7.689469202384331,
      "grad_norm": 12.01207447052002,
      "learning_rate": 9.231478853250071e-05,
      "loss": 0.10193637609481812,
      "step": 27090
    },
    {
      "epoch": 7.6923076923076925,
      "grad_norm": 8.483120918273926,
      "learning_rate": 9.231195004257735e-05,
      "loss": 0.07533005475997925,
      "step": 27100
    },
    {
      "epoch": 7.695146182231053,
      "grad_norm": 4.285157203674316,
      "learning_rate": 9.2309111552654e-05,
      "loss": 0.07308659553527833,
      "step": 27110
    },
    {
      "epoch": 7.697984672154414,
      "grad_norm": 13.469322204589844,
      "learning_rate": 9.230627306273064e-05,
      "loss": 0.11812728643417358,
      "step": 27120
    },
    {
      "epoch": 7.700823162077775,
      "grad_norm": 4.122330188751221,
      "learning_rate": 9.230343457280726e-05,
      "loss": 0.06805981993675232,
      "step": 27130
    },
    {
      "epoch": 7.703661652001135,
      "grad_norm": 8.195808410644531,
      "learning_rate": 9.23005960828839e-05,
      "loss": 0.10464376211166382,
      "step": 27140
    },
    {
      "epoch": 7.706500141924496,
      "grad_norm": 2.0590085983276367,
      "learning_rate": 9.229775759296055e-05,
      "loss": 0.05893133878707886,
      "step": 27150
    },
    {
      "epoch": 7.7093386318478565,
      "grad_norm": 6.839287757873535,
      "learning_rate": 9.229491910303719e-05,
      "loss": 0.07113541960716248,
      "step": 27160
    },
    {
      "epoch": 7.712177121771218,
      "grad_norm": 10.108999252319336,
      "learning_rate": 9.229208061311383e-05,
      "loss": 0.06880993843078613,
      "step": 27170
    },
    {
      "epoch": 7.715015611694579,
      "grad_norm": 7.260905742645264,
      "learning_rate": 9.228924212319047e-05,
      "loss": 0.07458056211471557,
      "step": 27180
    },
    {
      "epoch": 7.717854101617939,
      "grad_norm": 14.39460277557373,
      "learning_rate": 9.22864036332671e-05,
      "loss": 0.09929774403572082,
      "step": 27190
    },
    {
      "epoch": 7.7206925915413,
      "grad_norm": 8.72842025756836,
      "learning_rate": 9.228356514334374e-05,
      "loss": 0.09915777444839477,
      "step": 27200
    },
    {
      "epoch": 7.723531081464661,
      "grad_norm": 10.42147445678711,
      "learning_rate": 9.228072665342038e-05,
      "loss": 0.05212537050247192,
      "step": 27210
    },
    {
      "epoch": 7.726369571388021,
      "grad_norm": 8.22713565826416,
      "learning_rate": 9.227788816349702e-05,
      "loss": 0.07159795165061951,
      "step": 27220
    },
    {
      "epoch": 7.7292080613113825,
      "grad_norm": 4.017706394195557,
      "learning_rate": 9.227504967357366e-05,
      "loss": 0.06240721940994263,
      "step": 27230
    },
    {
      "epoch": 7.732046551234744,
      "grad_norm": 5.211182117462158,
      "learning_rate": 9.22722111836503e-05,
      "loss": 0.0729503571987152,
      "step": 27240
    },
    {
      "epoch": 7.734885041158104,
      "grad_norm": 10.763618469238281,
      "learning_rate": 9.226937269372693e-05,
      "loss": 0.06937309503555297,
      "step": 27250
    },
    {
      "epoch": 7.737723531081465,
      "grad_norm": 7.215002059936523,
      "learning_rate": 9.226653420380358e-05,
      "loss": 0.08836377263069153,
      "step": 27260
    },
    {
      "epoch": 7.740562021004825,
      "grad_norm": 12.056900024414062,
      "learning_rate": 9.226369571388022e-05,
      "loss": 0.11528704166412354,
      "step": 27270
    },
    {
      "epoch": 7.743400510928186,
      "grad_norm": 2.8099069595336914,
      "learning_rate": 9.226085722395686e-05,
      "loss": 0.0929962933063507,
      "step": 27280
    },
    {
      "epoch": 7.746239000851547,
      "grad_norm": 7.90421199798584,
      "learning_rate": 9.22580187340335e-05,
      "loss": 0.06480119228363038,
      "step": 27290
    },
    {
      "epoch": 7.749077490774908,
      "grad_norm": 5.393097400665283,
      "learning_rate": 9.225518024411014e-05,
      "loss": 0.05637313723564148,
      "step": 27300
    },
    {
      "epoch": 7.751915980698269,
      "grad_norm": 11.397273063659668,
      "learning_rate": 9.225234175418678e-05,
      "loss": 0.08215314745903016,
      "step": 27310
    },
    {
      "epoch": 7.754754470621629,
      "grad_norm": 10.50705623626709,
      "learning_rate": 9.224950326426341e-05,
      "loss": 0.0628060519695282,
      "step": 27320
    },
    {
      "epoch": 7.75759296054499,
      "grad_norm": 5.726906776428223,
      "learning_rate": 9.224666477434005e-05,
      "loss": 0.07564097046852111,
      "step": 27330
    },
    {
      "epoch": 7.760431450468351,
      "grad_norm": 4.556398868560791,
      "learning_rate": 9.224382628441669e-05,
      "loss": 0.0793717384338379,
      "step": 27340
    },
    {
      "epoch": 7.763269940391711,
      "grad_norm": 5.672148704528809,
      "learning_rate": 9.224098779449333e-05,
      "loss": 0.06801968216896057,
      "step": 27350
    },
    {
      "epoch": 7.7661084303150725,
      "grad_norm": 7.562779903411865,
      "learning_rate": 9.223814930456998e-05,
      "loss": 0.06255950331687928,
      "step": 27360
    },
    {
      "epoch": 7.768946920238433,
      "grad_norm": 4.586869239807129,
      "learning_rate": 9.223531081464662e-05,
      "loss": 0.07646456360816956,
      "step": 27370
    },
    {
      "epoch": 7.771785410161794,
      "grad_norm": 6.574797630310059,
      "learning_rate": 9.223247232472324e-05,
      "loss": 0.08726893067359924,
      "step": 27380
    },
    {
      "epoch": 7.774623900085155,
      "grad_norm": 11.990139961242676,
      "learning_rate": 9.222963383479989e-05,
      "loss": 0.10619642734527587,
      "step": 27390
    },
    {
      "epoch": 7.777462390008515,
      "grad_norm": 4.124191761016846,
      "learning_rate": 9.222679534487653e-05,
      "loss": 0.06511511206626892,
      "step": 27400
    },
    {
      "epoch": 7.780300879931876,
      "grad_norm": 11.842705726623535,
      "learning_rate": 9.222395685495317e-05,
      "loss": 0.07857871651649476,
      "step": 27410
    },
    {
      "epoch": 7.783139369855237,
      "grad_norm": 9.203516960144043,
      "learning_rate": 9.222111836502981e-05,
      "loss": 0.07252969741821289,
      "step": 27420
    },
    {
      "epoch": 7.785977859778598,
      "grad_norm": 7.54423713684082,
      "learning_rate": 9.221827987510645e-05,
      "loss": 0.06294538974761962,
      "step": 27430
    },
    {
      "epoch": 7.788816349701959,
      "grad_norm": 7.835594177246094,
      "learning_rate": 9.221544138518309e-05,
      "loss": 0.06422609686851502,
      "step": 27440
    },
    {
      "epoch": 7.79165483962532,
      "grad_norm": 10.204795837402344,
      "learning_rate": 9.221260289525972e-05,
      "loss": 0.07313728332519531,
      "step": 27450
    },
    {
      "epoch": 7.79449332954868,
      "grad_norm": 12.939783096313477,
      "learning_rate": 9.220976440533636e-05,
      "loss": 0.08524999022483826,
      "step": 27460
    },
    {
      "epoch": 7.797331819472041,
      "grad_norm": 12.29412841796875,
      "learning_rate": 9.2206925915413e-05,
      "loss": 0.0500885009765625,
      "step": 27470
    },
    {
      "epoch": 7.800170309395401,
      "grad_norm": 8.158082962036133,
      "learning_rate": 9.220408742548965e-05,
      "loss": 0.04963379502296448,
      "step": 27480
    },
    {
      "epoch": 7.8030087993187625,
      "grad_norm": 8.88704776763916,
      "learning_rate": 9.220124893556629e-05,
      "loss": 0.0734395980834961,
      "step": 27490
    },
    {
      "epoch": 7.805847289242124,
      "grad_norm": 3.642496347427368,
      "learning_rate": 9.219841044564293e-05,
      "loss": 0.05631873607635498,
      "step": 27500
    },
    {
      "epoch": 7.805847289242124,
      "eval_accuracy": 0.9385133846251669,
      "eval_loss": 0.17873375117778778,
      "eval_runtime": 31.7264,
      "eval_samples_per_second": 495.708,
      "eval_steps_per_second": 7.754,
      "step": 27500
    },
    {
      "epoch": 7.808685779165484,
      "grad_norm": 9.62083911895752,
      "learning_rate": 9.219557195571956e-05,
      "loss": 0.10155099630355835,
      "step": 27510
    },
    {
      "epoch": 7.811524269088845,
      "grad_norm": 9.295912742614746,
      "learning_rate": 9.21927334657962e-05,
      "loss": 0.0950278401374817,
      "step": 27520
    },
    {
      "epoch": 7.814362759012205,
      "grad_norm": 9.448451042175293,
      "learning_rate": 9.218989497587284e-05,
      "loss": 0.07441433668136596,
      "step": 27530
    },
    {
      "epoch": 7.817201248935566,
      "grad_norm": 14.678438186645508,
      "learning_rate": 9.218705648594948e-05,
      "loss": 0.0783860981464386,
      "step": 27540
    },
    {
      "epoch": 7.820039738858927,
      "grad_norm": 10.288312911987305,
      "learning_rate": 9.218421799602612e-05,
      "loss": 0.09926932454109191,
      "step": 27550
    },
    {
      "epoch": 7.822878228782288,
      "grad_norm": 15.592484474182129,
      "learning_rate": 9.218137950610276e-05,
      "loss": 0.07857686877250672,
      "step": 27560
    },
    {
      "epoch": 7.825716718705649,
      "grad_norm": 10.722553253173828,
      "learning_rate": 9.21785410161794e-05,
      "loss": 0.08373225927352905,
      "step": 27570
    },
    {
      "epoch": 7.828555208629009,
      "grad_norm": 21.01876449584961,
      "learning_rate": 9.217570252625603e-05,
      "loss": 0.1304453730583191,
      "step": 27580
    },
    {
      "epoch": 7.83139369855237,
      "grad_norm": 11.549372673034668,
      "learning_rate": 9.217286403633267e-05,
      "loss": 0.06365420818328857,
      "step": 27590
    },
    {
      "epoch": 7.834232188475731,
      "grad_norm": 7.73263692855835,
      "learning_rate": 9.217002554640931e-05,
      "loss": 0.07818495035171509,
      "step": 27600
    },
    {
      "epoch": 7.837070678399091,
      "grad_norm": 4.6471781730651855,
      "learning_rate": 9.216718705648594e-05,
      "loss": 0.11928218603134155,
      "step": 27610
    },
    {
      "epoch": 7.839909168322452,
      "grad_norm": 8.833198547363281,
      "learning_rate": 9.21643485665626e-05,
      "loss": 0.07078934907913208,
      "step": 27620
    },
    {
      "epoch": 7.8427476582458135,
      "grad_norm": 6.99755859375,
      "learning_rate": 9.216151007663924e-05,
      "loss": 0.06474398374557495,
      "step": 27630
    },
    {
      "epoch": 7.845586148169174,
      "grad_norm": 17.104679107666016,
      "learning_rate": 9.215867158671587e-05,
      "loss": 0.08713191747665405,
      "step": 27640
    },
    {
      "epoch": 7.848424638092535,
      "grad_norm": 12.586068153381348,
      "learning_rate": 9.215583309679251e-05,
      "loss": 0.09169603586196899,
      "step": 27650
    },
    {
      "epoch": 7.851263128015896,
      "grad_norm": 5.165184497833252,
      "learning_rate": 9.215299460686915e-05,
      "loss": 0.07519745230674743,
      "step": 27660
    },
    {
      "epoch": 7.854101617939256,
      "grad_norm": 6.144222736358643,
      "learning_rate": 9.215015611694579e-05,
      "loss": 0.08131247162818908,
      "step": 27670
    },
    {
      "epoch": 7.856940107862617,
      "grad_norm": 8.348605155944824,
      "learning_rate": 9.214731762702243e-05,
      "loss": 0.08561167716979981,
      "step": 27680
    },
    {
      "epoch": 7.8597785977859775,
      "grad_norm": 8.29742431640625,
      "learning_rate": 9.214447913709907e-05,
      "loss": 0.0758174180984497,
      "step": 27690
    },
    {
      "epoch": 7.862617087709339,
      "grad_norm": 4.052884101867676,
      "learning_rate": 9.214164064717571e-05,
      "loss": 0.05107748508453369,
      "step": 27700
    },
    {
      "epoch": 7.8654555776327,
      "grad_norm": 5.622657299041748,
      "learning_rate": 9.213880215725234e-05,
      "loss": 0.04839843213558197,
      "step": 27710
    },
    {
      "epoch": 7.86829406755606,
      "grad_norm": 8.668671607971191,
      "learning_rate": 9.213596366732898e-05,
      "loss": 0.07239719629287719,
      "step": 27720
    },
    {
      "epoch": 7.871132557479421,
      "grad_norm": 11.74494457244873,
      "learning_rate": 9.213312517740563e-05,
      "loss": 0.10596381425857544,
      "step": 27730
    },
    {
      "epoch": 7.873971047402781,
      "grad_norm": 8.483918190002441,
      "learning_rate": 9.213028668748225e-05,
      "loss": 0.08127362132072449,
      "step": 27740
    },
    {
      "epoch": 7.876809537326142,
      "grad_norm": 7.552752494812012,
      "learning_rate": 9.212744819755891e-05,
      "loss": 0.047620663046836854,
      "step": 27750
    },
    {
      "epoch": 7.8796480272495035,
      "grad_norm": 5.005000591278076,
      "learning_rate": 9.212460970763555e-05,
      "loss": 0.07075139880180359,
      "step": 27760
    },
    {
      "epoch": 7.882486517172864,
      "grad_norm": 7.97688102722168,
      "learning_rate": 9.212177121771218e-05,
      "loss": 0.06970332860946656,
      "step": 27770
    },
    {
      "epoch": 7.885325007096225,
      "grad_norm": 13.781211853027344,
      "learning_rate": 9.211893272778882e-05,
      "loss": 0.06361839771270753,
      "step": 27780
    },
    {
      "epoch": 7.888163497019586,
      "grad_norm": 9.488520622253418,
      "learning_rate": 9.211609423786546e-05,
      "loss": 0.07577272653579711,
      "step": 27790
    },
    {
      "epoch": 7.891001986942946,
      "grad_norm": 10.51278018951416,
      "learning_rate": 9.21132557479421e-05,
      "loss": 0.07654934525489807,
      "step": 27800
    },
    {
      "epoch": 7.893840476866307,
      "grad_norm": 7.967763423919678,
      "learning_rate": 9.211041725801873e-05,
      "loss": 0.075218665599823,
      "step": 27810
    },
    {
      "epoch": 7.8966789667896675,
      "grad_norm": 4.935878753662109,
      "learning_rate": 9.210757876809538e-05,
      "loss": 0.07743616700172425,
      "step": 27820
    },
    {
      "epoch": 7.899517456713029,
      "grad_norm": 14.868453979492188,
      "learning_rate": 9.210474027817203e-05,
      "loss": 0.08831820487976075,
      "step": 27830
    },
    {
      "epoch": 7.90235594663639,
      "grad_norm": 12.458539962768555,
      "learning_rate": 9.210190178824865e-05,
      "loss": 0.10484296083450317,
      "step": 27840
    },
    {
      "epoch": 7.90519443655975,
      "grad_norm": 8.402612686157227,
      "learning_rate": 9.20990632983253e-05,
      "loss": 0.0750116229057312,
      "step": 27850
    },
    {
      "epoch": 7.908032926483111,
      "grad_norm": 5.604598045349121,
      "learning_rate": 9.209622480840194e-05,
      "loss": 0.07691766023635864,
      "step": 27860
    },
    {
      "epoch": 7.910871416406472,
      "grad_norm": 9.31263256072998,
      "learning_rate": 9.209338631847856e-05,
      "loss": 0.08029322624206543,
      "step": 27870
    },
    {
      "epoch": 7.913709906329832,
      "grad_norm": 8.123852729797363,
      "learning_rate": 9.209054782855522e-05,
      "loss": 0.08302119970321656,
      "step": 27880
    },
    {
      "epoch": 7.9165483962531935,
      "grad_norm": 4.332707405090332,
      "learning_rate": 9.208770933863186e-05,
      "loss": 0.08256133794784545,
      "step": 27890
    },
    {
      "epoch": 7.919386886176554,
      "grad_norm": 10.565896034240723,
      "learning_rate": 9.208487084870849e-05,
      "loss": 0.06437480449676514,
      "step": 27900
    },
    {
      "epoch": 7.922225376099915,
      "grad_norm": 6.135404109954834,
      "learning_rate": 9.208203235878513e-05,
      "loss": 0.06738330125808716,
      "step": 27910
    },
    {
      "epoch": 7.925063866023276,
      "grad_norm": 4.180898189544678,
      "learning_rate": 9.207919386886177e-05,
      "loss": 0.06374726891517639,
      "step": 27920
    },
    {
      "epoch": 7.927902355946636,
      "grad_norm": 18.677400588989258,
      "learning_rate": 9.207635537893841e-05,
      "loss": 0.08649280071258544,
      "step": 27930
    },
    {
      "epoch": 7.930740845869997,
      "grad_norm": 19.61292839050293,
      "learning_rate": 9.207351688901504e-05,
      "loss": 0.1137991189956665,
      "step": 27940
    },
    {
      "epoch": 7.9335793357933575,
      "grad_norm": 8.618695259094238,
      "learning_rate": 9.20706783990917e-05,
      "loss": 0.09259873032569885,
      "step": 27950
    },
    {
      "epoch": 7.936417825716719,
      "grad_norm": 2.9410247802734375,
      "learning_rate": 9.206783990916834e-05,
      "loss": 0.06983211636543274,
      "step": 27960
    },
    {
      "epoch": 7.93925631564008,
      "grad_norm": 7.65075159072876,
      "learning_rate": 9.206500141924496e-05,
      "loss": 0.05612436532974243,
      "step": 27970
    },
    {
      "epoch": 7.94209480556344,
      "grad_norm": 13.11274528503418,
      "learning_rate": 9.20621629293216e-05,
      "loss": 0.06669648289680481,
      "step": 27980
    },
    {
      "epoch": 7.944933295486801,
      "grad_norm": 13.256776809692383,
      "learning_rate": 9.205932443939825e-05,
      "loss": 0.11692190170288086,
      "step": 27990
    },
    {
      "epoch": 7.947771785410162,
      "grad_norm": 4.469211578369141,
      "learning_rate": 9.205648594947487e-05,
      "loss": 0.07598663568496704,
      "step": 28000
    },
    {
      "epoch": 7.947771785410162,
      "eval_accuracy": 0.9387041393781395,
      "eval_loss": 0.18467707931995392,
      "eval_runtime": 31.9806,
      "eval_samples_per_second": 491.767,
      "eval_steps_per_second": 7.692,
      "step": 28000
    },
    {
      "epoch": 7.950610275333522,
      "grad_norm": 8.834558486938477,
      "learning_rate": 9.205364745955152e-05,
      "loss": 0.06588622331619262,
      "step": 28010
    },
    {
      "epoch": 7.9534487652568835,
      "grad_norm": 12.618034362792969,
      "learning_rate": 9.205080896962817e-05,
      "loss": 0.07139828205108642,
      "step": 28020
    },
    {
      "epoch": 7.956287255180245,
      "grad_norm": 10.613566398620605,
      "learning_rate": 9.20479704797048e-05,
      "loss": 0.07414588332176208,
      "step": 28030
    },
    {
      "epoch": 7.959125745103605,
      "grad_norm": 11.469747543334961,
      "learning_rate": 9.204513198978144e-05,
      "loss": 0.0742957592010498,
      "step": 28040
    },
    {
      "epoch": 7.961964235026966,
      "grad_norm": 4.077260971069336,
      "learning_rate": 9.204229349985808e-05,
      "loss": 0.06864491701126099,
      "step": 28050
    },
    {
      "epoch": 7.964802724950326,
      "grad_norm": 3.775050640106201,
      "learning_rate": 9.203945500993472e-05,
      "loss": 0.0753418266773224,
      "step": 28060
    },
    {
      "epoch": 7.967641214873687,
      "grad_norm": 1.6595792770385742,
      "learning_rate": 9.203661652001135e-05,
      "loss": 0.06280696988105774,
      "step": 28070
    },
    {
      "epoch": 7.970479704797048,
      "grad_norm": 7.248383522033691,
      "learning_rate": 9.2033778030088e-05,
      "loss": 0.06148518323898315,
      "step": 28080
    },
    {
      "epoch": 7.973318194720409,
      "grad_norm": 7.503737449645996,
      "learning_rate": 9.203093954016463e-05,
      "loss": 0.08175054788589478,
      "step": 28090
    },
    {
      "epoch": 7.97615668464377,
      "grad_norm": 6.599333763122559,
      "learning_rate": 9.202810105024127e-05,
      "loss": 0.06689843535423279,
      "step": 28100
    },
    {
      "epoch": 7.97899517456713,
      "grad_norm": 9.845291137695312,
      "learning_rate": 9.202526256031792e-05,
      "loss": 0.11563538312911988,
      "step": 28110
    },
    {
      "epoch": 7.981833664490491,
      "grad_norm": 7.681792259216309,
      "learning_rate": 9.202242407039456e-05,
      "loss": 0.08608756065368653,
      "step": 28120
    },
    {
      "epoch": 7.984672154413852,
      "grad_norm": 6.084494590759277,
      "learning_rate": 9.201958558047119e-05,
      "loss": 0.05544511079788208,
      "step": 28130
    },
    {
      "epoch": 7.987510644337212,
      "grad_norm": 13.77053165435791,
      "learning_rate": 9.201674709054783e-05,
      "loss": 0.08298943042755128,
      "step": 28140
    },
    {
      "epoch": 7.9903491342605735,
      "grad_norm": 26.657520294189453,
      "learning_rate": 9.201390860062448e-05,
      "loss": 0.11479487419128417,
      "step": 28150
    },
    {
      "epoch": 7.993187624183934,
      "grad_norm": 15.225860595703125,
      "learning_rate": 9.201107011070111e-05,
      "loss": 0.10137089490890502,
      "step": 28160
    },
    {
      "epoch": 7.996026114107295,
      "grad_norm": 4.095785140991211,
      "learning_rate": 9.200823162077775e-05,
      "loss": 0.09867247343063354,
      "step": 28170
    },
    {
      "epoch": 7.998864604030656,
      "grad_norm": 11.450681686401367,
      "learning_rate": 9.200539313085439e-05,
      "loss": 0.0811159610748291,
      "step": 28180
    },
    {
      "epoch": 8.001703093954017,
      "grad_norm": 4.138267993927002,
      "learning_rate": 9.200255464093102e-05,
      "loss": 0.08176198601722717,
      "step": 28190
    },
    {
      "epoch": 8.004541583877376,
      "grad_norm": 7.668015480041504,
      "learning_rate": 9.199971615100766e-05,
      "loss": 0.050739753246307376,
      "step": 28200
    },
    {
      "epoch": 8.007380073800737,
      "grad_norm": 18.287296295166016,
      "learning_rate": 9.199687766108432e-05,
      "loss": 0.08709633946418763,
      "step": 28210
    },
    {
      "epoch": 8.010218563724099,
      "grad_norm": 8.561509132385254,
      "learning_rate": 9.199403917116094e-05,
      "loss": 0.04590723216533661,
      "step": 28220
    },
    {
      "epoch": 8.01305705364746,
      "grad_norm": 5.986148834228516,
      "learning_rate": 9.199120068123759e-05,
      "loss": 0.06425231099128723,
      "step": 28230
    },
    {
      "epoch": 8.01589554357082,
      "grad_norm": 6.2063117027282715,
      "learning_rate": 9.198836219131423e-05,
      "loss": 0.05600226521492004,
      "step": 28240
    },
    {
      "epoch": 8.018734033494182,
      "grad_norm": 4.819368362426758,
      "learning_rate": 9.198552370139087e-05,
      "loss": 0.0593639612197876,
      "step": 28250
    },
    {
      "epoch": 8.021572523417541,
      "grad_norm": 8.739453315734863,
      "learning_rate": 9.19826852114675e-05,
      "loss": 0.10399658679962158,
      "step": 28260
    },
    {
      "epoch": 8.024411013340902,
      "grad_norm": 6.652143478393555,
      "learning_rate": 9.197984672154414e-05,
      "loss": 0.07376932501792907,
      "step": 28270
    },
    {
      "epoch": 8.027249503264263,
      "grad_norm": 6.728353977203369,
      "learning_rate": 9.197700823162079e-05,
      "loss": 0.05037533044815064,
      "step": 28280
    },
    {
      "epoch": 8.030087993187625,
      "grad_norm": 6.354711055755615,
      "learning_rate": 9.197416974169742e-05,
      "loss": 0.06774352788925171,
      "step": 28290
    },
    {
      "epoch": 8.032926483110986,
      "grad_norm": 7.2082061767578125,
      "learning_rate": 9.197133125177406e-05,
      "loss": 0.06029695272445679,
      "step": 28300
    },
    {
      "epoch": 8.035764973034345,
      "grad_norm": 4.566948413848877,
      "learning_rate": 9.19684927618507e-05,
      "loss": 0.07258274555206298,
      "step": 28310
    },
    {
      "epoch": 8.038603462957706,
      "grad_norm": 4.820693016052246,
      "learning_rate": 9.196565427192733e-05,
      "loss": 0.05814773440361023,
      "step": 28320
    },
    {
      "epoch": 8.041441952881067,
      "grad_norm": 8.485374450683594,
      "learning_rate": 9.196281578200397e-05,
      "loss": 0.05457184910774231,
      "step": 28330
    },
    {
      "epoch": 8.044280442804428,
      "grad_norm": 10.01970100402832,
      "learning_rate": 9.195997729208061e-05,
      "loss": 0.07092044353485108,
      "step": 28340
    },
    {
      "epoch": 8.04711893272779,
      "grad_norm": 5.771849632263184,
      "learning_rate": 9.195713880215726e-05,
      "loss": 0.06729248762130738,
      "step": 28350
    },
    {
      "epoch": 8.049957422651149,
      "grad_norm": 9.445206642150879,
      "learning_rate": 9.19543003122339e-05,
      "loss": 0.06455802917480469,
      "step": 28360
    },
    {
      "epoch": 8.05279591257451,
      "grad_norm": 5.551255226135254,
      "learning_rate": 9.195146182231054e-05,
      "loss": 0.0712252676486969,
      "step": 28370
    },
    {
      "epoch": 8.055634402497871,
      "grad_norm": 3.5603044033050537,
      "learning_rate": 9.194862333238718e-05,
      "loss": 0.09181907773017883,
      "step": 28380
    },
    {
      "epoch": 8.058472892421232,
      "grad_norm": 2.0351345539093018,
      "learning_rate": 9.194578484246381e-05,
      "loss": 0.07351831197738648,
      "step": 28390
    },
    {
      "epoch": 8.061311382344593,
      "grad_norm": 19.573463439941406,
      "learning_rate": 9.194294635254045e-05,
      "loss": 0.08498290181159973,
      "step": 28400
    },
    {
      "epoch": 8.064149872267954,
      "grad_norm": 12.892037391662598,
      "learning_rate": 9.19401078626171e-05,
      "loss": 0.07225368618965149,
      "step": 28410
    },
    {
      "epoch": 8.066988362191314,
      "grad_norm": 11.845551490783691,
      "learning_rate": 9.193726937269373e-05,
      "loss": 0.065793377161026,
      "step": 28420
    },
    {
      "epoch": 8.069826852114675,
      "grad_norm": 11.925344467163086,
      "learning_rate": 9.193443088277037e-05,
      "loss": 0.05553908944129944,
      "step": 28430
    },
    {
      "epoch": 8.072665342038036,
      "grad_norm": 12.499534606933594,
      "learning_rate": 9.193159239284701e-05,
      "loss": 0.06909813284873963,
      "step": 28440
    },
    {
      "epoch": 8.075503831961397,
      "grad_norm": 7.718222618103027,
      "learning_rate": 9.192875390292364e-05,
      "loss": 0.05608686208724976,
      "step": 28450
    },
    {
      "epoch": 8.078342321884758,
      "grad_norm": 12.7086820602417,
      "learning_rate": 9.192591541300028e-05,
      "loss": 0.05288501381874085,
      "step": 28460
    },
    {
      "epoch": 8.081180811808117,
      "grad_norm": 8.647099494934082,
      "learning_rate": 9.192307692307692e-05,
      "loss": 0.053787410259246826,
      "step": 28470
    },
    {
      "epoch": 8.084019301731479,
      "grad_norm": 9.51988410949707,
      "learning_rate": 9.192023843315357e-05,
      "loss": 0.08311404585838318,
      "step": 28480
    },
    {
      "epoch": 8.08685779165484,
      "grad_norm": 8.066847801208496,
      "learning_rate": 9.191739994323021e-05,
      "loss": 0.04294182658195496,
      "step": 28490
    },
    {
      "epoch": 8.0896962815782,
      "grad_norm": 5.057689189910889,
      "learning_rate": 9.191456145330685e-05,
      "loss": 0.049669665098190305,
      "step": 28500
    },
    {
      "epoch": 8.0896962815782,
      "eval_accuracy": 0.9378775354485916,
      "eval_loss": 0.1790970265865326,
      "eval_runtime": 31.1557,
      "eval_samples_per_second": 504.788,
      "eval_steps_per_second": 7.896,
      "step": 28500
    },
    {
      "epoch": 8.092534771501562,
      "grad_norm": 12.8748197555542,
      "learning_rate": 9.191172296338349e-05,
      "loss": 0.07101553678512573,
      "step": 28510
    },
    {
      "epoch": 8.095373261424921,
      "grad_norm": 2.00642728805542,
      "learning_rate": 9.190888447346012e-05,
      "loss": 0.04895746111869812,
      "step": 28520
    },
    {
      "epoch": 8.098211751348282,
      "grad_norm": 9.168696403503418,
      "learning_rate": 9.190604598353676e-05,
      "loss": 0.06775805354118347,
      "step": 28530
    },
    {
      "epoch": 8.101050241271643,
      "grad_norm": 6.544206619262695,
      "learning_rate": 9.19032074936134e-05,
      "loss": 0.05526372194290161,
      "step": 28540
    },
    {
      "epoch": 8.103888731195005,
      "grad_norm": 7.154520511627197,
      "learning_rate": 9.190036900369004e-05,
      "loss": 0.05225864052772522,
      "step": 28550
    },
    {
      "epoch": 8.106727221118366,
      "grad_norm": 8.968255996704102,
      "learning_rate": 9.189753051376668e-05,
      "loss": 0.04409788846969605,
      "step": 28560
    },
    {
      "epoch": 8.109565711041725,
      "grad_norm": 4.260835647583008,
      "learning_rate": 9.189469202384332e-05,
      "loss": 0.07063850164413452,
      "step": 28570
    },
    {
      "epoch": 8.112404200965086,
      "grad_norm": 11.586590766906738,
      "learning_rate": 9.189185353391995e-05,
      "loss": 0.05927335619926453,
      "step": 28580
    },
    {
      "epoch": 8.115242690888447,
      "grad_norm": 9.707047462463379,
      "learning_rate": 9.18890150439966e-05,
      "loss": 0.06357197761535645,
      "step": 28590
    },
    {
      "epoch": 8.118081180811808,
      "grad_norm": 8.005043029785156,
      "learning_rate": 9.188617655407324e-05,
      "loss": 0.05256074070930481,
      "step": 28600
    },
    {
      "epoch": 8.12091967073517,
      "grad_norm": 9.046860694885254,
      "learning_rate": 9.188333806414988e-05,
      "loss": 0.04542839527130127,
      "step": 28610
    },
    {
      "epoch": 8.12375816065853,
      "grad_norm": 2.8076603412628174,
      "learning_rate": 9.188049957422652e-05,
      "loss": 0.052019888162612916,
      "step": 28620
    },
    {
      "epoch": 8.12659665058189,
      "grad_norm": 6.1467204093933105,
      "learning_rate": 9.187766108430316e-05,
      "loss": 0.06769022941589356,
      "step": 28630
    },
    {
      "epoch": 8.129435140505251,
      "grad_norm": 5.422061920166016,
      "learning_rate": 9.18748225943798e-05,
      "loss": 0.06597175598144531,
      "step": 28640
    },
    {
      "epoch": 8.132273630428612,
      "grad_norm": 13.432371139526367,
      "learning_rate": 9.187198410445643e-05,
      "loss": 0.0722927987575531,
      "step": 28650
    },
    {
      "epoch": 8.135112120351973,
      "grad_norm": 5.142928600311279,
      "learning_rate": 9.186914561453307e-05,
      "loss": 0.06466186046600342,
      "step": 28660
    },
    {
      "epoch": 8.137950610275334,
      "grad_norm": 10.77366828918457,
      "learning_rate": 9.186630712460971e-05,
      "loss": 0.05489208698272705,
      "step": 28670
    },
    {
      "epoch": 8.140789100198694,
      "grad_norm": 18.801244735717773,
      "learning_rate": 9.186346863468635e-05,
      "loss": 0.07088869214057922,
      "step": 28680
    },
    {
      "epoch": 8.143627590122055,
      "grad_norm": 8.477251052856445,
      "learning_rate": 9.1860630144763e-05,
      "loss": 0.059319323301315306,
      "step": 28690
    },
    {
      "epoch": 8.146466080045416,
      "grad_norm": 15.10538101196289,
      "learning_rate": 9.185779165483964e-05,
      "loss": 0.05679097175598145,
      "step": 28700
    },
    {
      "epoch": 8.149304569968777,
      "grad_norm": 13.824700355529785,
      "learning_rate": 9.185495316491626e-05,
      "loss": 0.06503130793571472,
      "step": 28710
    },
    {
      "epoch": 8.152143059892138,
      "grad_norm": 6.980968952178955,
      "learning_rate": 9.18521146749929e-05,
      "loss": 0.048087412118911745,
      "step": 28720
    },
    {
      "epoch": 8.154981549815497,
      "grad_norm": 8.005522727966309,
      "learning_rate": 9.184927618506955e-05,
      "loss": 0.09411823153495788,
      "step": 28730
    },
    {
      "epoch": 8.157820039738858,
      "grad_norm": 8.153717041015625,
      "learning_rate": 9.184643769514619e-05,
      "loss": 0.050446927547454834,
      "step": 28740
    },
    {
      "epoch": 8.16065852966222,
      "grad_norm": 14.12834644317627,
      "learning_rate": 9.184359920522283e-05,
      "loss": 0.07808389663696289,
      "step": 28750
    },
    {
      "epoch": 8.16349701958558,
      "grad_norm": 7.5934367179870605,
      "learning_rate": 9.184076071529947e-05,
      "loss": 0.05060620307922363,
      "step": 28760
    },
    {
      "epoch": 8.166335509508942,
      "grad_norm": 1.4182673692703247,
      "learning_rate": 9.183792222537611e-05,
      "loss": 0.05429837703704834,
      "step": 28770
    },
    {
      "epoch": 8.169173999432301,
      "grad_norm": 4.7587361335754395,
      "learning_rate": 9.183508373545274e-05,
      "loss": 0.060312533378601076,
      "step": 28780
    },
    {
      "epoch": 8.172012489355662,
      "grad_norm": 2.2295658588409424,
      "learning_rate": 9.183224524552938e-05,
      "loss": 0.05230412483215332,
      "step": 28790
    },
    {
      "epoch": 8.174850979279023,
      "grad_norm": 8.094844818115234,
      "learning_rate": 9.182940675560602e-05,
      "loss": 0.06559881567955017,
      "step": 28800
    },
    {
      "epoch": 8.177689469202384,
      "grad_norm": 5.727514266967773,
      "learning_rate": 9.182656826568266e-05,
      "loss": 0.05493706464767456,
      "step": 28810
    },
    {
      "epoch": 8.180527959125746,
      "grad_norm": 9.533031463623047,
      "learning_rate": 9.18237297757593e-05,
      "loss": 0.05957727432250977,
      "step": 28820
    },
    {
      "epoch": 8.183366449049107,
      "grad_norm": 4.9983811378479,
      "learning_rate": 9.182089128583595e-05,
      "loss": 0.052783387899398806,
      "step": 28830
    },
    {
      "epoch": 8.186204938972466,
      "grad_norm": 6.4478607177734375,
      "learning_rate": 9.181805279591257e-05,
      "loss": 0.06514946818351745,
      "step": 28840
    },
    {
      "epoch": 8.189043428895827,
      "grad_norm": 4.356100082397461,
      "learning_rate": 9.181521430598922e-05,
      "loss": 0.05870385766029358,
      "step": 28850
    },
    {
      "epoch": 8.191881918819188,
      "grad_norm": 6.167853832244873,
      "learning_rate": 9.181237581606586e-05,
      "loss": 0.06018507480621338,
      "step": 28860
    },
    {
      "epoch": 8.19472040874255,
      "grad_norm": 11.0339994430542,
      "learning_rate": 9.18095373261425e-05,
      "loss": 0.0852200984954834,
      "step": 28870
    },
    {
      "epoch": 8.19755889866591,
      "grad_norm": 9.826255798339844,
      "learning_rate": 9.180669883621914e-05,
      "loss": 0.06306055784225464,
      "step": 28880
    },
    {
      "epoch": 8.20039738858927,
      "grad_norm": 7.294203758239746,
      "learning_rate": 9.180386034629578e-05,
      "loss": 0.08102589845657349,
      "step": 28890
    },
    {
      "epoch": 8.203235878512631,
      "grad_norm": 13.558586120605469,
      "learning_rate": 9.180102185637242e-05,
      "loss": 0.06372936964035034,
      "step": 28900
    },
    {
      "epoch": 8.206074368435992,
      "grad_norm": 6.082158088684082,
      "learning_rate": 9.179818336644905e-05,
      "loss": 0.07231396436691284,
      "step": 28910
    },
    {
      "epoch": 8.208912858359353,
      "grad_norm": 17.394548416137695,
      "learning_rate": 9.179534487652569e-05,
      "loss": 0.04497288167476654,
      "step": 28920
    },
    {
      "epoch": 8.211751348282714,
      "grad_norm": 3.8369526863098145,
      "learning_rate": 9.179250638660233e-05,
      "loss": 0.06243351697921753,
      "step": 28930
    },
    {
      "epoch": 8.214589838206074,
      "grad_norm": 7.327316761016846,
      "learning_rate": 9.178966789667896e-05,
      "loss": 0.04064055681228638,
      "step": 28940
    },
    {
      "epoch": 8.217428328129435,
      "grad_norm": 14.72048568725586,
      "learning_rate": 9.178682940675562e-05,
      "loss": 0.07328947186470032,
      "step": 28950
    },
    {
      "epoch": 8.220266818052796,
      "grad_norm": 9.032869338989258,
      "learning_rate": 9.178399091683226e-05,
      "loss": 0.06302196383476258,
      "step": 28960
    },
    {
      "epoch": 8.223105307976157,
      "grad_norm": 4.357512950897217,
      "learning_rate": 9.178115242690888e-05,
      "loss": 0.06843396425247192,
      "step": 28970
    },
    {
      "epoch": 8.225943797899518,
      "grad_norm": 6.721951007843018,
      "learning_rate": 9.177831393698553e-05,
      "loss": 0.0600142240524292,
      "step": 28980
    },
    {
      "epoch": 8.228782287822877,
      "grad_norm": 3.4084107875823975,
      "learning_rate": 9.17757592960545e-05,
      "loss": 0.07553284764289855,
      "step": 28990
    },
    {
      "epoch": 8.231620777746238,
      "grad_norm": 8.18293571472168,
      "learning_rate": 9.177292080613115e-05,
      "loss": 0.04516300559043884,
      "step": 29000
    },
    {
      "epoch": 8.231620777746238,
      "eval_accuracy": 0.943790932790742,
      "eval_loss": 0.1692141890525818,
      "eval_runtime": 30.9015,
      "eval_samples_per_second": 508.939,
      "eval_steps_per_second": 7.961,
      "step": 29000
    },
    {
      "epoch": 8.2344592676696,
      "grad_norm": 10.485038757324219,
      "learning_rate": 9.177008231620779e-05,
      "loss": 0.07149056792259216,
      "step": 29010
    },
    {
      "epoch": 8.23729775759296,
      "grad_norm": 4.556738376617432,
      "learning_rate": 9.176724382628442e-05,
      "loss": 0.06517475843429565,
      "step": 29020
    },
    {
      "epoch": 8.240136247516322,
      "grad_norm": 6.041149616241455,
      "learning_rate": 9.176440533636106e-05,
      "loss": 0.0854555606842041,
      "step": 29030
    },
    {
      "epoch": 8.242974737439683,
      "grad_norm": 5.430018901824951,
      "learning_rate": 9.17615668464377e-05,
      "loss": 0.07756624817848205,
      "step": 29040
    },
    {
      "epoch": 8.245813227363042,
      "grad_norm": 4.9481048583984375,
      "learning_rate": 9.175872835651434e-05,
      "loss": 0.07038364410400391,
      "step": 29050
    },
    {
      "epoch": 8.248651717286403,
      "grad_norm": 8.902161598205566,
      "learning_rate": 9.175588986659098e-05,
      "loss": 0.06555392742156982,
      "step": 29060
    },
    {
      "epoch": 8.251490207209764,
      "grad_norm": 8.555258750915527,
      "learning_rate": 9.175305137666762e-05,
      "loss": 0.05040958523750305,
      "step": 29070
    },
    {
      "epoch": 8.254328697133126,
      "grad_norm": 9.97897720336914,
      "learning_rate": 9.175021288674425e-05,
      "loss": 0.062173879146575926,
      "step": 29080
    },
    {
      "epoch": 8.257167187056487,
      "grad_norm": 5.084324836730957,
      "learning_rate": 9.174737439682089e-05,
      "loss": 0.06158629059791565,
      "step": 29090
    },
    {
      "epoch": 8.260005676979846,
      "grad_norm": 5.335748195648193,
      "learning_rate": 9.174453590689753e-05,
      "loss": 0.054487550258636476,
      "step": 29100
    },
    {
      "epoch": 8.262844166903207,
      "grad_norm": 4.3890228271484375,
      "learning_rate": 9.174169741697418e-05,
      "loss": 0.07088085412979125,
      "step": 29110
    },
    {
      "epoch": 8.265682656826568,
      "grad_norm": 10.776468276977539,
      "learning_rate": 9.17388589270508e-05,
      "loss": 0.05529135465621948,
      "step": 29120
    },
    {
      "epoch": 8.26852114674993,
      "grad_norm": 12.11426830291748,
      "learning_rate": 9.173602043712746e-05,
      "loss": 0.06611025333404541,
      "step": 29130
    },
    {
      "epoch": 8.27135963667329,
      "grad_norm": 7.765328884124756,
      "learning_rate": 9.17331819472041e-05,
      "loss": 0.05948413610458374,
      "step": 29140
    },
    {
      "epoch": 8.27419812659665,
      "grad_norm": 11.50073528289795,
      "learning_rate": 9.173034345728073e-05,
      "loss": 0.11590793132781982,
      "step": 29150
    },
    {
      "epoch": 8.27703661652001,
      "grad_norm": 2.4201109409332275,
      "learning_rate": 9.172750496735737e-05,
      "loss": 0.06780180931091309,
      "step": 29160
    },
    {
      "epoch": 8.279875106443372,
      "grad_norm": 10.273046493530273,
      "learning_rate": 9.172466647743401e-05,
      "loss": 0.050675714015960695,
      "step": 29170
    },
    {
      "epoch": 8.282713596366733,
      "grad_norm": 12.870804786682129,
      "learning_rate": 9.172182798751064e-05,
      "loss": 0.08268274068832397,
      "step": 29180
    },
    {
      "epoch": 8.285552086290094,
      "grad_norm": 10.322364807128906,
      "learning_rate": 9.171898949758729e-05,
      "loss": 0.06419029235839843,
      "step": 29190
    },
    {
      "epoch": 8.288390576213455,
      "grad_norm": 15.34072494506836,
      "learning_rate": 9.171615100766393e-05,
      "loss": 0.0902258574962616,
      "step": 29200
    },
    {
      "epoch": 8.291229066136815,
      "grad_norm": 12.890625953674316,
      "learning_rate": 9.171331251774056e-05,
      "loss": 0.07161271572113037,
      "step": 29210
    },
    {
      "epoch": 8.294067556060176,
      "grad_norm": 5.4624505043029785,
      "learning_rate": 9.17104740278172e-05,
      "loss": 0.06565551161766052,
      "step": 29220
    },
    {
      "epoch": 8.296906045983537,
      "grad_norm": 8.181920051574707,
      "learning_rate": 9.170763553789384e-05,
      "loss": 0.06308490633964539,
      "step": 29230
    },
    {
      "epoch": 8.299744535906898,
      "grad_norm": 3.0428454875946045,
      "learning_rate": 9.170479704797049e-05,
      "loss": 0.05067722201347351,
      "step": 29240
    },
    {
      "epoch": 8.302583025830259,
      "grad_norm": 5.010264873504639,
      "learning_rate": 9.170195855804711e-05,
      "loss": 0.04812777042388916,
      "step": 29250
    },
    {
      "epoch": 8.305421515753618,
      "grad_norm": 9.04201602935791,
      "learning_rate": 9.169912006812377e-05,
      "loss": 0.06117487549781799,
      "step": 29260
    },
    {
      "epoch": 8.30826000567698,
      "grad_norm": 9.197172164916992,
      "learning_rate": 9.169628157820041e-05,
      "loss": 0.062438076734542845,
      "step": 29270
    },
    {
      "epoch": 8.31109849560034,
      "grad_norm": 8.5620756149292,
      "learning_rate": 9.169344308827704e-05,
      "loss": 0.07440378665924072,
      "step": 29280
    },
    {
      "epoch": 8.313936985523702,
      "grad_norm": 5.338595390319824,
      "learning_rate": 9.169060459835368e-05,
      "loss": 0.05974234938621521,
      "step": 29290
    },
    {
      "epoch": 8.316775475447063,
      "grad_norm": 15.110782623291016,
      "learning_rate": 9.168776610843032e-05,
      "loss": 0.08244252800941468,
      "step": 29300
    },
    {
      "epoch": 8.319613965370422,
      "grad_norm": 17.116798400878906,
      "learning_rate": 9.168492761850695e-05,
      "loss": 0.06338911056518555,
      "step": 29310
    },
    {
      "epoch": 8.322452455293783,
      "grad_norm": 5.806769847869873,
      "learning_rate": 9.168208912858359e-05,
      "loss": 0.050322186946868894,
      "step": 29320
    },
    {
      "epoch": 8.325290945217144,
      "grad_norm": 9.002280235290527,
      "learning_rate": 9.167925063866025e-05,
      "loss": 0.05529931783676147,
      "step": 29330
    },
    {
      "epoch": 8.328129435140506,
      "grad_norm": 13.09355640411377,
      "learning_rate": 9.167641214873687e-05,
      "loss": 0.06681718230247498,
      "step": 29340
    },
    {
      "epoch": 8.330967925063867,
      "grad_norm": 7.570567607879639,
      "learning_rate": 9.167357365881351e-05,
      "loss": 0.08058772683143615,
      "step": 29350
    },
    {
      "epoch": 8.333806414987226,
      "grad_norm": 2.704773426055908,
      "learning_rate": 9.167073516889016e-05,
      "loss": 0.06773853898048401,
      "step": 29360
    },
    {
      "epoch": 8.336644904910587,
      "grad_norm": 3.5711374282836914,
      "learning_rate": 9.16678966789668e-05,
      "loss": 0.05489592552185059,
      "step": 29370
    },
    {
      "epoch": 8.339483394833948,
      "grad_norm": 5.974013328552246,
      "learning_rate": 9.166505818904342e-05,
      "loss": 0.05106773376464844,
      "step": 29380
    },
    {
      "epoch": 8.34232188475731,
      "grad_norm": 20.108932495117188,
      "learning_rate": 9.166221969912008e-05,
      "loss": 0.0673336923122406,
      "step": 29390
    },
    {
      "epoch": 8.34516037468067,
      "grad_norm": 18.985143661499023,
      "learning_rate": 9.165938120919672e-05,
      "loss": 0.05588265657424927,
      "step": 29400
    },
    {
      "epoch": 8.347998864604032,
      "grad_norm": 13.223711967468262,
      "learning_rate": 9.165654271927335e-05,
      "loss": 0.07614171504974365,
      "step": 29410
    },
    {
      "epoch": 8.35083735452739,
      "grad_norm": 4.557998180389404,
      "learning_rate": 9.165370422934999e-05,
      "loss": 0.04518270790576935,
      "step": 29420
    },
    {
      "epoch": 8.353675844450752,
      "grad_norm": 7.949966907501221,
      "learning_rate": 9.165086573942663e-05,
      "loss": 0.049933457374572755,
      "step": 29430
    },
    {
      "epoch": 8.356514334374113,
      "grad_norm": 7.423627853393555,
      "learning_rate": 9.164802724950326e-05,
      "loss": 0.07519654631614685,
      "step": 29440
    },
    {
      "epoch": 8.359352824297474,
      "grad_norm": 1.7148159742355347,
      "learning_rate": 9.16451887595799e-05,
      "loss": 0.05791993141174316,
      "step": 29450
    },
    {
      "epoch": 8.362191314220835,
      "grad_norm": 16.687705993652344,
      "learning_rate": 9.164235026965656e-05,
      "loss": 0.07638731598854065,
      "step": 29460
    },
    {
      "epoch": 8.365029804144195,
      "grad_norm": 5.448736190795898,
      "learning_rate": 9.163951177973318e-05,
      "loss": 0.058822745084762575,
      "step": 29470
    },
    {
      "epoch": 8.367868294067556,
      "grad_norm": 16.446035385131836,
      "learning_rate": 9.163667328980982e-05,
      "loss": 0.08842889070510865,
      "step": 29480
    },
    {
      "epoch": 8.370706783990917,
      "grad_norm": 17.496366500854492,
      "learning_rate": 9.163383479988647e-05,
      "loss": 0.06101993918418884,
      "step": 29490
    },
    {
      "epoch": 8.373545273914278,
      "grad_norm": 14.626163482666016,
      "learning_rate": 9.163099630996311e-05,
      "loss": 0.06540396213531494,
      "step": 29500
    },
    {
      "epoch": 8.373545273914278,
      "eval_accuracy": 0.938894894131112,
      "eval_loss": 0.1882154941558838,
      "eval_runtime": 31.3142,
      "eval_samples_per_second": 502.232,
      "eval_steps_per_second": 7.856,
      "step": 29500
    },
    {
      "epoch": 8.376383763837639,
      "grad_norm": 13.651522636413574,
      "learning_rate": 9.162815782003974e-05,
      "loss": 0.07866971492767334,
      "step": 29510
    },
    {
      "epoch": 8.379222253760998,
      "grad_norm": 10.000460624694824,
      "learning_rate": 9.162531933011638e-05,
      "loss": 0.06670709252357483,
      "step": 29520
    },
    {
      "epoch": 8.38206074368436,
      "grad_norm": 14.086359977722168,
      "learning_rate": 9.162248084019303e-05,
      "loss": 0.07480021119117737,
      "step": 29530
    },
    {
      "epoch": 8.38489923360772,
      "grad_norm": 4.935381889343262,
      "learning_rate": 9.161964235026966e-05,
      "loss": 0.06456262469291688,
      "step": 29540
    },
    {
      "epoch": 8.387737723531082,
      "grad_norm": 3.3612170219421387,
      "learning_rate": 9.16168038603463e-05,
      "loss": 0.05752266645431518,
      "step": 29550
    },
    {
      "epoch": 8.390576213454443,
      "grad_norm": 8.120400428771973,
      "learning_rate": 9.161396537042294e-05,
      "loss": 0.05767455101013184,
      "step": 29560
    },
    {
      "epoch": 8.393414703377802,
      "grad_norm": 6.8282880783081055,
      "learning_rate": 9.161112688049957e-05,
      "loss": 0.08247737884521485,
      "step": 29570
    },
    {
      "epoch": 8.396253193301163,
      "grad_norm": 17.539871215820312,
      "learning_rate": 9.160828839057621e-05,
      "loss": 0.05251510143280029,
      "step": 29580
    },
    {
      "epoch": 8.399091683224524,
      "grad_norm": 4.805350303649902,
      "learning_rate": 9.160544990065287e-05,
      "loss": 0.0473786473274231,
      "step": 29590
    },
    {
      "epoch": 8.401930173147885,
      "grad_norm": 10.188152313232422,
      "learning_rate": 9.16026114107295e-05,
      "loss": 0.08898124694824219,
      "step": 29600
    },
    {
      "epoch": 8.404768663071247,
      "grad_norm": 15.48466682434082,
      "learning_rate": 9.159977292080614e-05,
      "loss": 0.06543077230453491,
      "step": 29610
    },
    {
      "epoch": 8.407607152994608,
      "grad_norm": 12.283458709716797,
      "learning_rate": 9.159693443088278e-05,
      "loss": 0.09538312554359436,
      "step": 29620
    },
    {
      "epoch": 8.410445642917967,
      "grad_norm": 13.443482398986816,
      "learning_rate": 9.159409594095942e-05,
      "loss": 0.06765753626823426,
      "step": 29630
    },
    {
      "epoch": 8.413284132841328,
      "grad_norm": 11.031046867370605,
      "learning_rate": 9.159125745103605e-05,
      "loss": 0.08531753420829773,
      "step": 29640
    },
    {
      "epoch": 8.41612262276469,
      "grad_norm": 8.386163711547852,
      "learning_rate": 9.158841896111269e-05,
      "loss": 0.07212099432945251,
      "step": 29650
    },
    {
      "epoch": 8.41896111268805,
      "grad_norm": 5.738980293273926,
      "learning_rate": 9.158558047118934e-05,
      "loss": 0.07749001383781433,
      "step": 29660
    },
    {
      "epoch": 8.421799602611411,
      "grad_norm": 14.240184783935547,
      "learning_rate": 9.158274198126597e-05,
      "loss": 0.07213364839553833,
      "step": 29670
    },
    {
      "epoch": 8.42463809253477,
      "grad_norm": 10.006685256958008,
      "learning_rate": 9.157990349134261e-05,
      "loss": 0.06452087163925171,
      "step": 29680
    },
    {
      "epoch": 8.427476582458132,
      "grad_norm": 6.8914055824279785,
      "learning_rate": 9.157706500141925e-05,
      "loss": 0.04866733551025391,
      "step": 29690
    },
    {
      "epoch": 8.430315072381493,
      "grad_norm": 7.061382293701172,
      "learning_rate": 9.157422651149588e-05,
      "loss": 0.05150160789489746,
      "step": 29700
    },
    {
      "epoch": 8.433153562304854,
      "grad_norm": 14.592506408691406,
      "learning_rate": 9.157138802157252e-05,
      "loss": 0.0820405900478363,
      "step": 29710
    },
    {
      "epoch": 8.435992052228215,
      "grad_norm": 16.972883224487305,
      "learning_rate": 9.156854953164916e-05,
      "loss": 0.08067867755889893,
      "step": 29720
    },
    {
      "epoch": 8.438830542151575,
      "grad_norm": 6.584667205810547,
      "learning_rate": 9.15657110417258e-05,
      "loss": 0.07627401351928711,
      "step": 29730
    },
    {
      "epoch": 8.441669032074936,
      "grad_norm": 3.4748692512512207,
      "learning_rate": 9.156287255180245e-05,
      "loss": 0.05884630680084228,
      "step": 29740
    },
    {
      "epoch": 8.444507521998297,
      "grad_norm": 10.529985427856445,
      "learning_rate": 9.156003406187909e-05,
      "loss": 0.06511694788932801,
      "step": 29750
    },
    {
      "epoch": 8.447346011921658,
      "grad_norm": 7.230268955230713,
      "learning_rate": 9.155719557195573e-05,
      "loss": 0.06866088509559631,
      "step": 29760
    },
    {
      "epoch": 8.450184501845019,
      "grad_norm": 10.624114990234375,
      "learning_rate": 9.155435708203236e-05,
      "loss": 0.07058738470077515,
      "step": 29770
    },
    {
      "epoch": 8.453022991768378,
      "grad_norm": 6.953898906707764,
      "learning_rate": 9.1551518592109e-05,
      "loss": 0.057642459869384766,
      "step": 29780
    },
    {
      "epoch": 8.45586148169174,
      "grad_norm": 6.2965312004089355,
      "learning_rate": 9.154868010218564e-05,
      "loss": 0.07264969348907471,
      "step": 29790
    },
    {
      "epoch": 8.4586999716151,
      "grad_norm": 10.869126319885254,
      "learning_rate": 9.154584161226228e-05,
      "loss": 0.07255736589431763,
      "step": 29800
    },
    {
      "epoch": 8.461538461538462,
      "grad_norm": 6.415733337402344,
      "learning_rate": 9.154300312233892e-05,
      "loss": 0.04898870587348938,
      "step": 29810
    },
    {
      "epoch": 8.464376951461823,
      "grad_norm": 7.364101886749268,
      "learning_rate": 9.154016463241556e-05,
      "loss": 0.06439247727394104,
      "step": 29820
    },
    {
      "epoch": 8.467215441385184,
      "grad_norm": 6.468971252441406,
      "learning_rate": 9.153732614249219e-05,
      "loss": 0.06732615828514099,
      "step": 29830
    },
    {
      "epoch": 8.470053931308543,
      "grad_norm": 3.9857325553894043,
      "learning_rate": 9.153448765256883e-05,
      "loss": 0.06769530177116394,
      "step": 29840
    },
    {
      "epoch": 8.472892421231904,
      "grad_norm": 15.828653335571289,
      "learning_rate": 9.153164916264547e-05,
      "loss": 0.05919771790504456,
      "step": 29850
    },
    {
      "epoch": 8.475730911155265,
      "grad_norm": 17.331520080566406,
      "learning_rate": 9.152881067272212e-05,
      "loss": 0.08205833435058593,
      "step": 29860
    },
    {
      "epoch": 8.478569401078627,
      "grad_norm": 17.98587417602539,
      "learning_rate": 9.152597218279876e-05,
      "loss": 0.0660819411277771,
      "step": 29870
    },
    {
      "epoch": 8.481407891001988,
      "grad_norm": 3.2231264114379883,
      "learning_rate": 9.15231336928754e-05,
      "loss": 0.05351273417472839,
      "step": 29880
    },
    {
      "epoch": 8.484246380925347,
      "grad_norm": 4.374488353729248,
      "learning_rate": 9.152029520295204e-05,
      "loss": 0.06690125465393067,
      "step": 29890
    },
    {
      "epoch": 8.487084870848708,
      "grad_norm": 20.5726375579834,
      "learning_rate": 9.151745671302867e-05,
      "loss": 0.10268547534942626,
      "step": 29900
    },
    {
      "epoch": 8.48992336077207,
      "grad_norm": 9.212912559509277,
      "learning_rate": 9.151461822310531e-05,
      "loss": 0.07267470359802246,
      "step": 29910
    },
    {
      "epoch": 8.49276185069543,
      "grad_norm": 9.972260475158691,
      "learning_rate": 9.151177973318195e-05,
      "loss": 0.06243507862091065,
      "step": 29920
    },
    {
      "epoch": 8.495600340618791,
      "grad_norm": 9.37176513671875,
      "learning_rate": 9.150894124325859e-05,
      "loss": 0.07114928960800171,
      "step": 29930
    },
    {
      "epoch": 8.49843883054215,
      "grad_norm": 1.9846758842468262,
      "learning_rate": 9.150610275333523e-05,
      "loss": 0.06344807744026185,
      "step": 29940
    },
    {
      "epoch": 8.501277320465512,
      "grad_norm": 4.739149570465088,
      "learning_rate": 9.150326426341187e-05,
      "loss": 0.0565528929233551,
      "step": 29950
    },
    {
      "epoch": 8.504115810388873,
      "grad_norm": 6.5317254066467285,
      "learning_rate": 9.15004257734885e-05,
      "loss": 0.07424017786979675,
      "step": 29960
    },
    {
      "epoch": 8.506954300312234,
      "grad_norm": 9.611992835998535,
      "learning_rate": 9.149758728356514e-05,
      "loss": 0.08838900327682495,
      "step": 29970
    },
    {
      "epoch": 8.509792790235595,
      "grad_norm": 8.434793472290039,
      "learning_rate": 9.149474879364179e-05,
      "loss": 0.07297046780586243,
      "step": 29980
    },
    {
      "epoch": 8.512631280158956,
      "grad_norm": 3.4695658683776855,
      "learning_rate": 9.149191030371843e-05,
      "loss": 0.04186428189277649,
      "step": 29990
    },
    {
      "epoch": 8.515469770082316,
      "grad_norm": 9.727413177490234,
      "learning_rate": 9.148907181379507e-05,
      "loss": 0.05960800051689148,
      "step": 30000
    },
    {
      "epoch": 8.515469770082316,
      "eval_accuracy": 0.9387041393781395,
      "eval_loss": 0.189566969871521,
      "eval_runtime": 31.0727,
      "eval_samples_per_second": 506.135,
      "eval_steps_per_second": 7.917,
      "step": 30000
    },
    {
      "epoch": 8.518308260005677,
      "grad_norm": 6.594583511352539,
      "learning_rate": 9.148623332387171e-05,
      "loss": 0.058769834041595456,
      "step": 30010
    },
    {
      "epoch": 8.521146749929038,
      "grad_norm": 12.190415382385254,
      "learning_rate": 9.148339483394834e-05,
      "loss": 0.06734001636505127,
      "step": 30020
    },
    {
      "epoch": 8.523985239852399,
      "grad_norm": 9.373311042785645,
      "learning_rate": 9.148055634402498e-05,
      "loss": 0.065748530626297,
      "step": 30030
    },
    {
      "epoch": 8.52682372977576,
      "grad_norm": 3.8623650074005127,
      "learning_rate": 9.147771785410162e-05,
      "loss": 0.06235647201538086,
      "step": 30040
    },
    {
      "epoch": 8.52966221969912,
      "grad_norm": 6.599777698516846,
      "learning_rate": 9.147487936417826e-05,
      "loss": 0.04081343710422516,
      "step": 30050
    },
    {
      "epoch": 8.53250070962248,
      "grad_norm": 6.091607570648193,
      "learning_rate": 9.14720408742549e-05,
      "loss": 0.08156938552856445,
      "step": 30060
    },
    {
      "epoch": 8.535339199545842,
      "grad_norm": 10.18403434753418,
      "learning_rate": 9.146920238433154e-05,
      "loss": 0.053347927331924436,
      "step": 30070
    },
    {
      "epoch": 8.538177689469203,
      "grad_norm": 3.583003282546997,
      "learning_rate": 9.146636389440819e-05,
      "loss": 0.051168018579483034,
      "step": 30080
    },
    {
      "epoch": 8.541016179392564,
      "grad_norm": 5.291472911834717,
      "learning_rate": 9.146352540448481e-05,
      "loss": 0.05486744046211243,
      "step": 30090
    },
    {
      "epoch": 8.543854669315923,
      "grad_norm": 14.551093101501465,
      "learning_rate": 9.146068691456145e-05,
      "loss": 0.0818892240524292,
      "step": 30100
    },
    {
      "epoch": 8.546693159239284,
      "grad_norm": 7.438065528869629,
      "learning_rate": 9.14578484246381e-05,
      "loss": 0.053384286165237424,
      "step": 30110
    },
    {
      "epoch": 8.549531649162645,
      "grad_norm": 10.797991752624512,
      "learning_rate": 9.145500993471474e-05,
      "loss": 0.06102468371391297,
      "step": 30120
    },
    {
      "epoch": 8.552370139086007,
      "grad_norm": 9.833860397338867,
      "learning_rate": 9.145217144479138e-05,
      "loss": 0.0709060251712799,
      "step": 30130
    },
    {
      "epoch": 8.555208629009368,
      "grad_norm": 9.727563858032227,
      "learning_rate": 9.144933295486802e-05,
      "loss": 0.056763797998428345,
      "step": 30140
    },
    {
      "epoch": 8.558047118932727,
      "grad_norm": 12.185494422912598,
      "learning_rate": 9.144649446494465e-05,
      "loss": 0.057816308736801145,
      "step": 30150
    },
    {
      "epoch": 8.560885608856088,
      "grad_norm": 11.085196495056152,
      "learning_rate": 9.144365597502129e-05,
      "loss": 0.08229559659957886,
      "step": 30160
    },
    {
      "epoch": 8.56372409877945,
      "grad_norm": 13.693253517150879,
      "learning_rate": 9.144081748509793e-05,
      "loss": 0.08204988241195679,
      "step": 30170
    },
    {
      "epoch": 8.56656258870281,
      "grad_norm": 10.15668773651123,
      "learning_rate": 9.143797899517457e-05,
      "loss": 0.06671963930130005,
      "step": 30180
    },
    {
      "epoch": 8.569401078626171,
      "grad_norm": 8.673599243164062,
      "learning_rate": 9.143514050525121e-05,
      "loss": 0.04791229963302612,
      "step": 30190
    },
    {
      "epoch": 8.57223956854953,
      "grad_norm": 9.761473655700684,
      "learning_rate": 9.143230201532785e-05,
      "loss": 0.06229044198989868,
      "step": 30200
    },
    {
      "epoch": 8.575078058472892,
      "grad_norm": 6.504861831665039,
      "learning_rate": 9.14294635254045e-05,
      "loss": 0.07771341204643249,
      "step": 30210
    },
    {
      "epoch": 8.577916548396253,
      "grad_norm": 14.052194595336914,
      "learning_rate": 9.142662503548112e-05,
      "loss": 0.0685269832611084,
      "step": 30220
    },
    {
      "epoch": 8.580755038319614,
      "grad_norm": 9.634164810180664,
      "learning_rate": 9.142378654555777e-05,
      "loss": 0.09259766340255737,
      "step": 30230
    },
    {
      "epoch": 8.583593528242975,
      "grad_norm": 11.475071907043457,
      "learning_rate": 9.142094805563441e-05,
      "loss": 0.07257603406906128,
      "step": 30240
    },
    {
      "epoch": 8.586432018166336,
      "grad_norm": 5.292922496795654,
      "learning_rate": 9.141810956571103e-05,
      "loss": 0.07653214931488037,
      "step": 30250
    },
    {
      "epoch": 8.589270508089696,
      "grad_norm": 9.08526611328125,
      "learning_rate": 9.141527107578769e-05,
      "loss": 0.08168586492538452,
      "step": 30260
    },
    {
      "epoch": 8.592108998013057,
      "grad_norm": 5.779264450073242,
      "learning_rate": 9.141243258586433e-05,
      "loss": 0.05767713785171509,
      "step": 30270
    },
    {
      "epoch": 8.594947487936418,
      "grad_norm": 5.022219181060791,
      "learning_rate": 9.140959409594096e-05,
      "loss": 0.05964950919151306,
      "step": 30280
    },
    {
      "epoch": 8.597785977859779,
      "grad_norm": 5.085413455963135,
      "learning_rate": 9.14067556060176e-05,
      "loss": 0.04928558766841888,
      "step": 30290
    },
    {
      "epoch": 8.60062446778314,
      "grad_norm": 13.999133110046387,
      "learning_rate": 9.140391711609424e-05,
      "loss": 0.07592015266418457,
      "step": 30300
    },
    {
      "epoch": 8.6034629577065,
      "grad_norm": 10.314279556274414,
      "learning_rate": 9.140107862617088e-05,
      "loss": 0.08794921636581421,
      "step": 30310
    },
    {
      "epoch": 8.60630144762986,
      "grad_norm": 5.452056884765625,
      "learning_rate": 9.139824013624752e-05,
      "loss": 0.06838034391403199,
      "step": 30320
    },
    {
      "epoch": 8.609139937553222,
      "grad_norm": 12.394155502319336,
      "learning_rate": 9.139540164632417e-05,
      "loss": 0.05805119276046753,
      "step": 30330
    },
    {
      "epoch": 8.611978427476583,
      "grad_norm": 3.0079991817474365,
      "learning_rate": 9.139256315640081e-05,
      "loss": 0.057649272680282596,
      "step": 30340
    },
    {
      "epoch": 8.614816917399944,
      "grad_norm": 3.536961555480957,
      "learning_rate": 9.138972466647743e-05,
      "loss": 0.04464294612407684,
      "step": 30350
    },
    {
      "epoch": 8.617655407323305,
      "grad_norm": 12.767498970031738,
      "learning_rate": 9.138688617655408e-05,
      "loss": 0.07630480527877807,
      "step": 30360
    },
    {
      "epoch": 8.620493897246664,
      "grad_norm": 4.636853218078613,
      "learning_rate": 9.138404768663072e-05,
      "loss": 0.061662924289703366,
      "step": 30370
    },
    {
      "epoch": 8.623332387170025,
      "grad_norm": 11.533599853515625,
      "learning_rate": 9.138120919670735e-05,
      "loss": 0.057594358921051025,
      "step": 30380
    },
    {
      "epoch": 8.626170877093386,
      "grad_norm": 3.9659109115600586,
      "learning_rate": 9.1378370706784e-05,
      "loss": 0.034635528922080994,
      "step": 30390
    },
    {
      "epoch": 8.629009367016748,
      "grad_norm": 5.484599590301514,
      "learning_rate": 9.137553221686064e-05,
      "loss": 0.05701947212219238,
      "step": 30400
    },
    {
      "epoch": 8.631847856940109,
      "grad_norm": 11.841418266296387,
      "learning_rate": 9.137269372693727e-05,
      "loss": 0.05404433608055115,
      "step": 30410
    },
    {
      "epoch": 8.634686346863468,
      "grad_norm": 4.232126235961914,
      "learning_rate": 9.136985523701391e-05,
      "loss": 0.08561909198760986,
      "step": 30420
    },
    {
      "epoch": 8.63752483678683,
      "grad_norm": 4.9084014892578125,
      "learning_rate": 9.136701674709055e-05,
      "loss": 0.052912741899490356,
      "step": 30430
    },
    {
      "epoch": 8.64036332671019,
      "grad_norm": 2.036550760269165,
      "learning_rate": 9.13641782571672e-05,
      "loss": 0.07551093697547913,
      "step": 30440
    },
    {
      "epoch": 8.643201816633551,
      "grad_norm": 12.768614768981934,
      "learning_rate": 9.136133976724382e-05,
      "loss": 0.0871202826499939,
      "step": 30450
    },
    {
      "epoch": 8.646040306556912,
      "grad_norm": 10.173173904418945,
      "learning_rate": 9.135850127732048e-05,
      "loss": 0.0725557565689087,
      "step": 30460
    },
    {
      "epoch": 8.648878796480272,
      "grad_norm": 4.290877819061279,
      "learning_rate": 9.135566278739712e-05,
      "loss": 0.03578062057495117,
      "step": 30470
    },
    {
      "epoch": 8.651717286403633,
      "grad_norm": 10.177022933959961,
      "learning_rate": 9.135282429747375e-05,
      "loss": 0.07076327204704284,
      "step": 30480
    },
    {
      "epoch": 8.654555776326994,
      "grad_norm": 11.57591438293457,
      "learning_rate": 9.134998580755039e-05,
      "loss": 0.08516298532485962,
      "step": 30490
    },
    {
      "epoch": 8.657394266250355,
      "grad_norm": 6.045260906219482,
      "learning_rate": 9.134714731762703e-05,
      "loss": 0.06630693674087525,
      "step": 30500
    },
    {
      "epoch": 8.657394266250355,
      "eval_accuracy": 0.9411839511667832,
      "eval_loss": 0.1725481003522873,
      "eval_runtime": 31.1701,
      "eval_samples_per_second": 504.554,
      "eval_steps_per_second": 7.892,
      "step": 30500
    },
    {
      "epoch": 8.660232756173716,
      "grad_norm": 5.8542022705078125,
      "learning_rate": 9.134430882770366e-05,
      "loss": 0.07063242793083191,
      "step": 30510
    },
    {
      "epoch": 8.663071246097076,
      "grad_norm": 6.192205429077148,
      "learning_rate": 9.134147033778031e-05,
      "loss": 0.07235486507415771,
      "step": 30520
    },
    {
      "epoch": 8.665909736020437,
      "grad_norm": 7.132839679718018,
      "learning_rate": 9.133863184785695e-05,
      "loss": 0.07178015112876893,
      "step": 30530
    },
    {
      "epoch": 8.668748225943798,
      "grad_norm": 5.959991455078125,
      "learning_rate": 9.133579335793358e-05,
      "loss": 0.05568063259124756,
      "step": 30540
    },
    {
      "epoch": 8.671586715867159,
      "grad_norm": 10.190081596374512,
      "learning_rate": 9.133295486801022e-05,
      "loss": 0.06881506443023681,
      "step": 30550
    },
    {
      "epoch": 8.67442520579052,
      "grad_norm": 11.966773986816406,
      "learning_rate": 9.133011637808686e-05,
      "loss": 0.05807017683982849,
      "step": 30560
    },
    {
      "epoch": 8.67726369571388,
      "grad_norm": 8.806239128112793,
      "learning_rate": 9.13272778881635e-05,
      "loss": 0.0771269977092743,
      "step": 30570
    },
    {
      "epoch": 8.68010218563724,
      "grad_norm": 1.1770923137664795,
      "learning_rate": 9.132443939824013e-05,
      "loss": 0.03302620351314545,
      "step": 30580
    },
    {
      "epoch": 8.682940675560602,
      "grad_norm": 16.780710220336914,
      "learning_rate": 9.132160090831679e-05,
      "loss": 0.08350026607513428,
      "step": 30590
    },
    {
      "epoch": 8.685779165483963,
      "grad_norm": 6.750619888305664,
      "learning_rate": 9.131876241839343e-05,
      "loss": 0.09315245151519776,
      "step": 30600
    },
    {
      "epoch": 8.688617655407324,
      "grad_norm": 5.321597576141357,
      "learning_rate": 9.131592392847006e-05,
      "loss": 0.06254659295082092,
      "step": 30610
    },
    {
      "epoch": 8.691456145330685,
      "grad_norm": 6.383153915405273,
      "learning_rate": 9.13130854385467e-05,
      "loss": 0.07338873147964478,
      "step": 30620
    },
    {
      "epoch": 8.694294635254044,
      "grad_norm": 13.765811920166016,
      "learning_rate": 9.131024694862334e-05,
      "loss": 0.05769875645637512,
      "step": 30630
    },
    {
      "epoch": 8.697133125177405,
      "grad_norm": 8.510196685791016,
      "learning_rate": 9.130740845869997e-05,
      "loss": 0.05774034261703491,
      "step": 30640
    },
    {
      "epoch": 8.699971615100766,
      "grad_norm": 9.130453109741211,
      "learning_rate": 9.130456996877661e-05,
      "loss": 0.056436973810195926,
      "step": 30650
    },
    {
      "epoch": 8.702810105024128,
      "grad_norm": 11.908205032348633,
      "learning_rate": 9.130173147885326e-05,
      "loss": 0.0650800883769989,
      "step": 30660
    },
    {
      "epoch": 8.705648594947489,
      "grad_norm": 8.894450187683105,
      "learning_rate": 9.129889298892989e-05,
      "loss": 0.050325775146484376,
      "step": 30670
    },
    {
      "epoch": 8.708487084870848,
      "grad_norm": 4.6478447914123535,
      "learning_rate": 9.129605449900653e-05,
      "loss": 0.06203482151031494,
      "step": 30680
    },
    {
      "epoch": 8.711325574794209,
      "grad_norm": 12.28982162475586,
      "learning_rate": 9.129321600908317e-05,
      "loss": 0.06148065328598022,
      "step": 30690
    },
    {
      "epoch": 8.71416406471757,
      "grad_norm": 8.56557559967041,
      "learning_rate": 9.129037751915982e-05,
      "loss": 0.048805588483810426,
      "step": 30700
    },
    {
      "epoch": 8.717002554640931,
      "grad_norm": 11.844701766967773,
      "learning_rate": 9.128753902923644e-05,
      "loss": 0.07599793076515197,
      "step": 30710
    },
    {
      "epoch": 8.719841044564292,
      "grad_norm": 9.818061828613281,
      "learning_rate": 9.12847005393131e-05,
      "loss": 0.09255142211914062,
      "step": 30720
    },
    {
      "epoch": 8.722679534487652,
      "grad_norm": 8.864899635314941,
      "learning_rate": 9.128186204938973e-05,
      "loss": 0.05062950253486633,
      "step": 30730
    },
    {
      "epoch": 8.725518024411013,
      "grad_norm": 4.842836380004883,
      "learning_rate": 9.127902355946637e-05,
      "loss": 0.05957203507423401,
      "step": 30740
    },
    {
      "epoch": 8.728356514334374,
      "grad_norm": 10.169377326965332,
      "learning_rate": 9.127618506954301e-05,
      "loss": 0.07847896218299866,
      "step": 30750
    },
    {
      "epoch": 8.731195004257735,
      "grad_norm": 10.235345840454102,
      "learning_rate": 9.127334657961965e-05,
      "loss": 0.07809728384017944,
      "step": 30760
    },
    {
      "epoch": 8.734033494181096,
      "grad_norm": 14.762968063354492,
      "learning_rate": 9.127050808969628e-05,
      "loss": 0.07696836590766906,
      "step": 30770
    },
    {
      "epoch": 8.736871984104457,
      "grad_norm": 6.025543689727783,
      "learning_rate": 9.126766959977292e-05,
      "loss": 0.06058669090270996,
      "step": 30780
    },
    {
      "epoch": 8.739710474027817,
      "grad_norm": 11.903536796569824,
      "learning_rate": 9.126483110984957e-05,
      "loss": 0.07239351868629455,
      "step": 30790
    },
    {
      "epoch": 8.742548963951178,
      "grad_norm": 12.460436820983887,
      "learning_rate": 9.12619926199262e-05,
      "loss": 0.07071493268013,
      "step": 30800
    },
    {
      "epoch": 8.745387453874539,
      "grad_norm": 5.232187271118164,
      "learning_rate": 9.125915413000284e-05,
      "loss": 0.05923362970352173,
      "step": 30810
    },
    {
      "epoch": 8.7482259437979,
      "grad_norm": 14.459244728088379,
      "learning_rate": 9.125631564007948e-05,
      "loss": 0.06732058525085449,
      "step": 30820
    },
    {
      "epoch": 8.751064433721261,
      "grad_norm": 3.6833598613739014,
      "learning_rate": 9.125347715015613e-05,
      "loss": 0.058967185020446775,
      "step": 30830
    },
    {
      "epoch": 8.75390292364462,
      "grad_norm": 10.895611763000488,
      "learning_rate": 9.125063866023275e-05,
      "loss": 0.07326056957244872,
      "step": 30840
    },
    {
      "epoch": 8.756741413567982,
      "grad_norm": 11.13486099243164,
      "learning_rate": 9.12478001703094e-05,
      "loss": 0.06982390880584717,
      "step": 30850
    },
    {
      "epoch": 8.759579903491343,
      "grad_norm": 14.711895942687988,
      "learning_rate": 9.124496168038604e-05,
      "loss": 0.05212066173553467,
      "step": 30860
    },
    {
      "epoch": 8.762418393414704,
      "grad_norm": 17.846214294433594,
      "learning_rate": 9.124212319046268e-05,
      "loss": 0.08571950197219849,
      "step": 30870
    },
    {
      "epoch": 8.765256883338065,
      "grad_norm": 12.61149787902832,
      "learning_rate": 9.123928470053932e-05,
      "loss": 0.0673592209815979,
      "step": 30880
    },
    {
      "epoch": 8.768095373261424,
      "grad_norm": 5.6938605308532715,
      "learning_rate": 9.123644621061596e-05,
      "loss": 0.06559733152389527,
      "step": 30890
    },
    {
      "epoch": 8.770933863184785,
      "grad_norm": 6.9619927406311035,
      "learning_rate": 9.123360772069259e-05,
      "loss": 0.0866067111492157,
      "step": 30900
    },
    {
      "epoch": 8.773772353108146,
      "grad_norm": 8.645709037780762,
      "learning_rate": 9.123076923076923e-05,
      "loss": 0.08687164783477783,
      "step": 30910
    },
    {
      "epoch": 8.776610843031508,
      "grad_norm": 5.143220901489258,
      "learning_rate": 9.122793074084588e-05,
      "loss": 0.057608401775360106,
      "step": 30920
    },
    {
      "epoch": 8.779449332954869,
      "grad_norm": 14.712035179138184,
      "learning_rate": 9.122509225092251e-05,
      "loss": 0.06908709406852723,
      "step": 30930
    },
    {
      "epoch": 8.782287822878228,
      "grad_norm": 13.615184783935547,
      "learning_rate": 9.122225376099915e-05,
      "loss": 0.04762253761291504,
      "step": 30940
    },
    {
      "epoch": 8.785126312801589,
      "grad_norm": 2.48268723487854,
      "learning_rate": 9.12194152710758e-05,
      "loss": 0.0639740526676178,
      "step": 30950
    },
    {
      "epoch": 8.78796480272495,
      "grad_norm": 19.646148681640625,
      "learning_rate": 9.121657678115242e-05,
      "loss": 0.07524076700210572,
      "step": 30960
    },
    {
      "epoch": 8.790803292648311,
      "grad_norm": 5.18089485168457,
      "learning_rate": 9.121373829122906e-05,
      "loss": 0.08438354134559631,
      "step": 30970
    },
    {
      "epoch": 8.793641782571672,
      "grad_norm": 5.1031060218811035,
      "learning_rate": 9.12108998013057e-05,
      "loss": 0.05910661220550537,
      "step": 30980
    },
    {
      "epoch": 8.796480272495032,
      "grad_norm": 3.8738019466400146,
      "learning_rate": 9.120806131138235e-05,
      "loss": 0.07147717475891113,
      "step": 30990
    },
    {
      "epoch": 8.799318762418393,
      "grad_norm": 5.0825042724609375,
      "learning_rate": 9.120522282145899e-05,
      "loss": 0.07504554390907288,
      "step": 31000
    },
    {
      "epoch": 8.799318762418393,
      "eval_accuracy": 0.9465886691676735,
      "eval_loss": 0.15738148987293243,
      "eval_runtime": 31.6307,
      "eval_samples_per_second": 497.206,
      "eval_steps_per_second": 7.777,
      "step": 31000
    },
    {
      "epoch": 8.802157252341754,
      "grad_norm": 10.514293670654297,
      "learning_rate": 9.120238433153563e-05,
      "loss": 0.06329176425933838,
      "step": 31010
    },
    {
      "epoch": 8.804995742265115,
      "grad_norm": 9.732636451721191,
      "learning_rate": 9.119954584161227e-05,
      "loss": 0.06165165901184082,
      "step": 31020
    },
    {
      "epoch": 8.807834232188476,
      "grad_norm": 10.218039512634277,
      "learning_rate": 9.11967073516889e-05,
      "loss": 0.07378641963005066,
      "step": 31030
    },
    {
      "epoch": 8.810672722111837,
      "grad_norm": 20.488651275634766,
      "learning_rate": 9.119386886176554e-05,
      "loss": 0.07996610403060914,
      "step": 31040
    },
    {
      "epoch": 8.813511212035197,
      "grad_norm": 10.727422714233398,
      "learning_rate": 9.119103037184218e-05,
      "loss": 0.07219599485397339,
      "step": 31050
    },
    {
      "epoch": 8.816349701958558,
      "grad_norm": 10.90174388885498,
      "learning_rate": 9.118819188191882e-05,
      "loss": 0.06459282040596008,
      "step": 31060
    },
    {
      "epoch": 8.819188191881919,
      "grad_norm": 9.382819175720215,
      "learning_rate": 9.118535339199546e-05,
      "loss": 0.07012121677398682,
      "step": 31070
    },
    {
      "epoch": 8.82202668180528,
      "grad_norm": 4.837339401245117,
      "learning_rate": 9.11825149020721e-05,
      "loss": 0.05027868151664734,
      "step": 31080
    },
    {
      "epoch": 8.824865171728641,
      "grad_norm": 4.266017913818359,
      "learning_rate": 9.117967641214873e-05,
      "loss": 0.04700014591217041,
      "step": 31090
    },
    {
      "epoch": 8.827703661652,
      "grad_norm": 10.60668659210205,
      "learning_rate": 9.117683792222538e-05,
      "loss": 0.061681056022644044,
      "step": 31100
    },
    {
      "epoch": 8.830542151575361,
      "grad_norm": 4.687035083770752,
      "learning_rate": 9.117399943230202e-05,
      "loss": 0.08623862266540527,
      "step": 31110
    },
    {
      "epoch": 8.833380641498723,
      "grad_norm": 9.520359992980957,
      "learning_rate": 9.117116094237866e-05,
      "loss": 0.048452964425086974,
      "step": 31120
    },
    {
      "epoch": 8.836219131422084,
      "grad_norm": 7.0932841300964355,
      "learning_rate": 9.11683224524553e-05,
      "loss": 0.062233710289001466,
      "step": 31130
    },
    {
      "epoch": 8.839057621345445,
      "grad_norm": 6.89607048034668,
      "learning_rate": 9.116548396253194e-05,
      "loss": 0.06079059839248657,
      "step": 31140
    },
    {
      "epoch": 8.841896111268806,
      "grad_norm": 10.693526268005371,
      "learning_rate": 9.116264547260858e-05,
      "loss": 0.05661168098449707,
      "step": 31150
    },
    {
      "epoch": 8.844734601192165,
      "grad_norm": 12.32718563079834,
      "learning_rate": 9.115980698268521e-05,
      "loss": 0.043204781413078305,
      "step": 31160
    },
    {
      "epoch": 8.847573091115526,
      "grad_norm": 4.099190711975098,
      "learning_rate": 9.115696849276185e-05,
      "loss": 0.06582169532775879,
      "step": 31170
    },
    {
      "epoch": 8.850411581038887,
      "grad_norm": 10.13891887664795,
      "learning_rate": 9.115413000283849e-05,
      "loss": 0.09330734014511108,
      "step": 31180
    },
    {
      "epoch": 8.853250070962249,
      "grad_norm": 4.434497356414795,
      "learning_rate": 9.115129151291513e-05,
      "loss": 0.05179306268692017,
      "step": 31190
    },
    {
      "epoch": 8.85608856088561,
      "grad_norm": 7.782890796661377,
      "learning_rate": 9.114845302299178e-05,
      "loss": 0.06201858520507812,
      "step": 31200
    },
    {
      "epoch": 8.858927050808969,
      "grad_norm": 11.122188568115234,
      "learning_rate": 9.114561453306842e-05,
      "loss": 0.05274918079376221,
      "step": 31210
    },
    {
      "epoch": 8.86176554073233,
      "grad_norm": 9.690204620361328,
      "learning_rate": 9.114277604314504e-05,
      "loss": 0.06978282928466797,
      "step": 31220
    },
    {
      "epoch": 8.864604030655691,
      "grad_norm": 14.059823036193848,
      "learning_rate": 9.113993755322169e-05,
      "loss": 0.053569084405899046,
      "step": 31230
    },
    {
      "epoch": 8.867442520579052,
      "grad_norm": 9.883621215820312,
      "learning_rate": 9.113709906329833e-05,
      "loss": 0.08523458242416382,
      "step": 31240
    },
    {
      "epoch": 8.870281010502413,
      "grad_norm": 9.064912796020508,
      "learning_rate": 9.113426057337497e-05,
      "loss": 0.054639339447021484,
      "step": 31250
    },
    {
      "epoch": 8.873119500425773,
      "grad_norm": 17.543582916259766,
      "learning_rate": 9.113142208345161e-05,
      "loss": 0.05767755508422852,
      "step": 31260
    },
    {
      "epoch": 8.875957990349134,
      "grad_norm": 12.58343505859375,
      "learning_rate": 9.112858359352825e-05,
      "loss": 0.05623639822006225,
      "step": 31270
    },
    {
      "epoch": 8.878796480272495,
      "grad_norm": 14.024517059326172,
      "learning_rate": 9.112574510360489e-05,
      "loss": 0.08230687975883484,
      "step": 31280
    },
    {
      "epoch": 8.881634970195856,
      "grad_norm": 14.335714340209961,
      "learning_rate": 9.112290661368152e-05,
      "loss": 0.07999399900436402,
      "step": 31290
    },
    {
      "epoch": 8.884473460119217,
      "grad_norm": 6.547296047210693,
      "learning_rate": 9.112006812375816e-05,
      "loss": 0.07681872844696044,
      "step": 31300
    },
    {
      "epoch": 8.887311950042577,
      "grad_norm": 12.506217002868652,
      "learning_rate": 9.11172296338348e-05,
      "loss": 0.06401631832122803,
      "step": 31310
    },
    {
      "epoch": 8.890150439965938,
      "grad_norm": 3.479335308074951,
      "learning_rate": 9.111439114391144e-05,
      "loss": 0.06286961436271668,
      "step": 31320
    },
    {
      "epoch": 8.892988929889299,
      "grad_norm": 9.801219940185547,
      "learning_rate": 9.111155265398809e-05,
      "loss": 0.08583763837814332,
      "step": 31330
    },
    {
      "epoch": 8.89582741981266,
      "grad_norm": 12.058196067810059,
      "learning_rate": 9.110871416406473e-05,
      "loss": 0.09592686891555786,
      "step": 31340
    },
    {
      "epoch": 8.898665909736021,
      "grad_norm": 11.869256019592285,
      "learning_rate": 9.110587567414136e-05,
      "loss": 0.07629110813140869,
      "step": 31350
    },
    {
      "epoch": 8.90150439965938,
      "grad_norm": 1.2793084383010864,
      "learning_rate": 9.1103037184218e-05,
      "loss": 0.059134817123413085,
      "step": 31360
    },
    {
      "epoch": 8.904342889582741,
      "grad_norm": 14.228399276733398,
      "learning_rate": 9.110019869429464e-05,
      "loss": 0.08088141679763794,
      "step": 31370
    },
    {
      "epoch": 8.907181379506103,
      "grad_norm": 17.807186126708984,
      "learning_rate": 9.109736020437128e-05,
      "loss": 0.06523163318634033,
      "step": 31380
    },
    {
      "epoch": 8.910019869429464,
      "grad_norm": 13.873000144958496,
      "learning_rate": 9.109452171444792e-05,
      "loss": 0.07417306303977966,
      "step": 31390
    },
    {
      "epoch": 8.912858359352825,
      "grad_norm": 8.632597923278809,
      "learning_rate": 9.109168322452456e-05,
      "loss": 0.06533147096633911,
      "step": 31400
    },
    {
      "epoch": 8.915696849276186,
      "grad_norm": 8.661585807800293,
      "learning_rate": 9.10888447346012e-05,
      "loss": 0.07017588019371032,
      "step": 31410
    },
    {
      "epoch": 8.918535339199545,
      "grad_norm": 14.436474800109863,
      "learning_rate": 9.108600624467783e-05,
      "loss": 0.07135186791419983,
      "step": 31420
    },
    {
      "epoch": 8.921373829122906,
      "grad_norm": 5.942589282989502,
      "learning_rate": 9.108316775475447e-05,
      "loss": 0.07056043148040772,
      "step": 31430
    },
    {
      "epoch": 8.924212319046267,
      "grad_norm": 8.080760955810547,
      "learning_rate": 9.108032926483111e-05,
      "loss": 0.04907607734203338,
      "step": 31440
    },
    {
      "epoch": 8.927050808969629,
      "grad_norm": 8.866616249084473,
      "learning_rate": 9.107749077490776e-05,
      "loss": 0.061613088846206664,
      "step": 31450
    },
    {
      "epoch": 8.92988929889299,
      "grad_norm": 4.63857889175415,
      "learning_rate": 9.10746522849844e-05,
      "loss": 0.06444783210754394,
      "step": 31460
    },
    {
      "epoch": 8.932727788816349,
      "grad_norm": 19.896678924560547,
      "learning_rate": 9.107181379506104e-05,
      "loss": 0.07473809123039246,
      "step": 31470
    },
    {
      "epoch": 8.93556627873971,
      "grad_norm": 7.841392993927002,
      "learning_rate": 9.106897530513767e-05,
      "loss": 0.053362196683883666,
      "step": 31480
    },
    {
      "epoch": 8.938404768663071,
      "grad_norm": 16.415979385375977,
      "learning_rate": 9.106613681521431e-05,
      "loss": 0.07507288455963135,
      "step": 31490
    },
    {
      "epoch": 8.941243258586432,
      "grad_norm": 6.388270378112793,
      "learning_rate": 9.106329832529095e-05,
      "loss": 0.07571802735328674,
      "step": 31500
    },
    {
      "epoch": 8.941243258586432,
      "eval_accuracy": 0.9469701786736187,
      "eval_loss": 0.1587066799402237,
      "eval_runtime": 31.7736,
      "eval_samples_per_second": 494.971,
      "eval_steps_per_second": 7.742,
      "step": 31500
    },
    {
      "epoch": 8.944081748509793,
      "grad_norm": 7.662382125854492,
      "learning_rate": 9.106045983536759e-05,
      "loss": 0.03651776909828186,
      "step": 31510
    },
    {
      "epoch": 8.946920238433153,
      "grad_norm": 14.817533493041992,
      "learning_rate": 9.105762134544423e-05,
      "loss": 0.07683525085449219,
      "step": 31520
    },
    {
      "epoch": 8.949758728356514,
      "grad_norm": 6.961455345153809,
      "learning_rate": 9.105478285552087e-05,
      "loss": 0.0702526330947876,
      "step": 31530
    },
    {
      "epoch": 8.952597218279875,
      "grad_norm": 11.14152717590332,
      "learning_rate": 9.105194436559751e-05,
      "loss": 0.06614710092544555,
      "step": 31540
    },
    {
      "epoch": 8.955435708203236,
      "grad_norm": 15.313032150268555,
      "learning_rate": 9.104910587567414e-05,
      "loss": 0.04896753430366516,
      "step": 31550
    },
    {
      "epoch": 8.958274198126597,
      "grad_norm": 7.498221397399902,
      "learning_rate": 9.104626738575078e-05,
      "loss": 0.04692138135433197,
      "step": 31560
    },
    {
      "epoch": 8.961112688049958,
      "grad_norm": 8.560870170593262,
      "learning_rate": 9.104342889582743e-05,
      "loss": 0.05697489380836487,
      "step": 31570
    },
    {
      "epoch": 8.963951177973318,
      "grad_norm": 5.503041744232178,
      "learning_rate": 9.104059040590405e-05,
      "loss": 0.08263733983039856,
      "step": 31580
    },
    {
      "epoch": 8.966789667896679,
      "grad_norm": 2.8306033611297607,
      "learning_rate": 9.103775191598071e-05,
      "loss": 0.061256974935531616,
      "step": 31590
    },
    {
      "epoch": 8.96962815782004,
      "grad_norm": 5.988136291503906,
      "learning_rate": 9.103491342605735e-05,
      "loss": 0.060560721158981326,
      "step": 31600
    },
    {
      "epoch": 8.972466647743401,
      "grad_norm": 3.7903780937194824,
      "learning_rate": 9.103207493613398e-05,
      "loss": 0.05420640707015991,
      "step": 31610
    },
    {
      "epoch": 8.975305137666762,
      "grad_norm": 3.0052037239074707,
      "learning_rate": 9.102923644621062e-05,
      "loss": 0.044920125603675844,
      "step": 31620
    },
    {
      "epoch": 8.978143627590121,
      "grad_norm": 4.551146030426025,
      "learning_rate": 9.102639795628726e-05,
      "loss": 0.06018396019935608,
      "step": 31630
    },
    {
      "epoch": 8.980982117513483,
      "grad_norm": 4.385807514190674,
      "learning_rate": 9.102384331535624e-05,
      "loss": 0.057430309057235715,
      "step": 31640
    },
    {
      "epoch": 8.983820607436844,
      "grad_norm": 9.01045036315918,
      "learning_rate": 9.102100482543288e-05,
      "loss": 0.07642671465873718,
      "step": 31650
    },
    {
      "epoch": 8.986659097360205,
      "grad_norm": 15.248885154724121,
      "learning_rate": 9.101816633550951e-05,
      "loss": 0.06820961236953735,
      "step": 31660
    },
    {
      "epoch": 8.989497587283566,
      "grad_norm": 6.19985818862915,
      "learning_rate": 9.101532784558615e-05,
      "loss": 0.04741819202899933,
      "step": 31670
    },
    {
      "epoch": 8.992336077206925,
      "grad_norm": 26.0809326171875,
      "learning_rate": 9.101248935566279e-05,
      "loss": 0.05679764747619629,
      "step": 31680
    },
    {
      "epoch": 8.995174567130286,
      "grad_norm": 13.292773246765137,
      "learning_rate": 9.100965086573943e-05,
      "loss": 0.06299090385437012,
      "step": 31690
    },
    {
      "epoch": 8.998013057053647,
      "grad_norm": 5.700288772583008,
      "learning_rate": 9.100681237581607e-05,
      "loss": 0.08798794746398926,
      "step": 31700
    },
    {
      "epoch": 9.000851546977009,
      "grad_norm": 9.686714172363281,
      "learning_rate": 9.100397388589272e-05,
      "loss": 0.04719998836517334,
      "step": 31710
    },
    {
      "epoch": 9.00369003690037,
      "grad_norm": 1.6230088472366333,
      "learning_rate": 9.100113539596934e-05,
      "loss": 0.0502514123916626,
      "step": 31720
    },
    {
      "epoch": 9.006528526823729,
      "grad_norm": 7.6409592628479,
      "learning_rate": 9.099829690604598e-05,
      "loss": 0.05189587473869324,
      "step": 31730
    },
    {
      "epoch": 9.00936701674709,
      "grad_norm": 14.42635726928711,
      "learning_rate": 9.099545841612263e-05,
      "loss": 0.06049224138259888,
      "step": 31740
    },
    {
      "epoch": 9.012205506670451,
      "grad_norm": 12.75652027130127,
      "learning_rate": 9.099261992619927e-05,
      "loss": 0.06303011775016784,
      "step": 31750
    },
    {
      "epoch": 9.015043996593812,
      "grad_norm": 12.73111343383789,
      "learning_rate": 9.09897814362759e-05,
      "loss": 0.05389610528945923,
      "step": 31760
    },
    {
      "epoch": 9.017882486517173,
      "grad_norm": 11.562397003173828,
      "learning_rate": 9.098694294635255e-05,
      "loss": 0.06323632597923279,
      "step": 31770
    },
    {
      "epoch": 9.020720976440534,
      "grad_norm": 8.386837005615234,
      "learning_rate": 9.098410445642919e-05,
      "loss": 0.03240850865840912,
      "step": 31780
    },
    {
      "epoch": 9.023559466363894,
      "grad_norm": 7.020870685577393,
      "learning_rate": 9.098126596650582e-05,
      "loss": 0.06059011220932007,
      "step": 31790
    },
    {
      "epoch": 9.026397956287255,
      "grad_norm": 12.727704048156738,
      "learning_rate": 9.097842747658246e-05,
      "loss": 0.035716494917869566,
      "step": 31800
    },
    {
      "epoch": 9.029236446210616,
      "grad_norm": 12.288534164428711,
      "learning_rate": 9.09755889866591e-05,
      "loss": 0.06254916787147521,
      "step": 31810
    },
    {
      "epoch": 9.032074936133977,
      "grad_norm": 7.2569403648376465,
      "learning_rate": 9.097275049673573e-05,
      "loss": 0.05126778483390808,
      "step": 31820
    },
    {
      "epoch": 9.034913426057338,
      "grad_norm": 8.454544067382812,
      "learning_rate": 9.096991200681239e-05,
      "loss": 0.042856672406196596,
      "step": 31830
    },
    {
      "epoch": 9.037751915980698,
      "grad_norm": 13.909622192382812,
      "learning_rate": 9.096707351688903e-05,
      "loss": 0.056902194023132326,
      "step": 31840
    },
    {
      "epoch": 9.040590405904059,
      "grad_norm": 11.164444923400879,
      "learning_rate": 9.096423502696565e-05,
      "loss": 0.05609207153320313,
      "step": 31850
    },
    {
      "epoch": 9.04342889582742,
      "grad_norm": 2.568427801132202,
      "learning_rate": 9.09613965370423e-05,
      "loss": 0.039089816808700564,
      "step": 31860
    },
    {
      "epoch": 9.046267385750781,
      "grad_norm": 7.224270820617676,
      "learning_rate": 9.095855804711894e-05,
      "loss": 0.050432974100112916,
      "step": 31870
    },
    {
      "epoch": 9.049105875674142,
      "grad_norm": 4.621863842010498,
      "learning_rate": 9.095571955719558e-05,
      "loss": 0.04566062092781067,
      "step": 31880
    },
    {
      "epoch": 9.051944365597501,
      "grad_norm": 4.847370147705078,
      "learning_rate": 9.09528810672722e-05,
      "loss": 0.03859710693359375,
      "step": 31890
    },
    {
      "epoch": 9.054782855520862,
      "grad_norm": 6.624361991882324,
      "learning_rate": 9.095004257734886e-05,
      "loss": 0.04116395115852356,
      "step": 31900
    },
    {
      "epoch": 9.057621345444224,
      "grad_norm": 13.439946174621582,
      "learning_rate": 9.09472040874255e-05,
      "loss": 0.04496969282627106,
      "step": 31910
    },
    {
      "epoch": 9.060459835367585,
      "grad_norm": 7.528670787811279,
      "learning_rate": 9.094436559750213e-05,
      "loss": 0.03843676149845123,
      "step": 31920
    },
    {
      "epoch": 9.063298325290946,
      "grad_norm": 8.01745319366455,
      "learning_rate": 9.094152710757877e-05,
      "loss": 0.059415769577026364,
      "step": 31930
    },
    {
      "epoch": 9.066136815214305,
      "grad_norm": 0.7322397828102112,
      "learning_rate": 9.093868861765541e-05,
      "loss": 0.04222564697265625,
      "step": 31940
    },
    {
      "epoch": 9.068975305137666,
      "grad_norm": 3.2487258911132812,
      "learning_rate": 9.093585012773204e-05,
      "loss": 0.04935858547687531,
      "step": 31950
    },
    {
      "epoch": 9.071813795061027,
      "grad_norm": 5.705401420593262,
      "learning_rate": 9.093301163780868e-05,
      "loss": 0.03522341847419739,
      "step": 31960
    },
    {
      "epoch": 9.074652284984388,
      "grad_norm": 12.268989562988281,
      "learning_rate": 9.093017314788534e-05,
      "loss": 0.0877863883972168,
      "step": 31970
    },
    {
      "epoch": 9.07749077490775,
      "grad_norm": 3.4812443256378174,
      "learning_rate": 9.092733465796197e-05,
      "loss": 0.04092869162559509,
      "step": 31980
    },
    {
      "epoch": 9.08032926483111,
      "grad_norm": 3.5843749046325684,
      "learning_rate": 9.09244961680386e-05,
      "loss": 0.047165191173553465,
      "step": 31990
    },
    {
      "epoch": 9.08316775475447,
      "grad_norm": 7.594471454620361,
      "learning_rate": 9.092165767811525e-05,
      "loss": 0.050229012966156006,
      "step": 32000
    },
    {
      "epoch": 9.08316775475447,
      "eval_accuracy": 0.9454441406498378,
      "eval_loss": 0.16055257618427277,
      "eval_runtime": 31.6102,
      "eval_samples_per_second": 497.529,
      "eval_steps_per_second": 7.782,
      "step": 32000
    },
    {
      "epoch": 9.086006244677831,
      "grad_norm": 4.322275638580322,
      "learning_rate": 9.091881918819189e-05,
      "loss": 0.06291774511337281,
      "step": 32010
    },
    {
      "epoch": 9.088844734601192,
      "grad_norm": 9.507974624633789,
      "learning_rate": 9.091598069826852e-05,
      "loss": 0.049620908498764035,
      "step": 32020
    },
    {
      "epoch": 9.091683224524553,
      "grad_norm": 3.4975383281707764,
      "learning_rate": 9.091314220834517e-05,
      "loss": 0.05312735438346863,
      "step": 32030
    },
    {
      "epoch": 9.094521714447914,
      "grad_norm": 8.572249412536621,
      "learning_rate": 9.091030371842181e-05,
      "loss": 0.06213139295578003,
      "step": 32040
    },
    {
      "epoch": 9.097360204371274,
      "grad_norm": 12.718893051147461,
      "learning_rate": 9.090746522849844e-05,
      "loss": 0.0554810881614685,
      "step": 32050
    },
    {
      "epoch": 9.100198694294635,
      "grad_norm": 5.842963218688965,
      "learning_rate": 9.090462673857508e-05,
      "loss": 0.05603500008583069,
      "step": 32060
    },
    {
      "epoch": 9.103037184217996,
      "grad_norm": 11.779271125793457,
      "learning_rate": 9.090178824865172e-05,
      "loss": 0.05315329432487488,
      "step": 32070
    },
    {
      "epoch": 9.105875674141357,
      "grad_norm": 14.351799964904785,
      "learning_rate": 9.089894975872835e-05,
      "loss": 0.05615080595016479,
      "step": 32080
    },
    {
      "epoch": 9.108714164064718,
      "grad_norm": 2.0403223037719727,
      "learning_rate": 9.089611126880499e-05,
      "loss": 0.03992740511894226,
      "step": 32090
    },
    {
      "epoch": 9.111552653988078,
      "grad_norm": 5.808372497558594,
      "learning_rate": 9.089327277888165e-05,
      "loss": 0.05110638737678528,
      "step": 32100
    },
    {
      "epoch": 9.114391143911439,
      "grad_norm": 5.363349437713623,
      "learning_rate": 9.089043428895828e-05,
      "loss": 0.044322311878204346,
      "step": 32110
    },
    {
      "epoch": 9.1172296338348,
      "grad_norm": 13.359721183776855,
      "learning_rate": 9.088759579903492e-05,
      "loss": 0.042077115178108214,
      "step": 32120
    },
    {
      "epoch": 9.120068123758161,
      "grad_norm": 10.45478630065918,
      "learning_rate": 9.088475730911156e-05,
      "loss": 0.05763462781906128,
      "step": 32130
    },
    {
      "epoch": 9.122906613681522,
      "grad_norm": 15.195608139038086,
      "learning_rate": 9.08819188191882e-05,
      "loss": 0.06402850151062012,
      "step": 32140
    },
    {
      "epoch": 9.125745103604881,
      "grad_norm": 11.2763671875,
      "learning_rate": 9.087908032926483e-05,
      "loss": 0.06520623564720154,
      "step": 32150
    },
    {
      "epoch": 9.128583593528242,
      "grad_norm": 6.463578701019287,
      "learning_rate": 9.087624183934147e-05,
      "loss": 0.0754274606704712,
      "step": 32160
    },
    {
      "epoch": 9.131422083451604,
      "grad_norm": 3.423412322998047,
      "learning_rate": 9.087340334941812e-05,
      "loss": 0.041135784983634946,
      "step": 32170
    },
    {
      "epoch": 9.134260573374965,
      "grad_norm": 10.44631576538086,
      "learning_rate": 9.087056485949475e-05,
      "loss": 0.05105924606323242,
      "step": 32180
    },
    {
      "epoch": 9.137099063298326,
      "grad_norm": 13.660996437072754,
      "learning_rate": 9.08677263695714e-05,
      "loss": 0.0788940131664276,
      "step": 32190
    },
    {
      "epoch": 9.139937553221687,
      "grad_norm": 8.275731086730957,
      "learning_rate": 9.086488787964803e-05,
      "loss": 0.04814446866512299,
      "step": 32200
    },
    {
      "epoch": 9.142776043145046,
      "grad_norm": 6.043083667755127,
      "learning_rate": 9.086204938972466e-05,
      "loss": 0.05335206389427185,
      "step": 32210
    },
    {
      "epoch": 9.145614533068407,
      "grad_norm": 14.26015567779541,
      "learning_rate": 9.08592108998013e-05,
      "loss": 0.04197232127189636,
      "step": 32220
    },
    {
      "epoch": 9.148453022991768,
      "grad_norm": 7.298044204711914,
      "learning_rate": 9.085637240987796e-05,
      "loss": 0.05922012329101563,
      "step": 32230
    },
    {
      "epoch": 9.15129151291513,
      "grad_norm": 8.097308158874512,
      "learning_rate": 9.085353391995459e-05,
      "loss": 0.052116990089416504,
      "step": 32240
    },
    {
      "epoch": 9.15413000283849,
      "grad_norm": 4.339701175689697,
      "learning_rate": 9.085069543003123e-05,
      "loss": 0.05344792604446411,
      "step": 32250
    },
    {
      "epoch": 9.15696849276185,
      "grad_norm": 6.495077133178711,
      "learning_rate": 9.084785694010787e-05,
      "loss": 0.055792659521102905,
      "step": 32260
    },
    {
      "epoch": 9.159806982685211,
      "grad_norm": 2.3581790924072266,
      "learning_rate": 9.084501845018451e-05,
      "loss": 0.051355397701263426,
      "step": 32270
    },
    {
      "epoch": 9.162645472608572,
      "grad_norm": 10.864691734313965,
      "learning_rate": 9.084217996026114e-05,
      "loss": 0.07544721364974975,
      "step": 32280
    },
    {
      "epoch": 9.165483962531933,
      "grad_norm": 3.968423366546631,
      "learning_rate": 9.083934147033778e-05,
      "loss": 0.047580179572105405,
      "step": 32290
    },
    {
      "epoch": 9.168322452455294,
      "grad_norm": 4.634256362915039,
      "learning_rate": 9.083650298041443e-05,
      "loss": 0.04713979363441467,
      "step": 32300
    },
    {
      "epoch": 9.171160942378654,
      "grad_norm": 7.158009052276611,
      "learning_rate": 9.083366449049106e-05,
      "loss": 0.07584148645401001,
      "step": 32310
    },
    {
      "epoch": 9.173999432302015,
      "grad_norm": 15.426690101623535,
      "learning_rate": 9.08308260005677e-05,
      "loss": 0.06314955353736877,
      "step": 32320
    },
    {
      "epoch": 9.176837922225376,
      "grad_norm": 3.2448484897613525,
      "learning_rate": 9.082798751064435e-05,
      "loss": 0.03707973659038544,
      "step": 32330
    },
    {
      "epoch": 9.179676412148737,
      "grad_norm": 3.8451483249664307,
      "learning_rate": 9.082514902072097e-05,
      "loss": 0.04627393186092377,
      "step": 32340
    },
    {
      "epoch": 9.182514902072098,
      "grad_norm": 5.8484907150268555,
      "learning_rate": 9.082231053079761e-05,
      "loss": 0.046934878826141356,
      "step": 32350
    },
    {
      "epoch": 9.18535339199546,
      "grad_norm": 6.43878173828125,
      "learning_rate": 9.081947204087426e-05,
      "loss": 0.04995456337928772,
      "step": 32360
    },
    {
      "epoch": 9.188191881918819,
      "grad_norm": 13.413664817810059,
      "learning_rate": 9.08166335509509e-05,
      "loss": 0.06470543146133423,
      "step": 32370
    },
    {
      "epoch": 9.19103037184218,
      "grad_norm": 5.617048740386963,
      "learning_rate": 9.081379506102754e-05,
      "loss": 0.03900338113307953,
      "step": 32380
    },
    {
      "epoch": 9.19386886176554,
      "grad_norm": 12.342789649963379,
      "learning_rate": 9.081095657110418e-05,
      "loss": 0.07401307821273803,
      "step": 32390
    },
    {
      "epoch": 9.196707351688902,
      "grad_norm": 4.099523067474365,
      "learning_rate": 9.080811808118082e-05,
      "loss": 0.05901659727096557,
      "step": 32400
    },
    {
      "epoch": 9.199545841612263,
      "grad_norm": 1.8025932312011719,
      "learning_rate": 9.080527959125745e-05,
      "loss": 0.040928953886032106,
      "step": 32410
    },
    {
      "epoch": 9.202384331535622,
      "grad_norm": 7.9925994873046875,
      "learning_rate": 9.080244110133409e-05,
      "loss": 0.04325757920742035,
      "step": 32420
    },
    {
      "epoch": 9.205222821458984,
      "grad_norm": 13.091001510620117,
      "learning_rate": 9.079960261141075e-05,
      "loss": 0.05394256114959717,
      "step": 32430
    },
    {
      "epoch": 9.208061311382345,
      "grad_norm": 9.377684593200684,
      "learning_rate": 9.079676412148737e-05,
      "loss": 0.05074498653411865,
      "step": 32440
    },
    {
      "epoch": 9.210899801305706,
      "grad_norm": 12.04058837890625,
      "learning_rate": 9.079392563156401e-05,
      "loss": 0.054376494884490964,
      "step": 32450
    },
    {
      "epoch": 9.213738291229067,
      "grad_norm": 8.228126525878906,
      "learning_rate": 9.079108714164066e-05,
      "loss": 0.0486117035150528,
      "step": 32460
    },
    {
      "epoch": 9.216576781152426,
      "grad_norm": 12.644847869873047,
      "learning_rate": 9.078824865171728e-05,
      "loss": 0.05557645559310913,
      "step": 32470
    },
    {
      "epoch": 9.219415271075787,
      "grad_norm": 7.676670074462891,
      "learning_rate": 9.078541016179393e-05,
      "loss": 0.03925705552101135,
      "step": 32480
    },
    {
      "epoch": 9.222253760999148,
      "grad_norm": 12.223360061645508,
      "learning_rate": 9.078257167187057e-05,
      "loss": 0.049240988492965695,
      "step": 32490
    },
    {
      "epoch": 9.22509225092251,
      "grad_norm": 4.789877891540527,
      "learning_rate": 9.077973318194721e-05,
      "loss": 0.05932843089103699,
      "step": 32500
    },
    {
      "epoch": 9.22509225092251,
      "eval_accuracy": 0.9498314999682076,
      "eval_loss": 0.15024346113204956,
      "eval_runtime": 31.2491,
      "eval_samples_per_second": 503.278,
      "eval_steps_per_second": 7.872,
      "step": 32500
    },
    {
      "epoch": 9.22793074084587,
      "grad_norm": 8.23260498046875,
      "learning_rate": 9.077689469202385e-05,
      "loss": 0.04186435043811798,
      "step": 32510
    },
    {
      "epoch": 9.23076923076923,
      "grad_norm": 6.626314640045166,
      "learning_rate": 9.077405620210049e-05,
      "loss": 0.045269107818603514,
      "step": 32520
    },
    {
      "epoch": 9.233607720692591,
      "grad_norm": 9.466409683227539,
      "learning_rate": 9.077121771217713e-05,
      "loss": 0.08306272625923157,
      "step": 32530
    },
    {
      "epoch": 9.236446210615952,
      "grad_norm": 9.16460132598877,
      "learning_rate": 9.076837922225376e-05,
      "loss": 0.04933281540870667,
      "step": 32540
    },
    {
      "epoch": 9.239284700539313,
      "grad_norm": 10.519789695739746,
      "learning_rate": 9.07655407323304e-05,
      "loss": 0.055536919832229616,
      "step": 32550
    },
    {
      "epoch": 9.242123190462674,
      "grad_norm": 7.917647838592529,
      "learning_rate": 9.076270224240704e-05,
      "loss": 0.031090694665908813,
      "step": 32560
    },
    {
      "epoch": 9.244961680386035,
      "grad_norm": 6.369583606719971,
      "learning_rate": 9.075986375248368e-05,
      "loss": 0.06728399395942689,
      "step": 32570
    },
    {
      "epoch": 9.247800170309395,
      "grad_norm": 6.490670204162598,
      "learning_rate": 9.075702526256033e-05,
      "loss": 0.06461179852485657,
      "step": 32580
    },
    {
      "epoch": 9.250638660232756,
      "grad_norm": 1.9340163469314575,
      "learning_rate": 9.075418677263697e-05,
      "loss": 0.058265691995620726,
      "step": 32590
    },
    {
      "epoch": 9.253477150156117,
      "grad_norm": 12.019217491149902,
      "learning_rate": 9.07513482827136e-05,
      "loss": 0.0459725558757782,
      "step": 32600
    },
    {
      "epoch": 9.256315640079478,
      "grad_norm": 8.567548751831055,
      "learning_rate": 9.074850979279024e-05,
      "loss": 0.06528127193450928,
      "step": 32610
    },
    {
      "epoch": 9.25915413000284,
      "grad_norm": 10.61620807647705,
      "learning_rate": 9.074567130286688e-05,
      "loss": 0.07091454863548279,
      "step": 32620
    },
    {
      "epoch": 9.261992619926199,
      "grad_norm": 3.3433566093444824,
      "learning_rate": 9.074283281294352e-05,
      "loss": 0.05981482863426209,
      "step": 32630
    },
    {
      "epoch": 9.26483110984956,
      "grad_norm": 12.406513214111328,
      "learning_rate": 9.073999432302016e-05,
      "loss": 0.059479260444641115,
      "step": 32640
    },
    {
      "epoch": 9.26766959977292,
      "grad_norm": 3.91577410697937,
      "learning_rate": 9.07371558330968e-05,
      "loss": 0.06922081112861633,
      "step": 32650
    },
    {
      "epoch": 9.270508089696282,
      "grad_norm": 5.703088283538818,
      "learning_rate": 9.073431734317343e-05,
      "loss": 0.048567506670951846,
      "step": 32660
    },
    {
      "epoch": 9.273346579619643,
      "grad_norm": 13.659531593322754,
      "learning_rate": 9.073147885325007e-05,
      "loss": 0.05960379838943482,
      "step": 32670
    },
    {
      "epoch": 9.276185069543002,
      "grad_norm": 10.7163724899292,
      "learning_rate": 9.072864036332671e-05,
      "loss": 0.04003342390060425,
      "step": 32680
    },
    {
      "epoch": 9.279023559466363,
      "grad_norm": 4.531045913696289,
      "learning_rate": 9.072580187340335e-05,
      "loss": 0.046465244889259336,
      "step": 32690
    },
    {
      "epoch": 9.281862049389725,
      "grad_norm": 2.041239023208618,
      "learning_rate": 9.072296338348e-05,
      "loss": 0.04149287343025208,
      "step": 32700
    },
    {
      "epoch": 9.284700539313086,
      "grad_norm": 11.627174377441406,
      "learning_rate": 9.072012489355664e-05,
      "loss": 0.05404342412948608,
      "step": 32710
    },
    {
      "epoch": 9.287539029236447,
      "grad_norm": 7.9445481300354,
      "learning_rate": 9.071728640363328e-05,
      "loss": 0.05792953372001648,
      "step": 32720
    },
    {
      "epoch": 9.290377519159806,
      "grad_norm": 6.924063682556152,
      "learning_rate": 9.07144479137099e-05,
      "loss": 0.07755460143089295,
      "step": 32730
    },
    {
      "epoch": 9.293216009083167,
      "grad_norm": 6.975043773651123,
      "learning_rate": 9.071160942378655e-05,
      "loss": 0.04379885494709015,
      "step": 32740
    },
    {
      "epoch": 9.296054499006528,
      "grad_norm": 15.474916458129883,
      "learning_rate": 9.070877093386319e-05,
      "loss": 0.06385596990585327,
      "step": 32750
    },
    {
      "epoch": 9.29889298892989,
      "grad_norm": 4.132717609405518,
      "learning_rate": 9.070593244393982e-05,
      "loss": 0.0586206316947937,
      "step": 32760
    },
    {
      "epoch": 9.30173147885325,
      "grad_norm": 20.231489181518555,
      "learning_rate": 9.070309395401647e-05,
      "loss": 0.06682396531105042,
      "step": 32770
    },
    {
      "epoch": 9.304569968776612,
      "grad_norm": 9.4744873046875,
      "learning_rate": 9.070025546409311e-05,
      "loss": 0.05048024654388428,
      "step": 32780
    },
    {
      "epoch": 9.307408458699971,
      "grad_norm": 2.6996655464172363,
      "learning_rate": 9.069741697416974e-05,
      "loss": 0.052340883016586306,
      "step": 32790
    },
    {
      "epoch": 9.310246948623332,
      "grad_norm": 11.197591781616211,
      "learning_rate": 9.069457848424638e-05,
      "loss": 0.04950667023658752,
      "step": 32800
    },
    {
      "epoch": 9.313085438546693,
      "grad_norm": 7.194085597991943,
      "learning_rate": 9.069173999432302e-05,
      "loss": 0.04353128969669342,
      "step": 32810
    },
    {
      "epoch": 9.315923928470054,
      "grad_norm": 10.58753490447998,
      "learning_rate": 9.068890150439966e-05,
      "loss": 0.05287941694259644,
      "step": 32820
    },
    {
      "epoch": 9.318762418393415,
      "grad_norm": 5.493806838989258,
      "learning_rate": 9.06860630144763e-05,
      "loss": 0.04841550588607788,
      "step": 32830
    },
    {
      "epoch": 9.321600908316775,
      "grad_norm": 10.897929191589355,
      "learning_rate": 9.068322452455295e-05,
      "loss": 0.058511865139007566,
      "step": 32840
    },
    {
      "epoch": 9.324439398240136,
      "grad_norm": 8.787210464477539,
      "learning_rate": 9.068038603462959e-05,
      "loss": 0.04149903357028961,
      "step": 32850
    },
    {
      "epoch": 9.327277888163497,
      "grad_norm": 6.68208122253418,
      "learning_rate": 9.067754754470622e-05,
      "loss": 0.05815345048904419,
      "step": 32860
    },
    {
      "epoch": 9.330116378086858,
      "grad_norm": 13.823956489562988,
      "learning_rate": 9.067470905478286e-05,
      "loss": 0.04482123851776123,
      "step": 32870
    },
    {
      "epoch": 9.33295486801022,
      "grad_norm": 8.48888874053955,
      "learning_rate": 9.06718705648595e-05,
      "loss": 0.06808430552482606,
      "step": 32880
    },
    {
      "epoch": 9.335793357933579,
      "grad_norm": 10.772937774658203,
      "learning_rate": 9.066903207493613e-05,
      "loss": 0.05200949907302856,
      "step": 32890
    },
    {
      "epoch": 9.33863184785694,
      "grad_norm": 1.882460117340088,
      "learning_rate": 9.066619358501278e-05,
      "loss": 0.034025412797927854,
      "step": 32900
    },
    {
      "epoch": 9.3414703377803,
      "grad_norm": 6.333077430725098,
      "learning_rate": 9.066335509508942e-05,
      "loss": 0.04800796806812287,
      "step": 32910
    },
    {
      "epoch": 9.344308827703662,
      "grad_norm": 2.586176633834839,
      "learning_rate": 9.066051660516605e-05,
      "loss": 0.05263393521308899,
      "step": 32920
    },
    {
      "epoch": 9.347147317627023,
      "grad_norm": 9.180237770080566,
      "learning_rate": 9.065767811524269e-05,
      "loss": 0.031053465604782105,
      "step": 32930
    },
    {
      "epoch": 9.349985807550382,
      "grad_norm": 8.368448257446289,
      "learning_rate": 9.065483962531933e-05,
      "loss": 0.04113053679466248,
      "step": 32940
    },
    {
      "epoch": 9.352824297473743,
      "grad_norm": 3.159412145614624,
      "learning_rate": 9.065200113539598e-05,
      "loss": 0.05956799983978271,
      "step": 32950
    },
    {
      "epoch": 9.355662787397105,
      "grad_norm": 19.962419509887695,
      "learning_rate": 9.064916264547262e-05,
      "loss": 0.07392004132270813,
      "step": 32960
    },
    {
      "epoch": 9.358501277320466,
      "grad_norm": 9.92551040649414,
      "learning_rate": 9.064632415554926e-05,
      "loss": 0.055903536081314084,
      "step": 32970
    },
    {
      "epoch": 9.361339767243827,
      "grad_norm": 8.83159065246582,
      "learning_rate": 9.06434856656259e-05,
      "loss": 0.045094051957130434,
      "step": 32980
    },
    {
      "epoch": 9.364178257167188,
      "grad_norm": 9.9266996383667,
      "learning_rate": 9.064064717570253e-05,
      "loss": 0.058740997314453126,
      "step": 32990
    },
    {
      "epoch": 9.367016747090547,
      "grad_norm": 9.501269340515137,
      "learning_rate": 9.063780868577917e-05,
      "loss": 0.04406949281692505,
      "step": 33000
    },
    {
      "epoch": 9.367016747090547,
      "eval_accuracy": 0.9465886691676735,
      "eval_loss": 0.1648065596818924,
      "eval_runtime": 31.3549,
      "eval_samples_per_second": 501.581,
      "eval_steps_per_second": 7.846,
      "step": 33000
    },
    {
      "epoch": 9.369855237013908,
      "grad_norm": 9.468525886535645,
      "learning_rate": 9.063497019585581e-05,
      "loss": 0.048831897974014285,
      "step": 33010
    },
    {
      "epoch": 9.37269372693727,
      "grad_norm": 7.1258625984191895,
      "learning_rate": 9.063213170593244e-05,
      "loss": 0.05858026742935181,
      "step": 33020
    },
    {
      "epoch": 9.37553221686063,
      "grad_norm": 10.655607223510742,
      "learning_rate": 9.062929321600909e-05,
      "loss": 0.05381351709365845,
      "step": 33030
    },
    {
      "epoch": 9.378370706783992,
      "grad_norm": 12.196785926818848,
      "learning_rate": 9.062645472608573e-05,
      "loss": 0.06687752604484558,
      "step": 33040
    },
    {
      "epoch": 9.381209196707351,
      "grad_norm": 5.432117462158203,
      "learning_rate": 9.062361623616236e-05,
      "loss": 0.06028173565864563,
      "step": 33050
    },
    {
      "epoch": 9.384047686630712,
      "grad_norm": 8.175498962402344,
      "learning_rate": 9.0620777746239e-05,
      "loss": 0.0691449522972107,
      "step": 33060
    },
    {
      "epoch": 9.386886176554073,
      "grad_norm": 13.365309715270996,
      "learning_rate": 9.061793925631564e-05,
      "loss": 0.07226872444152832,
      "step": 33070
    },
    {
      "epoch": 9.389724666477434,
      "grad_norm": 14.634222030639648,
      "learning_rate": 9.061510076639229e-05,
      "loss": 0.07034959197044373,
      "step": 33080
    },
    {
      "epoch": 9.392563156400795,
      "grad_norm": 3.979567766189575,
      "learning_rate": 9.061226227646891e-05,
      "loss": 0.05294480323791504,
      "step": 33090
    },
    {
      "epoch": 9.395401646324155,
      "grad_norm": 13.566909790039062,
      "learning_rate": 9.060942378654557e-05,
      "loss": 0.06107956767082214,
      "step": 33100
    },
    {
      "epoch": 9.398240136247516,
      "grad_norm": 3.6039998531341553,
      "learning_rate": 9.060658529662221e-05,
      "loss": 0.062139981985092164,
      "step": 33110
    },
    {
      "epoch": 9.401078626170877,
      "grad_norm": 11.181325912475586,
      "learning_rate": 9.060374680669884e-05,
      "loss": 0.059360653162002563,
      "step": 33120
    },
    {
      "epoch": 9.403917116094238,
      "grad_norm": 6.055405139923096,
      "learning_rate": 9.060090831677548e-05,
      "loss": 0.07335441112518311,
      "step": 33130
    },
    {
      "epoch": 9.4067556060176,
      "grad_norm": 4.620993614196777,
      "learning_rate": 9.059806982685212e-05,
      "loss": 0.07250437140464783,
      "step": 33140
    },
    {
      "epoch": 9.40959409594096,
      "grad_norm": 1.5898244380950928,
      "learning_rate": 9.059523133692875e-05,
      "loss": 0.0533543586730957,
      "step": 33150
    },
    {
      "epoch": 9.41243258586432,
      "grad_norm": 10.768118858337402,
      "learning_rate": 9.05923928470054e-05,
      "loss": 0.05850784778594971,
      "step": 33160
    },
    {
      "epoch": 9.41527107578768,
      "grad_norm": 7.327179908752441,
      "learning_rate": 9.058955435708204e-05,
      "loss": 0.043669885396957396,
      "step": 33170
    },
    {
      "epoch": 9.418109565711042,
      "grad_norm": 8.013859748840332,
      "learning_rate": 9.058671586715867e-05,
      "loss": 0.06741069555282593,
      "step": 33180
    },
    {
      "epoch": 9.420948055634403,
      "grad_norm": 9.66065502166748,
      "learning_rate": 9.058387737723531e-05,
      "loss": 0.07110794782638549,
      "step": 33190
    },
    {
      "epoch": 9.423786545557764,
      "grad_norm": 13.936553001403809,
      "learning_rate": 9.058103888731196e-05,
      "loss": 0.07287751436233521,
      "step": 33200
    },
    {
      "epoch": 9.426625035481123,
      "grad_norm": 7.313204288482666,
      "learning_rate": 9.05782003973886e-05,
      "loss": 0.05224173665046692,
      "step": 33210
    },
    {
      "epoch": 9.429463525404485,
      "grad_norm": 11.710868835449219,
      "learning_rate": 9.057536190746522e-05,
      "loss": 0.056077378988265994,
      "step": 33220
    },
    {
      "epoch": 9.432302015327846,
      "grad_norm": 1.254156470298767,
      "learning_rate": 9.057252341754188e-05,
      "loss": 0.032727751135826114,
      "step": 33230
    },
    {
      "epoch": 9.435140505251207,
      "grad_norm": 1.8861327171325684,
      "learning_rate": 9.056968492761852e-05,
      "loss": 0.05641490817070007,
      "step": 33240
    },
    {
      "epoch": 9.437978995174568,
      "grad_norm": 9.412788391113281,
      "learning_rate": 9.056684643769515e-05,
      "loss": 0.053233766555786134,
      "step": 33250
    },
    {
      "epoch": 9.440817485097927,
      "grad_norm": 2.569307804107666,
      "learning_rate": 9.056400794777179e-05,
      "loss": 0.05567408800125122,
      "step": 33260
    },
    {
      "epoch": 9.443655975021288,
      "grad_norm": 7.04046630859375,
      "learning_rate": 9.056116945784843e-05,
      "loss": 0.04117984473705292,
      "step": 33270
    },
    {
      "epoch": 9.44649446494465,
      "grad_norm": 13.325481414794922,
      "learning_rate": 9.055833096792506e-05,
      "loss": 0.06516222953796387,
      "step": 33280
    },
    {
      "epoch": 9.44933295486801,
      "grad_norm": 5.740192413330078,
      "learning_rate": 9.05554924780017e-05,
      "loss": 0.05689252018928528,
      "step": 33290
    },
    {
      "epoch": 9.452171444791372,
      "grad_norm": 12.770515441894531,
      "learning_rate": 9.055265398807836e-05,
      "loss": 0.07921002507209778,
      "step": 33300
    },
    {
      "epoch": 9.455009934714731,
      "grad_norm": 13.46126937866211,
      "learning_rate": 9.054981549815498e-05,
      "loss": 0.04767411649227142,
      "step": 33310
    },
    {
      "epoch": 9.457848424638092,
      "grad_norm": 5.489933013916016,
      "learning_rate": 9.054697700823162e-05,
      "loss": 0.047058111429214476,
      "step": 33320
    },
    {
      "epoch": 9.460686914561453,
      "grad_norm": 9.937371253967285,
      "learning_rate": 9.054413851830827e-05,
      "loss": 0.07372314929962158,
      "step": 33330
    },
    {
      "epoch": 9.463525404484814,
      "grad_norm": 9.115100860595703,
      "learning_rate": 9.054130002838491e-05,
      "loss": 0.0654799461364746,
      "step": 33340
    },
    {
      "epoch": 9.466363894408175,
      "grad_norm": 3.8474485874176025,
      "learning_rate": 9.053846153846154e-05,
      "loss": 0.04887437522411346,
      "step": 33350
    },
    {
      "epoch": 9.469202384331535,
      "grad_norm": 11.678000450134277,
      "learning_rate": 9.053562304853819e-05,
      "loss": 0.07052799463272094,
      "step": 33360
    },
    {
      "epoch": 9.472040874254896,
      "grad_norm": 16.68780517578125,
      "learning_rate": 9.053278455861483e-05,
      "loss": 0.04608502984046936,
      "step": 33370
    },
    {
      "epoch": 9.474879364178257,
      "grad_norm": 3.5634920597076416,
      "learning_rate": 9.052994606869146e-05,
      "loss": 0.0610964834690094,
      "step": 33380
    },
    {
      "epoch": 9.477717854101618,
      "grad_norm": 7.769752025604248,
      "learning_rate": 9.05271075787681e-05,
      "loss": 0.05623708963394165,
      "step": 33390
    },
    {
      "epoch": 9.48055634402498,
      "grad_norm": 14.619163513183594,
      "learning_rate": 9.052426908884474e-05,
      "loss": 0.05393269658088684,
      "step": 33400
    },
    {
      "epoch": 9.48339483394834,
      "grad_norm": 9.452289581298828,
      "learning_rate": 9.052143059892137e-05,
      "loss": 0.058954554796218875,
      "step": 33410
    },
    {
      "epoch": 9.4862333238717,
      "grad_norm": 3.0336191654205322,
      "learning_rate": 9.051859210899801e-05,
      "loss": 0.04946458339691162,
      "step": 33420
    },
    {
      "epoch": 9.48907181379506,
      "grad_norm": 9.975210189819336,
      "learning_rate": 9.051575361907467e-05,
      "loss": 0.07156325578689575,
      "step": 33430
    },
    {
      "epoch": 9.491910303718422,
      "grad_norm": 9.630119323730469,
      "learning_rate": 9.05129151291513e-05,
      "loss": 0.05123568773269653,
      "step": 33440
    },
    {
      "epoch": 9.494748793641783,
      "grad_norm": 6.102807998657227,
      "learning_rate": 9.051007663922794e-05,
      "loss": 0.05549406409263611,
      "step": 33450
    },
    {
      "epoch": 9.497587283565144,
      "grad_norm": 0.8345799446105957,
      "learning_rate": 9.050723814930458e-05,
      "loss": 0.04638853669166565,
      "step": 33460
    },
    {
      "epoch": 9.500425773488503,
      "grad_norm": 6.340814113616943,
      "learning_rate": 9.050439965938122e-05,
      "loss": 0.06161945462226868,
      "step": 33470
    },
    {
      "epoch": 9.503264263411864,
      "grad_norm": 8.755582809448242,
      "learning_rate": 9.050156116945785e-05,
      "loss": 0.05433828234672546,
      "step": 33480
    },
    {
      "epoch": 9.506102753335226,
      "grad_norm": 8.362161636352539,
      "learning_rate": 9.049872267953449e-05,
      "loss": 0.06090704798698425,
      "step": 33490
    },
    {
      "epoch": 9.508941243258587,
      "grad_norm": 6.977360248565674,
      "learning_rate": 9.049588418961113e-05,
      "loss": 0.05114209055900574,
      "step": 33500
    },
    {
      "epoch": 9.508941243258587,
      "eval_accuracy": 0.9557448973103579,
      "eval_loss": 0.13861849904060364,
      "eval_runtime": 31.7276,
      "eval_samples_per_second": 495.688,
      "eval_steps_per_second": 7.754,
      "step": 33500
    },
    {
      "epoch": 9.511779733181948,
      "grad_norm": 14.03275203704834,
      "learning_rate": 9.049304569968777e-05,
      "loss": 0.04922421872615814,
      "step": 33510
    },
    {
      "epoch": 9.514618223105309,
      "grad_norm": 10.364019393920898,
      "learning_rate": 9.049020720976441e-05,
      "loss": 0.06954495906829834,
      "step": 33520
    },
    {
      "epoch": 9.517456713028668,
      "grad_norm": 5.165099143981934,
      "learning_rate": 9.048736871984105e-05,
      "loss": 0.05066389441490173,
      "step": 33530
    },
    {
      "epoch": 9.52029520295203,
      "grad_norm": 12.195990562438965,
      "learning_rate": 9.048453022991768e-05,
      "loss": 0.06648446917533875,
      "step": 33540
    },
    {
      "epoch": 9.52313369287539,
      "grad_norm": 4.532296180725098,
      "learning_rate": 9.048169173999432e-05,
      "loss": 0.04216625392436981,
      "step": 33550
    },
    {
      "epoch": 9.525972182798752,
      "grad_norm": 16.543903350830078,
      "learning_rate": 9.047885325007098e-05,
      "loss": 0.0777590274810791,
      "step": 33560
    },
    {
      "epoch": 9.528810672722113,
      "grad_norm": 7.024462699890137,
      "learning_rate": 9.04760147601476e-05,
      "loss": 0.05962986946105957,
      "step": 33570
    },
    {
      "epoch": 9.531649162645472,
      "grad_norm": 8.659052848815918,
      "learning_rate": 9.047317627022425e-05,
      "loss": 0.03709257543087006,
      "step": 33580
    },
    {
      "epoch": 9.534487652568833,
      "grad_norm": 3.186353921890259,
      "learning_rate": 9.047033778030089e-05,
      "loss": 0.05533115863800049,
      "step": 33590
    },
    {
      "epoch": 9.537326142492194,
      "grad_norm": 5.910744667053223,
      "learning_rate": 9.046749929037752e-05,
      "loss": 0.05815193653106689,
      "step": 33600
    },
    {
      "epoch": 9.540164632415555,
      "grad_norm": 12.982107162475586,
      "learning_rate": 9.046466080045416e-05,
      "loss": 0.06229819655418396,
      "step": 33610
    },
    {
      "epoch": 9.543003122338916,
      "grad_norm": 5.328309059143066,
      "learning_rate": 9.04618223105308e-05,
      "loss": 0.05786202549934387,
      "step": 33620
    },
    {
      "epoch": 9.545841612262276,
      "grad_norm": 8.800302505493164,
      "learning_rate": 9.045898382060744e-05,
      "loss": 0.06661202311515808,
      "step": 33630
    },
    {
      "epoch": 9.548680102185637,
      "grad_norm": 2.952171802520752,
      "learning_rate": 9.045614533068408e-05,
      "loss": 0.041045427322387695,
      "step": 33640
    },
    {
      "epoch": 9.551518592108998,
      "grad_norm": 9.967278480529785,
      "learning_rate": 9.045330684076072e-05,
      "loss": 0.09409021735191345,
      "step": 33650
    },
    {
      "epoch": 9.55435708203236,
      "grad_norm": 8.845858573913574,
      "learning_rate": 9.045046835083736e-05,
      "loss": 0.07691789865493774,
      "step": 33660
    },
    {
      "epoch": 9.55719557195572,
      "grad_norm": 4.441326141357422,
      "learning_rate": 9.044762986091399e-05,
      "loss": 0.03699399530887604,
      "step": 33670
    },
    {
      "epoch": 9.56003406187908,
      "grad_norm": 15.421638488769531,
      "learning_rate": 9.044479137099063e-05,
      "loss": 0.05435764789581299,
      "step": 33680
    },
    {
      "epoch": 9.56287255180244,
      "grad_norm": 10.45661735534668,
      "learning_rate": 9.044195288106727e-05,
      "loss": 0.04014127254486084,
      "step": 33690
    },
    {
      "epoch": 9.565711041725802,
      "grad_norm": 12.669248580932617,
      "learning_rate": 9.043911439114392e-05,
      "loss": 0.058434033393859865,
      "step": 33700
    },
    {
      "epoch": 9.568549531649163,
      "grad_norm": 7.6707305908203125,
      "learning_rate": 9.043627590122056e-05,
      "loss": 0.0526405930519104,
      "step": 33710
    },
    {
      "epoch": 9.571388021572524,
      "grad_norm": 7.266836643218994,
      "learning_rate": 9.04334374112972e-05,
      "loss": 0.05215244293212891,
      "step": 33720
    },
    {
      "epoch": 9.574226511495883,
      "grad_norm": 6.0221357345581055,
      "learning_rate": 9.043059892137383e-05,
      "loss": 0.06672795414924622,
      "step": 33730
    },
    {
      "epoch": 9.577065001419244,
      "grad_norm": 26.31820297241211,
      "learning_rate": 9.042776043145047e-05,
      "loss": 0.08382744789123535,
      "step": 33740
    },
    {
      "epoch": 9.579903491342606,
      "grad_norm": 10.193788528442383,
      "learning_rate": 9.042492194152711e-05,
      "loss": 0.06574383974075318,
      "step": 33750
    },
    {
      "epoch": 9.582741981265967,
      "grad_norm": 9.864354133605957,
      "learning_rate": 9.042208345160375e-05,
      "loss": 0.08076583743095397,
      "step": 33760
    },
    {
      "epoch": 9.585580471189328,
      "grad_norm": 7.901949405670166,
      "learning_rate": 9.041924496168039e-05,
      "loss": 0.07552215456962585,
      "step": 33770
    },
    {
      "epoch": 9.588418961112689,
      "grad_norm": 12.871236801147461,
      "learning_rate": 9.041640647175703e-05,
      "loss": 0.05840908288955689,
      "step": 33780
    },
    {
      "epoch": 9.591257451036048,
      "grad_norm": 9.155179023742676,
      "learning_rate": 9.041356798183367e-05,
      "loss": 0.03758316934108734,
      "step": 33790
    },
    {
      "epoch": 9.59409594095941,
      "grad_norm": 3.980708122253418,
      "learning_rate": 9.04107294919103e-05,
      "loss": 0.04662624001502991,
      "step": 33800
    },
    {
      "epoch": 9.59693443088277,
      "grad_norm": 9.541356086730957,
      "learning_rate": 9.040789100198694e-05,
      "loss": 0.04631110429763794,
      "step": 33810
    },
    {
      "epoch": 9.599772920806132,
      "grad_norm": 8.988570213317871,
      "learning_rate": 9.040505251206359e-05,
      "loss": 0.0743104338645935,
      "step": 33820
    },
    {
      "epoch": 9.602611410729493,
      "grad_norm": 9.99904727935791,
      "learning_rate": 9.040221402214023e-05,
      "loss": 0.04648450016975403,
      "step": 33830
    },
    {
      "epoch": 9.605449900652852,
      "grad_norm": 9.594966888427734,
      "learning_rate": 9.039937553221687e-05,
      "loss": 0.05980173349380493,
      "step": 33840
    },
    {
      "epoch": 9.608288390576213,
      "grad_norm": 14.48553466796875,
      "learning_rate": 9.039653704229351e-05,
      "loss": 0.05360459089279175,
      "step": 33850
    },
    {
      "epoch": 9.611126880499574,
      "grad_norm": 9.922717094421387,
      "learning_rate": 9.039369855237014e-05,
      "loss": 0.06500089168548584,
      "step": 33860
    },
    {
      "epoch": 9.613965370422935,
      "grad_norm": 4.235474586486816,
      "learning_rate": 9.039114391143912e-05,
      "loss": 0.07516565322875976,
      "step": 33870
    },
    {
      "epoch": 9.616803860346296,
      "grad_norm": 11.98862361907959,
      "learning_rate": 9.038830542151576e-05,
      "loss": 0.05682286620140076,
      "step": 33880
    },
    {
      "epoch": 9.619642350269656,
      "grad_norm": 16.991994857788086,
      "learning_rate": 9.03854669315924e-05,
      "loss": 0.07232513427734374,
      "step": 33890
    },
    {
      "epoch": 9.622480840193017,
      "grad_norm": 7.825690746307373,
      "learning_rate": 9.038262844166904e-05,
      "loss": 0.11338208913803101,
      "step": 33900
    },
    {
      "epoch": 9.625319330116378,
      "grad_norm": 14.34385871887207,
      "learning_rate": 9.037978995174567e-05,
      "loss": 0.06633058786392212,
      "step": 33910
    },
    {
      "epoch": 9.628157820039739,
      "grad_norm": 14.932887077331543,
      "learning_rate": 9.037695146182231e-05,
      "loss": 0.056937789916992186,
      "step": 33920
    },
    {
      "epoch": 9.6309963099631,
      "grad_norm": 14.428411483764648,
      "learning_rate": 9.037411297189895e-05,
      "loss": 0.058716845512390134,
      "step": 33930
    },
    {
      "epoch": 9.633834799886461,
      "grad_norm": 10.574261665344238,
      "learning_rate": 9.037127448197559e-05,
      "loss": 0.0521108090877533,
      "step": 33940
    },
    {
      "epoch": 9.63667328980982,
      "grad_norm": 8.565613746643066,
      "learning_rate": 9.036843599205223e-05,
      "loss": 0.04207794368267059,
      "step": 33950
    },
    {
      "epoch": 9.639511779733182,
      "grad_norm": 12.022870063781738,
      "learning_rate": 9.036559750212888e-05,
      "loss": 0.045295464992523196,
      "step": 33960
    },
    {
      "epoch": 9.642350269656543,
      "grad_norm": 5.8100738525390625,
      "learning_rate": 9.036275901220552e-05,
      "loss": 0.0781801164150238,
      "step": 33970
    },
    {
      "epoch": 9.645188759579904,
      "grad_norm": 4.381354331970215,
      "learning_rate": 9.035992052228214e-05,
      "loss": 0.05108208060264587,
      "step": 33980
    },
    {
      "epoch": 9.648027249503265,
      "grad_norm": 5.99945068359375,
      "learning_rate": 9.035708203235879e-05,
      "loss": 0.044146662950515746,
      "step": 33990
    },
    {
      "epoch": 9.650865739426624,
      "grad_norm": 11.849239349365234,
      "learning_rate": 9.035424354243543e-05,
      "loss": 0.05025023818016052,
      "step": 34000
    },
    {
      "epoch": 9.650865739426624,
      "eval_accuracy": 0.947606027850194,
      "eval_loss": 0.15701857209205627,
      "eval_runtime": 31.2038,
      "eval_samples_per_second": 504.01,
      "eval_steps_per_second": 7.884,
      "step": 34000
    },
    {
      "epoch": 9.653704229349986,
      "grad_norm": 4.164856910705566,
      "learning_rate": 9.035140505251207e-05,
      "loss": 0.052230226993560794,
      "step": 34010
    },
    {
      "epoch": 9.656542719273347,
      "grad_norm": 10.305496215820312,
      "learning_rate": 9.034856656258871e-05,
      "loss": 0.06024667024612427,
      "step": 34020
    },
    {
      "epoch": 9.659381209196708,
      "grad_norm": 8.605240821838379,
      "learning_rate": 9.034572807266535e-05,
      "loss": 0.05980364680290222,
      "step": 34030
    },
    {
      "epoch": 9.662219699120069,
      "grad_norm": 2.591973304748535,
      "learning_rate": 9.034288958274198e-05,
      "loss": 0.05305702686309814,
      "step": 34040
    },
    {
      "epoch": 9.665058189043428,
      "grad_norm": 12.5418119430542,
      "learning_rate": 9.034005109281862e-05,
      "loss": 0.0470085084438324,
      "step": 34050
    },
    {
      "epoch": 9.66789667896679,
      "grad_norm": 7.019689083099365,
      "learning_rate": 9.033721260289526e-05,
      "loss": 0.043849459290504454,
      "step": 34060
    },
    {
      "epoch": 9.67073516889015,
      "grad_norm": 1.362231969833374,
      "learning_rate": 9.03343741129719e-05,
      "loss": 0.04370751082897186,
      "step": 34070
    },
    {
      "epoch": 9.673573658813511,
      "grad_norm": 5.060967922210693,
      "learning_rate": 9.033153562304855e-05,
      "loss": 0.04780924916267395,
      "step": 34080
    },
    {
      "epoch": 9.676412148736873,
      "grad_norm": 9.045281410217285,
      "learning_rate": 9.032869713312519e-05,
      "loss": 0.06363303661346435,
      "step": 34090
    },
    {
      "epoch": 9.679250638660232,
      "grad_norm": 12.295304298400879,
      "learning_rate": 9.032585864320183e-05,
      "loss": 0.0758301079273224,
      "step": 34100
    },
    {
      "epoch": 9.682089128583593,
      "grad_norm": 10.919517517089844,
      "learning_rate": 9.032302015327846e-05,
      "loss": 0.05505637526512146,
      "step": 34110
    },
    {
      "epoch": 9.684927618506954,
      "grad_norm": 3.8185107707977295,
      "learning_rate": 9.03201816633551e-05,
      "loss": 0.05205299258232117,
      "step": 34120
    },
    {
      "epoch": 9.687766108430315,
      "grad_norm": 5.767138481140137,
      "learning_rate": 9.031734317343174e-05,
      "loss": 0.044072318077087405,
      "step": 34130
    },
    {
      "epoch": 9.690604598353676,
      "grad_norm": 10.324020385742188,
      "learning_rate": 9.031450468350838e-05,
      "loss": 0.049072015285491946,
      "step": 34140
    },
    {
      "epoch": 9.693443088277036,
      "grad_norm": 3.8596479892730713,
      "learning_rate": 9.031166619358502e-05,
      "loss": 0.040961083769798276,
      "step": 34150
    },
    {
      "epoch": 9.696281578200397,
      "grad_norm": 6.289798736572266,
      "learning_rate": 9.030882770366166e-05,
      "loss": 0.06501933336257934,
      "step": 34160
    },
    {
      "epoch": 9.699120068123758,
      "grad_norm": 9.705649375915527,
      "learning_rate": 9.030598921373829e-05,
      "loss": 0.049810847640037535,
      "step": 34170
    },
    {
      "epoch": 9.701958558047119,
      "grad_norm": 6.887966156005859,
      "learning_rate": 9.030315072381493e-05,
      "loss": 0.047608527541160586,
      "step": 34180
    },
    {
      "epoch": 9.70479704797048,
      "grad_norm": 6.2285075187683105,
      "learning_rate": 9.030031223389157e-05,
      "loss": 0.07033423781394958,
      "step": 34190
    },
    {
      "epoch": 9.707635537893841,
      "grad_norm": 14.393875122070312,
      "learning_rate": 9.029747374396821e-05,
      "loss": 0.06578875184059144,
      "step": 34200
    },
    {
      "epoch": 9.7104740278172,
      "grad_norm": 11.480899810791016,
      "learning_rate": 9.029463525404486e-05,
      "loss": 0.04567984938621521,
      "step": 34210
    },
    {
      "epoch": 9.713312517740562,
      "grad_norm": 23.58464241027832,
      "learning_rate": 9.02917967641215e-05,
      "loss": 0.08886916041374207,
      "step": 34220
    },
    {
      "epoch": 9.716151007663923,
      "grad_norm": 1.8707438707351685,
      "learning_rate": 9.028895827419814e-05,
      "loss": 0.07297916412353515,
      "step": 34230
    },
    {
      "epoch": 9.718989497587284,
      "grad_norm": 9.00274658203125,
      "learning_rate": 9.028611978427477e-05,
      "loss": 0.03945229649543762,
      "step": 34240
    },
    {
      "epoch": 9.721827987510645,
      "grad_norm": 3.684415102005005,
      "learning_rate": 9.028328129435141e-05,
      "loss": 0.06324402689933777,
      "step": 34250
    },
    {
      "epoch": 9.724666477434004,
      "grad_norm": 10.152517318725586,
      "learning_rate": 9.028044280442805e-05,
      "loss": 0.05714346766471863,
      "step": 34260
    },
    {
      "epoch": 9.727504967357365,
      "grad_norm": 10.9435453414917,
      "learning_rate": 9.027760431450468e-05,
      "loss": 0.0816379427909851,
      "step": 34270
    },
    {
      "epoch": 9.730343457280727,
      "grad_norm": 7.825002193450928,
      "learning_rate": 9.027476582458133e-05,
      "loss": 0.07541795969009399,
      "step": 34280
    },
    {
      "epoch": 9.733181947204088,
      "grad_norm": 2.87214994430542,
      "learning_rate": 9.027192733465797e-05,
      "loss": 0.06760282516479492,
      "step": 34290
    },
    {
      "epoch": 9.736020437127449,
      "grad_norm": 6.221203327178955,
      "learning_rate": 9.02690888447346e-05,
      "loss": 0.0756977915763855,
      "step": 34300
    },
    {
      "epoch": 9.73885892705081,
      "grad_norm": 8.128471374511719,
      "learning_rate": 9.026625035481124e-05,
      "loss": 0.0395295262336731,
      "step": 34310
    },
    {
      "epoch": 9.74169741697417,
      "grad_norm": 9.556492805480957,
      "learning_rate": 9.026341186488788e-05,
      "loss": 0.03885287344455719,
      "step": 34320
    },
    {
      "epoch": 9.74453590689753,
      "grad_norm": 10.452277183532715,
      "learning_rate": 9.026057337496453e-05,
      "loss": 0.0708901047706604,
      "step": 34330
    },
    {
      "epoch": 9.747374396820891,
      "grad_norm": 1.3048969507217407,
      "learning_rate": 9.025773488504117e-05,
      "loss": 0.03583475351333618,
      "step": 34340
    },
    {
      "epoch": 9.750212886744253,
      "grad_norm": 15.01528263092041,
      "learning_rate": 9.025489639511781e-05,
      "loss": 0.054342222213745114,
      "step": 34350
    },
    {
      "epoch": 9.753051376667614,
      "grad_norm": 14.763592720031738,
      "learning_rate": 9.025205790519444e-05,
      "loss": 0.0733025074005127,
      "step": 34360
    },
    {
      "epoch": 9.755889866590973,
      "grad_norm": 7.0914692878723145,
      "learning_rate": 9.024921941527108e-05,
      "loss": 0.051036524772644046,
      "step": 34370
    },
    {
      "epoch": 9.758728356514334,
      "grad_norm": 5.289018630981445,
      "learning_rate": 9.024638092534772e-05,
      "loss": 0.043183648586273195,
      "step": 34380
    },
    {
      "epoch": 9.761566846437695,
      "grad_norm": 2.602764844894409,
      "learning_rate": 9.024354243542436e-05,
      "loss": 0.06717556118965148,
      "step": 34390
    },
    {
      "epoch": 9.764405336361056,
      "grad_norm": 12.55064582824707,
      "learning_rate": 9.024070394550099e-05,
      "loss": 0.05532287359237671,
      "step": 34400
    },
    {
      "epoch": 9.767243826284417,
      "grad_norm": 4.019501686096191,
      "learning_rate": 9.023786545557764e-05,
      "loss": 0.08732059597969055,
      "step": 34410
    },
    {
      "epoch": 9.770082316207777,
      "grad_norm": 12.261041641235352,
      "learning_rate": 9.023502696565428e-05,
      "loss": 0.061834710836410525,
      "step": 34420
    },
    {
      "epoch": 9.772920806131138,
      "grad_norm": 11.338926315307617,
      "learning_rate": 9.023218847573091e-05,
      "loss": 0.05416260957717896,
      "step": 34430
    },
    {
      "epoch": 9.775759296054499,
      "grad_norm": 11.139217376708984,
      "learning_rate": 9.022934998580755e-05,
      "loss": 0.06826131343841553,
      "step": 34440
    },
    {
      "epoch": 9.77859778597786,
      "grad_norm": 5.79235315322876,
      "learning_rate": 9.02265114958842e-05,
      "loss": 0.046767657995224,
      "step": 34450
    },
    {
      "epoch": 9.781436275901221,
      "grad_norm": 3.5053915977478027,
      "learning_rate": 9.022367300596082e-05,
      "loss": 0.06353358030319214,
      "step": 34460
    },
    {
      "epoch": 9.78427476582458,
      "grad_norm": 5.582108020782471,
      "learning_rate": 9.022083451603748e-05,
      "loss": 0.04567739963531494,
      "step": 34470
    },
    {
      "epoch": 9.787113255747942,
      "grad_norm": 12.241418838500977,
      "learning_rate": 9.021799602611412e-05,
      "loss": 0.06244375705718994,
      "step": 34480
    },
    {
      "epoch": 9.789951745671303,
      "grad_norm": 9.93967056274414,
      "learning_rate": 9.021515753619075e-05,
      "loss": 0.08247674703598022,
      "step": 34490
    },
    {
      "epoch": 9.792790235594664,
      "grad_norm": 2.7331008911132812,
      "learning_rate": 9.021231904626739e-05,
      "loss": 0.055683857202529906,
      "step": 34500
    },
    {
      "epoch": 9.792790235594664,
      "eval_accuracy": 0.9530107458510841,
      "eval_loss": 0.1437760889530182,
      "eval_runtime": 31.1699,
      "eval_samples_per_second": 504.558,
      "eval_steps_per_second": 7.892,
      "step": 34500
    },
    {
      "epoch": 9.795628725518025,
      "grad_norm": 6.881160259246826,
      "learning_rate": 9.020948055634403e-05,
      "loss": 0.06169273853302002,
      "step": 34510
    },
    {
      "epoch": 9.798467215441384,
      "grad_norm": 11.642535209655762,
      "learning_rate": 9.020664206642067e-05,
      "loss": 0.04656406939029693,
      "step": 34520
    },
    {
      "epoch": 9.801305705364745,
      "grad_norm": 4.442540645599365,
      "learning_rate": 9.02038035764973e-05,
      "loss": 0.07033843994140625,
      "step": 34530
    },
    {
      "epoch": 9.804144195288107,
      "grad_norm": 13.053991317749023,
      "learning_rate": 9.020096508657395e-05,
      "loss": 0.08287585377693177,
      "step": 34540
    },
    {
      "epoch": 9.806982685211468,
      "grad_norm": 10.460387229919434,
      "learning_rate": 9.01981265966506e-05,
      "loss": 0.07373381853103637,
      "step": 34550
    },
    {
      "epoch": 9.809821175134829,
      "grad_norm": 2.3229660987854004,
      "learning_rate": 9.019528810672722e-05,
      "loss": 0.05174373984336853,
      "step": 34560
    },
    {
      "epoch": 9.81265966505819,
      "grad_norm": 4.232043266296387,
      "learning_rate": 9.019244961680386e-05,
      "loss": 0.03481191396713257,
      "step": 34570
    },
    {
      "epoch": 9.81549815498155,
      "grad_norm": 3.6204004287719727,
      "learning_rate": 9.01896111268805e-05,
      "loss": 0.043971121311187744,
      "step": 34580
    },
    {
      "epoch": 9.81833664490491,
      "grad_norm": 7.5256476402282715,
      "learning_rate": 9.018677263695713e-05,
      "loss": 0.05240855813026428,
      "step": 34590
    },
    {
      "epoch": 9.821175134828271,
      "grad_norm": 7.201149940490723,
      "learning_rate": 9.018393414703377e-05,
      "loss": 0.06972816586494446,
      "step": 34600
    },
    {
      "epoch": 9.824013624751633,
      "grad_norm": 9.160229682922363,
      "learning_rate": 9.018109565711043e-05,
      "loss": 0.08697982430458069,
      "step": 34610
    },
    {
      "epoch": 9.826852114674994,
      "grad_norm": 7.5745439529418945,
      "learning_rate": 9.017825716718706e-05,
      "loss": 0.06619174480438232,
      "step": 34620
    },
    {
      "epoch": 9.829690604598353,
      "grad_norm": 7.200782299041748,
      "learning_rate": 9.01754186772637e-05,
      "loss": 0.08137998580932618,
      "step": 34630
    },
    {
      "epoch": 9.832529094521714,
      "grad_norm": 7.1459174156188965,
      "learning_rate": 9.017258018734034e-05,
      "loss": 0.044330033659935,
      "step": 34640
    },
    {
      "epoch": 9.835367584445075,
      "grad_norm": 6.897623538970947,
      "learning_rate": 9.016974169741698e-05,
      "loss": 0.05968307256698609,
      "step": 34650
    },
    {
      "epoch": 9.838206074368436,
      "grad_norm": 8.819377899169922,
      "learning_rate": 9.016690320749361e-05,
      "loss": 0.060765206813812256,
      "step": 34660
    },
    {
      "epoch": 9.841044564291797,
      "grad_norm": 8.2949857711792,
      "learning_rate": 9.016406471757026e-05,
      "loss": 0.06788620352745056,
      "step": 34670
    },
    {
      "epoch": 9.843883054215157,
      "grad_norm": 16.896696090698242,
      "learning_rate": 9.01612262276469e-05,
      "loss": 0.06423289775848388,
      "step": 34680
    },
    {
      "epoch": 9.846721544138518,
      "grad_norm": 4.395873546600342,
      "learning_rate": 9.015838773772353e-05,
      "loss": 0.042915630340576175,
      "step": 34690
    },
    {
      "epoch": 9.849560034061879,
      "grad_norm": 6.8880743980407715,
      "learning_rate": 9.015554924780017e-05,
      "loss": 0.04221827983856201,
      "step": 34700
    },
    {
      "epoch": 9.85239852398524,
      "grad_norm": 8.773017883300781,
      "learning_rate": 9.015271075787682e-05,
      "loss": 0.042918425798416135,
      "step": 34710
    },
    {
      "epoch": 9.855237013908601,
      "grad_norm": 10.875904083251953,
      "learning_rate": 9.014987226795344e-05,
      "loss": 0.050799238681793216,
      "step": 34720
    },
    {
      "epoch": 9.858075503831962,
      "grad_norm": 4.369656085968018,
      "learning_rate": 9.014703377803009e-05,
      "loss": 0.05311837196350098,
      "step": 34730
    },
    {
      "epoch": 9.860913993755322,
      "grad_norm": 14.646661758422852,
      "learning_rate": 9.014419528810674e-05,
      "loss": 0.06273683905601501,
      "step": 34740
    },
    {
      "epoch": 9.863752483678683,
      "grad_norm": 22.416555404663086,
      "learning_rate": 9.014135679818337e-05,
      "loss": 0.055133724212646486,
      "step": 34750
    },
    {
      "epoch": 9.866590973602044,
      "grad_norm": 10.24942684173584,
      "learning_rate": 9.013851830826001e-05,
      "loss": 0.0659708857536316,
      "step": 34760
    },
    {
      "epoch": 9.869429463525405,
      "grad_norm": 7.025540351867676,
      "learning_rate": 9.013567981833665e-05,
      "loss": 0.07185924053192139,
      "step": 34770
    },
    {
      "epoch": 9.872267953448766,
      "grad_norm": 3.1399707794189453,
      "learning_rate": 9.013284132841329e-05,
      "loss": 0.04979588687419891,
      "step": 34780
    },
    {
      "epoch": 9.875106443372125,
      "grad_norm": 7.0206685066223145,
      "learning_rate": 9.013000283848992e-05,
      "loss": 0.06861185431480407,
      "step": 34790
    },
    {
      "epoch": 9.877944933295487,
      "grad_norm": 8.200695037841797,
      "learning_rate": 9.012716434856656e-05,
      "loss": 0.05065483450889587,
      "step": 34800
    },
    {
      "epoch": 9.880783423218848,
      "grad_norm": 2.2723278999328613,
      "learning_rate": 9.012432585864322e-05,
      "loss": 0.03970271944999695,
      "step": 34810
    },
    {
      "epoch": 9.883621913142209,
      "grad_norm": 8.67282772064209,
      "learning_rate": 9.012148736871984e-05,
      "loss": 0.04657282829284668,
      "step": 34820
    },
    {
      "epoch": 9.88646040306557,
      "grad_norm": 16.092105865478516,
      "learning_rate": 9.011864887879649e-05,
      "loss": 0.08139899969100953,
      "step": 34830
    },
    {
      "epoch": 9.88929889298893,
      "grad_norm": 4.557160377502441,
      "learning_rate": 9.011581038887313e-05,
      "loss": 0.06661387681961059,
      "step": 34840
    },
    {
      "epoch": 9.89213738291229,
      "grad_norm": 4.081960678100586,
      "learning_rate": 9.011297189894975e-05,
      "loss": 0.06317877769470215,
      "step": 34850
    },
    {
      "epoch": 9.894975872835651,
      "grad_norm": 5.8050642013549805,
      "learning_rate": 9.01101334090264e-05,
      "loss": 0.0409137099981308,
      "step": 34860
    },
    {
      "epoch": 9.897814362759012,
      "grad_norm": 11.1065092086792,
      "learning_rate": 9.010729491910305e-05,
      "loss": 0.0826742947101593,
      "step": 34870
    },
    {
      "epoch": 9.900652852682374,
      "grad_norm": 9.089666366577148,
      "learning_rate": 9.010445642917968e-05,
      "loss": 0.05835459232330322,
      "step": 34880
    },
    {
      "epoch": 9.903491342605733,
      "grad_norm": 14.420373916625977,
      "learning_rate": 9.010161793925632e-05,
      "loss": 0.061403590440750125,
      "step": 34890
    },
    {
      "epoch": 9.906329832529094,
      "grad_norm": 15.15878677368164,
      "learning_rate": 9.009877944933296e-05,
      "loss": 0.08295650482177734,
      "step": 34900
    },
    {
      "epoch": 9.909168322452455,
      "grad_norm": 2.6207170486450195,
      "learning_rate": 9.00959409594096e-05,
      "loss": 0.05734787583351135,
      "step": 34910
    },
    {
      "epoch": 9.912006812375816,
      "grad_norm": 18.246538162231445,
      "learning_rate": 9.009310246948623e-05,
      "loss": 0.08594343662261963,
      "step": 34920
    },
    {
      "epoch": 9.914845302299177,
      "grad_norm": 3.96736478805542,
      "learning_rate": 9.009026397956287e-05,
      "loss": 0.05625451803207397,
      "step": 34930
    },
    {
      "epoch": 9.917683792222537,
      "grad_norm": 7.367128849029541,
      "learning_rate": 9.008742548963953e-05,
      "loss": 0.046605801582336424,
      "step": 34940
    },
    {
      "epoch": 9.920522282145898,
      "grad_norm": 3.4135327339172363,
      "learning_rate": 9.008458699971616e-05,
      "loss": 0.05121656060218811,
      "step": 34950
    },
    {
      "epoch": 9.923360772069259,
      "grad_norm": 3.933467149734497,
      "learning_rate": 9.00817485097928e-05,
      "loss": 0.08033983707427979,
      "step": 34960
    },
    {
      "epoch": 9.92619926199262,
      "grad_norm": 5.143579959869385,
      "learning_rate": 9.007891001986944e-05,
      "loss": 0.03440910875797272,
      "step": 34970
    },
    {
      "epoch": 9.929037751915981,
      "grad_norm": 4.858419895172119,
      "learning_rate": 9.007607152994607e-05,
      "loss": 0.05421479344367981,
      "step": 34980
    },
    {
      "epoch": 9.931876241839342,
      "grad_norm": 8.035731315612793,
      "learning_rate": 9.007323304002271e-05,
      "loss": 0.05538780689239502,
      "step": 34990
    },
    {
      "epoch": 9.934714731762702,
      "grad_norm": 7.565187931060791,
      "learning_rate": 9.007039455009935e-05,
      "loss": 0.0592837929725647,
      "step": 35000
    },
    {
      "epoch": 9.934714731762702,
      "eval_accuracy": 0.9512303681566733,
      "eval_loss": 0.14887449145317078,
      "eval_runtime": 31.7605,
      "eval_samples_per_second": 495.176,
      "eval_steps_per_second": 7.745,
      "step": 35000
    },
    {
      "epoch": 9.937553221686063,
      "grad_norm": 18.523094177246094,
      "learning_rate": 9.006755606017599e-05,
      "loss": 0.06304702758789063,
      "step": 35010
    },
    {
      "epoch": 9.940391711609424,
      "grad_norm": 9.168621063232422,
      "learning_rate": 9.006471757025263e-05,
      "loss": 0.05922250747680664,
      "step": 35020
    },
    {
      "epoch": 9.943230201532785,
      "grad_norm": 14.038727760314941,
      "learning_rate": 9.006187908032927e-05,
      "loss": 0.08694047331809998,
      "step": 35030
    },
    {
      "epoch": 9.946068691456146,
      "grad_norm": 7.897758960723877,
      "learning_rate": 9.005904059040591e-05,
      "loss": 0.039450731873512265,
      "step": 35040
    },
    {
      "epoch": 9.948907181379505,
      "grad_norm": 8.153043746948242,
      "learning_rate": 9.005620210048254e-05,
      "loss": 0.0521111786365509,
      "step": 35050
    },
    {
      "epoch": 9.951745671302866,
      "grad_norm": 6.396288871765137,
      "learning_rate": 9.005336361055918e-05,
      "loss": 0.05624227523803711,
      "step": 35060
    },
    {
      "epoch": 9.954584161226228,
      "grad_norm": 19.761207580566406,
      "learning_rate": 9.005052512063584e-05,
      "loss": 0.057583165168762204,
      "step": 35070
    },
    {
      "epoch": 9.957422651149589,
      "grad_norm": 15.971322059631348,
      "learning_rate": 9.004768663071247e-05,
      "loss": 0.06497880220413207,
      "step": 35080
    },
    {
      "epoch": 9.96026114107295,
      "grad_norm": 10.363218307495117,
      "learning_rate": 9.004484814078911e-05,
      "loss": 0.05719610452651978,
      "step": 35090
    },
    {
      "epoch": 9.96309963099631,
      "grad_norm": 16.19901466369629,
      "learning_rate": 9.004200965086575e-05,
      "loss": 0.07287636399269104,
      "step": 35100
    },
    {
      "epoch": 9.96593812091967,
      "grad_norm": 3.8646068572998047,
      "learning_rate": 9.003917116094238e-05,
      "loss": 0.058908873796463014,
      "step": 35110
    },
    {
      "epoch": 9.968776610843031,
      "grad_norm": 20.06281852722168,
      "learning_rate": 9.003633267101902e-05,
      "loss": 0.05985050201416016,
      "step": 35120
    },
    {
      "epoch": 9.971615100766392,
      "grad_norm": 8.061190605163574,
      "learning_rate": 9.003349418109566e-05,
      "loss": 0.05187458992004394,
      "step": 35130
    },
    {
      "epoch": 9.974453590689754,
      "grad_norm": 16.898475646972656,
      "learning_rate": 9.00306556911723e-05,
      "loss": 0.07484169006347656,
      "step": 35140
    },
    {
      "epoch": 9.977292080613115,
      "grad_norm": 5.978675365447998,
      "learning_rate": 9.002781720124894e-05,
      "loss": 0.07348761558532715,
      "step": 35150
    },
    {
      "epoch": 9.980130570536474,
      "grad_norm": 12.11555290222168,
      "learning_rate": 9.002497871132558e-05,
      "loss": 0.07540331482887268,
      "step": 35160
    },
    {
      "epoch": 9.982969060459835,
      "grad_norm": 10.992325782775879,
      "learning_rate": 9.002214022140222e-05,
      "loss": 0.06647417545318604,
      "step": 35170
    },
    {
      "epoch": 9.985807550383196,
      "grad_norm": 10.29089069366455,
      "learning_rate": 9.001930173147885e-05,
      "loss": 0.06358823776245118,
      "step": 35180
    },
    {
      "epoch": 9.988646040306557,
      "grad_norm": 4.416968822479248,
      "learning_rate": 9.00164632415555e-05,
      "loss": 0.06774464845657349,
      "step": 35190
    },
    {
      "epoch": 9.991484530229918,
      "grad_norm": 5.965823650360107,
      "learning_rate": 9.001362475163214e-05,
      "loss": 0.05627605319023132,
      "step": 35200
    },
    {
      "epoch": 9.994323020153278,
      "grad_norm": 4.1874165534973145,
      "learning_rate": 9.001078626170878e-05,
      "loss": 0.051260560750961304,
      "step": 35210
    },
    {
      "epoch": 9.997161510076639,
      "grad_norm": 8.402924537658691,
      "learning_rate": 9.000794777178542e-05,
      "loss": 0.056137198209762575,
      "step": 35220
    },
    {
      "epoch": 10.0,
      "grad_norm": 7.2380051612854,
      "learning_rate": 9.000510928186206e-05,
      "loss": 0.06442593336105347,
      "step": 35230
    },
    {
      "epoch": 10.002838489923361,
      "grad_norm": 7.25870418548584,
      "learning_rate": 9.000227079193869e-05,
      "loss": 0.054992109537124634,
      "step": 35240
    },
    {
      "epoch": 10.005676979846722,
      "grad_norm": 6.037044048309326,
      "learning_rate": 8.999943230201533e-05,
      "loss": 0.04472480118274689,
      "step": 35250
    },
    {
      "epoch": 10.008515469770082,
      "grad_norm": 6.1710638999938965,
      "learning_rate": 8.999659381209197e-05,
      "loss": 0.05034240484237671,
      "step": 35260
    },
    {
      "epoch": 10.011353959693443,
      "grad_norm": 9.642723083496094,
      "learning_rate": 8.999375532216861e-05,
      "loss": 0.05321880578994751,
      "step": 35270
    },
    {
      "epoch": 10.014192449616804,
      "grad_norm": 10.24276065826416,
      "learning_rate": 8.999091683224525e-05,
      "loss": 0.04606838226318359,
      "step": 35280
    },
    {
      "epoch": 10.017030939540165,
      "grad_norm": 7.279782772064209,
      "learning_rate": 8.99880783423219e-05,
      "loss": 0.054281812906265256,
      "step": 35290
    },
    {
      "epoch": 10.019869429463526,
      "grad_norm": 6.006746768951416,
      "learning_rate": 8.998523985239852e-05,
      "loss": 0.03670320808887482,
      "step": 35300
    },
    {
      "epoch": 10.022707919386885,
      "grad_norm": 3.8634235858917236,
      "learning_rate": 8.998240136247516e-05,
      "loss": 0.05234031081199646,
      "step": 35310
    },
    {
      "epoch": 10.025546409310246,
      "grad_norm": 12.542618751525879,
      "learning_rate": 8.99795628725518e-05,
      "loss": 0.04267399311065674,
      "step": 35320
    },
    {
      "epoch": 10.028384899233608,
      "grad_norm": 12.190305709838867,
      "learning_rate": 8.997672438262845e-05,
      "loss": 0.05247676372528076,
      "step": 35330
    },
    {
      "epoch": 10.031223389156969,
      "grad_norm": 7.946407318115234,
      "learning_rate": 8.997388589270509e-05,
      "loss": 0.03265722393989563,
      "step": 35340
    },
    {
      "epoch": 10.03406187908033,
      "grad_norm": 11.76549243927002,
      "learning_rate": 8.997104740278173e-05,
      "loss": 0.04685068130493164,
      "step": 35350
    },
    {
      "epoch": 10.03690036900369,
      "grad_norm": 19.568191528320312,
      "learning_rate": 8.996820891285837e-05,
      "loss": 0.08826239109039306,
      "step": 35360
    },
    {
      "epoch": 10.03973885892705,
      "grad_norm": 3.0280404090881348,
      "learning_rate": 8.9965370422935e-05,
      "loss": 0.03310412466526032,
      "step": 35370
    },
    {
      "epoch": 10.042577348850411,
      "grad_norm": 8.630297660827637,
      "learning_rate": 8.996253193301164e-05,
      "loss": 0.05217688083648682,
      "step": 35380
    },
    {
      "epoch": 10.045415838773772,
      "grad_norm": 1.595406174659729,
      "learning_rate": 8.995969344308828e-05,
      "loss": 0.04221091270446777,
      "step": 35390
    },
    {
      "epoch": 10.048254328697134,
      "grad_norm": 6.737777233123779,
      "learning_rate": 8.995685495316491e-05,
      "loss": 0.05780994892120361,
      "step": 35400
    },
    {
      "epoch": 10.051092818620495,
      "grad_norm": 7.514589786529541,
      "learning_rate": 8.995401646324156e-05,
      "loss": 0.04562924802303314,
      "step": 35410
    },
    {
      "epoch": 10.053931308543854,
      "grad_norm": 4.597402095794678,
      "learning_rate": 8.99511779733182e-05,
      "loss": 0.0336977481842041,
      "step": 35420
    },
    {
      "epoch": 10.056769798467215,
      "grad_norm": 18.800081253051758,
      "learning_rate": 8.994833948339483e-05,
      "loss": 0.0389517605304718,
      "step": 35430
    },
    {
      "epoch": 10.059608288390576,
      "grad_norm": 1.9848864078521729,
      "learning_rate": 8.994550099347147e-05,
      "loss": 0.04523916840553284,
      "step": 35440
    },
    {
      "epoch": 10.062446778313937,
      "grad_norm": 7.036876201629639,
      "learning_rate": 8.994266250354812e-05,
      "loss": 0.0344100296497345,
      "step": 35450
    },
    {
      "epoch": 10.065285268237298,
      "grad_norm": 13.14783763885498,
      "learning_rate": 8.993982401362476e-05,
      "loss": 0.03282588124275208,
      "step": 35460
    },
    {
      "epoch": 10.068123758160658,
      "grad_norm": 10.074634552001953,
      "learning_rate": 8.99369855237014e-05,
      "loss": 0.05955502986907959,
      "step": 35470
    },
    {
      "epoch": 10.070962248084019,
      "grad_norm": 4.729062557220459,
      "learning_rate": 8.993414703377804e-05,
      "loss": 0.05340811610221863,
      "step": 35480
    },
    {
      "epoch": 10.07380073800738,
      "grad_norm": 6.622889518737793,
      "learning_rate": 8.993130854385468e-05,
      "loss": 0.03359088003635406,
      "step": 35490
    },
    {
      "epoch": 10.076639227930741,
      "grad_norm": 2.5799474716186523,
      "learning_rate": 8.992847005393131e-05,
      "loss": 0.04123450219631195,
      "step": 35500
    },
    {
      "epoch": 10.076639227930741,
      "eval_accuracy": 0.9471609334265912,
      "eval_loss": 0.16008706390857697,
      "eval_runtime": 31.296,
      "eval_samples_per_second": 502.525,
      "eval_steps_per_second": 7.86,
      "step": 35500
    },
    {
      "epoch": 10.079477717854102,
      "grad_norm": 8.8908109664917,
      "learning_rate": 8.992563156400795e-05,
      "loss": 0.05573850274085999,
      "step": 35510
    },
    {
      "epoch": 10.082316207777462,
      "grad_norm": 9.230525970458984,
      "learning_rate": 8.992279307408459e-05,
      "loss": 0.06340209245681763,
      "step": 35520
    },
    {
      "epoch": 10.085154697700823,
      "grad_norm": 5.9668498039245605,
      "learning_rate": 8.991995458416122e-05,
      "loss": 0.05195220708847046,
      "step": 35530
    },
    {
      "epoch": 10.087993187624184,
      "grad_norm": 7.181271553039551,
      "learning_rate": 8.991711609423787e-05,
      "loss": 0.0380649745464325,
      "step": 35540
    },
    {
      "epoch": 10.090831677547545,
      "grad_norm": 8.197202682495117,
      "learning_rate": 8.991427760431452e-05,
      "loss": 0.037612175941467284,
      "step": 35550
    },
    {
      "epoch": 10.093670167470906,
      "grad_norm": 2.697593927383423,
      "learning_rate": 8.991143911439114e-05,
      "loss": 0.022783589363098145,
      "step": 35560
    },
    {
      "epoch": 10.096508657394267,
      "grad_norm": 3.909217596054077,
      "learning_rate": 8.990860062446778e-05,
      "loss": 0.03313631415367126,
      "step": 35570
    },
    {
      "epoch": 10.099347147317626,
      "grad_norm": 6.560413360595703,
      "learning_rate": 8.990576213454443e-05,
      "loss": 0.03892926871776581,
      "step": 35580
    },
    {
      "epoch": 10.102185637240988,
      "grad_norm": 8.605551719665527,
      "learning_rate": 8.990292364462107e-05,
      "loss": 0.051356267929077146,
      "step": 35590
    },
    {
      "epoch": 10.105024127164349,
      "grad_norm": 8.99337100982666,
      "learning_rate": 8.99000851546977e-05,
      "loss": 0.03490999937057495,
      "step": 35600
    },
    {
      "epoch": 10.10786261708771,
      "grad_norm": 19.017131805419922,
      "learning_rate": 8.989724666477435e-05,
      "loss": 0.09872019290924072,
      "step": 35610
    },
    {
      "epoch": 10.11070110701107,
      "grad_norm": 11.235498428344727,
      "learning_rate": 8.989440817485099e-05,
      "loss": 0.05722121000289917,
      "step": 35620
    },
    {
      "epoch": 10.11353959693443,
      "grad_norm": 5.047369480133057,
      "learning_rate": 8.989156968492762e-05,
      "loss": 0.0776980996131897,
      "step": 35630
    },
    {
      "epoch": 10.116378086857791,
      "grad_norm": 6.9297895431518555,
      "learning_rate": 8.988873119500426e-05,
      "loss": 0.060690116882324216,
      "step": 35640
    },
    {
      "epoch": 10.119216576781152,
      "grad_norm": 17.668964385986328,
      "learning_rate": 8.98858927050809e-05,
      "loss": 0.06002585887908936,
      "step": 35650
    },
    {
      "epoch": 10.122055066704513,
      "grad_norm": 9.181960105895996,
      "learning_rate": 8.988305421515753e-05,
      "loss": 0.035371404886245725,
      "step": 35660
    },
    {
      "epoch": 10.124893556627875,
      "grad_norm": 7.714477062225342,
      "learning_rate": 8.988021572523418e-05,
      "loss": 0.024712955951690672,
      "step": 35670
    },
    {
      "epoch": 10.127732046551234,
      "grad_norm": 6.619862079620361,
      "learning_rate": 8.987737723531083e-05,
      "loss": 0.054236209392547606,
      "step": 35680
    },
    {
      "epoch": 10.130570536474595,
      "grad_norm": 11.290130615234375,
      "learning_rate": 8.987453874538745e-05,
      "loss": 0.06832056641578674,
      "step": 35690
    },
    {
      "epoch": 10.133409026397956,
      "grad_norm": 4.3995842933654785,
      "learning_rate": 8.98717002554641e-05,
      "loss": 0.04756172299385071,
      "step": 35700
    },
    {
      "epoch": 10.136247516321317,
      "grad_norm": 17.909027099609375,
      "learning_rate": 8.986886176554074e-05,
      "loss": 0.037564057111740115,
      "step": 35710
    },
    {
      "epoch": 10.139086006244678,
      "grad_norm": 6.395310401916504,
      "learning_rate": 8.986602327561738e-05,
      "loss": 0.0656251609325409,
      "step": 35720
    },
    {
      "epoch": 10.141924496168038,
      "grad_norm": 5.638888359069824,
      "learning_rate": 8.9863184785694e-05,
      "loss": 0.022350901365280153,
      "step": 35730
    },
    {
      "epoch": 10.144762986091399,
      "grad_norm": 15.563482284545898,
      "learning_rate": 8.986034629577066e-05,
      "loss": 0.07064988613128662,
      "step": 35740
    },
    {
      "epoch": 10.14760147601476,
      "grad_norm": 8.048022270202637,
      "learning_rate": 8.98575078058473e-05,
      "loss": 0.07858047485351563,
      "step": 35750
    },
    {
      "epoch": 10.150439965938121,
      "grad_norm": 4.078849792480469,
      "learning_rate": 8.985466931592393e-05,
      "loss": 0.0566254198551178,
      "step": 35760
    },
    {
      "epoch": 10.153278455861482,
      "grad_norm": 4.1883544921875,
      "learning_rate": 8.985183082600057e-05,
      "loss": 0.046141821146011355,
      "step": 35770
    },
    {
      "epoch": 10.156116945784843,
      "grad_norm": 7.745772361755371,
      "learning_rate": 8.984899233607721e-05,
      "loss": 0.035416656732559205,
      "step": 35780
    },
    {
      "epoch": 10.158955435708203,
      "grad_norm": 5.770241737365723,
      "learning_rate": 8.984615384615384e-05,
      "loss": 0.04654979407787323,
      "step": 35790
    },
    {
      "epoch": 10.161793925631564,
      "grad_norm": 3.8276517391204834,
      "learning_rate": 8.98433153562305e-05,
      "loss": 0.05311836004257202,
      "step": 35800
    },
    {
      "epoch": 10.164632415554925,
      "grad_norm": 15.73356819152832,
      "learning_rate": 8.984047686630714e-05,
      "loss": 0.02972656488418579,
      "step": 35810
    },
    {
      "epoch": 10.167470905478286,
      "grad_norm": 5.667628288269043,
      "learning_rate": 8.983763837638376e-05,
      "loss": 0.06465369462966919,
      "step": 35820
    },
    {
      "epoch": 10.170309395401647,
      "grad_norm": 6.022791862487793,
      "learning_rate": 8.98347998864604e-05,
      "loss": 0.048763447999954225,
      "step": 35830
    },
    {
      "epoch": 10.173147885325006,
      "grad_norm": 2.321946144104004,
      "learning_rate": 8.983196139653705e-05,
      "loss": 0.04847998023033142,
      "step": 35840
    },
    {
      "epoch": 10.175986375248367,
      "grad_norm": 5.7691192626953125,
      "learning_rate": 8.982912290661369e-05,
      "loss": 0.04806318581104278,
      "step": 35850
    },
    {
      "epoch": 10.178824865171729,
      "grad_norm": 13.540083885192871,
      "learning_rate": 8.982628441669032e-05,
      "loss": 0.05916212201118469,
      "step": 35860
    },
    {
      "epoch": 10.18166335509509,
      "grad_norm": 4.886026859283447,
      "learning_rate": 8.982344592676697e-05,
      "loss": 0.03068886399269104,
      "step": 35870
    },
    {
      "epoch": 10.18450184501845,
      "grad_norm": 14.19399356842041,
      "learning_rate": 8.982060743684361e-05,
      "loss": 0.06252155303955079,
      "step": 35880
    },
    {
      "epoch": 10.18734033494181,
      "grad_norm": 7.2224225997924805,
      "learning_rate": 8.981776894692024e-05,
      "loss": 0.056982392072677614,
      "step": 35890
    },
    {
      "epoch": 10.190178824865171,
      "grad_norm": 13.160146713256836,
      "learning_rate": 8.981493045699688e-05,
      "loss": 0.07169893980026246,
      "step": 35900
    },
    {
      "epoch": 10.193017314788532,
      "grad_norm": 3.4034552574157715,
      "learning_rate": 8.981209196707352e-05,
      "loss": 0.04664395749568939,
      "step": 35910
    },
    {
      "epoch": 10.195855804711893,
      "grad_norm": 7.455689430236816,
      "learning_rate": 8.980925347715015e-05,
      "loss": 0.05753688216209411,
      "step": 35920
    },
    {
      "epoch": 10.198694294635255,
      "grad_norm": 6.141243934631348,
      "learning_rate": 8.980641498722679e-05,
      "loss": 0.052104413509368896,
      "step": 35930
    },
    {
      "epoch": 10.201532784558616,
      "grad_norm": 3.608466863632202,
      "learning_rate": 8.980357649730345e-05,
      "loss": 0.028675973415374756,
      "step": 35940
    },
    {
      "epoch": 10.204371274481975,
      "grad_norm": 11.292566299438477,
      "learning_rate": 8.980073800738008e-05,
      "loss": 0.034777671098709106,
      "step": 35950
    },
    {
      "epoch": 10.207209764405336,
      "grad_norm": 6.170752048492432,
      "learning_rate": 8.979789951745672e-05,
      "loss": 0.04363607168197632,
      "step": 35960
    },
    {
      "epoch": 10.210048254328697,
      "grad_norm": 0.47235235571861267,
      "learning_rate": 8.979506102753336e-05,
      "loss": 0.01717611700296402,
      "step": 35970
    },
    {
      "epoch": 10.212886744252058,
      "grad_norm": 8.509514808654785,
      "learning_rate": 8.979222253761e-05,
      "loss": 0.03820683658123016,
      "step": 35980
    },
    {
      "epoch": 10.21572523417542,
      "grad_norm": 10.440407752990723,
      "learning_rate": 8.978938404768663e-05,
      "loss": 0.058907926082611084,
      "step": 35990
    },
    {
      "epoch": 10.218563724098779,
      "grad_norm": 8.241033554077148,
      "learning_rate": 8.978654555776328e-05,
      "loss": 0.03713009059429169,
      "step": 36000
    },
    {
      "epoch": 10.218563724098779,
      "eval_accuracy": 0.9473516881795638,
      "eval_loss": 0.15866875648498535,
      "eval_runtime": 31.5073,
      "eval_samples_per_second": 499.154,
      "eval_steps_per_second": 7.808,
      "step": 36000
    },
    {
      "epoch": 10.22140221402214,
      "grad_norm": 5.251601696014404,
      "learning_rate": 8.978370706783992e-05,
      "loss": 0.04426538050174713,
      "step": 36010
    },
    {
      "epoch": 10.224240703945501,
      "grad_norm": 1.979737401008606,
      "learning_rate": 8.978086857791655e-05,
      "loss": 0.055144065618515016,
      "step": 36020
    },
    {
      "epoch": 10.227079193868862,
      "grad_norm": 7.218305587768555,
      "learning_rate": 8.977803008799319e-05,
      "loss": 0.046082299947738645,
      "step": 36030
    },
    {
      "epoch": 10.229917683792223,
      "grad_norm": 9.494359970092773,
      "learning_rate": 8.977519159806983e-05,
      "loss": 0.0413273423910141,
      "step": 36040
    },
    {
      "epoch": 10.232756173715583,
      "grad_norm": 4.212758541107178,
      "learning_rate": 8.977263695713881e-05,
      "loss": 0.03304300308227539,
      "step": 36050
    },
    {
      "epoch": 10.235594663638944,
      "grad_norm": 13.917865753173828,
      "learning_rate": 8.976979846721546e-05,
      "loss": 0.03857312798500061,
      "step": 36060
    },
    {
      "epoch": 10.238433153562305,
      "grad_norm": 11.222973823547363,
      "learning_rate": 8.976695997729208e-05,
      "loss": 0.05476447343826294,
      "step": 36070
    },
    {
      "epoch": 10.241271643485666,
      "grad_norm": 2.1426548957824707,
      "learning_rate": 8.976412148736873e-05,
      "loss": 0.035970038175582884,
      "step": 36080
    },
    {
      "epoch": 10.244110133409027,
      "grad_norm": 5.713494777679443,
      "learning_rate": 8.976128299744537e-05,
      "loss": 0.04338274598121643,
      "step": 36090
    },
    {
      "epoch": 10.246948623332386,
      "grad_norm": 7.571118354797363,
      "learning_rate": 8.9758444507522e-05,
      "loss": 0.054502475261688235,
      "step": 36100
    },
    {
      "epoch": 10.249787113255747,
      "grad_norm": 9.001779556274414,
      "learning_rate": 8.975560601759864e-05,
      "loss": 0.04046744704246521,
      "step": 36110
    },
    {
      "epoch": 10.252625603179109,
      "grad_norm": 6.142360687255859,
      "learning_rate": 8.975276752767529e-05,
      "loss": 0.04563030898571015,
      "step": 36120
    },
    {
      "epoch": 10.25546409310247,
      "grad_norm": 8.738130569458008,
      "learning_rate": 8.974992903775192e-05,
      "loss": 0.05603446960449219,
      "step": 36130
    },
    {
      "epoch": 10.25830258302583,
      "grad_norm": 18.529918670654297,
      "learning_rate": 8.974709054782856e-05,
      "loss": 0.0697174310684204,
      "step": 36140
    },
    {
      "epoch": 10.261141072949192,
      "grad_norm": 14.157784461975098,
      "learning_rate": 8.97442520579052e-05,
      "loss": 0.05651613473892212,
      "step": 36150
    },
    {
      "epoch": 10.263979562872551,
      "grad_norm": 9.717578887939453,
      "learning_rate": 8.974141356798184e-05,
      "loss": 0.0654645323753357,
      "step": 36160
    },
    {
      "epoch": 10.266818052795912,
      "grad_norm": 9.141207695007324,
      "learning_rate": 8.973857507805847e-05,
      "loss": 0.03184970915317535,
      "step": 36170
    },
    {
      "epoch": 10.269656542719273,
      "grad_norm": 5.090808391571045,
      "learning_rate": 8.973573658813513e-05,
      "loss": 0.06937923431396484,
      "step": 36180
    },
    {
      "epoch": 10.272495032642635,
      "grad_norm": 4.167090892791748,
      "learning_rate": 8.973289809821175e-05,
      "loss": 0.08139961957931519,
      "step": 36190
    },
    {
      "epoch": 10.275333522565996,
      "grad_norm": 4.789121627807617,
      "learning_rate": 8.97300596082884e-05,
      "loss": 0.03616894781589508,
      "step": 36200
    },
    {
      "epoch": 10.278172012489355,
      "grad_norm": 5.434744358062744,
      "learning_rate": 8.972722111836504e-05,
      "loss": 0.03950951099395752,
      "step": 36210
    },
    {
      "epoch": 10.281010502412716,
      "grad_norm": 14.839851379394531,
      "learning_rate": 8.972438262844168e-05,
      "loss": 0.05882408618927002,
      "step": 36220
    },
    {
      "epoch": 10.283848992336077,
      "grad_norm": 5.861428737640381,
      "learning_rate": 8.97215441385183e-05,
      "loss": 0.08029335141181945,
      "step": 36230
    },
    {
      "epoch": 10.286687482259438,
      "grad_norm": 1.1029629707336426,
      "learning_rate": 8.971870564859495e-05,
      "loss": 0.058646482229232785,
      "step": 36240
    },
    {
      "epoch": 10.2895259721828,
      "grad_norm": 9.69312858581543,
      "learning_rate": 8.97158671586716e-05,
      "loss": 0.05503484606742859,
      "step": 36250
    },
    {
      "epoch": 10.292364462106159,
      "grad_norm": 9.076212882995605,
      "learning_rate": 8.971302866874823e-05,
      "loss": 0.045270389318466185,
      "step": 36260
    },
    {
      "epoch": 10.29520295202952,
      "grad_norm": 2.113004446029663,
      "learning_rate": 8.971019017882487e-05,
      "loss": 0.045760995149612425,
      "step": 36270
    },
    {
      "epoch": 10.298041441952881,
      "grad_norm": 6.438479423522949,
      "learning_rate": 8.970735168890151e-05,
      "loss": 0.053906583786010744,
      "step": 36280
    },
    {
      "epoch": 10.300879931876242,
      "grad_norm": 4.5841145515441895,
      "learning_rate": 8.970451319897814e-05,
      "loss": 0.043227142095565795,
      "step": 36290
    },
    {
      "epoch": 10.303718421799603,
      "grad_norm": 4.7565999031066895,
      "learning_rate": 8.970167470905478e-05,
      "loss": 0.032406175136566163,
      "step": 36300
    },
    {
      "epoch": 10.306556911722964,
      "grad_norm": 5.3222174644470215,
      "learning_rate": 8.969883621913142e-05,
      "loss": 0.058787685632705686,
      "step": 36310
    },
    {
      "epoch": 10.309395401646324,
      "grad_norm": 14.744622230529785,
      "learning_rate": 8.969599772920806e-05,
      "loss": 0.04638729691505432,
      "step": 36320
    },
    {
      "epoch": 10.312233891569685,
      "grad_norm": 4.511398792266846,
      "learning_rate": 8.96931592392847e-05,
      "loss": 0.056706970930099486,
      "step": 36330
    },
    {
      "epoch": 10.315072381493046,
      "grad_norm": 13.197816848754883,
      "learning_rate": 8.969032074936135e-05,
      "loss": 0.041135966777801514,
      "step": 36340
    },
    {
      "epoch": 10.317910871416407,
      "grad_norm": 6.950282096862793,
      "learning_rate": 8.968748225943799e-05,
      "loss": 0.04915453195571899,
      "step": 36350
    },
    {
      "epoch": 10.320749361339768,
      "grad_norm": 6.648165702819824,
      "learning_rate": 8.968464376951462e-05,
      "loss": 0.058501708507537845,
      "step": 36360
    },
    {
      "epoch": 10.323587851263127,
      "grad_norm": 9.410599708557129,
      "learning_rate": 8.968180527959126e-05,
      "loss": 0.03711932003498077,
      "step": 36370
    },
    {
      "epoch": 10.326426341186489,
      "grad_norm": 3.2461369037628174,
      "learning_rate": 8.967896678966791e-05,
      "loss": 0.029022285342216493,
      "step": 36380
    },
    {
      "epoch": 10.32926483110985,
      "grad_norm": 4.180446624755859,
      "learning_rate": 8.967612829974454e-05,
      "loss": 0.05456423759460449,
      "step": 36390
    },
    {
      "epoch": 10.33210332103321,
      "grad_norm": 8.58168888092041,
      "learning_rate": 8.967328980982118e-05,
      "loss": 0.06622806787490845,
      "step": 36400
    },
    {
      "epoch": 10.334941810956572,
      "grad_norm": 6.193541526794434,
      "learning_rate": 8.967045131989782e-05,
      "loss": 0.04797664284706116,
      "step": 36410
    },
    {
      "epoch": 10.337780300879931,
      "grad_norm": 2.4929521083831787,
      "learning_rate": 8.966761282997445e-05,
      "loss": 0.03953470289707184,
      "step": 36420
    },
    {
      "epoch": 10.340618790803292,
      "grad_norm": 12.940290451049805,
      "learning_rate": 8.966477434005109e-05,
      "loss": 0.05653942227363586,
      "step": 36430
    },
    {
      "epoch": 10.343457280726653,
      "grad_norm": 13.67895221710205,
      "learning_rate": 8.966193585012773e-05,
      "loss": 0.05468284487724304,
      "step": 36440
    },
    {
      "epoch": 10.346295770650014,
      "grad_norm": 8.446161270141602,
      "learning_rate": 8.965909736020437e-05,
      "loss": 0.05774714946746826,
      "step": 36450
    },
    {
      "epoch": 10.349134260573376,
      "grad_norm": 9.774473190307617,
      "learning_rate": 8.965625887028102e-05,
      "loss": 0.058855897188186644,
      "step": 36460
    },
    {
      "epoch": 10.351972750496735,
      "grad_norm": 13.697847366333008,
      "learning_rate": 8.965342038035766e-05,
      "loss": 0.06780917644500732,
      "step": 36470
    },
    {
      "epoch": 10.354811240420096,
      "grad_norm": 12.0314359664917,
      "learning_rate": 8.96505818904343e-05,
      "loss": 0.0725091814994812,
      "step": 36480
    },
    {
      "epoch": 10.357649730343457,
      "grad_norm": 4.78488302230835,
      "learning_rate": 8.964774340051093e-05,
      "loss": 0.04848609566688537,
      "step": 36490
    },
    {
      "epoch": 10.360488220266818,
      "grad_norm": 1.0983777046203613,
      "learning_rate": 8.964490491058757e-05,
      "loss": 0.04272754490375519,
      "step": 36500
    },
    {
      "epoch": 10.360488220266818,
      "eval_accuracy": 0.9542824442042348,
      "eval_loss": 0.1382163166999817,
      "eval_runtime": 31.338,
      "eval_samples_per_second": 501.85,
      "eval_steps_per_second": 7.85,
      "step": 36500
    },
    {
      "epoch": 10.36332671019018,
      "grad_norm": 16.599246978759766,
      "learning_rate": 8.964206642066421e-05,
      "loss": 0.055419743061065674,
      "step": 36510
    },
    {
      "epoch": 10.366165200113539,
      "grad_norm": 8.60208511352539,
      "learning_rate": 8.963922793074085e-05,
      "loss": 0.04853387475013733,
      "step": 36520
    },
    {
      "epoch": 10.3690036900369,
      "grad_norm": 6.653528213500977,
      "learning_rate": 8.963638944081749e-05,
      "loss": 0.0412818044424057,
      "step": 36530
    },
    {
      "epoch": 10.371842179960261,
      "grad_norm": 8.607937812805176,
      "learning_rate": 8.963355095089413e-05,
      "loss": 0.03893500566482544,
      "step": 36540
    },
    {
      "epoch": 10.374680669883622,
      "grad_norm": 15.344191551208496,
      "learning_rate": 8.963071246097076e-05,
      "loss": 0.04885843992233276,
      "step": 36550
    },
    {
      "epoch": 10.377519159806983,
      "grad_norm": 9.874051094055176,
      "learning_rate": 8.96278739710474e-05,
      "loss": 0.037526220083236694,
      "step": 36560
    },
    {
      "epoch": 10.380357649730344,
      "grad_norm": 4.276268005371094,
      "learning_rate": 8.962503548112404e-05,
      "loss": 0.04496805965900421,
      "step": 36570
    },
    {
      "epoch": 10.383196139653704,
      "grad_norm": 11.307729721069336,
      "learning_rate": 8.962219699120069e-05,
      "loss": 0.055699515342712405,
      "step": 36580
    },
    {
      "epoch": 10.386034629577065,
      "grad_norm": 14.329275131225586,
      "learning_rate": 8.961935850127733e-05,
      "loss": 0.06453256011009216,
      "step": 36590
    },
    {
      "epoch": 10.388873119500426,
      "grad_norm": 5.845245838165283,
      "learning_rate": 8.961652001135397e-05,
      "loss": 0.04669055938720703,
      "step": 36600
    },
    {
      "epoch": 10.391711609423787,
      "grad_norm": 6.348171234130859,
      "learning_rate": 8.961368152143061e-05,
      "loss": 0.0633388638496399,
      "step": 36610
    },
    {
      "epoch": 10.394550099347148,
      "grad_norm": 3.777928113937378,
      "learning_rate": 8.961084303150724e-05,
      "loss": 0.040178346633911136,
      "step": 36620
    },
    {
      "epoch": 10.397388589270507,
      "grad_norm": 7.331879138946533,
      "learning_rate": 8.960800454158388e-05,
      "loss": 0.05863572359085083,
      "step": 36630
    },
    {
      "epoch": 10.400227079193868,
      "grad_norm": 1.6967625617980957,
      "learning_rate": 8.960516605166052e-05,
      "loss": 0.07237986922264099,
      "step": 36640
    },
    {
      "epoch": 10.40306556911723,
      "grad_norm": 0.5049591064453125,
      "learning_rate": 8.960232756173716e-05,
      "loss": 0.037475234270095824,
      "step": 36650
    },
    {
      "epoch": 10.40590405904059,
      "grad_norm": 6.233887672424316,
      "learning_rate": 8.95994890718138e-05,
      "loss": 0.04201467633247376,
      "step": 36660
    },
    {
      "epoch": 10.408742548963952,
      "grad_norm": 9.84954833984375,
      "learning_rate": 8.959665058189044e-05,
      "loss": 0.035579648613929746,
      "step": 36670
    },
    {
      "epoch": 10.411581038887311,
      "grad_norm": 3.7128124237060547,
      "learning_rate": 8.959381209196707e-05,
      "loss": 0.05249812006950379,
      "step": 36680
    },
    {
      "epoch": 10.414419528810672,
      "grad_norm": 3.413320541381836,
      "learning_rate": 8.959097360204371e-05,
      "loss": 0.02419271171092987,
      "step": 36690
    },
    {
      "epoch": 10.417258018734033,
      "grad_norm": 12.171339988708496,
      "learning_rate": 8.958813511212035e-05,
      "loss": 0.04807780981063843,
      "step": 36700
    },
    {
      "epoch": 10.420096508657394,
      "grad_norm": 5.273982048034668,
      "learning_rate": 8.9585296622197e-05,
      "loss": 0.03695886433124542,
      "step": 36710
    },
    {
      "epoch": 10.422934998580756,
      "grad_norm": 6.233440399169922,
      "learning_rate": 8.958245813227364e-05,
      "loss": 0.06202154755592346,
      "step": 36720
    },
    {
      "epoch": 10.425773488504117,
      "grad_norm": 8.094334602355957,
      "learning_rate": 8.957961964235028e-05,
      "loss": 0.05624317526817322,
      "step": 36730
    },
    {
      "epoch": 10.428611978427476,
      "grad_norm": 3.4567596912384033,
      "learning_rate": 8.957678115242692e-05,
      "loss": 0.03419476449489593,
      "step": 36740
    },
    {
      "epoch": 10.431450468350837,
      "grad_norm": 10.846534729003906,
      "learning_rate": 8.957394266250355e-05,
      "loss": 0.05165817141532898,
      "step": 36750
    },
    {
      "epoch": 10.434288958274198,
      "grad_norm": 8.145037651062012,
      "learning_rate": 8.957110417258019e-05,
      "loss": 0.04305214285850525,
      "step": 36760
    },
    {
      "epoch": 10.43712744819756,
      "grad_norm": 6.392567157745361,
      "learning_rate": 8.956826568265683e-05,
      "loss": 0.029481494426727296,
      "step": 36770
    },
    {
      "epoch": 10.43996593812092,
      "grad_norm": 9.604814529418945,
      "learning_rate": 8.956542719273347e-05,
      "loss": 0.04259648323059082,
      "step": 36780
    },
    {
      "epoch": 10.44280442804428,
      "grad_norm": 6.068155288696289,
      "learning_rate": 8.956258870281011e-05,
      "loss": 0.031088033318519594,
      "step": 36790
    },
    {
      "epoch": 10.44564291796764,
      "grad_norm": 7.184687614440918,
      "learning_rate": 8.955975021288675e-05,
      "loss": 0.03532496094703674,
      "step": 36800
    },
    {
      "epoch": 10.448481407891002,
      "grad_norm": 7.030819892883301,
      "learning_rate": 8.955691172296338e-05,
      "loss": 0.06580129861831666,
      "step": 36810
    },
    {
      "epoch": 10.451319897814363,
      "grad_norm": 11.161382675170898,
      "learning_rate": 8.955407323304002e-05,
      "loss": 0.05144480466842651,
      "step": 36820
    },
    {
      "epoch": 10.454158387737724,
      "grad_norm": 13.454534530639648,
      "learning_rate": 8.955123474311667e-05,
      "loss": 0.04106142222881317,
      "step": 36830
    },
    {
      "epoch": 10.456996877661084,
      "grad_norm": 9.999848365783691,
      "learning_rate": 8.954839625319331e-05,
      "loss": 0.06666446924209594,
      "step": 36840
    },
    {
      "epoch": 10.459835367584445,
      "grad_norm": 13.241418838500977,
      "learning_rate": 8.954555776326995e-05,
      "loss": 0.04435061812400818,
      "step": 36850
    },
    {
      "epoch": 10.462673857507806,
      "grad_norm": 1.3495501279830933,
      "learning_rate": 8.954271927334659e-05,
      "loss": 0.043979603052139285,
      "step": 36860
    },
    {
      "epoch": 10.465512347431167,
      "grad_norm": 9.35275650024414,
      "learning_rate": 8.953988078342323e-05,
      "loss": 0.03838358819484711,
      "step": 36870
    },
    {
      "epoch": 10.468350837354528,
      "grad_norm": 5.111002445220947,
      "learning_rate": 8.953704229349986e-05,
      "loss": 0.04560145139694214,
      "step": 36880
    },
    {
      "epoch": 10.471189327277887,
      "grad_norm": 12.216764450073242,
      "learning_rate": 8.95342038035765e-05,
      "loss": 0.07290319204330445,
      "step": 36890
    },
    {
      "epoch": 10.474027817201248,
      "grad_norm": 11.165630340576172,
      "learning_rate": 8.953136531365314e-05,
      "loss": 0.05220510363578797,
      "step": 36900
    },
    {
      "epoch": 10.47686630712461,
      "grad_norm": 8.722389221191406,
      "learning_rate": 8.952852682372977e-05,
      "loss": 0.07873188853263854,
      "step": 36910
    },
    {
      "epoch": 10.47970479704797,
      "grad_norm": 11.220128059387207,
      "learning_rate": 8.952568833380642e-05,
      "loss": 0.03465230166912079,
      "step": 36920
    },
    {
      "epoch": 10.482543286971332,
      "grad_norm": 3.550487756729126,
      "learning_rate": 8.952284984388307e-05,
      "loss": 0.07557478547096252,
      "step": 36930
    },
    {
      "epoch": 10.485381776894693,
      "grad_norm": 3.807497262954712,
      "learning_rate": 8.95200113539597e-05,
      "loss": 0.05049041509628296,
      "step": 36940
    },
    {
      "epoch": 10.488220266818052,
      "grad_norm": 6.9589738845825195,
      "learning_rate": 8.951717286403633e-05,
      "loss": 0.05697675347328186,
      "step": 36950
    },
    {
      "epoch": 10.491058756741413,
      "grad_norm": 4.297712326049805,
      "learning_rate": 8.951433437411298e-05,
      "loss": 0.03934766054153442,
      "step": 36960
    },
    {
      "epoch": 10.493897246664774,
      "grad_norm": 1.7308541536331177,
      "learning_rate": 8.951149588418962e-05,
      "loss": 0.03359519839286804,
      "step": 36970
    },
    {
      "epoch": 10.496735736588136,
      "grad_norm": 2.608450412750244,
      "learning_rate": 8.950865739426626e-05,
      "loss": 0.04380088448524475,
      "step": 36980
    },
    {
      "epoch": 10.499574226511497,
      "grad_norm": 8.706777572631836,
      "learning_rate": 8.95058189043429e-05,
      "loss": 0.044136333465576175,
      "step": 36990
    },
    {
      "epoch": 10.502412716434856,
      "grad_norm": 3.4056553840637207,
      "learning_rate": 8.950298041441953e-05,
      "loss": 0.030855458974838258,
      "step": 37000
    },
    {
      "epoch": 10.502412716434856,
      "eval_accuracy": 0.9511667832390157,
      "eval_loss": 0.1493266224861145,
      "eval_runtime": 31.6631,
      "eval_samples_per_second": 496.698,
      "eval_steps_per_second": 7.769,
      "step": 37000
    },
    {
      "epoch": 10.505251206358217,
      "grad_norm": 11.527070045471191,
      "learning_rate": 8.950014192449617e-05,
      "loss": 0.04167925119400025,
      "step": 37010
    },
    {
      "epoch": 10.508089696281578,
      "grad_norm": 2.684386730194092,
      "learning_rate": 8.949730343457281e-05,
      "loss": 0.0481109082698822,
      "step": 37020
    },
    {
      "epoch": 10.51092818620494,
      "grad_norm": 9.902045249938965,
      "learning_rate": 8.949446494464945e-05,
      "loss": 0.05797825455665588,
      "step": 37030
    },
    {
      "epoch": 10.5137666761283,
      "grad_norm": 7.688483715057373,
      "learning_rate": 8.949162645472608e-05,
      "loss": 0.05305147171020508,
      "step": 37040
    },
    {
      "epoch": 10.51660516605166,
      "grad_norm": 5.112579345703125,
      "learning_rate": 8.948878796480274e-05,
      "loss": 0.05579475164413452,
      "step": 37050
    },
    {
      "epoch": 10.51944365597502,
      "grad_norm": 3.4778544902801514,
      "learning_rate": 8.948594947487938e-05,
      "loss": 0.052108395099639895,
      "step": 37060
    },
    {
      "epoch": 10.522282145898382,
      "grad_norm": 7.331381320953369,
      "learning_rate": 8.9483110984956e-05,
      "loss": 0.04809304475784302,
      "step": 37070
    },
    {
      "epoch": 10.525120635821743,
      "grad_norm": 7.230174541473389,
      "learning_rate": 8.948027249503265e-05,
      "loss": 0.04410632848739624,
      "step": 37080
    },
    {
      "epoch": 10.527959125745104,
      "grad_norm": 11.53337287902832,
      "learning_rate": 8.947743400510929e-05,
      "loss": 0.060776710510253906,
      "step": 37090
    },
    {
      "epoch": 10.530797615668465,
      "grad_norm": 12.495294570922852,
      "learning_rate": 8.947459551518593e-05,
      "loss": 0.0499268114566803,
      "step": 37100
    },
    {
      "epoch": 10.533636105591825,
      "grad_norm": 3.009577751159668,
      "learning_rate": 8.947175702526256e-05,
      "loss": 0.04827150106430054,
      "step": 37110
    },
    {
      "epoch": 10.536474595515186,
      "grad_norm": 10.95189094543457,
      "learning_rate": 8.946891853533921e-05,
      "loss": 0.05228626728057861,
      "step": 37120
    },
    {
      "epoch": 10.539313085438547,
      "grad_norm": 11.486112594604492,
      "learning_rate": 8.946608004541584e-05,
      "loss": 0.05806214213371277,
      "step": 37130
    },
    {
      "epoch": 10.542151575361908,
      "grad_norm": 2.045602321624756,
      "learning_rate": 8.946324155549248e-05,
      "loss": 0.04174984395503998,
      "step": 37140
    },
    {
      "epoch": 10.544990065285269,
      "grad_norm": 5.676380634307861,
      "learning_rate": 8.946040306556912e-05,
      "loss": 0.040227028727531436,
      "step": 37150
    },
    {
      "epoch": 10.547828555208628,
      "grad_norm": 5.284129619598389,
      "learning_rate": 8.945756457564576e-05,
      "loss": 0.038356566429138185,
      "step": 37160
    },
    {
      "epoch": 10.55066704513199,
      "grad_norm": 9.811860084533691,
      "learning_rate": 8.945472608572239e-05,
      "loss": 0.054844117164611815,
      "step": 37170
    },
    {
      "epoch": 10.55350553505535,
      "grad_norm": 7.599334716796875,
      "learning_rate": 8.945188759579905e-05,
      "loss": 0.05273175835609436,
      "step": 37180
    },
    {
      "epoch": 10.556344024978712,
      "grad_norm": 2.7173471450805664,
      "learning_rate": 8.944904910587569e-05,
      "loss": 0.03449954092502594,
      "step": 37190
    },
    {
      "epoch": 10.559182514902073,
      "grad_norm": 12.445903778076172,
      "learning_rate": 8.944621061595232e-05,
      "loss": 0.05224841833114624,
      "step": 37200
    },
    {
      "epoch": 10.562021004825432,
      "grad_norm": 5.940929412841797,
      "learning_rate": 8.944337212602896e-05,
      "loss": 0.04023993611335754,
      "step": 37210
    },
    {
      "epoch": 10.564859494748793,
      "grad_norm": 5.430938720703125,
      "learning_rate": 8.94405336361056e-05,
      "loss": 0.045115295052528384,
      "step": 37220
    },
    {
      "epoch": 10.567697984672154,
      "grad_norm": 14.923076629638672,
      "learning_rate": 8.943769514618223e-05,
      "loss": 0.04161341786384583,
      "step": 37230
    },
    {
      "epoch": 10.570536474595515,
      "grad_norm": 5.093540191650391,
      "learning_rate": 8.943485665625887e-05,
      "loss": 0.05185024738311768,
      "step": 37240
    },
    {
      "epoch": 10.573374964518877,
      "grad_norm": 9.676848411560059,
      "learning_rate": 8.943201816633552e-05,
      "loss": 0.07991297841072083,
      "step": 37250
    },
    {
      "epoch": 10.576213454442236,
      "grad_norm": 21.847732543945312,
      "learning_rate": 8.942917967641215e-05,
      "loss": 0.05551596283912659,
      "step": 37260
    },
    {
      "epoch": 10.579051944365597,
      "grad_norm": 10.302873611450195,
      "learning_rate": 8.942634118648879e-05,
      "loss": 0.0576478362083435,
      "step": 37270
    },
    {
      "epoch": 10.581890434288958,
      "grad_norm": 8.793778419494629,
      "learning_rate": 8.942350269656543e-05,
      "loss": 0.05904110670089722,
      "step": 37280
    },
    {
      "epoch": 10.58472892421232,
      "grad_norm": 3.1477136611938477,
      "learning_rate": 8.942066420664207e-05,
      "loss": 0.055383825302124025,
      "step": 37290
    },
    {
      "epoch": 10.58756741413568,
      "grad_norm": 17.405223846435547,
      "learning_rate": 8.94178257167187e-05,
      "loss": 0.05748081803321838,
      "step": 37300
    },
    {
      "epoch": 10.59040590405904,
      "grad_norm": 1.784849762916565,
      "learning_rate": 8.941498722679534e-05,
      "loss": 0.03512507379055023,
      "step": 37310
    },
    {
      "epoch": 10.5932443939824,
      "grad_norm": 1.0856764316558838,
      "learning_rate": 8.9412148736872e-05,
      "loss": 0.05085453987121582,
      "step": 37320
    },
    {
      "epoch": 10.596082883905762,
      "grad_norm": 7.579006195068359,
      "learning_rate": 8.940931024694863e-05,
      "loss": 0.058287763595581056,
      "step": 37330
    },
    {
      "epoch": 10.598921373829123,
      "grad_norm": 10.341796875,
      "learning_rate": 8.940647175702527e-05,
      "loss": 0.033180254697799685,
      "step": 37340
    },
    {
      "epoch": 10.601759863752484,
      "grad_norm": 13.001203536987305,
      "learning_rate": 8.940363326710191e-05,
      "loss": 0.03567807376384735,
      "step": 37350
    },
    {
      "epoch": 10.604598353675845,
      "grad_norm": 4.889222145080566,
      "learning_rate": 8.940079477717854e-05,
      "loss": 0.042106378078460696,
      "step": 37360
    },
    {
      "epoch": 10.607436843599205,
      "grad_norm": 5.18829870223999,
      "learning_rate": 8.939795628725518e-05,
      "loss": 0.057581347227096555,
      "step": 37370
    },
    {
      "epoch": 10.610275333522566,
      "grad_norm": 3.69331955909729,
      "learning_rate": 8.939511779733183e-05,
      "loss": 0.06446216702461242,
      "step": 37380
    },
    {
      "epoch": 10.613113823445927,
      "grad_norm": 2.446038246154785,
      "learning_rate": 8.939227930740846e-05,
      "loss": 0.03915868997573853,
      "step": 37390
    },
    {
      "epoch": 10.615952313369288,
      "grad_norm": 5.8372650146484375,
      "learning_rate": 8.93894408174851e-05,
      "loss": 0.03436077833175659,
      "step": 37400
    },
    {
      "epoch": 10.618790803292649,
      "grad_norm": 5.311520576477051,
      "learning_rate": 8.938660232756174e-05,
      "loss": 0.04923179149627686,
      "step": 37410
    },
    {
      "epoch": 10.621629293216008,
      "grad_norm": 10.053256034851074,
      "learning_rate": 8.938376383763838e-05,
      "loss": 0.052911609411239624,
      "step": 37420
    },
    {
      "epoch": 10.62446778313937,
      "grad_norm": 10.347527503967285,
      "learning_rate": 8.938092534771501e-05,
      "loss": 0.06862825751304627,
      "step": 37430
    },
    {
      "epoch": 10.62730627306273,
      "grad_norm": 2.2873501777648926,
      "learning_rate": 8.937808685779165e-05,
      "loss": 0.05170561671257019,
      "step": 37440
    },
    {
      "epoch": 10.630144762986092,
      "grad_norm": 3.918123245239258,
      "learning_rate": 8.937524836786831e-05,
      "loss": 0.0604759693145752,
      "step": 37450
    },
    {
      "epoch": 10.632983252909453,
      "grad_norm": 2.1962473392486572,
      "learning_rate": 8.937240987794494e-05,
      "loss": 0.08560603260993957,
      "step": 37460
    },
    {
      "epoch": 10.635821742832812,
      "grad_norm": 2.565603494644165,
      "learning_rate": 8.936957138802158e-05,
      "loss": 0.038162344694137575,
      "step": 37470
    },
    {
      "epoch": 10.638660232756173,
      "grad_norm": 17.403154373168945,
      "learning_rate": 8.936673289809822e-05,
      "loss": 0.040880581736564635,
      "step": 37480
    },
    {
      "epoch": 10.641498722679534,
      "grad_norm": 8.736148834228516,
      "learning_rate": 8.936389440817485e-05,
      "loss": 0.040731039643287656,
      "step": 37490
    },
    {
      "epoch": 10.644337212602895,
      "grad_norm": 6.361547946929932,
      "learning_rate": 8.936105591825149e-05,
      "loss": 0.036026608943939206,
      "step": 37500
    },
    {
      "epoch": 10.644337212602895,
      "eval_accuracy": 0.9541552743689197,
      "eval_loss": 0.13262097537517548,
      "eval_runtime": 31.3432,
      "eval_samples_per_second": 501.767,
      "eval_steps_per_second": 7.849,
      "step": 37500
    },
    {
      "epoch": 10.647175702526257,
      "grad_norm": 10.978135108947754,
      "learning_rate": 8.935821742832814e-05,
      "loss": 0.06716124415397644,
      "step": 37510
    },
    {
      "epoch": 10.650014192449618,
      "grad_norm": 4.695350170135498,
      "learning_rate": 8.935537893840477e-05,
      "loss": 0.05710494518280029,
      "step": 37520
    },
    {
      "epoch": 10.652852682372977,
      "grad_norm": 12.999497413635254,
      "learning_rate": 8.935254044848141e-05,
      "loss": 0.04939869046211243,
      "step": 37530
    },
    {
      "epoch": 10.655691172296338,
      "grad_norm": 4.029967308044434,
      "learning_rate": 8.934970195855805e-05,
      "loss": 0.045660588145256045,
      "step": 37540
    },
    {
      "epoch": 10.6585296622197,
      "grad_norm": 13.028033256530762,
      "learning_rate": 8.93468634686347e-05,
      "loss": 0.046709710359573366,
      "step": 37550
    },
    {
      "epoch": 10.66136815214306,
      "grad_norm": 11.557087898254395,
      "learning_rate": 8.934402497871132e-05,
      "loss": 0.07236909866333008,
      "step": 37560
    },
    {
      "epoch": 10.664206642066421,
      "grad_norm": 10.125807762145996,
      "learning_rate": 8.934118648878796e-05,
      "loss": 0.0551980197429657,
      "step": 37570
    },
    {
      "epoch": 10.66704513198978,
      "grad_norm": 14.962666511535645,
      "learning_rate": 8.933834799886462e-05,
      "loss": 0.04768535494804382,
      "step": 37580
    },
    {
      "epoch": 10.669883621913142,
      "grad_norm": 1.9633394479751587,
      "learning_rate": 8.933550950894125e-05,
      "loss": 0.03601632714271545,
      "step": 37590
    },
    {
      "epoch": 10.672722111836503,
      "grad_norm": 4.98323392868042,
      "learning_rate": 8.933267101901789e-05,
      "loss": 0.04608268141746521,
      "step": 37600
    },
    {
      "epoch": 10.675560601759864,
      "grad_norm": 11.655495643615723,
      "learning_rate": 8.932983252909453e-05,
      "loss": 0.05340247750282288,
      "step": 37610
    },
    {
      "epoch": 10.678399091683225,
      "grad_norm": 11.320283889770508,
      "learning_rate": 8.932699403917116e-05,
      "loss": 0.0428974986076355,
      "step": 37620
    },
    {
      "epoch": 10.681237581606585,
      "grad_norm": 3.8176426887512207,
      "learning_rate": 8.93241555492478e-05,
      "loss": 0.03072204887866974,
      "step": 37630
    },
    {
      "epoch": 10.684076071529946,
      "grad_norm": 15.627106666564941,
      "learning_rate": 8.932131705932444e-05,
      "loss": 0.05828758478164673,
      "step": 37640
    },
    {
      "epoch": 10.686914561453307,
      "grad_norm": 7.288362979888916,
      "learning_rate": 8.931847856940108e-05,
      "loss": 0.04542043209075928,
      "step": 37650
    },
    {
      "epoch": 10.689753051376668,
      "grad_norm": 11.484195709228516,
      "learning_rate": 8.931564007947772e-05,
      "loss": 0.0534932017326355,
      "step": 37660
    },
    {
      "epoch": 10.692591541300029,
      "grad_norm": 7.502041816711426,
      "learning_rate": 8.931280158955436e-05,
      "loss": 0.06448942422866821,
      "step": 37670
    },
    {
      "epoch": 10.695430031223388,
      "grad_norm": 7.892602920532227,
      "learning_rate": 8.9309963099631e-05,
      "loss": 0.053426170349121095,
      "step": 37680
    },
    {
      "epoch": 10.69826852114675,
      "grad_norm": 2.4152629375457764,
      "learning_rate": 8.930712460970763e-05,
      "loss": 0.043140605092048645,
      "step": 37690
    },
    {
      "epoch": 10.70110701107011,
      "grad_norm": 4.495814323425293,
      "learning_rate": 8.930428611978428e-05,
      "loss": 0.058317774534225465,
      "step": 37700
    },
    {
      "epoch": 10.703945500993472,
      "grad_norm": 7.092718601226807,
      "learning_rate": 8.930144762986093e-05,
      "loss": 0.04303047060966492,
      "step": 37710
    },
    {
      "epoch": 10.706783990916833,
      "grad_norm": 9.289896011352539,
      "learning_rate": 8.929860913993756e-05,
      "loss": 0.04771975576877594,
      "step": 37720
    },
    {
      "epoch": 10.709622480840192,
      "grad_norm": 8.999591827392578,
      "learning_rate": 8.92957706500142e-05,
      "loss": 0.0611985445022583,
      "step": 37730
    },
    {
      "epoch": 10.712460970763553,
      "grad_norm": 6.30220365524292,
      "learning_rate": 8.929293216009084e-05,
      "loss": 0.054555898904800414,
      "step": 37740
    },
    {
      "epoch": 10.715299460686914,
      "grad_norm": 7.283388614654541,
      "learning_rate": 8.929009367016747e-05,
      "loss": 0.0529313862323761,
      "step": 37750
    },
    {
      "epoch": 10.718137950610275,
      "grad_norm": 12.678055763244629,
      "learning_rate": 8.928725518024411e-05,
      "loss": 0.05128545761108398,
      "step": 37760
    },
    {
      "epoch": 10.720976440533637,
      "grad_norm": 7.386247634887695,
      "learning_rate": 8.928441669032075e-05,
      "loss": 0.03549208641052246,
      "step": 37770
    },
    {
      "epoch": 10.723814930456998,
      "grad_norm": 7.320062160491943,
      "learning_rate": 8.928157820039739e-05,
      "loss": 0.049683955311775205,
      "step": 37780
    },
    {
      "epoch": 10.726653420380357,
      "grad_norm": 7.607080936431885,
      "learning_rate": 8.927873971047403e-05,
      "loss": 0.0352347731590271,
      "step": 37790
    },
    {
      "epoch": 10.729491910303718,
      "grad_norm": 12.08772087097168,
      "learning_rate": 8.927590122055068e-05,
      "loss": 0.03941766023635864,
      "step": 37800
    },
    {
      "epoch": 10.73233040022708,
      "grad_norm": 4.185817241668701,
      "learning_rate": 8.927306273062732e-05,
      "loss": 0.0489218533039093,
      "step": 37810
    },
    {
      "epoch": 10.73516889015044,
      "grad_norm": 10.718232154846191,
      "learning_rate": 8.927022424070394e-05,
      "loss": 0.04533816874027252,
      "step": 37820
    },
    {
      "epoch": 10.738007380073801,
      "grad_norm": 11.412670135498047,
      "learning_rate": 8.926738575078059e-05,
      "loss": 0.05357496738433838,
      "step": 37830
    },
    {
      "epoch": 10.74084586999716,
      "grad_norm": 12.72150707244873,
      "learning_rate": 8.926454726085723e-05,
      "loss": 0.08674684166908264,
      "step": 37840
    },
    {
      "epoch": 10.743684359920522,
      "grad_norm": 4.283827781677246,
      "learning_rate": 8.926170877093387e-05,
      "loss": 0.0715071678161621,
      "step": 37850
    },
    {
      "epoch": 10.746522849843883,
      "grad_norm": 7.868562698364258,
      "learning_rate": 8.925887028101051e-05,
      "loss": 0.06395405530929565,
      "step": 37860
    },
    {
      "epoch": 10.749361339767244,
      "grad_norm": 2.7977356910705566,
      "learning_rate": 8.925603179108715e-05,
      "loss": 0.05910104513168335,
      "step": 37870
    },
    {
      "epoch": 10.752199829690605,
      "grad_norm": 8.621735572814941,
      "learning_rate": 8.925319330116378e-05,
      "loss": 0.06608824729919434,
      "step": 37880
    },
    {
      "epoch": 10.755038319613966,
      "grad_norm": 11.802695274353027,
      "learning_rate": 8.925035481124042e-05,
      "loss": 0.06655054688453674,
      "step": 37890
    },
    {
      "epoch": 10.757876809537326,
      "grad_norm": 0.9506636261940002,
      "learning_rate": 8.924751632131706e-05,
      "loss": 0.045469969511032104,
      "step": 37900
    },
    {
      "epoch": 10.760715299460687,
      "grad_norm": 9.88681411743164,
      "learning_rate": 8.92446778313937e-05,
      "loss": 0.04759877920150757,
      "step": 37910
    },
    {
      "epoch": 10.763553789384048,
      "grad_norm": 8.221586227416992,
      "learning_rate": 8.924183934147034e-05,
      "loss": 0.0642558753490448,
      "step": 37920
    },
    {
      "epoch": 10.766392279307409,
      "grad_norm": 4.3333234786987305,
      "learning_rate": 8.923900085154699e-05,
      "loss": 0.036686384677886964,
      "step": 37930
    },
    {
      "epoch": 10.76923076923077,
      "grad_norm": 11.051621437072754,
      "learning_rate": 8.923616236162361e-05,
      "loss": 0.037858805060386656,
      "step": 37940
    },
    {
      "epoch": 10.77206925915413,
      "grad_norm": 5.9207963943481445,
      "learning_rate": 8.923332387170026e-05,
      "loss": 0.04645982384681702,
      "step": 37950
    },
    {
      "epoch": 10.77490774907749,
      "grad_norm": 1.5580244064331055,
      "learning_rate": 8.92304853817769e-05,
      "loss": 0.041862720251083375,
      "step": 37960
    },
    {
      "epoch": 10.777746239000852,
      "grad_norm": 3.5849783420562744,
      "learning_rate": 8.922764689185354e-05,
      "loss": 0.04566018283367157,
      "step": 37970
    },
    {
      "epoch": 10.780584728924213,
      "grad_norm": 3.8744287490844727,
      "learning_rate": 8.922480840193018e-05,
      "loss": 0.045349889993667604,
      "step": 37980
    },
    {
      "epoch": 10.783423218847574,
      "grad_norm": 15.801223754882812,
      "learning_rate": 8.922196991200682e-05,
      "loss": 0.06152545213699341,
      "step": 37990
    },
    {
      "epoch": 10.786261708770933,
      "grad_norm": 4.121667385101318,
      "learning_rate": 8.921913142208346e-05,
      "loss": 0.0529982328414917,
      "step": 38000
    },
    {
      "epoch": 10.786261708770933,
      "eval_accuracy": 0.954791123545495,
      "eval_loss": 0.13791421055793762,
      "eval_runtime": 31.4972,
      "eval_samples_per_second": 499.314,
      "eval_steps_per_second": 7.81,
      "step": 38000
    },
    {
      "epoch": 10.789100198694294,
      "grad_norm": 8.214282989501953,
      "learning_rate": 8.921629293216009e-05,
      "loss": 0.05908235311508179,
      "step": 38010
    },
    {
      "epoch": 10.791938688617655,
      "grad_norm": 8.53370189666748,
      "learning_rate": 8.921345444223673e-05,
      "loss": 0.05178715586662293,
      "step": 38020
    },
    {
      "epoch": 10.794777178541016,
      "grad_norm": 13.354255676269531,
      "learning_rate": 8.921061595231337e-05,
      "loss": 0.05319744348526001,
      "step": 38030
    },
    {
      "epoch": 10.797615668464378,
      "grad_norm": 7.935073375701904,
      "learning_rate": 8.920777746239e-05,
      "loss": 0.04854918122291565,
      "step": 38040
    },
    {
      "epoch": 10.800454158387737,
      "grad_norm": 9.971450805664062,
      "learning_rate": 8.920493897246666e-05,
      "loss": 0.03835484087467193,
      "step": 38050
    },
    {
      "epoch": 10.803292648311098,
      "grad_norm": 3.2206015586853027,
      "learning_rate": 8.92021004825433e-05,
      "loss": 0.03390050232410431,
      "step": 38060
    },
    {
      "epoch": 10.80613113823446,
      "grad_norm": 5.002132415771484,
      "learning_rate": 8.919926199261992e-05,
      "loss": 0.06402709484100341,
      "step": 38070
    },
    {
      "epoch": 10.80896962815782,
      "grad_norm": 4.769713401794434,
      "learning_rate": 8.919642350269657e-05,
      "loss": 0.055002915859222415,
      "step": 38080
    },
    {
      "epoch": 10.811808118081181,
      "grad_norm": 12.758962631225586,
      "learning_rate": 8.919358501277321e-05,
      "loss": 0.04941333532333374,
      "step": 38090
    },
    {
      "epoch": 10.81464660800454,
      "grad_norm": 12.254110336303711,
      "learning_rate": 8.919074652284985e-05,
      "loss": 0.058733904361724855,
      "step": 38100
    },
    {
      "epoch": 10.817485097927902,
      "grad_norm": 10.186753273010254,
      "learning_rate": 8.918790803292649e-05,
      "loss": 0.039628738164901735,
      "step": 38110
    },
    {
      "epoch": 10.820323587851263,
      "grad_norm": 8.08736801147461,
      "learning_rate": 8.918506954300313e-05,
      "loss": 0.04123604893684387,
      "step": 38120
    },
    {
      "epoch": 10.823162077774624,
      "grad_norm": 6.497982501983643,
      "learning_rate": 8.918223105307977e-05,
      "loss": 0.06079375743865967,
      "step": 38130
    },
    {
      "epoch": 10.826000567697985,
      "grad_norm": 1.8209428787231445,
      "learning_rate": 8.91793925631564e-05,
      "loss": 0.031149780750274657,
      "step": 38140
    },
    {
      "epoch": 10.828839057621346,
      "grad_norm": 3.366511821746826,
      "learning_rate": 8.917655407323304e-05,
      "loss": 0.042412635684013364,
      "step": 38150
    },
    {
      "epoch": 10.831677547544706,
      "grad_norm": 3.608931303024292,
      "learning_rate": 8.917371558330968e-05,
      "loss": 0.045526671409606936,
      "step": 38160
    },
    {
      "epoch": 10.834516037468067,
      "grad_norm": 5.949339389801025,
      "learning_rate": 8.917087709338631e-05,
      "loss": 0.05724095106124878,
      "step": 38170
    },
    {
      "epoch": 10.837354527391428,
      "grad_norm": 6.7733049392700195,
      "learning_rate": 8.916803860346297e-05,
      "loss": 0.047005188465118405,
      "step": 38180
    },
    {
      "epoch": 10.840193017314789,
      "grad_norm": 25.709814071655273,
      "learning_rate": 8.916520011353961e-05,
      "loss": 0.08772512674331664,
      "step": 38190
    },
    {
      "epoch": 10.84303150723815,
      "grad_norm": 16.664751052856445,
      "learning_rate": 8.916236162361624e-05,
      "loss": 0.06233923435211182,
      "step": 38200
    },
    {
      "epoch": 10.84586999716151,
      "grad_norm": 5.98714017868042,
      "learning_rate": 8.915980698268522e-05,
      "loss": 0.06928134560585023,
      "step": 38210
    },
    {
      "epoch": 10.84870848708487,
      "grad_norm": 10.891114234924316,
      "learning_rate": 8.915696849276184e-05,
      "loss": 0.053616517782211305,
      "step": 38220
    },
    {
      "epoch": 10.851546977008232,
      "grad_norm": 3.9036974906921387,
      "learning_rate": 8.91541300028385e-05,
      "loss": 0.04729513525962829,
      "step": 38230
    },
    {
      "epoch": 10.854385466931593,
      "grad_norm": 13.764676094055176,
      "learning_rate": 8.915129151291514e-05,
      "loss": 0.05080227851867676,
      "step": 38240
    },
    {
      "epoch": 10.857223956854954,
      "grad_norm": 7.141643524169922,
      "learning_rate": 8.914845302299177e-05,
      "loss": 0.03850345313549042,
      "step": 38250
    },
    {
      "epoch": 10.860062446778313,
      "grad_norm": 6.551845073699951,
      "learning_rate": 8.914561453306841e-05,
      "loss": 0.044964417815208435,
      "step": 38260
    },
    {
      "epoch": 10.862900936701674,
      "grad_norm": 8.506628036499023,
      "learning_rate": 8.914277604314505e-05,
      "loss": 0.04852698743343353,
      "step": 38270
    },
    {
      "epoch": 10.865739426625035,
      "grad_norm": 4.149714469909668,
      "learning_rate": 8.913993755322169e-05,
      "loss": 0.04149608314037323,
      "step": 38280
    },
    {
      "epoch": 10.868577916548396,
      "grad_norm": 8.39784049987793,
      "learning_rate": 8.913709906329833e-05,
      "loss": 0.037533077597618106,
      "step": 38290
    },
    {
      "epoch": 10.871416406471758,
      "grad_norm": 5.319234848022461,
      "learning_rate": 8.913426057337497e-05,
      "loss": 0.038730254769325255,
      "step": 38300
    },
    {
      "epoch": 10.874254896395119,
      "grad_norm": 5.00916051864624,
      "learning_rate": 8.913142208345162e-05,
      "loss": 0.060316479206085204,
      "step": 38310
    },
    {
      "epoch": 10.877093386318478,
      "grad_norm": 9.620858192443848,
      "learning_rate": 8.912858359352824e-05,
      "loss": 0.043903243541717527,
      "step": 38320
    },
    {
      "epoch": 10.879931876241839,
      "grad_norm": 3.082885503768921,
      "learning_rate": 8.912574510360489e-05,
      "loss": 0.04673202931880951,
      "step": 38330
    },
    {
      "epoch": 10.8827703661652,
      "grad_norm": 2.890110969543457,
      "learning_rate": 8.912290661368153e-05,
      "loss": 0.04546935260295868,
      "step": 38340
    },
    {
      "epoch": 10.885608856088561,
      "grad_norm": 2.106387138366699,
      "learning_rate": 8.912006812375815e-05,
      "loss": 0.05234392285346985,
      "step": 38350
    },
    {
      "epoch": 10.888447346011922,
      "grad_norm": 10.52157974243164,
      "learning_rate": 8.911722963383481e-05,
      "loss": 0.044599348306655885,
      "step": 38360
    },
    {
      "epoch": 10.891285835935282,
      "grad_norm": 11.754111289978027,
      "learning_rate": 8.911439114391145e-05,
      "loss": 0.05273027420043945,
      "step": 38370
    },
    {
      "epoch": 10.894124325858643,
      "grad_norm": 7.142884731292725,
      "learning_rate": 8.911155265398808e-05,
      "loss": 0.05217101573944092,
      "step": 38380
    },
    {
      "epoch": 10.896962815782004,
      "grad_norm": 5.7373738288879395,
      "learning_rate": 8.910871416406472e-05,
      "loss": 0.041763636469841006,
      "step": 38390
    },
    {
      "epoch": 10.899801305705365,
      "grad_norm": 9.567590713500977,
      "learning_rate": 8.910587567414136e-05,
      "loss": 0.05150666832923889,
      "step": 38400
    },
    {
      "epoch": 10.902639795628726,
      "grad_norm": 10.401607513427734,
      "learning_rate": 8.9103037184218e-05,
      "loss": 0.050959455966949466,
      "step": 38410
    },
    {
      "epoch": 10.905478285552086,
      "grad_norm": 10.096796035766602,
      "learning_rate": 8.910019869429463e-05,
      "loss": 0.04512452483177185,
      "step": 38420
    },
    {
      "epoch": 10.908316775475447,
      "grad_norm": 4.9429931640625,
      "learning_rate": 8.909736020437129e-05,
      "loss": 0.04739703834056854,
      "step": 38430
    },
    {
      "epoch": 10.911155265398808,
      "grad_norm": 5.128699779510498,
      "learning_rate": 8.909452171444793e-05,
      "loss": 0.04005731642246246,
      "step": 38440
    },
    {
      "epoch": 10.913993755322169,
      "grad_norm": 11.468729019165039,
      "learning_rate": 8.909168322452455e-05,
      "loss": 0.05342747569084168,
      "step": 38450
    },
    {
      "epoch": 10.91683224524553,
      "grad_norm": 8.957573890686035,
      "learning_rate": 8.90888447346012e-05,
      "loss": 0.06143091320991516,
      "step": 38460
    },
    {
      "epoch": 10.91967073516889,
      "grad_norm": 10.171913146972656,
      "learning_rate": 8.908600624467784e-05,
      "loss": 0.051012343168258666,
      "step": 38470
    },
    {
      "epoch": 10.92250922509225,
      "grad_norm": 8.428825378417969,
      "learning_rate": 8.908316775475446e-05,
      "loss": 0.06785122156143189,
      "step": 38480
    },
    {
      "epoch": 10.925347715015612,
      "grad_norm": 8.689305305480957,
      "learning_rate": 8.908032926483112e-05,
      "loss": 0.059849923849105834,
      "step": 38490
    },
    {
      "epoch": 10.928186204938973,
      "grad_norm": 11.59469223022461,
      "learning_rate": 8.907749077490776e-05,
      "loss": 0.03448027074337005,
      "step": 38500
    },
    {
      "epoch": 10.928186204938973,
      "eval_accuracy": 0.9539645196159471,
      "eval_loss": 0.1362266093492508,
      "eval_runtime": 32.0371,
      "eval_samples_per_second": 490.899,
      "eval_steps_per_second": 7.679,
      "step": 38500
    },
    {
      "epoch": 10.931024694862334,
      "grad_norm": 12.533304214477539,
      "learning_rate": 8.907465228498439e-05,
      "loss": 0.06095793843269348,
      "step": 38510
    },
    {
      "epoch": 10.933863184785693,
      "grad_norm": 5.1569600105285645,
      "learning_rate": 8.907181379506103e-05,
      "loss": 0.04762200117111206,
      "step": 38520
    },
    {
      "epoch": 10.936701674709054,
      "grad_norm": 5.558366298675537,
      "learning_rate": 8.906897530513767e-05,
      "loss": 0.030347022414207458,
      "step": 38530
    },
    {
      "epoch": 10.939540164632415,
      "grad_norm": 1.4949798583984375,
      "learning_rate": 8.906613681521431e-05,
      "loss": 0.09603907465934754,
      "step": 38540
    },
    {
      "epoch": 10.942378654555776,
      "grad_norm": 4.198401927947998,
      "learning_rate": 8.906329832529094e-05,
      "loss": 0.03959104716777802,
      "step": 38550
    },
    {
      "epoch": 10.945217144479138,
      "grad_norm": 3.2333035469055176,
      "learning_rate": 8.90604598353676e-05,
      "loss": 0.06963757276535035,
      "step": 38560
    },
    {
      "epoch": 10.948055634402499,
      "grad_norm": 15.018879890441895,
      "learning_rate": 8.905762134544424e-05,
      "loss": 0.06735700368881226,
      "step": 38570
    },
    {
      "epoch": 10.950894124325858,
      "grad_norm": 1.3846895694732666,
      "learning_rate": 8.905478285552087e-05,
      "loss": 0.04106098413467407,
      "step": 38580
    },
    {
      "epoch": 10.953732614249219,
      "grad_norm": 7.6418890953063965,
      "learning_rate": 8.90519443655975e-05,
      "loss": 0.052422887086868285,
      "step": 38590
    },
    {
      "epoch": 10.95657110417258,
      "grad_norm": 7.431676387786865,
      "learning_rate": 8.904910587567415e-05,
      "loss": 0.057167887687683105,
      "step": 38600
    },
    {
      "epoch": 10.959409594095941,
      "grad_norm": 6.120459079742432,
      "learning_rate": 8.904626738575078e-05,
      "loss": 0.04599092602729797,
      "step": 38610
    },
    {
      "epoch": 10.962248084019302,
      "grad_norm": 3.009070873260498,
      "learning_rate": 8.904342889582742e-05,
      "loss": 0.03604561984539032,
      "step": 38620
    },
    {
      "epoch": 10.965086573942662,
      "grad_norm": 7.108126163482666,
      "learning_rate": 8.904059040590407e-05,
      "loss": 0.07322196960449219,
      "step": 38630
    },
    {
      "epoch": 10.967925063866023,
      "grad_norm": 6.833071708679199,
      "learning_rate": 8.90377519159807e-05,
      "loss": 0.034294241666793825,
      "step": 38640
    },
    {
      "epoch": 10.970763553789384,
      "grad_norm": 2.084627389907837,
      "learning_rate": 8.903491342605734e-05,
      "loss": 0.05035883784294128,
      "step": 38650
    },
    {
      "epoch": 10.973602043712745,
      "grad_norm": 7.055487632751465,
      "learning_rate": 8.903207493613398e-05,
      "loss": 0.030414706468582152,
      "step": 38660
    },
    {
      "epoch": 10.976440533636106,
      "grad_norm": 8.627908706665039,
      "learning_rate": 8.902923644621062e-05,
      "loss": 0.06591638922691345,
      "step": 38670
    },
    {
      "epoch": 10.979279023559467,
      "grad_norm": 4.9932403564453125,
      "learning_rate": 8.902639795628725e-05,
      "loss": 0.05395441055297852,
      "step": 38680
    },
    {
      "epoch": 10.982117513482827,
      "grad_norm": 15.169208526611328,
      "learning_rate": 8.90235594663639e-05,
      "loss": 0.05914019346237183,
      "step": 38690
    },
    {
      "epoch": 10.984956003406188,
      "grad_norm": 10.65866756439209,
      "learning_rate": 8.902072097644055e-05,
      "loss": 0.0431813508272171,
      "step": 38700
    },
    {
      "epoch": 10.987794493329549,
      "grad_norm": 20.115015029907227,
      "learning_rate": 8.901788248651718e-05,
      "loss": 0.08140392899513245,
      "step": 38710
    },
    {
      "epoch": 10.99063298325291,
      "grad_norm": 1.754321813583374,
      "learning_rate": 8.901504399659382e-05,
      "loss": 0.05766031742095947,
      "step": 38720
    },
    {
      "epoch": 10.993471473176271,
      "grad_norm": 3.1317625045776367,
      "learning_rate": 8.901220550667046e-05,
      "loss": 0.05524242520332336,
      "step": 38730
    },
    {
      "epoch": 10.99630996309963,
      "grad_norm": 3.33754301071167,
      "learning_rate": 8.900936701674709e-05,
      "loss": 0.02870774269104004,
      "step": 38740
    },
    {
      "epoch": 10.999148453022991,
      "grad_norm": 8.899383544921875,
      "learning_rate": 8.900652852682373e-05,
      "loss": 0.05789498686790466,
      "step": 38750
    },
    {
      "epoch": 11.001986942946353,
      "grad_norm": 14.599570274353027,
      "learning_rate": 8.900369003690038e-05,
      "loss": 0.06009508371353149,
      "step": 38760
    },
    {
      "epoch": 11.004825432869714,
      "grad_norm": 11.534079551696777,
      "learning_rate": 8.900085154697701e-05,
      "loss": 0.044374477863311765,
      "step": 38770
    },
    {
      "epoch": 11.007663922793075,
      "grad_norm": 6.6945576667785645,
      "learning_rate": 8.899801305705365e-05,
      "loss": 0.026109185814857484,
      "step": 38780
    },
    {
      "epoch": 11.010502412716434,
      "grad_norm": 5.227405071258545,
      "learning_rate": 8.89951745671303e-05,
      "loss": 0.02248488962650299,
      "step": 38790
    },
    {
      "epoch": 11.013340902639795,
      "grad_norm": 6.267428398132324,
      "learning_rate": 8.899233607720693e-05,
      "loss": 0.042588543891906736,
      "step": 38800
    },
    {
      "epoch": 11.016179392563156,
      "grad_norm": 10.655970573425293,
      "learning_rate": 8.898949758728356e-05,
      "loss": 0.05288074016571045,
      "step": 38810
    },
    {
      "epoch": 11.019017882486517,
      "grad_norm": 10.031039237976074,
      "learning_rate": 8.89866590973602e-05,
      "loss": 0.051262885332107544,
      "step": 38820
    },
    {
      "epoch": 11.021856372409879,
      "grad_norm": 3.097400426864624,
      "learning_rate": 8.898382060743685e-05,
      "loss": 0.05224065780639649,
      "step": 38830
    },
    {
      "epoch": 11.024694862333238,
      "grad_norm": 4.993659019470215,
      "learning_rate": 8.898098211751349e-05,
      "loss": 0.051405346393585204,
      "step": 38840
    },
    {
      "epoch": 11.027533352256599,
      "grad_norm": 6.353282928466797,
      "learning_rate": 8.897814362759013e-05,
      "loss": 0.03370220959186554,
      "step": 38850
    },
    {
      "epoch": 11.03037184217996,
      "grad_norm": 2.966917037963867,
      "learning_rate": 8.897530513766677e-05,
      "loss": 0.048281505703926086,
      "step": 38860
    },
    {
      "epoch": 11.033210332103321,
      "grad_norm": 21.087085723876953,
      "learning_rate": 8.89724666477434e-05,
      "loss": 0.041472843289375304,
      "step": 38870
    },
    {
      "epoch": 11.036048822026682,
      "grad_norm": 2.9310221672058105,
      "learning_rate": 8.896962815782004e-05,
      "loss": 0.03393231630325318,
      "step": 38880
    },
    {
      "epoch": 11.038887311950042,
      "grad_norm": 2.4367799758911133,
      "learning_rate": 8.89667896678967e-05,
      "loss": 0.0396744042634964,
      "step": 38890
    },
    {
      "epoch": 11.041725801873403,
      "grad_norm": 13.638497352600098,
      "learning_rate": 8.896395117797332e-05,
      "loss": 0.028379836678504945,
      "step": 38900
    },
    {
      "epoch": 11.044564291796764,
      "grad_norm": 2.574260711669922,
      "learning_rate": 8.896111268804996e-05,
      "loss": 0.04455780684947967,
      "step": 38910
    },
    {
      "epoch": 11.047402781720125,
      "grad_norm": 5.962393760681152,
      "learning_rate": 8.89582741981266e-05,
      "loss": 0.04642617702484131,
      "step": 38920
    },
    {
      "epoch": 11.050241271643486,
      "grad_norm": 0.777250349521637,
      "learning_rate": 8.895543570820323e-05,
      "loss": 0.033685031533241275,
      "step": 38930
    },
    {
      "epoch": 11.053079761566847,
      "grad_norm": 7.838201522827148,
      "learning_rate": 8.895259721827987e-05,
      "loss": 0.04278995990753174,
      "step": 38940
    },
    {
      "epoch": 11.055918251490207,
      "grad_norm": 2.831878185272217,
      "learning_rate": 8.894975872835651e-05,
      "loss": 0.03615584969520569,
      "step": 38950
    },
    {
      "epoch": 11.058756741413568,
      "grad_norm": 4.757225036621094,
      "learning_rate": 8.894692023843316e-05,
      "loss": 0.03551706671714783,
      "step": 38960
    },
    {
      "epoch": 11.061595231336929,
      "grad_norm": 5.683297634124756,
      "learning_rate": 8.89440817485098e-05,
      "loss": 0.039518555998802184,
      "step": 38970
    },
    {
      "epoch": 11.06443372126029,
      "grad_norm": 10.178922653198242,
      "learning_rate": 8.894124325858644e-05,
      "loss": 0.039110153913497925,
      "step": 38980
    },
    {
      "epoch": 11.067272211183651,
      "grad_norm": 1.9131416082382202,
      "learning_rate": 8.893840476866308e-05,
      "loss": 0.039188307523727414,
      "step": 38990
    },
    {
      "epoch": 11.07011070110701,
      "grad_norm": 17.690940856933594,
      "learning_rate": 8.893556627873971e-05,
      "loss": 0.059947621822357175,
      "step": 39000
    },
    {
      "epoch": 11.07011070110701,
      "eval_accuracy": 0.9514847078273033,
      "eval_loss": 0.14660021662712097,
      "eval_runtime": 31.1761,
      "eval_samples_per_second": 504.457,
      "eval_steps_per_second": 7.891,
      "step": 39000
    },
    {
      "epoch": 11.072949191030371,
      "grad_norm": 5.929471492767334,
      "learning_rate": 8.893272778881635e-05,
      "loss": 0.027347517013549805,
      "step": 39010
    },
    {
      "epoch": 11.075787680953733,
      "grad_norm": 6.718734264373779,
      "learning_rate": 8.892988929889299e-05,
      "loss": 0.056528902053833006,
      "step": 39020
    },
    {
      "epoch": 11.078626170877094,
      "grad_norm": 6.625723361968994,
      "learning_rate": 8.892705080896963e-05,
      "loss": 0.06551033854484559,
      "step": 39030
    },
    {
      "epoch": 11.081464660800455,
      "grad_norm": 3.869439125061035,
      "learning_rate": 8.892421231904627e-05,
      "loss": 0.039598909020423886,
      "step": 39040
    },
    {
      "epoch": 11.084303150723814,
      "grad_norm": 7.520358562469482,
      "learning_rate": 8.892137382912291e-05,
      "loss": 0.03328187465667724,
      "step": 39050
    },
    {
      "epoch": 11.087141640647175,
      "grad_norm": 2.2077038288116455,
      "learning_rate": 8.891853533919954e-05,
      "loss": 0.027495452761650087,
      "step": 39060
    },
    {
      "epoch": 11.089980130570536,
      "grad_norm": 5.891507625579834,
      "learning_rate": 8.891569684927618e-05,
      "loss": 0.030188310146331786,
      "step": 39070
    },
    {
      "epoch": 11.092818620493897,
      "grad_norm": 6.687758445739746,
      "learning_rate": 8.891285835935283e-05,
      "loss": 0.04934812784194946,
      "step": 39080
    },
    {
      "epoch": 11.095657110417259,
      "grad_norm": 9.199897766113281,
      "learning_rate": 8.891001986942947e-05,
      "loss": 0.042330962419509885,
      "step": 39090
    },
    {
      "epoch": 11.09849560034062,
      "grad_norm": 8.42823600769043,
      "learning_rate": 8.890718137950611e-05,
      "loss": 0.041123732924461365,
      "step": 39100
    },
    {
      "epoch": 11.101334090263979,
      "grad_norm": 7.519351959228516,
      "learning_rate": 8.890434288958275e-05,
      "loss": 0.04811866879463196,
      "step": 39110
    },
    {
      "epoch": 11.10417258018734,
      "grad_norm": 10.867979049682617,
      "learning_rate": 8.890150439965939e-05,
      "loss": 0.021816276013851166,
      "step": 39120
    },
    {
      "epoch": 11.107011070110701,
      "grad_norm": 11.573338508605957,
      "learning_rate": 8.889866590973602e-05,
      "loss": 0.0337225079536438,
      "step": 39130
    },
    {
      "epoch": 11.109849560034062,
      "grad_norm": 5.146324634552002,
      "learning_rate": 8.889582741981266e-05,
      "loss": 0.042096379399299624,
      "step": 39140
    },
    {
      "epoch": 11.112688049957423,
      "grad_norm": 12.381686210632324,
      "learning_rate": 8.88929889298893e-05,
      "loss": 0.035431957244873045,
      "step": 39150
    },
    {
      "epoch": 11.115526539880783,
      "grad_norm": 6.369025707244873,
      "learning_rate": 8.889015043996594e-05,
      "loss": 0.04335471987724304,
      "step": 39160
    },
    {
      "epoch": 11.118365029804144,
      "grad_norm": 1.6466095447540283,
      "learning_rate": 8.888731195004258e-05,
      "loss": 0.04059185683727264,
      "step": 39170
    },
    {
      "epoch": 11.121203519727505,
      "grad_norm": 5.768196105957031,
      "learning_rate": 8.888447346011923e-05,
      "loss": 0.037748593091964724,
      "step": 39180
    },
    {
      "epoch": 11.124042009650866,
      "grad_norm": 8.6279935836792,
      "learning_rate": 8.888163497019585e-05,
      "loss": 0.0365769624710083,
      "step": 39190
    },
    {
      "epoch": 11.126880499574227,
      "grad_norm": 11.250873565673828,
      "learning_rate": 8.88787964802725e-05,
      "loss": 0.04417133927345276,
      "step": 39200
    },
    {
      "epoch": 11.129718989497587,
      "grad_norm": 3.655747652053833,
      "learning_rate": 8.887595799034914e-05,
      "loss": 0.03420442938804626,
      "step": 39210
    },
    {
      "epoch": 11.132557479420948,
      "grad_norm": 1.911678433418274,
      "learning_rate": 8.887311950042578e-05,
      "loss": 0.04320805966854095,
      "step": 39220
    },
    {
      "epoch": 11.135395969344309,
      "grad_norm": 6.267022132873535,
      "learning_rate": 8.887028101050242e-05,
      "loss": 0.04816638827323914,
      "step": 39230
    },
    {
      "epoch": 11.13823445926767,
      "grad_norm": 5.782435417175293,
      "learning_rate": 8.886744252057906e-05,
      "loss": 0.04777747094631195,
      "step": 39240
    },
    {
      "epoch": 11.141072949191031,
      "grad_norm": 4.721435546875,
      "learning_rate": 8.88646040306557e-05,
      "loss": 0.03958534300327301,
      "step": 39250
    },
    {
      "epoch": 11.14391143911439,
      "grad_norm": 1.6922602653503418,
      "learning_rate": 8.886176554073233e-05,
      "loss": 0.03485212028026581,
      "step": 39260
    },
    {
      "epoch": 11.146749929037751,
      "grad_norm": 10.722707748413086,
      "learning_rate": 8.885892705080897e-05,
      "loss": 0.052032476663589476,
      "step": 39270
    },
    {
      "epoch": 11.149588418961113,
      "grad_norm": 11.775294303894043,
      "learning_rate": 8.885608856088561e-05,
      "loss": 0.028667646646499633,
      "step": 39280
    },
    {
      "epoch": 11.152426908884474,
      "grad_norm": 6.354204177856445,
      "learning_rate": 8.885325007096225e-05,
      "loss": 0.037648916244506836,
      "step": 39290
    },
    {
      "epoch": 11.155265398807835,
      "grad_norm": 13.509031295776367,
      "learning_rate": 8.88504115810389e-05,
      "loss": 0.04364720582962036,
      "step": 39300
    },
    {
      "epoch": 11.158103888731196,
      "grad_norm": 7.607676982879639,
      "learning_rate": 8.884757309111554e-05,
      "loss": 0.04972715973854065,
      "step": 39310
    },
    {
      "epoch": 11.160942378654555,
      "grad_norm": 11.344837188720703,
      "learning_rate": 8.884473460119216e-05,
      "loss": 0.05181915760040283,
      "step": 39320
    },
    {
      "epoch": 11.163780868577916,
      "grad_norm": 5.426929473876953,
      "learning_rate": 8.88418961112688e-05,
      "loss": 0.048836493492126466,
      "step": 39330
    },
    {
      "epoch": 11.166619358501277,
      "grad_norm": 1.4606432914733887,
      "learning_rate": 8.883905762134545e-05,
      "loss": 0.04444275200366974,
      "step": 39340
    },
    {
      "epoch": 11.169457848424639,
      "grad_norm": 5.54641580581665,
      "learning_rate": 8.883621913142209e-05,
      "loss": 0.04066909849643707,
      "step": 39350
    },
    {
      "epoch": 11.172296338348,
      "grad_norm": 7.7750773429870605,
      "learning_rate": 8.883338064149873e-05,
      "loss": 0.036909368634223935,
      "step": 39360
    },
    {
      "epoch": 11.175134828271359,
      "grad_norm": 9.151082992553711,
      "learning_rate": 8.883054215157537e-05,
      "loss": 0.036677679419517516,
      "step": 39370
    },
    {
      "epoch": 11.17797331819472,
      "grad_norm": 10.54290771484375,
      "learning_rate": 8.882770366165201e-05,
      "loss": 0.031220394372940063,
      "step": 39380
    },
    {
      "epoch": 11.180811808118081,
      "grad_norm": 2.6449501514434814,
      "learning_rate": 8.882486517172864e-05,
      "loss": 0.03766760528087616,
      "step": 39390
    },
    {
      "epoch": 11.183650298041442,
      "grad_norm": 6.63143253326416,
      "learning_rate": 8.882202668180528e-05,
      "loss": 0.0867281973361969,
      "step": 39400
    },
    {
      "epoch": 11.186488787964803,
      "grad_norm": 1.8736786842346191,
      "learning_rate": 8.881918819188192e-05,
      "loss": 0.04634508788585663,
      "step": 39410
    },
    {
      "epoch": 11.189327277888163,
      "grad_norm": 3.681365728378296,
      "learning_rate": 8.881634970195856e-05,
      "loss": 0.04647490978240967,
      "step": 39420
    },
    {
      "epoch": 11.192165767811524,
      "grad_norm": 8.258210182189941,
      "learning_rate": 8.88135112120352e-05,
      "loss": 0.03525558114051819,
      "step": 39430
    },
    {
      "epoch": 11.195004257734885,
      "grad_norm": 10.567083358764648,
      "learning_rate": 8.881067272211185e-05,
      "loss": 0.040261727571487424,
      "step": 39440
    },
    {
      "epoch": 11.197842747658246,
      "grad_norm": 1.8331791162490845,
      "learning_rate": 8.880783423218848e-05,
      "loss": 0.033540600538253786,
      "step": 39450
    },
    {
      "epoch": 11.200681237581607,
      "grad_norm": 2.7237627506256104,
      "learning_rate": 8.880499574226512e-05,
      "loss": 0.024020782113075255,
      "step": 39460
    },
    {
      "epoch": 11.203519727504966,
      "grad_norm": 3.1561853885650635,
      "learning_rate": 8.880215725234176e-05,
      "loss": 0.05169380307197571,
      "step": 39470
    },
    {
      "epoch": 11.206358217428328,
      "grad_norm": 17.208776473999023,
      "learning_rate": 8.87993187624184e-05,
      "loss": 0.06630215644836426,
      "step": 39480
    },
    {
      "epoch": 11.209196707351689,
      "grad_norm": 9.18853759765625,
      "learning_rate": 8.879648027249504e-05,
      "loss": 0.06524783372879028,
      "step": 39490
    },
    {
      "epoch": 11.21203519727505,
      "grad_norm": 8.582340240478516,
      "learning_rate": 8.879364178257168e-05,
      "loss": 0.03788497149944305,
      "step": 39500
    },
    {
      "epoch": 11.21203519727505,
      "eval_accuracy": 0.952756406180454,
      "eval_loss": 0.14106468856334686,
      "eval_runtime": 31.4128,
      "eval_samples_per_second": 500.655,
      "eval_steps_per_second": 7.831,
      "step": 39500
    },
    {
      "epoch": 11.214873687198411,
      "grad_norm": 2.1199471950531006,
      "learning_rate": 8.879080329264832e-05,
      "loss": 0.028016036748886107,
      "step": 39510
    },
    {
      "epoch": 11.217712177121772,
      "grad_norm": 7.91051721572876,
      "learning_rate": 8.878796480272495e-05,
      "loss": 0.03719348311424255,
      "step": 39520
    },
    {
      "epoch": 11.220550667045131,
      "grad_norm": 6.829957008361816,
      "learning_rate": 8.878512631280159e-05,
      "loss": 0.041284403204917906,
      "step": 39530
    },
    {
      "epoch": 11.223389156968492,
      "grad_norm": 6.979713439941406,
      "learning_rate": 8.878228782287823e-05,
      "loss": 0.033257138729095456,
      "step": 39540
    },
    {
      "epoch": 11.226227646891854,
      "grad_norm": 2.0938971042633057,
      "learning_rate": 8.877944933295486e-05,
      "loss": 0.04780513346195221,
      "step": 39550
    },
    {
      "epoch": 11.229066136815215,
      "grad_norm": 7.281595706939697,
      "learning_rate": 8.877661084303152e-05,
      "loss": 0.0397993266582489,
      "step": 39560
    },
    {
      "epoch": 11.231904626738576,
      "grad_norm": 3.8252663612365723,
      "learning_rate": 8.877377235310816e-05,
      "loss": 0.03965983390808105,
      "step": 39570
    },
    {
      "epoch": 11.234743116661935,
      "grad_norm": 13.741844177246094,
      "learning_rate": 8.877093386318479e-05,
      "loss": 0.03729905784130096,
      "step": 39580
    },
    {
      "epoch": 11.237581606585296,
      "grad_norm": 2.891458511352539,
      "learning_rate": 8.876809537326143e-05,
      "loss": 0.052103960514068605,
      "step": 39590
    },
    {
      "epoch": 11.240420096508657,
      "grad_norm": 5.4181671142578125,
      "learning_rate": 8.876525688333807e-05,
      "loss": 0.02109394073486328,
      "step": 39600
    },
    {
      "epoch": 11.243258586432018,
      "grad_norm": 7.526214599609375,
      "learning_rate": 8.876241839341471e-05,
      "loss": 0.02676447629928589,
      "step": 39610
    },
    {
      "epoch": 11.24609707635538,
      "grad_norm": 7.1598992347717285,
      "learning_rate": 8.875957990349135e-05,
      "loss": 0.044556647539138794,
      "step": 39620
    },
    {
      "epoch": 11.248935566278739,
      "grad_norm": 6.817886829376221,
      "learning_rate": 8.875674141356799e-05,
      "loss": 0.05024838447570801,
      "step": 39630
    },
    {
      "epoch": 11.2517740562021,
      "grad_norm": 11.896966934204102,
      "learning_rate": 8.875390292364463e-05,
      "loss": 0.03999115824699402,
      "step": 39640
    },
    {
      "epoch": 11.254612546125461,
      "grad_norm": 4.5129714012146,
      "learning_rate": 8.875106443372126e-05,
      "loss": 0.036044588685035704,
      "step": 39650
    },
    {
      "epoch": 11.257451036048822,
      "grad_norm": 4.467544078826904,
      "learning_rate": 8.87482259437979e-05,
      "loss": 0.04666455984115601,
      "step": 39660
    },
    {
      "epoch": 11.260289525972183,
      "grad_norm": 8.925542831420898,
      "learning_rate": 8.874538745387454e-05,
      "loss": 0.03652244806289673,
      "step": 39670
    },
    {
      "epoch": 11.263128015895543,
      "grad_norm": 9.0923433303833,
      "learning_rate": 8.874254896395117e-05,
      "loss": 0.03240195512771606,
      "step": 39680
    },
    {
      "epoch": 11.265966505818904,
      "grad_norm": 6.623198986053467,
      "learning_rate": 8.873971047402783e-05,
      "loss": 0.043272939324378965,
      "step": 39690
    },
    {
      "epoch": 11.268804995742265,
      "grad_norm": 2.8783748149871826,
      "learning_rate": 8.873687198410447e-05,
      "loss": 0.040272146463394165,
      "step": 39700
    },
    {
      "epoch": 11.271643485665626,
      "grad_norm": 6.466062068939209,
      "learning_rate": 8.87340334941811e-05,
      "loss": 0.03759041130542755,
      "step": 39710
    },
    {
      "epoch": 11.274481975588987,
      "grad_norm": 3.5749056339263916,
      "learning_rate": 8.873119500425774e-05,
      "loss": 0.043566498160362246,
      "step": 39720
    },
    {
      "epoch": 11.277320465512348,
      "grad_norm": 7.263559341430664,
      "learning_rate": 8.872835651433438e-05,
      "loss": 0.04249974191188812,
      "step": 39730
    },
    {
      "epoch": 11.280158955435708,
      "grad_norm": 11.27505111694336,
      "learning_rate": 8.872551802441102e-05,
      "loss": 0.037914568185806276,
      "step": 39740
    },
    {
      "epoch": 11.282997445359069,
      "grad_norm": 2.521247625350952,
      "learning_rate": 8.872267953448765e-05,
      "loss": 0.03940242528915405,
      "step": 39750
    },
    {
      "epoch": 11.28583593528243,
      "grad_norm": 1.8530833721160889,
      "learning_rate": 8.87198410445643e-05,
      "loss": 0.03865052461624145,
      "step": 39760
    },
    {
      "epoch": 11.288674425205791,
      "grad_norm": 5.247794151306152,
      "learning_rate": 8.871700255464093e-05,
      "loss": 0.036801236867904666,
      "step": 39770
    },
    {
      "epoch": 11.291512915129152,
      "grad_norm": 13.172710418701172,
      "learning_rate": 8.871416406471757e-05,
      "loss": 0.04191643297672272,
      "step": 39780
    },
    {
      "epoch": 11.294351405052511,
      "grad_norm": 3.5339529514312744,
      "learning_rate": 8.871132557479421e-05,
      "loss": 0.04157689809799194,
      "step": 39790
    },
    {
      "epoch": 11.297189894975872,
      "grad_norm": 2.7554502487182617,
      "learning_rate": 8.870848708487086e-05,
      "loss": 0.04689006805419922,
      "step": 39800
    },
    {
      "epoch": 11.300028384899234,
      "grad_norm": 11.842961311340332,
      "learning_rate": 8.870564859494748e-05,
      "loss": 0.04804868698120117,
      "step": 39810
    },
    {
      "epoch": 11.302866874822595,
      "grad_norm": 2.761676549911499,
      "learning_rate": 8.870281010502414e-05,
      "loss": 0.03384347856044769,
      "step": 39820
    },
    {
      "epoch": 11.305705364745956,
      "grad_norm": 6.327510833740234,
      "learning_rate": 8.869997161510078e-05,
      "loss": 0.042056670784950255,
      "step": 39830
    },
    {
      "epoch": 11.308543854669315,
      "grad_norm": 9.295647621154785,
      "learning_rate": 8.869713312517741e-05,
      "loss": 0.047100627422332765,
      "step": 39840
    },
    {
      "epoch": 11.311382344592676,
      "grad_norm": 5.03857421875,
      "learning_rate": 8.869429463525405e-05,
      "loss": 0.05792197585105896,
      "step": 39850
    },
    {
      "epoch": 11.314220834516037,
      "grad_norm": 5.200779438018799,
      "learning_rate": 8.869145614533069e-05,
      "loss": 0.02069302201271057,
      "step": 39860
    },
    {
      "epoch": 11.317059324439398,
      "grad_norm": 0.8214617967605591,
      "learning_rate": 8.868861765540732e-05,
      "loss": 0.04381552934646606,
      "step": 39870
    },
    {
      "epoch": 11.31989781436276,
      "grad_norm": 13.484682083129883,
      "learning_rate": 8.868577916548396e-05,
      "loss": 0.04238225817680359,
      "step": 39880
    },
    {
      "epoch": 11.32273630428612,
      "grad_norm": 11.78823184967041,
      "learning_rate": 8.868294067556061e-05,
      "loss": 0.041246157884597776,
      "step": 39890
    },
    {
      "epoch": 11.32557479420948,
      "grad_norm": 10.334183692932129,
      "learning_rate": 8.868010218563724e-05,
      "loss": 0.04436834454536438,
      "step": 39900
    },
    {
      "epoch": 11.328413284132841,
      "grad_norm": 8.74250316619873,
      "learning_rate": 8.867726369571388e-05,
      "loss": 0.04008035361766815,
      "step": 39910
    },
    {
      "epoch": 11.331251774056202,
      "grad_norm": 4.13367223739624,
      "learning_rate": 8.867442520579052e-05,
      "loss": 0.05199337601661682,
      "step": 39920
    },
    {
      "epoch": 11.334090263979563,
      "grad_norm": 17.059772491455078,
      "learning_rate": 8.867158671586717e-05,
      "loss": 0.04543254673480988,
      "step": 39930
    },
    {
      "epoch": 11.336928753902924,
      "grad_norm": 6.4024553298950195,
      "learning_rate": 8.86687482259438e-05,
      "loss": 0.039739423990249635,
      "step": 39940
    },
    {
      "epoch": 11.339767243826284,
      "grad_norm": 5.212245464324951,
      "learning_rate": 8.866590973602044e-05,
      "loss": 0.04636736512184143,
      "step": 39950
    },
    {
      "epoch": 11.342605733749645,
      "grad_norm": 17.039974212646484,
      "learning_rate": 8.866307124609709e-05,
      "loss": 0.06416319012641906,
      "step": 39960
    },
    {
      "epoch": 11.345444223673006,
      "grad_norm": 2.626824378967285,
      "learning_rate": 8.866023275617372e-05,
      "loss": 0.04492620229721069,
      "step": 39970
    },
    {
      "epoch": 11.348282713596367,
      "grad_norm": 7.594601154327393,
      "learning_rate": 8.865739426625036e-05,
      "loss": 0.028285348415374757,
      "step": 39980
    },
    {
      "epoch": 11.351121203519728,
      "grad_norm": 2.9567835330963135,
      "learning_rate": 8.8654555776327e-05,
      "loss": 0.05564141273498535,
      "step": 39990
    },
    {
      "epoch": 11.353959693443088,
      "grad_norm": 4.25455379486084,
      "learning_rate": 8.865171728640363e-05,
      "loss": 0.07398279905319213,
      "step": 40000
    },
    {
      "epoch": 11.353959693443088,
      "eval_accuracy": 0.9532015006040567,
      "eval_loss": 0.13763917982578278,
      "eval_runtime": 31.1135,
      "eval_samples_per_second": 505.472,
      "eval_steps_per_second": 7.907,
      "step": 40000
    },
    {
      "epoch": 11.356798183366449,
      "grad_norm": 4.23692512512207,
      "learning_rate": 8.864887879648027e-05,
      "loss": 0.03264350891113281,
      "step": 40010
    },
    {
      "epoch": 11.35963667328981,
      "grad_norm": 8.874746322631836,
      "learning_rate": 8.864604030655692e-05,
      "loss": 0.05038106441497803,
      "step": 40020
    },
    {
      "epoch": 11.36247516321317,
      "grad_norm": 1.852181077003479,
      "learning_rate": 8.864320181663355e-05,
      "loss": 0.030090701580047608,
      "step": 40030
    },
    {
      "epoch": 11.365313653136532,
      "grad_norm": 3.475825309753418,
      "learning_rate": 8.86403633267102e-05,
      "loss": 0.059383833408355714,
      "step": 40040
    },
    {
      "epoch": 11.368152143059891,
      "grad_norm": 8.16132926940918,
      "learning_rate": 8.863752483678684e-05,
      "loss": 0.03293721675872803,
      "step": 40050
    },
    {
      "epoch": 11.370990632983252,
      "grad_norm": 13.255269050598145,
      "learning_rate": 8.863468634686348e-05,
      "loss": 0.05026310086250305,
      "step": 40060
    },
    {
      "epoch": 11.373829122906614,
      "grad_norm": 10.392302513122559,
      "learning_rate": 8.86318478569401e-05,
      "loss": 0.04383312463760376,
      "step": 40070
    },
    {
      "epoch": 11.376667612829975,
      "grad_norm": 8.054374694824219,
      "learning_rate": 8.862900936701675e-05,
      "loss": 0.028009480237960814,
      "step": 40080
    },
    {
      "epoch": 11.379506102753336,
      "grad_norm": 7.8724541664123535,
      "learning_rate": 8.86261708770934e-05,
      "loss": 0.04316805005073547,
      "step": 40090
    },
    {
      "epoch": 11.382344592676697,
      "grad_norm": 4.537086486816406,
      "learning_rate": 8.862333238717003e-05,
      "loss": 0.030184608697891236,
      "step": 40100
    },
    {
      "epoch": 11.385183082600056,
      "grad_norm": 4.153220176696777,
      "learning_rate": 8.862049389724667e-05,
      "loss": 0.03144936561584473,
      "step": 40110
    },
    {
      "epoch": 11.388021572523417,
      "grad_norm": 2.012098550796509,
      "learning_rate": 8.861765540732331e-05,
      "loss": 0.03236268162727356,
      "step": 40120
    },
    {
      "epoch": 11.390860062446778,
      "grad_norm": 5.7497758865356445,
      "learning_rate": 8.861481691739994e-05,
      "loss": 0.02789943218231201,
      "step": 40130
    },
    {
      "epoch": 11.39369855237014,
      "grad_norm": 14.024553298950195,
      "learning_rate": 8.861197842747658e-05,
      "loss": 0.02832912802696228,
      "step": 40140
    },
    {
      "epoch": 11.3965370422935,
      "grad_norm": 8.662565231323242,
      "learning_rate": 8.860913993755322e-05,
      "loss": 0.035794562101364134,
      "step": 40150
    },
    {
      "epoch": 11.39937553221686,
      "grad_norm": 13.299927711486816,
      "learning_rate": 8.860630144762986e-05,
      "loss": 0.051793599128723146,
      "step": 40160
    },
    {
      "epoch": 11.402214022140221,
      "grad_norm": 3.974324941635132,
      "learning_rate": 8.86034629577065e-05,
      "loss": 0.03746732771396637,
      "step": 40170
    },
    {
      "epoch": 11.405052512063582,
      "grad_norm": 4.349684715270996,
      "learning_rate": 8.860062446778315e-05,
      "loss": 0.0361139714717865,
      "step": 40180
    },
    {
      "epoch": 11.407891001986943,
      "grad_norm": 6.070065021514893,
      "learning_rate": 8.859778597785979e-05,
      "loss": 0.034117507934570315,
      "step": 40190
    },
    {
      "epoch": 11.410729491910304,
      "grad_norm": 11.619451522827148,
      "learning_rate": 8.859494748793642e-05,
      "loss": 0.046603921055793765,
      "step": 40200
    },
    {
      "epoch": 11.413567981833664,
      "grad_norm": 5.385000705718994,
      "learning_rate": 8.859210899801306e-05,
      "loss": 0.04721275269985199,
      "step": 40210
    },
    {
      "epoch": 11.416406471757025,
      "grad_norm": 13.089354515075684,
      "learning_rate": 8.858927050808971e-05,
      "loss": 0.05168584585189819,
      "step": 40220
    },
    {
      "epoch": 11.419244961680386,
      "grad_norm": 12.567547798156738,
      "learning_rate": 8.858643201816634e-05,
      "loss": 0.04480060338973999,
      "step": 40230
    },
    {
      "epoch": 11.422083451603747,
      "grad_norm": 12.628218650817871,
      "learning_rate": 8.858359352824298e-05,
      "loss": 0.07853772640228271,
      "step": 40240
    },
    {
      "epoch": 11.424921941527108,
      "grad_norm": 9.429924964904785,
      "learning_rate": 8.858075503831962e-05,
      "loss": 0.0535646915435791,
      "step": 40250
    },
    {
      "epoch": 11.427760431450467,
      "grad_norm": 1.943774700164795,
      "learning_rate": 8.857791654839625e-05,
      "loss": 0.018581156432628632,
      "step": 40260
    },
    {
      "epoch": 11.430598921373829,
      "grad_norm": 5.487278938293457,
      "learning_rate": 8.857507805847289e-05,
      "loss": 0.05353713035583496,
      "step": 40270
    },
    {
      "epoch": 11.43343741129719,
      "grad_norm": 11.646852493286133,
      "learning_rate": 8.857223956854953e-05,
      "loss": 0.027602463960647583,
      "step": 40280
    },
    {
      "epoch": 11.43627590122055,
      "grad_norm": 11.33411693572998,
      "learning_rate": 8.856940107862617e-05,
      "loss": 0.04334660172462464,
      "step": 40290
    },
    {
      "epoch": 11.439114391143912,
      "grad_norm": 12.542768478393555,
      "learning_rate": 8.856656258870282e-05,
      "loss": 0.04223451018333435,
      "step": 40300
    },
    {
      "epoch": 11.441952881067273,
      "grad_norm": 6.214544296264648,
      "learning_rate": 8.856372409877946e-05,
      "loss": 0.04474785327911377,
      "step": 40310
    },
    {
      "epoch": 11.444791370990632,
      "grad_norm": 8.552118301391602,
      "learning_rate": 8.85608856088561e-05,
      "loss": 0.03829832077026367,
      "step": 40320
    },
    {
      "epoch": 11.447629860913993,
      "grad_norm": 3.506162405014038,
      "learning_rate": 8.855804711893273e-05,
      "loss": 0.03281140923500061,
      "step": 40330
    },
    {
      "epoch": 11.450468350837355,
      "grad_norm": 6.62778902053833,
      "learning_rate": 8.855520862900937e-05,
      "loss": 0.049500495195388794,
      "step": 40340
    },
    {
      "epoch": 11.453306840760716,
      "grad_norm": 5.414072036743164,
      "learning_rate": 8.855237013908601e-05,
      "loss": 0.03758763372898102,
      "step": 40350
    },
    {
      "epoch": 11.456145330684077,
      "grad_norm": 9.07391357421875,
      "learning_rate": 8.854953164916265e-05,
      "loss": 0.0444775402545929,
      "step": 40360
    },
    {
      "epoch": 11.458983820607436,
      "grad_norm": 5.526808738708496,
      "learning_rate": 8.854669315923929e-05,
      "loss": 0.03314472734928131,
      "step": 40370
    },
    {
      "epoch": 11.461822310530797,
      "grad_norm": 4.515206813812256,
      "learning_rate": 8.854385466931593e-05,
      "loss": 0.0357508659362793,
      "step": 40380
    },
    {
      "epoch": 11.464660800454158,
      "grad_norm": 17.794885635375977,
      "learning_rate": 8.854101617939256e-05,
      "loss": 0.040084469318389895,
      "step": 40390
    },
    {
      "epoch": 11.46749929037752,
      "grad_norm": 7.860543727874756,
      "learning_rate": 8.85381776894692e-05,
      "loss": 0.03683890998363495,
      "step": 40400
    },
    {
      "epoch": 11.47033778030088,
      "grad_norm": 2.1493349075317383,
      "learning_rate": 8.853533919954584e-05,
      "loss": 0.030355399847030638,
      "step": 40410
    },
    {
      "epoch": 11.47317627022424,
      "grad_norm": 2.1995296478271484,
      "learning_rate": 8.853250070962249e-05,
      "loss": 0.042979204654693605,
      "step": 40420
    },
    {
      "epoch": 11.476014760147601,
      "grad_norm": 12.870260238647461,
      "learning_rate": 8.852966221969913e-05,
      "loss": 0.053621649742126465,
      "step": 40430
    },
    {
      "epoch": 11.478853250070962,
      "grad_norm": 5.2994537353515625,
      "learning_rate": 8.852682372977577e-05,
      "loss": 0.042656701803207395,
      "step": 40440
    },
    {
      "epoch": 11.481691739994323,
      "grad_norm": 6.095134258270264,
      "learning_rate": 8.852398523985241e-05,
      "loss": 0.05943422317504883,
      "step": 40450
    },
    {
      "epoch": 11.484530229917684,
      "grad_norm": 10.917078971862793,
      "learning_rate": 8.852114674992904e-05,
      "loss": 0.04059193134307861,
      "step": 40460
    },
    {
      "epoch": 11.487368719841044,
      "grad_norm": 2.4780454635620117,
      "learning_rate": 8.851830826000568e-05,
      "loss": 0.03488376438617706,
      "step": 40470
    },
    {
      "epoch": 11.490207209764405,
      "grad_norm": 4.229562759399414,
      "learning_rate": 8.851546977008232e-05,
      "loss": 0.04805220365524292,
      "step": 40480
    },
    {
      "epoch": 11.493045699687766,
      "grad_norm": 7.648486137390137,
      "learning_rate": 8.851263128015896e-05,
      "loss": 0.037777215242385864,
      "step": 40490
    },
    {
      "epoch": 11.495884189611127,
      "grad_norm": 5.515402793884277,
      "learning_rate": 8.85097927902356e-05,
      "loss": 0.04882913529872894,
      "step": 40500
    },
    {
      "epoch": 11.495884189611127,
      "eval_accuracy": 0.9565715012399059,
      "eval_loss": 0.13253000378608704,
      "eval_runtime": 31.5222,
      "eval_samples_per_second": 498.918,
      "eval_steps_per_second": 7.804,
      "step": 40500
    },
    {
      "epoch": 11.498722679534488,
      "grad_norm": 12.569120407104492,
      "learning_rate": 8.850723814930457e-05,
      "loss": 0.06314729452133179,
      "step": 40510
    },
    {
      "epoch": 11.50156116945785,
      "grad_norm": 8.53803539276123,
      "learning_rate": 8.850439965938121e-05,
      "loss": 0.03850083947181702,
      "step": 40520
    },
    {
      "epoch": 11.504399659381209,
      "grad_norm": 5.943691730499268,
      "learning_rate": 8.850156116945785e-05,
      "loss": 0.05651187300682068,
      "step": 40530
    },
    {
      "epoch": 11.50723814930457,
      "grad_norm": 1.117270827293396,
      "learning_rate": 8.849872267953449e-05,
      "loss": 0.02911505401134491,
      "step": 40540
    },
    {
      "epoch": 11.51007663922793,
      "grad_norm": 0.5039180517196655,
      "learning_rate": 8.849588418961113e-05,
      "loss": 0.039834386110305785,
      "step": 40550
    },
    {
      "epoch": 11.512915129151292,
      "grad_norm": 1.643454909324646,
      "learning_rate": 8.849304569968778e-05,
      "loss": 0.04547125399112702,
      "step": 40560
    },
    {
      "epoch": 11.515753619074653,
      "grad_norm": 3.3448643684387207,
      "learning_rate": 8.84902072097644e-05,
      "loss": 0.03437337875366211,
      "step": 40570
    },
    {
      "epoch": 11.518592108998012,
      "grad_norm": 6.541136264801025,
      "learning_rate": 8.848736871984105e-05,
      "loss": 0.06071277260780335,
      "step": 40580
    },
    {
      "epoch": 11.521430598921373,
      "grad_norm": 7.1211466789245605,
      "learning_rate": 8.848453022991769e-05,
      "loss": 0.032107898592948915,
      "step": 40590
    },
    {
      "epoch": 11.524269088844735,
      "grad_norm": 10.989923477172852,
      "learning_rate": 8.848169173999433e-05,
      "loss": 0.0668204665184021,
      "step": 40600
    },
    {
      "epoch": 11.527107578768096,
      "grad_norm": 1.7694687843322754,
      "learning_rate": 8.847885325007097e-05,
      "loss": 0.07757980227470399,
      "step": 40610
    },
    {
      "epoch": 11.529946068691457,
      "grad_norm": 13.304360389709473,
      "learning_rate": 8.847601476014761e-05,
      "loss": 0.06403954029083252,
      "step": 40620
    },
    {
      "epoch": 11.532784558614816,
      "grad_norm": 8.41611099243164,
      "learning_rate": 8.847317627022424e-05,
      "loss": 0.07346206307411193,
      "step": 40630
    },
    {
      "epoch": 11.535623048538177,
      "grad_norm": 19.83196258544922,
      "learning_rate": 8.847033778030088e-05,
      "loss": 0.06302956342697144,
      "step": 40640
    },
    {
      "epoch": 11.538461538461538,
      "grad_norm": 1.5962491035461426,
      "learning_rate": 8.846749929037752e-05,
      "loss": 0.0424712061882019,
      "step": 40650
    },
    {
      "epoch": 11.5413000283849,
      "grad_norm": 4.162332057952881,
      "learning_rate": 8.846466080045416e-05,
      "loss": 0.05826054215431213,
      "step": 40660
    },
    {
      "epoch": 11.54413851830826,
      "grad_norm": 0.9698114991188049,
      "learning_rate": 8.84618223105308e-05,
      "loss": 0.04148352444171906,
      "step": 40670
    },
    {
      "epoch": 11.546977008231622,
      "grad_norm": 7.476339817047119,
      "learning_rate": 8.845898382060745e-05,
      "loss": 0.027741429209709168,
      "step": 40680
    },
    {
      "epoch": 11.549815498154981,
      "grad_norm": 4.600419998168945,
      "learning_rate": 8.845614533068409e-05,
      "loss": 0.02867766618728638,
      "step": 40690
    },
    {
      "epoch": 11.552653988078342,
      "grad_norm": 6.975192546844482,
      "learning_rate": 8.845330684076071e-05,
      "loss": 0.050774085521698,
      "step": 40700
    },
    {
      "epoch": 11.555492478001703,
      "grad_norm": 3.077288866043091,
      "learning_rate": 8.845046835083736e-05,
      "loss": 0.03793132603168488,
      "step": 40710
    },
    {
      "epoch": 11.558330967925064,
      "grad_norm": 12.623274803161621,
      "learning_rate": 8.8447629860914e-05,
      "loss": 0.049743345379829405,
      "step": 40720
    },
    {
      "epoch": 11.561169457848425,
      "grad_norm": 15.091360092163086,
      "learning_rate": 8.844479137099062e-05,
      "loss": 0.07584220170974731,
      "step": 40730
    },
    {
      "epoch": 11.564007947771785,
      "grad_norm": 10.169875144958496,
      "learning_rate": 8.844195288106728e-05,
      "loss": 0.04955956935882568,
      "step": 40740
    },
    {
      "epoch": 11.566846437695146,
      "grad_norm": 5.2755303382873535,
      "learning_rate": 8.843911439114392e-05,
      "loss": 0.034739372134208676,
      "step": 40750
    },
    {
      "epoch": 11.569684927618507,
      "grad_norm": 8.641897201538086,
      "learning_rate": 8.843627590122055e-05,
      "loss": 0.052517431974411014,
      "step": 40760
    },
    {
      "epoch": 11.572523417541868,
      "grad_norm": 4.512448310852051,
      "learning_rate": 8.843343741129719e-05,
      "loss": 0.061132752895355226,
      "step": 40770
    },
    {
      "epoch": 11.57536190746523,
      "grad_norm": 4.899333953857422,
      "learning_rate": 8.843059892137383e-05,
      "loss": 0.04364375174045563,
      "step": 40780
    },
    {
      "epoch": 11.578200397388589,
      "grad_norm": 10.793277740478516,
      "learning_rate": 8.842776043145047e-05,
      "loss": 0.048547714948654175,
      "step": 40790
    },
    {
      "epoch": 11.58103888731195,
      "grad_norm": 6.0333781242370605,
      "learning_rate": 8.842492194152711e-05,
      "loss": 0.032700303196907046,
      "step": 40800
    },
    {
      "epoch": 11.58387737723531,
      "grad_norm": 12.72266674041748,
      "learning_rate": 8.842208345160376e-05,
      "loss": 0.05314317345619202,
      "step": 40810
    },
    {
      "epoch": 11.586715867158672,
      "grad_norm": 1.3179000616073608,
      "learning_rate": 8.84192449616804e-05,
      "loss": 0.048350965976715087,
      "step": 40820
    },
    {
      "epoch": 11.589554357082033,
      "grad_norm": 6.280195713043213,
      "learning_rate": 8.841640647175703e-05,
      "loss": 0.04212665855884552,
      "step": 40830
    },
    {
      "epoch": 11.592392847005392,
      "grad_norm": 1.0754423141479492,
      "learning_rate": 8.841356798183367e-05,
      "loss": 0.05024091601371765,
      "step": 40840
    },
    {
      "epoch": 11.595231336928753,
      "grad_norm": 8.706792831420898,
      "learning_rate": 8.841072949191031e-05,
      "loss": 0.048531919717788696,
      "step": 40850
    },
    {
      "epoch": 11.598069826852115,
      "grad_norm": 11.477270126342773,
      "learning_rate": 8.840789100198694e-05,
      "loss": 0.0477634072303772,
      "step": 40860
    },
    {
      "epoch": 11.600908316775476,
      "grad_norm": 3.2627789974212646,
      "learning_rate": 8.840505251206359e-05,
      "loss": 0.057295024394989014,
      "step": 40870
    },
    {
      "epoch": 11.603746806698837,
      "grad_norm": 8.127674102783203,
      "learning_rate": 8.840221402214023e-05,
      "loss": 0.049380600452423096,
      "step": 40880
    },
    {
      "epoch": 11.606585296622196,
      "grad_norm": 11.47251033782959,
      "learning_rate": 8.839937553221686e-05,
      "loss": 0.052211332321166995,
      "step": 40890
    },
    {
      "epoch": 11.609423786545557,
      "grad_norm": 1.2437288761138916,
      "learning_rate": 8.83965370422935e-05,
      "loss": 0.0263483464717865,
      "step": 40900
    },
    {
      "epoch": 11.612262276468918,
      "grad_norm": 13.604060173034668,
      "learning_rate": 8.839369855237014e-05,
      "loss": 0.057996582984924314,
      "step": 40910
    },
    {
      "epoch": 11.61510076639228,
      "grad_norm": 11.502900123596191,
      "learning_rate": 8.839086006244678e-05,
      "loss": 0.03979346752166748,
      "step": 40920
    },
    {
      "epoch": 11.61793925631564,
      "grad_norm": 8.272281646728516,
      "learning_rate": 8.838802157252343e-05,
      "loss": 0.06818183660507202,
      "step": 40930
    },
    {
      "epoch": 11.620777746239002,
      "grad_norm": 12.573554039001465,
      "learning_rate": 8.838518308260007e-05,
      "loss": 0.04096316695213318,
      "step": 40940
    },
    {
      "epoch": 11.623616236162361,
      "grad_norm": 12.9218111038208,
      "learning_rate": 8.838234459267671e-05,
      "loss": 0.06615672111511231,
      "step": 40950
    },
    {
      "epoch": 11.626454726085722,
      "grad_norm": 12.730849266052246,
      "learning_rate": 8.837950610275334e-05,
      "loss": 0.06989012360572815,
      "step": 40960
    },
    {
      "epoch": 11.629293216009083,
      "grad_norm": 9.000524520874023,
      "learning_rate": 8.837666761282998e-05,
      "loss": 0.09680927991867065,
      "step": 40970
    },
    {
      "epoch": 11.632131705932444,
      "grad_norm": 5.213916778564453,
      "learning_rate": 8.837382912290662e-05,
      "loss": 0.061992645263671875,
      "step": 40980
    },
    {
      "epoch": 11.634970195855805,
      "grad_norm": 7.030336856842041,
      "learning_rate": 8.837099063298325e-05,
      "loss": 0.05134835243225098,
      "step": 40990
    },
    {
      "epoch": 11.637808685779165,
      "grad_norm": 1.3521445989608765,
      "learning_rate": 8.83681521430599e-05,
      "loss": 0.03942259848117828,
      "step": 41000
    },
    {
      "epoch": 11.637808685779165,
      "eval_accuracy": 0.9580975392636867,
      "eval_loss": 0.13014930486679077,
      "eval_runtime": 31.3658,
      "eval_samples_per_second": 501.407,
      "eval_steps_per_second": 7.843,
      "step": 41000
    },
    {
      "epoch": 11.640647175702526,
      "grad_norm": 3.281205415725708,
      "learning_rate": 8.836531365313654e-05,
      "loss": 0.03922618329524994,
      "step": 41010
    },
    {
      "epoch": 11.643485665625887,
      "grad_norm": 7.914754867553711,
      "learning_rate": 8.836247516321317e-05,
      "loss": 0.05260990858078003,
      "step": 41020
    },
    {
      "epoch": 11.646324155549248,
      "grad_norm": 3.968221664428711,
      "learning_rate": 8.835963667328981e-05,
      "loss": 0.0357495129108429,
      "step": 41030
    },
    {
      "epoch": 11.64916264547261,
      "grad_norm": 1.4710693359375,
      "learning_rate": 8.835679818336645e-05,
      "loss": 0.03923320174217224,
      "step": 41040
    },
    {
      "epoch": 11.65200113539597,
      "grad_norm": 4.021718978881836,
      "learning_rate": 8.83539596934431e-05,
      "loss": 0.03728710114955902,
      "step": 41050
    },
    {
      "epoch": 11.65483962531933,
      "grad_norm": 9.911172866821289,
      "learning_rate": 8.835112120351972e-05,
      "loss": 0.04307613372802734,
      "step": 41060
    },
    {
      "epoch": 11.65767811524269,
      "grad_norm": 3.696253538131714,
      "learning_rate": 8.834828271359638e-05,
      "loss": 0.04925021827220917,
      "step": 41070
    },
    {
      "epoch": 11.660516605166052,
      "grad_norm": 2.0492725372314453,
      "learning_rate": 8.834544422367302e-05,
      "loss": 0.041888803243637085,
      "step": 41080
    },
    {
      "epoch": 11.663355095089413,
      "grad_norm": 1.6307481527328491,
      "learning_rate": 8.834260573374965e-05,
      "loss": 0.042088651657104494,
      "step": 41090
    },
    {
      "epoch": 11.666193585012774,
      "grad_norm": 2.7641148567199707,
      "learning_rate": 8.833976724382629e-05,
      "loss": 0.06680274605751038,
      "step": 41100
    },
    {
      "epoch": 11.669032074936133,
      "grad_norm": 4.430089950561523,
      "learning_rate": 8.833692875390293e-05,
      "loss": 0.039932504296302795,
      "step": 41110
    },
    {
      "epoch": 11.671870564859494,
      "grad_norm": 5.572601318359375,
      "learning_rate": 8.833409026397956e-05,
      "loss": 0.0400811493396759,
      "step": 41120
    },
    {
      "epoch": 11.674709054782856,
      "grad_norm": 3.174543857574463,
      "learning_rate": 8.833125177405621e-05,
      "loss": 0.05859325528144836,
      "step": 41130
    },
    {
      "epoch": 11.677547544706217,
      "grad_norm": 0.4484679400920868,
      "learning_rate": 8.832841328413285e-05,
      "loss": 0.024863974750041963,
      "step": 41140
    },
    {
      "epoch": 11.680386034629578,
      "grad_norm": 8.200668334960938,
      "learning_rate": 8.832557479420948e-05,
      "loss": 0.04697154462337494,
      "step": 41150
    },
    {
      "epoch": 11.683224524552937,
      "grad_norm": 15.55133056640625,
      "learning_rate": 8.832273630428612e-05,
      "loss": 0.04557651579380036,
      "step": 41160
    },
    {
      "epoch": 11.686063014476298,
      "grad_norm": 3.048593521118164,
      "learning_rate": 8.831989781436276e-05,
      "loss": 0.04560634195804596,
      "step": 41170
    },
    {
      "epoch": 11.68890150439966,
      "grad_norm": 3.9336376190185547,
      "learning_rate": 8.83170593244394e-05,
      "loss": 0.06830593347549438,
      "step": 41180
    },
    {
      "epoch": 11.69173999432302,
      "grad_norm": 8.1607027053833,
      "learning_rate": 8.831422083451603e-05,
      "loss": 0.029866716265678404,
      "step": 41190
    },
    {
      "epoch": 11.694578484246382,
      "grad_norm": 1.0305315256118774,
      "learning_rate": 8.831138234459269e-05,
      "loss": 0.04140089750289917,
      "step": 41200
    },
    {
      "epoch": 11.697416974169741,
      "grad_norm": 2.2033443450927734,
      "learning_rate": 8.830854385466933e-05,
      "loss": 0.044012248516082764,
      "step": 41210
    },
    {
      "epoch": 11.700255464093102,
      "grad_norm": 8.77370548248291,
      "learning_rate": 8.830570536474596e-05,
      "loss": 0.045134469866752625,
      "step": 41220
    },
    {
      "epoch": 11.703093954016463,
      "grad_norm": 12.965953826904297,
      "learning_rate": 8.83028668748226e-05,
      "loss": 0.0631374478340149,
      "step": 41230
    },
    {
      "epoch": 11.705932443939824,
      "grad_norm": 2.009962797164917,
      "learning_rate": 8.830002838489924e-05,
      "loss": 0.03572629988193512,
      "step": 41240
    },
    {
      "epoch": 11.708770933863185,
      "grad_norm": 7.408824920654297,
      "learning_rate": 8.829718989497587e-05,
      "loss": 0.04673230350017547,
      "step": 41250
    },
    {
      "epoch": 11.711609423786545,
      "grad_norm": 3.6747851371765137,
      "learning_rate": 8.829435140505251e-05,
      "loss": 0.050879156589508055,
      "step": 41260
    },
    {
      "epoch": 11.714447913709906,
      "grad_norm": 4.7804059982299805,
      "learning_rate": 8.829151291512916e-05,
      "loss": 0.06859890222549439,
      "step": 41270
    },
    {
      "epoch": 11.717286403633267,
      "grad_norm": 12.197978973388672,
      "learning_rate": 8.828867442520579e-05,
      "loss": 0.06282455325126649,
      "step": 41280
    },
    {
      "epoch": 11.720124893556628,
      "grad_norm": 4.496255874633789,
      "learning_rate": 8.828583593528243e-05,
      "loss": 0.06147288680076599,
      "step": 41290
    },
    {
      "epoch": 11.72296338347999,
      "grad_norm": 6.720599174499512,
      "learning_rate": 8.828299744535907e-05,
      "loss": 0.041345641016960144,
      "step": 41300
    },
    {
      "epoch": 11.72580187340335,
      "grad_norm": 9.309507369995117,
      "learning_rate": 8.828015895543572e-05,
      "loss": 0.04281518161296845,
      "step": 41310
    },
    {
      "epoch": 11.72864036332671,
      "grad_norm": 11.473611831665039,
      "learning_rate": 8.827732046551234e-05,
      "loss": 0.04009150862693787,
      "step": 41320
    },
    {
      "epoch": 11.73147885325007,
      "grad_norm": 14.777087211608887,
      "learning_rate": 8.8274481975589e-05,
      "loss": 0.04862431287765503,
      "step": 41330
    },
    {
      "epoch": 11.734317343173432,
      "grad_norm": 5.718709468841553,
      "learning_rate": 8.827164348566564e-05,
      "loss": 0.03550182282924652,
      "step": 41340
    },
    {
      "epoch": 11.737155833096793,
      "grad_norm": 9.201991081237793,
      "learning_rate": 8.826880499574227e-05,
      "loss": 0.03538917601108551,
      "step": 41350
    },
    {
      "epoch": 11.739994323020154,
      "grad_norm": 3.1091489791870117,
      "learning_rate": 8.826596650581891e-05,
      "loss": 0.02286338061094284,
      "step": 41360
    },
    {
      "epoch": 11.742832812943513,
      "grad_norm": 5.587309837341309,
      "learning_rate": 8.826312801589555e-05,
      "loss": 0.04504280686378479,
      "step": 41370
    },
    {
      "epoch": 11.745671302866874,
      "grad_norm": 6.159370422363281,
      "learning_rate": 8.826028952597218e-05,
      "loss": 0.035818624496459964,
      "step": 41380
    },
    {
      "epoch": 11.748509792790236,
      "grad_norm": 6.3836517333984375,
      "learning_rate": 8.825745103604882e-05,
      "loss": 0.029144224524497987,
      "step": 41390
    },
    {
      "epoch": 11.751348282713597,
      "grad_norm": 3.7704317569732666,
      "learning_rate": 8.825461254612548e-05,
      "loss": 0.04859644770622253,
      "step": 41400
    },
    {
      "epoch": 11.754186772636958,
      "grad_norm": 5.243749618530273,
      "learning_rate": 8.82517740562021e-05,
      "loss": 0.05567106008529663,
      "step": 41410
    },
    {
      "epoch": 11.757025262560317,
      "grad_norm": 5.640360355377197,
      "learning_rate": 8.824893556627874e-05,
      "loss": 0.03157995641231537,
      "step": 41420
    },
    {
      "epoch": 11.759863752483678,
      "grad_norm": 5.861602306365967,
      "learning_rate": 8.824609707635539e-05,
      "loss": 0.052859604358673096,
      "step": 41430
    },
    {
      "epoch": 11.76270224240704,
      "grad_norm": 2.414142608642578,
      "learning_rate": 8.824325858643203e-05,
      "loss": 0.032310235500335696,
      "step": 41440
    },
    {
      "epoch": 11.7655407323304,
      "grad_norm": 3.531975030899048,
      "learning_rate": 8.824042009650865e-05,
      "loss": 0.04497153759002685,
      "step": 41450
    },
    {
      "epoch": 11.768379222253762,
      "grad_norm": 12.987493515014648,
      "learning_rate": 8.82375816065853e-05,
      "loss": 0.039786535501480105,
      "step": 41460
    },
    {
      "epoch": 11.771217712177123,
      "grad_norm": 5.657522678375244,
      "learning_rate": 8.823474311666194e-05,
      "loss": 0.04915342330932617,
      "step": 41470
    },
    {
      "epoch": 11.774056202100482,
      "grad_norm": 4.03167200088501,
      "learning_rate": 8.823190462673858e-05,
      "loss": 0.04671434760093689,
      "step": 41480
    },
    {
      "epoch": 11.776894692023843,
      "grad_norm": 4.126925945281982,
      "learning_rate": 8.822906613681522e-05,
      "loss": 0.041422408819198606,
      "step": 41490
    },
    {
      "epoch": 11.779733181947204,
      "grad_norm": 6.271815776824951,
      "learning_rate": 8.822622764689186e-05,
      "loss": 0.03411697447299957,
      "step": 41500
    },
    {
      "epoch": 11.779733181947204,
      "eval_accuracy": 0.9564443314045908,
      "eval_loss": 0.1370164006948471,
      "eval_runtime": 31.626,
      "eval_samples_per_second": 497.28,
      "eval_steps_per_second": 7.778,
      "step": 41500
    },
    {
      "epoch": 11.782571671870565,
      "grad_norm": 10.734105110168457,
      "learning_rate": 8.822338915696849e-05,
      "loss": 0.04735274910926819,
      "step": 41510
    },
    {
      "epoch": 11.785410161793926,
      "grad_norm": 7.924331188201904,
      "learning_rate": 8.822055066704513e-05,
      "loss": 0.03922942280769348,
      "step": 41520
    },
    {
      "epoch": 11.788248651717286,
      "grad_norm": 4.027884006500244,
      "learning_rate": 8.821771217712179e-05,
      "loss": 0.026624178886413573,
      "step": 41530
    },
    {
      "epoch": 11.791087141640647,
      "grad_norm": 8.639642715454102,
      "learning_rate": 8.821487368719841e-05,
      "loss": 0.03481380939483643,
      "step": 41540
    },
    {
      "epoch": 11.793925631564008,
      "grad_norm": 1.7185767889022827,
      "learning_rate": 8.821203519727506e-05,
      "loss": 0.03559336364269257,
      "step": 41550
    },
    {
      "epoch": 11.796764121487369,
      "grad_norm": 3.059953451156616,
      "learning_rate": 8.82091967073517e-05,
      "loss": 0.031420886516571045,
      "step": 41560
    },
    {
      "epoch": 11.79960261141073,
      "grad_norm": 11.784826278686523,
      "learning_rate": 8.820635821742832e-05,
      "loss": 0.05998876690864563,
      "step": 41570
    },
    {
      "epoch": 11.80244110133409,
      "grad_norm": 13.273045539855957,
      "learning_rate": 8.820351972750497e-05,
      "loss": 0.04332319796085358,
      "step": 41580
    },
    {
      "epoch": 11.80527959125745,
      "grad_norm": 7.318718910217285,
      "learning_rate": 8.820068123758161e-05,
      "loss": 0.06102036833763123,
      "step": 41590
    },
    {
      "epoch": 11.808118081180812,
      "grad_norm": 5.058545112609863,
      "learning_rate": 8.819784274765825e-05,
      "loss": 0.04044685959815979,
      "step": 41600
    },
    {
      "epoch": 11.810956571104173,
      "grad_norm": 7.207066059112549,
      "learning_rate": 8.819500425773489e-05,
      "loss": 0.03486702144145966,
      "step": 41610
    },
    {
      "epoch": 11.813795061027534,
      "grad_norm": 4.909163475036621,
      "learning_rate": 8.819216576781153e-05,
      "loss": 0.03199976980686188,
      "step": 41620
    },
    {
      "epoch": 11.816633550950893,
      "grad_norm": 2.056788921356201,
      "learning_rate": 8.818932727788817e-05,
      "loss": 0.03631947040557861,
      "step": 41630
    },
    {
      "epoch": 11.819472040874254,
      "grad_norm": 3.98041033744812,
      "learning_rate": 8.81864887879648e-05,
      "loss": 0.049060821533203125,
      "step": 41640
    },
    {
      "epoch": 11.822310530797616,
      "grad_norm": 3.0290110111236572,
      "learning_rate": 8.818365029804144e-05,
      "loss": 0.03193089067935943,
      "step": 41650
    },
    {
      "epoch": 11.825149020720977,
      "grad_norm": 2.7639479637145996,
      "learning_rate": 8.818081180811808e-05,
      "loss": 0.03945748507976532,
      "step": 41660
    },
    {
      "epoch": 11.827987510644338,
      "grad_norm": 12.770966529846191,
      "learning_rate": 8.817797331819472e-05,
      "loss": 0.04247860610485077,
      "step": 41670
    },
    {
      "epoch": 11.830826000567697,
      "grad_norm": 12.497370719909668,
      "learning_rate": 8.817513482827137e-05,
      "loss": 0.05721154808998108,
      "step": 41680
    },
    {
      "epoch": 11.833664490491058,
      "grad_norm": 5.532436370849609,
      "learning_rate": 8.817229633834801e-05,
      "loss": 0.04636166989803314,
      "step": 41690
    },
    {
      "epoch": 11.83650298041442,
      "grad_norm": 6.073328971862793,
      "learning_rate": 8.816945784842464e-05,
      "loss": 0.03356616199016571,
      "step": 41700
    },
    {
      "epoch": 11.83934147033778,
      "grad_norm": 4.38300085067749,
      "learning_rate": 8.816661935850128e-05,
      "loss": 0.053462475538253784,
      "step": 41710
    },
    {
      "epoch": 11.842179960261142,
      "grad_norm": 7.321256637573242,
      "learning_rate": 8.816378086857792e-05,
      "loss": 0.03292208909988403,
      "step": 41720
    },
    {
      "epoch": 11.845018450184503,
      "grad_norm": 8.405853271484375,
      "learning_rate": 8.816094237865456e-05,
      "loss": 0.03312258422374725,
      "step": 41730
    },
    {
      "epoch": 11.847856940107862,
      "grad_norm": 9.193829536437988,
      "learning_rate": 8.81581038887312e-05,
      "loss": 0.038640400767326354,
      "step": 41740
    },
    {
      "epoch": 11.850695430031223,
      "grad_norm": 6.199357032775879,
      "learning_rate": 8.815526539880784e-05,
      "loss": 0.037906485795974734,
      "step": 41750
    },
    {
      "epoch": 11.853533919954584,
      "grad_norm": 2.5867607593536377,
      "learning_rate": 8.815242690888448e-05,
      "loss": 0.04049636423587799,
      "step": 41760
    },
    {
      "epoch": 11.856372409877945,
      "grad_norm": 5.802853107452393,
      "learning_rate": 8.814958841896111e-05,
      "loss": 0.033687397837638855,
      "step": 41770
    },
    {
      "epoch": 11.859210899801306,
      "grad_norm": 10.755009651184082,
      "learning_rate": 8.814674992903775e-05,
      "loss": 0.047056376934051514,
      "step": 41780
    },
    {
      "epoch": 11.862049389724666,
      "grad_norm": 11.590856552124023,
      "learning_rate": 8.81439114391144e-05,
      "loss": 0.048236900568008424,
      "step": 41790
    },
    {
      "epoch": 11.864887879648027,
      "grad_norm": 5.509455680847168,
      "learning_rate": 8.814107294919104e-05,
      "loss": 0.04306095242500305,
      "step": 41800
    },
    {
      "epoch": 11.867726369571388,
      "grad_norm": 6.582401752471924,
      "learning_rate": 8.813823445926768e-05,
      "loss": 0.047679051756858826,
      "step": 41810
    },
    {
      "epoch": 11.870564859494749,
      "grad_norm": 9.859996795654297,
      "learning_rate": 8.813539596934432e-05,
      "loss": 0.03740607500076294,
      "step": 41820
    },
    {
      "epoch": 11.87340334941811,
      "grad_norm": 12.171548843383789,
      "learning_rate": 8.813255747942095e-05,
      "loss": 0.03052902817726135,
      "step": 41830
    },
    {
      "epoch": 11.876241839341471,
      "grad_norm": 2.2306880950927734,
      "learning_rate": 8.812971898949759e-05,
      "loss": 0.05683172345161438,
      "step": 41840
    },
    {
      "epoch": 11.87908032926483,
      "grad_norm": 12.165205001831055,
      "learning_rate": 8.812688049957423e-05,
      "loss": 0.05342947244644165,
      "step": 41850
    },
    {
      "epoch": 11.881918819188192,
      "grad_norm": 6.478231906890869,
      "learning_rate": 8.812404200965087e-05,
      "loss": 0.03707284927368164,
      "step": 41860
    },
    {
      "epoch": 11.884757309111553,
      "grad_norm": 22.78431510925293,
      "learning_rate": 8.812120351972751e-05,
      "loss": 0.05215840339660645,
      "step": 41870
    },
    {
      "epoch": 11.887595799034914,
      "grad_norm": 3.137918472290039,
      "learning_rate": 8.811836502980415e-05,
      "loss": 0.06916093230247497,
      "step": 41880
    },
    {
      "epoch": 11.890434288958275,
      "grad_norm": 8.645134925842285,
      "learning_rate": 8.81155265398808e-05,
      "loss": 0.04642211198806763,
      "step": 41890
    },
    {
      "epoch": 11.893272778881634,
      "grad_norm": 1.8387553691864014,
      "learning_rate": 8.811268804995742e-05,
      "loss": 0.0411649614572525,
      "step": 41900
    },
    {
      "epoch": 11.896111268804995,
      "grad_norm": 4.195104598999023,
      "learning_rate": 8.810984956003406e-05,
      "loss": 0.04706032574176788,
      "step": 41910
    },
    {
      "epoch": 11.898949758728357,
      "grad_norm": 4.468024253845215,
      "learning_rate": 8.81070110701107e-05,
      "loss": 0.04212497770786285,
      "step": 41920
    },
    {
      "epoch": 11.901788248651718,
      "grad_norm": 9.088099479675293,
      "learning_rate": 8.810417258018735e-05,
      "loss": 0.04539592862129212,
      "step": 41930
    },
    {
      "epoch": 11.904626738575079,
      "grad_norm": 7.755095958709717,
      "learning_rate": 8.810133409026399e-05,
      "loss": 0.040537679195404054,
      "step": 41940
    },
    {
      "epoch": 11.907465228498438,
      "grad_norm": 4.191405773162842,
      "learning_rate": 8.809849560034063e-05,
      "loss": 0.018843682110309602,
      "step": 41950
    },
    {
      "epoch": 11.9103037184218,
      "grad_norm": 1.374987006187439,
      "learning_rate": 8.809565711041726e-05,
      "loss": 0.043079504370689393,
      "step": 41960
    },
    {
      "epoch": 11.91314220834516,
      "grad_norm": 14.031201362609863,
      "learning_rate": 8.80928186204939e-05,
      "loss": 0.05393848419189453,
      "step": 41970
    },
    {
      "epoch": 11.915980698268521,
      "grad_norm": 14.838302612304688,
      "learning_rate": 8.808998013057054e-05,
      "loss": 0.050819283723831175,
      "step": 41980
    },
    {
      "epoch": 11.918819188191883,
      "grad_norm": 3.2832798957824707,
      "learning_rate": 8.808714164064718e-05,
      "loss": 0.020282982289791106,
      "step": 41990
    },
    {
      "epoch": 11.921657678115242,
      "grad_norm": 3.9569833278656006,
      "learning_rate": 8.808430315072382e-05,
      "loss": 0.017566296458244323,
      "step": 42000
    },
    {
      "epoch": 11.921657678115242,
      "eval_accuracy": 0.9522477268391938,
      "eval_loss": 0.14945776760578156,
      "eval_runtime": 31.1494,
      "eval_samples_per_second": 504.889,
      "eval_steps_per_second": 7.897,
      "step": 42000
    },
    {
      "epoch": 11.924496168038603,
      "grad_norm": 11.897820472717285,
      "learning_rate": 8.808146466080046e-05,
      "loss": 0.07478679418563842,
      "step": 42010
    },
    {
      "epoch": 11.927334657961964,
      "grad_norm": 6.896853923797607,
      "learning_rate": 8.80786261708771e-05,
      "loss": 0.0601233720779419,
      "step": 42020
    },
    {
      "epoch": 11.930173147885325,
      "grad_norm": 6.358935832977295,
      "learning_rate": 8.807578768095373e-05,
      "loss": 0.03607468008995056,
      "step": 42030
    },
    {
      "epoch": 11.933011637808686,
      "grad_norm": 2.915968656539917,
      "learning_rate": 8.807294919103037e-05,
      "loss": 0.025422203540802,
      "step": 42040
    },
    {
      "epoch": 11.935850127732046,
      "grad_norm": 22.498268127441406,
      "learning_rate": 8.807011070110702e-05,
      "loss": 0.07519700527191162,
      "step": 42050
    },
    {
      "epoch": 11.938688617655407,
      "grad_norm": 9.091548919677734,
      "learning_rate": 8.806727221118364e-05,
      "loss": 0.03408772349357605,
      "step": 42060
    },
    {
      "epoch": 11.941527107578768,
      "grad_norm": 12.65857219696045,
      "learning_rate": 8.80644337212603e-05,
      "loss": 0.06372144222259521,
      "step": 42070
    },
    {
      "epoch": 11.944365597502129,
      "grad_norm": 4.213757038116455,
      "learning_rate": 8.806159523133694e-05,
      "loss": 0.05281946063041687,
      "step": 42080
    },
    {
      "epoch": 11.94720408742549,
      "grad_norm": 4.296651840209961,
      "learning_rate": 8.805875674141357e-05,
      "loss": 0.03249916732311249,
      "step": 42090
    },
    {
      "epoch": 11.950042577348851,
      "grad_norm": 7.227084159851074,
      "learning_rate": 8.805591825149021e-05,
      "loss": 0.05526970028877258,
      "step": 42100
    },
    {
      "epoch": 11.95288106727221,
      "grad_norm": 2.970989465713501,
      "learning_rate": 8.805307976156685e-05,
      "loss": 0.08181440234184265,
      "step": 42110
    },
    {
      "epoch": 11.955719557195572,
      "grad_norm": 8.614218711853027,
      "learning_rate": 8.805024127164349e-05,
      "loss": 0.06890835165977478,
      "step": 42120
    },
    {
      "epoch": 11.958558047118933,
      "grad_norm": 11.872515678405762,
      "learning_rate": 8.804740278172013e-05,
      "loss": 0.04381870925426483,
      "step": 42130
    },
    {
      "epoch": 11.961396537042294,
      "grad_norm": 11.55391788482666,
      "learning_rate": 8.804456429179677e-05,
      "loss": 0.04326084852218628,
      "step": 42140
    },
    {
      "epoch": 11.964235026965655,
      "grad_norm": 3.883270025253296,
      "learning_rate": 8.804172580187342e-05,
      "loss": 0.05156497955322266,
      "step": 42150
    },
    {
      "epoch": 11.967073516889014,
      "grad_norm": 8.038182258605957,
      "learning_rate": 8.803888731195004e-05,
      "loss": 0.048931282758712766,
      "step": 42160
    },
    {
      "epoch": 11.969912006812375,
      "grad_norm": 11.797504425048828,
      "learning_rate": 8.803604882202668e-05,
      "loss": 0.047080081701278684,
      "step": 42170
    },
    {
      "epoch": 11.972750496735737,
      "grad_norm": 9.18109130859375,
      "learning_rate": 8.803321033210333e-05,
      "loss": 0.041204416751861574,
      "step": 42180
    },
    {
      "epoch": 11.975588986659098,
      "grad_norm": 6.056558609008789,
      "learning_rate": 8.803037184217995e-05,
      "loss": 0.059931010007858276,
      "step": 42190
    },
    {
      "epoch": 11.978427476582459,
      "grad_norm": 1.8931242227554321,
      "learning_rate": 8.802753335225661e-05,
      "loss": 0.031217482686042786,
      "step": 42200
    },
    {
      "epoch": 11.981265966505818,
      "grad_norm": 10.000553131103516,
      "learning_rate": 8.802469486233325e-05,
      "loss": 0.0313848614692688,
      "step": 42210
    },
    {
      "epoch": 11.98410445642918,
      "grad_norm": 17.2817325592041,
      "learning_rate": 8.802185637240988e-05,
      "loss": 0.03470008373260498,
      "step": 42220
    },
    {
      "epoch": 11.98694294635254,
      "grad_norm": 7.7406697273254395,
      "learning_rate": 8.801901788248652e-05,
      "loss": 0.03568956553936005,
      "step": 42230
    },
    {
      "epoch": 11.989781436275901,
      "grad_norm": 11.62606143951416,
      "learning_rate": 8.801617939256316e-05,
      "loss": 0.04242848157882691,
      "step": 42240
    },
    {
      "epoch": 11.992619926199263,
      "grad_norm": 12.246201515197754,
      "learning_rate": 8.80133409026398e-05,
      "loss": 0.04232434034347534,
      "step": 42250
    },
    {
      "epoch": 11.995458416122624,
      "grad_norm": 13.124924659729004,
      "learning_rate": 8.801050241271644e-05,
      "loss": 0.04008290469646454,
      "step": 42260
    },
    {
      "epoch": 11.998296906045983,
      "grad_norm": 4.904374599456787,
      "learning_rate": 8.800766392279308e-05,
      "loss": 0.033311459422111514,
      "step": 42270
    },
    {
      "epoch": 12.001135395969344,
      "grad_norm": 2.361665725708008,
      "learning_rate": 8.800482543286973e-05,
      "loss": 0.04348172843456268,
      "step": 42280
    },
    {
      "epoch": 12.003973885892705,
      "grad_norm": 6.2063984870910645,
      "learning_rate": 8.800198694294635e-05,
      "loss": 0.03905256688594818,
      "step": 42290
    },
    {
      "epoch": 12.006812375816066,
      "grad_norm": 15.71562671661377,
      "learning_rate": 8.7999148453023e-05,
      "loss": 0.04348371624946594,
      "step": 42300
    },
    {
      "epoch": 12.009650865739427,
      "grad_norm": 17.343069076538086,
      "learning_rate": 8.799630996309964e-05,
      "loss": 0.045898044109344484,
      "step": 42310
    },
    {
      "epoch": 12.012489355662787,
      "grad_norm": 3.102987766265869,
      "learning_rate": 8.799347147317626e-05,
      "loss": 0.035368016362190245,
      "step": 42320
    },
    {
      "epoch": 12.015327845586148,
      "grad_norm": 6.192180156707764,
      "learning_rate": 8.799063298325292e-05,
      "loss": 0.033758658170700076,
      "step": 42330
    },
    {
      "epoch": 12.018166335509509,
      "grad_norm": 3.8070037364959717,
      "learning_rate": 8.798779449332956e-05,
      "loss": 0.05028794407844543,
      "step": 42340
    },
    {
      "epoch": 12.02100482543287,
      "grad_norm": 2.517406463623047,
      "learning_rate": 8.798495600340619e-05,
      "loss": 0.03581957519054413,
      "step": 42350
    },
    {
      "epoch": 12.023843315356231,
      "grad_norm": 1.1406595706939697,
      "learning_rate": 8.798211751348283e-05,
      "loss": 0.01943010836839676,
      "step": 42360
    },
    {
      "epoch": 12.02668180527959,
      "grad_norm": 7.439162254333496,
      "learning_rate": 8.797927902355947e-05,
      "loss": 0.035764312744140624,
      "step": 42370
    },
    {
      "epoch": 12.029520295202952,
      "grad_norm": 7.404397010803223,
      "learning_rate": 8.797644053363611e-05,
      "loss": 0.03981121182441712,
      "step": 42380
    },
    {
      "epoch": 12.032358785126313,
      "grad_norm": 6.782740116119385,
      "learning_rate": 8.797360204371274e-05,
      "loss": 0.03376849591732025,
      "step": 42390
    },
    {
      "epoch": 12.035197275049674,
      "grad_norm": 9.032066345214844,
      "learning_rate": 8.79707635537894e-05,
      "loss": 0.03204431533813477,
      "step": 42400
    },
    {
      "epoch": 12.038035764973035,
      "grad_norm": 9.95860767364502,
      "learning_rate": 8.796792506386602e-05,
      "loss": 0.05169352889060974,
      "step": 42410
    },
    {
      "epoch": 12.040874254896394,
      "grad_norm": 10.233536720275879,
      "learning_rate": 8.796508657394266e-05,
      "loss": 0.03769796788692474,
      "step": 42420
    },
    {
      "epoch": 12.043712744819755,
      "grad_norm": 7.656820297241211,
      "learning_rate": 8.79622480840193e-05,
      "loss": 0.042575728893280027,
      "step": 42430
    },
    {
      "epoch": 12.046551234743117,
      "grad_norm": 0.9790419936180115,
      "learning_rate": 8.795940959409595e-05,
      "loss": 0.03717527687549591,
      "step": 42440
    },
    {
      "epoch": 12.049389724666478,
      "grad_norm": 5.362203598022461,
      "learning_rate": 8.795657110417258e-05,
      "loss": 0.023480720818042755,
      "step": 42450
    },
    {
      "epoch": 12.052228214589839,
      "grad_norm": 12.407505989074707,
      "learning_rate": 8.795373261424923e-05,
      "loss": 0.04921396374702454,
      "step": 42460
    },
    {
      "epoch": 12.0550667045132,
      "grad_norm": 4.909878730773926,
      "learning_rate": 8.795089412432587e-05,
      "loss": 0.03264098167419434,
      "step": 42470
    },
    {
      "epoch": 12.05790519443656,
      "grad_norm": 5.828683376312256,
      "learning_rate": 8.79480556344025e-05,
      "loss": 0.038746589422225954,
      "step": 42480
    },
    {
      "epoch": 12.06074368435992,
      "grad_norm": 5.357588291168213,
      "learning_rate": 8.794521714447914e-05,
      "loss": 0.03934710025787354,
      "step": 42490
    },
    {
      "epoch": 12.063582174283281,
      "grad_norm": 5.996522426605225,
      "learning_rate": 8.794237865455578e-05,
      "loss": 0.03904659152030945,
      "step": 42500
    },
    {
      "epoch": 12.063582174283281,
      "eval_accuracy": 0.9554269727220703,
      "eval_loss": 0.13257147371768951,
      "eval_runtime": 31.2505,
      "eval_samples_per_second": 503.256,
      "eval_steps_per_second": 7.872,
      "step": 42500
    },
    {
      "epoch": 12.066420664206642,
      "grad_norm": 4.982382297515869,
      "learning_rate": 8.793954016463241e-05,
      "loss": 0.03837008774280548,
      "step": 42510
    },
    {
      "epoch": 12.069259154130004,
      "grad_norm": 2.4558770656585693,
      "learning_rate": 8.793670167470905e-05,
      "loss": 0.037520408630371094,
      "step": 42520
    },
    {
      "epoch": 12.072097644053363,
      "grad_norm": 4.890630722045898,
      "learning_rate": 8.79338631847857e-05,
      "loss": 0.03202327191829681,
      "step": 42530
    },
    {
      "epoch": 12.074936133976724,
      "grad_norm": 1.131028413772583,
      "learning_rate": 8.793102469486233e-05,
      "loss": 0.020473235845565797,
      "step": 42540
    },
    {
      "epoch": 12.077774623900085,
      "grad_norm": 7.33089542388916,
      "learning_rate": 8.792818620493898e-05,
      "loss": 0.02888813614845276,
      "step": 42550
    },
    {
      "epoch": 12.080613113823446,
      "grad_norm": 8.652800559997559,
      "learning_rate": 8.792534771501562e-05,
      "loss": 0.02965686023235321,
      "step": 42560
    },
    {
      "epoch": 12.083451603746807,
      "grad_norm": 14.127680778503418,
      "learning_rate": 8.792250922509226e-05,
      "loss": 0.04270358085632324,
      "step": 42570
    },
    {
      "epoch": 12.086290093670167,
      "grad_norm": 1.539968490600586,
      "learning_rate": 8.791967073516889e-05,
      "loss": 0.037129050493240355,
      "step": 42580
    },
    {
      "epoch": 12.089128583593528,
      "grad_norm": 9.812592506408691,
      "learning_rate": 8.791683224524553e-05,
      "loss": 0.03342490792274475,
      "step": 42590
    },
    {
      "epoch": 12.091967073516889,
      "grad_norm": 7.126936912536621,
      "learning_rate": 8.791399375532218e-05,
      "loss": 0.029442226886749266,
      "step": 42600
    },
    {
      "epoch": 12.09480556344025,
      "grad_norm": 3.3653762340545654,
      "learning_rate": 8.791115526539881e-05,
      "loss": 0.029621991515159606,
      "step": 42610
    },
    {
      "epoch": 12.097644053363611,
      "grad_norm": 14.050369262695312,
      "learning_rate": 8.790831677547545e-05,
      "loss": 0.04310985803604126,
      "step": 42620
    },
    {
      "epoch": 12.10048254328697,
      "grad_norm": 6.579164981842041,
      "learning_rate": 8.790547828555209e-05,
      "loss": 0.030458983778953553,
      "step": 42630
    },
    {
      "epoch": 12.103321033210332,
      "grad_norm": 4.941464900970459,
      "learning_rate": 8.790263979562872e-05,
      "loss": 0.03447927534580231,
      "step": 42640
    },
    {
      "epoch": 12.106159523133693,
      "grad_norm": 10.615941047668457,
      "learning_rate": 8.789980130570536e-05,
      "loss": 0.0419352263212204,
      "step": 42650
    },
    {
      "epoch": 12.108998013057054,
      "grad_norm": 4.031155586242676,
      "learning_rate": 8.789696281578202e-05,
      "loss": 0.03987049460411072,
      "step": 42660
    },
    {
      "epoch": 12.111836502980415,
      "grad_norm": 1.352059245109558,
      "learning_rate": 8.789412432585865e-05,
      "loss": 0.03157951831817627,
      "step": 42670
    },
    {
      "epoch": 12.114674992903776,
      "grad_norm": 11.661965370178223,
      "learning_rate": 8.789128583593529e-05,
      "loss": 0.03795190453529358,
      "step": 42680
    },
    {
      "epoch": 12.117513482827135,
      "grad_norm": 1.8685001134872437,
      "learning_rate": 8.788844734601193e-05,
      "loss": 0.04514520764350891,
      "step": 42690
    },
    {
      "epoch": 12.120351972750496,
      "grad_norm": 5.572317600250244,
      "learning_rate": 8.788560885608857e-05,
      "loss": 0.048702746629714966,
      "step": 42700
    },
    {
      "epoch": 12.123190462673858,
      "grad_norm": 10.726156234741211,
      "learning_rate": 8.78827703661652e-05,
      "loss": 0.05781365633010864,
      "step": 42710
    },
    {
      "epoch": 12.126028952597219,
      "grad_norm": 7.369020938873291,
      "learning_rate": 8.787993187624184e-05,
      "loss": 0.05057991743087768,
      "step": 42720
    },
    {
      "epoch": 12.12886744252058,
      "grad_norm": 0.9827678203582764,
      "learning_rate": 8.78770933863185e-05,
      "loss": 0.024762935936450958,
      "step": 42730
    },
    {
      "epoch": 12.13170593244394,
      "grad_norm": 12.68351936340332,
      "learning_rate": 8.787425489639512e-05,
      "loss": 0.047663050889968875,
      "step": 42740
    },
    {
      "epoch": 12.1345444223673,
      "grad_norm": 13.245437622070312,
      "learning_rate": 8.787141640647176e-05,
      "loss": 0.04732508659362793,
      "step": 42750
    },
    {
      "epoch": 12.137382912290661,
      "grad_norm": 9.696499824523926,
      "learning_rate": 8.78685779165484e-05,
      "loss": 0.03219702243804932,
      "step": 42760
    },
    {
      "epoch": 12.140221402214022,
      "grad_norm": 6.578033924102783,
      "learning_rate": 8.786573942662503e-05,
      "loss": 0.02564353048801422,
      "step": 42770
    },
    {
      "epoch": 12.143059892137384,
      "grad_norm": 5.737947940826416,
      "learning_rate": 8.786290093670167e-05,
      "loss": 0.033503282070159915,
      "step": 42780
    },
    {
      "epoch": 12.145898382060743,
      "grad_norm": 9.456143379211426,
      "learning_rate": 8.786006244677831e-05,
      "loss": 0.03979947566986084,
      "step": 42790
    },
    {
      "epoch": 12.148736871984104,
      "grad_norm": 6.267122745513916,
      "learning_rate": 8.785722395685496e-05,
      "loss": 0.03283730149269104,
      "step": 42800
    },
    {
      "epoch": 12.151575361907465,
      "grad_norm": 7.741136074066162,
      "learning_rate": 8.78543854669316e-05,
      "loss": 0.031019583344459534,
      "step": 42810
    },
    {
      "epoch": 12.154413851830826,
      "grad_norm": 6.468270778656006,
      "learning_rate": 8.785154697700824e-05,
      "loss": 0.03639858365058899,
      "step": 42820
    },
    {
      "epoch": 12.157252341754187,
      "grad_norm": 10.36785888671875,
      "learning_rate": 8.784870848708488e-05,
      "loss": 0.037794700264930724,
      "step": 42830
    },
    {
      "epoch": 12.160090831677547,
      "grad_norm": 8.132712364196777,
      "learning_rate": 8.784586999716151e-05,
      "loss": 0.023357495665550232,
      "step": 42840
    },
    {
      "epoch": 12.162929321600908,
      "grad_norm": 1.8190107345581055,
      "learning_rate": 8.784303150723815e-05,
      "loss": 0.03351306617259979,
      "step": 42850
    },
    {
      "epoch": 12.165767811524269,
      "grad_norm": 1.7640764713287354,
      "learning_rate": 8.78401930173148e-05,
      "loss": 0.034797403216362,
      "step": 42860
    },
    {
      "epoch": 12.16860630144763,
      "grad_norm": 7.038665294647217,
      "learning_rate": 8.783735452739143e-05,
      "loss": 0.044648760557174684,
      "step": 42870
    },
    {
      "epoch": 12.171444791370991,
      "grad_norm": 8.174056053161621,
      "learning_rate": 8.783451603746807e-05,
      "loss": 0.033867499232292174,
      "step": 42880
    },
    {
      "epoch": 12.174283281294352,
      "grad_norm": 20.226119995117188,
      "learning_rate": 8.783167754754471e-05,
      "loss": 0.039821818470954895,
      "step": 42890
    },
    {
      "epoch": 12.177121771217712,
      "grad_norm": 7.061811923980713,
      "learning_rate": 8.782883905762134e-05,
      "loss": 0.030632388591766358,
      "step": 42900
    },
    {
      "epoch": 12.179960261141073,
      "grad_norm": 8.081908226013184,
      "learning_rate": 8.782600056769798e-05,
      "loss": 0.018213525414466858,
      "step": 42910
    },
    {
      "epoch": 12.182798751064434,
      "grad_norm": 3.5805306434631348,
      "learning_rate": 8.782316207777463e-05,
      "loss": 0.05772478580474853,
      "step": 42920
    },
    {
      "epoch": 12.185637240987795,
      "grad_norm": 5.696958065032959,
      "learning_rate": 8.782032358785127e-05,
      "loss": 0.022018913924694062,
      "step": 42930
    },
    {
      "epoch": 12.188475730911156,
      "grad_norm": 1.5079044103622437,
      "learning_rate": 8.781748509792791e-05,
      "loss": 0.024490922689437866,
      "step": 42940
    },
    {
      "epoch": 12.191314220834515,
      "grad_norm": 3.882221221923828,
      "learning_rate": 8.781464660800455e-05,
      "loss": 0.03649272918701172,
      "step": 42950
    },
    {
      "epoch": 12.194152710757876,
      "grad_norm": 4.7321882247924805,
      "learning_rate": 8.781180811808119e-05,
      "loss": 0.02632758319377899,
      "step": 42960
    },
    {
      "epoch": 12.196991200681238,
      "grad_norm": 6.319700241088867,
      "learning_rate": 8.780896962815782e-05,
      "loss": 0.02812599241733551,
      "step": 42970
    },
    {
      "epoch": 12.199829690604599,
      "grad_norm": 11.42015552520752,
      "learning_rate": 8.780613113823446e-05,
      "loss": 0.032302433252334596,
      "step": 42980
    },
    {
      "epoch": 12.20266818052796,
      "grad_norm": 5.789360523223877,
      "learning_rate": 8.78032926483111e-05,
      "loss": 0.024473686516284943,
      "step": 42990
    },
    {
      "epoch": 12.205506670451319,
      "grad_norm": 6.161420822143555,
      "learning_rate": 8.780045415838774e-05,
      "loss": 0.035588255524635314,
      "step": 43000
    },
    {
      "epoch": 12.205506670451319,
      "eval_accuracy": 0.9591784828638646,
      "eval_loss": 0.12536734342575073,
      "eval_runtime": 31.8222,
      "eval_samples_per_second": 494.215,
      "eval_steps_per_second": 7.73,
      "step": 43000
    },
    {
      "epoch": 12.20834516037468,
      "grad_norm": 4.745807647705078,
      "learning_rate": 8.779761566846438e-05,
      "loss": 0.02419144958257675,
      "step": 43010
    },
    {
      "epoch": 12.211183650298041,
      "grad_norm": 9.065601348876953,
      "learning_rate": 8.779477717854103e-05,
      "loss": 0.028294992446899415,
      "step": 43020
    },
    {
      "epoch": 12.214022140221402,
      "grad_norm": 0.9984960556030273,
      "learning_rate": 8.779193868861765e-05,
      "loss": 0.016124072670936584,
      "step": 43030
    },
    {
      "epoch": 12.216860630144764,
      "grad_norm": 9.727514266967773,
      "learning_rate": 8.77891001986943e-05,
      "loss": 0.028460302948951723,
      "step": 43040
    },
    {
      "epoch": 12.219699120068125,
      "grad_norm": 4.6544647216796875,
      "learning_rate": 8.778626170877094e-05,
      "loss": 0.015810102224349976,
      "step": 43050
    },
    {
      "epoch": 12.222537609991484,
      "grad_norm": 14.09741497039795,
      "learning_rate": 8.778342321884758e-05,
      "loss": 0.030832844972610473,
      "step": 43060
    },
    {
      "epoch": 12.225376099914845,
      "grad_norm": 1.9660956859588623,
      "learning_rate": 8.778058472892422e-05,
      "loss": 0.021106334030628206,
      "step": 43070
    },
    {
      "epoch": 12.228214589838206,
      "grad_norm": 6.233950138092041,
      "learning_rate": 8.777774623900086e-05,
      "loss": 0.030867934226989746,
      "step": 43080
    },
    {
      "epoch": 12.231053079761567,
      "grad_norm": 8.817118644714355,
      "learning_rate": 8.77749077490775e-05,
      "loss": 0.029362088441848753,
      "step": 43090
    },
    {
      "epoch": 12.233891569684928,
      "grad_norm": 2.01446270942688,
      "learning_rate": 8.777206925915413e-05,
      "loss": 0.0267075777053833,
      "step": 43100
    },
    {
      "epoch": 12.236730059608288,
      "grad_norm": 2.2910454273223877,
      "learning_rate": 8.776951461822311e-05,
      "loss": 0.028422459959983826,
      "step": 43110
    },
    {
      "epoch": 12.239568549531649,
      "grad_norm": 7.2257280349731445,
      "learning_rate": 8.776667612829975e-05,
      "loss": 0.05786190032958984,
      "step": 43120
    },
    {
      "epoch": 12.24240703945501,
      "grad_norm": 5.866209030151367,
      "learning_rate": 8.776383763837639e-05,
      "loss": 0.08099673986434937,
      "step": 43130
    },
    {
      "epoch": 12.245245529378371,
      "grad_norm": 4.997642993927002,
      "learning_rate": 8.776099914845303e-05,
      "loss": 0.032355567812919615,
      "step": 43140
    },
    {
      "epoch": 12.248084019301732,
      "grad_norm": 3.9322164058685303,
      "learning_rate": 8.775816065852966e-05,
      "loss": 0.028098559379577635,
      "step": 43150
    },
    {
      "epoch": 12.250922509225092,
      "grad_norm": 0.49283063411712646,
      "learning_rate": 8.77553221686063e-05,
      "loss": 0.020551927387714386,
      "step": 43160
    },
    {
      "epoch": 12.253760999148453,
      "grad_norm": 2.4942538738250732,
      "learning_rate": 8.775248367868294e-05,
      "loss": 0.03668495118618011,
      "step": 43170
    },
    {
      "epoch": 12.256599489071814,
      "grad_norm": 5.253053188323975,
      "learning_rate": 8.774964518875959e-05,
      "loss": 0.03662307262420654,
      "step": 43180
    },
    {
      "epoch": 12.259437978995175,
      "grad_norm": 5.2050275802612305,
      "learning_rate": 8.774680669883623e-05,
      "loss": 0.027386596798896788,
      "step": 43190
    },
    {
      "epoch": 12.262276468918536,
      "grad_norm": 2.7373571395874023,
      "learning_rate": 8.774396820891287e-05,
      "loss": 0.0306460440158844,
      "step": 43200
    },
    {
      "epoch": 12.265114958841895,
      "grad_norm": 4.104968070983887,
      "learning_rate": 8.77411297189895e-05,
      "loss": 0.03437112867832184,
      "step": 43210
    },
    {
      "epoch": 12.267953448765256,
      "grad_norm": 10.098012924194336,
      "learning_rate": 8.773829122906614e-05,
      "loss": 0.039316299557685855,
      "step": 43220
    },
    {
      "epoch": 12.270791938688618,
      "grad_norm": 8.756053924560547,
      "learning_rate": 8.773545273914278e-05,
      "loss": 0.048076283931732175,
      "step": 43230
    },
    {
      "epoch": 12.273630428611979,
      "grad_norm": 5.571841239929199,
      "learning_rate": 8.773261424921942e-05,
      "loss": 0.0457059919834137,
      "step": 43240
    },
    {
      "epoch": 12.27646891853534,
      "grad_norm": 6.984152793884277,
      "learning_rate": 8.772977575929606e-05,
      "loss": 0.04278313517570496,
      "step": 43250
    },
    {
      "epoch": 12.279307408458699,
      "grad_norm": 4.612057209014893,
      "learning_rate": 8.77269372693727e-05,
      "loss": 0.03646177053451538,
      "step": 43260
    },
    {
      "epoch": 12.28214589838206,
      "grad_norm": 10.539977073669434,
      "learning_rate": 8.772409877944934e-05,
      "loss": 0.028658172488212584,
      "step": 43270
    },
    {
      "epoch": 12.284984388305421,
      "grad_norm": 5.533712387084961,
      "learning_rate": 8.772126028952597e-05,
      "loss": 0.032504862546920775,
      "step": 43280
    },
    {
      "epoch": 12.287822878228782,
      "grad_norm": 3.1030781269073486,
      "learning_rate": 8.771842179960261e-05,
      "loss": 0.04522564709186554,
      "step": 43290
    },
    {
      "epoch": 12.290661368152143,
      "grad_norm": 5.694429874420166,
      "learning_rate": 8.771558330967925e-05,
      "loss": 0.027199763059616088,
      "step": 43300
    },
    {
      "epoch": 12.293499858075505,
      "grad_norm": 3.403394937515259,
      "learning_rate": 8.77127448197559e-05,
      "loss": 0.03828282952308655,
      "step": 43310
    },
    {
      "epoch": 12.296338347998864,
      "grad_norm": 4.743762493133545,
      "learning_rate": 8.770990632983254e-05,
      "loss": 0.05068618655204773,
      "step": 43320
    },
    {
      "epoch": 12.299176837922225,
      "grad_norm": 5.177332878112793,
      "learning_rate": 8.770706783990918e-05,
      "loss": 0.03021160066127777,
      "step": 43330
    },
    {
      "epoch": 12.302015327845586,
      "grad_norm": 1.2973488569259644,
      "learning_rate": 8.77042293499858e-05,
      "loss": 0.024166953563690186,
      "step": 43340
    },
    {
      "epoch": 12.304853817768947,
      "grad_norm": 10.911638259887695,
      "learning_rate": 8.770139086006245e-05,
      "loss": 0.04314410984516144,
      "step": 43350
    },
    {
      "epoch": 12.307692307692308,
      "grad_norm": 5.359420299530029,
      "learning_rate": 8.769855237013909e-05,
      "loss": 0.05115126371383667,
      "step": 43360
    },
    {
      "epoch": 12.310530797615668,
      "grad_norm": 9.825884819030762,
      "learning_rate": 8.769571388021573e-05,
      "loss": 0.037463635206222534,
      "step": 43370
    },
    {
      "epoch": 12.313369287539029,
      "grad_norm": 8.118050575256348,
      "learning_rate": 8.769287539029237e-05,
      "loss": 0.02044713944196701,
      "step": 43380
    },
    {
      "epoch": 12.31620777746239,
      "grad_norm": 5.115602016448975,
      "learning_rate": 8.769003690036901e-05,
      "loss": 0.03220890462398529,
      "step": 43390
    },
    {
      "epoch": 12.319046267385751,
      "grad_norm": 6.095489978790283,
      "learning_rate": 8.768719841044564e-05,
      "loss": 0.03606012463569641,
      "step": 43400
    },
    {
      "epoch": 12.321884757309112,
      "grad_norm": 1.2136930227279663,
      "learning_rate": 8.768435992052228e-05,
      "loss": 0.022545641660690306,
      "step": 43410
    },
    {
      "epoch": 12.324723247232471,
      "grad_norm": 3.3116722106933594,
      "learning_rate": 8.768152143059892e-05,
      "loss": 0.046520963311195374,
      "step": 43420
    },
    {
      "epoch": 12.327561737155833,
      "grad_norm": 5.2578444480896,
      "learning_rate": 8.767868294067557e-05,
      "loss": 0.02567618191242218,
      "step": 43430
    },
    {
      "epoch": 12.330400227079194,
      "grad_norm": 7.1209211349487305,
      "learning_rate": 8.767584445075221e-05,
      "loss": 0.03235772252082825,
      "step": 43440
    },
    {
      "epoch": 12.333238717002555,
      "grad_norm": 8.388750076293945,
      "learning_rate": 8.767300596082885e-05,
      "loss": 0.03658254742622376,
      "step": 43450
    },
    {
      "epoch": 12.336077206925916,
      "grad_norm": 6.390434265136719,
      "learning_rate": 8.767016747090549e-05,
      "loss": 0.04859638214111328,
      "step": 43460
    },
    {
      "epoch": 12.338915696849277,
      "grad_norm": 4.703362464904785,
      "learning_rate": 8.766732898098212e-05,
      "loss": 0.056396156549453735,
      "step": 43470
    },
    {
      "epoch": 12.341754186772636,
      "grad_norm": 6.984062671661377,
      "learning_rate": 8.766449049105876e-05,
      "loss": 0.047119522094726564,
      "step": 43480
    },
    {
      "epoch": 12.344592676695997,
      "grad_norm": 16.24080467224121,
      "learning_rate": 8.76616520011354e-05,
      "loss": 0.059197747707366945,
      "step": 43490
    },
    {
      "epoch": 12.347431166619359,
      "grad_norm": 4.961618900299072,
      "learning_rate": 8.765881351121203e-05,
      "loss": 0.043620502948760985,
      "step": 43500
    },
    {
      "epoch": 12.347431166619359,
      "eval_accuracy": 0.9575252750047689,
      "eval_loss": 0.12631969153881073,
      "eval_runtime": 32.0467,
      "eval_samples_per_second": 490.753,
      "eval_steps_per_second": 7.676,
      "step": 43500
    },
    {
      "epoch": 12.35026965654272,
      "grad_norm": 8.303855895996094,
      "learning_rate": 8.765597502128868e-05,
      "loss": 0.04292536377906799,
      "step": 43510
    },
    {
      "epoch": 12.35310814646608,
      "grad_norm": 3.528038501739502,
      "learning_rate": 8.765313653136532e-05,
      "loss": 0.025262534618377686,
      "step": 43520
    },
    {
      "epoch": 12.35594663638944,
      "grad_norm": 7.5579047203063965,
      "learning_rate": 8.765029804144195e-05,
      "loss": 0.037899512052536014,
      "step": 43530
    },
    {
      "epoch": 12.358785126312801,
      "grad_norm": 2.860135555267334,
      "learning_rate": 8.76474595515186e-05,
      "loss": 0.020976607501506806,
      "step": 43540
    },
    {
      "epoch": 12.361623616236162,
      "grad_norm": 10.304487228393555,
      "learning_rate": 8.764462106159523e-05,
      "loss": 0.021430324018001556,
      "step": 43550
    },
    {
      "epoch": 12.364462106159523,
      "grad_norm": 1.565558910369873,
      "learning_rate": 8.764178257167188e-05,
      "loss": 0.032142537832260135,
      "step": 43560
    },
    {
      "epoch": 12.367300596082885,
      "grad_norm": 3.965823173522949,
      "learning_rate": 8.76389440817485e-05,
      "loss": 0.03533589839935303,
      "step": 43570
    },
    {
      "epoch": 12.370139086006244,
      "grad_norm": 7.955080509185791,
      "learning_rate": 8.763610559182516e-05,
      "loss": 0.02849397957324982,
      "step": 43580
    },
    {
      "epoch": 12.372977575929605,
      "grad_norm": 3.1058242321014404,
      "learning_rate": 8.76332671019018e-05,
      "loss": 0.045594796538352966,
      "step": 43590
    },
    {
      "epoch": 12.375816065852966,
      "grad_norm": 9.424407958984375,
      "learning_rate": 8.763042861197843e-05,
      "loss": 0.027894610166549684,
      "step": 43600
    },
    {
      "epoch": 12.378654555776327,
      "grad_norm": 13.528703689575195,
      "learning_rate": 8.762759012205507e-05,
      "loss": 0.043479698896408084,
      "step": 43610
    },
    {
      "epoch": 12.381493045699688,
      "grad_norm": 3.8311986923217773,
      "learning_rate": 8.762475163213171e-05,
      "loss": 0.06389501690864563,
      "step": 43620
    },
    {
      "epoch": 12.384331535623048,
      "grad_norm": 5.446621417999268,
      "learning_rate": 8.762191314220834e-05,
      "loss": 0.05128917694091797,
      "step": 43630
    },
    {
      "epoch": 12.387170025546409,
      "grad_norm": 10.516417503356934,
      "learning_rate": 8.7619074652285e-05,
      "loss": 0.033826780319213864,
      "step": 43640
    },
    {
      "epoch": 12.39000851546977,
      "grad_norm": 9.974024772644043,
      "learning_rate": 8.761623616236164e-05,
      "loss": 0.027305030822753908,
      "step": 43650
    },
    {
      "epoch": 12.392847005393131,
      "grad_norm": 1.2486107349395752,
      "learning_rate": 8.761339767243826e-05,
      "loss": 0.031165832281112672,
      "step": 43660
    },
    {
      "epoch": 12.395685495316492,
      "grad_norm": 15.140544891357422,
      "learning_rate": 8.76105591825149e-05,
      "loss": 0.037294122576713565,
      "step": 43670
    },
    {
      "epoch": 12.398523985239853,
      "grad_norm": 1.334664225578308,
      "learning_rate": 8.760772069259155e-05,
      "loss": 0.03070785999298096,
      "step": 43680
    },
    {
      "epoch": 12.401362475163213,
      "grad_norm": 4.582605361938477,
      "learning_rate": 8.760488220266819e-05,
      "loss": 0.033212360739707944,
      "step": 43690
    },
    {
      "epoch": 12.404200965086574,
      "grad_norm": 4.622889995574951,
      "learning_rate": 8.760204371274481e-05,
      "loss": 0.025003960728645323,
      "step": 43700
    },
    {
      "epoch": 12.407039455009935,
      "grad_norm": 4.53710412979126,
      "learning_rate": 8.759920522282147e-05,
      "loss": 0.037190794944763184,
      "step": 43710
    },
    {
      "epoch": 12.409877944933296,
      "grad_norm": 14.40041732788086,
      "learning_rate": 8.759636673289811e-05,
      "loss": 0.03887277543544769,
      "step": 43720
    },
    {
      "epoch": 12.412716434856657,
      "grad_norm": 4.566652774810791,
      "learning_rate": 8.759352824297474e-05,
      "loss": 0.036245569586753845,
      "step": 43730
    },
    {
      "epoch": 12.415554924780016,
      "grad_norm": 1.31491219997406,
      "learning_rate": 8.759068975305138e-05,
      "loss": 0.04045813083648682,
      "step": 43740
    },
    {
      "epoch": 12.418393414703377,
      "grad_norm": 6.651790142059326,
      "learning_rate": 8.758785126312802e-05,
      "loss": 0.04301515221595764,
      "step": 43750
    },
    {
      "epoch": 12.421231904626739,
      "grad_norm": 0.6239034533500671,
      "learning_rate": 8.758501277320465e-05,
      "loss": 0.02968602180480957,
      "step": 43760
    },
    {
      "epoch": 12.4240703945501,
      "grad_norm": 14.26272964477539,
      "learning_rate": 8.758217428328129e-05,
      "loss": 0.050833702087402344,
      "step": 43770
    },
    {
      "epoch": 12.42690888447346,
      "grad_norm": 4.808349609375,
      "learning_rate": 8.757933579335795e-05,
      "loss": 0.04993874728679657,
      "step": 43780
    },
    {
      "epoch": 12.42974737439682,
      "grad_norm": 0.792005717754364,
      "learning_rate": 8.757649730343457e-05,
      "loss": 0.053781908750534055,
      "step": 43790
    },
    {
      "epoch": 12.432585864320181,
      "grad_norm": 15.564403533935547,
      "learning_rate": 8.757365881351122e-05,
      "loss": 0.031227922439575194,
      "step": 43800
    },
    {
      "epoch": 12.435424354243542,
      "grad_norm": 5.584966659545898,
      "learning_rate": 8.757082032358786e-05,
      "loss": 0.04057415425777435,
      "step": 43810
    },
    {
      "epoch": 12.438262844166903,
      "grad_norm": 10.339362144470215,
      "learning_rate": 8.75679818336645e-05,
      "loss": 0.047411096096038816,
      "step": 43820
    },
    {
      "epoch": 12.441101334090265,
      "grad_norm": 1.389992117881775,
      "learning_rate": 8.756514334374113e-05,
      "loss": 0.03012396991252899,
      "step": 43830
    },
    {
      "epoch": 12.443939824013626,
      "grad_norm": 12.567704200744629,
      "learning_rate": 8.756230485381778e-05,
      "loss": 0.036432310938835144,
      "step": 43840
    },
    {
      "epoch": 12.446778313936985,
      "grad_norm": 1.846071481704712,
      "learning_rate": 8.755946636389442e-05,
      "loss": 0.041059836745262146,
      "step": 43850
    },
    {
      "epoch": 12.449616803860346,
      "grad_norm": 1.4687913656234741,
      "learning_rate": 8.755662787397105e-05,
      "loss": 0.03178577125072479,
      "step": 43860
    },
    {
      "epoch": 12.452455293783707,
      "grad_norm": 8.927151679992676,
      "learning_rate": 8.755378938404769e-05,
      "loss": 0.05755612254142761,
      "step": 43870
    },
    {
      "epoch": 12.455293783707068,
      "grad_norm": 8.224190711975098,
      "learning_rate": 8.755095089412433e-05,
      "loss": 0.030533698201179505,
      "step": 43880
    },
    {
      "epoch": 12.45813227363043,
      "grad_norm": 3.2042739391326904,
      "learning_rate": 8.754811240420096e-05,
      "loss": 0.03339183330535889,
      "step": 43890
    },
    {
      "epoch": 12.460970763553789,
      "grad_norm": 0.938050389289856,
      "learning_rate": 8.75452739142776e-05,
      "loss": 0.03771317601203918,
      "step": 43900
    },
    {
      "epoch": 12.46380925347715,
      "grad_norm": 5.056128025054932,
      "learning_rate": 8.754243542435426e-05,
      "loss": 0.021968071162700654,
      "step": 43910
    },
    {
      "epoch": 12.466647743400511,
      "grad_norm": 2.449493169784546,
      "learning_rate": 8.753959693443088e-05,
      "loss": 0.021292699873447417,
      "step": 43920
    },
    {
      "epoch": 12.469486233323872,
      "grad_norm": 6.865640163421631,
      "learning_rate": 8.753675844450753e-05,
      "loss": 0.03274143636226654,
      "step": 43930
    },
    {
      "epoch": 12.472324723247233,
      "grad_norm": 6.458326816558838,
      "learning_rate": 8.753391995458417e-05,
      "loss": 0.02600388526916504,
      "step": 43940
    },
    {
      "epoch": 12.475163213170593,
      "grad_norm": 4.321523666381836,
      "learning_rate": 8.753108146466081e-05,
      "loss": 0.03519765734672546,
      "step": 43950
    },
    {
      "epoch": 12.478001703093954,
      "grad_norm": 3.655992269515991,
      "learning_rate": 8.752824297473744e-05,
      "loss": 0.02357748746871948,
      "step": 43960
    },
    {
      "epoch": 12.480840193017315,
      "grad_norm": 9.547161102294922,
      "learning_rate": 8.752540448481409e-05,
      "loss": 0.04760206341743469,
      "step": 43970
    },
    {
      "epoch": 12.483678682940676,
      "grad_norm": 7.882644176483154,
      "learning_rate": 8.752256599489073e-05,
      "loss": 0.06730462312698364,
      "step": 43980
    },
    {
      "epoch": 12.486517172864037,
      "grad_norm": 6.823720455169678,
      "learning_rate": 8.751972750496736e-05,
      "loss": 0.04396010637283325,
      "step": 43990
    },
    {
      "epoch": 12.489355662787396,
      "grad_norm": 0.9623273015022278,
      "learning_rate": 8.7516889015044e-05,
      "loss": 0.03538389205932617,
      "step": 44000
    },
    {
      "epoch": 12.489355662787396,
      "eval_accuracy": 0.9582882940166593,
      "eval_loss": 0.12650801241397858,
      "eval_runtime": 31.6181,
      "eval_samples_per_second": 497.405,
      "eval_steps_per_second": 7.78,
      "step": 44000
    },
    {
      "epoch": 12.492194152710757,
      "grad_norm": 14.151564598083496,
      "learning_rate": 8.751405052512064e-05,
      "loss": 0.043718361854553224,
      "step": 44010
    },
    {
      "epoch": 12.495032642634119,
      "grad_norm": 3.9960615634918213,
      "learning_rate": 8.751121203519727e-05,
      "loss": 0.04656648337841034,
      "step": 44020
    },
    {
      "epoch": 12.49787113255748,
      "grad_norm": 9.379133224487305,
      "learning_rate": 8.750837354527391e-05,
      "loss": 0.04604141414165497,
      "step": 44030
    },
    {
      "epoch": 12.50070962248084,
      "grad_norm": 10.999762535095215,
      "learning_rate": 8.750553505535057e-05,
      "loss": 0.06433385610580444,
      "step": 44040
    },
    {
      "epoch": 12.5035481124042,
      "grad_norm": 2.8640451431274414,
      "learning_rate": 8.75026965654272e-05,
      "loss": 0.05977180600166321,
      "step": 44050
    },
    {
      "epoch": 12.506386602327561,
      "grad_norm": 10.317655563354492,
      "learning_rate": 8.749985807550384e-05,
      "loss": 0.032905089855194095,
      "step": 44060
    },
    {
      "epoch": 12.509225092250922,
      "grad_norm": 7.531612873077393,
      "learning_rate": 8.749701958558048e-05,
      "loss": 0.0447359174489975,
      "step": 44070
    },
    {
      "epoch": 12.512063582174283,
      "grad_norm": 3.5741655826568604,
      "learning_rate": 8.749418109565712e-05,
      "loss": 0.0609171986579895,
      "step": 44080
    },
    {
      "epoch": 12.514902072097644,
      "grad_norm": 5.538876533508301,
      "learning_rate": 8.749134260573375e-05,
      "loss": 0.027841529250144957,
      "step": 44090
    },
    {
      "epoch": 12.517740562021006,
      "grad_norm": 2.073808431625366,
      "learning_rate": 8.748850411581039e-05,
      "loss": 0.038188064098358156,
      "step": 44100
    },
    {
      "epoch": 12.520579051944365,
      "grad_norm": 8.731266975402832,
      "learning_rate": 8.748566562588703e-05,
      "loss": 0.038576528429985046,
      "step": 44110
    },
    {
      "epoch": 12.523417541867726,
      "grad_norm": 18.402067184448242,
      "learning_rate": 8.748282713596367e-05,
      "loss": 0.043379539251327516,
      "step": 44120
    },
    {
      "epoch": 12.526256031791087,
      "grad_norm": 5.526108741760254,
      "learning_rate": 8.747998864604031e-05,
      "loss": 0.023209093511104582,
      "step": 44130
    },
    {
      "epoch": 12.529094521714448,
      "grad_norm": 11.312538146972656,
      "learning_rate": 8.747715015611695e-05,
      "loss": 0.02301383316516876,
      "step": 44140
    },
    {
      "epoch": 12.53193301163781,
      "grad_norm": 10.428009986877441,
      "learning_rate": 8.747431166619358e-05,
      "loss": 0.043978333473205566,
      "step": 44150
    },
    {
      "epoch": 12.534771501561169,
      "grad_norm": 10.228684425354004,
      "learning_rate": 8.747147317627022e-05,
      "loss": 0.035221797227859494,
      "step": 44160
    },
    {
      "epoch": 12.53760999148453,
      "grad_norm": 8.798989295959473,
      "learning_rate": 8.746863468634688e-05,
      "loss": 0.031769979000091556,
      "step": 44170
    },
    {
      "epoch": 12.540448481407891,
      "grad_norm": 6.162079334259033,
      "learning_rate": 8.74657961964235e-05,
      "loss": 0.027561104297637938,
      "step": 44180
    },
    {
      "epoch": 12.543286971331252,
      "grad_norm": 14.738848686218262,
      "learning_rate": 8.746295770650015e-05,
      "loss": 0.03986902236938476,
      "step": 44190
    },
    {
      "epoch": 12.546125461254613,
      "grad_norm": 7.156569480895996,
      "learning_rate": 8.746011921657679e-05,
      "loss": 0.04217462539672852,
      "step": 44200
    },
    {
      "epoch": 12.548963951177974,
      "grad_norm": 4.159841060638428,
      "learning_rate": 8.745728072665342e-05,
      "loss": 0.035502883791923526,
      "step": 44210
    },
    {
      "epoch": 12.551802441101334,
      "grad_norm": 7.1332316398620605,
      "learning_rate": 8.745444223673006e-05,
      "loss": 0.03097446858882904,
      "step": 44220
    },
    {
      "epoch": 12.554640931024695,
      "grad_norm": 8.165454864501953,
      "learning_rate": 8.74516037468067e-05,
      "loss": 0.04084550738334656,
      "step": 44230
    },
    {
      "epoch": 12.557479420948056,
      "grad_norm": 1.4880213737487793,
      "learning_rate": 8.744876525688334e-05,
      "loss": 0.026027166843414308,
      "step": 44240
    },
    {
      "epoch": 12.560317910871417,
      "grad_norm": 7.585758686065674,
      "learning_rate": 8.744592676695998e-05,
      "loss": 0.05606292486190796,
      "step": 44250
    },
    {
      "epoch": 12.563156400794778,
      "grad_norm": 14.649521827697754,
      "learning_rate": 8.744308827703662e-05,
      "loss": 0.03574174642562866,
      "step": 44260
    },
    {
      "epoch": 12.565994890718137,
      "grad_norm": 8.295636177062988,
      "learning_rate": 8.744024978711326e-05,
      "loss": 0.025662416219711305,
      "step": 44270
    },
    {
      "epoch": 12.568833380641498,
      "grad_norm": 6.584290981292725,
      "learning_rate": 8.743741129718989e-05,
      "loss": 0.045034149289131166,
      "step": 44280
    },
    {
      "epoch": 12.57167187056486,
      "grad_norm": 7.3886799812316895,
      "learning_rate": 8.743457280726653e-05,
      "loss": 0.039062914252281186,
      "step": 44290
    },
    {
      "epoch": 12.57451036048822,
      "grad_norm": 1.0931612253189087,
      "learning_rate": 8.743173431734318e-05,
      "loss": 0.026321253180503844,
      "step": 44300
    },
    {
      "epoch": 12.577348850411582,
      "grad_norm": 3.9734034538269043,
      "learning_rate": 8.742889582741982e-05,
      "loss": 0.030106019973754884,
      "step": 44310
    },
    {
      "epoch": 12.580187340334941,
      "grad_norm": 3.5984418392181396,
      "learning_rate": 8.742605733749646e-05,
      "loss": 0.038176172971725465,
      "step": 44320
    },
    {
      "epoch": 12.583025830258302,
      "grad_norm": 3.420219898223877,
      "learning_rate": 8.74232188475731e-05,
      "loss": 0.04225446283817291,
      "step": 44330
    },
    {
      "epoch": 12.585864320181663,
      "grad_norm": 6.645458221435547,
      "learning_rate": 8.742038035764973e-05,
      "loss": 0.04775756597518921,
      "step": 44340
    },
    {
      "epoch": 12.588702810105024,
      "grad_norm": 14.90196418762207,
      "learning_rate": 8.741754186772637e-05,
      "loss": 0.03470243215560913,
      "step": 44350
    },
    {
      "epoch": 12.591541300028386,
      "grad_norm": 6.199538230895996,
      "learning_rate": 8.741470337780301e-05,
      "loss": 0.028724318742752074,
      "step": 44360
    },
    {
      "epoch": 12.594379789951745,
      "grad_norm": 11.46839714050293,
      "learning_rate": 8.741186488787965e-05,
      "loss": 0.028799358010292053,
      "step": 44370
    },
    {
      "epoch": 12.597218279875106,
      "grad_norm": 6.085451126098633,
      "learning_rate": 8.740902639795629e-05,
      "loss": 0.044294270873069766,
      "step": 44380
    },
    {
      "epoch": 12.600056769798467,
      "grad_norm": 9.302290916442871,
      "learning_rate": 8.740618790803293e-05,
      "loss": 0.04389075338840485,
      "step": 44390
    },
    {
      "epoch": 12.602895259721828,
      "grad_norm": 9.839166641235352,
      "learning_rate": 8.740334941810958e-05,
      "loss": 0.04611421227455139,
      "step": 44400
    },
    {
      "epoch": 12.60573374964519,
      "grad_norm": 2.907285213470459,
      "learning_rate": 8.74005109281862e-05,
      "loss": 0.02291063964366913,
      "step": 44410
    },
    {
      "epoch": 12.608572239568549,
      "grad_norm": 15.603127479553223,
      "learning_rate": 8.739767243826284e-05,
      "loss": 0.050103431940078734,
      "step": 44420
    },
    {
      "epoch": 12.61141072949191,
      "grad_norm": 7.325162887573242,
      "learning_rate": 8.739483394833949e-05,
      "loss": 0.04359988272190094,
      "step": 44430
    },
    {
      "epoch": 12.61424921941527,
      "grad_norm": 5.135088920593262,
      "learning_rate": 8.739199545841613e-05,
      "loss": 0.02586413323879242,
      "step": 44440
    },
    {
      "epoch": 12.617087709338632,
      "grad_norm": 6.717268466949463,
      "learning_rate": 8.738915696849277e-05,
      "loss": 0.025001877546310426,
      "step": 44450
    },
    {
      "epoch": 12.619926199261993,
      "grad_norm": 5.10106897354126,
      "learning_rate": 8.738631847856941e-05,
      "loss": 0.033975788950920106,
      "step": 44460
    },
    {
      "epoch": 12.622764689185354,
      "grad_norm": 4.140856742858887,
      "learning_rate": 8.738347998864604e-05,
      "loss": 0.033056697249412535,
      "step": 44470
    },
    {
      "epoch": 12.625603179108714,
      "grad_norm": 6.44667911529541,
      "learning_rate": 8.738064149872268e-05,
      "loss": 0.036567941308021545,
      "step": 44480
    },
    {
      "epoch": 12.628441669032075,
      "grad_norm": 8.409709930419922,
      "learning_rate": 8.737780300879932e-05,
      "loss": 0.04436020851135254,
      "step": 44490
    },
    {
      "epoch": 12.631280158955436,
      "grad_norm": 7.989817142486572,
      "learning_rate": 8.737496451887596e-05,
      "loss": 0.05788024663925171,
      "step": 44500
    },
    {
      "epoch": 12.631280158955436,
      "eval_accuracy": 0.9514847078273033,
      "eval_loss": 0.15090706944465637,
      "eval_runtime": 31.5217,
      "eval_samples_per_second": 498.927,
      "eval_steps_per_second": 7.804,
      "step": 44500
    },
    {
      "epoch": 12.634118648878797,
      "grad_norm": 5.888620376586914,
      "learning_rate": 8.73721260289526e-05,
      "loss": 0.033754372596740724,
      "step": 44510
    },
    {
      "epoch": 12.636957138802158,
      "grad_norm": 13.206843376159668,
      "learning_rate": 8.736928753902924e-05,
      "loss": 0.051671010255813596,
      "step": 44520
    },
    {
      "epoch": 12.639795628725517,
      "grad_norm": 17.667720794677734,
      "learning_rate": 8.736644904910589e-05,
      "loss": 0.0664360761642456,
      "step": 44530
    },
    {
      "epoch": 12.642634118648878,
      "grad_norm": 5.94382905960083,
      "learning_rate": 8.736361055918251e-05,
      "loss": 0.046749609708786014,
      "step": 44540
    },
    {
      "epoch": 12.64547260857224,
      "grad_norm": 3.479749917984009,
      "learning_rate": 8.736077206925916e-05,
      "loss": 0.06297379732131958,
      "step": 44550
    },
    {
      "epoch": 12.6483110984956,
      "grad_norm": 2.689229726791382,
      "learning_rate": 8.73579335793358e-05,
      "loss": 0.05441311001777649,
      "step": 44560
    },
    {
      "epoch": 12.651149588418962,
      "grad_norm": 8.93832015991211,
      "learning_rate": 8.735509508941244e-05,
      "loss": 0.07238125801086426,
      "step": 44570
    },
    {
      "epoch": 12.653988078342321,
      "grad_norm": 1.7631980180740356,
      "learning_rate": 8.735225659948908e-05,
      "loss": 0.0652324378490448,
      "step": 44580
    },
    {
      "epoch": 12.656826568265682,
      "grad_norm": 12.822599411010742,
      "learning_rate": 8.734941810956572e-05,
      "loss": 0.062204664945602416,
      "step": 44590
    },
    {
      "epoch": 12.659665058189043,
      "grad_norm": 10.897828102111816,
      "learning_rate": 8.734657961964235e-05,
      "loss": 0.0740848183631897,
      "step": 44600
    },
    {
      "epoch": 12.662503548112404,
      "grad_norm": 10.319339752197266,
      "learning_rate": 8.734374112971899e-05,
      "loss": 0.04726337492465973,
      "step": 44610
    },
    {
      "epoch": 12.665342038035766,
      "grad_norm": 13.167956352233887,
      "learning_rate": 8.734090263979563e-05,
      "loss": 0.052084213495254515,
      "step": 44620
    },
    {
      "epoch": 12.668180527959127,
      "grad_norm": 12.470789909362793,
      "learning_rate": 8.733806414987227e-05,
      "loss": 0.03885383605957031,
      "step": 44630
    },
    {
      "epoch": 12.671019017882486,
      "grad_norm": 3.0231542587280273,
      "learning_rate": 8.733522565994891e-05,
      "loss": 0.02193387597799301,
      "step": 44640
    },
    {
      "epoch": 12.673857507805847,
      "grad_norm": 6.86521577835083,
      "learning_rate": 8.733238717002556e-05,
      "loss": 0.04313365817070007,
      "step": 44650
    },
    {
      "epoch": 12.676695997729208,
      "grad_norm": 6.823424339294434,
      "learning_rate": 8.73295486801022e-05,
      "loss": 0.04115325808525085,
      "step": 44660
    },
    {
      "epoch": 12.67953448765257,
      "grad_norm": 5.839014530181885,
      "learning_rate": 8.732671019017882e-05,
      "loss": 0.028002548217773437,
      "step": 44670
    },
    {
      "epoch": 12.68237297757593,
      "grad_norm": 3.50873064994812,
      "learning_rate": 8.732387170025547e-05,
      "loss": 0.04054447710514068,
      "step": 44680
    },
    {
      "epoch": 12.68521146749929,
      "grad_norm": 10.432472229003906,
      "learning_rate": 8.732103321033211e-05,
      "loss": 0.05989189743995667,
      "step": 44690
    },
    {
      "epoch": 12.68804995742265,
      "grad_norm": 3.8197381496429443,
      "learning_rate": 8.731819472040874e-05,
      "loss": 0.07042637467384338,
      "step": 44700
    },
    {
      "epoch": 12.690888447346012,
      "grad_norm": 18.913007736206055,
      "learning_rate": 8.731535623048539e-05,
      "loss": 0.03881549537181854,
      "step": 44710
    },
    {
      "epoch": 12.693726937269373,
      "grad_norm": 7.585214138031006,
      "learning_rate": 8.731251774056203e-05,
      "loss": 0.03703373670578003,
      "step": 44720
    },
    {
      "epoch": 12.696565427192734,
      "grad_norm": 10.077887535095215,
      "learning_rate": 8.730967925063866e-05,
      "loss": 0.04344882369041443,
      "step": 44730
    },
    {
      "epoch": 12.699403917116094,
      "grad_norm": 4.145294666290283,
      "learning_rate": 8.73068407607153e-05,
      "loss": 0.03734053671360016,
      "step": 44740
    },
    {
      "epoch": 12.702242407039455,
      "grad_norm": 13.83118724822998,
      "learning_rate": 8.730400227079194e-05,
      "loss": 0.03142139911651611,
      "step": 44750
    },
    {
      "epoch": 12.705080896962816,
      "grad_norm": 7.415091514587402,
      "learning_rate": 8.730116378086858e-05,
      "loss": 0.04637177884578705,
      "step": 44760
    },
    {
      "epoch": 12.707919386886177,
      "grad_norm": 2.674288511276245,
      "learning_rate": 8.729832529094523e-05,
      "loss": 0.036681899428367616,
      "step": 44770
    },
    {
      "epoch": 12.710757876809538,
      "grad_norm": 10.82300853729248,
      "learning_rate": 8.729548680102187e-05,
      "loss": 0.04347237348556519,
      "step": 44780
    },
    {
      "epoch": 12.713596366732897,
      "grad_norm": 2.6379778385162354,
      "learning_rate": 8.729264831109851e-05,
      "loss": 0.03218862116336822,
      "step": 44790
    },
    {
      "epoch": 12.716434856656258,
      "grad_norm": 2.0847623348236084,
      "learning_rate": 8.728980982117514e-05,
      "loss": 0.026503151655197142,
      "step": 44800
    },
    {
      "epoch": 12.71927334657962,
      "grad_norm": 10.228126525878906,
      "learning_rate": 8.728697133125178e-05,
      "loss": 0.03470951318740845,
      "step": 44810
    },
    {
      "epoch": 12.72211183650298,
      "grad_norm": 2.4183998107910156,
      "learning_rate": 8.728413284132842e-05,
      "loss": 0.04810204803943634,
      "step": 44820
    },
    {
      "epoch": 12.724950326426342,
      "grad_norm": 4.971852779388428,
      "learning_rate": 8.728129435140505e-05,
      "loss": 0.027870452404022215,
      "step": 44830
    },
    {
      "epoch": 12.727788816349701,
      "grad_norm": 2.172386884689331,
      "learning_rate": 8.72784558614817e-05,
      "loss": 0.052520227432250974,
      "step": 44840
    },
    {
      "epoch": 12.730627306273062,
      "grad_norm": 9.188197135925293,
      "learning_rate": 8.727561737155834e-05,
      "loss": 0.03296519219875336,
      "step": 44850
    },
    {
      "epoch": 12.733465796196423,
      "grad_norm": 12.632057189941406,
      "learning_rate": 8.727277888163497e-05,
      "loss": 0.0348456084728241,
      "step": 44860
    },
    {
      "epoch": 12.736304286119784,
      "grad_norm": 3.0140838623046875,
      "learning_rate": 8.726994039171161e-05,
      "loss": 0.040859383344650266,
      "step": 44870
    },
    {
      "epoch": 12.739142776043145,
      "grad_norm": 6.8895769119262695,
      "learning_rate": 8.726710190178825e-05,
      "loss": 0.03838675320148468,
      "step": 44880
    },
    {
      "epoch": 12.741981265966507,
      "grad_norm": 3.9290289878845215,
      "learning_rate": 8.72642634118649e-05,
      "loss": 0.043990492820739746,
      "step": 44890
    },
    {
      "epoch": 12.744819755889866,
      "grad_norm": 9.954632759094238,
      "learning_rate": 8.726142492194152e-05,
      "loss": 0.04762568473815918,
      "step": 44900
    },
    {
      "epoch": 12.747658245813227,
      "grad_norm": 12.930060386657715,
      "learning_rate": 8.725858643201818e-05,
      "loss": 0.04320020079612732,
      "step": 44910
    },
    {
      "epoch": 12.750496735736588,
      "grad_norm": 1.8095616102218628,
      "learning_rate": 8.725574794209482e-05,
      "loss": 0.0823807418346405,
      "step": 44920
    },
    {
      "epoch": 12.75333522565995,
      "grad_norm": 5.546041965484619,
      "learning_rate": 8.725290945217145e-05,
      "loss": 0.0340006411075592,
      "step": 44930
    },
    {
      "epoch": 12.75617371558331,
      "grad_norm": 8.760685920715332,
      "learning_rate": 8.725007096224809e-05,
      "loss": 0.02766079604625702,
      "step": 44940
    },
    {
      "epoch": 12.75901220550667,
      "grad_norm": 4.659869194030762,
      "learning_rate": 8.724723247232473e-05,
      "loss": 0.034121465682983396,
      "step": 44950
    },
    {
      "epoch": 12.76185069543003,
      "grad_norm": 7.385076999664307,
      "learning_rate": 8.724439398240136e-05,
      "loss": 0.04381527006626129,
      "step": 44960
    },
    {
      "epoch": 12.764689185353392,
      "grad_norm": 12.8385009765625,
      "learning_rate": 8.724155549247801e-05,
      "loss": 0.03534000217914581,
      "step": 44970
    },
    {
      "epoch": 12.767527675276753,
      "grad_norm": 5.491786956787109,
      "learning_rate": 8.723871700255465e-05,
      "loss": 0.03210706412792206,
      "step": 44980
    },
    {
      "epoch": 12.770366165200114,
      "grad_norm": 9.597651481628418,
      "learning_rate": 8.723587851263128e-05,
      "loss": 0.05301475524902344,
      "step": 44990
    },
    {
      "epoch": 12.773204655123475,
      "grad_norm": 9.693765640258789,
      "learning_rate": 8.723304002270792e-05,
      "loss": 0.047547003626823424,
      "step": 45000
    },
    {
      "epoch": 12.773204655123475,
      "eval_accuracy": 0.9508488586507281,
      "eval_loss": 0.1479937732219696,
      "eval_runtime": 31.7114,
      "eval_samples_per_second": 495.942,
      "eval_steps_per_second": 7.757,
      "step": 45000
    },
    {
      "epoch": 12.776043145046835,
      "grad_norm": 4.376029014587402,
      "learning_rate": 8.723020153278456e-05,
      "loss": 0.04740391373634338,
      "step": 45010
    },
    {
      "epoch": 12.778881634970196,
      "grad_norm": 3.5137078762054443,
      "learning_rate": 8.72273630428612e-05,
      "loss": 0.054237252473831175,
      "step": 45020
    },
    {
      "epoch": 12.781720124893557,
      "grad_norm": 4.819000244140625,
      "learning_rate": 8.722452455293783e-05,
      "loss": 0.04695221185684204,
      "step": 45030
    },
    {
      "epoch": 12.784558614816918,
      "grad_norm": 11.225417137145996,
      "learning_rate": 8.722168606301449e-05,
      "loss": 0.030767554044723512,
      "step": 45040
    },
    {
      "epoch": 12.787397104740279,
      "grad_norm": 2.993690252304077,
      "learning_rate": 8.721884757309112e-05,
      "loss": 0.04000074565410614,
      "step": 45050
    },
    {
      "epoch": 12.790235594663638,
      "grad_norm": 2.7759902477264404,
      "learning_rate": 8.721600908316776e-05,
      "loss": 0.061340874433517455,
      "step": 45060
    },
    {
      "epoch": 12.793074084587,
      "grad_norm": 12.015202522277832,
      "learning_rate": 8.72131705932444e-05,
      "loss": 0.06872159838676453,
      "step": 45070
    },
    {
      "epoch": 12.79591257451036,
      "grad_norm": 13.68930435180664,
      "learning_rate": 8.721033210332104e-05,
      "loss": 0.04538925886154175,
      "step": 45080
    },
    {
      "epoch": 12.798751064433722,
      "grad_norm": 16.635292053222656,
      "learning_rate": 8.720749361339767e-05,
      "loss": 0.08499961495399475,
      "step": 45090
    },
    {
      "epoch": 12.801589554357083,
      "grad_norm": 1.5720669031143188,
      "learning_rate": 8.720465512347431e-05,
      "loss": 0.07012171745300293,
      "step": 45100
    },
    {
      "epoch": 12.804428044280442,
      "grad_norm": 1.3354732990264893,
      "learning_rate": 8.720181663355096e-05,
      "loss": 0.025925347208976747,
      "step": 45110
    },
    {
      "epoch": 12.807266534203803,
      "grad_norm": 4.402794361114502,
      "learning_rate": 8.719897814362759e-05,
      "loss": 0.03301866352558136,
      "step": 45120
    },
    {
      "epoch": 12.810105024127164,
      "grad_norm": 1.6817296743392944,
      "learning_rate": 8.719613965370423e-05,
      "loss": 0.032939016819000244,
      "step": 45130
    },
    {
      "epoch": 12.812943514050525,
      "grad_norm": 14.497953414916992,
      "learning_rate": 8.719330116378087e-05,
      "loss": 0.04649432897567749,
      "step": 45140
    },
    {
      "epoch": 12.815782003973887,
      "grad_norm": 7.468667030334473,
      "learning_rate": 8.71904626738575e-05,
      "loss": 0.028683766722679138,
      "step": 45150
    },
    {
      "epoch": 12.818620493897246,
      "grad_norm": 6.453930854797363,
      "learning_rate": 8.718762418393414e-05,
      "loss": 0.06180890202522278,
      "step": 45160
    },
    {
      "epoch": 12.821458983820607,
      "grad_norm": 0.8599632382392883,
      "learning_rate": 8.71847856940108e-05,
      "loss": 0.02617986798286438,
      "step": 45170
    },
    {
      "epoch": 12.824297473743968,
      "grad_norm": 7.782673358917236,
      "learning_rate": 8.718194720408743e-05,
      "loss": 0.04447529911994934,
      "step": 45180
    },
    {
      "epoch": 12.82713596366733,
      "grad_norm": 3.7080676555633545,
      "learning_rate": 8.717910871416407e-05,
      "loss": 0.04249391853809357,
      "step": 45190
    },
    {
      "epoch": 12.82997445359069,
      "grad_norm": 1.1791934967041016,
      "learning_rate": 8.717627022424071e-05,
      "loss": 0.02329118549823761,
      "step": 45200
    },
    {
      "epoch": 12.83281294351405,
      "grad_norm": 9.215568542480469,
      "learning_rate": 8.717343173431735e-05,
      "loss": 0.03221278488636017,
      "step": 45210
    },
    {
      "epoch": 12.83565143343741,
      "grad_norm": 2.603409767150879,
      "learning_rate": 8.717059324439398e-05,
      "loss": 0.033879855275154115,
      "step": 45220
    },
    {
      "epoch": 12.838489923360772,
      "grad_norm": 11.32485294342041,
      "learning_rate": 8.716775475447062e-05,
      "loss": 0.041959017515182495,
      "step": 45230
    },
    {
      "epoch": 12.841328413284133,
      "grad_norm": 6.9285807609558105,
      "learning_rate": 8.716491626454727e-05,
      "loss": 0.03080936670303345,
      "step": 45240
    },
    {
      "epoch": 12.844166903207494,
      "grad_norm": 0.6139256358146667,
      "learning_rate": 8.71620777746239e-05,
      "loss": 0.04797856211662292,
      "step": 45250
    },
    {
      "epoch": 12.847005393130855,
      "grad_norm": 6.193779945373535,
      "learning_rate": 8.715923928470054e-05,
      "loss": 0.03794891238212585,
      "step": 45260
    },
    {
      "epoch": 12.849843883054215,
      "grad_norm": 9.690537452697754,
      "learning_rate": 8.715640079477719e-05,
      "loss": 0.044877105951309205,
      "step": 45270
    },
    {
      "epoch": 12.852682372977576,
      "grad_norm": 9.704392433166504,
      "learning_rate": 8.715356230485381e-05,
      "loss": 0.03184947371482849,
      "step": 45280
    },
    {
      "epoch": 12.855520862900937,
      "grad_norm": 5.39846658706665,
      "learning_rate": 8.715072381493045e-05,
      "loss": 0.04482846856117249,
      "step": 45290
    },
    {
      "epoch": 12.858359352824298,
      "grad_norm": 3.099754810333252,
      "learning_rate": 8.714788532500711e-05,
      "loss": 0.03206309378147125,
      "step": 45300
    },
    {
      "epoch": 12.861197842747659,
      "grad_norm": 3.275723934173584,
      "learning_rate": 8.714504683508374e-05,
      "loss": 0.04932217299938202,
      "step": 45310
    },
    {
      "epoch": 12.864036332671018,
      "grad_norm": 5.542270660400391,
      "learning_rate": 8.714220834516038e-05,
      "loss": 0.025083592534065245,
      "step": 45320
    },
    {
      "epoch": 12.86687482259438,
      "grad_norm": 16.415597915649414,
      "learning_rate": 8.713936985523702e-05,
      "loss": 0.05791771411895752,
      "step": 45330
    },
    {
      "epoch": 12.86971331251774,
      "grad_norm": 11.34880542755127,
      "learning_rate": 8.713653136531366e-05,
      "loss": 0.036676087975502016,
      "step": 45340
    },
    {
      "epoch": 12.872551802441102,
      "grad_norm": 6.134761810302734,
      "learning_rate": 8.713369287539029e-05,
      "loss": 0.03904784321784973,
      "step": 45350
    },
    {
      "epoch": 12.875390292364463,
      "grad_norm": 8.232940673828125,
      "learning_rate": 8.713085438546693e-05,
      "loss": 0.044435751438140866,
      "step": 45360
    },
    {
      "epoch": 12.878228782287822,
      "grad_norm": 13.713446617126465,
      "learning_rate": 8.712801589554359e-05,
      "loss": 0.0583861231803894,
      "step": 45370
    },
    {
      "epoch": 12.881067272211183,
      "grad_norm": 12.268309593200684,
      "learning_rate": 8.712517740562021e-05,
      "loss": 0.0572661817073822,
      "step": 45380
    },
    {
      "epoch": 12.883905762134544,
      "grad_norm": 1.7122182846069336,
      "learning_rate": 8.712233891569685e-05,
      "loss": 0.04821889400482178,
      "step": 45390
    },
    {
      "epoch": 12.886744252057905,
      "grad_norm": 9.639969825744629,
      "learning_rate": 8.71195004257735e-05,
      "loss": 0.04056214690208435,
      "step": 45400
    },
    {
      "epoch": 12.889582741981267,
      "grad_norm": 3.87674617767334,
      "learning_rate": 8.711666193585012e-05,
      "loss": 0.015631312131881715,
      "step": 45410
    },
    {
      "epoch": 12.892421231904628,
      "grad_norm": 5.597787857055664,
      "learning_rate": 8.711382344592677e-05,
      "loss": 0.042028391361236574,
      "step": 45420
    },
    {
      "epoch": 12.895259721827987,
      "grad_norm": 5.725367546081543,
      "learning_rate": 8.711098495600341e-05,
      "loss": 0.025868400931358337,
      "step": 45430
    },
    {
      "epoch": 12.898098211751348,
      "grad_norm": 1.6748632192611694,
      "learning_rate": 8.710814646608005e-05,
      "loss": 0.05726107954978943,
      "step": 45440
    },
    {
      "epoch": 12.90093670167471,
      "grad_norm": 7.724864482879639,
      "learning_rate": 8.710530797615669e-05,
      "loss": 0.028647717833518983,
      "step": 45450
    },
    {
      "epoch": 12.90377519159807,
      "grad_norm": 6.866396903991699,
      "learning_rate": 8.710246948623333e-05,
      "loss": 0.04249510169029236,
      "step": 45460
    },
    {
      "epoch": 12.906613681521431,
      "grad_norm": 3.0799500942230225,
      "learning_rate": 8.709963099630997e-05,
      "loss": 0.05019757151603699,
      "step": 45470
    },
    {
      "epoch": 12.90945217144479,
      "grad_norm": 3.3673219680786133,
      "learning_rate": 8.70967925063866e-05,
      "loss": 0.03434927761554718,
      "step": 45480
    },
    {
      "epoch": 12.912290661368152,
      "grad_norm": 1.4446579217910767,
      "learning_rate": 8.709395401646324e-05,
      "loss": 0.03518008589744568,
      "step": 45490
    },
    {
      "epoch": 12.915129151291513,
      "grad_norm": 11.165140151977539,
      "learning_rate": 8.70911155265399e-05,
      "loss": 0.03347097933292389,
      "step": 45500
    },
    {
      "epoch": 12.915129151291513,
      "eval_accuracy": 0.9593056526991798,
      "eval_loss": 0.1263011395931244,
      "eval_runtime": 31.3449,
      "eval_samples_per_second": 501.741,
      "eval_steps_per_second": 7.848,
      "step": 45500
    },
    {
      "epoch": 12.917967641214874,
      "grad_norm": 3.674799680709839,
      "learning_rate": 8.708856088560886e-05,
      "loss": 0.06423990130424499,
      "step": 45510
    },
    {
      "epoch": 12.920806131138235,
      "grad_norm": 6.9470062255859375,
      "learning_rate": 8.70857223956855e-05,
      "loss": 0.032919925451278684,
      "step": 45520
    },
    {
      "epoch": 12.923644621061595,
      "grad_norm": 0.6750406622886658,
      "learning_rate": 8.708288390576213e-05,
      "loss": 0.027924519777297974,
      "step": 45530
    },
    {
      "epoch": 12.926483110984956,
      "grad_norm": 2.5715408325195312,
      "learning_rate": 8.708004541583877e-05,
      "loss": 0.03935595750808716,
      "step": 45540
    },
    {
      "epoch": 12.929321600908317,
      "grad_norm": 5.05348539352417,
      "learning_rate": 8.707720692591543e-05,
      "loss": 0.0449282169342041,
      "step": 45550
    },
    {
      "epoch": 12.932160090831678,
      "grad_norm": 5.564386367797852,
      "learning_rate": 8.707436843599206e-05,
      "loss": 0.03764061629772186,
      "step": 45560
    },
    {
      "epoch": 12.934998580755039,
      "grad_norm": 21.975465774536133,
      "learning_rate": 8.70715299460687e-05,
      "loss": 0.052472901344299314,
      "step": 45570
    },
    {
      "epoch": 12.937837070678398,
      "grad_norm": 11.615373611450195,
      "learning_rate": 8.706869145614534e-05,
      "loss": 0.04490956962108612,
      "step": 45580
    },
    {
      "epoch": 12.94067556060176,
      "grad_norm": 5.858719348907471,
      "learning_rate": 8.706585296622197e-05,
      "loss": 0.038946935534477235,
      "step": 45590
    },
    {
      "epoch": 12.94351405052512,
      "grad_norm": 10.555013656616211,
      "learning_rate": 8.706301447629861e-05,
      "loss": 0.04604542851448059,
      "step": 45600
    },
    {
      "epoch": 12.946352540448482,
      "grad_norm": 12.783527374267578,
      "learning_rate": 8.706017598637525e-05,
      "loss": 0.04718633890151978,
      "step": 45610
    },
    {
      "epoch": 12.949191030371843,
      "grad_norm": 7.371231555938721,
      "learning_rate": 8.705733749645189e-05,
      "loss": 0.05113707780838013,
      "step": 45620
    },
    {
      "epoch": 12.952029520295202,
      "grad_norm": 7.177131652832031,
      "learning_rate": 8.705449900652853e-05,
      "loss": 0.029880782961845397,
      "step": 45630
    },
    {
      "epoch": 12.954868010218563,
      "grad_norm": 6.594571113586426,
      "learning_rate": 8.705166051660517e-05,
      "loss": 0.04537354707717896,
      "step": 45640
    },
    {
      "epoch": 12.957706500141924,
      "grad_norm": 3.1812551021575928,
      "learning_rate": 8.704882202668181e-05,
      "loss": 0.05719947814941406,
      "step": 45650
    },
    {
      "epoch": 12.960544990065285,
      "grad_norm": 7.866095066070557,
      "learning_rate": 8.704598353675844e-05,
      "loss": 0.0355503648519516,
      "step": 45660
    },
    {
      "epoch": 12.963383479988646,
      "grad_norm": 17.68269157409668,
      "learning_rate": 8.704314504683508e-05,
      "loss": 0.03570097386837005,
      "step": 45670
    },
    {
      "epoch": 12.966221969912008,
      "grad_norm": 1.6464214324951172,
      "learning_rate": 8.704030655691174e-05,
      "loss": 0.04330264925956726,
      "step": 45680
    },
    {
      "epoch": 12.969060459835367,
      "grad_norm": 1.3372726440429688,
      "learning_rate": 8.703746806698837e-05,
      "loss": 0.018665172159671783,
      "step": 45690
    },
    {
      "epoch": 12.971898949758728,
      "grad_norm": 10.106221199035645,
      "learning_rate": 8.703462957706501e-05,
      "loss": 0.031829485297203065,
      "step": 45700
    },
    {
      "epoch": 12.97473743968209,
      "grad_norm": 13.196824073791504,
      "learning_rate": 8.703179108714165e-05,
      "loss": 0.03434259295463562,
      "step": 45710
    },
    {
      "epoch": 12.97757592960545,
      "grad_norm": 3.910924196243286,
      "learning_rate": 8.702895259721828e-05,
      "loss": 0.05792022347450256,
      "step": 45720
    },
    {
      "epoch": 12.980414419528811,
      "grad_norm": 17.774368286132812,
      "learning_rate": 8.702611410729492e-05,
      "loss": 0.07785875797271728,
      "step": 45730
    },
    {
      "epoch": 12.98325290945217,
      "grad_norm": 8.901176452636719,
      "learning_rate": 8.702327561737156e-05,
      "loss": 0.037799912691116336,
      "step": 45740
    },
    {
      "epoch": 12.986091399375532,
      "grad_norm": 13.765517234802246,
      "learning_rate": 8.70204371274482e-05,
      "loss": 0.043500500917434695,
      "step": 45750
    },
    {
      "epoch": 12.988929889298893,
      "grad_norm": 7.7782745361328125,
      "learning_rate": 8.701759863752484e-05,
      "loss": 0.044987809658050534,
      "step": 45760
    },
    {
      "epoch": 12.991768379222254,
      "grad_norm": 13.591089248657227,
      "learning_rate": 8.701476014760148e-05,
      "loss": 0.03518059253692627,
      "step": 45770
    },
    {
      "epoch": 12.994606869145615,
      "grad_norm": 4.035161972045898,
      "learning_rate": 8.701192165767813e-05,
      "loss": 0.028557729721069337,
      "step": 45780
    },
    {
      "epoch": 12.997445359068976,
      "grad_norm": 4.575911521911621,
      "learning_rate": 8.700908316775475e-05,
      "loss": 0.020898987352848054,
      "step": 45790
    },
    {
      "epoch": 13.000283848992336,
      "grad_norm": 8.753043174743652,
      "learning_rate": 8.70062446778314e-05,
      "loss": 0.025048828125,
      "step": 45800
    },
    {
      "epoch": 13.003122338915697,
      "grad_norm": 3.86434268951416,
      "learning_rate": 8.700340618790804e-05,
      "loss": 0.030737340450286865,
      "step": 45810
    },
    {
      "epoch": 13.005960828839058,
      "grad_norm": 1.5299665927886963,
      "learning_rate": 8.700056769798468e-05,
      "loss": 0.019111914932727812,
      "step": 45820
    },
    {
      "epoch": 13.008799318762419,
      "grad_norm": 8.415980339050293,
      "learning_rate": 8.699772920806132e-05,
      "loss": 0.03076198399066925,
      "step": 45830
    },
    {
      "epoch": 13.01163780868578,
      "grad_norm": 11.649621963500977,
      "learning_rate": 8.699489071813796e-05,
      "loss": 0.03850152194499969,
      "step": 45840
    },
    {
      "epoch": 13.01447629860914,
      "grad_norm": 11.341511726379395,
      "learning_rate": 8.699205222821459e-05,
      "loss": 0.024037735164165498,
      "step": 45850
    },
    {
      "epoch": 13.0173147885325,
      "grad_norm": 6.691775798797607,
      "learning_rate": 8.698921373829123e-05,
      "loss": 0.022889155149459838,
      "step": 45860
    },
    {
      "epoch": 13.020153278455862,
      "grad_norm": 5.299643039703369,
      "learning_rate": 8.698637524836787e-05,
      "loss": 0.02528354525566101,
      "step": 45870
    },
    {
      "epoch": 13.022991768379223,
      "grad_norm": 5.357428073883057,
      "learning_rate": 8.698353675844451e-05,
      "loss": 0.030365380644798278,
      "step": 45880
    },
    {
      "epoch": 13.025830258302584,
      "grad_norm": 17.28185272216797,
      "learning_rate": 8.698069826852115e-05,
      "loss": 0.032807016372680665,
      "step": 45890
    },
    {
      "epoch": 13.028668748225943,
      "grad_norm": 10.43936824798584,
      "learning_rate": 8.69778597785978e-05,
      "loss": 0.03723643124103546,
      "step": 45900
    },
    {
      "epoch": 13.031507238149304,
      "grad_norm": 3.7319672107696533,
      "learning_rate": 8.697502128867444e-05,
      "loss": 0.03306630849838257,
      "step": 45910
    },
    {
      "epoch": 13.034345728072665,
      "grad_norm": 3.9782063961029053,
      "learning_rate": 8.697218279875106e-05,
      "loss": 0.016803576052188872,
      "step": 45920
    },
    {
      "epoch": 13.037184217996026,
      "grad_norm": 4.513489723205566,
      "learning_rate": 8.69693443088277e-05,
      "loss": 0.037658748030662534,
      "step": 45930
    },
    {
      "epoch": 13.040022707919388,
      "grad_norm": 5.186512470245361,
      "learning_rate": 8.696650581890435e-05,
      "loss": 0.023856021463871002,
      "step": 45940
    },
    {
      "epoch": 13.042861197842747,
      "grad_norm": 2.8630008697509766,
      "learning_rate": 8.696366732898099e-05,
      "loss": 0.029960185289382935,
      "step": 45950
    },
    {
      "epoch": 13.045699687766108,
      "grad_norm": 5.441481113433838,
      "learning_rate": 8.696082883905763e-05,
      "loss": 0.031141310930252075,
      "step": 45960
    },
    {
      "epoch": 13.048538177689469,
      "grad_norm": 3.809981107711792,
      "learning_rate": 8.695799034913427e-05,
      "loss": 0.035477432608604434,
      "step": 45970
    },
    {
      "epoch": 13.05137666761283,
      "grad_norm": 1.3065361976623535,
      "learning_rate": 8.69551518592109e-05,
      "loss": 0.01905675381422043,
      "step": 45980
    },
    {
      "epoch": 13.054215157536191,
      "grad_norm": 11.69388484954834,
      "learning_rate": 8.695231336928754e-05,
      "loss": 0.04160212874412537,
      "step": 45990
    },
    {
      "epoch": 13.05705364745955,
      "grad_norm": 9.860189437866211,
      "learning_rate": 8.694947487936418e-05,
      "loss": 0.03494201600551605,
      "step": 46000
    },
    {
      "epoch": 13.05705364745955,
      "eval_accuracy": 0.9600050867934126,
      "eval_loss": 0.12102089822292328,
      "eval_runtime": 31.4676,
      "eval_samples_per_second": 499.785,
      "eval_steps_per_second": 7.818,
      "step": 46000
    },
    {
      "epoch": 13.059892137382912,
      "grad_norm": 13.886919975280762,
      "learning_rate": 8.694663638944082e-05,
      "loss": 0.037542781233787535,
      "step": 46010
    },
    {
      "epoch": 13.062730627306273,
      "grad_norm": 2.0131142139434814,
      "learning_rate": 8.694379789951746e-05,
      "loss": 0.016535870730876923,
      "step": 46020
    },
    {
      "epoch": 13.065569117229634,
      "grad_norm": 7.188528537750244,
      "learning_rate": 8.69409594095941e-05,
      "loss": 0.027523231506347657,
      "step": 46030
    },
    {
      "epoch": 13.068407607152995,
      "grad_norm": 4.929324626922607,
      "learning_rate": 8.693812091967073e-05,
      "loss": 0.023347745835781097,
      "step": 46040
    },
    {
      "epoch": 13.071246097076356,
      "grad_norm": 10.391345024108887,
      "learning_rate": 8.693528242974738e-05,
      "loss": 0.026271986961364745,
      "step": 46050
    },
    {
      "epoch": 13.074084586999716,
      "grad_norm": 9.589155197143555,
      "learning_rate": 8.693244393982402e-05,
      "loss": 0.0391389787197113,
      "step": 46060
    },
    {
      "epoch": 13.076923076923077,
      "grad_norm": 5.854538440704346,
      "learning_rate": 8.692960544990066e-05,
      "loss": 0.02978052794933319,
      "step": 46070
    },
    {
      "epoch": 13.079761566846438,
      "grad_norm": 1.6398125886917114,
      "learning_rate": 8.69267669599773e-05,
      "loss": 0.02386342138051987,
      "step": 46080
    },
    {
      "epoch": 13.082600056769799,
      "grad_norm": 8.533074378967285,
      "learning_rate": 8.692392847005394e-05,
      "loss": 0.04604070782661438,
      "step": 46090
    },
    {
      "epoch": 13.08543854669316,
      "grad_norm": 8.954219818115234,
      "learning_rate": 8.692108998013058e-05,
      "loss": 0.03578729331493378,
      "step": 46100
    },
    {
      "epoch": 13.08827703661652,
      "grad_norm": 6.10044002532959,
      "learning_rate": 8.691825149020721e-05,
      "loss": 0.027187839150428772,
      "step": 46110
    },
    {
      "epoch": 13.09111552653988,
      "grad_norm": 10.100062370300293,
      "learning_rate": 8.691541300028385e-05,
      "loss": 0.021938730776309968,
      "step": 46120
    },
    {
      "epoch": 13.093954016463242,
      "grad_norm": 1.8479245901107788,
      "learning_rate": 8.691257451036049e-05,
      "loss": 0.04361025094985962,
      "step": 46130
    },
    {
      "epoch": 13.096792506386603,
      "grad_norm": 3.1001205444335938,
      "learning_rate": 8.690973602043712e-05,
      "loss": 0.0469401091337204,
      "step": 46140
    },
    {
      "epoch": 13.099630996309964,
      "grad_norm": 11.211882591247559,
      "learning_rate": 8.690689753051378e-05,
      "loss": 0.041463214159011844,
      "step": 46150
    },
    {
      "epoch": 13.102469486233323,
      "grad_norm": 9.55882740020752,
      "learning_rate": 8.690405904059042e-05,
      "loss": 0.029076778888702394,
      "step": 46160
    },
    {
      "epoch": 13.105307976156684,
      "grad_norm": 2.4896345138549805,
      "learning_rate": 8.690122055066704e-05,
      "loss": 0.03126168251037598,
      "step": 46170
    },
    {
      "epoch": 13.108146466080045,
      "grad_norm": 4.450166702270508,
      "learning_rate": 8.689838206074369e-05,
      "loss": 0.029432645440101622,
      "step": 46180
    },
    {
      "epoch": 13.110984956003406,
      "grad_norm": 7.018934726715088,
      "learning_rate": 8.689554357082033e-05,
      "loss": 0.024624615907669067,
      "step": 46190
    },
    {
      "epoch": 13.113823445926768,
      "grad_norm": 4.068532943725586,
      "learning_rate": 8.689270508089697e-05,
      "loss": 0.017405125498771667,
      "step": 46200
    },
    {
      "epoch": 13.116661935850127,
      "grad_norm": 24.175397872924805,
      "learning_rate": 8.68898665909736e-05,
      "loss": 0.03965692520141602,
      "step": 46210
    },
    {
      "epoch": 13.119500425773488,
      "grad_norm": 3.6043851375579834,
      "learning_rate": 8.688702810105025e-05,
      "loss": 0.03555071651935578,
      "step": 46220
    },
    {
      "epoch": 13.122338915696849,
      "grad_norm": 6.531484603881836,
      "learning_rate": 8.688418961112689e-05,
      "loss": 0.0271712988615036,
      "step": 46230
    },
    {
      "epoch": 13.12517740562021,
      "grad_norm": 10.330556869506836,
      "learning_rate": 8.688135112120352e-05,
      "loss": 0.036047860980033875,
      "step": 46240
    },
    {
      "epoch": 13.128015895543571,
      "grad_norm": 7.164229869842529,
      "learning_rate": 8.687851263128016e-05,
      "loss": 0.03753184676170349,
      "step": 46250
    },
    {
      "epoch": 13.130854385466932,
      "grad_norm": 9.894229888916016,
      "learning_rate": 8.68756741413568e-05,
      "loss": 0.044302555918693545,
      "step": 46260
    },
    {
      "epoch": 13.133692875390292,
      "grad_norm": 9.697216033935547,
      "learning_rate": 8.687283565143343e-05,
      "loss": 0.03176181614398956,
      "step": 46270
    },
    {
      "epoch": 13.136531365313653,
      "grad_norm": 12.86292552947998,
      "learning_rate": 8.686999716151009e-05,
      "loss": 0.027163279056549073,
      "step": 46280
    },
    {
      "epoch": 13.139369855237014,
      "grad_norm": 15.37137508392334,
      "learning_rate": 8.686715867158673e-05,
      "loss": 0.05186057090759277,
      "step": 46290
    },
    {
      "epoch": 13.142208345160375,
      "grad_norm": 5.393747329711914,
      "learning_rate": 8.686432018166336e-05,
      "loss": 0.03587070107460022,
      "step": 46300
    },
    {
      "epoch": 13.145046835083736,
      "grad_norm": 7.149489402770996,
      "learning_rate": 8.686148169174e-05,
      "loss": 0.03435201644897461,
      "step": 46310
    },
    {
      "epoch": 13.147885325007096,
      "grad_norm": 6.243102073669434,
      "learning_rate": 8.685864320181664e-05,
      "loss": 0.03733316957950592,
      "step": 46320
    },
    {
      "epoch": 13.150723814930457,
      "grad_norm": 10.496167182922363,
      "learning_rate": 8.685580471189328e-05,
      "loss": 0.05221510529518127,
      "step": 46330
    },
    {
      "epoch": 13.153562304853818,
      "grad_norm": 13.295903205871582,
      "learning_rate": 8.685296622196991e-05,
      "loss": 0.03955149352550506,
      "step": 46340
    },
    {
      "epoch": 13.156400794777179,
      "grad_norm": 6.255700588226318,
      "learning_rate": 8.685012773204656e-05,
      "loss": 0.03965249061584473,
      "step": 46350
    },
    {
      "epoch": 13.15923928470054,
      "grad_norm": 2.5287551879882812,
      "learning_rate": 8.68472892421232e-05,
      "loss": 0.03453528881072998,
      "step": 46360
    },
    {
      "epoch": 13.1620777746239,
      "grad_norm": 5.715973854064941,
      "learning_rate": 8.684445075219983e-05,
      "loss": 0.025282251834869384,
      "step": 46370
    },
    {
      "epoch": 13.16491626454726,
      "grad_norm": 7.935683250427246,
      "learning_rate": 8.684161226227647e-05,
      "loss": 0.033587193489074706,
      "step": 46380
    },
    {
      "epoch": 13.167754754470621,
      "grad_norm": 8.875100135803223,
      "learning_rate": 8.683877377235311e-05,
      "loss": 0.059543246030807497,
      "step": 46390
    },
    {
      "epoch": 13.170593244393983,
      "grad_norm": 2.1696767807006836,
      "learning_rate": 8.683593528242974e-05,
      "loss": 0.015411739051342011,
      "step": 46400
    },
    {
      "epoch": 13.173431734317344,
      "grad_norm": 8.039100646972656,
      "learning_rate": 8.683309679250638e-05,
      "loss": 0.026804310083389283,
      "step": 46410
    },
    {
      "epoch": 13.176270224240703,
      "grad_norm": 1.1770687103271484,
      "learning_rate": 8.683025830258304e-05,
      "loss": 0.029194334149360658,
      "step": 46420
    },
    {
      "epoch": 13.179108714164064,
      "grad_norm": 16.008943557739258,
      "learning_rate": 8.682741981265967e-05,
      "loss": 0.03047935664653778,
      "step": 46430
    },
    {
      "epoch": 13.181947204087425,
      "grad_norm": 6.653017520904541,
      "learning_rate": 8.682458132273631e-05,
      "loss": 0.03581974804401398,
      "step": 46440
    },
    {
      "epoch": 13.184785694010786,
      "grad_norm": 3.258111000061035,
      "learning_rate": 8.682174283281295e-05,
      "loss": 0.0355497270822525,
      "step": 46450
    },
    {
      "epoch": 13.187624183934147,
      "grad_norm": 4.663342475891113,
      "learning_rate": 8.681890434288959e-05,
      "loss": 0.02163380980491638,
      "step": 46460
    },
    {
      "epoch": 13.190462673857509,
      "grad_norm": 9.256166458129883,
      "learning_rate": 8.681606585296622e-05,
      "loss": 0.015953955054283143,
      "step": 46470
    },
    {
      "epoch": 13.193301163780868,
      "grad_norm": 9.308165550231934,
      "learning_rate": 8.681322736304287e-05,
      "loss": 0.021913789212703705,
      "step": 46480
    },
    {
      "epoch": 13.196139653704229,
      "grad_norm": 5.865832805633545,
      "learning_rate": 8.681038887311951e-05,
      "loss": 0.021701508760452272,
      "step": 46490
    },
    {
      "epoch": 13.19897814362759,
      "grad_norm": 3.9808759689331055,
      "learning_rate": 8.680755038319614e-05,
      "loss": 0.032222169637680056,
      "step": 46500
    },
    {
      "epoch": 13.19897814362759,
      "eval_accuracy": 0.9576524448400839,
      "eval_loss": 0.13142837584018707,
      "eval_runtime": 31.4104,
      "eval_samples_per_second": 500.693,
      "eval_steps_per_second": 7.832,
      "step": 46500
    },
    {
      "epoch": 13.201816633550951,
      "grad_norm": 4.899045467376709,
      "learning_rate": 8.680471189327278e-05,
      "loss": 0.048612141609191896,
      "step": 46510
    },
    {
      "epoch": 13.204655123474312,
      "grad_norm": 5.104890823364258,
      "learning_rate": 8.680187340334942e-05,
      "loss": 0.02812759280204773,
      "step": 46520
    },
    {
      "epoch": 13.207493613397672,
      "grad_norm": 0.8021270036697388,
      "learning_rate": 8.679903491342605e-05,
      "loss": 0.014854462444782257,
      "step": 46530
    },
    {
      "epoch": 13.210332103321033,
      "grad_norm": 2.0054574012756348,
      "learning_rate": 8.67961964235027e-05,
      "loss": 0.035123175382614134,
      "step": 46540
    },
    {
      "epoch": 13.213170593244394,
      "grad_norm": 14.444690704345703,
      "learning_rate": 8.679335793357935e-05,
      "loss": 0.03248676061630249,
      "step": 46550
    },
    {
      "epoch": 13.216009083167755,
      "grad_norm": 10.37210464477539,
      "learning_rate": 8.679051944365598e-05,
      "loss": 0.032960107922554015,
      "step": 46560
    },
    {
      "epoch": 13.218847573091116,
      "grad_norm": 8.02227783203125,
      "learning_rate": 8.678768095373262e-05,
      "loss": 0.025931930541992186,
      "step": 46570
    },
    {
      "epoch": 13.221686063014475,
      "grad_norm": 1.6116859912872314,
      "learning_rate": 8.678484246380926e-05,
      "loss": 0.025424587726593017,
      "step": 46580
    },
    {
      "epoch": 13.224524552937837,
      "grad_norm": 14.00739574432373,
      "learning_rate": 8.67820039738859e-05,
      "loss": 0.036745718121528624,
      "step": 46590
    },
    {
      "epoch": 13.227363042861198,
      "grad_norm": 3.6275362968444824,
      "learning_rate": 8.677916548396253e-05,
      "loss": 0.03769190013408661,
      "step": 46600
    },
    {
      "epoch": 13.230201532784559,
      "grad_norm": 7.274720191955566,
      "learning_rate": 8.677632699403917e-05,
      "loss": 0.017800626158714295,
      "step": 46610
    },
    {
      "epoch": 13.23304002270792,
      "grad_norm": 20.06517791748047,
      "learning_rate": 8.677348850411583e-05,
      "loss": 0.06622192859649659,
      "step": 46620
    },
    {
      "epoch": 13.235878512631281,
      "grad_norm": 9.146626472473145,
      "learning_rate": 8.677065001419245e-05,
      "loss": 0.022119684517383574,
      "step": 46630
    },
    {
      "epoch": 13.23871700255464,
      "grad_norm": 10.979401588439941,
      "learning_rate": 8.67678115242691e-05,
      "loss": 0.033708488941192626,
      "step": 46640
    },
    {
      "epoch": 13.241555492478001,
      "grad_norm": 8.480527877807617,
      "learning_rate": 8.676497303434574e-05,
      "loss": 0.031891119480133054,
      "step": 46650
    },
    {
      "epoch": 13.244393982401363,
      "grad_norm": 3.1861956119537354,
      "learning_rate": 8.676213454442236e-05,
      "loss": 0.02654237449169159,
      "step": 46660
    },
    {
      "epoch": 13.247232472324724,
      "grad_norm": 1.1386229991912842,
      "learning_rate": 8.6759296054499e-05,
      "loss": 0.029223036766052247,
      "step": 46670
    },
    {
      "epoch": 13.250070962248085,
      "grad_norm": 8.858532905578613,
      "learning_rate": 8.675645756457566e-05,
      "loss": 0.03646268248558045,
      "step": 46680
    },
    {
      "epoch": 13.252909452171444,
      "grad_norm": 7.6891632080078125,
      "learning_rate": 8.675361907465229e-05,
      "loss": 0.04000713527202606,
      "step": 46690
    },
    {
      "epoch": 13.255747942094805,
      "grad_norm": 8.544937133789062,
      "learning_rate": 8.675078058472893e-05,
      "loss": 0.04307003319263458,
      "step": 46700
    },
    {
      "epoch": 13.258586432018166,
      "grad_norm": 4.729654788970947,
      "learning_rate": 8.674794209480557e-05,
      "loss": 0.0224888414144516,
      "step": 46710
    },
    {
      "epoch": 13.261424921941527,
      "grad_norm": 10.592219352722168,
      "learning_rate": 8.674510360488221e-05,
      "loss": 0.017426466941833495,
      "step": 46720
    },
    {
      "epoch": 13.264263411864889,
      "grad_norm": 12.848734855651855,
      "learning_rate": 8.674226511495884e-05,
      "loss": 0.02888982892036438,
      "step": 46730
    },
    {
      "epoch": 13.267101901788248,
      "grad_norm": 5.484411716461182,
      "learning_rate": 8.673942662503548e-05,
      "loss": 0.03522112965583801,
      "step": 46740
    },
    {
      "epoch": 13.269940391711609,
      "grad_norm": 12.630756378173828,
      "learning_rate": 8.673658813511214e-05,
      "loss": 0.04130620062351227,
      "step": 46750
    },
    {
      "epoch": 13.27277888163497,
      "grad_norm": 8.3544921875,
      "learning_rate": 8.673374964518876e-05,
      "loss": 0.0466865062713623,
      "step": 46760
    },
    {
      "epoch": 13.275617371558331,
      "grad_norm": 3.942180633544922,
      "learning_rate": 8.67309111552654e-05,
      "loss": 0.03530599772930145,
      "step": 46770
    },
    {
      "epoch": 13.278455861481692,
      "grad_norm": 12.378708839416504,
      "learning_rate": 8.672807266534205e-05,
      "loss": 0.052664142847061154,
      "step": 46780
    },
    {
      "epoch": 13.281294351405052,
      "grad_norm": 3.9495184421539307,
      "learning_rate": 8.672523417541867e-05,
      "loss": 0.031207001209259032,
      "step": 46790
    },
    {
      "epoch": 13.284132841328413,
      "grad_norm": 0.7871926426887512,
      "learning_rate": 8.672239568549532e-05,
      "loss": 0.01120012104511261,
      "step": 46800
    },
    {
      "epoch": 13.286971331251774,
      "grad_norm": 2.352360963821411,
      "learning_rate": 8.671955719557196e-05,
      "loss": 0.02702072560787201,
      "step": 46810
    },
    {
      "epoch": 13.289809821175135,
      "grad_norm": 2.5599679946899414,
      "learning_rate": 8.67167187056486e-05,
      "loss": 0.04142013490200043,
      "step": 46820
    },
    {
      "epoch": 13.292648311098496,
      "grad_norm": 2.9239768981933594,
      "learning_rate": 8.671388021572524e-05,
      "loss": 0.022765138745307924,
      "step": 46830
    },
    {
      "epoch": 13.295486801021857,
      "grad_norm": 4.48128604888916,
      "learning_rate": 8.671104172580188e-05,
      "loss": 0.04166404902935028,
      "step": 46840
    },
    {
      "epoch": 13.298325290945217,
      "grad_norm": 6.070557117462158,
      "learning_rate": 8.670820323587852e-05,
      "loss": 0.047522330284118654,
      "step": 46850
    },
    {
      "epoch": 13.301163780868578,
      "grad_norm": 3.784759521484375,
      "learning_rate": 8.670536474595515e-05,
      "loss": 0.025988921523094177,
      "step": 46860
    },
    {
      "epoch": 13.304002270791939,
      "grad_norm": 2.189690351486206,
      "learning_rate": 8.670252625603179e-05,
      "loss": 0.030152735114097596,
      "step": 46870
    },
    {
      "epoch": 13.3068407607153,
      "grad_norm": 6.351118087768555,
      "learning_rate": 8.669968776610843e-05,
      "loss": 0.02525656819343567,
      "step": 46880
    },
    {
      "epoch": 13.309679250638661,
      "grad_norm": 4.462566375732422,
      "learning_rate": 8.669684927618507e-05,
      "loss": 0.036944061517715454,
      "step": 46890
    },
    {
      "epoch": 13.31251774056202,
      "grad_norm": 1.909093976020813,
      "learning_rate": 8.669401078626172e-05,
      "loss": 0.03044966757297516,
      "step": 46900
    },
    {
      "epoch": 13.315356230485381,
      "grad_norm": 9.567749977111816,
      "learning_rate": 8.669117229633836e-05,
      "loss": 0.0394818127155304,
      "step": 46910
    },
    {
      "epoch": 13.318194720408743,
      "grad_norm": 5.714660167694092,
      "learning_rate": 8.668833380641498e-05,
      "loss": 0.03719258904457092,
      "step": 46920
    },
    {
      "epoch": 13.321033210332104,
      "grad_norm": 10.4080810546875,
      "learning_rate": 8.668549531649163e-05,
      "loss": 0.04888079762458801,
      "step": 46930
    },
    {
      "epoch": 13.323871700255465,
      "grad_norm": 15.858975410461426,
      "learning_rate": 8.668265682656827e-05,
      "loss": 0.05152977108955383,
      "step": 46940
    },
    {
      "epoch": 13.326710190178824,
      "grad_norm": 5.539859294891357,
      "learning_rate": 8.667981833664491e-05,
      "loss": 0.029546111822128296,
      "step": 46950
    },
    {
      "epoch": 13.329548680102185,
      "grad_norm": 7.2826457023620605,
      "learning_rate": 8.667697984672155e-05,
      "loss": 0.03621213436126709,
      "step": 46960
    },
    {
      "epoch": 13.332387170025546,
      "grad_norm": 4.723855972290039,
      "learning_rate": 8.667414135679819e-05,
      "loss": 0.03703057169914246,
      "step": 46970
    },
    {
      "epoch": 13.335225659948907,
      "grad_norm": 9.292047500610352,
      "learning_rate": 8.667130286687482e-05,
      "loss": 0.035714578628540036,
      "step": 46980
    },
    {
      "epoch": 13.338064149872269,
      "grad_norm": 8.979728698730469,
      "learning_rate": 8.666846437695146e-05,
      "loss": 0.034203764796257016,
      "step": 46990
    },
    {
      "epoch": 13.34090263979563,
      "grad_norm": 1.1390995979309082,
      "learning_rate": 8.66656258870281e-05,
      "loss": 0.033992066979408264,
      "step": 47000
    },
    {
      "epoch": 13.34090263979563,
      "eval_accuracy": 0.9556813123927005,
      "eval_loss": 0.1364302933216095,
      "eval_runtime": 31.5685,
      "eval_samples_per_second": 498.186,
      "eval_steps_per_second": 7.793,
      "step": 47000
    },
    {
      "epoch": 13.343741129718989,
      "grad_norm": 1.6108620166778564,
      "learning_rate": 8.666278739710474e-05,
      "loss": 0.019977061450481413,
      "step": 47010
    },
    {
      "epoch": 13.34657961964235,
      "grad_norm": 3.0914905071258545,
      "learning_rate": 8.665994890718139e-05,
      "loss": 0.026553791761398316,
      "step": 47020
    },
    {
      "epoch": 13.349418109565711,
      "grad_norm": 2.484858989715576,
      "learning_rate": 8.665711041725803e-05,
      "loss": 0.01819160133600235,
      "step": 47030
    },
    {
      "epoch": 13.352256599489072,
      "grad_norm": 5.5316548347473145,
      "learning_rate": 8.665427192733467e-05,
      "loss": 0.029444748163223268,
      "step": 47040
    },
    {
      "epoch": 13.355095089412433,
      "grad_norm": 1.544081449508667,
      "learning_rate": 8.66514334374113e-05,
      "loss": 0.01586762219667435,
      "step": 47050
    },
    {
      "epoch": 13.357933579335793,
      "grad_norm": 6.189842700958252,
      "learning_rate": 8.664859494748794e-05,
      "loss": 0.040950334072113036,
      "step": 47060
    },
    {
      "epoch": 13.360772069259154,
      "grad_norm": 2.857926845550537,
      "learning_rate": 8.664575645756458e-05,
      "loss": 0.019795121252536775,
      "step": 47070
    },
    {
      "epoch": 13.363610559182515,
      "grad_norm": 8.21424674987793,
      "learning_rate": 8.664291796764122e-05,
      "loss": 0.05254107117652893,
      "step": 47080
    },
    {
      "epoch": 13.366449049105876,
      "grad_norm": 3.3173604011535645,
      "learning_rate": 8.664007947771786e-05,
      "loss": 0.048231422901153564,
      "step": 47090
    },
    {
      "epoch": 13.369287539029237,
      "grad_norm": 12.414484977722168,
      "learning_rate": 8.66372409877945e-05,
      "loss": 0.025853991508483887,
      "step": 47100
    },
    {
      "epoch": 13.372126028952596,
      "grad_norm": 8.269749641418457,
      "learning_rate": 8.663440249787113e-05,
      "loss": 0.043690598011016844,
      "step": 47110
    },
    {
      "epoch": 13.374964518875958,
      "grad_norm": 3.5122714042663574,
      "learning_rate": 8.663156400794777e-05,
      "loss": 0.038815534114837645,
      "step": 47120
    },
    {
      "epoch": 13.377803008799319,
      "grad_norm": 8.742683410644531,
      "learning_rate": 8.662872551802441e-05,
      "loss": 0.0601883590221405,
      "step": 47130
    },
    {
      "epoch": 13.38064149872268,
      "grad_norm": 1.2642333507537842,
      "learning_rate": 8.662588702810105e-05,
      "loss": 0.034104183316230774,
      "step": 47140
    },
    {
      "epoch": 13.383479988646041,
      "grad_norm": 3.2773542404174805,
      "learning_rate": 8.66230485381777e-05,
      "loss": 0.025511911511421202,
      "step": 47150
    },
    {
      "epoch": 13.3863184785694,
      "grad_norm": 16.95147705078125,
      "learning_rate": 8.662021004825434e-05,
      "loss": 0.03688313961029053,
      "step": 47160
    },
    {
      "epoch": 13.389156968492761,
      "grad_norm": 3.6043238639831543,
      "learning_rate": 8.661737155833098e-05,
      "loss": 0.034344923496246335,
      "step": 47170
    },
    {
      "epoch": 13.391995458416122,
      "grad_norm": 9.516392707824707,
      "learning_rate": 8.66145330684076e-05,
      "loss": 0.0326772004365921,
      "step": 47180
    },
    {
      "epoch": 13.394833948339484,
      "grad_norm": 3.5813658237457275,
      "learning_rate": 8.661169457848425e-05,
      "loss": 0.04920532703399658,
      "step": 47190
    },
    {
      "epoch": 13.397672438262845,
      "grad_norm": 12.189434051513672,
      "learning_rate": 8.660885608856089e-05,
      "loss": 0.039902889728546144,
      "step": 47200
    },
    {
      "epoch": 13.400510928186204,
      "grad_norm": 1.0895549058914185,
      "learning_rate": 8.660601759863753e-05,
      "loss": 0.011589239537715911,
      "step": 47210
    },
    {
      "epoch": 13.403349418109565,
      "grad_norm": 3.7724554538726807,
      "learning_rate": 8.660317910871417e-05,
      "loss": 0.02462998479604721,
      "step": 47220
    },
    {
      "epoch": 13.406187908032926,
      "grad_norm": 12.197464942932129,
      "learning_rate": 8.660034061879081e-05,
      "loss": 0.01989622414112091,
      "step": 47230
    },
    {
      "epoch": 13.409026397956287,
      "grad_norm": 10.303778648376465,
      "learning_rate": 8.659750212886744e-05,
      "loss": 0.03702112436294556,
      "step": 47240
    },
    {
      "epoch": 13.411864887879648,
      "grad_norm": 25.24805450439453,
      "learning_rate": 8.659466363894408e-05,
      "loss": 0.05495012998580932,
      "step": 47250
    },
    {
      "epoch": 13.41470337780301,
      "grad_norm": 10.352737426757812,
      "learning_rate": 8.659182514902072e-05,
      "loss": 0.04449354410171509,
      "step": 47260
    },
    {
      "epoch": 13.417541867726369,
      "grad_norm": 6.472108364105225,
      "learning_rate": 8.658898665909737e-05,
      "loss": 0.037041071057319644,
      "step": 47270
    },
    {
      "epoch": 13.42038035764973,
      "grad_norm": 4.82477331161499,
      "learning_rate": 8.6586148169174e-05,
      "loss": 0.027723467350006102,
      "step": 47280
    },
    {
      "epoch": 13.423218847573091,
      "grad_norm": 10.74465274810791,
      "learning_rate": 8.658330967925065e-05,
      "loss": 0.04451736807823181,
      "step": 47290
    },
    {
      "epoch": 13.426057337496452,
      "grad_norm": 9.49945068359375,
      "learning_rate": 8.658047118932729e-05,
      "loss": 0.03110710084438324,
      "step": 47300
    },
    {
      "epoch": 13.428895827419813,
      "grad_norm": 1.6850651502609253,
      "learning_rate": 8.657763269940392e-05,
      "loss": 0.02335614264011383,
      "step": 47310
    },
    {
      "epoch": 13.431734317343173,
      "grad_norm": 0.6513549089431763,
      "learning_rate": 8.657479420948056e-05,
      "loss": 0.027879971265792846,
      "step": 47320
    },
    {
      "epoch": 13.434572807266534,
      "grad_norm": 4.218456745147705,
      "learning_rate": 8.65719557195572e-05,
      "loss": 0.03128032088279724,
      "step": 47330
    },
    {
      "epoch": 13.437411297189895,
      "grad_norm": 2.6938302516937256,
      "learning_rate": 8.656911722963383e-05,
      "loss": 0.01789194643497467,
      "step": 47340
    },
    {
      "epoch": 13.440249787113256,
      "grad_norm": 4.033883571624756,
      "learning_rate": 8.656627873971048e-05,
      "loss": 0.019952358305454256,
      "step": 47350
    },
    {
      "epoch": 13.443088277036617,
      "grad_norm": 4.329926490783691,
      "learning_rate": 8.656344024978712e-05,
      "loss": 0.032864853739738464,
      "step": 47360
    },
    {
      "epoch": 13.445926766959976,
      "grad_norm": 7.312324523925781,
      "learning_rate": 8.656060175986375e-05,
      "loss": 0.03324364721775055,
      "step": 47370
    },
    {
      "epoch": 13.448765256883338,
      "grad_norm": 1.3990533351898193,
      "learning_rate": 8.65577632699404e-05,
      "loss": 0.029722386598587038,
      "step": 47380
    },
    {
      "epoch": 13.451603746806699,
      "grad_norm": 11.081133842468262,
      "learning_rate": 8.655492478001703e-05,
      "loss": 0.024282051622867583,
      "step": 47390
    },
    {
      "epoch": 13.45444223673006,
      "grad_norm": 5.652640342712402,
      "learning_rate": 8.655208629009368e-05,
      "loss": 0.035066595673561095,
      "step": 47400
    },
    {
      "epoch": 13.457280726653421,
      "grad_norm": 1.3877437114715576,
      "learning_rate": 8.654924780017032e-05,
      "loss": 0.04085328876972198,
      "step": 47410
    },
    {
      "epoch": 13.460119216576782,
      "grad_norm": 6.15867805480957,
      "learning_rate": 8.654640931024696e-05,
      "loss": 0.029024744033813478,
      "step": 47420
    },
    {
      "epoch": 13.462957706500141,
      "grad_norm": 3.4995944499969482,
      "learning_rate": 8.65435708203236e-05,
      "loss": 0.03490034639835358,
      "step": 47430
    },
    {
      "epoch": 13.465796196423502,
      "grad_norm": 12.462900161743164,
      "learning_rate": 8.654073233040023e-05,
      "loss": 0.04476391971111297,
      "step": 47440
    },
    {
      "epoch": 13.468634686346864,
      "grad_norm": 2.4970486164093018,
      "learning_rate": 8.653789384047687e-05,
      "loss": 0.03142915666103363,
      "step": 47450
    },
    {
      "epoch": 13.471473176270225,
      "grad_norm": 9.125443458557129,
      "learning_rate": 8.653505535055351e-05,
      "loss": 0.04043779075145722,
      "step": 47460
    },
    {
      "epoch": 13.474311666193586,
      "grad_norm": 0.5085991621017456,
      "learning_rate": 8.653221686063014e-05,
      "loss": 0.022030943632125856,
      "step": 47470
    },
    {
      "epoch": 13.477150156116945,
      "grad_norm": 7.457637310028076,
      "learning_rate": 8.65293783707068e-05,
      "loss": 0.01750267595052719,
      "step": 47480
    },
    {
      "epoch": 13.479988646040306,
      "grad_norm": 8.2711181640625,
      "learning_rate": 8.652653988078343e-05,
      "loss": 0.028965687751770018,
      "step": 47490
    },
    {
      "epoch": 13.482827135963667,
      "grad_norm": 8.768735885620117,
      "learning_rate": 8.652370139086006e-05,
      "loss": 0.04551350176334381,
      "step": 47500
    },
    {
      "epoch": 13.482827135963667,
      "eval_accuracy": 0.9582247090990017,
      "eval_loss": 0.1277717500925064,
      "eval_runtime": 31.6797,
      "eval_samples_per_second": 496.437,
      "eval_steps_per_second": 7.765,
      "step": 47500
    },
    {
      "epoch": 13.485665625887028,
      "grad_norm": 6.081392288208008,
      "learning_rate": 8.65208629009367e-05,
      "loss": 0.026894816756248476,
      "step": 47510
    },
    {
      "epoch": 13.48850411581039,
      "grad_norm": 9.793679237365723,
      "learning_rate": 8.651802441101335e-05,
      "loss": 0.02709413170814514,
      "step": 47520
    },
    {
      "epoch": 13.491342605733749,
      "grad_norm": 6.559260368347168,
      "learning_rate": 8.651518592108999e-05,
      "loss": 0.030215150117874144,
      "step": 47530
    },
    {
      "epoch": 13.49418109565711,
      "grad_norm": 6.889903545379639,
      "learning_rate": 8.651234743116661e-05,
      "loss": 0.040194612741470334,
      "step": 47540
    },
    {
      "epoch": 13.497019585580471,
      "grad_norm": 3.6317713260650635,
      "learning_rate": 8.650950894124327e-05,
      "loss": 0.038753736019134524,
      "step": 47550
    },
    {
      "epoch": 13.499858075503832,
      "grad_norm": 8.623577117919922,
      "learning_rate": 8.650667045131991e-05,
      "loss": 0.030288875102996826,
      "step": 47560
    },
    {
      "epoch": 13.502696565427193,
      "grad_norm": 7.882750988006592,
      "learning_rate": 8.650383196139654e-05,
      "loss": 0.03433360159397125,
      "step": 47570
    },
    {
      "epoch": 13.505535055350553,
      "grad_norm": 5.83099889755249,
      "learning_rate": 8.650099347147318e-05,
      "loss": 0.025126606225967407,
      "step": 47580
    },
    {
      "epoch": 13.508373545273914,
      "grad_norm": 16.601478576660156,
      "learning_rate": 8.649815498154982e-05,
      "loss": 0.03801619708538055,
      "step": 47590
    },
    {
      "epoch": 13.511212035197275,
      "grad_norm": 3.1400492191314697,
      "learning_rate": 8.649531649162645e-05,
      "loss": 0.02400256395339966,
      "step": 47600
    },
    {
      "epoch": 13.514050525120636,
      "grad_norm": 12.873468399047852,
      "learning_rate": 8.64924780017031e-05,
      "loss": 0.03426647186279297,
      "step": 47610
    },
    {
      "epoch": 13.516889015043997,
      "grad_norm": 1.4552162885665894,
      "learning_rate": 8.648963951177975e-05,
      "loss": 0.04324913024902344,
      "step": 47620
    },
    {
      "epoch": 13.519727504967356,
      "grad_norm": 5.839138507843018,
      "learning_rate": 8.648680102185637e-05,
      "loss": 0.02866055369377136,
      "step": 47630
    },
    {
      "epoch": 13.522565994890718,
      "grad_norm": 2.0074400901794434,
      "learning_rate": 8.648396253193301e-05,
      "loss": 0.044433844089508054,
      "step": 47640
    },
    {
      "epoch": 13.525404484814079,
      "grad_norm": 2.4012770652770996,
      "learning_rate": 8.648112404200966e-05,
      "loss": 0.031976637244224546,
      "step": 47650
    },
    {
      "epoch": 13.52824297473744,
      "grad_norm": 0.5516537427902222,
      "learning_rate": 8.647856940107864e-05,
      "loss": 0.032358679175376895,
      "step": 47660
    },
    {
      "epoch": 13.5310814646608,
      "grad_norm": 10.947912216186523,
      "learning_rate": 8.647573091115528e-05,
      "loss": 0.029296010732650757,
      "step": 47670
    },
    {
      "epoch": 13.533919954584162,
      "grad_norm": 4.325680255889893,
      "learning_rate": 8.64728924212319e-05,
      "loss": 0.03215175569057464,
      "step": 47680
    },
    {
      "epoch": 13.536758444507521,
      "grad_norm": 3.7877259254455566,
      "learning_rate": 8.647005393130855e-05,
      "loss": 0.021947005391120912,
      "step": 47690
    },
    {
      "epoch": 13.539596934430882,
      "grad_norm": 3.5999224185943604,
      "learning_rate": 8.646721544138519e-05,
      "loss": 0.028984886407852174,
      "step": 47700
    },
    {
      "epoch": 13.542435424354244,
      "grad_norm": 2.117652416229248,
      "learning_rate": 8.646437695146183e-05,
      "loss": 0.03015979528427124,
      "step": 47710
    },
    {
      "epoch": 13.545273914277605,
      "grad_norm": 6.415585994720459,
      "learning_rate": 8.646153846153846e-05,
      "loss": 0.02594228684902191,
      "step": 47720
    },
    {
      "epoch": 13.548112404200966,
      "grad_norm": 11.240501403808594,
      "learning_rate": 8.645869997161511e-05,
      "loss": 0.0444490909576416,
      "step": 47730
    },
    {
      "epoch": 13.550950894124325,
      "grad_norm": 4.122629642486572,
      "learning_rate": 8.645586148169174e-05,
      "loss": 0.027328258752822875,
      "step": 47740
    },
    {
      "epoch": 13.553789384047686,
      "grad_norm": 5.013462066650391,
      "learning_rate": 8.645302299176838e-05,
      "loss": 0.04120333194732666,
      "step": 47750
    },
    {
      "epoch": 13.556627873971047,
      "grad_norm": 0.9704323410987854,
      "learning_rate": 8.645018450184502e-05,
      "loss": 0.03185535073280334,
      "step": 47760
    },
    {
      "epoch": 13.559466363894408,
      "grad_norm": 11.820596694946289,
      "learning_rate": 8.644734601192166e-05,
      "loss": 0.025168603658676146,
      "step": 47770
    },
    {
      "epoch": 13.56230485381777,
      "grad_norm": 9.895241737365723,
      "learning_rate": 8.644450752199829e-05,
      "loss": 0.027022379636764526,
      "step": 47780
    },
    {
      "epoch": 13.56514334374113,
      "grad_norm": 9.648585319519043,
      "learning_rate": 8.644166903207495e-05,
      "loss": 0.04760119318962097,
      "step": 47790
    },
    {
      "epoch": 13.56798183366449,
      "grad_norm": 5.825608253479004,
      "learning_rate": 8.643883054215159e-05,
      "loss": 0.027638736367225646,
      "step": 47800
    },
    {
      "epoch": 13.570820323587851,
      "grad_norm": 3.1733558177948,
      "learning_rate": 8.643599205222822e-05,
      "loss": 0.01781163513660431,
      "step": 47810
    },
    {
      "epoch": 13.573658813511212,
      "grad_norm": 20.244632720947266,
      "learning_rate": 8.643315356230486e-05,
      "loss": 0.053033411502838135,
      "step": 47820
    },
    {
      "epoch": 13.576497303434573,
      "grad_norm": 3.861985683441162,
      "learning_rate": 8.64303150723815e-05,
      "loss": 0.040492516756057736,
      "step": 47830
    },
    {
      "epoch": 13.579335793357934,
      "grad_norm": 6.913362979888916,
      "learning_rate": 8.642747658245813e-05,
      "loss": 0.02677871584892273,
      "step": 47840
    },
    {
      "epoch": 13.582174283281294,
      "grad_norm": 4.416360378265381,
      "learning_rate": 8.642463809253477e-05,
      "loss": 0.02400323748588562,
      "step": 47850
    },
    {
      "epoch": 13.585012773204655,
      "grad_norm": 14.091269493103027,
      "learning_rate": 8.642179960261142e-05,
      "loss": 0.06146126985549927,
      "step": 47860
    },
    {
      "epoch": 13.587851263128016,
      "grad_norm": 9.991366386413574,
      "learning_rate": 8.641896111268805e-05,
      "loss": 0.04355473220348358,
      "step": 47870
    },
    {
      "epoch": 13.590689753051377,
      "grad_norm": 8.972782135009766,
      "learning_rate": 8.641612262276469e-05,
      "loss": 0.029344624280929564,
      "step": 47880
    },
    {
      "epoch": 13.593528242974738,
      "grad_norm": 7.045830726623535,
      "learning_rate": 8.641328413284133e-05,
      "loss": 0.04973292350769043,
      "step": 47890
    },
    {
      "epoch": 13.596366732898097,
      "grad_norm": 8.405793190002441,
      "learning_rate": 8.641044564291797e-05,
      "loss": 0.04243167638778687,
      "step": 47900
    },
    {
      "epoch": 13.599205222821459,
      "grad_norm": 5.957789421081543,
      "learning_rate": 8.64076071529946e-05,
      "loss": 0.02167646735906601,
      "step": 47910
    },
    {
      "epoch": 13.60204371274482,
      "grad_norm": 3.9297266006469727,
      "learning_rate": 8.640476866307124e-05,
      "loss": 0.041967970132827756,
      "step": 47920
    },
    {
      "epoch": 13.60488220266818,
      "grad_norm": 5.658904075622559,
      "learning_rate": 8.64019301731479e-05,
      "loss": 0.046526464819908145,
      "step": 47930
    },
    {
      "epoch": 13.607720692591542,
      "grad_norm": 3.7970118522644043,
      "learning_rate": 8.639909168322453e-05,
      "loss": 0.03723128139972687,
      "step": 47940
    },
    {
      "epoch": 13.610559182514901,
      "grad_norm": 4.04428768157959,
      "learning_rate": 8.639625319330117e-05,
      "loss": 0.03817231059074402,
      "step": 47950
    },
    {
      "epoch": 13.613397672438262,
      "grad_norm": 7.314342975616455,
      "learning_rate": 8.639341470337781e-05,
      "loss": 0.040563181042671204,
      "step": 47960
    },
    {
      "epoch": 13.616236162361623,
      "grad_norm": 4.703126430511475,
      "learning_rate": 8.639057621345444e-05,
      "loss": 0.042730242013931274,
      "step": 47970
    },
    {
      "epoch": 13.619074652284985,
      "grad_norm": 16.723527908325195,
      "learning_rate": 8.638773772353108e-05,
      "loss": 0.044729626178741454,
      "step": 47980
    },
    {
      "epoch": 13.621913142208346,
      "grad_norm": 8.587444305419922,
      "learning_rate": 8.638489923360773e-05,
      "loss": 0.036767077445983884,
      "step": 47990
    },
    {
      "epoch": 13.624751632131705,
      "grad_norm": 7.164139747619629,
      "learning_rate": 8.638206074368436e-05,
      "loss": 0.03381766676902771,
      "step": 48000
    },
    {
      "epoch": 13.624751632131705,
      "eval_accuracy": 0.9483690468620843,
      "eval_loss": 0.15916083753108978,
      "eval_runtime": 31.362,
      "eval_samples_per_second": 501.467,
      "eval_steps_per_second": 7.844,
      "step": 48000
    },
    {
      "epoch": 13.627590122055066,
      "grad_norm": 7.926647186279297,
      "learning_rate": 8.6379222253761e-05,
      "loss": 0.047017800807952884,
      "step": 48010
    },
    {
      "epoch": 13.630428611978427,
      "grad_norm": 5.58323335647583,
      "learning_rate": 8.637638376383764e-05,
      "loss": 0.020823168754577636,
      "step": 48020
    },
    {
      "epoch": 13.633267101901788,
      "grad_norm": 5.785126686096191,
      "learning_rate": 8.637354527391429e-05,
      "loss": 0.026754921674728392,
      "step": 48030
    },
    {
      "epoch": 13.63610559182515,
      "grad_norm": 4.36375093460083,
      "learning_rate": 8.637070678399091e-05,
      "loss": 0.026675713062286378,
      "step": 48040
    },
    {
      "epoch": 13.63894408174851,
      "grad_norm": 6.520670413970947,
      "learning_rate": 8.636786829406755e-05,
      "loss": 0.027671533823013305,
      "step": 48050
    },
    {
      "epoch": 13.64178257167187,
      "grad_norm": 9.967583656311035,
      "learning_rate": 8.636502980414421e-05,
      "loss": 0.037085187435150144,
      "step": 48060
    },
    {
      "epoch": 13.644621061595231,
      "grad_norm": 1.4123072624206543,
      "learning_rate": 8.636219131422084e-05,
      "loss": 0.03883474767208099,
      "step": 48070
    },
    {
      "epoch": 13.647459551518592,
      "grad_norm": 4.982833385467529,
      "learning_rate": 8.635935282429748e-05,
      "loss": 0.0395208865404129,
      "step": 48080
    },
    {
      "epoch": 13.650298041441953,
      "grad_norm": 11.74831771850586,
      "learning_rate": 8.635651433437412e-05,
      "loss": 0.04091322720050812,
      "step": 48090
    },
    {
      "epoch": 13.653136531365314,
      "grad_norm": 13.438674926757812,
      "learning_rate": 8.635367584445075e-05,
      "loss": 0.04151826798915863,
      "step": 48100
    },
    {
      "epoch": 13.655975021288674,
      "grad_norm": 6.555180549621582,
      "learning_rate": 8.635083735452739e-05,
      "loss": 0.052014333009719846,
      "step": 48110
    },
    {
      "epoch": 13.658813511212035,
      "grad_norm": 1.3131600618362427,
      "learning_rate": 8.634799886460403e-05,
      "loss": 0.05837852954864502,
      "step": 48120
    },
    {
      "epoch": 13.661652001135396,
      "grad_norm": 4.496669769287109,
      "learning_rate": 8.634516037468067e-05,
      "loss": 0.03982115089893341,
      "step": 48130
    },
    {
      "epoch": 13.664490491058757,
      "grad_norm": 6.10742712020874,
      "learning_rate": 8.634232188475731e-05,
      "loss": 0.043477493524551394,
      "step": 48140
    },
    {
      "epoch": 13.667328980982118,
      "grad_norm": 11.731135368347168,
      "learning_rate": 8.633948339483396e-05,
      "loss": 0.0458399623632431,
      "step": 48150
    },
    {
      "epoch": 13.670167470905477,
      "grad_norm": 5.114226341247559,
      "learning_rate": 8.63366449049106e-05,
      "loss": 0.029850488901138304,
      "step": 48160
    },
    {
      "epoch": 13.673005960828839,
      "grad_norm": 10.64398193359375,
      "learning_rate": 8.633380641498722e-05,
      "loss": 0.03504253327846527,
      "step": 48170
    },
    {
      "epoch": 13.6758444507522,
      "grad_norm": 10.427469253540039,
      "learning_rate": 8.633096792506387e-05,
      "loss": 0.04113469421863556,
      "step": 48180
    },
    {
      "epoch": 13.67868294067556,
      "grad_norm": 1.5694217681884766,
      "learning_rate": 8.632812943514052e-05,
      "loss": 0.033141395449638365,
      "step": 48190
    },
    {
      "epoch": 13.681521430598922,
      "grad_norm": 3.8541409969329834,
      "learning_rate": 8.632529094521715e-05,
      "loss": 0.032296416163444516,
      "step": 48200
    },
    {
      "epoch": 13.684359920522283,
      "grad_norm": 3.6853768825531006,
      "learning_rate": 8.632245245529379e-05,
      "loss": 0.04020425975322724,
      "step": 48210
    },
    {
      "epoch": 13.687198410445642,
      "grad_norm": 7.272515773773193,
      "learning_rate": 8.631961396537043e-05,
      "loss": 0.022356095910072326,
      "step": 48220
    },
    {
      "epoch": 13.690036900369003,
      "grad_norm": 0.4939178228378296,
      "learning_rate": 8.631677547544706e-05,
      "loss": 0.013650459051132203,
      "step": 48230
    },
    {
      "epoch": 13.692875390292365,
      "grad_norm": 5.019110679626465,
      "learning_rate": 8.63139369855237e-05,
      "loss": 0.049863725900650024,
      "step": 48240
    },
    {
      "epoch": 13.695713880215726,
      "grad_norm": 1.272499442100525,
      "learning_rate": 8.631109849560034e-05,
      "loss": 0.01859838664531708,
      "step": 48250
    },
    {
      "epoch": 13.698552370139087,
      "grad_norm": 3.420999526977539,
      "learning_rate": 8.630826000567698e-05,
      "loss": 0.022597207129001616,
      "step": 48260
    },
    {
      "epoch": 13.701390860062446,
      "grad_norm": 4.212942600250244,
      "learning_rate": 8.630542151575362e-05,
      "loss": 0.03907855451107025,
      "step": 48270
    },
    {
      "epoch": 13.704229349985807,
      "grad_norm": 13.82310676574707,
      "learning_rate": 8.630258302583027e-05,
      "loss": 0.052293556928634646,
      "step": 48280
    },
    {
      "epoch": 13.707067839909168,
      "grad_norm": 8.72931957244873,
      "learning_rate": 8.629974453590691e-05,
      "loss": 0.03531932830810547,
      "step": 48290
    },
    {
      "epoch": 13.70990632983253,
      "grad_norm": 4.780879974365234,
      "learning_rate": 8.629690604598354e-05,
      "loss": 0.023269952833652498,
      "step": 48300
    },
    {
      "epoch": 13.71274481975589,
      "grad_norm": 4.586771011352539,
      "learning_rate": 8.629406755606018e-05,
      "loss": 0.026098227500915526,
      "step": 48310
    },
    {
      "epoch": 13.71558330967925,
      "grad_norm": 2.3212790489196777,
      "learning_rate": 8.629122906613682e-05,
      "loss": 0.019361671805381776,
      "step": 48320
    },
    {
      "epoch": 13.718421799602611,
      "grad_norm": 9.564801216125488,
      "learning_rate": 8.628839057621346e-05,
      "loss": 0.057374054193496705,
      "step": 48330
    },
    {
      "epoch": 13.721260289525972,
      "grad_norm": 1.632908821105957,
      "learning_rate": 8.62855520862901e-05,
      "loss": 0.03448108434677124,
      "step": 48340
    },
    {
      "epoch": 13.724098779449333,
      "grad_norm": 1.9769229888916016,
      "learning_rate": 8.628271359636674e-05,
      "loss": 0.03277253210544586,
      "step": 48350
    },
    {
      "epoch": 13.726937269372694,
      "grad_norm": 2.660634756088257,
      "learning_rate": 8.627987510644337e-05,
      "loss": 0.030680391192436218,
      "step": 48360
    },
    {
      "epoch": 13.729775759296054,
      "grad_norm": 3.069620132446289,
      "learning_rate": 8.627703661652001e-05,
      "loss": 0.03792406022548676,
      "step": 48370
    },
    {
      "epoch": 13.732614249219415,
      "grad_norm": 3.5685558319091797,
      "learning_rate": 8.627419812659665e-05,
      "loss": 0.052587497234344485,
      "step": 48380
    },
    {
      "epoch": 13.735452739142776,
      "grad_norm": 5.610739707946777,
      "learning_rate": 8.62713596366733e-05,
      "loss": 0.028223451972007752,
      "step": 48390
    },
    {
      "epoch": 13.738291229066137,
      "grad_norm": 5.543591022491455,
      "learning_rate": 8.626852114674994e-05,
      "loss": 0.03342098891735077,
      "step": 48400
    },
    {
      "epoch": 13.741129718989498,
      "grad_norm": 7.765977382659912,
      "learning_rate": 8.626568265682658e-05,
      "loss": 0.03520514965057373,
      "step": 48410
    },
    {
      "epoch": 13.743968208912857,
      "grad_norm": 8.167157173156738,
      "learning_rate": 8.626284416690322e-05,
      "loss": 0.021751371026039124,
      "step": 48420
    },
    {
      "epoch": 13.746806698836219,
      "grad_norm": 11.61294937133789,
      "learning_rate": 8.626000567697985e-05,
      "loss": 0.04853320717811584,
      "step": 48430
    },
    {
      "epoch": 13.74964518875958,
      "grad_norm": 7.700771808624268,
      "learning_rate": 8.625716718705649e-05,
      "loss": 0.0263394832611084,
      "step": 48440
    },
    {
      "epoch": 13.75248367868294,
      "grad_norm": 13.254837989807129,
      "learning_rate": 8.625432869713313e-05,
      "loss": 0.03906325101852417,
      "step": 48450
    },
    {
      "epoch": 13.755322168606302,
      "grad_norm": 1.32351553440094,
      "learning_rate": 8.625149020720977e-05,
      "loss": 0.02241494506597519,
      "step": 48460
    },
    {
      "epoch": 13.758160658529663,
      "grad_norm": 2.231417417526245,
      "learning_rate": 8.624865171728641e-05,
      "loss": 0.03027019202709198,
      "step": 48470
    },
    {
      "epoch": 13.760999148453022,
      "grad_norm": 11.51838207244873,
      "learning_rate": 8.624581322736305e-05,
      "loss": 0.0392301082611084,
      "step": 48480
    },
    {
      "epoch": 13.763837638376383,
      "grad_norm": 6.8642354011535645,
      "learning_rate": 8.624297473743968e-05,
      "loss": 0.03570234477519989,
      "step": 48490
    },
    {
      "epoch": 13.766676128299745,
      "grad_norm": 12.227685928344727,
      "learning_rate": 8.624013624751632e-05,
      "loss": 0.03676712810993195,
      "step": 48500
    },
    {
      "epoch": 13.766676128299745,
      "eval_accuracy": 0.9556813123927005,
      "eval_loss": 0.13903740048408508,
      "eval_runtime": 31.1746,
      "eval_samples_per_second": 504.481,
      "eval_steps_per_second": 7.891,
      "step": 48500
    },
    {
      "epoch": 13.769514618223106,
      "grad_norm": 8.302947044372559,
      "learning_rate": 8.623729775759296e-05,
      "loss": 0.03121452033519745,
      "step": 48510
    },
    {
      "epoch": 13.772353108146467,
      "grad_norm": 0.5118089318275452,
      "learning_rate": 8.62344592676696e-05,
      "loss": 0.026290327310562134,
      "step": 48520
    },
    {
      "epoch": 13.775191598069826,
      "grad_norm": 8.581632614135742,
      "learning_rate": 8.623162077774625e-05,
      "loss": 0.029920867085456847,
      "step": 48530
    },
    {
      "epoch": 13.778030087993187,
      "grad_norm": 17.793174743652344,
      "learning_rate": 8.622878228782289e-05,
      "loss": 0.04763144850730896,
      "step": 48540
    },
    {
      "epoch": 13.780868577916548,
      "grad_norm": 5.864516258239746,
      "learning_rate": 8.622594379789953e-05,
      "loss": 0.03731409013271332,
      "step": 48550
    },
    {
      "epoch": 13.78370706783991,
      "grad_norm": 4.590426921844482,
      "learning_rate": 8.622310530797616e-05,
      "loss": 0.052971386909484865,
      "step": 48560
    },
    {
      "epoch": 13.78654555776327,
      "grad_norm": 7.735920429229736,
      "learning_rate": 8.62202668180528e-05,
      "loss": 0.029651179909706116,
      "step": 48570
    },
    {
      "epoch": 13.789384047686632,
      "grad_norm": 3.522223711013794,
      "learning_rate": 8.621742832812944e-05,
      "loss": 0.03489167392253876,
      "step": 48580
    },
    {
      "epoch": 13.792222537609991,
      "grad_norm": 7.527494430541992,
      "learning_rate": 8.621458983820608e-05,
      "loss": 0.05272212624549866,
      "step": 48590
    },
    {
      "epoch": 13.795061027533352,
      "grad_norm": 6.009422779083252,
      "learning_rate": 8.621175134828272e-05,
      "loss": 0.03795459270477295,
      "step": 48600
    },
    {
      "epoch": 13.797899517456713,
      "grad_norm": 2.8332602977752686,
      "learning_rate": 8.620891285835936e-05,
      "loss": 0.04317569136619568,
      "step": 48610
    },
    {
      "epoch": 13.800738007380074,
      "grad_norm": 6.636512756347656,
      "learning_rate": 8.620607436843599e-05,
      "loss": 0.0366673469543457,
      "step": 48620
    },
    {
      "epoch": 13.803576497303435,
      "grad_norm": 7.454467296600342,
      "learning_rate": 8.620323587851263e-05,
      "loss": 0.04124805927276611,
      "step": 48630
    },
    {
      "epoch": 13.806414987226795,
      "grad_norm": 3.7960667610168457,
      "learning_rate": 8.620039738858927e-05,
      "loss": 0.033261549472808835,
      "step": 48640
    },
    {
      "epoch": 13.809253477150156,
      "grad_norm": 2.6478302478790283,
      "learning_rate": 8.619755889866592e-05,
      "loss": 0.027139192819595336,
      "step": 48650
    },
    {
      "epoch": 13.812091967073517,
      "grad_norm": 6.87192964553833,
      "learning_rate": 8.619472040874256e-05,
      "loss": 0.0389889806509018,
      "step": 48660
    },
    {
      "epoch": 13.814930456996878,
      "grad_norm": 1.3228923082351685,
      "learning_rate": 8.61918819188192e-05,
      "loss": 0.04917533099651337,
      "step": 48670
    },
    {
      "epoch": 13.81776894692024,
      "grad_norm": 2.538217067718506,
      "learning_rate": 8.618904342889583e-05,
      "loss": 0.03401821255683899,
      "step": 48680
    },
    {
      "epoch": 13.820607436843598,
      "grad_norm": 6.82718563079834,
      "learning_rate": 8.618620493897247e-05,
      "loss": 0.0350192666053772,
      "step": 48690
    },
    {
      "epoch": 13.82344592676696,
      "grad_norm": 5.149806499481201,
      "learning_rate": 8.618336644904911e-05,
      "loss": 0.03552314043045044,
      "step": 48700
    },
    {
      "epoch": 13.82628441669032,
      "grad_norm": 8.446120262145996,
      "learning_rate": 8.618052795912575e-05,
      "loss": 0.05527779459953308,
      "step": 48710
    },
    {
      "epoch": 13.829122906613682,
      "grad_norm": 2.4945015907287598,
      "learning_rate": 8.617768946920239e-05,
      "loss": 0.021887503564357758,
      "step": 48720
    },
    {
      "epoch": 13.831961396537043,
      "grad_norm": 7.55051326751709,
      "learning_rate": 8.617485097927903e-05,
      "loss": 0.051308661699295044,
      "step": 48730
    },
    {
      "epoch": 13.834799886460402,
      "grad_norm": 11.434135437011719,
      "learning_rate": 8.617201248935567e-05,
      "loss": 0.03981300890445709,
      "step": 48740
    },
    {
      "epoch": 13.837638376383763,
      "grad_norm": 5.199482440948486,
      "learning_rate": 8.61691739994323e-05,
      "loss": 0.04285101294517517,
      "step": 48750
    },
    {
      "epoch": 13.840476866307124,
      "grad_norm": 2.992975950241089,
      "learning_rate": 8.616633550950894e-05,
      "loss": 0.03893276751041412,
      "step": 48760
    },
    {
      "epoch": 13.843315356230486,
      "grad_norm": 1.883261799812317,
      "learning_rate": 8.616349701958558e-05,
      "loss": 0.03400767743587494,
      "step": 48770
    },
    {
      "epoch": 13.846153846153847,
      "grad_norm": 13.214120864868164,
      "learning_rate": 8.616065852966221e-05,
      "loss": 0.03298504948616028,
      "step": 48780
    },
    {
      "epoch": 13.848992336077206,
      "grad_norm": 6.457499980926514,
      "learning_rate": 8.615782003973887e-05,
      "loss": 0.024342995882034302,
      "step": 48790
    },
    {
      "epoch": 13.851830826000567,
      "grad_norm": 4.725510120391846,
      "learning_rate": 8.615498154981551e-05,
      "loss": 0.044960802793502806,
      "step": 48800
    },
    {
      "epoch": 13.854669315923928,
      "grad_norm": 12.056913375854492,
      "learning_rate": 8.615214305989214e-05,
      "loss": 0.03085973858833313,
      "step": 48810
    },
    {
      "epoch": 13.85750780584729,
      "grad_norm": 1.6123108863830566,
      "learning_rate": 8.614930456996878e-05,
      "loss": 0.042486336827278134,
      "step": 48820
    },
    {
      "epoch": 13.86034629577065,
      "grad_norm": 10.329974174499512,
      "learning_rate": 8.614646608004542e-05,
      "loss": 0.028568169474601744,
      "step": 48830
    },
    {
      "epoch": 13.863184785694012,
      "grad_norm": 5.0873703956604,
      "learning_rate": 8.614362759012206e-05,
      "loss": 0.03121526539325714,
      "step": 48840
    },
    {
      "epoch": 13.866023275617371,
      "grad_norm": 8.78348445892334,
      "learning_rate": 8.614078910019869e-05,
      "loss": 0.03692586123943329,
      "step": 48850
    },
    {
      "epoch": 13.868861765540732,
      "grad_norm": 2.368893623352051,
      "learning_rate": 8.613795061027534e-05,
      "loss": 0.03682179450988769,
      "step": 48860
    },
    {
      "epoch": 13.871700255464093,
      "grad_norm": 3.834256172180176,
      "learning_rate": 8.613511212035199e-05,
      "loss": 0.042053830623626706,
      "step": 48870
    },
    {
      "epoch": 13.874538745387454,
      "grad_norm": 8.786736488342285,
      "learning_rate": 8.613227363042861e-05,
      "loss": 0.030211985111236572,
      "step": 48880
    },
    {
      "epoch": 13.877377235310815,
      "grad_norm": 6.813754558563232,
      "learning_rate": 8.612943514050525e-05,
      "loss": 0.04961362779140473,
      "step": 48890
    },
    {
      "epoch": 13.880215725234175,
      "grad_norm": 1.4872814416885376,
      "learning_rate": 8.61265966505819e-05,
      "loss": 0.03151536285877228,
      "step": 48900
    },
    {
      "epoch": 13.883054215157536,
      "grad_norm": 0.777333676815033,
      "learning_rate": 8.612375816065852e-05,
      "loss": 0.03926765620708465,
      "step": 48910
    },
    {
      "epoch": 13.885892705080897,
      "grad_norm": 11.57472038269043,
      "learning_rate": 8.612091967073518e-05,
      "loss": 0.04946111738681793,
      "step": 48920
    },
    {
      "epoch": 13.888731195004258,
      "grad_norm": 1.3169654607772827,
      "learning_rate": 8.611808118081182e-05,
      "loss": 0.05031781792640686,
      "step": 48930
    },
    {
      "epoch": 13.89156968492762,
      "grad_norm": 4.004415035247803,
      "learning_rate": 8.611524269088845e-05,
      "loss": 0.031546711921691895,
      "step": 48940
    },
    {
      "epoch": 13.894408174850978,
      "grad_norm": 1.268197774887085,
      "learning_rate": 8.611240420096509e-05,
      "loss": 0.034222853183746335,
      "step": 48950
    },
    {
      "epoch": 13.89724666477434,
      "grad_norm": 5.52115535736084,
      "learning_rate": 8.610956571104173e-05,
      "loss": 0.03151540458202362,
      "step": 48960
    },
    {
      "epoch": 13.9000851546977,
      "grad_norm": 1.7569721937179565,
      "learning_rate": 8.610672722111837e-05,
      "loss": 0.023706762492656706,
      "step": 48970
    },
    {
      "epoch": 13.902923644621062,
      "grad_norm": 6.592088222503662,
      "learning_rate": 8.6103888731195e-05,
      "loss": 0.057257479429244994,
      "step": 48980
    },
    {
      "epoch": 13.905762134544423,
      "grad_norm": 11.381346702575684,
      "learning_rate": 8.610105024127165e-05,
      "loss": 0.03457716703414917,
      "step": 48990
    },
    {
      "epoch": 13.908600624467784,
      "grad_norm": 3.538055658340454,
      "learning_rate": 8.60982117513483e-05,
      "loss": 0.037588676810264586,
      "step": 49000
    },
    {
      "epoch": 13.908600624467784,
      "eval_accuracy": 0.9500222547211802,
      "eval_loss": 0.1497591733932495,
      "eval_runtime": 31.5303,
      "eval_samples_per_second": 498.79,
      "eval_steps_per_second": 7.802,
      "step": 49000
    },
    {
      "epoch": 13.911439114391143,
      "grad_norm": 8.341750144958496,
      "learning_rate": 8.609537326142492e-05,
      "loss": 0.024016308784484863,
      "step": 49010
    },
    {
      "epoch": 13.914277604314504,
      "grad_norm": 3.7513880729675293,
      "learning_rate": 8.609253477150156e-05,
      "loss": 0.03671912550926208,
      "step": 49020
    },
    {
      "epoch": 13.917116094237866,
      "grad_norm": 7.933117389678955,
      "learning_rate": 8.60896962815782e-05,
      "loss": 0.05676103830337524,
      "step": 49030
    },
    {
      "epoch": 13.919954584161227,
      "grad_norm": 3.312819004058838,
      "learning_rate": 8.608685779165483e-05,
      "loss": 0.041959112882614134,
      "step": 49040
    },
    {
      "epoch": 13.922793074084588,
      "grad_norm": 2.7314186096191406,
      "learning_rate": 8.608401930173148e-05,
      "loss": 0.047778987884521486,
      "step": 49050
    },
    {
      "epoch": 13.925631564007947,
      "grad_norm": 9.731554985046387,
      "learning_rate": 8.608118081180813e-05,
      "loss": 0.043270617723464966,
      "step": 49060
    },
    {
      "epoch": 13.928470053931308,
      "grad_norm": 4.183832168579102,
      "learning_rate": 8.607834232188476e-05,
      "loss": 0.03043203353881836,
      "step": 49070
    },
    {
      "epoch": 13.93130854385467,
      "grad_norm": 6.057960510253906,
      "learning_rate": 8.60755038319614e-05,
      "loss": 0.026605433225631712,
      "step": 49080
    },
    {
      "epoch": 13.93414703377803,
      "grad_norm": 9.82526969909668,
      "learning_rate": 8.607266534203804e-05,
      "loss": 0.02724425494670868,
      "step": 49090
    },
    {
      "epoch": 13.936985523701392,
      "grad_norm": 1.08476984500885,
      "learning_rate": 8.606982685211468e-05,
      "loss": 0.020269884169101714,
      "step": 49100
    },
    {
      "epoch": 13.93982401362475,
      "grad_norm": 9.197391510009766,
      "learning_rate": 8.606698836219131e-05,
      "loss": 0.01588790714740753,
      "step": 49110
    },
    {
      "epoch": 13.942662503548112,
      "grad_norm": 3.962165594100952,
      "learning_rate": 8.606414987226797e-05,
      "loss": 0.018375831842422485,
      "step": 49120
    },
    {
      "epoch": 13.945500993471473,
      "grad_norm": 10.083041191101074,
      "learning_rate": 8.60613113823446e-05,
      "loss": 0.0471290111541748,
      "step": 49130
    },
    {
      "epoch": 13.948339483394834,
      "grad_norm": 7.544643878936768,
      "learning_rate": 8.605847289242123e-05,
      "loss": 0.03300432562828064,
      "step": 49140
    },
    {
      "epoch": 13.951177973318195,
      "grad_norm": 3.965888500213623,
      "learning_rate": 8.605563440249788e-05,
      "loss": 0.04498538374900818,
      "step": 49150
    },
    {
      "epoch": 13.954016463241555,
      "grad_norm": 2.322057008743286,
      "learning_rate": 8.605279591257452e-05,
      "loss": 0.028949755430221557,
      "step": 49160
    },
    {
      "epoch": 13.956854953164916,
      "grad_norm": 3.3506157398223877,
      "learning_rate": 8.604995742265114e-05,
      "loss": 0.02183050811290741,
      "step": 49170
    },
    {
      "epoch": 13.959693443088277,
      "grad_norm": 9.87710189819336,
      "learning_rate": 8.604711893272779e-05,
      "loss": 0.034973537921905516,
      "step": 49180
    },
    {
      "epoch": 13.962531933011638,
      "grad_norm": 0.7107105851173401,
      "learning_rate": 8.604428044280444e-05,
      "loss": 0.037807163596153257,
      "step": 49190
    },
    {
      "epoch": 13.965370422934999,
      "grad_norm": 0.5313303470611572,
      "learning_rate": 8.604144195288107e-05,
      "loss": 0.03104567527770996,
      "step": 49200
    },
    {
      "epoch": 13.968208912858358,
      "grad_norm": 11.003992080688477,
      "learning_rate": 8.603860346295771e-05,
      "loss": 0.023176269233226778,
      "step": 49210
    },
    {
      "epoch": 13.97104740278172,
      "grad_norm": 7.429601669311523,
      "learning_rate": 8.603576497303435e-05,
      "loss": 0.04297060966491699,
      "step": 49220
    },
    {
      "epoch": 13.97388589270508,
      "grad_norm": 4.123441696166992,
      "learning_rate": 8.603292648311099e-05,
      "loss": 0.04354064166545868,
      "step": 49230
    },
    {
      "epoch": 13.976724382628442,
      "grad_norm": 15.48652458190918,
      "learning_rate": 8.603008799318762e-05,
      "loss": 0.03613204956054687,
      "step": 49240
    },
    {
      "epoch": 13.979562872551803,
      "grad_norm": 8.313751220703125,
      "learning_rate": 8.602724950326426e-05,
      "loss": 0.036995112895965576,
      "step": 49250
    },
    {
      "epoch": 13.982401362475164,
      "grad_norm": 6.673551559448242,
      "learning_rate": 8.602441101334092e-05,
      "loss": 0.04898467361927032,
      "step": 49260
    },
    {
      "epoch": 13.985239852398523,
      "grad_norm": 12.521001815795898,
      "learning_rate": 8.602157252341755e-05,
      "loss": 0.04510078430175781,
      "step": 49270
    },
    {
      "epoch": 13.988078342321884,
      "grad_norm": 3.74715518951416,
      "learning_rate": 8.601873403349419e-05,
      "loss": 0.04523438215255737,
      "step": 49280
    },
    {
      "epoch": 13.990916832245246,
      "grad_norm": 4.469878196716309,
      "learning_rate": 8.601589554357083e-05,
      "loss": 0.023627981543540955,
      "step": 49290
    },
    {
      "epoch": 13.993755322168607,
      "grad_norm": 10.40145492553711,
      "learning_rate": 8.601305705364746e-05,
      "loss": 0.024804672598838805,
      "step": 49300
    },
    {
      "epoch": 13.996593812091968,
      "grad_norm": 11.39846420288086,
      "learning_rate": 8.60102185637241e-05,
      "loss": 0.04401984214782715,
      "step": 49310
    },
    {
      "epoch": 13.999432302015327,
      "grad_norm": 11.098444938659668,
      "learning_rate": 8.600738007380075e-05,
      "loss": 0.02608799934387207,
      "step": 49320
    },
    {
      "epoch": 14.002270791938688,
      "grad_norm": 15.914974212646484,
      "learning_rate": 8.600454158387738e-05,
      "loss": 0.03237663507461548,
      "step": 49330
    },
    {
      "epoch": 14.00510928186205,
      "grad_norm": 14.43706226348877,
      "learning_rate": 8.600170309395402e-05,
      "loss": 0.031493523716926576,
      "step": 49340
    },
    {
      "epoch": 14.00794777178541,
      "grad_norm": 10.505023002624512,
      "learning_rate": 8.599886460403066e-05,
      "loss": 0.034197115898132326,
      "step": 49350
    },
    {
      "epoch": 14.010786261708772,
      "grad_norm": 15.057801246643066,
      "learning_rate": 8.59960261141073e-05,
      "loss": 0.044925516843795775,
      "step": 49360
    },
    {
      "epoch": 14.01362475163213,
      "grad_norm": 10.624711036682129,
      "learning_rate": 8.599318762418393e-05,
      "loss": 0.027485841512680055,
      "step": 49370
    },
    {
      "epoch": 14.016463241555492,
      "grad_norm": 0.39117997884750366,
      "learning_rate": 8.599034913426057e-05,
      "loss": 0.02868073582649231,
      "step": 49380
    },
    {
      "epoch": 14.019301731478853,
      "grad_norm": 4.588068962097168,
      "learning_rate": 8.598751064433723e-05,
      "loss": 0.0261732816696167,
      "step": 49390
    },
    {
      "epoch": 14.022140221402214,
      "grad_norm": 3.7235233783721924,
      "learning_rate": 8.598467215441386e-05,
      "loss": 0.028425517678260803,
      "step": 49400
    },
    {
      "epoch": 14.024978711325575,
      "grad_norm": 8.271585464477539,
      "learning_rate": 8.59818336644905e-05,
      "loss": 0.019010022282600403,
      "step": 49410
    },
    {
      "epoch": 14.027817201248936,
      "grad_norm": 1.4294496774673462,
      "learning_rate": 8.597899517456714e-05,
      "loss": 0.03500679433345795,
      "step": 49420
    },
    {
      "epoch": 14.030655691172296,
      "grad_norm": 9.961371421813965,
      "learning_rate": 8.597615668464377e-05,
      "loss": 0.041388750076293945,
      "step": 49430
    },
    {
      "epoch": 14.033494181095657,
      "grad_norm": 4.820285797119141,
      "learning_rate": 8.597331819472041e-05,
      "loss": 0.022803471982479097,
      "step": 49440
    },
    {
      "epoch": 14.036332671019018,
      "grad_norm": 0.7603951692581177,
      "learning_rate": 8.597047970479705e-05,
      "loss": 0.02021336704492569,
      "step": 49450
    },
    {
      "epoch": 14.039171160942379,
      "grad_norm": 11.177534103393555,
      "learning_rate": 8.596764121487369e-05,
      "loss": 0.03464284539222717,
      "step": 49460
    },
    {
      "epoch": 14.04200965086574,
      "grad_norm": 16.69522476196289,
      "learning_rate": 8.596480272495033e-05,
      "loss": 0.042182856798172,
      "step": 49470
    },
    {
      "epoch": 14.0448481407891,
      "grad_norm": 15.361966133117676,
      "learning_rate": 8.596196423502697e-05,
      "loss": 0.03753726482391358,
      "step": 49480
    },
    {
      "epoch": 14.04768663071246,
      "grad_norm": 7.312578201293945,
      "learning_rate": 8.595912574510361e-05,
      "loss": 0.04405333399772644,
      "step": 49490
    },
    {
      "epoch": 14.050525120635822,
      "grad_norm": 5.694483757019043,
      "learning_rate": 8.595628725518024e-05,
      "loss": 0.018757201731204987,
      "step": 49500
    },
    {
      "epoch": 14.050525120635822,
      "eval_accuracy": 0.9628664080880015,
      "eval_loss": 0.11576772481203079,
      "eval_runtime": 31.4364,
      "eval_samples_per_second": 500.279,
      "eval_steps_per_second": 7.825,
      "step": 49500
    },
    {
      "epoch": 14.053363610559183,
      "grad_norm": 1.6199156045913696,
      "learning_rate": 8.595344876525688e-05,
      "loss": 0.02848471999168396,
      "step": 49510
    },
    {
      "epoch": 14.056202100482544,
      "grad_norm": 5.574838638305664,
      "learning_rate": 8.595061027533353e-05,
      "loss": 0.02811778783798218,
      "step": 49520
    },
    {
      "epoch": 14.059040590405903,
      "grad_norm": 2.209955930709839,
      "learning_rate": 8.594777178541017e-05,
      "loss": 0.028638684749603273,
      "step": 49530
    },
    {
      "epoch": 14.061879080329264,
      "grad_norm": 2.9624717235565186,
      "learning_rate": 8.594493329548681e-05,
      "loss": 0.0230121374130249,
      "step": 49540
    },
    {
      "epoch": 14.064717570252625,
      "grad_norm": 7.450040340423584,
      "learning_rate": 8.594209480556345e-05,
      "loss": 0.035555633902549746,
      "step": 49550
    },
    {
      "epoch": 14.067556060175987,
      "grad_norm": 6.268996715545654,
      "learning_rate": 8.593925631564008e-05,
      "loss": 0.031611236929893496,
      "step": 49560
    },
    {
      "epoch": 14.070394550099348,
      "grad_norm": 5.829379081726074,
      "learning_rate": 8.593641782571672e-05,
      "loss": 0.04754213988780975,
      "step": 49570
    },
    {
      "epoch": 14.073233040022707,
      "grad_norm": 12.128203392028809,
      "learning_rate": 8.593357933579336e-05,
      "loss": 0.040772438049316406,
      "step": 49580
    },
    {
      "epoch": 14.076071529946068,
      "grad_norm": 1.4006669521331787,
      "learning_rate": 8.593074084587e-05,
      "loss": 0.02982007563114166,
      "step": 49590
    },
    {
      "epoch": 14.07891001986943,
      "grad_norm": 4.344081401824951,
      "learning_rate": 8.592790235594664e-05,
      "loss": 0.015107987821102143,
      "step": 49600
    },
    {
      "epoch": 14.08174850979279,
      "grad_norm": 4.288851737976074,
      "learning_rate": 8.592506386602328e-05,
      "loss": 0.02266688793897629,
      "step": 49610
    },
    {
      "epoch": 14.084586999716151,
      "grad_norm": 9.014312744140625,
      "learning_rate": 8.592222537609991e-05,
      "loss": 0.04199548959732056,
      "step": 49620
    },
    {
      "epoch": 14.087425489639513,
      "grad_norm": 6.336841106414795,
      "learning_rate": 8.591938688617655e-05,
      "loss": 0.03385908305644989,
      "step": 49630
    },
    {
      "epoch": 14.090263979562872,
      "grad_norm": 4.437830924987793,
      "learning_rate": 8.59165483962532e-05,
      "loss": 0.03252928256988526,
      "step": 49640
    },
    {
      "epoch": 14.093102469486233,
      "grad_norm": 2.8051395416259766,
      "learning_rate": 8.591370990632984e-05,
      "loss": 0.02804200053215027,
      "step": 49650
    },
    {
      "epoch": 14.095940959409594,
      "grad_norm": 8.211349487304688,
      "learning_rate": 8.591087141640648e-05,
      "loss": 0.029553130269050598,
      "step": 49660
    },
    {
      "epoch": 14.098779449332955,
      "grad_norm": 6.522387981414795,
      "learning_rate": 8.590803292648312e-05,
      "loss": 0.041052433848381045,
      "step": 49670
    },
    {
      "epoch": 14.101617939256316,
      "grad_norm": 1.68451988697052,
      "learning_rate": 8.590519443655976e-05,
      "loss": 0.03737498223781586,
      "step": 49680
    },
    {
      "epoch": 14.104456429179676,
      "grad_norm": 0.94016432762146,
      "learning_rate": 8.590235594663639e-05,
      "loss": 0.024851275980472563,
      "step": 49690
    },
    {
      "epoch": 14.107294919103037,
      "grad_norm": 4.330970764160156,
      "learning_rate": 8.589951745671303e-05,
      "loss": 0.01902497559785843,
      "step": 49700
    },
    {
      "epoch": 14.110133409026398,
      "grad_norm": 9.700041770935059,
      "learning_rate": 8.589667896678967e-05,
      "loss": 0.01783207356929779,
      "step": 49710
    },
    {
      "epoch": 14.112971898949759,
      "grad_norm": 1.798171043395996,
      "learning_rate": 8.589384047686631e-05,
      "loss": 0.030052945017814636,
      "step": 49720
    },
    {
      "epoch": 14.11581038887312,
      "grad_norm": 0.37163740396499634,
      "learning_rate": 8.589100198694295e-05,
      "loss": 0.03718210458755493,
      "step": 49730
    },
    {
      "epoch": 14.11864887879648,
      "grad_norm": 1.8951197862625122,
      "learning_rate": 8.58881634970196e-05,
      "loss": 0.02399684488773346,
      "step": 49740
    },
    {
      "epoch": 14.12148736871984,
      "grad_norm": 5.541714668273926,
      "learning_rate": 8.588532500709622e-05,
      "loss": 0.03576144576072693,
      "step": 49750
    },
    {
      "epoch": 14.124325858643202,
      "grad_norm": 0.4724043905735016,
      "learning_rate": 8.588248651717286e-05,
      "loss": 0.025749751925468446,
      "step": 49760
    },
    {
      "epoch": 14.127164348566563,
      "grad_norm": 5.300692081451416,
      "learning_rate": 8.58796480272495e-05,
      "loss": 0.017452608048915862,
      "step": 49770
    },
    {
      "epoch": 14.130002838489924,
      "grad_norm": 0.691306471824646,
      "learning_rate": 8.587680953732615e-05,
      "loss": 0.014005993306636811,
      "step": 49780
    },
    {
      "epoch": 14.132841328413285,
      "grad_norm": 10.76020336151123,
      "learning_rate": 8.587397104740279e-05,
      "loss": 0.041131556034088135,
      "step": 49790
    },
    {
      "epoch": 14.135679818336644,
      "grad_norm": 3.684041976928711,
      "learning_rate": 8.587113255747943e-05,
      "loss": 0.023859602212905884,
      "step": 49800
    },
    {
      "epoch": 14.138518308260005,
      "grad_norm": 6.936976909637451,
      "learning_rate": 8.586829406755607e-05,
      "loss": 0.015928187966346742,
      "step": 49810
    },
    {
      "epoch": 14.141356798183367,
      "grad_norm": 6.7339091300964355,
      "learning_rate": 8.58654555776327e-05,
      "loss": 0.023181793093681336,
      "step": 49820
    },
    {
      "epoch": 14.144195288106728,
      "grad_norm": 9.926058769226074,
      "learning_rate": 8.586261708770934e-05,
      "loss": 0.031194770336151124,
      "step": 49830
    },
    {
      "epoch": 14.147033778030089,
      "grad_norm": 3.7394731044769287,
      "learning_rate": 8.585977859778598e-05,
      "loss": 0.03421114981174469,
      "step": 49840
    },
    {
      "epoch": 14.149872267953448,
      "grad_norm": 10.771072387695312,
      "learning_rate": 8.585694010786261e-05,
      "loss": 0.023525959253311156,
      "step": 49850
    },
    {
      "epoch": 14.15271075787681,
      "grad_norm": 3.9890058040618896,
      "learning_rate": 8.585410161793926e-05,
      "loss": 0.04136554598808288,
      "step": 49860
    },
    {
      "epoch": 14.15554924780017,
      "grad_norm": 3.086162567138672,
      "learning_rate": 8.58512631280159e-05,
      "loss": 0.033159708976745604,
      "step": 49870
    },
    {
      "epoch": 14.158387737723531,
      "grad_norm": 2.5958006381988525,
      "learning_rate": 8.584842463809253e-05,
      "loss": 0.03257180154323578,
      "step": 49880
    },
    {
      "epoch": 14.161226227646893,
      "grad_norm": 1.4766089916229248,
      "learning_rate": 8.584558614816917e-05,
      "loss": 0.025836512446403503,
      "step": 49890
    },
    {
      "epoch": 14.164064717570252,
      "grad_norm": 12.105684280395508,
      "learning_rate": 8.584274765824582e-05,
      "loss": 0.02869676351547241,
      "step": 49900
    },
    {
      "epoch": 14.166903207493613,
      "grad_norm": 6.916757583618164,
      "learning_rate": 8.583990916832246e-05,
      "loss": 0.01802630126476288,
      "step": 49910
    },
    {
      "epoch": 14.169741697416974,
      "grad_norm": 4.512959003448486,
      "learning_rate": 8.58370706783991e-05,
      "loss": 0.01660865545272827,
      "step": 49920
    },
    {
      "epoch": 14.172580187340335,
      "grad_norm": 6.5449981689453125,
      "learning_rate": 8.583423218847574e-05,
      "loss": 0.023651117086410524,
      "step": 49930
    },
    {
      "epoch": 14.175418677263696,
      "grad_norm": 1.095157265663147,
      "learning_rate": 8.583139369855238e-05,
      "loss": 0.027622419595718383,
      "step": 49940
    },
    {
      "epoch": 14.178257167187056,
      "grad_norm": 4.747420310974121,
      "learning_rate": 8.582855520862901e-05,
      "loss": 0.03191403746604919,
      "step": 49950
    },
    {
      "epoch": 14.181095657110417,
      "grad_norm": 2.007878065109253,
      "learning_rate": 8.582571671870565e-05,
      "loss": 0.04082639217376709,
      "step": 49960
    },
    {
      "epoch": 14.183934147033778,
      "grad_norm": 18.160432815551758,
      "learning_rate": 8.582287822878229e-05,
      "loss": 0.025690150260925294,
      "step": 49970
    },
    {
      "epoch": 14.186772636957139,
      "grad_norm": 4.28290319442749,
      "learning_rate": 8.582003973885892e-05,
      "loss": 0.02678082585334778,
      "step": 49980
    },
    {
      "epoch": 14.1896111268805,
      "grad_norm": 11.701045036315918,
      "learning_rate": 8.581720124893558e-05,
      "loss": 0.04468222260475159,
      "step": 49990
    },
    {
      "epoch": 14.192449616803861,
      "grad_norm": 7.335849761962891,
      "learning_rate": 8.581436275901222e-05,
      "loss": 0.027931097149848937,
      "step": 50000
    },
    {
      "epoch": 14.192449616803861,
      "eval_accuracy": 0.9580975392636867,
      "eval_loss": 0.12964147329330444,
      "eval_runtime": 31.689,
      "eval_samples_per_second": 496.293,
      "eval_steps_per_second": 7.763,
      "step": 50000
    },
    {
      "epoch": 14.19528810672722,
      "grad_norm": 3.745678663253784,
      "learning_rate": 8.581152426908884e-05,
      "loss": 0.01775626391172409,
      "step": 50010
    },
    {
      "epoch": 14.198126596650582,
      "grad_norm": 3.5942907333374023,
      "learning_rate": 8.580868577916549e-05,
      "loss": 0.01917008012533188,
      "step": 50020
    },
    {
      "epoch": 14.200965086573943,
      "grad_norm": 12.183575630187988,
      "learning_rate": 8.580584728924213e-05,
      "loss": 0.031027162075042726,
      "step": 50030
    },
    {
      "epoch": 14.203803576497304,
      "grad_norm": 1.2091981172561646,
      "learning_rate": 8.580300879931877e-05,
      "loss": 0.00841534286737442,
      "step": 50040
    },
    {
      "epoch": 14.206642066420665,
      "grad_norm": 11.461080551147461,
      "learning_rate": 8.580017030939541e-05,
      "loss": 0.017339488863945006,
      "step": 50050
    },
    {
      "epoch": 14.209480556344024,
      "grad_norm": 5.73613166809082,
      "learning_rate": 8.579733181947205e-05,
      "loss": 0.032194784283638,
      "step": 50060
    },
    {
      "epoch": 14.212319046267385,
      "grad_norm": 3.822580099105835,
      "learning_rate": 8.579449332954869e-05,
      "loss": 0.03832382559776306,
      "step": 50070
    },
    {
      "epoch": 14.215157536190747,
      "grad_norm": 6.792261123657227,
      "learning_rate": 8.579165483962532e-05,
      "loss": 0.02406049370765686,
      "step": 50080
    },
    {
      "epoch": 14.217996026114108,
      "grad_norm": 6.343382358551025,
      "learning_rate": 8.578881634970196e-05,
      "loss": 0.021263013780117034,
      "step": 50090
    },
    {
      "epoch": 14.220834516037469,
      "grad_norm": 4.553318977355957,
      "learning_rate": 8.57859778597786e-05,
      "loss": 0.02685590982437134,
      "step": 50100
    },
    {
      "epoch": 14.223673005960828,
      "grad_norm": 7.7648844718933105,
      "learning_rate": 8.578313936985523e-05,
      "loss": 0.02674403488636017,
      "step": 50110
    },
    {
      "epoch": 14.22651149588419,
      "grad_norm": 1.3327938318252563,
      "learning_rate": 8.578030087993189e-05,
      "loss": 0.03879241645336151,
      "step": 50120
    },
    {
      "epoch": 14.22934998580755,
      "grad_norm": 4.96824312210083,
      "learning_rate": 8.577746239000853e-05,
      "loss": 0.017473468184471132,
      "step": 50130
    },
    {
      "epoch": 14.232188475730911,
      "grad_norm": 4.059240818023682,
      "learning_rate": 8.577462390008515e-05,
      "loss": 0.019765456020832062,
      "step": 50140
    },
    {
      "epoch": 14.235026965654273,
      "grad_norm": 1.62040114402771,
      "learning_rate": 8.57717854101618e-05,
      "loss": 0.029139450192451476,
      "step": 50150
    },
    {
      "epoch": 14.237865455577632,
      "grad_norm": 13.126379013061523,
      "learning_rate": 8.576894692023844e-05,
      "loss": 0.03385429084300995,
      "step": 50160
    },
    {
      "epoch": 14.240703945500993,
      "grad_norm": 18.521568298339844,
      "learning_rate": 8.576610843031508e-05,
      "loss": 0.032480287551879886,
      "step": 50170
    },
    {
      "epoch": 14.243542435424354,
      "grad_norm": 1.7817473411560059,
      "learning_rate": 8.576326994039171e-05,
      "loss": 0.03137780725955963,
      "step": 50180
    },
    {
      "epoch": 14.246380925347715,
      "grad_norm": 2.8078908920288086,
      "learning_rate": 8.576043145046836e-05,
      "loss": 0.029249498248100282,
      "step": 50190
    },
    {
      "epoch": 14.249219415271076,
      "grad_norm": 0.9359973669052124,
      "learning_rate": 8.5757592960545e-05,
      "loss": 0.02133913040161133,
      "step": 50200
    },
    {
      "epoch": 14.252057905194437,
      "grad_norm": 8.117232322692871,
      "learning_rate": 8.575475447062163e-05,
      "loss": 0.03519972562789917,
      "step": 50210
    },
    {
      "epoch": 14.254896395117797,
      "grad_norm": 10.54971694946289,
      "learning_rate": 8.575191598069827e-05,
      "loss": 0.041789859533309937,
      "step": 50220
    },
    {
      "epoch": 14.257734885041158,
      "grad_norm": 9.854226112365723,
      "learning_rate": 8.574907749077491e-05,
      "loss": 0.026445215940475462,
      "step": 50230
    },
    {
      "epoch": 14.260573374964519,
      "grad_norm": 8.530778884887695,
      "learning_rate": 8.574623900085154e-05,
      "loss": 0.025764849781990052,
      "step": 50240
    },
    {
      "epoch": 14.26341186488788,
      "grad_norm": 7.195922374725342,
      "learning_rate": 8.57434005109282e-05,
      "loss": 0.022102949023246766,
      "step": 50250
    },
    {
      "epoch": 14.266250354811241,
      "grad_norm": 5.125481128692627,
      "learning_rate": 8.574056202100484e-05,
      "loss": 0.03041970729827881,
      "step": 50260
    },
    {
      "epoch": 14.2690888447346,
      "grad_norm": 12.072976112365723,
      "learning_rate": 8.573772353108147e-05,
      "loss": 0.043194171786308286,
      "step": 50270
    },
    {
      "epoch": 14.271927334657962,
      "grad_norm": 0.5271238684654236,
      "learning_rate": 8.573488504115811e-05,
      "loss": 0.021397794783115386,
      "step": 50280
    },
    {
      "epoch": 14.274765824581323,
      "grad_norm": 7.865839958190918,
      "learning_rate": 8.573204655123475e-05,
      "loss": 0.015236221253871918,
      "step": 50290
    },
    {
      "epoch": 14.277604314504684,
      "grad_norm": 3.6244921684265137,
      "learning_rate": 8.572920806131139e-05,
      "loss": 0.02463509887456894,
      "step": 50300
    },
    {
      "epoch": 14.280442804428045,
      "grad_norm": 17.5566463470459,
      "learning_rate": 8.572636957138802e-05,
      "loss": 0.058706659078598025,
      "step": 50310
    },
    {
      "epoch": 14.283281294351404,
      "grad_norm": 2.5470423698425293,
      "learning_rate": 8.572353108146467e-05,
      "loss": 0.027238631248474122,
      "step": 50320
    },
    {
      "epoch": 14.286119784274765,
      "grad_norm": 14.131685256958008,
      "learning_rate": 8.572069259154131e-05,
      "loss": 0.02821727395057678,
      "step": 50330
    },
    {
      "epoch": 14.288958274198126,
      "grad_norm": 9.750001907348633,
      "learning_rate": 8.571785410161794e-05,
      "loss": 0.0328499972820282,
      "step": 50340
    },
    {
      "epoch": 14.291796764121488,
      "grad_norm": 6.270340919494629,
      "learning_rate": 8.571501561169458e-05,
      "loss": 0.042112496495246884,
      "step": 50350
    },
    {
      "epoch": 14.294635254044849,
      "grad_norm": 0.8371354341506958,
      "learning_rate": 8.571217712177122e-05,
      "loss": 0.018076685070991517,
      "step": 50360
    },
    {
      "epoch": 14.297473743968208,
      "grad_norm": 8.482423782348633,
      "learning_rate": 8.570933863184785e-05,
      "loss": 0.02399584650993347,
      "step": 50370
    },
    {
      "epoch": 14.30031223389157,
      "grad_norm": 18.69969367980957,
      "learning_rate": 8.57065001419245e-05,
      "loss": 0.02841760218143463,
      "step": 50380
    },
    {
      "epoch": 14.30315072381493,
      "grad_norm": 1.6346039772033691,
      "learning_rate": 8.570366165200115e-05,
      "loss": 0.021412229537963866,
      "step": 50390
    },
    {
      "epoch": 14.305989213738291,
      "grad_norm": 12.068750381469727,
      "learning_rate": 8.570082316207778e-05,
      "loss": 0.03126122355461121,
      "step": 50400
    },
    {
      "epoch": 14.308827703661652,
      "grad_norm": 12.208056449890137,
      "learning_rate": 8.569798467215442e-05,
      "loss": 0.03616892397403717,
      "step": 50410
    },
    {
      "epoch": 14.311666193585014,
      "grad_norm": 6.67946195602417,
      "learning_rate": 8.569514618223106e-05,
      "loss": 0.029627156257629395,
      "step": 50420
    },
    {
      "epoch": 14.314504683508373,
      "grad_norm": 3.8765785694122314,
      "learning_rate": 8.56923076923077e-05,
      "loss": 0.032170730829238894,
      "step": 50430
    },
    {
      "epoch": 14.317343173431734,
      "grad_norm": 10.998615264892578,
      "learning_rate": 8.568946920238433e-05,
      "loss": 0.05253680944442749,
      "step": 50440
    },
    {
      "epoch": 14.320181663355095,
      "grad_norm": 0.28679242730140686,
      "learning_rate": 8.568663071246098e-05,
      "loss": 0.01667790859937668,
      "step": 50450
    },
    {
      "epoch": 14.323020153278456,
      "grad_norm": 1.6101802587509155,
      "learning_rate": 8.568379222253761e-05,
      "loss": 0.05359309911727905,
      "step": 50460
    },
    {
      "epoch": 14.325858643201817,
      "grad_norm": 0.8244699239730835,
      "learning_rate": 8.568095373261425e-05,
      "loss": 0.02687566578388214,
      "step": 50470
    },
    {
      "epoch": 14.328697133125177,
      "grad_norm": 15.569474220275879,
      "learning_rate": 8.56781152426909e-05,
      "loss": 0.023008763790130615,
      "step": 50480
    },
    {
      "epoch": 14.331535623048538,
      "grad_norm": 3.032259225845337,
      "learning_rate": 8.567527675276754e-05,
      "loss": 0.03624162673950195,
      "step": 50490
    },
    {
      "epoch": 14.334374112971899,
      "grad_norm": 14.733590126037598,
      "learning_rate": 8.567272211183652e-05,
      "loss": 0.08000153303146362,
      "step": 50500
    },
    {
      "epoch": 14.334374112971899,
      "eval_accuracy": 0.9556177274750429,
      "eval_loss": 0.1320401430130005,
      "eval_runtime": 31.5257,
      "eval_samples_per_second": 498.862,
      "eval_steps_per_second": 7.803,
      "step": 50500
    },
    {
      "epoch": 14.33721260289526,
      "grad_norm": 4.095160007476807,
      "learning_rate": 8.566988362191314e-05,
      "loss": 0.013829153776168824,
      "step": 50510
    },
    {
      "epoch": 14.340051092818621,
      "grad_norm": 5.885178565979004,
      "learning_rate": 8.566704513198978e-05,
      "loss": 0.019609487056732176,
      "step": 50520
    },
    {
      "epoch": 14.34288958274198,
      "grad_norm": 1.6446106433868408,
      "learning_rate": 8.566420664206643e-05,
      "loss": 0.029877638816833495,
      "step": 50530
    },
    {
      "epoch": 14.345728072665342,
      "grad_norm": 1.0699890851974487,
      "learning_rate": 8.566136815214307e-05,
      "loss": 0.02531459629535675,
      "step": 50540
    },
    {
      "epoch": 14.348566562588703,
      "grad_norm": 2.2033169269561768,
      "learning_rate": 8.56585296622197e-05,
      "loss": 0.027001941204071046,
      "step": 50550
    },
    {
      "epoch": 14.351405052512064,
      "grad_norm": 17.86330795288086,
      "learning_rate": 8.565569117229634e-05,
      "loss": 0.04457869231700897,
      "step": 50560
    },
    {
      "epoch": 14.354243542435425,
      "grad_norm": 7.273069381713867,
      "learning_rate": 8.565285268237299e-05,
      "loss": 0.04515500962734222,
      "step": 50570
    },
    {
      "epoch": 14.357082032358786,
      "grad_norm": 2.293480396270752,
      "learning_rate": 8.565001419244962e-05,
      "loss": 0.02725212275981903,
      "step": 50580
    },
    {
      "epoch": 14.359920522282145,
      "grad_norm": 6.626486301422119,
      "learning_rate": 8.564717570252626e-05,
      "loss": 0.03408137559890747,
      "step": 50590
    },
    {
      "epoch": 14.362759012205506,
      "grad_norm": 12.899566650390625,
      "learning_rate": 8.56443372126029e-05,
      "loss": 0.03901633620262146,
      "step": 50600
    },
    {
      "epoch": 14.365597502128868,
      "grad_norm": 5.514700412750244,
      "learning_rate": 8.564149872267953e-05,
      "loss": 0.0323235034942627,
      "step": 50610
    },
    {
      "epoch": 14.368435992052229,
      "grad_norm": 0.6008784174919128,
      "learning_rate": 8.563866023275617e-05,
      "loss": 0.020973995327949524,
      "step": 50620
    },
    {
      "epoch": 14.37127448197559,
      "grad_norm": 8.451521873474121,
      "learning_rate": 8.563582174283283e-05,
      "loss": 0.0300495445728302,
      "step": 50630
    },
    {
      "epoch": 14.374112971898949,
      "grad_norm": 4.776422500610352,
      "learning_rate": 8.563298325290945e-05,
      "loss": 0.03752357959747314,
      "step": 50640
    },
    {
      "epoch": 14.37695146182231,
      "grad_norm": 8.96488094329834,
      "learning_rate": 8.56301447629861e-05,
      "loss": 0.04631670713424683,
      "step": 50650
    },
    {
      "epoch": 14.379789951745671,
      "grad_norm": 12.579450607299805,
      "learning_rate": 8.562730627306274e-05,
      "loss": 0.03514888882637024,
      "step": 50660
    },
    {
      "epoch": 14.382628441669032,
      "grad_norm": 6.835667610168457,
      "learning_rate": 8.562446778313938e-05,
      "loss": 0.031214013695716858,
      "step": 50670
    },
    {
      "epoch": 14.385466931592394,
      "grad_norm": 10.591228485107422,
      "learning_rate": 8.5621629293216e-05,
      "loss": 0.045433837175369265,
      "step": 50680
    },
    {
      "epoch": 14.388305421515753,
      "grad_norm": 4.52405309677124,
      "learning_rate": 8.561879080329265e-05,
      "loss": 0.014749705791473389,
      "step": 50690
    },
    {
      "epoch": 14.391143911439114,
      "grad_norm": 5.818645477294922,
      "learning_rate": 8.56159523133693e-05,
      "loss": 0.030107370018959044,
      "step": 50700
    },
    {
      "epoch": 14.393982401362475,
      "grad_norm": 13.904401779174805,
      "learning_rate": 8.561311382344593e-05,
      "loss": 0.039076763391494754,
      "step": 50710
    },
    {
      "epoch": 14.396820891285836,
      "grad_norm": 6.508571624755859,
      "learning_rate": 8.561027533352257e-05,
      "loss": 0.06043229103088379,
      "step": 50720
    },
    {
      "epoch": 14.399659381209197,
      "grad_norm": 5.219316482543945,
      "learning_rate": 8.560743684359921e-05,
      "loss": 0.02476014941930771,
      "step": 50730
    },
    {
      "epoch": 14.402497871132557,
      "grad_norm": 12.049295425415039,
      "learning_rate": 8.560459835367584e-05,
      "loss": 0.02449103593826294,
      "step": 50740
    },
    {
      "epoch": 14.405336361055918,
      "grad_norm": 5.143094062805176,
      "learning_rate": 8.560175986375248e-05,
      "loss": 0.019943231344223024,
      "step": 50750
    },
    {
      "epoch": 14.408174850979279,
      "grad_norm": 14.276522636413574,
      "learning_rate": 8.559892137382912e-05,
      "loss": 0.034413126111030576,
      "step": 50760
    },
    {
      "epoch": 14.41101334090264,
      "grad_norm": 14.80979061126709,
      "learning_rate": 8.559608288390576e-05,
      "loss": 0.03124769628047943,
      "step": 50770
    },
    {
      "epoch": 14.413851830826001,
      "grad_norm": 13.902080535888672,
      "learning_rate": 8.55932443939824e-05,
      "loss": 0.03595017194747925,
      "step": 50780
    },
    {
      "epoch": 14.416690320749362,
      "grad_norm": 4.783239364624023,
      "learning_rate": 8.559040590405905e-05,
      "loss": 0.018854564428329466,
      "step": 50790
    },
    {
      "epoch": 14.419528810672722,
      "grad_norm": 11.942090034484863,
      "learning_rate": 8.558756741413569e-05,
      "loss": 0.046934318542480466,
      "step": 50800
    },
    {
      "epoch": 14.422367300596083,
      "grad_norm": 6.899314880371094,
      "learning_rate": 8.558472892421232e-05,
      "loss": 0.020881219208240508,
      "step": 50810
    },
    {
      "epoch": 14.425205790519444,
      "grad_norm": 1.4761505126953125,
      "learning_rate": 8.558189043428896e-05,
      "loss": 0.04355973303318024,
      "step": 50820
    },
    {
      "epoch": 14.428044280442805,
      "grad_norm": 9.936765670776367,
      "learning_rate": 8.557905194436561e-05,
      "loss": 0.025480246543884276,
      "step": 50830
    },
    {
      "epoch": 14.430882770366166,
      "grad_norm": 7.196719646453857,
      "learning_rate": 8.557621345444224e-05,
      "loss": 0.051013869047164914,
      "step": 50840
    },
    {
      "epoch": 14.433721260289525,
      "grad_norm": 7.83380651473999,
      "learning_rate": 8.557337496451888e-05,
      "loss": 0.026437598466873168,
      "step": 50850
    },
    {
      "epoch": 14.436559750212886,
      "grad_norm": 2.26269793510437,
      "learning_rate": 8.557053647459552e-05,
      "loss": 0.029989293217658995,
      "step": 50860
    },
    {
      "epoch": 14.439398240136248,
      "grad_norm": 1.4525343179702759,
      "learning_rate": 8.556769798467215e-05,
      "loss": 0.033813050389289855,
      "step": 50870
    },
    {
      "epoch": 14.442236730059609,
      "grad_norm": 8.670069694519043,
      "learning_rate": 8.556485949474879e-05,
      "loss": 0.03728064000606537,
      "step": 50880
    },
    {
      "epoch": 14.44507521998297,
      "grad_norm": 7.66602897644043,
      "learning_rate": 8.556202100482543e-05,
      "loss": 0.047035437822341916,
      "step": 50890
    },
    {
      "epoch": 14.447913709906329,
      "grad_norm": 0.837008535861969,
      "learning_rate": 8.555918251490208e-05,
      "loss": 0.025792202353477477,
      "step": 50900
    },
    {
      "epoch": 14.45075219982969,
      "grad_norm": 0.4187609553337097,
      "learning_rate": 8.555634402497872e-05,
      "loss": 0.032220259308815,
      "step": 50910
    },
    {
      "epoch": 14.453590689753051,
      "grad_norm": 3.26013445854187,
      "learning_rate": 8.555350553505536e-05,
      "loss": 0.034147176146507266,
      "step": 50920
    },
    {
      "epoch": 14.456429179676412,
      "grad_norm": 14.086835861206055,
      "learning_rate": 8.5550667045132e-05,
      "loss": 0.023838698863983154,
      "step": 50930
    },
    {
      "epoch": 14.459267669599773,
      "grad_norm": 2.3969225883483887,
      "learning_rate": 8.554782855520863e-05,
      "loss": 0.025146037340164185,
      "step": 50940
    },
    {
      "epoch": 14.462106159523135,
      "grad_norm": 1.6130523681640625,
      "learning_rate": 8.554499006528527e-05,
      "loss": 0.02198980748653412,
      "step": 50950
    },
    {
      "epoch": 14.464944649446494,
      "grad_norm": 2.830883502960205,
      "learning_rate": 8.554215157536191e-05,
      "loss": 0.02476453483104706,
      "step": 50960
    },
    {
      "epoch": 14.467783139369855,
      "grad_norm": 13.475018501281738,
      "learning_rate": 8.553931308543855e-05,
      "loss": 0.034607219696044925,
      "step": 50970
    },
    {
      "epoch": 14.470621629293216,
      "grad_norm": 2.6382336616516113,
      "learning_rate": 8.553647459551519e-05,
      "loss": 0.016006499528884888,
      "step": 50980
    },
    {
      "epoch": 14.473460119216577,
      "grad_norm": 1.152177333831787,
      "learning_rate": 8.553363610559183e-05,
      "loss": 0.037303170561790465,
      "step": 50990
    },
    {
      "epoch": 14.476298609139938,
      "grad_norm": 0.9025798439979553,
      "learning_rate": 8.553079761566846e-05,
      "loss": 0.03282157778739929,
      "step": 51000
    },
    {
      "epoch": 14.476298609139938,
      "eval_accuracy": 0.9616582946525084,
      "eval_loss": 0.11275384575128555,
      "eval_runtime": 31.983,
      "eval_samples_per_second": 491.729,
      "eval_steps_per_second": 7.692,
      "step": 51000
    },
    {
      "epoch": 14.479137099063298,
      "grad_norm": 4.254899501800537,
      "learning_rate": 8.55279591257451e-05,
      "loss": 0.0281696081161499,
      "step": 51010
    },
    {
      "epoch": 14.481975588986659,
      "grad_norm": 3.0414316654205322,
      "learning_rate": 8.552512063582174e-05,
      "loss": 0.02458951771259308,
      "step": 51020
    },
    {
      "epoch": 14.48481407891002,
      "grad_norm": 3.1042959690093994,
      "learning_rate": 8.552228214589839e-05,
      "loss": 0.02192099541425705,
      "step": 51030
    },
    {
      "epoch": 14.487652568833381,
      "grad_norm": 0.6739271283149719,
      "learning_rate": 8.551944365597503e-05,
      "loss": 0.02606382668018341,
      "step": 51040
    },
    {
      "epoch": 14.490491058756742,
      "grad_norm": 19.049753189086914,
      "learning_rate": 8.551660516605167e-05,
      "loss": 0.02903587222099304,
      "step": 51050
    },
    {
      "epoch": 14.493329548680101,
      "grad_norm": 12.988877296447754,
      "learning_rate": 8.551376667612831e-05,
      "loss": 0.030290287733078004,
      "step": 51060
    },
    {
      "epoch": 14.496168038603463,
      "grad_norm": 19.155982971191406,
      "learning_rate": 8.551092818620494e-05,
      "loss": 0.05030482411384583,
      "step": 51070
    },
    {
      "epoch": 14.499006528526824,
      "grad_norm": 14.54448413848877,
      "learning_rate": 8.550808969628158e-05,
      "loss": 0.03043851852416992,
      "step": 51080
    },
    {
      "epoch": 14.501845018450185,
      "grad_norm": 3.08514666557312,
      "learning_rate": 8.550525120635822e-05,
      "loss": 0.02116159498691559,
      "step": 51090
    },
    {
      "epoch": 14.504683508373546,
      "grad_norm": 10.747303009033203,
      "learning_rate": 8.550241271643486e-05,
      "loss": 0.05740382671356201,
      "step": 51100
    },
    {
      "epoch": 14.507521998296905,
      "grad_norm": 10.09619426727295,
      "learning_rate": 8.54995742265115e-05,
      "loss": 0.043869960308074954,
      "step": 51110
    },
    {
      "epoch": 14.510360488220266,
      "grad_norm": 5.170281887054443,
      "learning_rate": 8.549673573658814e-05,
      "loss": 0.05161537528038025,
      "step": 51120
    },
    {
      "epoch": 14.513198978143627,
      "grad_norm": 2.299725294113159,
      "learning_rate": 8.549389724666477e-05,
      "loss": 0.037432721257209776,
      "step": 51130
    },
    {
      "epoch": 14.516037468066989,
      "grad_norm": 0.9760419130325317,
      "learning_rate": 8.549105875674141e-05,
      "loss": 0.022002269327640534,
      "step": 51140
    },
    {
      "epoch": 14.51887595799035,
      "grad_norm": 4.733566761016846,
      "learning_rate": 8.548822026681806e-05,
      "loss": 0.017149904370307924,
      "step": 51150
    },
    {
      "epoch": 14.521714447913709,
      "grad_norm": 5.653111457824707,
      "learning_rate": 8.54853817768947e-05,
      "loss": 0.019655078649520874,
      "step": 51160
    },
    {
      "epoch": 14.52455293783707,
      "grad_norm": 3.013505220413208,
      "learning_rate": 8.548254328697134e-05,
      "loss": 0.038420644402503965,
      "step": 51170
    },
    {
      "epoch": 14.527391427760431,
      "grad_norm": 6.026484966278076,
      "learning_rate": 8.547970479704798e-05,
      "loss": 0.03479229509830475,
      "step": 51180
    },
    {
      "epoch": 14.530229917683792,
      "grad_norm": 9.362969398498535,
      "learning_rate": 8.547686630712462e-05,
      "loss": 0.02241542041301727,
      "step": 51190
    },
    {
      "epoch": 14.533068407607153,
      "grad_norm": 2.119218587875366,
      "learning_rate": 8.547402781720125e-05,
      "loss": 0.030665215849876405,
      "step": 51200
    },
    {
      "epoch": 14.535906897530515,
      "grad_norm": 3.773319959640503,
      "learning_rate": 8.547118932727789e-05,
      "loss": 0.0242320716381073,
      "step": 51210
    },
    {
      "epoch": 14.538745387453874,
      "grad_norm": 12.859066009521484,
      "learning_rate": 8.546835083735453e-05,
      "loss": 0.04029773473739624,
      "step": 51220
    },
    {
      "epoch": 14.541583877377235,
      "grad_norm": 8.100462913513184,
      "learning_rate": 8.546551234743117e-05,
      "loss": 0.028128573298454286,
      "step": 51230
    },
    {
      "epoch": 14.544422367300596,
      "grad_norm": 6.918891429901123,
      "learning_rate": 8.546267385750781e-05,
      "loss": 0.035101976990699765,
      "step": 51240
    },
    {
      "epoch": 14.547260857223957,
      "grad_norm": 12.470224380493164,
      "learning_rate": 8.545983536758446e-05,
      "loss": 0.04570910930633545,
      "step": 51250
    },
    {
      "epoch": 14.550099347147318,
      "grad_norm": 11.28945255279541,
      "learning_rate": 8.545699687766108e-05,
      "loss": 0.057882821559906004,
      "step": 51260
    },
    {
      "epoch": 14.552937837070678,
      "grad_norm": 7.4758405685424805,
      "learning_rate": 8.545415838773772e-05,
      "loss": 0.017592470347881316,
      "step": 51270
    },
    {
      "epoch": 14.555776326994039,
      "grad_norm": 3.6670124530792236,
      "learning_rate": 8.545131989781437e-05,
      "loss": 0.01690032333135605,
      "step": 51280
    },
    {
      "epoch": 14.5586148169174,
      "grad_norm": 6.950485706329346,
      "learning_rate": 8.544848140789101e-05,
      "loss": 0.02661212682723999,
      "step": 51290
    },
    {
      "epoch": 14.561453306840761,
      "grad_norm": 12.661957740783691,
      "learning_rate": 8.544564291796765e-05,
      "loss": 0.04555579125881195,
      "step": 51300
    },
    {
      "epoch": 14.564291796764122,
      "grad_norm": 2.024977445602417,
      "learning_rate": 8.544280442804429e-05,
      "loss": 0.03390985727310181,
      "step": 51310
    },
    {
      "epoch": 14.567130286687481,
      "grad_norm": 2.6837615966796875,
      "learning_rate": 8.543996593812092e-05,
      "loss": 0.021190357208251954,
      "step": 51320
    },
    {
      "epoch": 14.569968776610843,
      "grad_norm": 7.549190521240234,
      "learning_rate": 8.543712744819756e-05,
      "loss": 0.04251736998558044,
      "step": 51330
    },
    {
      "epoch": 14.572807266534204,
      "grad_norm": 8.728250503540039,
      "learning_rate": 8.54342889582742e-05,
      "loss": 0.024599093198776244,
      "step": 51340
    },
    {
      "epoch": 14.575645756457565,
      "grad_norm": 0.6689748167991638,
      "learning_rate": 8.543145046835084e-05,
      "loss": 0.029670712351799012,
      "step": 51350
    },
    {
      "epoch": 14.578484246380926,
      "grad_norm": 8.635516166687012,
      "learning_rate": 8.542861197842747e-05,
      "loss": 0.037997931241989136,
      "step": 51360
    },
    {
      "epoch": 14.581322736304287,
      "grad_norm": 1.4663504362106323,
      "learning_rate": 8.542577348850413e-05,
      "loss": 0.020631946623325348,
      "step": 51370
    },
    {
      "epoch": 14.584161226227646,
      "grad_norm": 1.7536824941635132,
      "learning_rate": 8.542293499858077e-05,
      "loss": 0.023515698313713074,
      "step": 51380
    },
    {
      "epoch": 14.586999716151007,
      "grad_norm": 3.2846157550811768,
      "learning_rate": 8.54200965086574e-05,
      "loss": 0.031092366576194762,
      "step": 51390
    },
    {
      "epoch": 14.589838206074369,
      "grad_norm": 4.611798286437988,
      "learning_rate": 8.541725801873404e-05,
      "loss": 0.027603542804718016,
      "step": 51400
    },
    {
      "epoch": 14.59267669599773,
      "grad_norm": 2.8955039978027344,
      "learning_rate": 8.541441952881068e-05,
      "loss": 0.027316337823867796,
      "step": 51410
    },
    {
      "epoch": 14.59551518592109,
      "grad_norm": 1.4328004121780396,
      "learning_rate": 8.54115810388873e-05,
      "loss": 0.010398831963539124,
      "step": 51420
    },
    {
      "epoch": 14.59835367584445,
      "grad_norm": 6.431270599365234,
      "learning_rate": 8.540874254896396e-05,
      "loss": 0.020113934576511384,
      "step": 51430
    },
    {
      "epoch": 14.601192165767811,
      "grad_norm": 8.832716941833496,
      "learning_rate": 8.54059040590406e-05,
      "loss": 0.03382132351398468,
      "step": 51440
    },
    {
      "epoch": 14.604030655691172,
      "grad_norm": 7.1022233963012695,
      "learning_rate": 8.540306556911723e-05,
      "loss": 0.04306955337524414,
      "step": 51450
    },
    {
      "epoch": 14.606869145614533,
      "grad_norm": 8.230582237243652,
      "learning_rate": 8.540022707919387e-05,
      "loss": 0.042462009191513064,
      "step": 51460
    },
    {
      "epoch": 14.609707635537895,
      "grad_norm": 1.2157502174377441,
      "learning_rate": 8.539738858927051e-05,
      "loss": 0.022740651667118073,
      "step": 51470
    },
    {
      "epoch": 14.612546125461254,
      "grad_norm": 7.07900857925415,
      "learning_rate": 8.539455009934715e-05,
      "loss": 0.045801737904548646,
      "step": 51480
    },
    {
      "epoch": 14.615384615384615,
      "grad_norm": 9.022019386291504,
      "learning_rate": 8.539171160942378e-05,
      "loss": 0.020640844106674196,
      "step": 51490
    },
    {
      "epoch": 14.618223105307976,
      "grad_norm": 2.2961812019348145,
      "learning_rate": 8.538887311950044e-05,
      "loss": 0.012333531677722932,
      "step": 51500
    },
    {
      "epoch": 14.618223105307976,
      "eval_accuracy": 0.9654733897119603,
      "eval_loss": 0.11064077168703079,
      "eval_runtime": 31.5427,
      "eval_samples_per_second": 498.594,
      "eval_steps_per_second": 7.799,
      "step": 51500
    },
    {
      "epoch": 14.621061595231337,
      "grad_norm": 2.7893013954162598,
      "learning_rate": 8.538603462957708e-05,
      "loss": 0.022663894295692443,
      "step": 51510
    },
    {
      "epoch": 14.623900085154698,
      "grad_norm": 9.522071838378906,
      "learning_rate": 8.53831961396537e-05,
      "loss": 0.05162531733512878,
      "step": 51520
    },
    {
      "epoch": 14.626738575078058,
      "grad_norm": 1.3576302528381348,
      "learning_rate": 8.538035764973035e-05,
      "loss": 0.0386125922203064,
      "step": 51530
    },
    {
      "epoch": 14.629577065001419,
      "grad_norm": 15.340320587158203,
      "learning_rate": 8.537751915980699e-05,
      "loss": 0.0510531485080719,
      "step": 51540
    },
    {
      "epoch": 14.63241555492478,
      "grad_norm": 2.0894927978515625,
      "learning_rate": 8.537468066988362e-05,
      "loss": 0.028602179884910584,
      "step": 51550
    },
    {
      "epoch": 14.635254044848141,
      "grad_norm": 9.491701126098633,
      "learning_rate": 8.537184217996026e-05,
      "loss": 0.027068281173706056,
      "step": 51560
    },
    {
      "epoch": 14.638092534771502,
      "grad_norm": 2.4531548023223877,
      "learning_rate": 8.536900369003691e-05,
      "loss": 0.020249664783477783,
      "step": 51570
    },
    {
      "epoch": 14.640931024694861,
      "grad_norm": 9.986639976501465,
      "learning_rate": 8.536616520011354e-05,
      "loss": 0.025724044442176817,
      "step": 51580
    },
    {
      "epoch": 14.643769514618223,
      "grad_norm": 2.639873743057251,
      "learning_rate": 8.536332671019018e-05,
      "loss": 0.021254521608352662,
      "step": 51590
    },
    {
      "epoch": 14.646608004541584,
      "grad_norm": 6.323326587677002,
      "learning_rate": 8.536048822026682e-05,
      "loss": 0.03456567525863648,
      "step": 51600
    },
    {
      "epoch": 14.649446494464945,
      "grad_norm": 1.1075998544692993,
      "learning_rate": 8.535764973034346e-05,
      "loss": 0.039362508058547976,
      "step": 51610
    },
    {
      "epoch": 14.652284984388306,
      "grad_norm": 11.35267162322998,
      "learning_rate": 8.535481124042009e-05,
      "loss": 0.03138631880283356,
      "step": 51620
    },
    {
      "epoch": 14.655123474311667,
      "grad_norm": 0.9547427892684937,
      "learning_rate": 8.535197275049675e-05,
      "loss": 0.01574273854494095,
      "step": 51630
    },
    {
      "epoch": 14.657961964235026,
      "grad_norm": 5.166176795959473,
      "learning_rate": 8.534913426057339e-05,
      "loss": 0.018043191730976106,
      "step": 51640
    },
    {
      "epoch": 14.660800454158387,
      "grad_norm": 3.543492317199707,
      "learning_rate": 8.534629577065002e-05,
      "loss": 0.02168959379196167,
      "step": 51650
    },
    {
      "epoch": 14.663638944081749,
      "grad_norm": 3.625410318374634,
      "learning_rate": 8.534345728072666e-05,
      "loss": 0.026066002249717713,
      "step": 51660
    },
    {
      "epoch": 14.66647743400511,
      "grad_norm": 14.819214820861816,
      "learning_rate": 8.53406187908033e-05,
      "loss": 0.04708296358585358,
      "step": 51670
    },
    {
      "epoch": 14.66931592392847,
      "grad_norm": 1.5611768960952759,
      "learning_rate": 8.533778030087993e-05,
      "loss": 0.021326422691345215,
      "step": 51680
    },
    {
      "epoch": 14.67215441385183,
      "grad_norm": 6.022402286529541,
      "learning_rate": 8.533494181095657e-05,
      "loss": 0.008427245914936066,
      "step": 51690
    },
    {
      "epoch": 14.674992903775191,
      "grad_norm": 9.686131477355957,
      "learning_rate": 8.533210332103322e-05,
      "loss": 0.03838449716567993,
      "step": 51700
    },
    {
      "epoch": 14.677831393698552,
      "grad_norm": 2.243096113204956,
      "learning_rate": 8.532926483110985e-05,
      "loss": 0.03304242491722107,
      "step": 51710
    },
    {
      "epoch": 14.680669883621913,
      "grad_norm": 4.648524284362793,
      "learning_rate": 8.532642634118649e-05,
      "loss": 0.025969645380973815,
      "step": 51720
    },
    {
      "epoch": 14.683508373545274,
      "grad_norm": 4.643424034118652,
      "learning_rate": 8.532358785126313e-05,
      "loss": 0.026057317852973938,
      "step": 51730
    },
    {
      "epoch": 14.686346863468636,
      "grad_norm": 3.315474510192871,
      "learning_rate": 8.532074936133977e-05,
      "loss": 0.03135319650173187,
      "step": 51740
    },
    {
      "epoch": 14.689185353391995,
      "grad_norm": 11.527090072631836,
      "learning_rate": 8.53179108714164e-05,
      "loss": 0.042044630646705626,
      "step": 51750
    },
    {
      "epoch": 14.692023843315356,
      "grad_norm": 7.199404239654541,
      "learning_rate": 8.531507238149306e-05,
      "loss": 0.029575473070144652,
      "step": 51760
    },
    {
      "epoch": 14.694862333238717,
      "grad_norm": 1.290221095085144,
      "learning_rate": 8.53122338915697e-05,
      "loss": 0.01553715169429779,
      "step": 51770
    },
    {
      "epoch": 14.697700823162078,
      "grad_norm": 2.0115106105804443,
      "learning_rate": 8.530939540164633e-05,
      "loss": 0.023492184281349183,
      "step": 51780
    },
    {
      "epoch": 14.70053931308544,
      "grad_norm": 6.614366054534912,
      "learning_rate": 8.530655691172297e-05,
      "loss": 0.03776212334632874,
      "step": 51790
    },
    {
      "epoch": 14.703377803008799,
      "grad_norm": 0.5551867485046387,
      "learning_rate": 8.530371842179961e-05,
      "loss": 0.02782229483127594,
      "step": 51800
    },
    {
      "epoch": 14.70621629293216,
      "grad_norm": 1.761235237121582,
      "learning_rate": 8.530087993187624e-05,
      "loss": 0.02025478333234787,
      "step": 51810
    },
    {
      "epoch": 14.709054782855521,
      "grad_norm": 4.280798435211182,
      "learning_rate": 8.529804144195288e-05,
      "loss": 0.033472859859466554,
      "step": 51820
    },
    {
      "epoch": 14.711893272778882,
      "grad_norm": 7.388118267059326,
      "learning_rate": 8.529520295202953e-05,
      "loss": 0.028340888023376466,
      "step": 51830
    },
    {
      "epoch": 14.714731762702243,
      "grad_norm": 8.572880744934082,
      "learning_rate": 8.529236446210616e-05,
      "loss": 0.03093830347061157,
      "step": 51840
    },
    {
      "epoch": 14.717570252625602,
      "grad_norm": 12.007716178894043,
      "learning_rate": 8.52895259721828e-05,
      "loss": 0.04730663299560547,
      "step": 51850
    },
    {
      "epoch": 14.720408742548964,
      "grad_norm": 1.922437310218811,
      "learning_rate": 8.528668748225944e-05,
      "loss": 0.03615111708641052,
      "step": 51860
    },
    {
      "epoch": 14.723247232472325,
      "grad_norm": 10.655746459960938,
      "learning_rate": 8.528384899233609e-05,
      "loss": 0.05345605611801148,
      "step": 51870
    },
    {
      "epoch": 14.726085722395686,
      "grad_norm": 1.8940304517745972,
      "learning_rate": 8.528101050241271e-05,
      "loss": 0.027235192060470582,
      "step": 51880
    },
    {
      "epoch": 14.728924212319047,
      "grad_norm": 12.067229270935059,
      "learning_rate": 8.527817201248935e-05,
      "loss": 0.03236901462078094,
      "step": 51890
    },
    {
      "epoch": 14.731762702242406,
      "grad_norm": 8.673604011535645,
      "learning_rate": 8.527533352256601e-05,
      "loss": 0.02623067796230316,
      "step": 51900
    },
    {
      "epoch": 14.734601192165767,
      "grad_norm": 10.754587173461914,
      "learning_rate": 8.527249503264264e-05,
      "loss": 0.05044602751731873,
      "step": 51910
    },
    {
      "epoch": 14.737439682089128,
      "grad_norm": 8.14383602142334,
      "learning_rate": 8.526965654271928e-05,
      "loss": 0.03298507928848267,
      "step": 51920
    },
    {
      "epoch": 14.74027817201249,
      "grad_norm": 9.571702003479004,
      "learning_rate": 8.526681805279592e-05,
      "loss": 0.020658814907073976,
      "step": 51930
    },
    {
      "epoch": 14.74311666193585,
      "grad_norm": 1.7532601356506348,
      "learning_rate": 8.526397956287255e-05,
      "loss": 0.02656387984752655,
      "step": 51940
    },
    {
      "epoch": 14.74595515185921,
      "grad_norm": 4.6879987716674805,
      "learning_rate": 8.526114107294919e-05,
      "loss": 0.04317150413990021,
      "step": 51950
    },
    {
      "epoch": 14.748793641782571,
      "grad_norm": 16.127050399780273,
      "learning_rate": 8.525830258302584e-05,
      "loss": 0.05169779658317566,
      "step": 51960
    },
    {
      "epoch": 14.751632131705932,
      "grad_norm": 9.529219627380371,
      "learning_rate": 8.525546409310247e-05,
      "loss": 0.022693523764610292,
      "step": 51970
    },
    {
      "epoch": 14.754470621629293,
      "grad_norm": 1.4467220306396484,
      "learning_rate": 8.525262560317911e-05,
      "loss": 0.05085247755050659,
      "step": 51980
    },
    {
      "epoch": 14.757309111552654,
      "grad_norm": 3.411275863647461,
      "learning_rate": 8.524978711325575e-05,
      "loss": 0.017076879739761353,
      "step": 51990
    },
    {
      "epoch": 14.760147601476016,
      "grad_norm": 5.0562744140625,
      "learning_rate": 8.52469486233324e-05,
      "loss": 0.01956384479999542,
      "step": 52000
    },
    {
      "epoch": 14.760147601476016,
      "eval_accuracy": 0.9601958415463852,
      "eval_loss": 0.12304113060235977,
      "eval_runtime": 31.6736,
      "eval_samples_per_second": 496.533,
      "eval_steps_per_second": 7.767,
      "step": 52000
    },
    {
      "epoch": 14.762986091399375,
      "grad_norm": 2.034628391265869,
      "learning_rate": 8.524411013340902e-05,
      "loss": 0.038484308123588565,
      "step": 52010
    },
    {
      "epoch": 14.765824581322736,
      "grad_norm": 15.556130409240723,
      "learning_rate": 8.524127164348567e-05,
      "loss": 0.03666593432426453,
      "step": 52020
    },
    {
      "epoch": 14.768663071246097,
      "grad_norm": 1.5747390985488892,
      "learning_rate": 8.523843315356232e-05,
      "loss": 0.014248223602771759,
      "step": 52030
    },
    {
      "epoch": 14.771501561169458,
      "grad_norm": 11.546631813049316,
      "learning_rate": 8.523559466363895e-05,
      "loss": 0.03246625065803528,
      "step": 52040
    },
    {
      "epoch": 14.77434005109282,
      "grad_norm": 1.5567431449890137,
      "learning_rate": 8.523275617371559e-05,
      "loss": 0.03479932546615601,
      "step": 52050
    },
    {
      "epoch": 14.777178541016179,
      "grad_norm": 8.006046295166016,
      "learning_rate": 8.522991768379223e-05,
      "loss": 0.036957567930221556,
      "step": 52060
    },
    {
      "epoch": 14.78001703093954,
      "grad_norm": 10.67141342163086,
      "learning_rate": 8.522707919386886e-05,
      "loss": 0.03419145047664642,
      "step": 52070
    },
    {
      "epoch": 14.782855520862901,
      "grad_norm": 2.0818207263946533,
      "learning_rate": 8.52242407039455e-05,
      "loss": 0.029305773973464965,
      "step": 52080
    },
    {
      "epoch": 14.785694010786262,
      "grad_norm": 6.951576232910156,
      "learning_rate": 8.522140221402214e-05,
      "loss": 0.05066871643066406,
      "step": 52090
    },
    {
      "epoch": 14.788532500709623,
      "grad_norm": 12.597771644592285,
      "learning_rate": 8.521856372409878e-05,
      "loss": 0.02722790539264679,
      "step": 52100
    },
    {
      "epoch": 14.791370990632982,
      "grad_norm": 3.6698906421661377,
      "learning_rate": 8.521572523417542e-05,
      "loss": 0.03936859667301178,
      "step": 52110
    },
    {
      "epoch": 14.794209480556344,
      "grad_norm": 0.5758851766586304,
      "learning_rate": 8.521288674425207e-05,
      "loss": 0.031905907392501834,
      "step": 52120
    },
    {
      "epoch": 14.797047970479705,
      "grad_norm": 4.625484466552734,
      "learning_rate": 8.521004825432871e-05,
      "loss": 0.032159218192100526,
      "step": 52130
    },
    {
      "epoch": 14.799886460403066,
      "grad_norm": 0.46332302689552307,
      "learning_rate": 8.520720976440533e-05,
      "loss": 0.023548538982868194,
      "step": 52140
    },
    {
      "epoch": 14.802724950326427,
      "grad_norm": 0.3681604266166687,
      "learning_rate": 8.520437127448198e-05,
      "loss": 0.018060213327407836,
      "step": 52150
    },
    {
      "epoch": 14.805563440249788,
      "grad_norm": 5.4887847900390625,
      "learning_rate": 8.520153278455862e-05,
      "loss": 0.02133317440748215,
      "step": 52160
    },
    {
      "epoch": 14.808401930173147,
      "grad_norm": 9.38009262084961,
      "learning_rate": 8.519869429463526e-05,
      "loss": 0.026021358370780946,
      "step": 52170
    },
    {
      "epoch": 14.811240420096508,
      "grad_norm": 11.541468620300293,
      "learning_rate": 8.51958558047119e-05,
      "loss": 0.03665421307086945,
      "step": 52180
    },
    {
      "epoch": 14.81407891001987,
      "grad_norm": 4.257431507110596,
      "learning_rate": 8.519301731478854e-05,
      "loss": 0.032706519961357115,
      "step": 52190
    },
    {
      "epoch": 14.81691739994323,
      "grad_norm": 14.1207914352417,
      "learning_rate": 8.519017882486517e-05,
      "loss": 0.029098379611968993,
      "step": 52200
    },
    {
      "epoch": 14.819755889866592,
      "grad_norm": 6.371778964996338,
      "learning_rate": 8.518734033494181e-05,
      "loss": 0.03482284843921661,
      "step": 52210
    },
    {
      "epoch": 14.822594379789951,
      "grad_norm": 6.763216018676758,
      "learning_rate": 8.518450184501845e-05,
      "loss": 0.03519497513771057,
      "step": 52220
    },
    {
      "epoch": 14.825432869713312,
      "grad_norm": 2.672987699508667,
      "learning_rate": 8.51816633550951e-05,
      "loss": 0.034593862295150754,
      "step": 52230
    },
    {
      "epoch": 14.828271359636673,
      "grad_norm": 11.970619201660156,
      "learning_rate": 8.517882486517174e-05,
      "loss": 0.0287896990776062,
      "step": 52240
    },
    {
      "epoch": 14.831109849560034,
      "grad_norm": 5.577654838562012,
      "learning_rate": 8.517598637524838e-05,
      "loss": 0.030020585656166075,
      "step": 52250
    },
    {
      "epoch": 14.833948339483396,
      "grad_norm": 1.1137555837631226,
      "learning_rate": 8.5173147885325e-05,
      "loss": 0.023542656004428862,
      "step": 52260
    },
    {
      "epoch": 14.836786829406755,
      "grad_norm": 6.946108818054199,
      "learning_rate": 8.517030939540165e-05,
      "loss": 0.025078311562538147,
      "step": 52270
    },
    {
      "epoch": 14.839625319330116,
      "grad_norm": 1.5836098194122314,
      "learning_rate": 8.516747090547829e-05,
      "loss": 0.01837528347969055,
      "step": 52280
    },
    {
      "epoch": 14.842463809253477,
      "grad_norm": 6.4721221923828125,
      "learning_rate": 8.516463241555493e-05,
      "loss": 0.027233871817588805,
      "step": 52290
    },
    {
      "epoch": 14.845302299176838,
      "grad_norm": 2.855774402618408,
      "learning_rate": 8.516179392563157e-05,
      "loss": 0.019256198406219484,
      "step": 52300
    },
    {
      "epoch": 14.8481407891002,
      "grad_norm": 4.033708095550537,
      "learning_rate": 8.515895543570821e-05,
      "loss": 0.024883374571800232,
      "step": 52310
    },
    {
      "epoch": 14.850979279023559,
      "grad_norm": 14.830466270446777,
      "learning_rate": 8.515611694578485e-05,
      "loss": 0.03544933795928955,
      "step": 52320
    },
    {
      "epoch": 14.85381776894692,
      "grad_norm": 3.2022197246551514,
      "learning_rate": 8.515327845586148e-05,
      "loss": 0.03954409956932068,
      "step": 52330
    },
    {
      "epoch": 14.85665625887028,
      "grad_norm": 5.517317771911621,
      "learning_rate": 8.515043996593812e-05,
      "loss": 0.03756464123725891,
      "step": 52340
    },
    {
      "epoch": 14.859494748793642,
      "grad_norm": 5.024097919464111,
      "learning_rate": 8.514760147601476e-05,
      "loss": 0.027983418107032774,
      "step": 52350
    },
    {
      "epoch": 14.862333238717003,
      "grad_norm": 4.13739538192749,
      "learning_rate": 8.51447629860914e-05,
      "loss": 0.027565941214561462,
      "step": 52360
    },
    {
      "epoch": 14.865171728640362,
      "grad_norm": 7.301817893981934,
      "learning_rate": 8.514192449616805e-05,
      "loss": 0.04110962748527527,
      "step": 52370
    },
    {
      "epoch": 14.868010218563724,
      "grad_norm": 4.641166687011719,
      "learning_rate": 8.513908600624469e-05,
      "loss": 0.015780039131641388,
      "step": 52380
    },
    {
      "epoch": 14.870848708487085,
      "grad_norm": 1.6691021919250488,
      "learning_rate": 8.513624751632131e-05,
      "loss": 0.02632930874824524,
      "step": 52390
    },
    {
      "epoch": 14.873687198410446,
      "grad_norm": 6.865642070770264,
      "learning_rate": 8.513340902639796e-05,
      "loss": 0.06053704023361206,
      "step": 52400
    },
    {
      "epoch": 14.876525688333807,
      "grad_norm": 11.731450080871582,
      "learning_rate": 8.51305705364746e-05,
      "loss": 0.05539149045944214,
      "step": 52410
    },
    {
      "epoch": 14.879364178257168,
      "grad_norm": 8.948708534240723,
      "learning_rate": 8.512773204655124e-05,
      "loss": 0.05718296766281128,
      "step": 52420
    },
    {
      "epoch": 14.882202668180527,
      "grad_norm": 1.5404996871948242,
      "learning_rate": 8.512489355662788e-05,
      "loss": 0.04857009053230286,
      "step": 52430
    },
    {
      "epoch": 14.885041158103888,
      "grad_norm": 6.185203552246094,
      "learning_rate": 8.512205506670452e-05,
      "loss": 0.022160325944423676,
      "step": 52440
    },
    {
      "epoch": 14.88787964802725,
      "grad_norm": 0.450327605009079,
      "learning_rate": 8.511921657678116e-05,
      "loss": 0.023737695813179017,
      "step": 52450
    },
    {
      "epoch": 14.89071813795061,
      "grad_norm": 8.882545471191406,
      "learning_rate": 8.511637808685779e-05,
      "loss": 0.05116015076637268,
      "step": 52460
    },
    {
      "epoch": 14.893556627873972,
      "grad_norm": 3.0069682598114014,
      "learning_rate": 8.511353959693443e-05,
      "loss": 0.028014230728149413,
      "step": 52470
    },
    {
      "epoch": 14.896395117797331,
      "grad_norm": 1.16648530960083,
      "learning_rate": 8.511070110701107e-05,
      "loss": 0.02635745108127594,
      "step": 52480
    },
    {
      "epoch": 14.899233607720692,
      "grad_norm": 11.484435081481934,
      "learning_rate": 8.51078626170877e-05,
      "loss": 0.03438654243946075,
      "step": 52490
    },
    {
      "epoch": 14.902072097644053,
      "grad_norm": 7.604098320007324,
      "learning_rate": 8.510502412716436e-05,
      "loss": 0.026794984936714172,
      "step": 52500
    },
    {
      "epoch": 14.902072097644053,
      "eval_accuracy": 0.9615947097348508,
      "eval_loss": 0.11910422146320343,
      "eval_runtime": 31.0208,
      "eval_samples_per_second": 506.982,
      "eval_steps_per_second": 7.93,
      "step": 52500
    },
    {
      "epoch": 14.904910587567414,
      "grad_norm": 0.1136249452829361,
      "learning_rate": 8.5102185637241e-05,
      "loss": 0.02782406210899353,
      "step": 52510
    },
    {
      "epoch": 14.907749077490775,
      "grad_norm": 0.9823296070098877,
      "learning_rate": 8.509934714731763e-05,
      "loss": 0.03094485104084015,
      "step": 52520
    },
    {
      "epoch": 14.910587567414137,
      "grad_norm": 6.141584873199463,
      "learning_rate": 8.509650865739427e-05,
      "loss": 0.03172458112239838,
      "step": 52530
    },
    {
      "epoch": 14.913426057337496,
      "grad_norm": 11.04259204864502,
      "learning_rate": 8.509367016747091e-05,
      "loss": 0.04151504337787628,
      "step": 52540
    },
    {
      "epoch": 14.916264547260857,
      "grad_norm": 9.582223892211914,
      "learning_rate": 8.509083167754755e-05,
      "loss": 0.029722175002098082,
      "step": 52550
    },
    {
      "epoch": 14.919103037184218,
      "grad_norm": 7.273066997528076,
      "learning_rate": 8.508799318762419e-05,
      "loss": 0.023678822815418242,
      "step": 52560
    },
    {
      "epoch": 14.92194152710758,
      "grad_norm": 1.9548935890197754,
      "learning_rate": 8.508515469770083e-05,
      "loss": 0.033037933707237246,
      "step": 52570
    },
    {
      "epoch": 14.92478001703094,
      "grad_norm": 1.294035792350769,
      "learning_rate": 8.508231620777747e-05,
      "loss": 0.040040749311447146,
      "step": 52580
    },
    {
      "epoch": 14.9276185069543,
      "grad_norm": 13.968194961547852,
      "learning_rate": 8.50794777178541e-05,
      "loss": 0.050274431705474854,
      "step": 52590
    },
    {
      "epoch": 14.93045699687766,
      "grad_norm": 14.652900695800781,
      "learning_rate": 8.507663922793074e-05,
      "loss": 0.02755894958972931,
      "step": 52600
    },
    {
      "epoch": 14.933295486801022,
      "grad_norm": 11.293218612670898,
      "learning_rate": 8.507380073800738e-05,
      "loss": 0.06112414598464966,
      "step": 52610
    },
    {
      "epoch": 14.936133976724383,
      "grad_norm": 4.71428108215332,
      "learning_rate": 8.507096224808401e-05,
      "loss": 0.05159469246864319,
      "step": 52620
    },
    {
      "epoch": 14.938972466647744,
      "grad_norm": 9.22663688659668,
      "learning_rate": 8.506812375816067e-05,
      "loss": 0.04534577429294586,
      "step": 52630
    },
    {
      "epoch": 14.941810956571103,
      "grad_norm": 3.193680763244629,
      "learning_rate": 8.506528526823731e-05,
      "loss": 0.05434118509292603,
      "step": 52640
    },
    {
      "epoch": 14.944649446494465,
      "grad_norm": 3.10563063621521,
      "learning_rate": 8.506244677831394e-05,
      "loss": 0.04572090208530426,
      "step": 52650
    },
    {
      "epoch": 14.947487936417826,
      "grad_norm": 10.260764122009277,
      "learning_rate": 8.505960828839058e-05,
      "loss": 0.04893912971019745,
      "step": 52660
    },
    {
      "epoch": 14.950326426341187,
      "grad_norm": 4.654698848724365,
      "learning_rate": 8.505676979846722e-05,
      "loss": 0.05549478530883789,
      "step": 52670
    },
    {
      "epoch": 14.953164916264548,
      "grad_norm": 9.661906242370605,
      "learning_rate": 8.505393130854386e-05,
      "loss": 0.03315366804599762,
      "step": 52680
    },
    {
      "epoch": 14.956003406187907,
      "grad_norm": 3.403984546661377,
      "learning_rate": 8.505109281862049e-05,
      "loss": 0.031098783016204834,
      "step": 52690
    },
    {
      "epoch": 14.958841896111268,
      "grad_norm": 3.113482713699341,
      "learning_rate": 8.504825432869714e-05,
      "loss": 0.04761813879013062,
      "step": 52700
    },
    {
      "epoch": 14.96168038603463,
      "grad_norm": 5.0839009284973145,
      "learning_rate": 8.504541583877378e-05,
      "loss": 0.02483803331851959,
      "step": 52710
    },
    {
      "epoch": 14.96451887595799,
      "grad_norm": 10.606212615966797,
      "learning_rate": 8.504257734885041e-05,
      "loss": 0.03670084178447723,
      "step": 52720
    },
    {
      "epoch": 14.967357365881352,
      "grad_norm": 7.447025775909424,
      "learning_rate": 8.503973885892705e-05,
      "loss": 0.01570524275302887,
      "step": 52730
    },
    {
      "epoch": 14.970195855804711,
      "grad_norm": 2.658383369445801,
      "learning_rate": 8.50369003690037e-05,
      "loss": 0.06445603370666504,
      "step": 52740
    },
    {
      "epoch": 14.973034345728072,
      "grad_norm": 4.66203498840332,
      "learning_rate": 8.503406187908032e-05,
      "loss": 0.014038428664207458,
      "step": 52750
    },
    {
      "epoch": 14.975872835651433,
      "grad_norm": 1.796512246131897,
      "learning_rate": 8.503122338915698e-05,
      "loss": 0.043055981397628784,
      "step": 52760
    },
    {
      "epoch": 14.978711325574794,
      "grad_norm": 3.048551559448242,
      "learning_rate": 8.502838489923362e-05,
      "loss": 0.035670405626296996,
      "step": 52770
    },
    {
      "epoch": 14.981549815498155,
      "grad_norm": 2.936340093612671,
      "learning_rate": 8.502554640931025e-05,
      "loss": 0.02280242145061493,
      "step": 52780
    },
    {
      "epoch": 14.984388305421517,
      "grad_norm": 1.5961838960647583,
      "learning_rate": 8.502270791938689e-05,
      "loss": 0.05337930917739868,
      "step": 52790
    },
    {
      "epoch": 14.987226795344876,
      "grad_norm": 4.143683910369873,
      "learning_rate": 8.501986942946353e-05,
      "loss": 0.032545506954193115,
      "step": 52800
    },
    {
      "epoch": 14.990065285268237,
      "grad_norm": 3.4018266201019287,
      "learning_rate": 8.501703093954017e-05,
      "loss": 0.02733646333217621,
      "step": 52810
    },
    {
      "epoch": 14.992903775191598,
      "grad_norm": 1.1462609767913818,
      "learning_rate": 8.50141924496168e-05,
      "loss": 0.012983845174312591,
      "step": 52820
    },
    {
      "epoch": 14.99574226511496,
      "grad_norm": 9.14090633392334,
      "learning_rate": 8.501135395969345e-05,
      "loss": 0.0259353905916214,
      "step": 52830
    },
    {
      "epoch": 14.99858075503832,
      "grad_norm": 5.760972023010254,
      "learning_rate": 8.50085154697701e-05,
      "loss": 0.026335883140563964,
      "step": 52840
    },
    {
      "epoch": 15.00141924496168,
      "grad_norm": 11.581393241882324,
      "learning_rate": 8.500567697984672e-05,
      "loss": 0.028559917211532594,
      "step": 52850
    },
    {
      "epoch": 15.00425773488504,
      "grad_norm": 6.410826206207275,
      "learning_rate": 8.500283848992336e-05,
      "loss": 0.0230854332447052,
      "step": 52860
    },
    {
      "epoch": 15.007096224808402,
      "grad_norm": 6.874464988708496,
      "learning_rate": 8.5e-05,
      "loss": 0.016904139518737794,
      "step": 52870
    },
    {
      "epoch": 15.009934714731763,
      "grad_norm": 10.186416625976562,
      "learning_rate": 8.499716151007663e-05,
      "loss": 0.02468867152929306,
      "step": 52880
    },
    {
      "epoch": 15.012773204655124,
      "grad_norm": 8.226420402526855,
      "learning_rate": 8.499432302015329e-05,
      "loss": 0.026024246215820314,
      "step": 52890
    },
    {
      "epoch": 15.015611694578483,
      "grad_norm": 3.9801876544952393,
      "learning_rate": 8.499148453022993e-05,
      "loss": 0.023551414906978606,
      "step": 52900
    },
    {
      "epoch": 15.018450184501845,
      "grad_norm": 4.600092887878418,
      "learning_rate": 8.498864604030656e-05,
      "loss": 0.025222939252853394,
      "step": 52910
    },
    {
      "epoch": 15.021288674425206,
      "grad_norm": 0.8289909958839417,
      "learning_rate": 8.49858075503832e-05,
      "loss": 0.025393322110176086,
      "step": 52920
    },
    {
      "epoch": 15.024127164348567,
      "grad_norm": 5.408454418182373,
      "learning_rate": 8.498296906045984e-05,
      "loss": 0.03954998254776001,
      "step": 52930
    },
    {
      "epoch": 15.026965654271928,
      "grad_norm": 0.8199604749679565,
      "learning_rate": 8.498013057053648e-05,
      "loss": 0.023568087816238405,
      "step": 52940
    },
    {
      "epoch": 15.029804144195289,
      "grad_norm": 15.933345794677734,
      "learning_rate": 8.497729208061311e-05,
      "loss": 0.0334645688533783,
      "step": 52950
    },
    {
      "epoch": 15.032642634118648,
      "grad_norm": 3.764054536819458,
      "learning_rate": 8.497445359068976e-05,
      "loss": 0.014909225702285766,
      "step": 52960
    },
    {
      "epoch": 15.03548112404201,
      "grad_norm": 11.115058898925781,
      "learning_rate": 8.49716151007664e-05,
      "loss": 0.02324901521205902,
      "step": 52970
    },
    {
      "epoch": 15.03831961396537,
      "grad_norm": 0.42527395486831665,
      "learning_rate": 8.496877661084303e-05,
      "loss": 0.01743047833442688,
      "step": 52980
    },
    {
      "epoch": 15.041158103888732,
      "grad_norm": 0.5302337408065796,
      "learning_rate": 8.496593812091968e-05,
      "loss": 0.008684401214122773,
      "step": 52990
    },
    {
      "epoch": 15.043996593812093,
      "grad_norm": 3.21989107131958,
      "learning_rate": 8.496309963099632e-05,
      "loss": 0.014872929453849793,
      "step": 53000
    },
    {
      "epoch": 15.043996593812093,
      "eval_accuracy": 0.9634386723469193,
      "eval_loss": 0.11136753112077713,
      "eval_runtime": 31.3226,
      "eval_samples_per_second": 502.098,
      "eval_steps_per_second": 7.854,
      "step": 53000
    },
    {
      "epoch": 15.046835083735452,
      "grad_norm": 2.858579397201538,
      "learning_rate": 8.496026114107294e-05,
      "loss": 0.02118138372898102,
      "step": 53010
    },
    {
      "epoch": 15.049673573658813,
      "grad_norm": 0.9128031134605408,
      "learning_rate": 8.495742265114959e-05,
      "loss": 0.013003897666931153,
      "step": 53020
    },
    {
      "epoch": 15.052512063582174,
      "grad_norm": 9.113799095153809,
      "learning_rate": 8.495458416122624e-05,
      "loss": 0.041965633630752563,
      "step": 53030
    },
    {
      "epoch": 15.055350553505535,
      "grad_norm": 2.1945927143096924,
      "learning_rate": 8.495174567130287e-05,
      "loss": 0.0187469482421875,
      "step": 53040
    },
    {
      "epoch": 15.058189043428897,
      "grad_norm": 2.8984506130218506,
      "learning_rate": 8.494890718137951e-05,
      "loss": 0.03352283835411072,
      "step": 53050
    },
    {
      "epoch": 15.061027533352256,
      "grad_norm": 2.4544174671173096,
      "learning_rate": 8.494606869145615e-05,
      "loss": 0.012337620556354522,
      "step": 53060
    },
    {
      "epoch": 15.063866023275617,
      "grad_norm": 6.353875160217285,
      "learning_rate": 8.494323020153279e-05,
      "loss": 0.03873030245304108,
      "step": 53070
    },
    {
      "epoch": 15.066704513198978,
      "grad_norm": 1.4572876691818237,
      "learning_rate": 8.494039171160942e-05,
      "loss": 0.026953238248825073,
      "step": 53080
    },
    {
      "epoch": 15.06954300312234,
      "grad_norm": 13.734784126281738,
      "learning_rate": 8.493755322168608e-05,
      "loss": 0.024659039080142976,
      "step": 53090
    },
    {
      "epoch": 15.0723814930457,
      "grad_norm": 3.3762965202331543,
      "learning_rate": 8.49347147317627e-05,
      "loss": 0.01574217975139618,
      "step": 53100
    },
    {
      "epoch": 15.07521998296906,
      "grad_norm": 4.7035956382751465,
      "learning_rate": 8.493187624183934e-05,
      "loss": 0.024960589408874512,
      "step": 53110
    },
    {
      "epoch": 15.07805847289242,
      "grad_norm": 11.74179744720459,
      "learning_rate": 8.492903775191599e-05,
      "loss": 0.02742968499660492,
      "step": 53120
    },
    {
      "epoch": 15.080896962815782,
      "grad_norm": 3.068833351135254,
      "learning_rate": 8.492619926199263e-05,
      "loss": 0.02853260636329651,
      "step": 53130
    },
    {
      "epoch": 15.083735452739143,
      "grad_norm": 4.965625762939453,
      "learning_rate": 8.492336077206926e-05,
      "loss": 0.018844975531101225,
      "step": 53140
    },
    {
      "epoch": 15.086573942662504,
      "grad_norm": 17.976947784423828,
      "learning_rate": 8.49205222821459e-05,
      "loss": 0.027571189403533935,
      "step": 53150
    },
    {
      "epoch": 15.089412432585865,
      "grad_norm": 4.006868839263916,
      "learning_rate": 8.491768379222255e-05,
      "loss": 0.018645693361759186,
      "step": 53160
    },
    {
      "epoch": 15.092250922509225,
      "grad_norm": 2.344867467880249,
      "learning_rate": 8.491484530229918e-05,
      "loss": 0.01034977361559868,
      "step": 53170
    },
    {
      "epoch": 15.095089412432586,
      "grad_norm": 4.115565776824951,
      "learning_rate": 8.491200681237582e-05,
      "loss": 0.0225346639752388,
      "step": 53180
    },
    {
      "epoch": 15.097927902355947,
      "grad_norm": 3.8482604026794434,
      "learning_rate": 8.490916832245246e-05,
      "loss": 0.017920659482479097,
      "step": 53190
    },
    {
      "epoch": 15.100766392279308,
      "grad_norm": 2.0540499687194824,
      "learning_rate": 8.490632983252909e-05,
      "loss": 0.016692161560058594,
      "step": 53200
    },
    {
      "epoch": 15.103604882202669,
      "grad_norm": 5.664206504821777,
      "learning_rate": 8.490349134260573e-05,
      "loss": 0.020225676894187927,
      "step": 53210
    },
    {
      "epoch": 15.106443372126028,
      "grad_norm": 1.425145149230957,
      "learning_rate": 8.490065285268237e-05,
      "loss": 0.01829567402601242,
      "step": 53220
    },
    {
      "epoch": 15.10928186204939,
      "grad_norm": 4.028144836425781,
      "learning_rate": 8.489781436275901e-05,
      "loss": 0.029241713881492614,
      "step": 53230
    },
    {
      "epoch": 15.11212035197275,
      "grad_norm": 8.185257911682129,
      "learning_rate": 8.489497587283566e-05,
      "loss": 0.01918734759092331,
      "step": 53240
    },
    {
      "epoch": 15.114958841896112,
      "grad_norm": 6.857207775115967,
      "learning_rate": 8.48921373829123e-05,
      "loss": 0.030198761820793153,
      "step": 53250
    },
    {
      "epoch": 15.117797331819473,
      "grad_norm": 2.811000347137451,
      "learning_rate": 8.488929889298894e-05,
      "loss": 0.024341876804828643,
      "step": 53260
    },
    {
      "epoch": 15.120635821742832,
      "grad_norm": 7.477023124694824,
      "learning_rate": 8.488646040306557e-05,
      "loss": 0.029148364067077638,
      "step": 53270
    },
    {
      "epoch": 15.123474311666193,
      "grad_norm": 1.9254635572433472,
      "learning_rate": 8.488362191314221e-05,
      "loss": 0.032135629653930665,
      "step": 53280
    },
    {
      "epoch": 15.126312801589554,
      "grad_norm": 12.80935287475586,
      "learning_rate": 8.488078342321886e-05,
      "loss": 0.01927918493747711,
      "step": 53290
    },
    {
      "epoch": 15.129151291512915,
      "grad_norm": 14.936166763305664,
      "learning_rate": 8.487794493329549e-05,
      "loss": 0.02948797345161438,
      "step": 53300
    },
    {
      "epoch": 15.131989781436276,
      "grad_norm": 2.925687789916992,
      "learning_rate": 8.487510644337213e-05,
      "loss": 0.028370019793510438,
      "step": 53310
    },
    {
      "epoch": 15.134828271359636,
      "grad_norm": 4.047255039215088,
      "learning_rate": 8.487226795344877e-05,
      "loss": 0.026260924339294434,
      "step": 53320
    },
    {
      "epoch": 15.137666761282997,
      "grad_norm": 6.451000213623047,
      "learning_rate": 8.48694294635254e-05,
      "loss": 0.017392192780971528,
      "step": 53330
    },
    {
      "epoch": 15.140505251206358,
      "grad_norm": 6.477872848510742,
      "learning_rate": 8.486659097360204e-05,
      "loss": 0.022298693656921387,
      "step": 53340
    },
    {
      "epoch": 15.14334374112972,
      "grad_norm": 2.901130437850952,
      "learning_rate": 8.486375248367868e-05,
      "loss": 0.037961924076080324,
      "step": 53350
    },
    {
      "epoch": 15.14618223105308,
      "grad_norm": 5.307724952697754,
      "learning_rate": 8.486091399375533e-05,
      "loss": 0.024045148491859437,
      "step": 53360
    },
    {
      "epoch": 15.149020720976441,
      "grad_norm": 1.505552053451538,
      "learning_rate": 8.485807550383197e-05,
      "loss": 0.027203640341758727,
      "step": 53370
    },
    {
      "epoch": 15.1518592108998,
      "grad_norm": 3.445411443710327,
      "learning_rate": 8.485523701390861e-05,
      "loss": 0.017229658365249634,
      "step": 53380
    },
    {
      "epoch": 15.154697700823162,
      "grad_norm": 3.822455883026123,
      "learning_rate": 8.485239852398525e-05,
      "loss": 0.031219345331192017,
      "step": 53390
    },
    {
      "epoch": 15.157536190746523,
      "grad_norm": 2.3856401443481445,
      "learning_rate": 8.484956003406188e-05,
      "loss": 0.0303028404712677,
      "step": 53400
    },
    {
      "epoch": 15.160374680669884,
      "grad_norm": 14.556412696838379,
      "learning_rate": 8.484672154413852e-05,
      "loss": 0.050366705656051634,
      "step": 53410
    },
    {
      "epoch": 15.163213170593245,
      "grad_norm": 7.2573652267456055,
      "learning_rate": 8.484388305421516e-05,
      "loss": 0.014065055549144745,
      "step": 53420
    },
    {
      "epoch": 15.166051660516604,
      "grad_norm": 2.4567673206329346,
      "learning_rate": 8.48410445642918e-05,
      "loss": 0.014977070689201354,
      "step": 53430
    },
    {
      "epoch": 15.168890150439966,
      "grad_norm": 4.0306878089904785,
      "learning_rate": 8.483820607436844e-05,
      "loss": 0.02399582862854004,
      "step": 53440
    },
    {
      "epoch": 15.171728640363327,
      "grad_norm": 2.532724380493164,
      "learning_rate": 8.483536758444508e-05,
      "loss": 0.011108486354351044,
      "step": 53450
    },
    {
      "epoch": 15.174567130286688,
      "grad_norm": 4.917425632476807,
      "learning_rate": 8.483252909452171e-05,
      "loss": 0.017559367418289184,
      "step": 53460
    },
    {
      "epoch": 15.177405620210049,
      "grad_norm": 2.377610445022583,
      "learning_rate": 8.48299744535907e-05,
      "loss": 0.02271555960178375,
      "step": 53470
    },
    {
      "epoch": 15.180244110133408,
      "grad_norm": 7.6386823654174805,
      "learning_rate": 8.482713596366733e-05,
      "loss": 0.019404828548431396,
      "step": 53480
    },
    {
      "epoch": 15.18308260005677,
      "grad_norm": 7.318428039550781,
      "learning_rate": 8.482429747374397e-05,
      "loss": 0.01831010729074478,
      "step": 53490
    },
    {
      "epoch": 15.18592108998013,
      "grad_norm": 13.758485794067383,
      "learning_rate": 8.482145898382062e-05,
      "loss": 0.045812129974365234,
      "step": 53500
    },
    {
      "epoch": 15.18592108998013,
      "eval_accuracy": 0.9558720671456731,
      "eval_loss": 0.13972848653793335,
      "eval_runtime": 31.1871,
      "eval_samples_per_second": 504.279,
      "eval_steps_per_second": 7.888,
      "step": 53500
    },
    {
      "epoch": 15.188759579903492,
      "grad_norm": 3.0384340286254883,
      "learning_rate": 8.481862049389724e-05,
      "loss": 0.03476028740406036,
      "step": 53510
    },
    {
      "epoch": 15.191598069826853,
      "grad_norm": 8.93148422241211,
      "learning_rate": 8.481578200397388e-05,
      "loss": 0.037573397159576416,
      "step": 53520
    },
    {
      "epoch": 15.194436559750212,
      "grad_norm": 6.068247318267822,
      "learning_rate": 8.481294351405053e-05,
      "loss": 0.0215740829706192,
      "step": 53530
    },
    {
      "epoch": 15.197275049673573,
      "grad_norm": 4.60642147064209,
      "learning_rate": 8.481010502412717e-05,
      "loss": 0.03872728049755096,
      "step": 53540
    },
    {
      "epoch": 15.200113539596934,
      "grad_norm": 3.9938864707946777,
      "learning_rate": 8.480726653420381e-05,
      "loss": 0.029471418261528014,
      "step": 53550
    },
    {
      "epoch": 15.202952029520295,
      "grad_norm": 6.822102069854736,
      "learning_rate": 8.480442804428045e-05,
      "loss": 0.023526379466056825,
      "step": 53560
    },
    {
      "epoch": 15.205790519443656,
      "grad_norm": 2.222734212875366,
      "learning_rate": 8.480158955435709e-05,
      "loss": 0.01734481304883957,
      "step": 53570
    },
    {
      "epoch": 15.208629009367018,
      "grad_norm": 1.5611189603805542,
      "learning_rate": 8.479875106443372e-05,
      "loss": 0.045385193824768064,
      "step": 53580
    },
    {
      "epoch": 15.211467499290377,
      "grad_norm": 4.646595001220703,
      "learning_rate": 8.479591257451036e-05,
      "loss": 0.03704947233200073,
      "step": 53590
    },
    {
      "epoch": 15.214305989213738,
      "grad_norm": 4.166716575622559,
      "learning_rate": 8.4793074084587e-05,
      "loss": 0.02110726684331894,
      "step": 53600
    },
    {
      "epoch": 15.217144479137099,
      "grad_norm": 4.020148754119873,
      "learning_rate": 8.479023559466364e-05,
      "loss": 0.024294754862785338,
      "step": 53610
    },
    {
      "epoch": 15.21998296906046,
      "grad_norm": 0.2567417621612549,
      "learning_rate": 8.478739710474029e-05,
      "loss": 0.018196302652359008,
      "step": 53620
    },
    {
      "epoch": 15.222821458983821,
      "grad_norm": 6.354002952575684,
      "learning_rate": 8.478455861481693e-05,
      "loss": 0.031344667077064514,
      "step": 53630
    },
    {
      "epoch": 15.22565994890718,
      "grad_norm": 1.950029969215393,
      "learning_rate": 8.478172012489355e-05,
      "loss": 0.01933310478925705,
      "step": 53640
    },
    {
      "epoch": 15.228498438830542,
      "grad_norm": 12.206835746765137,
      "learning_rate": 8.47788816349702e-05,
      "loss": 0.03785330653190613,
      "step": 53650
    },
    {
      "epoch": 15.231336928753903,
      "grad_norm": 15.296836853027344,
      "learning_rate": 8.477604314504684e-05,
      "loss": 0.03131443858146667,
      "step": 53660
    },
    {
      "epoch": 15.234175418677264,
      "grad_norm": 6.091537952423096,
      "learning_rate": 8.477320465512348e-05,
      "loss": 0.02741886079311371,
      "step": 53670
    },
    {
      "epoch": 15.237013908600625,
      "grad_norm": 3.2014384269714355,
      "learning_rate": 8.477036616520012e-05,
      "loss": 0.012884004414081574,
      "step": 53680
    },
    {
      "epoch": 15.239852398523984,
      "grad_norm": 8.4446382522583,
      "learning_rate": 8.476752767527676e-05,
      "loss": 0.029363781213760376,
      "step": 53690
    },
    {
      "epoch": 15.242690888447346,
      "grad_norm": 13.960970878601074,
      "learning_rate": 8.47646891853534e-05,
      "loss": 0.026152205467224122,
      "step": 53700
    },
    {
      "epoch": 15.245529378370707,
      "grad_norm": 1.270023226737976,
      "learning_rate": 8.476185069543003e-05,
      "loss": 0.018164780735969544,
      "step": 53710
    },
    {
      "epoch": 15.248367868294068,
      "grad_norm": 0.8989373445510864,
      "learning_rate": 8.475901220550667e-05,
      "loss": 0.02691385746002197,
      "step": 53720
    },
    {
      "epoch": 15.251206358217429,
      "grad_norm": 10.848620414733887,
      "learning_rate": 8.475617371558331e-05,
      "loss": 0.01985987424850464,
      "step": 53730
    },
    {
      "epoch": 15.25404484814079,
      "grad_norm": 7.492671012878418,
      "learning_rate": 8.475333522565995e-05,
      "loss": 0.03982653319835663,
      "step": 53740
    },
    {
      "epoch": 15.25688333806415,
      "grad_norm": 6.352535247802734,
      "learning_rate": 8.47504967357366e-05,
      "loss": 0.022130705416202545,
      "step": 53750
    },
    {
      "epoch": 15.25972182798751,
      "grad_norm": 4.262563705444336,
      "learning_rate": 8.474765824581324e-05,
      "loss": 0.02525729238986969,
      "step": 53760
    },
    {
      "epoch": 15.262560317910872,
      "grad_norm": 0.6703700423240662,
      "learning_rate": 8.474481975588987e-05,
      "loss": 0.016026677191257478,
      "step": 53770
    },
    {
      "epoch": 15.265398807834233,
      "grad_norm": 13.317543029785156,
      "learning_rate": 8.47419812659665e-05,
      "loss": 0.021447065472602844,
      "step": 53780
    },
    {
      "epoch": 15.268237297757594,
      "grad_norm": 5.439362049102783,
      "learning_rate": 8.473914277604315e-05,
      "loss": 0.022591273486614227,
      "step": 53790
    },
    {
      "epoch": 15.271075787680953,
      "grad_norm": 0.9924070835113525,
      "learning_rate": 8.473630428611979e-05,
      "loss": 0.013536173105239867,
      "step": 53800
    },
    {
      "epoch": 15.273914277604314,
      "grad_norm": 1.9649178981781006,
      "learning_rate": 8.473346579619643e-05,
      "loss": 0.019028496742248536,
      "step": 53810
    },
    {
      "epoch": 15.276752767527675,
      "grad_norm": 5.477230072021484,
      "learning_rate": 8.473062730627307e-05,
      "loss": 0.04178065359592438,
      "step": 53820
    },
    {
      "epoch": 15.279591257451036,
      "grad_norm": 7.263782501220703,
      "learning_rate": 8.472778881634971e-05,
      "loss": 0.02689291536808014,
      "step": 53830
    },
    {
      "epoch": 15.282429747374398,
      "grad_norm": 0.9777436852455139,
      "learning_rate": 8.472495032642634e-05,
      "loss": 0.019415967166423798,
      "step": 53840
    },
    {
      "epoch": 15.285268237297757,
      "grad_norm": 6.003324508666992,
      "learning_rate": 8.472211183650298e-05,
      "loss": 0.022689004242420197,
      "step": 53850
    },
    {
      "epoch": 15.288106727221118,
      "grad_norm": 9.073564529418945,
      "learning_rate": 8.471927334657962e-05,
      "loss": 0.02712928056716919,
      "step": 53860
    },
    {
      "epoch": 15.290945217144479,
      "grad_norm": 1.9147794246673584,
      "learning_rate": 8.471643485665627e-05,
      "loss": 0.03310129344463349,
      "step": 53870
    },
    {
      "epoch": 15.29378370706784,
      "grad_norm": 1.485855221748352,
      "learning_rate": 8.47135963667329e-05,
      "loss": 0.03053884506225586,
      "step": 53880
    },
    {
      "epoch": 15.296622196991201,
      "grad_norm": 4.232726097106934,
      "learning_rate": 8.471075787680955e-05,
      "loss": 0.023817449808120728,
      "step": 53890
    },
    {
      "epoch": 15.29946068691456,
      "grad_norm": 12.246906280517578,
      "learning_rate": 8.470791938688618e-05,
      "loss": 0.033807724714279175,
      "step": 53900
    },
    {
      "epoch": 15.302299176837922,
      "grad_norm": 5.464107990264893,
      "learning_rate": 8.470508089696282e-05,
      "loss": 0.024816635251045226,
      "step": 53910
    },
    {
      "epoch": 15.305137666761283,
      "grad_norm": 5.937697887420654,
      "learning_rate": 8.470224240703946e-05,
      "loss": 0.04006586968898773,
      "step": 53920
    },
    {
      "epoch": 15.307976156684644,
      "grad_norm": 1.50701904296875,
      "learning_rate": 8.46994039171161e-05,
      "loss": 0.017366278171539306,
      "step": 53930
    },
    {
      "epoch": 15.310814646608005,
      "grad_norm": 4.248182773590088,
      "learning_rate": 8.469656542719274e-05,
      "loss": 0.024303600192070007,
      "step": 53940
    },
    {
      "epoch": 15.313653136531366,
      "grad_norm": 2.161425828933716,
      "learning_rate": 8.469372693726938e-05,
      "loss": 0.01707477420568466,
      "step": 53950
    },
    {
      "epoch": 15.316491626454726,
      "grad_norm": 7.709280490875244,
      "learning_rate": 8.469088844734601e-05,
      "loss": 0.018223875761032106,
      "step": 53960
    },
    {
      "epoch": 15.319330116378087,
      "grad_norm": 8.556921005249023,
      "learning_rate": 8.468804995742265e-05,
      "loss": 0.03257838785648346,
      "step": 53970
    },
    {
      "epoch": 15.322168606301448,
      "grad_norm": 7.517209529876709,
      "learning_rate": 8.46852114674993e-05,
      "loss": 0.027686682343482972,
      "step": 53980
    },
    {
      "epoch": 15.325007096224809,
      "grad_norm": 1.236788034439087,
      "learning_rate": 8.468237297757593e-05,
      "loss": 0.01426074355840683,
      "step": 53990
    },
    {
      "epoch": 15.32784558614817,
      "grad_norm": 3.0803632736206055,
      "learning_rate": 8.467953448765256e-05,
      "loss": 0.02618541717529297,
      "step": 54000
    },
    {
      "epoch": 15.32784558614817,
      "eval_accuracy": 0.9631843326762892,
      "eval_loss": 0.11433736234903336,
      "eval_runtime": 31.7487,
      "eval_samples_per_second": 495.359,
      "eval_steps_per_second": 7.748,
      "step": 54000
    },
    {
      "epoch": 15.33068407607153,
      "grad_norm": 3.7513182163238525,
      "learning_rate": 8.467669599772922e-05,
      "loss": 0.017883220314979555,
      "step": 54010
    },
    {
      "epoch": 15.33352256599489,
      "grad_norm": 0.6390168070793152,
      "learning_rate": 8.467385750780586e-05,
      "loss": 0.04821707010269165,
      "step": 54020
    },
    {
      "epoch": 15.336361055918251,
      "grad_norm": 10.022395133972168,
      "learning_rate": 8.467101901788249e-05,
      "loss": 0.018045893311500548,
      "step": 54030
    },
    {
      "epoch": 15.339199545841613,
      "grad_norm": 5.511784076690674,
      "learning_rate": 8.466818052795913e-05,
      "loss": 0.025122317671775817,
      "step": 54040
    },
    {
      "epoch": 15.342038035764974,
      "grad_norm": 7.796726226806641,
      "learning_rate": 8.466534203803577e-05,
      "loss": 0.04648490250110626,
      "step": 54050
    },
    {
      "epoch": 15.344876525688333,
      "grad_norm": 3.3285627365112305,
      "learning_rate": 8.466250354811241e-05,
      "loss": 0.024082866311073304,
      "step": 54060
    },
    {
      "epoch": 15.347715015611694,
      "grad_norm": 2.4231514930725098,
      "learning_rate": 8.465966505818905e-05,
      "loss": 0.029388442635536194,
      "step": 54070
    },
    {
      "epoch": 15.350553505535055,
      "grad_norm": 11.6298828125,
      "learning_rate": 8.46568265682657e-05,
      "loss": 0.03692953884601593,
      "step": 54080
    },
    {
      "epoch": 15.353391995458416,
      "grad_norm": 3.9724791049957275,
      "learning_rate": 8.465398807834232e-05,
      "loss": 0.022999082505702973,
      "step": 54090
    },
    {
      "epoch": 15.356230485381777,
      "grad_norm": 3.0545289516448975,
      "learning_rate": 8.465114958841896e-05,
      "loss": 0.028710386157035826,
      "step": 54100
    },
    {
      "epoch": 15.359068975305137,
      "grad_norm": 3.168630361557007,
      "learning_rate": 8.46483110984956e-05,
      "loss": 0.02780725955963135,
      "step": 54110
    },
    {
      "epoch": 15.361907465228498,
      "grad_norm": 0.7495129704475403,
      "learning_rate": 8.464547260857225e-05,
      "loss": 0.008597607910633086,
      "step": 54120
    },
    {
      "epoch": 15.364745955151859,
      "grad_norm": 0.25669223070144653,
      "learning_rate": 8.464263411864887e-05,
      "loss": 0.019466498494148256,
      "step": 54130
    },
    {
      "epoch": 15.36758444507522,
      "grad_norm": 5.771591663360596,
      "learning_rate": 8.463979562872553e-05,
      "loss": 0.029212260246276857,
      "step": 54140
    },
    {
      "epoch": 15.370422934998581,
      "grad_norm": 2.0156846046447754,
      "learning_rate": 8.463695713880217e-05,
      "loss": 0.03037104308605194,
      "step": 54150
    },
    {
      "epoch": 15.373261424921942,
      "grad_norm": 2.4536187648773193,
      "learning_rate": 8.46341186488788e-05,
      "loss": 0.023899687826633452,
      "step": 54160
    },
    {
      "epoch": 15.376099914845302,
      "grad_norm": 1.5051302909851074,
      "learning_rate": 8.463128015895544e-05,
      "loss": 0.016792848706245422,
      "step": 54170
    },
    {
      "epoch": 15.378938404768663,
      "grad_norm": 12.736126899719238,
      "learning_rate": 8.462844166903208e-05,
      "loss": 0.035494789481163025,
      "step": 54180
    },
    {
      "epoch": 15.381776894692024,
      "grad_norm": 4.557518482208252,
      "learning_rate": 8.462560317910871e-05,
      "loss": 0.025106295943260193,
      "step": 54190
    },
    {
      "epoch": 15.384615384615385,
      "grad_norm": 11.785857200622559,
      "learning_rate": 8.462276468918535e-05,
      "loss": 0.044714349508285525,
      "step": 54200
    },
    {
      "epoch": 15.387453874538746,
      "grad_norm": 3.202505588531494,
      "learning_rate": 8.4619926199262e-05,
      "loss": 0.02049727886915207,
      "step": 54210
    },
    {
      "epoch": 15.390292364462105,
      "grad_norm": 2.1362318992614746,
      "learning_rate": 8.461708770933863e-05,
      "loss": 0.013331833481788635,
      "step": 54220
    },
    {
      "epoch": 15.393130854385467,
      "grad_norm": 3.6306371688842773,
      "learning_rate": 8.461424921941527e-05,
      "loss": 0.012118640542030334,
      "step": 54230
    },
    {
      "epoch": 15.395969344308828,
      "grad_norm": 3.8929033279418945,
      "learning_rate": 8.461141072949191e-05,
      "loss": 0.02974376380443573,
      "step": 54240
    },
    {
      "epoch": 15.398807834232189,
      "grad_norm": 9.148716926574707,
      "learning_rate": 8.460857223956856e-05,
      "loss": 0.018864558637142183,
      "step": 54250
    },
    {
      "epoch": 15.40164632415555,
      "grad_norm": 14.939241409301758,
      "learning_rate": 8.460573374964518e-05,
      "loss": 0.049407050013542175,
      "step": 54260
    },
    {
      "epoch": 15.40448481407891,
      "grad_norm": 5.407922744750977,
      "learning_rate": 8.460289525972184e-05,
      "loss": 0.014245152473449707,
      "step": 54270
    },
    {
      "epoch": 15.40732330400227,
      "grad_norm": 11.050570487976074,
      "learning_rate": 8.460005676979848e-05,
      "loss": 0.03409420251846314,
      "step": 54280
    },
    {
      "epoch": 15.410161793925631,
      "grad_norm": 8.218360900878906,
      "learning_rate": 8.459721827987511e-05,
      "loss": 0.020531482994556427,
      "step": 54290
    },
    {
      "epoch": 15.413000283848993,
      "grad_norm": 0.9123543500900269,
      "learning_rate": 8.459437978995175e-05,
      "loss": 0.03399288058280945,
      "step": 54300
    },
    {
      "epoch": 15.415838773772354,
      "grad_norm": 1.9880450963974,
      "learning_rate": 8.459154130002839e-05,
      "loss": 0.02451392710208893,
      "step": 54310
    },
    {
      "epoch": 15.418677263695713,
      "grad_norm": 2.1012115478515625,
      "learning_rate": 8.458870281010502e-05,
      "loss": 0.026399418711662292,
      "step": 54320
    },
    {
      "epoch": 15.421515753619074,
      "grad_norm": 3.144155740737915,
      "learning_rate": 8.458586432018166e-05,
      "loss": 0.027680817246437072,
      "step": 54330
    },
    {
      "epoch": 15.424354243542435,
      "grad_norm": 1.895432710647583,
      "learning_rate": 8.458302583025832e-05,
      "loss": 0.021748249232769013,
      "step": 54340
    },
    {
      "epoch": 15.427192733465796,
      "grad_norm": 7.695284366607666,
      "learning_rate": 8.458018734033494e-05,
      "loss": 0.02895972728729248,
      "step": 54350
    },
    {
      "epoch": 15.430031223389157,
      "grad_norm": 1.6202532052993774,
      "learning_rate": 8.457734885041158e-05,
      "loss": 0.01592489331960678,
      "step": 54360
    },
    {
      "epoch": 15.432869713312519,
      "grad_norm": 3.5763144493103027,
      "learning_rate": 8.457451036048823e-05,
      "loss": 0.04506300091743469,
      "step": 54370
    },
    {
      "epoch": 15.435708203235878,
      "grad_norm": 0.7253667712211609,
      "learning_rate": 8.457167187056487e-05,
      "loss": 0.02437674254179001,
      "step": 54380
    },
    {
      "epoch": 15.438546693159239,
      "grad_norm": 10.76182746887207,
      "learning_rate": 8.45688333806415e-05,
      "loss": 0.04124364852905273,
      "step": 54390
    },
    {
      "epoch": 15.4413851830826,
      "grad_norm": 0.7431447505950928,
      "learning_rate": 8.456599489071814e-05,
      "loss": 0.029517734050750734,
      "step": 54400
    },
    {
      "epoch": 15.444223673005961,
      "grad_norm": 15.151039123535156,
      "learning_rate": 8.456315640079479e-05,
      "loss": 0.04049486517906189,
      "step": 54410
    },
    {
      "epoch": 15.447062162929322,
      "grad_norm": 3.646165370941162,
      "learning_rate": 8.456031791087142e-05,
      "loss": 0.02199735939502716,
      "step": 54420
    },
    {
      "epoch": 15.449900652852682,
      "grad_norm": 7.934549331665039,
      "learning_rate": 8.455747942094806e-05,
      "loss": 0.03410484790802002,
      "step": 54430
    },
    {
      "epoch": 15.452739142776043,
      "grad_norm": 14.28009033203125,
      "learning_rate": 8.45546409310247e-05,
      "loss": 0.045516836643219,
      "step": 54440
    },
    {
      "epoch": 15.455577632699404,
      "grad_norm": 9.804656028747559,
      "learning_rate": 8.455180244110133e-05,
      "loss": 0.0263959139585495,
      "step": 54450
    },
    {
      "epoch": 15.458416122622765,
      "grad_norm": 3.664119005203247,
      "learning_rate": 8.454896395117797e-05,
      "loss": 0.02227897047996521,
      "step": 54460
    },
    {
      "epoch": 15.461254612546126,
      "grad_norm": 8.203856468200684,
      "learning_rate": 8.454612546125463e-05,
      "loss": 0.020636704564094544,
      "step": 54470
    },
    {
      "epoch": 15.464093102469485,
      "grad_norm": 3.1648714542388916,
      "learning_rate": 8.454328697133125e-05,
      "loss": 0.03570373356342316,
      "step": 54480
    },
    {
      "epoch": 15.466931592392847,
      "grad_norm": 2.6219470500946045,
      "learning_rate": 8.45404484814079e-05,
      "loss": 0.03272389769554138,
      "step": 54490
    },
    {
      "epoch": 15.469770082316208,
      "grad_norm": 10.881244659423828,
      "learning_rate": 8.453760999148454e-05,
      "loss": 0.05371968746185303,
      "step": 54500
    },
    {
      "epoch": 15.469770082316208,
      "eval_accuracy": 0.9592420677815222,
      "eval_loss": 0.13407358527183533,
      "eval_runtime": 31.4472,
      "eval_samples_per_second": 500.108,
      "eval_steps_per_second": 7.823,
      "step": 54500
    },
    {
      "epoch": 15.472608572239569,
      "grad_norm": 10.07542610168457,
      "learning_rate": 8.453477150156118e-05,
      "loss": 0.04315173923969269,
      "step": 54510
    },
    {
      "epoch": 15.47544706216293,
      "grad_norm": 7.209736347198486,
      "learning_rate": 8.45319330116378e-05,
      "loss": 0.03973647058010101,
      "step": 54520
    },
    {
      "epoch": 15.478285552086291,
      "grad_norm": 3.3896737098693848,
      "learning_rate": 8.452909452171445e-05,
      "loss": 0.02482490837574005,
      "step": 54530
    },
    {
      "epoch": 15.48112404200965,
      "grad_norm": 0.6716392040252686,
      "learning_rate": 8.45262560317911e-05,
      "loss": 0.04154448807239532,
      "step": 54540
    },
    {
      "epoch": 15.483962531933011,
      "grad_norm": 8.893574714660645,
      "learning_rate": 8.452341754186773e-05,
      "loss": 0.03340011239051819,
      "step": 54550
    },
    {
      "epoch": 15.486801021856373,
      "grad_norm": 6.536942481994629,
      "learning_rate": 8.452057905194437e-05,
      "loss": 0.014489671587944031,
      "step": 54560
    },
    {
      "epoch": 15.489639511779734,
      "grad_norm": 2.938420534133911,
      "learning_rate": 8.451774056202101e-05,
      "loss": 0.02795588970184326,
      "step": 54570
    },
    {
      "epoch": 15.492478001703095,
      "grad_norm": 8.389741897583008,
      "learning_rate": 8.451490207209764e-05,
      "loss": 0.029088133573532106,
      "step": 54580
    },
    {
      "epoch": 15.495316491626454,
      "grad_norm": 1.0364203453063965,
      "learning_rate": 8.451206358217428e-05,
      "loss": 0.016360311210155486,
      "step": 54590
    },
    {
      "epoch": 15.498154981549815,
      "grad_norm": 2.46856689453125,
      "learning_rate": 8.450922509225094e-05,
      "loss": 0.01980867236852646,
      "step": 54600
    },
    {
      "epoch": 15.500993471473176,
      "grad_norm": 8.08748722076416,
      "learning_rate": 8.450638660232756e-05,
      "loss": 0.033151456713676454,
      "step": 54610
    },
    {
      "epoch": 15.503831961396537,
      "grad_norm": 12.501373291015625,
      "learning_rate": 8.45035481124042e-05,
      "loss": 0.018897634744644166,
      "step": 54620
    },
    {
      "epoch": 15.506670451319899,
      "grad_norm": 7.797924518585205,
      "learning_rate": 8.450070962248085e-05,
      "loss": 0.024894019961357115,
      "step": 54630
    },
    {
      "epoch": 15.509508941243258,
      "grad_norm": 11.26822280883789,
      "learning_rate": 8.449787113255749e-05,
      "loss": 0.02593979239463806,
      "step": 54640
    },
    {
      "epoch": 15.512347431166619,
      "grad_norm": 5.739212989807129,
      "learning_rate": 8.449503264263412e-05,
      "loss": 0.02486390620470047,
      "step": 54650
    },
    {
      "epoch": 15.51518592108998,
      "grad_norm": 5.1653642654418945,
      "learning_rate": 8.449219415271076e-05,
      "loss": 0.031486839056015015,
      "step": 54660
    },
    {
      "epoch": 15.518024411013341,
      "grad_norm": 5.867390155792236,
      "learning_rate": 8.448935566278741e-05,
      "loss": 0.04039141535758972,
      "step": 54670
    },
    {
      "epoch": 15.520862900936702,
      "grad_norm": 7.855792045593262,
      "learning_rate": 8.448651717286404e-05,
      "loss": 0.043772509694099425,
      "step": 54680
    },
    {
      "epoch": 15.523701390860062,
      "grad_norm": 6.4003496170043945,
      "learning_rate": 8.448367868294068e-05,
      "loss": 0.033394047617912294,
      "step": 54690
    },
    {
      "epoch": 15.526539880783423,
      "grad_norm": 3.44984769821167,
      "learning_rate": 8.448084019301732e-05,
      "loss": 0.02004280537366867,
      "step": 54700
    },
    {
      "epoch": 15.529378370706784,
      "grad_norm": 0.9175532460212708,
      "learning_rate": 8.447800170309395e-05,
      "loss": 0.014948099851608276,
      "step": 54710
    },
    {
      "epoch": 15.532216860630145,
      "grad_norm": 5.156901836395264,
      "learning_rate": 8.447516321317059e-05,
      "loss": 0.022969256341457366,
      "step": 54720
    },
    {
      "epoch": 15.535055350553506,
      "grad_norm": 10.004914283752441,
      "learning_rate": 8.447232472324723e-05,
      "loss": 0.021813562512397765,
      "step": 54730
    },
    {
      "epoch": 15.537893840476865,
      "grad_norm": 1.4644076824188232,
      "learning_rate": 8.446948623332388e-05,
      "loss": 0.02282424122095108,
      "step": 54740
    },
    {
      "epoch": 15.540732330400227,
      "grad_norm": 9.742193222045898,
      "learning_rate": 8.446664774340052e-05,
      "loss": 0.02879665493965149,
      "step": 54750
    },
    {
      "epoch": 15.543570820323588,
      "grad_norm": 4.354197978973389,
      "learning_rate": 8.446380925347716e-05,
      "loss": 0.03276260197162628,
      "step": 54760
    },
    {
      "epoch": 15.546409310246949,
      "grad_norm": 3.7555503845214844,
      "learning_rate": 8.44609707635538e-05,
      "loss": 0.015233895182609558,
      "step": 54770
    },
    {
      "epoch": 15.54924780017031,
      "grad_norm": 6.296515941619873,
      "learning_rate": 8.445813227363043e-05,
      "loss": 0.017288655042648315,
      "step": 54780
    },
    {
      "epoch": 15.552086290093671,
      "grad_norm": 1.4583958387374878,
      "learning_rate": 8.445529378370707e-05,
      "loss": 0.040284544229507446,
      "step": 54790
    },
    {
      "epoch": 15.55492478001703,
      "grad_norm": 3.4453704357147217,
      "learning_rate": 8.445245529378371e-05,
      "loss": 0.024253936111927034,
      "step": 54800
    },
    {
      "epoch": 15.557763269940391,
      "grad_norm": 4.137217998504639,
      "learning_rate": 8.444961680386035e-05,
      "loss": 0.031335747241973876,
      "step": 54810
    },
    {
      "epoch": 15.560601759863752,
      "grad_norm": 3.6918506622314453,
      "learning_rate": 8.444677831393699e-05,
      "loss": 0.012523356080055236,
      "step": 54820
    },
    {
      "epoch": 15.563440249787114,
      "grad_norm": 11.648348808288574,
      "learning_rate": 8.444393982401363e-05,
      "loss": 0.02171445041894913,
      "step": 54830
    },
    {
      "epoch": 15.566278739710475,
      "grad_norm": 2.579878807067871,
      "learning_rate": 8.444110133409026e-05,
      "loss": 0.03140862584114075,
      "step": 54840
    },
    {
      "epoch": 15.569117229633834,
      "grad_norm": 8.60331916809082,
      "learning_rate": 8.44382628441669e-05,
      "loss": 0.034834560751914975,
      "step": 54850
    },
    {
      "epoch": 15.571955719557195,
      "grad_norm": 2.809657096862793,
      "learning_rate": 8.443542435424354e-05,
      "loss": 0.055110251903533934,
      "step": 54860
    },
    {
      "epoch": 15.574794209480556,
      "grad_norm": 4.75227689743042,
      "learning_rate": 8.443258586432019e-05,
      "loss": 0.04067589640617371,
      "step": 54870
    },
    {
      "epoch": 15.577632699403917,
      "grad_norm": 5.746793270111084,
      "learning_rate": 8.442974737439683e-05,
      "loss": 0.019145742058753967,
      "step": 54880
    },
    {
      "epoch": 15.580471189327278,
      "grad_norm": 1.67842435836792,
      "learning_rate": 8.442690888447347e-05,
      "loss": 0.02290988713502884,
      "step": 54890
    },
    {
      "epoch": 15.58330967925064,
      "grad_norm": 9.37649154663086,
      "learning_rate": 8.44240703945501e-05,
      "loss": 0.01786271333694458,
      "step": 54900
    },
    {
      "epoch": 15.586148169173999,
      "grad_norm": 8.962061882019043,
      "learning_rate": 8.442123190462674e-05,
      "loss": 0.04273948967456818,
      "step": 54910
    },
    {
      "epoch": 15.58898665909736,
      "grad_norm": 8.013415336608887,
      "learning_rate": 8.441839341470338e-05,
      "loss": 0.03322238624095917,
      "step": 54920
    },
    {
      "epoch": 15.591825149020721,
      "grad_norm": 5.692356109619141,
      "learning_rate": 8.441555492478002e-05,
      "loss": 0.02370072901248932,
      "step": 54930
    },
    {
      "epoch": 15.594663638944082,
      "grad_norm": 8.148560523986816,
      "learning_rate": 8.441271643485666e-05,
      "loss": 0.0259392648935318,
      "step": 54940
    },
    {
      "epoch": 15.597502128867443,
      "grad_norm": 5.666191577911377,
      "learning_rate": 8.44098779449333e-05,
      "loss": 0.022132568061351776,
      "step": 54950
    },
    {
      "epoch": 15.600340618790803,
      "grad_norm": 10.347456932067871,
      "learning_rate": 8.440703945500994e-05,
      "loss": 0.029416689276695253,
      "step": 54960
    },
    {
      "epoch": 15.603179108714164,
      "grad_norm": 4.348835468292236,
      "learning_rate": 8.440420096508657e-05,
      "loss": 0.014265596866607666,
      "step": 54970
    },
    {
      "epoch": 15.606017598637525,
      "grad_norm": 17.19972038269043,
      "learning_rate": 8.440136247516321e-05,
      "loss": 0.022851523756980897,
      "step": 54980
    },
    {
      "epoch": 15.608856088560886,
      "grad_norm": 7.24250602722168,
      "learning_rate": 8.439852398523986e-05,
      "loss": 0.031148865818977356,
      "step": 54990
    },
    {
      "epoch": 15.611694578484247,
      "grad_norm": 2.0472657680511475,
      "learning_rate": 8.43956854953165e-05,
      "loss": 0.045480182766914366,
      "step": 55000
    },
    {
      "epoch": 15.611694578484247,
      "eval_accuracy": 0.9544096140395498,
      "eval_loss": 0.13611218333244324,
      "eval_runtime": 31.4101,
      "eval_samples_per_second": 500.699,
      "eval_steps_per_second": 7.832,
      "step": 55000
    },
    {
      "epoch": 15.614533068407606,
      "grad_norm": 0.725804328918457,
      "learning_rate": 8.439284700539314e-05,
      "loss": 0.023694561421871187,
      "step": 55010
    },
    {
      "epoch": 15.617371558330968,
      "grad_norm": 3.9917492866516113,
      "learning_rate": 8.439000851546978e-05,
      "loss": 0.027720621228218077,
      "step": 55020
    },
    {
      "epoch": 15.620210048254329,
      "grad_norm": 3.7484641075134277,
      "learning_rate": 8.438717002554641e-05,
      "loss": 0.011597867310047149,
      "step": 55030
    },
    {
      "epoch": 15.62304853817769,
      "grad_norm": 6.442530155181885,
      "learning_rate": 8.438433153562305e-05,
      "loss": 0.02480282336473465,
      "step": 55040
    },
    {
      "epoch": 15.625887028101051,
      "grad_norm": 0.9566351175308228,
      "learning_rate": 8.438149304569969e-05,
      "loss": 0.03094039261341095,
      "step": 55050
    },
    {
      "epoch": 15.62872551802441,
      "grad_norm": 9.80335521697998,
      "learning_rate": 8.437865455577633e-05,
      "loss": 0.04263542592525482,
      "step": 55060
    },
    {
      "epoch": 15.631564007947771,
      "grad_norm": 0.9957788586616516,
      "learning_rate": 8.437581606585297e-05,
      "loss": 0.01051989123225212,
      "step": 55070
    },
    {
      "epoch": 15.634402497871132,
      "grad_norm": 12.291481971740723,
      "learning_rate": 8.437297757592961e-05,
      "loss": 0.016485369205474852,
      "step": 55080
    },
    {
      "epoch": 15.637240987794494,
      "grad_norm": 1.9296355247497559,
      "learning_rate": 8.437013908600626e-05,
      "loss": 0.025524410605430602,
      "step": 55090
    },
    {
      "epoch": 15.640079477717855,
      "grad_norm": 2.2079684734344482,
      "learning_rate": 8.436730059608288e-05,
      "loss": 0.02186250388622284,
      "step": 55100
    },
    {
      "epoch": 15.642917967641214,
      "grad_norm": 14.009074211120605,
      "learning_rate": 8.436446210615952e-05,
      "loss": 0.043946081399917604,
      "step": 55110
    },
    {
      "epoch": 15.645756457564575,
      "grad_norm": 6.56678581237793,
      "learning_rate": 8.436162361623617e-05,
      "loss": 0.014149743318557739,
      "step": 55120
    },
    {
      "epoch": 15.648594947487936,
      "grad_norm": 0.9099812507629395,
      "learning_rate": 8.43587851263128e-05,
      "loss": 0.033816903829574585,
      "step": 55130
    },
    {
      "epoch": 15.651433437411297,
      "grad_norm": 1.2540795803070068,
      "learning_rate": 8.435594663638945e-05,
      "loss": 0.012104620784521103,
      "step": 55140
    },
    {
      "epoch": 15.654271927334658,
      "grad_norm": 8.036087989807129,
      "learning_rate": 8.435310814646609e-05,
      "loss": 0.031072506308555604,
      "step": 55150
    },
    {
      "epoch": 15.65711041725802,
      "grad_norm": 5.32780647277832,
      "learning_rate": 8.435026965654272e-05,
      "loss": 0.03889361619949341,
      "step": 55160
    },
    {
      "epoch": 15.659948907181379,
      "grad_norm": 5.735215187072754,
      "learning_rate": 8.434743116661936e-05,
      "loss": 0.0414595752954483,
      "step": 55170
    },
    {
      "epoch": 15.66278739710474,
      "grad_norm": 6.035196781158447,
      "learning_rate": 8.4344592676696e-05,
      "loss": 0.020461562275886535,
      "step": 55180
    },
    {
      "epoch": 15.665625887028101,
      "grad_norm": 0.9485716223716736,
      "learning_rate": 8.434175418677264e-05,
      "loss": 0.02586342990398407,
      "step": 55190
    },
    {
      "epoch": 15.668464376951462,
      "grad_norm": 9.054306030273438,
      "learning_rate": 8.433891569684928e-05,
      "loss": 0.02514852285385132,
      "step": 55200
    },
    {
      "epoch": 15.671302866874823,
      "grad_norm": 0.6084137558937073,
      "learning_rate": 8.433607720692592e-05,
      "loss": 0.02879243791103363,
      "step": 55210
    },
    {
      "epoch": 15.674141356798183,
      "grad_norm": 1.6793603897094727,
      "learning_rate": 8.433323871700257e-05,
      "loss": 0.010233908891677856,
      "step": 55220
    },
    {
      "epoch": 15.676979846721544,
      "grad_norm": 1.3449037075042725,
      "learning_rate": 8.43304002270792e-05,
      "loss": 0.021717290580272674,
      "step": 55230
    },
    {
      "epoch": 15.679818336644905,
      "grad_norm": 6.453930854797363,
      "learning_rate": 8.432756173715584e-05,
      "loss": 0.024816076457500457,
      "step": 55240
    },
    {
      "epoch": 15.682656826568266,
      "grad_norm": 6.76559591293335,
      "learning_rate": 8.432472324723248e-05,
      "loss": 0.012948113679885864,
      "step": 55250
    },
    {
      "epoch": 15.685495316491627,
      "grad_norm": 3.3474678993225098,
      "learning_rate": 8.43218847573091e-05,
      "loss": 0.029218363761901855,
      "step": 55260
    },
    {
      "epoch": 15.688333806414986,
      "grad_norm": 8.351326942443848,
      "learning_rate": 8.431904626738576e-05,
      "loss": 0.026746863126754762,
      "step": 55270
    },
    {
      "epoch": 15.691172296338348,
      "grad_norm": 10.07962417602539,
      "learning_rate": 8.43162077774624e-05,
      "loss": 0.0370498925447464,
      "step": 55280
    },
    {
      "epoch": 15.694010786261709,
      "grad_norm": 1.937565803527832,
      "learning_rate": 8.431336928753903e-05,
      "loss": 0.02836006283760071,
      "step": 55290
    },
    {
      "epoch": 15.69684927618507,
      "grad_norm": 2.2049403190612793,
      "learning_rate": 8.431053079761567e-05,
      "loss": 0.024386924505233765,
      "step": 55300
    },
    {
      "epoch": 15.69968776610843,
      "grad_norm": 11.942442893981934,
      "learning_rate": 8.430769230769231e-05,
      "loss": 0.025810697674751283,
      "step": 55310
    },
    {
      "epoch": 15.702526256031792,
      "grad_norm": 15.803805351257324,
      "learning_rate": 8.430485381776895e-05,
      "loss": 0.0414841890335083,
      "step": 55320
    },
    {
      "epoch": 15.705364745955151,
      "grad_norm": 2.537261962890625,
      "learning_rate": 8.430201532784558e-05,
      "loss": 0.019336915016174315,
      "step": 55330
    },
    {
      "epoch": 15.708203235878512,
      "grad_norm": 2.325127363204956,
      "learning_rate": 8.429917683792224e-05,
      "loss": 0.013833549618721009,
      "step": 55340
    },
    {
      "epoch": 15.711041725801874,
      "grad_norm": 3.4864227771759033,
      "learning_rate": 8.429633834799888e-05,
      "loss": 0.035663875937461856,
      "step": 55350
    },
    {
      "epoch": 15.713880215725235,
      "grad_norm": 2.592761516571045,
      "learning_rate": 8.42934998580755e-05,
      "loss": 0.028659674525260925,
      "step": 55360
    },
    {
      "epoch": 15.716718705648596,
      "grad_norm": 10.83828353881836,
      "learning_rate": 8.429066136815215e-05,
      "loss": 0.03129350543022156,
      "step": 55370
    },
    {
      "epoch": 15.719557195571955,
      "grad_norm": 3.1333580017089844,
      "learning_rate": 8.428782287822879e-05,
      "loss": 0.019594284892082214,
      "step": 55380
    },
    {
      "epoch": 15.722395685495316,
      "grad_norm": 14.388250350952148,
      "learning_rate": 8.428498438830542e-05,
      "loss": 0.01917848885059357,
      "step": 55390
    },
    {
      "epoch": 15.725234175418677,
      "grad_norm": 12.465145111083984,
      "learning_rate": 8.428214589838207e-05,
      "loss": 0.034659892320632935,
      "step": 55400
    },
    {
      "epoch": 15.728072665342038,
      "grad_norm": 2.8771822452545166,
      "learning_rate": 8.427930740845871e-05,
      "loss": 0.025818976759910583,
      "step": 55410
    },
    {
      "epoch": 15.7309111552654,
      "grad_norm": 10.621774673461914,
      "learning_rate": 8.427646891853534e-05,
      "loss": 0.0350786954164505,
      "step": 55420
    },
    {
      "epoch": 15.733749645188759,
      "grad_norm": 4.423511505126953,
      "learning_rate": 8.427363042861198e-05,
      "loss": 0.03854709267616272,
      "step": 55430
    },
    {
      "epoch": 15.73658813511212,
      "grad_norm": 1.963225245475769,
      "learning_rate": 8.427079193868862e-05,
      "loss": 0.03947713673114776,
      "step": 55440
    },
    {
      "epoch": 15.739426625035481,
      "grad_norm": 3.383223056793213,
      "learning_rate": 8.426795344876526e-05,
      "loss": 0.021287114918231965,
      "step": 55450
    },
    {
      "epoch": 15.742265114958842,
      "grad_norm": 2.511078119277954,
      "learning_rate": 8.426511495884189e-05,
      "loss": 0.016826075315475465,
      "step": 55460
    },
    {
      "epoch": 15.745103604882203,
      "grad_norm": 8.995229721069336,
      "learning_rate": 8.426227646891855e-05,
      "loss": 0.03690308928489685,
      "step": 55470
    },
    {
      "epoch": 15.747942094805563,
      "grad_norm": 11.649239540100098,
      "learning_rate": 8.425943797899519e-05,
      "loss": 0.02400638908147812,
      "step": 55480
    },
    {
      "epoch": 15.750780584728924,
      "grad_norm": 12.552679061889648,
      "learning_rate": 8.425659948907182e-05,
      "loss": 0.05099513530731201,
      "step": 55490
    },
    {
      "epoch": 15.753619074652285,
      "grad_norm": 2.8938004970550537,
      "learning_rate": 8.425376099914846e-05,
      "loss": 0.020146849751472472,
      "step": 55500
    },
    {
      "epoch": 15.753619074652285,
      "eval_accuracy": 0.9609588605582756,
      "eval_loss": 0.12461204826831818,
      "eval_runtime": 31.6124,
      "eval_samples_per_second": 497.494,
      "eval_steps_per_second": 7.782,
      "step": 55500
    },
    {
      "epoch": 15.756457564575646,
      "grad_norm": 0.6011388301849365,
      "learning_rate": 8.42509225092251e-05,
      "loss": 0.014214943349361419,
      "step": 55510
    },
    {
      "epoch": 15.759296054499007,
      "grad_norm": 6.048890590667725,
      "learning_rate": 8.424808401930173e-05,
      "loss": 0.035259172320365906,
      "step": 55520
    },
    {
      "epoch": 15.762134544422366,
      "grad_norm": 2.1683619022369385,
      "learning_rate": 8.424524552937837e-05,
      "loss": 0.026011309027671813,
      "step": 55530
    },
    {
      "epoch": 15.764973034345727,
      "grad_norm": 4.43717098236084,
      "learning_rate": 8.424240703945502e-05,
      "loss": 0.020434191823005675,
      "step": 55540
    },
    {
      "epoch": 15.767811524269089,
      "grad_norm": 4.325998306274414,
      "learning_rate": 8.423956854953165e-05,
      "loss": 0.03220485746860504,
      "step": 55550
    },
    {
      "epoch": 15.77065001419245,
      "grad_norm": 2.4194178581237793,
      "learning_rate": 8.423673005960829e-05,
      "loss": 0.02003835141658783,
      "step": 55560
    },
    {
      "epoch": 15.77348850411581,
      "grad_norm": 4.828169345855713,
      "learning_rate": 8.423389156968493e-05,
      "loss": 0.016806913912296294,
      "step": 55570
    },
    {
      "epoch": 15.776326994039172,
      "grad_norm": 3.605966091156006,
      "learning_rate": 8.423105307976157e-05,
      "loss": 0.02392922043800354,
      "step": 55580
    },
    {
      "epoch": 15.779165483962531,
      "grad_norm": 3.2854690551757812,
      "learning_rate": 8.42282145898382e-05,
      "loss": 0.01690668612718582,
      "step": 55590
    },
    {
      "epoch": 15.782003973885892,
      "grad_norm": 8.527112007141113,
      "learning_rate": 8.422537609991486e-05,
      "loss": 0.02466445565223694,
      "step": 55600
    },
    {
      "epoch": 15.784842463809253,
      "grad_norm": 15.029330253601074,
      "learning_rate": 8.42225376099915e-05,
      "loss": 0.026534241437911988,
      "step": 55610
    },
    {
      "epoch": 15.787680953732615,
      "grad_norm": 18.220033645629883,
      "learning_rate": 8.421969912006813e-05,
      "loss": 0.04281447231769562,
      "step": 55620
    },
    {
      "epoch": 15.790519443655976,
      "grad_norm": 17.181987762451172,
      "learning_rate": 8.421686063014477e-05,
      "loss": 0.018392178416252136,
      "step": 55630
    },
    {
      "epoch": 15.793357933579335,
      "grad_norm": 9.275490760803223,
      "learning_rate": 8.421402214022141e-05,
      "loss": 0.020517879724502565,
      "step": 55640
    },
    {
      "epoch": 15.796196423502696,
      "grad_norm": 0.9653279185295105,
      "learning_rate": 8.421118365029804e-05,
      "loss": 0.03457046747207641,
      "step": 55650
    },
    {
      "epoch": 15.799034913426057,
      "grad_norm": 4.480401515960693,
      "learning_rate": 8.420834516037468e-05,
      "loss": 0.02772296667098999,
      "step": 55660
    },
    {
      "epoch": 15.801873403349418,
      "grad_norm": 7.011767387390137,
      "learning_rate": 8.420550667045133e-05,
      "loss": 0.04459327757358551,
      "step": 55670
    },
    {
      "epoch": 15.80471189327278,
      "grad_norm": 0.9134606122970581,
      "learning_rate": 8.420266818052796e-05,
      "loss": 0.014770305156707764,
      "step": 55680
    },
    {
      "epoch": 15.80755038319614,
      "grad_norm": 3.7655885219573975,
      "learning_rate": 8.41998296906046e-05,
      "loss": 0.0340694934129715,
      "step": 55690
    },
    {
      "epoch": 15.8103888731195,
      "grad_norm": 2.3053646087646484,
      "learning_rate": 8.419699120068124e-05,
      "loss": 0.03109254240989685,
      "step": 55700
    },
    {
      "epoch": 15.813227363042861,
      "grad_norm": 1.0620684623718262,
      "learning_rate": 8.419415271075789e-05,
      "loss": 0.01757873147726059,
      "step": 55710
    },
    {
      "epoch": 15.816065852966222,
      "grad_norm": 5.15792179107666,
      "learning_rate": 8.419131422083451e-05,
      "loss": 0.027486610412597656,
      "step": 55720
    },
    {
      "epoch": 15.818904342889583,
      "grad_norm": 6.560937404632568,
      "learning_rate": 8.418847573091115e-05,
      "loss": 0.016028845310211183,
      "step": 55730
    },
    {
      "epoch": 15.821742832812944,
      "grad_norm": 16.208751678466797,
      "learning_rate": 8.41856372409878e-05,
      "loss": 0.026366108655929567,
      "step": 55740
    },
    {
      "epoch": 15.824581322736304,
      "grad_norm": 1.8737319707870483,
      "learning_rate": 8.418279875106444e-05,
      "loss": 0.04411182701587677,
      "step": 55750
    },
    {
      "epoch": 15.827419812659665,
      "grad_norm": 12.58658504486084,
      "learning_rate": 8.417996026114108e-05,
      "loss": 0.033492755889892575,
      "step": 55760
    },
    {
      "epoch": 15.830258302583026,
      "grad_norm": 7.875770568847656,
      "learning_rate": 8.417712177121772e-05,
      "loss": 0.025324517488479616,
      "step": 55770
    },
    {
      "epoch": 15.833096792506387,
      "grad_norm": 1.199244499206543,
      "learning_rate": 8.417428328129435e-05,
      "loss": 0.022986453771591187,
      "step": 55780
    },
    {
      "epoch": 15.835935282429748,
      "grad_norm": 3.8129796981811523,
      "learning_rate": 8.417144479137099e-05,
      "loss": 0.02322043925523758,
      "step": 55790
    },
    {
      "epoch": 15.838773772353107,
      "grad_norm": 0.5839950442314148,
      "learning_rate": 8.416860630144764e-05,
      "loss": 0.02498307228088379,
      "step": 55800
    },
    {
      "epoch": 15.841612262276469,
      "grad_norm": 12.658780097961426,
      "learning_rate": 8.416576781152427e-05,
      "loss": 0.03638279438018799,
      "step": 55810
    },
    {
      "epoch": 15.84445075219983,
      "grad_norm": 0.9448693990707397,
      "learning_rate": 8.416292932160091e-05,
      "loss": 0.0205583393573761,
      "step": 55820
    },
    {
      "epoch": 15.84728924212319,
      "grad_norm": 4.23764705657959,
      "learning_rate": 8.416009083167755e-05,
      "loss": 0.023648782074451445,
      "step": 55830
    },
    {
      "epoch": 15.850127732046552,
      "grad_norm": 6.288824558258057,
      "learning_rate": 8.415725234175418e-05,
      "loss": 0.02515784800052643,
      "step": 55840
    },
    {
      "epoch": 15.852966221969911,
      "grad_norm": 0.7288370728492737,
      "learning_rate": 8.415441385183082e-05,
      "loss": 0.012754830718040466,
      "step": 55850
    },
    {
      "epoch": 15.855804711893272,
      "grad_norm": 0.5737016797065735,
      "learning_rate": 8.415157536190747e-05,
      "loss": 0.04049540162086487,
      "step": 55860
    },
    {
      "epoch": 15.858643201816633,
      "grad_norm": 8.069538116455078,
      "learning_rate": 8.41487368719841e-05,
      "loss": 0.04047139883041382,
      "step": 55870
    },
    {
      "epoch": 15.861481691739995,
      "grad_norm": 1.805380940437317,
      "learning_rate": 8.414589838206075e-05,
      "loss": 0.01842280328273773,
      "step": 55880
    },
    {
      "epoch": 15.864320181663356,
      "grad_norm": 10.589919090270996,
      "learning_rate": 8.414305989213739e-05,
      "loss": 0.02534562349319458,
      "step": 55890
    },
    {
      "epoch": 15.867158671586715,
      "grad_norm": 3.550118923187256,
      "learning_rate": 8.414022140221403e-05,
      "loss": 0.0255268394947052,
      "step": 55900
    },
    {
      "epoch": 15.869997161510076,
      "grad_norm": 14.242570877075195,
      "learning_rate": 8.413738291229066e-05,
      "loss": 0.01810208261013031,
      "step": 55910
    },
    {
      "epoch": 15.872835651433437,
      "grad_norm": 10.449312210083008,
      "learning_rate": 8.41345444223673e-05,
      "loss": 0.026680964231491088,
      "step": 55920
    },
    {
      "epoch": 15.875674141356798,
      "grad_norm": 8.877686500549316,
      "learning_rate": 8.413170593244395e-05,
      "loss": 0.022716453671455382,
      "step": 55930
    },
    {
      "epoch": 15.87851263128016,
      "grad_norm": 2.263636589050293,
      "learning_rate": 8.412886744252058e-05,
      "loss": 0.023564670979976655,
      "step": 55940
    },
    {
      "epoch": 15.88135112120352,
      "grad_norm": 5.455941677093506,
      "learning_rate": 8.412602895259722e-05,
      "loss": 0.026516324281692503,
      "step": 55950
    },
    {
      "epoch": 15.88418961112688,
      "grad_norm": 6.262031078338623,
      "learning_rate": 8.412319046267387e-05,
      "loss": 0.0171290248632431,
      "step": 55960
    },
    {
      "epoch": 15.887028101050241,
      "grad_norm": 10.699884414672852,
      "learning_rate": 8.412035197275049e-05,
      "loss": 0.02896207869052887,
      "step": 55970
    },
    {
      "epoch": 15.889866590973602,
      "grad_norm": 4.022065162658691,
      "learning_rate": 8.411751348282713e-05,
      "loss": 0.031614893674850465,
      "step": 55980
    },
    {
      "epoch": 15.892705080896963,
      "grad_norm": 11.237079620361328,
      "learning_rate": 8.411467499290378e-05,
      "loss": 0.025946211814880372,
      "step": 55990
    },
    {
      "epoch": 15.895543570820324,
      "grad_norm": Infinity,
      "learning_rate": 8.411183650298042e-05,
      "loss": 0.030976390838623045,
      "step": 56000
    },
    {
      "epoch": 15.895543570820324,
      "eval_accuracy": 0.9621033890761111,
      "eval_loss": 0.11983194202184677,
      "eval_runtime": 31.3136,
      "eval_samples_per_second": 502.241,
      "eval_steps_per_second": 7.856,
      "step": 56000
    },
    {
      "epoch": 15.898382060743684,
      "grad_norm": 4.911223411560059,
      "learning_rate": 8.41092818620494e-05,
      "loss": 0.03257195055484772,
      "step": 56010
    },
    {
      "epoch": 15.901220550667045,
      "grad_norm": 3.5249905586242676,
      "learning_rate": 8.410644337212603e-05,
      "loss": 0.020603126287460326,
      "step": 56020
    },
    {
      "epoch": 15.904059040590406,
      "grad_norm": 4.776357650756836,
      "learning_rate": 8.410360488220267e-05,
      "loss": 0.01293005645275116,
      "step": 56030
    },
    {
      "epoch": 15.906897530513767,
      "grad_norm": 16.306373596191406,
      "learning_rate": 8.410076639227931e-05,
      "loss": 0.028395670652389526,
      "step": 56040
    },
    {
      "epoch": 15.909736020437128,
      "grad_norm": 0.15443992614746094,
      "learning_rate": 8.409792790235595e-05,
      "loss": 0.049327808618545535,
      "step": 56050
    },
    {
      "epoch": 15.912574510360487,
      "grad_norm": 7.249589443206787,
      "learning_rate": 8.409508941243259e-05,
      "loss": 0.04538487792015076,
      "step": 56060
    },
    {
      "epoch": 15.915413000283849,
      "grad_norm": 14.804144859313965,
      "learning_rate": 8.409225092250923e-05,
      "loss": 0.04028477370738983,
      "step": 56070
    },
    {
      "epoch": 15.91825149020721,
      "grad_norm": 12.310279846191406,
      "learning_rate": 8.408941243258587e-05,
      "loss": 0.025865823030471802,
      "step": 56080
    },
    {
      "epoch": 15.92108998013057,
      "grad_norm": 0.782041609287262,
      "learning_rate": 8.40865739426625e-05,
      "loss": 0.017394444346427916,
      "step": 56090
    },
    {
      "epoch": 15.923928470053932,
      "grad_norm": 1.530235767364502,
      "learning_rate": 8.408373545273914e-05,
      "loss": 0.014441056549549103,
      "step": 56100
    },
    {
      "epoch": 15.926766959977293,
      "grad_norm": 9.699479103088379,
      "learning_rate": 8.408089696281578e-05,
      "loss": 0.01742193102836609,
      "step": 56110
    },
    {
      "epoch": 15.929605449900652,
      "grad_norm": 0.7000200152397156,
      "learning_rate": 8.407805847289243e-05,
      "loss": 0.014637525379657745,
      "step": 56120
    },
    {
      "epoch": 15.932443939824013,
      "grad_norm": 1.259110450744629,
      "learning_rate": 8.407521998296907e-05,
      "loss": 0.019371147453784942,
      "step": 56130
    },
    {
      "epoch": 15.935282429747375,
      "grad_norm": 2.1443381309509277,
      "learning_rate": 8.407238149304571e-05,
      "loss": 0.0268601655960083,
      "step": 56140
    },
    {
      "epoch": 15.938120919670736,
      "grad_norm": 3.9176294803619385,
      "learning_rate": 8.406954300312234e-05,
      "loss": 0.02974343001842499,
      "step": 56150
    },
    {
      "epoch": 15.940959409594097,
      "grad_norm": 1.5984960794448853,
      "learning_rate": 8.406670451319898e-05,
      "loss": 0.00898154228925705,
      "step": 56160
    },
    {
      "epoch": 15.943797899517456,
      "grad_norm": 2.9393582344055176,
      "learning_rate": 8.406386602327562e-05,
      "loss": 0.01995365172624588,
      "step": 56170
    },
    {
      "epoch": 15.946636389440817,
      "grad_norm": 2.334904670715332,
      "learning_rate": 8.406102753335226e-05,
      "loss": 0.022166568040847778,
      "step": 56180
    },
    {
      "epoch": 15.949474879364178,
      "grad_norm": 4.428810119628906,
      "learning_rate": 8.40581890434289e-05,
      "loss": 0.02068374752998352,
      "step": 56190
    },
    {
      "epoch": 15.95231336928754,
      "grad_norm": 0.9818922877311707,
      "learning_rate": 8.405535055350554e-05,
      "loss": 0.0217548206448555,
      "step": 56200
    },
    {
      "epoch": 15.9551518592109,
      "grad_norm": 9.420214653015137,
      "learning_rate": 8.405251206358218e-05,
      "loss": 0.025222313404083253,
      "step": 56210
    },
    {
      "epoch": 15.95799034913426,
      "grad_norm": 5.281280517578125,
      "learning_rate": 8.404967357365881e-05,
      "loss": 0.025636616349220275,
      "step": 56220
    },
    {
      "epoch": 15.960828839057621,
      "grad_norm": 6.161774635314941,
      "learning_rate": 8.404683508373545e-05,
      "loss": 0.029733723402023314,
      "step": 56230
    },
    {
      "epoch": 15.963667328980982,
      "grad_norm": 5.249983787536621,
      "learning_rate": 8.40439965938121e-05,
      "loss": 0.020673768222332002,
      "step": 56240
    },
    {
      "epoch": 15.966505818904343,
      "grad_norm": 0.741504967212677,
      "learning_rate": 8.404115810388874e-05,
      "loss": 0.02231096476316452,
      "step": 56250
    },
    {
      "epoch": 15.969344308827704,
      "grad_norm": 2.4594666957855225,
      "learning_rate": 8.403831961396538e-05,
      "loss": 0.03507919013500214,
      "step": 56260
    },
    {
      "epoch": 15.972182798751064,
      "grad_norm": 8.734623908996582,
      "learning_rate": 8.403548112404202e-05,
      "loss": 0.014858177304267884,
      "step": 56270
    },
    {
      "epoch": 15.975021288674425,
      "grad_norm": 3.020758867263794,
      "learning_rate": 8.403264263411865e-05,
      "loss": 0.027657362818717956,
      "step": 56280
    },
    {
      "epoch": 15.977859778597786,
      "grad_norm": 0.1763543337583542,
      "learning_rate": 8.402980414419529e-05,
      "loss": 0.01789838671684265,
      "step": 56290
    },
    {
      "epoch": 15.980698268521147,
      "grad_norm": 5.489733695983887,
      "learning_rate": 8.402696565427193e-05,
      "loss": 0.022160105407238007,
      "step": 56300
    },
    {
      "epoch": 15.983536758444508,
      "grad_norm": 4.364232063293457,
      "learning_rate": 8.402412716434857e-05,
      "loss": 0.02629002630710602,
      "step": 56310
    },
    {
      "epoch": 15.986375248367867,
      "grad_norm": 14.296595573425293,
      "learning_rate": 8.402128867442521e-05,
      "loss": 0.03456106185913086,
      "step": 56320
    },
    {
      "epoch": 15.989213738291228,
      "grad_norm": 9.274201393127441,
      "learning_rate": 8.401845018450185e-05,
      "loss": 0.03412656188011169,
      "step": 56330
    },
    {
      "epoch": 15.99205222821459,
      "grad_norm": 7.401856899261475,
      "learning_rate": 8.40156116945785e-05,
      "loss": 0.031022083759307862,
      "step": 56340
    },
    {
      "epoch": 15.99489071813795,
      "grad_norm": 7.272182464599609,
      "learning_rate": 8.401277320465512e-05,
      "loss": 0.046731477975845336,
      "step": 56350
    },
    {
      "epoch": 15.997729208061312,
      "grad_norm": 7.696033000946045,
      "learning_rate": 8.400993471473176e-05,
      "loss": 0.018336690962314606,
      "step": 56360
    },
    {
      "epoch": 16.00056769798467,
      "grad_norm": 3.114326238632202,
      "learning_rate": 8.40070962248084e-05,
      "loss": 0.022405683994293213,
      "step": 56370
    },
    {
      "epoch": 16.003406187908034,
      "grad_norm": 0.5919771790504456,
      "learning_rate": 8.400425773488505e-05,
      "loss": 0.010076545178890228,
      "step": 56380
    },
    {
      "epoch": 16.006244677831393,
      "grad_norm": 4.794750690460205,
      "learning_rate": 8.400141924496169e-05,
      "loss": 0.029783767461776734,
      "step": 56390
    },
    {
      "epoch": 16.009083167754753,
      "grad_norm": 1.4368441104888916,
      "learning_rate": 8.399858075503833e-05,
      "loss": 0.01736864298582077,
      "step": 56400
    },
    {
      "epoch": 16.011921657678116,
      "grad_norm": 1.3042086362838745,
      "learning_rate": 8.399574226511496e-05,
      "loss": 0.008252985030412673,
      "step": 56410
    },
    {
      "epoch": 16.014760147601475,
      "grad_norm": 12.989331245422363,
      "learning_rate": 8.39929037751916e-05,
      "loss": 0.04838740825653076,
      "step": 56420
    },
    {
      "epoch": 16.017598637524838,
      "grad_norm": 1.1730929613113403,
      "learning_rate": 8.399006528526824e-05,
      "loss": 0.01593453735113144,
      "step": 56430
    },
    {
      "epoch": 16.020437127448197,
      "grad_norm": 14.540125846862793,
      "learning_rate": 8.398722679534488e-05,
      "loss": 0.015429207682609558,
      "step": 56440
    },
    {
      "epoch": 16.02327561737156,
      "grad_norm": 1.9196276664733887,
      "learning_rate": 8.398438830542152e-05,
      "loss": 0.03693382143974304,
      "step": 56450
    },
    {
      "epoch": 16.02611410729492,
      "grad_norm": 3.64133358001709,
      "learning_rate": 8.398154981549816e-05,
      "loss": 0.01034165620803833,
      "step": 56460
    },
    {
      "epoch": 16.02895259721828,
      "grad_norm": 0.6410951018333435,
      "learning_rate": 8.39787113255748e-05,
      "loss": 0.020341362059116363,
      "step": 56470
    },
    {
      "epoch": 16.03179108714164,
      "grad_norm": 0.3791448175907135,
      "learning_rate": 8.397587283565143e-05,
      "loss": 0.020900492370128632,
      "step": 56480
    },
    {
      "epoch": 16.034629577065,
      "grad_norm": 2.473736047744751,
      "learning_rate": 8.397303434572807e-05,
      "loss": 0.019682198762893677,
      "step": 56490
    },
    {
      "epoch": 16.037468066988364,
      "grad_norm": 0.5253610014915466,
      "learning_rate": 8.397019585580472e-05,
      "loss": 0.02010924816131592,
      "step": 56500
    },
    {
      "epoch": 16.037468066988364,
      "eval_accuracy": 0.9591148979462072,
      "eval_loss": 0.12914350628852844,
      "eval_runtime": 31.935,
      "eval_samples_per_second": 492.469,
      "eval_steps_per_second": 7.703,
      "step": 56500
    },
    {
      "epoch": 16.040306556911723,
      "grad_norm": 2.5677573680877686,
      "learning_rate": 8.396735736588136e-05,
      "loss": 0.017707046866416932,
      "step": 56510
    },
    {
      "epoch": 16.043145046835082,
      "grad_norm": 12.400981903076172,
      "learning_rate": 8.3964518875958e-05,
      "loss": 0.02374941110610962,
      "step": 56520
    },
    {
      "epoch": 16.045983536758445,
      "grad_norm": 5.282382488250732,
      "learning_rate": 8.396168038603464e-05,
      "loss": 0.01573721468448639,
      "step": 56530
    },
    {
      "epoch": 16.048822026681805,
      "grad_norm": 7.666718006134033,
      "learning_rate": 8.395884189611127e-05,
      "loss": 0.010613903403282166,
      "step": 56540
    },
    {
      "epoch": 16.051660516605168,
      "grad_norm": 3.9401490688323975,
      "learning_rate": 8.395600340618791e-05,
      "loss": 0.016375108063220976,
      "step": 56550
    },
    {
      "epoch": 16.054499006528527,
      "grad_norm": 3.0923590660095215,
      "learning_rate": 8.395316491626455e-05,
      "loss": 0.024221356213092803,
      "step": 56560
    },
    {
      "epoch": 16.057337496451886,
      "grad_norm": 8.81050968170166,
      "learning_rate": 8.395032642634119e-05,
      "loss": 0.012352322041988373,
      "step": 56570
    },
    {
      "epoch": 16.06017598637525,
      "grad_norm": 3.850731611251831,
      "learning_rate": 8.394748793641783e-05,
      "loss": 0.010475626587867737,
      "step": 56580
    },
    {
      "epoch": 16.06301447629861,
      "grad_norm": 5.724727630615234,
      "learning_rate": 8.394464944649448e-05,
      "loss": 0.011408202350139618,
      "step": 56590
    },
    {
      "epoch": 16.06585296622197,
      "grad_norm": 2.9450466632843018,
      "learning_rate": 8.394181095657112e-05,
      "loss": 0.009185663610696792,
      "step": 56600
    },
    {
      "epoch": 16.06869145614533,
      "grad_norm": 0.4280485510826111,
      "learning_rate": 8.393897246664774e-05,
      "loss": 0.013979765772819518,
      "step": 56610
    },
    {
      "epoch": 16.07152994606869,
      "grad_norm": 0.3558211922645569,
      "learning_rate": 8.393613397672439e-05,
      "loss": 0.019109795987606048,
      "step": 56620
    },
    {
      "epoch": 16.074368435992053,
      "grad_norm": 5.710763931274414,
      "learning_rate": 8.393329548680103e-05,
      "loss": 0.01879764497280121,
      "step": 56630
    },
    {
      "epoch": 16.077206925915412,
      "grad_norm": 2.8225626945495605,
      "learning_rate": 8.393045699687765e-05,
      "loss": 0.014234524965286256,
      "step": 56640
    },
    {
      "epoch": 16.080045415838775,
      "grad_norm": 1.5015414953231812,
      "learning_rate": 8.392761850695431e-05,
      "loss": 0.006666857749223709,
      "step": 56650
    },
    {
      "epoch": 16.082883905762134,
      "grad_norm": 1.0233074426651,
      "learning_rate": 8.392478001703095e-05,
      "loss": 0.022152113914489745,
      "step": 56660
    },
    {
      "epoch": 16.085722395685494,
      "grad_norm": 1.9204943180084229,
      "learning_rate": 8.392194152710758e-05,
      "loss": 0.019176481664180754,
      "step": 56670
    },
    {
      "epoch": 16.088560885608857,
      "grad_norm": 2.049398183822632,
      "learning_rate": 8.391910303718422e-05,
      "loss": 0.013513170182704926,
      "step": 56680
    },
    {
      "epoch": 16.091399375532216,
      "grad_norm": 5.297300338745117,
      "learning_rate": 8.391626454726086e-05,
      "loss": 0.02534930408000946,
      "step": 56690
    },
    {
      "epoch": 16.09423786545558,
      "grad_norm": 3.6562960147857666,
      "learning_rate": 8.39134260573375e-05,
      "loss": 0.020561394095420838,
      "step": 56700
    },
    {
      "epoch": 16.097076355378938,
      "grad_norm": 9.539783477783203,
      "learning_rate": 8.391058756741414e-05,
      "loss": 0.02547207772731781,
      "step": 56710
    },
    {
      "epoch": 16.099914845302298,
      "grad_norm": 9.702581405639648,
      "learning_rate": 8.390774907749079e-05,
      "loss": 0.03833670914173126,
      "step": 56720
    },
    {
      "epoch": 16.10275333522566,
      "grad_norm": 6.7406840324401855,
      "learning_rate": 8.390491058756741e-05,
      "loss": 0.01657630205154419,
      "step": 56730
    },
    {
      "epoch": 16.10559182514902,
      "grad_norm": 1.8221476078033447,
      "learning_rate": 8.390207209764406e-05,
      "loss": 0.016314813494682313,
      "step": 56740
    },
    {
      "epoch": 16.108430315072383,
      "grad_norm": 1.935363531112671,
      "learning_rate": 8.38992336077207e-05,
      "loss": 0.009916795045137405,
      "step": 56750
    },
    {
      "epoch": 16.111268804995742,
      "grad_norm": 1.9656075239181519,
      "learning_rate": 8.389639511779734e-05,
      "loss": 0.02160351872444153,
      "step": 56760
    },
    {
      "epoch": 16.1141072949191,
      "grad_norm": 10.291141510009766,
      "learning_rate": 8.389355662787397e-05,
      "loss": 0.036981242895126346,
      "step": 56770
    },
    {
      "epoch": 16.116945784842464,
      "grad_norm": 8.028336524963379,
      "learning_rate": 8.389071813795062e-05,
      "loss": 0.02217787057161331,
      "step": 56780
    },
    {
      "epoch": 16.119784274765824,
      "grad_norm": 10.198317527770996,
      "learning_rate": 8.388787964802726e-05,
      "loss": 0.020279482007026672,
      "step": 56790
    },
    {
      "epoch": 16.122622764689186,
      "grad_norm": 8.702027320861816,
      "learning_rate": 8.388504115810389e-05,
      "loss": 0.023785114288330078,
      "step": 56800
    },
    {
      "epoch": 16.125461254612546,
      "grad_norm": 9.861729621887207,
      "learning_rate": 8.388220266818053e-05,
      "loss": 0.02164105325937271,
      "step": 56810
    },
    {
      "epoch": 16.12829974453591,
      "grad_norm": 2.009347915649414,
      "learning_rate": 8.387936417825717e-05,
      "loss": 0.015425843000411988,
      "step": 56820
    },
    {
      "epoch": 16.131138234459268,
      "grad_norm": 4.039822578430176,
      "learning_rate": 8.38765256883338e-05,
      "loss": 0.02053661197423935,
      "step": 56830
    },
    {
      "epoch": 16.133976724382627,
      "grad_norm": 0.49414023756980896,
      "learning_rate": 8.387368719841044e-05,
      "loss": 0.01741681694984436,
      "step": 56840
    },
    {
      "epoch": 16.13681521430599,
      "grad_norm": 12.763054847717285,
      "learning_rate": 8.38708487084871e-05,
      "loss": 0.0228745698928833,
      "step": 56850
    },
    {
      "epoch": 16.13965370422935,
      "grad_norm": 7.280646324157715,
      "learning_rate": 8.386801021856372e-05,
      "loss": 0.014351931214332581,
      "step": 56860
    },
    {
      "epoch": 16.142492194152712,
      "grad_norm": 5.627215385437012,
      "learning_rate": 8.386517172864037e-05,
      "loss": 0.01622956097126007,
      "step": 56870
    },
    {
      "epoch": 16.14533068407607,
      "grad_norm": 4.212461948394775,
      "learning_rate": 8.386233323871701e-05,
      "loss": 0.01635368764400482,
      "step": 56880
    },
    {
      "epoch": 16.14816917399943,
      "grad_norm": 7.619419097900391,
      "learning_rate": 8.385949474879365e-05,
      "loss": 0.025797906517982482,
      "step": 56890
    },
    {
      "epoch": 16.151007663922794,
      "grad_norm": 0.8816103935241699,
      "learning_rate": 8.385665625887028e-05,
      "loss": 0.015209449827671051,
      "step": 56900
    },
    {
      "epoch": 16.153846153846153,
      "grad_norm": 1.5230587720870972,
      "learning_rate": 8.385381776894693e-05,
      "loss": 0.020472793281078337,
      "step": 56910
    },
    {
      "epoch": 16.156684643769516,
      "grad_norm": 14.331717491149902,
      "learning_rate": 8.385097927902357e-05,
      "loss": 0.028105777502059937,
      "step": 56920
    },
    {
      "epoch": 16.159523133692876,
      "grad_norm": 1.133758306503296,
      "learning_rate": 8.38481407891002e-05,
      "loss": 0.01614253968000412,
      "step": 56930
    },
    {
      "epoch": 16.162361623616235,
      "grad_norm": 10.500707626342773,
      "learning_rate": 8.384530229917684e-05,
      "loss": 0.027049621939659117,
      "step": 56940
    },
    {
      "epoch": 16.165200113539598,
      "grad_norm": 9.087316513061523,
      "learning_rate": 8.384246380925348e-05,
      "loss": 0.030893945693969728,
      "step": 56950
    },
    {
      "epoch": 16.168038603462957,
      "grad_norm": 10.016810417175293,
      "learning_rate": 8.383962531933011e-05,
      "loss": 0.019363540410995483,
      "step": 56960
    },
    {
      "epoch": 16.17087709338632,
      "grad_norm": 1.9782463312149048,
      "learning_rate": 8.383678682940675e-05,
      "loss": 0.020191127061843873,
      "step": 56970
    },
    {
      "epoch": 16.17371558330968,
      "grad_norm": 10.08120059967041,
      "learning_rate": 8.383394833948341e-05,
      "loss": 0.033452361822128296,
      "step": 56980
    },
    {
      "epoch": 16.17655407323304,
      "grad_norm": 6.860118865966797,
      "learning_rate": 8.383110984956004e-05,
      "loss": 0.02506357729434967,
      "step": 56990
    },
    {
      "epoch": 16.1793925631564,
      "grad_norm": 14.535609245300293,
      "learning_rate": 8.382827135963668e-05,
      "loss": 0.035907596349716187,
      "step": 57000
    },
    {
      "epoch": 16.1793925631564,
      "eval_accuracy": 0.9640745215234946,
      "eval_loss": 0.11561466753482819,
      "eval_runtime": 31.6189,
      "eval_samples_per_second": 497.392,
      "eval_steps_per_second": 7.78,
      "step": 57000
    },
    {
      "epoch": 16.18223105307976,
      "grad_norm": 3.973383903503418,
      "learning_rate": 8.382543286971332e-05,
      "loss": 0.025158500671386717,
      "step": 57010
    },
    {
      "epoch": 16.185069543003124,
      "grad_norm": 4.1740312576293945,
      "learning_rate": 8.382259437978996e-05,
      "loss": 0.011600501090288162,
      "step": 57020
    },
    {
      "epoch": 16.187908032926483,
      "grad_norm": 2.513554096221924,
      "learning_rate": 8.381975588986659e-05,
      "loss": 0.021694007515907287,
      "step": 57030
    },
    {
      "epoch": 16.190746522849842,
      "grad_norm": 1.7866590023040771,
      "learning_rate": 8.381691739994323e-05,
      "loss": 0.014931215345859528,
      "step": 57040
    },
    {
      "epoch": 16.193585012773205,
      "grad_norm": 2.149397611618042,
      "learning_rate": 8.381407891001988e-05,
      "loss": 0.02089792340993881,
      "step": 57050
    },
    {
      "epoch": 16.196423502696565,
      "grad_norm": 9.424501419067383,
      "learning_rate": 8.381124042009651e-05,
      "loss": 0.02378803789615631,
      "step": 57060
    },
    {
      "epoch": 16.199261992619927,
      "grad_norm": 1.4272844791412354,
      "learning_rate": 8.380840193017315e-05,
      "loss": 0.032920295000076295,
      "step": 57070
    },
    {
      "epoch": 16.202100482543287,
      "grad_norm": 4.679747104644775,
      "learning_rate": 8.38055634402498e-05,
      "loss": 0.031467515230178836,
      "step": 57080
    },
    {
      "epoch": 16.204938972466646,
      "grad_norm": 0.3514966368675232,
      "learning_rate": 8.380272495032642e-05,
      "loss": 0.016771838068962097,
      "step": 57090
    },
    {
      "epoch": 16.20777746239001,
      "grad_norm": 8.899028778076172,
      "learning_rate": 8.379988646040306e-05,
      "loss": 0.03600646853446961,
      "step": 57100
    },
    {
      "epoch": 16.21061595231337,
      "grad_norm": 10.548735618591309,
      "learning_rate": 8.379704797047972e-05,
      "loss": 0.022897185385227205,
      "step": 57110
    },
    {
      "epoch": 16.21345444223673,
      "grad_norm": 6.314229965209961,
      "learning_rate": 8.379420948055635e-05,
      "loss": 0.03035309910774231,
      "step": 57120
    },
    {
      "epoch": 16.21629293216009,
      "grad_norm": 3.9411020278930664,
      "learning_rate": 8.379137099063299e-05,
      "loss": 0.01534806489944458,
      "step": 57130
    },
    {
      "epoch": 16.21913142208345,
      "grad_norm": 1.3754514455795288,
      "learning_rate": 8.378853250070963e-05,
      "loss": 0.017643196880817412,
      "step": 57140
    },
    {
      "epoch": 16.221969912006813,
      "grad_norm": 5.0410614013671875,
      "learning_rate": 8.378569401078627e-05,
      "loss": 0.010111761093139649,
      "step": 57150
    },
    {
      "epoch": 16.224808401930172,
      "grad_norm": 4.362121105194092,
      "learning_rate": 8.37828555208629e-05,
      "loss": 0.03830148577690125,
      "step": 57160
    },
    {
      "epoch": 16.227646891853535,
      "grad_norm": 11.422591209411621,
      "learning_rate": 8.378001703093954e-05,
      "loss": 0.02887851595878601,
      "step": 57170
    },
    {
      "epoch": 16.230485381776894,
      "grad_norm": 1.5751036405563354,
      "learning_rate": 8.37771785410162e-05,
      "loss": 0.04523343145847321,
      "step": 57180
    },
    {
      "epoch": 16.233323871700254,
      "grad_norm": 0.8377181887626648,
      "learning_rate": 8.377434005109282e-05,
      "loss": 0.026569089293479918,
      "step": 57190
    },
    {
      "epoch": 16.236162361623617,
      "grad_norm": 4.400909423828125,
      "learning_rate": 8.377150156116946e-05,
      "loss": 0.016935981810092926,
      "step": 57200
    },
    {
      "epoch": 16.239000851546976,
      "grad_norm": 11.719658851623535,
      "learning_rate": 8.37686630712461e-05,
      "loss": 0.014204981923103332,
      "step": 57210
    },
    {
      "epoch": 16.24183934147034,
      "grad_norm": 4.207752704620361,
      "learning_rate": 8.376582458132273e-05,
      "loss": 0.01809924840927124,
      "step": 57220
    },
    {
      "epoch": 16.244677831393698,
      "grad_norm": 7.823772430419922,
      "learning_rate": 8.376298609139937e-05,
      "loss": 0.018288904428482057,
      "step": 57230
    },
    {
      "epoch": 16.24751632131706,
      "grad_norm": 3.182079553604126,
      "learning_rate": 8.376014760147602e-05,
      "loss": 0.04123339354991913,
      "step": 57240
    },
    {
      "epoch": 16.25035481124042,
      "grad_norm": 4.608572959899902,
      "learning_rate": 8.375730911155266e-05,
      "loss": 0.025735455751419067,
      "step": 57250
    },
    {
      "epoch": 16.25319330116378,
      "grad_norm": 11.234136581420898,
      "learning_rate": 8.37544706216293e-05,
      "loss": 0.029076090455055235,
      "step": 57260
    },
    {
      "epoch": 16.256031791087143,
      "grad_norm": 18.861520767211914,
      "learning_rate": 8.375163213170594e-05,
      "loss": 0.02638309299945831,
      "step": 57270
    },
    {
      "epoch": 16.258870281010502,
      "grad_norm": 6.439869403839111,
      "learning_rate": 8.374879364178258e-05,
      "loss": 0.03639518618583679,
      "step": 57280
    },
    {
      "epoch": 16.261708770933865,
      "grad_norm": 0.38243794441223145,
      "learning_rate": 8.374595515185921e-05,
      "loss": 0.03366603255271912,
      "step": 57290
    },
    {
      "epoch": 16.264547260857224,
      "grad_norm": 3.2586543560028076,
      "learning_rate": 8.374311666193585e-05,
      "loss": 0.018071483075618743,
      "step": 57300
    },
    {
      "epoch": 16.267385750780583,
      "grad_norm": 17.52294921875,
      "learning_rate": 8.37402781720125e-05,
      "loss": 0.022135978937149046,
      "step": 57310
    },
    {
      "epoch": 16.270224240703946,
      "grad_norm": 3.7334818840026855,
      "learning_rate": 8.373743968208913e-05,
      "loss": 0.019068683683872222,
      "step": 57320
    },
    {
      "epoch": 16.273062730627306,
      "grad_norm": 1.8096365928649902,
      "learning_rate": 8.373460119216577e-05,
      "loss": 0.03038056194782257,
      "step": 57330
    },
    {
      "epoch": 16.27590122055067,
      "grad_norm": 1.243525505065918,
      "learning_rate": 8.373176270224242e-05,
      "loss": 0.020485574007034303,
      "step": 57340
    },
    {
      "epoch": 16.278739710474028,
      "grad_norm": 8.22391128540039,
      "learning_rate": 8.372892421231904e-05,
      "loss": 0.03525983989238739,
      "step": 57350
    },
    {
      "epoch": 16.281578200397387,
      "grad_norm": 9.366171836853027,
      "learning_rate": 8.372608572239568e-05,
      "loss": 0.037952476739883424,
      "step": 57360
    },
    {
      "epoch": 16.28441669032075,
      "grad_norm": 3.075228691101074,
      "learning_rate": 8.372324723247233e-05,
      "loss": 0.025078901648521425,
      "step": 57370
    },
    {
      "epoch": 16.28725518024411,
      "grad_norm": 5.465268611907959,
      "learning_rate": 8.372040874254897e-05,
      "loss": 0.036406916379928586,
      "step": 57380
    },
    {
      "epoch": 16.290093670167472,
      "grad_norm": 2.449941873550415,
      "learning_rate": 8.371757025262561e-05,
      "loss": 0.012875911593437196,
      "step": 57390
    },
    {
      "epoch": 16.29293216009083,
      "grad_norm": 7.457828521728516,
      "learning_rate": 8.371473176270225e-05,
      "loss": 0.019717252254486083,
      "step": 57400
    },
    {
      "epoch": 16.29577065001419,
      "grad_norm": 8.481057167053223,
      "learning_rate": 8.371189327277889e-05,
      "loss": 0.015264761447906495,
      "step": 57410
    },
    {
      "epoch": 16.298609139937554,
      "grad_norm": 3.624656915664673,
      "learning_rate": 8.370905478285552e-05,
      "loss": 0.018616043031215668,
      "step": 57420
    },
    {
      "epoch": 16.301447629860913,
      "grad_norm": 5.0557403564453125,
      "learning_rate": 8.370621629293216e-05,
      "loss": 0.020565032958984375,
      "step": 57430
    },
    {
      "epoch": 16.304286119784276,
      "grad_norm": 2.042957067489624,
      "learning_rate": 8.37033778030088e-05,
      "loss": 0.024635624885559083,
      "step": 57440
    },
    {
      "epoch": 16.307124609707635,
      "grad_norm": 7.351503372192383,
      "learning_rate": 8.370053931308544e-05,
      "loss": 0.037080830335617064,
      "step": 57450
    },
    {
      "epoch": 16.309963099630995,
      "grad_norm": 10.515963554382324,
      "learning_rate": 8.369770082316208e-05,
      "loss": 0.025804191827774048,
      "step": 57460
    },
    {
      "epoch": 16.312801589554358,
      "grad_norm": 3.777642250061035,
      "learning_rate": 8.369486233323873e-05,
      "loss": 0.011177933216094971,
      "step": 57470
    },
    {
      "epoch": 16.315640079477717,
      "grad_norm": 6.366386413574219,
      "learning_rate": 8.369202384331535e-05,
      "loss": 0.031482675671577455,
      "step": 57480
    },
    {
      "epoch": 16.31847856940108,
      "grad_norm": 1.4926985502243042,
      "learning_rate": 8.3689185353392e-05,
      "loss": 0.018231646716594697,
      "step": 57490
    },
    {
      "epoch": 16.32131705932444,
      "grad_norm": 5.067385196685791,
      "learning_rate": 8.368634686346864e-05,
      "loss": 0.016429007053375244,
      "step": 57500
    },
    {
      "epoch": 16.32131705932444,
      "eval_accuracy": 0.9640745215234946,
      "eval_loss": 0.11266815662384033,
      "eval_runtime": 31.5059,
      "eval_samples_per_second": 499.176,
      "eval_steps_per_second": 7.808,
      "step": 57500
    },
    {
      "epoch": 16.3241555492478,
      "grad_norm": 3.4043490886688232,
      "learning_rate": 8.368350837354528e-05,
      "loss": 0.03504315614700317,
      "step": 57510
    },
    {
      "epoch": 16.32699403917116,
      "grad_norm": 5.386589527130127,
      "learning_rate": 8.368066988362192e-05,
      "loss": 0.013249245285987855,
      "step": 57520
    },
    {
      "epoch": 16.32983252909452,
      "grad_norm": 2.719717025756836,
      "learning_rate": 8.367783139369856e-05,
      "loss": 0.016923144459724426,
      "step": 57530
    },
    {
      "epoch": 16.332671019017884,
      "grad_norm": 1.1243789196014404,
      "learning_rate": 8.36749929037752e-05,
      "loss": 0.021266916394233705,
      "step": 57540
    },
    {
      "epoch": 16.335509508941243,
      "grad_norm": 3.2733981609344482,
      "learning_rate": 8.367215441385183e-05,
      "loss": 0.029911166429519652,
      "step": 57550
    },
    {
      "epoch": 16.338347998864602,
      "grad_norm": 3.686540126800537,
      "learning_rate": 8.366931592392847e-05,
      "loss": 0.01913428455591202,
      "step": 57560
    },
    {
      "epoch": 16.341186488787965,
      "grad_norm": 8.905865669250488,
      "learning_rate": 8.366647743400511e-05,
      "loss": 0.019753411412239075,
      "step": 57570
    },
    {
      "epoch": 16.344024978711325,
      "grad_norm": 2.0028085708618164,
      "learning_rate": 8.366363894408175e-05,
      "loss": 0.014767123758792878,
      "step": 57580
    },
    {
      "epoch": 16.346863468634687,
      "grad_norm": 8.34296989440918,
      "learning_rate": 8.36608004541584e-05,
      "loss": 0.01722279340028763,
      "step": 57590
    },
    {
      "epoch": 16.349701958558047,
      "grad_norm": 1.5266138315200806,
      "learning_rate": 8.365796196423504e-05,
      "loss": 0.02117514908313751,
      "step": 57600
    },
    {
      "epoch": 16.352540448481406,
      "grad_norm": 6.234114170074463,
      "learning_rate": 8.365512347431166e-05,
      "loss": 0.02072383016347885,
      "step": 57610
    },
    {
      "epoch": 16.35537893840477,
      "grad_norm": 7.617072582244873,
      "learning_rate": 8.36522849843883e-05,
      "loss": 0.02148745656013489,
      "step": 57620
    },
    {
      "epoch": 16.35821742832813,
      "grad_norm": 9.410679817199707,
      "learning_rate": 8.364944649446495e-05,
      "loss": 0.019561731815338136,
      "step": 57630
    },
    {
      "epoch": 16.36105591825149,
      "grad_norm": 3.7705769538879395,
      "learning_rate": 8.364660800454159e-05,
      "loss": 0.01389913707971573,
      "step": 57640
    },
    {
      "epoch": 16.36389440817485,
      "grad_norm": 4.646562576293945,
      "learning_rate": 8.364376951461823e-05,
      "loss": 0.02539682686328888,
      "step": 57650
    },
    {
      "epoch": 16.366732898098213,
      "grad_norm": 4.98703670501709,
      "learning_rate": 8.364093102469487e-05,
      "loss": 0.01383742094039917,
      "step": 57660
    },
    {
      "epoch": 16.369571388021573,
      "grad_norm": 7.859958648681641,
      "learning_rate": 8.36380925347715e-05,
      "loss": 0.04223350584506989,
      "step": 57670
    },
    {
      "epoch": 16.372409877944932,
      "grad_norm": 0.5775028467178345,
      "learning_rate": 8.363525404484814e-05,
      "loss": 0.03109874129295349,
      "step": 57680
    },
    {
      "epoch": 16.375248367868295,
      "grad_norm": 8.082518577575684,
      "learning_rate": 8.363241555492478e-05,
      "loss": 0.024701231718063356,
      "step": 57690
    },
    {
      "epoch": 16.378086857791654,
      "grad_norm": 3.0618207454681396,
      "learning_rate": 8.362957706500142e-05,
      "loss": 0.029514485597610475,
      "step": 57700
    },
    {
      "epoch": 16.380925347715017,
      "grad_norm": 3.8841583728790283,
      "learning_rate": 8.362673857507807e-05,
      "loss": 0.03482574224472046,
      "step": 57710
    },
    {
      "epoch": 16.383763837638377,
      "grad_norm": 7.379438877105713,
      "learning_rate": 8.36239000851547e-05,
      "loss": 0.021716126799583436,
      "step": 57720
    },
    {
      "epoch": 16.386602327561736,
      "grad_norm": 7.598274230957031,
      "learning_rate": 8.362106159523135e-05,
      "loss": 0.015579144656658172,
      "step": 57730
    },
    {
      "epoch": 16.3894408174851,
      "grad_norm": 1.0270344018936157,
      "learning_rate": 8.361822310530798e-05,
      "loss": 0.015401656925678252,
      "step": 57740
    },
    {
      "epoch": 16.392279307408458,
      "grad_norm": 3.3055825233459473,
      "learning_rate": 8.361538461538462e-05,
      "loss": 0.014878253638744354,
      "step": 57750
    },
    {
      "epoch": 16.39511779733182,
      "grad_norm": 0.4192140996456146,
      "learning_rate": 8.361254612546126e-05,
      "loss": 0.028828465938568117,
      "step": 57760
    },
    {
      "epoch": 16.39795628725518,
      "grad_norm": 1.9138195514678955,
      "learning_rate": 8.360970763553789e-05,
      "loss": 0.0379542350769043,
      "step": 57770
    },
    {
      "epoch": 16.40079477717854,
      "grad_norm": 8.491437911987305,
      "learning_rate": 8.360686914561454e-05,
      "loss": 0.03638689815998077,
      "step": 57780
    },
    {
      "epoch": 16.403633267101903,
      "grad_norm": 5.084245681762695,
      "learning_rate": 8.360403065569118e-05,
      "loss": 0.01824689209461212,
      "step": 57790
    },
    {
      "epoch": 16.406471757025262,
      "grad_norm": 12.565242767333984,
      "learning_rate": 8.360119216576781e-05,
      "loss": 0.017689278721809386,
      "step": 57800
    },
    {
      "epoch": 16.409310246948625,
      "grad_norm": 2.6246073246002197,
      "learning_rate": 8.359835367584445e-05,
      "loss": 0.013861703872680663,
      "step": 57810
    },
    {
      "epoch": 16.412148736871984,
      "grad_norm": 2.0474650859832764,
      "learning_rate": 8.359551518592109e-05,
      "loss": 0.024439796805381775,
      "step": 57820
    },
    {
      "epoch": 16.414987226795343,
      "grad_norm": 0.5992058515548706,
      "learning_rate": 8.359267669599773e-05,
      "loss": 0.009769780933856964,
      "step": 57830
    },
    {
      "epoch": 16.417825716718706,
      "grad_norm": 0.8119294047355652,
      "learning_rate": 8.358983820607438e-05,
      "loss": 0.016229116916656496,
      "step": 57840
    },
    {
      "epoch": 16.420664206642066,
      "grad_norm": 2.0341272354125977,
      "learning_rate": 8.358699971615102e-05,
      "loss": 0.021486197412014008,
      "step": 57850
    },
    {
      "epoch": 16.42350269656543,
      "grad_norm": 9.836889266967773,
      "learning_rate": 8.358416122622766e-05,
      "loss": 0.02454804927110672,
      "step": 57860
    },
    {
      "epoch": 16.426341186488788,
      "grad_norm": 0.10809176415205002,
      "learning_rate": 8.358132273630429e-05,
      "loss": 0.029806384444236757,
      "step": 57870
    },
    {
      "epoch": 16.429179676412147,
      "grad_norm": 2.020871639251709,
      "learning_rate": 8.357848424638093e-05,
      "loss": 0.03331070840358734,
      "step": 57880
    },
    {
      "epoch": 16.43201816633551,
      "grad_norm": 3.713773250579834,
      "learning_rate": 8.357564575645757e-05,
      "loss": 0.03311811685562134,
      "step": 57890
    },
    {
      "epoch": 16.43485665625887,
      "grad_norm": 7.971744537353516,
      "learning_rate": 8.35728072665342e-05,
      "loss": 0.023117053508758544,
      "step": 57900
    },
    {
      "epoch": 16.437695146182232,
      "grad_norm": 12.512771606445312,
      "learning_rate": 8.356996877661085e-05,
      "loss": 0.015941639244556428,
      "step": 57910
    },
    {
      "epoch": 16.44053363610559,
      "grad_norm": 3.032975435256958,
      "learning_rate": 8.35671302866875e-05,
      "loss": 0.026855844259262084,
      "step": 57920
    },
    {
      "epoch": 16.44337212602895,
      "grad_norm": 4.288434028625488,
      "learning_rate": 8.356429179676412e-05,
      "loss": 0.009644306451082229,
      "step": 57930
    },
    {
      "epoch": 16.446210615952314,
      "grad_norm": 7.594199180603027,
      "learning_rate": 8.356145330684076e-05,
      "loss": 0.021636776626110077,
      "step": 57940
    },
    {
      "epoch": 16.449049105875673,
      "grad_norm": 0.735028088092804,
      "learning_rate": 8.35586148169174e-05,
      "loss": 0.04598793387413025,
      "step": 57950
    },
    {
      "epoch": 16.451887595799036,
      "grad_norm": 5.356010913848877,
      "learning_rate": 8.355577632699405e-05,
      "loss": 0.026205438375473022,
      "step": 57960
    },
    {
      "epoch": 16.454726085722395,
      "grad_norm": 0.9242346286773682,
      "learning_rate": 8.355293783707067e-05,
      "loss": 0.023983070254325868,
      "step": 57970
    },
    {
      "epoch": 16.457564575645755,
      "grad_norm": 10.547995567321777,
      "learning_rate": 8.355009934714733e-05,
      "loss": 0.02581058442592621,
      "step": 57980
    },
    {
      "epoch": 16.460403065569118,
      "grad_norm": 17.382551193237305,
      "learning_rate": 8.354726085722397e-05,
      "loss": 0.02995608150959015,
      "step": 57990
    },
    {
      "epoch": 16.463241555492477,
      "grad_norm": 10.663006782531738,
      "learning_rate": 8.35444223673006e-05,
      "loss": 0.039972490072250365,
      "step": 58000
    },
    {
      "epoch": 16.463241555492477,
      "eval_accuracy": 0.956698671075221,
      "eval_loss": 0.1387191116809845,
      "eval_runtime": 32.4613,
      "eval_samples_per_second": 484.484,
      "eval_steps_per_second": 7.578,
      "step": 58000
    },
    {
      "epoch": 16.46608004541584,
      "grad_norm": 2.595625877380371,
      "learning_rate": 8.354158387737724e-05,
      "loss": 0.026548388600349426,
      "step": 58010
    },
    {
      "epoch": 16.4689185353392,
      "grad_norm": 3.061826705932617,
      "learning_rate": 8.353874538745388e-05,
      "loss": 0.028435802459716795,
      "step": 58020
    },
    {
      "epoch": 16.471757025262562,
      "grad_norm": 3.808105945587158,
      "learning_rate": 8.353590689753051e-05,
      "loss": 0.021903225779533388,
      "step": 58030
    },
    {
      "epoch": 16.47459551518592,
      "grad_norm": 14.7695951461792,
      "learning_rate": 8.353306840760716e-05,
      "loss": 0.03147459626197815,
      "step": 58040
    },
    {
      "epoch": 16.47743400510928,
      "grad_norm": 0.28551092743873596,
      "learning_rate": 8.35302299176838e-05,
      "loss": 0.028281348943710326,
      "step": 58050
    },
    {
      "epoch": 16.480272495032644,
      "grad_norm": 10.403939247131348,
      "learning_rate": 8.352739142776043e-05,
      "loss": 0.037682768702507016,
      "step": 58060
    },
    {
      "epoch": 16.483110984956003,
      "grad_norm": 8.957832336425781,
      "learning_rate": 8.352455293783707e-05,
      "loss": 0.024968945980072023,
      "step": 58070
    },
    {
      "epoch": 16.485949474879366,
      "grad_norm": 11.300315856933594,
      "learning_rate": 8.352171444791371e-05,
      "loss": 0.018435643613338472,
      "step": 58080
    },
    {
      "epoch": 16.488787964802725,
      "grad_norm": 9.459074020385742,
      "learning_rate": 8.351887595799036e-05,
      "loss": 0.030626952648162842,
      "step": 58090
    },
    {
      "epoch": 16.491626454726084,
      "grad_norm": 1.922043800354004,
      "learning_rate": 8.351603746806698e-05,
      "loss": 0.007774235308170318,
      "step": 58100
    },
    {
      "epoch": 16.494464944649447,
      "grad_norm": 4.1167731285095215,
      "learning_rate": 8.351319897814364e-05,
      "loss": 0.018003752827644347,
      "step": 58110
    },
    {
      "epoch": 16.497303434572807,
      "grad_norm": 7.29007625579834,
      "learning_rate": 8.351036048822028e-05,
      "loss": 0.02688470184803009,
      "step": 58120
    },
    {
      "epoch": 16.50014192449617,
      "grad_norm": 4.435036659240723,
      "learning_rate": 8.350752199829691e-05,
      "loss": 0.013380326330661774,
      "step": 58130
    },
    {
      "epoch": 16.50298041441953,
      "grad_norm": 0.9292892217636108,
      "learning_rate": 8.350468350837355e-05,
      "loss": 0.022972561419010162,
      "step": 58140
    },
    {
      "epoch": 16.50581890434289,
      "grad_norm": 6.01042366027832,
      "learning_rate": 8.350184501845019e-05,
      "loss": 0.0166106715798378,
      "step": 58150
    },
    {
      "epoch": 16.50865739426625,
      "grad_norm": 13.110054016113281,
      "learning_rate": 8.349900652852682e-05,
      "loss": 0.03879846334457397,
      "step": 58160
    },
    {
      "epoch": 16.51149588418961,
      "grad_norm": 0.42446956038475037,
      "learning_rate": 8.349616803860346e-05,
      "loss": 0.021194837987422943,
      "step": 58170
    },
    {
      "epoch": 16.514334374112973,
      "grad_norm": 6.163670063018799,
      "learning_rate": 8.349332954868011e-05,
      "loss": 0.03990270495414734,
      "step": 58180
    },
    {
      "epoch": 16.517172864036333,
      "grad_norm": 1.6155263185501099,
      "learning_rate": 8.349049105875674e-05,
      "loss": 0.023068951070308687,
      "step": 58190
    },
    {
      "epoch": 16.520011353959692,
      "grad_norm": 2.1065757274627686,
      "learning_rate": 8.348765256883338e-05,
      "loss": 0.03105611801147461,
      "step": 58200
    },
    {
      "epoch": 16.522849843883055,
      "grad_norm": 6.374539375305176,
      "learning_rate": 8.348481407891003e-05,
      "loss": 0.023350685834884644,
      "step": 58210
    },
    {
      "epoch": 16.525688333806414,
      "grad_norm": 8.94660472869873,
      "learning_rate": 8.348197558898667e-05,
      "loss": 0.026402127742767335,
      "step": 58220
    },
    {
      "epoch": 16.528526823729777,
      "grad_norm": 0.7009008526802063,
      "learning_rate": 8.34791370990633e-05,
      "loss": 0.02437014877796173,
      "step": 58230
    },
    {
      "epoch": 16.531365313653136,
      "grad_norm": 2.0632848739624023,
      "learning_rate": 8.347629860913995e-05,
      "loss": 0.015159733593463898,
      "step": 58240
    },
    {
      "epoch": 16.534203803576496,
      "grad_norm": 6.802942752838135,
      "learning_rate": 8.347346011921659e-05,
      "loss": 0.025207504630088806,
      "step": 58250
    },
    {
      "epoch": 16.53704229349986,
      "grad_norm": 2.2284159660339355,
      "learning_rate": 8.347062162929322e-05,
      "loss": 0.028232869505882264,
      "step": 58260
    },
    {
      "epoch": 16.539880783423218,
      "grad_norm": 6.927571773529053,
      "learning_rate": 8.346778313936986e-05,
      "loss": 0.011921193450689316,
      "step": 58270
    },
    {
      "epoch": 16.54271927334658,
      "grad_norm": 10.382787704467773,
      "learning_rate": 8.34649446494465e-05,
      "loss": 0.021317094564437866,
      "step": 58280
    },
    {
      "epoch": 16.54555776326994,
      "grad_norm": 13.224753379821777,
      "learning_rate": 8.346210615952313e-05,
      "loss": 0.041150495409965515,
      "step": 58290
    },
    {
      "epoch": 16.5483962531933,
      "grad_norm": 8.97059154510498,
      "learning_rate": 8.345926766959977e-05,
      "loss": 0.021298563480377196,
      "step": 58300
    },
    {
      "epoch": 16.551234743116662,
      "grad_norm": 3.087956666946411,
      "learning_rate": 8.345642917967643e-05,
      "loss": 0.017362378537654877,
      "step": 58310
    },
    {
      "epoch": 16.55407323304002,
      "grad_norm": 0.6908025145530701,
      "learning_rate": 8.345359068975305e-05,
      "loss": 0.010384636372327805,
      "step": 58320
    },
    {
      "epoch": 16.556911722963385,
      "grad_norm": 11.758752822875977,
      "learning_rate": 8.34507521998297e-05,
      "loss": 0.014615438878536224,
      "step": 58330
    },
    {
      "epoch": 16.559750212886744,
      "grad_norm": 7.739632606506348,
      "learning_rate": 8.344791370990634e-05,
      "loss": 0.02874714732170105,
      "step": 58340
    },
    {
      "epoch": 16.562588702810103,
      "grad_norm": 8.88198471069336,
      "learning_rate": 8.344507521998298e-05,
      "loss": 0.021249204874038696,
      "step": 58350
    },
    {
      "epoch": 16.565427192733466,
      "grad_norm": 1.5235759019851685,
      "learning_rate": 8.34422367300596e-05,
      "loss": 0.029982563853263856,
      "step": 58360
    },
    {
      "epoch": 16.568265682656826,
      "grad_norm": 0.500368058681488,
      "learning_rate": 8.343939824013625e-05,
      "loss": 0.020798631012439728,
      "step": 58370
    },
    {
      "epoch": 16.57110417258019,
      "grad_norm": 0.6884090304374695,
      "learning_rate": 8.343655975021289e-05,
      "loss": 0.013786274194717407,
      "step": 58380
    },
    {
      "epoch": 16.573942662503548,
      "grad_norm": 7.296133995056152,
      "learning_rate": 8.343372126028953e-05,
      "loss": 0.025480160117149354,
      "step": 58390
    },
    {
      "epoch": 16.57678115242691,
      "grad_norm": 5.508730888366699,
      "learning_rate": 8.343088277036617e-05,
      "loss": 0.018541747331619264,
      "step": 58400
    },
    {
      "epoch": 16.57961964235027,
      "grad_norm": 10.622841835021973,
      "learning_rate": 8.342804428044281e-05,
      "loss": 0.013938701152801514,
      "step": 58410
    },
    {
      "epoch": 16.58245813227363,
      "grad_norm": 7.3429999351501465,
      "learning_rate": 8.342520579051944e-05,
      "loss": 0.033525124192237854,
      "step": 58420
    },
    {
      "epoch": 16.585296622196992,
      "grad_norm": 1.1720824241638184,
      "learning_rate": 8.342236730059608e-05,
      "loss": 0.026029011607170104,
      "step": 58430
    },
    {
      "epoch": 16.58813511212035,
      "grad_norm": 1.4224919080734253,
      "learning_rate": 8.341952881067274e-05,
      "loss": 0.03980385959148407,
      "step": 58440
    },
    {
      "epoch": 16.590973602043714,
      "grad_norm": 7.893923282623291,
      "learning_rate": 8.341669032074936e-05,
      "loss": 0.047191083431243896,
      "step": 58450
    },
    {
      "epoch": 16.593812091967074,
      "grad_norm": 1.7177177667617798,
      "learning_rate": 8.3413851830826e-05,
      "loss": 0.01860113739967346,
      "step": 58460
    },
    {
      "epoch": 16.596650581890433,
      "grad_norm": 1.3261648416519165,
      "learning_rate": 8.341101334090265e-05,
      "loss": 0.027608656883239747,
      "step": 58470
    },
    {
      "epoch": 16.599489071813796,
      "grad_norm": 11.521811485290527,
      "learning_rate": 8.340845869997161e-05,
      "loss": 0.03972205221652984,
      "step": 58480
    },
    {
      "epoch": 16.602327561737155,
      "grad_norm": 8.218220710754395,
      "learning_rate": 8.340562021004827e-05,
      "loss": 0.02715291976928711,
      "step": 58490
    },
    {
      "epoch": 16.605166051660518,
      "grad_norm": 6.369820594787598,
      "learning_rate": 8.34027817201249e-05,
      "loss": 0.030585858225822448,
      "step": 58500
    },
    {
      "epoch": 16.605166051660518,
      "eval_accuracy": 0.9649647103707001,
      "eval_loss": 0.11148511618375778,
      "eval_runtime": 31.7598,
      "eval_samples_per_second": 495.186,
      "eval_steps_per_second": 7.746,
      "step": 58500
    },
    {
      "epoch": 16.608004541583878,
      "grad_norm": 5.023155212402344,
      "learning_rate": 8.339994323020154e-05,
      "loss": 0.009741977602243424,
      "step": 58510
    },
    {
      "epoch": 16.610843031507237,
      "grad_norm": 11.585160255432129,
      "learning_rate": 8.339710474027818e-05,
      "loss": 0.02634880244731903,
      "step": 58520
    },
    {
      "epoch": 16.6136815214306,
      "grad_norm": 6.228874683380127,
      "learning_rate": 8.33942662503548e-05,
      "loss": 0.01261507123708725,
      "step": 58530
    },
    {
      "epoch": 16.61652001135396,
      "grad_norm": 0.7988203167915344,
      "learning_rate": 8.339142776043145e-05,
      "loss": 0.0111713707447052,
      "step": 58540
    },
    {
      "epoch": 16.619358501277322,
      "grad_norm": 11.321148872375488,
      "learning_rate": 8.338858927050809e-05,
      "loss": 0.0320526659488678,
      "step": 58550
    },
    {
      "epoch": 16.62219699120068,
      "grad_norm": 9.522676467895508,
      "learning_rate": 8.338575078058473e-05,
      "loss": 0.015048983693122863,
      "step": 58560
    },
    {
      "epoch": 16.62503548112404,
      "grad_norm": 17.335908889770508,
      "learning_rate": 8.338291229066137e-05,
      "loss": 0.028058427572250366,
      "step": 58570
    },
    {
      "epoch": 16.627873971047404,
      "grad_norm": 2.0620009899139404,
      "learning_rate": 8.338007380073801e-05,
      "loss": 0.032004249095916745,
      "step": 58580
    },
    {
      "epoch": 16.630712460970763,
      "grad_norm": 4.046170711517334,
      "learning_rate": 8.337723531081465e-05,
      "loss": 0.0275095134973526,
      "step": 58590
    },
    {
      "epoch": 16.633550950894126,
      "grad_norm": 6.5045061111450195,
      "learning_rate": 8.337439682089128e-05,
      "loss": 0.03998938202857971,
      "step": 58600
    },
    {
      "epoch": 16.636389440817485,
      "grad_norm": 1.0320584774017334,
      "learning_rate": 8.337155833096792e-05,
      "loss": 0.029005658626556397,
      "step": 58610
    },
    {
      "epoch": 16.639227930740844,
      "grad_norm": 8.144086837768555,
      "learning_rate": 8.336871984104458e-05,
      "loss": 0.03329362571239471,
      "step": 58620
    },
    {
      "epoch": 16.642066420664207,
      "grad_norm": 1.1369993686676025,
      "learning_rate": 8.336588135112121e-05,
      "loss": 0.007962089776992799,
      "step": 58630
    },
    {
      "epoch": 16.644904910587567,
      "grad_norm": 9.268597602844238,
      "learning_rate": 8.336304286119785e-05,
      "loss": 0.028556400537490846,
      "step": 58640
    },
    {
      "epoch": 16.64774340051093,
      "grad_norm": 8.676448822021484,
      "learning_rate": 8.336020437127449e-05,
      "loss": 0.01715867519378662,
      "step": 58650
    },
    {
      "epoch": 16.65058189043429,
      "grad_norm": 7.841867446899414,
      "learning_rate": 8.335736588135112e-05,
      "loss": 0.021545688807964324,
      "step": 58660
    },
    {
      "epoch": 16.653420380357648,
      "grad_norm": 0.9473310112953186,
      "learning_rate": 8.335452739142776e-05,
      "loss": 0.011715412139892578,
      "step": 58670
    },
    {
      "epoch": 16.65625887028101,
      "grad_norm": 1.8944321870803833,
      "learning_rate": 8.33516889015044e-05,
      "loss": 0.032149726152420045,
      "step": 58680
    },
    {
      "epoch": 16.65909736020437,
      "grad_norm": 1.2174142599105835,
      "learning_rate": 8.334885041158104e-05,
      "loss": 0.02364833503961563,
      "step": 58690
    },
    {
      "epoch": 16.661935850127733,
      "grad_norm": 0.9931562542915344,
      "learning_rate": 8.334601192165768e-05,
      "loss": 0.020965445041656493,
      "step": 58700
    },
    {
      "epoch": 16.664774340051093,
      "grad_norm": 2.5448696613311768,
      "learning_rate": 8.334317343173432e-05,
      "loss": 0.019027411937713623,
      "step": 58710
    },
    {
      "epoch": 16.667612829974452,
      "grad_norm": 1.6934963464736938,
      "learning_rate": 8.334033494181097e-05,
      "loss": 0.02984304428100586,
      "step": 58720
    },
    {
      "epoch": 16.670451319897815,
      "grad_norm": 7.586284637451172,
      "learning_rate": 8.33374964518876e-05,
      "loss": 0.021623696386814117,
      "step": 58730
    },
    {
      "epoch": 16.673289809821174,
      "grad_norm": 4.527450084686279,
      "learning_rate": 8.333465796196423e-05,
      "loss": 0.026447987556457518,
      "step": 58740
    },
    {
      "epoch": 16.676128299744537,
      "grad_norm": 7.088357925415039,
      "learning_rate": 8.333181947204088e-05,
      "loss": 0.027219468355178834,
      "step": 58750
    },
    {
      "epoch": 16.678966789667896,
      "grad_norm": 7.283918380737305,
      "learning_rate": 8.332898098211752e-05,
      "loss": 0.017597493529319764,
      "step": 58760
    },
    {
      "epoch": 16.68180527959126,
      "grad_norm": 0.988620400428772,
      "learning_rate": 8.332614249219416e-05,
      "loss": 0.04683432281017304,
      "step": 58770
    },
    {
      "epoch": 16.68464376951462,
      "grad_norm": 4.270298957824707,
      "learning_rate": 8.33233040022708e-05,
      "loss": 0.01735730916261673,
      "step": 58780
    },
    {
      "epoch": 16.687482259437978,
      "grad_norm": 4.433078289031982,
      "learning_rate": 8.332046551234743e-05,
      "loss": 0.018961824476718903,
      "step": 58790
    },
    {
      "epoch": 16.69032074936134,
      "grad_norm": 7.815520763397217,
      "learning_rate": 8.331762702242407e-05,
      "loss": 0.020434993505477905,
      "step": 58800
    },
    {
      "epoch": 16.6931592392847,
      "grad_norm": 9.323116302490234,
      "learning_rate": 8.331478853250071e-05,
      "loss": 0.03925870954990387,
      "step": 58810
    },
    {
      "epoch": 16.695997729208063,
      "grad_norm": 6.909570693969727,
      "learning_rate": 8.331195004257735e-05,
      "loss": 0.027740225195884705,
      "step": 58820
    },
    {
      "epoch": 16.698836219131422,
      "grad_norm": 7.167376518249512,
      "learning_rate": 8.3309111552654e-05,
      "loss": 0.028341460227966308,
      "step": 58830
    },
    {
      "epoch": 16.70167470905478,
      "grad_norm": 6.4705705642700195,
      "learning_rate": 8.330627306273064e-05,
      "loss": 0.018236668407917024,
      "step": 58840
    },
    {
      "epoch": 16.704513198978145,
      "grad_norm": 4.500127792358398,
      "learning_rate": 8.330343457280728e-05,
      "loss": 0.020041263103485106,
      "step": 58850
    },
    {
      "epoch": 16.707351688901504,
      "grad_norm": 11.18964672088623,
      "learning_rate": 8.33005960828839e-05,
      "loss": 0.022896376252174378,
      "step": 58860
    },
    {
      "epoch": 16.710190178824867,
      "grad_norm": 5.560362339019775,
      "learning_rate": 8.329775759296055e-05,
      "loss": 0.012195299565792083,
      "step": 58870
    },
    {
      "epoch": 16.713028668748226,
      "grad_norm": 10.360980033874512,
      "learning_rate": 8.329491910303719e-05,
      "loss": 0.04283861517906189,
      "step": 58880
    },
    {
      "epoch": 16.715867158671585,
      "grad_norm": 3.417398691177368,
      "learning_rate": 8.329208061311383e-05,
      "loss": 0.028645554184913637,
      "step": 58890
    },
    {
      "epoch": 16.71870564859495,
      "grad_norm": 9.850347518920898,
      "learning_rate": 8.328924212319047e-05,
      "loss": 0.01909451186656952,
      "step": 58900
    },
    {
      "epoch": 16.721544138518308,
      "grad_norm": 1.0660990476608276,
      "learning_rate": 8.328640363326711e-05,
      "loss": 0.016973789036273956,
      "step": 58910
    },
    {
      "epoch": 16.72438262844167,
      "grad_norm": 2.129889965057373,
      "learning_rate": 8.328356514334374e-05,
      "loss": 0.014286068081855775,
      "step": 58920
    },
    {
      "epoch": 16.72722111836503,
      "grad_norm": 16.273263931274414,
      "learning_rate": 8.328072665342038e-05,
      "loss": 0.0229129821062088,
      "step": 58930
    },
    {
      "epoch": 16.73005960828839,
      "grad_norm": 2.727794885635376,
      "learning_rate": 8.327788816349702e-05,
      "loss": 0.02416025400161743,
      "step": 58940
    },
    {
      "epoch": 16.732898098211752,
      "grad_norm": 1.6632486581802368,
      "learning_rate": 8.327504967357366e-05,
      "loss": 0.03493282496929169,
      "step": 58950
    },
    {
      "epoch": 16.73573658813511,
      "grad_norm": 0.533372700214386,
      "learning_rate": 8.32722111836503e-05,
      "loss": 0.021981509029865266,
      "step": 58960
    },
    {
      "epoch": 16.738575078058474,
      "grad_norm": 3.5226762294769287,
      "learning_rate": 8.326937269372695e-05,
      "loss": 0.046072971820831296,
      "step": 58970
    },
    {
      "epoch": 16.741413567981834,
      "grad_norm": 5.532339572906494,
      "learning_rate": 8.326653420380359e-05,
      "loss": 0.02512103021144867,
      "step": 58980
    },
    {
      "epoch": 16.744252057905193,
      "grad_norm": 2.8009884357452393,
      "learning_rate": 8.326369571388022e-05,
      "loss": 0.01772029250860214,
      "step": 58990
    },
    {
      "epoch": 16.747090547828556,
      "grad_norm": 2.3903415203094482,
      "learning_rate": 8.326085722395686e-05,
      "loss": 0.012354165315628052,
      "step": 59000
    },
    {
      "epoch": 16.747090547828556,
      "eval_accuracy": 0.9645832008647549,
      "eval_loss": 0.10953931510448456,
      "eval_runtime": 31.5988,
      "eval_samples_per_second": 497.709,
      "eval_steps_per_second": 7.785,
      "step": 59000
    },
    {
      "epoch": 16.749929037751915,
      "grad_norm": 11.503759384155273,
      "learning_rate": 8.32580187340335e-05,
      "loss": 0.04351683557033539,
      "step": 59010
    },
    {
      "epoch": 16.752767527675278,
      "grad_norm": 3.0906810760498047,
      "learning_rate": 8.325518024411014e-05,
      "loss": 0.023315902054309844,
      "step": 59020
    },
    {
      "epoch": 16.755606017598637,
      "grad_norm": 6.008513927459717,
      "learning_rate": 8.325234175418678e-05,
      "loss": 0.01908901333808899,
      "step": 59030
    },
    {
      "epoch": 16.758444507521997,
      "grad_norm": 3.565109968185425,
      "learning_rate": 8.324950326426342e-05,
      "loss": 0.02006048709154129,
      "step": 59040
    },
    {
      "epoch": 16.76128299744536,
      "grad_norm": 2.8562307357788086,
      "learning_rate": 8.324666477434005e-05,
      "loss": 0.015707996487617493,
      "step": 59050
    },
    {
      "epoch": 16.76412148736872,
      "grad_norm": 2.4045372009277344,
      "learning_rate": 8.324382628441669e-05,
      "loss": 0.02519112527370453,
      "step": 59060
    },
    {
      "epoch": 16.766959977292082,
      "grad_norm": 14.199237823486328,
      "learning_rate": 8.324098779449333e-05,
      "loss": 0.0317169189453125,
      "step": 59070
    },
    {
      "epoch": 16.76979846721544,
      "grad_norm": 2.1875901222229004,
      "learning_rate": 8.323814930456997e-05,
      "loss": 0.04812244176864624,
      "step": 59080
    },
    {
      "epoch": 16.7726369571388,
      "grad_norm": 18.521930694580078,
      "learning_rate": 8.323531081464662e-05,
      "loss": 0.0313480406999588,
      "step": 59090
    },
    {
      "epoch": 16.775475447062163,
      "grad_norm": 8.342741966247559,
      "learning_rate": 8.323247232472326e-05,
      "loss": 0.02497076541185379,
      "step": 59100
    },
    {
      "epoch": 16.778313936985523,
      "grad_norm": 4.784858703613281,
      "learning_rate": 8.32296338347999e-05,
      "loss": 0.04681958556175232,
      "step": 59110
    },
    {
      "epoch": 16.781152426908886,
      "grad_norm": 0.9078540802001953,
      "learning_rate": 8.322679534487653e-05,
      "loss": 0.01849210411310196,
      "step": 59120
    },
    {
      "epoch": 16.783990916832245,
      "grad_norm": 5.355259895324707,
      "learning_rate": 8.322395685495317e-05,
      "loss": 0.025255858898162842,
      "step": 59130
    },
    {
      "epoch": 16.786829406755604,
      "grad_norm": 1.1002379655838013,
      "learning_rate": 8.322111836502981e-05,
      "loss": 0.023817498981952668,
      "step": 59140
    },
    {
      "epoch": 16.789667896678967,
      "grad_norm": 3.684565544128418,
      "learning_rate": 8.321827987510644e-05,
      "loss": 0.02081771641969681,
      "step": 59150
    },
    {
      "epoch": 16.792506386602327,
      "grad_norm": 3.4217519760131836,
      "learning_rate": 8.321544138518309e-05,
      "loss": 0.0380407840013504,
      "step": 59160
    },
    {
      "epoch": 16.79534487652569,
      "grad_norm": 4.720285415649414,
      "learning_rate": 8.321260289525973e-05,
      "loss": 0.028555679321289062,
      "step": 59170
    },
    {
      "epoch": 16.79818336644905,
      "grad_norm": 0.4961705207824707,
      "learning_rate": 8.320976440533636e-05,
      "loss": 0.030113762617111205,
      "step": 59180
    },
    {
      "epoch": 16.801021856372408,
      "grad_norm": 0.9919694662094116,
      "learning_rate": 8.3206925915413e-05,
      "loss": 0.030449867248535156,
      "step": 59190
    },
    {
      "epoch": 16.80386034629577,
      "grad_norm": 1.8493183851242065,
      "learning_rate": 8.320408742548964e-05,
      "loss": 0.015259429812431335,
      "step": 59200
    },
    {
      "epoch": 16.80669883621913,
      "grad_norm": 1.274405598640442,
      "learning_rate": 8.320124893556628e-05,
      "loss": 0.026192039251327515,
      "step": 59210
    },
    {
      "epoch": 16.809537326142493,
      "grad_norm": 0.2808459997177124,
      "learning_rate": 8.319841044564293e-05,
      "loss": 0.012450910359621047,
      "step": 59220
    },
    {
      "epoch": 16.812375816065853,
      "grad_norm": 1.5280746221542358,
      "learning_rate": 8.319557195571957e-05,
      "loss": 0.025003212690353393,
      "step": 59230
    },
    {
      "epoch": 16.815214305989215,
      "grad_norm": 3.0192575454711914,
      "learning_rate": 8.319273346579621e-05,
      "loss": 0.029531455039978026,
      "step": 59240
    },
    {
      "epoch": 16.818052795912575,
      "grad_norm": 5.311906814575195,
      "learning_rate": 8.318989497587284e-05,
      "loss": 0.02772926092147827,
      "step": 59250
    },
    {
      "epoch": 16.820891285835934,
      "grad_norm": 20.13922882080078,
      "learning_rate": 8.318705648594948e-05,
      "loss": 0.057452791929244997,
      "step": 59260
    },
    {
      "epoch": 16.823729775759297,
      "grad_norm": 9.292319297790527,
      "learning_rate": 8.318421799602612e-05,
      "loss": 0.02266717404127121,
      "step": 59270
    },
    {
      "epoch": 16.826568265682656,
      "grad_norm": 1.6498178243637085,
      "learning_rate": 8.318137950610275e-05,
      "loss": 0.03798863887786865,
      "step": 59280
    },
    {
      "epoch": 16.82940675560602,
      "grad_norm": 4.63967752456665,
      "learning_rate": 8.31785410161794e-05,
      "loss": 0.01955742686986923,
      "step": 59290
    },
    {
      "epoch": 16.83224524552938,
      "grad_norm": 0.9928451776504517,
      "learning_rate": 8.317570252625604e-05,
      "loss": 0.016254188120365144,
      "step": 59300
    },
    {
      "epoch": 16.835083735452738,
      "grad_norm": 7.91848611831665,
      "learning_rate": 8.317286403633267e-05,
      "loss": 0.024498085677623748,
      "step": 59310
    },
    {
      "epoch": 16.8379222253761,
      "grad_norm": 2.515704870223999,
      "learning_rate": 8.317002554640931e-05,
      "loss": 0.020629404485225676,
      "step": 59320
    },
    {
      "epoch": 16.84076071529946,
      "grad_norm": 1.1646966934204102,
      "learning_rate": 8.316718705648595e-05,
      "loss": 0.021324965357780456,
      "step": 59330
    },
    {
      "epoch": 16.843599205222823,
      "grad_norm": 5.102069854736328,
      "learning_rate": 8.31643485665626e-05,
      "loss": 0.02417541891336441,
      "step": 59340
    },
    {
      "epoch": 16.846437695146182,
      "grad_norm": 3.9034411907196045,
      "learning_rate": 8.316151007663924e-05,
      "loss": 0.04414777159690857,
      "step": 59350
    },
    {
      "epoch": 16.84927618506954,
      "grad_norm": 7.37410306930542,
      "learning_rate": 8.315867158671588e-05,
      "loss": 0.04508165717124939,
      "step": 59360
    },
    {
      "epoch": 16.852114674992904,
      "grad_norm": 4.48972749710083,
      "learning_rate": 8.31558330967925e-05,
      "loss": 0.026401233673095704,
      "step": 59370
    },
    {
      "epoch": 16.854953164916264,
      "grad_norm": 6.299874782562256,
      "learning_rate": 8.315299460686915e-05,
      "loss": 0.02049678862094879,
      "step": 59380
    },
    {
      "epoch": 16.857791654839627,
      "grad_norm": 3.2093350887298584,
      "learning_rate": 8.315015611694579e-05,
      "loss": 0.018880945444107056,
      "step": 59390
    },
    {
      "epoch": 16.860630144762986,
      "grad_norm": 6.709545135498047,
      "learning_rate": 8.314731762702243e-05,
      "loss": 0.015604633092880248,
      "step": 59400
    },
    {
      "epoch": 16.863468634686345,
      "grad_norm": 6.909293174743652,
      "learning_rate": 8.314447913709906e-05,
      "loss": 0.017441083490848542,
      "step": 59410
    },
    {
      "epoch": 16.86630712460971,
      "grad_norm": 2.27356219291687,
      "learning_rate": 8.314164064717571e-05,
      "loss": 0.01717013567686081,
      "step": 59420
    },
    {
      "epoch": 16.869145614533068,
      "grad_norm": 2.114701271057129,
      "learning_rate": 8.313880215725235e-05,
      "loss": 0.03882797956466675,
      "step": 59430
    },
    {
      "epoch": 16.87198410445643,
      "grad_norm": 1.572272777557373,
      "learning_rate": 8.313596366732898e-05,
      "loss": 0.022096604108810425,
      "step": 59440
    },
    {
      "epoch": 16.87482259437979,
      "grad_norm": 3.547701358795166,
      "learning_rate": 8.313312517740562e-05,
      "loss": 0.025479802489280702,
      "step": 59450
    },
    {
      "epoch": 16.87766108430315,
      "grad_norm": 8.331896781921387,
      "learning_rate": 8.313028668748226e-05,
      "loss": 0.027176687121391298,
      "step": 59460
    },
    {
      "epoch": 16.880499574226512,
      "grad_norm": 11.463362693786621,
      "learning_rate": 8.312744819755889e-05,
      "loss": 0.03849641978740692,
      "step": 59470
    },
    {
      "epoch": 16.88333806414987,
      "grad_norm": 6.4528489112854,
      "learning_rate": 8.312460970763553e-05,
      "loss": 0.029990372061729432,
      "step": 59480
    },
    {
      "epoch": 16.886176554073234,
      "grad_norm": 7.1564483642578125,
      "learning_rate": 8.312177121771219e-05,
      "loss": 0.023881636559963226,
      "step": 59490
    },
    {
      "epoch": 16.889015043996594,
      "grad_norm": 9.189780235290527,
      "learning_rate": 8.311893272778882e-05,
      "loss": 0.030628982186317443,
      "step": 59500
    },
    {
      "epoch": 16.889015043996594,
      "eval_accuracy": 0.965791314300248,
      "eval_loss": 0.11050871759653091,
      "eval_runtime": 31.5946,
      "eval_samples_per_second": 497.775,
      "eval_steps_per_second": 7.786,
      "step": 59500
    },
    {
      "epoch": 16.891853533919953,
      "grad_norm": 2.0772604942321777,
      "learning_rate": 8.311609423786546e-05,
      "loss": 0.030345064401626588,
      "step": 59510
    },
    {
      "epoch": 16.894692023843316,
      "grad_norm": 14.330068588256836,
      "learning_rate": 8.31132557479421e-05,
      "loss": 0.02902843654155731,
      "step": 59520
    },
    {
      "epoch": 16.897530513766675,
      "grad_norm": 4.2019243240356445,
      "learning_rate": 8.311041725801874e-05,
      "loss": 0.012346556037664413,
      "step": 59530
    },
    {
      "epoch": 16.900369003690038,
      "grad_norm": 10.613985061645508,
      "learning_rate": 8.310757876809537e-05,
      "loss": 0.04536116123199463,
      "step": 59540
    },
    {
      "epoch": 16.903207493613397,
      "grad_norm": 8.715611457824707,
      "learning_rate": 8.310474027817202e-05,
      "loss": 0.02634459137916565,
      "step": 59550
    },
    {
      "epoch": 16.906045983536757,
      "grad_norm": 9.356816291809082,
      "learning_rate": 8.310190178824866e-05,
      "loss": 0.033313590288162234,
      "step": 59560
    },
    {
      "epoch": 16.90888447346012,
      "grad_norm": 9.63582992553711,
      "learning_rate": 8.309906329832529e-05,
      "loss": 0.03300425112247467,
      "step": 59570
    },
    {
      "epoch": 16.91172296338348,
      "grad_norm": 4.4836320877075195,
      "learning_rate": 8.309622480840193e-05,
      "loss": 0.02680702805519104,
      "step": 59580
    },
    {
      "epoch": 16.914561453306842,
      "grad_norm": 8.611139297485352,
      "learning_rate": 8.309338631847858e-05,
      "loss": 0.020978954434394837,
      "step": 59590
    },
    {
      "epoch": 16.9173999432302,
      "grad_norm": 15.433649063110352,
      "learning_rate": 8.30905478285552e-05,
      "loss": 0.027812206745147706,
      "step": 59600
    },
    {
      "epoch": 16.920238433153564,
      "grad_norm": 5.864755153656006,
      "learning_rate": 8.308770933863184e-05,
      "loss": 0.029373759031295778,
      "step": 59610
    },
    {
      "epoch": 16.923076923076923,
      "grad_norm": 7.978265762329102,
      "learning_rate": 8.30848708487085e-05,
      "loss": 0.024450643360614775,
      "step": 59620
    },
    {
      "epoch": 16.925915413000283,
      "grad_norm": 3.213543176651001,
      "learning_rate": 8.308203235878513e-05,
      "loss": 0.018979364633560182,
      "step": 59630
    },
    {
      "epoch": 16.928753902923646,
      "grad_norm": 3.1002628803253174,
      "learning_rate": 8.307919386886177e-05,
      "loss": 0.02413661330938339,
      "step": 59640
    },
    {
      "epoch": 16.931592392847005,
      "grad_norm": 1.833674430847168,
      "learning_rate": 8.307635537893841e-05,
      "loss": 0.025863069295883178,
      "step": 59650
    },
    {
      "epoch": 16.934430882770368,
      "grad_norm": 0.3098338544368744,
      "learning_rate": 8.307351688901505e-05,
      "loss": 0.02430594861507416,
      "step": 59660
    },
    {
      "epoch": 16.937269372693727,
      "grad_norm": 2.6017730236053467,
      "learning_rate": 8.307067839909168e-05,
      "loss": 0.02899605631828308,
      "step": 59670
    },
    {
      "epoch": 16.940107862617086,
      "grad_norm": 11.811752319335938,
      "learning_rate": 8.306783990916832e-05,
      "loss": 0.051928967237472534,
      "step": 59680
    },
    {
      "epoch": 16.94294635254045,
      "grad_norm": 3.9172732830047607,
      "learning_rate": 8.306500141924498e-05,
      "loss": 0.012934400141239167,
      "step": 59690
    },
    {
      "epoch": 16.94578484246381,
      "grad_norm": 3.5357184410095215,
      "learning_rate": 8.30621629293216e-05,
      "loss": 0.022454433143138885,
      "step": 59700
    },
    {
      "epoch": 16.94862333238717,
      "grad_norm": 15.8103666305542,
      "learning_rate": 8.305932443939824e-05,
      "loss": 0.03347194194793701,
      "step": 59710
    },
    {
      "epoch": 16.95146182231053,
      "grad_norm": 5.368538856506348,
      "learning_rate": 8.305648594947489e-05,
      "loss": 0.018606145679950715,
      "step": 59720
    },
    {
      "epoch": 16.95430031223389,
      "grad_norm": 10.236196517944336,
      "learning_rate": 8.305364745955151e-05,
      "loss": 0.020950442552566527,
      "step": 59730
    },
    {
      "epoch": 16.957138802157253,
      "grad_norm": 6.344552516937256,
      "learning_rate": 8.305080896962816e-05,
      "loss": 0.02093493938446045,
      "step": 59740
    },
    {
      "epoch": 16.959977292080612,
      "grad_norm": 1.1742557287216187,
      "learning_rate": 8.304797047970481e-05,
      "loss": 0.012659642100334167,
      "step": 59750
    },
    {
      "epoch": 16.962815782003975,
      "grad_norm": 3.148052215576172,
      "learning_rate": 8.304513198978144e-05,
      "loss": 0.012905122339725494,
      "step": 59760
    },
    {
      "epoch": 16.965654271927335,
      "grad_norm": 2.044912576675415,
      "learning_rate": 8.304229349985808e-05,
      "loss": 0.012452203035354614,
      "step": 59770
    },
    {
      "epoch": 16.968492761850694,
      "grad_norm": 8.290250778198242,
      "learning_rate": 8.303945500993472e-05,
      "loss": 0.02918228507041931,
      "step": 59780
    },
    {
      "epoch": 16.971331251774057,
      "grad_norm": 0.7254201769828796,
      "learning_rate": 8.303661652001136e-05,
      "loss": 0.01370764970779419,
      "step": 59790
    },
    {
      "epoch": 16.974169741697416,
      "grad_norm": 1.0992618799209595,
      "learning_rate": 8.303377803008799e-05,
      "loss": 0.016495805978775025,
      "step": 59800
    },
    {
      "epoch": 16.97700823162078,
      "grad_norm": 8.768839836120605,
      "learning_rate": 8.303093954016463e-05,
      "loss": 0.020415471494197847,
      "step": 59810
    },
    {
      "epoch": 16.97984672154414,
      "grad_norm": 8.609142303466797,
      "learning_rate": 8.302810105024129e-05,
      "loss": 0.021872419118881225,
      "step": 59820
    },
    {
      "epoch": 16.982685211467498,
      "grad_norm": 9.52709674835205,
      "learning_rate": 8.302526256031791e-05,
      "loss": 0.036979538202285764,
      "step": 59830
    },
    {
      "epoch": 16.98552370139086,
      "grad_norm": 5.274742126464844,
      "learning_rate": 8.302242407039456e-05,
      "loss": 0.040659642219543456,
      "step": 59840
    },
    {
      "epoch": 16.98836219131422,
      "grad_norm": 1.3396081924438477,
      "learning_rate": 8.30195855804712e-05,
      "loss": 0.016694553196430206,
      "step": 59850
    },
    {
      "epoch": 16.991200681237583,
      "grad_norm": 12.425445556640625,
      "learning_rate": 8.301674709054782e-05,
      "loss": 0.023501750826835633,
      "step": 59860
    },
    {
      "epoch": 16.994039171160942,
      "grad_norm": 10.03705883026123,
      "learning_rate": 8.301390860062447e-05,
      "loss": 0.021814729273319244,
      "step": 59870
    },
    {
      "epoch": 16.9968776610843,
      "grad_norm": 8.44933032989502,
      "learning_rate": 8.301107011070111e-05,
      "loss": 0.049596580862998965,
      "step": 59880
    },
    {
      "epoch": 16.999716151007664,
      "grad_norm": 0.7388323545455933,
      "learning_rate": 8.300823162077775e-05,
      "loss": 0.027832764387130737,
      "step": 59890
    },
    {
      "epoch": 17.002554640931024,
      "grad_norm": 0.48154428601264954,
      "learning_rate": 8.300539313085439e-05,
      "loss": 0.02041080445051193,
      "step": 59900
    },
    {
      "epoch": 17.005393130854387,
      "grad_norm": 2.1173973083496094,
      "learning_rate": 8.300255464093103e-05,
      "loss": 0.007772276550531388,
      "step": 59910
    },
    {
      "epoch": 17.008231620777746,
      "grad_norm": 10.271995544433594,
      "learning_rate": 8.299971615100767e-05,
      "loss": 0.018062648177146912,
      "step": 59920
    },
    {
      "epoch": 17.011070110701105,
      "grad_norm": 5.4751787185668945,
      "learning_rate": 8.29968776610843e-05,
      "loss": 0.0222108393907547,
      "step": 59930
    },
    {
      "epoch": 17.013908600624468,
      "grad_norm": 10.109040260314941,
      "learning_rate": 8.299403917116094e-05,
      "loss": 0.011389323323965073,
      "step": 59940
    },
    {
      "epoch": 17.016747090547828,
      "grad_norm": 0.5950800180435181,
      "learning_rate": 8.29912006812376e-05,
      "loss": 0.022434996068477632,
      "step": 59950
    },
    {
      "epoch": 17.01958558047119,
      "grad_norm": 1.5476775169372559,
      "learning_rate": 8.298836219131423e-05,
      "loss": 0.011746164411306381,
      "step": 59960
    },
    {
      "epoch": 17.02242407039455,
      "grad_norm": 0.9059267640113831,
      "learning_rate": 8.298552370139087e-05,
      "loss": 0.01673087626695633,
      "step": 59970
    },
    {
      "epoch": 17.025262560317913,
      "grad_norm": 2.857586622238159,
      "learning_rate": 8.298268521146751e-05,
      "loss": 0.016063447296619415,
      "step": 59980
    },
    {
      "epoch": 17.028101050241272,
      "grad_norm": 7.521078586578369,
      "learning_rate": 8.297984672154414e-05,
      "loss": 0.010403391718864442,
      "step": 59990
    },
    {
      "epoch": 17.03093954016463,
      "grad_norm": 13.28499984741211,
      "learning_rate": 8.297700823162078e-05,
      "loss": 0.048198947310447694,
      "step": 60000
    },
    {
      "epoch": 17.03093954016463,
      "eval_accuracy": 0.9584154638519743,
      "eval_loss": 0.1367514282464981,
      "eval_runtime": 31.5703,
      "eval_samples_per_second": 498.158,
      "eval_steps_per_second": 7.792,
      "step": 60000
    },
    {
      "epoch": 17.033778030087994,
      "grad_norm": 1.726984977722168,
      "learning_rate": 8.297416974169742e-05,
      "loss": 0.02558598220348358,
      "step": 60010
    },
    {
      "epoch": 17.036616520011354,
      "grad_norm": 8.808225631713867,
      "learning_rate": 8.297133125177406e-05,
      "loss": 0.010977664589881897,
      "step": 60020
    },
    {
      "epoch": 17.039455009934716,
      "grad_norm": 0.8192585110664368,
      "learning_rate": 8.29684927618507e-05,
      "loss": 0.01847737729549408,
      "step": 60030
    },
    {
      "epoch": 17.042293499858076,
      "grad_norm": 12.757583618164062,
      "learning_rate": 8.296565427192734e-05,
      "loss": 0.036665529012680054,
      "step": 60040
    },
    {
      "epoch": 17.045131989781435,
      "grad_norm": 6.71480131149292,
      "learning_rate": 8.296281578200398e-05,
      "loss": 0.03365083038806915,
      "step": 60050
    },
    {
      "epoch": 17.047970479704798,
      "grad_norm": 0.3680950403213501,
      "learning_rate": 8.295997729208061e-05,
      "loss": 0.011141233146190643,
      "step": 60060
    },
    {
      "epoch": 17.050808969628157,
      "grad_norm": 7.116525650024414,
      "learning_rate": 8.295713880215725e-05,
      "loss": 0.018178504705429078,
      "step": 60070
    },
    {
      "epoch": 17.05364745955152,
      "grad_norm": 5.741809368133545,
      "learning_rate": 8.29543003122339e-05,
      "loss": 0.017730489373207092,
      "step": 60080
    },
    {
      "epoch": 17.05648594947488,
      "grad_norm": 10.484217643737793,
      "learning_rate": 8.295146182231054e-05,
      "loss": 0.03090069890022278,
      "step": 60090
    },
    {
      "epoch": 17.05932443939824,
      "grad_norm": 1.1893121004104614,
      "learning_rate": 8.294862333238718e-05,
      "loss": 0.02439778298139572,
      "step": 60100
    },
    {
      "epoch": 17.0621629293216,
      "grad_norm": 4.7010931968688965,
      "learning_rate": 8.294578484246382e-05,
      "loss": 0.017384180426597597,
      "step": 60110
    },
    {
      "epoch": 17.06500141924496,
      "grad_norm": 5.869739055633545,
      "learning_rate": 8.294294635254045e-05,
      "loss": 0.02081870138645172,
      "step": 60120
    },
    {
      "epoch": 17.067839909168324,
      "grad_norm": 3.771984100341797,
      "learning_rate": 8.294010786261709e-05,
      "loss": 0.01817036122083664,
      "step": 60130
    },
    {
      "epoch": 17.070678399091683,
      "grad_norm": 7.8905558586120605,
      "learning_rate": 8.293726937269373e-05,
      "loss": 0.014063933491706848,
      "step": 60140
    },
    {
      "epoch": 17.073516889015043,
      "grad_norm": 7.925786972045898,
      "learning_rate": 8.293443088277037e-05,
      "loss": 0.027467098832130433,
      "step": 60150
    },
    {
      "epoch": 17.076355378938405,
      "grad_norm": 2.6028711795806885,
      "learning_rate": 8.293159239284701e-05,
      "loss": 0.009261946380138397,
      "step": 60160
    },
    {
      "epoch": 17.079193868861765,
      "grad_norm": 2.34194016456604,
      "learning_rate": 8.292875390292365e-05,
      "loss": 0.02392832189798355,
      "step": 60170
    },
    {
      "epoch": 17.082032358785128,
      "grad_norm": 6.850749969482422,
      "learning_rate": 8.29259154130003e-05,
      "loss": 0.016964365541934968,
      "step": 60180
    },
    {
      "epoch": 17.084870848708487,
      "grad_norm": 0.5283069610595703,
      "learning_rate": 8.292307692307692e-05,
      "loss": 0.01907171607017517,
      "step": 60190
    },
    {
      "epoch": 17.087709338631846,
      "grad_norm": 0.7016083002090454,
      "learning_rate": 8.292023843315356e-05,
      "loss": 0.022131794691085817,
      "step": 60200
    },
    {
      "epoch": 17.09054782855521,
      "grad_norm": 11.678545951843262,
      "learning_rate": 8.29173999432302e-05,
      "loss": 0.0329552173614502,
      "step": 60210
    },
    {
      "epoch": 17.09338631847857,
      "grad_norm": 8.187173843383789,
      "learning_rate": 8.291456145330685e-05,
      "loss": 0.014009331166744233,
      "step": 60220
    },
    {
      "epoch": 17.09622480840193,
      "grad_norm": 1.4075108766555786,
      "learning_rate": 8.291172296338349e-05,
      "loss": 0.023814243078231812,
      "step": 60230
    },
    {
      "epoch": 17.09906329832529,
      "grad_norm": 13.14627742767334,
      "learning_rate": 8.290888447346013e-05,
      "loss": 0.023088133335113524,
      "step": 60240
    },
    {
      "epoch": 17.10190178824865,
      "grad_norm": 15.659075736999512,
      "learning_rate": 8.290604598353676e-05,
      "loss": 0.02848309576511383,
      "step": 60250
    },
    {
      "epoch": 17.104740278172013,
      "grad_norm": 19.55657958984375,
      "learning_rate": 8.29032074936134e-05,
      "loss": 0.0240742027759552,
      "step": 60260
    },
    {
      "epoch": 17.107578768095372,
      "grad_norm": 5.172979831695557,
      "learning_rate": 8.290036900369004e-05,
      "loss": 0.022654667496681213,
      "step": 60270
    },
    {
      "epoch": 17.110417258018735,
      "grad_norm": 4.127310276031494,
      "learning_rate": 8.289753051376668e-05,
      "loss": 0.013410001993179321,
      "step": 60280
    },
    {
      "epoch": 17.113255747942095,
      "grad_norm": 3.83300518989563,
      "learning_rate": 8.289469202384332e-05,
      "loss": 0.03008730411529541,
      "step": 60290
    },
    {
      "epoch": 17.116094237865454,
      "grad_norm": 3.6131479740142822,
      "learning_rate": 8.289185353391996e-05,
      "loss": 0.015287838876247406,
      "step": 60300
    },
    {
      "epoch": 17.118932727788817,
      "grad_norm": 0.22699904441833496,
      "learning_rate": 8.288901504399659e-05,
      "loss": 0.01297130435705185,
      "step": 60310
    },
    {
      "epoch": 17.121771217712176,
      "grad_norm": 2.5407752990722656,
      "learning_rate": 8.288617655407323e-05,
      "loss": 0.02008923441171646,
      "step": 60320
    },
    {
      "epoch": 17.12460970763554,
      "grad_norm": 7.36152458190918,
      "learning_rate": 8.288333806414987e-05,
      "loss": 0.01682742089033127,
      "step": 60330
    },
    {
      "epoch": 17.1274481975589,
      "grad_norm": 12.616373062133789,
      "learning_rate": 8.288049957422652e-05,
      "loss": 0.032228341698646544,
      "step": 60340
    },
    {
      "epoch": 17.130286687482258,
      "grad_norm": 1.2286903858184814,
      "learning_rate": 8.287766108430316e-05,
      "loss": 0.008339907974004745,
      "step": 60350
    },
    {
      "epoch": 17.13312517740562,
      "grad_norm": 0.852236270904541,
      "learning_rate": 8.28748225943798e-05,
      "loss": 0.02148095816373825,
      "step": 60360
    },
    {
      "epoch": 17.13596366732898,
      "grad_norm": 0.7790113091468811,
      "learning_rate": 8.287198410445644e-05,
      "loss": 0.020625939965248107,
      "step": 60370
    },
    {
      "epoch": 17.138802157252343,
      "grad_norm": 7.01544189453125,
      "learning_rate": 8.286914561453307e-05,
      "loss": 0.02250695526599884,
      "step": 60380
    },
    {
      "epoch": 17.141640647175702,
      "grad_norm": 0.8989993929862976,
      "learning_rate": 8.286630712460971e-05,
      "loss": 0.01921136528253555,
      "step": 60390
    },
    {
      "epoch": 17.144479137099065,
      "grad_norm": 10.793028831481934,
      "learning_rate": 8.286346863468635e-05,
      "loss": 0.02690858244895935,
      "step": 60400
    },
    {
      "epoch": 17.147317627022424,
      "grad_norm": 13.074443817138672,
      "learning_rate": 8.286063014476298e-05,
      "loss": 0.019533056020736694,
      "step": 60410
    },
    {
      "epoch": 17.150156116945784,
      "grad_norm": 14.613844871520996,
      "learning_rate": 8.285779165483963e-05,
      "loss": 0.03272394835948944,
      "step": 60420
    },
    {
      "epoch": 17.152994606869147,
      "grad_norm": 8.377421379089355,
      "learning_rate": 8.285495316491627e-05,
      "loss": 0.027012503147125243,
      "step": 60430
    },
    {
      "epoch": 17.155833096792506,
      "grad_norm": 2.1425633430480957,
      "learning_rate": 8.28521146749929e-05,
      "loss": 0.0074956871569156645,
      "step": 60440
    },
    {
      "epoch": 17.15867158671587,
      "grad_norm": 3.3529205322265625,
      "learning_rate": 8.284927618506954e-05,
      "loss": 0.017431148886680604,
      "step": 60450
    },
    {
      "epoch": 17.161510076639228,
      "grad_norm": 1.3230326175689697,
      "learning_rate": 8.284643769514619e-05,
      "loss": 0.011084027588367462,
      "step": 60460
    },
    {
      "epoch": 17.164348566562587,
      "grad_norm": 12.417771339416504,
      "learning_rate": 8.284359920522283e-05,
      "loss": 0.028391575813293456,
      "step": 60470
    },
    {
      "epoch": 17.16718705648595,
      "grad_norm": 3.665861129760742,
      "learning_rate": 8.284076071529945e-05,
      "loss": 0.01258855015039444,
      "step": 60480
    },
    {
      "epoch": 17.17002554640931,
      "grad_norm": 6.475068092346191,
      "learning_rate": 8.283792222537611e-05,
      "loss": 0.017808590829372407,
      "step": 60490
    },
    {
      "epoch": 17.172864036332673,
      "grad_norm": 2.4653563499450684,
      "learning_rate": 8.283508373545275e-05,
      "loss": 0.017405638098716737,
      "step": 60500
    },
    {
      "epoch": 17.172864036332673,
      "eval_accuracy": 0.9638201818528646,
      "eval_loss": 0.11213576048612595,
      "eval_runtime": 31.6708,
      "eval_samples_per_second": 496.577,
      "eval_steps_per_second": 7.767,
      "step": 60500
    },
    {
      "epoch": 17.175702526256032,
      "grad_norm": 6.46360445022583,
      "learning_rate": 8.283224524552938e-05,
      "loss": 0.008883007615804673,
      "step": 60510
    },
    {
      "epoch": 17.17854101617939,
      "grad_norm": 3.651374340057373,
      "learning_rate": 8.282940675560602e-05,
      "loss": 0.013101382553577423,
      "step": 60520
    },
    {
      "epoch": 17.181379506102754,
      "grad_norm": 8.114255905151367,
      "learning_rate": 8.282656826568266e-05,
      "loss": 0.014031532406806945,
      "step": 60530
    },
    {
      "epoch": 17.184217996026113,
      "grad_norm": 6.263001441955566,
      "learning_rate": 8.282372977575929e-05,
      "loss": 0.010283049196004868,
      "step": 60540
    },
    {
      "epoch": 17.187056485949476,
      "grad_norm": 2.2577359676361084,
      "learning_rate": 8.282089128583594e-05,
      "loss": 0.00542941614985466,
      "step": 60550
    },
    {
      "epoch": 17.189894975872836,
      "grad_norm": 3.672323703765869,
      "learning_rate": 8.281805279591259e-05,
      "loss": 0.008926591277122498,
      "step": 60560
    },
    {
      "epoch": 17.192733465796195,
      "grad_norm": 1.2772222757339478,
      "learning_rate": 8.281521430598921e-05,
      "loss": 0.018028415739536285,
      "step": 60570
    },
    {
      "epoch": 17.195571955719558,
      "grad_norm": 8.611686706542969,
      "learning_rate": 8.281237581606585e-05,
      "loss": 0.011967145651578904,
      "step": 60580
    },
    {
      "epoch": 17.198410445642917,
      "grad_norm": 5.633968830108643,
      "learning_rate": 8.28095373261425e-05,
      "loss": 0.01403355747461319,
      "step": 60590
    },
    {
      "epoch": 17.20124893556628,
      "grad_norm": 7.445531368255615,
      "learning_rate": 8.280669883621914e-05,
      "loss": 0.01686886101961136,
      "step": 60600
    },
    {
      "epoch": 17.20408742548964,
      "grad_norm": 9.879779815673828,
      "learning_rate": 8.280386034629577e-05,
      "loss": 0.020007339119911195,
      "step": 60610
    },
    {
      "epoch": 17.206925915413,
      "grad_norm": 11.010483741760254,
      "learning_rate": 8.280102185637242e-05,
      "loss": 0.02129928916692734,
      "step": 60620
    },
    {
      "epoch": 17.20976440533636,
      "grad_norm": 6.090151786804199,
      "learning_rate": 8.279818336644906e-05,
      "loss": 0.015851506590843202,
      "step": 60630
    },
    {
      "epoch": 17.21260289525972,
      "grad_norm": 2.9903645515441895,
      "learning_rate": 8.279534487652569e-05,
      "loss": 0.023469410836696625,
      "step": 60640
    },
    {
      "epoch": 17.215441385183084,
      "grad_norm": 0.5268586874008179,
      "learning_rate": 8.279250638660233e-05,
      "loss": 0.016330859065055846,
      "step": 60650
    },
    {
      "epoch": 17.218279875106443,
      "grad_norm": 1.6142635345458984,
      "learning_rate": 8.278966789667897e-05,
      "loss": 0.009772557765245438,
      "step": 60660
    },
    {
      "epoch": 17.221118365029803,
      "grad_norm": 0.33875951170921326,
      "learning_rate": 8.27868294067556e-05,
      "loss": 0.01213638037443161,
      "step": 60670
    },
    {
      "epoch": 17.223956854953165,
      "grad_norm": 0.6342658400535583,
      "learning_rate": 8.278399091683225e-05,
      "loss": 0.015492451190948487,
      "step": 60680
    },
    {
      "epoch": 17.226795344876525,
      "grad_norm": 1.8617217540740967,
      "learning_rate": 8.27811524269089e-05,
      "loss": 0.021076194941997528,
      "step": 60690
    },
    {
      "epoch": 17.229633834799888,
      "grad_norm": 6.909392356872559,
      "learning_rate": 8.277831393698552e-05,
      "loss": 0.02256089895963669,
      "step": 60700
    },
    {
      "epoch": 17.232472324723247,
      "grad_norm": 0.8374917507171631,
      "learning_rate": 8.277547544706217e-05,
      "loss": 0.018785883486270905,
      "step": 60710
    },
    {
      "epoch": 17.235310814646606,
      "grad_norm": 7.933974266052246,
      "learning_rate": 8.277263695713881e-05,
      "loss": 0.013684138655662537,
      "step": 60720
    },
    {
      "epoch": 17.23814930456997,
      "grad_norm": 3.9869275093078613,
      "learning_rate": 8.276979846721545e-05,
      "loss": 0.018250496685504915,
      "step": 60730
    },
    {
      "epoch": 17.24098779449333,
      "grad_norm": 4.982598781585693,
      "learning_rate": 8.276695997729208e-05,
      "loss": 0.02122266888618469,
      "step": 60740
    },
    {
      "epoch": 17.24382628441669,
      "grad_norm": 0.5410777926445007,
      "learning_rate": 8.276412148736873e-05,
      "loss": 0.02262820452451706,
      "step": 60750
    },
    {
      "epoch": 17.24666477434005,
      "grad_norm": 4.1117377281188965,
      "learning_rate": 8.276128299744537e-05,
      "loss": 0.021995536983013153,
      "step": 60760
    },
    {
      "epoch": 17.249503264263414,
      "grad_norm": 13.128637313842773,
      "learning_rate": 8.2758444507522e-05,
      "loss": 0.024110031127929688,
      "step": 60770
    },
    {
      "epoch": 17.252341754186773,
      "grad_norm": 6.246779441833496,
      "learning_rate": 8.275560601759864e-05,
      "loss": 0.02954556345939636,
      "step": 60780
    },
    {
      "epoch": 17.255180244110132,
      "grad_norm": 10.835840225219727,
      "learning_rate": 8.275276752767528e-05,
      "loss": 0.014489531517028809,
      "step": 60790
    },
    {
      "epoch": 17.258018734033495,
      "grad_norm": 1.1595449447631836,
      "learning_rate": 8.274992903775191e-05,
      "loss": 0.028317677974700927,
      "step": 60800
    },
    {
      "epoch": 17.260857223956855,
      "grad_norm": 6.88871955871582,
      "learning_rate": 8.274709054782855e-05,
      "loss": 0.03034435510635376,
      "step": 60810
    },
    {
      "epoch": 17.263695713880217,
      "grad_norm": 9.165369033813477,
      "learning_rate": 8.274425205790521e-05,
      "loss": 0.026477175951004028,
      "step": 60820
    },
    {
      "epoch": 17.266534203803577,
      "grad_norm": 4.424760341644287,
      "learning_rate": 8.274141356798183e-05,
      "loss": 0.016534657776355745,
      "step": 60830
    },
    {
      "epoch": 17.269372693726936,
      "grad_norm": 2.463466167449951,
      "learning_rate": 8.273857507805848e-05,
      "loss": 0.016133719682693483,
      "step": 60840
    },
    {
      "epoch": 17.2722111836503,
      "grad_norm": 1.105623722076416,
      "learning_rate": 8.273573658813512e-05,
      "loss": 0.009032898396253587,
      "step": 60850
    },
    {
      "epoch": 17.27504967357366,
      "grad_norm": 8.224363327026367,
      "learning_rate": 8.273289809821176e-05,
      "loss": 0.014842145144939423,
      "step": 60860
    },
    {
      "epoch": 17.27788816349702,
      "grad_norm": 2.3137259483337402,
      "learning_rate": 8.273005960828839e-05,
      "loss": 0.02080821692943573,
      "step": 60870
    },
    {
      "epoch": 17.28072665342038,
      "grad_norm": 4.120140552520752,
      "learning_rate": 8.272722111836504e-05,
      "loss": 0.022870764136314392,
      "step": 60880
    },
    {
      "epoch": 17.28356514334374,
      "grad_norm": 8.517062187194824,
      "learning_rate": 8.272438262844168e-05,
      "loss": 0.02370109558105469,
      "step": 60890
    },
    {
      "epoch": 17.286403633267103,
      "grad_norm": 0.7718416452407837,
      "learning_rate": 8.272154413851831e-05,
      "loss": 0.028670358657836913,
      "step": 60900
    },
    {
      "epoch": 17.289242123190462,
      "grad_norm": 0.37946709990501404,
      "learning_rate": 8.271870564859495e-05,
      "loss": 0.007799313962459564,
      "step": 60910
    },
    {
      "epoch": 17.292080613113825,
      "grad_norm": 9.967166900634766,
      "learning_rate": 8.27158671586716e-05,
      "loss": 0.023309630155563355,
      "step": 60920
    },
    {
      "epoch": 17.294919103037184,
      "grad_norm": 9.929471015930176,
      "learning_rate": 8.271302866874822e-05,
      "loss": 0.01647820174694061,
      "step": 60930
    },
    {
      "epoch": 17.297757592960544,
      "grad_norm": 7.87197208404541,
      "learning_rate": 8.271019017882486e-05,
      "loss": 0.02176760733127594,
      "step": 60940
    },
    {
      "epoch": 17.300596082883906,
      "grad_norm": 4.767561912536621,
      "learning_rate": 8.270735168890152e-05,
      "loss": 0.014724665880203247,
      "step": 60950
    },
    {
      "epoch": 17.303434572807266,
      "grad_norm": 6.339470863342285,
      "learning_rate": 8.270451319897815e-05,
      "loss": 0.028527584671974183,
      "step": 60960
    },
    {
      "epoch": 17.30627306273063,
      "grad_norm": 1.07589590549469,
      "learning_rate": 8.270167470905479e-05,
      "loss": 0.01153789609670639,
      "step": 60970
    },
    {
      "epoch": 17.309111552653988,
      "grad_norm": 14.57190990447998,
      "learning_rate": 8.269883621913143e-05,
      "loss": 0.0345747172832489,
      "step": 60980
    },
    {
      "epoch": 17.311950042577347,
      "grad_norm": 10.745390892028809,
      "learning_rate": 8.269599772920807e-05,
      "loss": 0.021194887161254884,
      "step": 60990
    },
    {
      "epoch": 17.31478853250071,
      "grad_norm": 5.401939392089844,
      "learning_rate": 8.26931592392847e-05,
      "loss": 0.026307469606399535,
      "step": 61000
    },
    {
      "epoch": 17.31478853250071,
      "eval_accuracy": 0.961849049405481,
      "eval_loss": 0.12669186294078827,
      "eval_runtime": 31.7504,
      "eval_samples_per_second": 495.333,
      "eval_steps_per_second": 7.748,
      "step": 61000
    },
    {
      "epoch": 17.31762702242407,
      "grad_norm": 1.6662150621414185,
      "learning_rate": 8.269032074936134e-05,
      "loss": 0.019942404329776765,
      "step": 61010
    },
    {
      "epoch": 17.320465512347432,
      "grad_norm": 2.3525197505950928,
      "learning_rate": 8.2687482259438e-05,
      "loss": 0.02618812918663025,
      "step": 61020
    },
    {
      "epoch": 17.323304002270792,
      "grad_norm": 5.391884803771973,
      "learning_rate": 8.268464376951462e-05,
      "loss": 0.023497243225574494,
      "step": 61030
    },
    {
      "epoch": 17.32614249219415,
      "grad_norm": 3.5315847396850586,
      "learning_rate": 8.268180527959126e-05,
      "loss": 0.02437012642621994,
      "step": 61040
    },
    {
      "epoch": 17.328980982117514,
      "grad_norm": 8.045187950134277,
      "learning_rate": 8.26789667896679e-05,
      "loss": 0.024712073802947997,
      "step": 61050
    },
    {
      "epoch": 17.331819472040873,
      "grad_norm": 1.0701186656951904,
      "learning_rate": 8.267612829974453e-05,
      "loss": 0.011471447348594666,
      "step": 61060
    },
    {
      "epoch": 17.334657961964236,
      "grad_norm": 2.743116617202759,
      "learning_rate": 8.267328980982117e-05,
      "loss": 0.01637566536664963,
      "step": 61070
    },
    {
      "epoch": 17.337496451887596,
      "grad_norm": 17.422468185424805,
      "learning_rate": 8.267045131989783e-05,
      "loss": 0.020722249150276185,
      "step": 61080
    },
    {
      "epoch": 17.340334941810955,
      "grad_norm": 7.825224876403809,
      "learning_rate": 8.266761282997446e-05,
      "loss": 0.02752777934074402,
      "step": 61090
    },
    {
      "epoch": 17.343173431734318,
      "grad_norm": 1.663003921508789,
      "learning_rate": 8.26647743400511e-05,
      "loss": 0.014432349801063537,
      "step": 61100
    },
    {
      "epoch": 17.346011921657677,
      "grad_norm": 1.9922808408737183,
      "learning_rate": 8.266193585012774e-05,
      "loss": 0.017636868357658386,
      "step": 61110
    },
    {
      "epoch": 17.34885041158104,
      "grad_norm": 2.2063710689544678,
      "learning_rate": 8.265909736020438e-05,
      "loss": 0.026861193776130676,
      "step": 61120
    },
    {
      "epoch": 17.3516889015044,
      "grad_norm": 14.567687034606934,
      "learning_rate": 8.265625887028101e-05,
      "loss": 0.02321614623069763,
      "step": 61130
    },
    {
      "epoch": 17.35452739142776,
      "grad_norm": 2.5034542083740234,
      "learning_rate": 8.265342038035765e-05,
      "loss": 0.03582393527030945,
      "step": 61140
    },
    {
      "epoch": 17.35736588135112,
      "grad_norm": 12.30094051361084,
      "learning_rate": 8.265058189043429e-05,
      "loss": 0.03541815876960754,
      "step": 61150
    },
    {
      "epoch": 17.36020437127448,
      "grad_norm": 4.182841777801514,
      "learning_rate": 8.264774340051093e-05,
      "loss": 0.023144432902336122,
      "step": 61160
    },
    {
      "epoch": 17.363042861197844,
      "grad_norm": Infinity,
      "learning_rate": 8.264490491058757e-05,
      "loss": 0.033010005950927734,
      "step": 61170
    },
    {
      "epoch": 17.365881351121203,
      "grad_norm": 5.141995429992676,
      "learning_rate": 8.264235026965654e-05,
      "loss": 0.014861783385276795,
      "step": 61180
    },
    {
      "epoch": 17.368719841044566,
      "grad_norm": 2.766414165496826,
      "learning_rate": 8.263951177973318e-05,
      "loss": 0.026583638787269593,
      "step": 61190
    },
    {
      "epoch": 17.371558330967925,
      "grad_norm": 4.854646682739258,
      "learning_rate": 8.263667328980982e-05,
      "loss": 0.020185372233390807,
      "step": 61200
    },
    {
      "epoch": 17.374396820891285,
      "grad_norm": 4.532007217407227,
      "learning_rate": 8.263383479988646e-05,
      "loss": 0.044744473695755,
      "step": 61210
    },
    {
      "epoch": 17.377235310814648,
      "grad_norm": 8.546868324279785,
      "learning_rate": 8.26309963099631e-05,
      "loss": 0.016836707293987275,
      "step": 61220
    },
    {
      "epoch": 17.380073800738007,
      "grad_norm": 8.586074829101562,
      "learning_rate": 8.262815782003975e-05,
      "loss": 0.012118186801671982,
      "step": 61230
    },
    {
      "epoch": 17.38291229066137,
      "grad_norm": 3.301542043685913,
      "learning_rate": 8.262531933011638e-05,
      "loss": 0.033820819854736325,
      "step": 61240
    },
    {
      "epoch": 17.38575078058473,
      "grad_norm": 5.653260231018066,
      "learning_rate": 8.262248084019302e-05,
      "loss": 0.026761633157730103,
      "step": 61250
    },
    {
      "epoch": 17.38858927050809,
      "grad_norm": 1.653430461883545,
      "learning_rate": 8.261964235026967e-05,
      "loss": 0.007496782392263412,
      "step": 61260
    },
    {
      "epoch": 17.39142776043145,
      "grad_norm": 5.211973667144775,
      "learning_rate": 8.26168038603463e-05,
      "loss": 0.018194952607154848,
      "step": 61270
    },
    {
      "epoch": 17.39426625035481,
      "grad_norm": 2.7129406929016113,
      "learning_rate": 8.261396537042294e-05,
      "loss": 0.013098260760307312,
      "step": 61280
    },
    {
      "epoch": 17.397104740278174,
      "grad_norm": 8.766242027282715,
      "learning_rate": 8.261112688049958e-05,
      "loss": 0.029319095611572265,
      "step": 61290
    },
    {
      "epoch": 17.399943230201533,
      "grad_norm": 6.793268203735352,
      "learning_rate": 8.260828839057621e-05,
      "loss": 0.021523159742355347,
      "step": 61300
    },
    {
      "epoch": 17.402781720124892,
      "grad_norm": 7.518362998962402,
      "learning_rate": 8.260544990065285e-05,
      "loss": 0.021864983439445495,
      "step": 61310
    },
    {
      "epoch": 17.405620210048255,
      "grad_norm": 0.4228460192680359,
      "learning_rate": 8.260261141072949e-05,
      "loss": 0.013299587368965148,
      "step": 61320
    },
    {
      "epoch": 17.408458699971614,
      "grad_norm": 4.653689861297607,
      "learning_rate": 8.259977292080613e-05,
      "loss": 0.009516909718513489,
      "step": 61330
    },
    {
      "epoch": 17.411297189894977,
      "grad_norm": 0.4042937159538269,
      "learning_rate": 8.259693443088278e-05,
      "loss": 0.02425500899553299,
      "step": 61340
    },
    {
      "epoch": 17.414135679818337,
      "grad_norm": 9.169136047363281,
      "learning_rate": 8.259409594095942e-05,
      "loss": 0.010941586643457412,
      "step": 61350
    },
    {
      "epoch": 17.416974169741696,
      "grad_norm": 2.9483869075775146,
      "learning_rate": 8.259125745103606e-05,
      "loss": 0.03379940986633301,
      "step": 61360
    },
    {
      "epoch": 17.41981265966506,
      "grad_norm": 9.949241638183594,
      "learning_rate": 8.258841896111269e-05,
      "loss": 0.02597987651824951,
      "step": 61370
    },
    {
      "epoch": 17.422651149588418,
      "grad_norm": 4.888527870178223,
      "learning_rate": 8.258558047118933e-05,
      "loss": 0.023670709133148192,
      "step": 61380
    },
    {
      "epoch": 17.42548963951178,
      "grad_norm": 8.181744575500488,
      "learning_rate": 8.258274198126597e-05,
      "loss": 0.017273417115211485,
      "step": 61390
    },
    {
      "epoch": 17.42832812943514,
      "grad_norm": 3.6558029651641846,
      "learning_rate": 8.257990349134261e-05,
      "loss": 0.01796172559261322,
      "step": 61400
    },
    {
      "epoch": 17.4311666193585,
      "grad_norm": 2.3941311836242676,
      "learning_rate": 8.257706500141925e-05,
      "loss": 0.02153703719377518,
      "step": 61410
    },
    {
      "epoch": 17.434005109281863,
      "grad_norm": 9.692488670349121,
      "learning_rate": 8.257422651149589e-05,
      "loss": 0.020052766799926756,
      "step": 61420
    },
    {
      "epoch": 17.436843599205222,
      "grad_norm": 8.038256645202637,
      "learning_rate": 8.257138802157252e-05,
      "loss": 0.018956753611564636,
      "step": 61430
    },
    {
      "epoch": 17.439682089128585,
      "grad_norm": 9.291525840759277,
      "learning_rate": 8.256854953164916e-05,
      "loss": 0.029993498325347902,
      "step": 61440
    },
    {
      "epoch": 17.442520579051944,
      "grad_norm": 8.019403457641602,
      "learning_rate": 8.25657110417258e-05,
      "loss": 0.03004841208457947,
      "step": 61450
    },
    {
      "epoch": 17.445359068975304,
      "grad_norm": 3.3659279346466064,
      "learning_rate": 8.256287255180244e-05,
      "loss": 0.029397329688072203,
      "step": 61460
    },
    {
      "epoch": 17.448197558898666,
      "grad_norm": 5.011568069458008,
      "learning_rate": 8.256003406187909e-05,
      "loss": 0.022481609880924226,
      "step": 61470
    },
    {
      "epoch": 17.451036048822026,
      "grad_norm": 7.6243391036987305,
      "learning_rate": 8.255719557195573e-05,
      "loss": 0.017621365189552308,
      "step": 61480
    },
    {
      "epoch": 17.45387453874539,
      "grad_norm": 2.629657030105591,
      "learning_rate": 8.255435708203237e-05,
      "loss": 0.011807309836149216,
      "step": 61490
    },
    {
      "epoch": 17.456713028668748,
      "grad_norm": 1.587230920791626,
      "learning_rate": 8.2551518592109e-05,
      "loss": 0.019284310936927795,
      "step": 61500
    },
    {
      "epoch": 17.456713028668748,
      "eval_accuracy": 0.9609588605582756,
      "eval_loss": 0.1191679984331131,
      "eval_runtime": 31.766,
      "eval_samples_per_second": 495.09,
      "eval_steps_per_second": 7.744,
      "step": 61500
    },
    {
      "epoch": 17.459551518592107,
      "grad_norm": 2.1209075450897217,
      "learning_rate": 8.254868010218564e-05,
      "loss": 0.012888598442077636,
      "step": 61510
    },
    {
      "epoch": 17.46239000851547,
      "grad_norm": 7.053223609924316,
      "learning_rate": 8.254584161226228e-05,
      "loss": 0.023036378622055053,
      "step": 61520
    },
    {
      "epoch": 17.46522849843883,
      "grad_norm": 6.634057998657227,
      "learning_rate": 8.254300312233892e-05,
      "loss": 0.03232641220092773,
      "step": 61530
    },
    {
      "epoch": 17.468066988362192,
      "grad_norm": 1.1502784490585327,
      "learning_rate": 8.254016463241556e-05,
      "loss": 0.017886632680892946,
      "step": 61540
    },
    {
      "epoch": 17.47090547828555,
      "grad_norm": 0.5109356045722961,
      "learning_rate": 8.25373261424922e-05,
      "loss": 0.016124421358108522,
      "step": 61550
    },
    {
      "epoch": 17.473743968208915,
      "grad_norm": 8.057290077209473,
      "learning_rate": 8.253448765256883e-05,
      "loss": 0.026202934980392455,
      "step": 61560
    },
    {
      "epoch": 17.476582458132274,
      "grad_norm": 8.621732711791992,
      "learning_rate": 8.253164916264547e-05,
      "loss": 0.010917245596647262,
      "step": 61570
    },
    {
      "epoch": 17.479420948055633,
      "grad_norm": 5.579722881317139,
      "learning_rate": 8.252881067272211e-05,
      "loss": 0.01770494282245636,
      "step": 61580
    },
    {
      "epoch": 17.482259437978996,
      "grad_norm": 1.4923181533813477,
      "learning_rate": 8.252597218279876e-05,
      "loss": 0.02207198888063431,
      "step": 61590
    },
    {
      "epoch": 17.485097927902356,
      "grad_norm": 0.5258405804634094,
      "learning_rate": 8.25231336928754e-05,
      "loss": 0.015277965366840363,
      "step": 61600
    },
    {
      "epoch": 17.48793641782572,
      "grad_norm": 0.5790354609489441,
      "learning_rate": 8.252029520295204e-05,
      "loss": 0.01770160347223282,
      "step": 61610
    },
    {
      "epoch": 17.490774907749078,
      "grad_norm": 2.4939751625061035,
      "learning_rate": 8.251745671302868e-05,
      "loss": 0.016967292129993438,
      "step": 61620
    },
    {
      "epoch": 17.493613397672437,
      "grad_norm": 7.325229167938232,
      "learning_rate": 8.251461822310531e-05,
      "loss": 0.02095828950405121,
      "step": 61630
    },
    {
      "epoch": 17.4964518875958,
      "grad_norm": 7.570591926574707,
      "learning_rate": 8.251177973318195e-05,
      "loss": 0.027062812447547914,
      "step": 61640
    },
    {
      "epoch": 17.49929037751916,
      "grad_norm": 2.952631950378418,
      "learning_rate": 8.250894124325859e-05,
      "loss": 0.015240280330181122,
      "step": 61650
    },
    {
      "epoch": 17.502128867442522,
      "grad_norm": 0.49329328536987305,
      "learning_rate": 8.250610275333523e-05,
      "loss": 0.028195285797119142,
      "step": 61660
    },
    {
      "epoch": 17.50496735736588,
      "grad_norm": 4.536864757537842,
      "learning_rate": 8.250326426341187e-05,
      "loss": 0.025210756063461303,
      "step": 61670
    },
    {
      "epoch": 17.50780584728924,
      "grad_norm": 3.4256057739257812,
      "learning_rate": 8.250042577348851e-05,
      "loss": 0.03867365717887879,
      "step": 61680
    },
    {
      "epoch": 17.510644337212604,
      "grad_norm": 13.087890625,
      "learning_rate": 8.249758728356514e-05,
      "loss": 0.04873683452606201,
      "step": 61690
    },
    {
      "epoch": 17.513482827135963,
      "grad_norm": 5.287320613861084,
      "learning_rate": 8.249474879364178e-05,
      "loss": 0.020358946919441224,
      "step": 61700
    },
    {
      "epoch": 17.516321317059326,
      "grad_norm": 6.921557903289795,
      "learning_rate": 8.249191030371842e-05,
      "loss": 0.01727771610021591,
      "step": 61710
    },
    {
      "epoch": 17.519159806982685,
      "grad_norm": 8.656535148620605,
      "learning_rate": 8.248907181379507e-05,
      "loss": 0.031967931985855104,
      "step": 61720
    },
    {
      "epoch": 17.521998296906045,
      "grad_norm": 9.841416358947754,
      "learning_rate": 8.248623332387171e-05,
      "loss": 0.010060474276542664,
      "step": 61730
    },
    {
      "epoch": 17.524836786829407,
      "grad_norm": 0.8050335645675659,
      "learning_rate": 8.248339483394835e-05,
      "loss": 0.025714984536170958,
      "step": 61740
    },
    {
      "epoch": 17.527675276752767,
      "grad_norm": 2.155386447906494,
      "learning_rate": 8.248055634402499e-05,
      "loss": 0.020784711837768553,
      "step": 61750
    },
    {
      "epoch": 17.53051376667613,
      "grad_norm": 6.113373279571533,
      "learning_rate": 8.247771785410162e-05,
      "loss": 0.02968502938747406,
      "step": 61760
    },
    {
      "epoch": 17.53335225659949,
      "grad_norm": 3.792987585067749,
      "learning_rate": 8.247487936417826e-05,
      "loss": 0.014387741684913635,
      "step": 61770
    },
    {
      "epoch": 17.53619074652285,
      "grad_norm": 2.1913392543792725,
      "learning_rate": 8.24720408742549e-05,
      "loss": 0.030770033597946167,
      "step": 61780
    },
    {
      "epoch": 17.53902923644621,
      "grad_norm": 6.313526630401611,
      "learning_rate": 8.246920238433153e-05,
      "loss": 0.029643353819847108,
      "step": 61790
    },
    {
      "epoch": 17.54186772636957,
      "grad_norm": 9.234797477722168,
      "learning_rate": 8.246636389440818e-05,
      "loss": 0.0283705472946167,
      "step": 61800
    },
    {
      "epoch": 17.544706216292933,
      "grad_norm": 6.143775463104248,
      "learning_rate": 8.246352540448482e-05,
      "loss": 0.0158234179019928,
      "step": 61810
    },
    {
      "epoch": 17.547544706216293,
      "grad_norm": 1.9553626775741577,
      "learning_rate": 8.246068691456145e-05,
      "loss": 0.01241091787815094,
      "step": 61820
    },
    {
      "epoch": 17.550383196139652,
      "grad_norm": 1.0569223165512085,
      "learning_rate": 8.24578484246381e-05,
      "loss": 0.01063622534275055,
      "step": 61830
    },
    {
      "epoch": 17.553221686063015,
      "grad_norm": 2.694061756134033,
      "learning_rate": 8.245500993471474e-05,
      "loss": 0.01957698315382004,
      "step": 61840
    },
    {
      "epoch": 17.556060175986374,
      "grad_norm": 3.3244571685791016,
      "learning_rate": 8.245217144479138e-05,
      "loss": 0.027434754371643066,
      "step": 61850
    },
    {
      "epoch": 17.558898665909737,
      "grad_norm": 5.469682693481445,
      "learning_rate": 8.244933295486802e-05,
      "loss": 0.010187521576881409,
      "step": 61860
    },
    {
      "epoch": 17.561737155833097,
      "grad_norm": 0.5937946438789368,
      "learning_rate": 8.244649446494466e-05,
      "loss": 0.01589364856481552,
      "step": 61870
    },
    {
      "epoch": 17.564575645756456,
      "grad_norm": 13.531807899475098,
      "learning_rate": 8.24436559750213e-05,
      "loss": 0.039757069945335385,
      "step": 61880
    },
    {
      "epoch": 17.56741413567982,
      "grad_norm": 1.991186261177063,
      "learning_rate": 8.244081748509793e-05,
      "loss": 0.022796808183193205,
      "step": 61890
    },
    {
      "epoch": 17.570252625603178,
      "grad_norm": 2.8076136112213135,
      "learning_rate": 8.243797899517457e-05,
      "loss": 0.01887902766466141,
      "step": 61900
    },
    {
      "epoch": 17.57309111552654,
      "grad_norm": 7.56357479095459,
      "learning_rate": 8.243514050525121e-05,
      "loss": 0.014006641507148743,
      "step": 61910
    },
    {
      "epoch": 17.5759296054499,
      "grad_norm": 10.527331352233887,
      "learning_rate": 8.243230201532784e-05,
      "loss": 0.033383351564407346,
      "step": 61920
    },
    {
      "epoch": 17.578768095373263,
      "grad_norm": 3.0198590755462646,
      "learning_rate": 8.24294635254045e-05,
      "loss": 0.016942545771598816,
      "step": 61930
    },
    {
      "epoch": 17.581606585296623,
      "grad_norm": 2.1270992755889893,
      "learning_rate": 8.242662503548114e-05,
      "loss": 0.010510009527206422,
      "step": 61940
    },
    {
      "epoch": 17.584445075219982,
      "grad_norm": 0.9897298812866211,
      "learning_rate": 8.242378654555776e-05,
      "loss": 0.01847278028726578,
      "step": 61950
    },
    {
      "epoch": 17.587283565143345,
      "grad_norm": 8.407822608947754,
      "learning_rate": 8.24209480556344e-05,
      "loss": 0.025774875283241273,
      "step": 61960
    },
    {
      "epoch": 17.590122055066704,
      "grad_norm": 0.4995296597480774,
      "learning_rate": 8.241810956571105e-05,
      "loss": 0.01796213984489441,
      "step": 61970
    },
    {
      "epoch": 17.592960544990063,
      "grad_norm": 9.558378219604492,
      "learning_rate": 8.241527107578769e-05,
      "loss": 0.012067146599292755,
      "step": 61980
    },
    {
      "epoch": 17.595799034913426,
      "grad_norm": 8.898117065429688,
      "learning_rate": 8.241243258586432e-05,
      "loss": 0.01925307810306549,
      "step": 61990
    },
    {
      "epoch": 17.598637524836786,
      "grad_norm": 1.8054234981536865,
      "learning_rate": 8.240959409594097e-05,
      "loss": 0.019535760581493377,
      "step": 62000
    },
    {
      "epoch": 17.598637524836786,
      "eval_accuracy": 0.9594328225344948,
      "eval_loss": 0.12758444249629974,
      "eval_runtime": 32.1311,
      "eval_samples_per_second": 489.463,
      "eval_steps_per_second": 7.656,
      "step": 62000
    },
    {
      "epoch": 17.60147601476015,
      "grad_norm": 7.3236775398254395,
      "learning_rate": 8.24067556060176e-05,
      "loss": 0.022417618334293364,
      "step": 62010
    },
    {
      "epoch": 17.604314504683508,
      "grad_norm": 2.7041015625,
      "learning_rate": 8.240391711609424e-05,
      "loss": 0.027331614494323732,
      "step": 62020
    },
    {
      "epoch": 17.60715299460687,
      "grad_norm": 7.156198501586914,
      "learning_rate": 8.240107862617088e-05,
      "loss": 0.029489129781723022,
      "step": 62030
    },
    {
      "epoch": 17.60999148453023,
      "grad_norm": 6.282800674438477,
      "learning_rate": 8.239824013624752e-05,
      "loss": 0.018814331293106078,
      "step": 62040
    },
    {
      "epoch": 17.61282997445359,
      "grad_norm": 11.916584968566895,
      "learning_rate": 8.239540164632415e-05,
      "loss": 0.038588258624076846,
      "step": 62050
    },
    {
      "epoch": 17.615668464376952,
      "grad_norm": 11.690528869628906,
      "learning_rate": 8.23925631564008e-05,
      "loss": 0.0323014110326767,
      "step": 62060
    },
    {
      "epoch": 17.61850695430031,
      "grad_norm": 1.9180601835250854,
      "learning_rate": 8.238972466647745e-05,
      "loss": 0.02899269163608551,
      "step": 62070
    },
    {
      "epoch": 17.621345444223675,
      "grad_norm": 1.3286105394363403,
      "learning_rate": 8.238688617655407e-05,
      "loss": 0.020975494384765626,
      "step": 62080
    },
    {
      "epoch": 17.624183934147034,
      "grad_norm": 3.077463388442993,
      "learning_rate": 8.238404768663072e-05,
      "loss": 0.019228251278400423,
      "step": 62090
    },
    {
      "epoch": 17.627022424070393,
      "grad_norm": 5.05003547668457,
      "learning_rate": 8.238120919670736e-05,
      "loss": 0.034769049286842345,
      "step": 62100
    },
    {
      "epoch": 17.629860913993756,
      "grad_norm": 1.4105678796768188,
      "learning_rate": 8.237837070678398e-05,
      "loss": 0.022635513544082643,
      "step": 62110
    },
    {
      "epoch": 17.632699403917115,
      "grad_norm": 3.652216911315918,
      "learning_rate": 8.237553221686063e-05,
      "loss": 0.04560632407665253,
      "step": 62120
    },
    {
      "epoch": 17.63553789384048,
      "grad_norm": 5.339707851409912,
      "learning_rate": 8.237269372693728e-05,
      "loss": 0.011022308468818664,
      "step": 62130
    },
    {
      "epoch": 17.638376383763838,
      "grad_norm": 2.7681965827941895,
      "learning_rate": 8.236985523701391e-05,
      "loss": 0.03116375207901001,
      "step": 62140
    },
    {
      "epoch": 17.641214873687197,
      "grad_norm": 0.3593093752861023,
      "learning_rate": 8.236701674709055e-05,
      "loss": 0.01997450739145279,
      "step": 62150
    },
    {
      "epoch": 17.64405336361056,
      "grad_norm": 5.6336588859558105,
      "learning_rate": 8.236417825716719e-05,
      "loss": 0.04349226951599121,
      "step": 62160
    },
    {
      "epoch": 17.64689185353392,
      "grad_norm": 8.167986869812012,
      "learning_rate": 8.236133976724383e-05,
      "loss": 0.021750986576080322,
      "step": 62170
    },
    {
      "epoch": 17.649730343457282,
      "grad_norm": 4.768847942352295,
      "learning_rate": 8.235850127732046e-05,
      "loss": 0.04157846570014954,
      "step": 62180
    },
    {
      "epoch": 17.65256883338064,
      "grad_norm": 10.957707405090332,
      "learning_rate": 8.23556627873971e-05,
      "loss": 0.03963684439659119,
      "step": 62190
    },
    {
      "epoch": 17.655407323304,
      "grad_norm": 4.948668956756592,
      "learning_rate": 8.235282429747376e-05,
      "loss": 0.03166250288486481,
      "step": 62200
    },
    {
      "epoch": 17.658245813227364,
      "grad_norm": 2.6129262447357178,
      "learning_rate": 8.234998580755039e-05,
      "loss": 0.011470156908035278,
      "step": 62210
    },
    {
      "epoch": 17.661084303150723,
      "grad_norm": 0.8243390917778015,
      "learning_rate": 8.234714731762703e-05,
      "loss": 0.014020107686519623,
      "step": 62220
    },
    {
      "epoch": 17.663922793074086,
      "grad_norm": 1.3467164039611816,
      "learning_rate": 8.234430882770367e-05,
      "loss": 0.01807920038700104,
      "step": 62230
    },
    {
      "epoch": 17.666761282997445,
      "grad_norm": 1.9685744047164917,
      "learning_rate": 8.23414703377803e-05,
      "loss": 0.018755415081977846,
      "step": 62240
    },
    {
      "epoch": 17.669599772920805,
      "grad_norm": 0.6170058846473694,
      "learning_rate": 8.233863184785694e-05,
      "loss": 0.018535336852073668,
      "step": 62250
    },
    {
      "epoch": 17.672438262844167,
      "grad_norm": 6.998648166656494,
      "learning_rate": 8.233579335793359e-05,
      "loss": 0.014197832345962525,
      "step": 62260
    },
    {
      "epoch": 17.675276752767527,
      "grad_norm": 11.751276016235352,
      "learning_rate": 8.233295486801022e-05,
      "loss": 0.024789305031299592,
      "step": 62270
    },
    {
      "epoch": 17.67811524269089,
      "grad_norm": 7.251413822174072,
      "learning_rate": 8.233011637808686e-05,
      "loss": 0.014999964833259582,
      "step": 62280
    },
    {
      "epoch": 17.68095373261425,
      "grad_norm": 6.438571929931641,
      "learning_rate": 8.23272778881635e-05,
      "loss": 0.01924816071987152,
      "step": 62290
    },
    {
      "epoch": 17.68379222253761,
      "grad_norm": 5.512231826782227,
      "learning_rate": 8.232443939824014e-05,
      "loss": 0.016895389556884764,
      "step": 62300
    },
    {
      "epoch": 17.68663071246097,
      "grad_norm": 2.3840599060058594,
      "learning_rate": 8.232160090831677e-05,
      "loss": 0.013796734809875488,
      "step": 62310
    },
    {
      "epoch": 17.68946920238433,
      "grad_norm": 5.094207763671875,
      "learning_rate": 8.231876241839341e-05,
      "loss": 0.02423170506954193,
      "step": 62320
    },
    {
      "epoch": 17.692307692307693,
      "grad_norm": 9.66804313659668,
      "learning_rate": 8.231592392847007e-05,
      "loss": 0.029379761219024657,
      "step": 62330
    },
    {
      "epoch": 17.695146182231053,
      "grad_norm": 3.0670926570892334,
      "learning_rate": 8.23130854385467e-05,
      "loss": 0.017469339072704315,
      "step": 62340
    },
    {
      "epoch": 17.697984672154412,
      "grad_norm": 7.919584274291992,
      "learning_rate": 8.231024694862334e-05,
      "loss": 0.040009519457817076,
      "step": 62350
    },
    {
      "epoch": 17.700823162077775,
      "grad_norm": 0.6892955303192139,
      "learning_rate": 8.230740845869998e-05,
      "loss": 0.02099369019269943,
      "step": 62360
    },
    {
      "epoch": 17.703661652001134,
      "grad_norm": 3.5605952739715576,
      "learning_rate": 8.23045699687766e-05,
      "loss": 0.02239793837070465,
      "step": 62370
    },
    {
      "epoch": 17.706500141924497,
      "grad_norm": 4.261983394622803,
      "learning_rate": 8.230173147885325e-05,
      "loss": 0.012834864854812621,
      "step": 62380
    },
    {
      "epoch": 17.709338631847857,
      "grad_norm": 3.5467593669891357,
      "learning_rate": 8.22988929889299e-05,
      "loss": 0.020592191815376283,
      "step": 62390
    },
    {
      "epoch": 17.71217712177122,
      "grad_norm": 4.0290021896362305,
      "learning_rate": 8.229605449900653e-05,
      "loss": 0.01588050276041031,
      "step": 62400
    },
    {
      "epoch": 17.71501561169458,
      "grad_norm": 5.264271259307861,
      "learning_rate": 8.229321600908317e-05,
      "loss": 0.026083704829216004,
      "step": 62410
    },
    {
      "epoch": 17.717854101617938,
      "grad_norm": 0.9289742708206177,
      "learning_rate": 8.229037751915981e-05,
      "loss": 0.020295150578022003,
      "step": 62420
    },
    {
      "epoch": 17.7206925915413,
      "grad_norm": 0.3363112211227417,
      "learning_rate": 8.228753902923645e-05,
      "loss": 0.021733446419239043,
      "step": 62430
    },
    {
      "epoch": 17.72353108146466,
      "grad_norm": 1.5515199899673462,
      "learning_rate": 8.228470053931308e-05,
      "loss": 0.03489026129245758,
      "step": 62440
    },
    {
      "epoch": 17.726369571388023,
      "grad_norm": 5.499934673309326,
      "learning_rate": 8.228186204938972e-05,
      "loss": 0.01474495530128479,
      "step": 62450
    },
    {
      "epoch": 17.729208061311382,
      "grad_norm": 9.99197006225586,
      "learning_rate": 8.227902355946638e-05,
      "loss": 0.031265202164649966,
      "step": 62460
    },
    {
      "epoch": 17.732046551234742,
      "grad_norm": 6.757200241088867,
      "learning_rate": 8.2276185069543e-05,
      "loss": 0.021450319886207582,
      "step": 62470
    },
    {
      "epoch": 17.734885041158105,
      "grad_norm": 6.411030292510986,
      "learning_rate": 8.227334657961965e-05,
      "loss": 0.009881447255611419,
      "step": 62480
    },
    {
      "epoch": 17.737723531081464,
      "grad_norm": 1.2655224800109863,
      "learning_rate": 8.227050808969629e-05,
      "loss": 0.023757727444171907,
      "step": 62490
    },
    {
      "epoch": 17.740562021004827,
      "grad_norm": 5.832747936248779,
      "learning_rate": 8.226766959977292e-05,
      "loss": 0.006980141997337342,
      "step": 62500
    },
    {
      "epoch": 17.740562021004827,
      "eval_accuracy": 0.9597507471227825,
      "eval_loss": 0.12602688372135162,
      "eval_runtime": 32.1713,
      "eval_samples_per_second": 488.852,
      "eval_steps_per_second": 7.647,
      "step": 62500
    },
    {
      "epoch": 17.743400510928186,
      "grad_norm": 0.4959080219268799,
      "learning_rate": 8.226483110984956e-05,
      "loss": 0.018655620515346527,
      "step": 62510
    },
    {
      "epoch": 17.746239000851546,
      "grad_norm": 5.3150763511657715,
      "learning_rate": 8.22619926199262e-05,
      "loss": 0.018986122310161592,
      "step": 62520
    },
    {
      "epoch": 17.74907749077491,
      "grad_norm": 10.762018203735352,
      "learning_rate": 8.225915413000284e-05,
      "loss": 0.01721494346857071,
      "step": 62530
    },
    {
      "epoch": 17.751915980698268,
      "grad_norm": 12.646697044372559,
      "learning_rate": 8.225631564007948e-05,
      "loss": 0.03775920569896698,
      "step": 62540
    },
    {
      "epoch": 17.75475447062163,
      "grad_norm": 5.9773850440979,
      "learning_rate": 8.225347715015612e-05,
      "loss": 0.02860093116760254,
      "step": 62550
    },
    {
      "epoch": 17.75759296054499,
      "grad_norm": 0.8533363938331604,
      "learning_rate": 8.225063866023277e-05,
      "loss": 0.028066220879554748,
      "step": 62560
    },
    {
      "epoch": 17.76043145046835,
      "grad_norm": 5.873940944671631,
      "learning_rate": 8.22478001703094e-05,
      "loss": 0.03162427544593811,
      "step": 62570
    },
    {
      "epoch": 17.763269940391712,
      "grad_norm": 0.8610804677009583,
      "learning_rate": 8.224496168038603e-05,
      "loss": 0.018642157316207886,
      "step": 62580
    },
    {
      "epoch": 17.76610843031507,
      "grad_norm": 5.9034953117370605,
      "learning_rate": 8.224212319046269e-05,
      "loss": 0.03293103873729706,
      "step": 62590
    },
    {
      "epoch": 17.768946920238434,
      "grad_norm": 12.166813850402832,
      "learning_rate": 8.223928470053932e-05,
      "loss": 0.026655450463294983,
      "step": 62600
    },
    {
      "epoch": 17.771785410161794,
      "grad_norm": 0.6069245338439941,
      "learning_rate": 8.223644621061596e-05,
      "loss": 0.024833826720714568,
      "step": 62610
    },
    {
      "epoch": 17.774623900085153,
      "grad_norm": 0.3689430058002472,
      "learning_rate": 8.22336077206926e-05,
      "loss": 0.025038906931877138,
      "step": 62620
    },
    {
      "epoch": 17.777462390008516,
      "grad_norm": 0.8548313975334167,
      "learning_rate": 8.223076923076923e-05,
      "loss": 0.018029791116714478,
      "step": 62630
    },
    {
      "epoch": 17.780300879931875,
      "grad_norm": 1.6226696968078613,
      "learning_rate": 8.222793074084587e-05,
      "loss": 0.028565794229507446,
      "step": 62640
    },
    {
      "epoch": 17.78313936985524,
      "grad_norm": 13.490941047668457,
      "learning_rate": 8.222509225092251e-05,
      "loss": 0.02153516560792923,
      "step": 62650
    },
    {
      "epoch": 17.785977859778598,
      "grad_norm": 9.21764850616455,
      "learning_rate": 8.222225376099915e-05,
      "loss": 0.017442944645881652,
      "step": 62660
    },
    {
      "epoch": 17.788816349701957,
      "grad_norm": 2.681166887283325,
      "learning_rate": 8.22194152710758e-05,
      "loss": 0.013702800869941712,
      "step": 62670
    },
    {
      "epoch": 17.79165483962532,
      "grad_norm": 1.2050832509994507,
      "learning_rate": 8.221657678115243e-05,
      "loss": 0.01701520085334778,
      "step": 62680
    },
    {
      "epoch": 17.79449332954868,
      "grad_norm": 2.4795150756835938,
      "learning_rate": 8.221373829122908e-05,
      "loss": 0.013615964353084565,
      "step": 62690
    },
    {
      "epoch": 17.797331819472042,
      "grad_norm": 3.0875420570373535,
      "learning_rate": 8.22108998013057e-05,
      "loss": 0.012938465178012847,
      "step": 62700
    },
    {
      "epoch": 17.8001703093954,
      "grad_norm": 11.013349533081055,
      "learning_rate": 8.220806131138235e-05,
      "loss": 0.031111732125282288,
      "step": 62710
    },
    {
      "epoch": 17.80300879931876,
      "grad_norm": 10.807975769042969,
      "learning_rate": 8.220522282145899e-05,
      "loss": 0.035066303610801694,
      "step": 62720
    },
    {
      "epoch": 17.805847289242124,
      "grad_norm": 10.930746078491211,
      "learning_rate": 8.220238433153563e-05,
      "loss": 0.013575820624828339,
      "step": 62730
    },
    {
      "epoch": 17.808685779165483,
      "grad_norm": 3.0772478580474854,
      "learning_rate": 8.219954584161227e-05,
      "loss": 0.018264994025230408,
      "step": 62740
    },
    {
      "epoch": 17.811524269088846,
      "grad_norm": 1.3442137241363525,
      "learning_rate": 8.219670735168891e-05,
      "loss": 0.02007365822792053,
      "step": 62750
    },
    {
      "epoch": 17.814362759012205,
      "grad_norm": 7.879894733428955,
      "learning_rate": 8.219386886176554e-05,
      "loss": 0.01938418745994568,
      "step": 62760
    },
    {
      "epoch": 17.817201248935568,
      "grad_norm": 1.2253329753875732,
      "learning_rate": 8.219103037184218e-05,
      "loss": 0.039791280031204225,
      "step": 62770
    },
    {
      "epoch": 17.820039738858927,
      "grad_norm": 5.614813804626465,
      "learning_rate": 8.218819188191882e-05,
      "loss": 0.015275192260742188,
      "step": 62780
    },
    {
      "epoch": 17.822878228782287,
      "grad_norm": 6.799874305725098,
      "learning_rate": 8.218535339199546e-05,
      "loss": 0.01871945261955261,
      "step": 62790
    },
    {
      "epoch": 17.82571671870565,
      "grad_norm": 7.841037750244141,
      "learning_rate": 8.21825149020721e-05,
      "loss": 0.028096556663513184,
      "step": 62800
    },
    {
      "epoch": 17.82855520862901,
      "grad_norm": 1.605942964553833,
      "learning_rate": 8.217967641214875e-05,
      "loss": 0.02815968096256256,
      "step": 62810
    },
    {
      "epoch": 17.83139369855237,
      "grad_norm": 10.052791595458984,
      "learning_rate": 8.217683792222539e-05,
      "loss": 0.06668474674224853,
      "step": 62820
    },
    {
      "epoch": 17.83423218847573,
      "grad_norm": 23.35610580444336,
      "learning_rate": 8.217399943230201e-05,
      "loss": 0.08080804347991943,
      "step": 62830
    },
    {
      "epoch": 17.83707067839909,
      "grad_norm": 10.331356048583984,
      "learning_rate": 8.217116094237866e-05,
      "loss": 0.03516271412372589,
      "step": 62840
    },
    {
      "epoch": 17.839909168322453,
      "grad_norm": 12.676920890808105,
      "learning_rate": 8.21683224524553e-05,
      "loss": 0.03169257640838623,
      "step": 62850
    },
    {
      "epoch": 17.842747658245813,
      "grad_norm": 12.567449569702148,
      "learning_rate": 8.216548396253194e-05,
      "loss": 0.018966612219810487,
      "step": 62860
    },
    {
      "epoch": 17.845586148169176,
      "grad_norm": 1.1577986478805542,
      "learning_rate": 8.216264547260858e-05,
      "loss": 0.03630762100219727,
      "step": 62870
    },
    {
      "epoch": 17.848424638092535,
      "grad_norm": 1.5280566215515137,
      "learning_rate": 8.215980698268522e-05,
      "loss": 0.02344632297754288,
      "step": 62880
    },
    {
      "epoch": 17.851263128015894,
      "grad_norm": 13.967700004577637,
      "learning_rate": 8.215696849276185e-05,
      "loss": 0.026290476322174072,
      "step": 62890
    },
    {
      "epoch": 17.854101617939257,
      "grad_norm": 4.650578022003174,
      "learning_rate": 8.215413000283849e-05,
      "loss": 0.04005018174648285,
      "step": 62900
    },
    {
      "epoch": 17.856940107862616,
      "grad_norm": 0.17366883158683777,
      "learning_rate": 8.215129151291513e-05,
      "loss": 0.008091751486063004,
      "step": 62910
    },
    {
      "epoch": 17.85977859778598,
      "grad_norm": 2.0971131324768066,
      "learning_rate": 8.214845302299177e-05,
      "loss": 0.04074928164482117,
      "step": 62920
    },
    {
      "epoch": 17.86261708770934,
      "grad_norm": 4.856375694274902,
      "learning_rate": 8.214561453306841e-05,
      "loss": 0.02438105344772339,
      "step": 62930
    },
    {
      "epoch": 17.865455577632698,
      "grad_norm": 2.2290196418762207,
      "learning_rate": 8.214277604314506e-05,
      "loss": 0.040464848279953,
      "step": 62940
    },
    {
      "epoch": 17.86829406755606,
      "grad_norm": 0.29619914293289185,
      "learning_rate": 8.213993755322168e-05,
      "loss": 0.023798368871212006,
      "step": 62950
    },
    {
      "epoch": 17.87113255747942,
      "grad_norm": 8.350001335144043,
      "learning_rate": 8.213709906329833e-05,
      "loss": 0.032443073391914365,
      "step": 62960
    },
    {
      "epoch": 17.873971047402783,
      "grad_norm": 2.495300769805908,
      "learning_rate": 8.213426057337497e-05,
      "loss": 0.025326359272003173,
      "step": 62970
    },
    {
      "epoch": 17.876809537326142,
      "grad_norm": 6.034106731414795,
      "learning_rate": 8.213142208345161e-05,
      "loss": 0.02383596748113632,
      "step": 62980
    },
    {
      "epoch": 17.8796480272495,
      "grad_norm": 10.183959007263184,
      "learning_rate": 8.212858359352825e-05,
      "loss": 0.045653107762336734,
      "step": 62990
    },
    {
      "epoch": 17.882486517172865,
      "grad_norm": 9.86632251739502,
      "learning_rate": 8.212574510360489e-05,
      "loss": 0.020216721296310424,
      "step": 63000
    },
    {
      "epoch": 17.882486517172865,
      "eval_accuracy": 0.9662364087238507,
      "eval_loss": 0.10448950529098511,
      "eval_runtime": 31.5176,
      "eval_samples_per_second": 498.991,
      "eval_steps_per_second": 7.805,
      "step": 63000
    },
    {
      "epoch": 17.885325007096224,
      "grad_norm": 5.937552452087402,
      "learning_rate": 8.212290661368153e-05,
      "loss": 0.020594467222690583,
      "step": 63010
    },
    {
      "epoch": 17.888163497019587,
      "grad_norm": 10.451207160949707,
      "learning_rate": 8.212006812375816e-05,
      "loss": 0.024574863910675048,
      "step": 63020
    },
    {
      "epoch": 17.891001986942946,
      "grad_norm": 1.551293969154358,
      "learning_rate": 8.21172296338348e-05,
      "loss": 0.024712464213371275,
      "step": 63030
    },
    {
      "epoch": 17.893840476866306,
      "grad_norm": 3.0373482704162598,
      "learning_rate": 8.211439114391144e-05,
      "loss": 0.03469589948654175,
      "step": 63040
    },
    {
      "epoch": 17.89667896678967,
      "grad_norm": 0.7828530073165894,
      "learning_rate": 8.211155265398807e-05,
      "loss": 0.017641761898994447,
      "step": 63050
    },
    {
      "epoch": 17.899517456713028,
      "grad_norm": 3.3779284954071045,
      "learning_rate": 8.210871416406473e-05,
      "loss": 0.01700296550989151,
      "step": 63060
    },
    {
      "epoch": 17.90235594663639,
      "grad_norm": 2.051417112350464,
      "learning_rate": 8.210587567414137e-05,
      "loss": 0.015218839049339294,
      "step": 63070
    },
    {
      "epoch": 17.90519443655975,
      "grad_norm": 9.505125999450684,
      "learning_rate": 8.2103037184218e-05,
      "loss": 0.04159405827522278,
      "step": 63080
    },
    {
      "epoch": 17.90803292648311,
      "grad_norm": 3.0860536098480225,
      "learning_rate": 8.210019869429464e-05,
      "loss": 0.020009826123714446,
      "step": 63090
    },
    {
      "epoch": 17.910871416406472,
      "grad_norm": 1.8598318099975586,
      "learning_rate": 8.209736020437128e-05,
      "loss": 0.0435604989528656,
      "step": 63100
    },
    {
      "epoch": 17.91370990632983,
      "grad_norm": 1.0046916007995605,
      "learning_rate": 8.209452171444792e-05,
      "loss": 0.032159548997879026,
      "step": 63110
    },
    {
      "epoch": 17.916548396253194,
      "grad_norm": 6.161089897155762,
      "learning_rate": 8.209168322452455e-05,
      "loss": 0.01534636914730072,
      "step": 63120
    },
    {
      "epoch": 17.919386886176554,
      "grad_norm": 0.5256252288818359,
      "learning_rate": 8.20888447346012e-05,
      "loss": 0.02654860317707062,
      "step": 63130
    },
    {
      "epoch": 17.922225376099917,
      "grad_norm": 5.088385105133057,
      "learning_rate": 8.208600624467784e-05,
      "loss": 0.018932031095027925,
      "step": 63140
    },
    {
      "epoch": 17.925063866023276,
      "grad_norm": 8.788341522216797,
      "learning_rate": 8.208316775475447e-05,
      "loss": 0.0427392840385437,
      "step": 63150
    },
    {
      "epoch": 17.927902355946635,
      "grad_norm": 11.259965896606445,
      "learning_rate": 8.208032926483111e-05,
      "loss": 0.019377776980400087,
      "step": 63160
    },
    {
      "epoch": 17.930740845869998,
      "grad_norm": 7.890923500061035,
      "learning_rate": 8.207749077490775e-05,
      "loss": 0.029963275790214537,
      "step": 63170
    },
    {
      "epoch": 17.933579335793358,
      "grad_norm": 1.4626001119613647,
      "learning_rate": 8.207465228498438e-05,
      "loss": 0.01323719322681427,
      "step": 63180
    },
    {
      "epoch": 17.93641782571672,
      "grad_norm": 6.506709098815918,
      "learning_rate": 8.207181379506104e-05,
      "loss": 0.017372867465019225,
      "step": 63190
    },
    {
      "epoch": 17.93925631564008,
      "grad_norm": 1.2769954204559326,
      "learning_rate": 8.206897530513768e-05,
      "loss": 0.012759195268154144,
      "step": 63200
    },
    {
      "epoch": 17.94209480556344,
      "grad_norm": 7.2475481033325195,
      "learning_rate": 8.20661368152143e-05,
      "loss": 0.015059612691402435,
      "step": 63210
    },
    {
      "epoch": 17.944933295486802,
      "grad_norm": 2.2666776180267334,
      "learning_rate": 8.206329832529095e-05,
      "loss": 0.04562419652938843,
      "step": 63220
    },
    {
      "epoch": 17.94777178541016,
      "grad_norm": 3.9278175830841064,
      "learning_rate": 8.206045983536759e-05,
      "loss": 0.041058939695358274,
      "step": 63230
    },
    {
      "epoch": 17.950610275333524,
      "grad_norm": 1.6852155923843384,
      "learning_rate": 8.205762134544423e-05,
      "loss": 0.021528658270835877,
      "step": 63240
    },
    {
      "epoch": 17.953448765256883,
      "grad_norm": 10.25830364227295,
      "learning_rate": 8.205478285552086e-05,
      "loss": 0.019893506169319154,
      "step": 63250
    },
    {
      "epoch": 17.956287255180243,
      "grad_norm": 4.412455081939697,
      "learning_rate": 8.205194436559751e-05,
      "loss": 0.015544518828392029,
      "step": 63260
    },
    {
      "epoch": 17.959125745103606,
      "grad_norm": 9.273332595825195,
      "learning_rate": 8.204910587567415e-05,
      "loss": 0.014378215372562408,
      "step": 63270
    },
    {
      "epoch": 17.961964235026965,
      "grad_norm": 0.8240272402763367,
      "learning_rate": 8.204626738575078e-05,
      "loss": 0.02420433610677719,
      "step": 63280
    },
    {
      "epoch": 17.964802724950328,
      "grad_norm": 4.694282531738281,
      "learning_rate": 8.204342889582742e-05,
      "loss": 0.017586009204387666,
      "step": 63290
    },
    {
      "epoch": 17.967641214873687,
      "grad_norm": 3.016824960708618,
      "learning_rate": 8.204059040590406e-05,
      "loss": 0.029995223879814147,
      "step": 63300
    },
    {
      "epoch": 17.970479704797047,
      "grad_norm": 11.36194133758545,
      "learning_rate": 8.203775191598069e-05,
      "loss": 0.04072842001914978,
      "step": 63310
    },
    {
      "epoch": 17.97331819472041,
      "grad_norm": 4.692788600921631,
      "learning_rate": 8.203491342605733e-05,
      "loss": 0.02683298885822296,
      "step": 63320
    },
    {
      "epoch": 17.97615668464377,
      "grad_norm": 1.0070034265518188,
      "learning_rate": 8.203207493613399e-05,
      "loss": 0.031165337562561034,
      "step": 63330
    },
    {
      "epoch": 17.97899517456713,
      "grad_norm": 0.6627662181854248,
      "learning_rate": 8.202923644621062e-05,
      "loss": 0.024088594317436218,
      "step": 63340
    },
    {
      "epoch": 17.98183366449049,
      "grad_norm": 9.662945747375488,
      "learning_rate": 8.202639795628726e-05,
      "loss": 0.03549680709838867,
      "step": 63350
    },
    {
      "epoch": 17.98467215441385,
      "grad_norm": 3.9854066371917725,
      "learning_rate": 8.20235594663639e-05,
      "loss": 0.014979708194732665,
      "step": 63360
    },
    {
      "epoch": 17.987510644337213,
      "grad_norm": 3.163055181503296,
      "learning_rate": 8.202072097644054e-05,
      "loss": 0.023752792179584502,
      "step": 63370
    },
    {
      "epoch": 17.990349134260573,
      "grad_norm": 5.344048976898193,
      "learning_rate": 8.201788248651717e-05,
      "loss": 0.012189032137393951,
      "step": 63380
    },
    {
      "epoch": 17.993187624183935,
      "grad_norm": 3.682119131088257,
      "learning_rate": 8.201504399659382e-05,
      "loss": 0.01623266488313675,
      "step": 63390
    },
    {
      "epoch": 17.996026114107295,
      "grad_norm": 8.711835861206055,
      "learning_rate": 8.201220550667046e-05,
      "loss": 0.0282275527715683,
      "step": 63400
    },
    {
      "epoch": 17.998864604030654,
      "grad_norm": 1.8471035957336426,
      "learning_rate": 8.200936701674709e-05,
      "loss": 0.01651301383972168,
      "step": 63410
    },
    {
      "epoch": 18.001703093954017,
      "grad_norm": 12.883648872375488,
      "learning_rate": 8.200652852682373e-05,
      "loss": 0.028872612118721008,
      "step": 63420
    },
    {
      "epoch": 18.004541583877376,
      "grad_norm": 11.164959907531738,
      "learning_rate": 8.200369003690038e-05,
      "loss": 0.015454404056072235,
      "step": 63430
    },
    {
      "epoch": 18.00738007380074,
      "grad_norm": 9.231427192687988,
      "learning_rate": 8.2000851546977e-05,
      "loss": 0.030927425622940062,
      "step": 63440
    },
    {
      "epoch": 18.0102185637241,
      "grad_norm": 7.0846076011657715,
      "learning_rate": 8.199801305705364e-05,
      "loss": 0.026499417424201966,
      "step": 63450
    },
    {
      "epoch": 18.013057053647458,
      "grad_norm": 2.6806135177612305,
      "learning_rate": 8.19951745671303e-05,
      "loss": 0.017173588275909424,
      "step": 63460
    },
    {
      "epoch": 18.01589554357082,
      "grad_norm": 9.149405479431152,
      "learning_rate": 8.199233607720693e-05,
      "loss": 0.0320206880569458,
      "step": 63470
    },
    {
      "epoch": 18.01873403349418,
      "grad_norm": 4.1694769859313965,
      "learning_rate": 8.198949758728357e-05,
      "loss": 0.01917231231927872,
      "step": 63480
    },
    {
      "epoch": 18.021572523417543,
      "grad_norm": 1.1832677125930786,
      "learning_rate": 8.198665909736021e-05,
      "loss": 0.024472203850746155,
      "step": 63490
    },
    {
      "epoch": 18.024411013340902,
      "grad_norm": 13.837799072265625,
      "learning_rate": 8.198382060743685e-05,
      "loss": 0.02485620677471161,
      "step": 63500
    },
    {
      "epoch": 18.024411013340902,
      "eval_accuracy": 0.9628664080880015,
      "eval_loss": 0.11966855823993683,
      "eval_runtime": 31.2298,
      "eval_samples_per_second": 503.59,
      "eval_steps_per_second": 7.877,
      "step": 63500
    },
    {
      "epoch": 18.02724950326426,
      "grad_norm": 0.33790045976638794,
      "learning_rate": 8.198098211751348e-05,
      "loss": 0.011388207972049712,
      "step": 63510
    },
    {
      "epoch": 18.030087993187625,
      "grad_norm": 6.030017375946045,
      "learning_rate": 8.197842747658246e-05,
      "loss": 0.03063642680644989,
      "step": 63520
    },
    {
      "epoch": 18.032926483110984,
      "grad_norm": 5.377939224243164,
      "learning_rate": 8.19755889866591e-05,
      "loss": 0.0260456919670105,
      "step": 63530
    },
    {
      "epoch": 18.035764973034347,
      "grad_norm": 0.9391288161277771,
      "learning_rate": 8.197275049673574e-05,
      "loss": 0.018294480443000794,
      "step": 63540
    },
    {
      "epoch": 18.038603462957706,
      "grad_norm": 10.59805965423584,
      "learning_rate": 8.196991200681238e-05,
      "loss": 0.026727724075317382,
      "step": 63550
    },
    {
      "epoch": 18.04144195288107,
      "grad_norm": 9.906193733215332,
      "learning_rate": 8.196707351688901e-05,
      "loss": 0.017150397598743438,
      "step": 63560
    },
    {
      "epoch": 18.04428044280443,
      "grad_norm": 1.0210798978805542,
      "learning_rate": 8.196423502696567e-05,
      "loss": 0.013381093740463257,
      "step": 63570
    },
    {
      "epoch": 18.047118932727788,
      "grad_norm": 8.017005920410156,
      "learning_rate": 8.196139653704231e-05,
      "loss": 0.019440135359764098,
      "step": 63580
    },
    {
      "epoch": 18.04995742265115,
      "grad_norm": 1.6258970499038696,
      "learning_rate": 8.195855804711894e-05,
      "loss": 0.01793234944343567,
      "step": 63590
    },
    {
      "epoch": 18.05279591257451,
      "grad_norm": 2.851628065109253,
      "learning_rate": 8.195571955719558e-05,
      "loss": 0.005626097694039345,
      "step": 63600
    },
    {
      "epoch": 18.055634402497873,
      "grad_norm": 8.741817474365234,
      "learning_rate": 8.195288106727222e-05,
      "loss": 0.018557682633399963,
      "step": 63610
    },
    {
      "epoch": 18.058472892421232,
      "grad_norm": 2.799314022064209,
      "learning_rate": 8.195004257734885e-05,
      "loss": 0.029077717661857606,
      "step": 63620
    },
    {
      "epoch": 18.06131138234459,
      "grad_norm": 0.7993476986885071,
      "learning_rate": 8.194720408742549e-05,
      "loss": 0.016263784468173982,
      "step": 63630
    },
    {
      "epoch": 18.064149872267954,
      "grad_norm": 7.793503284454346,
      "learning_rate": 8.194436559750214e-05,
      "loss": 0.0186955064535141,
      "step": 63640
    },
    {
      "epoch": 18.066988362191314,
      "grad_norm": 1.565314769744873,
      "learning_rate": 8.194152710757877e-05,
      "loss": 0.02875252068042755,
      "step": 63650
    },
    {
      "epoch": 18.069826852114677,
      "grad_norm": 0.37805306911468506,
      "learning_rate": 8.193868861765541e-05,
      "loss": 0.010606154799461365,
      "step": 63660
    },
    {
      "epoch": 18.072665342038036,
      "grad_norm": 3.1161158084869385,
      "learning_rate": 8.193585012773205e-05,
      "loss": 0.007773499935865402,
      "step": 63670
    },
    {
      "epoch": 18.075503831961395,
      "grad_norm": 2.411614179611206,
      "learning_rate": 8.19330116378087e-05,
      "loss": 0.02073639929294586,
      "step": 63680
    },
    {
      "epoch": 18.078342321884758,
      "grad_norm": 3.6368751525878906,
      "learning_rate": 8.193017314788532e-05,
      "loss": 0.006334617733955383,
      "step": 63690
    },
    {
      "epoch": 18.081180811808117,
      "grad_norm": 10.789783477783203,
      "learning_rate": 8.192733465796196e-05,
      "loss": 0.025414007902145385,
      "step": 63700
    },
    {
      "epoch": 18.08401930173148,
      "grad_norm": 2.523940324783325,
      "learning_rate": 8.192449616803862e-05,
      "loss": 0.01137855276465416,
      "step": 63710
    },
    {
      "epoch": 18.08685779165484,
      "grad_norm": 0.5612166523933411,
      "learning_rate": 8.192165767811525e-05,
      "loss": 0.012892305850982666,
      "step": 63720
    },
    {
      "epoch": 18.0896962815782,
      "grad_norm": 9.74189281463623,
      "learning_rate": 8.191881918819189e-05,
      "loss": 0.019851094484329222,
      "step": 63730
    },
    {
      "epoch": 18.092534771501562,
      "grad_norm": 1.339308500289917,
      "learning_rate": 8.191598069826853e-05,
      "loss": 0.012660935521125793,
      "step": 63740
    },
    {
      "epoch": 18.09537326142492,
      "grad_norm": 2.1699862480163574,
      "learning_rate": 8.191314220834516e-05,
      "loss": 0.014389137923717498,
      "step": 63750
    },
    {
      "epoch": 18.098211751348284,
      "grad_norm": 3.613295078277588,
      "learning_rate": 8.19103037184218e-05,
      "loss": 0.01403622180223465,
      "step": 63760
    },
    {
      "epoch": 18.101050241271643,
      "grad_norm": 0.926080584526062,
      "learning_rate": 8.190746522849845e-05,
      "loss": 0.011268796026706695,
      "step": 63770
    },
    {
      "epoch": 18.103888731195003,
      "grad_norm": 7.519463062286377,
      "learning_rate": 8.190462673857508e-05,
      "loss": 0.024682824313640595,
      "step": 63780
    },
    {
      "epoch": 18.106727221118366,
      "grad_norm": 6.180432319641113,
      "learning_rate": 8.190178824865172e-05,
      "loss": 0.022163230180740356,
      "step": 63790
    },
    {
      "epoch": 18.109565711041725,
      "grad_norm": 2.1202762126922607,
      "learning_rate": 8.189894975872836e-05,
      "loss": 0.013387003540992736,
      "step": 63800
    },
    {
      "epoch": 18.112404200965088,
      "grad_norm": 3.738833427429199,
      "learning_rate": 8.1896111268805e-05,
      "loss": 0.02549264132976532,
      "step": 63810
    },
    {
      "epoch": 18.115242690888447,
      "grad_norm": 5.993077278137207,
      "learning_rate": 8.189327277888163e-05,
      "loss": 0.024049943685531615,
      "step": 63820
    },
    {
      "epoch": 18.118081180811807,
      "grad_norm": 2.1975038051605225,
      "learning_rate": 8.189043428895827e-05,
      "loss": 0.014769497513771056,
      "step": 63830
    },
    {
      "epoch": 18.12091967073517,
      "grad_norm": 2.6579856872558594,
      "learning_rate": 8.188759579903492e-05,
      "loss": 0.014962542057037353,
      "step": 63840
    },
    {
      "epoch": 18.12375816065853,
      "grad_norm": 0.9042843580245972,
      "learning_rate": 8.188475730911156e-05,
      "loss": 0.02386534810066223,
      "step": 63850
    },
    {
      "epoch": 18.12659665058189,
      "grad_norm": 1.5423873662948608,
      "learning_rate": 8.18819188191882e-05,
      "loss": 0.0172099232673645,
      "step": 63860
    },
    {
      "epoch": 18.12943514050525,
      "grad_norm": 5.565915584564209,
      "learning_rate": 8.187908032926484e-05,
      "loss": 0.022585429251194,
      "step": 63870
    },
    {
      "epoch": 18.13227363042861,
      "grad_norm": 5.982133865356445,
      "learning_rate": 8.187624183934147e-05,
      "loss": 0.020662890374660493,
      "step": 63880
    },
    {
      "epoch": 18.135112120351973,
      "grad_norm": 2.7468416690826416,
      "learning_rate": 8.187340334941811e-05,
      "loss": 0.02313216179609299,
      "step": 63890
    },
    {
      "epoch": 18.137950610275333,
      "grad_norm": 4.347878932952881,
      "learning_rate": 8.187056485949475e-05,
      "loss": 0.03242829442024231,
      "step": 63900
    },
    {
      "epoch": 18.140789100198695,
      "grad_norm": 6.778963088989258,
      "learning_rate": 8.186772636957139e-05,
      "loss": 0.03919451236724854,
      "step": 63910
    },
    {
      "epoch": 18.143627590122055,
      "grad_norm": 13.00778865814209,
      "learning_rate": 8.186488787964803e-05,
      "loss": 0.0203878253698349,
      "step": 63920
    },
    {
      "epoch": 18.146466080045414,
      "grad_norm": 13.537951469421387,
      "learning_rate": 8.186204938972467e-05,
      "loss": 0.023000206053256988,
      "step": 63930
    },
    {
      "epoch": 18.149304569968777,
      "grad_norm": 6.750326156616211,
      "learning_rate": 8.18592108998013e-05,
      "loss": 0.018214805424213408,
      "step": 63940
    },
    {
      "epoch": 18.152143059892136,
      "grad_norm": 1.235145092010498,
      "learning_rate": 8.185637240987794e-05,
      "loss": 0.025742003321647645,
      "step": 63950
    },
    {
      "epoch": 18.1549815498155,
      "grad_norm": 1.247154712677002,
      "learning_rate": 8.185353391995458e-05,
      "loss": 0.010801458358764648,
      "step": 63960
    },
    {
      "epoch": 18.15782003973886,
      "grad_norm": 6.5609660148620605,
      "learning_rate": 8.185069543003123e-05,
      "loss": 0.033099958300590517,
      "step": 63970
    },
    {
      "epoch": 18.16065852966222,
      "grad_norm": 7.6868205070495605,
      "learning_rate": 8.184785694010787e-05,
      "loss": 0.015451616048812867,
      "step": 63980
    },
    {
      "epoch": 18.16349701958558,
      "grad_norm": 0.7117218971252441,
      "learning_rate": 8.184501845018451e-05,
      "loss": 0.015498486161231995,
      "step": 63990
    },
    {
      "epoch": 18.16633550950894,
      "grad_norm": 0.5023856163024902,
      "learning_rate": 8.184217996026115e-05,
      "loss": 0.019790877401828767,
      "step": 64000
    },
    {
      "epoch": 18.16633550950894,
      "eval_accuracy": 0.9602594264640427,
      "eval_loss": 0.12741507589817047,
      "eval_runtime": 31.7025,
      "eval_samples_per_second": 496.08,
      "eval_steps_per_second": 7.76,
      "step": 64000
    },
    {
      "epoch": 18.169173999432303,
      "grad_norm": 2.1337358951568604,
      "learning_rate": 8.183934147033778e-05,
      "loss": 0.01412832885980606,
      "step": 64010
    },
    {
      "epoch": 18.172012489355662,
      "grad_norm": 5.427219390869141,
      "learning_rate": 8.183650298041442e-05,
      "loss": 0.021152403950691224,
      "step": 64020
    },
    {
      "epoch": 18.174850979279025,
      "grad_norm": 1.6672537326812744,
      "learning_rate": 8.183366449049106e-05,
      "loss": 0.009095242619514466,
      "step": 64030
    },
    {
      "epoch": 18.177689469202384,
      "grad_norm": 0.5499343276023865,
      "learning_rate": 8.18308260005677e-05,
      "loss": 0.00586022213101387,
      "step": 64040
    },
    {
      "epoch": 18.180527959125744,
      "grad_norm": 1.3497560024261475,
      "learning_rate": 8.182798751064434e-05,
      "loss": 0.026206356287002564,
      "step": 64050
    },
    {
      "epoch": 18.183366449049107,
      "grad_norm": 8.766846656799316,
      "learning_rate": 8.182514902072098e-05,
      "loss": 0.022936439514160155,
      "step": 64060
    },
    {
      "epoch": 18.186204938972466,
      "grad_norm": 2.5361690521240234,
      "learning_rate": 8.182231053079761e-05,
      "loss": 0.012121029198169708,
      "step": 64070
    },
    {
      "epoch": 18.18904342889583,
      "grad_norm": 5.867077827453613,
      "learning_rate": 8.181947204087425e-05,
      "loss": 0.012035675346851349,
      "step": 64080
    },
    {
      "epoch": 18.19188191881919,
      "grad_norm": 5.768041610717773,
      "learning_rate": 8.18166335509509e-05,
      "loss": 0.015632781386375427,
      "step": 64090
    },
    {
      "epoch": 18.194720408742548,
      "grad_norm": 4.355159282684326,
      "learning_rate": 8.181379506102754e-05,
      "loss": 0.012428981065750123,
      "step": 64100
    },
    {
      "epoch": 18.19755889866591,
      "grad_norm": 1.3644630908966064,
      "learning_rate": 8.181095657110418e-05,
      "loss": 0.011507098376750947,
      "step": 64110
    },
    {
      "epoch": 18.20039738858927,
      "grad_norm": 1.4660484790802002,
      "learning_rate": 8.180811808118082e-05,
      "loss": 0.012022342532873154,
      "step": 64120
    },
    {
      "epoch": 18.203235878512633,
      "grad_norm": 5.857143402099609,
      "learning_rate": 8.180527959125746e-05,
      "loss": 0.028668275475502013,
      "step": 64130
    },
    {
      "epoch": 18.206074368435992,
      "grad_norm": 7.809715747833252,
      "learning_rate": 8.180244110133409e-05,
      "loss": 0.016119451820850374,
      "step": 64140
    },
    {
      "epoch": 18.20891285835935,
      "grad_norm": 11.552522659301758,
      "learning_rate": 8.179960261141073e-05,
      "loss": 0.022437827289104463,
      "step": 64150
    },
    {
      "epoch": 18.211751348282714,
      "grad_norm": 7.198179244995117,
      "learning_rate": 8.179676412148737e-05,
      "loss": 0.023779657483100892,
      "step": 64160
    },
    {
      "epoch": 18.214589838206074,
      "grad_norm": 0.9052576422691345,
      "learning_rate": 8.179392563156401e-05,
      "loss": 0.014769582450389862,
      "step": 64170
    },
    {
      "epoch": 18.217428328129436,
      "grad_norm": 5.4091339111328125,
      "learning_rate": 8.179108714164065e-05,
      "loss": 0.014927726984024049,
      "step": 64180
    },
    {
      "epoch": 18.220266818052796,
      "grad_norm": 13.912561416625977,
      "learning_rate": 8.17882486517173e-05,
      "loss": 0.017953263223171236,
      "step": 64190
    },
    {
      "epoch": 18.223105307976155,
      "grad_norm": 3.58343768119812,
      "learning_rate": 8.178541016179392e-05,
      "loss": 0.02186986953020096,
      "step": 64200
    },
    {
      "epoch": 18.225943797899518,
      "grad_norm": 6.5197038650512695,
      "learning_rate": 8.178257167187056e-05,
      "loss": 0.021269217133522034,
      "step": 64210
    },
    {
      "epoch": 18.228782287822877,
      "grad_norm": 0.6840488910675049,
      "learning_rate": 8.17797331819472e-05,
      "loss": 0.0069411441683769224,
      "step": 64220
    },
    {
      "epoch": 18.23162077774624,
      "grad_norm": 2.928089141845703,
      "learning_rate": 8.177689469202385e-05,
      "loss": 0.025230550765991212,
      "step": 64230
    },
    {
      "epoch": 18.2344592676696,
      "grad_norm": 10.981921195983887,
      "learning_rate": 8.177405620210049e-05,
      "loss": 0.024006845057010652,
      "step": 64240
    },
    {
      "epoch": 18.23729775759296,
      "grad_norm": 4.477245807647705,
      "learning_rate": 8.177121771217713e-05,
      "loss": 0.010527391731739045,
      "step": 64250
    },
    {
      "epoch": 18.240136247516322,
      "grad_norm": 1.1175553798675537,
      "learning_rate": 8.176837922225377e-05,
      "loss": 0.03738897442817688,
      "step": 64260
    },
    {
      "epoch": 18.24297473743968,
      "grad_norm": 2.9653480052948,
      "learning_rate": 8.17655407323304e-05,
      "loss": 0.014497557282447815,
      "step": 64270
    },
    {
      "epoch": 18.245813227363044,
      "grad_norm": 11.807784080505371,
      "learning_rate": 8.176270224240704e-05,
      "loss": 0.026821306347846983,
      "step": 64280
    },
    {
      "epoch": 18.248651717286403,
      "grad_norm": 8.044933319091797,
      "learning_rate": 8.175986375248368e-05,
      "loss": 0.016881439089775085,
      "step": 64290
    },
    {
      "epoch": 18.251490207209763,
      "grad_norm": 3.5477685928344727,
      "learning_rate": 8.175702526256032e-05,
      "loss": 0.008084752410650254,
      "step": 64300
    },
    {
      "epoch": 18.254328697133126,
      "grad_norm": 11.40770149230957,
      "learning_rate": 8.175418677263697e-05,
      "loss": 0.024506354331970216,
      "step": 64310
    },
    {
      "epoch": 18.257167187056485,
      "grad_norm": 6.7736897468566895,
      "learning_rate": 8.17513482827136e-05,
      "loss": 0.026788806915283202,
      "step": 64320
    },
    {
      "epoch": 18.260005676979848,
      "grad_norm": 5.127542495727539,
      "learning_rate": 8.174850979279023e-05,
      "loss": 0.013246673345565795,
      "step": 64330
    },
    {
      "epoch": 18.262844166903207,
      "grad_norm": 3.1091461181640625,
      "learning_rate": 8.174567130286688e-05,
      "loss": 0.027608779072761536,
      "step": 64340
    },
    {
      "epoch": 18.26568265682657,
      "grad_norm": 1.223820686340332,
      "learning_rate": 8.174283281294352e-05,
      "loss": 0.029618015885353087,
      "step": 64350
    },
    {
      "epoch": 18.26852114674993,
      "grad_norm": 1.600553035736084,
      "learning_rate": 8.173999432302016e-05,
      "loss": 0.02370784431695938,
      "step": 64360
    },
    {
      "epoch": 18.27135963667329,
      "grad_norm": 5.467772483825684,
      "learning_rate": 8.17371558330968e-05,
      "loss": 0.009500958770513535,
      "step": 64370
    },
    {
      "epoch": 18.27419812659665,
      "grad_norm": 11.989056587219238,
      "learning_rate": 8.173431734317344e-05,
      "loss": 0.02884567677974701,
      "step": 64380
    },
    {
      "epoch": 18.27703661652001,
      "grad_norm": 0.37605565786361694,
      "learning_rate": 8.173147885325008e-05,
      "loss": 0.016412146389484406,
      "step": 64390
    },
    {
      "epoch": 18.279875106443374,
      "grad_norm": 9.794072151184082,
      "learning_rate": 8.172864036332671e-05,
      "loss": 0.01564871668815613,
      "step": 64400
    },
    {
      "epoch": 18.282713596366733,
      "grad_norm": 7.608578681945801,
      "learning_rate": 8.172580187340335e-05,
      "loss": 0.05414324402809143,
      "step": 64410
    },
    {
      "epoch": 18.285552086290092,
      "grad_norm": 21.11903190612793,
      "learning_rate": 8.172296338347999e-05,
      "loss": 0.035235723853111266,
      "step": 64420
    },
    {
      "epoch": 18.288390576213455,
      "grad_norm": 9.885835647583008,
      "learning_rate": 8.172012489355662e-05,
      "loss": 0.03422803580760956,
      "step": 64430
    },
    {
      "epoch": 18.291229066136815,
      "grad_norm": 5.371190071105957,
      "learning_rate": 8.171728640363328e-05,
      "loss": 0.021089106798171997,
      "step": 64440
    },
    {
      "epoch": 18.294067556060178,
      "grad_norm": 0.27432265877723694,
      "learning_rate": 8.171444791370992e-05,
      "loss": 0.013120771944522857,
      "step": 64450
    },
    {
      "epoch": 18.296906045983537,
      "grad_norm": 2.571268081665039,
      "learning_rate": 8.171160942378655e-05,
      "loss": 0.02113320529460907,
      "step": 64460
    },
    {
      "epoch": 18.299744535906896,
      "grad_norm": 11.56346321105957,
      "learning_rate": 8.170877093386319e-05,
      "loss": 0.01973898708820343,
      "step": 64470
    },
    {
      "epoch": 18.30258302583026,
      "grad_norm": 3.8079206943511963,
      "learning_rate": 8.170593244393983e-05,
      "loss": 0.0114520862698555,
      "step": 64480
    },
    {
      "epoch": 18.30542151575362,
      "grad_norm": 10.881686210632324,
      "learning_rate": 8.170309395401647e-05,
      "loss": 0.028375059366226196,
      "step": 64490
    },
    {
      "epoch": 18.30826000567698,
      "grad_norm": 3.4686203002929688,
      "learning_rate": 8.170025546409311e-05,
      "loss": 0.03203393816947937,
      "step": 64500
    },
    {
      "epoch": 18.30826000567698,
      "eval_accuracy": 0.9640109366058371,
      "eval_loss": 0.11452274024486542,
      "eval_runtime": 31.9861,
      "eval_samples_per_second": 491.682,
      "eval_steps_per_second": 7.691,
      "step": 64500
    },
    {
      "epoch": 18.31109849560034,
      "grad_norm": 7.38455867767334,
      "learning_rate": 8.169741697416975e-05,
      "loss": 0.03204857707023621,
      "step": 64510
    },
    {
      "epoch": 18.3139369855237,
      "grad_norm": 3.299268960952759,
      "learning_rate": 8.16945784842464e-05,
      "loss": 0.020198746025562285,
      "step": 64520
    },
    {
      "epoch": 18.316775475447063,
      "grad_norm": 1.4064335823059082,
      "learning_rate": 8.169173999432302e-05,
      "loss": 0.023576533794403075,
      "step": 64530
    },
    {
      "epoch": 18.319613965370422,
      "grad_norm": 0.4955042600631714,
      "learning_rate": 8.168890150439966e-05,
      "loss": 0.0181998610496521,
      "step": 64540
    },
    {
      "epoch": 18.322452455293785,
      "grad_norm": 9.189884185791016,
      "learning_rate": 8.16860630144763e-05,
      "loss": 0.02198520749807358,
      "step": 64550
    },
    {
      "epoch": 18.325290945217144,
      "grad_norm": 2.911696672439575,
      "learning_rate": 8.168322452455293e-05,
      "loss": 0.008375386148691178,
      "step": 64560
    },
    {
      "epoch": 18.328129435140504,
      "grad_norm": 1.7371827363967896,
      "learning_rate": 8.168038603462959e-05,
      "loss": 0.020772762596607208,
      "step": 64570
    },
    {
      "epoch": 18.330967925063867,
      "grad_norm": 7.886782169342041,
      "learning_rate": 8.167754754470623e-05,
      "loss": 0.027090001106262206,
      "step": 64580
    },
    {
      "epoch": 18.333806414987226,
      "grad_norm": 2.4615044593811035,
      "learning_rate": 8.167470905478286e-05,
      "loss": 0.029046925902366637,
      "step": 64590
    },
    {
      "epoch": 18.33664490491059,
      "grad_norm": 4.161962985992432,
      "learning_rate": 8.16718705648595e-05,
      "loss": 0.016161419451236725,
      "step": 64600
    },
    {
      "epoch": 18.339483394833948,
      "grad_norm": 7.789259433746338,
      "learning_rate": 8.166903207493614e-05,
      "loss": 0.015282918512821198,
      "step": 64610
    },
    {
      "epoch": 18.342321884757308,
      "grad_norm": 3.1386585235595703,
      "learning_rate": 8.166619358501278e-05,
      "loss": 0.015604965388774872,
      "step": 64620
    },
    {
      "epoch": 18.34516037468067,
      "grad_norm": 0.9156437516212463,
      "learning_rate": 8.166335509508941e-05,
      "loss": 0.007730305194854736,
      "step": 64630
    },
    {
      "epoch": 18.34799886460403,
      "grad_norm": 0.6245124936103821,
      "learning_rate": 8.166051660516606e-05,
      "loss": 0.01292480230331421,
      "step": 64640
    },
    {
      "epoch": 18.350837354527393,
      "grad_norm": 7.491996765136719,
      "learning_rate": 8.165767811524269e-05,
      "loss": 0.018374304473400115,
      "step": 64650
    },
    {
      "epoch": 18.353675844450752,
      "grad_norm": 6.406768798828125,
      "learning_rate": 8.165483962531933e-05,
      "loss": 0.0387495219707489,
      "step": 64660
    },
    {
      "epoch": 18.35651433437411,
      "grad_norm": 0.1988161951303482,
      "learning_rate": 8.165200113539597e-05,
      "loss": 0.02027973085641861,
      "step": 64670
    },
    {
      "epoch": 18.359352824297474,
      "grad_norm": 4.635152339935303,
      "learning_rate": 8.164916264547261e-05,
      "loss": 0.013459804654121398,
      "step": 64680
    },
    {
      "epoch": 18.362191314220834,
      "grad_norm": 15.537267684936523,
      "learning_rate": 8.164632415554924e-05,
      "loss": 0.03918320536613464,
      "step": 64690
    },
    {
      "epoch": 18.365029804144196,
      "grad_norm": 2.991279125213623,
      "learning_rate": 8.16434856656259e-05,
      "loss": 0.02377077043056488,
      "step": 64700
    },
    {
      "epoch": 18.367868294067556,
      "grad_norm": 3.7001233100891113,
      "learning_rate": 8.164064717570254e-05,
      "loss": 0.01345890760421753,
      "step": 64710
    },
    {
      "epoch": 18.37070678399092,
      "grad_norm": 2.105233907699585,
      "learning_rate": 8.163780868577917e-05,
      "loss": 0.011597125977277755,
      "step": 64720
    },
    {
      "epoch": 18.373545273914278,
      "grad_norm": 2.74676513671875,
      "learning_rate": 8.163497019585581e-05,
      "loss": 0.024632632732391357,
      "step": 64730
    },
    {
      "epoch": 18.376383763837637,
      "grad_norm": 1.438855528831482,
      "learning_rate": 8.163213170593245e-05,
      "loss": 0.021962599456310274,
      "step": 64740
    },
    {
      "epoch": 18.379222253761,
      "grad_norm": 8.500609397888184,
      "learning_rate": 8.162929321600909e-05,
      "loss": 0.02144559323787689,
      "step": 64750
    },
    {
      "epoch": 18.38206074368436,
      "grad_norm": 2.9104247093200684,
      "learning_rate": 8.162645472608572e-05,
      "loss": 0.019889961183071136,
      "step": 64760
    },
    {
      "epoch": 18.384899233607722,
      "grad_norm": 0.9933539032936096,
      "learning_rate": 8.162361623616237e-05,
      "loss": 0.025526612997055054,
      "step": 64770
    },
    {
      "epoch": 18.38773772353108,
      "grad_norm": 5.836310386657715,
      "learning_rate": 8.1620777746239e-05,
      "loss": 0.012993815541267394,
      "step": 64780
    },
    {
      "epoch": 18.39057621345444,
      "grad_norm": 1.755178451538086,
      "learning_rate": 8.161793925631564e-05,
      "loss": 0.02501714825630188,
      "step": 64790
    },
    {
      "epoch": 18.393414703377804,
      "grad_norm": 14.083462715148926,
      "learning_rate": 8.161510076639228e-05,
      "loss": 0.01563563197851181,
      "step": 64800
    },
    {
      "epoch": 18.396253193301163,
      "grad_norm": 7.616891384124756,
      "learning_rate": 8.161226227646893e-05,
      "loss": 0.02151015102863312,
      "step": 64810
    },
    {
      "epoch": 18.399091683224526,
      "grad_norm": 3.3544535636901855,
      "learning_rate": 8.160942378654555e-05,
      "loss": 0.016945125162601472,
      "step": 64820
    },
    {
      "epoch": 18.401930173147885,
      "grad_norm": 6.8202595710754395,
      "learning_rate": 8.16065852966222e-05,
      "loss": 0.014925295114517212,
      "step": 64830
    },
    {
      "epoch": 18.404768663071245,
      "grad_norm": 2.5886168479919434,
      "learning_rate": 8.160374680669885e-05,
      "loss": 0.01073727160692215,
      "step": 64840
    },
    {
      "epoch": 18.407607152994608,
      "grad_norm": 8.829425811767578,
      "learning_rate": 8.160090831677548e-05,
      "loss": 0.02141517847776413,
      "step": 64850
    },
    {
      "epoch": 18.410445642917967,
      "grad_norm": 1.5394178628921509,
      "learning_rate": 8.159806982685212e-05,
      "loss": 0.028698107600212096,
      "step": 64860
    },
    {
      "epoch": 18.41328413284133,
      "grad_norm": 2.6815719604492188,
      "learning_rate": 8.159523133692876e-05,
      "loss": 0.017949622869491578,
      "step": 64870
    },
    {
      "epoch": 18.41612262276469,
      "grad_norm": 2.280712842941284,
      "learning_rate": 8.159239284700539e-05,
      "loss": 0.026585522294044494,
      "step": 64880
    },
    {
      "epoch": 18.41896111268805,
      "grad_norm": 6.153099060058594,
      "learning_rate": 8.158955435708203e-05,
      "loss": 0.03667968809604645,
      "step": 64890
    },
    {
      "epoch": 18.42179960261141,
      "grad_norm": 3.3398218154907227,
      "learning_rate": 8.158671586715868e-05,
      "loss": 0.008441707491874695,
      "step": 64900
    },
    {
      "epoch": 18.42463809253477,
      "grad_norm": 1.0757747888565063,
      "learning_rate": 8.158387737723531e-05,
      "loss": 0.02430875599384308,
      "step": 64910
    },
    {
      "epoch": 18.427476582458134,
      "grad_norm": 11.187719345092773,
      "learning_rate": 8.158103888731195e-05,
      "loss": 0.013773807883262634,
      "step": 64920
    },
    {
      "epoch": 18.430315072381493,
      "grad_norm": 6.259658336639404,
      "learning_rate": 8.15782003973886e-05,
      "loss": 0.02490702420473099,
      "step": 64930
    },
    {
      "epoch": 18.433153562304852,
      "grad_norm": 17.728290557861328,
      "learning_rate": 8.157536190746524e-05,
      "loss": 0.022846725583076478,
      "step": 64940
    },
    {
      "epoch": 18.435992052228215,
      "grad_norm": 2.0085222721099854,
      "learning_rate": 8.157252341754186e-05,
      "loss": 0.011582992970943451,
      "step": 64950
    },
    {
      "epoch": 18.438830542151575,
      "grad_norm": 1.6925238370895386,
      "learning_rate": 8.15696849276185e-05,
      "loss": 0.015410149097442627,
      "step": 64960
    },
    {
      "epoch": 18.441669032074937,
      "grad_norm": 4.890226364135742,
      "learning_rate": 8.156684643769516e-05,
      "loss": 0.03174746334552765,
      "step": 64970
    },
    {
      "epoch": 18.444507521998297,
      "grad_norm": 7.692101955413818,
      "learning_rate": 8.156400794777179e-05,
      "loss": 0.028371542692184448,
      "step": 64980
    },
    {
      "epoch": 18.447346011921656,
      "grad_norm": 2.4370031356811523,
      "learning_rate": 8.156116945784843e-05,
      "loss": 0.02862280011177063,
      "step": 64990
    },
    {
      "epoch": 18.45018450184502,
      "grad_norm": 4.500633716583252,
      "learning_rate": 8.155833096792507e-05,
      "loss": 0.03583382368087769,
      "step": 65000
    },
    {
      "epoch": 18.45018450184502,
      "eval_accuracy": 0.9598779169580975,
      "eval_loss": 0.12365498393774033,
      "eval_runtime": 32.0445,
      "eval_samples_per_second": 490.786,
      "eval_steps_per_second": 7.677,
      "step": 65000
    },
    {
      "epoch": 18.45302299176838,
      "grad_norm": 0.15921132266521454,
      "learning_rate": 8.15554924780017e-05,
      "loss": 0.01450244039297104,
      "step": 65010
    },
    {
      "epoch": 18.45586148169174,
      "grad_norm": 2.5077812671661377,
      "learning_rate": 8.155265398807834e-05,
      "loss": 0.025313353538513182,
      "step": 65020
    },
    {
      "epoch": 18.4586999716151,
      "grad_norm": 3.5951077938079834,
      "learning_rate": 8.154981549815498e-05,
      "loss": 0.019084098935127258,
      "step": 65030
    },
    {
      "epoch": 18.46153846153846,
      "grad_norm": 2.805438280105591,
      "learning_rate": 8.154697700823162e-05,
      "loss": 0.028533899784088136,
      "step": 65040
    },
    {
      "epoch": 18.464376951461823,
      "grad_norm": 8.419462203979492,
      "learning_rate": 8.154413851830826e-05,
      "loss": 0.018583706021308898,
      "step": 65050
    },
    {
      "epoch": 18.467215441385182,
      "grad_norm": 4.189601421356201,
      "learning_rate": 8.15413000283849e-05,
      "loss": 0.016246131062507628,
      "step": 65060
    },
    {
      "epoch": 18.470053931308545,
      "grad_norm": 0.6846535801887512,
      "learning_rate": 8.153846153846155e-05,
      "loss": 0.014149273931980132,
      "step": 65070
    },
    {
      "epoch": 18.472892421231904,
      "grad_norm": 1.6015697717666626,
      "learning_rate": 8.153562304853817e-05,
      "loss": 0.013975532352924347,
      "step": 65080
    },
    {
      "epoch": 18.475730911155264,
      "grad_norm": 1.1435742378234863,
      "learning_rate": 8.153278455861482e-05,
      "loss": 0.01093505248427391,
      "step": 65090
    },
    {
      "epoch": 18.478569401078627,
      "grad_norm": 1.3598769903182983,
      "learning_rate": 8.152994606869147e-05,
      "loss": 0.015113480389118195,
      "step": 65100
    },
    {
      "epoch": 18.481407891001986,
      "grad_norm": 5.738377571105957,
      "learning_rate": 8.15271075787681e-05,
      "loss": 0.024793604016304018,
      "step": 65110
    },
    {
      "epoch": 18.48424638092535,
      "grad_norm": 2.031440019607544,
      "learning_rate": 8.152426908884474e-05,
      "loss": 0.018024921417236328,
      "step": 65120
    },
    {
      "epoch": 18.487084870848708,
      "grad_norm": 4.7869462966918945,
      "learning_rate": 8.152143059892138e-05,
      "loss": 0.01986464262008667,
      "step": 65130
    },
    {
      "epoch": 18.48992336077207,
      "grad_norm": 3.7340545654296875,
      "learning_rate": 8.151859210899801e-05,
      "loss": 0.015004071593284606,
      "step": 65140
    },
    {
      "epoch": 18.49276185069543,
      "grad_norm": 2.174328565597534,
      "learning_rate": 8.151575361907465e-05,
      "loss": 0.0166971817612648,
      "step": 65150
    },
    {
      "epoch": 18.49560034061879,
      "grad_norm": 11.978513717651367,
      "learning_rate": 8.151291512915129e-05,
      "loss": 0.030484890937805174,
      "step": 65160
    },
    {
      "epoch": 18.498438830542153,
      "grad_norm": 6.7397990226745605,
      "learning_rate": 8.151007663922793e-05,
      "loss": 0.0289802223443985,
      "step": 65170
    },
    {
      "epoch": 18.501277320465512,
      "grad_norm": 2.4745960235595703,
      "learning_rate": 8.150723814930457e-05,
      "loss": 0.014822545647621154,
      "step": 65180
    },
    {
      "epoch": 18.504115810388875,
      "grad_norm": 10.138374328613281,
      "learning_rate": 8.150439965938122e-05,
      "loss": 0.023067659139633177,
      "step": 65190
    },
    {
      "epoch": 18.506954300312234,
      "grad_norm": 9.499871253967285,
      "learning_rate": 8.150156116945786e-05,
      "loss": 0.015490874648094177,
      "step": 65200
    },
    {
      "epoch": 18.509792790235593,
      "grad_norm": 16.6196231842041,
      "learning_rate": 8.149872267953449e-05,
      "loss": 0.03123834729194641,
      "step": 65210
    },
    {
      "epoch": 18.512631280158956,
      "grad_norm": 6.706479072570801,
      "learning_rate": 8.149588418961113e-05,
      "loss": 0.009177811443805695,
      "step": 65220
    },
    {
      "epoch": 18.515469770082316,
      "grad_norm": 11.436570167541504,
      "learning_rate": 8.149304569968777e-05,
      "loss": 0.025717338919639586,
      "step": 65230
    },
    {
      "epoch": 18.51830826000568,
      "grad_norm": 1.1007322072982788,
      "learning_rate": 8.149020720976441e-05,
      "loss": 0.008171102404594422,
      "step": 65240
    },
    {
      "epoch": 18.521146749929038,
      "grad_norm": 0.3315299451351166,
      "learning_rate": 8.148736871984105e-05,
      "loss": 0.006200124323368072,
      "step": 65250
    },
    {
      "epoch": 18.523985239852397,
      "grad_norm": 11.3115873336792,
      "learning_rate": 8.148453022991769e-05,
      "loss": 0.01734861433506012,
      "step": 65260
    },
    {
      "epoch": 18.52682372977576,
      "grad_norm": 19.757917404174805,
      "learning_rate": 8.148169173999432e-05,
      "loss": 0.03324038684368134,
      "step": 65270
    },
    {
      "epoch": 18.52966221969912,
      "grad_norm": 2.587345838546753,
      "learning_rate": 8.147885325007096e-05,
      "loss": 0.019647155702114106,
      "step": 65280
    },
    {
      "epoch": 18.532500709622482,
      "grad_norm": 4.350657939910889,
      "learning_rate": 8.14760147601476e-05,
      "loss": 0.018183624744415282,
      "step": 65290
    },
    {
      "epoch": 18.53533919954584,
      "grad_norm": 1.657505989074707,
      "learning_rate": 8.147317627022424e-05,
      "loss": 0.01055358499288559,
      "step": 65300
    },
    {
      "epoch": 18.5381776894692,
      "grad_norm": 1.2367732524871826,
      "learning_rate": 8.147033778030089e-05,
      "loss": 0.013584132492542266,
      "step": 65310
    },
    {
      "epoch": 18.541016179392564,
      "grad_norm": 0.5722014904022217,
      "learning_rate": 8.146749929037753e-05,
      "loss": 0.013419798016548157,
      "step": 65320
    },
    {
      "epoch": 18.543854669315923,
      "grad_norm": 3.7260754108428955,
      "learning_rate": 8.146466080045417e-05,
      "loss": 0.013349471986293793,
      "step": 65330
    },
    {
      "epoch": 18.546693159239286,
      "grad_norm": 0.7379252910614014,
      "learning_rate": 8.14618223105308e-05,
      "loss": 0.013547423481941222,
      "step": 65340
    },
    {
      "epoch": 18.549531649162645,
      "grad_norm": 7.842015743255615,
      "learning_rate": 8.145898382060744e-05,
      "loss": 0.026204794645309448,
      "step": 65350
    },
    {
      "epoch": 18.552370139086005,
      "grad_norm": 1.1814987659454346,
      "learning_rate": 8.145614533068408e-05,
      "loss": 0.019396062195301055,
      "step": 65360
    },
    {
      "epoch": 18.555208629009368,
      "grad_norm": 2.6612191200256348,
      "learning_rate": 8.145330684076072e-05,
      "loss": 0.011805374920368195,
      "step": 65370
    },
    {
      "epoch": 18.558047118932727,
      "grad_norm": 8.61876392364502,
      "learning_rate": 8.145046835083736e-05,
      "loss": 0.017999276518821716,
      "step": 65380
    },
    {
      "epoch": 18.56088560885609,
      "grad_norm": 13.031587600708008,
      "learning_rate": 8.1447629860914e-05,
      "loss": 0.028682655096054076,
      "step": 65390
    },
    {
      "epoch": 18.56372409877945,
      "grad_norm": 3.239630699157715,
      "learning_rate": 8.144479137099063e-05,
      "loss": 0.02604726850986481,
      "step": 65400
    },
    {
      "epoch": 18.56656258870281,
      "grad_norm": 5.915991306304932,
      "learning_rate": 8.144195288106727e-05,
      "loss": 0.018662264943122862,
      "step": 65410
    },
    {
      "epoch": 18.56940107862617,
      "grad_norm": 3.464056968688965,
      "learning_rate": 8.143911439114391e-05,
      "loss": 0.013923452794551849,
      "step": 65420
    },
    {
      "epoch": 18.57223956854953,
      "grad_norm": 1.9145278930664062,
      "learning_rate": 8.143627590122056e-05,
      "loss": 0.01487433910369873,
      "step": 65430
    },
    {
      "epoch": 18.575078058472894,
      "grad_norm": 13.171759605407715,
      "learning_rate": 8.14334374112972e-05,
      "loss": 0.011693964898586272,
      "step": 65440
    },
    {
      "epoch": 18.577916548396253,
      "grad_norm": 10.689671516418457,
      "learning_rate": 8.143059892137384e-05,
      "loss": 0.020308999717235564,
      "step": 65450
    },
    {
      "epoch": 18.580755038319612,
      "grad_norm": 5.099935531616211,
      "learning_rate": 8.142776043145048e-05,
      "loss": 0.018070951104164124,
      "step": 65460
    },
    {
      "epoch": 18.583593528242975,
      "grad_norm": 1.1860604286193848,
      "learning_rate": 8.142492194152711e-05,
      "loss": 0.007710698246955872,
      "step": 65470
    },
    {
      "epoch": 18.586432018166335,
      "grad_norm": 5.3310723304748535,
      "learning_rate": 8.142208345160375e-05,
      "loss": 0.01653881072998047,
      "step": 65480
    },
    {
      "epoch": 18.589270508089697,
      "grad_norm": 3.0680739879608154,
      "learning_rate": 8.141924496168039e-05,
      "loss": 0.020436204969882965,
      "step": 65490
    },
    {
      "epoch": 18.592108998013057,
      "grad_norm": 0.19042538106441498,
      "learning_rate": 8.141640647175703e-05,
      "loss": 0.020362281799316408,
      "step": 65500
    },
    {
      "epoch": 18.592108998013057,
      "eval_accuracy": 0.9671265975710561,
      "eval_loss": 0.10823927074670792,
      "eval_runtime": 32.0506,
      "eval_samples_per_second": 490.692,
      "eval_steps_per_second": 7.675,
      "step": 65500
    },
    {
      "epoch": 18.594947487936416,
      "grad_norm": 0.9644738435745239,
      "learning_rate": 8.141356798183367e-05,
      "loss": 0.019485700130462646,
      "step": 65510
    },
    {
      "epoch": 18.59778597785978,
      "grad_norm": 0.5033319592475891,
      "learning_rate": 8.141072949191031e-05,
      "loss": 0.013396617770195008,
      "step": 65520
    },
    {
      "epoch": 18.60062446778314,
      "grad_norm": 3.123328924179077,
      "learning_rate": 8.140789100198694e-05,
      "loss": 0.011749409139156342,
      "step": 65530
    },
    {
      "epoch": 18.6034629577065,
      "grad_norm": 15.192595481872559,
      "learning_rate": 8.140505251206358e-05,
      "loss": 0.031108838319778443,
      "step": 65540
    },
    {
      "epoch": 18.60630144762986,
      "grad_norm": 0.7971093654632568,
      "learning_rate": 8.140221402214022e-05,
      "loss": 0.008194833993911743,
      "step": 65550
    },
    {
      "epoch": 18.609139937553223,
      "grad_norm": 2.494683027267456,
      "learning_rate": 8.139937553221687e-05,
      "loss": 0.017994403839111328,
      "step": 65560
    },
    {
      "epoch": 18.611978427476583,
      "grad_norm": 3.4145376682281494,
      "learning_rate": 8.139653704229351e-05,
      "loss": 0.012980231642723083,
      "step": 65570
    },
    {
      "epoch": 18.614816917399942,
      "grad_norm": 13.31275749206543,
      "learning_rate": 8.139369855237015e-05,
      "loss": 0.011567264050245284,
      "step": 65580
    },
    {
      "epoch": 18.617655407323305,
      "grad_norm": 3.671393632888794,
      "learning_rate": 8.139086006244678e-05,
      "loss": 0.03883555829524994,
      "step": 65590
    },
    {
      "epoch": 18.620493897246664,
      "grad_norm": 21.00157928466797,
      "learning_rate": 8.138802157252342e-05,
      "loss": 0.034800782799720764,
      "step": 65600
    },
    {
      "epoch": 18.623332387170027,
      "grad_norm": 13.980853080749512,
      "learning_rate": 8.138518308260006e-05,
      "loss": 0.02740713059902191,
      "step": 65610
    },
    {
      "epoch": 18.626170877093386,
      "grad_norm": 2.1741952896118164,
      "learning_rate": 8.13823445926767e-05,
      "loss": 0.027539867162704467,
      "step": 65620
    },
    {
      "epoch": 18.629009367016746,
      "grad_norm": 1.7575544118881226,
      "learning_rate": 8.137950610275334e-05,
      "loss": 0.019827230274677275,
      "step": 65630
    },
    {
      "epoch": 18.63184785694011,
      "grad_norm": 12.440897941589355,
      "learning_rate": 8.137666761282998e-05,
      "loss": 0.033045870065689084,
      "step": 65640
    },
    {
      "epoch": 18.634686346863468,
      "grad_norm": 8.510391235351562,
      "learning_rate": 8.137382912290662e-05,
      "loss": 0.021538083255290986,
      "step": 65650
    },
    {
      "epoch": 18.63752483678683,
      "grad_norm": 2.480590343475342,
      "learning_rate": 8.137099063298325e-05,
      "loss": 0.012605838477611542,
      "step": 65660
    },
    {
      "epoch": 18.64036332671019,
      "grad_norm": 2.146552801132202,
      "learning_rate": 8.13681521430599e-05,
      "loss": 0.010300079733133316,
      "step": 65670
    },
    {
      "epoch": 18.64320181663355,
      "grad_norm": 12.616165161132812,
      "learning_rate": 8.136531365313654e-05,
      "loss": 0.017979632318019866,
      "step": 65680
    },
    {
      "epoch": 18.646040306556912,
      "grad_norm": 1.010509729385376,
      "learning_rate": 8.136247516321316e-05,
      "loss": 0.025213581323623658,
      "step": 65690
    },
    {
      "epoch": 18.648878796480272,
      "grad_norm": 0.7157605290412903,
      "learning_rate": 8.135963667328982e-05,
      "loss": 0.015295769274234771,
      "step": 65700
    },
    {
      "epoch": 18.651717286403635,
      "grad_norm": 3.0977420806884766,
      "learning_rate": 8.135679818336646e-05,
      "loss": 0.009039697796106338,
      "step": 65710
    },
    {
      "epoch": 18.654555776326994,
      "grad_norm": 1.2158349752426147,
      "learning_rate": 8.135395969344309e-05,
      "loss": 0.011754780262708663,
      "step": 65720
    },
    {
      "epoch": 18.657394266250353,
      "grad_norm": 4.4673638343811035,
      "learning_rate": 8.135112120351973e-05,
      "loss": 0.0157989576458931,
      "step": 65730
    },
    {
      "epoch": 18.660232756173716,
      "grad_norm": 0.3597680628299713,
      "learning_rate": 8.134828271359637e-05,
      "loss": 0.020639072358608245,
      "step": 65740
    },
    {
      "epoch": 18.663071246097076,
      "grad_norm": 3.230433702468872,
      "learning_rate": 8.134544422367301e-05,
      "loss": 0.009913558512926102,
      "step": 65750
    },
    {
      "epoch": 18.66590973602044,
      "grad_norm": 2.2301106452941895,
      "learning_rate": 8.134260573374964e-05,
      "loss": 0.019906213879585265,
      "step": 65760
    },
    {
      "epoch": 18.668748225943798,
      "grad_norm": 11.6491117477417,
      "learning_rate": 8.13397672438263e-05,
      "loss": 0.028836414217948914,
      "step": 65770
    },
    {
      "epoch": 18.671586715867157,
      "grad_norm": 3.1171281337738037,
      "learning_rate": 8.133692875390294e-05,
      "loss": 0.009900050610303879,
      "step": 65780
    },
    {
      "epoch": 18.67442520579052,
      "grad_norm": 0.781021773815155,
      "learning_rate": 8.133409026397956e-05,
      "loss": 0.03698730170726776,
      "step": 65790
    },
    {
      "epoch": 18.67726369571388,
      "grad_norm": 10.685922622680664,
      "learning_rate": 8.13312517740562e-05,
      "loss": 0.030243587493896485,
      "step": 65800
    },
    {
      "epoch": 18.680102185637242,
      "grad_norm": 5.416703224182129,
      "learning_rate": 8.132841328413285e-05,
      "loss": 0.01442263126373291,
      "step": 65810
    },
    {
      "epoch": 18.6829406755606,
      "grad_norm": 7.635152339935303,
      "learning_rate": 8.132557479420947e-05,
      "loss": 0.025737428665161134,
      "step": 65820
    },
    {
      "epoch": 18.68577916548396,
      "grad_norm": 0.8116246461868286,
      "learning_rate": 8.132273630428613e-05,
      "loss": 0.01686970591545105,
      "step": 65830
    },
    {
      "epoch": 18.688617655407324,
      "grad_norm": 0.4055250287055969,
      "learning_rate": 8.131989781436277e-05,
      "loss": 0.014805729687213897,
      "step": 65840
    },
    {
      "epoch": 18.691456145330683,
      "grad_norm": 4.816773891448975,
      "learning_rate": 8.13170593244394e-05,
      "loss": 0.011856859922409058,
      "step": 65850
    },
    {
      "epoch": 18.694294635254046,
      "grad_norm": 3.883345603942871,
      "learning_rate": 8.131422083451604e-05,
      "loss": 0.015941402316093443,
      "step": 65860
    },
    {
      "epoch": 18.697133125177405,
      "grad_norm": 1.9226481914520264,
      "learning_rate": 8.131138234459268e-05,
      "loss": 0.011912903189659119,
      "step": 65870
    },
    {
      "epoch": 18.699971615100765,
      "grad_norm": 2.6523337364196777,
      "learning_rate": 8.130854385466932e-05,
      "loss": 0.009482324868440629,
      "step": 65880
    },
    {
      "epoch": 18.702810105024128,
      "grad_norm": 3.491548776626587,
      "learning_rate": 8.130570536474595e-05,
      "loss": 0.02799041271209717,
      "step": 65890
    },
    {
      "epoch": 18.705648594947487,
      "grad_norm": 1.8625236749649048,
      "learning_rate": 8.13028668748226e-05,
      "loss": 0.007141588628292084,
      "step": 65900
    },
    {
      "epoch": 18.70848708487085,
      "grad_norm": 10.894309997558594,
      "learning_rate": 8.130002838489925e-05,
      "loss": 0.020209994912147523,
      "step": 65910
    },
    {
      "epoch": 18.71132557479421,
      "grad_norm": 5.925015926361084,
      "learning_rate": 8.129718989497587e-05,
      "loss": 0.010579054057598115,
      "step": 65920
    },
    {
      "epoch": 18.714164064717572,
      "grad_norm": 6.101390361785889,
      "learning_rate": 8.129435140505252e-05,
      "loss": 0.010245073586702347,
      "step": 65930
    },
    {
      "epoch": 18.71700255464093,
      "grad_norm": 4.651159763336182,
      "learning_rate": 8.129151291512916e-05,
      "loss": 0.017720875144004822,
      "step": 65940
    },
    {
      "epoch": 18.71984104456429,
      "grad_norm": 4.947421073913574,
      "learning_rate": 8.128867442520578e-05,
      "loss": 0.018265710771083833,
      "step": 65950
    },
    {
      "epoch": 18.722679534487654,
      "grad_norm": 7.7095746994018555,
      "learning_rate": 8.128583593528243e-05,
      "loss": 0.01981608271598816,
      "step": 65960
    },
    {
      "epoch": 18.725518024411013,
      "grad_norm": 1.5604537725448608,
      "learning_rate": 8.128299744535908e-05,
      "loss": 0.015255461633205413,
      "step": 65970
    },
    {
      "epoch": 18.728356514334376,
      "grad_norm": 4.790753364562988,
      "learning_rate": 8.128015895543571e-05,
      "loss": 0.024638724327087403,
      "step": 65980
    },
    {
      "epoch": 18.731195004257735,
      "grad_norm": 1.3473572731018066,
      "learning_rate": 8.127732046551235e-05,
      "loss": 0.014015762507915497,
      "step": 65990
    },
    {
      "epoch": 18.734033494181094,
      "grad_norm": 1.5037555694580078,
      "learning_rate": 8.127476582458132e-05,
      "loss": 0.05017166137695313,
      "step": 66000
    },
    {
      "epoch": 18.734033494181094,
      "eval_accuracy": 0.964837540535385,
      "eval_loss": 0.11509203910827637,
      "eval_runtime": 31.6327,
      "eval_samples_per_second": 497.176,
      "eval_steps_per_second": 7.777,
      "step": 66000
    },
    {
      "epoch": 18.736871984104457,
      "grad_norm": 17.559885025024414,
      "learning_rate": 8.127192733465797e-05,
      "loss": 0.036193746328353885,
      "step": 66010
    },
    {
      "epoch": 18.739710474027817,
      "grad_norm": 18.118698120117188,
      "learning_rate": 8.126908884473461e-05,
      "loss": 0.023644256591796874,
      "step": 66020
    },
    {
      "epoch": 18.74254896395118,
      "grad_norm": 2.231119155883789,
      "learning_rate": 8.126625035481124e-05,
      "loss": 0.045569732785224915,
      "step": 66030
    },
    {
      "epoch": 18.74538745387454,
      "grad_norm": 1.703681468963623,
      "learning_rate": 8.126341186488788e-05,
      "loss": 0.012719261646270751,
      "step": 66040
    },
    {
      "epoch": 18.748225943797898,
      "grad_norm": 5.024110317230225,
      "learning_rate": 8.126057337496452e-05,
      "loss": 0.012549507617950439,
      "step": 66050
    },
    {
      "epoch": 18.75106443372126,
      "grad_norm": 3.1849050521850586,
      "learning_rate": 8.125773488504116e-05,
      "loss": 0.01445792019367218,
      "step": 66060
    },
    {
      "epoch": 18.75390292364462,
      "grad_norm": 9.782297134399414,
      "learning_rate": 8.125489639511779e-05,
      "loss": 0.01607317179441452,
      "step": 66070
    },
    {
      "epoch": 18.756741413567983,
      "grad_norm": 2.452918291091919,
      "learning_rate": 8.125205790519445e-05,
      "loss": 0.03378154039382934,
      "step": 66080
    },
    {
      "epoch": 18.759579903491343,
      "grad_norm": 2.312612295150757,
      "learning_rate": 8.124921941527109e-05,
      "loss": 0.012735034525394439,
      "step": 66090
    },
    {
      "epoch": 18.762418393414702,
      "grad_norm": 7.722110271453857,
      "learning_rate": 8.124638092534772e-05,
      "loss": 0.021268382668495178,
      "step": 66100
    },
    {
      "epoch": 18.765256883338065,
      "grad_norm": 11.429648399353027,
      "learning_rate": 8.124354243542436e-05,
      "loss": 0.020001563429832458,
      "step": 66110
    },
    {
      "epoch": 18.768095373261424,
      "grad_norm": 2.895681858062744,
      "learning_rate": 8.1240703945501e-05,
      "loss": 0.02193840593099594,
      "step": 66120
    },
    {
      "epoch": 18.770933863184787,
      "grad_norm": 7.935652256011963,
      "learning_rate": 8.123786545557763e-05,
      "loss": 0.04819098711013794,
      "step": 66130
    },
    {
      "epoch": 18.773772353108146,
      "grad_norm": 11.845431327819824,
      "learning_rate": 8.123502696565427e-05,
      "loss": 0.021462681889533996,
      "step": 66140
    },
    {
      "epoch": 18.776610843031506,
      "grad_norm": 2.0651376247406006,
      "learning_rate": 8.123218847573092e-05,
      "loss": 0.018990828096866606,
      "step": 66150
    },
    {
      "epoch": 18.77944933295487,
      "grad_norm": 2.8424670696258545,
      "learning_rate": 8.122934998580755e-05,
      "loss": 0.02519553303718567,
      "step": 66160
    },
    {
      "epoch": 18.782287822878228,
      "grad_norm": 0.7492388486862183,
      "learning_rate": 8.122651149588419e-05,
      "loss": 0.006743628531694412,
      "step": 66170
    },
    {
      "epoch": 18.78512631280159,
      "grad_norm": 8.703424453735352,
      "learning_rate": 8.122367300596083e-05,
      "loss": 0.014591945707798004,
      "step": 66180
    },
    {
      "epoch": 18.78796480272495,
      "grad_norm": 1.5100408792495728,
      "learning_rate": 8.122083451603748e-05,
      "loss": 0.010236608982086181,
      "step": 66190
    },
    {
      "epoch": 18.79080329264831,
      "grad_norm": 7.783727645874023,
      "learning_rate": 8.12179960261141e-05,
      "loss": 0.019305528700351716,
      "step": 66200
    },
    {
      "epoch": 18.793641782571672,
      "grad_norm": 6.598199367523193,
      "learning_rate": 8.121515753619076e-05,
      "loss": 0.01837824434041977,
      "step": 66210
    },
    {
      "epoch": 18.79648027249503,
      "grad_norm": 5.368304252624512,
      "learning_rate": 8.12123190462674e-05,
      "loss": 0.027355998754501343,
      "step": 66220
    },
    {
      "epoch": 18.799318762418395,
      "grad_norm": 3.307619333267212,
      "learning_rate": 8.120948055634403e-05,
      "loss": 0.0271921843290329,
      "step": 66230
    },
    {
      "epoch": 18.802157252341754,
      "grad_norm": 2.775135040283203,
      "learning_rate": 8.120664206642067e-05,
      "loss": 0.025950881838798522,
      "step": 66240
    },
    {
      "epoch": 18.804995742265113,
      "grad_norm": 5.855447769165039,
      "learning_rate": 8.120380357649731e-05,
      "loss": 0.036981263756752016,
      "step": 66250
    },
    {
      "epoch": 18.807834232188476,
      "grad_norm": 7.363571643829346,
      "learning_rate": 8.120096508657394e-05,
      "loss": 0.04288973212242127,
      "step": 66260
    },
    {
      "epoch": 18.810672722111835,
      "grad_norm": 4.994757175445557,
      "learning_rate": 8.119812659665058e-05,
      "loss": 0.01854916363954544,
      "step": 66270
    },
    {
      "epoch": 18.8135112120352,
      "grad_norm": 6.26474142074585,
      "learning_rate": 8.119528810672723e-05,
      "loss": 0.012716639041900634,
      "step": 66280
    },
    {
      "epoch": 18.816349701958558,
      "grad_norm": 2.603522539138794,
      "learning_rate": 8.119244961680386e-05,
      "loss": 0.02817658185958862,
      "step": 66290
    },
    {
      "epoch": 18.81918819188192,
      "grad_norm": 7.4962639808654785,
      "learning_rate": 8.11896111268805e-05,
      "loss": 0.03160146474838257,
      "step": 66300
    },
    {
      "epoch": 18.82202668180528,
      "grad_norm": 4.705212116241455,
      "learning_rate": 8.118677263695714e-05,
      "loss": 0.023783554136753083,
      "step": 66310
    },
    {
      "epoch": 18.82486517172864,
      "grad_norm": 3.021087169647217,
      "learning_rate": 8.118393414703379e-05,
      "loss": 0.013346585631370544,
      "step": 66320
    },
    {
      "epoch": 18.827703661652002,
      "grad_norm": 3.620481252670288,
      "learning_rate": 8.118109565711041e-05,
      "loss": 0.030485108494758606,
      "step": 66330
    },
    {
      "epoch": 18.83054215157536,
      "grad_norm": 8.69963264465332,
      "learning_rate": 8.117825716718706e-05,
      "loss": 0.02241632789373398,
      "step": 66340
    },
    {
      "epoch": 18.833380641498724,
      "grad_norm": 2.1544408798217773,
      "learning_rate": 8.117541867726371e-05,
      "loss": 0.037103688716888426,
      "step": 66350
    },
    {
      "epoch": 18.836219131422084,
      "grad_norm": 0.8527352809906006,
      "learning_rate": 8.117258018734034e-05,
      "loss": 0.02223041355609894,
      "step": 66360
    },
    {
      "epoch": 18.839057621345443,
      "grad_norm": 0.90654456615448,
      "learning_rate": 8.116974169741698e-05,
      "loss": 0.014562907814979553,
      "step": 66370
    },
    {
      "epoch": 18.841896111268806,
      "grad_norm": 3.5208656787872314,
      "learning_rate": 8.116690320749362e-05,
      "loss": 0.00752825140953064,
      "step": 66380
    },
    {
      "epoch": 18.844734601192165,
      "grad_norm": 1.0482949018478394,
      "learning_rate": 8.116406471757025e-05,
      "loss": 0.010385850816965103,
      "step": 66390
    },
    {
      "epoch": 18.847573091115528,
      "grad_norm": 4.282619476318359,
      "learning_rate": 8.116122622764689e-05,
      "loss": 0.019163095951080324,
      "step": 66400
    },
    {
      "epoch": 18.850411581038887,
      "grad_norm": 11.027481079101562,
      "learning_rate": 8.115838773772355e-05,
      "loss": 0.03325257301330566,
      "step": 66410
    },
    {
      "epoch": 18.853250070962247,
      "grad_norm": 1.6965198516845703,
      "learning_rate": 8.115554924780017e-05,
      "loss": 0.016721759736537934,
      "step": 66420
    },
    {
      "epoch": 18.85608856088561,
      "grad_norm": 1.1206939220428467,
      "learning_rate": 8.115271075787681e-05,
      "loss": 0.0070926487445831295,
      "step": 66430
    },
    {
      "epoch": 18.85892705080897,
      "grad_norm": 0.26779115200042725,
      "learning_rate": 8.114987226795346e-05,
      "loss": 0.025253993272781373,
      "step": 66440
    },
    {
      "epoch": 18.861765540732332,
      "grad_norm": 0.7530791163444519,
      "learning_rate": 8.11470337780301e-05,
      "loss": 0.020818400382995605,
      "step": 66450
    },
    {
      "epoch": 18.86460403065569,
      "grad_norm": 3.3261451721191406,
      "learning_rate": 8.114419528810672e-05,
      "loss": 0.022585004568099976,
      "step": 66460
    },
    {
      "epoch": 18.86744252057905,
      "grad_norm": 6.81146240234375,
      "learning_rate": 8.114135679818337e-05,
      "loss": 0.016673743724822998,
      "step": 66470
    },
    {
      "epoch": 18.870281010502413,
      "grad_norm": 1.1730128526687622,
      "learning_rate": 8.113851830826001e-05,
      "loss": 0.015111240744590759,
      "step": 66480
    },
    {
      "epoch": 18.873119500425773,
      "grad_norm": 6.115340232849121,
      "learning_rate": 8.113567981833665e-05,
      "loss": 0.021538375318050383,
      "step": 66490
    },
    {
      "epoch": 18.875957990349136,
      "grad_norm": 2.6992621421813965,
      "learning_rate": 8.113284132841329e-05,
      "loss": 0.020711855590343477,
      "step": 66500
    },
    {
      "epoch": 18.875957990349136,
      "eval_accuracy": 0.9640109366058371,
      "eval_loss": 0.12057431787252426,
      "eval_runtime": 31.6811,
      "eval_samples_per_second": 496.415,
      "eval_steps_per_second": 7.765,
      "step": 66500
    },
    {
      "epoch": 18.878796480272495,
      "grad_norm": 9.589396476745605,
      "learning_rate": 8.113000283848993e-05,
      "loss": 0.047894620895385744,
      "step": 66510
    },
    {
      "epoch": 18.881634970195854,
      "grad_norm": 1.3040424585342407,
      "learning_rate": 8.112716434856656e-05,
      "loss": 0.026215952634811402,
      "step": 66520
    },
    {
      "epoch": 18.884473460119217,
      "grad_norm": 4.642782688140869,
      "learning_rate": 8.11243258586432e-05,
      "loss": 0.0335545539855957,
      "step": 66530
    },
    {
      "epoch": 18.887311950042577,
      "grad_norm": 15.508852005004883,
      "learning_rate": 8.112148736871984e-05,
      "loss": 0.0157991424202919,
      "step": 66540
    },
    {
      "epoch": 18.89015043996594,
      "grad_norm": 5.99678373336792,
      "learning_rate": 8.111864887879648e-05,
      "loss": 0.03736512362957001,
      "step": 66550
    },
    {
      "epoch": 18.8929889298893,
      "grad_norm": 4.121882438659668,
      "learning_rate": 8.111581038887313e-05,
      "loss": 0.02517453134059906,
      "step": 66560
    },
    {
      "epoch": 18.895827419812658,
      "grad_norm": 3.6094770431518555,
      "learning_rate": 8.111297189894977e-05,
      "loss": 0.027205502986907958,
      "step": 66570
    },
    {
      "epoch": 18.89866590973602,
      "grad_norm": 12.144388198852539,
      "learning_rate": 8.11101334090264e-05,
      "loss": 0.020541377365589142,
      "step": 66580
    },
    {
      "epoch": 18.90150439965938,
      "grad_norm": 1.298830509185791,
      "learning_rate": 8.110729491910304e-05,
      "loss": 0.03413808345794678,
      "step": 66590
    },
    {
      "epoch": 18.904342889582743,
      "grad_norm": 7.557719707489014,
      "learning_rate": 8.110445642917968e-05,
      "loss": 0.01710568219423294,
      "step": 66600
    },
    {
      "epoch": 18.907181379506103,
      "grad_norm": 3.173307418823242,
      "learning_rate": 8.110161793925632e-05,
      "loss": 0.02641441524028778,
      "step": 66610
    },
    {
      "epoch": 18.910019869429462,
      "grad_norm": 0.7265740633010864,
      "learning_rate": 8.109877944933296e-05,
      "loss": 0.012198587507009506,
      "step": 66620
    },
    {
      "epoch": 18.912858359352825,
      "grad_norm": 17.35437774658203,
      "learning_rate": 8.10959409594096e-05,
      "loss": 0.030594530701637267,
      "step": 66630
    },
    {
      "epoch": 18.915696849276184,
      "grad_norm": 3.9852724075317383,
      "learning_rate": 8.109310246948624e-05,
      "loss": 0.02220355123281479,
      "step": 66640
    },
    {
      "epoch": 18.918535339199547,
      "grad_norm": 0.9912310242652893,
      "learning_rate": 8.109026397956287e-05,
      "loss": 0.02435201108455658,
      "step": 66650
    },
    {
      "epoch": 18.921373829122906,
      "grad_norm": 7.441756725311279,
      "learning_rate": 8.108742548963951e-05,
      "loss": 0.039067605137825014,
      "step": 66660
    },
    {
      "epoch": 18.92421231904627,
      "grad_norm": 1.6327534914016724,
      "learning_rate": 8.108458699971615e-05,
      "loss": 0.023820698261260986,
      "step": 66670
    },
    {
      "epoch": 18.92705080896963,
      "grad_norm": 5.430531024932861,
      "learning_rate": 8.10817485097928e-05,
      "loss": 0.024405109882354736,
      "step": 66680
    },
    {
      "epoch": 18.929889298892988,
      "grad_norm": 13.121145248413086,
      "learning_rate": 8.107891001986944e-05,
      "loss": 0.02522510290145874,
      "step": 66690
    },
    {
      "epoch": 18.93272778881635,
      "grad_norm": 0.9832379221916199,
      "learning_rate": 8.107607152994608e-05,
      "loss": 0.01745489239692688,
      "step": 66700
    },
    {
      "epoch": 18.93556627873971,
      "grad_norm": 17.05642318725586,
      "learning_rate": 8.10732330400227e-05,
      "loss": 0.02806151509284973,
      "step": 66710
    },
    {
      "epoch": 18.93840476866307,
      "grad_norm": 0.6314168572425842,
      "learning_rate": 8.107039455009935e-05,
      "loss": 0.03050752282142639,
      "step": 66720
    },
    {
      "epoch": 18.941243258586432,
      "grad_norm": 2.886826515197754,
      "learning_rate": 8.106755606017599e-05,
      "loss": 0.01392044723033905,
      "step": 66730
    },
    {
      "epoch": 18.94408174850979,
      "grad_norm": 1.569759488105774,
      "learning_rate": 8.106471757025263e-05,
      "loss": 0.016734452545642854,
      "step": 66740
    },
    {
      "epoch": 18.946920238433155,
      "grad_norm": 9.503257751464844,
      "learning_rate": 8.106187908032927e-05,
      "loss": 0.01802990436553955,
      "step": 66750
    },
    {
      "epoch": 18.949758728356514,
      "grad_norm": 2.374508857727051,
      "learning_rate": 8.105904059040591e-05,
      "loss": 0.030473613739013673,
      "step": 66760
    },
    {
      "epoch": 18.952597218279877,
      "grad_norm": 2.867720127105713,
      "learning_rate": 8.105620210048255e-05,
      "loss": 0.02514689564704895,
      "step": 66770
    },
    {
      "epoch": 18.955435708203236,
      "grad_norm": 10.971470832824707,
      "learning_rate": 8.105336361055918e-05,
      "loss": 0.021542105078697204,
      "step": 66780
    },
    {
      "epoch": 18.958274198126595,
      "grad_norm": 3.179478406906128,
      "learning_rate": 8.105052512063582e-05,
      "loss": 0.021673890948295593,
      "step": 66790
    },
    {
      "epoch": 18.96111268804996,
      "grad_norm": 4.309968948364258,
      "learning_rate": 8.104768663071246e-05,
      "loss": 0.014268122613430023,
      "step": 66800
    },
    {
      "epoch": 18.963951177973318,
      "grad_norm": 7.718191146850586,
      "learning_rate": 8.10448481407891e-05,
      "loss": 0.02030482143163681,
      "step": 66810
    },
    {
      "epoch": 18.96678966789668,
      "grad_norm": 2.393521308898926,
      "learning_rate": 8.104200965086575e-05,
      "loss": 0.01478860229253769,
      "step": 66820
    },
    {
      "epoch": 18.96962815782004,
      "grad_norm": 7.92020845413208,
      "learning_rate": 8.103917116094239e-05,
      "loss": 0.011255685240030289,
      "step": 66830
    },
    {
      "epoch": 18.9724666477434,
      "grad_norm": 14.875160217285156,
      "learning_rate": 8.103633267101902e-05,
      "loss": 0.026578035950660706,
      "step": 66840
    },
    {
      "epoch": 18.975305137666762,
      "grad_norm": 0.6984143853187561,
      "learning_rate": 8.103349418109566e-05,
      "loss": 0.029433563351631165,
      "step": 66850
    },
    {
      "epoch": 18.97814362759012,
      "grad_norm": 2.8797895908355713,
      "learning_rate": 8.10306556911723e-05,
      "loss": 0.01952017694711685,
      "step": 66860
    },
    {
      "epoch": 18.980982117513484,
      "grad_norm": 5.100076198577881,
      "learning_rate": 8.102781720124894e-05,
      "loss": 0.022707422077655793,
      "step": 66870
    },
    {
      "epoch": 18.983820607436844,
      "grad_norm": 0.7674856185913086,
      "learning_rate": 8.102497871132558e-05,
      "loss": 0.014295876026153564,
      "step": 66880
    },
    {
      "epoch": 18.986659097360203,
      "grad_norm": 12.749394416809082,
      "learning_rate": 8.102214022140222e-05,
      "loss": 0.021755361557006837,
      "step": 66890
    },
    {
      "epoch": 18.989497587283566,
      "grad_norm": 0.5015447735786438,
      "learning_rate": 8.101930173147886e-05,
      "loss": 0.014195278286933899,
      "step": 66900
    },
    {
      "epoch": 18.992336077206925,
      "grad_norm": 5.684432029724121,
      "learning_rate": 8.101646324155549e-05,
      "loss": 0.026375368237495422,
      "step": 66910
    },
    {
      "epoch": 18.995174567130288,
      "grad_norm": 8.181107521057129,
      "learning_rate": 8.101362475163213e-05,
      "loss": 0.02488187253475189,
      "step": 66920
    },
    {
      "epoch": 18.998013057053647,
      "grad_norm": 8.39255428314209,
      "learning_rate": 8.101078626170877e-05,
      "loss": 0.029263943433761597,
      "step": 66930
    },
    {
      "epoch": 19.000851546977007,
      "grad_norm": 9.619133949279785,
      "learning_rate": 8.10079477717854e-05,
      "loss": 0.025025132298469543,
      "step": 66940
    },
    {
      "epoch": 19.00369003690037,
      "grad_norm": 2.117828607559204,
      "learning_rate": 8.100510928186206e-05,
      "loss": 0.033998480439186095,
      "step": 66950
    },
    {
      "epoch": 19.00652852682373,
      "grad_norm": 3.756300449371338,
      "learning_rate": 8.10022707919387e-05,
      "loss": 0.015250203013420106,
      "step": 66960
    },
    {
      "epoch": 19.009367016747092,
      "grad_norm": 2.767303705215454,
      "learning_rate": 8.099943230201533e-05,
      "loss": 0.005189386755228042,
      "step": 66970
    },
    {
      "epoch": 19.01220550667045,
      "grad_norm": 16.027650833129883,
      "learning_rate": 8.099659381209197e-05,
      "loss": 0.03313213288784027,
      "step": 66980
    },
    {
      "epoch": 19.01504399659381,
      "grad_norm": 12.516337394714355,
      "learning_rate": 8.099375532216861e-05,
      "loss": 0.022537827491760254,
      "step": 66990
    },
    {
      "epoch": 19.017882486517173,
      "grad_norm": 3.3282933235168457,
      "learning_rate": 8.099091683224525e-05,
      "loss": 0.018492187559604644,
      "step": 67000
    },
    {
      "epoch": 19.017882486517173,
      "eval_accuracy": 0.9661728238061932,
      "eval_loss": 0.10742823779582977,
      "eval_runtime": 31.8494,
      "eval_samples_per_second": 493.792,
      "eval_steps_per_second": 7.724,
      "step": 67000
    },
    {
      "epoch": 19.020720976440533,
      "grad_norm": 0.7128183841705322,
      "learning_rate": 8.098807834232189e-05,
      "loss": 0.02031441926956177,
      "step": 67010
    },
    {
      "epoch": 19.023559466363896,
      "grad_norm": 5.169584274291992,
      "learning_rate": 8.098523985239853e-05,
      "loss": 0.018671661615371704,
      "step": 67020
    },
    {
      "epoch": 19.026397956287255,
      "grad_norm": 2.880197763442993,
      "learning_rate": 8.098240136247517e-05,
      "loss": 0.02809334099292755,
      "step": 67030
    },
    {
      "epoch": 19.029236446210614,
      "grad_norm": 3.767319679260254,
      "learning_rate": 8.09795628725518e-05,
      "loss": 0.009692888706922531,
      "step": 67040
    },
    {
      "epoch": 19.032074936133977,
      "grad_norm": 7.517703056335449,
      "learning_rate": 8.097672438262844e-05,
      "loss": 0.024506066739559174,
      "step": 67050
    },
    {
      "epoch": 19.034913426057336,
      "grad_norm": 0.9031392335891724,
      "learning_rate": 8.097388589270509e-05,
      "loss": 0.0055957984179258345,
      "step": 67060
    },
    {
      "epoch": 19.0377519159807,
      "grad_norm": 5.282407283782959,
      "learning_rate": 8.097104740278171e-05,
      "loss": 0.012705722451210022,
      "step": 67070
    },
    {
      "epoch": 19.04059040590406,
      "grad_norm": 6.722054481506348,
      "learning_rate": 8.096820891285837e-05,
      "loss": 0.016569796204566955,
      "step": 67080
    },
    {
      "epoch": 19.043428895827418,
      "grad_norm": 2.3723578453063965,
      "learning_rate": 8.096537042293501e-05,
      "loss": 0.009938886761665345,
      "step": 67090
    },
    {
      "epoch": 19.04626738575078,
      "grad_norm": 2.4302635192871094,
      "learning_rate": 8.096253193301164e-05,
      "loss": 0.010542555898427963,
      "step": 67100
    },
    {
      "epoch": 19.04910587567414,
      "grad_norm": 15.297137260437012,
      "learning_rate": 8.095969344308828e-05,
      "loss": 0.013072320818901062,
      "step": 67110
    },
    {
      "epoch": 19.051944365597503,
      "grad_norm": 2.111640691757202,
      "learning_rate": 8.095685495316492e-05,
      "loss": 0.013304086029529571,
      "step": 67120
    },
    {
      "epoch": 19.054782855520862,
      "grad_norm": 0.5536283254623413,
      "learning_rate": 8.095401646324156e-05,
      "loss": 0.0153475821018219,
      "step": 67130
    },
    {
      "epoch": 19.057621345444225,
      "grad_norm": 3.7595982551574707,
      "learning_rate": 8.09511779733182e-05,
      "loss": 0.008560802042484283,
      "step": 67140
    },
    {
      "epoch": 19.060459835367585,
      "grad_norm": 0.7707223892211914,
      "learning_rate": 8.094833948339484e-05,
      "loss": 0.01635643094778061,
      "step": 67150
    },
    {
      "epoch": 19.063298325290944,
      "grad_norm": 3.6618599891662598,
      "learning_rate": 8.094550099347149e-05,
      "loss": 0.03128852248191834,
      "step": 67160
    },
    {
      "epoch": 19.066136815214307,
      "grad_norm": 3.980419635772705,
      "learning_rate": 8.094266250354811e-05,
      "loss": 0.023784548044204712,
      "step": 67170
    },
    {
      "epoch": 19.068975305137666,
      "grad_norm": 5.53279447555542,
      "learning_rate": 8.093982401362475e-05,
      "loss": 0.017350564897060394,
      "step": 67180
    },
    {
      "epoch": 19.07181379506103,
      "grad_norm": 1.5272711515426636,
      "learning_rate": 8.09369855237014e-05,
      "loss": 0.009153299778699876,
      "step": 67190
    },
    {
      "epoch": 19.07465228498439,
      "grad_norm": 9.337668418884277,
      "learning_rate": 8.093414703377802e-05,
      "loss": 0.01281513124704361,
      "step": 67200
    },
    {
      "epoch": 19.077490774907748,
      "grad_norm": 3.136249303817749,
      "learning_rate": 8.093130854385468e-05,
      "loss": 0.006052959710359573,
      "step": 67210
    },
    {
      "epoch": 19.08032926483111,
      "grad_norm": 0.6747031211853027,
      "learning_rate": 8.092847005393132e-05,
      "loss": 0.013777542114257812,
      "step": 67220
    },
    {
      "epoch": 19.08316775475447,
      "grad_norm": 7.410195827484131,
      "learning_rate": 8.092563156400795e-05,
      "loss": 0.016757294535636902,
      "step": 67230
    },
    {
      "epoch": 19.086006244677833,
      "grad_norm": 1.0990087985992432,
      "learning_rate": 8.092279307408459e-05,
      "loss": 0.009844745695590972,
      "step": 67240
    },
    {
      "epoch": 19.088844734601192,
      "grad_norm": 2.769007682800293,
      "learning_rate": 8.091995458416123e-05,
      "loss": 0.007458788901567459,
      "step": 67250
    },
    {
      "epoch": 19.09168322452455,
      "grad_norm": 1.658687710762024,
      "learning_rate": 8.091711609423787e-05,
      "loss": 0.011466432362794876,
      "step": 67260
    },
    {
      "epoch": 19.094521714447914,
      "grad_norm": 1.5208580493927002,
      "learning_rate": 8.09142776043145e-05,
      "loss": 0.008203256875276566,
      "step": 67270
    },
    {
      "epoch": 19.097360204371274,
      "grad_norm": 0.12263114750385284,
      "learning_rate": 8.091143911439116e-05,
      "loss": 0.015834109485149385,
      "step": 67280
    },
    {
      "epoch": 19.100198694294637,
      "grad_norm": 0.8305667042732239,
      "learning_rate": 8.09086006244678e-05,
      "loss": 0.007405736297369003,
      "step": 67290
    },
    {
      "epoch": 19.103037184217996,
      "grad_norm": 0.3608134686946869,
      "learning_rate": 8.090576213454442e-05,
      "loss": 0.014245216548442841,
      "step": 67300
    },
    {
      "epoch": 19.105875674141355,
      "grad_norm": 1.3385645151138306,
      "learning_rate": 8.090292364462107e-05,
      "loss": 0.011370453983545303,
      "step": 67310
    },
    {
      "epoch": 19.10871416406472,
      "grad_norm": 10.860981941223145,
      "learning_rate": 8.090008515469771e-05,
      "loss": 0.031423071026802064,
      "step": 67320
    },
    {
      "epoch": 19.111552653988078,
      "grad_norm": 2.063197135925293,
      "learning_rate": 8.089724666477433e-05,
      "loss": 0.01963375210762024,
      "step": 67330
    },
    {
      "epoch": 19.11439114391144,
      "grad_norm": 8.667553901672363,
      "learning_rate": 8.089440817485099e-05,
      "loss": 0.020614391565322875,
      "step": 67340
    },
    {
      "epoch": 19.1172296338348,
      "grad_norm": 2.4544875621795654,
      "learning_rate": 8.089156968492763e-05,
      "loss": 0.010786810517311096,
      "step": 67350
    },
    {
      "epoch": 19.12006812375816,
      "grad_norm": 5.245517730712891,
      "learning_rate": 8.088873119500426e-05,
      "loss": 0.014381052553653717,
      "step": 67360
    },
    {
      "epoch": 19.122906613681522,
      "grad_norm": 6.7719950675964355,
      "learning_rate": 8.08858927050809e-05,
      "loss": 0.014396995306015015,
      "step": 67370
    },
    {
      "epoch": 19.12574510360488,
      "grad_norm": 10.405839920043945,
      "learning_rate": 8.088305421515754e-05,
      "loss": 0.018640251457691194,
      "step": 67380
    },
    {
      "epoch": 19.128583593528244,
      "grad_norm": 0.6709728837013245,
      "learning_rate": 8.088021572523418e-05,
      "loss": 0.0157562181353569,
      "step": 67390
    },
    {
      "epoch": 19.131422083451604,
      "grad_norm": 2.6977698802948,
      "learning_rate": 8.087737723531081e-05,
      "loss": 0.010232111811637879,
      "step": 67400
    },
    {
      "epoch": 19.134260573374963,
      "grad_norm": 11.183005332946777,
      "learning_rate": 8.087453874538747e-05,
      "loss": 0.011555835604667664,
      "step": 67410
    },
    {
      "epoch": 19.137099063298326,
      "grad_norm": 11.41781997680664,
      "learning_rate": 8.08717002554641e-05,
      "loss": 0.029315918684005737,
      "step": 67420
    },
    {
      "epoch": 19.139937553221685,
      "grad_norm": 1.9492456912994385,
      "learning_rate": 8.086886176554073e-05,
      "loss": 0.018679100275039672,
      "step": 67430
    },
    {
      "epoch": 19.142776043145048,
      "grad_norm": 1.1353530883789062,
      "learning_rate": 8.086602327561738e-05,
      "loss": 0.01314205229282379,
      "step": 67440
    },
    {
      "epoch": 19.145614533068407,
      "grad_norm": 1.0492925643920898,
      "learning_rate": 8.086318478569402e-05,
      "loss": 0.024735867977142334,
      "step": 67450
    },
    {
      "epoch": 19.148453022991767,
      "grad_norm": 3.6296498775482178,
      "learning_rate": 8.086034629577065e-05,
      "loss": 0.021997874975204466,
      "step": 67460
    },
    {
      "epoch": 19.15129151291513,
      "grad_norm": 0.559613049030304,
      "learning_rate": 8.085750780584729e-05,
      "loss": 0.02434665262699127,
      "step": 67470
    },
    {
      "epoch": 19.15413000283849,
      "grad_norm": 4.4978556632995605,
      "learning_rate": 8.085466931592394e-05,
      "loss": 0.012002518028020858,
      "step": 67480
    },
    {
      "epoch": 19.15696849276185,
      "grad_norm": 1.7960566282272339,
      "learning_rate": 8.085183082600057e-05,
      "loss": 0.01233261376619339,
      "step": 67490
    },
    {
      "epoch": 19.15980698268521,
      "grad_norm": 3.5794947147369385,
      "learning_rate": 8.084899233607721e-05,
      "loss": 0.008021872490644455,
      "step": 67500
    },
    {
      "epoch": 19.15980698268521,
      "eval_accuracy": 0.9592420677815222,
      "eval_loss": 0.12999430298805237,
      "eval_runtime": 31.6083,
      "eval_samples_per_second": 497.56,
      "eval_steps_per_second": 7.783,
      "step": 67500
    },
    {
      "epoch": 19.162645472608574,
      "grad_norm": 1.2775534391403198,
      "learning_rate": 8.084615384615385e-05,
      "loss": 0.014560607075691224,
      "step": 67510
    },
    {
      "epoch": 19.165483962531933,
      "grad_norm": 1.8521090745925903,
      "learning_rate": 8.084331535623048e-05,
      "loss": 0.02767760157585144,
      "step": 67520
    },
    {
      "epoch": 19.168322452455293,
      "grad_norm": 0.9166882038116455,
      "learning_rate": 8.084047686630712e-05,
      "loss": 0.010507382452487946,
      "step": 67530
    },
    {
      "epoch": 19.171160942378656,
      "grad_norm": 5.8645830154418945,
      "learning_rate": 8.083763837638378e-05,
      "loss": 0.025987327098846436,
      "step": 67540
    },
    {
      "epoch": 19.173999432302015,
      "grad_norm": 2.1302120685577393,
      "learning_rate": 8.08347998864604e-05,
      "loss": 0.012335435301065446,
      "step": 67550
    },
    {
      "epoch": 19.176837922225378,
      "grad_norm": 10.883798599243164,
      "learning_rate": 8.083196139653705e-05,
      "loss": 0.007305572181940079,
      "step": 67560
    },
    {
      "epoch": 19.179676412148737,
      "grad_norm": 0.6601458787918091,
      "learning_rate": 8.082912290661369e-05,
      "loss": 0.01694016009569168,
      "step": 67570
    },
    {
      "epoch": 19.182514902072096,
      "grad_norm": 9.057695388793945,
      "learning_rate": 8.082628441669033e-05,
      "loss": 0.013481160998344422,
      "step": 67580
    },
    {
      "epoch": 19.18535339199546,
      "grad_norm": 1.173474669456482,
      "learning_rate": 8.082344592676696e-05,
      "loss": 0.020177052915096284,
      "step": 67590
    },
    {
      "epoch": 19.18819188191882,
      "grad_norm": 6.526787281036377,
      "learning_rate": 8.08206074368436e-05,
      "loss": 0.01318686306476593,
      "step": 67600
    },
    {
      "epoch": 19.19103037184218,
      "grad_norm": 2.1249332427978516,
      "learning_rate": 8.081776894692025e-05,
      "loss": 0.006008858978748322,
      "step": 67610
    },
    {
      "epoch": 19.19386886176554,
      "grad_norm": 0.16272367537021637,
      "learning_rate": 8.081493045699688e-05,
      "loss": 0.010288356989622115,
      "step": 67620
    },
    {
      "epoch": 19.1967073516889,
      "grad_norm": 0.6398535370826721,
      "learning_rate": 8.081209196707352e-05,
      "loss": 0.018589666485786437,
      "step": 67630
    },
    {
      "epoch": 19.199545841612263,
      "grad_norm": 0.6388561725616455,
      "learning_rate": 8.080925347715016e-05,
      "loss": 0.01729421615600586,
      "step": 67640
    },
    {
      "epoch": 19.202384331535622,
      "grad_norm": 1.0805052518844604,
      "learning_rate": 8.080641498722679e-05,
      "loss": 0.012442007660865784,
      "step": 67650
    },
    {
      "epoch": 19.205222821458985,
      "grad_norm": 6.400195598602295,
      "learning_rate": 8.080357649730343e-05,
      "loss": 0.019930429756641388,
      "step": 67660
    },
    {
      "epoch": 19.208061311382345,
      "grad_norm": 9.064736366271973,
      "learning_rate": 8.080073800738007e-05,
      "loss": 0.03814596235752106,
      "step": 67670
    },
    {
      "epoch": 19.210899801305704,
      "grad_norm": 0.1559750884771347,
      "learning_rate": 8.079789951745672e-05,
      "loss": 0.014321824908256531,
      "step": 67680
    },
    {
      "epoch": 19.213738291229067,
      "grad_norm": 1.3881678581237793,
      "learning_rate": 8.079506102753336e-05,
      "loss": 0.009494108706712722,
      "step": 67690
    },
    {
      "epoch": 19.216576781152426,
      "grad_norm": 2.005899429321289,
      "learning_rate": 8.079222253761e-05,
      "loss": 0.02582116723060608,
      "step": 67700
    },
    {
      "epoch": 19.21941527107579,
      "grad_norm": 0.5148826241493225,
      "learning_rate": 8.078938404768664e-05,
      "loss": 0.008900617063045502,
      "step": 67710
    },
    {
      "epoch": 19.22225376099915,
      "grad_norm": 5.008853912353516,
      "learning_rate": 8.078654555776327e-05,
      "loss": 0.021077825129032134,
      "step": 67720
    },
    {
      "epoch": 19.225092250922508,
      "grad_norm": 0.9692829847335815,
      "learning_rate": 8.078370706783991e-05,
      "loss": 0.023941811919212342,
      "step": 67730
    },
    {
      "epoch": 19.22793074084587,
      "grad_norm": 17.336599349975586,
      "learning_rate": 8.078086857791656e-05,
      "loss": 0.014363613724708558,
      "step": 67740
    },
    {
      "epoch": 19.23076923076923,
      "grad_norm": 3.8277463912963867,
      "learning_rate": 8.077803008799319e-05,
      "loss": 0.018578089773654938,
      "step": 67750
    },
    {
      "epoch": 19.233607720692593,
      "grad_norm": 7.376985549926758,
      "learning_rate": 8.077519159806983e-05,
      "loss": 0.01765679270029068,
      "step": 67760
    },
    {
      "epoch": 19.236446210615952,
      "grad_norm": 3.771419048309326,
      "learning_rate": 8.077235310814647e-05,
      "loss": 0.010608582198619843,
      "step": 67770
    },
    {
      "epoch": 19.23928470053931,
      "grad_norm": 10.57840347290039,
      "learning_rate": 8.07695146182231e-05,
      "loss": 0.03190801739692688,
      "step": 67780
    },
    {
      "epoch": 19.242123190462674,
      "grad_norm": 0.2734811305999756,
      "learning_rate": 8.076667612829974e-05,
      "loss": 0.02803638279438019,
      "step": 67790
    },
    {
      "epoch": 19.244961680386034,
      "grad_norm": 5.873191833496094,
      "learning_rate": 8.076383763837638e-05,
      "loss": 0.021638505160808563,
      "step": 67800
    },
    {
      "epoch": 19.247800170309397,
      "grad_norm": 2.8172056674957275,
      "learning_rate": 8.076099914845303e-05,
      "loss": 0.00960276871919632,
      "step": 67810
    },
    {
      "epoch": 19.250638660232756,
      "grad_norm": 1.21724534034729,
      "learning_rate": 8.075816065852967e-05,
      "loss": 0.020796990394592284,
      "step": 67820
    },
    {
      "epoch": 19.253477150156115,
      "grad_norm": 0.6179468631744385,
      "learning_rate": 8.075532216860631e-05,
      "loss": 0.009797509759664536,
      "step": 67830
    },
    {
      "epoch": 19.256315640079478,
      "grad_norm": 5.0413818359375,
      "learning_rate": 8.075248367868295e-05,
      "loss": 0.02460653930902481,
      "step": 67840
    },
    {
      "epoch": 19.259154130002837,
      "grad_norm": 8.126354217529297,
      "learning_rate": 8.074964518875958e-05,
      "loss": 0.016722001135349274,
      "step": 67850
    },
    {
      "epoch": 19.2619926199262,
      "grad_norm": 0.4409915804862976,
      "learning_rate": 8.074680669883622e-05,
      "loss": 0.008022047579288483,
      "step": 67860
    },
    {
      "epoch": 19.26483110984956,
      "grad_norm": 8.339662551879883,
      "learning_rate": 8.074396820891286e-05,
      "loss": 0.014708764851093292,
      "step": 67870
    },
    {
      "epoch": 19.267669599772923,
      "grad_norm": 14.161458969116211,
      "learning_rate": 8.07411297189895e-05,
      "loss": 0.023420609533786774,
      "step": 67880
    },
    {
      "epoch": 19.270508089696282,
      "grad_norm": 5.470942974090576,
      "learning_rate": 8.073829122906614e-05,
      "loss": 0.017591893672943115,
      "step": 67890
    },
    {
      "epoch": 19.27334657961964,
      "grad_norm": 2.383192777633667,
      "learning_rate": 8.073545273914278e-05,
      "loss": 0.016281697154045104,
      "step": 67900
    },
    {
      "epoch": 19.276185069543004,
      "grad_norm": 4.8740973472595215,
      "learning_rate": 8.073261424921941e-05,
      "loss": 0.030304166674613952,
      "step": 67910
    },
    {
      "epoch": 19.279023559466363,
      "grad_norm": 7.9033684730529785,
      "learning_rate": 8.072977575929605e-05,
      "loss": 0.017916691303253175,
      "step": 67920
    },
    {
      "epoch": 19.281862049389726,
      "grad_norm": 16.310129165649414,
      "learning_rate": 8.07269372693727e-05,
      "loss": 0.017711320519447328,
      "step": 67930
    },
    {
      "epoch": 19.284700539313086,
      "grad_norm": 2.117884397506714,
      "learning_rate": 8.072409877944934e-05,
      "loss": 0.01555851399898529,
      "step": 67940
    },
    {
      "epoch": 19.287539029236445,
      "grad_norm": 3.372568130493164,
      "learning_rate": 8.072126028952598e-05,
      "loss": 0.007022402435541153,
      "step": 67950
    },
    {
      "epoch": 19.290377519159808,
      "grad_norm": 0.4804041087627411,
      "learning_rate": 8.071842179960262e-05,
      "loss": 0.009142833948135375,
      "step": 67960
    },
    {
      "epoch": 19.293216009083167,
      "grad_norm": 8.029982566833496,
      "learning_rate": 8.071558330967926e-05,
      "loss": 0.018873147666454315,
      "step": 67970
    },
    {
      "epoch": 19.29605449900653,
      "grad_norm": 0.36158791184425354,
      "learning_rate": 8.071274481975589e-05,
      "loss": 0.009248950332403184,
      "step": 67980
    },
    {
      "epoch": 19.29889298892989,
      "grad_norm": 10.57125186920166,
      "learning_rate": 8.070990632983253e-05,
      "loss": 0.017669256031513213,
      "step": 67990
    },
    {
      "epoch": 19.30173147885325,
      "grad_norm": 2.9398984909057617,
      "learning_rate": 8.070706783990917e-05,
      "loss": 0.010069814324378968,
      "step": 68000
    },
    {
      "epoch": 19.30173147885325,
      "eval_accuracy": 0.9672537674063713,
      "eval_loss": 0.10566136986017227,
      "eval_runtime": 32.0337,
      "eval_samples_per_second": 490.951,
      "eval_steps_per_second": 7.679,
      "step": 68000
    },
    {
      "epoch": 19.30456996877661,
      "grad_norm": 0.36136746406555176,
      "learning_rate": 8.070422934998581e-05,
      "loss": 0.020288358628749847,
      "step": 68010
    },
    {
      "epoch": 19.30740845869997,
      "grad_norm": 6.821678161621094,
      "learning_rate": 8.070139086006245e-05,
      "loss": 0.02235458791255951,
      "step": 68020
    },
    {
      "epoch": 19.310246948623334,
      "grad_norm": 1.2413313388824463,
      "learning_rate": 8.06985523701391e-05,
      "loss": 0.005752924084663391,
      "step": 68030
    },
    {
      "epoch": 19.313085438546693,
      "grad_norm": 1.815874695777893,
      "learning_rate": 8.069571388021572e-05,
      "loss": 0.011402224749326706,
      "step": 68040
    },
    {
      "epoch": 19.315923928470053,
      "grad_norm": 4.294982433319092,
      "learning_rate": 8.069287539029236e-05,
      "loss": 0.024358630180358887,
      "step": 68050
    },
    {
      "epoch": 19.318762418393415,
      "grad_norm": 16.838899612426758,
      "learning_rate": 8.0690036900369e-05,
      "loss": 0.014200448989868164,
      "step": 68060
    },
    {
      "epoch": 19.321600908316775,
      "grad_norm": 5.7158050537109375,
      "learning_rate": 8.068719841044565e-05,
      "loss": 0.023680722713470458,
      "step": 68070
    },
    {
      "epoch": 19.324439398240138,
      "grad_norm": 2.2217957973480225,
      "learning_rate": 8.068435992052229e-05,
      "loss": 0.02281351536512375,
      "step": 68080
    },
    {
      "epoch": 19.327277888163497,
      "grad_norm": 0.7452578544616699,
      "learning_rate": 8.068152143059893e-05,
      "loss": 0.013897983729839325,
      "step": 68090
    },
    {
      "epoch": 19.330116378086856,
      "grad_norm": 2.279170274734497,
      "learning_rate": 8.067868294067557e-05,
      "loss": 0.018126335740089417,
      "step": 68100
    },
    {
      "epoch": 19.33295486801022,
      "grad_norm": 1.5987281799316406,
      "learning_rate": 8.06758444507522e-05,
      "loss": 0.017642366886138915,
      "step": 68110
    },
    {
      "epoch": 19.33579335793358,
      "grad_norm": 0.5822604298591614,
      "learning_rate": 8.067300596082884e-05,
      "loss": 0.01753823906183243,
      "step": 68120
    },
    {
      "epoch": 19.33863184785694,
      "grad_norm": 2.722149133682251,
      "learning_rate": 8.067016747090548e-05,
      "loss": 0.010779988765716553,
      "step": 68130
    },
    {
      "epoch": 19.3414703377803,
      "grad_norm": 0.3484378159046173,
      "learning_rate": 8.066732898098212e-05,
      "loss": 0.016800422966480256,
      "step": 68140
    },
    {
      "epoch": 19.34430882770366,
      "grad_norm": 1.1629972457885742,
      "learning_rate": 8.066449049105876e-05,
      "loss": 0.016008684039115907,
      "step": 68150
    },
    {
      "epoch": 19.347147317627023,
      "grad_norm": 2.117344379425049,
      "learning_rate": 8.06616520011354e-05,
      "loss": 0.012802721560001373,
      "step": 68160
    },
    {
      "epoch": 19.349985807550382,
      "grad_norm": 1.9862496852874756,
      "learning_rate": 8.065881351121203e-05,
      "loss": 0.022111015021800996,
      "step": 68170
    },
    {
      "epoch": 19.352824297473745,
      "grad_norm": 8.672078132629395,
      "learning_rate": 8.065597502128868e-05,
      "loss": 0.014741088449954986,
      "step": 68180
    },
    {
      "epoch": 19.355662787397105,
      "grad_norm": 1.0371514558792114,
      "learning_rate": 8.065313653136532e-05,
      "loss": 0.007393895089626313,
      "step": 68190
    },
    {
      "epoch": 19.358501277320464,
      "grad_norm": 3.1068577766418457,
      "learning_rate": 8.065029804144196e-05,
      "loss": 0.010084652155637742,
      "step": 68200
    },
    {
      "epoch": 19.361339767243827,
      "grad_norm": 1.4325002431869507,
      "learning_rate": 8.06474595515186e-05,
      "loss": 0.02285253703594208,
      "step": 68210
    },
    {
      "epoch": 19.364178257167186,
      "grad_norm": 1.3104113340377808,
      "learning_rate": 8.064462106159524e-05,
      "loss": 0.012064166367053986,
      "step": 68220
    },
    {
      "epoch": 19.36701674709055,
      "grad_norm": 5.637192726135254,
      "learning_rate": 8.064178257167188e-05,
      "loss": 0.014824016392230988,
      "step": 68230
    },
    {
      "epoch": 19.36985523701391,
      "grad_norm": 1.9618721008300781,
      "learning_rate": 8.063894408174851e-05,
      "loss": 0.014382511377334595,
      "step": 68240
    },
    {
      "epoch": 19.372693726937268,
      "grad_norm": 0.652323305606842,
      "learning_rate": 8.063610559182515e-05,
      "loss": 0.01997605860233307,
      "step": 68250
    },
    {
      "epoch": 19.37553221686063,
      "grad_norm": 1.9782042503356934,
      "learning_rate": 8.063355095089413e-05,
      "loss": 0.03351330459117889,
      "step": 68260
    },
    {
      "epoch": 19.37837070678399,
      "grad_norm": 3.842339277267456,
      "learning_rate": 8.063071246097077e-05,
      "loss": 0.017319580912590025,
      "step": 68270
    },
    {
      "epoch": 19.381209196707353,
      "grad_norm": 3.4852116107940674,
      "learning_rate": 8.06278739710474e-05,
      "loss": 0.013397429883480073,
      "step": 68280
    },
    {
      "epoch": 19.384047686630712,
      "grad_norm": 3.6600663661956787,
      "learning_rate": 8.062503548112404e-05,
      "loss": 0.018483200669288637,
      "step": 68290
    },
    {
      "epoch": 19.386886176554075,
      "grad_norm": 6.97262716293335,
      "learning_rate": 8.062219699120068e-05,
      "loss": 0.016976676881313324,
      "step": 68300
    },
    {
      "epoch": 19.389724666477434,
      "grad_norm": 14.758373260498047,
      "learning_rate": 8.061935850127732e-05,
      "loss": 0.020726034045219423,
      "step": 68310
    },
    {
      "epoch": 19.392563156400794,
      "grad_norm": 5.3285651206970215,
      "learning_rate": 8.061652001135397e-05,
      "loss": 0.007838141173124313,
      "step": 68320
    },
    {
      "epoch": 19.395401646324157,
      "grad_norm": 0.40738874673843384,
      "learning_rate": 8.061368152143061e-05,
      "loss": 0.007996352016925811,
      "step": 68330
    },
    {
      "epoch": 19.398240136247516,
      "grad_norm": 8.883698463439941,
      "learning_rate": 8.061084303150725e-05,
      "loss": 0.029583722352981567,
      "step": 68340
    },
    {
      "epoch": 19.40107862617088,
      "grad_norm": 1.2910749912261963,
      "learning_rate": 8.060800454158388e-05,
      "loss": 0.015966296195983887,
      "step": 68350
    },
    {
      "epoch": 19.403917116094238,
      "grad_norm": 4.066921234130859,
      "learning_rate": 8.060516605166052e-05,
      "loss": 0.017843152582645416,
      "step": 68360
    },
    {
      "epoch": 19.406755606017597,
      "grad_norm": 10.746665000915527,
      "learning_rate": 8.060232756173716e-05,
      "loss": 0.015083371102809906,
      "step": 68370
    },
    {
      "epoch": 19.40959409594096,
      "grad_norm": 1.8514008522033691,
      "learning_rate": 8.059948907181379e-05,
      "loss": 0.010478422045707703,
      "step": 68380
    },
    {
      "epoch": 19.41243258586432,
      "grad_norm": 1.0810657739639282,
      "learning_rate": 8.059665058189044e-05,
      "loss": 0.007983805984258652,
      "step": 68390
    },
    {
      "epoch": 19.415271075787683,
      "grad_norm": 1.9075071811676025,
      "learning_rate": 8.059381209196708e-05,
      "loss": 0.01799801141023636,
      "step": 68400
    },
    {
      "epoch": 19.418109565711042,
      "grad_norm": 1.0062581300735474,
      "learning_rate": 8.059097360204371e-05,
      "loss": 0.024819447100162505,
      "step": 68410
    },
    {
      "epoch": 19.4209480556344,
      "grad_norm": 11.236578941345215,
      "learning_rate": 8.058813511212035e-05,
      "loss": 0.011575204879045486,
      "step": 68420
    },
    {
      "epoch": 19.423786545557764,
      "grad_norm": 9.953205108642578,
      "learning_rate": 8.0585296622197e-05,
      "loss": 0.019811382889747618,
      "step": 68430
    },
    {
      "epoch": 19.426625035481123,
      "grad_norm": 2.9195587635040283,
      "learning_rate": 8.058245813227364e-05,
      "loss": 0.0164811834692955,
      "step": 68440
    },
    {
      "epoch": 19.429463525404486,
      "grad_norm": 3.004770040512085,
      "learning_rate": 8.057961964235026e-05,
      "loss": 0.014394600689411164,
      "step": 68450
    },
    {
      "epoch": 19.432302015327846,
      "grad_norm": 4.019406795501709,
      "learning_rate": 8.057678115242692e-05,
      "loss": 0.013678045570850372,
      "step": 68460
    },
    {
      "epoch": 19.435140505251205,
      "grad_norm": 1.4142735004425049,
      "learning_rate": 8.057394266250356e-05,
      "loss": 0.01651117354631424,
      "step": 68470
    },
    {
      "epoch": 19.437978995174568,
      "grad_norm": 6.617769241333008,
      "learning_rate": 8.057110417258019e-05,
      "loss": 0.014800643920898438,
      "step": 68480
    },
    {
      "epoch": 19.440817485097927,
      "grad_norm": 12.624940872192383,
      "learning_rate": 8.056826568265683e-05,
      "loss": 0.011644546687602998,
      "step": 68490
    },
    {
      "epoch": 19.44365597502129,
      "grad_norm": 1.4502366781234741,
      "learning_rate": 8.056542719273347e-05,
      "loss": 0.02547231912612915,
      "step": 68500
    },
    {
      "epoch": 19.44365597502129,
      "eval_accuracy": 0.9623577287467413,
      "eval_loss": 0.12027938663959503,
      "eval_runtime": 31.8958,
      "eval_samples_per_second": 493.074,
      "eval_steps_per_second": 7.713,
      "step": 68500
    },
    {
      "epoch": 19.44649446494465,
      "grad_norm": 2.7482173442840576,
      "learning_rate": 8.05625887028101e-05,
      "loss": 0.025292092561721803,
      "step": 68510
    },
    {
      "epoch": 19.44933295486801,
      "grad_norm": 15.106111526489258,
      "learning_rate": 8.055975021288675e-05,
      "loss": 0.03591836988925934,
      "step": 68520
    },
    {
      "epoch": 19.45217144479137,
      "grad_norm": 4.529141902923584,
      "learning_rate": 8.05569117229634e-05,
      "loss": 0.015529727935791016,
      "step": 68530
    },
    {
      "epoch": 19.45500993471473,
      "grad_norm": 13.03493595123291,
      "learning_rate": 8.055407323304002e-05,
      "loss": 0.016385802626609804,
      "step": 68540
    },
    {
      "epoch": 19.457848424638094,
      "grad_norm": 1.7573965787887573,
      "learning_rate": 8.055123474311666e-05,
      "loss": 0.023104332387447357,
      "step": 68550
    },
    {
      "epoch": 19.460686914561453,
      "grad_norm": 1.6557745933532715,
      "learning_rate": 8.05483962531933e-05,
      "loss": 0.0061702758073806764,
      "step": 68560
    },
    {
      "epoch": 19.463525404484812,
      "grad_norm": 2.6736443042755127,
      "learning_rate": 8.054555776326995e-05,
      "loss": 0.011512688547372817,
      "step": 68570
    },
    {
      "epoch": 19.466363894408175,
      "grad_norm": 6.430649280548096,
      "learning_rate": 8.054271927334657e-05,
      "loss": 0.021578523516654968,
      "step": 68580
    },
    {
      "epoch": 19.469202384331535,
      "grad_norm": 10.6021089553833,
      "learning_rate": 8.053988078342323e-05,
      "loss": 0.012772506475448609,
      "step": 68590
    },
    {
      "epoch": 19.472040874254898,
      "grad_norm": 2.297525644302368,
      "learning_rate": 8.053704229349987e-05,
      "loss": 0.01806233674287796,
      "step": 68600
    },
    {
      "epoch": 19.474879364178257,
      "grad_norm": 0.6878420114517212,
      "learning_rate": 8.05342038035765e-05,
      "loss": 0.01141258180141449,
      "step": 68610
    },
    {
      "epoch": 19.477717854101616,
      "grad_norm": 1.08841872215271,
      "learning_rate": 8.053136531365314e-05,
      "loss": 0.011026586592197418,
      "step": 68620
    },
    {
      "epoch": 19.48055634402498,
      "grad_norm": 8.123515129089355,
      "learning_rate": 8.052852682372978e-05,
      "loss": 0.022044964134693146,
      "step": 68630
    },
    {
      "epoch": 19.48339483394834,
      "grad_norm": 1.9518377780914307,
      "learning_rate": 8.052568833380641e-05,
      "loss": 0.008070898056030274,
      "step": 68640
    },
    {
      "epoch": 19.4862333238717,
      "grad_norm": 2.4339699745178223,
      "learning_rate": 8.052284984388305e-05,
      "loss": 0.034346505999565125,
      "step": 68650
    },
    {
      "epoch": 19.48907181379506,
      "grad_norm": 1.3761080503463745,
      "learning_rate": 8.05200113539597e-05,
      "loss": 0.015331877768039704,
      "step": 68660
    },
    {
      "epoch": 19.49191030371842,
      "grad_norm": 5.9419684410095215,
      "learning_rate": 8.051717286403633e-05,
      "loss": 0.026616877317428587,
      "step": 68670
    },
    {
      "epoch": 19.494748793641783,
      "grad_norm": 11.315434455871582,
      "learning_rate": 8.051433437411297e-05,
      "loss": 0.014980483055114745,
      "step": 68680
    },
    {
      "epoch": 19.497587283565142,
      "grad_norm": 1.7705806493759155,
      "learning_rate": 8.051149588418962e-05,
      "loss": 0.02223304808139801,
      "step": 68690
    },
    {
      "epoch": 19.500425773488505,
      "grad_norm": 0.7497650384902954,
      "learning_rate": 8.050865739426626e-05,
      "loss": 0.023962581157684328,
      "step": 68700
    },
    {
      "epoch": 19.503264263411864,
      "grad_norm": 2.380741596221924,
      "learning_rate": 8.050581890434288e-05,
      "loss": 0.02096230834722519,
      "step": 68710
    },
    {
      "epoch": 19.506102753335227,
      "grad_norm": 3.907726287841797,
      "learning_rate": 8.050298041441954e-05,
      "loss": 0.03797944188117981,
      "step": 68720
    },
    {
      "epoch": 19.508941243258587,
      "grad_norm": 0.4190168082714081,
      "learning_rate": 8.050014192449618e-05,
      "loss": 0.009597226977348328,
      "step": 68730
    },
    {
      "epoch": 19.511779733181946,
      "grad_norm": 8.626215934753418,
      "learning_rate": 8.049730343457281e-05,
      "loss": 0.035277745127677916,
      "step": 68740
    },
    {
      "epoch": 19.51461822310531,
      "grad_norm": 1.285332202911377,
      "learning_rate": 8.049446494464945e-05,
      "loss": 0.029807767271995543,
      "step": 68750
    },
    {
      "epoch": 19.51745671302867,
      "grad_norm": 4.7375688552856445,
      "learning_rate": 8.049162645472609e-05,
      "loss": 0.03547104001045227,
      "step": 68760
    },
    {
      "epoch": 19.52029520295203,
      "grad_norm": 3.6163253784179688,
      "learning_rate": 8.048878796480272e-05,
      "loss": 0.012176578491926193,
      "step": 68770
    },
    {
      "epoch": 19.52313369287539,
      "grad_norm": 1.5618643760681152,
      "learning_rate": 8.048594947487936e-05,
      "loss": 0.009652698040008545,
      "step": 68780
    },
    {
      "epoch": 19.52597218279875,
      "grad_norm": 0.4965190291404724,
      "learning_rate": 8.048311098495602e-05,
      "loss": 0.020688846707344055,
      "step": 68790
    },
    {
      "epoch": 19.528810672722113,
      "grad_norm": 16.246986389160156,
      "learning_rate": 8.048027249503264e-05,
      "loss": 0.07054977416992188,
      "step": 68800
    },
    {
      "epoch": 19.531649162645472,
      "grad_norm": 0.8282925486564636,
      "learning_rate": 8.047743400510929e-05,
      "loss": 0.01789591759443283,
      "step": 68810
    },
    {
      "epoch": 19.534487652568835,
      "grad_norm": 6.707474231719971,
      "learning_rate": 8.047459551518593e-05,
      "loss": 0.016325725615024565,
      "step": 68820
    },
    {
      "epoch": 19.537326142492194,
      "grad_norm": 0.343199223279953,
      "learning_rate": 8.047175702526257e-05,
      "loss": 0.028074952960014343,
      "step": 68830
    },
    {
      "epoch": 19.540164632415554,
      "grad_norm": 1.5699849128723145,
      "learning_rate": 8.04689185353392e-05,
      "loss": 0.023118598759174346,
      "step": 68840
    },
    {
      "epoch": 19.543003122338916,
      "grad_norm": 1.3957401514053345,
      "learning_rate": 8.046608004541585e-05,
      "loss": 0.017350849509239197,
      "step": 68850
    },
    {
      "epoch": 19.545841612262276,
      "grad_norm": 5.136393070220947,
      "learning_rate": 8.046324155549249e-05,
      "loss": 0.022270841896533965,
      "step": 68860
    },
    {
      "epoch": 19.54868010218564,
      "grad_norm": 2.1466891765594482,
      "learning_rate": 8.046040306556912e-05,
      "loss": 0.012609714269638061,
      "step": 68870
    },
    {
      "epoch": 19.551518592108998,
      "grad_norm": 5.206986427307129,
      "learning_rate": 8.045756457564576e-05,
      "loss": 0.019027863442897797,
      "step": 68880
    },
    {
      "epoch": 19.554357082032357,
      "grad_norm": 6.954113960266113,
      "learning_rate": 8.04547260857224e-05,
      "loss": 0.013817405700683594,
      "step": 68890
    },
    {
      "epoch": 19.55719557195572,
      "grad_norm": 1.9676868915557861,
      "learning_rate": 8.045188759579903e-05,
      "loss": 0.014141739904880523,
      "step": 68900
    },
    {
      "epoch": 19.56003406187908,
      "grad_norm": 13.528081893920898,
      "learning_rate": 8.044904910587567e-05,
      "loss": 0.026004457473754884,
      "step": 68910
    },
    {
      "epoch": 19.562872551802442,
      "grad_norm": 14.59493350982666,
      "learning_rate": 8.044621061595233e-05,
      "loss": 0.03138493895530701,
      "step": 68920
    },
    {
      "epoch": 19.565711041725802,
      "grad_norm": 1.2393336296081543,
      "learning_rate": 8.044337212602895e-05,
      "loss": 0.026313084363937377,
      "step": 68930
    },
    {
      "epoch": 19.56854953164916,
      "grad_norm": 13.831196784973145,
      "learning_rate": 8.04405336361056e-05,
      "loss": 0.02973119616508484,
      "step": 68940
    },
    {
      "epoch": 19.571388021572524,
      "grad_norm": 9.97261905670166,
      "learning_rate": 8.043769514618224e-05,
      "loss": 0.018589992821216584,
      "step": 68950
    },
    {
      "epoch": 19.574226511495883,
      "grad_norm": 1.2044906616210938,
      "learning_rate": 8.043485665625888e-05,
      "loss": 0.02998480498790741,
      "step": 68960
    },
    {
      "epoch": 19.577065001419246,
      "grad_norm": 20.05248260498047,
      "learning_rate": 8.04320181663355e-05,
      "loss": 0.02988308072090149,
      "step": 68970
    },
    {
      "epoch": 19.579903491342606,
      "grad_norm": 15.3196439743042,
      "learning_rate": 8.042917967641215e-05,
      "loss": 0.014380241930484771,
      "step": 68980
    },
    {
      "epoch": 19.582741981265965,
      "grad_norm": 12.566217422485352,
      "learning_rate": 8.04263411864888e-05,
      "loss": 0.016055886447429658,
      "step": 68990
    },
    {
      "epoch": 19.585580471189328,
      "grad_norm": 0.3085486590862274,
      "learning_rate": 8.042350269656543e-05,
      "loss": 0.013671569526195526,
      "step": 69000
    },
    {
      "epoch": 19.585580471189328,
      "eval_accuracy": 0.9593692376168372,
      "eval_loss": 0.13230754435062408,
      "eval_runtime": 31.8925,
      "eval_samples_per_second": 493.126,
      "eval_steps_per_second": 7.713,
      "step": 69000
    },
    {
      "epoch": 19.588418961112687,
      "grad_norm": 0.4700465500354767,
      "learning_rate": 8.042066420664207e-05,
      "loss": 0.026226577162742615,
      "step": 69010
    },
    {
      "epoch": 19.59125745103605,
      "grad_norm": 5.278825759887695,
      "learning_rate": 8.041782571671871e-05,
      "loss": 0.00673285573720932,
      "step": 69020
    },
    {
      "epoch": 19.59409594095941,
      "grad_norm": 2.240048408508301,
      "learning_rate": 8.041498722679534e-05,
      "loss": 0.030140620470046998,
      "step": 69030
    },
    {
      "epoch": 19.59693443088277,
      "grad_norm": 1.9747363328933716,
      "learning_rate": 8.041214873687198e-05,
      "loss": 0.01521899402141571,
      "step": 69040
    },
    {
      "epoch": 19.59977292080613,
      "grad_norm": 13.838067054748535,
      "learning_rate": 8.040931024694864e-05,
      "loss": 0.013614904880523682,
      "step": 69050
    },
    {
      "epoch": 19.60261141072949,
      "grad_norm": 1.7056885957717896,
      "learning_rate": 8.040647175702527e-05,
      "loss": 0.016552548110485076,
      "step": 69060
    },
    {
      "epoch": 19.605449900652854,
      "grad_norm": 2.1976730823516846,
      "learning_rate": 8.04036332671019e-05,
      "loss": 0.02237391322851181,
      "step": 69070
    },
    {
      "epoch": 19.608288390576213,
      "grad_norm": 1.5214805603027344,
      "learning_rate": 8.040079477717855e-05,
      "loss": 0.01581161469221115,
      "step": 69080
    },
    {
      "epoch": 19.611126880499576,
      "grad_norm": 1.4361978769302368,
      "learning_rate": 8.039795628725519e-05,
      "loss": 0.014688368141651153,
      "step": 69090
    },
    {
      "epoch": 19.613965370422935,
      "grad_norm": 5.594478607177734,
      "learning_rate": 8.039511779733182e-05,
      "loss": 0.014558854699134826,
      "step": 69100
    },
    {
      "epoch": 19.616803860346295,
      "grad_norm": 10.732208251953125,
      "learning_rate": 8.039227930740846e-05,
      "loss": 0.02499958574771881,
      "step": 69110
    },
    {
      "epoch": 19.619642350269658,
      "grad_norm": 4.799899101257324,
      "learning_rate": 8.03894408174851e-05,
      "loss": 0.02225075364112854,
      "step": 69120
    },
    {
      "epoch": 19.622480840193017,
      "grad_norm": 4.448549747467041,
      "learning_rate": 8.038660232756174e-05,
      "loss": 0.035136446356773376,
      "step": 69130
    },
    {
      "epoch": 19.62531933011638,
      "grad_norm": 0.06879843026399612,
      "learning_rate": 8.038376383763838e-05,
      "loss": 0.022039005160331727,
      "step": 69140
    },
    {
      "epoch": 19.62815782003974,
      "grad_norm": 14.831464767456055,
      "learning_rate": 8.038092534771502e-05,
      "loss": 0.026927202939987183,
      "step": 69150
    },
    {
      "epoch": 19.6309963099631,
      "grad_norm": 3.4224581718444824,
      "learning_rate": 8.037808685779165e-05,
      "loss": 0.03298492729663849,
      "step": 69160
    },
    {
      "epoch": 19.63383479988646,
      "grad_norm": 10.462454795837402,
      "learning_rate": 8.03752483678683e-05,
      "loss": 0.02494146078824997,
      "step": 69170
    },
    {
      "epoch": 19.63667328980982,
      "grad_norm": 2.098823308944702,
      "learning_rate": 8.037240987794493e-05,
      "loss": 0.032711532711982724,
      "step": 69180
    },
    {
      "epoch": 19.639511779733184,
      "grad_norm": 1.413672685623169,
      "learning_rate": 8.036957138802158e-05,
      "loss": 0.020758078992366792,
      "step": 69190
    },
    {
      "epoch": 19.642350269656543,
      "grad_norm": 0.10148569941520691,
      "learning_rate": 8.036673289809822e-05,
      "loss": 0.009158961474895477,
      "step": 69200
    },
    {
      "epoch": 19.645188759579902,
      "grad_norm": 7.626468658447266,
      "learning_rate": 8.036389440817486e-05,
      "loss": 0.023182570934295654,
      "step": 69210
    },
    {
      "epoch": 19.648027249503265,
      "grad_norm": 2.296173334121704,
      "learning_rate": 8.036105591825149e-05,
      "loss": 0.008154605329036713,
      "step": 69220
    },
    {
      "epoch": 19.650865739426624,
      "grad_norm": 2.589481830596924,
      "learning_rate": 8.035821742832813e-05,
      "loss": 0.02610575556755066,
      "step": 69230
    },
    {
      "epoch": 19.653704229349987,
      "grad_norm": 7.635892868041992,
      "learning_rate": 8.035537893840477e-05,
      "loss": 0.012900859117507935,
      "step": 69240
    },
    {
      "epoch": 19.656542719273347,
      "grad_norm": 2.0708858966827393,
      "learning_rate": 8.035254044848141e-05,
      "loss": 0.012569186091423035,
      "step": 69250
    },
    {
      "epoch": 19.659381209196706,
      "grad_norm": 6.0787787437438965,
      "learning_rate": 8.034970195855805e-05,
      "loss": 0.012606383860111236,
      "step": 69260
    },
    {
      "epoch": 19.66221969912007,
      "grad_norm": 2.274489402770996,
      "learning_rate": 8.03468634686347e-05,
      "loss": 0.022255726158618927,
      "step": 69270
    },
    {
      "epoch": 19.665058189043428,
      "grad_norm": 1.3657938241958618,
      "learning_rate": 8.034402497871133e-05,
      "loss": 0.014625145494937897,
      "step": 69280
    },
    {
      "epoch": 19.66789667896679,
      "grad_norm": 4.934107303619385,
      "learning_rate": 8.034118648878796e-05,
      "loss": 0.009336713701486588,
      "step": 69290
    },
    {
      "epoch": 19.67073516889015,
      "grad_norm": 0.7426437139511108,
      "learning_rate": 8.03383479988646e-05,
      "loss": 0.011014585942029953,
      "step": 69300
    },
    {
      "epoch": 19.67357365881351,
      "grad_norm": 0.8515448570251465,
      "learning_rate": 8.033550950894125e-05,
      "loss": 0.013521890342235564,
      "step": 69310
    },
    {
      "epoch": 19.676412148736873,
      "grad_norm": 0.4950335621833801,
      "learning_rate": 8.033267101901789e-05,
      "loss": 0.011190605908632278,
      "step": 69320
    },
    {
      "epoch": 19.679250638660232,
      "grad_norm": 8.686479568481445,
      "learning_rate": 8.032983252909453e-05,
      "loss": 0.0067523129284381865,
      "step": 69330
    },
    {
      "epoch": 19.682089128583595,
      "grad_norm": 1.1687695980072021,
      "learning_rate": 8.032699403917117e-05,
      "loss": 0.01453712284564972,
      "step": 69340
    },
    {
      "epoch": 19.684927618506954,
      "grad_norm": 6.46031379699707,
      "learning_rate": 8.03241555492478e-05,
      "loss": 0.010612066835165024,
      "step": 69350
    },
    {
      "epoch": 19.687766108430313,
      "grad_norm": 7.104761600494385,
      "learning_rate": 8.032131705932444e-05,
      "loss": 0.018320584297180177,
      "step": 69360
    },
    {
      "epoch": 19.690604598353676,
      "grad_norm": 7.480422496795654,
      "learning_rate": 8.031847856940108e-05,
      "loss": 0.024844080209732056,
      "step": 69370
    },
    {
      "epoch": 19.693443088277036,
      "grad_norm": 1.9221047163009644,
      "learning_rate": 8.031564007947772e-05,
      "loss": 0.0582061767578125,
      "step": 69380
    },
    {
      "epoch": 19.6962815782004,
      "grad_norm": 0.8026178479194641,
      "learning_rate": 8.031280158955436e-05,
      "loss": 0.007932065427303315,
      "step": 69390
    },
    {
      "epoch": 19.699120068123758,
      "grad_norm": 5.988377094268799,
      "learning_rate": 8.0309963099631e-05,
      "loss": 0.008467693626880646,
      "step": 69400
    },
    {
      "epoch": 19.701958558047117,
      "grad_norm": 1.679275631904602,
      "learning_rate": 8.030712460970765e-05,
      "loss": 0.03526282012462616,
      "step": 69410
    },
    {
      "epoch": 19.70479704797048,
      "grad_norm": 11.361992835998535,
      "learning_rate": 8.030428611978427e-05,
      "loss": 0.01715822219848633,
      "step": 69420
    },
    {
      "epoch": 19.70763553789384,
      "grad_norm": 0.5744953751564026,
      "learning_rate": 8.030144762986091e-05,
      "loss": 0.02265937626361847,
      "step": 69430
    },
    {
      "epoch": 19.710474027817202,
      "grad_norm": 1.8528063297271729,
      "learning_rate": 8.029860913993756e-05,
      "loss": 0.013503438234329224,
      "step": 69440
    },
    {
      "epoch": 19.71331251774056,
      "grad_norm": 2.933100938796997,
      "learning_rate": 8.02957706500142e-05,
      "loss": 0.017415373027324675,
      "step": 69450
    },
    {
      "epoch": 19.716151007663925,
      "grad_norm": 5.731636047363281,
      "learning_rate": 8.029293216009084e-05,
      "loss": 0.015204900503158569,
      "step": 69460
    },
    {
      "epoch": 19.718989497587284,
      "grad_norm": 2.7832765579223633,
      "learning_rate": 8.029009367016748e-05,
      "loss": 0.030335879325866698,
      "step": 69470
    },
    {
      "epoch": 19.721827987510643,
      "grad_norm": 6.3767900466918945,
      "learning_rate": 8.028725518024411e-05,
      "loss": 0.018050846457481385,
      "step": 69480
    },
    {
      "epoch": 19.724666477434006,
      "grad_norm": 3.4176506996154785,
      "learning_rate": 8.028441669032075e-05,
      "loss": 0.017700466513633727,
      "step": 69490
    },
    {
      "epoch": 19.727504967357365,
      "grad_norm": 4.745724678039551,
      "learning_rate": 8.028157820039739e-05,
      "loss": 0.01572309136390686,
      "step": 69500
    },
    {
      "epoch": 19.727504967357365,
      "eval_accuracy": 0.9616582946525084,
      "eval_loss": 0.12550418078899384,
      "eval_runtime": 31.6521,
      "eval_samples_per_second": 496.87,
      "eval_steps_per_second": 7.772,
      "step": 69500
    },
    {
      "epoch": 19.73034345728073,
      "grad_norm": 3.1548526287078857,
      "learning_rate": 8.027873971047403e-05,
      "loss": 0.028902781009674073,
      "step": 69510
    },
    {
      "epoch": 19.733181947204088,
      "grad_norm": 1.549743890762329,
      "learning_rate": 8.027590122055067e-05,
      "loss": 0.024157786369323732,
      "step": 69520
    },
    {
      "epoch": 19.736020437127447,
      "grad_norm": 0.5650414228439331,
      "learning_rate": 8.027306273062732e-05,
      "loss": 0.01640947163105011,
      "step": 69530
    },
    {
      "epoch": 19.73885892705081,
      "grad_norm": 2.0185089111328125,
      "learning_rate": 8.027022424070396e-05,
      "loss": 0.02525729835033417,
      "step": 69540
    },
    {
      "epoch": 19.74169741697417,
      "grad_norm": 3.5437769889831543,
      "learning_rate": 8.026738575078058e-05,
      "loss": 0.01942034512758255,
      "step": 69550
    },
    {
      "epoch": 19.744535906897532,
      "grad_norm": 7.796629428863525,
      "learning_rate": 8.026454726085723e-05,
      "loss": 0.014662231504917144,
      "step": 69560
    },
    {
      "epoch": 19.74737439682089,
      "grad_norm": 1.618914246559143,
      "learning_rate": 8.026170877093387e-05,
      "loss": 0.025937393307685852,
      "step": 69570
    },
    {
      "epoch": 19.75021288674425,
      "grad_norm": 4.295910835266113,
      "learning_rate": 8.02588702810105e-05,
      "loss": 0.014453108608722686,
      "step": 69580
    },
    {
      "epoch": 19.753051376667614,
      "grad_norm": 2.167728900909424,
      "learning_rate": 8.025603179108715e-05,
      "loss": 0.012874837219715118,
      "step": 69590
    },
    {
      "epoch": 19.755889866590973,
      "grad_norm": 12.813432693481445,
      "learning_rate": 8.025319330116379e-05,
      "loss": 0.021997025609016417,
      "step": 69600
    },
    {
      "epoch": 19.758728356514336,
      "grad_norm": 9.523062705993652,
      "learning_rate": 8.025035481124042e-05,
      "loss": 0.01497279703617096,
      "step": 69610
    },
    {
      "epoch": 19.761566846437695,
      "grad_norm": 5.072718620300293,
      "learning_rate": 8.024751632131706e-05,
      "loss": 0.02813688516616821,
      "step": 69620
    },
    {
      "epoch": 19.764405336361055,
      "grad_norm": 4.230207920074463,
      "learning_rate": 8.02446778313937e-05,
      "loss": 0.015277157723903655,
      "step": 69630
    },
    {
      "epoch": 19.767243826284417,
      "grad_norm": 1.1602848768234253,
      "learning_rate": 8.024183934147034e-05,
      "loss": 0.02338716685771942,
      "step": 69640
    },
    {
      "epoch": 19.770082316207777,
      "grad_norm": 1.6782395839691162,
      "learning_rate": 8.023900085154698e-05,
      "loss": 0.012429679185152054,
      "step": 69650
    },
    {
      "epoch": 19.77292080613114,
      "grad_norm": 10.235223770141602,
      "learning_rate": 8.023616236162363e-05,
      "loss": 0.014614701271057129,
      "step": 69660
    },
    {
      "epoch": 19.7757592960545,
      "grad_norm": 5.014599323272705,
      "learning_rate": 8.023332387170027e-05,
      "loss": 0.013847139477729798,
      "step": 69670
    },
    {
      "epoch": 19.77859778597786,
      "grad_norm": 9.123796463012695,
      "learning_rate": 8.02304853817769e-05,
      "loss": 0.01719532459974289,
      "step": 69680
    },
    {
      "epoch": 19.78143627590122,
      "grad_norm": 1.3546220064163208,
      "learning_rate": 8.022764689185354e-05,
      "loss": 0.005158590525388718,
      "step": 69690
    },
    {
      "epoch": 19.78427476582458,
      "grad_norm": 6.31652307510376,
      "learning_rate": 8.022480840193018e-05,
      "loss": 0.015220491588115692,
      "step": 69700
    },
    {
      "epoch": 19.787113255747943,
      "grad_norm": 9.803191184997559,
      "learning_rate": 8.02219699120068e-05,
      "loss": 0.019742248952388762,
      "step": 69710
    },
    {
      "epoch": 19.789951745671303,
      "grad_norm": 0.5209247469902039,
      "learning_rate": 8.021913142208346e-05,
      "loss": 0.00744052529335022,
      "step": 69720
    },
    {
      "epoch": 19.792790235594662,
      "grad_norm": 2.5495564937591553,
      "learning_rate": 8.02162929321601e-05,
      "loss": 0.01600113958120346,
      "step": 69730
    },
    {
      "epoch": 19.795628725518025,
      "grad_norm": 2.423266887664795,
      "learning_rate": 8.021345444223673e-05,
      "loss": 0.009242112934589385,
      "step": 69740
    },
    {
      "epoch": 19.798467215441384,
      "grad_norm": 11.537529945373535,
      "learning_rate": 8.021061595231337e-05,
      "loss": 0.01511542797088623,
      "step": 69750
    },
    {
      "epoch": 19.801305705364747,
      "grad_norm": 0.7733875513076782,
      "learning_rate": 8.020777746239001e-05,
      "loss": 0.019382932782173158,
      "step": 69760
    },
    {
      "epoch": 19.804144195288107,
      "grad_norm": 7.190354824066162,
      "learning_rate": 8.020493897246665e-05,
      "loss": 0.016523183882236482,
      "step": 69770
    },
    {
      "epoch": 19.806982685211466,
      "grad_norm": 1.3888078927993774,
      "learning_rate": 8.020210048254328e-05,
      "loss": 0.016259367763996124,
      "step": 69780
    },
    {
      "epoch": 19.80982117513483,
      "grad_norm": 4.544417381286621,
      "learning_rate": 8.019926199261994e-05,
      "loss": 0.03132299184799194,
      "step": 69790
    },
    {
      "epoch": 19.812659665058188,
      "grad_norm": 1.9963366985321045,
      "learning_rate": 8.019642350269658e-05,
      "loss": 0.007616961002349853,
      "step": 69800
    },
    {
      "epoch": 19.81549815498155,
      "grad_norm": 4.9850993156433105,
      "learning_rate": 8.01935850127732e-05,
      "loss": 0.01671280264854431,
      "step": 69810
    },
    {
      "epoch": 19.81833664490491,
      "grad_norm": 2.2570724487304688,
      "learning_rate": 8.019074652284985e-05,
      "loss": 0.02907361090183258,
      "step": 69820
    },
    {
      "epoch": 19.82117513482827,
      "grad_norm": 9.15283489227295,
      "learning_rate": 8.018790803292649e-05,
      "loss": 0.00781503990292549,
      "step": 69830
    },
    {
      "epoch": 19.824013624751633,
      "grad_norm": 0.6080517768859863,
      "learning_rate": 8.018506954300312e-05,
      "loss": 0.014565247297286987,
      "step": 69840
    },
    {
      "epoch": 19.826852114674992,
      "grad_norm": 10.303339004516602,
      "learning_rate": 8.018223105307977e-05,
      "loss": 0.0220239520072937,
      "step": 69850
    },
    {
      "epoch": 19.829690604598355,
      "grad_norm": 6.275253772735596,
      "learning_rate": 8.017939256315641e-05,
      "loss": 0.013669410347938537,
      "step": 69860
    },
    {
      "epoch": 19.832529094521714,
      "grad_norm": 0.2501889169216156,
      "learning_rate": 8.017655407323304e-05,
      "loss": 0.017661070823669432,
      "step": 69870
    },
    {
      "epoch": 19.835367584445073,
      "grad_norm": 1.7205308675765991,
      "learning_rate": 8.017371558330968e-05,
      "loss": 0.030998238921165468,
      "step": 69880
    },
    {
      "epoch": 19.838206074368436,
      "grad_norm": 2.869626760482788,
      "learning_rate": 8.017087709338632e-05,
      "loss": 0.01047343760728836,
      "step": 69890
    },
    {
      "epoch": 19.841044564291796,
      "grad_norm": 2.089573621749878,
      "learning_rate": 8.016803860346296e-05,
      "loss": 0.01836358904838562,
      "step": 69900
    },
    {
      "epoch": 19.84388305421516,
      "grad_norm": 0.397853285074234,
      "learning_rate": 8.016520011353959e-05,
      "loss": 0.012542331218719482,
      "step": 69910
    },
    {
      "epoch": 19.846721544138518,
      "grad_norm": 10.944541931152344,
      "learning_rate": 8.016236162361625e-05,
      "loss": 0.017021429538726807,
      "step": 69920
    },
    {
      "epoch": 19.84956003406188,
      "grad_norm": 1.6689707040786743,
      "learning_rate": 8.015952313369289e-05,
      "loss": 0.017763373255729676,
      "step": 69930
    },
    {
      "epoch": 19.85239852398524,
      "grad_norm": 1.750494122505188,
      "learning_rate": 8.015668464376952e-05,
      "loss": 0.009762341529130936,
      "step": 69940
    },
    {
      "epoch": 19.8552370139086,
      "grad_norm": 3.302927017211914,
      "learning_rate": 8.015384615384616e-05,
      "loss": 0.03229142427444458,
      "step": 69950
    },
    {
      "epoch": 19.858075503831962,
      "grad_norm": 3.7164931297302246,
      "learning_rate": 8.01510076639228e-05,
      "loss": 0.024825701117515565,
      "step": 69960
    },
    {
      "epoch": 19.86091399375532,
      "grad_norm": 0.9139042496681213,
      "learning_rate": 8.014816917399943e-05,
      "loss": 0.014837440848350526,
      "step": 69970
    },
    {
      "epoch": 19.863752483678685,
      "grad_norm": 6.463283061981201,
      "learning_rate": 8.014533068407608e-05,
      "loss": 0.02670870125293732,
      "step": 69980
    },
    {
      "epoch": 19.866590973602044,
      "grad_norm": 9.262341499328613,
      "learning_rate": 8.014249219415272e-05,
      "loss": 0.03314664363861084,
      "step": 69990
    },
    {
      "epoch": 19.869429463525403,
      "grad_norm": 0.8242834210395813,
      "learning_rate": 8.013965370422935e-05,
      "loss": 0.007536008954048157,
      "step": 70000
    },
    {
      "epoch": 19.869429463525403,
      "eval_accuracy": 0.962802823170344,
      "eval_loss": 0.12460049986839294,
      "eval_runtime": 31.5585,
      "eval_samples_per_second": 498.344,
      "eval_steps_per_second": 7.795,
      "step": 70000
    },
    {
      "epoch": 19.872267953448766,
      "grad_norm": 6.0503315925598145,
      "learning_rate": 8.013681521430599e-05,
      "loss": 0.01712670028209686,
      "step": 70010
    },
    {
      "epoch": 19.875106443372125,
      "grad_norm": 0.711719810962677,
      "learning_rate": 8.013397672438263e-05,
      "loss": 0.018879419565200804,
      "step": 70020
    },
    {
      "epoch": 19.87794493329549,
      "grad_norm": 6.612565517425537,
      "learning_rate": 8.013113823445928e-05,
      "loss": 0.015574388206005096,
      "step": 70030
    },
    {
      "epoch": 19.880783423218848,
      "grad_norm": 1.5285192728042603,
      "learning_rate": 8.01282997445359e-05,
      "loss": 0.01017269566655159,
      "step": 70040
    },
    {
      "epoch": 19.883621913142207,
      "grad_norm": 0.736617922782898,
      "learning_rate": 8.012546125461256e-05,
      "loss": 0.018401379883289336,
      "step": 70050
    },
    {
      "epoch": 19.88646040306557,
      "grad_norm": 8.752188682556152,
      "learning_rate": 8.012262276468919e-05,
      "loss": 0.021568451821804047,
      "step": 70060
    },
    {
      "epoch": 19.88929889298893,
      "grad_norm": 16.844388961791992,
      "learning_rate": 8.011978427476583e-05,
      "loss": 0.022290828824043273,
      "step": 70070
    },
    {
      "epoch": 19.892137382912292,
      "grad_norm": 1.6637965440750122,
      "learning_rate": 8.011694578484247e-05,
      "loss": 0.013622608780860902,
      "step": 70080
    },
    {
      "epoch": 19.89497587283565,
      "grad_norm": 10.465700149536133,
      "learning_rate": 8.011410729491911e-05,
      "loss": 0.029503101110458375,
      "step": 70090
    },
    {
      "epoch": 19.89781436275901,
      "grad_norm": 1.5931602716445923,
      "learning_rate": 8.011126880499574e-05,
      "loss": 0.006964816898107529,
      "step": 70100
    },
    {
      "epoch": 19.900652852682374,
      "grad_norm": 16.223539352416992,
      "learning_rate": 8.010843031507238e-05,
      "loss": 0.023791635036468507,
      "step": 70110
    },
    {
      "epoch": 19.903491342605733,
      "grad_norm": 0.9101476073265076,
      "learning_rate": 8.010559182514903e-05,
      "loss": 0.028920650482177734,
      "step": 70120
    },
    {
      "epoch": 19.906329832529096,
      "grad_norm": 2.0515339374542236,
      "learning_rate": 8.010275333522566e-05,
      "loss": 0.01323588639497757,
      "step": 70130
    },
    {
      "epoch": 19.909168322452455,
      "grad_norm": 5.446404933929443,
      "learning_rate": 8.00999148453023e-05,
      "loss": 0.00625767856836319,
      "step": 70140
    },
    {
      "epoch": 19.912006812375814,
      "grad_norm": 5.3146467208862305,
      "learning_rate": 8.009707635537894e-05,
      "loss": 0.0213604599237442,
      "step": 70150
    },
    {
      "epoch": 19.914845302299177,
      "grad_norm": 1.8785371780395508,
      "learning_rate": 8.009423786545557e-05,
      "loss": 0.033167564868927,
      "step": 70160
    },
    {
      "epoch": 19.917683792222537,
      "grad_norm": 0.9094081521034241,
      "learning_rate": 8.009139937553221e-05,
      "loss": 0.0048977967351675035,
      "step": 70170
    },
    {
      "epoch": 19.9205222821459,
      "grad_norm": 0.49379199743270874,
      "learning_rate": 8.008856088560887e-05,
      "loss": 0.026901325583457945,
      "step": 70180
    },
    {
      "epoch": 19.92336077206926,
      "grad_norm": 10.196534156799316,
      "learning_rate": 8.00857223956855e-05,
      "loss": 0.017138615250587463,
      "step": 70190
    },
    {
      "epoch": 19.92619926199262,
      "grad_norm": 8.169548988342285,
      "learning_rate": 8.008288390576214e-05,
      "loss": 0.02445916533470154,
      "step": 70200
    },
    {
      "epoch": 19.92903775191598,
      "grad_norm": 2.5194480419158936,
      "learning_rate": 8.008004541583878e-05,
      "loss": 0.008446452021598817,
      "step": 70210
    },
    {
      "epoch": 19.93187624183934,
      "grad_norm": 1.267454981803894,
      "learning_rate": 8.007720692591542e-05,
      "loss": 0.0110305555164814,
      "step": 70220
    },
    {
      "epoch": 19.934714731762703,
      "grad_norm": 2.1513097286224365,
      "learning_rate": 8.007436843599205e-05,
      "loss": 0.014187799394130706,
      "step": 70230
    },
    {
      "epoch": 19.937553221686063,
      "grad_norm": 6.587697505950928,
      "learning_rate": 8.007152994606869e-05,
      "loss": 0.013854707777500152,
      "step": 70240
    },
    {
      "epoch": 19.940391711609422,
      "grad_norm": 0.5239924192428589,
      "learning_rate": 8.006869145614534e-05,
      "loss": 0.0256253182888031,
      "step": 70250
    },
    {
      "epoch": 19.943230201532785,
      "grad_norm": 0.8720202445983887,
      "learning_rate": 8.006585296622197e-05,
      "loss": 0.006948457658290863,
      "step": 70260
    },
    {
      "epoch": 19.946068691456144,
      "grad_norm": 2.0155763626098633,
      "learning_rate": 8.006301447629861e-05,
      "loss": 0.013918130099773407,
      "step": 70270
    },
    {
      "epoch": 19.948907181379507,
      "grad_norm": 5.3240861892700195,
      "learning_rate": 8.006017598637526e-05,
      "loss": 0.01625025421380997,
      "step": 70280
    },
    {
      "epoch": 19.951745671302866,
      "grad_norm": 3.647378921508789,
      "learning_rate": 8.005733749645188e-05,
      "loss": 0.025415289402008056,
      "step": 70290
    },
    {
      "epoch": 19.95458416122623,
      "grad_norm": 5.258601665496826,
      "learning_rate": 8.005449900652852e-05,
      "loss": 0.020264589786529542,
      "step": 70300
    },
    {
      "epoch": 19.95742265114959,
      "grad_norm": 0.9246200919151306,
      "learning_rate": 8.005166051660517e-05,
      "loss": 0.010485036671161652,
      "step": 70310
    },
    {
      "epoch": 19.960261141072948,
      "grad_norm": 8.737808227539062,
      "learning_rate": 8.004882202668181e-05,
      "loss": 0.020587009191513062,
      "step": 70320
    },
    {
      "epoch": 19.96309963099631,
      "grad_norm": 4.19546365737915,
      "learning_rate": 8.004598353675845e-05,
      "loss": 0.022262267768383026,
      "step": 70330
    },
    {
      "epoch": 19.96593812091967,
      "grad_norm": 0.8448171019554138,
      "learning_rate": 8.004314504683509e-05,
      "loss": 0.005003838986158371,
      "step": 70340
    },
    {
      "epoch": 19.968776610843033,
      "grad_norm": 1.529058814048767,
      "learning_rate": 8.004030655691173e-05,
      "loss": 0.016503478586673736,
      "step": 70350
    },
    {
      "epoch": 19.971615100766392,
      "grad_norm": 2.432154655456543,
      "learning_rate": 8.003746806698836e-05,
      "loss": 0.02536535859107971,
      "step": 70360
    },
    {
      "epoch": 19.974453590689752,
      "grad_norm": 3.5854225158691406,
      "learning_rate": 8.0034629577065e-05,
      "loss": 0.016874317824840546,
      "step": 70370
    },
    {
      "epoch": 19.977292080613115,
      "grad_norm": 1.7768503427505493,
      "learning_rate": 8.003179108714166e-05,
      "loss": 0.011259014159440995,
      "step": 70380
    },
    {
      "epoch": 19.980130570536474,
      "grad_norm": 2.8667845726013184,
      "learning_rate": 8.002895259721828e-05,
      "loss": 0.015071055293083191,
      "step": 70390
    },
    {
      "epoch": 19.982969060459837,
      "grad_norm": 2.7725303173065186,
      "learning_rate": 8.002611410729492e-05,
      "loss": 0.012452834844589233,
      "step": 70400
    },
    {
      "epoch": 19.985807550383196,
      "grad_norm": 3.431821584701538,
      "learning_rate": 8.002327561737157e-05,
      "loss": 0.016131222248077393,
      "step": 70410
    },
    {
      "epoch": 19.988646040306556,
      "grad_norm": 3.3145828247070312,
      "learning_rate": 8.00204371274482e-05,
      "loss": 0.013110698759555816,
      "step": 70420
    },
    {
      "epoch": 19.99148453022992,
      "grad_norm": 0.8674981594085693,
      "learning_rate": 8.001759863752484e-05,
      "loss": 0.012944674491882325,
      "step": 70430
    },
    {
      "epoch": 19.994323020153278,
      "grad_norm": 5.62935209274292,
      "learning_rate": 8.001476014760148e-05,
      "loss": 0.009259013086557388,
      "step": 70440
    },
    {
      "epoch": 19.99716151007664,
      "grad_norm": 2.8510632514953613,
      "learning_rate": 8.001192165767812e-05,
      "loss": 0.013773289322853089,
      "step": 70450
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.5516908168792725,
      "learning_rate": 8.000908316775476e-05,
      "loss": 0.024528275430202483,
      "step": 70460
    },
    {
      "epoch": 20.00283848992336,
      "grad_norm": 7.436856746673584,
      "learning_rate": 8.00062446778314e-05,
      "loss": 0.016245432198047638,
      "step": 70470
    },
    {
      "epoch": 20.005676979846722,
      "grad_norm": 1.6136023998260498,
      "learning_rate": 8.000340618790804e-05,
      "loss": 0.011430624127388,
      "step": 70480
    },
    {
      "epoch": 20.00851546977008,
      "grad_norm": 0.8578248620033264,
      "learning_rate": 8.000056769798467e-05,
      "loss": 0.007121416926383972,
      "step": 70490
    },
    {
      "epoch": 20.011353959693444,
      "grad_norm": 8.089055061340332,
      "learning_rate": 7.999772920806131e-05,
      "loss": 0.016089314222335817,
      "step": 70500
    },
    {
      "epoch": 20.011353959693444,
      "eval_accuracy": 0.9650282952883577,
      "eval_loss": 0.112466961145401,
      "eval_runtime": 31.6424,
      "eval_samples_per_second": 497.023,
      "eval_steps_per_second": 7.774,
      "step": 70500
    },
    {
      "epoch": 20.014192449616804,
      "grad_norm": 0.6011362671852112,
      "learning_rate": 7.999489071813795e-05,
      "loss": 0.008259302377700806,
      "step": 70510
    },
    {
      "epoch": 20.017030939540163,
      "grad_norm": 2.6536293029785156,
      "learning_rate": 7.99920522282146e-05,
      "loss": 0.020915262401103973,
      "step": 70520
    },
    {
      "epoch": 20.019869429463526,
      "grad_norm": 5.806552410125732,
      "learning_rate": 7.998921373829124e-05,
      "loss": 0.010254844278097152,
      "step": 70530
    },
    {
      "epoch": 20.022707919386885,
      "grad_norm": 1.3889726400375366,
      "learning_rate": 7.998637524836788e-05,
      "loss": 0.008887986093759537,
      "step": 70540
    },
    {
      "epoch": 20.025546409310248,
      "grad_norm": 0.1703825443983078,
      "learning_rate": 7.99835367584445e-05,
      "loss": 0.013272418081760407,
      "step": 70550
    },
    {
      "epoch": 20.028384899233608,
      "grad_norm": 2.1090433597564697,
      "learning_rate": 7.998069826852115e-05,
      "loss": 0.006426654756069183,
      "step": 70560
    },
    {
      "epoch": 20.031223389156967,
      "grad_norm": 1.681519627571106,
      "learning_rate": 7.997785977859779e-05,
      "loss": 0.015022775530815125,
      "step": 70570
    },
    {
      "epoch": 20.03406187908033,
      "grad_norm": 4.431528091430664,
      "learning_rate": 7.997502128867443e-05,
      "loss": 0.019916574656963348,
      "step": 70580
    },
    {
      "epoch": 20.03690036900369,
      "grad_norm": 0.5745396018028259,
      "learning_rate": 7.997218279875107e-05,
      "loss": 0.011699742078781128,
      "step": 70590
    },
    {
      "epoch": 20.039738858927052,
      "grad_norm": 9.14401912689209,
      "learning_rate": 7.996934430882771e-05,
      "loss": 0.01926853507757187,
      "step": 70600
    },
    {
      "epoch": 20.04257734885041,
      "grad_norm": 9.073312759399414,
      "learning_rate": 7.996650581890435e-05,
      "loss": 0.015802636742591858,
      "step": 70610
    },
    {
      "epoch": 20.04541583877377,
      "grad_norm": 8.742959022521973,
      "learning_rate": 7.996366732898098e-05,
      "loss": 0.022052106261253358,
      "step": 70620
    },
    {
      "epoch": 20.048254328697134,
      "grad_norm": 0.25531259179115295,
      "learning_rate": 7.996082883905762e-05,
      "loss": 0.009354813396930695,
      "step": 70630
    },
    {
      "epoch": 20.051092818620493,
      "grad_norm": 1.383939504623413,
      "learning_rate": 7.995799034913426e-05,
      "loss": 0.018287202715873717,
      "step": 70640
    },
    {
      "epoch": 20.053931308543856,
      "grad_norm": 2.431122303009033,
      "learning_rate": 7.99551518592109e-05,
      "loss": 0.010712487995624543,
      "step": 70650
    },
    {
      "epoch": 20.056769798467215,
      "grad_norm": 0.8003904819488525,
      "learning_rate": 7.995231336928755e-05,
      "loss": 0.012613195180892944,
      "step": 70660
    },
    {
      "epoch": 20.059608288390578,
      "grad_norm": 4.319883346557617,
      "learning_rate": 7.994947487936419e-05,
      "loss": 0.0128926619887352,
      "step": 70670
    },
    {
      "epoch": 20.062446778313937,
      "grad_norm": 0.8970504403114319,
      "learning_rate": 7.994663638944082e-05,
      "loss": 0.008167168498039246,
      "step": 70680
    },
    {
      "epoch": 20.065285268237297,
      "grad_norm": 3.034276247024536,
      "learning_rate": 7.994379789951746e-05,
      "loss": 0.013646291196346283,
      "step": 70690
    },
    {
      "epoch": 20.06812375816066,
      "grad_norm": 12.887155532836914,
      "learning_rate": 7.99409594095941e-05,
      "loss": 0.018387052416801452,
      "step": 70700
    },
    {
      "epoch": 20.07096224808402,
      "grad_norm": 3.731816291809082,
      "learning_rate": 7.993812091967074e-05,
      "loss": 0.007870706915855407,
      "step": 70710
    },
    {
      "epoch": 20.07380073800738,
      "grad_norm": 0.980358898639679,
      "learning_rate": 7.993528242974738e-05,
      "loss": 0.00835033431649208,
      "step": 70720
    },
    {
      "epoch": 20.07663922793074,
      "grad_norm": 0.06576377153396606,
      "learning_rate": 7.993244393982402e-05,
      "loss": 0.00818023383617401,
      "step": 70730
    },
    {
      "epoch": 20.0794777178541,
      "grad_norm": 1.553149700164795,
      "learning_rate": 7.992960544990066e-05,
      "loss": 0.019675710797309877,
      "step": 70740
    },
    {
      "epoch": 20.082316207777463,
      "grad_norm": 0.5809218287467957,
      "learning_rate": 7.992676695997729e-05,
      "loss": 0.009852318465709687,
      "step": 70750
    },
    {
      "epoch": 20.085154697700823,
      "grad_norm": 5.539701461791992,
      "learning_rate": 7.992392847005393e-05,
      "loss": 0.017030355334281922,
      "step": 70760
    },
    {
      "epoch": 20.087993187624186,
      "grad_norm": 19.671899795532227,
      "learning_rate": 7.992108998013057e-05,
      "loss": 0.01973271667957306,
      "step": 70770
    },
    {
      "epoch": 20.090831677547545,
      "grad_norm": 0.4359962046146393,
      "learning_rate": 7.991825149020722e-05,
      "loss": 0.016238373517990113,
      "step": 70780
    },
    {
      "epoch": 20.093670167470904,
      "grad_norm": 1.2678791284561157,
      "learning_rate": 7.991541300028386e-05,
      "loss": 0.01931024491786957,
      "step": 70790
    },
    {
      "epoch": 20.096508657394267,
      "grad_norm": 0.18864713609218597,
      "learning_rate": 7.99125745103605e-05,
      "loss": 0.016872239112854005,
      "step": 70800
    },
    {
      "epoch": 20.099347147317626,
      "grad_norm": 1.9875191450119019,
      "learning_rate": 7.990973602043713e-05,
      "loss": 0.011762619018554688,
      "step": 70810
    },
    {
      "epoch": 20.10218563724099,
      "grad_norm": 2.685734272003174,
      "learning_rate": 7.990689753051377e-05,
      "loss": 0.012342245131731034,
      "step": 70820
    },
    {
      "epoch": 20.10502412716435,
      "grad_norm": 4.351536750793457,
      "learning_rate": 7.990405904059041e-05,
      "loss": 0.021676430106163026,
      "step": 70830
    },
    {
      "epoch": 20.107862617087708,
      "grad_norm": 3.1585872173309326,
      "learning_rate": 7.990122055066705e-05,
      "loss": 0.010295553505420685,
      "step": 70840
    },
    {
      "epoch": 20.11070110701107,
      "grad_norm": 0.46891143918037415,
      "learning_rate": 7.989838206074369e-05,
      "loss": 0.007197104394435883,
      "step": 70850
    },
    {
      "epoch": 20.11353959693443,
      "grad_norm": 9.117058753967285,
      "learning_rate": 7.989554357082033e-05,
      "loss": 0.026901519298553465,
      "step": 70860
    },
    {
      "epoch": 20.116378086857793,
      "grad_norm": 7.406280517578125,
      "learning_rate": 7.989270508089697e-05,
      "loss": 0.017835670709609987,
      "step": 70870
    },
    {
      "epoch": 20.119216576781152,
      "grad_norm": 0.4278368055820465,
      "learning_rate": 7.98898665909736e-05,
      "loss": 0.008642089366912842,
      "step": 70880
    },
    {
      "epoch": 20.12205506670451,
      "grad_norm": 4.760260105133057,
      "learning_rate": 7.988702810105024e-05,
      "loss": 0.017266520857810976,
      "step": 70890
    },
    {
      "epoch": 20.124893556627875,
      "grad_norm": 3.2486774921417236,
      "learning_rate": 7.988418961112689e-05,
      "loss": 0.021109884977340697,
      "step": 70900
    },
    {
      "epoch": 20.127732046551234,
      "grad_norm": 7.284914016723633,
      "learning_rate": 7.988135112120351e-05,
      "loss": 0.02543071210384369,
      "step": 70910
    },
    {
      "epoch": 20.130570536474597,
      "grad_norm": 3.347695827484131,
      "learning_rate": 7.987851263128017e-05,
      "loss": 0.013283728063106537,
      "step": 70920
    },
    {
      "epoch": 20.133409026397956,
      "grad_norm": 2.3983588218688965,
      "learning_rate": 7.987567414135681e-05,
      "loss": 0.010482189059257508,
      "step": 70930
    },
    {
      "epoch": 20.136247516321315,
      "grad_norm": 2.587460994720459,
      "learning_rate": 7.987283565143344e-05,
      "loss": 0.012927180528640747,
      "step": 70940
    },
    {
      "epoch": 20.13908600624468,
      "grad_norm": 1.217547059059143,
      "learning_rate": 7.986999716151008e-05,
      "loss": 0.01064080446958542,
      "step": 70950
    },
    {
      "epoch": 20.141924496168038,
      "grad_norm": 4.618635654449463,
      "learning_rate": 7.986715867158672e-05,
      "loss": 0.016633853316307068,
      "step": 70960
    },
    {
      "epoch": 20.1447629860914,
      "grad_norm": 4.599666595458984,
      "learning_rate": 7.986432018166336e-05,
      "loss": 0.007615604996681213,
      "step": 70970
    },
    {
      "epoch": 20.14760147601476,
      "grad_norm": 7.990434646606445,
      "learning_rate": 7.986148169174e-05,
      "loss": 0.035049337148666385,
      "step": 70980
    },
    {
      "epoch": 20.15043996593812,
      "grad_norm": 4.115412712097168,
      "learning_rate": 7.985864320181664e-05,
      "loss": 0.02891141176223755,
      "step": 70990
    },
    {
      "epoch": 20.153278455861482,
      "grad_norm": 9.228787422180176,
      "learning_rate": 7.985580471189327e-05,
      "loss": 0.023379792273044587,
      "step": 71000
    },
    {
      "epoch": 20.153278455861482,
      "eval_accuracy": 0.9644560310294398,
      "eval_loss": 0.1151220053434372,
      "eval_runtime": 31.5462,
      "eval_samples_per_second": 498.538,
      "eval_steps_per_second": 7.798,
      "step": 71000
    },
    {
      "epoch": 20.15611694578484,
      "grad_norm": 0.9651784300804138,
      "learning_rate": 7.985296622196991e-05,
      "loss": 0.008450717478990556,
      "step": 71010
    },
    {
      "epoch": 20.158955435708204,
      "grad_norm": 1.4010860919952393,
      "learning_rate": 7.985012773204655e-05,
      "loss": 0.026739388704299927,
      "step": 71020
    },
    {
      "epoch": 20.161793925631564,
      "grad_norm": 0.8490244746208191,
      "learning_rate": 7.98472892421232e-05,
      "loss": 0.016658060252666473,
      "step": 71030
    },
    {
      "epoch": 20.164632415554923,
      "grad_norm": 0.5647145509719849,
      "learning_rate": 7.984445075219982e-05,
      "loss": 0.005608601123094558,
      "step": 71040
    },
    {
      "epoch": 20.167470905478286,
      "grad_norm": 7.153414249420166,
      "learning_rate": 7.984161226227648e-05,
      "loss": 0.010309536755084992,
      "step": 71050
    },
    {
      "epoch": 20.170309395401645,
      "grad_norm": 4.291512966156006,
      "learning_rate": 7.983877377235312e-05,
      "loss": 0.013422916829586028,
      "step": 71060
    },
    {
      "epoch": 20.173147885325008,
      "grad_norm": 8.023852348327637,
      "learning_rate": 7.983593528242975e-05,
      "loss": 0.010602827370166778,
      "step": 71070
    },
    {
      "epoch": 20.175986375248367,
      "grad_norm": 0.68719482421875,
      "learning_rate": 7.983309679250639e-05,
      "loss": 0.0068502120673656465,
      "step": 71080
    },
    {
      "epoch": 20.17882486517173,
      "grad_norm": 8.918876647949219,
      "learning_rate": 7.983025830258303e-05,
      "loss": 0.016236656904220582,
      "step": 71090
    },
    {
      "epoch": 20.18166335509509,
      "grad_norm": 0.6813021302223206,
      "learning_rate": 7.982741981265966e-05,
      "loss": 0.023597031831741333,
      "step": 71100
    },
    {
      "epoch": 20.18450184501845,
      "grad_norm": 0.4900694191455841,
      "learning_rate": 7.98245813227363e-05,
      "loss": 0.011836731433868408,
      "step": 71110
    },
    {
      "epoch": 20.187340334941812,
      "grad_norm": 10.9681978225708,
      "learning_rate": 7.982174283281295e-05,
      "loss": 0.020834103226661682,
      "step": 71120
    },
    {
      "epoch": 20.19017882486517,
      "grad_norm": 5.577882289886475,
      "learning_rate": 7.981890434288958e-05,
      "loss": 0.0135760098695755,
      "step": 71130
    },
    {
      "epoch": 20.193017314788534,
      "grad_norm": 0.5071020722389221,
      "learning_rate": 7.981606585296622e-05,
      "loss": 0.008869817852973938,
      "step": 71140
    },
    {
      "epoch": 20.195855804711893,
      "grad_norm": 6.140706539154053,
      "learning_rate": 7.981322736304287e-05,
      "loss": 0.015114164352416993,
      "step": 71150
    },
    {
      "epoch": 20.198694294635253,
      "grad_norm": 0.5593485236167908,
      "learning_rate": 7.98103888731195e-05,
      "loss": 0.015102194249629974,
      "step": 71160
    },
    {
      "epoch": 20.201532784558616,
      "grad_norm": 1.271063208580017,
      "learning_rate": 7.980755038319613e-05,
      "loss": 0.014655356109142304,
      "step": 71170
    },
    {
      "epoch": 20.204371274481975,
      "grad_norm": 2.1790623664855957,
      "learning_rate": 7.980471189327279e-05,
      "loss": 0.007778779417276382,
      "step": 71180
    },
    {
      "epoch": 20.207209764405338,
      "grad_norm": 1.5992292165756226,
      "learning_rate": 7.980187340334943e-05,
      "loss": 0.01679815649986267,
      "step": 71190
    },
    {
      "epoch": 20.210048254328697,
      "grad_norm": 1.2398967742919922,
      "learning_rate": 7.979903491342606e-05,
      "loss": 0.01423579454421997,
      "step": 71200
    },
    {
      "epoch": 20.212886744252057,
      "grad_norm": 0.20961888134479523,
      "learning_rate": 7.97961964235027e-05,
      "loss": 0.013106675446033477,
      "step": 71210
    },
    {
      "epoch": 20.21572523417542,
      "grad_norm": 7.693957805633545,
      "learning_rate": 7.979335793357934e-05,
      "loss": 0.018496458232402802,
      "step": 71220
    },
    {
      "epoch": 20.21856372409878,
      "grad_norm": 1.2590605020523071,
      "learning_rate": 7.979051944365597e-05,
      "loss": 0.01798439621925354,
      "step": 71230
    },
    {
      "epoch": 20.22140221402214,
      "grad_norm": 3.11205792427063,
      "learning_rate": 7.978768095373261e-05,
      "loss": 0.01695810705423355,
      "step": 71240
    },
    {
      "epoch": 20.2242407039455,
      "grad_norm": 1.706710934638977,
      "learning_rate": 7.978484246380927e-05,
      "loss": 0.009011100977659225,
      "step": 71250
    },
    {
      "epoch": 20.22707919386886,
      "grad_norm": 1.7847572565078735,
      "learning_rate": 7.97820039738859e-05,
      "loss": 0.013674171268939972,
      "step": 71260
    },
    {
      "epoch": 20.229917683792223,
      "grad_norm": 0.8101648092269897,
      "learning_rate": 7.977916548396253e-05,
      "loss": 0.017009571194648743,
      "step": 71270
    },
    {
      "epoch": 20.232756173715583,
      "grad_norm": 2.692492961883545,
      "learning_rate": 7.977632699403918e-05,
      "loss": 0.009096066653728484,
      "step": 71280
    },
    {
      "epoch": 20.235594663638945,
      "grad_norm": 10.445018768310547,
      "learning_rate": 7.977348850411582e-05,
      "loss": 0.01693447232246399,
      "step": 71290
    },
    {
      "epoch": 20.238433153562305,
      "grad_norm": 9.105262756347656,
      "learning_rate": 7.977065001419245e-05,
      "loss": 0.015661458671092986,
      "step": 71300
    },
    {
      "epoch": 20.241271643485664,
      "grad_norm": 12.29133415222168,
      "learning_rate": 7.97678115242691e-05,
      "loss": 0.01562212109565735,
      "step": 71310
    },
    {
      "epoch": 20.244110133409027,
      "grad_norm": 7.209648609161377,
      "learning_rate": 7.976497303434574e-05,
      "loss": 0.014545503258705138,
      "step": 71320
    },
    {
      "epoch": 20.246948623332386,
      "grad_norm": 1.1836363077163696,
      "learning_rate": 7.976213454442237e-05,
      "loss": 0.019958382844924925,
      "step": 71330
    },
    {
      "epoch": 20.24978711325575,
      "grad_norm": 2.5779213905334473,
      "learning_rate": 7.975929605449901e-05,
      "loss": 0.008409702777862548,
      "step": 71340
    },
    {
      "epoch": 20.25262560317911,
      "grad_norm": 16.579957962036133,
      "learning_rate": 7.975645756457565e-05,
      "loss": 0.028837484121322633,
      "step": 71350
    },
    {
      "epoch": 20.255464093102468,
      "grad_norm": 0.8662018775939941,
      "learning_rate": 7.975361907465228e-05,
      "loss": 0.012503746151924133,
      "step": 71360
    },
    {
      "epoch": 20.25830258302583,
      "grad_norm": 1.3396071195602417,
      "learning_rate": 7.975078058472892e-05,
      "loss": 0.009117329120635986,
      "step": 71370
    },
    {
      "epoch": 20.26114107294919,
      "grad_norm": 6.912156105041504,
      "learning_rate": 7.974794209480558e-05,
      "loss": 0.021028077602386473,
      "step": 71380
    },
    {
      "epoch": 20.263979562872553,
      "grad_norm": 1.0493494272232056,
      "learning_rate": 7.97451036048822e-05,
      "loss": 0.008940963447093964,
      "step": 71390
    },
    {
      "epoch": 20.266818052795912,
      "grad_norm": 3.1840806007385254,
      "learning_rate": 7.974226511495885e-05,
      "loss": 0.00422750860452652,
      "step": 71400
    },
    {
      "epoch": 20.26965654271927,
      "grad_norm": 5.137458324432373,
      "learning_rate": 7.973942662503549e-05,
      "loss": 0.009950690716505051,
      "step": 71410
    },
    {
      "epoch": 20.272495032642635,
      "grad_norm": 1.6262131929397583,
      "learning_rate": 7.973658813511213e-05,
      "loss": 0.011805801093578339,
      "step": 71420
    },
    {
      "epoch": 20.275333522565994,
      "grad_norm": 10.116072654724121,
      "learning_rate": 7.973374964518876e-05,
      "loss": 0.015828925371170043,
      "step": 71430
    },
    {
      "epoch": 20.278172012489357,
      "grad_norm": 12.012731552124023,
      "learning_rate": 7.97309111552654e-05,
      "loss": 0.028188514709472656,
      "step": 71440
    },
    {
      "epoch": 20.281010502412716,
      "grad_norm": 1.8385461568832397,
      "learning_rate": 7.972807266534205e-05,
      "loss": 0.013013137876987458,
      "step": 71450
    },
    {
      "epoch": 20.283848992336075,
      "grad_norm": 1.95732581615448,
      "learning_rate": 7.972523417541868e-05,
      "loss": 0.007443028688430786,
      "step": 71460
    },
    {
      "epoch": 20.28668748225944,
      "grad_norm": 5.327107906341553,
      "learning_rate": 7.972239568549532e-05,
      "loss": 0.008409370481967927,
      "step": 71470
    },
    {
      "epoch": 20.289525972182798,
      "grad_norm": 2.0625081062316895,
      "learning_rate": 7.971955719557196e-05,
      "loss": 0.018064771592617036,
      "step": 71480
    },
    {
      "epoch": 20.29236446210616,
      "grad_norm": 10.508517265319824,
      "learning_rate": 7.971671870564859e-05,
      "loss": 0.011872991174459457,
      "step": 71490
    },
    {
      "epoch": 20.29520295202952,
      "grad_norm": 8.100379943847656,
      "learning_rate": 7.971388021572523e-05,
      "loss": 0.024738314747810363,
      "step": 71500
    },
    {
      "epoch": 20.29520295202952,
      "eval_accuracy": 0.9661728238061932,
      "eval_loss": 0.10654716193675995,
      "eval_runtime": 31.8601,
      "eval_samples_per_second": 493.627,
      "eval_steps_per_second": 7.721,
      "step": 71500
    },
    {
      "epoch": 20.298041441952883,
      "grad_norm": 6.455258369445801,
      "learning_rate": 7.971104172580189e-05,
      "loss": 0.03311143219470978,
      "step": 71510
    },
    {
      "epoch": 20.300879931876242,
      "grad_norm": 3.1452577114105225,
      "learning_rate": 7.970820323587851e-05,
      "loss": 0.016617275774478912,
      "step": 71520
    },
    {
      "epoch": 20.3037184217996,
      "grad_norm": 13.45455265045166,
      "learning_rate": 7.970536474595516e-05,
      "loss": 0.031100407242774963,
      "step": 71530
    },
    {
      "epoch": 20.306556911722964,
      "grad_norm": 2.1702699661254883,
      "learning_rate": 7.970281010502412e-05,
      "loss": 0.019925877451896667,
      "step": 71540
    },
    {
      "epoch": 20.309395401646324,
      "grad_norm": 0.5071220397949219,
      "learning_rate": 7.969997161510076e-05,
      "loss": 0.01536392718553543,
      "step": 71550
    },
    {
      "epoch": 20.312233891569687,
      "grad_norm": 3.798987865447998,
      "learning_rate": 7.969713312517742e-05,
      "loss": 0.018707436323165894,
      "step": 71560
    },
    {
      "epoch": 20.315072381493046,
      "grad_norm": 16.85640525817871,
      "learning_rate": 7.969429463525405e-05,
      "loss": 0.018195712566375734,
      "step": 71570
    },
    {
      "epoch": 20.317910871416405,
      "grad_norm": 16.111982345581055,
      "learning_rate": 7.969145614533069e-05,
      "loss": 0.017200061678886415,
      "step": 71580
    },
    {
      "epoch": 20.320749361339768,
      "grad_norm": 13.741687774658203,
      "learning_rate": 7.968861765540733e-05,
      "loss": 0.016931161284446716,
      "step": 71590
    },
    {
      "epoch": 20.323587851263127,
      "grad_norm": 6.790251731872559,
      "learning_rate": 7.968577916548397e-05,
      "loss": 0.015344303846359254,
      "step": 71600
    },
    {
      "epoch": 20.32642634118649,
      "grad_norm": 1.0825552940368652,
      "learning_rate": 7.96829406755606e-05,
      "loss": 0.01314530372619629,
      "step": 71610
    },
    {
      "epoch": 20.32926483110985,
      "grad_norm": 1.0809060335159302,
      "learning_rate": 7.968010218563724e-05,
      "loss": 0.024292610585689545,
      "step": 71620
    },
    {
      "epoch": 20.33210332103321,
      "grad_norm": 6.901791572570801,
      "learning_rate": 7.96772636957139e-05,
      "loss": 0.013001339137554168,
      "step": 71630
    },
    {
      "epoch": 20.334941810956572,
      "grad_norm": 10.481952667236328,
      "learning_rate": 7.967442520579052e-05,
      "loss": 0.02448507398366928,
      "step": 71640
    },
    {
      "epoch": 20.33778030087993,
      "grad_norm": 10.30474853515625,
      "learning_rate": 7.967158671586716e-05,
      "loss": 0.05966736674308777,
      "step": 71650
    },
    {
      "epoch": 20.340618790803294,
      "grad_norm": 1.1744650602340698,
      "learning_rate": 7.96687482259438e-05,
      "loss": 0.03822956085205078,
      "step": 71660
    },
    {
      "epoch": 20.343457280726653,
      "grad_norm": 12.141825675964355,
      "learning_rate": 7.966590973602043e-05,
      "loss": 0.01278379261493683,
      "step": 71670
    },
    {
      "epoch": 20.346295770650013,
      "grad_norm": 0.8915175795555115,
      "learning_rate": 7.966307124609707e-05,
      "loss": 0.024162687361240387,
      "step": 71680
    },
    {
      "epoch": 20.349134260573376,
      "grad_norm": 4.455752849578857,
      "learning_rate": 7.966023275617373e-05,
      "loss": 0.016721682250499727,
      "step": 71690
    },
    {
      "epoch": 20.351972750496735,
      "grad_norm": 3.03383731842041,
      "learning_rate": 7.965739426625036e-05,
      "loss": 0.028349217772483826,
      "step": 71700
    },
    {
      "epoch": 20.354811240420098,
      "grad_norm": 3.06288480758667,
      "learning_rate": 7.9654555776327e-05,
      "loss": 0.01646275520324707,
      "step": 71710
    },
    {
      "epoch": 20.357649730343457,
      "grad_norm": 1.9586575031280518,
      "learning_rate": 7.965171728640364e-05,
      "loss": 0.005636728554964066,
      "step": 71720
    },
    {
      "epoch": 20.360488220266816,
      "grad_norm": 2.7683238983154297,
      "learning_rate": 7.964887879648028e-05,
      "loss": 0.01677453815937042,
      "step": 71730
    },
    {
      "epoch": 20.36332671019018,
      "grad_norm": 0.9296435713768005,
      "learning_rate": 7.964604030655691e-05,
      "loss": 0.0143295019865036,
      "step": 71740
    },
    {
      "epoch": 20.36616520011354,
      "grad_norm": 5.427598476409912,
      "learning_rate": 7.964320181663355e-05,
      "loss": 0.015647386014461518,
      "step": 71750
    },
    {
      "epoch": 20.3690036900369,
      "grad_norm": 2.572540760040283,
      "learning_rate": 7.964036332671019e-05,
      "loss": 0.016397115588188172,
      "step": 71760
    },
    {
      "epoch": 20.37184217996026,
      "grad_norm": 14.900172233581543,
      "learning_rate": 7.963752483678683e-05,
      "loss": 0.01482781171798706,
      "step": 71770
    },
    {
      "epoch": 20.37468066988362,
      "grad_norm": 6.146739959716797,
      "learning_rate": 7.963468634686348e-05,
      "loss": 0.022622056305408478,
      "step": 71780
    },
    {
      "epoch": 20.377519159806983,
      "grad_norm": 0.31290292739868164,
      "learning_rate": 7.963184785694012e-05,
      "loss": 0.002663133107125759,
      "step": 71790
    },
    {
      "epoch": 20.380357649730342,
      "grad_norm": 0.7098491191864014,
      "learning_rate": 7.962900936701674e-05,
      "loss": 0.007530221343040466,
      "step": 71800
    },
    {
      "epoch": 20.383196139653705,
      "grad_norm": 8.688029289245605,
      "learning_rate": 7.962617087709339e-05,
      "loss": 0.02052065432071686,
      "step": 71810
    },
    {
      "epoch": 20.386034629577065,
      "grad_norm": 5.2097649574279785,
      "learning_rate": 7.962333238717003e-05,
      "loss": 0.01468944400548935,
      "step": 71820
    },
    {
      "epoch": 20.388873119500424,
      "grad_norm": 2.7574174404144287,
      "learning_rate": 7.962049389724667e-05,
      "loss": 0.008304913341999055,
      "step": 71830
    },
    {
      "epoch": 20.391711609423787,
      "grad_norm": 3.9530792236328125,
      "learning_rate": 7.961765540732331e-05,
      "loss": 0.006035934388637543,
      "step": 71840
    },
    {
      "epoch": 20.394550099347146,
      "grad_norm": 1.443293571472168,
      "learning_rate": 7.961481691739995e-05,
      "loss": 0.015250900387763977,
      "step": 71850
    },
    {
      "epoch": 20.39738858927051,
      "grad_norm": 7.124764442443848,
      "learning_rate": 7.961197842747658e-05,
      "loss": 0.009171187132596969,
      "step": 71860
    },
    {
      "epoch": 20.40022707919387,
      "grad_norm": 0.6649703979492188,
      "learning_rate": 7.960913993755322e-05,
      "loss": 0.017971611022949217,
      "step": 71870
    },
    {
      "epoch": 20.40306556911723,
      "grad_norm": 2.4708728790283203,
      "learning_rate": 7.960630144762986e-05,
      "loss": 0.018482764065265656,
      "step": 71880
    },
    {
      "epoch": 20.40590405904059,
      "grad_norm": 6.341677188873291,
      "learning_rate": 7.96034629577065e-05,
      "loss": 0.01345614492893219,
      "step": 71890
    },
    {
      "epoch": 20.40874254896395,
      "grad_norm": 1.9383199214935303,
      "learning_rate": 7.960062446778314e-05,
      "loss": 0.029498499631881715,
      "step": 71900
    },
    {
      "epoch": 20.411581038887313,
      "grad_norm": 10.272661209106445,
      "learning_rate": 7.959778597785979e-05,
      "loss": 0.019552084803581237,
      "step": 71910
    },
    {
      "epoch": 20.414419528810672,
      "grad_norm": 0.3369351327419281,
      "learning_rate": 7.959494748793643e-05,
      "loss": 0.016626468300819396,
      "step": 71920
    },
    {
      "epoch": 20.417258018734035,
      "grad_norm": 11.512466430664062,
      "learning_rate": 7.959210899801305e-05,
      "loss": 0.02207489013671875,
      "step": 71930
    },
    {
      "epoch": 20.420096508657394,
      "grad_norm": 1.1375325918197632,
      "learning_rate": 7.95892705080897e-05,
      "loss": 0.019103460013866425,
      "step": 71940
    },
    {
      "epoch": 20.422934998580754,
      "grad_norm": 9.873573303222656,
      "learning_rate": 7.958643201816634e-05,
      "loss": 0.019082090258598326,
      "step": 71950
    },
    {
      "epoch": 20.425773488504117,
      "grad_norm": 1.6722835302352905,
      "learning_rate": 7.958359352824298e-05,
      "loss": 0.017862191796302794,
      "step": 71960
    },
    {
      "epoch": 20.428611978427476,
      "grad_norm": 0.7831377387046814,
      "learning_rate": 7.958075503831962e-05,
      "loss": 0.010959525406360627,
      "step": 71970
    },
    {
      "epoch": 20.43145046835084,
      "grad_norm": 0.2760637402534485,
      "learning_rate": 7.957791654839626e-05,
      "loss": 0.016851162910461424,
      "step": 71980
    },
    {
      "epoch": 20.434288958274198,
      "grad_norm": 2.167778253555298,
      "learning_rate": 7.957507805847289e-05,
      "loss": 0.021981926262378694,
      "step": 71990
    },
    {
      "epoch": 20.437127448197558,
      "grad_norm": 1.963661789894104,
      "learning_rate": 7.957223956854953e-05,
      "loss": 0.027234962582588194,
      "step": 72000
    },
    {
      "epoch": 20.437127448197558,
      "eval_accuracy": 0.9650918802060151,
      "eval_loss": 0.11085671931505203,
      "eval_runtime": 32.2812,
      "eval_samples_per_second": 487.188,
      "eval_steps_per_second": 7.621,
      "step": 72000
    },
    {
      "epoch": 20.43996593812092,
      "grad_norm": 2.861565351486206,
      "learning_rate": 7.956940107862617e-05,
      "loss": 0.014334313571453094,
      "step": 72010
    },
    {
      "epoch": 20.44280442804428,
      "grad_norm": 0.9935620427131653,
      "learning_rate": 7.956656258870281e-05,
      "loss": 0.023954546451568602,
      "step": 72020
    },
    {
      "epoch": 20.445642917967643,
      "grad_norm": 1.6452908515930176,
      "learning_rate": 7.956372409877946e-05,
      "loss": 0.011409955471754074,
      "step": 72030
    },
    {
      "epoch": 20.448481407891002,
      "grad_norm": 3.152693510055542,
      "learning_rate": 7.95608856088561e-05,
      "loss": 0.007183098793029785,
      "step": 72040
    },
    {
      "epoch": 20.45131989781436,
      "grad_norm": 1.5322866439819336,
      "learning_rate": 7.955804711893274e-05,
      "loss": 0.007374898344278335,
      "step": 72050
    },
    {
      "epoch": 20.454158387737724,
      "grad_norm": 3.7526168823242188,
      "learning_rate": 7.955520862900937e-05,
      "loss": 0.012207742035388946,
      "step": 72060
    },
    {
      "epoch": 20.456996877661084,
      "grad_norm": 12.91372013092041,
      "learning_rate": 7.955237013908601e-05,
      "loss": 0.012936016917228699,
      "step": 72070
    },
    {
      "epoch": 20.459835367584446,
      "grad_norm": 0.5344229340553284,
      "learning_rate": 7.954953164916265e-05,
      "loss": 0.018308672308921813,
      "step": 72080
    },
    {
      "epoch": 20.462673857507806,
      "grad_norm": 1.184501051902771,
      "learning_rate": 7.954669315923929e-05,
      "loss": 0.02302212417125702,
      "step": 72090
    },
    {
      "epoch": 20.465512347431165,
      "grad_norm": 2.184994697570801,
      "learning_rate": 7.954385466931593e-05,
      "loss": 0.034254243969917296,
      "step": 72100
    },
    {
      "epoch": 20.468350837354528,
      "grad_norm": 9.629460334777832,
      "learning_rate": 7.954101617939257e-05,
      "loss": 0.0381091833114624,
      "step": 72110
    },
    {
      "epoch": 20.471189327277887,
      "grad_norm": 0.7039865255355835,
      "learning_rate": 7.95381776894692e-05,
      "loss": 0.004887149482965469,
      "step": 72120
    },
    {
      "epoch": 20.47402781720125,
      "grad_norm": 8.712130546569824,
      "learning_rate": 7.953533919954584e-05,
      "loss": 0.007143816351890564,
      "step": 72130
    },
    {
      "epoch": 20.47686630712461,
      "grad_norm": 3.7989861965179443,
      "learning_rate": 7.953250070962248e-05,
      "loss": 0.029581409692764283,
      "step": 72140
    },
    {
      "epoch": 20.47970479704797,
      "grad_norm": 2.2428572177886963,
      "learning_rate": 7.952966221969912e-05,
      "loss": 0.019566893577575684,
      "step": 72150
    },
    {
      "epoch": 20.48254328697133,
      "grad_norm": 8.025857925415039,
      "learning_rate": 7.952682372977577e-05,
      "loss": 0.023201459646224977,
      "step": 72160
    },
    {
      "epoch": 20.48538177689469,
      "grad_norm": 0.7326478958129883,
      "learning_rate": 7.952398523985241e-05,
      "loss": 0.01832791864871979,
      "step": 72170
    },
    {
      "epoch": 20.488220266818054,
      "grad_norm": 7.723518371582031,
      "learning_rate": 7.952114674992905e-05,
      "loss": 0.012235413491725921,
      "step": 72180
    },
    {
      "epoch": 20.491058756741413,
      "grad_norm": 0.8384155035018921,
      "learning_rate": 7.951830826000568e-05,
      "loss": 0.01628859043121338,
      "step": 72190
    },
    {
      "epoch": 20.493897246664773,
      "grad_norm": 0.47885945439338684,
      "learning_rate": 7.951546977008232e-05,
      "loss": 0.011692804098129273,
      "step": 72200
    },
    {
      "epoch": 20.496735736588136,
      "grad_norm": 1.621778130531311,
      "learning_rate": 7.951263128015896e-05,
      "loss": 0.01482357382774353,
      "step": 72210
    },
    {
      "epoch": 20.499574226511495,
      "grad_norm": 1.566728949546814,
      "learning_rate": 7.950979279023559e-05,
      "loss": 0.019316935539245607,
      "step": 72220
    },
    {
      "epoch": 20.502412716434858,
      "grad_norm": 8.736160278320312,
      "learning_rate": 7.950695430031224e-05,
      "loss": 0.03253134787082672,
      "step": 72230
    },
    {
      "epoch": 20.505251206358217,
      "grad_norm": 1.2695624828338623,
      "learning_rate": 7.950411581038888e-05,
      "loss": 0.012108281999826432,
      "step": 72240
    },
    {
      "epoch": 20.50808969628158,
      "grad_norm": 0.06632253527641296,
      "learning_rate": 7.950127732046551e-05,
      "loss": 0.014710676670074464,
      "step": 72250
    },
    {
      "epoch": 20.51092818620494,
      "grad_norm": 7.1121625900268555,
      "learning_rate": 7.949843883054215e-05,
      "loss": 0.014168836176395416,
      "step": 72260
    },
    {
      "epoch": 20.5137666761283,
      "grad_norm": 4.94971227645874,
      "learning_rate": 7.94956003406188e-05,
      "loss": 0.011790807545185088,
      "step": 72270
    },
    {
      "epoch": 20.51660516605166,
      "grad_norm": 19.84990119934082,
      "learning_rate": 7.949276185069544e-05,
      "loss": 0.01199587732553482,
      "step": 72280
    },
    {
      "epoch": 20.51944365597502,
      "grad_norm": 1.730820655822754,
      "learning_rate": 7.948992336077208e-05,
      "loss": 0.01737298220396042,
      "step": 72290
    },
    {
      "epoch": 20.522282145898384,
      "grad_norm": 0.3394525647163391,
      "learning_rate": 7.948708487084872e-05,
      "loss": 0.012470385432243348,
      "step": 72300
    },
    {
      "epoch": 20.525120635821743,
      "grad_norm": 2.8416941165924072,
      "learning_rate": 7.948424638092536e-05,
      "loss": 0.010157190263271332,
      "step": 72310
    },
    {
      "epoch": 20.527959125745102,
      "grad_norm": 5.284004211425781,
      "learning_rate": 7.948140789100199e-05,
      "loss": 0.011029845476150513,
      "step": 72320
    },
    {
      "epoch": 20.530797615668465,
      "grad_norm": 5.133011817932129,
      "learning_rate": 7.947856940107863e-05,
      "loss": 0.025474944710731508,
      "step": 72330
    },
    {
      "epoch": 20.533636105591825,
      "grad_norm": 0.8183083534240723,
      "learning_rate": 7.947573091115527e-05,
      "loss": 0.02821999490261078,
      "step": 72340
    },
    {
      "epoch": 20.536474595515188,
      "grad_norm": 4.689294338226318,
      "learning_rate": 7.94728924212319e-05,
      "loss": 0.033982190489768985,
      "step": 72350
    },
    {
      "epoch": 20.539313085438547,
      "grad_norm": 1.7934982776641846,
      "learning_rate": 7.947005393130855e-05,
      "loss": 0.01371917873620987,
      "step": 72360
    },
    {
      "epoch": 20.542151575361906,
      "grad_norm": 1.614597201347351,
      "learning_rate": 7.94672154413852e-05,
      "loss": 0.009933491051197053,
      "step": 72370
    },
    {
      "epoch": 20.54499006528527,
      "grad_norm": 9.98063850402832,
      "learning_rate": 7.946437695146182e-05,
      "loss": 0.03616088330745697,
      "step": 72380
    },
    {
      "epoch": 20.54782855520863,
      "grad_norm": 10.279441833496094,
      "learning_rate": 7.946153846153846e-05,
      "loss": 0.015198688209056854,
      "step": 72390
    },
    {
      "epoch": 20.55066704513199,
      "grad_norm": 7.6630730628967285,
      "learning_rate": 7.94586999716151e-05,
      "loss": 0.01767825335264206,
      "step": 72400
    },
    {
      "epoch": 20.55350553505535,
      "grad_norm": 11.035643577575684,
      "learning_rate": 7.945586148169175e-05,
      "loss": 0.011328937858343125,
      "step": 72410
    },
    {
      "epoch": 20.55634402497871,
      "grad_norm": 3.737757682800293,
      "learning_rate": 7.945302299176837e-05,
      "loss": 0.011093376576900483,
      "step": 72420
    },
    {
      "epoch": 20.559182514902073,
      "grad_norm": 5.453857898712158,
      "learning_rate": 7.945018450184503e-05,
      "loss": 0.014840850234031677,
      "step": 72430
    },
    {
      "epoch": 20.562021004825432,
      "grad_norm": 2.1554341316223145,
      "learning_rate": 7.944734601192167e-05,
      "loss": 0.009334130585193634,
      "step": 72440
    },
    {
      "epoch": 20.564859494748795,
      "grad_norm": 4.127039432525635,
      "learning_rate": 7.94445075219983e-05,
      "loss": 0.014179280400276184,
      "step": 72450
    },
    {
      "epoch": 20.567697984672154,
      "grad_norm": 1.2953083515167236,
      "learning_rate": 7.944166903207494e-05,
      "loss": 0.012179459631443023,
      "step": 72460
    },
    {
      "epoch": 20.570536474595514,
      "grad_norm": 1.1966592073440552,
      "learning_rate": 7.943883054215158e-05,
      "loss": 0.010234113037586211,
      "step": 72470
    },
    {
      "epoch": 20.573374964518877,
      "grad_norm": 1.6623189449310303,
      "learning_rate": 7.943599205222821e-05,
      "loss": 0.007945075631141663,
      "step": 72480
    },
    {
      "epoch": 20.576213454442236,
      "grad_norm": 1.375534176826477,
      "learning_rate": 7.943315356230486e-05,
      "loss": 0.0140870600938797,
      "step": 72490
    },
    {
      "epoch": 20.5790519443656,
      "grad_norm": 3.179778575897217,
      "learning_rate": 7.94303150723815e-05,
      "loss": 0.026906004548072814,
      "step": 72500
    },
    {
      "epoch": 20.5790519443656,
      "eval_accuracy": 0.9663635785591658,
      "eval_loss": 0.11196941882371902,
      "eval_runtime": 31.55,
      "eval_samples_per_second": 498.479,
      "eval_steps_per_second": 7.797,
      "step": 72500
    },
    {
      "epoch": 20.581890434288958,
      "grad_norm": 0.11507121473550797,
      "learning_rate": 7.942747658245813e-05,
      "loss": 0.011539269983768464,
      "step": 72510
    },
    {
      "epoch": 20.584728924212317,
      "grad_norm": 8.79147720336914,
      "learning_rate": 7.942463809253477e-05,
      "loss": 0.01557164490222931,
      "step": 72520
    },
    {
      "epoch": 20.58756741413568,
      "grad_norm": 0.8278812766075134,
      "learning_rate": 7.942179960261142e-05,
      "loss": 0.012269531190395356,
      "step": 72530
    },
    {
      "epoch": 20.59040590405904,
      "grad_norm": 20.315732955932617,
      "learning_rate": 7.941896111268806e-05,
      "loss": 0.021502546966075897,
      "step": 72540
    },
    {
      "epoch": 20.593244393982403,
      "grad_norm": 1.9884330034255981,
      "learning_rate": 7.941612262276468e-05,
      "loss": 0.017670203745365144,
      "step": 72550
    },
    {
      "epoch": 20.596082883905762,
      "grad_norm": 1.3758779764175415,
      "learning_rate": 7.941328413284134e-05,
      "loss": 0.018218497931957244,
      "step": 72560
    },
    {
      "epoch": 20.59892137382912,
      "grad_norm": 4.684302806854248,
      "learning_rate": 7.941044564291798e-05,
      "loss": 0.006948423385620117,
      "step": 72570
    },
    {
      "epoch": 20.601759863752484,
      "grad_norm": 4.039017200469971,
      "learning_rate": 7.940760715299461e-05,
      "loss": 0.015477192401885987,
      "step": 72580
    },
    {
      "epoch": 20.604598353675843,
      "grad_norm": 4.502068042755127,
      "learning_rate": 7.940476866307125e-05,
      "loss": 0.01728564202785492,
      "step": 72590
    },
    {
      "epoch": 20.607436843599206,
      "grad_norm": 10.67024040222168,
      "learning_rate": 7.940193017314789e-05,
      "loss": 0.014483582973480225,
      "step": 72600
    },
    {
      "epoch": 20.610275333522566,
      "grad_norm": 21.431047439575195,
      "learning_rate": 7.939909168322452e-05,
      "loss": 0.023728176951408386,
      "step": 72610
    },
    {
      "epoch": 20.61311382344593,
      "grad_norm": 2.125991106033325,
      "learning_rate": 7.939625319330116e-05,
      "loss": 0.009162011742591857,
      "step": 72620
    },
    {
      "epoch": 20.615952313369288,
      "grad_norm": 3.0115318298339844,
      "learning_rate": 7.939341470337782e-05,
      "loss": 0.010436663031578064,
      "step": 72630
    },
    {
      "epoch": 20.618790803292647,
      "grad_norm": 1.3310275077819824,
      "learning_rate": 7.939057621345444e-05,
      "loss": 0.0088840551674366,
      "step": 72640
    },
    {
      "epoch": 20.62162929321601,
      "grad_norm": 14.502790451049805,
      "learning_rate": 7.938773772353108e-05,
      "loss": 0.01096387803554535,
      "step": 72650
    },
    {
      "epoch": 20.62446778313937,
      "grad_norm": 2.0159082412719727,
      "learning_rate": 7.938489923360773e-05,
      "loss": 0.020039521157741547,
      "step": 72660
    },
    {
      "epoch": 20.627306273062732,
      "grad_norm": 11.093973159790039,
      "learning_rate": 7.938206074368437e-05,
      "loss": 0.01522703468799591,
      "step": 72670
    },
    {
      "epoch": 20.63014476298609,
      "grad_norm": 17.858243942260742,
      "learning_rate": 7.9379222253761e-05,
      "loss": 0.020603251457214356,
      "step": 72680
    },
    {
      "epoch": 20.63298325290945,
      "grad_norm": 6.4766621589660645,
      "learning_rate": 7.937638376383765e-05,
      "loss": 0.025670135021209718,
      "step": 72690
    },
    {
      "epoch": 20.635821742832814,
      "grad_norm": 9.239554405212402,
      "learning_rate": 7.937354527391428e-05,
      "loss": 0.017160232365131377,
      "step": 72700
    },
    {
      "epoch": 20.638660232756173,
      "grad_norm": 11.623858451843262,
      "learning_rate": 7.937070678399092e-05,
      "loss": 0.023585006594657898,
      "step": 72710
    },
    {
      "epoch": 20.641498722679536,
      "grad_norm": 1.536014437675476,
      "learning_rate": 7.936786829406756e-05,
      "loss": 0.014897760748863221,
      "step": 72720
    },
    {
      "epoch": 20.644337212602895,
      "grad_norm": 3.0355446338653564,
      "learning_rate": 7.93650298041442e-05,
      "loss": 0.016392450034618377,
      "step": 72730
    },
    {
      "epoch": 20.647175702526255,
      "grad_norm": 2.6414432525634766,
      "learning_rate": 7.936219131422083e-05,
      "loss": 0.008637484163045883,
      "step": 72740
    },
    {
      "epoch": 20.650014192449618,
      "grad_norm": 6.1766862869262695,
      "learning_rate": 7.935935282429747e-05,
      "loss": 0.02251293361186981,
      "step": 72750
    },
    {
      "epoch": 20.652852682372977,
      "grad_norm": 8.996399879455566,
      "learning_rate": 7.935651433437413e-05,
      "loss": 0.01978897154331207,
      "step": 72760
    },
    {
      "epoch": 20.65569117229634,
      "grad_norm": 8.1782865524292,
      "learning_rate": 7.935367584445075e-05,
      "loss": 0.01567893773317337,
      "step": 72770
    },
    {
      "epoch": 20.6585296622197,
      "grad_norm": 0.18052145838737488,
      "learning_rate": 7.93508373545274e-05,
      "loss": 0.015609405934810638,
      "step": 72780
    },
    {
      "epoch": 20.66136815214306,
      "grad_norm": 0.7489326596260071,
      "learning_rate": 7.934799886460404e-05,
      "loss": 0.016970519721508027,
      "step": 72790
    },
    {
      "epoch": 20.66420664206642,
      "grad_norm": 2.3605284690856934,
      "learning_rate": 7.934516037468066e-05,
      "loss": 0.026287132501602174,
      "step": 72800
    },
    {
      "epoch": 20.66704513198978,
      "grad_norm": 1.7063428163528442,
      "learning_rate": 7.93423218847573e-05,
      "loss": 0.018955835700035097,
      "step": 72810
    },
    {
      "epoch": 20.669883621913144,
      "grad_norm": 1.76027512550354,
      "learning_rate": 7.933948339483395e-05,
      "loss": 0.013470149040222168,
      "step": 72820
    },
    {
      "epoch": 20.672722111836503,
      "grad_norm": 0.7529674172401428,
      "learning_rate": 7.933664490491059e-05,
      "loss": 0.03139452338218689,
      "step": 72830
    },
    {
      "epoch": 20.675560601759862,
      "grad_norm": 0.45633724331855774,
      "learning_rate": 7.933380641498723e-05,
      "loss": 0.012667980790138245,
      "step": 72840
    },
    {
      "epoch": 20.678399091683225,
      "grad_norm": 0.46502992510795593,
      "learning_rate": 7.933096792506387e-05,
      "loss": 0.038240844011306764,
      "step": 72850
    },
    {
      "epoch": 20.681237581606585,
      "grad_norm": 2.5293679237365723,
      "learning_rate": 7.932812943514051e-05,
      "loss": 0.015347310900688171,
      "step": 72860
    },
    {
      "epoch": 20.684076071529947,
      "grad_norm": 0.2258562445640564,
      "learning_rate": 7.932529094521714e-05,
      "loss": 0.022741031646728516,
      "step": 72870
    },
    {
      "epoch": 20.686914561453307,
      "grad_norm": 0.45324718952178955,
      "learning_rate": 7.932245245529378e-05,
      "loss": 0.013174809515476227,
      "step": 72880
    },
    {
      "epoch": 20.689753051376666,
      "grad_norm": 4.431980133056641,
      "learning_rate": 7.931961396537044e-05,
      "loss": 0.019937348365783692,
      "step": 72890
    },
    {
      "epoch": 20.69259154130003,
      "grad_norm": 6.658142566680908,
      "learning_rate": 7.931677547544707e-05,
      "loss": 0.02352796643972397,
      "step": 72900
    },
    {
      "epoch": 20.69543003122339,
      "grad_norm": 1.5227582454681396,
      "learning_rate": 7.93139369855237e-05,
      "loss": 0.011709966510534287,
      "step": 72910
    },
    {
      "epoch": 20.69826852114675,
      "grad_norm": 2.4257354736328125,
      "learning_rate": 7.931109849560035e-05,
      "loss": 0.014188449084758758,
      "step": 72920
    },
    {
      "epoch": 20.70110701107011,
      "grad_norm": 0.863277792930603,
      "learning_rate": 7.930826000567698e-05,
      "loss": 0.014848138391971587,
      "step": 72930
    },
    {
      "epoch": 20.70394550099347,
      "grad_norm": 6.3147125244140625,
      "learning_rate": 7.930542151575362e-05,
      "loss": 0.015090684592723846,
      "step": 72940
    },
    {
      "epoch": 20.706783990916833,
      "grad_norm": 1.402535080909729,
      "learning_rate": 7.930258302583026e-05,
      "loss": 0.009539633244276046,
      "step": 72950
    },
    {
      "epoch": 20.709622480840192,
      "grad_norm": 1.1036694049835205,
      "learning_rate": 7.92997445359069e-05,
      "loss": 0.02125362902879715,
      "step": 72960
    },
    {
      "epoch": 20.712460970763555,
      "grad_norm": 10.724868774414062,
      "learning_rate": 7.929690604598354e-05,
      "loss": 0.014958055317401886,
      "step": 72970
    },
    {
      "epoch": 20.715299460686914,
      "grad_norm": 0.3165739178657532,
      "learning_rate": 7.929406755606018e-05,
      "loss": 0.01848827600479126,
      "step": 72980
    },
    {
      "epoch": 20.718137950610274,
      "grad_norm": 7.2359185218811035,
      "learning_rate": 7.929122906613682e-05,
      "loss": 0.01082410141825676,
      "step": 72990
    },
    {
      "epoch": 20.720976440533637,
      "grad_norm": 3.24294114112854,
      "learning_rate": 7.928839057621345e-05,
      "loss": 0.021345661580562593,
      "step": 73000
    },
    {
      "epoch": 20.720976440533637,
      "eval_accuracy": 0.9629299930056591,
      "eval_loss": 0.12116753309965134,
      "eval_runtime": 31.339,
      "eval_samples_per_second": 501.834,
      "eval_steps_per_second": 7.85,
      "step": 73000
    },
    {
      "epoch": 20.723814930456996,
      "grad_norm": 0.4712117910385132,
      "learning_rate": 7.928555208629009e-05,
      "loss": 0.015656279027462007,
      "step": 73010
    },
    {
      "epoch": 20.72665342038036,
      "grad_norm": 1.1789573431015015,
      "learning_rate": 7.928271359636675e-05,
      "loss": 0.013817785680294037,
      "step": 73020
    },
    {
      "epoch": 20.729491910303718,
      "grad_norm": 1.4754494428634644,
      "learning_rate": 7.927987510644338e-05,
      "loss": 0.033756738901138304,
      "step": 73030
    },
    {
      "epoch": 20.732330400227077,
      "grad_norm": 5.361126899719238,
      "learning_rate": 7.927703661652002e-05,
      "loss": 0.032455235719680786,
      "step": 73040
    },
    {
      "epoch": 20.73516889015044,
      "grad_norm": 10.095773696899414,
      "learning_rate": 7.927419812659666e-05,
      "loss": 0.03645889163017273,
      "step": 73050
    },
    {
      "epoch": 20.7380073800738,
      "grad_norm": 0.8621581196784973,
      "learning_rate": 7.927135963667329e-05,
      "loss": 0.016415764391422272,
      "step": 73060
    },
    {
      "epoch": 20.740845869997163,
      "grad_norm": 6.6164374351501465,
      "learning_rate": 7.926852114674993e-05,
      "loss": 0.014220431447029114,
      "step": 73070
    },
    {
      "epoch": 20.743684359920522,
      "grad_norm": 8.197409629821777,
      "learning_rate": 7.926568265682657e-05,
      "loss": 0.00881960690021515,
      "step": 73080
    },
    {
      "epoch": 20.746522849843885,
      "grad_norm": 13.695781707763672,
      "learning_rate": 7.926284416690321e-05,
      "loss": 0.019520682096481324,
      "step": 73090
    },
    {
      "epoch": 20.749361339767244,
      "grad_norm": 13.915450096130371,
      "learning_rate": 7.926000567697985e-05,
      "loss": 0.012441352009773254,
      "step": 73100
    },
    {
      "epoch": 20.752199829690603,
      "grad_norm": 12.172597885131836,
      "learning_rate": 7.92571671870565e-05,
      "loss": 0.027883368730545043,
      "step": 73110
    },
    {
      "epoch": 20.755038319613966,
      "grad_norm": 2.490008592605591,
      "learning_rate": 7.925432869713313e-05,
      "loss": 0.00721658319234848,
      "step": 73120
    },
    {
      "epoch": 20.757876809537326,
      "grad_norm": 0.8751887679100037,
      "learning_rate": 7.925149020720976e-05,
      "loss": 0.010434217751026154,
      "step": 73130
    },
    {
      "epoch": 20.76071529946069,
      "grad_norm": 6.381956577301025,
      "learning_rate": 7.92486517172864e-05,
      "loss": 0.008940516412258149,
      "step": 73140
    },
    {
      "epoch": 20.763553789384048,
      "grad_norm": 2.7959420680999756,
      "learning_rate": 7.924581322736305e-05,
      "loss": 0.026629617810249327,
      "step": 73150
    },
    {
      "epoch": 20.766392279307407,
      "grad_norm": 9.569304466247559,
      "learning_rate": 7.924297473743969e-05,
      "loss": 0.019592656195163725,
      "step": 73160
    },
    {
      "epoch": 20.76923076923077,
      "grad_norm": 1.1693205833435059,
      "learning_rate": 7.924013624751633e-05,
      "loss": 0.019284830987453462,
      "step": 73170
    },
    {
      "epoch": 20.77206925915413,
      "grad_norm": 1.6337071657180786,
      "learning_rate": 7.923729775759297e-05,
      "loss": 0.027567052841186525,
      "step": 73180
    },
    {
      "epoch": 20.774907749077492,
      "grad_norm": 8.254040718078613,
      "learning_rate": 7.92344592676696e-05,
      "loss": 0.007244700193405151,
      "step": 73190
    },
    {
      "epoch": 20.77774623900085,
      "grad_norm": 9.274359703063965,
      "learning_rate": 7.923162077774624e-05,
      "loss": 0.023554137349128722,
      "step": 73200
    },
    {
      "epoch": 20.78058472892421,
      "grad_norm": 0.4924163520336151,
      "learning_rate": 7.922878228782288e-05,
      "loss": 0.015244534611701966,
      "step": 73210
    },
    {
      "epoch": 20.783423218847574,
      "grad_norm": 3.1781094074249268,
      "learning_rate": 7.922594379789952e-05,
      "loss": 0.021272359788417815,
      "step": 73220
    },
    {
      "epoch": 20.786261708770933,
      "grad_norm": 10.52193832397461,
      "learning_rate": 7.922310530797616e-05,
      "loss": 0.024418361485004425,
      "step": 73230
    },
    {
      "epoch": 20.789100198694296,
      "grad_norm": 2.9453516006469727,
      "learning_rate": 7.92202668180528e-05,
      "loss": 0.030804479122161867,
      "step": 73240
    },
    {
      "epoch": 20.791938688617655,
      "grad_norm": 0.8992425203323364,
      "learning_rate": 7.921742832812945e-05,
      "loss": 0.019534553587436675,
      "step": 73250
    },
    {
      "epoch": 20.794777178541015,
      "grad_norm": 2.0761821269989014,
      "learning_rate": 7.921458983820607e-05,
      "loss": 0.01790963262319565,
      "step": 73260
    },
    {
      "epoch": 20.797615668464378,
      "grad_norm": 5.022449970245361,
      "learning_rate": 7.921175134828271e-05,
      "loss": 0.00907110795378685,
      "step": 73270
    },
    {
      "epoch": 20.800454158387737,
      "grad_norm": 2.6330373287200928,
      "learning_rate": 7.920891285835936e-05,
      "loss": 0.016720624268054964,
      "step": 73280
    },
    {
      "epoch": 20.8032926483111,
      "grad_norm": 8.316458702087402,
      "learning_rate": 7.9206074368436e-05,
      "loss": 0.01833339035511017,
      "step": 73290
    },
    {
      "epoch": 20.80613113823446,
      "grad_norm": 10.399785995483398,
      "learning_rate": 7.920323587851264e-05,
      "loss": 0.01224859207868576,
      "step": 73300
    },
    {
      "epoch": 20.80896962815782,
      "grad_norm": 3.4011497497558594,
      "learning_rate": 7.920039738858928e-05,
      "loss": 0.01687619388103485,
      "step": 73310
    },
    {
      "epoch": 20.81180811808118,
      "grad_norm": 6.77141809463501,
      "learning_rate": 7.919755889866591e-05,
      "loss": 0.032311511039733884,
      "step": 73320
    },
    {
      "epoch": 20.81464660800454,
      "grad_norm": 4.743729114532471,
      "learning_rate": 7.919472040874255e-05,
      "loss": 0.017578643560409547,
      "step": 73330
    },
    {
      "epoch": 20.817485097927904,
      "grad_norm": 12.768263816833496,
      "learning_rate": 7.919188191881919e-05,
      "loss": 0.022203050553798676,
      "step": 73340
    },
    {
      "epoch": 20.820323587851263,
      "grad_norm": 2.953514814376831,
      "learning_rate": 7.918904342889583e-05,
      "loss": 0.035276031494140624,
      "step": 73350
    },
    {
      "epoch": 20.823162077774622,
      "grad_norm": 1.0464810132980347,
      "learning_rate": 7.918620493897247e-05,
      "loss": 0.019856755435466767,
      "step": 73360
    },
    {
      "epoch": 20.826000567697985,
      "grad_norm": 0.7617159485816956,
      "learning_rate": 7.918336644904911e-05,
      "loss": 0.014259248971939087,
      "step": 73370
    },
    {
      "epoch": 20.828839057621344,
      "grad_norm": 5.238072395324707,
      "learning_rate": 7.918052795912576e-05,
      "loss": 0.028229644894599913,
      "step": 73380
    },
    {
      "epoch": 20.831677547544707,
      "grad_norm": 1.6024333238601685,
      "learning_rate": 7.917768946920238e-05,
      "loss": 0.01618278622627258,
      "step": 73390
    },
    {
      "epoch": 20.834516037468067,
      "grad_norm": 2.3969497680664062,
      "learning_rate": 7.917485097927903e-05,
      "loss": 0.02306237518787384,
      "step": 73400
    },
    {
      "epoch": 20.837354527391426,
      "grad_norm": 1.034471869468689,
      "learning_rate": 7.917201248935567e-05,
      "loss": 0.011136702448129653,
      "step": 73410
    },
    {
      "epoch": 20.84019301731479,
      "grad_norm": 4.866801738739014,
      "learning_rate": 7.916917399943231e-05,
      "loss": 0.01119556576013565,
      "step": 73420
    },
    {
      "epoch": 20.84303150723815,
      "grad_norm": 15.34523868560791,
      "learning_rate": 7.916633550950895e-05,
      "loss": 0.019770808517932892,
      "step": 73430
    },
    {
      "epoch": 20.84586999716151,
      "grad_norm": 6.16182279586792,
      "learning_rate": 7.916349701958559e-05,
      "loss": 0.03611786663532257,
      "step": 73440
    },
    {
      "epoch": 20.84870848708487,
      "grad_norm": 9.896342277526855,
      "learning_rate": 7.916065852966222e-05,
      "loss": 0.026860374212265014,
      "step": 73450
    },
    {
      "epoch": 20.851546977008233,
      "grad_norm": 3.893913984298706,
      "learning_rate": 7.915782003973886e-05,
      "loss": 0.016549192368984222,
      "step": 73460
    },
    {
      "epoch": 20.854385466931593,
      "grad_norm": 0.46682605147361755,
      "learning_rate": 7.91549815498155e-05,
      "loss": 0.022315067052841187,
      "step": 73470
    },
    {
      "epoch": 20.857223956854952,
      "grad_norm": 2.7312629222869873,
      "learning_rate": 7.915214305989214e-05,
      "loss": 0.01133921667933464,
      "step": 73480
    },
    {
      "epoch": 20.860062446778315,
      "grad_norm": 7.671429634094238,
      "learning_rate": 7.914930456996878e-05,
      "loss": 0.010831889510154725,
      "step": 73490
    },
    {
      "epoch": 20.862900936701674,
      "grad_norm": 1.0821009874343872,
      "learning_rate": 7.914646608004543e-05,
      "loss": 0.018389439582824706,
      "step": 73500
    },
    {
      "epoch": 20.862900936701674,
      "eval_accuracy": 0.9700515037833026,
      "eval_loss": 0.09482727199792862,
      "eval_runtime": 31.7055,
      "eval_samples_per_second": 496.033,
      "eval_steps_per_second": 7.759,
      "step": 73500
    },
    {
      "epoch": 20.865739426625037,
      "grad_norm": 9.925872802734375,
      "learning_rate": 7.914362759012207e-05,
      "loss": 0.027100187540054322,
      "step": 73510
    },
    {
      "epoch": 20.868577916548396,
      "grad_norm": 2.02506160736084,
      "learning_rate": 7.91407891001987e-05,
      "loss": 0.008456361293792725,
      "step": 73520
    },
    {
      "epoch": 20.871416406471756,
      "grad_norm": 2.9364569187164307,
      "learning_rate": 7.913795061027534e-05,
      "loss": 0.02375752031803131,
      "step": 73530
    },
    {
      "epoch": 20.87425489639512,
      "grad_norm": 10.55201244354248,
      "learning_rate": 7.913511212035198e-05,
      "loss": 0.01722322702407837,
      "step": 73540
    },
    {
      "epoch": 20.877093386318478,
      "grad_norm": 2.3054146766662598,
      "learning_rate": 7.91322736304286e-05,
      "loss": 0.0069063156843185425,
      "step": 73550
    },
    {
      "epoch": 20.87993187624184,
      "grad_norm": 7.215479850769043,
      "learning_rate": 7.912943514050526e-05,
      "loss": 0.012965303659439088,
      "step": 73560
    },
    {
      "epoch": 20.8827703661652,
      "grad_norm": 4.540820121765137,
      "learning_rate": 7.91265966505819e-05,
      "loss": 0.012083213031291961,
      "step": 73570
    },
    {
      "epoch": 20.88560885608856,
      "grad_norm": 0.2832660377025604,
      "learning_rate": 7.912375816065853e-05,
      "loss": 0.01356612890958786,
      "step": 73580
    },
    {
      "epoch": 20.888447346011922,
      "grad_norm": 2.2344326972961426,
      "learning_rate": 7.912091967073517e-05,
      "loss": 0.015434113144874573,
      "step": 73590
    },
    {
      "epoch": 20.89128583593528,
      "grad_norm": 9.897329330444336,
      "learning_rate": 7.911808118081181e-05,
      "loss": 0.014597289264202118,
      "step": 73600
    },
    {
      "epoch": 20.894124325858645,
      "grad_norm": 13.89268684387207,
      "learning_rate": 7.911524269088845e-05,
      "loss": 0.04495354294776917,
      "step": 73610
    },
    {
      "epoch": 20.896962815782004,
      "grad_norm": 1.9589704275131226,
      "learning_rate": 7.91124042009651e-05,
      "loss": 0.013670009374618531,
      "step": 73620
    },
    {
      "epoch": 20.899801305705363,
      "grad_norm": 0.9423201680183411,
      "learning_rate": 7.910956571104174e-05,
      "loss": 0.01837254613637924,
      "step": 73630
    },
    {
      "epoch": 20.902639795628726,
      "grad_norm": 12.29433822631836,
      "learning_rate": 7.910672722111836e-05,
      "loss": 0.02250507175922394,
      "step": 73640
    },
    {
      "epoch": 20.905478285552086,
      "grad_norm": 5.166441440582275,
      "learning_rate": 7.9103888731195e-05,
      "loss": 0.012467760592699051,
      "step": 73650
    },
    {
      "epoch": 20.90831677547545,
      "grad_norm": 4.902466773986816,
      "learning_rate": 7.910105024127165e-05,
      "loss": 0.01773456037044525,
      "step": 73660
    },
    {
      "epoch": 20.911155265398808,
      "grad_norm": 5.229700088500977,
      "learning_rate": 7.909821175134829e-05,
      "loss": 0.021667641401290894,
      "step": 73670
    },
    {
      "epoch": 20.913993755322167,
      "grad_norm": 0.29768115282058716,
      "learning_rate": 7.909565711041727e-05,
      "loss": 0.01981286108493805,
      "step": 73680
    },
    {
      "epoch": 20.91683224524553,
      "grad_norm": 8.50903034210205,
      "learning_rate": 7.90928186204939e-05,
      "loss": 0.03741942048072815,
      "step": 73690
    },
    {
      "epoch": 20.91967073516889,
      "grad_norm": 7.114592552185059,
      "learning_rate": 7.908998013057054e-05,
      "loss": 0.02943575382232666,
      "step": 73700
    },
    {
      "epoch": 20.922509225092252,
      "grad_norm": 0.8609477877616882,
      "learning_rate": 7.908714164064718e-05,
      "loss": 0.012257280200719834,
      "step": 73710
    },
    {
      "epoch": 20.92534771501561,
      "grad_norm": 4.012287616729736,
      "learning_rate": 7.908430315072382e-05,
      "loss": 0.012139906734228134,
      "step": 73720
    },
    {
      "epoch": 20.92818620493897,
      "grad_norm": 20.529264450073242,
      "learning_rate": 7.908146466080045e-05,
      "loss": 0.024718427658081056,
      "step": 73730
    },
    {
      "epoch": 20.931024694862334,
      "grad_norm": 4.075937271118164,
      "learning_rate": 7.90786261708771e-05,
      "loss": 0.02088976800441742,
      "step": 73740
    },
    {
      "epoch": 20.933863184785693,
      "grad_norm": 1.309096336364746,
      "learning_rate": 7.907578768095374e-05,
      "loss": 0.015802517533302307,
      "step": 73750
    },
    {
      "epoch": 20.936701674709056,
      "grad_norm": 7.873183727264404,
      "learning_rate": 7.907294919103037e-05,
      "loss": 0.015766820311546324,
      "step": 73760
    },
    {
      "epoch": 20.939540164632415,
      "grad_norm": 6.351379871368408,
      "learning_rate": 7.907011070110701e-05,
      "loss": 0.027875053882598876,
      "step": 73770
    },
    {
      "epoch": 20.942378654555775,
      "grad_norm": 2.4245989322662354,
      "learning_rate": 7.906727221118365e-05,
      "loss": 0.013500745594501495,
      "step": 73780
    },
    {
      "epoch": 20.945217144479138,
      "grad_norm": 5.598823547363281,
      "learning_rate": 7.906443372126028e-05,
      "loss": 0.018315429985523223,
      "step": 73790
    },
    {
      "epoch": 20.948055634402497,
      "grad_norm": 9.185812950134277,
      "learning_rate": 7.906159523133694e-05,
      "loss": 0.03035212755203247,
      "step": 73800
    },
    {
      "epoch": 20.95089412432586,
      "grad_norm": 7.553031921386719,
      "learning_rate": 7.905875674141358e-05,
      "loss": 0.018949438631534577,
      "step": 73810
    },
    {
      "epoch": 20.95373261424922,
      "grad_norm": 9.329947471618652,
      "learning_rate": 7.905591825149021e-05,
      "loss": 0.025003620982170106,
      "step": 73820
    },
    {
      "epoch": 20.956571104172582,
      "grad_norm": 6.53037691116333,
      "learning_rate": 7.905307976156685e-05,
      "loss": 0.022636470198631287,
      "step": 73830
    },
    {
      "epoch": 20.95940959409594,
      "grad_norm": 5.085198879241943,
      "learning_rate": 7.905024127164349e-05,
      "loss": 0.02745906710624695,
      "step": 73840
    },
    {
      "epoch": 20.9622480840193,
      "grad_norm": 11.349584579467773,
      "learning_rate": 7.904740278172013e-05,
      "loss": 0.017695842683315276,
      "step": 73850
    },
    {
      "epoch": 20.965086573942664,
      "grad_norm": 0.44437509775161743,
      "learning_rate": 7.904456429179676e-05,
      "loss": 0.037027180194854736,
      "step": 73860
    },
    {
      "epoch": 20.967925063866023,
      "grad_norm": 0.7721370458602905,
      "learning_rate": 7.904172580187341e-05,
      "loss": 0.03285959661006928,
      "step": 73870
    },
    {
      "epoch": 20.970763553789386,
      "grad_norm": 4.44058895111084,
      "learning_rate": 7.903888731195006e-05,
      "loss": 0.012057169526815414,
      "step": 73880
    },
    {
      "epoch": 20.973602043712745,
      "grad_norm": 12.873336791992188,
      "learning_rate": 7.903604882202668e-05,
      "loss": 0.042155617475509645,
      "step": 73890
    },
    {
      "epoch": 20.976440533636104,
      "grad_norm": 2.300752639770508,
      "learning_rate": 7.903321033210332e-05,
      "loss": 0.02772943377494812,
      "step": 73900
    },
    {
      "epoch": 20.979279023559467,
      "grad_norm": 9.823734283447266,
      "learning_rate": 7.903037184217997e-05,
      "loss": 0.0104764923453331,
      "step": 73910
    },
    {
      "epoch": 20.982117513482827,
      "grad_norm": 2.656264305114746,
      "learning_rate": 7.90275333522566e-05,
      "loss": 0.01471453458070755,
      "step": 73920
    },
    {
      "epoch": 20.98495600340619,
      "grad_norm": 1.9895806312561035,
      "learning_rate": 7.902469486233323e-05,
      "loss": 0.010947420448064803,
      "step": 73930
    },
    {
      "epoch": 20.98779449332955,
      "grad_norm": 10.514595985412598,
      "learning_rate": 7.902185637240989e-05,
      "loss": 0.014849673211574554,
      "step": 73940
    },
    {
      "epoch": 20.990632983252908,
      "grad_norm": 5.333828449249268,
      "learning_rate": 7.901901788248652e-05,
      "loss": 0.012930047512054444,
      "step": 73950
    },
    {
      "epoch": 20.99347147317627,
      "grad_norm": 3.018665313720703,
      "learning_rate": 7.901617939256316e-05,
      "loss": 0.018447376787662506,
      "step": 73960
    },
    {
      "epoch": 20.99630996309963,
      "grad_norm": 1.616911768913269,
      "learning_rate": 7.90133409026398e-05,
      "loss": 0.027783718705177308,
      "step": 73970
    },
    {
      "epoch": 20.999148453022993,
      "grad_norm": 0.4653075039386749,
      "learning_rate": 7.901050241271644e-05,
      "loss": 0.015703950822353364,
      "step": 73980
    },
    {
      "epoch": 21.001986942946353,
      "grad_norm": 0.5165928602218628,
      "learning_rate": 7.900766392279307e-05,
      "loss": 0.00799490585923195,
      "step": 73990
    },
    {
      "epoch": 21.004825432869712,
      "grad_norm": 3.2757325172424316,
      "learning_rate": 7.900482543286972e-05,
      "loss": 0.00518273264169693,
      "step": 74000
    },
    {
      "epoch": 21.004825432869712,
      "eval_accuracy": 0.9689705601831246,
      "eval_loss": 0.10108955949544907,
      "eval_runtime": 31.1336,
      "eval_samples_per_second": 505.146,
      "eval_steps_per_second": 7.901,
      "step": 74000
    },
    {
      "epoch": 21.007663922793075,
      "grad_norm": 2.6399831771850586,
      "learning_rate": 7.900198694294637e-05,
      "loss": 0.006677420437335968,
      "step": 74010
    },
    {
      "epoch": 21.010502412716434,
      "grad_norm": 1.0280345678329468,
      "learning_rate": 7.8999148453023e-05,
      "loss": 0.003554373234510422,
      "step": 74020
    },
    {
      "epoch": 21.013340902639797,
      "grad_norm": 2.7186496257781982,
      "learning_rate": 7.899630996309963e-05,
      "loss": 0.010536560416221618,
      "step": 74030
    },
    {
      "epoch": 21.016179392563156,
      "grad_norm": 2.8670501708984375,
      "learning_rate": 7.899347147317628e-05,
      "loss": 0.011020071804523468,
      "step": 74040
    },
    {
      "epoch": 21.019017882486516,
      "grad_norm": 4.692769527435303,
      "learning_rate": 7.89906329832529e-05,
      "loss": 0.008445170521736146,
      "step": 74050
    },
    {
      "epoch": 21.02185637240988,
      "grad_norm": 1.6817808151245117,
      "learning_rate": 7.898779449332955e-05,
      "loss": 0.00845225602388382,
      "step": 74060
    },
    {
      "epoch": 21.024694862333238,
      "grad_norm": 14.013290405273438,
      "learning_rate": 7.89849560034062e-05,
      "loss": 0.015586079657077789,
      "step": 74070
    },
    {
      "epoch": 21.0275333522566,
      "grad_norm": 0.6257701516151428,
      "learning_rate": 7.898211751348283e-05,
      "loss": 0.011401771008968354,
      "step": 74080
    },
    {
      "epoch": 21.03037184217996,
      "grad_norm": 1.6471327543258667,
      "learning_rate": 7.897927902355947e-05,
      "loss": 0.007209101319313049,
      "step": 74090
    },
    {
      "epoch": 21.03321033210332,
      "grad_norm": 3.179908037185669,
      "learning_rate": 7.897644053363611e-05,
      "loss": 0.011152968555688859,
      "step": 74100
    },
    {
      "epoch": 21.036048822026682,
      "grad_norm": 0.4030959904193878,
      "learning_rate": 7.897360204371275e-05,
      "loss": 0.009203150868415833,
      "step": 74110
    },
    {
      "epoch": 21.03888731195004,
      "grad_norm": 0.9386562705039978,
      "learning_rate": 7.897076355378938e-05,
      "loss": 0.010810703039169312,
      "step": 74120
    },
    {
      "epoch": 21.041725801873405,
      "grad_norm": 2.520800828933716,
      "learning_rate": 7.896792506386602e-05,
      "loss": 0.006613048911094666,
      "step": 74130
    },
    {
      "epoch": 21.044564291796764,
      "grad_norm": 6.6119866371154785,
      "learning_rate": 7.896508657394268e-05,
      "loss": 0.02997497618198395,
      "step": 74140
    },
    {
      "epoch": 21.047402781720123,
      "grad_norm": 2.750290870666504,
      "learning_rate": 7.89622480840193e-05,
      "loss": 0.01177423596382141,
      "step": 74150
    },
    {
      "epoch": 21.050241271643486,
      "grad_norm": 4.717784881591797,
      "learning_rate": 7.895940959409595e-05,
      "loss": 0.02524755597114563,
      "step": 74160
    },
    {
      "epoch": 21.053079761566845,
      "grad_norm": 9.544171333312988,
      "learning_rate": 7.895657110417259e-05,
      "loss": 0.012386641651391982,
      "step": 74170
    },
    {
      "epoch": 21.05591825149021,
      "grad_norm": 18.918048858642578,
      "learning_rate": 7.895373261424921e-05,
      "loss": 0.04360443949699402,
      "step": 74180
    },
    {
      "epoch": 21.058756741413568,
      "grad_norm": 2.1022586822509766,
      "learning_rate": 7.895089412432586e-05,
      "loss": 0.01563587784767151,
      "step": 74190
    },
    {
      "epoch": 21.061595231336927,
      "grad_norm": 0.31142160296440125,
      "learning_rate": 7.894805563440251e-05,
      "loss": 0.005332668125629425,
      "step": 74200
    },
    {
      "epoch": 21.06443372126029,
      "grad_norm": 1.6475532054901123,
      "learning_rate": 7.894521714447914e-05,
      "loss": 0.01780291348695755,
      "step": 74210
    },
    {
      "epoch": 21.06727221118365,
      "grad_norm": 6.250597953796387,
      "learning_rate": 7.894237865455578e-05,
      "loss": 0.01200171336531639,
      "step": 74220
    },
    {
      "epoch": 21.070110701107012,
      "grad_norm": 0.46693089604377747,
      "learning_rate": 7.893954016463242e-05,
      "loss": 0.007612199336290359,
      "step": 74230
    },
    {
      "epoch": 21.07294919103037,
      "grad_norm": 0.7868378758430481,
      "learning_rate": 7.893670167470906e-05,
      "loss": 0.009841600060462951,
      "step": 74240
    },
    {
      "epoch": 21.075787680953734,
      "grad_norm": 0.13300862908363342,
      "learning_rate": 7.893386318478569e-05,
      "loss": 0.014477512240409851,
      "step": 74250
    },
    {
      "epoch": 21.078626170877094,
      "grad_norm": 2.9693217277526855,
      "learning_rate": 7.893102469486233e-05,
      "loss": 0.028074046969413756,
      "step": 74260
    },
    {
      "epoch": 21.081464660800453,
      "grad_norm": 0.18693657219409943,
      "learning_rate": 7.892818620493899e-05,
      "loss": 0.012653912603855132,
      "step": 74270
    },
    {
      "epoch": 21.084303150723816,
      "grad_norm": 0.18800175189971924,
      "learning_rate": 7.892534771501562e-05,
      "loss": 0.01829739660024643,
      "step": 74280
    },
    {
      "epoch": 21.087141640647175,
      "grad_norm": 9.543168067932129,
      "learning_rate": 7.892250922509226e-05,
      "loss": 0.01922588050365448,
      "step": 74290
    },
    {
      "epoch": 21.089980130570538,
      "grad_norm": 5.368625640869141,
      "learning_rate": 7.89196707351689e-05,
      "loss": 0.022974665462970733,
      "step": 74300
    },
    {
      "epoch": 21.092818620493897,
      "grad_norm": 9.876455307006836,
      "learning_rate": 7.891683224524553e-05,
      "loss": 0.011523884534835816,
      "step": 74310
    },
    {
      "epoch": 21.095657110417257,
      "grad_norm": 3.1807303428649902,
      "learning_rate": 7.891399375532217e-05,
      "loss": 0.010734608024358749,
      "step": 74320
    },
    {
      "epoch": 21.09849560034062,
      "grad_norm": 10.888519287109375,
      "learning_rate": 7.891115526539881e-05,
      "loss": 0.018534770607948302,
      "step": 74330
    },
    {
      "epoch": 21.10133409026398,
      "grad_norm": 0.78104567527771,
      "learning_rate": 7.890831677547545e-05,
      "loss": 0.024481944739818573,
      "step": 74340
    },
    {
      "epoch": 21.104172580187342,
      "grad_norm": 6.779789924621582,
      "learning_rate": 7.890547828555209e-05,
      "loss": 0.01251528412103653,
      "step": 74350
    },
    {
      "epoch": 21.1070110701107,
      "grad_norm": 0.4543623924255371,
      "learning_rate": 7.890263979562873e-05,
      "loss": 0.0240698367357254,
      "step": 74360
    },
    {
      "epoch": 21.10984956003406,
      "grad_norm": 9.469886779785156,
      "learning_rate": 7.889980130570537e-05,
      "loss": 0.02046073377132416,
      "step": 74370
    },
    {
      "epoch": 21.112688049957423,
      "grad_norm": 14.509222984313965,
      "learning_rate": 7.8896962815782e-05,
      "loss": 0.016410830616950988,
      "step": 74380
    },
    {
      "epoch": 21.115526539880783,
      "grad_norm": 0.5220840573310852,
      "learning_rate": 7.889412432585864e-05,
      "loss": 0.009171589463949203,
      "step": 74390
    },
    {
      "epoch": 21.118365029804146,
      "grad_norm": 7.434703826904297,
      "learning_rate": 7.889128583593528e-05,
      "loss": 0.02177475243806839,
      "step": 74400
    },
    {
      "epoch": 21.121203519727505,
      "grad_norm": 1.199174404144287,
      "learning_rate": 7.888844734601193e-05,
      "loss": 0.016218411922454833,
      "step": 74410
    },
    {
      "epoch": 21.124042009650864,
      "grad_norm": 2.3787662982940674,
      "learning_rate": 7.888560885608857e-05,
      "loss": 0.007263926416635513,
      "step": 74420
    },
    {
      "epoch": 21.126880499574227,
      "grad_norm": 7.946519374847412,
      "learning_rate": 7.888277036616521e-05,
      "loss": 0.02057839035987854,
      "step": 74430
    },
    {
      "epoch": 21.129718989497587,
      "grad_norm": 8.170157432556152,
      "learning_rate": 7.887993187624184e-05,
      "loss": 0.007064300030469895,
      "step": 74440
    },
    {
      "epoch": 21.13255747942095,
      "grad_norm": 6.6253767013549805,
      "learning_rate": 7.887709338631848e-05,
      "loss": 0.020754867792129518,
      "step": 74450
    },
    {
      "epoch": 21.13539596934431,
      "grad_norm": 9.66815185546875,
      "learning_rate": 7.887425489639512e-05,
      "loss": 0.009997677057981491,
      "step": 74460
    },
    {
      "epoch": 21.138234459267668,
      "grad_norm": 4.227406978607178,
      "learning_rate": 7.887141640647176e-05,
      "loss": 0.010682257264852524,
      "step": 74470
    },
    {
      "epoch": 21.14107294919103,
      "grad_norm": 0.5994613170623779,
      "learning_rate": 7.88685779165484e-05,
      "loss": 0.013827003538608551,
      "step": 74480
    },
    {
      "epoch": 21.14391143911439,
      "grad_norm": 1.6911752223968506,
      "learning_rate": 7.886573942662504e-05,
      "loss": 0.01313171088695526,
      "step": 74490
    },
    {
      "epoch": 21.146749929037753,
      "grad_norm": 4.3524346351623535,
      "learning_rate": 7.886290093670168e-05,
      "loss": 0.008445474505424499,
      "step": 74500
    },
    {
      "epoch": 21.146749929037753,
      "eval_accuracy": 0.9672537674063713,
      "eval_loss": 0.10577034205198288,
      "eval_runtime": 31.4875,
      "eval_samples_per_second": 499.468,
      "eval_steps_per_second": 7.813,
      "step": 74500
    },
    {
      "epoch": 21.149588418961113,
      "grad_norm": 1.578529953956604,
      "learning_rate": 7.886006244677831e-05,
      "loss": 0.01444295197725296,
      "step": 74510
    },
    {
      "epoch": 21.152426908884472,
      "grad_norm": 1.9007748365402222,
      "learning_rate": 7.885722395685495e-05,
      "loss": 0.00819685235619545,
      "step": 74520
    },
    {
      "epoch": 21.155265398807835,
      "grad_norm": 2.874579906463623,
      "learning_rate": 7.88543854669316e-05,
      "loss": 0.009656573832035064,
      "step": 74530
    },
    {
      "epoch": 21.158103888731194,
      "grad_norm": 10.027885437011719,
      "learning_rate": 7.885154697700824e-05,
      "loss": 0.011105933040380479,
      "step": 74540
    },
    {
      "epoch": 21.160942378654557,
      "grad_norm": 5.321685791015625,
      "learning_rate": 7.884870848708488e-05,
      "loss": 0.007541386038064956,
      "step": 74550
    },
    {
      "epoch": 21.163780868577916,
      "grad_norm": 6.998576641082764,
      "learning_rate": 7.884586999716152e-05,
      "loss": 0.007888945937156677,
      "step": 74560
    },
    {
      "epoch": 21.166619358501276,
      "grad_norm": 11.888397216796875,
      "learning_rate": 7.884303150723815e-05,
      "loss": 0.018642354011535644,
      "step": 74570
    },
    {
      "epoch": 21.16945784842464,
      "grad_norm": 2.0892932415008545,
      "learning_rate": 7.884019301731479e-05,
      "loss": 0.011233103275299073,
      "step": 74580
    },
    {
      "epoch": 21.172296338347998,
      "grad_norm": 2.484281301498413,
      "learning_rate": 7.883735452739143e-05,
      "loss": 0.00315704345703125,
      "step": 74590
    },
    {
      "epoch": 21.17513482827136,
      "grad_norm": 5.515682697296143,
      "learning_rate": 7.883451603746807e-05,
      "loss": 0.0116035558283329,
      "step": 74600
    },
    {
      "epoch": 21.17797331819472,
      "grad_norm": 12.491768836975098,
      "learning_rate": 7.883167754754471e-05,
      "loss": 0.019374684989452363,
      "step": 74610
    },
    {
      "epoch": 21.18081180811808,
      "grad_norm": 5.670478343963623,
      "learning_rate": 7.882883905762135e-05,
      "loss": 0.008849841356277467,
      "step": 74620
    },
    {
      "epoch": 21.183650298041442,
      "grad_norm": 1.0811883211135864,
      "learning_rate": 7.882600056769798e-05,
      "loss": 0.00990903303027153,
      "step": 74630
    },
    {
      "epoch": 21.1864887879648,
      "grad_norm": 1.6259132623672485,
      "learning_rate": 7.882316207777462e-05,
      "loss": 0.008647838979959488,
      "step": 74640
    },
    {
      "epoch": 21.189327277888165,
      "grad_norm": 3.8051600456237793,
      "learning_rate": 7.882032358785126e-05,
      "loss": 0.015223966538906097,
      "step": 74650
    },
    {
      "epoch": 21.192165767811524,
      "grad_norm": 1.9164929389953613,
      "learning_rate": 7.88174850979279e-05,
      "loss": 0.030438461899757387,
      "step": 74660
    },
    {
      "epoch": 21.195004257734887,
      "grad_norm": 0.6392512321472168,
      "learning_rate": 7.881464660800455e-05,
      "loss": 0.012022168934345245,
      "step": 74670
    },
    {
      "epoch": 21.197842747658246,
      "grad_norm": 6.6565399169921875,
      "learning_rate": 7.881180811808119e-05,
      "loss": 0.01997986137866974,
      "step": 74680
    },
    {
      "epoch": 21.200681237581605,
      "grad_norm": 0.23077349364757538,
      "learning_rate": 7.880896962815783e-05,
      "loss": 0.008937107771635056,
      "step": 74690
    },
    {
      "epoch": 21.20351972750497,
      "grad_norm": 1.7519068717956543,
      "learning_rate": 7.880613113823446e-05,
      "loss": 0.01710183620452881,
      "step": 74700
    },
    {
      "epoch": 21.206358217428328,
      "grad_norm": 2.4098291397094727,
      "learning_rate": 7.88032926483111e-05,
      "loss": 0.012014935910701751,
      "step": 74710
    },
    {
      "epoch": 21.20919670735169,
      "grad_norm": 0.6433802843093872,
      "learning_rate": 7.880045415838774e-05,
      "loss": 0.029374882578849792,
      "step": 74720
    },
    {
      "epoch": 21.21203519727505,
      "grad_norm": 5.814181327819824,
      "learning_rate": 7.879761566846438e-05,
      "loss": 0.006299564242362976,
      "step": 74730
    },
    {
      "epoch": 21.21487368719841,
      "grad_norm": 6.569656848907471,
      "learning_rate": 7.879477717854102e-05,
      "loss": 0.01947879195213318,
      "step": 74740
    },
    {
      "epoch": 21.217712177121772,
      "grad_norm": 2.881561517715454,
      "learning_rate": 7.879193868861766e-05,
      "loss": 0.010992929339408875,
      "step": 74750
    },
    {
      "epoch": 21.22055066704513,
      "grad_norm": 1.2301818132400513,
      "learning_rate": 7.878910019869429e-05,
      "loss": 0.009652945399284362,
      "step": 74760
    },
    {
      "epoch": 21.223389156968494,
      "grad_norm": 4.122570037841797,
      "learning_rate": 7.878626170877093e-05,
      "loss": 0.03406794667243958,
      "step": 74770
    },
    {
      "epoch": 21.226227646891854,
      "grad_norm": 3.4331257343292236,
      "learning_rate": 7.878342321884758e-05,
      "loss": 0.008916236460208893,
      "step": 74780
    },
    {
      "epoch": 21.229066136815213,
      "grad_norm": 4.2541351318359375,
      "learning_rate": 7.878058472892422e-05,
      "loss": 0.021498420834541322,
      "step": 74790
    },
    {
      "epoch": 21.231904626738576,
      "grad_norm": 1.6433011293411255,
      "learning_rate": 7.877774623900086e-05,
      "loss": 0.009510713815689086,
      "step": 74800
    },
    {
      "epoch": 21.234743116661935,
      "grad_norm": 0.45566901564598083,
      "learning_rate": 7.87749077490775e-05,
      "loss": 0.008892115950584412,
      "step": 74810
    },
    {
      "epoch": 21.237581606585298,
      "grad_norm": 1.3600140810012817,
      "learning_rate": 7.877206925915414e-05,
      "loss": 0.012219396978616714,
      "step": 74820
    },
    {
      "epoch": 21.240420096508657,
      "grad_norm": 5.141730785369873,
      "learning_rate": 7.876923076923077e-05,
      "loss": 0.017458860576152802,
      "step": 74830
    },
    {
      "epoch": 21.243258586432017,
      "grad_norm": 0.5994536280632019,
      "learning_rate": 7.876639227930741e-05,
      "loss": 0.010760480165481567,
      "step": 74840
    },
    {
      "epoch": 21.24609707635538,
      "grad_norm": 0.10164127498865128,
      "learning_rate": 7.876355378938405e-05,
      "loss": 0.006358616054058075,
      "step": 74850
    },
    {
      "epoch": 21.24893556627874,
      "grad_norm": 0.5450071692466736,
      "learning_rate": 7.876071529946068e-05,
      "loss": 0.004813630878925323,
      "step": 74860
    },
    {
      "epoch": 21.251774056202102,
      "grad_norm": 1.0310561656951904,
      "learning_rate": 7.875787680953733e-05,
      "loss": 0.025302216410636902,
      "step": 74870
    },
    {
      "epoch": 21.25461254612546,
      "grad_norm": 5.286652088165283,
      "learning_rate": 7.875503831961398e-05,
      "loss": 0.0180318683385849,
      "step": 74880
    },
    {
      "epoch": 21.25745103604882,
      "grad_norm": 7.939635753631592,
      "learning_rate": 7.87521998296906e-05,
      "loss": 0.01227591633796692,
      "step": 74890
    },
    {
      "epoch": 21.260289525972183,
      "grad_norm": 0.2786083221435547,
      "learning_rate": 7.874936133976724e-05,
      "loss": 0.023888009786605834,
      "step": 74900
    },
    {
      "epoch": 21.263128015895543,
      "grad_norm": 4.865163326263428,
      "learning_rate": 7.874652284984389e-05,
      "loss": 0.017589816451072694,
      "step": 74910
    },
    {
      "epoch": 21.265966505818906,
      "grad_norm": 6.7929182052612305,
      "learning_rate": 7.874368435992053e-05,
      "loss": 0.022551374137401582,
      "step": 74920
    },
    {
      "epoch": 21.268804995742265,
      "grad_norm": 3.380335807800293,
      "learning_rate": 7.874084586999717e-05,
      "loss": 0.007152558863162994,
      "step": 74930
    },
    {
      "epoch": 21.271643485665624,
      "grad_norm": 7.091317653656006,
      "learning_rate": 7.873800738007381e-05,
      "loss": 0.012429316341876984,
      "step": 74940
    },
    {
      "epoch": 21.274481975588987,
      "grad_norm": 0.5314913988113403,
      "learning_rate": 7.873516889015045e-05,
      "loss": 0.010797836631536484,
      "step": 74950
    },
    {
      "epoch": 21.277320465512346,
      "grad_norm": 13.70671272277832,
      "learning_rate": 7.873233040022708e-05,
      "loss": 0.030415144562721253,
      "step": 74960
    },
    {
      "epoch": 21.28015895543571,
      "grad_norm": 5.643704891204834,
      "learning_rate": 7.872949191030372e-05,
      "loss": 0.01302773654460907,
      "step": 74970
    },
    {
      "epoch": 21.28299744535907,
      "grad_norm": 1.379393219947815,
      "learning_rate": 7.872665342038036e-05,
      "loss": 0.011286456882953644,
      "step": 74980
    },
    {
      "epoch": 21.285835935282428,
      "grad_norm": 4.563117027282715,
      "learning_rate": 7.872381493045699e-05,
      "loss": 0.02213556319475174,
      "step": 74990
    },
    {
      "epoch": 21.28867442520579,
      "grad_norm": 8.047471046447754,
      "learning_rate": 7.872097644053365e-05,
      "loss": 0.021155691146850585,
      "step": 75000
    },
    {
      "epoch": 21.28867442520579,
      "eval_accuracy": 0.9647739556177275,
      "eval_loss": 0.1150110587477684,
      "eval_runtime": 32.1865,
      "eval_samples_per_second": 488.62,
      "eval_steps_per_second": 7.643,
      "step": 75000
    },
    {
      "epoch": 21.29151291512915,
      "grad_norm": 2.0192182064056396,
      "learning_rate": 7.871813795061029e-05,
      "loss": 0.011174393445253372,
      "step": 75010
    },
    {
      "epoch": 21.294351405052513,
      "grad_norm": 12.190423011779785,
      "learning_rate": 7.871529946068691e-05,
      "loss": 0.024303445219993593,
      "step": 75020
    },
    {
      "epoch": 21.297189894975872,
      "grad_norm": 0.7670546174049377,
      "learning_rate": 7.871246097076356e-05,
      "loss": 0.010821713507175446,
      "step": 75030
    },
    {
      "epoch": 21.300028384899235,
      "grad_norm": 9.03890323638916,
      "learning_rate": 7.87096224808402e-05,
      "loss": 0.019540603458881377,
      "step": 75040
    },
    {
      "epoch": 21.302866874822595,
      "grad_norm": 2.6298537254333496,
      "learning_rate": 7.870678399091684e-05,
      "loss": 0.02876976728439331,
      "step": 75050
    },
    {
      "epoch": 21.305705364745954,
      "grad_norm": 8.473215103149414,
      "learning_rate": 7.870394550099347e-05,
      "loss": 0.01282089501619339,
      "step": 75060
    },
    {
      "epoch": 21.308543854669317,
      "grad_norm": 9.589628219604492,
      "learning_rate": 7.870110701107012e-05,
      "loss": 0.029595237970352174,
      "step": 75070
    },
    {
      "epoch": 21.311382344592676,
      "grad_norm": 0.7744223475456238,
      "learning_rate": 7.869826852114676e-05,
      "loss": 0.012272786349058151,
      "step": 75080
    },
    {
      "epoch": 21.31422083451604,
      "grad_norm": 9.22041130065918,
      "learning_rate": 7.869543003122339e-05,
      "loss": 0.010503837466239929,
      "step": 75090
    },
    {
      "epoch": 21.3170593244394,
      "grad_norm": 3.693596839904785,
      "learning_rate": 7.869259154130003e-05,
      "loss": 0.017556154727935792,
      "step": 75100
    },
    {
      "epoch": 21.319897814362758,
      "grad_norm": 1.0030608177185059,
      "learning_rate": 7.868975305137667e-05,
      "loss": 0.0142797589302063,
      "step": 75110
    },
    {
      "epoch": 21.32273630428612,
      "grad_norm": 4.09512186050415,
      "learning_rate": 7.86869145614533e-05,
      "loss": 0.017878469824790955,
      "step": 75120
    },
    {
      "epoch": 21.32557479420948,
      "grad_norm": 8.11027717590332,
      "learning_rate": 7.868407607152996e-05,
      "loss": 0.011373965442180634,
      "step": 75130
    },
    {
      "epoch": 21.328413284132843,
      "grad_norm": 8.891484260559082,
      "learning_rate": 7.86812375816066e-05,
      "loss": 0.011668568849563599,
      "step": 75140
    },
    {
      "epoch": 21.331251774056202,
      "grad_norm": 3.9640254974365234,
      "learning_rate": 7.867839909168323e-05,
      "loss": 0.020840416848659515,
      "step": 75150
    },
    {
      "epoch": 21.33409026397956,
      "grad_norm": 5.166890621185303,
      "learning_rate": 7.867556060175987e-05,
      "loss": 0.025284305214881897,
      "step": 75160
    },
    {
      "epoch": 21.336928753902924,
      "grad_norm": 0.3052218556404114,
      "learning_rate": 7.867272211183651e-05,
      "loss": 0.011096090078353882,
      "step": 75170
    },
    {
      "epoch": 21.339767243826284,
      "grad_norm": 0.2023923248052597,
      "learning_rate": 7.866988362191315e-05,
      "loss": 0.03064195215702057,
      "step": 75180
    },
    {
      "epoch": 21.342605733749647,
      "grad_norm": 0.2708665132522583,
      "learning_rate": 7.866704513198978e-05,
      "loss": 0.01771521717309952,
      "step": 75190
    },
    {
      "epoch": 21.345444223673006,
      "grad_norm": 3.1364245414733887,
      "learning_rate": 7.866420664206643e-05,
      "loss": 0.017460963129997252,
      "step": 75200
    },
    {
      "epoch": 21.348282713596365,
      "grad_norm": 0.32325777411460876,
      "learning_rate": 7.866136815214307e-05,
      "loss": 0.01415797770023346,
      "step": 75210
    },
    {
      "epoch": 21.351121203519728,
      "grad_norm": 12.002775192260742,
      "learning_rate": 7.86585296622197e-05,
      "loss": 0.016395658254623413,
      "step": 75220
    },
    {
      "epoch": 21.353959693443088,
      "grad_norm": 26.531234741210938,
      "learning_rate": 7.865569117229634e-05,
      "loss": 0.04341956377029419,
      "step": 75230
    },
    {
      "epoch": 21.35679818336645,
      "grad_norm": 0.35819247364997864,
      "learning_rate": 7.865285268237298e-05,
      "loss": 0.037037086486816403,
      "step": 75240
    },
    {
      "epoch": 21.35963667328981,
      "grad_norm": 5.870786190032959,
      "learning_rate": 7.865001419244961e-05,
      "loss": 0.01768689453601837,
      "step": 75250
    },
    {
      "epoch": 21.36247516321317,
      "grad_norm": 1.8148654699325562,
      "learning_rate": 7.864717570252625e-05,
      "loss": 0.013791662454605103,
      "step": 75260
    },
    {
      "epoch": 21.365313653136532,
      "grad_norm": 1.9761031866073608,
      "learning_rate": 7.864433721260291e-05,
      "loss": 0.012117818742990494,
      "step": 75270
    },
    {
      "epoch": 21.36815214305989,
      "grad_norm": 3.0222856998443604,
      "learning_rate": 7.864149872267954e-05,
      "loss": 0.015822088718414305,
      "step": 75280
    },
    {
      "epoch": 21.370990632983254,
      "grad_norm": 2.839536666870117,
      "learning_rate": 7.863866023275618e-05,
      "loss": 0.014929421246051788,
      "step": 75290
    },
    {
      "epoch": 21.373829122906614,
      "grad_norm": 1.188350796699524,
      "learning_rate": 7.863582174283282e-05,
      "loss": 0.011323277652263642,
      "step": 75300
    },
    {
      "epoch": 21.376667612829973,
      "grad_norm": 2.121244430541992,
      "learning_rate": 7.863298325290946e-05,
      "loss": 0.028906035423278808,
      "step": 75310
    },
    {
      "epoch": 21.379506102753336,
      "grad_norm": 0.7676653861999512,
      "learning_rate": 7.863014476298609e-05,
      "loss": 0.019589702785015106,
      "step": 75320
    },
    {
      "epoch": 21.382344592676695,
      "grad_norm": 5.872739791870117,
      "learning_rate": 7.862730627306274e-05,
      "loss": 0.01680319309234619,
      "step": 75330
    },
    {
      "epoch": 21.385183082600058,
      "grad_norm": 2.170499563217163,
      "learning_rate": 7.862446778313937e-05,
      "loss": 0.010884857922792434,
      "step": 75340
    },
    {
      "epoch": 21.388021572523417,
      "grad_norm": 4.428328037261963,
      "learning_rate": 7.862162929321601e-05,
      "loss": 0.02672119140625,
      "step": 75350
    },
    {
      "epoch": 21.390860062446777,
      "grad_norm": 1.1998437643051147,
      "learning_rate": 7.861879080329265e-05,
      "loss": 0.01797333061695099,
      "step": 75360
    },
    {
      "epoch": 21.39369855237014,
      "grad_norm": 2.891197443008423,
      "learning_rate": 7.86159523133693e-05,
      "loss": 0.009053567051887512,
      "step": 75370
    },
    {
      "epoch": 21.3965370422935,
      "grad_norm": 2.026702404022217,
      "learning_rate": 7.861311382344592e-05,
      "loss": 0.01224260926246643,
      "step": 75380
    },
    {
      "epoch": 21.39937553221686,
      "grad_norm": 2.7068800926208496,
      "learning_rate": 7.861027533352256e-05,
      "loss": 0.026593488454818726,
      "step": 75390
    },
    {
      "epoch": 21.40221402214022,
      "grad_norm": 4.440280437469482,
      "learning_rate": 7.860743684359922e-05,
      "loss": 0.017116546630859375,
      "step": 75400
    },
    {
      "epoch": 21.405052512063584,
      "grad_norm": 1.9180214405059814,
      "learning_rate": 7.860459835367585e-05,
      "loss": 0.015582957863807678,
      "step": 75410
    },
    {
      "epoch": 21.407891001986943,
      "grad_norm": 7.818751811981201,
      "learning_rate": 7.860175986375249e-05,
      "loss": 0.01647305190563202,
      "step": 75420
    },
    {
      "epoch": 21.410729491910303,
      "grad_norm": 0.6835595965385437,
      "learning_rate": 7.859892137382913e-05,
      "loss": 0.00738518089056015,
      "step": 75430
    },
    {
      "epoch": 21.413567981833665,
      "grad_norm": 3.5683164596557617,
      "learning_rate": 7.859608288390577e-05,
      "loss": 0.013958995044231415,
      "step": 75440
    },
    {
      "epoch": 21.416406471757025,
      "grad_norm": 0.7420746684074402,
      "learning_rate": 7.85932443939824e-05,
      "loss": 0.016392260789871216,
      "step": 75450
    },
    {
      "epoch": 21.419244961680388,
      "grad_norm": 14.469867706298828,
      "learning_rate": 7.859040590405904e-05,
      "loss": 0.022926747798919678,
      "step": 75460
    },
    {
      "epoch": 21.422083451603747,
      "grad_norm": 0.4592805504798889,
      "learning_rate": 7.858756741413568e-05,
      "loss": 0.011049892753362656,
      "step": 75470
    },
    {
      "epoch": 21.424921941527106,
      "grad_norm": 3.571254253387451,
      "learning_rate": 7.858472892421232e-05,
      "loss": 0.015177573263645171,
      "step": 75480
    },
    {
      "epoch": 21.42776043145047,
      "grad_norm": 3.2654356956481934,
      "learning_rate": 7.858189043428896e-05,
      "loss": 0.017203959822654723,
      "step": 75490
    },
    {
      "epoch": 21.43059892137383,
      "grad_norm": 6.545628070831299,
      "learning_rate": 7.85790519443656e-05,
      "loss": 0.014694875478744507,
      "step": 75500
    },
    {
      "epoch": 21.43059892137383,
      "eval_accuracy": 0.9673809372416863,
      "eval_loss": 0.10844043642282486,
      "eval_runtime": 31.977,
      "eval_samples_per_second": 491.822,
      "eval_steps_per_second": 7.693,
      "step": 75500
    },
    {
      "epoch": 21.43343741129719,
      "grad_norm": 3.4959089756011963,
      "learning_rate": 7.857621345444223e-05,
      "loss": 0.017986363172531127,
      "step": 75510
    },
    {
      "epoch": 21.43627590122055,
      "grad_norm": 15.093456268310547,
      "learning_rate": 7.857337496451887e-05,
      "loss": 0.024652181565761565,
      "step": 75520
    },
    {
      "epoch": 21.43911439114391,
      "grad_norm": 2.2097418308258057,
      "learning_rate": 7.857053647459553e-05,
      "loss": 0.02050362229347229,
      "step": 75530
    },
    {
      "epoch": 21.441952881067273,
      "grad_norm": 0.3023211658000946,
      "learning_rate": 7.856769798467216e-05,
      "loss": 0.009076818823814392,
      "step": 75540
    },
    {
      "epoch": 21.444791370990632,
      "grad_norm": 4.824226379394531,
      "learning_rate": 7.85648594947488e-05,
      "loss": 0.01609097868204117,
      "step": 75550
    },
    {
      "epoch": 21.447629860913995,
      "grad_norm": 11.725981712341309,
      "learning_rate": 7.856202100482544e-05,
      "loss": 0.008646267652511596,
      "step": 75560
    },
    {
      "epoch": 21.450468350837355,
      "grad_norm": 2.0100808143615723,
      "learning_rate": 7.855918251490207e-05,
      "loss": 0.01589919477701187,
      "step": 75570
    },
    {
      "epoch": 21.453306840760714,
      "grad_norm": 0.2560100257396698,
      "learning_rate": 7.855634402497871e-05,
      "loss": 0.011624252796173096,
      "step": 75580
    },
    {
      "epoch": 21.456145330684077,
      "grad_norm": 6.72741174697876,
      "learning_rate": 7.855350553505535e-05,
      "loss": 0.017058047652244567,
      "step": 75590
    },
    {
      "epoch": 21.458983820607436,
      "grad_norm": 0.8524512648582458,
      "learning_rate": 7.855066704513199e-05,
      "loss": 0.02553977072238922,
      "step": 75600
    },
    {
      "epoch": 21.4618223105308,
      "grad_norm": 1.851595163345337,
      "learning_rate": 7.854782855520863e-05,
      "loss": 0.014857091009616852,
      "step": 75610
    },
    {
      "epoch": 21.46466080045416,
      "grad_norm": 2.2471249103546143,
      "learning_rate": 7.854499006528527e-05,
      "loss": 0.0066909223794937136,
      "step": 75620
    },
    {
      "epoch": 21.467499290377518,
      "grad_norm": 1.8378700017929077,
      "learning_rate": 7.854215157536192e-05,
      "loss": 0.01038406416773796,
      "step": 75630
    },
    {
      "epoch": 21.47033778030088,
      "grad_norm": 0.9276632070541382,
      "learning_rate": 7.853931308543854e-05,
      "loss": 0.006019279360771179,
      "step": 75640
    },
    {
      "epoch": 21.47317627022424,
      "grad_norm": 4.08128547668457,
      "learning_rate": 7.853647459551519e-05,
      "loss": 0.013660694658756255,
      "step": 75650
    },
    {
      "epoch": 21.476014760147603,
      "grad_norm": 0.3786870539188385,
      "learning_rate": 7.853363610559183e-05,
      "loss": 0.013316331803798676,
      "step": 75660
    },
    {
      "epoch": 21.478853250070962,
      "grad_norm": 6.131454944610596,
      "learning_rate": 7.853079761566847e-05,
      "loss": 0.02022210359573364,
      "step": 75670
    },
    {
      "epoch": 21.48169173999432,
      "grad_norm": 18.037982940673828,
      "learning_rate": 7.852795912574511e-05,
      "loss": 0.027197617292404174,
      "step": 75680
    },
    {
      "epoch": 21.484530229917684,
      "grad_norm": 0.18578213453292847,
      "learning_rate": 7.852512063582175e-05,
      "loss": 0.02393554449081421,
      "step": 75690
    },
    {
      "epoch": 21.487368719841044,
      "grad_norm": 8.948965072631836,
      "learning_rate": 7.852228214589838e-05,
      "loss": 0.015443667769432068,
      "step": 75700
    },
    {
      "epoch": 21.490207209764407,
      "grad_norm": 1.9039208889007568,
      "learning_rate": 7.851944365597502e-05,
      "loss": 0.013566416501998902,
      "step": 75710
    },
    {
      "epoch": 21.493045699687766,
      "grad_norm": 0.17158998548984528,
      "learning_rate": 7.851660516605166e-05,
      "loss": 0.013597056269645691,
      "step": 75720
    },
    {
      "epoch": 21.495884189611125,
      "grad_norm": 8.39345645904541,
      "learning_rate": 7.85137666761283e-05,
      "loss": 0.014701043069362641,
      "step": 75730
    },
    {
      "epoch": 21.498722679534488,
      "grad_norm": 0.5644033551216125,
      "learning_rate": 7.851092818620494e-05,
      "loss": 0.010521616786718369,
      "step": 75740
    },
    {
      "epoch": 21.501561169457847,
      "grad_norm": 8.533385276794434,
      "learning_rate": 7.850808969628159e-05,
      "loss": 0.007334190607070923,
      "step": 75750
    },
    {
      "epoch": 21.50439965938121,
      "grad_norm": 3.7465977668762207,
      "learning_rate": 7.850525120635823e-05,
      "loss": 0.013029253482818604,
      "step": 75760
    },
    {
      "epoch": 21.50723814930457,
      "grad_norm": 6.337586879730225,
      "learning_rate": 7.850241271643485e-05,
      "loss": 0.024566058814525605,
      "step": 75770
    },
    {
      "epoch": 21.510076639227933,
      "grad_norm": 1.4087105989456177,
      "learning_rate": 7.84995742265115e-05,
      "loss": 0.016794337332248686,
      "step": 75780
    },
    {
      "epoch": 21.512915129151292,
      "grad_norm": 15.563828468322754,
      "learning_rate": 7.849673573658814e-05,
      "loss": 0.02238583266735077,
      "step": 75790
    },
    {
      "epoch": 21.51575361907465,
      "grad_norm": 3.300431251525879,
      "learning_rate": 7.849389724666478e-05,
      "loss": 0.011002079397439957,
      "step": 75800
    },
    {
      "epoch": 21.518592108998014,
      "grad_norm": 5.7201128005981445,
      "learning_rate": 7.849105875674142e-05,
      "loss": 0.011888142675161362,
      "step": 75810
    },
    {
      "epoch": 21.521430598921373,
      "grad_norm": 20.418607711791992,
      "learning_rate": 7.848822026681806e-05,
      "loss": 0.03444123864173889,
      "step": 75820
    },
    {
      "epoch": 21.524269088844733,
      "grad_norm": 0.3705821931362152,
      "learning_rate": 7.848538177689469e-05,
      "loss": 0.024283920228481293,
      "step": 75830
    },
    {
      "epoch": 21.527107578768096,
      "grad_norm": 3.9132163524627686,
      "learning_rate": 7.848254328697133e-05,
      "loss": 0.015902027487754822,
      "step": 75840
    },
    {
      "epoch": 21.529946068691455,
      "grad_norm": 1.588884949684143,
      "learning_rate": 7.847970479704797e-05,
      "loss": 0.00878513902425766,
      "step": 75850
    },
    {
      "epoch": 21.532784558614818,
      "grad_norm": 0.15484026074409485,
      "learning_rate": 7.847686630712461e-05,
      "loss": 0.015248262882232666,
      "step": 75860
    },
    {
      "epoch": 21.535623048538177,
      "grad_norm": 5.4234209060668945,
      "learning_rate": 7.847402781720125e-05,
      "loss": 0.0068693958222866055,
      "step": 75870
    },
    {
      "epoch": 21.53846153846154,
      "grad_norm": 4.74229097366333,
      "learning_rate": 7.84711893272779e-05,
      "loss": 0.02267167568206787,
      "step": 75880
    },
    {
      "epoch": 21.5413000283849,
      "grad_norm": 0.520041286945343,
      "learning_rate": 7.846835083735454e-05,
      "loss": 0.020007678866386415,
      "step": 75890
    },
    {
      "epoch": 21.54413851830826,
      "grad_norm": 4.989255428314209,
      "learning_rate": 7.846551234743117e-05,
      "loss": 0.016596060991287232,
      "step": 75900
    },
    {
      "epoch": 21.54697700823162,
      "grad_norm": 7.046559810638428,
      "learning_rate": 7.846267385750781e-05,
      "loss": 0.017507560551166534,
      "step": 75910
    },
    {
      "epoch": 21.54981549815498,
      "grad_norm": 0.6137560606002808,
      "learning_rate": 7.845983536758445e-05,
      "loss": 0.0036479607224464417,
      "step": 75920
    },
    {
      "epoch": 21.552653988078344,
      "grad_norm": 1.9928137063980103,
      "learning_rate": 7.845699687766109e-05,
      "loss": 0.005956060439348221,
      "step": 75930
    },
    {
      "epoch": 21.555492478001703,
      "grad_norm": 0.5439964532852173,
      "learning_rate": 7.845415838773773e-05,
      "loss": 0.00621982216835022,
      "step": 75940
    },
    {
      "epoch": 21.558330967925063,
      "grad_norm": 1.5522284507751465,
      "learning_rate": 7.845131989781437e-05,
      "loss": 0.013819333910942078,
      "step": 75950
    },
    {
      "epoch": 21.561169457848425,
      "grad_norm": 0.7793036699295044,
      "learning_rate": 7.8448481407891e-05,
      "loss": 0.013180595636367799,
      "step": 75960
    },
    {
      "epoch": 21.564007947771785,
      "grad_norm": 7.142220973968506,
      "learning_rate": 7.844564291796764e-05,
      "loss": 0.025642618536949158,
      "step": 75970
    },
    {
      "epoch": 21.566846437695148,
      "grad_norm": 0.406651109457016,
      "learning_rate": 7.844280442804428e-05,
      "loss": 0.010029391199350358,
      "step": 75980
    },
    {
      "epoch": 21.569684927618507,
      "grad_norm": 0.28016993403434753,
      "learning_rate": 7.843996593812092e-05,
      "loss": 0.009296253323554993,
      "step": 75990
    },
    {
      "epoch": 21.572523417541866,
      "grad_norm": 3.4080655574798584,
      "learning_rate": 7.843712744819757e-05,
      "loss": 0.016548486053943635,
      "step": 76000
    },
    {
      "epoch": 21.572523417541866,
      "eval_accuracy": 0.9671265975710561,
      "eval_loss": 0.10753749310970306,
      "eval_runtime": 31.5089,
      "eval_samples_per_second": 499.128,
      "eval_steps_per_second": 7.807,
      "step": 76000
    },
    {
      "epoch": 21.57536190746523,
      "grad_norm": 5.59756326675415,
      "learning_rate": 7.843428895827421e-05,
      "loss": 0.01600130498409271,
      "step": 76010
    },
    {
      "epoch": 21.57820039738859,
      "grad_norm": 9.081891059875488,
      "learning_rate": 7.843145046835085e-05,
      "loss": 0.03320396840572357,
      "step": 76020
    },
    {
      "epoch": 21.58103888731195,
      "grad_norm": 0.8349407911300659,
      "learning_rate": 7.842861197842748e-05,
      "loss": 0.032330819964408876,
      "step": 76030
    },
    {
      "epoch": 21.58387737723531,
      "grad_norm": 3.4547901153564453,
      "learning_rate": 7.842577348850412e-05,
      "loss": 0.012268371880054474,
      "step": 76040
    },
    {
      "epoch": 21.58671586715867,
      "grad_norm": 1.637893557548523,
      "learning_rate": 7.842293499858076e-05,
      "loss": 0.012229278683662415,
      "step": 76050
    },
    {
      "epoch": 21.589554357082033,
      "grad_norm": 3.288370132446289,
      "learning_rate": 7.84200965086574e-05,
      "loss": 0.010865146666765213,
      "step": 76060
    },
    {
      "epoch": 21.592392847005392,
      "grad_norm": 3.339815616607666,
      "learning_rate": 7.841725801873404e-05,
      "loss": 0.0044495463371276855,
      "step": 76070
    },
    {
      "epoch": 21.595231336928755,
      "grad_norm": 1.7163646221160889,
      "learning_rate": 7.841441952881068e-05,
      "loss": 0.013310173153877258,
      "step": 76080
    },
    {
      "epoch": 21.598069826852115,
      "grad_norm": 12.90844440460205,
      "learning_rate": 7.841158103888731e-05,
      "loss": 0.03339901566505432,
      "step": 76090
    },
    {
      "epoch": 21.600908316775474,
      "grad_norm": 10.757219314575195,
      "learning_rate": 7.840874254896395e-05,
      "loss": 0.019683653116226198,
      "step": 76100
    },
    {
      "epoch": 21.603746806698837,
      "grad_norm": 6.217966079711914,
      "learning_rate": 7.84059040590406e-05,
      "loss": 0.018769189715385437,
      "step": 76110
    },
    {
      "epoch": 21.606585296622196,
      "grad_norm": 3.0087974071502686,
      "learning_rate": 7.840306556911724e-05,
      "loss": 0.019456428289413453,
      "step": 76120
    },
    {
      "epoch": 21.60942378654556,
      "grad_norm": 0.22842000424861908,
      "learning_rate": 7.840022707919388e-05,
      "loss": 0.01441936194896698,
      "step": 76130
    },
    {
      "epoch": 21.61226227646892,
      "grad_norm": 11.709714889526367,
      "learning_rate": 7.839738858927052e-05,
      "loss": 0.009755165874958038,
      "step": 76140
    },
    {
      "epoch": 21.615100766392278,
      "grad_norm": 5.41524076461792,
      "learning_rate": 7.839455009934716e-05,
      "loss": 0.015011473000049591,
      "step": 76150
    },
    {
      "epoch": 21.61793925631564,
      "grad_norm": 0.8947502970695496,
      "learning_rate": 7.839171160942379e-05,
      "loss": 0.018856336176395417,
      "step": 76160
    },
    {
      "epoch": 21.620777746239,
      "grad_norm": 0.9825923442840576,
      "learning_rate": 7.838887311950043e-05,
      "loss": 0.02439998686313629,
      "step": 76170
    },
    {
      "epoch": 21.623616236162363,
      "grad_norm": 5.807397842407227,
      "learning_rate": 7.838603462957707e-05,
      "loss": 0.01727648973464966,
      "step": 76180
    },
    {
      "epoch": 21.626454726085722,
      "grad_norm": 0.9970225691795349,
      "learning_rate": 7.83831961396537e-05,
      "loss": 0.006820657104253769,
      "step": 76190
    },
    {
      "epoch": 21.62929321600908,
      "grad_norm": 2.366140365600586,
      "learning_rate": 7.838035764973035e-05,
      "loss": 0.017223221063613892,
      "step": 76200
    },
    {
      "epoch": 21.632131705932444,
      "grad_norm": 2.6090049743652344,
      "learning_rate": 7.8377519159807e-05,
      "loss": 0.017718872427940367,
      "step": 76210
    },
    {
      "epoch": 21.634970195855804,
      "grad_norm": 13.620538711547852,
      "learning_rate": 7.837468066988362e-05,
      "loss": 0.019850121438503267,
      "step": 76220
    },
    {
      "epoch": 21.637808685779166,
      "grad_norm": 4.612888336181641,
      "learning_rate": 7.837184217996026e-05,
      "loss": 0.013288548588752747,
      "step": 76230
    },
    {
      "epoch": 21.640647175702526,
      "grad_norm": 4.95379638671875,
      "learning_rate": 7.83690036900369e-05,
      "loss": 0.030544334650039674,
      "step": 76240
    },
    {
      "epoch": 21.64348566562589,
      "grad_norm": 15.60219669342041,
      "learning_rate": 7.836616520011355e-05,
      "loss": 0.016857418417930602,
      "step": 76250
    },
    {
      "epoch": 21.646324155549248,
      "grad_norm": 2.1820733547210693,
      "learning_rate": 7.836332671019019e-05,
      "loss": 0.021499449014663698,
      "step": 76260
    },
    {
      "epoch": 21.649162645472607,
      "grad_norm": 3.8691420555114746,
      "learning_rate": 7.836048822026683e-05,
      "loss": 0.03640146553516388,
      "step": 76270
    },
    {
      "epoch": 21.65200113539597,
      "grad_norm": 17.362730026245117,
      "learning_rate": 7.835764973034346e-05,
      "loss": 0.03667094707489014,
      "step": 76280
    },
    {
      "epoch": 21.65483962531933,
      "grad_norm": 2.120270252227783,
      "learning_rate": 7.83548112404201e-05,
      "loss": 0.024965274333953857,
      "step": 76290
    },
    {
      "epoch": 21.657678115242692,
      "grad_norm": 3.7044076919555664,
      "learning_rate": 7.835197275049674e-05,
      "loss": 0.01726778894662857,
      "step": 76300
    },
    {
      "epoch": 21.660516605166052,
      "grad_norm": 2.1240944862365723,
      "learning_rate": 7.834913426057338e-05,
      "loss": 0.028606876730918884,
      "step": 76310
    },
    {
      "epoch": 21.66335509508941,
      "grad_norm": 6.846428394317627,
      "learning_rate": 7.834629577065001e-05,
      "loss": 0.025026938319206236,
      "step": 76320
    },
    {
      "epoch": 21.666193585012774,
      "grad_norm": 0.16483888030052185,
      "learning_rate": 7.834345728072666e-05,
      "loss": 0.016480317711830138,
      "step": 76330
    },
    {
      "epoch": 21.669032074936133,
      "grad_norm": 1.0717039108276367,
      "learning_rate": 7.83406187908033e-05,
      "loss": 0.005578938871622086,
      "step": 76340
    },
    {
      "epoch": 21.671870564859496,
      "grad_norm": 3.6805405616760254,
      "learning_rate": 7.833778030087993e-05,
      "loss": 0.010481615364551545,
      "step": 76350
    },
    {
      "epoch": 21.674709054782856,
      "grad_norm": 1.2032544612884521,
      "learning_rate": 7.833494181095657e-05,
      "loss": 0.009577855467796326,
      "step": 76360
    },
    {
      "epoch": 21.677547544706215,
      "grad_norm": 1.5188395977020264,
      "learning_rate": 7.833210332103322e-05,
      "loss": 0.010832367837429047,
      "step": 76370
    },
    {
      "epoch": 21.680386034629578,
      "grad_norm": 0.5624155402183533,
      "learning_rate": 7.832926483110984e-05,
      "loss": 0.010810618102550507,
      "step": 76380
    },
    {
      "epoch": 21.683224524552937,
      "grad_norm": 3.3656253814697266,
      "learning_rate": 7.832642634118648e-05,
      "loss": 0.011495617032051087,
      "step": 76390
    },
    {
      "epoch": 21.6860630144763,
      "grad_norm": 1.0415982007980347,
      "learning_rate": 7.832358785126314e-05,
      "loss": 0.010781669616699218,
      "step": 76400
    },
    {
      "epoch": 21.68890150439966,
      "grad_norm": 2.0982673168182373,
      "learning_rate": 7.832074936133977e-05,
      "loss": 0.04109894633293152,
      "step": 76410
    },
    {
      "epoch": 21.69173999432302,
      "grad_norm": 2.817380666732788,
      "learning_rate": 7.831791087141641e-05,
      "loss": 0.016175878047943116,
      "step": 76420
    },
    {
      "epoch": 21.69457848424638,
      "grad_norm": 8.560664176940918,
      "learning_rate": 7.831507238149305e-05,
      "loss": 0.011419643461704255,
      "step": 76430
    },
    {
      "epoch": 21.69741697416974,
      "grad_norm": 7.277251720428467,
      "learning_rate": 7.831223389156969e-05,
      "loss": 0.017267069220542906,
      "step": 76440
    },
    {
      "epoch": 21.700255464093104,
      "grad_norm": 2.763134717941284,
      "learning_rate": 7.830939540164632e-05,
      "loss": 0.01617017537355423,
      "step": 76450
    },
    {
      "epoch": 21.703093954016463,
      "grad_norm": 3.0343339443206787,
      "learning_rate": 7.830655691172297e-05,
      "loss": 0.01750296652317047,
      "step": 76460
    },
    {
      "epoch": 21.705932443939822,
      "grad_norm": 3.3168580532073975,
      "learning_rate": 7.830371842179962e-05,
      "loss": 0.0069194786250591275,
      "step": 76470
    },
    {
      "epoch": 21.708770933863185,
      "grad_norm": 1.5890346765518188,
      "learning_rate": 7.830087993187624e-05,
      "loss": 0.005515266954898834,
      "step": 76480
    },
    {
      "epoch": 21.711609423786545,
      "grad_norm": 4.021758556365967,
      "learning_rate": 7.829804144195288e-05,
      "loss": 0.009052052348852157,
      "step": 76490
    },
    {
      "epoch": 21.714447913709908,
      "grad_norm": 9.180389404296875,
      "learning_rate": 7.829520295202953e-05,
      "loss": 0.02535584568977356,
      "step": 76500
    },
    {
      "epoch": 21.714447913709908,
      "eval_accuracy": 0.9638201818528646,
      "eval_loss": 0.12102220207452774,
      "eval_runtime": 31.7455,
      "eval_samples_per_second": 495.409,
      "eval_steps_per_second": 7.749,
      "step": 76500
    },
    {
      "epoch": 21.717286403633267,
      "grad_norm": 2.0180230140686035,
      "learning_rate": 7.829236446210615e-05,
      "loss": 0.02152651846408844,
      "step": 76510
    },
    {
      "epoch": 21.720124893556626,
      "grad_norm": 0.6894105076789856,
      "learning_rate": 7.82895259721828e-05,
      "loss": 0.017346447706222533,
      "step": 76520
    },
    {
      "epoch": 21.72296338347999,
      "grad_norm": 0.2653830945491791,
      "learning_rate": 7.828668748225945e-05,
      "loss": 0.006541232764720917,
      "step": 76530
    },
    {
      "epoch": 21.72580187340335,
      "grad_norm": 4.41383695602417,
      "learning_rate": 7.828384899233608e-05,
      "loss": 0.015062984824180604,
      "step": 76540
    },
    {
      "epoch": 21.72864036332671,
      "grad_norm": 16.263296127319336,
      "learning_rate": 7.828101050241272e-05,
      "loss": 0.026714929938316347,
      "step": 76550
    },
    {
      "epoch": 21.73147885325007,
      "grad_norm": 4.507694721221924,
      "learning_rate": 7.827817201248936e-05,
      "loss": 0.01858513355255127,
      "step": 76560
    },
    {
      "epoch": 21.73431734317343,
      "grad_norm": 3.594895362854004,
      "learning_rate": 7.8275333522566e-05,
      "loss": 0.010094435513019561,
      "step": 76570
    },
    {
      "epoch": 21.737155833096793,
      "grad_norm": 11.130035400390625,
      "learning_rate": 7.827249503264263e-05,
      "loss": 0.016239157319068907,
      "step": 76580
    },
    {
      "epoch": 21.739994323020152,
      "grad_norm": 12.336492538452148,
      "learning_rate": 7.826965654271927e-05,
      "loss": 0.015177685022354125,
      "step": 76590
    },
    {
      "epoch": 21.742832812943515,
      "grad_norm": 0.9573445916175842,
      "learning_rate": 7.826681805279593e-05,
      "loss": 0.015393544733524323,
      "step": 76600
    },
    {
      "epoch": 21.745671302866874,
      "grad_norm": 3.006967067718506,
      "learning_rate": 7.826397956287255e-05,
      "loss": 0.010941926389932632,
      "step": 76610
    },
    {
      "epoch": 21.748509792790237,
      "grad_norm": 8.52607250213623,
      "learning_rate": 7.82611410729492e-05,
      "loss": 0.01392655223608017,
      "step": 76620
    },
    {
      "epoch": 21.751348282713597,
      "grad_norm": 0.6521673202514648,
      "learning_rate": 7.825830258302584e-05,
      "loss": 0.016238363087177278,
      "step": 76630
    },
    {
      "epoch": 21.754186772636956,
      "grad_norm": 2.1173746585845947,
      "learning_rate": 7.825546409310246e-05,
      "loss": 0.02088761180639267,
      "step": 76640
    },
    {
      "epoch": 21.75702526256032,
      "grad_norm": 15.88673210144043,
      "learning_rate": 7.82526256031791e-05,
      "loss": 0.02807828187942505,
      "step": 76650
    },
    {
      "epoch": 21.759863752483678,
      "grad_norm": 11.676653861999512,
      "learning_rate": 7.825007096224809e-05,
      "loss": 0.0297097772359848,
      "step": 76660
    },
    {
      "epoch": 21.76270224240704,
      "grad_norm": 11.488948822021484,
      "learning_rate": 7.824723247232473e-05,
      "loss": 0.039601188898086545,
      "step": 76670
    },
    {
      "epoch": 21.7655407323304,
      "grad_norm": 1.1914842128753662,
      "learning_rate": 7.824439398240137e-05,
      "loss": 0.019672128558158874,
      "step": 76680
    },
    {
      "epoch": 21.76837922225376,
      "grad_norm": 1.257739543914795,
      "learning_rate": 7.8241555492478e-05,
      "loss": 0.007462108135223388,
      "step": 76690
    },
    {
      "epoch": 21.771217712177123,
      "grad_norm": 4.431422710418701,
      "learning_rate": 7.823871700255464e-05,
      "loss": 0.018943874537944792,
      "step": 76700
    },
    {
      "epoch": 21.774056202100482,
      "grad_norm": 2.9444966316223145,
      "learning_rate": 7.823587851263129e-05,
      "loss": 0.03041023313999176,
      "step": 76710
    },
    {
      "epoch": 21.776894692023845,
      "grad_norm": 0.32359835505485535,
      "learning_rate": 7.823304002270792e-05,
      "loss": 0.022947846353054045,
      "step": 76720
    },
    {
      "epoch": 21.779733181947204,
      "grad_norm": 2.0393660068511963,
      "learning_rate": 7.823020153278456e-05,
      "loss": 0.01531689167022705,
      "step": 76730
    },
    {
      "epoch": 21.782571671870564,
      "grad_norm": 0.7711515426635742,
      "learning_rate": 7.82273630428612e-05,
      "loss": 0.01182211861014366,
      "step": 76740
    },
    {
      "epoch": 21.785410161793926,
      "grad_norm": 2.697129249572754,
      "learning_rate": 7.822452455293784e-05,
      "loss": 0.003196723759174347,
      "step": 76750
    },
    {
      "epoch": 21.788248651717286,
      "grad_norm": 8.89659309387207,
      "learning_rate": 7.822168606301447e-05,
      "loss": 0.027492699027061463,
      "step": 76760
    },
    {
      "epoch": 21.79108714164065,
      "grad_norm": 0.5973318219184875,
      "learning_rate": 7.821884757309111e-05,
      "loss": 0.009495507925748825,
      "step": 76770
    },
    {
      "epoch": 21.793925631564008,
      "grad_norm": 3.4102282524108887,
      "learning_rate": 7.821600908316777e-05,
      "loss": 0.011741966009140015,
      "step": 76780
    },
    {
      "epoch": 21.796764121487367,
      "grad_norm": 8.232495307922363,
      "learning_rate": 7.82131705932444e-05,
      "loss": 0.018143440783023834,
      "step": 76790
    },
    {
      "epoch": 21.79960261141073,
      "grad_norm": 2.0071282386779785,
      "learning_rate": 7.821033210332104e-05,
      "loss": 0.03609492182731629,
      "step": 76800
    },
    {
      "epoch": 21.80244110133409,
      "grad_norm": 12.894474983215332,
      "learning_rate": 7.820749361339768e-05,
      "loss": 0.03749408721923828,
      "step": 76810
    },
    {
      "epoch": 21.805279591257452,
      "grad_norm": 9.95533561706543,
      "learning_rate": 7.820465512347431e-05,
      "loss": 0.037414824962615965,
      "step": 76820
    },
    {
      "epoch": 21.80811808118081,
      "grad_norm": 1.8562756776809692,
      "learning_rate": 7.820181663355095e-05,
      "loss": 0.011846522241830826,
      "step": 76830
    },
    {
      "epoch": 21.81095657110417,
      "grad_norm": 1.3606544733047485,
      "learning_rate": 7.81989781436276e-05,
      "loss": 0.015765571594238283,
      "step": 76840
    },
    {
      "epoch": 21.813795061027534,
      "grad_norm": 5.749678134918213,
      "learning_rate": 7.819613965370423e-05,
      "loss": 0.017157140374183654,
      "step": 76850
    },
    {
      "epoch": 21.816633550950893,
      "grad_norm": 10.163677215576172,
      "learning_rate": 7.819330116378087e-05,
      "loss": 0.01743210256099701,
      "step": 76860
    },
    {
      "epoch": 21.819472040874256,
      "grad_norm": 0.16774751245975494,
      "learning_rate": 7.819046267385751e-05,
      "loss": 0.004885649681091309,
      "step": 76870
    },
    {
      "epoch": 21.822310530797616,
      "grad_norm": 1.4132473468780518,
      "learning_rate": 7.818762418393416e-05,
      "loss": 0.006147371232509613,
      "step": 76880
    },
    {
      "epoch": 21.825149020720975,
      "grad_norm": 3.599931240081787,
      "learning_rate": 7.818478569401078e-05,
      "loss": 0.019440756738185884,
      "step": 76890
    },
    {
      "epoch": 21.827987510644338,
      "grad_norm": 0.451099693775177,
      "learning_rate": 7.818194720408742e-05,
      "loss": 0.012780244648456573,
      "step": 76900
    },
    {
      "epoch": 21.830826000567697,
      "grad_norm": 5.143447399139404,
      "learning_rate": 7.817910871416408e-05,
      "loss": 0.010682998597621918,
      "step": 76910
    },
    {
      "epoch": 21.83366449049106,
      "grad_norm": 4.042118549346924,
      "learning_rate": 7.817627022424071e-05,
      "loss": 0.022549442946910858,
      "step": 76920
    },
    {
      "epoch": 21.83650298041442,
      "grad_norm": 3.9703948497772217,
      "learning_rate": 7.817343173431735e-05,
      "loss": 0.02118019461631775,
      "step": 76930
    },
    {
      "epoch": 21.83934147033778,
      "grad_norm": 8.298737525939941,
      "learning_rate": 7.817059324439399e-05,
      "loss": 0.007871933281421661,
      "step": 76940
    },
    {
      "epoch": 21.84217996026114,
      "grad_norm": 1.2813085317611694,
      "learning_rate": 7.816775475447062e-05,
      "loss": 0.009709246456623077,
      "step": 76950
    },
    {
      "epoch": 21.8450184501845,
      "grad_norm": 6.624592304229736,
      "learning_rate": 7.816491626454726e-05,
      "loss": 0.019441741704940795,
      "step": 76960
    },
    {
      "epoch": 21.847856940107864,
      "grad_norm": 3.284335136413574,
      "learning_rate": 7.81620777746239e-05,
      "loss": 0.023923635482788086,
      "step": 76970
    },
    {
      "epoch": 21.850695430031223,
      "grad_norm": 0.8073540925979614,
      "learning_rate": 7.815923928470054e-05,
      "loss": 0.009224055707454682,
      "step": 76980
    },
    {
      "epoch": 21.853533919954586,
      "grad_norm": 3.46307110786438,
      "learning_rate": 7.815640079477718e-05,
      "loss": 0.015401774644851684,
      "step": 76990
    },
    {
      "epoch": 21.856372409877945,
      "grad_norm": 0.9773533940315247,
      "learning_rate": 7.815356230485382e-05,
      "loss": 0.012726616859436036,
      "step": 77000
    },
    {
      "epoch": 21.856372409877945,
      "eval_accuracy": 0.964837540535385,
      "eval_loss": 0.10746651887893677,
      "eval_runtime": 31.4459,
      "eval_samples_per_second": 500.129,
      "eval_steps_per_second": 7.823,
      "step": 77000
    },
    {
      "epoch": 21.859210899801305,
      "grad_norm": 0.2333589494228363,
      "learning_rate": 7.815072381493047e-05,
      "loss": 0.009247449785470962,
      "step": 77010
    },
    {
      "epoch": 21.862049389724667,
      "grad_norm": 12.921958923339844,
      "learning_rate": 7.81478853250071e-05,
      "loss": 0.025931236147880555,
      "step": 77020
    },
    {
      "epoch": 21.864887879648027,
      "grad_norm": 8.462388038635254,
      "learning_rate": 7.814504683508374e-05,
      "loss": 0.020560833811759948,
      "step": 77030
    },
    {
      "epoch": 21.86772636957139,
      "grad_norm": 1.0064833164215088,
      "learning_rate": 7.814220834516039e-05,
      "loss": 0.02224023938179016,
      "step": 77040
    },
    {
      "epoch": 21.87056485949475,
      "grad_norm": 4.009249210357666,
      "learning_rate": 7.813936985523702e-05,
      "loss": 0.016633112728595734,
      "step": 77050
    },
    {
      "epoch": 21.87340334941811,
      "grad_norm": 9.028077125549316,
      "learning_rate": 7.813653136531366e-05,
      "loss": 0.037223371863365176,
      "step": 77060
    },
    {
      "epoch": 21.87624183934147,
      "grad_norm": 2.489049196243286,
      "learning_rate": 7.81336928753903e-05,
      "loss": 0.020155759155750276,
      "step": 77070
    },
    {
      "epoch": 21.87908032926483,
      "grad_norm": 10.08251667022705,
      "learning_rate": 7.813085438546693e-05,
      "loss": 0.01674887239933014,
      "step": 77080
    },
    {
      "epoch": 21.881918819188193,
      "grad_norm": 0.4034408628940582,
      "learning_rate": 7.812801589554357e-05,
      "loss": 0.0044997379183769224,
      "step": 77090
    },
    {
      "epoch": 21.884757309111553,
      "grad_norm": 1.4603570699691772,
      "learning_rate": 7.812517740562021e-05,
      "loss": 0.007852521538734437,
      "step": 77100
    },
    {
      "epoch": 21.887595799034912,
      "grad_norm": 0.9303902983665466,
      "learning_rate": 7.812233891569685e-05,
      "loss": 0.027444884181022644,
      "step": 77110
    },
    {
      "epoch": 21.890434288958275,
      "grad_norm": 1.6801143884658813,
      "learning_rate": 7.81195004257735e-05,
      "loss": 0.014386960864067077,
      "step": 77120
    },
    {
      "epoch": 21.893272778881634,
      "grad_norm": 4.740293979644775,
      "learning_rate": 7.811666193585014e-05,
      "loss": 0.013454185426235199,
      "step": 77130
    },
    {
      "epoch": 21.896111268804997,
      "grad_norm": 1.0744178295135498,
      "learning_rate": 7.811382344592678e-05,
      "loss": 0.009902026504278183,
      "step": 77140
    },
    {
      "epoch": 21.898949758728357,
      "grad_norm": 11.454477310180664,
      "learning_rate": 7.81109849560034e-05,
      "loss": 0.014002050459384918,
      "step": 77150
    },
    {
      "epoch": 21.901788248651716,
      "grad_norm": 6.228581428527832,
      "learning_rate": 7.810814646608005e-05,
      "loss": 0.024948759377002715,
      "step": 77160
    },
    {
      "epoch": 21.90462673857508,
      "grad_norm": 0.2029803991317749,
      "learning_rate": 7.810530797615669e-05,
      "loss": 0.012694242596626281,
      "step": 77170
    },
    {
      "epoch": 21.907465228498438,
      "grad_norm": 0.6979201436042786,
      "learning_rate": 7.810246948623333e-05,
      "loss": 0.005986906960606575,
      "step": 77180
    },
    {
      "epoch": 21.9103037184218,
      "grad_norm": 4.384439468383789,
      "learning_rate": 7.809963099630997e-05,
      "loss": 0.016067472100257874,
      "step": 77190
    },
    {
      "epoch": 21.91314220834516,
      "grad_norm": 4.829434394836426,
      "learning_rate": 7.809679250638661e-05,
      "loss": 0.008404189348220825,
      "step": 77200
    },
    {
      "epoch": 21.91598069826852,
      "grad_norm": 0.5981588959693909,
      "learning_rate": 7.809395401646324e-05,
      "loss": 0.016064085066318512,
      "step": 77210
    },
    {
      "epoch": 21.918819188191883,
      "grad_norm": 7.731984615325928,
      "learning_rate": 7.809111552653988e-05,
      "loss": 0.02272556722164154,
      "step": 77220
    },
    {
      "epoch": 21.921657678115242,
      "grad_norm": 1.8182883262634277,
      "learning_rate": 7.808827703661652e-05,
      "loss": 0.010691514611244202,
      "step": 77230
    },
    {
      "epoch": 21.924496168038605,
      "grad_norm": 19.29737091064453,
      "learning_rate": 7.808543854669316e-05,
      "loss": 0.030672550201416016,
      "step": 77240
    },
    {
      "epoch": 21.927334657961964,
      "grad_norm": 1.276174783706665,
      "learning_rate": 7.80826000567698e-05,
      "loss": 0.010022606700658798,
      "step": 77250
    },
    {
      "epoch": 21.930173147885323,
      "grad_norm": 8.187736511230469,
      "learning_rate": 7.807976156684645e-05,
      "loss": 0.010281653702259063,
      "step": 77260
    },
    {
      "epoch": 21.933011637808686,
      "grad_norm": 11.26250171661377,
      "learning_rate": 7.807692307692307e-05,
      "loss": 0.02555321455001831,
      "step": 77270
    },
    {
      "epoch": 21.935850127732046,
      "grad_norm": 0.2935706377029419,
      "learning_rate": 7.807408458699972e-05,
      "loss": 0.016476118564605714,
      "step": 77280
    },
    {
      "epoch": 21.93868861765541,
      "grad_norm": 9.60158634185791,
      "learning_rate": 7.807124609707636e-05,
      "loss": 0.01854208707809448,
      "step": 77290
    },
    {
      "epoch": 21.941527107578768,
      "grad_norm": 1.9438365697860718,
      "learning_rate": 7.8068407607153e-05,
      "loss": 0.011578445881605148,
      "step": 77300
    },
    {
      "epoch": 21.944365597502127,
      "grad_norm": 1.3686802387237549,
      "learning_rate": 7.806556911722964e-05,
      "loss": 0.00876007303595543,
      "step": 77310
    },
    {
      "epoch": 21.94720408742549,
      "grad_norm": 1.525789499282837,
      "learning_rate": 7.806273062730628e-05,
      "loss": 0.02529314160346985,
      "step": 77320
    },
    {
      "epoch": 21.95004257734885,
      "grad_norm": 4.271810531616211,
      "learning_rate": 7.805989213738292e-05,
      "loss": 0.02377483993768692,
      "step": 77330
    },
    {
      "epoch": 21.952881067272212,
      "grad_norm": 5.852486610412598,
      "learning_rate": 7.805705364745955e-05,
      "loss": 0.034498760104179384,
      "step": 77340
    },
    {
      "epoch": 21.95571955719557,
      "grad_norm": 0.8101273775100708,
      "learning_rate": 7.805421515753619e-05,
      "loss": 0.008817987143993377,
      "step": 77350
    },
    {
      "epoch": 21.958558047118935,
      "grad_norm": 0.2558135688304901,
      "learning_rate": 7.805137666761283e-05,
      "loss": 0.024744048714637756,
      "step": 77360
    },
    {
      "epoch": 21.961396537042294,
      "grad_norm": 15.887232780456543,
      "learning_rate": 7.804853817768946e-05,
      "loss": 0.012068048864603043,
      "step": 77370
    },
    {
      "epoch": 21.964235026965653,
      "grad_norm": 2.8790669441223145,
      "learning_rate": 7.804569968776612e-05,
      "loss": 0.012857344746589661,
      "step": 77380
    },
    {
      "epoch": 21.967073516889016,
      "grad_norm": 4.863633155822754,
      "learning_rate": 7.804286119784276e-05,
      "loss": 0.036144202947616576,
      "step": 77390
    },
    {
      "epoch": 21.969912006812375,
      "grad_norm": 2.471874713897705,
      "learning_rate": 7.804002270791939e-05,
      "loss": 0.02833322584629059,
      "step": 77400
    },
    {
      "epoch": 21.972750496735735,
      "grad_norm": 5.715687274932861,
      "learning_rate": 7.803718421799603e-05,
      "loss": 0.009659874439239501,
      "step": 77410
    },
    {
      "epoch": 21.975588986659098,
      "grad_norm": 0.931130051612854,
      "learning_rate": 7.803434572807267e-05,
      "loss": 0.01696839779615402,
      "step": 77420
    },
    {
      "epoch": 21.978427476582457,
      "grad_norm": 5.212191581726074,
      "learning_rate": 7.803150723814931e-05,
      "loss": 0.026097530126571657,
      "step": 77430
    },
    {
      "epoch": 21.98126596650582,
      "grad_norm": 6.8784942626953125,
      "learning_rate": 7.802866874822595e-05,
      "loss": 0.027269402146339418,
      "step": 77440
    },
    {
      "epoch": 21.98410445642918,
      "grad_norm": 0.5826104283332825,
      "learning_rate": 7.802583025830259e-05,
      "loss": 0.008418075740337372,
      "step": 77450
    },
    {
      "epoch": 21.986942946352542,
      "grad_norm": 0.7270312905311584,
      "learning_rate": 7.802299176837923e-05,
      "loss": 0.008091999590396881,
      "step": 77460
    },
    {
      "epoch": 21.9897814362759,
      "grad_norm": 2.320132255554199,
      "learning_rate": 7.802015327845586e-05,
      "loss": 0.006875626742839813,
      "step": 77470
    },
    {
      "epoch": 21.99261992619926,
      "grad_norm": 3.2568306922912598,
      "learning_rate": 7.80173147885325e-05,
      "loss": 0.007874561101198196,
      "step": 77480
    },
    {
      "epoch": 21.995458416122624,
      "grad_norm": 0.14644795656204224,
      "learning_rate": 7.801447629860914e-05,
      "loss": 0.0060274101793766025,
      "step": 77490
    },
    {
      "epoch": 21.998296906045983,
      "grad_norm": 8.998992919921875,
      "learning_rate": 7.801163780868577e-05,
      "loss": 0.02764994204044342,
      "step": 77500
    },
    {
      "epoch": 21.998296906045983,
      "eval_accuracy": 0.9660456539708782,
      "eval_loss": 0.11295709013938904,
      "eval_runtime": 31.6057,
      "eval_samples_per_second": 497.6,
      "eval_steps_per_second": 7.783,
      "step": 77500
    },
    {
      "epoch": 22.001135395969346,
      "grad_norm": 0.8961758017539978,
      "learning_rate": 7.800879931876243e-05,
      "loss": 0.008397797495126725,
      "step": 77510
    },
    {
      "epoch": 22.003973885892705,
      "grad_norm": 1.009544849395752,
      "learning_rate": 7.800596082883907e-05,
      "loss": 0.021177010238170625,
      "step": 77520
    },
    {
      "epoch": 22.006812375816065,
      "grad_norm": 6.588967323303223,
      "learning_rate": 7.80031223389157e-05,
      "loss": 0.03147798776626587,
      "step": 77530
    },
    {
      "epoch": 22.009650865739427,
      "grad_norm": 0.9545502066612244,
      "learning_rate": 7.800028384899234e-05,
      "loss": 0.018263402581214904,
      "step": 77540
    },
    {
      "epoch": 22.012489355662787,
      "grad_norm": 2.773300886154175,
      "learning_rate": 7.799744535906898e-05,
      "loss": 0.008683116734027862,
      "step": 77550
    },
    {
      "epoch": 22.01532784558615,
      "grad_norm": 3.157545328140259,
      "learning_rate": 7.799460686914562e-05,
      "loss": 0.011756989359855651,
      "step": 77560
    },
    {
      "epoch": 22.01816633550951,
      "grad_norm": 0.31751975417137146,
      "learning_rate": 7.799176837922225e-05,
      "loss": 0.021091081202030182,
      "step": 77570
    },
    {
      "epoch": 22.02100482543287,
      "grad_norm": 0.3000395894050598,
      "learning_rate": 7.79889298892989e-05,
      "loss": 0.0038685090839862823,
      "step": 77580
    },
    {
      "epoch": 22.02384331535623,
      "grad_norm": 0.4833454489707947,
      "learning_rate": 7.798609139937554e-05,
      "loss": 0.008476876467466355,
      "step": 77590
    },
    {
      "epoch": 22.02668180527959,
      "grad_norm": 0.2859414517879486,
      "learning_rate": 7.798325290945217e-05,
      "loss": 0.010607036203145981,
      "step": 77600
    },
    {
      "epoch": 22.029520295202953,
      "grad_norm": 0.2562928795814514,
      "learning_rate": 7.798041441952881e-05,
      "loss": 0.008344996720552444,
      "step": 77610
    },
    {
      "epoch": 22.032358785126313,
      "grad_norm": 5.452681064605713,
      "learning_rate": 7.797757592960545e-05,
      "loss": 0.012287989258766174,
      "step": 77620
    },
    {
      "epoch": 22.035197275049672,
      "grad_norm": 0.38039126992225647,
      "learning_rate": 7.797473743968208e-05,
      "loss": 0.02315075695514679,
      "step": 77630
    },
    {
      "epoch": 22.038035764973035,
      "grad_norm": 8.661182403564453,
      "learning_rate": 7.797189894975874e-05,
      "loss": 0.00963817909359932,
      "step": 77640
    },
    {
      "epoch": 22.040874254896394,
      "grad_norm": 10.70436954498291,
      "learning_rate": 7.796906045983538e-05,
      "loss": 0.01076717972755432,
      "step": 77650
    },
    {
      "epoch": 22.043712744819757,
      "grad_norm": 0.6941260099411011,
      "learning_rate": 7.7966221969912e-05,
      "loss": 0.01910896450281143,
      "step": 77660
    },
    {
      "epoch": 22.046551234743117,
      "grad_norm": 13.742024421691895,
      "learning_rate": 7.796338347998865e-05,
      "loss": 0.010995467752218246,
      "step": 77670
    },
    {
      "epoch": 22.049389724666476,
      "grad_norm": 8.066434860229492,
      "learning_rate": 7.796054499006529e-05,
      "loss": 0.013952425122261048,
      "step": 77680
    },
    {
      "epoch": 22.05222821458984,
      "grad_norm": 2.0535006523132324,
      "learning_rate": 7.795770650014193e-05,
      "loss": 0.009541848301887512,
      "step": 77690
    },
    {
      "epoch": 22.055066704513198,
      "grad_norm": 1.1072839498519897,
      "learning_rate": 7.795486801021856e-05,
      "loss": 0.010959744453430176,
      "step": 77700
    },
    {
      "epoch": 22.05790519443656,
      "grad_norm": 0.14257793128490448,
      "learning_rate": 7.795202952029521e-05,
      "loss": 0.018217116594314575,
      "step": 77710
    },
    {
      "epoch": 22.06074368435992,
      "grad_norm": 1.2833832502365112,
      "learning_rate": 7.794919103037185e-05,
      "loss": 0.004118860885500908,
      "step": 77720
    },
    {
      "epoch": 22.06358217428328,
      "grad_norm": 0.1957813948392868,
      "learning_rate": 7.794635254044848e-05,
      "loss": 0.015275177359580994,
      "step": 77730
    },
    {
      "epoch": 22.066420664206642,
      "grad_norm": 1.174944519996643,
      "learning_rate": 7.794351405052512e-05,
      "loss": 0.006693316996097565,
      "step": 77740
    },
    {
      "epoch": 22.069259154130002,
      "grad_norm": 1.5817304849624634,
      "learning_rate": 7.794067556060177e-05,
      "loss": 0.02046239227056503,
      "step": 77750
    },
    {
      "epoch": 22.072097644053365,
      "grad_norm": 8.418137550354004,
      "learning_rate": 7.793783707067839e-05,
      "loss": 0.027798721194267274,
      "step": 77760
    },
    {
      "epoch": 22.074936133976724,
      "grad_norm": 5.5687336921691895,
      "learning_rate": 7.793499858075505e-05,
      "loss": 0.008561762422323227,
      "step": 77770
    },
    {
      "epoch": 22.077774623900083,
      "grad_norm": 6.29403018951416,
      "learning_rate": 7.793216009083169e-05,
      "loss": 0.025265347957611085,
      "step": 77780
    },
    {
      "epoch": 22.080613113823446,
      "grad_norm": 0.5373438000679016,
      "learning_rate": 7.792932160090832e-05,
      "loss": 0.01051487699151039,
      "step": 77790
    },
    {
      "epoch": 22.083451603746806,
      "grad_norm": 1.3243863582611084,
      "learning_rate": 7.792648311098496e-05,
      "loss": 0.006049828231334686,
      "step": 77800
    },
    {
      "epoch": 22.08629009367017,
      "grad_norm": 1.5712735652923584,
      "learning_rate": 7.79236446210616e-05,
      "loss": 0.013195675611495972,
      "step": 77810
    },
    {
      "epoch": 22.089128583593528,
      "grad_norm": 0.260774165391922,
      "learning_rate": 7.792080613113824e-05,
      "loss": 0.015640179812908172,
      "step": 77820
    },
    {
      "epoch": 22.09196707351689,
      "grad_norm": 1.150002360343933,
      "learning_rate": 7.791796764121487e-05,
      "loss": 0.012659856677055359,
      "step": 77830
    },
    {
      "epoch": 22.09480556344025,
      "grad_norm": 12.027323722839355,
      "learning_rate": 7.791512915129152e-05,
      "loss": 0.011055758595466614,
      "step": 77840
    },
    {
      "epoch": 22.09764405336361,
      "grad_norm": 0.07149851322174072,
      "learning_rate": 7.791229066136817e-05,
      "loss": 0.009861954301595689,
      "step": 77850
    },
    {
      "epoch": 22.100482543286972,
      "grad_norm": 1.3277777433395386,
      "learning_rate": 7.79094521714448e-05,
      "loss": 0.0067562595009803775,
      "step": 77860
    },
    {
      "epoch": 22.10332103321033,
      "grad_norm": 0.8369110226631165,
      "learning_rate": 7.790661368152143e-05,
      "loss": 0.012772098183631897,
      "step": 77870
    },
    {
      "epoch": 22.106159523133694,
      "grad_norm": 7.399005889892578,
      "learning_rate": 7.790377519159808e-05,
      "loss": 0.00803334340453148,
      "step": 77880
    },
    {
      "epoch": 22.108998013057054,
      "grad_norm": 10.434040069580078,
      "learning_rate": 7.79009367016747e-05,
      "loss": 0.010350019484758378,
      "step": 77890
    },
    {
      "epoch": 22.111836502980413,
      "grad_norm": 4.601600646972656,
      "learning_rate": 7.789809821175135e-05,
      "loss": 0.012400143593549729,
      "step": 77900
    },
    {
      "epoch": 22.114674992903776,
      "grad_norm": 4.953150272369385,
      "learning_rate": 7.7895259721828e-05,
      "loss": 0.017364446818828583,
      "step": 77910
    },
    {
      "epoch": 22.117513482827135,
      "grad_norm": 1.8699755668640137,
      "learning_rate": 7.789242123190463e-05,
      "loss": 0.01287461370229721,
      "step": 77920
    },
    {
      "epoch": 22.1203519727505,
      "grad_norm": 6.1669230461120605,
      "learning_rate": 7.788958274198127e-05,
      "loss": 0.018391843140125274,
      "step": 77930
    },
    {
      "epoch": 22.123190462673858,
      "grad_norm": 1.3146318197250366,
      "learning_rate": 7.788674425205791e-05,
      "loss": 0.013735891878604889,
      "step": 77940
    },
    {
      "epoch": 22.126028952597217,
      "grad_norm": 6.031404972076416,
      "learning_rate": 7.788390576213455e-05,
      "loss": 0.016026775538921356,
      "step": 77950
    },
    {
      "epoch": 22.12886744252058,
      "grad_norm": 12.811196327209473,
      "learning_rate": 7.788106727221118e-05,
      "loss": 0.018257009983062743,
      "step": 77960
    },
    {
      "epoch": 22.13170593244394,
      "grad_norm": 11.070796012878418,
      "learning_rate": 7.787822878228783e-05,
      "loss": 0.017827069759368895,
      "step": 77970
    },
    {
      "epoch": 22.134544422367302,
      "grad_norm": 4.018449306488037,
      "learning_rate": 7.787539029236448e-05,
      "loss": 0.013120408356189727,
      "step": 77980
    },
    {
      "epoch": 22.13738291229066,
      "grad_norm": 3.4805305004119873,
      "learning_rate": 7.78725518024411e-05,
      "loss": 0.009723368287086486,
      "step": 77990
    },
    {
      "epoch": 22.14022140221402,
      "grad_norm": 3.1845874786376953,
      "learning_rate": 7.786971331251775e-05,
      "loss": 0.021074341237545015,
      "step": 78000
    },
    {
      "epoch": 22.14022140221402,
      "eval_accuracy": 0.9673809372416863,
      "eval_loss": 0.10670884698629379,
      "eval_runtime": 31.4212,
      "eval_samples_per_second": 500.523,
      "eval_steps_per_second": 7.829,
      "step": 78000
    },
    {
      "epoch": 22.143059892137384,
      "grad_norm": 5.378195762634277,
      "learning_rate": 7.786687482259439e-05,
      "loss": 0.011688557267189027,
      "step": 78010
    },
    {
      "epoch": 22.145898382060743,
      "grad_norm": 0.1714269071817398,
      "learning_rate": 7.786403633267101e-05,
      "loss": 0.006844799965620041,
      "step": 78020
    },
    {
      "epoch": 22.148736871984106,
      "grad_norm": 0.8463479280471802,
      "learning_rate": 7.786119784274766e-05,
      "loss": 0.007517029345035553,
      "step": 78030
    },
    {
      "epoch": 22.151575361907465,
      "grad_norm": 3.1646523475646973,
      "learning_rate": 7.785835935282431e-05,
      "loss": 0.011973054707050323,
      "step": 78040
    },
    {
      "epoch": 22.154413851830824,
      "grad_norm": 2.057929754257202,
      "learning_rate": 7.785552086290094e-05,
      "loss": 0.00851019024848938,
      "step": 78050
    },
    {
      "epoch": 22.157252341754187,
      "grad_norm": 0.6269627213478088,
      "learning_rate": 7.785268237297758e-05,
      "loss": 0.013988684117794036,
      "step": 78060
    },
    {
      "epoch": 22.160090831677547,
      "grad_norm": 0.35166695713996887,
      "learning_rate": 7.784984388305422e-05,
      "loss": 0.020649422705173493,
      "step": 78070
    },
    {
      "epoch": 22.16292932160091,
      "grad_norm": 2.507573127746582,
      "learning_rate": 7.784700539313086e-05,
      "loss": 0.009508591145277023,
      "step": 78080
    },
    {
      "epoch": 22.16576781152427,
      "grad_norm": 0.4796569347381592,
      "learning_rate": 7.784416690320749e-05,
      "loss": 0.013066285848617553,
      "step": 78090
    },
    {
      "epoch": 22.16860630144763,
      "grad_norm": 0.13566182553768158,
      "learning_rate": 7.784132841328413e-05,
      "loss": 0.022416481375694276,
      "step": 78100
    },
    {
      "epoch": 22.17144479137099,
      "grad_norm": 10.376028060913086,
      "learning_rate": 7.783848992336077e-05,
      "loss": 0.009210991114377976,
      "step": 78110
    },
    {
      "epoch": 22.17428328129435,
      "grad_norm": 5.886260032653809,
      "learning_rate": 7.783565143343741e-05,
      "loss": 0.012082375586032867,
      "step": 78120
    },
    {
      "epoch": 22.177121771217713,
      "grad_norm": 2.1102240085601807,
      "learning_rate": 7.783281294351406e-05,
      "loss": 0.011650773137807846,
      "step": 78130
    },
    {
      "epoch": 22.179960261141073,
      "grad_norm": 1.5243607759475708,
      "learning_rate": 7.78299744535907e-05,
      "loss": 0.0038802415132522584,
      "step": 78140
    },
    {
      "epoch": 22.182798751064432,
      "grad_norm": 6.211618423461914,
      "learning_rate": 7.782713596366733e-05,
      "loss": 0.008468639105558395,
      "step": 78150
    },
    {
      "epoch": 22.185637240987795,
      "grad_norm": 4.821821212768555,
      "learning_rate": 7.782429747374397e-05,
      "loss": 0.007791467010974884,
      "step": 78160
    },
    {
      "epoch": 22.188475730911154,
      "grad_norm": 0.13932007551193237,
      "learning_rate": 7.782145898382062e-05,
      "loss": 0.00805671289563179,
      "step": 78170
    },
    {
      "epoch": 22.191314220834517,
      "grad_norm": 0.13139668107032776,
      "learning_rate": 7.781862049389725e-05,
      "loss": 0.021581993997097017,
      "step": 78180
    },
    {
      "epoch": 22.194152710757876,
      "grad_norm": 4.98887825012207,
      "learning_rate": 7.781578200397389e-05,
      "loss": 0.009293796122074127,
      "step": 78190
    },
    {
      "epoch": 22.19699120068124,
      "grad_norm": 3.6963984966278076,
      "learning_rate": 7.781294351405053e-05,
      "loss": 0.010300829261541366,
      "step": 78200
    },
    {
      "epoch": 22.1998296906046,
      "grad_norm": 4.2909722328186035,
      "learning_rate": 7.781010502412716e-05,
      "loss": 0.005210285633802414,
      "step": 78210
    },
    {
      "epoch": 22.202668180527958,
      "grad_norm": 0.45989683270454407,
      "learning_rate": 7.78072665342038e-05,
      "loss": 0.01618835926055908,
      "step": 78220
    },
    {
      "epoch": 22.20550667045132,
      "grad_norm": 7.520296573638916,
      "learning_rate": 7.780442804428044e-05,
      "loss": 0.01901959478855133,
      "step": 78230
    },
    {
      "epoch": 22.20834516037468,
      "grad_norm": 0.5820671916007996,
      "learning_rate": 7.780158955435708e-05,
      "loss": 0.003843599557876587,
      "step": 78240
    },
    {
      "epoch": 22.211183650298043,
      "grad_norm": 0.19811807572841644,
      "learning_rate": 7.779875106443373e-05,
      "loss": 0.005912791192531586,
      "step": 78250
    },
    {
      "epoch": 22.214022140221402,
      "grad_norm": 3.562718391418457,
      "learning_rate": 7.779591257451037e-05,
      "loss": 0.016668358445167543,
      "step": 78260
    },
    {
      "epoch": 22.21686063014476,
      "grad_norm": 10.126795768737793,
      "learning_rate": 7.779307408458701e-05,
      "loss": 0.01387522667646408,
      "step": 78270
    },
    {
      "epoch": 22.219699120068125,
      "grad_norm": 3.5295963287353516,
      "learning_rate": 7.779023559466364e-05,
      "loss": 0.00795711725950241,
      "step": 78280
    },
    {
      "epoch": 22.222537609991484,
      "grad_norm": 6.15455961227417,
      "learning_rate": 7.778739710474028e-05,
      "loss": 0.01316492110490799,
      "step": 78290
    },
    {
      "epoch": 22.225376099914847,
      "grad_norm": 11.097175598144531,
      "learning_rate": 7.778455861481692e-05,
      "loss": 0.013008816540241242,
      "step": 78300
    },
    {
      "epoch": 22.228214589838206,
      "grad_norm": 12.18593692779541,
      "learning_rate": 7.778172012489356e-05,
      "loss": 0.013775190711021424,
      "step": 78310
    },
    {
      "epoch": 22.231053079761566,
      "grad_norm": 8.522257804870605,
      "learning_rate": 7.77788816349702e-05,
      "loss": 0.010103340446949004,
      "step": 78320
    },
    {
      "epoch": 22.23389156968493,
      "grad_norm": 2.110915422439575,
      "learning_rate": 7.777604314504684e-05,
      "loss": 0.012848296761512756,
      "step": 78330
    },
    {
      "epoch": 22.236730059608288,
      "grad_norm": 1.4709959030151367,
      "learning_rate": 7.777320465512347e-05,
      "loss": 0.018046993017196655,
      "step": 78340
    },
    {
      "epoch": 22.23956854953165,
      "grad_norm": 9.303194046020508,
      "learning_rate": 7.777036616520011e-05,
      "loss": 0.020788265764713286,
      "step": 78350
    },
    {
      "epoch": 22.24240703945501,
      "grad_norm": 14.899702072143555,
      "learning_rate": 7.776752767527675e-05,
      "loss": 0.023821789026260375,
      "step": 78360
    },
    {
      "epoch": 22.24524552937837,
      "grad_norm": 4.227825164794922,
      "learning_rate": 7.77646891853534e-05,
      "loss": 0.009640121459960937,
      "step": 78370
    },
    {
      "epoch": 22.248084019301732,
      "grad_norm": 1.8510937690734863,
      "learning_rate": 7.776185069543004e-05,
      "loss": 0.010585528612136842,
      "step": 78380
    },
    {
      "epoch": 22.25092250922509,
      "grad_norm": 0.7175607681274414,
      "learning_rate": 7.775901220550668e-05,
      "loss": 0.008308710902929306,
      "step": 78390
    },
    {
      "epoch": 22.253760999148454,
      "grad_norm": 5.19725227355957,
      "learning_rate": 7.775617371558332e-05,
      "loss": 0.012205660343170166,
      "step": 78400
    },
    {
      "epoch": 22.256599489071814,
      "grad_norm": 0.04399823397397995,
      "learning_rate": 7.775333522565995e-05,
      "loss": 0.007741999626159668,
      "step": 78410
    },
    {
      "epoch": 22.259437978995173,
      "grad_norm": 2.996161460876465,
      "learning_rate": 7.775049673573659e-05,
      "loss": 0.024648337066173552,
      "step": 78420
    },
    {
      "epoch": 22.262276468918536,
      "grad_norm": 8.189443588256836,
      "learning_rate": 7.774765824581323e-05,
      "loss": 0.007954753190279006,
      "step": 78430
    },
    {
      "epoch": 22.265114958841895,
      "grad_norm": 0.22202037274837494,
      "learning_rate": 7.774481975588987e-05,
      "loss": 0.018301871418952943,
      "step": 78440
    },
    {
      "epoch": 22.267953448765258,
      "grad_norm": 1.2551482915878296,
      "learning_rate": 7.774198126596651e-05,
      "loss": 0.009818053245544434,
      "step": 78450
    },
    {
      "epoch": 22.270791938688618,
      "grad_norm": 0.8292374610900879,
      "learning_rate": 7.773914277604315e-05,
      "loss": 0.01662033349275589,
      "step": 78460
    },
    {
      "epoch": 22.273630428611977,
      "grad_norm": 0.10649089515209198,
      "learning_rate": 7.773630428611978e-05,
      "loss": 0.009980617463588715,
      "step": 78470
    },
    {
      "epoch": 22.27646891853534,
      "grad_norm": 5.147066593170166,
      "learning_rate": 7.773346579619642e-05,
      "loss": 0.006492535769939423,
      "step": 78480
    },
    {
      "epoch": 22.2793074084587,
      "grad_norm": 5.389790058135986,
      "learning_rate": 7.773062730627306e-05,
      "loss": 0.010049711167812347,
      "step": 78490
    },
    {
      "epoch": 22.282145898382062,
      "grad_norm": 2.226016044616699,
      "learning_rate": 7.77277888163497e-05,
      "loss": 0.005699679255485535,
      "step": 78500
    },
    {
      "epoch": 22.282145898382062,
      "eval_accuracy": 0.9643288611941248,
      "eval_loss": 0.11839230358600616,
      "eval_runtime": 31.8517,
      "eval_samples_per_second": 493.757,
      "eval_steps_per_second": 7.723,
      "step": 78500
    },
    {
      "epoch": 22.28498438830542,
      "grad_norm": 2.348543643951416,
      "learning_rate": 7.772495032642635e-05,
      "loss": 0.015471228957176208,
      "step": 78510
    },
    {
      "epoch": 22.28782287822878,
      "grad_norm": 3.2751684188842773,
      "learning_rate": 7.772211183650299e-05,
      "loss": 0.007494095712900162,
      "step": 78520
    },
    {
      "epoch": 22.290661368152143,
      "grad_norm": 5.831447601318359,
      "learning_rate": 7.771927334657963e-05,
      "loss": 0.03796030282974243,
      "step": 78530
    },
    {
      "epoch": 22.293499858075503,
      "grad_norm": 10.852279663085938,
      "learning_rate": 7.771643485665626e-05,
      "loss": 0.024220287799835205,
      "step": 78540
    },
    {
      "epoch": 22.296338347998866,
      "grad_norm": 3.8469982147216797,
      "learning_rate": 7.77135963667329e-05,
      "loss": 0.023365892469882965,
      "step": 78550
    },
    {
      "epoch": 22.299176837922225,
      "grad_norm": 2.7278900146484375,
      "learning_rate": 7.771075787680954e-05,
      "loss": 0.025482177734375,
      "step": 78560
    },
    {
      "epoch": 22.302015327845588,
      "grad_norm": 4.788492202758789,
      "learning_rate": 7.770791938688618e-05,
      "loss": 0.01673136353492737,
      "step": 78570
    },
    {
      "epoch": 22.304853817768947,
      "grad_norm": 0.524034321308136,
      "learning_rate": 7.770508089696282e-05,
      "loss": 0.013599051535129547,
      "step": 78580
    },
    {
      "epoch": 22.307692307692307,
      "grad_norm": 2.589271068572998,
      "learning_rate": 7.770224240703946e-05,
      "loss": 0.020609650015830993,
      "step": 78590
    },
    {
      "epoch": 22.31053079761567,
      "grad_norm": 0.9688955545425415,
      "learning_rate": 7.769940391711609e-05,
      "loss": 0.014199270308017731,
      "step": 78600
    },
    {
      "epoch": 22.31336928753903,
      "grad_norm": 7.938648223876953,
      "learning_rate": 7.769656542719273e-05,
      "loss": 0.009601452201604844,
      "step": 78610
    },
    {
      "epoch": 22.31620777746239,
      "grad_norm": 11.757878303527832,
      "learning_rate": 7.769372693726938e-05,
      "loss": 0.028433749079704286,
      "step": 78620
    },
    {
      "epoch": 22.31904626738575,
      "grad_norm": 15.40138053894043,
      "learning_rate": 7.769088844734602e-05,
      "loss": 0.03010331988334656,
      "step": 78630
    },
    {
      "epoch": 22.32188475730911,
      "grad_norm": 0.9149933457374573,
      "learning_rate": 7.768804995742266e-05,
      "loss": 0.014781475067138672,
      "step": 78640
    },
    {
      "epoch": 22.324723247232473,
      "grad_norm": 1.3284035921096802,
      "learning_rate": 7.76852114674993e-05,
      "loss": 0.013837674260139465,
      "step": 78650
    },
    {
      "epoch": 22.327561737155833,
      "grad_norm": 5.499685764312744,
      "learning_rate": 7.768237297757594e-05,
      "loss": 0.010042565315961838,
      "step": 78660
    },
    {
      "epoch": 22.330400227079195,
      "grad_norm": 4.1156744956970215,
      "learning_rate": 7.767953448765257e-05,
      "loss": 0.005651995167136192,
      "step": 78670
    },
    {
      "epoch": 22.333238717002555,
      "grad_norm": 9.972570419311523,
      "learning_rate": 7.767669599772921e-05,
      "loss": 0.010193221271038055,
      "step": 78680
    },
    {
      "epoch": 22.336077206925914,
      "grad_norm": 12.723579406738281,
      "learning_rate": 7.767385750780585e-05,
      "loss": 0.016563312709331514,
      "step": 78690
    },
    {
      "epoch": 22.338915696849277,
      "grad_norm": 8.280366897583008,
      "learning_rate": 7.767101901788248e-05,
      "loss": 0.011286767572164536,
      "step": 78700
    },
    {
      "epoch": 22.341754186772636,
      "grad_norm": 0.6984289288520813,
      "learning_rate": 7.766818052795913e-05,
      "loss": 0.007436839491128921,
      "step": 78710
    },
    {
      "epoch": 22.344592676696,
      "grad_norm": 3.3087446689605713,
      "learning_rate": 7.766534203803578e-05,
      "loss": 0.006595052033662796,
      "step": 78720
    },
    {
      "epoch": 22.34743116661936,
      "grad_norm": 10.459646224975586,
      "learning_rate": 7.76625035481124e-05,
      "loss": 0.028175845742225647,
      "step": 78730
    },
    {
      "epoch": 22.350269656542718,
      "grad_norm": 9.923874855041504,
      "learning_rate": 7.765966505818904e-05,
      "loss": 0.008939434587955476,
      "step": 78740
    },
    {
      "epoch": 22.35310814646608,
      "grad_norm": 1.7071095705032349,
      "learning_rate": 7.765682656826569e-05,
      "loss": 0.009886356443166733,
      "step": 78750
    },
    {
      "epoch": 22.35594663638944,
      "grad_norm": 0.7185766100883484,
      "learning_rate": 7.765398807834233e-05,
      "loss": 0.005316883325576782,
      "step": 78760
    },
    {
      "epoch": 22.358785126312803,
      "grad_norm": 12.837071418762207,
      "learning_rate": 7.765114958841897e-05,
      "loss": 0.022282302379608154,
      "step": 78770
    },
    {
      "epoch": 22.361623616236162,
      "grad_norm": 2.1667263507843018,
      "learning_rate": 7.764831109849561e-05,
      "loss": 0.011915234476327896,
      "step": 78780
    },
    {
      "epoch": 22.36446210615952,
      "grad_norm": 22.122661590576172,
      "learning_rate": 7.764547260857225e-05,
      "loss": 0.031076237559318542,
      "step": 78790
    },
    {
      "epoch": 22.367300596082885,
      "grad_norm": 0.27172213792800903,
      "learning_rate": 7.764291796764122e-05,
      "loss": 0.016910012066364288,
      "step": 78800
    },
    {
      "epoch": 22.370139086006244,
      "grad_norm": 0.5888198614120483,
      "learning_rate": 7.764007947771786e-05,
      "loss": 0.013816308975219727,
      "step": 78810
    },
    {
      "epoch": 22.372977575929607,
      "grad_norm": 10.406205177307129,
      "learning_rate": 7.76372409877945e-05,
      "loss": 0.023664560914039613,
      "step": 78820
    },
    {
      "epoch": 22.375816065852966,
      "grad_norm": 2.306295394897461,
      "learning_rate": 7.763440249787114e-05,
      "loss": 0.01593463122844696,
      "step": 78830
    },
    {
      "epoch": 22.378654555776325,
      "grad_norm": 9.898871421813965,
      "learning_rate": 7.763156400794778e-05,
      "loss": 0.012129470705986023,
      "step": 78840
    },
    {
      "epoch": 22.38149304569969,
      "grad_norm": 3.3095803260803223,
      "learning_rate": 7.762872551802441e-05,
      "loss": 0.007985438406467437,
      "step": 78850
    },
    {
      "epoch": 22.384331535623048,
      "grad_norm": 3.818220615386963,
      "learning_rate": 7.762588702810105e-05,
      "loss": 0.016553662717342377,
      "step": 78860
    },
    {
      "epoch": 22.38717002554641,
      "grad_norm": 0.08684609085321426,
      "learning_rate": 7.76230485381777e-05,
      "loss": 0.005004122480750084,
      "step": 78870
    },
    {
      "epoch": 22.39000851546977,
      "grad_norm": 8.991494178771973,
      "learning_rate": 7.762021004825432e-05,
      "loss": 0.012231387197971344,
      "step": 78880
    },
    {
      "epoch": 22.39284700539313,
      "grad_norm": 2.045503854751587,
      "learning_rate": 7.761737155833098e-05,
      "loss": 0.013507908582687378,
      "step": 78890
    },
    {
      "epoch": 22.395685495316492,
      "grad_norm": 5.527192115783691,
      "learning_rate": 7.761453306840762e-05,
      "loss": 0.027555057406425477,
      "step": 78900
    },
    {
      "epoch": 22.39852398523985,
      "grad_norm": 2.264983892440796,
      "learning_rate": 7.761169457848425e-05,
      "loss": 0.01494077742099762,
      "step": 78910
    },
    {
      "epoch": 22.401362475163214,
      "grad_norm": 10.577933311462402,
      "learning_rate": 7.760885608856089e-05,
      "loss": 0.016612935066223144,
      "step": 78920
    },
    {
      "epoch": 22.404200965086574,
      "grad_norm": 11.011502265930176,
      "learning_rate": 7.760601759863753e-05,
      "loss": 0.011415421217679977,
      "step": 78930
    },
    {
      "epoch": 22.407039455009933,
      "grad_norm": 1.5920131206512451,
      "learning_rate": 7.760317910871417e-05,
      "loss": 0.013965202867984772,
      "step": 78940
    },
    {
      "epoch": 22.409877944933296,
      "grad_norm": 2.83778977394104,
      "learning_rate": 7.760034061879081e-05,
      "loss": 0.011010801792144776,
      "step": 78950
    },
    {
      "epoch": 22.412716434856655,
      "grad_norm": 1.3154271841049194,
      "learning_rate": 7.759750212886745e-05,
      "loss": 0.010433544963598251,
      "step": 78960
    },
    {
      "epoch": 22.415554924780018,
      "grad_norm": 1.393776774406433,
      "learning_rate": 7.759466363894408e-05,
      "loss": 0.007394035905599594,
      "step": 78970
    },
    {
      "epoch": 22.418393414703377,
      "grad_norm": 2.478628158569336,
      "learning_rate": 7.759182514902072e-05,
      "loss": 0.012498383969068527,
      "step": 78980
    },
    {
      "epoch": 22.42123190462674,
      "grad_norm": 0.6744087338447571,
      "learning_rate": 7.758898665909736e-05,
      "loss": 0.007888572663068772,
      "step": 78990
    },
    {
      "epoch": 22.4240703945501,
      "grad_norm": 1.3194026947021484,
      "learning_rate": 7.7586148169174e-05,
      "loss": 0.020094528794288635,
      "step": 79000
    },
    {
      "epoch": 22.4240703945501,
      "eval_accuracy": 0.9644560310294398,
      "eval_loss": 0.11648497730493546,
      "eval_runtime": 32.4787,
      "eval_samples_per_second": 484.225,
      "eval_steps_per_second": 7.574,
      "step": 79000
    },
    {
      "epoch": 22.42690888447346,
      "grad_norm": 0.1950353980064392,
      "learning_rate": 7.758330967925063e-05,
      "loss": 0.005496396869421005,
      "step": 79010
    },
    {
      "epoch": 22.429747374396822,
      "grad_norm": 8.227417945861816,
      "learning_rate": 7.758047118932729e-05,
      "loss": 0.011399309337139129,
      "step": 79020
    },
    {
      "epoch": 22.43258586432018,
      "grad_norm": 1.3837053775787354,
      "learning_rate": 7.757763269940393e-05,
      "loss": 0.00926767811179161,
      "step": 79030
    },
    {
      "epoch": 22.435424354243544,
      "grad_norm": 5.21128511428833,
      "learning_rate": 7.757479420948056e-05,
      "loss": 0.021765249967575073,
      "step": 79040
    },
    {
      "epoch": 22.438262844166903,
      "grad_norm": 0.8473485112190247,
      "learning_rate": 7.75719557195572e-05,
      "loss": 0.024957311153411866,
      "step": 79050
    },
    {
      "epoch": 22.441101334090263,
      "grad_norm": 1.3490263223648071,
      "learning_rate": 7.756911722963384e-05,
      "loss": 0.009740283340215683,
      "step": 79060
    },
    {
      "epoch": 22.443939824013626,
      "grad_norm": 1.744531512260437,
      "learning_rate": 7.756627873971047e-05,
      "loss": 0.006131973490118981,
      "step": 79070
    },
    {
      "epoch": 22.446778313936985,
      "grad_norm": 0.1797332912683487,
      "learning_rate": 7.756344024978711e-05,
      "loss": 0.010336843132972718,
      "step": 79080
    },
    {
      "epoch": 22.449616803860348,
      "grad_norm": 1.2307103872299194,
      "learning_rate": 7.756060175986376e-05,
      "loss": 0.005235159397125244,
      "step": 79090
    },
    {
      "epoch": 22.452455293783707,
      "grad_norm": 2.4683594703674316,
      "learning_rate": 7.755776326994039e-05,
      "loss": 0.008046117424964905,
      "step": 79100
    },
    {
      "epoch": 22.455293783707067,
      "grad_norm": 1.6268317699432373,
      "learning_rate": 7.755492478001703e-05,
      "loss": 0.012852539122104645,
      "step": 79110
    },
    {
      "epoch": 22.45813227363043,
      "grad_norm": 4.865121841430664,
      "learning_rate": 7.755208629009367e-05,
      "loss": 0.011758530884981156,
      "step": 79120
    },
    {
      "epoch": 22.46097076355379,
      "grad_norm": 0.6189286112785339,
      "learning_rate": 7.754924780017032e-05,
      "loss": 0.01459445059299469,
      "step": 79130
    },
    {
      "epoch": 22.46380925347715,
      "grad_norm": 0.846178412437439,
      "learning_rate": 7.754640931024694e-05,
      "loss": 0.019893331825733183,
      "step": 79140
    },
    {
      "epoch": 22.46664774340051,
      "grad_norm": 6.491055011749268,
      "learning_rate": 7.75435708203236e-05,
      "loss": 0.007758082449436187,
      "step": 79150
    },
    {
      "epoch": 22.46948623332387,
      "grad_norm": 1.9458593130111694,
      "learning_rate": 7.754073233040024e-05,
      "loss": 0.01711057424545288,
      "step": 79160
    },
    {
      "epoch": 22.472324723247233,
      "grad_norm": 4.260765552520752,
      "learning_rate": 7.753789384047687e-05,
      "loss": 0.012027893960475922,
      "step": 79170
    },
    {
      "epoch": 22.475163213170593,
      "grad_norm": 2.313204526901245,
      "learning_rate": 7.753505535055351e-05,
      "loss": 0.0062425538897514345,
      "step": 79180
    },
    {
      "epoch": 22.478001703093955,
      "grad_norm": 0.30379486083984375,
      "learning_rate": 7.753221686063015e-05,
      "loss": 0.012911394238471985,
      "step": 79190
    },
    {
      "epoch": 22.480840193017315,
      "grad_norm": 3.386138439178467,
      "learning_rate": 7.752937837070678e-05,
      "loss": 0.007911705225706101,
      "step": 79200
    },
    {
      "epoch": 22.483678682940674,
      "grad_norm": 4.456545829772949,
      "learning_rate": 7.752653988078342e-05,
      "loss": 0.012407016754150391,
      "step": 79210
    },
    {
      "epoch": 22.486517172864037,
      "grad_norm": 0.7037520408630371,
      "learning_rate": 7.752370139086007e-05,
      "loss": 0.014980792999267578,
      "step": 79220
    },
    {
      "epoch": 22.489355662787396,
      "grad_norm": 3.615569829940796,
      "learning_rate": 7.75208629009367e-05,
      "loss": 0.013668224215507507,
      "step": 79230
    },
    {
      "epoch": 22.49219415271076,
      "grad_norm": 20.066904067993164,
      "learning_rate": 7.751802441101334e-05,
      "loss": 0.020690937340259553,
      "step": 79240
    },
    {
      "epoch": 22.49503264263412,
      "grad_norm": 5.913091659545898,
      "learning_rate": 7.751518592108998e-05,
      "loss": 0.012933464348316192,
      "step": 79250
    },
    {
      "epoch": 22.497871132557478,
      "grad_norm": 2.0230982303619385,
      "learning_rate": 7.751234743116663e-05,
      "loss": 0.013022205233573914,
      "step": 79260
    },
    {
      "epoch": 22.50070962248084,
      "grad_norm": 4.580300807952881,
      "learning_rate": 7.750950894124325e-05,
      "loss": 0.01799226552248001,
      "step": 79270
    },
    {
      "epoch": 22.5035481124042,
      "grad_norm": 2.7023584842681885,
      "learning_rate": 7.75066704513199e-05,
      "loss": 0.017762008309364318,
      "step": 79280
    },
    {
      "epoch": 22.506386602327563,
      "grad_norm": 2.3168201446533203,
      "learning_rate": 7.750383196139655e-05,
      "loss": 0.010609734058380126,
      "step": 79290
    },
    {
      "epoch": 22.509225092250922,
      "grad_norm": 0.5185960531234741,
      "learning_rate": 7.750099347147318e-05,
      "loss": 0.01253051608800888,
      "step": 79300
    },
    {
      "epoch": 22.51206358217428,
      "grad_norm": 0.44416698813438416,
      "learning_rate": 7.749815498154982e-05,
      "loss": 0.01206279695034027,
      "step": 79310
    },
    {
      "epoch": 22.514902072097644,
      "grad_norm": 8.609683990478516,
      "learning_rate": 7.749531649162646e-05,
      "loss": 0.009859508275985718,
      "step": 79320
    },
    {
      "epoch": 22.517740562021004,
      "grad_norm": 3.055227041244507,
      "learning_rate": 7.749247800170309e-05,
      "loss": 0.010765496641397476,
      "step": 79330
    },
    {
      "epoch": 22.520579051944367,
      "grad_norm": 11.13996696472168,
      "learning_rate": 7.748963951177973e-05,
      "loss": 0.01065712422132492,
      "step": 79340
    },
    {
      "epoch": 22.523417541867726,
      "grad_norm": 10.12613296508789,
      "learning_rate": 7.748680102185639e-05,
      "loss": 0.011258310079574585,
      "step": 79350
    },
    {
      "epoch": 22.526256031791085,
      "grad_norm": 0.39234694838523865,
      "learning_rate": 7.748396253193301e-05,
      "loss": 0.026453298330307008,
      "step": 79360
    },
    {
      "epoch": 22.52909452171445,
      "grad_norm": 4.60313606262207,
      "learning_rate": 7.748112404200965e-05,
      "loss": 0.008643917739391327,
      "step": 79370
    },
    {
      "epoch": 22.531933011637808,
      "grad_norm": 1.3526878356933594,
      "learning_rate": 7.74782855520863e-05,
      "loss": 0.012584066390991211,
      "step": 79380
    },
    {
      "epoch": 22.53477150156117,
      "grad_norm": 6.417731761932373,
      "learning_rate": 7.747544706216294e-05,
      "loss": 0.03642912209033966,
      "step": 79390
    },
    {
      "epoch": 22.53760999148453,
      "grad_norm": 7.170996189117432,
      "learning_rate": 7.747260857223956e-05,
      "loss": 0.007970795780420304,
      "step": 79400
    },
    {
      "epoch": 22.540448481407893,
      "grad_norm": 1.17007577419281,
      "learning_rate": 7.74697700823162e-05,
      "loss": 0.012048873305320739,
      "step": 79410
    },
    {
      "epoch": 22.543286971331252,
      "grad_norm": 4.373727798461914,
      "learning_rate": 7.746693159239286e-05,
      "loss": 0.010239073634147644,
      "step": 79420
    },
    {
      "epoch": 22.54612546125461,
      "grad_norm": 3.7238335609436035,
      "learning_rate": 7.746409310246949e-05,
      "loss": 0.010963058471679688,
      "step": 79430
    },
    {
      "epoch": 22.548963951177974,
      "grad_norm": 1.5594240427017212,
      "learning_rate": 7.746125461254613e-05,
      "loss": 0.013627929985523224,
      "step": 79440
    },
    {
      "epoch": 22.551802441101334,
      "grad_norm": 0.36623379588127136,
      "learning_rate": 7.745841612262277e-05,
      "loss": 0.01595919281244278,
      "step": 79450
    },
    {
      "epoch": 22.554640931024696,
      "grad_norm": 9.257342338562012,
      "learning_rate": 7.74555776326994e-05,
      "loss": 0.010426081717014313,
      "step": 79460
    },
    {
      "epoch": 22.557479420948056,
      "grad_norm": 2.940206289291382,
      "learning_rate": 7.745273914277604e-05,
      "loss": 0.011334303766489029,
      "step": 79470
    },
    {
      "epoch": 22.560317910871415,
      "grad_norm": 0.15435819327831268,
      "learning_rate": 7.74499006528527e-05,
      "loss": 0.02565968930721283,
      "step": 79480
    },
    {
      "epoch": 22.563156400794778,
      "grad_norm": 1.4230142831802368,
      "learning_rate": 7.744706216292932e-05,
      "loss": 0.013086535036563873,
      "step": 79490
    },
    {
      "epoch": 22.565994890718137,
      "grad_norm": 0.9359275698661804,
      "learning_rate": 7.744422367300597e-05,
      "loss": 0.04905348718166351,
      "step": 79500
    },
    {
      "epoch": 22.565994890718137,
      "eval_accuracy": 0.9649647103707001,
      "eval_loss": 0.11553753167390823,
      "eval_runtime": 31.8675,
      "eval_samples_per_second": 493.512,
      "eval_steps_per_second": 7.719,
      "step": 79500
    },
    {
      "epoch": 22.5688333806415,
      "grad_norm": 6.085573196411133,
      "learning_rate": 7.74413851830826e-05,
      "loss": 0.026369887590408325,
      "step": 79510
    },
    {
      "epoch": 22.57167187056486,
      "grad_norm": 5.694607257843018,
      "learning_rate": 7.743854669315925e-05,
      "loss": 0.0267105758190155,
      "step": 79520
    },
    {
      "epoch": 22.57451036048822,
      "grad_norm": 12.553425788879395,
      "learning_rate": 7.743570820323588e-05,
      "loss": 0.021405410766601563,
      "step": 79530
    },
    {
      "epoch": 22.577348850411582,
      "grad_norm": 1.0821653604507446,
      "learning_rate": 7.743286971331252e-05,
      "loss": 0.009584519267082214,
      "step": 79540
    },
    {
      "epoch": 22.58018734033494,
      "grad_norm": 0.2032950222492218,
      "learning_rate": 7.743003122338917e-05,
      "loss": 0.012168647348880767,
      "step": 79550
    },
    {
      "epoch": 22.583025830258304,
      "grad_norm": 2.5928611755371094,
      "learning_rate": 7.74271927334658e-05,
      "loss": 0.01846584677696228,
      "step": 79560
    },
    {
      "epoch": 22.585864320181663,
      "grad_norm": 1.6427046060562134,
      "learning_rate": 7.742435424354244e-05,
      "loss": 0.0032516315579414366,
      "step": 79570
    },
    {
      "epoch": 22.588702810105023,
      "grad_norm": 1.98064124584198,
      "learning_rate": 7.742151575361908e-05,
      "loss": 0.0148907870054245,
      "step": 79580
    },
    {
      "epoch": 22.591541300028386,
      "grad_norm": 0.0847540870308876,
      "learning_rate": 7.741867726369571e-05,
      "loss": 0.008065237104892731,
      "step": 79590
    },
    {
      "epoch": 22.594379789951745,
      "grad_norm": 2.2634873390197754,
      "learning_rate": 7.741583877377235e-05,
      "loss": 0.020466260612010956,
      "step": 79600
    },
    {
      "epoch": 22.597218279875108,
      "grad_norm": 7.790076732635498,
      "learning_rate": 7.741300028384899e-05,
      "loss": 0.011237717419862746,
      "step": 79610
    },
    {
      "epoch": 22.600056769798467,
      "grad_norm": 6.0409255027771,
      "learning_rate": 7.741016179392563e-05,
      "loss": 0.012408192455768585,
      "step": 79620
    },
    {
      "epoch": 22.602895259721826,
      "grad_norm": 3.8379149436950684,
      "learning_rate": 7.740732330400228e-05,
      "loss": 0.010703689604997634,
      "step": 79630
    },
    {
      "epoch": 22.60573374964519,
      "grad_norm": 0.6396515369415283,
      "learning_rate": 7.740448481407892e-05,
      "loss": 0.012553629279136658,
      "step": 79640
    },
    {
      "epoch": 22.60857223956855,
      "grad_norm": 2.2229232788085938,
      "learning_rate": 7.740164632415556e-05,
      "loss": 0.015364086627960205,
      "step": 79650
    },
    {
      "epoch": 22.61141072949191,
      "grad_norm": 8.918983459472656,
      "learning_rate": 7.739880783423219e-05,
      "loss": 0.01649310886859894,
      "step": 79660
    },
    {
      "epoch": 22.61424921941527,
      "grad_norm": 0.3912106454372406,
      "learning_rate": 7.739596934430883e-05,
      "loss": 0.005555889755487442,
      "step": 79670
    },
    {
      "epoch": 22.61708770933863,
      "grad_norm": 1.362387776374817,
      "learning_rate": 7.739313085438548e-05,
      "loss": 0.005939261242747307,
      "step": 79680
    },
    {
      "epoch": 22.619926199261993,
      "grad_norm": 1.3112120628356934,
      "learning_rate": 7.739029236446211e-05,
      "loss": 0.009702064096927643,
      "step": 79690
    },
    {
      "epoch": 22.622764689185352,
      "grad_norm": 2.9736695289611816,
      "learning_rate": 7.738745387453875e-05,
      "loss": 0.021935120224952698,
      "step": 79700
    },
    {
      "epoch": 22.625603179108715,
      "grad_norm": 8.930765151977539,
      "learning_rate": 7.73846153846154e-05,
      "loss": 0.01166784092783928,
      "step": 79710
    },
    {
      "epoch": 22.628441669032075,
      "grad_norm": 0.084261953830719,
      "learning_rate": 7.738177689469202e-05,
      "loss": 0.020249782502651213,
      "step": 79720
    },
    {
      "epoch": 22.631280158955434,
      "grad_norm": 7.59883975982666,
      "learning_rate": 7.737893840476866e-05,
      "loss": 0.011867715418338776,
      "step": 79730
    },
    {
      "epoch": 22.634118648878797,
      "grad_norm": 9.584271430969238,
      "learning_rate": 7.73760999148453e-05,
      "loss": 0.01911238580942154,
      "step": 79740
    },
    {
      "epoch": 22.636957138802156,
      "grad_norm": 2.8228602409362793,
      "learning_rate": 7.737326142492195e-05,
      "loss": 0.015226860344409943,
      "step": 79750
    },
    {
      "epoch": 22.63979562872552,
      "grad_norm": 1.3170843124389648,
      "learning_rate": 7.737042293499859e-05,
      "loss": 0.0049213763326406475,
      "step": 79760
    },
    {
      "epoch": 22.64263411864888,
      "grad_norm": 0.6893202066421509,
      "learning_rate": 7.736758444507523e-05,
      "loss": 0.005104249715805054,
      "step": 79770
    },
    {
      "epoch": 22.64547260857224,
      "grad_norm": 4.823670387268066,
      "learning_rate": 7.736474595515187e-05,
      "loss": 0.004693111777305603,
      "step": 79780
    },
    {
      "epoch": 22.6483110984956,
      "grad_norm": 11.441197395324707,
      "learning_rate": 7.73619074652285e-05,
      "loss": 0.012336860597133636,
      "step": 79790
    },
    {
      "epoch": 22.65114958841896,
      "grad_norm": 1.8496366739273071,
      "learning_rate": 7.735906897530514e-05,
      "loss": 0.004601847380399704,
      "step": 79800
    },
    {
      "epoch": 22.653988078342323,
      "grad_norm": 6.232671737670898,
      "learning_rate": 7.735623048538178e-05,
      "loss": 0.013932222127914428,
      "step": 79810
    },
    {
      "epoch": 22.656826568265682,
      "grad_norm": 0.7702204585075378,
      "learning_rate": 7.735339199545842e-05,
      "loss": 0.0047489553689956665,
      "step": 79820
    },
    {
      "epoch": 22.659665058189045,
      "grad_norm": 1.2361254692077637,
      "learning_rate": 7.735055350553506e-05,
      "loss": 0.007621980458498001,
      "step": 79830
    },
    {
      "epoch": 22.662503548112404,
      "grad_norm": 2.8007898330688477,
      "learning_rate": 7.73477150156117e-05,
      "loss": 0.023316282033920287,
      "step": 79840
    },
    {
      "epoch": 22.665342038035764,
      "grad_norm": 6.725745677947998,
      "learning_rate": 7.734487652568833e-05,
      "loss": 0.024811407923698424,
      "step": 79850
    },
    {
      "epoch": 22.668180527959127,
      "grad_norm": 0.1743684709072113,
      "learning_rate": 7.734203803576497e-05,
      "loss": 0.009672833979129792,
      "step": 79860
    },
    {
      "epoch": 22.671019017882486,
      "grad_norm": 3.275254011154175,
      "learning_rate": 7.733919954584161e-05,
      "loss": 0.02893396019935608,
      "step": 79870
    },
    {
      "epoch": 22.67385750780585,
      "grad_norm": 9.358637809753418,
      "learning_rate": 7.733636105591826e-05,
      "loss": 0.02172216922044754,
      "step": 79880
    },
    {
      "epoch": 22.676695997729208,
      "grad_norm": 7.674645900726318,
      "learning_rate": 7.73335225659949e-05,
      "loss": 0.008159933984279633,
      "step": 79890
    },
    {
      "epoch": 22.679534487652568,
      "grad_norm": 10.914156913757324,
      "learning_rate": 7.733068407607154e-05,
      "loss": 0.021128436923027037,
      "step": 79900
    },
    {
      "epoch": 22.68237297757593,
      "grad_norm": 0.7474072575569153,
      "learning_rate": 7.732784558614817e-05,
      "loss": 0.018513143062591553,
      "step": 79910
    },
    {
      "epoch": 22.68521146749929,
      "grad_norm": 13.689690589904785,
      "learning_rate": 7.732500709622481e-05,
      "loss": 0.01652028560638428,
      "step": 79920
    },
    {
      "epoch": 22.688049957422653,
      "grad_norm": 4.518795490264893,
      "learning_rate": 7.732216860630145e-05,
      "loss": 0.009361089020967484,
      "step": 79930
    },
    {
      "epoch": 22.690888447346012,
      "grad_norm": 16.592924118041992,
      "learning_rate": 7.731933011637809e-05,
      "loss": 0.014007166028022766,
      "step": 79940
    },
    {
      "epoch": 22.69372693726937,
      "grad_norm": 1.0795841217041016,
      "learning_rate": 7.731649162645473e-05,
      "loss": 0.00787409469485283,
      "step": 79950
    },
    {
      "epoch": 22.696565427192734,
      "grad_norm": 1.2775568962097168,
      "learning_rate": 7.731365313653137e-05,
      "loss": 0.007690049707889557,
      "step": 79960
    },
    {
      "epoch": 22.699403917116094,
      "grad_norm": 2.7688798904418945,
      "learning_rate": 7.731081464660801e-05,
      "loss": 0.03022007644176483,
      "step": 79970
    },
    {
      "epoch": 22.702242407039456,
      "grad_norm": 0.8021807074546814,
      "learning_rate": 7.730797615668464e-05,
      "loss": 0.016617977619171144,
      "step": 79980
    },
    {
      "epoch": 22.705080896962816,
      "grad_norm": 7.414029121398926,
      "learning_rate": 7.730513766676128e-05,
      "loss": 0.011338051408529282,
      "step": 79990
    },
    {
      "epoch": 22.707919386886175,
      "grad_norm": 9.380739212036133,
      "learning_rate": 7.730229917683793e-05,
      "loss": 0.009740841388702393,
      "step": 80000
    },
    {
      "epoch": 22.707919386886175,
      "eval_accuracy": 0.9673173523240287,
      "eval_loss": 0.10911644995212555,
      "eval_runtime": 31.7886,
      "eval_samples_per_second": 494.737,
      "eval_steps_per_second": 7.739,
      "step": 80000
    },
    {
      "epoch": 22.710757876809538,
      "grad_norm": 3.3060810565948486,
      "learning_rate": 7.729946068691455e-05,
      "loss": 0.02486606240272522,
      "step": 80010
    },
    {
      "epoch": 22.713596366732897,
      "grad_norm": 3.00077486038208,
      "learning_rate": 7.729662219699121e-05,
      "loss": 0.011626887321472167,
      "step": 80020
    },
    {
      "epoch": 22.71643485665626,
      "grad_norm": 9.370033264160156,
      "learning_rate": 7.729378370706785e-05,
      "loss": 0.018975050747394563,
      "step": 80030
    },
    {
      "epoch": 22.71927334657962,
      "grad_norm": 1.828291893005371,
      "learning_rate": 7.729094521714448e-05,
      "loss": 0.012643803656101228,
      "step": 80040
    },
    {
      "epoch": 22.72211183650298,
      "grad_norm": 2.541023015975952,
      "learning_rate": 7.728810672722112e-05,
      "loss": 0.010191821306943894,
      "step": 80050
    },
    {
      "epoch": 22.72495032642634,
      "grad_norm": 1.8205264806747437,
      "learning_rate": 7.728526823729776e-05,
      "loss": 0.012982244789600372,
      "step": 80060
    },
    {
      "epoch": 22.7277888163497,
      "grad_norm": 3.614253520965576,
      "learning_rate": 7.72824297473744e-05,
      "loss": 0.024577914178371428,
      "step": 80070
    },
    {
      "epoch": 22.730627306273064,
      "grad_norm": 8.686661720275879,
      "learning_rate": 7.727959125745104e-05,
      "loss": 0.010043241083621979,
      "step": 80080
    },
    {
      "epoch": 22.733465796196423,
      "grad_norm": 0.15997886657714844,
      "learning_rate": 7.727675276752768e-05,
      "loss": 0.008157923817634583,
      "step": 80090
    },
    {
      "epoch": 22.736304286119783,
      "grad_norm": 0.9059934616088867,
      "learning_rate": 7.727391427760433e-05,
      "loss": 0.018460023403167724,
      "step": 80100
    },
    {
      "epoch": 22.739142776043145,
      "grad_norm": 4.328021049499512,
      "learning_rate": 7.727107578768095e-05,
      "loss": 0.0299703449010849,
      "step": 80110
    },
    {
      "epoch": 22.741981265966505,
      "grad_norm": 14.030584335327148,
      "learning_rate": 7.72682372977576e-05,
      "loss": 0.01339331567287445,
      "step": 80120
    },
    {
      "epoch": 22.744819755889868,
      "grad_norm": 1.5305132865905762,
      "learning_rate": 7.726539880783424e-05,
      "loss": 0.008145565539598465,
      "step": 80130
    },
    {
      "epoch": 22.747658245813227,
      "grad_norm": 1.293609619140625,
      "learning_rate": 7.726256031791086e-05,
      "loss": 0.012378089874982835,
      "step": 80140
    },
    {
      "epoch": 22.75049673573659,
      "grad_norm": 2.0629069805145264,
      "learning_rate": 7.725972182798752e-05,
      "loss": 0.008963125944137573,
      "step": 80150
    },
    {
      "epoch": 22.75333522565995,
      "grad_norm": 2.436915397644043,
      "learning_rate": 7.725688333806416e-05,
      "loss": 0.01439805030822754,
      "step": 80160
    },
    {
      "epoch": 22.75617371558331,
      "grad_norm": 7.195567607879639,
      "learning_rate": 7.725404484814079e-05,
      "loss": 0.011921728402376175,
      "step": 80170
    },
    {
      "epoch": 22.75901220550667,
      "grad_norm": 10.314437866210938,
      "learning_rate": 7.725120635821743e-05,
      "loss": 0.014929772913455963,
      "step": 80180
    },
    {
      "epoch": 22.76185069543003,
      "grad_norm": 8.415282249450684,
      "learning_rate": 7.724836786829407e-05,
      "loss": 0.012924760580062866,
      "step": 80190
    },
    {
      "epoch": 22.764689185353394,
      "grad_norm": 1.7107425928115845,
      "learning_rate": 7.724552937837071e-05,
      "loss": 0.021162857115268708,
      "step": 80200
    },
    {
      "epoch": 22.767527675276753,
      "grad_norm": 0.7087059020996094,
      "learning_rate": 7.724269088844734e-05,
      "loss": 0.024776445329189302,
      "step": 80210
    },
    {
      "epoch": 22.770366165200112,
      "grad_norm": 1.5928897857666016,
      "learning_rate": 7.7239852398524e-05,
      "loss": 0.011905179917812347,
      "step": 80220
    },
    {
      "epoch": 22.773204655123475,
      "grad_norm": 4.587005138397217,
      "learning_rate": 7.723701390860064e-05,
      "loss": 0.010252158343791961,
      "step": 80230
    },
    {
      "epoch": 22.776043145046835,
      "grad_norm": 1.1359758377075195,
      "learning_rate": 7.723417541867726e-05,
      "loss": 0.008689039200544358,
      "step": 80240
    },
    {
      "epoch": 22.778881634970197,
      "grad_norm": 1.0572501420974731,
      "learning_rate": 7.72313369287539e-05,
      "loss": 0.01852968633174896,
      "step": 80250
    },
    {
      "epoch": 22.781720124893557,
      "grad_norm": 9.264588356018066,
      "learning_rate": 7.722849843883055e-05,
      "loss": 0.010110127925872802,
      "step": 80260
    },
    {
      "epoch": 22.784558614816916,
      "grad_norm": 0.22735106945037842,
      "learning_rate": 7.722565994890717e-05,
      "loss": 0.007475157082080841,
      "step": 80270
    },
    {
      "epoch": 22.78739710474028,
      "grad_norm": 11.317435264587402,
      "learning_rate": 7.722282145898383e-05,
      "loss": 0.012948741018772126,
      "step": 80280
    },
    {
      "epoch": 22.79023559466364,
      "grad_norm": 4.13758659362793,
      "learning_rate": 7.721998296906047e-05,
      "loss": 0.012177083641290665,
      "step": 80290
    },
    {
      "epoch": 22.793074084587,
      "grad_norm": 4.182050704956055,
      "learning_rate": 7.72171444791371e-05,
      "loss": 0.013347941637039184,
      "step": 80300
    },
    {
      "epoch": 22.79591257451036,
      "grad_norm": 0.21450981497764587,
      "learning_rate": 7.721430598921374e-05,
      "loss": 0.004410708323121071,
      "step": 80310
    },
    {
      "epoch": 22.79875106443372,
      "grad_norm": 12.97705078125,
      "learning_rate": 7.721146749929038e-05,
      "loss": 0.01112605631351471,
      "step": 80320
    },
    {
      "epoch": 22.801589554357083,
      "grad_norm": 13.333818435668945,
      "learning_rate": 7.720862900936702e-05,
      "loss": 0.01949972063302994,
      "step": 80330
    },
    {
      "epoch": 22.804428044280442,
      "grad_norm": 1.2529797554016113,
      "learning_rate": 7.720579051944365e-05,
      "loss": 0.009480949491262436,
      "step": 80340
    },
    {
      "epoch": 22.807266534203805,
      "grad_norm": 2.1950793266296387,
      "learning_rate": 7.72029520295203e-05,
      "loss": 0.008737755566835403,
      "step": 80350
    },
    {
      "epoch": 22.810105024127164,
      "grad_norm": 10.06026554107666,
      "learning_rate": 7.720011353959695e-05,
      "loss": 0.01917884349822998,
      "step": 80360
    },
    {
      "epoch": 22.812943514050524,
      "grad_norm": 0.45100489258766174,
      "learning_rate": 7.719727504967357e-05,
      "loss": 0.014648696780204773,
      "step": 80370
    },
    {
      "epoch": 22.815782003973887,
      "grad_norm": 10.042503356933594,
      "learning_rate": 7.719443655975022e-05,
      "loss": 0.009955741465091705,
      "step": 80380
    },
    {
      "epoch": 22.818620493897246,
      "grad_norm": 3.5598833560943604,
      "learning_rate": 7.719159806982686e-05,
      "loss": 0.027786251902580262,
      "step": 80390
    },
    {
      "epoch": 22.82145898382061,
      "grad_norm": 1.6881523132324219,
      "learning_rate": 7.718875957990349e-05,
      "loss": 0.018742060661315917,
      "step": 80400
    },
    {
      "epoch": 22.824297473743968,
      "grad_norm": 0.460720032453537,
      "learning_rate": 7.718592108998013e-05,
      "loss": 0.011539719253778457,
      "step": 80410
    },
    {
      "epoch": 22.827135963667327,
      "grad_norm": 0.5719653367996216,
      "learning_rate": 7.718308260005678e-05,
      "loss": 0.006723494082689285,
      "step": 80420
    },
    {
      "epoch": 22.82997445359069,
      "grad_norm": 10.23875617980957,
      "learning_rate": 7.718024411013341e-05,
      "loss": 0.021205466985702515,
      "step": 80430
    },
    {
      "epoch": 22.83281294351405,
      "grad_norm": 0.2245953530073166,
      "learning_rate": 7.717740562021005e-05,
      "loss": 0.010038424283266068,
      "step": 80440
    },
    {
      "epoch": 22.835651433437413,
      "grad_norm": 15.705665588378906,
      "learning_rate": 7.717456713028669e-05,
      "loss": 0.014079716801643372,
      "step": 80450
    },
    {
      "epoch": 22.838489923360772,
      "grad_norm": 0.5934808850288391,
      "learning_rate": 7.717172864036333e-05,
      "loss": 0.010980446636676789,
      "step": 80460
    },
    {
      "epoch": 22.84132841328413,
      "grad_norm": 4.694215774536133,
      "learning_rate": 7.716889015043996e-05,
      "loss": 0.004700841754674912,
      "step": 80470
    },
    {
      "epoch": 22.844166903207494,
      "grad_norm": 4.17569637298584,
      "learning_rate": 7.716605166051662e-05,
      "loss": 0.007723515480756759,
      "step": 80480
    },
    {
      "epoch": 22.847005393130853,
      "grad_norm": 0.8966252207756042,
      "learning_rate": 7.716321317059326e-05,
      "loss": 0.019952893257141113,
      "step": 80490
    },
    {
      "epoch": 22.849843883054216,
      "grad_norm": 0.3975766599178314,
      "learning_rate": 7.716037468066989e-05,
      "loss": 0.027705812454223634,
      "step": 80500
    },
    {
      "epoch": 22.849843883054216,
      "eval_accuracy": 0.9701786736186176,
      "eval_loss": 0.09653698652982712,
      "eval_runtime": 31.3457,
      "eval_samples_per_second": 501.727,
      "eval_steps_per_second": 7.848,
      "step": 80500
    },
    {
      "epoch": 22.852682372977576,
      "grad_norm": 0.948364794254303,
      "learning_rate": 7.715753619074653e-05,
      "loss": 0.023232600092887877,
      "step": 80510
    },
    {
      "epoch": 22.855520862900935,
      "grad_norm": 1.0897440910339355,
      "learning_rate": 7.715469770082317e-05,
      "loss": 0.018813584744930268,
      "step": 80520
    },
    {
      "epoch": 22.858359352824298,
      "grad_norm": 2.28645396232605,
      "learning_rate": 7.71518592108998e-05,
      "loss": 0.010080874711275101,
      "step": 80530
    },
    {
      "epoch": 22.861197842747657,
      "grad_norm": 4.140142440795898,
      "learning_rate": 7.714902072097644e-05,
      "loss": 0.009485715627670288,
      "step": 80540
    },
    {
      "epoch": 22.86403633267102,
      "grad_norm": 1.3476910591125488,
      "learning_rate": 7.714618223105309e-05,
      "loss": 0.013161341845989227,
      "step": 80550
    },
    {
      "epoch": 22.86687482259438,
      "grad_norm": 11.631701469421387,
      "learning_rate": 7.714334374112972e-05,
      "loss": 0.01815873831510544,
      "step": 80560
    },
    {
      "epoch": 22.86971331251774,
      "grad_norm": 6.01510763168335,
      "learning_rate": 7.714050525120636e-05,
      "loss": 0.026488906145095824,
      "step": 80570
    },
    {
      "epoch": 22.8725518024411,
      "grad_norm": 0.8410020470619202,
      "learning_rate": 7.7137666761283e-05,
      "loss": 0.008303754031658173,
      "step": 80580
    },
    {
      "epoch": 22.87539029236446,
      "grad_norm": 6.778632640838623,
      "learning_rate": 7.713482827135964e-05,
      "loss": 0.010488261282444001,
      "step": 80590
    },
    {
      "epoch": 22.878228782287824,
      "grad_norm": 12.870098114013672,
      "learning_rate": 7.713198978143627e-05,
      "loss": 0.021042497456073762,
      "step": 80600
    },
    {
      "epoch": 22.881067272211183,
      "grad_norm": 5.46492862701416,
      "learning_rate": 7.712915129151291e-05,
      "loss": 0.03602103590965271,
      "step": 80610
    },
    {
      "epoch": 22.883905762134546,
      "grad_norm": 7.759884357452393,
      "learning_rate": 7.712631280158957e-05,
      "loss": 0.012081490457057953,
      "step": 80620
    },
    {
      "epoch": 22.886744252057905,
      "grad_norm": 2.7611868381500244,
      "learning_rate": 7.71234743116662e-05,
      "loss": 0.021846266090869905,
      "step": 80630
    },
    {
      "epoch": 22.889582741981265,
      "grad_norm": 0.18728521466255188,
      "learning_rate": 7.712063582174284e-05,
      "loss": 0.009469299763441085,
      "step": 80640
    },
    {
      "epoch": 22.892421231904628,
      "grad_norm": 1.714592695236206,
      "learning_rate": 7.711779733181948e-05,
      "loss": 0.006551323831081391,
      "step": 80650
    },
    {
      "epoch": 22.895259721827987,
      "grad_norm": 0.7882659435272217,
      "learning_rate": 7.711495884189611e-05,
      "loss": 0.0070470161736011505,
      "step": 80660
    },
    {
      "epoch": 22.89809821175135,
      "grad_norm": 3.7027416229248047,
      "learning_rate": 7.711212035197275e-05,
      "loss": 0.011835765093564987,
      "step": 80670
    },
    {
      "epoch": 22.90093670167471,
      "grad_norm": 1.3493903875350952,
      "learning_rate": 7.71092818620494e-05,
      "loss": 0.007410074770450592,
      "step": 80680
    },
    {
      "epoch": 22.90377519159807,
      "grad_norm": 6.727886199951172,
      "learning_rate": 7.710644337212603e-05,
      "loss": 0.009225580096244811,
      "step": 80690
    },
    {
      "epoch": 22.90661368152143,
      "grad_norm": 3.2748970985412598,
      "learning_rate": 7.710360488220267e-05,
      "loss": 0.012339530885219574,
      "step": 80700
    },
    {
      "epoch": 22.90945217144479,
      "grad_norm": 0.43072709441185,
      "learning_rate": 7.710076639227931e-05,
      "loss": 0.008979567140340806,
      "step": 80710
    },
    {
      "epoch": 22.912290661368154,
      "grad_norm": 5.980490684509277,
      "learning_rate": 7.709792790235596e-05,
      "loss": 0.013184870779514312,
      "step": 80720
    },
    {
      "epoch": 22.915129151291513,
      "grad_norm": 1.4979661703109741,
      "learning_rate": 7.709508941243258e-05,
      "loss": 0.009309562295675278,
      "step": 80730
    },
    {
      "epoch": 22.917967641214872,
      "grad_norm": 19.444664001464844,
      "learning_rate": 7.709225092250922e-05,
      "loss": 0.029952794313430786,
      "step": 80740
    },
    {
      "epoch": 22.920806131138235,
      "grad_norm": 1.7137722969055176,
      "learning_rate": 7.708941243258587e-05,
      "loss": 0.02423321157693863,
      "step": 80750
    },
    {
      "epoch": 22.923644621061595,
      "grad_norm": 11.211395263671875,
      "learning_rate": 7.708657394266251e-05,
      "loss": 0.012348046153783798,
      "step": 80760
    },
    {
      "epoch": 22.926483110984957,
      "grad_norm": 7.594901084899902,
      "learning_rate": 7.708373545273915e-05,
      "loss": 0.023171013593673705,
      "step": 80770
    },
    {
      "epoch": 22.929321600908317,
      "grad_norm": 0.4991413950920105,
      "learning_rate": 7.708089696281579e-05,
      "loss": 0.01727769672870636,
      "step": 80780
    },
    {
      "epoch": 22.932160090831676,
      "grad_norm": 2.3327393531799316,
      "learning_rate": 7.707805847289242e-05,
      "loss": 0.011949050426483154,
      "step": 80790
    },
    {
      "epoch": 22.93499858075504,
      "grad_norm": 1.4672796726226807,
      "learning_rate": 7.707521998296906e-05,
      "loss": 0.012447305768728257,
      "step": 80800
    },
    {
      "epoch": 22.9378370706784,
      "grad_norm": 1.2329938411712646,
      "learning_rate": 7.707238149304571e-05,
      "loss": 0.01845271438360214,
      "step": 80810
    },
    {
      "epoch": 22.94067556060176,
      "grad_norm": 4.342756748199463,
      "learning_rate": 7.706954300312234e-05,
      "loss": 0.019174408912658692,
      "step": 80820
    },
    {
      "epoch": 22.94351405052512,
      "grad_norm": 0.26236632466316223,
      "learning_rate": 7.706670451319898e-05,
      "loss": 0.02667504847049713,
      "step": 80830
    },
    {
      "epoch": 22.94635254044848,
      "grad_norm": 0.35015568137168884,
      "learning_rate": 7.706386602327562e-05,
      "loss": 0.029777154326438904,
      "step": 80840
    },
    {
      "epoch": 22.949191030371843,
      "grad_norm": 1.522350788116455,
      "learning_rate": 7.706102753335225e-05,
      "loss": 0.03382187187671661,
      "step": 80850
    },
    {
      "epoch": 22.952029520295202,
      "grad_norm": 0.41056889295578003,
      "learning_rate": 7.70581890434289e-05,
      "loss": 0.020494188368320464,
      "step": 80860
    },
    {
      "epoch": 22.954868010218565,
      "grad_norm": 12.632144927978516,
      "learning_rate": 7.705535055350554e-05,
      "loss": 0.014747911691665649,
      "step": 80870
    },
    {
      "epoch": 22.957706500141924,
      "grad_norm": 3.3428947925567627,
      "learning_rate": 7.705251206358218e-05,
      "loss": 0.006142091006040573,
      "step": 80880
    },
    {
      "epoch": 22.960544990065284,
      "grad_norm": 12.096625328063965,
      "learning_rate": 7.704967357365882e-05,
      "loss": 0.010761167854070663,
      "step": 80890
    },
    {
      "epoch": 22.963383479988646,
      "grad_norm": 8.952106475830078,
      "learning_rate": 7.704683508373546e-05,
      "loss": 0.005761110782623291,
      "step": 80900
    },
    {
      "epoch": 22.966221969912006,
      "grad_norm": 5.995619297027588,
      "learning_rate": 7.70439965938121e-05,
      "loss": 0.006663285940885544,
      "step": 80910
    },
    {
      "epoch": 22.96906045983537,
      "grad_norm": 3.710517406463623,
      "learning_rate": 7.704115810388873e-05,
      "loss": 0.010477218031883239,
      "step": 80920
    },
    {
      "epoch": 22.971898949758728,
      "grad_norm": 0.8890381455421448,
      "learning_rate": 7.703831961396537e-05,
      "loss": 0.007119788229465485,
      "step": 80930
    },
    {
      "epoch": 22.974737439682087,
      "grad_norm": 7.485180854797363,
      "learning_rate": 7.703548112404201e-05,
      "loss": 0.010250549763441086,
      "step": 80940
    },
    {
      "epoch": 22.97757592960545,
      "grad_norm": 5.8570098876953125,
      "learning_rate": 7.703264263411865e-05,
      "loss": 0.01283009946346283,
      "step": 80950
    },
    {
      "epoch": 22.98041441952881,
      "grad_norm": 1.0446549654006958,
      "learning_rate": 7.70298041441953e-05,
      "loss": 0.012663394212722778,
      "step": 80960
    },
    {
      "epoch": 22.983252909452172,
      "grad_norm": Infinity,
      "learning_rate": 7.702696565427194e-05,
      "loss": 0.02184543162584305,
      "step": 80970
    },
    {
      "epoch": 22.986091399375532,
      "grad_norm": 1.4425677061080933,
      "learning_rate": 7.70244110133409e-05,
      "loss": 0.011099603772163392,
      "step": 80980
    },
    {
      "epoch": 22.988929889298895,
      "grad_norm": 4.527901649475098,
      "learning_rate": 7.702157252341754e-05,
      "loss": 0.011553056538105011,
      "step": 80990
    },
    {
      "epoch": 22.991768379222254,
      "grad_norm": 1.7087217569351196,
      "learning_rate": 7.701873403349418e-05,
      "loss": 0.012025050818920135,
      "step": 81000
    },
    {
      "epoch": 22.991768379222254,
      "eval_accuracy": 0.9655369746296179,
      "eval_loss": 0.11839976161718369,
      "eval_runtime": 31.5879,
      "eval_samples_per_second": 497.881,
      "eval_steps_per_second": 7.788,
      "step": 81000
    },
    {
      "epoch": 22.994606869145613,
      "grad_norm": 1.620803713798523,
      "learning_rate": 7.701589554357083e-05,
      "loss": 0.03272469043731689,
      "step": 81010
    },
    {
      "epoch": 22.997445359068976,
      "grad_norm": 14.319477081298828,
      "learning_rate": 7.701305705364747e-05,
      "loss": 0.027868452668190002,
      "step": 81020
    },
    {
      "epoch": 23.000283848992336,
      "grad_norm": 6.962742805480957,
      "learning_rate": 7.70102185637241e-05,
      "loss": 0.04558496177196503,
      "step": 81030
    },
    {
      "epoch": 23.0031223389157,
      "grad_norm": 0.3507782816886902,
      "learning_rate": 7.700738007380074e-05,
      "loss": 0.007761963456869125,
      "step": 81040
    },
    {
      "epoch": 23.005960828839058,
      "grad_norm": 3.6918137073516846,
      "learning_rate": 7.700454158387738e-05,
      "loss": 0.02295752465724945,
      "step": 81050
    },
    {
      "epoch": 23.008799318762417,
      "grad_norm": 1.276634931564331,
      "learning_rate": 7.700170309395402e-05,
      "loss": 0.006325936317443848,
      "step": 81060
    },
    {
      "epoch": 23.01163780868578,
      "grad_norm": 0.3400377631187439,
      "learning_rate": 7.699886460403066e-05,
      "loss": 0.010025095939636231,
      "step": 81070
    },
    {
      "epoch": 23.01447629860914,
      "grad_norm": 1.0349035263061523,
      "learning_rate": 7.69960261141073e-05,
      "loss": 0.0036687493324279786,
      "step": 81080
    },
    {
      "epoch": 23.017314788532502,
      "grad_norm": 0.08968742936849594,
      "learning_rate": 7.699318762418394e-05,
      "loss": 0.01273488849401474,
      "step": 81090
    },
    {
      "epoch": 23.02015327845586,
      "grad_norm": 0.22282001376152039,
      "learning_rate": 7.699034913426057e-05,
      "loss": 0.02432103157043457,
      "step": 81100
    },
    {
      "epoch": 23.02299176837922,
      "grad_norm": 0.25596579909324646,
      "learning_rate": 7.698751064433721e-05,
      "loss": 0.0030750010162591936,
      "step": 81110
    },
    {
      "epoch": 23.025830258302584,
      "grad_norm": 0.02531314454972744,
      "learning_rate": 7.698467215441385e-05,
      "loss": 0.003897952288389206,
      "step": 81120
    },
    {
      "epoch": 23.028668748225943,
      "grad_norm": 0.28874582052230835,
      "learning_rate": 7.69818336644905e-05,
      "loss": 0.004992254450917244,
      "step": 81130
    },
    {
      "epoch": 23.031507238149306,
      "grad_norm": 2.429277181625366,
      "learning_rate": 7.697899517456714e-05,
      "loss": 0.005841082334518433,
      "step": 81140
    },
    {
      "epoch": 23.034345728072665,
      "grad_norm": 2.9904825687408447,
      "learning_rate": 7.697615668464378e-05,
      "loss": 0.01131460964679718,
      "step": 81150
    },
    {
      "epoch": 23.037184217996025,
      "grad_norm": 3.2538156509399414,
      "learning_rate": 7.69733181947204e-05,
      "loss": 0.016667813062667847,
      "step": 81160
    },
    {
      "epoch": 23.040022707919388,
      "grad_norm": 0.5200117230415344,
      "learning_rate": 7.697047970479705e-05,
      "loss": 0.011638278514146805,
      "step": 81170
    },
    {
      "epoch": 23.042861197842747,
      "grad_norm": 0.9085693955421448,
      "learning_rate": 7.696764121487369e-05,
      "loss": 0.00712493434548378,
      "step": 81180
    },
    {
      "epoch": 23.04569968776611,
      "grad_norm": 0.18917958438396454,
      "learning_rate": 7.696480272495033e-05,
      "loss": 0.0066272996366024016,
      "step": 81190
    },
    {
      "epoch": 23.04853817768947,
      "grad_norm": 1.7186013460159302,
      "learning_rate": 7.696196423502697e-05,
      "loss": 0.019600804150104522,
      "step": 81200
    },
    {
      "epoch": 23.05137666761283,
      "grad_norm": 2.469651460647583,
      "learning_rate": 7.695912574510361e-05,
      "loss": 0.006852841377258301,
      "step": 81210
    },
    {
      "epoch": 23.05421515753619,
      "grad_norm": 0.803154468536377,
      "learning_rate": 7.695628725518025e-05,
      "loss": 0.008562416583299638,
      "step": 81220
    },
    {
      "epoch": 23.05705364745955,
      "grad_norm": 0.532713770866394,
      "learning_rate": 7.695344876525688e-05,
      "loss": 0.006589559465646743,
      "step": 81230
    },
    {
      "epoch": 23.059892137382914,
      "grad_norm": 0.40946492552757263,
      "learning_rate": 7.695061027533352e-05,
      "loss": 0.01059492826461792,
      "step": 81240
    },
    {
      "epoch": 23.062730627306273,
      "grad_norm": 0.6528635621070862,
      "learning_rate": 7.694777178541016e-05,
      "loss": 0.013476558029651642,
      "step": 81250
    },
    {
      "epoch": 23.065569117229632,
      "grad_norm": 9.170567512512207,
      "learning_rate": 7.69449332954868e-05,
      "loss": 0.0057659059762954715,
      "step": 81260
    },
    {
      "epoch": 23.068407607152995,
      "grad_norm": 6.036689281463623,
      "learning_rate": 7.694209480556345e-05,
      "loss": 0.010828778147697449,
      "step": 81270
    },
    {
      "epoch": 23.071246097076354,
      "grad_norm": 0.28587600588798523,
      "learning_rate": 7.693925631564009e-05,
      "loss": 0.009224681556224823,
      "step": 81280
    },
    {
      "epoch": 23.074084586999717,
      "grad_norm": 13.310474395751953,
      "learning_rate": 7.693641782571672e-05,
      "loss": 0.014253759384155273,
      "step": 81290
    },
    {
      "epoch": 23.076923076923077,
      "grad_norm": 14.087088584899902,
      "learning_rate": 7.693357933579336e-05,
      "loss": 0.020697185397148134,
      "step": 81300
    },
    {
      "epoch": 23.079761566846436,
      "grad_norm": 4.609619617462158,
      "learning_rate": 7.693074084587e-05,
      "loss": 0.004944352060556411,
      "step": 81310
    },
    {
      "epoch": 23.0826000567698,
      "grad_norm": 3.9731431007385254,
      "learning_rate": 7.692790235594664e-05,
      "loss": 0.010432165861129761,
      "step": 81320
    },
    {
      "epoch": 23.085438546693158,
      "grad_norm": 1.303425669670105,
      "learning_rate": 7.692506386602328e-05,
      "loss": 0.01547238528728485,
      "step": 81330
    },
    {
      "epoch": 23.08827703661652,
      "grad_norm": 0.16182425618171692,
      "learning_rate": 7.692222537609992e-05,
      "loss": 0.007608227431774139,
      "step": 81340
    },
    {
      "epoch": 23.09111552653988,
      "grad_norm": 0.09139491617679596,
      "learning_rate": 7.691938688617656e-05,
      "loss": 0.008348201215267182,
      "step": 81350
    },
    {
      "epoch": 23.093954016463243,
      "grad_norm": 1.6499582529067993,
      "learning_rate": 7.691654839625319e-05,
      "loss": 0.008794358372688294,
      "step": 81360
    },
    {
      "epoch": 23.096792506386603,
      "grad_norm": 2.591325283050537,
      "learning_rate": 7.691370990632983e-05,
      "loss": 0.02172408401966095,
      "step": 81370
    },
    {
      "epoch": 23.099630996309962,
      "grad_norm": 0.26487675309181213,
      "learning_rate": 7.691087141640648e-05,
      "loss": 0.016495434939861296,
      "step": 81380
    },
    {
      "epoch": 23.102469486233325,
      "grad_norm": 1.375809669494629,
      "learning_rate": 7.690803292648312e-05,
      "loss": 0.0035038746893405913,
      "step": 81390
    },
    {
      "epoch": 23.105307976156684,
      "grad_norm": 0.26093676686286926,
      "learning_rate": 7.690519443655976e-05,
      "loss": 0.00615517757833004,
      "step": 81400
    },
    {
      "epoch": 23.108146466080047,
      "grad_norm": 0.34218186140060425,
      "learning_rate": 7.69023559466364e-05,
      "loss": 0.01887122541666031,
      "step": 81410
    },
    {
      "epoch": 23.110984956003406,
      "grad_norm": 8.678123474121094,
      "learning_rate": 7.689951745671303e-05,
      "loss": 0.005620415136218071,
      "step": 81420
    },
    {
      "epoch": 23.113823445926766,
      "grad_norm": 0.11254508048295975,
      "learning_rate": 7.689667896678967e-05,
      "loss": 0.013050979375839234,
      "step": 81430
    },
    {
      "epoch": 23.11666193585013,
      "grad_norm": 1.4598323106765747,
      "learning_rate": 7.689384047686631e-05,
      "loss": 0.007474815100431442,
      "step": 81440
    },
    {
      "epoch": 23.119500425773488,
      "grad_norm": 2.1692874431610107,
      "learning_rate": 7.689100198694295e-05,
      "loss": 0.009286851435899735,
      "step": 81450
    },
    {
      "epoch": 23.12233891569685,
      "grad_norm": 2.1892526149749756,
      "learning_rate": 7.688816349701959e-05,
      "loss": 0.011077985167503357,
      "step": 81460
    },
    {
      "epoch": 23.12517740562021,
      "grad_norm": 3.690589666366577,
      "learning_rate": 7.688532500709623e-05,
      "loss": 0.00821208581328392,
      "step": 81470
    },
    {
      "epoch": 23.12801589554357,
      "grad_norm": 1.410402774810791,
      "learning_rate": 7.688248651717288e-05,
      "loss": 0.014732782542705537,
      "step": 81480
    },
    {
      "epoch": 23.130854385466932,
      "grad_norm": 17.127286911010742,
      "learning_rate": 7.68796480272495e-05,
      "loss": 0.025893032550811768,
      "step": 81490
    },
    {
      "epoch": 23.13369287539029,
      "grad_norm": 4.727377891540527,
      "learning_rate": 7.687680953732614e-05,
      "loss": 0.04660629630088806,
      "step": 81500
    },
    {
      "epoch": 23.13369287539029,
      "eval_accuracy": 0.9610860303935906,
      "eval_loss": 0.1265936642885208,
      "eval_runtime": 32.1294,
      "eval_samples_per_second": 489.489,
      "eval_steps_per_second": 7.657,
      "step": 81500
    },
    {
      "epoch": 23.136531365313655,
      "grad_norm": 3.576752185821533,
      "learning_rate": 7.687397104740279e-05,
      "loss": 0.016522106528282166,
      "step": 81510
    },
    {
      "epoch": 23.139369855237014,
      "grad_norm": 0.2805086374282837,
      "learning_rate": 7.687113255747941e-05,
      "loss": 0.006738145649433136,
      "step": 81520
    },
    {
      "epoch": 23.142208345160373,
      "grad_norm": 0.22854863107204437,
      "learning_rate": 7.686829406755607e-05,
      "loss": 0.019515399634838105,
      "step": 81530
    },
    {
      "epoch": 23.145046835083736,
      "grad_norm": 1.3174718618392944,
      "learning_rate": 7.686545557763271e-05,
      "loss": 0.00976240262389183,
      "step": 81540
    },
    {
      "epoch": 23.147885325007096,
      "grad_norm": 0.4594900906085968,
      "learning_rate": 7.686261708770934e-05,
      "loss": 0.013127994537353516,
      "step": 81550
    },
    {
      "epoch": 23.15072381493046,
      "grad_norm": 8.161598205566406,
      "learning_rate": 7.685977859778598e-05,
      "loss": 0.021325382590293884,
      "step": 81560
    },
    {
      "epoch": 23.153562304853818,
      "grad_norm": 0.4259471297264099,
      "learning_rate": 7.685694010786262e-05,
      "loss": 0.006512487679719925,
      "step": 81570
    },
    {
      "epoch": 23.156400794777177,
      "grad_norm": 1.1085305213928223,
      "learning_rate": 7.685410161793926e-05,
      "loss": 0.007955869287252426,
      "step": 81580
    },
    {
      "epoch": 23.15923928470054,
      "grad_norm": 0.5901926755905151,
      "learning_rate": 7.68512631280159e-05,
      "loss": 0.01664751172065735,
      "step": 81590
    },
    {
      "epoch": 23.1620777746239,
      "grad_norm": 1.5823017358779907,
      "learning_rate": 7.684842463809255e-05,
      "loss": 0.00969500169157982,
      "step": 81600
    },
    {
      "epoch": 23.164916264547262,
      "grad_norm": 10.561538696289062,
      "learning_rate": 7.684558614816917e-05,
      "loss": 0.02339756041765213,
      "step": 81610
    },
    {
      "epoch": 23.16775475447062,
      "grad_norm": 1.017353892326355,
      "learning_rate": 7.684274765824581e-05,
      "loss": 0.007292459905147553,
      "step": 81620
    },
    {
      "epoch": 23.17059324439398,
      "grad_norm": 1.3734478950500488,
      "learning_rate": 7.683990916832246e-05,
      "loss": 0.018131014704704285,
      "step": 81630
    },
    {
      "epoch": 23.173431734317344,
      "grad_norm": 10.98176097869873,
      "learning_rate": 7.68370706783991e-05,
      "loss": 0.01684893071651459,
      "step": 81640
    },
    {
      "epoch": 23.176270224240703,
      "grad_norm": 0.8957962989807129,
      "learning_rate": 7.683423218847572e-05,
      "loss": 0.014210113883018493,
      "step": 81650
    },
    {
      "epoch": 23.179108714164066,
      "grad_norm": 5.073215007781982,
      "learning_rate": 7.683139369855238e-05,
      "loss": 0.008887016773223877,
      "step": 81660
    },
    {
      "epoch": 23.181947204087425,
      "grad_norm": 2.3756263256073,
      "learning_rate": 7.682855520862902e-05,
      "loss": 0.00913379043340683,
      "step": 81670
    },
    {
      "epoch": 23.184785694010785,
      "grad_norm": 6.557875156402588,
      "learning_rate": 7.682571671870565e-05,
      "loss": 0.010885653644800186,
      "step": 81680
    },
    {
      "epoch": 23.187624183934147,
      "grad_norm": 13.57228946685791,
      "learning_rate": 7.682287822878229e-05,
      "loss": 0.023547247052192688,
      "step": 81690
    },
    {
      "epoch": 23.190462673857507,
      "grad_norm": 0.2959260046482086,
      "learning_rate": 7.682003973885893e-05,
      "loss": 0.02730979323387146,
      "step": 81700
    },
    {
      "epoch": 23.19330116378087,
      "grad_norm": 4.152402877807617,
      "learning_rate": 7.681720124893557e-05,
      "loss": 0.011736023426055908,
      "step": 81710
    },
    {
      "epoch": 23.19613965370423,
      "grad_norm": 1.0124796628952026,
      "learning_rate": 7.68143627590122e-05,
      "loss": 0.010967476665973664,
      "step": 81720
    },
    {
      "epoch": 23.19897814362759,
      "grad_norm": 0.7735602259635925,
      "learning_rate": 7.681152426908886e-05,
      "loss": 0.011045131087303161,
      "step": 81730
    },
    {
      "epoch": 23.20181663355095,
      "grad_norm": 2.0713863372802734,
      "learning_rate": 7.680868577916548e-05,
      "loss": 0.005089344829320908,
      "step": 81740
    },
    {
      "epoch": 23.20465512347431,
      "grad_norm": 3.1303160190582275,
      "learning_rate": 7.680584728924213e-05,
      "loss": 0.019877766072750092,
      "step": 81750
    },
    {
      "epoch": 23.207493613397673,
      "grad_norm": 13.594548225402832,
      "learning_rate": 7.680300879931877e-05,
      "loss": 0.019787566363811494,
      "step": 81760
    },
    {
      "epoch": 23.210332103321033,
      "grad_norm": 9.464159965515137,
      "learning_rate": 7.680017030939541e-05,
      "loss": 0.02696545422077179,
      "step": 81770
    },
    {
      "epoch": 23.213170593244396,
      "grad_norm": 6.040079593658447,
      "learning_rate": 7.679733181947204e-05,
      "loss": 0.012206007540225983,
      "step": 81780
    },
    {
      "epoch": 23.216009083167755,
      "grad_norm": 0.13113054633140564,
      "learning_rate": 7.679449332954869e-05,
      "loss": 0.00852060467004776,
      "step": 81790
    },
    {
      "epoch": 23.218847573091114,
      "grad_norm": 0.1292133927345276,
      "learning_rate": 7.679165483962533e-05,
      "loss": 0.007242535054683685,
      "step": 81800
    },
    {
      "epoch": 23.221686063014477,
      "grad_norm": 6.790336608886719,
      "learning_rate": 7.678881634970196e-05,
      "loss": 0.016596157848834992,
      "step": 81810
    },
    {
      "epoch": 23.224524552937837,
      "grad_norm": 6.135738849639893,
      "learning_rate": 7.67859778597786e-05,
      "loss": 0.011655677855014802,
      "step": 81820
    },
    {
      "epoch": 23.2273630428612,
      "grad_norm": 1.4554401636123657,
      "learning_rate": 7.678313936985524e-05,
      "loss": 0.01121780127286911,
      "step": 81830
    },
    {
      "epoch": 23.23020153278456,
      "grad_norm": 3.1679046154022217,
      "learning_rate": 7.678030087993187e-05,
      "loss": 0.014025050401687621,
      "step": 81840
    },
    {
      "epoch": 23.233040022707918,
      "grad_norm": 3.1480331420898438,
      "learning_rate": 7.677746239000851e-05,
      "loss": 0.014249539375305176,
      "step": 81850
    },
    {
      "epoch": 23.23587851263128,
      "grad_norm": 5.555353164672852,
      "learning_rate": 7.677462390008517e-05,
      "loss": 0.01360018253326416,
      "step": 81860
    },
    {
      "epoch": 23.23871700255464,
      "grad_norm": 0.6712464690208435,
      "learning_rate": 7.67717854101618e-05,
      "loss": 0.029358720779418944,
      "step": 81870
    },
    {
      "epoch": 23.241555492478003,
      "grad_norm": 0.22279907763004303,
      "learning_rate": 7.676894692023844e-05,
      "loss": 0.0040431000292301175,
      "step": 81880
    },
    {
      "epoch": 23.244393982401363,
      "grad_norm": 0.14023631811141968,
      "learning_rate": 7.676610843031508e-05,
      "loss": 0.024998120963573456,
      "step": 81890
    },
    {
      "epoch": 23.247232472324722,
      "grad_norm": 8.533802032470703,
      "learning_rate": 7.676326994039172e-05,
      "loss": 0.008667520433664321,
      "step": 81900
    },
    {
      "epoch": 23.250070962248085,
      "grad_norm": 0.43664854764938354,
      "learning_rate": 7.676043145046835e-05,
      "loss": 0.004219681024551392,
      "step": 81910
    },
    {
      "epoch": 23.252909452171444,
      "grad_norm": 7.283153057098389,
      "learning_rate": 7.675759296054499e-05,
      "loss": 0.01028265655040741,
      "step": 81920
    },
    {
      "epoch": 23.255747942094807,
      "grad_norm": 1.0547778606414795,
      "learning_rate": 7.675475447062164e-05,
      "loss": 0.0037733543664216994,
      "step": 81930
    },
    {
      "epoch": 23.258586432018166,
      "grad_norm": 1.406489372253418,
      "learning_rate": 7.675191598069827e-05,
      "loss": 0.009622058272361756,
      "step": 81940
    },
    {
      "epoch": 23.261424921941526,
      "grad_norm": 0.505281925201416,
      "learning_rate": 7.674907749077491e-05,
      "loss": 0.008259361237287521,
      "step": 81950
    },
    {
      "epoch": 23.26426341186489,
      "grad_norm": 0.05560813471674919,
      "learning_rate": 7.674623900085155e-05,
      "loss": 0.016153261065483093,
      "step": 81960
    },
    {
      "epoch": 23.267101901788248,
      "grad_norm": 0.7451133728027344,
      "learning_rate": 7.674340051092818e-05,
      "loss": 0.019797845184803008,
      "step": 81970
    },
    {
      "epoch": 23.26994039171161,
      "grad_norm": 0.3206908404827118,
      "learning_rate": 7.674056202100482e-05,
      "loss": 0.011257698386907577,
      "step": 81980
    },
    {
      "epoch": 23.27277888163497,
      "grad_norm": 1.20969557762146,
      "learning_rate": 7.673772353108148e-05,
      "loss": 0.01275096982717514,
      "step": 81990
    },
    {
      "epoch": 23.27561737155833,
      "grad_norm": 2.338411331176758,
      "learning_rate": 7.67348850411581e-05,
      "loss": 0.006433916836977005,
      "step": 82000
    },
    {
      "epoch": 23.27561737155833,
      "eval_accuracy": 0.9689705601831246,
      "eval_loss": 0.10624296218156815,
      "eval_runtime": 31.7804,
      "eval_samples_per_second": 494.865,
      "eval_steps_per_second": 7.741,
      "step": 82000
    },
    {
      "epoch": 23.278455861481692,
      "grad_norm": 1.3388887643814087,
      "learning_rate": 7.673204655123475e-05,
      "loss": 0.00449761301279068,
      "step": 82010
    },
    {
      "epoch": 23.28129435140505,
      "grad_norm": 11.14245891571045,
      "learning_rate": 7.672920806131139e-05,
      "loss": 0.01533648520708084,
      "step": 82020
    },
    {
      "epoch": 23.284132841328415,
      "grad_norm": 0.45252472162246704,
      "learning_rate": 7.672636957138803e-05,
      "loss": 0.01781100332736969,
      "step": 82030
    },
    {
      "epoch": 23.286971331251774,
      "grad_norm": 9.260944366455078,
      "learning_rate": 7.672353108146466e-05,
      "loss": 0.01715707629919052,
      "step": 82040
    },
    {
      "epoch": 23.289809821175133,
      "grad_norm": 3.249415159225464,
      "learning_rate": 7.67206925915413e-05,
      "loss": 0.011982394009828567,
      "step": 82050
    },
    {
      "epoch": 23.292648311098496,
      "grad_norm": 0.23406581580638885,
      "learning_rate": 7.671785410161795e-05,
      "loss": 0.019793927669525146,
      "step": 82060
    },
    {
      "epoch": 23.295486801021855,
      "grad_norm": 0.7506312131881714,
      "learning_rate": 7.671501561169458e-05,
      "loss": 0.009187953174114227,
      "step": 82070
    },
    {
      "epoch": 23.29832529094522,
      "grad_norm": 2.8571550846099854,
      "learning_rate": 7.671217712177122e-05,
      "loss": 0.010929618775844575,
      "step": 82080
    },
    {
      "epoch": 23.301163780868578,
      "grad_norm": 0.7757213115692139,
      "learning_rate": 7.670933863184786e-05,
      "loss": 0.014920903742313385,
      "step": 82090
    },
    {
      "epoch": 23.304002270791937,
      "grad_norm": 8.449934959411621,
      "learning_rate": 7.670650014192449e-05,
      "loss": 0.027359050512313843,
      "step": 82100
    },
    {
      "epoch": 23.3068407607153,
      "grad_norm": 1.6749534606933594,
      "learning_rate": 7.670366165200113e-05,
      "loss": 0.006395766139030456,
      "step": 82110
    },
    {
      "epoch": 23.30967925063866,
      "grad_norm": 5.584567546844482,
      "learning_rate": 7.670082316207777e-05,
      "loss": 0.025453031063079834,
      "step": 82120
    },
    {
      "epoch": 23.312517740562022,
      "grad_norm": 3.466580867767334,
      "learning_rate": 7.669798467215442e-05,
      "loss": 0.005917467921972275,
      "step": 82130
    },
    {
      "epoch": 23.31535623048538,
      "grad_norm": 4.849684238433838,
      "learning_rate": 7.669514618223106e-05,
      "loss": 0.010397037863731385,
      "step": 82140
    },
    {
      "epoch": 23.31819472040874,
      "grad_norm": 0.607302725315094,
      "learning_rate": 7.66923076923077e-05,
      "loss": 0.007649130374193192,
      "step": 82150
    },
    {
      "epoch": 23.321033210332104,
      "grad_norm": 0.4653611481189728,
      "learning_rate": 7.668946920238434e-05,
      "loss": 0.010015854239463806,
      "step": 82160
    },
    {
      "epoch": 23.323871700255463,
      "grad_norm": 0.5790054798126221,
      "learning_rate": 7.668663071246097e-05,
      "loss": 0.016525769233703615,
      "step": 82170
    },
    {
      "epoch": 23.326710190178826,
      "grad_norm": 0.34577104449272156,
      "learning_rate": 7.668379222253761e-05,
      "loss": 0.006699518859386444,
      "step": 82180
    },
    {
      "epoch": 23.329548680102185,
      "grad_norm": 0.6061055660247803,
      "learning_rate": 7.668095373261426e-05,
      "loss": 0.02122640162706375,
      "step": 82190
    },
    {
      "epoch": 23.332387170025548,
      "grad_norm": 5.8752121925354,
      "learning_rate": 7.667811524269089e-05,
      "loss": 0.013380686938762664,
      "step": 82200
    },
    {
      "epoch": 23.335225659948907,
      "grad_norm": 1.414131760597229,
      "learning_rate": 7.667527675276753e-05,
      "loss": 0.008331533521413803,
      "step": 82210
    },
    {
      "epoch": 23.338064149872267,
      "grad_norm": 1.5979766845703125,
      "learning_rate": 7.667243826284417e-05,
      "loss": 0.022481757402420043,
      "step": 82220
    },
    {
      "epoch": 23.34090263979563,
      "grad_norm": 1.1774921417236328,
      "learning_rate": 7.66695997729208e-05,
      "loss": 0.007056374847888946,
      "step": 82230
    },
    {
      "epoch": 23.34374112971899,
      "grad_norm": 6.198848724365234,
      "learning_rate": 7.666676128299744e-05,
      "loss": 0.008384213596582413,
      "step": 82240
    },
    {
      "epoch": 23.346579619642352,
      "grad_norm": 0.41497403383255005,
      "learning_rate": 7.666392279307409e-05,
      "loss": 0.009454632550477982,
      "step": 82250
    },
    {
      "epoch": 23.34941810956571,
      "grad_norm": 5.006596565246582,
      "learning_rate": 7.666108430315073e-05,
      "loss": 0.021467940509319307,
      "step": 82260
    },
    {
      "epoch": 23.35225659948907,
      "grad_norm": 1.8049062490463257,
      "learning_rate": 7.665824581322737e-05,
      "loss": 0.017208920419216157,
      "step": 82270
    },
    {
      "epoch": 23.355095089412433,
      "grad_norm": 0.6289519667625427,
      "learning_rate": 7.665540732330401e-05,
      "loss": 0.011082616448402405,
      "step": 82280
    },
    {
      "epoch": 23.357933579335793,
      "grad_norm": 5.964437484741211,
      "learning_rate": 7.665256883338065e-05,
      "loss": 0.01135619729757309,
      "step": 82290
    },
    {
      "epoch": 23.360772069259156,
      "grad_norm": 2.3437535762786865,
      "learning_rate": 7.664973034345728e-05,
      "loss": 0.019390083849430084,
      "step": 82300
    },
    {
      "epoch": 23.363610559182515,
      "grad_norm": 0.46483907103538513,
      "learning_rate": 7.664689185353392e-05,
      "loss": 0.007107363641262054,
      "step": 82310
    },
    {
      "epoch": 23.366449049105874,
      "grad_norm": 0.26846811175346375,
      "learning_rate": 7.664405336361056e-05,
      "loss": 0.01215272694826126,
      "step": 82320
    },
    {
      "epoch": 23.369287539029237,
      "grad_norm": 13.927918434143066,
      "learning_rate": 7.66412148736872e-05,
      "loss": 0.017893636226654054,
      "step": 82330
    },
    {
      "epoch": 23.372126028952596,
      "grad_norm": 1.4418696165084839,
      "learning_rate": 7.663837638376384e-05,
      "loss": 0.009810656309127808,
      "step": 82340
    },
    {
      "epoch": 23.37496451887596,
      "grad_norm": 13.951860427856445,
      "learning_rate": 7.663553789384049e-05,
      "loss": 0.015971386432647706,
      "step": 82350
    },
    {
      "epoch": 23.37780300879932,
      "grad_norm": 0.137238010764122,
      "learning_rate": 7.663269940391711e-05,
      "loss": 0.00865924134850502,
      "step": 82360
    },
    {
      "epoch": 23.380641498722678,
      "grad_norm": 3.564608335494995,
      "learning_rate": 7.662986091399375e-05,
      "loss": 0.01751369684934616,
      "step": 82370
    },
    {
      "epoch": 23.38347998864604,
      "grad_norm": 0.31712964177131653,
      "learning_rate": 7.66270224240704e-05,
      "loss": 0.008308090269565582,
      "step": 82380
    },
    {
      "epoch": 23.3863184785694,
      "grad_norm": 0.785385012626648,
      "learning_rate": 7.662418393414704e-05,
      "loss": 0.009592992812395095,
      "step": 82390
    },
    {
      "epoch": 23.389156968492763,
      "grad_norm": 6.185344219207764,
      "learning_rate": 7.662134544422368e-05,
      "loss": 0.01427169144153595,
      "step": 82400
    },
    {
      "epoch": 23.391995458416122,
      "grad_norm": 9.924135208129883,
      "learning_rate": 7.661850695430032e-05,
      "loss": 0.02664813995361328,
      "step": 82410
    },
    {
      "epoch": 23.394833948339482,
      "grad_norm": 7.843605041503906,
      "learning_rate": 7.661566846437696e-05,
      "loss": 0.01773083657026291,
      "step": 82420
    },
    {
      "epoch": 23.397672438262845,
      "grad_norm": 4.13249397277832,
      "learning_rate": 7.661282997445359e-05,
      "loss": 0.013245417177677155,
      "step": 82430
    },
    {
      "epoch": 23.400510928186204,
      "grad_norm": 6.57637357711792,
      "learning_rate": 7.660999148453023e-05,
      "loss": 0.007513901591300965,
      "step": 82440
    },
    {
      "epoch": 23.403349418109567,
      "grad_norm": 12.128868103027344,
      "learning_rate": 7.660715299460687e-05,
      "loss": 0.02394697517156601,
      "step": 82450
    },
    {
      "epoch": 23.406187908032926,
      "grad_norm": 0.25987422466278076,
      "learning_rate": 7.660431450468351e-05,
      "loss": 0.009269306808710099,
      "step": 82460
    },
    {
      "epoch": 23.409026397956286,
      "grad_norm": 5.1541666984558105,
      "learning_rate": 7.660147601476015e-05,
      "loss": 0.031182122230529786,
      "step": 82470
    },
    {
      "epoch": 23.41186488787965,
      "grad_norm": 0.9312660098075867,
      "learning_rate": 7.65986375248368e-05,
      "loss": 0.002501872181892395,
      "step": 82480
    },
    {
      "epoch": 23.414703377803008,
      "grad_norm": 4.570296764373779,
      "learning_rate": 7.659579903491342e-05,
      "loss": 0.02177284359931946,
      "step": 82490
    },
    {
      "epoch": 23.41754186772637,
      "grad_norm": 7.330686092376709,
      "learning_rate": 7.659296054499007e-05,
      "loss": 0.026332417130470277,
      "step": 82500
    },
    {
      "epoch": 23.41754186772637,
      "eval_accuracy": 0.9671901824887137,
      "eval_loss": 0.10547531396150589,
      "eval_runtime": 32.4629,
      "eval_samples_per_second": 484.46,
      "eval_steps_per_second": 7.578,
      "step": 82500
    },
    {
      "epoch": 23.42038035764973,
      "grad_norm": 3.181257486343384,
      "learning_rate": 7.659012205506671e-05,
      "loss": 0.029849618673324585,
      "step": 82510
    },
    {
      "epoch": 23.42321884757309,
      "grad_norm": 1.6969406604766846,
      "learning_rate": 7.658728356514335e-05,
      "loss": 0.010424987971782684,
      "step": 82520
    },
    {
      "epoch": 23.426057337496452,
      "grad_norm": 2.0591225624084473,
      "learning_rate": 7.658444507521999e-05,
      "loss": 0.00893688276410103,
      "step": 82530
    },
    {
      "epoch": 23.42889582741981,
      "grad_norm": 3.565380096435547,
      "learning_rate": 7.658160658529663e-05,
      "loss": 0.006332745403051376,
      "step": 82540
    },
    {
      "epoch": 23.431734317343174,
      "grad_norm": 1.5215063095092773,
      "learning_rate": 7.657876809537326e-05,
      "loss": 0.014168977737426758,
      "step": 82550
    },
    {
      "epoch": 23.434572807266534,
      "grad_norm": 0.7097628712654114,
      "learning_rate": 7.65759296054499e-05,
      "loss": 0.006420505046844482,
      "step": 82560
    },
    {
      "epoch": 23.437411297189897,
      "grad_norm": 0.589506208896637,
      "learning_rate": 7.657309111552654e-05,
      "loss": 0.008945582807064057,
      "step": 82570
    },
    {
      "epoch": 23.440249787113256,
      "grad_norm": 2.0901496410369873,
      "learning_rate": 7.657025262560318e-05,
      "loss": 0.006495766341686249,
      "step": 82580
    },
    {
      "epoch": 23.443088277036615,
      "grad_norm": 0.704718828201294,
      "learning_rate": 7.656741413567982e-05,
      "loss": 0.013205070793628693,
      "step": 82590
    },
    {
      "epoch": 23.44592676695998,
      "grad_norm": 8.177169799804688,
      "learning_rate": 7.656457564575647e-05,
      "loss": 0.009578696638345718,
      "step": 82600
    },
    {
      "epoch": 23.448765256883338,
      "grad_norm": 3.218440294265747,
      "learning_rate": 7.656173715583311e-05,
      "loss": 0.01101265922188759,
      "step": 82610
    },
    {
      "epoch": 23.4516037468067,
      "grad_norm": 6.081578731536865,
      "learning_rate": 7.655889866590973e-05,
      "loss": 0.023344169557094573,
      "step": 82620
    },
    {
      "epoch": 23.45444223673006,
      "grad_norm": 0.12123202532529831,
      "learning_rate": 7.655606017598638e-05,
      "loss": 0.010875825583934785,
      "step": 82630
    },
    {
      "epoch": 23.45728072665342,
      "grad_norm": 10.686674118041992,
      "learning_rate": 7.655322168606302e-05,
      "loss": 0.01307520568370819,
      "step": 82640
    },
    {
      "epoch": 23.460119216576782,
      "grad_norm": 8.951855659484863,
      "learning_rate": 7.655038319613965e-05,
      "loss": 0.019513271749019623,
      "step": 82650
    },
    {
      "epoch": 23.46295770650014,
      "grad_norm": 7.39168643951416,
      "learning_rate": 7.65475447062163e-05,
      "loss": 0.010536402463912964,
      "step": 82660
    },
    {
      "epoch": 23.465796196423504,
      "grad_norm": 1.263788104057312,
      "learning_rate": 7.654470621629294e-05,
      "loss": 0.015612158179283141,
      "step": 82670
    },
    {
      "epoch": 23.468634686346864,
      "grad_norm": 1.5506490468978882,
      "learning_rate": 7.654186772636957e-05,
      "loss": 0.006364137679338455,
      "step": 82680
    },
    {
      "epoch": 23.471473176270223,
      "grad_norm": 0.6718373894691467,
      "learning_rate": 7.653902923644621e-05,
      "loss": 0.00988832265138626,
      "step": 82690
    },
    {
      "epoch": 23.474311666193586,
      "grad_norm": 13.08192253112793,
      "learning_rate": 7.653619074652285e-05,
      "loss": 0.010516923666000367,
      "step": 82700
    },
    {
      "epoch": 23.477150156116945,
      "grad_norm": 0.6148474812507629,
      "learning_rate": 7.65333522565995e-05,
      "loss": 0.020795610547065736,
      "step": 82710
    },
    {
      "epoch": 23.479988646040308,
      "grad_norm": 1.006664514541626,
      "learning_rate": 7.653051376667614e-05,
      "loss": 0.017482875287532805,
      "step": 82720
    },
    {
      "epoch": 23.482827135963667,
      "grad_norm": 4.439058780670166,
      "learning_rate": 7.652767527675278e-05,
      "loss": 0.007084453850984574,
      "step": 82730
    },
    {
      "epoch": 23.485665625887027,
      "grad_norm": 11.166254997253418,
      "learning_rate": 7.652483678682942e-05,
      "loss": 0.036588022112846376,
      "step": 82740
    },
    {
      "epoch": 23.48850411581039,
      "grad_norm": 3.744107484817505,
      "learning_rate": 7.652199829690605e-05,
      "loss": 0.012581223249435424,
      "step": 82750
    },
    {
      "epoch": 23.49134260573375,
      "grad_norm": 3.978107213973999,
      "learning_rate": 7.651915980698269e-05,
      "loss": 0.01503506600856781,
      "step": 82760
    },
    {
      "epoch": 23.49418109565711,
      "grad_norm": 0.8000993728637695,
      "learning_rate": 7.651632131705933e-05,
      "loss": 0.006816352903842926,
      "step": 82770
    },
    {
      "epoch": 23.49701958558047,
      "grad_norm": 0.15485237538814545,
      "learning_rate": 7.651348282713596e-05,
      "loss": 0.0071754463016986845,
      "step": 82780
    },
    {
      "epoch": 23.49985807550383,
      "grad_norm": 4.076370716094971,
      "learning_rate": 7.651064433721261e-05,
      "loss": 0.01530846357345581,
      "step": 82790
    },
    {
      "epoch": 23.502696565427193,
      "grad_norm": 0.2965930104255676,
      "learning_rate": 7.650780584728925e-05,
      "loss": 0.009167153388261795,
      "step": 82800
    },
    {
      "epoch": 23.505535055350553,
      "grad_norm": 3.043975353240967,
      "learning_rate": 7.650496735736588e-05,
      "loss": 0.011548572778701782,
      "step": 82810
    },
    {
      "epoch": 23.508373545273916,
      "grad_norm": 10.682815551757812,
      "learning_rate": 7.650212886744252e-05,
      "loss": 0.02089568376541138,
      "step": 82820
    },
    {
      "epoch": 23.511212035197275,
      "grad_norm": 10.54460620880127,
      "learning_rate": 7.649929037751916e-05,
      "loss": 0.012385944277048111,
      "step": 82830
    },
    {
      "epoch": 23.514050525120634,
      "grad_norm": 5.183022499084473,
      "learning_rate": 7.64964518875958e-05,
      "loss": 0.009924705326557159,
      "step": 82840
    },
    {
      "epoch": 23.516889015043997,
      "grad_norm": 0.6913396716117859,
      "learning_rate": 7.649361339767243e-05,
      "loss": 0.010891494899988174,
      "step": 82850
    },
    {
      "epoch": 23.519727504967356,
      "grad_norm": 0.08964705467224121,
      "learning_rate": 7.649077490774909e-05,
      "loss": 0.00865996554493904,
      "step": 82860
    },
    {
      "epoch": 23.52256599489072,
      "grad_norm": 1.1144227981567383,
      "learning_rate": 7.648793641782573e-05,
      "loss": 0.009030211716890335,
      "step": 82870
    },
    {
      "epoch": 23.52540448481408,
      "grad_norm": 1.8332144021987915,
      "learning_rate": 7.648509792790236e-05,
      "loss": 0.007814111560583115,
      "step": 82880
    },
    {
      "epoch": 23.528242974737438,
      "grad_norm": 3.369371175765991,
      "learning_rate": 7.6482259437979e-05,
      "loss": 0.02521330714225769,
      "step": 82890
    },
    {
      "epoch": 23.5310814646608,
      "grad_norm": 13.359131813049316,
      "learning_rate": 7.647942094805564e-05,
      "loss": 0.009891249239444733,
      "step": 82900
    },
    {
      "epoch": 23.53391995458416,
      "grad_norm": 2.5486323833465576,
      "learning_rate": 7.647658245813227e-05,
      "loss": 0.02335374504327774,
      "step": 82910
    },
    {
      "epoch": 23.536758444507523,
      "grad_norm": 6.208741664886475,
      "learning_rate": 7.647374396820892e-05,
      "loss": 0.008053935319185256,
      "step": 82920
    },
    {
      "epoch": 23.539596934430882,
      "grad_norm": 0.5290101766586304,
      "learning_rate": 7.647090547828556e-05,
      "loss": 0.006270058453083038,
      "step": 82930
    },
    {
      "epoch": 23.542435424354245,
      "grad_norm": 6.628600120544434,
      "learning_rate": 7.646806698836219e-05,
      "loss": 0.01876838207244873,
      "step": 82940
    },
    {
      "epoch": 23.545273914277605,
      "grad_norm": 0.41733840107917786,
      "learning_rate": 7.646522849843883e-05,
      "loss": 0.006512662768363953,
      "step": 82950
    },
    {
      "epoch": 23.548112404200964,
      "grad_norm": 0.6664360761642456,
      "learning_rate": 7.646239000851547e-05,
      "loss": 0.013772030174732209,
      "step": 82960
    },
    {
      "epoch": 23.550950894124327,
      "grad_norm": 2.4472177028656006,
      "learning_rate": 7.645955151859212e-05,
      "loss": 0.006162246316671371,
      "step": 82970
    },
    {
      "epoch": 23.553789384047686,
      "grad_norm": 7.343294620513916,
      "learning_rate": 7.645671302866874e-05,
      "loss": 0.026304754614830016,
      "step": 82980
    },
    {
      "epoch": 23.55662787397105,
      "grad_norm": 0.9492427110671997,
      "learning_rate": 7.64538745387454e-05,
      "loss": 0.029026344418525696,
      "step": 82990
    },
    {
      "epoch": 23.55946636389441,
      "grad_norm": 2.043377161026001,
      "learning_rate": 7.645103604882204e-05,
      "loss": 0.014937584102153779,
      "step": 83000
    },
    {
      "epoch": 23.55946636389441,
      "eval_accuracy": 0.9701150887009601,
      "eval_loss": 0.09734461456537247,
      "eval_runtime": 31.7902,
      "eval_samples_per_second": 494.713,
      "eval_steps_per_second": 7.738,
      "step": 83000
    },
    {
      "epoch": 23.562304853817768,
      "grad_norm": 2.5440618991851807,
      "learning_rate": 7.644819755889867e-05,
      "loss": 0.006747973710298538,
      "step": 83010
    },
    {
      "epoch": 23.56514334374113,
      "grad_norm": 4.450223922729492,
      "learning_rate": 7.644535906897531e-05,
      "loss": 0.009841623902320861,
      "step": 83020
    },
    {
      "epoch": 23.56798183366449,
      "grad_norm": 8.121411323547363,
      "learning_rate": 7.644252057905195e-05,
      "loss": 0.019003406167030334,
      "step": 83030
    },
    {
      "epoch": 23.570820323587853,
      "grad_norm": 0.99057537317276,
      "learning_rate": 7.643968208912858e-05,
      "loss": 0.017102931439876557,
      "step": 83040
    },
    {
      "epoch": 23.573658813511212,
      "grad_norm": 6.358065605163574,
      "learning_rate": 7.643684359920522e-05,
      "loss": 0.0262486070394516,
      "step": 83050
    },
    {
      "epoch": 23.57649730343457,
      "grad_norm": 0.3345797657966614,
      "learning_rate": 7.643400510928187e-05,
      "loss": 0.019595248997211455,
      "step": 83060
    },
    {
      "epoch": 23.579335793357934,
      "grad_norm": 11.32795238494873,
      "learning_rate": 7.64311666193585e-05,
      "loss": 0.009392689168453216,
      "step": 83070
    },
    {
      "epoch": 23.582174283281294,
      "grad_norm": 5.22819709777832,
      "learning_rate": 7.642832812943514e-05,
      "loss": 0.012082602083683013,
      "step": 83080
    },
    {
      "epoch": 23.585012773204657,
      "grad_norm": 0.9010908603668213,
      "learning_rate": 7.642548963951178e-05,
      "loss": 0.01505519300699234,
      "step": 83090
    },
    {
      "epoch": 23.587851263128016,
      "grad_norm": 1.0775946378707886,
      "learning_rate": 7.642265114958843e-05,
      "loss": 0.007733301818370819,
      "step": 83100
    },
    {
      "epoch": 23.590689753051375,
      "grad_norm": 1.7868670225143433,
      "learning_rate": 7.641981265966505e-05,
      "loss": 0.009365884959697724,
      "step": 83110
    },
    {
      "epoch": 23.593528242974738,
      "grad_norm": 1.393751621246338,
      "learning_rate": 7.641697416974171e-05,
      "loss": 0.010777691006660461,
      "step": 83120
    },
    {
      "epoch": 23.596366732898097,
      "grad_norm": 0.1793908327817917,
      "learning_rate": 7.641413567981835e-05,
      "loss": 0.01588108390569687,
      "step": 83130
    },
    {
      "epoch": 23.59920522282146,
      "grad_norm": 2.337538480758667,
      "learning_rate": 7.641129718989498e-05,
      "loss": 0.010045894980430603,
      "step": 83140
    },
    {
      "epoch": 23.60204371274482,
      "grad_norm": 0.7882490754127502,
      "learning_rate": 7.640845869997162e-05,
      "loss": 0.007647889852523804,
      "step": 83150
    },
    {
      "epoch": 23.60488220266818,
      "grad_norm": 5.196844577789307,
      "learning_rate": 7.640562021004826e-05,
      "loss": 0.01271662414073944,
      "step": 83160
    },
    {
      "epoch": 23.607720692591542,
      "grad_norm": 2.0521433353424072,
      "learning_rate": 7.640278172012489e-05,
      "loss": 0.018307459354400635,
      "step": 83170
    },
    {
      "epoch": 23.6105591825149,
      "grad_norm": 14.771334648132324,
      "learning_rate": 7.639994323020153e-05,
      "loss": 0.034636908769607545,
      "step": 83180
    },
    {
      "epoch": 23.613397672438264,
      "grad_norm": 2.3762192726135254,
      "learning_rate": 7.639710474027818e-05,
      "loss": 0.013587671518325805,
      "step": 83190
    },
    {
      "epoch": 23.616236162361623,
      "grad_norm": 13.212779998779297,
      "learning_rate": 7.639426625035481e-05,
      "loss": 0.009491260349750518,
      "step": 83200
    },
    {
      "epoch": 23.619074652284983,
      "grad_norm": 3.265249013900757,
      "learning_rate": 7.639142776043145e-05,
      "loss": 0.017671296000480653,
      "step": 83210
    },
    {
      "epoch": 23.621913142208346,
      "grad_norm": 0.253572940826416,
      "learning_rate": 7.63885892705081e-05,
      "loss": 0.011944358795881271,
      "step": 83220
    },
    {
      "epoch": 23.624751632131705,
      "grad_norm": 10.773809432983398,
      "learning_rate": 7.638575078058474e-05,
      "loss": 0.020746909081935883,
      "step": 83230
    },
    {
      "epoch": 23.627590122055068,
      "grad_norm": 6.981840133666992,
      "learning_rate": 7.638291229066136e-05,
      "loss": 0.019993923604488373,
      "step": 83240
    },
    {
      "epoch": 23.630428611978427,
      "grad_norm": 0.41949746012687683,
      "learning_rate": 7.6380073800738e-05,
      "loss": 0.005101189762353897,
      "step": 83250
    },
    {
      "epoch": 23.633267101901787,
      "grad_norm": 2.4229013919830322,
      "learning_rate": 7.637723531081466e-05,
      "loss": 0.03231497406959534,
      "step": 83260
    },
    {
      "epoch": 23.63610559182515,
      "grad_norm": 4.9027862548828125,
      "learning_rate": 7.637439682089129e-05,
      "loss": 0.02451559603214264,
      "step": 83270
    },
    {
      "epoch": 23.63894408174851,
      "grad_norm": 0.41542407870292664,
      "learning_rate": 7.637155833096793e-05,
      "loss": 0.00812191590666771,
      "step": 83280
    },
    {
      "epoch": 23.64178257167187,
      "grad_norm": 3.698788642883301,
      "learning_rate": 7.636871984104457e-05,
      "loss": 0.003953716158866883,
      "step": 83290
    },
    {
      "epoch": 23.64462106159523,
      "grad_norm": 2.8746862411499023,
      "learning_rate": 7.63658813511212e-05,
      "loss": 0.00715617835521698,
      "step": 83300
    },
    {
      "epoch": 23.647459551518594,
      "grad_norm": 0.34903642535209656,
      "learning_rate": 7.636304286119784e-05,
      "loss": 0.0038349974900484087,
      "step": 83310
    },
    {
      "epoch": 23.650298041441953,
      "grad_norm": 0.8546062111854553,
      "learning_rate": 7.63602043712745e-05,
      "loss": 0.0067264877259731294,
      "step": 83320
    },
    {
      "epoch": 23.653136531365313,
      "grad_norm": 3.083432912826538,
      "learning_rate": 7.635736588135112e-05,
      "loss": 0.00889369398355484,
      "step": 83330
    },
    {
      "epoch": 23.655975021288675,
      "grad_norm": 4.731603145599365,
      "learning_rate": 7.635452739142776e-05,
      "loss": 0.020127904415130616,
      "step": 83340
    },
    {
      "epoch": 23.658813511212035,
      "grad_norm": 1.9799318313598633,
      "learning_rate": 7.63516889015044e-05,
      "loss": 0.008436944335699081,
      "step": 83350
    },
    {
      "epoch": 23.661652001135398,
      "grad_norm": 7.74187707901001,
      "learning_rate": 7.634885041158105e-05,
      "loss": 0.008479569852352143,
      "step": 83360
    },
    {
      "epoch": 23.664490491058757,
      "grad_norm": 0.1893157809972763,
      "learning_rate": 7.634601192165768e-05,
      "loss": 0.0064016193151474,
      "step": 83370
    },
    {
      "epoch": 23.667328980982116,
      "grad_norm": 1.0991312265396118,
      "learning_rate": 7.634317343173432e-05,
      "loss": 0.007808512449264527,
      "step": 83380
    },
    {
      "epoch": 23.67016747090548,
      "grad_norm": 0.1031964123249054,
      "learning_rate": 7.634033494181096e-05,
      "loss": 0.00815063640475273,
      "step": 83390
    },
    {
      "epoch": 23.67300596082884,
      "grad_norm": 14.569587707519531,
      "learning_rate": 7.63374964518876e-05,
      "loss": 0.01709076464176178,
      "step": 83400
    },
    {
      "epoch": 23.6758444507522,
      "grad_norm": 6.105042457580566,
      "learning_rate": 7.633465796196424e-05,
      "loss": 0.007482413202524185,
      "step": 83410
    },
    {
      "epoch": 23.67868294067556,
      "grad_norm": 8.330972671508789,
      "learning_rate": 7.633181947204088e-05,
      "loss": 0.01764088124036789,
      "step": 83420
    },
    {
      "epoch": 23.68152143059892,
      "grad_norm": 7.741170883178711,
      "learning_rate": 7.632898098211751e-05,
      "loss": 0.014732329547405243,
      "step": 83430
    },
    {
      "epoch": 23.684359920522283,
      "grad_norm": 3.4004404544830322,
      "learning_rate": 7.632614249219415e-05,
      "loss": 0.016580626368522644,
      "step": 83440
    },
    {
      "epoch": 23.687198410445642,
      "grad_norm": 1.7495700120925903,
      "learning_rate": 7.632330400227079e-05,
      "loss": 0.008380424231290817,
      "step": 83450
    },
    {
      "epoch": 23.690036900369005,
      "grad_norm": 4.892657279968262,
      "learning_rate": 7.632046551234743e-05,
      "loss": 0.014952945709228515,
      "step": 83460
    },
    {
      "epoch": 23.692875390292365,
      "grad_norm": 1.9817546606063843,
      "learning_rate": 7.631762702242408e-05,
      "loss": 0.011090509593486786,
      "step": 83470
    },
    {
      "epoch": 23.695713880215724,
      "grad_norm": 0.4966922998428345,
      "learning_rate": 7.631478853250072e-05,
      "loss": 0.015612806379795074,
      "step": 83480
    },
    {
      "epoch": 23.698552370139087,
      "grad_norm": 3.459414482116699,
      "learning_rate": 7.631195004257734e-05,
      "loss": 0.016526661813259125,
      "step": 83490
    },
    {
      "epoch": 23.701390860062446,
      "grad_norm": 0.08895218372344971,
      "learning_rate": 7.630911155265399e-05,
      "loss": 0.003358781337738037,
      "step": 83500
    },
    {
      "epoch": 23.701390860062446,
      "eval_accuracy": 0.9682075411712342,
      "eval_loss": 0.09861993044614792,
      "eval_runtime": 32.4496,
      "eval_samples_per_second": 484.66,
      "eval_steps_per_second": 7.581,
      "step": 83500
    },
    {
      "epoch": 23.70422934998581,
      "grad_norm": 12.85888671875,
      "learning_rate": 7.630627306273063e-05,
      "loss": 0.012155049294233323,
      "step": 83510
    },
    {
      "epoch": 23.70706783990917,
      "grad_norm": 2.528745174407959,
      "learning_rate": 7.630343457280727e-05,
      "loss": 0.007482944428920746,
      "step": 83520
    },
    {
      "epoch": 23.709906329832528,
      "grad_norm": 2.1727800369262695,
      "learning_rate": 7.630059608288391e-05,
      "loss": 0.004269128292798996,
      "step": 83530
    },
    {
      "epoch": 23.71274481975589,
      "grad_norm": 0.17869335412979126,
      "learning_rate": 7.629775759296055e-05,
      "loss": 0.01526753455400467,
      "step": 83540
    },
    {
      "epoch": 23.71558330967925,
      "grad_norm": 12.845808982849121,
      "learning_rate": 7.629491910303719e-05,
      "loss": 0.012925630807876587,
      "step": 83550
    },
    {
      "epoch": 23.718421799602613,
      "grad_norm": 0.3440016210079193,
      "learning_rate": 7.629208061311382e-05,
      "loss": 0.019649922847747803,
      "step": 83560
    },
    {
      "epoch": 23.721260289525972,
      "grad_norm": 2.505288600921631,
      "learning_rate": 7.628924212319046e-05,
      "loss": 0.016786137223243715,
      "step": 83570
    },
    {
      "epoch": 23.72409877944933,
      "grad_norm": 1.8294084072113037,
      "learning_rate": 7.62864036332671e-05,
      "loss": 0.013619786500930786,
      "step": 83580
    },
    {
      "epoch": 23.726937269372694,
      "grad_norm": 2.873342752456665,
      "learning_rate": 7.628356514334374e-05,
      "loss": 0.019265981018543245,
      "step": 83590
    },
    {
      "epoch": 23.729775759296054,
      "grad_norm": 0.1916784644126892,
      "learning_rate": 7.628072665342039e-05,
      "loss": 0.031936776638031,
      "step": 83600
    },
    {
      "epoch": 23.732614249219417,
      "grad_norm": 4.064939022064209,
      "learning_rate": 7.627788816349703e-05,
      "loss": 0.008780252933502198,
      "step": 83610
    },
    {
      "epoch": 23.735452739142776,
      "grad_norm": 0.621260941028595,
      "learning_rate": 7.627504967357366e-05,
      "loss": 0.015223242342472076,
      "step": 83620
    },
    {
      "epoch": 23.738291229066135,
      "grad_norm": 0.5598052144050598,
      "learning_rate": 7.62722111836503e-05,
      "loss": 0.007516104727983475,
      "step": 83630
    },
    {
      "epoch": 23.741129718989498,
      "grad_norm": 0.9211838245391846,
      "learning_rate": 7.626937269372694e-05,
      "loss": 0.00517229363322258,
      "step": 83640
    },
    {
      "epoch": 23.743968208912857,
      "grad_norm": 4.200852394104004,
      "learning_rate": 7.626653420380358e-05,
      "loss": 0.005148629099130631,
      "step": 83650
    },
    {
      "epoch": 23.74680669883622,
      "grad_norm": 5.853733062744141,
      "learning_rate": 7.626369571388022e-05,
      "loss": 0.015529228746891022,
      "step": 83660
    },
    {
      "epoch": 23.74964518875958,
      "grad_norm": 8.323141098022461,
      "learning_rate": 7.626085722395686e-05,
      "loss": 0.01382160782814026,
      "step": 83670
    },
    {
      "epoch": 23.75248367868294,
      "grad_norm": 2.699948787689209,
      "learning_rate": 7.62580187340335e-05,
      "loss": 0.020977148413658143,
      "step": 83680
    },
    {
      "epoch": 23.755322168606302,
      "grad_norm": 0.8940754532814026,
      "learning_rate": 7.625518024411013e-05,
      "loss": 0.02008947730064392,
      "step": 83690
    },
    {
      "epoch": 23.75816065852966,
      "grad_norm": 10.38927936553955,
      "learning_rate": 7.625234175418677e-05,
      "loss": 0.012113240361213685,
      "step": 83700
    },
    {
      "epoch": 23.760999148453024,
      "grad_norm": 2.431267499923706,
      "learning_rate": 7.624950326426341e-05,
      "loss": 0.004374124109745026,
      "step": 83710
    },
    {
      "epoch": 23.763837638376383,
      "grad_norm": 2.9984331130981445,
      "learning_rate": 7.624666477434006e-05,
      "loss": 0.014554980397224426,
      "step": 83720
    },
    {
      "epoch": 23.766676128299743,
      "grad_norm": 4.978708744049072,
      "learning_rate": 7.62438262844167e-05,
      "loss": 0.015243619680404663,
      "step": 83730
    },
    {
      "epoch": 23.769514618223106,
      "grad_norm": 0.329736590385437,
      "learning_rate": 7.624098779449334e-05,
      "loss": 0.01855323016643524,
      "step": 83740
    },
    {
      "epoch": 23.772353108146465,
      "grad_norm": 12.14116382598877,
      "learning_rate": 7.623814930456997e-05,
      "loss": 0.03478938341140747,
      "step": 83750
    },
    {
      "epoch": 23.775191598069828,
      "grad_norm": 2.0323081016540527,
      "learning_rate": 7.623531081464661e-05,
      "loss": 0.011597354710102082,
      "step": 83760
    },
    {
      "epoch": 23.778030087993187,
      "grad_norm": 2.683112382888794,
      "learning_rate": 7.623275617371559e-05,
      "loss": 0.015265840291976928,
      "step": 83770
    },
    {
      "epoch": 23.78086857791655,
      "grad_norm": 1.1147269010543823,
      "learning_rate": 7.622991768379223e-05,
      "loss": 0.006794743984937668,
      "step": 83780
    },
    {
      "epoch": 23.78370706783991,
      "grad_norm": 4.140469074249268,
      "learning_rate": 7.622707919386887e-05,
      "loss": 0.0138153538107872,
      "step": 83790
    },
    {
      "epoch": 23.78654555776327,
      "grad_norm": 0.9724971055984497,
      "learning_rate": 7.62242407039455e-05,
      "loss": 0.012181229144334792,
      "step": 83800
    },
    {
      "epoch": 23.78938404768663,
      "grad_norm": 1.8344225883483887,
      "learning_rate": 7.622140221402214e-05,
      "loss": 0.017166543006896972,
      "step": 83810
    },
    {
      "epoch": 23.79222253760999,
      "grad_norm": 4.065749645233154,
      "learning_rate": 7.621856372409878e-05,
      "loss": 0.025070038437843323,
      "step": 83820
    },
    {
      "epoch": 23.795061027533354,
      "grad_norm": 0.6657940745353699,
      "learning_rate": 7.621572523417542e-05,
      "loss": 0.0223751038312912,
      "step": 83830
    },
    {
      "epoch": 23.797899517456713,
      "grad_norm": 14.120619773864746,
      "learning_rate": 7.621288674425206e-05,
      "loss": 0.02350504994392395,
      "step": 83840
    },
    {
      "epoch": 23.800738007380073,
      "grad_norm": 1.6944475173950195,
      "learning_rate": 7.62100482543287e-05,
      "loss": 0.014007046818733215,
      "step": 83850
    },
    {
      "epoch": 23.803576497303435,
      "grad_norm": 6.3545732498168945,
      "learning_rate": 7.620720976440535e-05,
      "loss": 0.014083229005336761,
      "step": 83860
    },
    {
      "epoch": 23.806414987226795,
      "grad_norm": 0.9429182410240173,
      "learning_rate": 7.620437127448197e-05,
      "loss": 0.00722980797290802,
      "step": 83870
    },
    {
      "epoch": 23.809253477150158,
      "grad_norm": 1.3470393419265747,
      "learning_rate": 7.620153278455862e-05,
      "loss": 0.015278784930706025,
      "step": 83880
    },
    {
      "epoch": 23.812091967073517,
      "grad_norm": 3.320622205734253,
      "learning_rate": 7.619869429463526e-05,
      "loss": 0.004866759479045868,
      "step": 83890
    },
    {
      "epoch": 23.814930456996876,
      "grad_norm": 0.35066312551498413,
      "learning_rate": 7.61958558047119e-05,
      "loss": 0.009989409148693085,
      "step": 83900
    },
    {
      "epoch": 23.81776894692024,
      "grad_norm": 0.5345789194107056,
      "learning_rate": 7.619301731478854e-05,
      "loss": 0.003192286565899849,
      "step": 83910
    },
    {
      "epoch": 23.8206074368436,
      "grad_norm": 2.0772011280059814,
      "learning_rate": 7.619017882486518e-05,
      "loss": 0.00603991262614727,
      "step": 83920
    },
    {
      "epoch": 23.82344592676696,
      "grad_norm": 12.612524032592773,
      "learning_rate": 7.618734033494181e-05,
      "loss": 0.018510225415229797,
      "step": 83930
    },
    {
      "epoch": 23.82628441669032,
      "grad_norm": 1.5009267330169678,
      "learning_rate": 7.618450184501845e-05,
      "loss": 0.007368786633014679,
      "step": 83940
    },
    {
      "epoch": 23.82912290661368,
      "grad_norm": 2.1319992542266846,
      "learning_rate": 7.618166335509509e-05,
      "loss": 0.02300973832607269,
      "step": 83950
    },
    {
      "epoch": 23.831961396537043,
      "grad_norm": 4.3605055809021,
      "learning_rate": 7.617882486517173e-05,
      "loss": 0.007141160219907761,
      "step": 83960
    },
    {
      "epoch": 23.834799886460402,
      "grad_norm": 0.17979754507541656,
      "learning_rate": 7.617598637524837e-05,
      "loss": 0.013762891292572021,
      "step": 83970
    },
    {
      "epoch": 23.837638376383765,
      "grad_norm": 1.785412311553955,
      "learning_rate": 7.617314788532502e-05,
      "loss": 0.01592242568731308,
      "step": 83980
    },
    {
      "epoch": 23.840476866307124,
      "grad_norm": 0.3671169877052307,
      "learning_rate": 7.617030939540166e-05,
      "loss": 0.013170200586318969,
      "step": 83990
    },
    {
      "epoch": 23.843315356230484,
      "grad_norm": 6.230311870574951,
      "learning_rate": 7.616747090547829e-05,
      "loss": 0.022347329556941985,
      "step": 84000
    },
    {
      "epoch": 23.843315356230484,
      "eval_accuracy": 0.965791314300248,
      "eval_loss": 0.11082004010677338,
      "eval_runtime": 32.7698,
      "eval_samples_per_second": 479.924,
      "eval_steps_per_second": 7.507,
      "step": 84000
    },
    {
      "epoch": 23.846153846153847,
      "grad_norm": 0.759524405002594,
      "learning_rate": 7.616463241555493e-05,
      "loss": 0.008841366320848466,
      "step": 84010
    },
    {
      "epoch": 23.848992336077206,
      "grad_norm": 0.2547348141670227,
      "learning_rate": 7.616179392563157e-05,
      "loss": 0.004966495186090469,
      "step": 84020
    },
    {
      "epoch": 23.85183082600057,
      "grad_norm": 0.886723518371582,
      "learning_rate": 7.61589554357082e-05,
      "loss": 0.005022808164358139,
      "step": 84030
    },
    {
      "epoch": 23.85466931592393,
      "grad_norm": 0.229450523853302,
      "learning_rate": 7.615611694578485e-05,
      "loss": 0.01749954968690872,
      "step": 84040
    },
    {
      "epoch": 23.857507805847288,
      "grad_norm": 1.2417163848876953,
      "learning_rate": 7.615327845586149e-05,
      "loss": 0.004567822068929672,
      "step": 84050
    },
    {
      "epoch": 23.86034629577065,
      "grad_norm": 3.470907688140869,
      "learning_rate": 7.615043996593812e-05,
      "loss": 0.020869632065296174,
      "step": 84060
    },
    {
      "epoch": 23.86318478569401,
      "grad_norm": 0.49826672673225403,
      "learning_rate": 7.614760147601476e-05,
      "loss": 0.008900497853755952,
      "step": 84070
    },
    {
      "epoch": 23.866023275617373,
      "grad_norm": 2.1810293197631836,
      "learning_rate": 7.61447629860914e-05,
      "loss": 0.011838518828153611,
      "step": 84080
    },
    {
      "epoch": 23.868861765540732,
      "grad_norm": 6.328418731689453,
      "learning_rate": 7.614192449616804e-05,
      "loss": 0.011463573575019837,
      "step": 84090
    },
    {
      "epoch": 23.87170025546409,
      "grad_norm": 1.2754799127578735,
      "learning_rate": 7.613908600624469e-05,
      "loss": 0.01238202080130577,
      "step": 84100
    },
    {
      "epoch": 23.874538745387454,
      "grad_norm": 1.2204383611679077,
      "learning_rate": 7.613624751632133e-05,
      "loss": 0.010947038233280183,
      "step": 84110
    },
    {
      "epoch": 23.877377235310814,
      "grad_norm": 0.5982845425605774,
      "learning_rate": 7.613340902639797e-05,
      "loss": 0.018706832826137543,
      "step": 84120
    },
    {
      "epoch": 23.880215725234176,
      "grad_norm": 10.973430633544922,
      "learning_rate": 7.61305705364746e-05,
      "loss": 0.03058081567287445,
      "step": 84130
    },
    {
      "epoch": 23.883054215157536,
      "grad_norm": 7.948242664337158,
      "learning_rate": 7.612773204655124e-05,
      "loss": 0.0066550873219966885,
      "step": 84140
    },
    {
      "epoch": 23.8858927050809,
      "grad_norm": 1.1548420190811157,
      "learning_rate": 7.612489355662788e-05,
      "loss": 0.010352612286806107,
      "step": 84150
    },
    {
      "epoch": 23.888731195004258,
      "grad_norm": 1.1410589218139648,
      "learning_rate": 7.61220550667045e-05,
      "loss": 0.005702554434537888,
      "step": 84160
    },
    {
      "epoch": 23.891569684927617,
      "grad_norm": 8.575799942016602,
      "learning_rate": 7.611921657678116e-05,
      "loss": 0.024443334341049193,
      "step": 84170
    },
    {
      "epoch": 23.89440817485098,
      "grad_norm": 1.1583493947982788,
      "learning_rate": 7.61163780868578e-05,
      "loss": 0.0034597612917423247,
      "step": 84180
    },
    {
      "epoch": 23.89724666477434,
      "grad_norm": 9.375972747802734,
      "learning_rate": 7.611353959693443e-05,
      "loss": 0.009630904346704484,
      "step": 84190
    },
    {
      "epoch": 23.900085154697702,
      "grad_norm": 3.15425968170166,
      "learning_rate": 7.611070110701107e-05,
      "loss": 0.018268083035945893,
      "step": 84200
    },
    {
      "epoch": 23.902923644621062,
      "grad_norm": 0.2165924459695816,
      "learning_rate": 7.610786261708771e-05,
      "loss": 0.009226091951131821,
      "step": 84210
    },
    {
      "epoch": 23.90576213454442,
      "grad_norm": 0.3494337797164917,
      "learning_rate": 7.610502412716435e-05,
      "loss": 0.008549485355615616,
      "step": 84220
    },
    {
      "epoch": 23.908600624467784,
      "grad_norm": 2.8992931842803955,
      "learning_rate": 7.6102185637241e-05,
      "loss": 0.009287688881158829,
      "step": 84230
    },
    {
      "epoch": 23.911439114391143,
      "grad_norm": 4.939039707183838,
      "learning_rate": 7.609934714731764e-05,
      "loss": 0.019577112793922425,
      "step": 84240
    },
    {
      "epoch": 23.914277604314506,
      "grad_norm": 1.534653902053833,
      "learning_rate": 7.609650865739428e-05,
      "loss": 0.01008881777524948,
      "step": 84250
    },
    {
      "epoch": 23.917116094237866,
      "grad_norm": 3.809917449951172,
      "learning_rate": 7.60936701674709e-05,
      "loss": 0.02685166001319885,
      "step": 84260
    },
    {
      "epoch": 23.919954584161225,
      "grad_norm": 14.699606895446777,
      "learning_rate": 7.609083167754755e-05,
      "loss": 0.02274409234523773,
      "step": 84270
    },
    {
      "epoch": 23.922793074084588,
      "grad_norm": 14.207786560058594,
      "learning_rate": 7.608799318762419e-05,
      "loss": 0.03492521643638611,
      "step": 84280
    },
    {
      "epoch": 23.925631564007947,
      "grad_norm": 1.874414324760437,
      "learning_rate": 7.608515469770082e-05,
      "loss": 0.009038679301738739,
      "step": 84290
    },
    {
      "epoch": 23.92847005393131,
      "grad_norm": 13.659541130065918,
      "learning_rate": 7.608231620777747e-05,
      "loss": 0.011327023059129715,
      "step": 84300
    },
    {
      "epoch": 23.93130854385467,
      "grad_norm": 7.141719341278076,
      "learning_rate": 7.607947771785411e-05,
      "loss": 0.008892245590686798,
      "step": 84310
    },
    {
      "epoch": 23.93414703377803,
      "grad_norm": 1.0897291898727417,
      "learning_rate": 7.607663922793074e-05,
      "loss": 0.01939873993396759,
      "step": 84320
    },
    {
      "epoch": 23.93698552370139,
      "grad_norm": 9.336360931396484,
      "learning_rate": 7.607380073800738e-05,
      "loss": 0.012104091048240662,
      "step": 84330
    },
    {
      "epoch": 23.93982401362475,
      "grad_norm": 7.535465717315674,
      "learning_rate": 7.607096224808402e-05,
      "loss": 0.01173306182026863,
      "step": 84340
    },
    {
      "epoch": 23.942662503548114,
      "grad_norm": 1.202559471130371,
      "learning_rate": 7.606812375816067e-05,
      "loss": 0.03088620901107788,
      "step": 84350
    },
    {
      "epoch": 23.945500993471473,
      "grad_norm": 3.900709390640259,
      "learning_rate": 7.606528526823729e-05,
      "loss": 0.03329380750656128,
      "step": 84360
    },
    {
      "epoch": 23.948339483394832,
      "grad_norm": 7.0889716148376465,
      "learning_rate": 7.606244677831395e-05,
      "loss": 0.007630067318677903,
      "step": 84370
    },
    {
      "epoch": 23.951177973318195,
      "grad_norm": 0.25361502170562744,
      "learning_rate": 7.605960828839058e-05,
      "loss": 0.01660737246274948,
      "step": 84380
    },
    {
      "epoch": 23.954016463241555,
      "grad_norm": 0.27367648482322693,
      "learning_rate": 7.605676979846722e-05,
      "loss": 0.018328465521335602,
      "step": 84390
    },
    {
      "epoch": 23.956854953164918,
      "grad_norm": 2.3243510723114014,
      "learning_rate": 7.605393130854386e-05,
      "loss": 0.013505256175994873,
      "step": 84400
    },
    {
      "epoch": 23.959693443088277,
      "grad_norm": 9.808854103088379,
      "learning_rate": 7.60510928186205e-05,
      "loss": 0.027745193243026732,
      "step": 84410
    },
    {
      "epoch": 23.962531933011636,
      "grad_norm": 1.3662803173065186,
      "learning_rate": 7.604825432869713e-05,
      "loss": 0.00739012211561203,
      "step": 84420
    },
    {
      "epoch": 23.965370422935,
      "grad_norm": 5.0183281898498535,
      "learning_rate": 7.604541583877378e-05,
      "loss": 0.013295121490955353,
      "step": 84430
    },
    {
      "epoch": 23.96820891285836,
      "grad_norm": 0.8835544586181641,
      "learning_rate": 7.604257734885042e-05,
      "loss": 0.02955119013786316,
      "step": 84440
    },
    {
      "epoch": 23.97104740278172,
      "grad_norm": 13.038615226745605,
      "learning_rate": 7.603973885892705e-05,
      "loss": 0.016930481791496275,
      "step": 84450
    },
    {
      "epoch": 23.97388589270508,
      "grad_norm": 0.285456120967865,
      "learning_rate": 7.60369003690037e-05,
      "loss": 0.010375700145959853,
      "step": 84460
    },
    {
      "epoch": 23.97672438262844,
      "grad_norm": 0.826026976108551,
      "learning_rate": 7.603406187908033e-05,
      "loss": 0.005427590012550354,
      "step": 84470
    },
    {
      "epoch": 23.979562872551803,
      "grad_norm": 0.5882053971290588,
      "learning_rate": 7.603122338915696e-05,
      "loss": 0.031773269176483154,
      "step": 84480
    },
    {
      "epoch": 23.982401362475162,
      "grad_norm": 0.34798431396484375,
      "learning_rate": 7.60283848992336e-05,
      "loss": 0.017507794499397277,
      "step": 84490
    },
    {
      "epoch": 23.985239852398525,
      "grad_norm": 18.65087890625,
      "learning_rate": 7.602554640931026e-05,
      "loss": 0.010562290251255036,
      "step": 84500
    },
    {
      "epoch": 23.985239852398525,
      "eval_accuracy": 0.9675081070770013,
      "eval_loss": 0.10572259873151779,
      "eval_runtime": 31.9737,
      "eval_samples_per_second": 491.874,
      "eval_steps_per_second": 7.694,
      "step": 84500
    },
    {
      "epoch": 23.988078342321884,
      "grad_norm": 1.9048060178756714,
      "learning_rate": 7.602270791938689e-05,
      "loss": 0.021375086903572083,
      "step": 84510
    },
    {
      "epoch": 23.990916832245247,
      "grad_norm": 0.3980805575847626,
      "learning_rate": 7.601986942946353e-05,
      "loss": 0.026813524961471557,
      "step": 84520
    },
    {
      "epoch": 23.993755322168607,
      "grad_norm": 2.8301875591278076,
      "learning_rate": 7.601703093954017e-05,
      "loss": 0.012315025180578231,
      "step": 84530
    },
    {
      "epoch": 23.996593812091966,
      "grad_norm": 0.46884873509407043,
      "learning_rate": 7.601419244961681e-05,
      "loss": 0.021125704050064087,
      "step": 84540
    },
    {
      "epoch": 23.99943230201533,
      "grad_norm": 3.358616352081299,
      "learning_rate": 7.601135395969344e-05,
      "loss": 0.013842548429965972,
      "step": 84550
    },
    {
      "epoch": 24.002270791938688,
      "grad_norm": 2.1204473972320557,
      "learning_rate": 7.600851546977008e-05,
      "loss": 0.013447676599025727,
      "step": 84560
    },
    {
      "epoch": 24.00510928186205,
      "grad_norm": 0.10345333814620972,
      "learning_rate": 7.600567697984673e-05,
      "loss": 0.01847100257873535,
      "step": 84570
    },
    {
      "epoch": 24.00794777178541,
      "grad_norm": 0.26038622856140137,
      "learning_rate": 7.600283848992336e-05,
      "loss": 0.012046968936920166,
      "step": 84580
    },
    {
      "epoch": 24.01078626170877,
      "grad_norm": 3.4501569271087646,
      "learning_rate": 7.6e-05,
      "loss": 0.010232838988304137,
      "step": 84590
    },
    {
      "epoch": 24.013624751632133,
      "grad_norm": 0.2959158718585968,
      "learning_rate": 7.599716151007665e-05,
      "loss": 0.004404805973172188,
      "step": 84600
    },
    {
      "epoch": 24.016463241555492,
      "grad_norm": 0.6889228224754333,
      "learning_rate": 7.599432302015327e-05,
      "loss": 0.014449702203273773,
      "step": 84610
    },
    {
      "epoch": 24.019301731478855,
      "grad_norm": 0.9879194498062134,
      "learning_rate": 7.599148453022991e-05,
      "loss": 0.005400035530328751,
      "step": 84620
    },
    {
      "epoch": 24.022140221402214,
      "grad_norm": 3.965278148651123,
      "learning_rate": 7.598864604030657e-05,
      "loss": 0.026307129859924318,
      "step": 84630
    },
    {
      "epoch": 24.024978711325573,
      "grad_norm": 2.2091922760009766,
      "learning_rate": 7.59858075503832e-05,
      "loss": 0.01141185313463211,
      "step": 84640
    },
    {
      "epoch": 24.027817201248936,
      "grad_norm": 0.5622738003730774,
      "learning_rate": 7.598296906045984e-05,
      "loss": 0.006513230502605438,
      "step": 84650
    },
    {
      "epoch": 24.030655691172296,
      "grad_norm": 0.21021191775798798,
      "learning_rate": 7.598013057053648e-05,
      "loss": 0.0035309791564941406,
      "step": 84660
    },
    {
      "epoch": 24.03349418109566,
      "grad_norm": 2.49345064163208,
      "learning_rate": 7.597729208061312e-05,
      "loss": 0.007036945223808289,
      "step": 84670
    },
    {
      "epoch": 24.036332671019018,
      "grad_norm": 9.031028747558594,
      "learning_rate": 7.597445359068975e-05,
      "loss": 0.008424460887908936,
      "step": 84680
    },
    {
      "epoch": 24.039171160942377,
      "grad_norm": 3.8753223419189453,
      "learning_rate": 7.597161510076639e-05,
      "loss": 0.012846311926841736,
      "step": 84690
    },
    {
      "epoch": 24.04200965086574,
      "grad_norm": 3.0023934841156006,
      "learning_rate": 7.596877661084305e-05,
      "loss": 0.007488427311182022,
      "step": 84700
    },
    {
      "epoch": 24.0448481407891,
      "grad_norm": 8.386985778808594,
      "learning_rate": 7.596593812091967e-05,
      "loss": 0.019355453550815582,
      "step": 84710
    },
    {
      "epoch": 24.047686630712462,
      "grad_norm": 0.48925352096557617,
      "learning_rate": 7.596309963099631e-05,
      "loss": 0.0070985168218612674,
      "step": 84720
    },
    {
      "epoch": 24.05052512063582,
      "grad_norm": 5.458119869232178,
      "learning_rate": 7.596026114107296e-05,
      "loss": 0.009399396181106568,
      "step": 84730
    },
    {
      "epoch": 24.05336361055918,
      "grad_norm": 3.70332670211792,
      "learning_rate": 7.595742265114958e-05,
      "loss": 0.017363426089286805,
      "step": 84740
    },
    {
      "epoch": 24.056202100482544,
      "grad_norm": 5.7898430824279785,
      "learning_rate": 7.595458416122623e-05,
      "loss": 0.02179422527551651,
      "step": 84750
    },
    {
      "epoch": 24.059040590405903,
      "grad_norm": 0.608582615852356,
      "learning_rate": 7.595174567130287e-05,
      "loss": 0.009582144767045974,
      "step": 84760
    },
    {
      "epoch": 24.061879080329266,
      "grad_norm": 0.713959276676178,
      "learning_rate": 7.594890718137951e-05,
      "loss": 0.005781042575836182,
      "step": 84770
    },
    {
      "epoch": 24.064717570252625,
      "grad_norm": 14.982464790344238,
      "learning_rate": 7.594606869145615e-05,
      "loss": 0.010399052500724792,
      "step": 84780
    },
    {
      "epoch": 24.067556060175985,
      "grad_norm": 0.11601941287517548,
      "learning_rate": 7.594323020153279e-05,
      "loss": 0.0038609974086284637,
      "step": 84790
    },
    {
      "epoch": 24.070394550099348,
      "grad_norm": 0.48062312602996826,
      "learning_rate": 7.594039171160943e-05,
      "loss": 0.00273037888109684,
      "step": 84800
    },
    {
      "epoch": 24.073233040022707,
      "grad_norm": 3.332939386367798,
      "learning_rate": 7.593755322168606e-05,
      "loss": 0.006208663806319237,
      "step": 84810
    },
    {
      "epoch": 24.07607152994607,
      "grad_norm": 5.165525436401367,
      "learning_rate": 7.59347147317627e-05,
      "loss": 0.0037791263312101364,
      "step": 84820
    },
    {
      "epoch": 24.07891001986943,
      "grad_norm": 1.2610677480697632,
      "learning_rate": 7.593187624183936e-05,
      "loss": 0.006203094124794006,
      "step": 84830
    },
    {
      "epoch": 24.08174850979279,
      "grad_norm": 0.13150855898857117,
      "learning_rate": 7.592903775191598e-05,
      "loss": 0.0036239050328731536,
      "step": 84840
    },
    {
      "epoch": 24.08458699971615,
      "grad_norm": 0.5227026343345642,
      "learning_rate": 7.592619926199263e-05,
      "loss": 0.007079766690731048,
      "step": 84850
    },
    {
      "epoch": 24.08742548963951,
      "grad_norm": 8.103550910949707,
      "learning_rate": 7.592336077206927e-05,
      "loss": 0.008645490556955338,
      "step": 84860
    },
    {
      "epoch": 24.090263979562874,
      "grad_norm": 0.18947942554950714,
      "learning_rate": 7.59205222821459e-05,
      "loss": 0.011797967553138732,
      "step": 84870
    },
    {
      "epoch": 24.093102469486233,
      "grad_norm": 0.38045382499694824,
      "learning_rate": 7.591768379222254e-05,
      "loss": 0.011091934144496917,
      "step": 84880
    },
    {
      "epoch": 24.095940959409592,
      "grad_norm": 1.7371768951416016,
      "learning_rate": 7.591484530229918e-05,
      "loss": 0.010108402371406556,
      "step": 84890
    },
    {
      "epoch": 24.098779449332955,
      "grad_norm": 3.0963540077209473,
      "learning_rate": 7.591200681237582e-05,
      "loss": 0.004754865169525146,
      "step": 84900
    },
    {
      "epoch": 24.101617939256315,
      "grad_norm": 0.43404924869537354,
      "learning_rate": 7.590916832245246e-05,
      "loss": 0.004050199687480926,
      "step": 84910
    },
    {
      "epoch": 24.104456429179677,
      "grad_norm": 0.383226603269577,
      "learning_rate": 7.59063298325291e-05,
      "loss": 0.005064722150564194,
      "step": 84920
    },
    {
      "epoch": 24.107294919103037,
      "grad_norm": 5.937305450439453,
      "learning_rate": 7.590349134260574e-05,
      "loss": 0.01660972833633423,
      "step": 84930
    },
    {
      "epoch": 24.1101334090264,
      "grad_norm": 6.189237117767334,
      "learning_rate": 7.590065285268237e-05,
      "loss": 0.005902718007564545,
      "step": 84940
    },
    {
      "epoch": 24.11297189894976,
      "grad_norm": 3.1979804039001465,
      "learning_rate": 7.589781436275901e-05,
      "loss": 0.012903961539268493,
      "step": 84950
    },
    {
      "epoch": 24.11581038887312,
      "grad_norm": 1.1772693395614624,
      "learning_rate": 7.589497587283565e-05,
      "loss": 0.01269942820072174,
      "step": 84960
    },
    {
      "epoch": 24.11864887879648,
      "grad_norm": 1.1022330522537231,
      "learning_rate": 7.58921373829123e-05,
      "loss": 0.007412086427211762,
      "step": 84970
    },
    {
      "epoch": 24.12148736871984,
      "grad_norm": 1.8838292360305786,
      "learning_rate": 7.588929889298894e-05,
      "loss": 0.005235916748642921,
      "step": 84980
    },
    {
      "epoch": 24.124325858643203,
      "grad_norm": 2.305962085723877,
      "learning_rate": 7.588646040306558e-05,
      "loss": 0.01819613128900528,
      "step": 84990
    },
    {
      "epoch": 24.127164348566563,
      "grad_norm": 1.5014597177505493,
      "learning_rate": 7.58836219131422e-05,
      "loss": 0.005534688383340836,
      "step": 85000
    },
    {
      "epoch": 24.127164348566563,
      "eval_accuracy": 0.9657277293825904,
      "eval_loss": 0.1106887087225914,
      "eval_runtime": 32.0008,
      "eval_samples_per_second": 491.457,
      "eval_steps_per_second": 7.687,
      "step": 85000
    },
    {
      "epoch": 24.130002838489922,
      "grad_norm": 7.664631366729736,
      "learning_rate": 7.588078342321885e-05,
      "loss": 0.008475860953330994,
      "step": 85010
    },
    {
      "epoch": 24.132841328413285,
      "grad_norm": 7.520578384399414,
      "learning_rate": 7.587794493329549e-05,
      "loss": 0.024843305349349976,
      "step": 85020
    },
    {
      "epoch": 24.135679818336644,
      "grad_norm": 5.945399761199951,
      "learning_rate": 7.587510644337213e-05,
      "loss": 0.01612308919429779,
      "step": 85030
    },
    {
      "epoch": 24.138518308260007,
      "grad_norm": 2.361400842666626,
      "learning_rate": 7.587226795344877e-05,
      "loss": 0.01494961678981781,
      "step": 85040
    },
    {
      "epoch": 24.141356798183367,
      "grad_norm": 11.86889362335205,
      "learning_rate": 7.586942946352541e-05,
      "loss": 0.01900387406349182,
      "step": 85050
    },
    {
      "epoch": 24.144195288106726,
      "grad_norm": 0.2923731803894043,
      "learning_rate": 7.586659097360205e-05,
      "loss": 0.0044320538640022274,
      "step": 85060
    },
    {
      "epoch": 24.14703377803009,
      "grad_norm": 0.5926135182380676,
      "learning_rate": 7.586375248367868e-05,
      "loss": 0.006633836030960083,
      "step": 85070
    },
    {
      "epoch": 24.149872267953448,
      "grad_norm": 2.749441146850586,
      "learning_rate": 7.586091399375532e-05,
      "loss": 0.017066454887390135,
      "step": 85080
    },
    {
      "epoch": 24.15271075787681,
      "grad_norm": 0.2300490438938141,
      "learning_rate": 7.585807550383196e-05,
      "loss": 0.009375964105129243,
      "step": 85090
    },
    {
      "epoch": 24.15554924780017,
      "grad_norm": 1.776023507118225,
      "learning_rate": 7.58552370139086e-05,
      "loss": 0.013631683588027955,
      "step": 85100
    },
    {
      "epoch": 24.15838773772353,
      "grad_norm": 11.041183471679688,
      "learning_rate": 7.585239852398525e-05,
      "loss": 0.008951046317815781,
      "step": 85110
    },
    {
      "epoch": 24.161226227646893,
      "grad_norm": 2.2577743530273438,
      "learning_rate": 7.584956003406189e-05,
      "loss": 0.009594593942165375,
      "step": 85120
    },
    {
      "epoch": 24.164064717570252,
      "grad_norm": 1.3612792491912842,
      "learning_rate": 7.584672154413852e-05,
      "loss": 0.008569695055484772,
      "step": 85130
    },
    {
      "epoch": 24.166903207493615,
      "grad_norm": 11.403969764709473,
      "learning_rate": 7.584388305421516e-05,
      "loss": 0.0077738739550113675,
      "step": 85140
    },
    {
      "epoch": 24.169741697416974,
      "grad_norm": 2.2700397968292236,
      "learning_rate": 7.58410445642918e-05,
      "loss": 0.004367386549711227,
      "step": 85150
    },
    {
      "epoch": 24.172580187340333,
      "grad_norm": 3.2118048667907715,
      "learning_rate": 7.583820607436844e-05,
      "loss": 0.004065866023302079,
      "step": 85160
    },
    {
      "epoch": 24.175418677263696,
      "grad_norm": 1.2803608179092407,
      "learning_rate": 7.583536758444508e-05,
      "loss": 0.016695743799209593,
      "step": 85170
    },
    {
      "epoch": 24.178257167187056,
      "grad_norm": 3.2404940128326416,
      "learning_rate": 7.583252909452172e-05,
      "loss": 0.013581687211990356,
      "step": 85180
    },
    {
      "epoch": 24.18109565711042,
      "grad_norm": 2.400435209274292,
      "learning_rate": 7.582969060459836e-05,
      "loss": 0.00726730227470398,
      "step": 85190
    },
    {
      "epoch": 24.183934147033778,
      "grad_norm": 12.427816390991211,
      "learning_rate": 7.582685211467499e-05,
      "loss": 0.02270824760198593,
      "step": 85200
    },
    {
      "epoch": 24.186772636957137,
      "grad_norm": 3.7422256469726562,
      "learning_rate": 7.582401362475163e-05,
      "loss": 0.022049340605735778,
      "step": 85210
    },
    {
      "epoch": 24.1896111268805,
      "grad_norm": 0.34341269731521606,
      "learning_rate": 7.582117513482828e-05,
      "loss": 0.019231058657169342,
      "step": 85220
    },
    {
      "epoch": 24.19244961680386,
      "grad_norm": 16.335460662841797,
      "learning_rate": 7.581833664490492e-05,
      "loss": 0.03085731267929077,
      "step": 85230
    },
    {
      "epoch": 24.195288106727222,
      "grad_norm": 1.7565821409225464,
      "learning_rate": 7.581549815498156e-05,
      "loss": 0.01612974554300308,
      "step": 85240
    },
    {
      "epoch": 24.19812659665058,
      "grad_norm": 12.592902183532715,
      "learning_rate": 7.58126596650582e-05,
      "loss": 0.011208910495042801,
      "step": 85250
    },
    {
      "epoch": 24.20096508657394,
      "grad_norm": 6.833895206451416,
      "learning_rate": 7.580982117513483e-05,
      "loss": 0.010031253844499589,
      "step": 85260
    },
    {
      "epoch": 24.203803576497304,
      "grad_norm": 1.4420963525772095,
      "learning_rate": 7.580698268521147e-05,
      "loss": 0.01829663962125778,
      "step": 85270
    },
    {
      "epoch": 24.206642066420663,
      "grad_norm": 2.397555351257324,
      "learning_rate": 7.580414419528811e-05,
      "loss": 0.01912621855735779,
      "step": 85280
    },
    {
      "epoch": 24.209480556344026,
      "grad_norm": 2.0013787746429443,
      "learning_rate": 7.580130570536475e-05,
      "loss": 0.010009374469518661,
      "step": 85290
    },
    {
      "epoch": 24.212319046267385,
      "grad_norm": 13.157902717590332,
      "learning_rate": 7.579846721544139e-05,
      "loss": 0.022078612446784975,
      "step": 85300
    },
    {
      "epoch": 24.215157536190745,
      "grad_norm": 0.7067900896072388,
      "learning_rate": 7.579562872551803e-05,
      "loss": 0.015771131217479705,
      "step": 85310
    },
    {
      "epoch": 24.217996026114108,
      "grad_norm": 11.601316452026367,
      "learning_rate": 7.579279023559466e-05,
      "loss": 0.014154157042503357,
      "step": 85320
    },
    {
      "epoch": 24.220834516037467,
      "grad_norm": 2.4811770915985107,
      "learning_rate": 7.57899517456713e-05,
      "loss": 0.006140285357832909,
      "step": 85330
    },
    {
      "epoch": 24.22367300596083,
      "grad_norm": 0.33110931515693665,
      "learning_rate": 7.578711325574794e-05,
      "loss": 0.01483476310968399,
      "step": 85340
    },
    {
      "epoch": 24.22651149588419,
      "grad_norm": 0.7763314247131348,
      "learning_rate": 7.578427476582459e-05,
      "loss": 0.022756025195121765,
      "step": 85350
    },
    {
      "epoch": 24.229349985807552,
      "grad_norm": 1.5477824211120605,
      "learning_rate": 7.578143627590121e-05,
      "loss": 0.004848732054233551,
      "step": 85360
    },
    {
      "epoch": 24.23218847573091,
      "grad_norm": 0.3319757878780365,
      "learning_rate": 7.577859778597787e-05,
      "loss": 0.011671453714370728,
      "step": 85370
    },
    {
      "epoch": 24.23502696565427,
      "grad_norm": 0.3363244831562042,
      "learning_rate": 7.577575929605451e-05,
      "loss": 0.012435142695903779,
      "step": 85380
    },
    {
      "epoch": 24.237865455577634,
      "grad_norm": 0.23122678697109222,
      "learning_rate": 7.577292080613114e-05,
      "loss": 0.017498072981834412,
      "step": 85390
    },
    {
      "epoch": 24.240703945500993,
      "grad_norm": 10.567215919494629,
      "learning_rate": 7.577008231620778e-05,
      "loss": 0.007359525561332703,
      "step": 85400
    },
    {
      "epoch": 24.243542435424356,
      "grad_norm": 2.3252737522125244,
      "learning_rate": 7.576724382628442e-05,
      "loss": 0.006452349573373794,
      "step": 85410
    },
    {
      "epoch": 24.246380925347715,
      "grad_norm": 8.361401557922363,
      "learning_rate": 7.576440533636105e-05,
      "loss": 0.018063643574714662,
      "step": 85420
    },
    {
      "epoch": 24.249219415271074,
      "grad_norm": 3.543668270111084,
      "learning_rate": 7.57615668464377e-05,
      "loss": 0.015395312011241913,
      "step": 85430
    },
    {
      "epoch": 24.252057905194437,
      "grad_norm": 0.7056883573532104,
      "learning_rate": 7.575872835651434e-05,
      "loss": 0.01463746577501297,
      "step": 85440
    },
    {
      "epoch": 24.254896395117797,
      "grad_norm": 10.48837661743164,
      "learning_rate": 7.575588986659097e-05,
      "loss": 0.010624726861715316,
      "step": 85450
    },
    {
      "epoch": 24.25773488504116,
      "grad_norm": 2.8735995292663574,
      "learning_rate": 7.575305137666761e-05,
      "loss": 0.009501266479492187,
      "step": 85460
    },
    {
      "epoch": 24.26057337496452,
      "grad_norm": 0.900643527507782,
      "learning_rate": 7.575021288674426e-05,
      "loss": 0.00846024751663208,
      "step": 85470
    },
    {
      "epoch": 24.26341186488788,
      "grad_norm": 0.6239857077598572,
      "learning_rate": 7.57473743968209e-05,
      "loss": 0.007527311146259308,
      "step": 85480
    },
    {
      "epoch": 24.26625035481124,
      "grad_norm": 6.811006546020508,
      "learning_rate": 7.574453590689752e-05,
      "loss": 0.007985348254442215,
      "step": 85490
    },
    {
      "epoch": 24.2690888447346,
      "grad_norm": 0.8450534343719482,
      "learning_rate": 7.574169741697418e-05,
      "loss": 0.005784529075026512,
      "step": 85500
    },
    {
      "epoch": 24.2690888447346,
      "eval_accuracy": 0.9721498060660011,
      "eval_loss": 0.08942712843418121,
      "eval_runtime": 32.454,
      "eval_samples_per_second": 484.594,
      "eval_steps_per_second": 7.58,
      "step": 85500
    },
    {
      "epoch": 24.271927334657963,
      "grad_norm": 3.869919538497925,
      "learning_rate": 7.573885892705082e-05,
      "loss": 0.008161671459674835,
      "step": 85510
    },
    {
      "epoch": 24.274765824581323,
      "grad_norm": 2.0412440299987793,
      "learning_rate": 7.573602043712745e-05,
      "loss": 0.015182547271251678,
      "step": 85520
    },
    {
      "epoch": 24.277604314504682,
      "grad_norm": 0.23042835295200348,
      "learning_rate": 7.573318194720409e-05,
      "loss": 0.010144972801208496,
      "step": 85530
    },
    {
      "epoch": 24.280442804428045,
      "grad_norm": 3.158698081970215,
      "learning_rate": 7.573034345728073e-05,
      "loss": 0.01009099930524826,
      "step": 85540
    },
    {
      "epoch": 24.283281294351404,
      "grad_norm": 1.1883310079574585,
      "learning_rate": 7.572750496735736e-05,
      "loss": 0.006799289584159851,
      "step": 85550
    },
    {
      "epoch": 24.286119784274767,
      "grad_norm": 1.3850857019424438,
      "learning_rate": 7.572466647743401e-05,
      "loss": 0.007283708453178406,
      "step": 85560
    },
    {
      "epoch": 24.288958274198126,
      "grad_norm": 0.08487583696842194,
      "learning_rate": 7.572182798751066e-05,
      "loss": 0.012294325232505798,
      "step": 85570
    },
    {
      "epoch": 24.291796764121486,
      "grad_norm": 2.192869186401367,
      "learning_rate": 7.571898949758728e-05,
      "loss": 0.015506701171398162,
      "step": 85580
    },
    {
      "epoch": 24.29463525404485,
      "grad_norm": 0.125902459025383,
      "learning_rate": 7.571615100766392e-05,
      "loss": 0.0205858051776886,
      "step": 85590
    },
    {
      "epoch": 24.297473743968208,
      "grad_norm": 9.355557441711426,
      "learning_rate": 7.571331251774057e-05,
      "loss": 0.018552133440971376,
      "step": 85600
    },
    {
      "epoch": 24.30031223389157,
      "grad_norm": 2.395425796508789,
      "learning_rate": 7.571047402781721e-05,
      "loss": 0.004945536330342293,
      "step": 85610
    },
    {
      "epoch": 24.30315072381493,
      "grad_norm": 1.101117730140686,
      "learning_rate": 7.570763553789384e-05,
      "loss": 0.013471022248268127,
      "step": 85620
    },
    {
      "epoch": 24.30598921373829,
      "grad_norm": 0.2889711856842041,
      "learning_rate": 7.570479704797049e-05,
      "loss": 0.0149894580245018,
      "step": 85630
    },
    {
      "epoch": 24.308827703661652,
      "grad_norm": 4.994829177856445,
      "learning_rate": 7.570195855804713e-05,
      "loss": 0.009349507838487625,
      "step": 85640
    },
    {
      "epoch": 24.311666193585012,
      "grad_norm": 1.0628548860549927,
      "learning_rate": 7.569912006812376e-05,
      "loss": 0.01000620350241661,
      "step": 85650
    },
    {
      "epoch": 24.314504683508375,
      "grad_norm": 11.063276290893555,
      "learning_rate": 7.56962815782004e-05,
      "loss": 0.018959511816501618,
      "step": 85660
    },
    {
      "epoch": 24.317343173431734,
      "grad_norm": 0.08882137387990952,
      "learning_rate": 7.569344308827704e-05,
      "loss": 0.007786166667938232,
      "step": 85670
    },
    {
      "epoch": 24.320181663355093,
      "grad_norm": 2.3768861293792725,
      "learning_rate": 7.569060459835367e-05,
      "loss": 0.008636796474456787,
      "step": 85680
    },
    {
      "epoch": 24.323020153278456,
      "grad_norm": 5.162334442138672,
      "learning_rate": 7.568776610843031e-05,
      "loss": 0.011789308488368988,
      "step": 85690
    },
    {
      "epoch": 24.325858643201816,
      "grad_norm": 1.7718873023986816,
      "learning_rate": 7.568492761850697e-05,
      "loss": 0.011643117666244507,
      "step": 85700
    },
    {
      "epoch": 24.32869713312518,
      "grad_norm": 3.10215163230896,
      "learning_rate": 7.56820891285836e-05,
      "loss": 0.011649016290903091,
      "step": 85710
    },
    {
      "epoch": 24.331535623048538,
      "grad_norm": 1.8847545385360718,
      "learning_rate": 7.567925063866024e-05,
      "loss": 0.008386296778917312,
      "step": 85720
    },
    {
      "epoch": 24.3343741129719,
      "grad_norm": 0.15926982462406158,
      "learning_rate": 7.567641214873688e-05,
      "loss": 0.017521002888679506,
      "step": 85730
    },
    {
      "epoch": 24.33721260289526,
      "grad_norm": 1.864924669265747,
      "learning_rate": 7.567357365881352e-05,
      "loss": 0.009884700179100037,
      "step": 85740
    },
    {
      "epoch": 24.34005109281862,
      "grad_norm": 0.6104827523231506,
      "learning_rate": 7.567073516889015e-05,
      "loss": 0.00510796457529068,
      "step": 85750
    },
    {
      "epoch": 24.342889582741982,
      "grad_norm": 5.961177349090576,
      "learning_rate": 7.56678966789668e-05,
      "loss": 0.014258350431919097,
      "step": 85760
    },
    {
      "epoch": 24.34572807266534,
      "grad_norm": 1.9542880058288574,
      "learning_rate": 7.566505818904344e-05,
      "loss": 0.013351155817508698,
      "step": 85770
    },
    {
      "epoch": 24.348566562588704,
      "grad_norm": 1.545385718345642,
      "learning_rate": 7.566221969912007e-05,
      "loss": 0.00917430892586708,
      "step": 85780
    },
    {
      "epoch": 24.351405052512064,
      "grad_norm": 0.3673706352710724,
      "learning_rate": 7.565938120919671e-05,
      "loss": 0.008816666156053542,
      "step": 85790
    },
    {
      "epoch": 24.354243542435423,
      "grad_norm": 0.5473952889442444,
      "learning_rate": 7.565654271927335e-05,
      "loss": 0.0052477136254310604,
      "step": 85800
    },
    {
      "epoch": 24.357082032358786,
      "grad_norm": 1.6478450298309326,
      "learning_rate": 7.565370422934998e-05,
      "loss": 0.008570219576358794,
      "step": 85810
    },
    {
      "epoch": 24.359920522282145,
      "grad_norm": 1.4293564558029175,
      "learning_rate": 7.565086573942662e-05,
      "loss": 0.011023954302072526,
      "step": 85820
    },
    {
      "epoch": 24.362759012205508,
      "grad_norm": 0.10506630688905716,
      "learning_rate": 7.564802724950328e-05,
      "loss": 0.010192238539457322,
      "step": 85830
    },
    {
      "epoch": 24.365597502128868,
      "grad_norm": 0.05637767165899277,
      "learning_rate": 7.56451887595799e-05,
      "loss": 0.016657623648643493,
      "step": 85840
    },
    {
      "epoch": 24.368435992052227,
      "grad_norm": 0.24487879872322083,
      "learning_rate": 7.564235026965655e-05,
      "loss": 0.017416645586490632,
      "step": 85850
    },
    {
      "epoch": 24.37127448197559,
      "grad_norm": 0.6654809713363647,
      "learning_rate": 7.563951177973319e-05,
      "loss": 0.007884503901004791,
      "step": 85860
    },
    {
      "epoch": 24.37411297189895,
      "grad_norm": 8.374706268310547,
      "learning_rate": 7.563667328980983e-05,
      "loss": 0.014122214913368226,
      "step": 85870
    },
    {
      "epoch": 24.376951461822312,
      "grad_norm": 0.3887140452861786,
      "learning_rate": 7.563383479988646e-05,
      "loss": 0.012857778370380402,
      "step": 85880
    },
    {
      "epoch": 24.37978995174567,
      "grad_norm": 10.01579761505127,
      "learning_rate": 7.56309963099631e-05,
      "loss": 0.03637037575244904,
      "step": 85890
    },
    {
      "epoch": 24.38262844166903,
      "grad_norm": 3.512822151184082,
      "learning_rate": 7.562815782003975e-05,
      "loss": 0.005749260261654854,
      "step": 85900
    },
    {
      "epoch": 24.385466931592394,
      "grad_norm": 0.3181319832801819,
      "learning_rate": 7.562531933011638e-05,
      "loss": 0.015750199556350708,
      "step": 85910
    },
    {
      "epoch": 24.388305421515753,
      "grad_norm": 6.515655040740967,
      "learning_rate": 7.562248084019302e-05,
      "loss": 0.010526826977729798,
      "step": 85920
    },
    {
      "epoch": 24.391143911439116,
      "grad_norm": 13.040922164916992,
      "learning_rate": 7.561964235026966e-05,
      "loss": 0.01679864376783371,
      "step": 85930
    },
    {
      "epoch": 24.393982401362475,
      "grad_norm": 7.477162837982178,
      "learning_rate": 7.561680386034629e-05,
      "loss": 0.009164389967918397,
      "step": 85940
    },
    {
      "epoch": 24.396820891285834,
      "grad_norm": 1.548324704170227,
      "learning_rate": 7.561396537042293e-05,
      "loss": 0.01286749392747879,
      "step": 85950
    },
    {
      "epoch": 24.399659381209197,
      "grad_norm": 10.464897155761719,
      "learning_rate": 7.561112688049959e-05,
      "loss": 0.01696442663669586,
      "step": 85960
    },
    {
      "epoch": 24.402497871132557,
      "grad_norm": 5.819425106048584,
      "learning_rate": 7.560828839057622e-05,
      "loss": 0.0108443021774292,
      "step": 85970
    },
    {
      "epoch": 24.40533636105592,
      "grad_norm": 1.6916236877441406,
      "learning_rate": 7.560544990065286e-05,
      "loss": 0.016887083649635315,
      "step": 85980
    },
    {
      "epoch": 24.40817485097928,
      "grad_norm": 0.6641570329666138,
      "learning_rate": 7.56026114107295e-05,
      "loss": 0.008201391994953155,
      "step": 85990
    },
    {
      "epoch": 24.411013340902638,
      "grad_norm": 0.45300954580307007,
      "learning_rate": 7.559977292080614e-05,
      "loss": 0.01769823282957077,
      "step": 86000
    },
    {
      "epoch": 24.411013340902638,
      "eval_accuracy": 0.9689705601831246,
      "eval_loss": 0.10774319618940353,
      "eval_runtime": 31.7164,
      "eval_samples_per_second": 495.864,
      "eval_steps_per_second": 7.756,
      "step": 86000
    },
    {
      "epoch": 24.413851830826,
      "grad_norm": 1.4352697134017944,
      "learning_rate": 7.559693443088277e-05,
      "loss": 0.018337222933769225,
      "step": 86010
    },
    {
      "epoch": 24.41669032074936,
      "grad_norm": 2.8338675498962402,
      "learning_rate": 7.559409594095941e-05,
      "loss": 0.008044353127479554,
      "step": 86020
    },
    {
      "epoch": 24.419528810672723,
      "grad_norm": 0.3395408093929291,
      "learning_rate": 7.559125745103605e-05,
      "loss": 0.005279507488012314,
      "step": 86030
    },
    {
      "epoch": 24.422367300596083,
      "grad_norm": 0.31454989314079285,
      "learning_rate": 7.558841896111269e-05,
      "loss": 0.004294072836637497,
      "step": 86040
    },
    {
      "epoch": 24.425205790519442,
      "grad_norm": 0.6733565330505371,
      "learning_rate": 7.558558047118933e-05,
      "loss": 0.006210532039403915,
      "step": 86050
    },
    {
      "epoch": 24.428044280442805,
      "grad_norm": 1.6580836772918701,
      "learning_rate": 7.558274198126597e-05,
      "loss": 0.03039458692073822,
      "step": 86060
    },
    {
      "epoch": 24.430882770366164,
      "grad_norm": 7.571953296661377,
      "learning_rate": 7.55799034913426e-05,
      "loss": 0.016900870203971862,
      "step": 86070
    },
    {
      "epoch": 24.433721260289527,
      "grad_norm": 0.49733617901802063,
      "learning_rate": 7.557706500141924e-05,
      "loss": 0.013136042654514313,
      "step": 86080
    },
    {
      "epoch": 24.436559750212886,
      "grad_norm": 9.622514724731445,
      "learning_rate": 7.557422651149589e-05,
      "loss": 0.015646675229072572,
      "step": 86090
    },
    {
      "epoch": 24.43939824013625,
      "grad_norm": 1.1060527563095093,
      "learning_rate": 7.557138802157253e-05,
      "loss": 0.014757946133613586,
      "step": 86100
    },
    {
      "epoch": 24.44223673005961,
      "grad_norm": 7.923277854919434,
      "learning_rate": 7.556854953164917e-05,
      "loss": 0.014649665355682373,
      "step": 86110
    },
    {
      "epoch": 24.445075219982968,
      "grad_norm": 0.30258503556251526,
      "learning_rate": 7.556571104172581e-05,
      "loss": 0.008989807218313217,
      "step": 86120
    },
    {
      "epoch": 24.44791370990633,
      "grad_norm": 1.4158624410629272,
      "learning_rate": 7.556287255180244e-05,
      "loss": 0.007803232967853546,
      "step": 86130
    },
    {
      "epoch": 24.45075219982969,
      "grad_norm": 3.5515754222869873,
      "learning_rate": 7.556003406187908e-05,
      "loss": 0.010949216783046722,
      "step": 86140
    },
    {
      "epoch": 24.453590689753053,
      "grad_norm": 1.6793653964996338,
      "learning_rate": 7.555719557195572e-05,
      "loss": 0.006334037333726883,
      "step": 86150
    },
    {
      "epoch": 24.456429179676412,
      "grad_norm": 1.6839778423309326,
      "learning_rate": 7.555435708203236e-05,
      "loss": 0.01614852696657181,
      "step": 86160
    },
    {
      "epoch": 24.45926766959977,
      "grad_norm": 0.6528506278991699,
      "learning_rate": 7.5551518592109e-05,
      "loss": 0.00482315868139267,
      "step": 86170
    },
    {
      "epoch": 24.462106159523135,
      "grad_norm": 6.182528495788574,
      "learning_rate": 7.554868010218564e-05,
      "loss": 0.010558417439460755,
      "step": 86180
    },
    {
      "epoch": 24.464944649446494,
      "grad_norm": 0.49513840675354004,
      "learning_rate": 7.554584161226229e-05,
      "loss": 0.005963114276528359,
      "step": 86190
    },
    {
      "epoch": 24.467783139369857,
      "grad_norm": 0.9860437512397766,
      "learning_rate": 7.554300312233891e-05,
      "loss": 0.010787472873926163,
      "step": 86200
    },
    {
      "epoch": 24.470621629293216,
      "grad_norm": 10.727680206298828,
      "learning_rate": 7.554016463241555e-05,
      "loss": 0.02604767680168152,
      "step": 86210
    },
    {
      "epoch": 24.473460119216575,
      "grad_norm": 0.22291924059391022,
      "learning_rate": 7.55373261424922e-05,
      "loss": 0.007957303524017334,
      "step": 86220
    },
    {
      "epoch": 24.47629860913994,
      "grad_norm": 4.049028396606445,
      "learning_rate": 7.553448765256884e-05,
      "loss": 0.0058005165308713915,
      "step": 86230
    },
    {
      "epoch": 24.479137099063298,
      "grad_norm": 2.097445487976074,
      "learning_rate": 7.553164916264548e-05,
      "loss": 0.009019745886325837,
      "step": 86240
    },
    {
      "epoch": 24.48197558898666,
      "grad_norm": 0.2356591373682022,
      "learning_rate": 7.552881067272212e-05,
      "loss": 0.012569110095500945,
      "step": 86250
    },
    {
      "epoch": 24.48481407891002,
      "grad_norm": 1.7650319337844849,
      "learning_rate": 7.552597218279875e-05,
      "loss": 0.018574845790863038,
      "step": 86260
    },
    {
      "epoch": 24.48765256883338,
      "grad_norm": 0.629223644733429,
      "learning_rate": 7.552313369287539e-05,
      "loss": 0.015525272488594056,
      "step": 86270
    },
    {
      "epoch": 24.490491058756742,
      "grad_norm": 0.13371798396110535,
      "learning_rate": 7.552029520295203e-05,
      "loss": 0.0018377602100372314,
      "step": 86280
    },
    {
      "epoch": 24.4933295486801,
      "grad_norm": 0.17614972591400146,
      "learning_rate": 7.551745671302867e-05,
      "loss": 0.001714678481221199,
      "step": 86290
    },
    {
      "epoch": 24.496168038603464,
      "grad_norm": 0.1152481734752655,
      "learning_rate": 7.551461822310531e-05,
      "loss": 0.010191304236650467,
      "step": 86300
    },
    {
      "epoch": 24.499006528526824,
      "grad_norm": 1.1194592714309692,
      "learning_rate": 7.551177973318195e-05,
      "loss": 0.0097679503262043,
      "step": 86310
    },
    {
      "epoch": 24.501845018450183,
      "grad_norm": 1.1617859601974487,
      "learning_rate": 7.55089412432586e-05,
      "loss": 0.008482585102319718,
      "step": 86320
    },
    {
      "epoch": 24.504683508373546,
      "grad_norm": 0.48891130089759827,
      "learning_rate": 7.550610275333522e-05,
      "loss": 0.013673518598079682,
      "step": 86330
    },
    {
      "epoch": 24.507521998296905,
      "grad_norm": 0.48462042212486267,
      "learning_rate": 7.550326426341187e-05,
      "loss": 0.005935866013169288,
      "step": 86340
    },
    {
      "epoch": 24.510360488220268,
      "grad_norm": 7.163588047027588,
      "learning_rate": 7.55004257734885e-05,
      "loss": 0.0077618852257728575,
      "step": 86350
    },
    {
      "epoch": 24.513198978143627,
      "grad_norm": 0.7101739645004272,
      "learning_rate": 7.549758728356515e-05,
      "loss": 0.004610227048397064,
      "step": 86360
    },
    {
      "epoch": 24.516037468066987,
      "grad_norm": 3.297109603881836,
      "learning_rate": 7.549474879364179e-05,
      "loss": 0.011734139919281007,
      "step": 86370
    },
    {
      "epoch": 24.51887595799035,
      "grad_norm": 3.4683430194854736,
      "learning_rate": 7.549191030371843e-05,
      "loss": 0.008321138471364975,
      "step": 86380
    },
    {
      "epoch": 24.52171444791371,
      "grad_norm": 0.12119369953870773,
      "learning_rate": 7.548907181379506e-05,
      "loss": 0.006652224808931351,
      "step": 86390
    },
    {
      "epoch": 24.524552937837072,
      "grad_norm": 0.6051621437072754,
      "learning_rate": 7.54862333238717e-05,
      "loss": 0.012994924187660217,
      "step": 86400
    },
    {
      "epoch": 24.52739142776043,
      "grad_norm": 2.225515127182007,
      "learning_rate": 7.548339483394834e-05,
      "loss": 0.014857205748558044,
      "step": 86410
    },
    {
      "epoch": 24.53022991768379,
      "grad_norm": 6.695326805114746,
      "learning_rate": 7.548055634402498e-05,
      "loss": 0.007468993961811066,
      "step": 86420
    },
    {
      "epoch": 24.533068407607153,
      "grad_norm": 2.067754030227661,
      "learning_rate": 7.547771785410162e-05,
      "loss": 0.004249143972992897,
      "step": 86430
    },
    {
      "epoch": 24.535906897530513,
      "grad_norm": 3.1433768272399902,
      "learning_rate": 7.547516321317059e-05,
      "loss": 0.009402300417423248,
      "step": 86440
    },
    {
      "epoch": 24.538745387453876,
      "grad_norm": 1.1617540121078491,
      "learning_rate": 7.547232472324723e-05,
      "loss": 0.008847872912883758,
      "step": 86450
    },
    {
      "epoch": 24.541583877377235,
      "grad_norm": 0.5742031931877136,
      "learning_rate": 7.546948623332387e-05,
      "loss": 0.012604688107967377,
      "step": 86460
    },
    {
      "epoch": 24.544422367300598,
      "grad_norm": 0.5153142213821411,
      "learning_rate": 7.546664774340051e-05,
      "loss": 0.008601532131433488,
      "step": 86470
    },
    {
      "epoch": 24.547260857223957,
      "grad_norm": 5.297637939453125,
      "learning_rate": 7.546380925347716e-05,
      "loss": 0.008239265531301498,
      "step": 86480
    },
    {
      "epoch": 24.550099347147317,
      "grad_norm": 1.1708295345306396,
      "learning_rate": 7.54609707635538e-05,
      "loss": 0.012036576122045516,
      "step": 86490
    },
    {
      "epoch": 24.55293783707068,
      "grad_norm": 1.2955753803253174,
      "learning_rate": 7.545813227363044e-05,
      "loss": 0.007438194751739502,
      "step": 86500
    },
    {
      "epoch": 24.55293783707068,
      "eval_accuracy": 0.967826031665289,
      "eval_loss": 0.1085750162601471,
      "eval_runtime": 32.1874,
      "eval_samples_per_second": 488.607,
      "eval_steps_per_second": 7.643,
      "step": 86500
    },
    {
      "epoch": 24.55577632699404,
      "grad_norm": 4.013537883758545,
      "learning_rate": 7.545529378370707e-05,
      "loss": 0.020248191058635713,
      "step": 86510
    },
    {
      "epoch": 24.558614816917398,
      "grad_norm": 3.099640369415283,
      "learning_rate": 7.545245529378371e-05,
      "loss": 0.01388494074344635,
      "step": 86520
    },
    {
      "epoch": 24.56145330684076,
      "grad_norm": 8.307168006896973,
      "learning_rate": 7.544961680386035e-05,
      "loss": 0.012331354618072509,
      "step": 86530
    },
    {
      "epoch": 24.56429179676412,
      "grad_norm": 11.281428337097168,
      "learning_rate": 7.544677831393699e-05,
      "loss": 0.015896648168563843,
      "step": 86540
    },
    {
      "epoch": 24.567130286687483,
      "grad_norm": 4.183461666107178,
      "learning_rate": 7.544393982401363e-05,
      "loss": 0.01052619218826294,
      "step": 86550
    },
    {
      "epoch": 24.569968776610843,
      "grad_norm": 1.3841561079025269,
      "learning_rate": 7.544110133409027e-05,
      "loss": 0.013000310957431793,
      "step": 86560
    },
    {
      "epoch": 24.572807266534205,
      "grad_norm": 0.22603771090507507,
      "learning_rate": 7.54382628441669e-05,
      "loss": 0.02425757497549057,
      "step": 86570
    },
    {
      "epoch": 24.575645756457565,
      "grad_norm": 1.0649877786636353,
      "learning_rate": 7.543542435424354e-05,
      "loss": 0.018128839135169984,
      "step": 86580
    },
    {
      "epoch": 24.578484246380924,
      "grad_norm": 0.38050633668899536,
      "learning_rate": 7.543258586432018e-05,
      "loss": 0.010467040538787841,
      "step": 86590
    },
    {
      "epoch": 24.581322736304287,
      "grad_norm": 1.8373584747314453,
      "learning_rate": 7.542974737439683e-05,
      "loss": 0.004486091062426567,
      "step": 86600
    },
    {
      "epoch": 24.584161226227646,
      "grad_norm": 0.9621447920799255,
      "learning_rate": 7.542690888447347e-05,
      "loss": 0.014976632595062257,
      "step": 86610
    },
    {
      "epoch": 24.58699971615101,
      "grad_norm": 3.740422487258911,
      "learning_rate": 7.542407039455011e-05,
      "loss": 0.006465326249599457,
      "step": 86620
    },
    {
      "epoch": 24.58983820607437,
      "grad_norm": 7.454321384429932,
      "learning_rate": 7.542123190462675e-05,
      "loss": 0.012022820860147476,
      "step": 86630
    },
    {
      "epoch": 24.592676695997728,
      "grad_norm": 12.778980255126953,
      "learning_rate": 7.541839341470338e-05,
      "loss": 0.012108631432056427,
      "step": 86640
    },
    {
      "epoch": 24.59551518592109,
      "grad_norm": 10.574054718017578,
      "learning_rate": 7.541555492478002e-05,
      "loss": 0.023545390367507933,
      "step": 86650
    },
    {
      "epoch": 24.59835367584445,
      "grad_norm": 0.3041205406188965,
      "learning_rate": 7.541271643485666e-05,
      "loss": 0.009798835963010788,
      "step": 86660
    },
    {
      "epoch": 24.601192165767813,
      "grad_norm": 5.590033531188965,
      "learning_rate": 7.540987794493329e-05,
      "loss": 0.01951398253440857,
      "step": 86670
    },
    {
      "epoch": 24.604030655691172,
      "grad_norm": 0.42919260263442993,
      "learning_rate": 7.540703945500994e-05,
      "loss": 0.012195106595754623,
      "step": 86680
    },
    {
      "epoch": 24.60686914561453,
      "grad_norm": 10.508100509643555,
      "learning_rate": 7.540420096508658e-05,
      "loss": 0.026058754324913024,
      "step": 86690
    },
    {
      "epoch": 24.609707635537895,
      "grad_norm": 1.0591177940368652,
      "learning_rate": 7.540136247516321e-05,
      "loss": 0.01330510675907135,
      "step": 86700
    },
    {
      "epoch": 24.612546125461254,
      "grad_norm": 0.8153659701347351,
      "learning_rate": 7.539852398523985e-05,
      "loss": 0.011104600131511688,
      "step": 86710
    },
    {
      "epoch": 24.615384615384617,
      "grad_norm": 0.6737568974494934,
      "learning_rate": 7.53956854953165e-05,
      "loss": 0.0065693393349647525,
      "step": 86720
    },
    {
      "epoch": 24.618223105307976,
      "grad_norm": 16.24028205871582,
      "learning_rate": 7.539284700539314e-05,
      "loss": 0.02482734024524689,
      "step": 86730
    },
    {
      "epoch": 24.621061595231335,
      "grad_norm": 8.424254417419434,
      "learning_rate": 7.539000851546978e-05,
      "loss": 0.01015494391322136,
      "step": 86740
    },
    {
      "epoch": 24.6239000851547,
      "grad_norm": 0.5216388702392578,
      "learning_rate": 7.538717002554642e-05,
      "loss": 0.007353351265192032,
      "step": 86750
    },
    {
      "epoch": 24.626738575078058,
      "grad_norm": 0.18451029062271118,
      "learning_rate": 7.538433153562306e-05,
      "loss": 0.010811682790517807,
      "step": 86760
    },
    {
      "epoch": 24.62957706500142,
      "grad_norm": 3.698284387588501,
      "learning_rate": 7.538149304569969e-05,
      "loss": 0.012761187553405762,
      "step": 86770
    },
    {
      "epoch": 24.63241555492478,
      "grad_norm": 1.8929272890090942,
      "learning_rate": 7.537865455577633e-05,
      "loss": 0.003917547687888145,
      "step": 86780
    },
    {
      "epoch": 24.63525404484814,
      "grad_norm": 3.80242919921875,
      "learning_rate": 7.537581606585297e-05,
      "loss": 0.010679098963737487,
      "step": 86790
    },
    {
      "epoch": 24.638092534771502,
      "grad_norm": 8.86025333404541,
      "learning_rate": 7.53729775759296e-05,
      "loss": 0.014045929908752442,
      "step": 86800
    },
    {
      "epoch": 24.64093102469486,
      "grad_norm": 9.523038864135742,
      "learning_rate": 7.537013908600625e-05,
      "loss": 0.006660276651382446,
      "step": 86810
    },
    {
      "epoch": 24.643769514618224,
      "grad_norm": 14.693415641784668,
      "learning_rate": 7.53673005960829e-05,
      "loss": 0.009134665876626969,
      "step": 86820
    },
    {
      "epoch": 24.646608004541584,
      "grad_norm": 3.314321279525757,
      "learning_rate": 7.536446210615952e-05,
      "loss": 0.014549160003662109,
      "step": 86830
    },
    {
      "epoch": 24.649446494464943,
      "grad_norm": 0.21771349012851715,
      "learning_rate": 7.536162361623616e-05,
      "loss": 0.010577455163002014,
      "step": 86840
    },
    {
      "epoch": 24.652284984388306,
      "grad_norm": 1.2537258863449097,
      "learning_rate": 7.53587851263128e-05,
      "loss": 0.006577695906162262,
      "step": 86850
    },
    {
      "epoch": 24.655123474311665,
      "grad_norm": 0.24269087612628937,
      "learning_rate": 7.535594663638945e-05,
      "loss": 0.008163337409496308,
      "step": 86860
    },
    {
      "epoch": 24.657961964235028,
      "grad_norm": 2.7925302982330322,
      "learning_rate": 7.535310814646607e-05,
      "loss": 0.006752020120620728,
      "step": 86870
    },
    {
      "epoch": 24.660800454158387,
      "grad_norm": 0.33916938304901123,
      "learning_rate": 7.535026965654273e-05,
      "loss": 0.008962725847959518,
      "step": 86880
    },
    {
      "epoch": 24.663638944081747,
      "grad_norm": 0.2434057593345642,
      "learning_rate": 7.534743116661937e-05,
      "loss": 0.014952491223812103,
      "step": 86890
    },
    {
      "epoch": 24.66647743400511,
      "grad_norm": 0.4776949882507324,
      "learning_rate": 7.5344592676696e-05,
      "loss": 0.00804145187139511,
      "step": 86900
    },
    {
      "epoch": 24.66931592392847,
      "grad_norm": 0.43567776679992676,
      "learning_rate": 7.534175418677264e-05,
      "loss": 0.004113298654556274,
      "step": 86910
    },
    {
      "epoch": 24.672154413851832,
      "grad_norm": 0.11134946346282959,
      "learning_rate": 7.533891569684928e-05,
      "loss": 0.003929298743605614,
      "step": 86920
    },
    {
      "epoch": 24.67499290377519,
      "grad_norm": 1.0919134616851807,
      "learning_rate": 7.533607720692591e-05,
      "loss": 0.005093041807413101,
      "step": 86930
    },
    {
      "epoch": 24.677831393698554,
      "grad_norm": 0.731681764125824,
      "learning_rate": 7.533323871700256e-05,
      "loss": 0.009381788223981858,
      "step": 86940
    },
    {
      "epoch": 24.680669883621913,
      "grad_norm": 2.7037811279296875,
      "learning_rate": 7.53304002270792e-05,
      "loss": 0.007820196449756622,
      "step": 86950
    },
    {
      "epoch": 24.683508373545273,
      "grad_norm": 1.4343582391738892,
      "learning_rate": 7.532756173715583e-05,
      "loss": 0.006610675156116486,
      "step": 86960
    },
    {
      "epoch": 24.686346863468636,
      "grad_norm": 2.605112314224243,
      "learning_rate": 7.532472324723247e-05,
      "loss": 0.004300115257501602,
      "step": 86970
    },
    {
      "epoch": 24.689185353391995,
      "grad_norm": 7.131167411804199,
      "learning_rate": 7.532188475730912e-05,
      "loss": 0.011153773218393326,
      "step": 86980
    },
    {
      "epoch": 24.692023843315358,
      "grad_norm": 1.4395395517349243,
      "learning_rate": 7.531904626738576e-05,
      "loss": 0.020947591960430147,
      "step": 86990
    },
    {
      "epoch": 24.694862333238717,
      "grad_norm": 18.303829193115234,
      "learning_rate": 7.531620777746239e-05,
      "loss": 0.047566074132919314,
      "step": 87000
    },
    {
      "epoch": 24.694862333238717,
      "eval_accuracy": 0.9602594264640427,
      "eval_loss": 0.1317024827003479,
      "eval_runtime": 31.7111,
      "eval_samples_per_second": 495.946,
      "eval_steps_per_second": 7.758,
      "step": 87000
    },
    {
      "epoch": 24.697700823162076,
      "grad_norm": 2.497554063796997,
      "learning_rate": 7.531336928753904e-05,
      "loss": 0.016774526238441466,
      "step": 87010
    },
    {
      "epoch": 24.70053931308544,
      "grad_norm": 0.4982735514640808,
      "learning_rate": 7.531053079761567e-05,
      "loss": 0.011550667881965637,
      "step": 87020
    },
    {
      "epoch": 24.7033778030088,
      "grad_norm": 1.5645840167999268,
      "learning_rate": 7.530769230769231e-05,
      "loss": 0.012465214729309082,
      "step": 87030
    },
    {
      "epoch": 24.70621629293216,
      "grad_norm": 0.8627891540527344,
      "learning_rate": 7.530485381776895e-05,
      "loss": 0.008143007755279541,
      "step": 87040
    },
    {
      "epoch": 24.70905478285552,
      "grad_norm": 0.4362296462059021,
      "learning_rate": 7.530201532784559e-05,
      "loss": 0.007987241446971893,
      "step": 87050
    },
    {
      "epoch": 24.71189327277888,
      "grad_norm": 0.1727224439382553,
      "learning_rate": 7.529917683792222e-05,
      "loss": 0.012553539872169495,
      "step": 87060
    },
    {
      "epoch": 24.714731762702243,
      "grad_norm": 0.4889027178287506,
      "learning_rate": 7.529633834799886e-05,
      "loss": 0.008621846884489059,
      "step": 87070
    },
    {
      "epoch": 24.717570252625602,
      "grad_norm": 11.209785461425781,
      "learning_rate": 7.529349985807552e-05,
      "loss": 0.020931655168533327,
      "step": 87080
    },
    {
      "epoch": 24.720408742548965,
      "grad_norm": 14.347349166870117,
      "learning_rate": 7.529066136815214e-05,
      "loss": 0.034314191341400145,
      "step": 87090
    },
    {
      "epoch": 24.723247232472325,
      "grad_norm": 0.2758462429046631,
      "learning_rate": 7.528782287822879e-05,
      "loss": 0.009108667075634003,
      "step": 87100
    },
    {
      "epoch": 24.726085722395684,
      "grad_norm": 0.9600092172622681,
      "learning_rate": 7.528498438830543e-05,
      "loss": 0.009231778979301452,
      "step": 87110
    },
    {
      "epoch": 24.728924212319047,
      "grad_norm": 0.2973717451095581,
      "learning_rate": 7.528214589838205e-05,
      "loss": 0.008982735872268676,
      "step": 87120
    },
    {
      "epoch": 24.731762702242406,
      "grad_norm": 4.576530456542969,
      "learning_rate": 7.52793074084587e-05,
      "loss": 0.013734231889247894,
      "step": 87130
    },
    {
      "epoch": 24.73460119216577,
      "grad_norm": 0.5495737791061401,
      "learning_rate": 7.527646891853535e-05,
      "loss": 0.008714700490236283,
      "step": 87140
    },
    {
      "epoch": 24.73743968208913,
      "grad_norm": 13.219279289245605,
      "learning_rate": 7.527363042861198e-05,
      "loss": 0.008818300068378448,
      "step": 87150
    },
    {
      "epoch": 24.740278172012488,
      "grad_norm": 2.2898061275482178,
      "learning_rate": 7.527079193868862e-05,
      "loss": 0.005736418813467026,
      "step": 87160
    },
    {
      "epoch": 24.74311666193585,
      "grad_norm": 0.3078395128250122,
      "learning_rate": 7.526795344876526e-05,
      "loss": 0.007464403659105301,
      "step": 87170
    },
    {
      "epoch": 24.74595515185921,
      "grad_norm": 0.4191669523715973,
      "learning_rate": 7.52651149588419e-05,
      "loss": 0.019647635519504547,
      "step": 87180
    },
    {
      "epoch": 24.748793641782573,
      "grad_norm": 0.894784688949585,
      "learning_rate": 7.526227646891853e-05,
      "loss": 0.015010680258274078,
      "step": 87190
    },
    {
      "epoch": 24.751632131705932,
      "grad_norm": 0.36110028624534607,
      "learning_rate": 7.525943797899517e-05,
      "loss": 0.005522892251610756,
      "step": 87200
    },
    {
      "epoch": 24.75447062162929,
      "grad_norm": 0.1097046509385109,
      "learning_rate": 7.525659948907183e-05,
      "loss": 0.013405486941337585,
      "step": 87210
    },
    {
      "epoch": 24.757309111552654,
      "grad_norm": 8.6362886428833,
      "learning_rate": 7.525376099914846e-05,
      "loss": 0.008329806476831436,
      "step": 87220
    },
    {
      "epoch": 24.760147601476014,
      "grad_norm": 2.5056357383728027,
      "learning_rate": 7.52509225092251e-05,
      "loss": 0.016384705901145935,
      "step": 87230
    },
    {
      "epoch": 24.762986091399377,
      "grad_norm": 3.5006537437438965,
      "learning_rate": 7.524808401930174e-05,
      "loss": 0.01273140162229538,
      "step": 87240
    },
    {
      "epoch": 24.765824581322736,
      "grad_norm": 3.437854528427124,
      "learning_rate": 7.524524552937837e-05,
      "loss": 0.007156369090080261,
      "step": 87250
    },
    {
      "epoch": 24.768663071246095,
      "grad_norm": 3.800734043121338,
      "learning_rate": 7.524240703945501e-05,
      "loss": 0.03642425537109375,
      "step": 87260
    },
    {
      "epoch": 24.77150156116946,
      "grad_norm": 5.916914939880371,
      "learning_rate": 7.523956854953166e-05,
      "loss": 0.02490185797214508,
      "step": 87270
    },
    {
      "epoch": 24.774340051092818,
      "grad_norm": 14.271767616271973,
      "learning_rate": 7.523673005960829e-05,
      "loss": 0.0245846226811409,
      "step": 87280
    },
    {
      "epoch": 24.77717854101618,
      "grad_norm": 6.42150354385376,
      "learning_rate": 7.523389156968493e-05,
      "loss": 0.009956157207489014,
      "step": 87290
    },
    {
      "epoch": 24.78001703093954,
      "grad_norm": 9.239266395568848,
      "learning_rate": 7.523105307976157e-05,
      "loss": 0.014026257395744323,
      "step": 87300
    },
    {
      "epoch": 24.782855520862903,
      "grad_norm": 5.918678283691406,
      "learning_rate": 7.522821458983821e-05,
      "loss": 0.015704232454299926,
      "step": 87310
    },
    {
      "epoch": 24.785694010786262,
      "grad_norm": 6.271981239318848,
      "learning_rate": 7.522537609991484e-05,
      "loss": 0.01067735105752945,
      "step": 87320
    },
    {
      "epoch": 24.78853250070962,
      "grad_norm": 1.248551368713379,
      "learning_rate": 7.522253760999148e-05,
      "loss": 0.0186002254486084,
      "step": 87330
    },
    {
      "epoch": 24.791370990632984,
      "grad_norm": 0.4113011062145233,
      "learning_rate": 7.521969912006814e-05,
      "loss": 0.007843995094299316,
      "step": 87340
    },
    {
      "epoch": 24.794209480556344,
      "grad_norm": 0.37523746490478516,
      "learning_rate": 7.521686063014477e-05,
      "loss": 0.011750002205371857,
      "step": 87350
    },
    {
      "epoch": 24.797047970479706,
      "grad_norm": 7.022216796875,
      "learning_rate": 7.521402214022141e-05,
      "loss": 0.007191299647092819,
      "step": 87360
    },
    {
      "epoch": 24.799886460403066,
      "grad_norm": 1.983604907989502,
      "learning_rate": 7.521118365029805e-05,
      "loss": 0.005740867555141449,
      "step": 87370
    },
    {
      "epoch": 24.802724950326425,
      "grad_norm": 2.143331289291382,
      "learning_rate": 7.520834516037468e-05,
      "loss": 0.005615520477294922,
      "step": 87380
    },
    {
      "epoch": 24.805563440249788,
      "grad_norm": 1.0639060735702515,
      "learning_rate": 7.520550667045132e-05,
      "loss": 0.009142071008682251,
      "step": 87390
    },
    {
      "epoch": 24.808401930173147,
      "grad_norm": 8.75391674041748,
      "learning_rate": 7.520266818052796e-05,
      "loss": 0.005440958589315414,
      "step": 87400
    },
    {
      "epoch": 24.81124042009651,
      "grad_norm": 4.223254680633545,
      "learning_rate": 7.51998296906046e-05,
      "loss": 0.03077690601348877,
      "step": 87410
    },
    {
      "epoch": 24.81407891001987,
      "grad_norm": 4.01494836807251,
      "learning_rate": 7.519699120068124e-05,
      "loss": 0.008579573035240174,
      "step": 87420
    },
    {
      "epoch": 24.81691739994323,
      "grad_norm": 3.50024676322937,
      "learning_rate": 7.519415271075788e-05,
      "loss": 0.02002474069595337,
      "step": 87430
    },
    {
      "epoch": 24.81975588986659,
      "grad_norm": 0.917782187461853,
      "learning_rate": 7.519131422083452e-05,
      "loss": 0.013988485932350159,
      "step": 87440
    },
    {
      "epoch": 24.82259437978995,
      "grad_norm": 1.4966775178909302,
      "learning_rate": 7.518847573091115e-05,
      "loss": 0.026089850068092345,
      "step": 87450
    },
    {
      "epoch": 24.825432869713314,
      "grad_norm": 8.906994819641113,
      "learning_rate": 7.51856372409878e-05,
      "loss": 0.025699731707572938,
      "step": 87460
    },
    {
      "epoch": 24.828271359636673,
      "grad_norm": 1.5569522380828857,
      "learning_rate": 7.518279875106445e-05,
      "loss": 0.020185469090938567,
      "step": 87470
    },
    {
      "epoch": 24.831109849560033,
      "grad_norm": 16.2556095123291,
      "learning_rate": 7.517996026114108e-05,
      "loss": 0.015798908472061158,
      "step": 87480
    },
    {
      "epoch": 24.833948339483396,
      "grad_norm": 0.08828189224004745,
      "learning_rate": 7.517712177121772e-05,
      "loss": 0.006485775113105774,
      "step": 87490
    },
    {
      "epoch": 24.836786829406755,
      "grad_norm": 0.5685157179832458,
      "learning_rate": 7.517428328129436e-05,
      "loss": 0.01690186411142349,
      "step": 87500
    },
    {
      "epoch": 24.836786829406755,
      "eval_accuracy": 0.9669994277357411,
      "eval_loss": 0.1103290244936943,
      "eval_runtime": 31.9648,
      "eval_samples_per_second": 492.009,
      "eval_steps_per_second": 7.696,
      "step": 87500
    },
    {
      "epoch": 24.839625319330118,
      "grad_norm": 0.6033041477203369,
      "learning_rate": 7.517144479137099e-05,
      "loss": 0.012880977988243104,
      "step": 87510
    },
    {
      "epoch": 24.842463809253477,
      "grad_norm": 0.755061149597168,
      "learning_rate": 7.516860630144763e-05,
      "loss": 0.022327718138694764,
      "step": 87520
    },
    {
      "epoch": 24.845302299176836,
      "grad_norm": 6.705821990966797,
      "learning_rate": 7.516576781152427e-05,
      "loss": 0.0173881396651268,
      "step": 87530
    },
    {
      "epoch": 24.8481407891002,
      "grad_norm": 0.8221394419670105,
      "learning_rate": 7.516292932160091e-05,
      "loss": 0.004737811535596848,
      "step": 87540
    },
    {
      "epoch": 24.85097927902356,
      "grad_norm": 14.285408020019531,
      "learning_rate": 7.516009083167755e-05,
      "loss": 0.011455235630273819,
      "step": 87550
    },
    {
      "epoch": 24.85381776894692,
      "grad_norm": 4.1569504737854,
      "learning_rate": 7.51572523417542e-05,
      "loss": 0.026419785618782044,
      "step": 87560
    },
    {
      "epoch": 24.85665625887028,
      "grad_norm": 2.0025784969329834,
      "learning_rate": 7.515441385183084e-05,
      "loss": 0.02092153877019882,
      "step": 87570
    },
    {
      "epoch": 24.85949474879364,
      "grad_norm": 0.9641667008399963,
      "learning_rate": 7.515157536190746e-05,
      "loss": 0.02503083348274231,
      "step": 87580
    },
    {
      "epoch": 24.862333238717003,
      "grad_norm": 7.7629923820495605,
      "learning_rate": 7.51487368719841e-05,
      "loss": 0.014247119426727295,
      "step": 87590
    },
    {
      "epoch": 24.865171728640362,
      "grad_norm": 12.19746208190918,
      "learning_rate": 7.514589838206075e-05,
      "loss": 0.024246279895305634,
      "step": 87600
    },
    {
      "epoch": 24.868010218563725,
      "grad_norm": 4.187434196472168,
      "learning_rate": 7.514305989213739e-05,
      "loss": 0.01857968270778656,
      "step": 87610
    },
    {
      "epoch": 24.870848708487085,
      "grad_norm": 7.167263984680176,
      "learning_rate": 7.514022140221403e-05,
      "loss": 0.006836596131324768,
      "step": 87620
    },
    {
      "epoch": 24.873687198410444,
      "grad_norm": 0.4774487018585205,
      "learning_rate": 7.513738291229067e-05,
      "loss": 0.007520785927772522,
      "step": 87630
    },
    {
      "epoch": 24.876525688333807,
      "grad_norm": 12.680917739868164,
      "learning_rate": 7.51345444223673e-05,
      "loss": 0.010874718427658081,
      "step": 87640
    },
    {
      "epoch": 24.879364178257166,
      "grad_norm": 1.2957574129104614,
      "learning_rate": 7.513170593244394e-05,
      "loss": 0.01659461557865143,
      "step": 87650
    },
    {
      "epoch": 24.88220266818053,
      "grad_norm": 2.0649008750915527,
      "learning_rate": 7.512886744252058e-05,
      "loss": 0.012110823392868042,
      "step": 87660
    },
    {
      "epoch": 24.88504115810389,
      "grad_norm": 0.5235140323638916,
      "learning_rate": 7.512602895259722e-05,
      "loss": 0.00854002609848976,
      "step": 87670
    },
    {
      "epoch": 24.88787964802725,
      "grad_norm": 1.5249601602554321,
      "learning_rate": 7.512319046267386e-05,
      "loss": 0.015473225712776184,
      "step": 87680
    },
    {
      "epoch": 24.89071813795061,
      "grad_norm": 0.29476678371429443,
      "learning_rate": 7.51203519727505e-05,
      "loss": 0.021198800206184386,
      "step": 87690
    },
    {
      "epoch": 24.89355662787397,
      "grad_norm": 12.52983283996582,
      "learning_rate": 7.511751348282715e-05,
      "loss": 0.013080716133117676,
      "step": 87700
    },
    {
      "epoch": 24.896395117797333,
      "grad_norm": 9.88040828704834,
      "learning_rate": 7.511467499290377e-05,
      "loss": 0.019578444957733154,
      "step": 87710
    },
    {
      "epoch": 24.899233607720692,
      "grad_norm": 1.0633220672607422,
      "learning_rate": 7.511183650298042e-05,
      "loss": 0.018388885259628295,
      "step": 87720
    },
    {
      "epoch": 24.902072097644055,
      "grad_norm": 2.0224289894104004,
      "learning_rate": 7.510899801305706e-05,
      "loss": 0.019063127040863038,
      "step": 87730
    },
    {
      "epoch": 24.904910587567414,
      "grad_norm": 13.229560852050781,
      "learning_rate": 7.51061595231337e-05,
      "loss": 0.02253347635269165,
      "step": 87740
    },
    {
      "epoch": 24.907749077490774,
      "grad_norm": 1.544021725654602,
      "learning_rate": 7.510332103321034e-05,
      "loss": 0.011013183742761612,
      "step": 87750
    },
    {
      "epoch": 24.910587567414137,
      "grad_norm": 0.08206012845039368,
      "learning_rate": 7.510048254328698e-05,
      "loss": 0.009166860580444336,
      "step": 87760
    },
    {
      "epoch": 24.913426057337496,
      "grad_norm": 4.057959079742432,
      "learning_rate": 7.509764405336361e-05,
      "loss": 0.006913983821868896,
      "step": 87770
    },
    {
      "epoch": 24.91626454726086,
      "grad_norm": 3.6529219150543213,
      "learning_rate": 7.509480556344025e-05,
      "loss": 0.007982385903596878,
      "step": 87780
    },
    {
      "epoch": 24.919103037184218,
      "grad_norm": 3.55979061126709,
      "learning_rate": 7.509196707351689e-05,
      "loss": 0.006905844062566757,
      "step": 87790
    },
    {
      "epoch": 24.921941527107577,
      "grad_norm": 4.727871417999268,
      "learning_rate": 7.508912858359353e-05,
      "loss": 0.008420443534851075,
      "step": 87800
    },
    {
      "epoch": 24.92478001703094,
      "grad_norm": 2.0958633422851562,
      "learning_rate": 7.508629009367017e-05,
      "loss": 0.016092444956302642,
      "step": 87810
    },
    {
      "epoch": 24.9276185069543,
      "grad_norm": 0.15478816628456116,
      "learning_rate": 7.508345160374682e-05,
      "loss": 0.02559421956539154,
      "step": 87820
    },
    {
      "epoch": 24.930456996877663,
      "grad_norm": 0.5389772653579712,
      "learning_rate": 7.508061311382346e-05,
      "loss": 0.017667108774185182,
      "step": 87830
    },
    {
      "epoch": 24.933295486801022,
      "grad_norm": 18.557151794433594,
      "learning_rate": 7.507777462390008e-05,
      "loss": 0.02704583704471588,
      "step": 87840
    },
    {
      "epoch": 24.93613397672438,
      "grad_norm": 12.760278701782227,
      "learning_rate": 7.507493613397673e-05,
      "loss": 0.03062024712562561,
      "step": 87850
    },
    {
      "epoch": 24.938972466647744,
      "grad_norm": 1.5880898237228394,
      "learning_rate": 7.507209764405337e-05,
      "loss": 0.052779597043991086,
      "step": 87860
    },
    {
      "epoch": 24.941810956571103,
      "grad_norm": 0.4904594123363495,
      "learning_rate": 7.506925915413001e-05,
      "loss": 0.033505970239639284,
      "step": 87870
    },
    {
      "epoch": 24.944649446494466,
      "grad_norm": 0.7134040594100952,
      "learning_rate": 7.506642066420665e-05,
      "loss": 0.018302568793296815,
      "step": 87880
    },
    {
      "epoch": 24.947487936417826,
      "grad_norm": 1.6387419700622559,
      "learning_rate": 7.506358217428329e-05,
      "loss": 0.009172136336565018,
      "step": 87890
    },
    {
      "epoch": 24.950326426341185,
      "grad_norm": 10.225381851196289,
      "learning_rate": 7.506074368435992e-05,
      "loss": 0.015201254189014435,
      "step": 87900
    },
    {
      "epoch": 24.953164916264548,
      "grad_norm": 11.05975341796875,
      "learning_rate": 7.505790519443656e-05,
      "loss": 0.013270682096481324,
      "step": 87910
    },
    {
      "epoch": 24.956003406187907,
      "grad_norm": 0.05077755078673363,
      "learning_rate": 7.50550667045132e-05,
      "loss": 0.010212510824203491,
      "step": 87920
    },
    {
      "epoch": 24.95884189611127,
      "grad_norm": 1.3177310228347778,
      "learning_rate": 7.505222821458984e-05,
      "loss": 0.007656729966402054,
      "step": 87930
    },
    {
      "epoch": 24.96168038603463,
      "grad_norm": 6.527772903442383,
      "learning_rate": 7.504938972466649e-05,
      "loss": 0.011381358653306962,
      "step": 87940
    },
    {
      "epoch": 24.96451887595799,
      "grad_norm": 1.5577088594436646,
      "learning_rate": 7.504655123474313e-05,
      "loss": 0.011264510452747345,
      "step": 87950
    },
    {
      "epoch": 24.96735736588135,
      "grad_norm": 1.3065193891525269,
      "learning_rate": 7.504371274481975e-05,
      "loss": 0.009608546644449234,
      "step": 87960
    },
    {
      "epoch": 24.97019585580471,
      "grad_norm": 1.0861492156982422,
      "learning_rate": 7.50408742548964e-05,
      "loss": 0.007268063724040985,
      "step": 87970
    },
    {
      "epoch": 24.973034345728074,
      "grad_norm": 4.9401021003723145,
      "learning_rate": 7.503803576497304e-05,
      "loss": 0.012372027337551116,
      "step": 87980
    },
    {
      "epoch": 24.975872835651433,
      "grad_norm": 3.4493188858032227,
      "learning_rate": 7.503519727504968e-05,
      "loss": 0.02445839047431946,
      "step": 87990
    },
    {
      "epoch": 24.978711325574793,
      "grad_norm": 1.6203080415725708,
      "learning_rate": 7.50323587851263e-05,
      "loss": 0.006531777232885361,
      "step": 88000
    },
    {
      "epoch": 24.978711325574793,
      "eval_accuracy": 0.9694156546067273,
      "eval_loss": 0.10140075534582138,
      "eval_runtime": 31.776,
      "eval_samples_per_second": 494.933,
      "eval_steps_per_second": 7.742,
      "step": 88000
    },
    {
      "epoch": 24.981549815498155,
      "grad_norm": 0.5066255927085876,
      "learning_rate": 7.502952029520296e-05,
      "loss": 0.009429792314767838,
      "step": 88010
    },
    {
      "epoch": 24.984388305421515,
      "grad_norm": 3.279327392578125,
      "learning_rate": 7.50266818052796e-05,
      "loss": 0.0029561664909124375,
      "step": 88020
    },
    {
      "epoch": 24.987226795344878,
      "grad_norm": 0.3823559880256653,
      "learning_rate": 7.502384331535623e-05,
      "loss": 0.005222257226705551,
      "step": 88030
    },
    {
      "epoch": 24.990065285268237,
      "grad_norm": 0.5083175897598267,
      "learning_rate": 7.502100482543287e-05,
      "loss": 0.012538023293018341,
      "step": 88040
    },
    {
      "epoch": 24.9929037751916,
      "grad_norm": 2.0581490993499756,
      "learning_rate": 7.501816633550951e-05,
      "loss": 0.014437662065029144,
      "step": 88050
    },
    {
      "epoch": 24.99574226511496,
      "grad_norm": 2.9934422969818115,
      "learning_rate": 7.501532784558614e-05,
      "loss": 0.01310369074344635,
      "step": 88060
    },
    {
      "epoch": 24.99858075503832,
      "grad_norm": 0.717076301574707,
      "learning_rate": 7.50124893556628e-05,
      "loss": 0.004772055149078369,
      "step": 88070
    },
    {
      "epoch": 25.00141924496168,
      "grad_norm": 0.9498530030250549,
      "learning_rate": 7.500965086573944e-05,
      "loss": 0.01948847621679306,
      "step": 88080
    },
    {
      "epoch": 25.00425773488504,
      "grad_norm": 3.247642755508423,
      "learning_rate": 7.500681237581606e-05,
      "loss": 0.014774627983570099,
      "step": 88090
    },
    {
      "epoch": 25.007096224808404,
      "grad_norm": 0.13476575911045074,
      "learning_rate": 7.50039738858927e-05,
      "loss": 0.007771135866641998,
      "step": 88100
    },
    {
      "epoch": 25.009934714731763,
      "grad_norm": 0.2716066837310791,
      "learning_rate": 7.500113539596935e-05,
      "loss": 0.012376495450735093,
      "step": 88110
    },
    {
      "epoch": 25.012773204655122,
      "grad_norm": 3.709446668624878,
      "learning_rate": 7.499829690604599e-05,
      "loss": 0.011676845699548721,
      "step": 88120
    },
    {
      "epoch": 25.015611694578485,
      "grad_norm": 0.708116888999939,
      "learning_rate": 7.499545841612262e-05,
      "loss": 0.01151844710111618,
      "step": 88130
    },
    {
      "epoch": 25.018450184501845,
      "grad_norm": 1.56290602684021,
      "learning_rate": 7.499261992619927e-05,
      "loss": 0.015797659754753113,
      "step": 88140
    },
    {
      "epoch": 25.021288674425207,
      "grad_norm": 0.46768295764923096,
      "learning_rate": 7.498978143627591e-05,
      "loss": 0.012793682515621185,
      "step": 88150
    },
    {
      "epoch": 25.024127164348567,
      "grad_norm": 2.678234815597534,
      "learning_rate": 7.498694294635254e-05,
      "loss": 0.00877302885055542,
      "step": 88160
    },
    {
      "epoch": 25.026965654271926,
      "grad_norm": 2.4953036308288574,
      "learning_rate": 7.498410445642918e-05,
      "loss": 0.004702713340520859,
      "step": 88170
    },
    {
      "epoch": 25.02980414419529,
      "grad_norm": 0.37608787417411804,
      "learning_rate": 7.498126596650582e-05,
      "loss": 0.008555131405591965,
      "step": 88180
    },
    {
      "epoch": 25.03264263411865,
      "grad_norm": 0.352338045835495,
      "learning_rate": 7.497842747658245e-05,
      "loss": 0.007688891887664795,
      "step": 88190
    },
    {
      "epoch": 25.03548112404201,
      "grad_norm": 0.968505322933197,
      "learning_rate": 7.497558898665909e-05,
      "loss": 0.027637457847595213,
      "step": 88200
    },
    {
      "epoch": 25.03831961396537,
      "grad_norm": 3.6665990352630615,
      "learning_rate": 7.497275049673575e-05,
      "loss": 0.004941928759217262,
      "step": 88210
    },
    {
      "epoch": 25.04115810388873,
      "grad_norm": 0.2770722508430481,
      "learning_rate": 7.496991200681238e-05,
      "loss": 0.00776611715555191,
      "step": 88220
    },
    {
      "epoch": 25.043996593812093,
      "grad_norm": 12.582157135009766,
      "learning_rate": 7.496707351688902e-05,
      "loss": 0.013689103722572326,
      "step": 88230
    },
    {
      "epoch": 25.046835083735452,
      "grad_norm": 0.20681189000606537,
      "learning_rate": 7.496423502696566e-05,
      "loss": 0.00542277954518795,
      "step": 88240
    },
    {
      "epoch": 25.049673573658815,
      "grad_norm": 0.433368980884552,
      "learning_rate": 7.49613965370423e-05,
      "loss": 0.0042979147285223005,
      "step": 88250
    },
    {
      "epoch": 25.052512063582174,
      "grad_norm": 1.0875904560089111,
      "learning_rate": 7.495855804711893e-05,
      "loss": 0.0037341035902500154,
      "step": 88260
    },
    {
      "epoch": 25.055350553505534,
      "grad_norm": 0.049867842346429825,
      "learning_rate": 7.495571955719558e-05,
      "loss": 0.00936647430062294,
      "step": 88270
    },
    {
      "epoch": 25.058189043428897,
      "grad_norm": 0.23676122725009918,
      "learning_rate": 7.495288106727222e-05,
      "loss": 0.004690884053707123,
      "step": 88280
    },
    {
      "epoch": 25.061027533352256,
      "grad_norm": 0.16013887524604797,
      "learning_rate": 7.495004257734885e-05,
      "loss": 0.003253427892923355,
      "step": 88290
    },
    {
      "epoch": 25.06386602327562,
      "grad_norm": 0.33957648277282715,
      "learning_rate": 7.494720408742549e-05,
      "loss": 0.005886750295758247,
      "step": 88300
    },
    {
      "epoch": 25.066704513198978,
      "grad_norm": 0.0957634374499321,
      "learning_rate": 7.494436559750213e-05,
      "loss": 0.007570482790470123,
      "step": 88310
    },
    {
      "epoch": 25.069543003122337,
      "grad_norm": 0.1333116888999939,
      "learning_rate": 7.494152710757876e-05,
      "loss": 0.01260690689086914,
      "step": 88320
    },
    {
      "epoch": 25.0723814930457,
      "grad_norm": 0.7813993692398071,
      "learning_rate": 7.49386886176554e-05,
      "loss": 0.004560505971312523,
      "step": 88330
    },
    {
      "epoch": 25.07521998296906,
      "grad_norm": 0.28840896487236023,
      "learning_rate": 7.493585012773206e-05,
      "loss": 0.01157001256942749,
      "step": 88340
    },
    {
      "epoch": 25.078058472892423,
      "grad_norm": 6.713897228240967,
      "learning_rate": 7.493301163780869e-05,
      "loss": 0.009289898723363877,
      "step": 88350
    },
    {
      "epoch": 25.080896962815782,
      "grad_norm": 3.9016716480255127,
      "learning_rate": 7.493017314788533e-05,
      "loss": 0.0029821325093507766,
      "step": 88360
    },
    {
      "epoch": 25.08373545273914,
      "grad_norm": 0.06385046988725662,
      "learning_rate": 7.492733465796197e-05,
      "loss": 0.00582755021750927,
      "step": 88370
    },
    {
      "epoch": 25.086573942662504,
      "grad_norm": 0.6981302499771118,
      "learning_rate": 7.492449616803861e-05,
      "loss": 0.0291093111038208,
      "step": 88380
    },
    {
      "epoch": 25.089412432585863,
      "grad_norm": 2.3380746841430664,
      "learning_rate": 7.492165767811524e-05,
      "loss": 0.003324177861213684,
      "step": 88390
    },
    {
      "epoch": 25.092250922509226,
      "grad_norm": 12.919342041015625,
      "learning_rate": 7.49188191881919e-05,
      "loss": 0.019175425171852112,
      "step": 88400
    },
    {
      "epoch": 25.095089412432586,
      "grad_norm": 0.44890886545181274,
      "learning_rate": 7.491598069826853e-05,
      "loss": 0.010593345761299134,
      "step": 88410
    },
    {
      "epoch": 25.097927902355945,
      "grad_norm": 2.9193708896636963,
      "learning_rate": 7.491314220834516e-05,
      "loss": 0.015274964272975922,
      "step": 88420
    },
    {
      "epoch": 25.100766392279308,
      "grad_norm": 0.8115370273590088,
      "learning_rate": 7.49103037184218e-05,
      "loss": 0.010110214352607727,
      "step": 88430
    },
    {
      "epoch": 25.103604882202667,
      "grad_norm": 3.675935745239258,
      "learning_rate": 7.490746522849845e-05,
      "loss": 0.008279177546501159,
      "step": 88440
    },
    {
      "epoch": 25.10644337212603,
      "grad_norm": 1.4785747528076172,
      "learning_rate": 7.490462673857507e-05,
      "loss": 0.00602576658129692,
      "step": 88450
    },
    {
      "epoch": 25.10928186204939,
      "grad_norm": 3.8742661476135254,
      "learning_rate": 7.490178824865171e-05,
      "loss": 0.004725802689790726,
      "step": 88460
    },
    {
      "epoch": 25.11212035197275,
      "grad_norm": 2.847621440887451,
      "learning_rate": 7.489894975872837e-05,
      "loss": 0.008736441284418106,
      "step": 88470
    },
    {
      "epoch": 25.11495884189611,
      "grad_norm": 3.212477445602417,
      "learning_rate": 7.4896111268805e-05,
      "loss": 0.010048354417085648,
      "step": 88480
    },
    {
      "epoch": 25.11779733181947,
      "grad_norm": 0.6197217106819153,
      "learning_rate": 7.489327277888164e-05,
      "loss": 0.013915348052978515,
      "step": 88490
    },
    {
      "epoch": 25.120635821742834,
      "grad_norm": 1.7774198055267334,
      "learning_rate": 7.489043428895828e-05,
      "loss": 0.01153482049703598,
      "step": 88500
    },
    {
      "epoch": 25.120635821742834,
      "eval_accuracy": 0.970687352959878,
      "eval_loss": 0.09431147575378418,
      "eval_runtime": 31.6779,
      "eval_samples_per_second": 496.466,
      "eval_steps_per_second": 7.766,
      "step": 88500
    },
    {
      "epoch": 25.123474311666193,
      "grad_norm": 16.670610427856445,
      "learning_rate": 7.488759579903492e-05,
      "loss": 0.013590759038925171,
      "step": 88510
    },
    {
      "epoch": 25.126312801589556,
      "grad_norm": 1.0112979412078857,
      "learning_rate": 7.488475730911155e-05,
      "loss": 0.012445001304149628,
      "step": 88520
    },
    {
      "epoch": 25.129151291512915,
      "grad_norm": 0.8229496479034424,
      "learning_rate": 7.488191881918819e-05,
      "loss": 0.013478408753871917,
      "step": 88530
    },
    {
      "epoch": 25.131989781436275,
      "grad_norm": 0.2367406189441681,
      "learning_rate": 7.487908032926485e-05,
      "loss": 0.004234018921852112,
      "step": 88540
    },
    {
      "epoch": 25.134828271359638,
      "grad_norm": 9.110761642456055,
      "learning_rate": 7.487624183934147e-05,
      "loss": 0.021476611495018005,
      "step": 88550
    },
    {
      "epoch": 25.137666761282997,
      "grad_norm": 1.043081283569336,
      "learning_rate": 7.487340334941811e-05,
      "loss": 0.021034453809261323,
      "step": 88560
    },
    {
      "epoch": 25.14050525120636,
      "grad_norm": 0.14409944415092468,
      "learning_rate": 7.487056485949476e-05,
      "loss": 0.005038371682167054,
      "step": 88570
    },
    {
      "epoch": 25.14334374112972,
      "grad_norm": 0.3085535764694214,
      "learning_rate": 7.486772636957138e-05,
      "loss": 0.009268808364868163,
      "step": 88580
    },
    {
      "epoch": 25.14618223105308,
      "grad_norm": 0.8214060068130493,
      "learning_rate": 7.486488787964803e-05,
      "loss": 0.005521495640277862,
      "step": 88590
    },
    {
      "epoch": 25.14902072097644,
      "grad_norm": 0.34801214933395386,
      "learning_rate": 7.486204938972468e-05,
      "loss": 0.007418334484100342,
      "step": 88600
    },
    {
      "epoch": 25.1518592108998,
      "grad_norm": 0.8509289026260376,
      "learning_rate": 7.485921089980131e-05,
      "loss": 0.010123103857040405,
      "step": 88610
    },
    {
      "epoch": 25.154697700823164,
      "grad_norm": 6.638089656829834,
      "learning_rate": 7.485637240987795e-05,
      "loss": 0.009859584271907806,
      "step": 88620
    },
    {
      "epoch": 25.157536190746523,
      "grad_norm": 1.032214641571045,
      "learning_rate": 7.485353391995459e-05,
      "loss": 0.007451426237821579,
      "step": 88630
    },
    {
      "epoch": 25.160374680669882,
      "grad_norm": 10.31036376953125,
      "learning_rate": 7.485069543003123e-05,
      "loss": 0.009381026029586792,
      "step": 88640
    },
    {
      "epoch": 25.163213170593245,
      "grad_norm": 0.5044496059417725,
      "learning_rate": 7.484785694010786e-05,
      "loss": 0.004036794230341911,
      "step": 88650
    },
    {
      "epoch": 25.166051660516604,
      "grad_norm": 0.5299307107925415,
      "learning_rate": 7.48450184501845e-05,
      "loss": 0.0037077143788337708,
      "step": 88660
    },
    {
      "epoch": 25.168890150439967,
      "grad_norm": 0.14194254577159882,
      "learning_rate": 7.484217996026116e-05,
      "loss": 0.009735964238643646,
      "step": 88670
    },
    {
      "epoch": 25.171728640363327,
      "grad_norm": 0.8430927395820618,
      "learning_rate": 7.483934147033778e-05,
      "loss": 0.017762945592403413,
      "step": 88680
    },
    {
      "epoch": 25.174567130286686,
      "grad_norm": 7.64051628112793,
      "learning_rate": 7.483650298041443e-05,
      "loss": 0.016383124887943266,
      "step": 88690
    },
    {
      "epoch": 25.17740562021005,
      "grad_norm": 0.7499608397483826,
      "learning_rate": 7.483366449049107e-05,
      "loss": 0.004283434897661209,
      "step": 88700
    },
    {
      "epoch": 25.18024411013341,
      "grad_norm": 1.6921086311340332,
      "learning_rate": 7.48308260005677e-05,
      "loss": 0.013006976246833802,
      "step": 88710
    },
    {
      "epoch": 25.18308260005677,
      "grad_norm": 0.12373252213001251,
      "learning_rate": 7.482798751064434e-05,
      "loss": 0.004560244455933571,
      "step": 88720
    },
    {
      "epoch": 25.18592108998013,
      "grad_norm": 0.4559331238269806,
      "learning_rate": 7.482514902072098e-05,
      "loss": 0.011958794295787811,
      "step": 88730
    },
    {
      "epoch": 25.18875957990349,
      "grad_norm": 8.311619758605957,
      "learning_rate": 7.482231053079762e-05,
      "loss": 0.010458940267562866,
      "step": 88740
    },
    {
      "epoch": 25.191598069826853,
      "grad_norm": 6.972181797027588,
      "learning_rate": 7.481947204087426e-05,
      "loss": 0.01580372601747513,
      "step": 88750
    },
    {
      "epoch": 25.194436559750212,
      "grad_norm": 0.32099586725234985,
      "learning_rate": 7.48166335509509e-05,
      "loss": 0.009928276389837265,
      "step": 88760
    },
    {
      "epoch": 25.197275049673575,
      "grad_norm": 0.2579554319381714,
      "learning_rate": 7.481379506102754e-05,
      "loss": 0.006549473851919174,
      "step": 88770
    },
    {
      "epoch": 25.200113539596934,
      "grad_norm": 0.3874220550060272,
      "learning_rate": 7.481095657110417e-05,
      "loss": 0.007567853480577469,
      "step": 88780
    },
    {
      "epoch": 25.202952029520294,
      "grad_norm": 0.2313355654478073,
      "learning_rate": 7.480811808118081e-05,
      "loss": 0.0017976256087422371,
      "step": 88790
    },
    {
      "epoch": 25.205790519443656,
      "grad_norm": 17.096664428710938,
      "learning_rate": 7.480527959125745e-05,
      "loss": 0.014226937294006347,
      "step": 88800
    },
    {
      "epoch": 25.208629009367016,
      "grad_norm": 1.0677471160888672,
      "learning_rate": 7.48024411013341e-05,
      "loss": 0.008298756182193756,
      "step": 88810
    },
    {
      "epoch": 25.21146749929038,
      "grad_norm": 0.26424217224121094,
      "learning_rate": 7.479960261141074e-05,
      "loss": 0.007903221249580383,
      "step": 88820
    },
    {
      "epoch": 25.214305989213738,
      "grad_norm": 5.477983474731445,
      "learning_rate": 7.479676412148738e-05,
      "loss": 0.006830044090747833,
      "step": 88830
    },
    {
      "epoch": 25.217144479137097,
      "grad_norm": 0.6820372343063354,
      "learning_rate": 7.4793925631564e-05,
      "loss": 0.0071021273732185366,
      "step": 88840
    },
    {
      "epoch": 25.21998296906046,
      "grad_norm": 0.3963971436023712,
      "learning_rate": 7.479108714164065e-05,
      "loss": 0.005699237436056137,
      "step": 88850
    },
    {
      "epoch": 25.22282145898382,
      "grad_norm": 3.97796893119812,
      "learning_rate": 7.478824865171729e-05,
      "loss": 0.009984759241342544,
      "step": 88860
    },
    {
      "epoch": 25.225659948907182,
      "grad_norm": 3.2888407707214355,
      "learning_rate": 7.478541016179393e-05,
      "loss": 0.012810561060905456,
      "step": 88870
    },
    {
      "epoch": 25.22849843883054,
      "grad_norm": 10.101506233215332,
      "learning_rate": 7.478257167187057e-05,
      "loss": 0.006711839139461518,
      "step": 88880
    },
    {
      "epoch": 25.231336928753905,
      "grad_norm": 0.14629694819450378,
      "learning_rate": 7.477973318194721e-05,
      "loss": 0.008156283944845199,
      "step": 88890
    },
    {
      "epoch": 25.234175418677264,
      "grad_norm": 0.4291971027851105,
      "learning_rate": 7.477689469202384e-05,
      "loss": 0.010949918627738952,
      "step": 88900
    },
    {
      "epoch": 25.237013908600623,
      "grad_norm": 0.7845060229301453,
      "learning_rate": 7.477405620210048e-05,
      "loss": 0.003921416774392128,
      "step": 88910
    },
    {
      "epoch": 25.239852398523986,
      "grad_norm": 2.5105197429656982,
      "learning_rate": 7.477121771217712e-05,
      "loss": 0.023205177485942842,
      "step": 88920
    },
    {
      "epoch": 25.242690888447346,
      "grad_norm": 0.8168386816978455,
      "learning_rate": 7.476837922225376e-05,
      "loss": 0.0068645179271698,
      "step": 88930
    },
    {
      "epoch": 25.24552937837071,
      "grad_norm": 7.170585632324219,
      "learning_rate": 7.47655407323304e-05,
      "loss": 0.017696772515773774,
      "step": 88940
    },
    {
      "epoch": 25.248367868294068,
      "grad_norm": 1.9250679016113281,
      "learning_rate": 7.476270224240705e-05,
      "loss": 0.00790131688117981,
      "step": 88950
    },
    {
      "epoch": 25.251206358217427,
      "grad_norm": 0.5847057104110718,
      "learning_rate": 7.475986375248369e-05,
      "loss": 0.005492392182350159,
      "step": 88960
    },
    {
      "epoch": 25.25404484814079,
      "grad_norm": 0.412710964679718,
      "learning_rate": 7.475702526256032e-05,
      "loss": 0.008156956732273101,
      "step": 88970
    },
    {
      "epoch": 25.25688333806415,
      "grad_norm": 3.0751490592956543,
      "learning_rate": 7.475418677263696e-05,
      "loss": 0.006053932383656502,
      "step": 88980
    },
    {
      "epoch": 25.259721827987512,
      "grad_norm": 2.2361323833465576,
      "learning_rate": 7.47513482827136e-05,
      "loss": 0.010091498494148254,
      "step": 88990
    },
    {
      "epoch": 25.26256031791087,
      "grad_norm": 2.423135995864868,
      "learning_rate": 7.474850979279024e-05,
      "loss": 0.007437251508235931,
      "step": 89000
    },
    {
      "epoch": 25.26256031791087,
      "eval_accuracy": 0.9684618808418644,
      "eval_loss": 0.1019994467496872,
      "eval_runtime": 31.1684,
      "eval_samples_per_second": 504.582,
      "eval_steps_per_second": 7.893,
      "step": 89000
    },
    {
      "epoch": 25.26539880783423,
      "grad_norm": 1.7267898321151733,
      "learning_rate": 7.474567130286688e-05,
      "loss": 0.008317448943853379,
      "step": 89010
    },
    {
      "epoch": 25.268237297757594,
      "grad_norm": 8.916865348815918,
      "learning_rate": 7.474283281294352e-05,
      "loss": 0.012136688083410263,
      "step": 89020
    },
    {
      "epoch": 25.271075787680953,
      "grad_norm": 0.3587147295475006,
      "learning_rate": 7.473999432302015e-05,
      "loss": 0.015295733511447907,
      "step": 89030
    },
    {
      "epoch": 25.273914277604316,
      "grad_norm": 13.232930183410645,
      "learning_rate": 7.473715583309679e-05,
      "loss": 0.008462396264076234,
      "step": 89040
    },
    {
      "epoch": 25.276752767527675,
      "grad_norm": 2.6565041542053223,
      "learning_rate": 7.473431734317343e-05,
      "loss": 0.006883489340543747,
      "step": 89050
    },
    {
      "epoch": 25.279591257451035,
      "grad_norm": 1.7084414958953857,
      "learning_rate": 7.473147885325008e-05,
      "loss": 0.019288837909698486,
      "step": 89060
    },
    {
      "epoch": 25.282429747374398,
      "grad_norm": 0.9162612557411194,
      "learning_rate": 7.472864036332672e-05,
      "loss": 0.01957191973924637,
      "step": 89070
    },
    {
      "epoch": 25.285268237297757,
      "grad_norm": 0.14706148207187653,
      "learning_rate": 7.472580187340336e-05,
      "loss": 0.010636127740144729,
      "step": 89080
    },
    {
      "epoch": 25.28810672722112,
      "grad_norm": 0.7329239845275879,
      "learning_rate": 7.472296338348e-05,
      "loss": 0.003111882507801056,
      "step": 89090
    },
    {
      "epoch": 25.29094521714448,
      "grad_norm": 1.0475722551345825,
      "learning_rate": 7.472012489355663e-05,
      "loss": 0.006907813996076584,
      "step": 89100
    },
    {
      "epoch": 25.29378370706784,
      "grad_norm": 3.6275408267974854,
      "learning_rate": 7.471728640363327e-05,
      "loss": 0.006338354200124741,
      "step": 89110
    },
    {
      "epoch": 25.2966221969912,
      "grad_norm": 0.8380832076072693,
      "learning_rate": 7.471444791370991e-05,
      "loss": 0.008261603116989136,
      "step": 89120
    },
    {
      "epoch": 25.29946068691456,
      "grad_norm": 8.231410026550293,
      "learning_rate": 7.471160942378654e-05,
      "loss": 0.018148744106292726,
      "step": 89130
    },
    {
      "epoch": 25.302299176837924,
      "grad_norm": 9.821887016296387,
      "learning_rate": 7.470877093386319e-05,
      "loss": 0.01976998746395111,
      "step": 89140
    },
    {
      "epoch": 25.305137666761283,
      "grad_norm": 2.143378973007202,
      "learning_rate": 7.470593244393983e-05,
      "loss": 0.014833828806877137,
      "step": 89150
    },
    {
      "epoch": 25.307976156684642,
      "grad_norm": 0.8944421410560608,
      "learning_rate": 7.470309395401646e-05,
      "loss": 0.008533579856157303,
      "step": 89160
    },
    {
      "epoch": 25.310814646608005,
      "grad_norm": 4.758836269378662,
      "learning_rate": 7.47002554640931e-05,
      "loss": 0.010344582051038742,
      "step": 89170
    },
    {
      "epoch": 25.313653136531364,
      "grad_norm": 5.470677375793457,
      "learning_rate": 7.469741697416974e-05,
      "loss": 0.008317559957504272,
      "step": 89180
    },
    {
      "epoch": 25.316491626454727,
      "grad_norm": 5.148582935333252,
      "learning_rate": 7.469457848424639e-05,
      "loss": 0.00568288154900074,
      "step": 89190
    },
    {
      "epoch": 25.319330116378087,
      "grad_norm": 0.5280429720878601,
      "learning_rate": 7.469173999432303e-05,
      "loss": 0.014339965581893922,
      "step": 89200
    },
    {
      "epoch": 25.322168606301446,
      "grad_norm": 0.13519060611724854,
      "learning_rate": 7.468890150439967e-05,
      "loss": 0.0030381591990590096,
      "step": 89210
    },
    {
      "epoch": 25.32500709622481,
      "grad_norm": 0.805263876914978,
      "learning_rate": 7.468606301447631e-05,
      "loss": 0.006089410930871964,
      "step": 89220
    },
    {
      "epoch": 25.327845586148168,
      "grad_norm": 10.490825653076172,
      "learning_rate": 7.468322452455294e-05,
      "loss": 0.00943363755941391,
      "step": 89230
    },
    {
      "epoch": 25.33068407607153,
      "grad_norm": 9.392050743103027,
      "learning_rate": 7.468038603462958e-05,
      "loss": 0.019899250566959382,
      "step": 89240
    },
    {
      "epoch": 25.33352256599489,
      "grad_norm": 0.14821508526802063,
      "learning_rate": 7.467754754470622e-05,
      "loss": 0.01045924872159958,
      "step": 89250
    },
    {
      "epoch": 25.336361055918253,
      "grad_norm": 11.088156700134277,
      "learning_rate": 7.467470905478285e-05,
      "loss": 0.01059696450829506,
      "step": 89260
    },
    {
      "epoch": 25.339199545841613,
      "grad_norm": 2.051734209060669,
      "learning_rate": 7.46718705648595e-05,
      "loss": 0.004252380132675171,
      "step": 89270
    },
    {
      "epoch": 25.342038035764972,
      "grad_norm": 5.370089054107666,
      "learning_rate": 7.466903207493614e-05,
      "loss": 0.0049885731190443035,
      "step": 89280
    },
    {
      "epoch": 25.344876525688335,
      "grad_norm": 0.7628191113471985,
      "learning_rate": 7.466619358501277e-05,
      "loss": 0.017790311574935914,
      "step": 89290
    },
    {
      "epoch": 25.347715015611694,
      "grad_norm": 2.1906838417053223,
      "learning_rate": 7.466335509508941e-05,
      "loss": 0.009002625942230225,
      "step": 89300
    },
    {
      "epoch": 25.350553505535057,
      "grad_norm": 0.9409029483795166,
      "learning_rate": 7.466051660516606e-05,
      "loss": 0.010196210443973541,
      "step": 89310
    },
    {
      "epoch": 25.353391995458416,
      "grad_norm": 10.652228355407715,
      "learning_rate": 7.46576781152427e-05,
      "loss": 0.009210561960935592,
      "step": 89320
    },
    {
      "epoch": 25.356230485381776,
      "grad_norm": 0.5737007260322571,
      "learning_rate": 7.465483962531932e-05,
      "loss": 0.027372828125953673,
      "step": 89330
    },
    {
      "epoch": 25.35906897530514,
      "grad_norm": 1.1164965629577637,
      "learning_rate": 7.46522849843883e-05,
      "loss": 0.027994710206985473,
      "step": 89340
    },
    {
      "epoch": 25.361907465228498,
      "grad_norm": 1.7253034114837646,
      "learning_rate": 7.464944649446495e-05,
      "loss": 0.013909734785556793,
      "step": 89350
    },
    {
      "epoch": 25.36474595515186,
      "grad_norm": 6.285429954528809,
      "learning_rate": 7.464660800454159e-05,
      "loss": 0.0035476937890052797,
      "step": 89360
    },
    {
      "epoch": 25.36758444507522,
      "grad_norm": 0.17200380563735962,
      "learning_rate": 7.464376951461823e-05,
      "loss": 0.004537386819720268,
      "step": 89370
    },
    {
      "epoch": 25.37042293499858,
      "grad_norm": 0.19730165600776672,
      "learning_rate": 7.464093102469487e-05,
      "loss": 0.011376288533210755,
      "step": 89380
    },
    {
      "epoch": 25.373261424921942,
      "grad_norm": 1.7278491258621216,
      "learning_rate": 7.463809253477151e-05,
      "loss": 0.01840505748987198,
      "step": 89390
    },
    {
      "epoch": 25.3760999148453,
      "grad_norm": 6.291707515716553,
      "learning_rate": 7.463525404484815e-05,
      "loss": 0.02354396879673004,
      "step": 89400
    },
    {
      "epoch": 25.378938404768665,
      "grad_norm": 0.22226478159427643,
      "learning_rate": 7.463241555492478e-05,
      "loss": 0.004655932635068893,
      "step": 89410
    },
    {
      "epoch": 25.381776894692024,
      "grad_norm": 0.7492541670799255,
      "learning_rate": 7.462957706500142e-05,
      "loss": 0.010103525966405869,
      "step": 89420
    },
    {
      "epoch": 25.384615384615383,
      "grad_norm": 10.672445297241211,
      "learning_rate": 7.462673857507806e-05,
      "loss": 0.005995461717247963,
      "step": 89430
    },
    {
      "epoch": 25.387453874538746,
      "grad_norm": 0.5955822467803955,
      "learning_rate": 7.462390008515469e-05,
      "loss": 0.0016033813357353211,
      "step": 89440
    },
    {
      "epoch": 25.390292364462105,
      "grad_norm": 10.716609001159668,
      "learning_rate": 7.462106159523135e-05,
      "loss": 0.008373571932315827,
      "step": 89450
    },
    {
      "epoch": 25.39313085438547,
      "grad_norm": 0.0685468316078186,
      "learning_rate": 7.461822310530799e-05,
      "loss": 0.01330282837152481,
      "step": 89460
    },
    {
      "epoch": 25.395969344308828,
      "grad_norm": 0.5304935574531555,
      "learning_rate": 7.461538461538462e-05,
      "loss": 0.007710562646389007,
      "step": 89470
    },
    {
      "epoch": 25.398807834232187,
      "grad_norm": 2.178284168243408,
      "learning_rate": 7.461254612546126e-05,
      "loss": 0.012113523483276368,
      "step": 89480
    },
    {
      "epoch": 25.40164632415555,
      "grad_norm": 2.3107974529266357,
      "learning_rate": 7.46097076355379e-05,
      "loss": 0.004521171748638153,
      "step": 89490
    },
    {
      "epoch": 25.40448481407891,
      "grad_norm": 1.8620744943618774,
      "learning_rate": 7.460686914561454e-05,
      "loss": 0.011105257272720336,
      "step": 89500
    },
    {
      "epoch": 25.40448481407891,
      "eval_accuracy": 0.968779805430152,
      "eval_loss": 0.09675829112529755,
      "eval_runtime": 32.0178,
      "eval_samples_per_second": 491.196,
      "eval_steps_per_second": 7.683,
      "step": 89500
    },
    {
      "epoch": 25.407323304002272,
      "grad_norm": 0.47470375895500183,
      "learning_rate": 7.460403065569117e-05,
      "loss": 0.014409217238426208,
      "step": 89510
    },
    {
      "epoch": 25.41016179392563,
      "grad_norm": 9.425232887268066,
      "learning_rate": 7.460119216576782e-05,
      "loss": 0.018158724904060362,
      "step": 89520
    },
    {
      "epoch": 25.41300028384899,
      "grad_norm": 0.690945029258728,
      "learning_rate": 7.459835367584446e-05,
      "loss": 0.008444498479366302,
      "step": 89530
    },
    {
      "epoch": 25.415838773772354,
      "grad_norm": 0.20545555651187897,
      "learning_rate": 7.459551518592109e-05,
      "loss": 0.008665291965007782,
      "step": 89540
    },
    {
      "epoch": 25.418677263695713,
      "grad_norm": 0.12948764860630035,
      "learning_rate": 7.459267669599773e-05,
      "loss": 0.014690090715885163,
      "step": 89550
    },
    {
      "epoch": 25.421515753619076,
      "grad_norm": 0.17304100096225739,
      "learning_rate": 7.458983820607437e-05,
      "loss": 0.007998335361480712,
      "step": 89560
    },
    {
      "epoch": 25.424354243542435,
      "grad_norm": 0.2681643068790436,
      "learning_rate": 7.4586999716151e-05,
      "loss": 0.012142705917358398,
      "step": 89570
    },
    {
      "epoch": 25.427192733465795,
      "grad_norm": 1.4312011003494263,
      "learning_rate": 7.458416122622766e-05,
      "loss": 0.013508948683738708,
      "step": 89580
    },
    {
      "epoch": 25.430031223389157,
      "grad_norm": 4.251752853393555,
      "learning_rate": 7.45813227363043e-05,
      "loss": 0.03405413031578064,
      "step": 89590
    },
    {
      "epoch": 25.432869713312517,
      "grad_norm": 0.7419965267181396,
      "learning_rate": 7.457848424638093e-05,
      "loss": 0.021492759883403777,
      "step": 89600
    },
    {
      "epoch": 25.43570820323588,
      "grad_norm": 12.000972747802734,
      "learning_rate": 7.457564575645757e-05,
      "loss": 0.00889514535665512,
      "step": 89610
    },
    {
      "epoch": 25.43854669315924,
      "grad_norm": 0.7974718809127808,
      "learning_rate": 7.457280726653421e-05,
      "loss": 0.007174207270145417,
      "step": 89620
    },
    {
      "epoch": 25.4413851830826,
      "grad_norm": 1.3605254888534546,
      "learning_rate": 7.456996877661085e-05,
      "loss": 0.024742935597896577,
      "step": 89630
    },
    {
      "epoch": 25.44422367300596,
      "grad_norm": 2.2025234699249268,
      "learning_rate": 7.456713028668748e-05,
      "loss": 0.009747720509767532,
      "step": 89640
    },
    {
      "epoch": 25.44706216292932,
      "grad_norm": 13.632218360900879,
      "learning_rate": 7.456429179676413e-05,
      "loss": 0.016755563020706177,
      "step": 89650
    },
    {
      "epoch": 25.449900652852683,
      "grad_norm": 5.678091049194336,
      "learning_rate": 7.456145330684076e-05,
      "loss": 0.00833926498889923,
      "step": 89660
    },
    {
      "epoch": 25.452739142776043,
      "grad_norm": 0.250432550907135,
      "learning_rate": 7.45586148169174e-05,
      "loss": 0.00542074404656887,
      "step": 89670
    },
    {
      "epoch": 25.455577632699406,
      "grad_norm": 0.35750293731689453,
      "learning_rate": 7.455577632699404e-05,
      "loss": 0.004305268079042435,
      "step": 89680
    },
    {
      "epoch": 25.458416122622765,
      "grad_norm": 1.9380242824554443,
      "learning_rate": 7.455293783707068e-05,
      "loss": 0.0034865133464336397,
      "step": 89690
    },
    {
      "epoch": 25.461254612546124,
      "grad_norm": 1.7275127172470093,
      "learning_rate": 7.455009934714731e-05,
      "loss": 0.01373683512210846,
      "step": 89700
    },
    {
      "epoch": 25.464093102469487,
      "grad_norm": 2.633692502975464,
      "learning_rate": 7.454726085722395e-05,
      "loss": 0.007697141170501709,
      "step": 89710
    },
    {
      "epoch": 25.466931592392847,
      "grad_norm": 0.37864720821380615,
      "learning_rate": 7.454442236730061e-05,
      "loss": 0.006617267429828644,
      "step": 89720
    },
    {
      "epoch": 25.46977008231621,
      "grad_norm": 0.46586552262306213,
      "learning_rate": 7.454158387737724e-05,
      "loss": 0.015822188556194307,
      "step": 89730
    },
    {
      "epoch": 25.47260857223957,
      "grad_norm": 6.255487442016602,
      "learning_rate": 7.453874538745388e-05,
      "loss": 0.019852426648139954,
      "step": 89740
    },
    {
      "epoch": 25.475447062162928,
      "grad_norm": 1.8029837608337402,
      "learning_rate": 7.453590689753052e-05,
      "loss": 0.007124315202236176,
      "step": 89750
    },
    {
      "epoch": 25.47828555208629,
      "grad_norm": 5.727311134338379,
      "learning_rate": 7.453306840760715e-05,
      "loss": 0.007385125011205673,
      "step": 89760
    },
    {
      "epoch": 25.48112404200965,
      "grad_norm": 1.282725214958191,
      "learning_rate": 7.453022991768379e-05,
      "loss": 0.004679073393344879,
      "step": 89770
    },
    {
      "epoch": 25.483962531933013,
      "grad_norm": 0.2311718761920929,
      "learning_rate": 7.452739142776044e-05,
      "loss": 0.0028640016913414,
      "step": 89780
    },
    {
      "epoch": 25.486801021856373,
      "grad_norm": 0.18721166253089905,
      "learning_rate": 7.452455293783707e-05,
      "loss": 0.002139225974678993,
      "step": 89790
    },
    {
      "epoch": 25.489639511779732,
      "grad_norm": 0.42017048597335815,
      "learning_rate": 7.452171444791371e-05,
      "loss": 0.010600708425045013,
      "step": 89800
    },
    {
      "epoch": 25.492478001703095,
      "grad_norm": 0.6089770197868347,
      "learning_rate": 7.451887595799035e-05,
      "loss": 0.003857284039258957,
      "step": 89810
    },
    {
      "epoch": 25.495316491626454,
      "grad_norm": 12.837237358093262,
      "learning_rate": 7.4516037468067e-05,
      "loss": 0.011503615230321885,
      "step": 89820
    },
    {
      "epoch": 25.498154981549817,
      "grad_norm": 2.9708023071289062,
      "learning_rate": 7.451319897814362e-05,
      "loss": 0.010358942300081253,
      "step": 89830
    },
    {
      "epoch": 25.500993471473176,
      "grad_norm": 0.6784881949424744,
      "learning_rate": 7.451036048822026e-05,
      "loss": 0.012020077556371689,
      "step": 89840
    },
    {
      "epoch": 25.503831961396536,
      "grad_norm": 5.127619743347168,
      "learning_rate": 7.450752199829692e-05,
      "loss": 0.017242418229579927,
      "step": 89850
    },
    {
      "epoch": 25.5066704513199,
      "grad_norm": 6.843567848205566,
      "learning_rate": 7.450468350837355e-05,
      "loss": 0.018272146582603455,
      "step": 89860
    },
    {
      "epoch": 25.509508941243258,
      "grad_norm": 2.332331895828247,
      "learning_rate": 7.450184501845019e-05,
      "loss": 0.007542164623737335,
      "step": 89870
    },
    {
      "epoch": 25.51234743116662,
      "grad_norm": 4.345371246337891,
      "learning_rate": 7.449900652852683e-05,
      "loss": 0.009639744460582734,
      "step": 89880
    },
    {
      "epoch": 25.51518592108998,
      "grad_norm": 12.709461212158203,
      "learning_rate": 7.449616803860346e-05,
      "loss": 0.014072233438491821,
      "step": 89890
    },
    {
      "epoch": 25.51802441101334,
      "grad_norm": 0.20448452234268188,
      "learning_rate": 7.44933295486801e-05,
      "loss": 0.007963302731513976,
      "step": 89900
    },
    {
      "epoch": 25.520862900936702,
      "grad_norm": 13.476292610168457,
      "learning_rate": 7.449049105875674e-05,
      "loss": 0.019671632349491118,
      "step": 89910
    },
    {
      "epoch": 25.52370139086006,
      "grad_norm": 0.4545043110847473,
      "learning_rate": 7.448765256883338e-05,
      "loss": 0.0026844840496778486,
      "step": 89920
    },
    {
      "epoch": 25.526539880783425,
      "grad_norm": 0.33921900391578674,
      "learning_rate": 7.448481407891002e-05,
      "loss": 0.010314853489398956,
      "step": 89930
    },
    {
      "epoch": 25.529378370706784,
      "grad_norm": 4.510886192321777,
      "learning_rate": 7.448197558898666e-05,
      "loss": 0.009906703233718872,
      "step": 89940
    },
    {
      "epoch": 25.532216860630143,
      "grad_norm": 4.679142951965332,
      "learning_rate": 7.44791370990633e-05,
      "loss": 0.006335891783237457,
      "step": 89950
    },
    {
      "epoch": 25.535055350553506,
      "grad_norm": 10.095640182495117,
      "learning_rate": 7.447629860913993e-05,
      "loss": 0.006000934541225434,
      "step": 89960
    },
    {
      "epoch": 25.537893840476865,
      "grad_norm": 2.538888692855835,
      "learning_rate": 7.447346011921658e-05,
      "loss": 0.010891618579626084,
      "step": 89970
    },
    {
      "epoch": 25.54073233040023,
      "grad_norm": 1.4497588872909546,
      "learning_rate": 7.447062162929323e-05,
      "loss": 0.013632899522781372,
      "step": 89980
    },
    {
      "epoch": 25.543570820323588,
      "grad_norm": 14.750211715698242,
      "learning_rate": 7.446778313936986e-05,
      "loss": 0.021562840044498443,
      "step": 89990
    },
    {
      "epoch": 25.546409310246947,
      "grad_norm": 0.5189008712768555,
      "learning_rate": 7.44649446494465e-05,
      "loss": 0.008615458756685257,
      "step": 90000
    },
    {
      "epoch": 25.546409310246947,
      "eval_accuracy": 0.9682711260888918,
      "eval_loss": 0.10716607421636581,
      "eval_runtime": 32.1401,
      "eval_samples_per_second": 489.327,
      "eval_steps_per_second": 7.654,
      "step": 90000
    },
    {
      "epoch": 25.54924780017031,
      "grad_norm": 0.38540369272232056,
      "learning_rate": 7.446210615952314e-05,
      "loss": 0.011609244346618652,
      "step": 90010
    },
    {
      "epoch": 25.55208629009367,
      "grad_norm": 1.0785837173461914,
      "learning_rate": 7.445926766959977e-05,
      "loss": 0.008320166915655135,
      "step": 90020
    },
    {
      "epoch": 25.554924780017032,
      "grad_norm": 2.063633441925049,
      "learning_rate": 7.445642917967641e-05,
      "loss": 0.010965348035097123,
      "step": 90030
    },
    {
      "epoch": 25.55776326994039,
      "grad_norm": 1.066653847694397,
      "learning_rate": 7.445359068975305e-05,
      "loss": 0.004671579971909523,
      "step": 90040
    },
    {
      "epoch": 25.56060175986375,
      "grad_norm": 8.457413673400879,
      "learning_rate": 7.445075219982969e-05,
      "loss": 0.021729104220867157,
      "step": 90050
    },
    {
      "epoch": 25.563440249787114,
      "grad_norm": 8.392313003540039,
      "learning_rate": 7.444791370990633e-05,
      "loss": 0.013003981113433838,
      "step": 90060
    },
    {
      "epoch": 25.566278739710473,
      "grad_norm": 4.624108791351318,
      "learning_rate": 7.444507521998298e-05,
      "loss": 0.013246238231658936,
      "step": 90070
    },
    {
      "epoch": 25.569117229633836,
      "grad_norm": 9.892694473266602,
      "learning_rate": 7.444223673005962e-05,
      "loss": 0.01721630394458771,
      "step": 90080
    },
    {
      "epoch": 25.571955719557195,
      "grad_norm": 1.4449228048324585,
      "learning_rate": 7.443939824013624e-05,
      "loss": 0.010924828052520753,
      "step": 90090
    },
    {
      "epoch": 25.574794209480558,
      "grad_norm": 2.0210015773773193,
      "learning_rate": 7.443655975021289e-05,
      "loss": 0.008521143347024918,
      "step": 90100
    },
    {
      "epoch": 25.577632699403917,
      "grad_norm": 0.676954448223114,
      "learning_rate": 7.443372126028954e-05,
      "loss": 0.01687990576028824,
      "step": 90110
    },
    {
      "epoch": 25.580471189327277,
      "grad_norm": 0.1836266964673996,
      "learning_rate": 7.443088277036617e-05,
      "loss": 0.001654258742928505,
      "step": 90120
    },
    {
      "epoch": 25.58330967925064,
      "grad_norm": 2.2601585388183594,
      "learning_rate": 7.442804428044281e-05,
      "loss": 0.02085067331790924,
      "step": 90130
    },
    {
      "epoch": 25.586148169174,
      "grad_norm": 0.49517562985420227,
      "learning_rate": 7.442520579051945e-05,
      "loss": 0.00798189789056778,
      "step": 90140
    },
    {
      "epoch": 25.588986659097362,
      "grad_norm": 1.1464717388153076,
      "learning_rate": 7.442236730059608e-05,
      "loss": 0.003812873363494873,
      "step": 90150
    },
    {
      "epoch": 25.59182514902072,
      "grad_norm": 0.11964661628007889,
      "learning_rate": 7.441952881067272e-05,
      "loss": 0.0021910477429628373,
      "step": 90160
    },
    {
      "epoch": 25.59466363894408,
      "grad_norm": 7.753471374511719,
      "learning_rate": 7.441669032074936e-05,
      "loss": 0.0059756219387054445,
      "step": 90170
    },
    {
      "epoch": 25.597502128867443,
      "grad_norm": 0.2821861505508423,
      "learning_rate": 7.4413851830826e-05,
      "loss": 0.004846497625112534,
      "step": 90180
    },
    {
      "epoch": 25.600340618790803,
      "grad_norm": 3.1929540634155273,
      "learning_rate": 7.441101334090265e-05,
      "loss": 0.013284030556678771,
      "step": 90190
    },
    {
      "epoch": 25.603179108714166,
      "grad_norm": 2.5116124153137207,
      "learning_rate": 7.440817485097929e-05,
      "loss": 0.004954066127538681,
      "step": 90200
    },
    {
      "epoch": 25.606017598637525,
      "grad_norm": 0.8399869203567505,
      "learning_rate": 7.440533636105593e-05,
      "loss": 0.003288187086582184,
      "step": 90210
    },
    {
      "epoch": 25.608856088560884,
      "grad_norm": 3.6743085384368896,
      "learning_rate": 7.440249787113256e-05,
      "loss": 0.007133663445711136,
      "step": 90220
    },
    {
      "epoch": 25.611694578484247,
      "grad_norm": 0.6636117696762085,
      "learning_rate": 7.43996593812092e-05,
      "loss": 0.003915078938007355,
      "step": 90230
    },
    {
      "epoch": 25.614533068407606,
      "grad_norm": 2.7770304679870605,
      "learning_rate": 7.439682089128584e-05,
      "loss": 0.006891221553087234,
      "step": 90240
    },
    {
      "epoch": 25.61737155833097,
      "grad_norm": 12.426870346069336,
      "learning_rate": 7.439398240136248e-05,
      "loss": 0.013340204954147339,
      "step": 90250
    },
    {
      "epoch": 25.62021004825433,
      "grad_norm": 4.590137481689453,
      "learning_rate": 7.439114391143912e-05,
      "loss": 0.020398013293743134,
      "step": 90260
    },
    {
      "epoch": 25.623048538177688,
      "grad_norm": 0.472206711769104,
      "learning_rate": 7.438830542151576e-05,
      "loss": 0.009042491763830185,
      "step": 90270
    },
    {
      "epoch": 25.62588702810105,
      "grad_norm": 1.9922901391983032,
      "learning_rate": 7.438546693159239e-05,
      "loss": 0.009527622908353805,
      "step": 90280
    },
    {
      "epoch": 25.62872551802441,
      "grad_norm": 0.19202233850955963,
      "learning_rate": 7.438262844166903e-05,
      "loss": 0.016148528456687926,
      "step": 90290
    },
    {
      "epoch": 25.631564007947773,
      "grad_norm": 0.6510984897613525,
      "learning_rate": 7.437978995174567e-05,
      "loss": 0.011894938349723817,
      "step": 90300
    },
    {
      "epoch": 25.634402497871132,
      "grad_norm": 9.376468658447266,
      "learning_rate": 7.437695146182231e-05,
      "loss": 0.02315758615732193,
      "step": 90310
    },
    {
      "epoch": 25.637240987794492,
      "grad_norm": 1.9915847778320312,
      "learning_rate": 7.437411297189896e-05,
      "loss": 0.013033168017864227,
      "step": 90320
    },
    {
      "epoch": 25.640079477717855,
      "grad_norm": 0.18826636672019958,
      "learning_rate": 7.43712744819756e-05,
      "loss": 0.004606899619102478,
      "step": 90330
    },
    {
      "epoch": 25.642917967641214,
      "grad_norm": 2.8383915424346924,
      "learning_rate": 7.436843599205224e-05,
      "loss": 0.015871141850948334,
      "step": 90340
    },
    {
      "epoch": 25.645756457564577,
      "grad_norm": 3.208726644515991,
      "learning_rate": 7.436559750212887e-05,
      "loss": 0.009737040847539902,
      "step": 90350
    },
    {
      "epoch": 25.648594947487936,
      "grad_norm": 0.5642299056053162,
      "learning_rate": 7.436275901220551e-05,
      "loss": 0.0032096564769744872,
      "step": 90360
    },
    {
      "epoch": 25.651433437411296,
      "grad_norm": 2.0800364017486572,
      "learning_rate": 7.435992052228215e-05,
      "loss": 0.005873514711856842,
      "step": 90370
    },
    {
      "epoch": 25.65427192733466,
      "grad_norm": 1.5962680578231812,
      "learning_rate": 7.435708203235879e-05,
      "loss": 0.007971050590276718,
      "step": 90380
    },
    {
      "epoch": 25.657110417258018,
      "grad_norm": 1.4835530519485474,
      "learning_rate": 7.435424354243543e-05,
      "loss": 0.019242051243782043,
      "step": 90390
    },
    {
      "epoch": 25.65994890718138,
      "grad_norm": 0.503690242767334,
      "learning_rate": 7.435140505251207e-05,
      "loss": 0.0186344712972641,
      "step": 90400
    },
    {
      "epoch": 25.66278739710474,
      "grad_norm": 1.3899058103561401,
      "learning_rate": 7.43485665625887e-05,
      "loss": 0.008043307811021805,
      "step": 90410
    },
    {
      "epoch": 25.6656258870281,
      "grad_norm": 0.07085864245891571,
      "learning_rate": 7.434572807266534e-05,
      "loss": 0.003644735366106033,
      "step": 90420
    },
    {
      "epoch": 25.668464376951462,
      "grad_norm": 0.8334052562713623,
      "learning_rate": 7.434288958274198e-05,
      "loss": 0.007996916770935059,
      "step": 90430
    },
    {
      "epoch": 25.67130286687482,
      "grad_norm": 5.77651309967041,
      "learning_rate": 7.434005109281863e-05,
      "loss": 0.014565524458885194,
      "step": 90440
    },
    {
      "epoch": 25.674141356798184,
      "grad_norm": 3.1890127658843994,
      "learning_rate": 7.433721260289527e-05,
      "loss": 0.003922684863209724,
      "step": 90450
    },
    {
      "epoch": 25.676979846721544,
      "grad_norm": 14.084465980529785,
      "learning_rate": 7.433437411297191e-05,
      "loss": 0.015540170669555663,
      "step": 90460
    },
    {
      "epoch": 25.679818336644907,
      "grad_norm": 0.9925637245178223,
      "learning_rate": 7.433153562304855e-05,
      "loss": 0.007200963795185089,
      "step": 90470
    },
    {
      "epoch": 25.682656826568266,
      "grad_norm": 1.2706853151321411,
      "learning_rate": 7.432869713312518e-05,
      "loss": 0.013453133404254913,
      "step": 90480
    },
    {
      "epoch": 25.685495316491625,
      "grad_norm": 2.0845508575439453,
      "learning_rate": 7.432585864320182e-05,
      "loss": 0.017751027643680573,
      "step": 90490
    },
    {
      "epoch": 25.688333806414988,
      "grad_norm": 1.5505220890045166,
      "learning_rate": 7.432302015327846e-05,
      "loss": 0.00464121550321579,
      "step": 90500
    },
    {
      "epoch": 25.688333806414988,
      "eval_accuracy": 0.9701786736186176,
      "eval_loss": 0.10010858625173569,
      "eval_runtime": 31.4936,
      "eval_samples_per_second": 499.371,
      "eval_steps_per_second": 7.811,
      "step": 90500
    },
    {
      "epoch": 25.691172296338348,
      "grad_norm": 0.6962722539901733,
      "learning_rate": 7.43201816633551e-05,
      "loss": 0.007226504385471344,
      "step": 90510
    },
    {
      "epoch": 25.69401078626171,
      "grad_norm": 14.827392578125,
      "learning_rate": 7.431734317343174e-05,
      "loss": 0.012569423019886016,
      "step": 90520
    },
    {
      "epoch": 25.69684927618507,
      "grad_norm": 0.6234973669052124,
      "learning_rate": 7.431450468350838e-05,
      "loss": 0.005498065799474716,
      "step": 90530
    },
    {
      "epoch": 25.69968776610843,
      "grad_norm": 4.237894058227539,
      "learning_rate": 7.431166619358501e-05,
      "loss": 0.007270877063274383,
      "step": 90540
    },
    {
      "epoch": 25.702526256031792,
      "grad_norm": 3.3369369506835938,
      "learning_rate": 7.430882770366165e-05,
      "loss": 0.010328102111816406,
      "step": 90550
    },
    {
      "epoch": 25.70536474595515,
      "grad_norm": 0.22364307940006256,
      "learning_rate": 7.43059892137383e-05,
      "loss": 0.010292800515890122,
      "step": 90560
    },
    {
      "epoch": 25.708203235878514,
      "grad_norm": 0.38995879888534546,
      "learning_rate": 7.430315072381494e-05,
      "loss": 0.009616436064243316,
      "step": 90570
    },
    {
      "epoch": 25.711041725801874,
      "grad_norm": 0.827457070350647,
      "learning_rate": 7.430031223389158e-05,
      "loss": 0.011696966737508774,
      "step": 90580
    },
    {
      "epoch": 25.713880215725233,
      "grad_norm": 12.106224060058594,
      "learning_rate": 7.429747374396822e-05,
      "loss": 0.014079192280769348,
      "step": 90590
    },
    {
      "epoch": 25.716718705648596,
      "grad_norm": 3.9957568645477295,
      "learning_rate": 7.429463525404485e-05,
      "loss": 0.014243847131729126,
      "step": 90600
    },
    {
      "epoch": 25.719557195571955,
      "grad_norm": 0.22086453437805176,
      "learning_rate": 7.429179676412149e-05,
      "loss": 0.012242984771728516,
      "step": 90610
    },
    {
      "epoch": 25.722395685495318,
      "grad_norm": 0.5101515054702759,
      "learning_rate": 7.428895827419813e-05,
      "loss": 0.007430297881364822,
      "step": 90620
    },
    {
      "epoch": 25.725234175418677,
      "grad_norm": 3.7444708347320557,
      "learning_rate": 7.428611978427477e-05,
      "loss": 0.007301381975412369,
      "step": 90630
    },
    {
      "epoch": 25.728072665342037,
      "grad_norm": 8.901936531066895,
      "learning_rate": 7.42832812943514e-05,
      "loss": 0.025249657034873963,
      "step": 90640
    },
    {
      "epoch": 25.7309111552654,
      "grad_norm": 3.650114059448242,
      "learning_rate": 7.428044280442805e-05,
      "loss": 0.012544295191764832,
      "step": 90650
    },
    {
      "epoch": 25.73374964518876,
      "grad_norm": 9.632060050964355,
      "learning_rate": 7.42776043145047e-05,
      "loss": 0.017318019270896913,
      "step": 90660
    },
    {
      "epoch": 25.73658813511212,
      "grad_norm": 6.719638347625732,
      "learning_rate": 7.427476582458132e-05,
      "loss": 0.017960447072982787,
      "step": 90670
    },
    {
      "epoch": 25.73942662503548,
      "grad_norm": 2.021409511566162,
      "learning_rate": 7.427192733465796e-05,
      "loss": 0.013045002520084382,
      "step": 90680
    },
    {
      "epoch": 25.74226511495884,
      "grad_norm": 1.2915898561477661,
      "learning_rate": 7.42690888447346e-05,
      "loss": 0.008352828025817872,
      "step": 90690
    },
    {
      "epoch": 25.745103604882203,
      "grad_norm": 10.643007278442383,
      "learning_rate": 7.426625035481123e-05,
      "loss": 0.012357166409492493,
      "step": 90700
    },
    {
      "epoch": 25.747942094805563,
      "grad_norm": 0.26296836137771606,
      "learning_rate": 7.426341186488789e-05,
      "loss": 0.01130620613694191,
      "step": 90710
    },
    {
      "epoch": 25.750780584728926,
      "grad_norm": 1.054850459098816,
      "learning_rate": 7.426057337496453e-05,
      "loss": 0.006417820602655411,
      "step": 90720
    },
    {
      "epoch": 25.753619074652285,
      "grad_norm": 2.2031800746917725,
      "learning_rate": 7.425773488504116e-05,
      "loss": 0.010071641206741333,
      "step": 90730
    },
    {
      "epoch": 25.756457564575644,
      "grad_norm": 3.632169246673584,
      "learning_rate": 7.42548963951178e-05,
      "loss": 0.004650980234146118,
      "step": 90740
    },
    {
      "epoch": 25.759296054499007,
      "grad_norm": 4.0324625968933105,
      "learning_rate": 7.425205790519444e-05,
      "loss": 0.004930272698402405,
      "step": 90750
    },
    {
      "epoch": 25.762134544422366,
      "grad_norm": 0.16914105415344238,
      "learning_rate": 7.424921941527108e-05,
      "loss": 0.0028475372120738028,
      "step": 90760
    },
    {
      "epoch": 25.76497303434573,
      "grad_norm": 0.7319206595420837,
      "learning_rate": 7.424638092534771e-05,
      "loss": 0.003315746784210205,
      "step": 90770
    },
    {
      "epoch": 25.76781152426909,
      "grad_norm": 0.6170145869255066,
      "learning_rate": 7.424354243542436e-05,
      "loss": 0.006154341995716095,
      "step": 90780
    },
    {
      "epoch": 25.770650014192448,
      "grad_norm": 3.030618667602539,
      "learning_rate": 7.4240703945501e-05,
      "loss": 0.009194561839103698,
      "step": 90790
    },
    {
      "epoch": 25.77348850411581,
      "grad_norm": 8.204113006591797,
      "learning_rate": 7.423786545557763e-05,
      "loss": 0.012365908920764923,
      "step": 90800
    },
    {
      "epoch": 25.77632699403917,
      "grad_norm": 0.29321184754371643,
      "learning_rate": 7.423502696565427e-05,
      "loss": 0.01490156501531601,
      "step": 90810
    },
    {
      "epoch": 25.779165483962533,
      "grad_norm": 2.304220199584961,
      "learning_rate": 7.423218847573092e-05,
      "loss": 0.008847502619028091,
      "step": 90820
    },
    {
      "epoch": 25.782003973885892,
      "grad_norm": 0.2627812325954437,
      "learning_rate": 7.422934998580754e-05,
      "loss": 0.0071276560425758365,
      "step": 90830
    },
    {
      "epoch": 25.784842463809255,
      "grad_norm": 2.2214231491088867,
      "learning_rate": 7.422651149588419e-05,
      "loss": 0.017530369758605956,
      "step": 90840
    },
    {
      "epoch": 25.787680953732615,
      "grad_norm": 0.30617737770080566,
      "learning_rate": 7.422367300596084e-05,
      "loss": 0.015485936403274536,
      "step": 90850
    },
    {
      "epoch": 25.790519443655974,
      "grad_norm": 7.192709445953369,
      "learning_rate": 7.422083451603747e-05,
      "loss": 0.021380624175071715,
      "step": 90860
    },
    {
      "epoch": 25.793357933579337,
      "grad_norm": 0.9066954851150513,
      "learning_rate": 7.421799602611411e-05,
      "loss": 0.03239285945892334,
      "step": 90870
    },
    {
      "epoch": 25.796196423502696,
      "grad_norm": 0.6990506052970886,
      "learning_rate": 7.421515753619075e-05,
      "loss": 0.018890266120433808,
      "step": 90880
    },
    {
      "epoch": 25.79903491342606,
      "grad_norm": 9.746612548828125,
      "learning_rate": 7.421231904626739e-05,
      "loss": 0.01640550196170807,
      "step": 90890
    },
    {
      "epoch": 25.80187340334942,
      "grad_norm": 0.8654089570045471,
      "learning_rate": 7.420948055634402e-05,
      "loss": 0.004177383705973625,
      "step": 90900
    },
    {
      "epoch": 25.804711893272778,
      "grad_norm": 6.421217918395996,
      "learning_rate": 7.420664206642067e-05,
      "loss": 0.005990825593471527,
      "step": 90910
    },
    {
      "epoch": 25.80755038319614,
      "grad_norm": 0.46521976590156555,
      "learning_rate": 7.420380357649732e-05,
      "loss": 0.019810928404331206,
      "step": 90920
    },
    {
      "epoch": 25.8103888731195,
      "grad_norm": 7.606039524078369,
      "learning_rate": 7.420096508657394e-05,
      "loss": 0.01322605162858963,
      "step": 90930
    },
    {
      "epoch": 25.813227363042863,
      "grad_norm": 18.673248291015625,
      "learning_rate": 7.419812659665059e-05,
      "loss": 0.013413110375404358,
      "step": 90940
    },
    {
      "epoch": 25.816065852966222,
      "grad_norm": 4.864353656768799,
      "learning_rate": 7.419528810672723e-05,
      "loss": 0.01897050142288208,
      "step": 90950
    },
    {
      "epoch": 25.81890434288958,
      "grad_norm": 2.738835096359253,
      "learning_rate": 7.419244961680385e-05,
      "loss": 0.005047548562288284,
      "step": 90960
    },
    {
      "epoch": 25.821742832812944,
      "grad_norm": 2.254920244216919,
      "learning_rate": 7.41896111268805e-05,
      "loss": 0.02924007773399353,
      "step": 90970
    },
    {
      "epoch": 25.824581322736304,
      "grad_norm": 7.6654863357543945,
      "learning_rate": 7.418677263695715e-05,
      "loss": 0.019563350081443786,
      "step": 90980
    },
    {
      "epoch": 25.827419812659667,
      "grad_norm": 1.1490751504898071,
      "learning_rate": 7.418393414703378e-05,
      "loss": 0.017334794998168944,
      "step": 90990
    },
    {
      "epoch": 25.830258302583026,
      "grad_norm": 3.7733311653137207,
      "learning_rate": 7.418109565711042e-05,
      "loss": 0.004975069314241409,
      "step": 91000
    },
    {
      "epoch": 25.830258302583026,
      "eval_accuracy": 0.967953201500604,
      "eval_loss": 0.10677766054868698,
      "eval_runtime": 32.3298,
      "eval_samples_per_second": 486.456,
      "eval_steps_per_second": 7.609,
      "step": 91000
    },
    {
      "epoch": 25.833096792506385,
      "grad_norm": 6.386855125427246,
      "learning_rate": 7.417825716718706e-05,
      "loss": 0.01648752987384796,
      "step": 91010
    },
    {
      "epoch": 25.835935282429748,
      "grad_norm": 10.907753944396973,
      "learning_rate": 7.41754186772637e-05,
      "loss": 0.009514281153678894,
      "step": 91020
    },
    {
      "epoch": 25.838773772353107,
      "grad_norm": 0.3697199821472168,
      "learning_rate": 7.417258018734033e-05,
      "loss": 0.008957795798778534,
      "step": 91030
    },
    {
      "epoch": 25.84161226227647,
      "grad_norm": 23.928796768188477,
      "learning_rate": 7.416974169741697e-05,
      "loss": 0.026893222332000734,
      "step": 91040
    },
    {
      "epoch": 25.84445075219983,
      "grad_norm": 14.434755325317383,
      "learning_rate": 7.416690320749363e-05,
      "loss": 0.013743598759174348,
      "step": 91050
    },
    {
      "epoch": 25.84728924212319,
      "grad_norm": 1.8203034400939941,
      "learning_rate": 7.416406471757025e-05,
      "loss": 0.012014111876487732,
      "step": 91060
    },
    {
      "epoch": 25.850127732046552,
      "grad_norm": 4.547553062438965,
      "learning_rate": 7.41612262276469e-05,
      "loss": 0.008239923417568207,
      "step": 91070
    },
    {
      "epoch": 25.85296622196991,
      "grad_norm": 0.6611328125,
      "learning_rate": 7.415838773772354e-05,
      "loss": 0.007966189086437226,
      "step": 91080
    },
    {
      "epoch": 25.855804711893274,
      "grad_norm": 0.7000671029090881,
      "learning_rate": 7.415554924780017e-05,
      "loss": 0.015919029712677002,
      "step": 91090
    },
    {
      "epoch": 25.858643201816633,
      "grad_norm": 2.305426597595215,
      "learning_rate": 7.415271075787681e-05,
      "loss": 0.030028027296066285,
      "step": 91100
    },
    {
      "epoch": 25.861481691739993,
      "grad_norm": 15.570962905883789,
      "learning_rate": 7.414987226795346e-05,
      "loss": 0.011820004880428314,
      "step": 91110
    },
    {
      "epoch": 25.864320181663356,
      "grad_norm": 4.1458258628845215,
      "learning_rate": 7.414703377803009e-05,
      "loss": 0.013002432882785797,
      "step": 91120
    },
    {
      "epoch": 25.867158671586715,
      "grad_norm": 1.4667894840240479,
      "learning_rate": 7.414419528810673e-05,
      "loss": 0.007879237830638885,
      "step": 91130
    },
    {
      "epoch": 25.869997161510078,
      "grad_norm": 0.21614038944244385,
      "learning_rate": 7.414135679818337e-05,
      "loss": 0.021773914992809295,
      "step": 91140
    },
    {
      "epoch": 25.872835651433437,
      "grad_norm": 0.7875774502754211,
      "learning_rate": 7.413851830826001e-05,
      "loss": 0.013640430569648743,
      "step": 91150
    },
    {
      "epoch": 25.875674141356797,
      "grad_norm": 3.9141488075256348,
      "learning_rate": 7.413567981833664e-05,
      "loss": 0.00488295815885067,
      "step": 91160
    },
    {
      "epoch": 25.87851263128016,
      "grad_norm": 0.9543119072914124,
      "learning_rate": 7.413284132841328e-05,
      "loss": 0.00411306694149971,
      "step": 91170
    },
    {
      "epoch": 25.88135112120352,
      "grad_norm": 0.23275023698806763,
      "learning_rate": 7.413000283848994e-05,
      "loss": 0.010814174264669418,
      "step": 91180
    },
    {
      "epoch": 25.88418961112688,
      "grad_norm": 0.147043839097023,
      "learning_rate": 7.412716434856657e-05,
      "loss": 0.0070279560983181,
      "step": 91190
    },
    {
      "epoch": 25.88702810105024,
      "grad_norm": 1.1091636419296265,
      "learning_rate": 7.412432585864321e-05,
      "loss": 0.010818889737129212,
      "step": 91200
    },
    {
      "epoch": 25.8898665909736,
      "grad_norm": 10.605774879455566,
      "learning_rate": 7.412148736871985e-05,
      "loss": 0.023306429386138916,
      "step": 91210
    },
    {
      "epoch": 25.892705080896963,
      "grad_norm": 0.2075026035308838,
      "learning_rate": 7.411864887879648e-05,
      "loss": 0.007010768353939057,
      "step": 91220
    },
    {
      "epoch": 25.895543570820323,
      "grad_norm": 1.0023187398910522,
      "learning_rate": 7.411581038887312e-05,
      "loss": 0.0034771181643009188,
      "step": 91230
    },
    {
      "epoch": 25.898382060743685,
      "grad_norm": 9.463793754577637,
      "learning_rate": 7.411297189894976e-05,
      "loss": 0.007321682572364807,
      "step": 91240
    },
    {
      "epoch": 25.901220550667045,
      "grad_norm": 8.556441307067871,
      "learning_rate": 7.41101334090264e-05,
      "loss": 0.016090008616447448,
      "step": 91250
    },
    {
      "epoch": 25.904059040590404,
      "grad_norm": 2.1930532455444336,
      "learning_rate": 7.410729491910304e-05,
      "loss": 0.02759735882282257,
      "step": 91260
    },
    {
      "epoch": 25.906897530513767,
      "grad_norm": 0.5320106744766235,
      "learning_rate": 7.410445642917968e-05,
      "loss": 0.0117863230407238,
      "step": 91270
    },
    {
      "epoch": 25.909736020437126,
      "grad_norm": 7.868771553039551,
      "learning_rate": 7.410161793925632e-05,
      "loss": 0.022496986389160156,
      "step": 91280
    },
    {
      "epoch": 25.91257451036049,
      "grad_norm": 2.2077040672302246,
      "learning_rate": 7.409877944933295e-05,
      "loss": 0.009529747068881989,
      "step": 91290
    },
    {
      "epoch": 25.91541300028385,
      "grad_norm": 0.7100173830986023,
      "learning_rate": 7.40959409594096e-05,
      "loss": 0.0024851012974977495,
      "step": 91300
    },
    {
      "epoch": 25.91825149020721,
      "grad_norm": 4.5609869956970215,
      "learning_rate": 7.409310246948625e-05,
      "loss": 0.006191249564290046,
      "step": 91310
    },
    {
      "epoch": 25.92108998013057,
      "grad_norm": 0.27032625675201416,
      "learning_rate": 7.409026397956288e-05,
      "loss": 0.008818237483501435,
      "step": 91320
    },
    {
      "epoch": 25.92392847005393,
      "grad_norm": 12.286656379699707,
      "learning_rate": 7.408742548963952e-05,
      "loss": 0.013966916501522065,
      "step": 91330
    },
    {
      "epoch": 25.926766959977293,
      "grad_norm": 0.16965515911579132,
      "learning_rate": 7.408458699971616e-05,
      "loss": 0.013837824761867522,
      "step": 91340
    },
    {
      "epoch": 25.929605449900652,
      "grad_norm": 3.1021642684936523,
      "learning_rate": 7.408174850979279e-05,
      "loss": 0.012317148596048355,
      "step": 91350
    },
    {
      "epoch": 25.932443939824015,
      "grad_norm": 0.5863439440727234,
      "learning_rate": 7.407891001986943e-05,
      "loss": 0.017285771667957306,
      "step": 91360
    },
    {
      "epoch": 25.935282429747375,
      "grad_norm": 3.3516483306884766,
      "learning_rate": 7.407607152994607e-05,
      "loss": 0.019114471971988678,
      "step": 91370
    },
    {
      "epoch": 25.938120919670734,
      "grad_norm": 12.952999114990234,
      "learning_rate": 7.407323304002271e-05,
      "loss": 0.02923743426799774,
      "step": 91380
    },
    {
      "epoch": 25.940959409594097,
      "grad_norm": 10.030466079711914,
      "learning_rate": 7.407039455009935e-05,
      "loss": 0.013706576824188233,
      "step": 91390
    },
    {
      "epoch": 25.943797899517456,
      "grad_norm": 2.9724695682525635,
      "learning_rate": 7.4067556060176e-05,
      "loss": 0.005817212909460068,
      "step": 91400
    },
    {
      "epoch": 25.94663638944082,
      "grad_norm": 3.005795478820801,
      "learning_rate": 7.406471757025264e-05,
      "loss": 0.00826607272028923,
      "step": 91410
    },
    {
      "epoch": 25.94947487936418,
      "grad_norm": 6.576315879821777,
      "learning_rate": 7.406187908032926e-05,
      "loss": 0.009742242097854615,
      "step": 91420
    },
    {
      "epoch": 25.952313369287538,
      "grad_norm": 1.3519794940948486,
      "learning_rate": 7.40590405904059e-05,
      "loss": 0.016540224850177764,
      "step": 91430
    },
    {
      "epoch": 25.9551518592109,
      "grad_norm": 0.4042956829071045,
      "learning_rate": 7.405620210048255e-05,
      "loss": 0.009927792102098465,
      "step": 91440
    },
    {
      "epoch": 25.95799034913426,
      "grad_norm": 8.763338088989258,
      "learning_rate": 7.405336361055919e-05,
      "loss": 0.00847296342253685,
      "step": 91450
    },
    {
      "epoch": 25.960828839057623,
      "grad_norm": 2.534849166870117,
      "learning_rate": 7.405052512063583e-05,
      "loss": 0.007318933308124542,
      "step": 91460
    },
    {
      "epoch": 25.963667328980982,
      "grad_norm": 1.0958232879638672,
      "learning_rate": 7.404768663071247e-05,
      "loss": 0.0032197430729866026,
      "step": 91470
    },
    {
      "epoch": 25.96650581890434,
      "grad_norm": 1.2913508415222168,
      "learning_rate": 7.40448481407891e-05,
      "loss": 0.009135988354682923,
      "step": 91480
    },
    {
      "epoch": 25.969344308827704,
      "grad_norm": 15.695176124572754,
      "learning_rate": 7.404200965086574e-05,
      "loss": 0.01782803386449814,
      "step": 91490
    },
    {
      "epoch": 25.972182798751064,
      "grad_norm": 21.704504013061523,
      "learning_rate": 7.403917116094238e-05,
      "loss": 0.028137663006782533,
      "step": 91500
    },
    {
      "epoch": 25.972182798751064,
      "eval_accuracy": 0.9702422585362752,
      "eval_loss": 0.09414780884981155,
      "eval_runtime": 31.6903,
      "eval_samples_per_second": 496.272,
      "eval_steps_per_second": 7.763,
      "step": 91500
    },
    {
      "epoch": 25.975021288674427,
      "grad_norm": 0.9988896250724792,
      "learning_rate": 7.403633267101902e-05,
      "loss": 0.01075546145439148,
      "step": 91510
    },
    {
      "epoch": 25.977859778597786,
      "grad_norm": 0.19585177302360535,
      "learning_rate": 7.403349418109566e-05,
      "loss": 0.02346855103969574,
      "step": 91520
    },
    {
      "epoch": 25.980698268521145,
      "grad_norm": 1.9649263620376587,
      "learning_rate": 7.40306556911723e-05,
      "loss": 0.009406203776597977,
      "step": 91530
    },
    {
      "epoch": 25.983536758444508,
      "grad_norm": 11.722123146057129,
      "learning_rate": 7.402781720124893e-05,
      "loss": 0.013320703804492951,
      "step": 91540
    },
    {
      "epoch": 25.986375248367867,
      "grad_norm": 0.3345646560192108,
      "learning_rate": 7.402497871132557e-05,
      "loss": 0.005006463080644607,
      "step": 91550
    },
    {
      "epoch": 25.98921373829123,
      "grad_norm": 0.31834325194358826,
      "learning_rate": 7.402214022140222e-05,
      "loss": 0.003969578072428703,
      "step": 91560
    },
    {
      "epoch": 25.99205222821459,
      "grad_norm": 14.432562828063965,
      "learning_rate": 7.401930173147886e-05,
      "loss": 0.024895551800727844,
      "step": 91570
    },
    {
      "epoch": 25.99489071813795,
      "grad_norm": 6.374305248260498,
      "learning_rate": 7.40164632415555e-05,
      "loss": 0.0059796363115310665,
      "step": 91580
    },
    {
      "epoch": 25.997729208061312,
      "grad_norm": 6.072107791900635,
      "learning_rate": 7.401362475163214e-05,
      "loss": 0.010428088903427123,
      "step": 91590
    },
    {
      "epoch": 26.00056769798467,
      "grad_norm": 2.508728504180908,
      "learning_rate": 7.40110701107011e-05,
      "loss": 0.026856860518455504,
      "step": 91600
    },
    {
      "epoch": 26.003406187908034,
      "grad_norm": 0.47993969917297363,
      "learning_rate": 7.400823162077775e-05,
      "loss": 0.011805834621191025,
      "step": 91610
    },
    {
      "epoch": 26.006244677831393,
      "grad_norm": 4.719772815704346,
      "learning_rate": 7.400539313085439e-05,
      "loss": 0.01024458035826683,
      "step": 91620
    },
    {
      "epoch": 26.009083167754753,
      "grad_norm": 0.16518598794937134,
      "learning_rate": 7.400255464093103e-05,
      "loss": 0.016178898513317108,
      "step": 91630
    },
    {
      "epoch": 26.011921657678116,
      "grad_norm": 0.16494183242321014,
      "learning_rate": 7.399971615100767e-05,
      "loss": 0.00660051554441452,
      "step": 91640
    },
    {
      "epoch": 26.014760147601475,
      "grad_norm": 1.6402392387390137,
      "learning_rate": 7.399687766108431e-05,
      "loss": 0.012804946303367615,
      "step": 91650
    },
    {
      "epoch": 26.017598637524838,
      "grad_norm": 6.249334812164307,
      "learning_rate": 7.399403917116094e-05,
      "loss": 0.004940034821629524,
      "step": 91660
    },
    {
      "epoch": 26.020437127448197,
      "grad_norm": 0.16886022686958313,
      "learning_rate": 7.399120068123758e-05,
      "loss": 0.004770774394273758,
      "step": 91670
    },
    {
      "epoch": 26.02327561737156,
      "grad_norm": 0.3433725833892822,
      "learning_rate": 7.398836219131422e-05,
      "loss": 0.0034532219171524046,
      "step": 91680
    },
    {
      "epoch": 26.02611410729492,
      "grad_norm": 0.6225762367248535,
      "learning_rate": 7.398552370139086e-05,
      "loss": 0.006844887882471085,
      "step": 91690
    },
    {
      "epoch": 26.02895259721828,
      "grad_norm": 8.108752250671387,
      "learning_rate": 7.39826852114675e-05,
      "loss": 0.006304467469453812,
      "step": 91700
    },
    {
      "epoch": 26.03179108714164,
      "grad_norm": 0.42007920145988464,
      "learning_rate": 7.397984672154415e-05,
      "loss": 0.023453757166862488,
      "step": 91710
    },
    {
      "epoch": 26.034629577065,
      "grad_norm": 1.4228780269622803,
      "learning_rate": 7.397700823162078e-05,
      "loss": 0.004242456704378128,
      "step": 91720
    },
    {
      "epoch": 26.037468066988364,
      "grad_norm": 6.449555397033691,
      "learning_rate": 7.397416974169742e-05,
      "loss": 0.007444438338279724,
      "step": 91730
    },
    {
      "epoch": 26.040306556911723,
      "grad_norm": 0.423074334859848,
      "learning_rate": 7.397133125177406e-05,
      "loss": 0.00402703583240509,
      "step": 91740
    },
    {
      "epoch": 26.043145046835082,
      "grad_norm": 3.628984212875366,
      "learning_rate": 7.39684927618507e-05,
      "loss": 0.008005397766828537,
      "step": 91750
    },
    {
      "epoch": 26.045983536758445,
      "grad_norm": 1.4003269672393799,
      "learning_rate": 7.396565427192734e-05,
      "loss": 0.0035007480531930924,
      "step": 91760
    },
    {
      "epoch": 26.048822026681805,
      "grad_norm": 16.51129722595215,
      "learning_rate": 7.396281578200398e-05,
      "loss": 0.015429350733757018,
      "step": 91770
    },
    {
      "epoch": 26.051660516605168,
      "grad_norm": 0.6514703035354614,
      "learning_rate": 7.395997729208062e-05,
      "loss": 0.0046922177076339725,
      "step": 91780
    },
    {
      "epoch": 26.054499006528527,
      "grad_norm": 0.48423317074775696,
      "learning_rate": 7.395713880215725e-05,
      "loss": 0.008311672508716584,
      "step": 91790
    },
    {
      "epoch": 26.057337496451886,
      "grad_norm": 2.4846277236938477,
      "learning_rate": 7.395430031223389e-05,
      "loss": 0.012605038285255433,
      "step": 91800
    },
    {
      "epoch": 26.06017598637525,
      "grad_norm": 10.705912590026855,
      "learning_rate": 7.395146182231053e-05,
      "loss": 0.007300843298435211,
      "step": 91810
    },
    {
      "epoch": 26.06301447629861,
      "grad_norm": 6.153450965881348,
      "learning_rate": 7.394862333238718e-05,
      "loss": 0.012901200354099274,
      "step": 91820
    },
    {
      "epoch": 26.06585296622197,
      "grad_norm": 1.408206820487976,
      "learning_rate": 7.394578484246382e-05,
      "loss": 0.0029639896005392075,
      "step": 91830
    },
    {
      "epoch": 26.06869145614533,
      "grad_norm": 4.78284215927124,
      "learning_rate": 7.394294635254046e-05,
      "loss": 0.015924523770809173,
      "step": 91840
    },
    {
      "epoch": 26.07152994606869,
      "grad_norm": 4.065502643585205,
      "learning_rate": 7.394010786261709e-05,
      "loss": 0.015618054568767548,
      "step": 91850
    },
    {
      "epoch": 26.074368435992053,
      "grad_norm": 0.44722357392311096,
      "learning_rate": 7.393726937269373e-05,
      "loss": 0.005499398708343506,
      "step": 91860
    },
    {
      "epoch": 26.077206925915412,
      "grad_norm": 1.260283350944519,
      "learning_rate": 7.393443088277037e-05,
      "loss": 0.004512446001172066,
      "step": 91870
    },
    {
      "epoch": 26.080045415838775,
      "grad_norm": 3.7341675758361816,
      "learning_rate": 7.393159239284701e-05,
      "loss": 0.008506526798009872,
      "step": 91880
    },
    {
      "epoch": 26.082883905762134,
      "grad_norm": 2.3325865268707275,
      "learning_rate": 7.392875390292365e-05,
      "loss": 0.005642106756567955,
      "step": 91890
    },
    {
      "epoch": 26.085722395685494,
      "grad_norm": 12.285056114196777,
      "learning_rate": 7.392591541300029e-05,
      "loss": 0.010328821837902069,
      "step": 91900
    },
    {
      "epoch": 26.088560885608857,
      "grad_norm": 1.7200316190719604,
      "learning_rate": 7.392307692307693e-05,
      "loss": 0.01295812577009201,
      "step": 91910
    },
    {
      "epoch": 26.091399375532216,
      "grad_norm": 1.1561529636383057,
      "learning_rate": 7.392023843315356e-05,
      "loss": 0.021177276968955994,
      "step": 91920
    },
    {
      "epoch": 26.09423786545558,
      "grad_norm": 0.7578274011611938,
      "learning_rate": 7.39173999432302e-05,
      "loss": 0.0057920094579458235,
      "step": 91930
    },
    {
      "epoch": 26.097076355378938,
      "grad_norm": 5.519168376922607,
      "learning_rate": 7.391456145330684e-05,
      "loss": 0.005496088415384293,
      "step": 91940
    },
    {
      "epoch": 26.099914845302298,
      "grad_norm": 7.024682998657227,
      "learning_rate": 7.391172296338347e-05,
      "loss": 0.012977916002273559,
      "step": 91950
    },
    {
      "epoch": 26.10275333522566,
      "grad_norm": 0.22377611696720123,
      "learning_rate": 7.390888447346013e-05,
      "loss": 0.010054262727499009,
      "step": 91960
    },
    {
      "epoch": 26.10559182514902,
      "grad_norm": 0.5555391907691956,
      "learning_rate": 7.390604598353677e-05,
      "loss": 0.006316271424293518,
      "step": 91970
    },
    {
      "epoch": 26.108430315072383,
      "grad_norm": 0.6258983612060547,
      "learning_rate": 7.39032074936134e-05,
      "loss": 0.006910897791385651,
      "step": 91980
    },
    {
      "epoch": 26.111268804995742,
      "grad_norm": 2.155669927597046,
      "learning_rate": 7.390036900369004e-05,
      "loss": 0.006577058136463166,
      "step": 91990
    },
    {
      "epoch": 26.1141072949191,
      "grad_norm": 8.577763557434082,
      "learning_rate": 7.389753051376668e-05,
      "loss": 0.007209164649248123,
      "step": 92000
    },
    {
      "epoch": 26.1141072949191,
      "eval_accuracy": 0.9699243339479875,
      "eval_loss": 0.0958784893155098,
      "eval_runtime": 31.4863,
      "eval_samples_per_second": 499.487,
      "eval_steps_per_second": 7.813,
      "step": 92000
    },
    {
      "epoch": 26.116945784842464,
      "grad_norm": 0.41550689935684204,
      "learning_rate": 7.389469202384332e-05,
      "loss": 0.017158791422843933,
      "step": 92010
    },
    {
      "epoch": 26.119784274765824,
      "grad_norm": 8.108226776123047,
      "learning_rate": 7.389185353391996e-05,
      "loss": 0.007350982725620269,
      "step": 92020
    },
    {
      "epoch": 26.122622764689186,
      "grad_norm": 0.9515445232391357,
      "learning_rate": 7.38890150439966e-05,
      "loss": 0.006752476841211319,
      "step": 92030
    },
    {
      "epoch": 26.125461254612546,
      "grad_norm": 2.1425278186798096,
      "learning_rate": 7.388617655407324e-05,
      "loss": 0.010766242444515229,
      "step": 92040
    },
    {
      "epoch": 26.12829974453591,
      "grad_norm": 0.5390787720680237,
      "learning_rate": 7.388333806414987e-05,
      "loss": 0.0026796359568834307,
      "step": 92050
    },
    {
      "epoch": 26.131138234459268,
      "grad_norm": 1.4580254554748535,
      "learning_rate": 7.388049957422651e-05,
      "loss": 0.007669779658317566,
      "step": 92060
    },
    {
      "epoch": 26.133976724382627,
      "grad_norm": 0.5257467031478882,
      "learning_rate": 7.387766108430316e-05,
      "loss": 0.002690298855304718,
      "step": 92070
    },
    {
      "epoch": 26.13681521430599,
      "grad_norm": 1.2211860418319702,
      "learning_rate": 7.387482259437978e-05,
      "loss": 0.008357351273298263,
      "step": 92080
    },
    {
      "epoch": 26.13965370422935,
      "grad_norm": 0.7747055888175964,
      "learning_rate": 7.387198410445644e-05,
      "loss": 0.010955923795700073,
      "step": 92090
    },
    {
      "epoch": 26.142492194152712,
      "grad_norm": 0.18109197914600372,
      "learning_rate": 7.386914561453308e-05,
      "loss": 0.00810195952653885,
      "step": 92100
    },
    {
      "epoch": 26.14533068407607,
      "grad_norm": 5.393846035003662,
      "learning_rate": 7.386630712460971e-05,
      "loss": 0.010102920979261399,
      "step": 92110
    },
    {
      "epoch": 26.14816917399943,
      "grad_norm": 0.796149492263794,
      "learning_rate": 7.386346863468635e-05,
      "loss": 0.007447591423988343,
      "step": 92120
    },
    {
      "epoch": 26.151007663922794,
      "grad_norm": 0.0904274582862854,
      "learning_rate": 7.386063014476299e-05,
      "loss": 0.01581808179616928,
      "step": 92130
    },
    {
      "epoch": 26.153846153846153,
      "grad_norm": 1.426159143447876,
      "learning_rate": 7.385779165483963e-05,
      "loss": 0.017684692144393922,
      "step": 92140
    },
    {
      "epoch": 26.156684643769516,
      "grad_norm": 10.58442211151123,
      "learning_rate": 7.385495316491626e-05,
      "loss": 0.010682049393653869,
      "step": 92150
    },
    {
      "epoch": 26.159523133692876,
      "grad_norm": 0.8104249238967896,
      "learning_rate": 7.385211467499291e-05,
      "loss": 0.005755631625652314,
      "step": 92160
    },
    {
      "epoch": 26.162361623616235,
      "grad_norm": 11.096292495727539,
      "learning_rate": 7.384927618506956e-05,
      "loss": 0.011455899477005005,
      "step": 92170
    },
    {
      "epoch": 26.165200113539598,
      "grad_norm": 0.4139510989189148,
      "learning_rate": 7.384643769514618e-05,
      "loss": 0.016855058073997498,
      "step": 92180
    },
    {
      "epoch": 26.168038603462957,
      "grad_norm": 14.637990951538086,
      "learning_rate": 7.384359920522282e-05,
      "loss": 0.016529493033885956,
      "step": 92190
    },
    {
      "epoch": 26.17087709338632,
      "grad_norm": 3.83351993560791,
      "learning_rate": 7.384076071529947e-05,
      "loss": 0.007639339566230774,
      "step": 92200
    },
    {
      "epoch": 26.17371558330968,
      "grad_norm": 0.5502344369888306,
      "learning_rate": 7.38379222253761e-05,
      "loss": 0.006361270695924759,
      "step": 92210
    },
    {
      "epoch": 26.17655407323304,
      "grad_norm": 0.3495018184185028,
      "learning_rate": 7.383508373545275e-05,
      "loss": 0.009624318033456803,
      "step": 92220
    },
    {
      "epoch": 26.1793925631564,
      "grad_norm": 3.0911316871643066,
      "learning_rate": 7.383224524552939e-05,
      "loss": 0.004666218906641007,
      "step": 92230
    },
    {
      "epoch": 26.18223105307976,
      "grad_norm": 0.8983182311058044,
      "learning_rate": 7.382940675560602e-05,
      "loss": 0.017222529649734496,
      "step": 92240
    },
    {
      "epoch": 26.185069543003124,
      "grad_norm": 8.949037551879883,
      "learning_rate": 7.382656826568266e-05,
      "loss": 0.01280592679977417,
      "step": 92250
    },
    {
      "epoch": 26.187908032926483,
      "grad_norm": 0.21194879710674286,
      "learning_rate": 7.38237297757593e-05,
      "loss": 0.006116367876529694,
      "step": 92260
    },
    {
      "epoch": 26.190746522849842,
      "grad_norm": 0.7844346761703491,
      "learning_rate": 7.382089128583594e-05,
      "loss": 0.020743103325366975,
      "step": 92270
    },
    {
      "epoch": 26.193585012773205,
      "grad_norm": 6.67525577545166,
      "learning_rate": 7.381805279591257e-05,
      "loss": 0.010026507079601288,
      "step": 92280
    },
    {
      "epoch": 26.196423502696565,
      "grad_norm": 0.6282540559768677,
      "learning_rate": 7.381521430598923e-05,
      "loss": 0.002783839404582977,
      "step": 92290
    },
    {
      "epoch": 26.199261992619927,
      "grad_norm": 0.37662559747695923,
      "learning_rate": 7.381237581606585e-05,
      "loss": 0.00458422601222992,
      "step": 92300
    },
    {
      "epoch": 26.202100482543287,
      "grad_norm": 0.2306385189294815,
      "learning_rate": 7.38095373261425e-05,
      "loss": 0.002305847965180874,
      "step": 92310
    },
    {
      "epoch": 26.204938972466646,
      "grad_norm": 2.8592166900634766,
      "learning_rate": 7.380669883621914e-05,
      "loss": 0.006137551739811897,
      "step": 92320
    },
    {
      "epoch": 26.20777746239001,
      "grad_norm": 2.4240221977233887,
      "learning_rate": 7.380386034629578e-05,
      "loss": 0.009489794820547104,
      "step": 92330
    },
    {
      "epoch": 26.21061595231337,
      "grad_norm": 0.41663840413093567,
      "learning_rate": 7.38010218563724e-05,
      "loss": 0.012977346777915955,
      "step": 92340
    },
    {
      "epoch": 26.21345444223673,
      "grad_norm": 0.6482568383216858,
      "learning_rate": 7.379818336644905e-05,
      "loss": 0.0019741088151931764,
      "step": 92350
    },
    {
      "epoch": 26.21629293216009,
      "grad_norm": 1.2553253173828125,
      "learning_rate": 7.37953448765257e-05,
      "loss": 0.03158543407917023,
      "step": 92360
    },
    {
      "epoch": 26.21913142208345,
      "grad_norm": 1.1307393312454224,
      "learning_rate": 7.379250638660233e-05,
      "loss": 0.01929493099451065,
      "step": 92370
    },
    {
      "epoch": 26.221969912006813,
      "grad_norm": 0.2109479010105133,
      "learning_rate": 7.378966789667897e-05,
      "loss": 0.005219285190105438,
      "step": 92380
    },
    {
      "epoch": 26.224808401930172,
      "grad_norm": 6.8526716232299805,
      "learning_rate": 7.378682940675561e-05,
      "loss": 0.007986192405223847,
      "step": 92390
    },
    {
      "epoch": 26.227646891853535,
      "grad_norm": 0.7104370594024658,
      "learning_rate": 7.378399091683225e-05,
      "loss": 0.010788311064243317,
      "step": 92400
    },
    {
      "epoch": 26.230485381776894,
      "grad_norm": 7.885170936584473,
      "learning_rate": 7.378115242690888e-05,
      "loss": 0.007539589703083038,
      "step": 92410
    },
    {
      "epoch": 26.233323871700254,
      "grad_norm": 1.2500799894332886,
      "learning_rate": 7.377831393698554e-05,
      "loss": 0.002973645552992821,
      "step": 92420
    },
    {
      "epoch": 26.236162361623617,
      "grad_norm": 10.949283599853516,
      "learning_rate": 7.377547544706216e-05,
      "loss": 0.014808160066604615,
      "step": 92430
    },
    {
      "epoch": 26.239000851546976,
      "grad_norm": 0.8921816349029541,
      "learning_rate": 7.37726369571388e-05,
      "loss": 0.010337941348552704,
      "step": 92440
    },
    {
      "epoch": 26.24183934147034,
      "grad_norm": 1.4611961841583252,
      "learning_rate": 7.376979846721545e-05,
      "loss": 0.015223929286003112,
      "step": 92450
    },
    {
      "epoch": 26.244677831393698,
      "grad_norm": 1.7603025436401367,
      "learning_rate": 7.376695997729209e-05,
      "loss": 0.009551189839839935,
      "step": 92460
    },
    {
      "epoch": 26.24751632131706,
      "grad_norm": 5.045895099639893,
      "learning_rate": 7.376412148736872e-05,
      "loss": 0.020971786975860596,
      "step": 92470
    },
    {
      "epoch": 26.25035481124042,
      "grad_norm": 0.08690071851015091,
      "learning_rate": 7.376128299744536e-05,
      "loss": 0.00686119869351387,
      "step": 92480
    },
    {
      "epoch": 26.25319330116378,
      "grad_norm": 0.9654287695884705,
      "learning_rate": 7.375844450752201e-05,
      "loss": 0.006131289154291153,
      "step": 92490
    },
    {
      "epoch": 26.256031791087143,
      "grad_norm": 0.0503830760717392,
      "learning_rate": 7.375560601759864e-05,
      "loss": 0.012299502640962601,
      "step": 92500
    },
    {
      "epoch": 26.256031791087143,
      "eval_accuracy": 0.9703058434539328,
      "eval_loss": 0.09468699246644974,
      "eval_runtime": 31.8091,
      "eval_samples_per_second": 494.419,
      "eval_steps_per_second": 7.734,
      "step": 92500
    },
    {
      "epoch": 26.258870281010502,
      "grad_norm": 3.025846004486084,
      "learning_rate": 7.375276752767528e-05,
      "loss": 0.004975470900535584,
      "step": 92510
    },
    {
      "epoch": 26.261708770933865,
      "grad_norm": 0.2321932464838028,
      "learning_rate": 7.374992903775192e-05,
      "loss": 0.006643175333738327,
      "step": 92520
    },
    {
      "epoch": 26.264547260857224,
      "grad_norm": 2.2461955547332764,
      "learning_rate": 7.374709054782855e-05,
      "loss": 0.011097977310419083,
      "step": 92530
    },
    {
      "epoch": 26.267385750780583,
      "grad_norm": 3.1445255279541016,
      "learning_rate": 7.374425205790519e-05,
      "loss": 0.005190606415271759,
      "step": 92540
    },
    {
      "epoch": 26.270224240703946,
      "grad_norm": 0.3949303925037384,
      "learning_rate": 7.374141356798183e-05,
      "loss": 0.012363631278276443,
      "step": 92550
    },
    {
      "epoch": 26.273062730627306,
      "grad_norm": 1.0306131839752197,
      "learning_rate": 7.373857507805847e-05,
      "loss": 0.007492886483669281,
      "step": 92560
    },
    {
      "epoch": 26.27590122055067,
      "grad_norm": 6.044046401977539,
      "learning_rate": 7.373573658813512e-05,
      "loss": 0.00794539675116539,
      "step": 92570
    },
    {
      "epoch": 26.278739710474028,
      "grad_norm": 0.3525402843952179,
      "learning_rate": 7.373289809821176e-05,
      "loss": 0.009858734905719757,
      "step": 92580
    },
    {
      "epoch": 26.281578200397387,
      "grad_norm": 3.033472776412964,
      "learning_rate": 7.37300596082884e-05,
      "loss": 0.015811192989349365,
      "step": 92590
    },
    {
      "epoch": 26.28441669032075,
      "grad_norm": 1.378052830696106,
      "learning_rate": 7.372722111836503e-05,
      "loss": 0.006582745164632797,
      "step": 92600
    },
    {
      "epoch": 26.28725518024411,
      "grad_norm": 0.9024652242660522,
      "learning_rate": 7.372438262844167e-05,
      "loss": 0.028896653652191163,
      "step": 92610
    },
    {
      "epoch": 26.290093670167472,
      "grad_norm": 0.32481107115745544,
      "learning_rate": 7.372154413851832e-05,
      "loss": 0.026616260409355164,
      "step": 92620
    },
    {
      "epoch": 26.29293216009083,
      "grad_norm": 5.929981708526611,
      "learning_rate": 7.371870564859495e-05,
      "loss": 0.019439248740673064,
      "step": 92630
    },
    {
      "epoch": 26.29577065001419,
      "grad_norm": 2.0967047214508057,
      "learning_rate": 7.371586715867159e-05,
      "loss": 0.005662263557314873,
      "step": 92640
    },
    {
      "epoch": 26.298609139937554,
      "grad_norm": 11.044548988342285,
      "learning_rate": 7.371302866874823e-05,
      "loss": 0.03227064609527588,
      "step": 92650
    },
    {
      "epoch": 26.301447629860913,
      "grad_norm": 0.08486620336771011,
      "learning_rate": 7.371019017882486e-05,
      "loss": 0.005920019745826721,
      "step": 92660
    },
    {
      "epoch": 26.304286119784276,
      "grad_norm": 6.801551342010498,
      "learning_rate": 7.37073516889015e-05,
      "loss": 0.012377683073282242,
      "step": 92670
    },
    {
      "epoch": 26.307124609707635,
      "grad_norm": 2.0847713947296143,
      "learning_rate": 7.370451319897814e-05,
      "loss": 0.021713931858539582,
      "step": 92680
    },
    {
      "epoch": 26.309963099630995,
      "grad_norm": 1.3447253704071045,
      "learning_rate": 7.370167470905479e-05,
      "loss": 0.006275822967290878,
      "step": 92690
    },
    {
      "epoch": 26.312801589554358,
      "grad_norm": 5.674437046051025,
      "learning_rate": 7.369883621913143e-05,
      "loss": 0.01807156056165695,
      "step": 92700
    },
    {
      "epoch": 26.315640079477717,
      "grad_norm": 1.2775100469589233,
      "learning_rate": 7.369599772920807e-05,
      "loss": 0.030078983306884764,
      "step": 92710
    },
    {
      "epoch": 26.31847856940108,
      "grad_norm": 7.216073036193848,
      "learning_rate": 7.369315923928471e-05,
      "loss": 0.02365185022354126,
      "step": 92720
    },
    {
      "epoch": 26.32131705932444,
      "grad_norm": 14.173765182495117,
      "learning_rate": 7.369032074936134e-05,
      "loss": 0.025605848431587218,
      "step": 92730
    },
    {
      "epoch": 26.3241555492478,
      "grad_norm": 2.7671217918395996,
      "learning_rate": 7.368748225943798e-05,
      "loss": 0.013354629278182983,
      "step": 92740
    },
    {
      "epoch": 26.32699403917116,
      "grad_norm": 4.251099109649658,
      "learning_rate": 7.368464376951462e-05,
      "loss": 0.02026963084936142,
      "step": 92750
    },
    {
      "epoch": 26.32983252909452,
      "grad_norm": 3.0159120559692383,
      "learning_rate": 7.368180527959126e-05,
      "loss": 0.010386209189891814,
      "step": 92760
    },
    {
      "epoch": 26.332671019017884,
      "grad_norm": 3.779182195663452,
      "learning_rate": 7.36789667896679e-05,
      "loss": 0.004496930167078972,
      "step": 92770
    },
    {
      "epoch": 26.335509508941243,
      "grad_norm": 12.402325630187988,
      "learning_rate": 7.367612829974454e-05,
      "loss": 0.025400730967521667,
      "step": 92780
    },
    {
      "epoch": 26.338347998864602,
      "grad_norm": 8.638562202453613,
      "learning_rate": 7.367328980982117e-05,
      "loss": 0.013978950679302216,
      "step": 92790
    },
    {
      "epoch": 26.341186488787965,
      "grad_norm": 9.683276176452637,
      "learning_rate": 7.367045131989781e-05,
      "loss": 0.007374817878007889,
      "step": 92800
    },
    {
      "epoch": 26.344024978711325,
      "grad_norm": 0.45772796869277954,
      "learning_rate": 7.366761282997445e-05,
      "loss": 0.006616039574146271,
      "step": 92810
    },
    {
      "epoch": 26.346863468634687,
      "grad_norm": 3.851959228515625,
      "learning_rate": 7.36647743400511e-05,
      "loss": 0.012069138884544372,
      "step": 92820
    },
    {
      "epoch": 26.349701958558047,
      "grad_norm": 14.085392951965332,
      "learning_rate": 7.366193585012774e-05,
      "loss": 0.01765160858631134,
      "step": 92830
    },
    {
      "epoch": 26.352540448481406,
      "grad_norm": 1.4482502937316895,
      "learning_rate": 7.365909736020438e-05,
      "loss": 0.004070385918021202,
      "step": 92840
    },
    {
      "epoch": 26.35537893840477,
      "grad_norm": 2.6822054386138916,
      "learning_rate": 7.365625887028102e-05,
      "loss": 0.013836394250392913,
      "step": 92850
    },
    {
      "epoch": 26.35821742832813,
      "grad_norm": 0.6110975742340088,
      "learning_rate": 7.365342038035765e-05,
      "loss": 0.005951684340834618,
      "step": 92860
    },
    {
      "epoch": 26.36105591825149,
      "grad_norm": 0.13471630215644836,
      "learning_rate": 7.365058189043429e-05,
      "loss": 0.0041008956730365755,
      "step": 92870
    },
    {
      "epoch": 26.36389440817485,
      "grad_norm": 1.1595858335494995,
      "learning_rate": 7.364774340051093e-05,
      "loss": 0.005022244900465012,
      "step": 92880
    },
    {
      "epoch": 26.366732898098213,
      "grad_norm": 5.212616443634033,
      "learning_rate": 7.364490491058757e-05,
      "loss": 0.012870852649211884,
      "step": 92890
    },
    {
      "epoch": 26.369571388021573,
      "grad_norm": 0.7942333221435547,
      "learning_rate": 7.364206642066421e-05,
      "loss": 0.004189717397093773,
      "step": 92900
    },
    {
      "epoch": 26.372409877944932,
      "grad_norm": 0.6024128794670105,
      "learning_rate": 7.363922793074085e-05,
      "loss": 0.007008887082338333,
      "step": 92910
    },
    {
      "epoch": 26.375248367868295,
      "grad_norm": 0.3002496063709259,
      "learning_rate": 7.363638944081748e-05,
      "loss": 0.009779290109872819,
      "step": 92920
    },
    {
      "epoch": 26.378086857791654,
      "grad_norm": 0.3000771403312683,
      "learning_rate": 7.363355095089412e-05,
      "loss": 0.00793919637799263,
      "step": 92930
    },
    {
      "epoch": 26.380925347715017,
      "grad_norm": 1.5796071290969849,
      "learning_rate": 7.363071246097077e-05,
      "loss": 0.014450131356716156,
      "step": 92940
    },
    {
      "epoch": 26.383763837638377,
      "grad_norm": 1.6405445337295532,
      "learning_rate": 7.36278739710474e-05,
      "loss": 0.008795601129531861,
      "step": 92950
    },
    {
      "epoch": 26.386602327561736,
      "grad_norm": 0.2510904371738434,
      "learning_rate": 7.362503548112405e-05,
      "loss": 0.005699639394879341,
      "step": 92960
    },
    {
      "epoch": 26.3894408174851,
      "grad_norm": 0.09752581268548965,
      "learning_rate": 7.362219699120069e-05,
      "loss": 0.020186567306518556,
      "step": 92970
    },
    {
      "epoch": 26.392279307408458,
      "grad_norm": 0.5659540295600891,
      "learning_rate": 7.361935850127733e-05,
      "loss": 0.010608004033565521,
      "step": 92980
    },
    {
      "epoch": 26.39511779733182,
      "grad_norm": 1.1968358755111694,
      "learning_rate": 7.361652001135396e-05,
      "loss": 0.006119243800640106,
      "step": 92990
    },
    {
      "epoch": 26.39795628725518,
      "grad_norm": 3.1396477222442627,
      "learning_rate": 7.36136815214306e-05,
      "loss": 0.004807562381029129,
      "step": 93000
    },
    {
      "epoch": 26.39795628725518,
      "eval_accuracy": 0.9650282952883577,
      "eval_loss": 0.1161697655916214,
      "eval_runtime": 32.189,
      "eval_samples_per_second": 488.584,
      "eval_steps_per_second": 7.642,
      "step": 93000
    },
    {
      "epoch": 26.40079477717854,
      "grad_norm": 1.2163498401641846,
      "learning_rate": 7.361084303150724e-05,
      "loss": 0.008039128780364991,
      "step": 93010
    },
    {
      "epoch": 26.403633267101903,
      "grad_norm": 2.8897817134857178,
      "learning_rate": 7.360800454158388e-05,
      "loss": 0.0057464692741632465,
      "step": 93020
    },
    {
      "epoch": 26.406471757025262,
      "grad_norm": 1.4638886451721191,
      "learning_rate": 7.360516605166052e-05,
      "loss": 0.003186637908220291,
      "step": 93030
    },
    {
      "epoch": 26.409310246948625,
      "grad_norm": 0.8826212286949158,
      "learning_rate": 7.360232756173717e-05,
      "loss": 0.010672713071107865,
      "step": 93040
    },
    {
      "epoch": 26.412148736871984,
      "grad_norm": 3.4137396812438965,
      "learning_rate": 7.35994890718138e-05,
      "loss": 0.00811089649796486,
      "step": 93050
    },
    {
      "epoch": 26.414987226795343,
      "grad_norm": 1.3350623846054077,
      "learning_rate": 7.359665058189043e-05,
      "loss": 0.017855803668498992,
      "step": 93060
    },
    {
      "epoch": 26.417825716718706,
      "grad_norm": 4.353815078735352,
      "learning_rate": 7.359381209196708e-05,
      "loss": 0.006519340723752975,
      "step": 93070
    },
    {
      "epoch": 26.420664206642066,
      "grad_norm": 2.5659937858581543,
      "learning_rate": 7.359097360204372e-05,
      "loss": 0.005372990667819977,
      "step": 93080
    },
    {
      "epoch": 26.42350269656543,
      "grad_norm": 3.4872448444366455,
      "learning_rate": 7.358813511212036e-05,
      "loss": 0.010218745470046997,
      "step": 93090
    },
    {
      "epoch": 26.426341186488788,
      "grad_norm": 1.6558328866958618,
      "learning_rate": 7.3585296622197e-05,
      "loss": 0.011450881510972977,
      "step": 93100
    },
    {
      "epoch": 26.429179676412147,
      "grad_norm": 0.8963851928710938,
      "learning_rate": 7.358245813227364e-05,
      "loss": 0.009582974761724473,
      "step": 93110
    },
    {
      "epoch": 26.43201816633551,
      "grad_norm": 2.569427013397217,
      "learning_rate": 7.357961964235027e-05,
      "loss": 0.005919871479272842,
      "step": 93120
    },
    {
      "epoch": 26.43485665625887,
      "grad_norm": 2.3696396350860596,
      "learning_rate": 7.357678115242691e-05,
      "loss": 0.007329928129911423,
      "step": 93130
    },
    {
      "epoch": 26.437695146182232,
      "grad_norm": 0.1995571106672287,
      "learning_rate": 7.357394266250355e-05,
      "loss": 0.01079520806670189,
      "step": 93140
    },
    {
      "epoch": 26.44053363610559,
      "grad_norm": 0.35779473185539246,
      "learning_rate": 7.35711041725802e-05,
      "loss": 0.01520516574382782,
      "step": 93150
    },
    {
      "epoch": 26.44337212602895,
      "grad_norm": 0.9381704330444336,
      "learning_rate": 7.356826568265683e-05,
      "loss": 0.01937687546014786,
      "step": 93160
    },
    {
      "epoch": 26.446210615952314,
      "grad_norm": 0.7077054381370544,
      "learning_rate": 7.356542719273348e-05,
      "loss": 0.010111543536186218,
      "step": 93170
    },
    {
      "epoch": 26.449049105875673,
      "grad_norm": 4.909936904907227,
      "learning_rate": 7.35625887028101e-05,
      "loss": 0.008020897209644318,
      "step": 93180
    },
    {
      "epoch": 26.451887595799036,
      "grad_norm": 1.5370436906814575,
      "learning_rate": 7.355975021288675e-05,
      "loss": 0.006582596898078918,
      "step": 93190
    },
    {
      "epoch": 26.454726085722395,
      "grad_norm": 2.325643301010132,
      "learning_rate": 7.355691172296339e-05,
      "loss": 0.007035617530345917,
      "step": 93200
    },
    {
      "epoch": 26.457564575645755,
      "grad_norm": 1.2454723119735718,
      "learning_rate": 7.355407323304003e-05,
      "loss": 0.008420885354280472,
      "step": 93210
    },
    {
      "epoch": 26.460403065569118,
      "grad_norm": 1.3416330814361572,
      "learning_rate": 7.355123474311667e-05,
      "loss": 0.005542892217636109,
      "step": 93220
    },
    {
      "epoch": 26.463241555492477,
      "grad_norm": 4.251852512359619,
      "learning_rate": 7.354839625319331e-05,
      "loss": 0.01014450117945671,
      "step": 93230
    },
    {
      "epoch": 26.46608004541584,
      "grad_norm": 0.07669439166784286,
      "learning_rate": 7.354555776326994e-05,
      "loss": 0.010270357877016068,
      "step": 93240
    },
    {
      "epoch": 26.4689185353392,
      "grad_norm": 10.482711791992188,
      "learning_rate": 7.354271927334658e-05,
      "loss": 0.012404431402683259,
      "step": 93250
    },
    {
      "epoch": 26.471757025262562,
      "grad_norm": 1.408994197845459,
      "learning_rate": 7.353988078342322e-05,
      "loss": 0.012409591674804687,
      "step": 93260
    },
    {
      "epoch": 26.47459551518592,
      "grad_norm": 5.073132514953613,
      "learning_rate": 7.353704229349986e-05,
      "loss": 0.0165861576795578,
      "step": 93270
    },
    {
      "epoch": 26.47743400510928,
      "grad_norm": 0.9293218851089478,
      "learning_rate": 7.353420380357649e-05,
      "loss": 0.022351078689098358,
      "step": 93280
    },
    {
      "epoch": 26.480272495032644,
      "grad_norm": 0.6898579597473145,
      "learning_rate": 7.353136531365315e-05,
      "loss": 0.023683179914951325,
      "step": 93290
    },
    {
      "epoch": 26.483110984956003,
      "grad_norm": 3.2098238468170166,
      "learning_rate": 7.352852682372979e-05,
      "loss": 0.006058626249432564,
      "step": 93300
    },
    {
      "epoch": 26.485949474879366,
      "grad_norm": 5.955200672149658,
      "learning_rate": 7.352568833380641e-05,
      "loss": 0.005017219111323357,
      "step": 93310
    },
    {
      "epoch": 26.488787964802725,
      "grad_norm": 2.9740405082702637,
      "learning_rate": 7.352284984388306e-05,
      "loss": 0.007307432591915131,
      "step": 93320
    },
    {
      "epoch": 26.491626454726084,
      "grad_norm": 4.516443729400635,
      "learning_rate": 7.35200113539597e-05,
      "loss": 0.02282538264989853,
      "step": 93330
    },
    {
      "epoch": 26.494464944649447,
      "grad_norm": 0.15894350409507751,
      "learning_rate": 7.351717286403633e-05,
      "loss": 0.002300383523106575,
      "step": 93340
    },
    {
      "epoch": 26.497303434572807,
      "grad_norm": 2.02738618850708,
      "learning_rate": 7.351433437411298e-05,
      "loss": 0.01436062753200531,
      "step": 93350
    },
    {
      "epoch": 26.50014192449617,
      "grad_norm": 1.5969582796096802,
      "learning_rate": 7.351149588418962e-05,
      "loss": 0.017239491641521453,
      "step": 93360
    },
    {
      "epoch": 26.50298041441953,
      "grad_norm": 5.777594089508057,
      "learning_rate": 7.350865739426625e-05,
      "loss": 0.01588103175163269,
      "step": 93370
    },
    {
      "epoch": 26.50581890434289,
      "grad_norm": 6.048971652984619,
      "learning_rate": 7.350581890434289e-05,
      "loss": 0.015367710590362548,
      "step": 93380
    },
    {
      "epoch": 26.50865739426625,
      "grad_norm": 6.3774800300598145,
      "learning_rate": 7.350298041441953e-05,
      "loss": 0.01360400766134262,
      "step": 93390
    },
    {
      "epoch": 26.51149588418961,
      "grad_norm": 7.105537414550781,
      "learning_rate": 7.350014192449617e-05,
      "loss": 0.01069067120552063,
      "step": 93400
    },
    {
      "epoch": 26.514334374112973,
      "grad_norm": 4.433438301086426,
      "learning_rate": 7.34973034345728e-05,
      "loss": 0.01765124499797821,
      "step": 93410
    },
    {
      "epoch": 26.517172864036333,
      "grad_norm": 3.893505573272705,
      "learning_rate": 7.349446494464946e-05,
      "loss": 0.009969960153102874,
      "step": 93420
    },
    {
      "epoch": 26.520011353959692,
      "grad_norm": 0.6545042395591736,
      "learning_rate": 7.34916264547261e-05,
      "loss": 0.014891240000724792,
      "step": 93430
    },
    {
      "epoch": 26.522849843883055,
      "grad_norm": 6.841866493225098,
      "learning_rate": 7.348878796480273e-05,
      "loss": 0.00988268256187439,
      "step": 93440
    },
    {
      "epoch": 26.525688333806414,
      "grad_norm": 0.42013347148895264,
      "learning_rate": 7.348594947487937e-05,
      "loss": 0.022642302513122558,
      "step": 93450
    },
    {
      "epoch": 26.528526823729777,
      "grad_norm": 6.23939323425293,
      "learning_rate": 7.348311098495601e-05,
      "loss": 0.019336767494678497,
      "step": 93460
    },
    {
      "epoch": 26.531365313653136,
      "grad_norm": 0.7334332466125488,
      "learning_rate": 7.348027249503264e-05,
      "loss": 0.011577697098255157,
      "step": 93470
    },
    {
      "epoch": 26.534203803576496,
      "grad_norm": 0.7589626312255859,
      "learning_rate": 7.347743400510928e-05,
      "loss": 0.013959628343582154,
      "step": 93480
    },
    {
      "epoch": 26.53704229349986,
      "grad_norm": 0.8063339591026306,
      "learning_rate": 7.347459551518593e-05,
      "loss": 0.020280875265598297,
      "step": 93490
    },
    {
      "epoch": 26.539880783423218,
      "grad_norm": 4.919368267059326,
      "learning_rate": 7.347175702526256e-05,
      "loss": 0.011336325854063033,
      "step": 93500
    },
    {
      "epoch": 26.539880783423218,
      "eval_accuracy": 0.9685890506771794,
      "eval_loss": 0.09968402981758118,
      "eval_runtime": 32.4782,
      "eval_samples_per_second": 484.232,
      "eval_steps_per_second": 7.574,
      "step": 93500
    },
    {
      "epoch": 26.54271927334658,
      "grad_norm": 2.1325113773345947,
      "learning_rate": 7.34689185353392e-05,
      "loss": 0.018076063692569734,
      "step": 93510
    },
    {
      "epoch": 26.54555776326994,
      "grad_norm": 0.5887179970741272,
      "learning_rate": 7.346608004541584e-05,
      "loss": 0.011239643394947051,
      "step": 93520
    },
    {
      "epoch": 26.5483962531933,
      "grad_norm": 7.1910481452941895,
      "learning_rate": 7.346324155549248e-05,
      "loss": 0.006685967743396759,
      "step": 93530
    },
    {
      "epoch": 26.551234743116662,
      "grad_norm": 4.799663543701172,
      "learning_rate": 7.346040306556911e-05,
      "loss": 0.007635510712862015,
      "step": 93540
    },
    {
      "epoch": 26.55407323304002,
      "grad_norm": 9.017477035522461,
      "learning_rate": 7.345756457564577e-05,
      "loss": 0.011153483390808105,
      "step": 93550
    },
    {
      "epoch": 26.556911722963385,
      "grad_norm": 2.9574196338653564,
      "learning_rate": 7.345472608572241e-05,
      "loss": 0.005689339712262154,
      "step": 93560
    },
    {
      "epoch": 26.559750212886744,
      "grad_norm": 4.893571853637695,
      "learning_rate": 7.345188759579904e-05,
      "loss": 0.020778751373291014,
      "step": 93570
    },
    {
      "epoch": 26.562588702810103,
      "grad_norm": 0.6927777528762817,
      "learning_rate": 7.344904910587568e-05,
      "loss": 0.02228154391050339,
      "step": 93580
    },
    {
      "epoch": 26.565427192733466,
      "grad_norm": 1.3354644775390625,
      "learning_rate": 7.344621061595232e-05,
      "loss": 0.010807336121797562,
      "step": 93590
    },
    {
      "epoch": 26.568265682656826,
      "grad_norm": 0.41370514035224915,
      "learning_rate": 7.344337212602895e-05,
      "loss": 0.01049211323261261,
      "step": 93600
    },
    {
      "epoch": 26.57110417258019,
      "grad_norm": 8.193758964538574,
      "learning_rate": 7.344053363610559e-05,
      "loss": 0.023290959000587464,
      "step": 93610
    },
    {
      "epoch": 26.573942662503548,
      "grad_norm": 16.485464096069336,
      "learning_rate": 7.343797899517457e-05,
      "loss": 0.01803134083747864,
      "step": 93620
    },
    {
      "epoch": 26.57678115242691,
      "grad_norm": 1.6572365760803223,
      "learning_rate": 7.343514050525121e-05,
      "loss": 0.009075594693422317,
      "step": 93630
    },
    {
      "epoch": 26.57961964235027,
      "grad_norm": 0.6884151697158813,
      "learning_rate": 7.343230201532785e-05,
      "loss": 0.006136306002736092,
      "step": 93640
    },
    {
      "epoch": 26.58245813227363,
      "grad_norm": 4.5465826988220215,
      "learning_rate": 7.342946352540448e-05,
      "loss": 0.003441450744867325,
      "step": 93650
    },
    {
      "epoch": 26.585296622196992,
      "grad_norm": 1.3059022426605225,
      "learning_rate": 7.342662503548112e-05,
      "loss": 0.006587804853916168,
      "step": 93660
    },
    {
      "epoch": 26.58813511212035,
      "grad_norm": 1.2390974760055542,
      "learning_rate": 7.342378654555778e-05,
      "loss": 0.007151448726654052,
      "step": 93670
    },
    {
      "epoch": 26.590973602043714,
      "grad_norm": 6.941291809082031,
      "learning_rate": 7.34209480556344e-05,
      "loss": 0.009516268968582153,
      "step": 93680
    },
    {
      "epoch": 26.593812091967074,
      "grad_norm": 0.8465082049369812,
      "learning_rate": 7.341810956571104e-05,
      "loss": 0.01101771742105484,
      "step": 93690
    },
    {
      "epoch": 26.596650581890433,
      "grad_norm": 0.4537831246852875,
      "learning_rate": 7.341527107578769e-05,
      "loss": 0.008472254127264022,
      "step": 93700
    },
    {
      "epoch": 26.599489071813796,
      "grad_norm": 0.1561071276664734,
      "learning_rate": 7.341243258586433e-05,
      "loss": 0.008398085832595825,
      "step": 93710
    },
    {
      "epoch": 26.602327561737155,
      "grad_norm": 0.3406132161617279,
      "learning_rate": 7.340959409594095e-05,
      "loss": 0.005939728766679764,
      "step": 93720
    },
    {
      "epoch": 26.605166051660518,
      "grad_norm": 0.6351897716522217,
      "learning_rate": 7.340675560601761e-05,
      "loss": 0.019451333582401274,
      "step": 93730
    },
    {
      "epoch": 26.608004541583878,
      "grad_norm": 1.1513665914535522,
      "learning_rate": 7.340391711609425e-05,
      "loss": 0.021116769313812254,
      "step": 93740
    },
    {
      "epoch": 26.610843031507237,
      "grad_norm": 4.844072341918945,
      "learning_rate": 7.340107862617088e-05,
      "loss": 0.007307896018028259,
      "step": 93750
    },
    {
      "epoch": 26.6136815214306,
      "grad_norm": 0.8704673051834106,
      "learning_rate": 7.339824013624752e-05,
      "loss": 0.0052663017064332966,
      "step": 93760
    },
    {
      "epoch": 26.61652001135396,
      "grad_norm": 0.5843260884284973,
      "learning_rate": 7.339540164632416e-05,
      "loss": 0.00935194343328476,
      "step": 93770
    },
    {
      "epoch": 26.619358501277322,
      "grad_norm": 0.16241100430488586,
      "learning_rate": 7.339256315640079e-05,
      "loss": 0.00642252042889595,
      "step": 93780
    },
    {
      "epoch": 26.62219699120068,
      "grad_norm": 2.4427473545074463,
      "learning_rate": 7.338972466647743e-05,
      "loss": 0.006610368192195892,
      "step": 93790
    },
    {
      "epoch": 26.62503548112404,
      "grad_norm": 0.04928050562739372,
      "learning_rate": 7.338688617655409e-05,
      "loss": 0.02486005425453186,
      "step": 93800
    },
    {
      "epoch": 26.627873971047404,
      "grad_norm": 3.9031119346618652,
      "learning_rate": 7.338404768663071e-05,
      "loss": 0.008999011665582656,
      "step": 93810
    },
    {
      "epoch": 26.630712460970763,
      "grad_norm": 0.9540936946868896,
      "learning_rate": 7.338120919670736e-05,
      "loss": 0.005932574719190597,
      "step": 93820
    },
    {
      "epoch": 26.633550950894126,
      "grad_norm": 5.391296863555908,
      "learning_rate": 7.3378370706784e-05,
      "loss": 0.01105673685669899,
      "step": 93830
    },
    {
      "epoch": 26.636389440817485,
      "grad_norm": 1.161635160446167,
      "learning_rate": 7.337553221686064e-05,
      "loss": 0.01124252676963806,
      "step": 93840
    },
    {
      "epoch": 26.639227930740844,
      "grad_norm": 0.6210116744041443,
      "learning_rate": 7.337269372693727e-05,
      "loss": 0.008226653933525086,
      "step": 93850
    },
    {
      "epoch": 26.642066420664207,
      "grad_norm": 4.83935546875,
      "learning_rate": 7.336985523701391e-05,
      "loss": 0.02346726953983307,
      "step": 93860
    },
    {
      "epoch": 26.644904910587567,
      "grad_norm": 0.6457901000976562,
      "learning_rate": 7.336701674709056e-05,
      "loss": 0.006627421826124191,
      "step": 93870
    },
    {
      "epoch": 26.64774340051093,
      "grad_norm": 0.08087478578090668,
      "learning_rate": 7.336417825716719e-05,
      "loss": 0.0095731683075428,
      "step": 93880
    },
    {
      "epoch": 26.65058189043429,
      "grad_norm": 0.9883803129196167,
      "learning_rate": 7.336133976724383e-05,
      "loss": 0.00721057578921318,
      "step": 93890
    },
    {
      "epoch": 26.653420380357648,
      "grad_norm": 7.240014553070068,
      "learning_rate": 7.335850127732047e-05,
      "loss": 0.010792773962020875,
      "step": 93900
    },
    {
      "epoch": 26.65625887028101,
      "grad_norm": 0.3851408064365387,
      "learning_rate": 7.33556627873971e-05,
      "loss": 0.005291806533932686,
      "step": 93910
    },
    {
      "epoch": 26.65909736020437,
      "grad_norm": 0.4661642909049988,
      "learning_rate": 7.335282429747374e-05,
      "loss": 0.005909577012062073,
      "step": 93920
    },
    {
      "epoch": 26.661935850127733,
      "grad_norm": 0.12833087146282196,
      "learning_rate": 7.33499858075504e-05,
      "loss": 0.008726454526185989,
      "step": 93930
    },
    {
      "epoch": 26.664774340051093,
      "grad_norm": 0.6580816507339478,
      "learning_rate": 7.334714731762702e-05,
      "loss": 0.011630985140800475,
      "step": 93940
    },
    {
      "epoch": 26.667612829974452,
      "grad_norm": 3.5707263946533203,
      "learning_rate": 7.334430882770367e-05,
      "loss": 0.010473309457302094,
      "step": 93950
    },
    {
      "epoch": 26.670451319897815,
      "grad_norm": 0.9508419036865234,
      "learning_rate": 7.334147033778031e-05,
      "loss": 0.005357644706964493,
      "step": 93960
    },
    {
      "epoch": 26.673289809821174,
      "grad_norm": 12.329163551330566,
      "learning_rate": 7.333863184785695e-05,
      "loss": 0.03086712956428528,
      "step": 93970
    },
    {
      "epoch": 26.676128299744537,
      "grad_norm": 2.383697271347046,
      "learning_rate": 7.333579335793358e-05,
      "loss": 0.020872431993484496,
      "step": 93980
    },
    {
      "epoch": 26.678966789667896,
      "grad_norm": 1.512812852859497,
      "learning_rate": 7.333295486801022e-05,
      "loss": 0.016036182641983032,
      "step": 93990
    },
    {
      "epoch": 26.68180527959126,
      "grad_norm": 0.9780462384223938,
      "learning_rate": 7.333011637808687e-05,
      "loss": 0.010519023239612579,
      "step": 94000
    },
    {
      "epoch": 26.68180527959126,
      "eval_accuracy": 0.9692248998537547,
      "eval_loss": 0.1048494279384613,
      "eval_runtime": 31.7743,
      "eval_samples_per_second": 494.96,
      "eval_steps_per_second": 7.742,
      "step": 94000
    },
    {
      "epoch": 26.68464376951462,
      "grad_norm": 2.6364002227783203,
      "learning_rate": 7.33272778881635e-05,
      "loss": 0.013288049399852753,
      "step": 94010
    },
    {
      "epoch": 26.687482259437978,
      "grad_norm": 7.415493965148926,
      "learning_rate": 7.332443939824014e-05,
      "loss": 0.02603104114532471,
      "step": 94020
    },
    {
      "epoch": 26.69032074936134,
      "grad_norm": 0.13083277642726898,
      "learning_rate": 7.332160090831678e-05,
      "loss": 0.013575693964958191,
      "step": 94030
    },
    {
      "epoch": 26.6931592392847,
      "grad_norm": 2.9462966918945312,
      "learning_rate": 7.331876241839341e-05,
      "loss": 0.007019470632076264,
      "step": 94040
    },
    {
      "epoch": 26.695997729208063,
      "grad_norm": 1.0577255487442017,
      "learning_rate": 7.331592392847005e-05,
      "loss": 0.006899809837341309,
      "step": 94050
    },
    {
      "epoch": 26.698836219131422,
      "grad_norm": 1.240348219871521,
      "learning_rate": 7.33130854385467e-05,
      "loss": 0.0093466617166996,
      "step": 94060
    },
    {
      "epoch": 26.70167470905478,
      "grad_norm": 0.13549132645130157,
      "learning_rate": 7.331024694862334e-05,
      "loss": 0.01021496281027794,
      "step": 94070
    },
    {
      "epoch": 26.704513198978145,
      "grad_norm": 0.4940224885940552,
      "learning_rate": 7.330740845869998e-05,
      "loss": 0.0046342164278030396,
      "step": 94080
    },
    {
      "epoch": 26.707351688901504,
      "grad_norm": 0.7037240266799927,
      "learning_rate": 7.330456996877662e-05,
      "loss": 0.007728366553783417,
      "step": 94090
    },
    {
      "epoch": 26.710190178824867,
      "grad_norm": 1.1176986694335938,
      "learning_rate": 7.330173147885326e-05,
      "loss": 0.0026960244402289392,
      "step": 94100
    },
    {
      "epoch": 26.713028668748226,
      "grad_norm": 0.25860780477523804,
      "learning_rate": 7.329889298892989e-05,
      "loss": 0.0065338902175426485,
      "step": 94110
    },
    {
      "epoch": 26.715867158671585,
      "grad_norm": 2.9327640533447266,
      "learning_rate": 7.329605449900653e-05,
      "loss": 0.0034998197108507155,
      "step": 94120
    },
    {
      "epoch": 26.71870564859495,
      "grad_norm": 0.9977213144302368,
      "learning_rate": 7.329321600908317e-05,
      "loss": 0.01593065559864044,
      "step": 94130
    },
    {
      "epoch": 26.721544138518308,
      "grad_norm": 3.589890718460083,
      "learning_rate": 7.329037751915981e-05,
      "loss": 0.005348076671361923,
      "step": 94140
    },
    {
      "epoch": 26.72438262844167,
      "grad_norm": 1.153486728668213,
      "learning_rate": 7.328753902923645e-05,
      "loss": 0.016149234771728516,
      "step": 94150
    },
    {
      "epoch": 26.72722111836503,
      "grad_norm": 0.31281599402427673,
      "learning_rate": 7.32847005393131e-05,
      "loss": 0.0035915620625019073,
      "step": 94160
    },
    {
      "epoch": 26.73005960828839,
      "grad_norm": 2.5461089611053467,
      "learning_rate": 7.328186204938972e-05,
      "loss": 0.0040773067623376845,
      "step": 94170
    },
    {
      "epoch": 26.732898098211752,
      "grad_norm": 4.219906330108643,
      "learning_rate": 7.327902355946636e-05,
      "loss": 0.009428070485591888,
      "step": 94180
    },
    {
      "epoch": 26.73573658813511,
      "grad_norm": 1.0715082883834839,
      "learning_rate": 7.3276185069543e-05,
      "loss": 0.003371432051062584,
      "step": 94190
    },
    {
      "epoch": 26.738575078058474,
      "grad_norm": 0.19039398431777954,
      "learning_rate": 7.327334657961965e-05,
      "loss": 0.00784885436296463,
      "step": 94200
    },
    {
      "epoch": 26.741413567981834,
      "grad_norm": 7.357936859130859,
      "learning_rate": 7.327050808969629e-05,
      "loss": 0.011466562747955322,
      "step": 94210
    },
    {
      "epoch": 26.744252057905193,
      "grad_norm": 1.2628207206726074,
      "learning_rate": 7.326766959977293e-05,
      "loss": 0.0031861763447523115,
      "step": 94220
    },
    {
      "epoch": 26.747090547828556,
      "grad_norm": 0.24524575471878052,
      "learning_rate": 7.326483110984956e-05,
      "loss": 0.0031773947179317475,
      "step": 94230
    },
    {
      "epoch": 26.749929037751915,
      "grad_norm": 1.2926634550094604,
      "learning_rate": 7.32619926199262e-05,
      "loss": 0.010897711664438248,
      "step": 94240
    },
    {
      "epoch": 26.752767527675278,
      "grad_norm": 5.313648223876953,
      "learning_rate": 7.325915413000284e-05,
      "loss": 0.0052309799939394,
      "step": 94250
    },
    {
      "epoch": 26.755606017598637,
      "grad_norm": 11.837434768676758,
      "learning_rate": 7.325631564007948e-05,
      "loss": 0.006376129388809204,
      "step": 94260
    },
    {
      "epoch": 26.758444507521997,
      "grad_norm": 1.498656153678894,
      "learning_rate": 7.325347715015612e-05,
      "loss": 0.020749929547309875,
      "step": 94270
    },
    {
      "epoch": 26.76128299744536,
      "grad_norm": 1.3998619318008423,
      "learning_rate": 7.325063866023276e-05,
      "loss": 0.005323236435651779,
      "step": 94280
    },
    {
      "epoch": 26.76412148736872,
      "grad_norm": 10.666291236877441,
      "learning_rate": 7.32478001703094e-05,
      "loss": 0.010934248566627502,
      "step": 94290
    },
    {
      "epoch": 26.766959977292082,
      "grad_norm": 15.526649475097656,
      "learning_rate": 7.324496168038603e-05,
      "loss": 0.01293966919183731,
      "step": 94300
    },
    {
      "epoch": 26.76979846721544,
      "grad_norm": 6.155138969421387,
      "learning_rate": 7.324212319046267e-05,
      "loss": 0.011608079075813293,
      "step": 94310
    },
    {
      "epoch": 26.7726369571388,
      "grad_norm": 0.2257644385099411,
      "learning_rate": 7.323928470053932e-05,
      "loss": 0.01860720068216324,
      "step": 94320
    },
    {
      "epoch": 26.775475447062163,
      "grad_norm": 0.31138718128204346,
      "learning_rate": 7.323644621061596e-05,
      "loss": 0.017933444678783418,
      "step": 94330
    },
    {
      "epoch": 26.778313936985523,
      "grad_norm": 0.22316892445087433,
      "learning_rate": 7.32336077206926e-05,
      "loss": 0.01873452216386795,
      "step": 94340
    },
    {
      "epoch": 26.781152426908886,
      "grad_norm": 1.287440299987793,
      "learning_rate": 7.323076923076924e-05,
      "loss": 0.0024506187066435812,
      "step": 94350
    },
    {
      "epoch": 26.783990916832245,
      "grad_norm": 6.362514019012451,
      "learning_rate": 7.322793074084587e-05,
      "loss": 0.012238384038209916,
      "step": 94360
    },
    {
      "epoch": 26.786829406755604,
      "grad_norm": 1.06751549243927,
      "learning_rate": 7.322509225092251e-05,
      "loss": 0.009376539289951325,
      "step": 94370
    },
    {
      "epoch": 26.789667896678967,
      "grad_norm": 2.0011606216430664,
      "learning_rate": 7.322225376099915e-05,
      "loss": 0.010755889117717743,
      "step": 94380
    },
    {
      "epoch": 26.792506386602327,
      "grad_norm": 0.18323780596256256,
      "learning_rate": 7.321941527107579e-05,
      "loss": 0.0080840103328228,
      "step": 94390
    },
    {
      "epoch": 26.79534487652569,
      "grad_norm": 0.35479533672332764,
      "learning_rate": 7.321657678115243e-05,
      "loss": 0.002134854905307293,
      "step": 94400
    },
    {
      "epoch": 26.79818336644905,
      "grad_norm": 4.576927185058594,
      "learning_rate": 7.321373829122907e-05,
      "loss": 0.01145692765712738,
      "step": 94410
    },
    {
      "epoch": 26.801021856372408,
      "grad_norm": 0.4212914705276489,
      "learning_rate": 7.321089980130572e-05,
      "loss": 0.005034103989601135,
      "step": 94420
    },
    {
      "epoch": 26.80386034629577,
      "grad_norm": 0.6421583890914917,
      "learning_rate": 7.320806131138234e-05,
      "loss": 0.00903690904378891,
      "step": 94430
    },
    {
      "epoch": 26.80669883621913,
      "grad_norm": 12.132715225219727,
      "learning_rate": 7.320522282145898e-05,
      "loss": 0.016330939531326295,
      "step": 94440
    },
    {
      "epoch": 26.809537326142493,
      "grad_norm": 10.826515197753906,
      "learning_rate": 7.320238433153563e-05,
      "loss": 0.01073497161269188,
      "step": 94450
    },
    {
      "epoch": 26.812375816065853,
      "grad_norm": 1.2761679887771606,
      "learning_rate": 7.319954584161225e-05,
      "loss": 0.021025915443897248,
      "step": 94460
    },
    {
      "epoch": 26.815214305989215,
      "grad_norm": 0.7034673094749451,
      "learning_rate": 7.319670735168891e-05,
      "loss": 0.017810511589050292,
      "step": 94470
    },
    {
      "epoch": 26.818052795912575,
      "grad_norm": 0.815653920173645,
      "learning_rate": 7.319386886176555e-05,
      "loss": 0.005450914800167084,
      "step": 94480
    },
    {
      "epoch": 26.820891285835934,
      "grad_norm": 1.0747979879379272,
      "learning_rate": 7.319103037184218e-05,
      "loss": 0.009490350633859635,
      "step": 94490
    },
    {
      "epoch": 26.823729775759297,
      "grad_norm": 0.16881529986858368,
      "learning_rate": 7.318819188191882e-05,
      "loss": 0.009227895736694336,
      "step": 94500
    },
    {
      "epoch": 26.823729775759297,
      "eval_accuracy": 0.9666815031474534,
      "eval_loss": 0.11548859626054764,
      "eval_runtime": 31.8408,
      "eval_samples_per_second": 493.925,
      "eval_steps_per_second": 7.726,
      "step": 94500
    },
    {
      "epoch": 26.826568265682656,
      "grad_norm": 6.397447109222412,
      "learning_rate": 7.318535339199546e-05,
      "loss": 0.020643207430839538,
      "step": 94510
    },
    {
      "epoch": 26.82940675560602,
      "grad_norm": 0.5587457418441772,
      "learning_rate": 7.31825149020721e-05,
      "loss": 0.011985748261213302,
      "step": 94520
    },
    {
      "epoch": 26.83224524552938,
      "grad_norm": 1.2035893201828003,
      "learning_rate": 7.317967641214874e-05,
      "loss": 0.008251894265413284,
      "step": 94530
    },
    {
      "epoch": 26.835083735452738,
      "grad_norm": 0.9046289324760437,
      "learning_rate": 7.317683792222539e-05,
      "loss": 0.01099552884697914,
      "step": 94540
    },
    {
      "epoch": 26.8379222253761,
      "grad_norm": 9.948662757873535,
      "learning_rate": 7.317399943230203e-05,
      "loss": 0.016390039026737212,
      "step": 94550
    },
    {
      "epoch": 26.84076071529946,
      "grad_norm": 2.8673548698425293,
      "learning_rate": 7.317116094237865e-05,
      "loss": 0.0057214625179767605,
      "step": 94560
    },
    {
      "epoch": 26.843599205222823,
      "grad_norm": 16.781118392944336,
      "learning_rate": 7.31683224524553e-05,
      "loss": 0.0250559002161026,
      "step": 94570
    },
    {
      "epoch": 26.846437695146182,
      "grad_norm": 6.041494369506836,
      "learning_rate": 7.316548396253194e-05,
      "loss": 0.005941551178693771,
      "step": 94580
    },
    {
      "epoch": 26.84927618506954,
      "grad_norm": 2.3877108097076416,
      "learning_rate": 7.316264547260856e-05,
      "loss": 0.007182009518146515,
      "step": 94590
    },
    {
      "epoch": 26.852114674992904,
      "grad_norm": 1.4107893705368042,
      "learning_rate": 7.315980698268522e-05,
      "loss": 0.004748644679784775,
      "step": 94600
    },
    {
      "epoch": 26.854953164916264,
      "grad_norm": 2.2655131816864014,
      "learning_rate": 7.315696849276186e-05,
      "loss": 0.020558658242225646,
      "step": 94610
    },
    {
      "epoch": 26.857791654839627,
      "grad_norm": 0.6342401504516602,
      "learning_rate": 7.315413000283849e-05,
      "loss": 0.01361614465713501,
      "step": 94620
    },
    {
      "epoch": 26.860630144762986,
      "grad_norm": 14.508710861206055,
      "learning_rate": 7.315129151291513e-05,
      "loss": 0.02947966754436493,
      "step": 94630
    },
    {
      "epoch": 26.863468634686345,
      "grad_norm": 4.156998157501221,
      "learning_rate": 7.314845302299177e-05,
      "loss": 0.006452225148677826,
      "step": 94640
    },
    {
      "epoch": 26.86630712460971,
      "grad_norm": 1.8338624238967896,
      "learning_rate": 7.314561453306841e-05,
      "loss": 0.012524004280567168,
      "step": 94650
    },
    {
      "epoch": 26.869145614533068,
      "grad_norm": 1.187557339668274,
      "learning_rate": 7.314277604314504e-05,
      "loss": 0.02244972586631775,
      "step": 94660
    },
    {
      "epoch": 26.87198410445643,
      "grad_norm": 8.91844367980957,
      "learning_rate": 7.31399375532217e-05,
      "loss": 0.01820616126060486,
      "step": 94670
    },
    {
      "epoch": 26.87482259437979,
      "grad_norm": 9.856206893920898,
      "learning_rate": 7.313709906329834e-05,
      "loss": 0.012397131323814392,
      "step": 94680
    },
    {
      "epoch": 26.87766108430315,
      "grad_norm": 8.526193618774414,
      "learning_rate": 7.313426057337497e-05,
      "loss": 0.01205759346485138,
      "step": 94690
    },
    {
      "epoch": 26.880499574226512,
      "grad_norm": 1.1396818161010742,
      "learning_rate": 7.31314220834516e-05,
      "loss": 0.008211804926395417,
      "step": 94700
    },
    {
      "epoch": 26.88333806414987,
      "grad_norm": 5.030377388000488,
      "learning_rate": 7.312858359352825e-05,
      "loss": 0.010646968334913253,
      "step": 94710
    },
    {
      "epoch": 26.886176554073234,
      "grad_norm": 0.252395898103714,
      "learning_rate": 7.312574510360488e-05,
      "loss": 0.004068059474229812,
      "step": 94720
    },
    {
      "epoch": 26.889015043996594,
      "grad_norm": 0.7781500816345215,
      "learning_rate": 7.312290661368153e-05,
      "loss": 0.008444656431674958,
      "step": 94730
    },
    {
      "epoch": 26.891853533919953,
      "grad_norm": 4.12952184677124,
      "learning_rate": 7.312006812375817e-05,
      "loss": 0.004265264049172402,
      "step": 94740
    },
    {
      "epoch": 26.894692023843316,
      "grad_norm": 0.4212832748889923,
      "learning_rate": 7.31172296338348e-05,
      "loss": 0.012350338697433471,
      "step": 94750
    },
    {
      "epoch": 26.897530513766675,
      "grad_norm": 0.3568253219127655,
      "learning_rate": 7.311439114391144e-05,
      "loss": 0.013381332159042358,
      "step": 94760
    },
    {
      "epoch": 26.900369003690038,
      "grad_norm": 0.46841901540756226,
      "learning_rate": 7.311155265398808e-05,
      "loss": 0.014715728163719178,
      "step": 94770
    },
    {
      "epoch": 26.903207493613397,
      "grad_norm": 3.224759578704834,
      "learning_rate": 7.310871416406472e-05,
      "loss": 0.006854061782360077,
      "step": 94780
    },
    {
      "epoch": 26.906045983536757,
      "grad_norm": 0.8850780725479126,
      "learning_rate": 7.310587567414135e-05,
      "loss": 0.017140981554985047,
      "step": 94790
    },
    {
      "epoch": 26.90888447346012,
      "grad_norm": 2.9493415355682373,
      "learning_rate": 7.3103037184218e-05,
      "loss": 0.006350439041852951,
      "step": 94800
    },
    {
      "epoch": 26.91172296338348,
      "grad_norm": 0.2581605315208435,
      "learning_rate": 7.310019869429465e-05,
      "loss": 0.009014574438333511,
      "step": 94810
    },
    {
      "epoch": 26.914561453306842,
      "grad_norm": 7.52855920791626,
      "learning_rate": 7.309736020437128e-05,
      "loss": 0.01122516170144081,
      "step": 94820
    },
    {
      "epoch": 26.9173999432302,
      "grad_norm": 9.842864036560059,
      "learning_rate": 7.309452171444792e-05,
      "loss": 0.010161265730857849,
      "step": 94830
    },
    {
      "epoch": 26.920238433153564,
      "grad_norm": 8.342184066772461,
      "learning_rate": 7.309168322452456e-05,
      "loss": 0.025299814343452454,
      "step": 94840
    },
    {
      "epoch": 26.923076923076923,
      "grad_norm": 14.028326988220215,
      "learning_rate": 7.308884473460119e-05,
      "loss": 0.010789446532726288,
      "step": 94850
    },
    {
      "epoch": 26.925915413000283,
      "grad_norm": 5.466632843017578,
      "learning_rate": 7.308600624467784e-05,
      "loss": 0.02698662579059601,
      "step": 94860
    },
    {
      "epoch": 26.928753902923646,
      "grad_norm": 9.788919448852539,
      "learning_rate": 7.308316775475448e-05,
      "loss": 0.008938751369714736,
      "step": 94870
    },
    {
      "epoch": 26.931592392847005,
      "grad_norm": 3.878929376602173,
      "learning_rate": 7.308032926483111e-05,
      "loss": 0.0148302361369133,
      "step": 94880
    },
    {
      "epoch": 26.934430882770368,
      "grad_norm": 0.4376879334449768,
      "learning_rate": 7.307749077490775e-05,
      "loss": 0.015847574174404144,
      "step": 94890
    },
    {
      "epoch": 26.937269372693727,
      "grad_norm": 10.621622085571289,
      "learning_rate": 7.307465228498439e-05,
      "loss": 0.01660303622484207,
      "step": 94900
    },
    {
      "epoch": 26.940107862617086,
      "grad_norm": 0.3777320981025696,
      "learning_rate": 7.307181379506103e-05,
      "loss": 0.008840611577033997,
      "step": 94910
    },
    {
      "epoch": 26.94294635254045,
      "grad_norm": 1.948773741722107,
      "learning_rate": 7.306897530513766e-05,
      "loss": 0.011807572841644288,
      "step": 94920
    },
    {
      "epoch": 26.94578484246381,
      "grad_norm": 11.493670463562012,
      "learning_rate": 7.306613681521432e-05,
      "loss": 0.013804540038108826,
      "step": 94930
    },
    {
      "epoch": 26.94862333238717,
      "grad_norm": 0.6795633435249329,
      "learning_rate": 7.306329832529096e-05,
      "loss": 0.031068021059036256,
      "step": 94940
    },
    {
      "epoch": 26.95146182231053,
      "grad_norm": 12.200681686401367,
      "learning_rate": 7.306045983536759e-05,
      "loss": 0.015989182889461516,
      "step": 94950
    },
    {
      "epoch": 26.95430031223389,
      "grad_norm": 5.475698947906494,
      "learning_rate": 7.305762134544423e-05,
      "loss": 0.018589550256729127,
      "step": 94960
    },
    {
      "epoch": 26.957138802157253,
      "grad_norm": 10.37171459197998,
      "learning_rate": 7.305478285552087e-05,
      "loss": 0.014926382899284362,
      "step": 94970
    },
    {
      "epoch": 26.959977292080612,
      "grad_norm": 1.591007947921753,
      "learning_rate": 7.30519443655975e-05,
      "loss": 0.00547194704413414,
      "step": 94980
    },
    {
      "epoch": 26.962815782003975,
      "grad_norm": 1.946585774421692,
      "learning_rate": 7.304910587567414e-05,
      "loss": 0.010621996968984604,
      "step": 94990
    },
    {
      "epoch": 26.965654271927335,
      "grad_norm": 3.2840471267700195,
      "learning_rate": 7.30462673857508e-05,
      "loss": 0.022071157395839692,
      "step": 95000
    },
    {
      "epoch": 26.965654271927335,
      "eval_accuracy": 0.9651554651236727,
      "eval_loss": 0.1152285560965538,
      "eval_runtime": 31.8897,
      "eval_samples_per_second": 493.169,
      "eval_steps_per_second": 7.714,
      "step": 95000
    },
    {
      "epoch": 26.968492761850694,
      "grad_norm": 5.742368698120117,
      "learning_rate": 7.304342889582742e-05,
      "loss": 0.009496699273586272,
      "step": 95010
    },
    {
      "epoch": 26.971331251774057,
      "grad_norm": 0.09833352267742157,
      "learning_rate": 7.304059040590406e-05,
      "loss": 0.0046347551047801975,
      "step": 95020
    },
    {
      "epoch": 26.974169741697416,
      "grad_norm": 0.25711163878440857,
      "learning_rate": 7.30377519159807e-05,
      "loss": 0.017611576616764067,
      "step": 95030
    },
    {
      "epoch": 26.97700823162078,
      "grad_norm": 8.331297874450684,
      "learning_rate": 7.303491342605735e-05,
      "loss": 0.008422775566577912,
      "step": 95040
    },
    {
      "epoch": 26.97984672154414,
      "grad_norm": 1.734641671180725,
      "learning_rate": 7.303207493613397e-05,
      "loss": 0.01211327239871025,
      "step": 95050
    },
    {
      "epoch": 26.982685211467498,
      "grad_norm": 0.5438372492790222,
      "learning_rate": 7.302923644621063e-05,
      "loss": 0.008285184949636459,
      "step": 95060
    },
    {
      "epoch": 26.98552370139086,
      "grad_norm": 13.321969032287598,
      "learning_rate": 7.302639795628726e-05,
      "loss": 0.008095437288284301,
      "step": 95070
    },
    {
      "epoch": 26.98836219131422,
      "grad_norm": 5.428834438323975,
      "learning_rate": 7.30235594663639e-05,
      "loss": 0.014718478918075562,
      "step": 95080
    },
    {
      "epoch": 26.991200681237583,
      "grad_norm": 1.988488793373108,
      "learning_rate": 7.302072097644054e-05,
      "loss": 0.014892590045928956,
      "step": 95090
    },
    {
      "epoch": 26.994039171160942,
      "grad_norm": 0.21988387405872345,
      "learning_rate": 7.301788248651718e-05,
      "loss": 0.015247519314289092,
      "step": 95100
    },
    {
      "epoch": 26.9968776610843,
      "grad_norm": 4.11231803894043,
      "learning_rate": 7.301504399659381e-05,
      "loss": 0.012525960803031921,
      "step": 95110
    },
    {
      "epoch": 26.999716151007664,
      "grad_norm": 1.8023399114608765,
      "learning_rate": 7.301220550667045e-05,
      "loss": 0.01647813767194748,
      "step": 95120
    },
    {
      "epoch": 27.002554640931024,
      "grad_norm": 7.9026780128479,
      "learning_rate": 7.30093670167471e-05,
      "loss": 0.014436399936676026,
      "step": 95130
    },
    {
      "epoch": 27.005393130854387,
      "grad_norm": 1.0764693021774292,
      "learning_rate": 7.300652852682373e-05,
      "loss": 0.007888266444206237,
      "step": 95140
    },
    {
      "epoch": 27.008231620777746,
      "grad_norm": 0.08524254709482193,
      "learning_rate": 7.300369003690037e-05,
      "loss": 0.01325371116399765,
      "step": 95150
    },
    {
      "epoch": 27.011070110701105,
      "grad_norm": 0.18291603028774261,
      "learning_rate": 7.300085154697701e-05,
      "loss": 0.014587007462978363,
      "step": 95160
    },
    {
      "epoch": 27.013908600624468,
      "grad_norm": 0.2014232873916626,
      "learning_rate": 7.299801305705364e-05,
      "loss": 0.007328209280967712,
      "step": 95170
    },
    {
      "epoch": 27.016747090547828,
      "grad_norm": 0.8993664979934692,
      "learning_rate": 7.299517456713028e-05,
      "loss": 0.006159491091966629,
      "step": 95180
    },
    {
      "epoch": 27.01958558047119,
      "grad_norm": 7.0965495109558105,
      "learning_rate": 7.299233607720693e-05,
      "loss": 0.0031824085861444475,
      "step": 95190
    },
    {
      "epoch": 27.02242407039455,
      "grad_norm": 9.556612014770508,
      "learning_rate": 7.298949758728357e-05,
      "loss": 0.012323324382305146,
      "step": 95200
    },
    {
      "epoch": 27.025262560317913,
      "grad_norm": 0.1286274492740631,
      "learning_rate": 7.298665909736021e-05,
      "loss": 0.00985255092382431,
      "step": 95210
    },
    {
      "epoch": 27.028101050241272,
      "grad_norm": 7.862353801727295,
      "learning_rate": 7.298382060743685e-05,
      "loss": 0.0120895616710186,
      "step": 95220
    },
    {
      "epoch": 27.03093954016463,
      "grad_norm": 8.758325576782227,
      "learning_rate": 7.298098211751349e-05,
      "loss": 0.006636605411767959,
      "step": 95230
    },
    {
      "epoch": 27.033778030087994,
      "grad_norm": 1.5534446239471436,
      "learning_rate": 7.297814362759012e-05,
      "loss": 0.002974515035748482,
      "step": 95240
    },
    {
      "epoch": 27.036616520011354,
      "grad_norm": 0.06047556549310684,
      "learning_rate": 7.297530513766676e-05,
      "loss": 0.002357245981693268,
      "step": 95250
    },
    {
      "epoch": 27.039455009934716,
      "grad_norm": 0.07717142254114151,
      "learning_rate": 7.297246664774341e-05,
      "loss": 0.0028319453820586206,
      "step": 95260
    },
    {
      "epoch": 27.042293499858076,
      "grad_norm": 1.5141997337341309,
      "learning_rate": 7.296962815782004e-05,
      "loss": 0.012891288101673126,
      "step": 95270
    },
    {
      "epoch": 27.045131989781435,
      "grad_norm": 2.8872857093811035,
      "learning_rate": 7.296678966789668e-05,
      "loss": 0.005435745045542717,
      "step": 95280
    },
    {
      "epoch": 27.047970479704798,
      "grad_norm": 7.0280961990356445,
      "learning_rate": 7.296395117797333e-05,
      "loss": 0.012304079532623292,
      "step": 95290
    },
    {
      "epoch": 27.050808969628157,
      "grad_norm": 1.1121705770492554,
      "learning_rate": 7.296111268804995e-05,
      "loss": 0.003240145742893219,
      "step": 95300
    },
    {
      "epoch": 27.05364745955152,
      "grad_norm": 0.7756054997444153,
      "learning_rate": 7.29582741981266e-05,
      "loss": 0.00594131276011467,
      "step": 95310
    },
    {
      "epoch": 27.05648594947488,
      "grad_norm": 0.334761381149292,
      "learning_rate": 7.295543570820324e-05,
      "loss": 0.019674669206142425,
      "step": 95320
    },
    {
      "epoch": 27.05932443939824,
      "grad_norm": 0.1335884928703308,
      "learning_rate": 7.295259721827988e-05,
      "loss": 0.0056454546749591826,
      "step": 95330
    },
    {
      "epoch": 27.0621629293216,
      "grad_norm": 1.4134819507598877,
      "learning_rate": 7.294975872835652e-05,
      "loss": 0.014025644958019256,
      "step": 95340
    },
    {
      "epoch": 27.06500141924496,
      "grad_norm": 0.21472875773906708,
      "learning_rate": 7.294692023843316e-05,
      "loss": 0.00498523935675621,
      "step": 95350
    },
    {
      "epoch": 27.067839909168324,
      "grad_norm": 2.6188158988952637,
      "learning_rate": 7.29440817485098e-05,
      "loss": 0.015001380443572998,
      "step": 95360
    },
    {
      "epoch": 27.070678399091683,
      "grad_norm": 0.24918845295906067,
      "learning_rate": 7.294124325858643e-05,
      "loss": 0.0036949679255485536,
      "step": 95370
    },
    {
      "epoch": 27.073516889015043,
      "grad_norm": 3.7165369987487793,
      "learning_rate": 7.293840476866307e-05,
      "loss": 0.0031681805849075317,
      "step": 95380
    },
    {
      "epoch": 27.076355378938405,
      "grad_norm": 0.15804407000541687,
      "learning_rate": 7.293556627873971e-05,
      "loss": 0.0019795143976807593,
      "step": 95390
    },
    {
      "epoch": 27.079193868861765,
      "grad_norm": 0.23503343760967255,
      "learning_rate": 7.293272778881635e-05,
      "loss": 0.004706960916519165,
      "step": 95400
    },
    {
      "epoch": 27.082032358785128,
      "grad_norm": 0.255169153213501,
      "learning_rate": 7.2929889298893e-05,
      "loss": 0.004814198240637779,
      "step": 95410
    },
    {
      "epoch": 27.084870848708487,
      "grad_norm": 2.0719711780548096,
      "learning_rate": 7.292705080896964e-05,
      "loss": 0.009840811043977738,
      "step": 95420
    },
    {
      "epoch": 27.087709338631846,
      "grad_norm": 1.1336169242858887,
      "learning_rate": 7.292421231904626e-05,
      "loss": 0.0035200797021389006,
      "step": 95430
    },
    {
      "epoch": 27.09054782855521,
      "grad_norm": 1.4868299961090088,
      "learning_rate": 7.29213738291229e-05,
      "loss": 0.006213813275098801,
      "step": 95440
    },
    {
      "epoch": 27.09338631847857,
      "grad_norm": 0.1265774965286255,
      "learning_rate": 7.291853533919955e-05,
      "loss": 0.021602602303028108,
      "step": 95450
    },
    {
      "epoch": 27.09622480840193,
      "grad_norm": 3.412343978881836,
      "learning_rate": 7.291569684927619e-05,
      "loss": 0.003488282114267349,
      "step": 95460
    },
    {
      "epoch": 27.09906329832529,
      "grad_norm": 4.044987201690674,
      "learning_rate": 7.291285835935283e-05,
      "loss": 0.005457519739866257,
      "step": 95470
    },
    {
      "epoch": 27.10190178824865,
      "grad_norm": 8.39084243774414,
      "learning_rate": 7.291001986942947e-05,
      "loss": 0.015611542761325837,
      "step": 95480
    },
    {
      "epoch": 27.104740278172013,
      "grad_norm": 4.072351932525635,
      "learning_rate": 7.290718137950611e-05,
      "loss": 0.00747731626033783,
      "step": 95490
    },
    {
      "epoch": 27.107578768095372,
      "grad_norm": 0.2045382261276245,
      "learning_rate": 7.290434288958274e-05,
      "loss": 0.005929672718048095,
      "step": 95500
    },
    {
      "epoch": 27.107578768095372,
      "eval_accuracy": 0.96986074903033,
      "eval_loss": 0.10046815872192383,
      "eval_runtime": 31.8048,
      "eval_samples_per_second": 494.485,
      "eval_steps_per_second": 7.735,
      "step": 95500
    },
    {
      "epoch": 27.110417258018735,
      "grad_norm": 3.092857837677002,
      "learning_rate": 7.290150439965938e-05,
      "loss": 0.007351582497358322,
      "step": 95510
    },
    {
      "epoch": 27.113255747942095,
      "grad_norm": 0.35463252663612366,
      "learning_rate": 7.289866590973602e-05,
      "loss": 0.0043140958994627,
      "step": 95520
    },
    {
      "epoch": 27.116094237865454,
      "grad_norm": 0.03940659761428833,
      "learning_rate": 7.289582741981266e-05,
      "loss": 0.005683646351099014,
      "step": 95530
    },
    {
      "epoch": 27.118932727788817,
      "grad_norm": 0.4777376055717468,
      "learning_rate": 7.28929889298893e-05,
      "loss": 0.002036338672041893,
      "step": 95540
    },
    {
      "epoch": 27.121771217712176,
      "grad_norm": 2.1356241703033447,
      "learning_rate": 7.289015043996595e-05,
      "loss": 0.0062846653163433075,
      "step": 95550
    },
    {
      "epoch": 27.12460970763554,
      "grad_norm": 0.8136447668075562,
      "learning_rate": 7.288731195004257e-05,
      "loss": 0.018706507980823517,
      "step": 95560
    },
    {
      "epoch": 27.1274481975589,
      "grad_norm": 7.297195911407471,
      "learning_rate": 7.288447346011922e-05,
      "loss": 0.010456858575344086,
      "step": 95570
    },
    {
      "epoch": 27.130286687482258,
      "grad_norm": 3.5316708087921143,
      "learning_rate": 7.288163497019586e-05,
      "loss": 0.02014935314655304,
      "step": 95580
    },
    {
      "epoch": 27.13312517740562,
      "grad_norm": 1.3246920108795166,
      "learning_rate": 7.28787964802725e-05,
      "loss": 0.012094598263502121,
      "step": 95590
    },
    {
      "epoch": 27.13596366732898,
      "grad_norm": 0.6021990776062012,
      "learning_rate": 7.287595799034914e-05,
      "loss": 0.0029250387102365493,
      "step": 95600
    },
    {
      "epoch": 27.138802157252343,
      "grad_norm": 16.04469871520996,
      "learning_rate": 7.287311950042578e-05,
      "loss": 0.005827989429235458,
      "step": 95610
    },
    {
      "epoch": 27.141640647175702,
      "grad_norm": 8.3602933883667,
      "learning_rate": 7.287028101050242e-05,
      "loss": 0.011805259436368943,
      "step": 95620
    },
    {
      "epoch": 27.144479137099065,
      "grad_norm": 0.07793623954057693,
      "learning_rate": 7.286744252057905e-05,
      "loss": 0.006363669037818908,
      "step": 95630
    },
    {
      "epoch": 27.147317627022424,
      "grad_norm": 0.9708279371261597,
      "learning_rate": 7.286460403065569e-05,
      "loss": 0.002565213479101658,
      "step": 95640
    },
    {
      "epoch": 27.150156116945784,
      "grad_norm": 11.768292427062988,
      "learning_rate": 7.286176554073233e-05,
      "loss": 0.012795543670654297,
      "step": 95650
    },
    {
      "epoch": 27.152994606869147,
      "grad_norm": 4.34949254989624,
      "learning_rate": 7.285892705080898e-05,
      "loss": 0.007452624291181565,
      "step": 95660
    },
    {
      "epoch": 27.155833096792506,
      "grad_norm": 0.5058379173278809,
      "learning_rate": 7.285608856088562e-05,
      "loss": 0.0036213286221027374,
      "step": 95670
    },
    {
      "epoch": 27.15867158671587,
      "grad_norm": 0.32105812430381775,
      "learning_rate": 7.285325007096226e-05,
      "loss": 0.0041717179119586945,
      "step": 95680
    },
    {
      "epoch": 27.161510076639228,
      "grad_norm": 14.32698917388916,
      "learning_rate": 7.285041158103889e-05,
      "loss": 0.012550786137580872,
      "step": 95690
    },
    {
      "epoch": 27.164348566562587,
      "grad_norm": 4.324656009674072,
      "learning_rate": 7.284757309111553e-05,
      "loss": 0.011185474693775177,
      "step": 95700
    },
    {
      "epoch": 27.16718705648595,
      "grad_norm": 6.045734882354736,
      "learning_rate": 7.284473460119217e-05,
      "loss": 0.008705034852027893,
      "step": 95710
    },
    {
      "epoch": 27.17002554640931,
      "grad_norm": 3.5558059215545654,
      "learning_rate": 7.284189611126881e-05,
      "loss": 0.006051670387387275,
      "step": 95720
    },
    {
      "epoch": 27.172864036332673,
      "grad_norm": 3.1671321392059326,
      "learning_rate": 7.283905762134545e-05,
      "loss": 0.01106426641345024,
      "step": 95730
    },
    {
      "epoch": 27.175702526256032,
      "grad_norm": 0.912066638469696,
      "learning_rate": 7.283621913142209e-05,
      "loss": 0.006127223372459412,
      "step": 95740
    },
    {
      "epoch": 27.17854101617939,
      "grad_norm": 14.557927131652832,
      "learning_rate": 7.283338064149873e-05,
      "loss": 0.016689641773700713,
      "step": 95750
    },
    {
      "epoch": 27.181379506102754,
      "grad_norm": 0.20242708921432495,
      "learning_rate": 7.283054215157536e-05,
      "loss": 0.010514412820339204,
      "step": 95760
    },
    {
      "epoch": 27.184217996026113,
      "grad_norm": 0.16144487261772156,
      "learning_rate": 7.2827703661652e-05,
      "loss": 0.0036367315798997877,
      "step": 95770
    },
    {
      "epoch": 27.187056485949476,
      "grad_norm": 9.16048526763916,
      "learning_rate": 7.282486517172864e-05,
      "loss": 0.008228586614131927,
      "step": 95780
    },
    {
      "epoch": 27.189894975872836,
      "grad_norm": 0.05781473591923714,
      "learning_rate": 7.282202668180527e-05,
      "loss": 0.0021053383126854895,
      "step": 95790
    },
    {
      "epoch": 27.192733465796195,
      "grad_norm": 0.7909797430038452,
      "learning_rate": 7.281918819188193e-05,
      "loss": 0.004358413070440293,
      "step": 95800
    },
    {
      "epoch": 27.195571955719558,
      "grad_norm": 1.9347373247146606,
      "learning_rate": 7.281634970195857e-05,
      "loss": 0.01045796126127243,
      "step": 95810
    },
    {
      "epoch": 27.198410445642917,
      "grad_norm": 1.1858774423599243,
      "learning_rate": 7.28135112120352e-05,
      "loss": 0.008381951600313187,
      "step": 95820
    },
    {
      "epoch": 27.20124893556628,
      "grad_norm": 7.8907270431518555,
      "learning_rate": 7.281067272211184e-05,
      "loss": 0.009133132547140122,
      "step": 95830
    },
    {
      "epoch": 27.20408742548964,
      "grad_norm": 5.41282844543457,
      "learning_rate": 7.280783423218848e-05,
      "loss": 0.016681912541389465,
      "step": 95840
    },
    {
      "epoch": 27.206925915413,
      "grad_norm": 1.2343826293945312,
      "learning_rate": 7.280499574226512e-05,
      "loss": 0.005586374178528785,
      "step": 95850
    },
    {
      "epoch": 27.20976440533636,
      "grad_norm": 0.14001768827438354,
      "learning_rate": 7.280215725234176e-05,
      "loss": 0.004759392887353897,
      "step": 95860
    },
    {
      "epoch": 27.21260289525972,
      "grad_norm": 0.3972637951374054,
      "learning_rate": 7.27993187624184e-05,
      "loss": 0.0048266701400279995,
      "step": 95870
    },
    {
      "epoch": 27.215441385183084,
      "grad_norm": 1.9308353662490845,
      "learning_rate": 7.279648027249504e-05,
      "loss": 0.003910179436206818,
      "step": 95880
    },
    {
      "epoch": 27.218279875106443,
      "grad_norm": 2.325427770614624,
      "learning_rate": 7.279364178257167e-05,
      "loss": 0.01410355269908905,
      "step": 95890
    },
    {
      "epoch": 27.221118365029803,
      "grad_norm": 0.1319456845521927,
      "learning_rate": 7.279080329264831e-05,
      "loss": 0.00962822139263153,
      "step": 95900
    },
    {
      "epoch": 27.223956854953165,
      "grad_norm": 0.4218675196170807,
      "learning_rate": 7.278796480272496e-05,
      "loss": 0.02063506543636322,
      "step": 95910
    },
    {
      "epoch": 27.226795344876525,
      "grad_norm": 3.877767562866211,
      "learning_rate": 7.278512631280158e-05,
      "loss": 0.003137446567416191,
      "step": 95920
    },
    {
      "epoch": 27.229633834799888,
      "grad_norm": 1.1283106803894043,
      "learning_rate": 7.278228782287824e-05,
      "loss": 0.0037359926849603654,
      "step": 95930
    },
    {
      "epoch": 27.232472324723247,
      "grad_norm": 0.5127775073051453,
      "learning_rate": 7.277944933295488e-05,
      "loss": 0.006305716931819916,
      "step": 95940
    },
    {
      "epoch": 27.235310814646606,
      "grad_norm": 0.24632003903388977,
      "learning_rate": 7.277661084303151e-05,
      "loss": 0.019400307536125184,
      "step": 95950
    },
    {
      "epoch": 27.23814930456997,
      "grad_norm": 1.3541440963745117,
      "learning_rate": 7.277377235310815e-05,
      "loss": 0.0021309757605195044,
      "step": 95960
    },
    {
      "epoch": 27.24098779449333,
      "grad_norm": 1.2853630781173706,
      "learning_rate": 7.277093386318479e-05,
      "loss": 0.006213001161813736,
      "step": 95970
    },
    {
      "epoch": 27.24382628441669,
      "grad_norm": 0.6624603271484375,
      "learning_rate": 7.276809537326143e-05,
      "loss": 0.005244841054081917,
      "step": 95980
    },
    {
      "epoch": 27.24666477434005,
      "grad_norm": 1.2028666734695435,
      "learning_rate": 7.276525688333806e-05,
      "loss": 0.0059256177395582196,
      "step": 95990
    },
    {
      "epoch": 27.249503264263414,
      "grad_norm": 1.048482060432434,
      "learning_rate": 7.276241839341471e-05,
      "loss": 0.02292678952217102,
      "step": 96000
    },
    {
      "epoch": 27.249503264263414,
      "eval_accuracy": 0.9691613149360971,
      "eval_loss": 0.1033172532916069,
      "eval_runtime": 31.4942,
      "eval_samples_per_second": 499.362,
      "eval_steps_per_second": 7.811,
      "step": 96000
    },
    {
      "epoch": 27.252341754186773,
      "grad_norm": 13.947169303894043,
      "learning_rate": 7.275957990349134e-05,
      "loss": 0.012080571055412293,
      "step": 96010
    },
    {
      "epoch": 27.255180244110132,
      "grad_norm": 4.547450065612793,
      "learning_rate": 7.275674141356798e-05,
      "loss": 0.006088655069470405,
      "step": 96020
    },
    {
      "epoch": 27.258018734033495,
      "grad_norm": 0.33254486322402954,
      "learning_rate": 7.275390292364462e-05,
      "loss": 0.01574893295764923,
      "step": 96030
    },
    {
      "epoch": 27.260857223956855,
      "grad_norm": 0.09293801337480545,
      "learning_rate": 7.275106443372127e-05,
      "loss": 0.009039023518562317,
      "step": 96040
    },
    {
      "epoch": 27.263695713880217,
      "grad_norm": 1.611240029335022,
      "learning_rate": 7.27482259437979e-05,
      "loss": 0.010182559490203857,
      "step": 96050
    },
    {
      "epoch": 27.266534203803577,
      "grad_norm": 0.7767118811607361,
      "learning_rate": 7.274538745387455e-05,
      "loss": 0.0025655966252088546,
      "step": 96060
    },
    {
      "epoch": 27.269372693726936,
      "grad_norm": 7.622793197631836,
      "learning_rate": 7.274254896395119e-05,
      "loss": 0.005044136568903923,
      "step": 96070
    },
    {
      "epoch": 27.2722111836503,
      "grad_norm": 0.8604416847229004,
      "learning_rate": 7.273971047402782e-05,
      "loss": 0.0035683788359165193,
      "step": 96080
    },
    {
      "epoch": 27.27504967357366,
      "grad_norm": 0.7299647927284241,
      "learning_rate": 7.273687198410446e-05,
      "loss": 0.01007835790514946,
      "step": 96090
    },
    {
      "epoch": 27.27788816349702,
      "grad_norm": 0.3364797830581665,
      "learning_rate": 7.27340334941811e-05,
      "loss": 0.0036778807640075683,
      "step": 96100
    },
    {
      "epoch": 27.28072665342038,
      "grad_norm": 0.30109700560569763,
      "learning_rate": 7.273119500425773e-05,
      "loss": 0.007915738970041275,
      "step": 96110
    },
    {
      "epoch": 27.28356514334374,
      "grad_norm": 6.812478065490723,
      "learning_rate": 7.272835651433437e-05,
      "loss": 0.007474875450134278,
      "step": 96120
    },
    {
      "epoch": 27.286403633267103,
      "grad_norm": 1.2262887954711914,
      "learning_rate": 7.272551802441102e-05,
      "loss": 0.0050614744424819945,
      "step": 96130
    },
    {
      "epoch": 27.289242123190462,
      "grad_norm": 9.31180191040039,
      "learning_rate": 7.272267953448765e-05,
      "loss": 0.0058542326092720035,
      "step": 96140
    },
    {
      "epoch": 27.292080613113825,
      "grad_norm": 0.45088517665863037,
      "learning_rate": 7.27198410445643e-05,
      "loss": 0.013272187113761902,
      "step": 96150
    },
    {
      "epoch": 27.294919103037184,
      "grad_norm": 0.4800405502319336,
      "learning_rate": 7.271700255464094e-05,
      "loss": 0.004820101708173752,
      "step": 96160
    },
    {
      "epoch": 27.297757592960544,
      "grad_norm": 0.3897879123687744,
      "learning_rate": 7.271416406471758e-05,
      "loss": 0.005838307738304138,
      "step": 96170
    },
    {
      "epoch": 27.300596082883906,
      "grad_norm": 3.394780158996582,
      "learning_rate": 7.27113255747942e-05,
      "loss": 0.003043442964553833,
      "step": 96180
    },
    {
      "epoch": 27.303434572807266,
      "grad_norm": 2.188339948654175,
      "learning_rate": 7.270848708487086e-05,
      "loss": 0.004591661319136619,
      "step": 96190
    },
    {
      "epoch": 27.30627306273063,
      "grad_norm": 3.340996026992798,
      "learning_rate": 7.27056485949475e-05,
      "loss": 0.0065595611929893495,
      "step": 96200
    },
    {
      "epoch": 27.309111552653988,
      "grad_norm": 1.8544424772262573,
      "learning_rate": 7.270281010502413e-05,
      "loss": 0.022549913823604585,
      "step": 96210
    },
    {
      "epoch": 27.311950042577347,
      "grad_norm": 1.800384521484375,
      "learning_rate": 7.269997161510077e-05,
      "loss": 0.00701739564538002,
      "step": 96220
    },
    {
      "epoch": 27.31478853250071,
      "grad_norm": 0.809044361114502,
      "learning_rate": 7.269713312517741e-05,
      "loss": 0.005051366239786148,
      "step": 96230
    },
    {
      "epoch": 27.31762702242407,
      "grad_norm": 0.43807655572891235,
      "learning_rate": 7.269429463525404e-05,
      "loss": 0.014744056761264801,
      "step": 96240
    },
    {
      "epoch": 27.320465512347432,
      "grad_norm": 0.2330809086561203,
      "learning_rate": 7.269145614533068e-05,
      "loss": 0.008742526173591614,
      "step": 96250
    },
    {
      "epoch": 27.323304002270792,
      "grad_norm": 0.2933183014392853,
      "learning_rate": 7.268861765540734e-05,
      "loss": 0.0083245649933815,
      "step": 96260
    },
    {
      "epoch": 27.32614249219415,
      "grad_norm": 2.33992862701416,
      "learning_rate": 7.268577916548396e-05,
      "loss": 0.009417203068733216,
      "step": 96270
    },
    {
      "epoch": 27.328980982117514,
      "grad_norm": 1.5231910943984985,
      "learning_rate": 7.26829406755606e-05,
      "loss": 0.009498282521963119,
      "step": 96280
    },
    {
      "epoch": 27.331819472040873,
      "grad_norm": 0.9619318246841431,
      "learning_rate": 7.268010218563725e-05,
      "loss": 0.012492598593235016,
      "step": 96290
    },
    {
      "epoch": 27.334657961964236,
      "grad_norm": 3.6967458724975586,
      "learning_rate": 7.267726369571389e-05,
      "loss": 0.007103332877159118,
      "step": 96300
    },
    {
      "epoch": 27.337496451887596,
      "grad_norm": 18.862524032592773,
      "learning_rate": 7.267442520579052e-05,
      "loss": 0.006693925708532333,
      "step": 96310
    },
    {
      "epoch": 27.340334941810955,
      "grad_norm": 0.29030734300613403,
      "learning_rate": 7.267158671586716e-05,
      "loss": 0.007195537537336349,
      "step": 96320
    },
    {
      "epoch": 27.343173431734318,
      "grad_norm": 1.9510637521743774,
      "learning_rate": 7.266874822594381e-05,
      "loss": 0.012335879355669021,
      "step": 96330
    },
    {
      "epoch": 27.346011921657677,
      "grad_norm": 0.15836979448795319,
      "learning_rate": 7.266590973602044e-05,
      "loss": 0.0038990002125501634,
      "step": 96340
    },
    {
      "epoch": 27.34885041158104,
      "grad_norm": 0.12012756615877151,
      "learning_rate": 7.266307124609708e-05,
      "loss": 0.007017889618873596,
      "step": 96350
    },
    {
      "epoch": 27.3516889015044,
      "grad_norm": 0.25763124227523804,
      "learning_rate": 7.266023275617372e-05,
      "loss": 0.016184821724891663,
      "step": 96360
    },
    {
      "epoch": 27.35452739142776,
      "grad_norm": 0.27734827995300293,
      "learning_rate": 7.265739426625035e-05,
      "loss": 0.0028782114386558533,
      "step": 96370
    },
    {
      "epoch": 27.35736588135112,
      "grad_norm": 5.8750200271606445,
      "learning_rate": 7.265455577632699e-05,
      "loss": 0.011175237596035004,
      "step": 96380
    },
    {
      "epoch": 27.36020437127448,
      "grad_norm": 0.6503676772117615,
      "learning_rate": 7.265171728640365e-05,
      "loss": 0.0018696736544370652,
      "step": 96390
    },
    {
      "epoch": 27.363042861197844,
      "grad_norm": 1.7437611818313599,
      "learning_rate": 7.264887879648027e-05,
      "loss": 0.009328977018594743,
      "step": 96400
    },
    {
      "epoch": 27.365881351121203,
      "grad_norm": 3.5311877727508545,
      "learning_rate": 7.264604030655692e-05,
      "loss": 0.013546699285507202,
      "step": 96410
    },
    {
      "epoch": 27.368719841044566,
      "grad_norm": 1.6429803371429443,
      "learning_rate": 7.264320181663356e-05,
      "loss": 0.00697207897901535,
      "step": 96420
    },
    {
      "epoch": 27.371558330967925,
      "grad_norm": 4.738337993621826,
      "learning_rate": 7.26403633267102e-05,
      "loss": 0.005013350024819374,
      "step": 96430
    },
    {
      "epoch": 27.374396820891285,
      "grad_norm": 0.4073253870010376,
      "learning_rate": 7.263752483678683e-05,
      "loss": 0.006968820840120316,
      "step": 96440
    },
    {
      "epoch": 27.377235310814648,
      "grad_norm": 2.3746659755706787,
      "learning_rate": 7.263468634686347e-05,
      "loss": 0.0054799336940050125,
      "step": 96450
    },
    {
      "epoch": 27.380073800738007,
      "grad_norm": 1.6901803016662598,
      "learning_rate": 7.263184785694012e-05,
      "loss": 0.009037511050701141,
      "step": 96460
    },
    {
      "epoch": 27.38291229066137,
      "grad_norm": 2.0042309761047363,
      "learning_rate": 7.262900936701675e-05,
      "loss": 0.022585903108119965,
      "step": 96470
    },
    {
      "epoch": 27.38575078058473,
      "grad_norm": 7.392411231994629,
      "learning_rate": 7.262617087709339e-05,
      "loss": 0.01852399706840515,
      "step": 96480
    },
    {
      "epoch": 27.38858927050809,
      "grad_norm": 1.6687690019607544,
      "learning_rate": 7.262361623616236e-05,
      "loss": 0.03147360980510712,
      "step": 96490
    },
    {
      "epoch": 27.39142776043145,
      "grad_norm": 8.044534683227539,
      "learning_rate": 7.2620777746239e-05,
      "loss": 0.013714486360549926,
      "step": 96500
    },
    {
      "epoch": 27.39142776043145,
      "eval_accuracy": 0.9504037642271254,
      "eval_loss": 0.17266052961349487,
      "eval_runtime": 31.8253,
      "eval_samples_per_second": 494.166,
      "eval_steps_per_second": 7.73,
      "step": 96500
    },
    {
      "epoch": 27.39426625035481,
      "grad_norm": 2.395951509475708,
      "learning_rate": 7.261793925631565e-05,
      "loss": 0.012301383912563324,
      "step": 96510
    },
    {
      "epoch": 27.397104740278174,
      "grad_norm": 12.739520072937012,
      "learning_rate": 7.261510076639228e-05,
      "loss": 0.007749860733747482,
      "step": 96520
    },
    {
      "epoch": 27.399943230201533,
      "grad_norm": 7.558526039123535,
      "learning_rate": 7.261226227646892e-05,
      "loss": 0.00884302332997322,
      "step": 96530
    },
    {
      "epoch": 27.402781720124892,
      "grad_norm": 3.3410837650299072,
      "learning_rate": 7.260942378654556e-05,
      "loss": 0.0058293651789426805,
      "step": 96540
    },
    {
      "epoch": 27.405620210048255,
      "grad_norm": 1.2156295776367188,
      "learning_rate": 7.260658529662219e-05,
      "loss": 0.010729898512363435,
      "step": 96550
    },
    {
      "epoch": 27.408458699971614,
      "grad_norm": 2.8206207752227783,
      "learning_rate": 7.260374680669883e-05,
      "loss": 0.003312871605157852,
      "step": 96560
    },
    {
      "epoch": 27.411297189894977,
      "grad_norm": 0.6392220258712769,
      "learning_rate": 7.260090831677549e-05,
      "loss": 0.004106239229440689,
      "step": 96570
    },
    {
      "epoch": 27.414135679818337,
      "grad_norm": 0.2573968768119812,
      "learning_rate": 7.259806982685212e-05,
      "loss": 0.002267581783235073,
      "step": 96580
    },
    {
      "epoch": 27.416974169741696,
      "grad_norm": 1.4451062679290771,
      "learning_rate": 7.259523133692876e-05,
      "loss": 0.011018510162830352,
      "step": 96590
    },
    {
      "epoch": 27.41981265966506,
      "grad_norm": 1.4824426174163818,
      "learning_rate": 7.25923928470054e-05,
      "loss": 0.013237187266349792,
      "step": 96600
    },
    {
      "epoch": 27.422651149588418,
      "grad_norm": 4.649725437164307,
      "learning_rate": 7.258955435708204e-05,
      "loss": 0.007501885294914246,
      "step": 96610
    },
    {
      "epoch": 27.42548963951178,
      "grad_norm": 0.8873780369758606,
      "learning_rate": 7.258671586715867e-05,
      "loss": 0.003422384709119797,
      "step": 96620
    },
    {
      "epoch": 27.42832812943514,
      "grad_norm": 1.164281964302063,
      "learning_rate": 7.258387737723531e-05,
      "loss": 0.011814720183610915,
      "step": 96630
    },
    {
      "epoch": 27.4311666193585,
      "grad_norm": 0.29692983627319336,
      "learning_rate": 7.258103888731197e-05,
      "loss": 0.012984825670719147,
      "step": 96640
    },
    {
      "epoch": 27.434005109281863,
      "grad_norm": 0.19270507991313934,
      "learning_rate": 7.257820039738859e-05,
      "loss": 0.0064699709415435794,
      "step": 96650
    },
    {
      "epoch": 27.436843599205222,
      "grad_norm": 1.0714445114135742,
      "learning_rate": 7.257536190746523e-05,
      "loss": 0.005831456929445267,
      "step": 96660
    },
    {
      "epoch": 27.439682089128585,
      "grad_norm": 0.09459691494703293,
      "learning_rate": 7.257252341754188e-05,
      "loss": 0.007966187596321107,
      "step": 96670
    },
    {
      "epoch": 27.442520579051944,
      "grad_norm": 3.170504570007324,
      "learning_rate": 7.25696849276185e-05,
      "loss": 0.009935998916625976,
      "step": 96680
    },
    {
      "epoch": 27.445359068975304,
      "grad_norm": 0.16680097579956055,
      "learning_rate": 7.256684643769514e-05,
      "loss": 0.007530912756919861,
      "step": 96690
    },
    {
      "epoch": 27.448197558898666,
      "grad_norm": 0.6157312393188477,
      "learning_rate": 7.256400794777179e-05,
      "loss": 0.004374706372618675,
      "step": 96700
    },
    {
      "epoch": 27.451036048822026,
      "grad_norm": 1.2958533763885498,
      "learning_rate": 7.256116945784843e-05,
      "loss": 0.0033854272216558455,
      "step": 96710
    },
    {
      "epoch": 27.45387453874539,
      "grad_norm": 3.427262306213379,
      "learning_rate": 7.255833096792507e-05,
      "loss": 0.008452561497688294,
      "step": 96720
    },
    {
      "epoch": 27.456713028668748,
      "grad_norm": 3.8613510131835938,
      "learning_rate": 7.255549247800171e-05,
      "loss": 0.02232562154531479,
      "step": 96730
    },
    {
      "epoch": 27.459551518592107,
      "grad_norm": 0.5082535147666931,
      "learning_rate": 7.255265398807835e-05,
      "loss": 0.012889789044857025,
      "step": 96740
    },
    {
      "epoch": 27.46239000851547,
      "grad_norm": 3.989868640899658,
      "learning_rate": 7.254981549815498e-05,
      "loss": 0.011557345092296601,
      "step": 96750
    },
    {
      "epoch": 27.46522849843883,
      "grad_norm": 4.326202392578125,
      "learning_rate": 7.254697700823162e-05,
      "loss": 0.011441760510206223,
      "step": 96760
    },
    {
      "epoch": 27.468066988362192,
      "grad_norm": 7.229315280914307,
      "learning_rate": 7.254413851830826e-05,
      "loss": 0.013257811963558196,
      "step": 96770
    },
    {
      "epoch": 27.47090547828555,
      "grad_norm": 5.238214492797852,
      "learning_rate": 7.25413000283849e-05,
      "loss": 0.006614356487989426,
      "step": 96780
    },
    {
      "epoch": 27.473743968208915,
      "grad_norm": 1.003938913345337,
      "learning_rate": 7.253846153846155e-05,
      "loss": 0.005573833733797074,
      "step": 96790
    },
    {
      "epoch": 27.476582458132274,
      "grad_norm": 1.3977997303009033,
      "learning_rate": 7.253562304853819e-05,
      "loss": 0.012365708500146866,
      "step": 96800
    },
    {
      "epoch": 27.479420948055633,
      "grad_norm": 0.26060131192207336,
      "learning_rate": 7.253278455861481e-05,
      "loss": 0.01858411878347397,
      "step": 96810
    },
    {
      "epoch": 27.482259437978996,
      "grad_norm": 0.3627377152442932,
      "learning_rate": 7.252994606869146e-05,
      "loss": 0.00517183244228363,
      "step": 96820
    },
    {
      "epoch": 27.485097927902356,
      "grad_norm": 1.1966164112091064,
      "learning_rate": 7.25271075787681e-05,
      "loss": 0.013920573890209198,
      "step": 96830
    },
    {
      "epoch": 27.48793641782572,
      "grad_norm": 0.3464977741241455,
      "learning_rate": 7.252426908884474e-05,
      "loss": 0.013054509460926057,
      "step": 96840
    },
    {
      "epoch": 27.490774907749078,
      "grad_norm": 0.2093365341424942,
      "learning_rate": 7.252143059892138e-05,
      "loss": 0.012960982322692872,
      "step": 96850
    },
    {
      "epoch": 27.493613397672437,
      "grad_norm": 0.6205979585647583,
      "learning_rate": 7.251859210899802e-05,
      "loss": 0.004815002530813217,
      "step": 96860
    },
    {
      "epoch": 27.4964518875958,
      "grad_norm": 0.08560879528522491,
      "learning_rate": 7.251575361907465e-05,
      "loss": 0.00298139713704586,
      "step": 96870
    },
    {
      "epoch": 27.49929037751916,
      "grad_norm": 21.027992248535156,
      "learning_rate": 7.251291512915129e-05,
      "loss": 0.01402672678232193,
      "step": 96880
    },
    {
      "epoch": 27.502128867442522,
      "grad_norm": 6.23898458480835,
      "learning_rate": 7.251007663922793e-05,
      "loss": 0.006502272188663482,
      "step": 96890
    },
    {
      "epoch": 27.50496735736588,
      "grad_norm": 0.41462770104408264,
      "learning_rate": 7.250723814930457e-05,
      "loss": 0.005933268740773201,
      "step": 96900
    },
    {
      "epoch": 27.50780584728924,
      "grad_norm": 9.051358222961426,
      "learning_rate": 7.250439965938121e-05,
      "loss": 0.0063718177378177645,
      "step": 96910
    },
    {
      "epoch": 27.510644337212604,
      "grad_norm": 2.805727481842041,
      "learning_rate": 7.250156116945786e-05,
      "loss": 0.023745051026344298,
      "step": 96920
    },
    {
      "epoch": 27.513482827135963,
      "grad_norm": 0.05529703572392464,
      "learning_rate": 7.24987226795345e-05,
      "loss": 0.006505998969078064,
      "step": 96930
    },
    {
      "epoch": 27.516321317059326,
      "grad_norm": 1.3401886224746704,
      "learning_rate": 7.249588418961112e-05,
      "loss": 0.002583126723766327,
      "step": 96940
    },
    {
      "epoch": 27.519159806982685,
      "grad_norm": 18.591110229492188,
      "learning_rate": 7.249304569968777e-05,
      "loss": 0.007697752863168717,
      "step": 96950
    },
    {
      "epoch": 27.521998296906045,
      "grad_norm": 11.396615028381348,
      "learning_rate": 7.249020720976441e-05,
      "loss": 0.0189493864774704,
      "step": 96960
    },
    {
      "epoch": 27.524836786829407,
      "grad_norm": 0.5214499831199646,
      "learning_rate": 7.248736871984105e-05,
      "loss": 0.011895643919706345,
      "step": 96970
    },
    {
      "epoch": 27.527675276752767,
      "grad_norm": 6.899120807647705,
      "learning_rate": 7.248453022991769e-05,
      "loss": 0.013189561665058136,
      "step": 96980
    },
    {
      "epoch": 27.53051376667613,
      "grad_norm": 0.1787477284669876,
      "learning_rate": 7.248169173999433e-05,
      "loss": 0.008279111981391907,
      "step": 96990
    },
    {
      "epoch": 27.53335225659949,
      "grad_norm": 4.751768112182617,
      "learning_rate": 7.247885325007096e-05,
      "loss": 0.0037852950394153596,
      "step": 97000
    },
    {
      "epoch": 27.53335225659949,
      "eval_accuracy": 0.9696064093596999,
      "eval_loss": 0.10103029012680054,
      "eval_runtime": 31.6708,
      "eval_samples_per_second": 496.577,
      "eval_steps_per_second": 7.767,
      "step": 97000
    },
    {
      "epoch": 27.53619074652285,
      "grad_norm": 18.309608459472656,
      "learning_rate": 7.24760147601476e-05,
      "loss": 0.005369041860103607,
      "step": 97010
    },
    {
      "epoch": 27.53902923644621,
      "grad_norm": 0.3141149580478668,
      "learning_rate": 7.247317627022424e-05,
      "loss": 0.008644135296344757,
      "step": 97020
    },
    {
      "epoch": 27.54186772636957,
      "grad_norm": 1.2588982582092285,
      "learning_rate": 7.247033778030088e-05,
      "loss": 0.009364360570907592,
      "step": 97030
    },
    {
      "epoch": 27.544706216292933,
      "grad_norm": 2.368772268295288,
      "learning_rate": 7.246749929037753e-05,
      "loss": 0.008384592831134796,
      "step": 97040
    },
    {
      "epoch": 27.547544706216293,
      "grad_norm": 0.9589358568191528,
      "learning_rate": 7.246466080045417e-05,
      "loss": 0.015506300330162048,
      "step": 97050
    },
    {
      "epoch": 27.550383196139652,
      "grad_norm": 4.463979244232178,
      "learning_rate": 7.246182231053081e-05,
      "loss": 0.009839032590389252,
      "step": 97060
    },
    {
      "epoch": 27.553221686063015,
      "grad_norm": 1.2548481225967407,
      "learning_rate": 7.245898382060744e-05,
      "loss": 0.008721734583377837,
      "step": 97070
    },
    {
      "epoch": 27.556060175986374,
      "grad_norm": 0.9108835458755493,
      "learning_rate": 7.245614533068408e-05,
      "loss": 0.007576054334640503,
      "step": 97080
    },
    {
      "epoch": 27.558898665909737,
      "grad_norm": 0.2655707597732544,
      "learning_rate": 7.245330684076072e-05,
      "loss": 0.017939600348472595,
      "step": 97090
    },
    {
      "epoch": 27.561737155833097,
      "grad_norm": 3.9697265625,
      "learning_rate": 7.245046835083735e-05,
      "loss": 0.004625979065895081,
      "step": 97100
    },
    {
      "epoch": 27.564575645756456,
      "grad_norm": 15.890666961669922,
      "learning_rate": 7.2447629860914e-05,
      "loss": 0.0073037870228290554,
      "step": 97110
    },
    {
      "epoch": 27.56741413567982,
      "grad_norm": 0.27114689350128174,
      "learning_rate": 7.244479137099064e-05,
      "loss": 0.004322763532400131,
      "step": 97120
    },
    {
      "epoch": 27.570252625603178,
      "grad_norm": 0.1193581074476242,
      "learning_rate": 7.244195288106727e-05,
      "loss": 0.016020789742469788,
      "step": 97130
    },
    {
      "epoch": 27.57309111552654,
      "grad_norm": 3.3606345653533936,
      "learning_rate": 7.243911439114391e-05,
      "loss": 0.005299486964941025,
      "step": 97140
    },
    {
      "epoch": 27.5759296054499,
      "grad_norm": 0.4245653450489044,
      "learning_rate": 7.243627590122055e-05,
      "loss": 0.01874551922082901,
      "step": 97150
    },
    {
      "epoch": 27.578768095373263,
      "grad_norm": 0.6381334662437439,
      "learning_rate": 7.24334374112972e-05,
      "loss": 0.0030078617855906487,
      "step": 97160
    },
    {
      "epoch": 27.581606585296623,
      "grad_norm": 3.3288047313690186,
      "learning_rate": 7.243059892137384e-05,
      "loss": 0.002357511967420578,
      "step": 97170
    },
    {
      "epoch": 27.584445075219982,
      "grad_norm": 0.6359197497367859,
      "learning_rate": 7.242776043145048e-05,
      "loss": 0.006189919263124466,
      "step": 97180
    },
    {
      "epoch": 27.587283565143345,
      "grad_norm": 3.2643861770629883,
      "learning_rate": 7.242492194152712e-05,
      "loss": 0.0044521156698465346,
      "step": 97190
    },
    {
      "epoch": 27.590122055066704,
      "grad_norm": 3.789705276489258,
      "learning_rate": 7.242208345160375e-05,
      "loss": 0.006668069958686828,
      "step": 97200
    },
    {
      "epoch": 27.592960544990063,
      "grad_norm": 0.311458945274353,
      "learning_rate": 7.241924496168039e-05,
      "loss": 0.010240865498781204,
      "step": 97210
    },
    {
      "epoch": 27.595799034913426,
      "grad_norm": 6.262993812561035,
      "learning_rate": 7.241640647175703e-05,
      "loss": 0.01112165004014969,
      "step": 97220
    },
    {
      "epoch": 27.598637524836786,
      "grad_norm": 1.1777467727661133,
      "learning_rate": 7.241356798183366e-05,
      "loss": 0.008434039354324342,
      "step": 97230
    },
    {
      "epoch": 27.60147601476015,
      "grad_norm": 8.886853218078613,
      "learning_rate": 7.241072949191031e-05,
      "loss": 0.009245367348194122,
      "step": 97240
    },
    {
      "epoch": 27.604314504683508,
      "grad_norm": 7.137858867645264,
      "learning_rate": 7.240789100198695e-05,
      "loss": 0.009209196269512176,
      "step": 97250
    },
    {
      "epoch": 27.60715299460687,
      "grad_norm": 1.8496016263961792,
      "learning_rate": 7.240505251206358e-05,
      "loss": 0.008311323821544647,
      "step": 97260
    },
    {
      "epoch": 27.60999148453023,
      "grad_norm": 2.5030462741851807,
      "learning_rate": 7.240221402214022e-05,
      "loss": 0.00514330193400383,
      "step": 97270
    },
    {
      "epoch": 27.61282997445359,
      "grad_norm": 2.34675931930542,
      "learning_rate": 7.239937553221686e-05,
      "loss": 0.007117930054664612,
      "step": 97280
    },
    {
      "epoch": 27.615668464376952,
      "grad_norm": 2.6385257244110107,
      "learning_rate": 7.23965370422935e-05,
      "loss": 0.014711697399616242,
      "step": 97290
    },
    {
      "epoch": 27.61850695430031,
      "grad_norm": 12.356476783752441,
      "learning_rate": 7.239369855237013e-05,
      "loss": 0.01791999936103821,
      "step": 97300
    },
    {
      "epoch": 27.621345444223675,
      "grad_norm": 12.133975982666016,
      "learning_rate": 7.239086006244679e-05,
      "loss": 0.015779873728752135,
      "step": 97310
    },
    {
      "epoch": 27.624183934147034,
      "grad_norm": 8.408624649047852,
      "learning_rate": 7.238802157252343e-05,
      "loss": 0.016854201257228852,
      "step": 97320
    },
    {
      "epoch": 27.627022424070393,
      "grad_norm": 1.9350236654281616,
      "learning_rate": 7.238518308260006e-05,
      "loss": 0.02021499127149582,
      "step": 97330
    },
    {
      "epoch": 27.629860913993756,
      "grad_norm": 0.8486632704734802,
      "learning_rate": 7.23823445926767e-05,
      "loss": 0.009477576613426209,
      "step": 97340
    },
    {
      "epoch": 27.632699403917115,
      "grad_norm": 9.853727340698242,
      "learning_rate": 7.237950610275334e-05,
      "loss": 0.007922462373971938,
      "step": 97350
    },
    {
      "epoch": 27.63553789384048,
      "grad_norm": 13.207910537719727,
      "learning_rate": 7.237666761282997e-05,
      "loss": 0.012919804453849793,
      "step": 97360
    },
    {
      "epoch": 27.638376383763838,
      "grad_norm": 13.077064514160156,
      "learning_rate": 7.237382912290662e-05,
      "loss": 0.025103533267974855,
      "step": 97370
    },
    {
      "epoch": 27.641214873687197,
      "grad_norm": 2.627812385559082,
      "learning_rate": 7.237099063298326e-05,
      "loss": 0.005594475194811821,
      "step": 97380
    },
    {
      "epoch": 27.64405336361056,
      "grad_norm": 0.34933528304100037,
      "learning_rate": 7.236815214305989e-05,
      "loss": 0.009236873686313629,
      "step": 97390
    },
    {
      "epoch": 27.64689185353392,
      "grad_norm": 0.6580194234848022,
      "learning_rate": 7.236531365313653e-05,
      "loss": 0.004521681368350983,
      "step": 97400
    },
    {
      "epoch": 27.649730343457282,
      "grad_norm": 11.513921737670898,
      "learning_rate": 7.236247516321317e-05,
      "loss": 0.007708260416984558,
      "step": 97410
    },
    {
      "epoch": 27.65256883338064,
      "grad_norm": 5.798037528991699,
      "learning_rate": 7.235963667328982e-05,
      "loss": 0.020144805312156677,
      "step": 97420
    },
    {
      "epoch": 27.655407323304,
      "grad_norm": 5.320600986480713,
      "learning_rate": 7.235679818336644e-05,
      "loss": 0.006030211597681046,
      "step": 97430
    },
    {
      "epoch": 27.658245813227364,
      "grad_norm": 4.088475704193115,
      "learning_rate": 7.23539596934431e-05,
      "loss": 0.014188669621944427,
      "step": 97440
    },
    {
      "epoch": 27.661084303150723,
      "grad_norm": 0.15569989383220673,
      "learning_rate": 7.235112120351974e-05,
      "loss": 0.017789986729621888,
      "step": 97450
    },
    {
      "epoch": 27.663922793074086,
      "grad_norm": 3.991321086883545,
      "learning_rate": 7.234828271359637e-05,
      "loss": 0.007727222144603729,
      "step": 97460
    },
    {
      "epoch": 27.666761282997445,
      "grad_norm": 0.4398612678050995,
      "learning_rate": 7.234544422367301e-05,
      "loss": 0.028758913278579712,
      "step": 97470
    },
    {
      "epoch": 27.669599772920805,
      "grad_norm": 0.6409338116645813,
      "learning_rate": 7.234260573374965e-05,
      "loss": 0.00842859074473381,
      "step": 97480
    },
    {
      "epoch": 27.672438262844167,
      "grad_norm": 3.174285650253296,
      "learning_rate": 7.233976724382628e-05,
      "loss": 0.012332449853420257,
      "step": 97490
    },
    {
      "epoch": 27.675276752767527,
      "grad_norm": 10.691227912902832,
      "learning_rate": 7.233692875390292e-05,
      "loss": 0.02360384613275528,
      "step": 97500
    },
    {
      "epoch": 27.675276752767527,
      "eval_accuracy": 0.9672537674063713,
      "eval_loss": 0.10689795017242432,
      "eval_runtime": 31.9797,
      "eval_samples_per_second": 491.78,
      "eval_steps_per_second": 7.692,
      "step": 97500
    },
    {
      "epoch": 27.67811524269089,
      "grad_norm": 2.829873561859131,
      "learning_rate": 7.233409026397957e-05,
      "loss": 0.030706974864006042,
      "step": 97510
    },
    {
      "epoch": 27.68095373261425,
      "grad_norm": 0.17624731361865997,
      "learning_rate": 7.23312517740562e-05,
      "loss": 0.003634941950440407,
      "step": 97520
    },
    {
      "epoch": 27.68379222253761,
      "grad_norm": 9.34074592590332,
      "learning_rate": 7.232841328413284e-05,
      "loss": 0.021944473683834075,
      "step": 97530
    },
    {
      "epoch": 27.68663071246097,
      "grad_norm": 0.44940418004989624,
      "learning_rate": 7.232557479420949e-05,
      "loss": 0.013528761267662049,
      "step": 97540
    },
    {
      "epoch": 27.68946920238433,
      "grad_norm": 0.11034413427114487,
      "learning_rate": 7.232273630428613e-05,
      "loss": 0.009645035117864608,
      "step": 97550
    },
    {
      "epoch": 27.692307692307693,
      "grad_norm": 4.107700347900391,
      "learning_rate": 7.231989781436275e-05,
      "loss": 0.016584350168704985,
      "step": 97560
    },
    {
      "epoch": 27.695146182231053,
      "grad_norm": 3.4550509452819824,
      "learning_rate": 7.231705932443941e-05,
      "loss": 0.00828484669327736,
      "step": 97570
    },
    {
      "epoch": 27.697984672154412,
      "grad_norm": 13.39214038848877,
      "learning_rate": 7.231422083451605e-05,
      "loss": 0.009620623290538787,
      "step": 97580
    },
    {
      "epoch": 27.700823162077775,
      "grad_norm": 2.5609512329101562,
      "learning_rate": 7.231138234459268e-05,
      "loss": 0.005959572270512581,
      "step": 97590
    },
    {
      "epoch": 27.703661652001134,
      "grad_norm": 4.887593746185303,
      "learning_rate": 7.230854385466932e-05,
      "loss": 0.013908010721206666,
      "step": 97600
    },
    {
      "epoch": 27.706500141924497,
      "grad_norm": 1.9685144424438477,
      "learning_rate": 7.230570536474596e-05,
      "loss": 0.005428294837474823,
      "step": 97610
    },
    {
      "epoch": 27.709338631847857,
      "grad_norm": 6.592779636383057,
      "learning_rate": 7.230286687482259e-05,
      "loss": 0.006298057734966278,
      "step": 97620
    },
    {
      "epoch": 27.71217712177122,
      "grad_norm": 2.8147764205932617,
      "learning_rate": 7.230002838489923e-05,
      "loss": 0.007245076447725296,
      "step": 97630
    },
    {
      "epoch": 27.71501561169458,
      "grad_norm": 12.905450820922852,
      "learning_rate": 7.229718989497589e-05,
      "loss": 0.019087037444114684,
      "step": 97640
    },
    {
      "epoch": 27.717854101617938,
      "grad_norm": 3.3887698650360107,
      "learning_rate": 7.229435140505251e-05,
      "loss": 0.01091318279504776,
      "step": 97650
    },
    {
      "epoch": 27.7206925915413,
      "grad_norm": 8.123041152954102,
      "learning_rate": 7.229151291512915e-05,
      "loss": 0.006569728255271912,
      "step": 97660
    },
    {
      "epoch": 27.72353108146466,
      "grad_norm": 0.401538610458374,
      "learning_rate": 7.22886744252058e-05,
      "loss": 0.005447173863649369,
      "step": 97670
    },
    {
      "epoch": 27.726369571388023,
      "grad_norm": 1.5301367044448853,
      "learning_rate": 7.228583593528244e-05,
      "loss": 0.004692605882883072,
      "step": 97680
    },
    {
      "epoch": 27.729208061311382,
      "grad_norm": 1.930240273475647,
      "learning_rate": 7.228299744535907e-05,
      "loss": 0.013348780572414398,
      "step": 97690
    },
    {
      "epoch": 27.732046551234742,
      "grad_norm": 3.3197882175445557,
      "learning_rate": 7.228015895543571e-05,
      "loss": 0.011002518236637115,
      "step": 97700
    },
    {
      "epoch": 27.734885041158105,
      "grad_norm": 1.3957546949386597,
      "learning_rate": 7.227732046551235e-05,
      "loss": 0.008149507641792297,
      "step": 97710
    },
    {
      "epoch": 27.737723531081464,
      "grad_norm": 1.0319628715515137,
      "learning_rate": 7.227448197558899e-05,
      "loss": 0.0028079690411686896,
      "step": 97720
    },
    {
      "epoch": 27.740562021004827,
      "grad_norm": 0.13723233342170715,
      "learning_rate": 7.227164348566563e-05,
      "loss": 0.004231739044189453,
      "step": 97730
    },
    {
      "epoch": 27.743400510928186,
      "grad_norm": 1.7648617029190063,
      "learning_rate": 7.226880499574227e-05,
      "loss": 0.0059420645236969,
      "step": 97740
    },
    {
      "epoch": 27.746239000851546,
      "grad_norm": 0.7952282428741455,
      "learning_rate": 7.22659665058189e-05,
      "loss": 0.00470990426838398,
      "step": 97750
    },
    {
      "epoch": 27.74907749077491,
      "grad_norm": 0.6019049286842346,
      "learning_rate": 7.226312801589554e-05,
      "loss": 0.005932306498289108,
      "step": 97760
    },
    {
      "epoch": 27.751915980698268,
      "grad_norm": 8.901731491088867,
      "learning_rate": 7.22602895259722e-05,
      "loss": 0.007958200573921204,
      "step": 97770
    },
    {
      "epoch": 27.75475447062163,
      "grad_norm": 1.9121177196502686,
      "learning_rate": 7.225745103604882e-05,
      "loss": 0.006304142624139785,
      "step": 97780
    },
    {
      "epoch": 27.75759296054499,
      "grad_norm": 1.8885475397109985,
      "learning_rate": 7.225461254612547e-05,
      "loss": 0.00800783783197403,
      "step": 97790
    },
    {
      "epoch": 27.76043145046835,
      "grad_norm": 0.6312928795814514,
      "learning_rate": 7.225177405620211e-05,
      "loss": 0.013156653940677642,
      "step": 97800
    },
    {
      "epoch": 27.763269940391712,
      "grad_norm": 0.661371648311615,
      "learning_rate": 7.224893556627873e-05,
      "loss": 0.013505423069000244,
      "step": 97810
    },
    {
      "epoch": 27.76610843031507,
      "grad_norm": 2.2893810272216797,
      "learning_rate": 7.224609707635538e-05,
      "loss": 0.005285640805959701,
      "step": 97820
    },
    {
      "epoch": 27.768946920238434,
      "grad_norm": 0.1757335662841797,
      "learning_rate": 7.224325858643202e-05,
      "loss": 0.0034962423145771027,
      "step": 97830
    },
    {
      "epoch": 27.771785410161794,
      "grad_norm": 11.619396209716797,
      "learning_rate": 7.224042009650866e-05,
      "loss": 0.009888204932212829,
      "step": 97840
    },
    {
      "epoch": 27.774623900085153,
      "grad_norm": 11.598549842834473,
      "learning_rate": 7.22375816065853e-05,
      "loss": 0.007115699350833893,
      "step": 97850
    },
    {
      "epoch": 27.777462390008516,
      "grad_norm": 3.7941131591796875,
      "learning_rate": 7.223474311666194e-05,
      "loss": 0.005165163800120354,
      "step": 97860
    },
    {
      "epoch": 27.780300879931875,
      "grad_norm": 9.735224723815918,
      "learning_rate": 7.223190462673858e-05,
      "loss": 0.009645286202430724,
      "step": 97870
    },
    {
      "epoch": 27.78313936985524,
      "grad_norm": 4.729522228240967,
      "learning_rate": 7.222906613681521e-05,
      "loss": 0.01726793944835663,
      "step": 97880
    },
    {
      "epoch": 27.785977859778598,
      "grad_norm": 1.9284528493881226,
      "learning_rate": 7.222622764689185e-05,
      "loss": 0.014305844902992249,
      "step": 97890
    },
    {
      "epoch": 27.788816349701957,
      "grad_norm": 0.05202716961503029,
      "learning_rate": 7.222338915696851e-05,
      "loss": 0.010578495264053345,
      "step": 97900
    },
    {
      "epoch": 27.79165483962532,
      "grad_norm": 0.9352652430534363,
      "learning_rate": 7.222055066704514e-05,
      "loss": 0.010015155375003814,
      "step": 97910
    },
    {
      "epoch": 27.79449332954868,
      "grad_norm": 5.734329700469971,
      "learning_rate": 7.221771217712178e-05,
      "loss": 0.008656066656112672,
      "step": 97920
    },
    {
      "epoch": 27.797331819472042,
      "grad_norm": 1.7903918027877808,
      "learning_rate": 7.221487368719842e-05,
      "loss": 0.012787604331970214,
      "step": 97930
    },
    {
      "epoch": 27.8001703093954,
      "grad_norm": 3.4531805515289307,
      "learning_rate": 7.221203519727505e-05,
      "loss": 0.028377822041511534,
      "step": 97940
    },
    {
      "epoch": 27.80300879931876,
      "grad_norm": 0.14791043102741241,
      "learning_rate": 7.220919670735169e-05,
      "loss": 0.008879422396421432,
      "step": 97950
    },
    {
      "epoch": 27.805847289242124,
      "grad_norm": 0.09765651077032089,
      "learning_rate": 7.220635821742833e-05,
      "loss": 0.02921227514743805,
      "step": 97960
    },
    {
      "epoch": 27.808685779165483,
      "grad_norm": 10.292367935180664,
      "learning_rate": 7.220351972750497e-05,
      "loss": 0.013348039984703065,
      "step": 97970
    },
    {
      "epoch": 27.811524269088846,
      "grad_norm": 2.124948024749756,
      "learning_rate": 7.220068123758161e-05,
      "loss": 0.009312741458415985,
      "step": 97980
    },
    {
      "epoch": 27.814362759012205,
      "grad_norm": 3.3016645908355713,
      "learning_rate": 7.219784274765825e-05,
      "loss": 0.008218172937631607,
      "step": 97990
    },
    {
      "epoch": 27.817201248935568,
      "grad_norm": 0.9870637059211731,
      "learning_rate": 7.21950042577349e-05,
      "loss": 0.008779406547546387,
      "step": 98000
    },
    {
      "epoch": 27.817201248935568,
      "eval_accuracy": 0.9703058434539328,
      "eval_loss": 0.10532072931528091,
      "eval_runtime": 31.3547,
      "eval_samples_per_second": 501.583,
      "eval_steps_per_second": 7.846,
      "step": 98000
    },
    {
      "epoch": 27.820039738858927,
      "grad_norm": 2.913940191268921,
      "learning_rate": 7.219216576781152e-05,
      "loss": 0.010377797484397887,
      "step": 98010
    },
    {
      "epoch": 27.822878228782287,
      "grad_norm": 1.9477972984313965,
      "learning_rate": 7.218932727788816e-05,
      "loss": 0.011614929884672165,
      "step": 98020
    },
    {
      "epoch": 27.82571671870565,
      "grad_norm": 0.7664371132850647,
      "learning_rate": 7.21864887879648e-05,
      "loss": 0.011967381834983826,
      "step": 98030
    },
    {
      "epoch": 27.82855520862901,
      "grad_norm": 0.37905406951904297,
      "learning_rate": 7.218365029804145e-05,
      "loss": 0.014957410097122193,
      "step": 98040
    },
    {
      "epoch": 27.83139369855237,
      "grad_norm": 0.3093266785144806,
      "learning_rate": 7.218081180811809e-05,
      "loss": 0.004982839524745941,
      "step": 98050
    },
    {
      "epoch": 27.83423218847573,
      "grad_norm": 0.48155394196510315,
      "learning_rate": 7.217797331819473e-05,
      "loss": 0.007398340106010437,
      "step": 98060
    },
    {
      "epoch": 27.83707067839909,
      "grad_norm": 2.741950511932373,
      "learning_rate": 7.217513482827136e-05,
      "loss": 0.01253604143857956,
      "step": 98070
    },
    {
      "epoch": 27.839909168322453,
      "grad_norm": 13.374955177307129,
      "learning_rate": 7.2172296338348e-05,
      "loss": 0.01176898181438446,
      "step": 98080
    },
    {
      "epoch": 27.842747658245813,
      "grad_norm": 2.3230221271514893,
      "learning_rate": 7.216945784842464e-05,
      "loss": 0.008635807037353515,
      "step": 98090
    },
    {
      "epoch": 27.845586148169176,
      "grad_norm": 4.1186676025390625,
      "learning_rate": 7.216661935850128e-05,
      "loss": 0.004837186262011528,
      "step": 98100
    },
    {
      "epoch": 27.848424638092535,
      "grad_norm": 4.813791275024414,
      "learning_rate": 7.216378086857792e-05,
      "loss": 0.03244186639785766,
      "step": 98110
    },
    {
      "epoch": 27.851263128015894,
      "grad_norm": 4.337273120880127,
      "learning_rate": 7.216094237865456e-05,
      "loss": 0.00572916567325592,
      "step": 98120
    },
    {
      "epoch": 27.854101617939257,
      "grad_norm": 1.7730696201324463,
      "learning_rate": 7.21581038887312e-05,
      "loss": 0.011466015875339509,
      "step": 98130
    },
    {
      "epoch": 27.856940107862616,
      "grad_norm": 0.16428756713867188,
      "learning_rate": 7.215526539880783e-05,
      "loss": 0.015670016407966614,
      "step": 98140
    },
    {
      "epoch": 27.85977859778598,
      "grad_norm": 0.12462593615055084,
      "learning_rate": 7.215242690888447e-05,
      "loss": 0.013828511536121368,
      "step": 98150
    },
    {
      "epoch": 27.86261708770934,
      "grad_norm": 0.19113485515117645,
      "learning_rate": 7.214958841896112e-05,
      "loss": 0.0016321105882525443,
      "step": 98160
    },
    {
      "epoch": 27.865455577632698,
      "grad_norm": 8.633868217468262,
      "learning_rate": 7.214674992903776e-05,
      "loss": 0.012891767919063568,
      "step": 98170
    },
    {
      "epoch": 27.86829406755606,
      "grad_norm": 3.787313461303711,
      "learning_rate": 7.21439114391144e-05,
      "loss": 0.01167978197336197,
      "step": 98180
    },
    {
      "epoch": 27.87113255747942,
      "grad_norm": 1.3536145687103271,
      "learning_rate": 7.214107294919104e-05,
      "loss": 0.0035085935145616532,
      "step": 98190
    },
    {
      "epoch": 27.873971047402783,
      "grad_norm": 13.450508117675781,
      "learning_rate": 7.213823445926767e-05,
      "loss": 0.008535811305046081,
      "step": 98200
    },
    {
      "epoch": 27.876809537326142,
      "grad_norm": 1.455994725227356,
      "learning_rate": 7.213539596934431e-05,
      "loss": 0.013846945762634278,
      "step": 98210
    },
    {
      "epoch": 27.8796480272495,
      "grad_norm": 6.51689338684082,
      "learning_rate": 7.213255747942095e-05,
      "loss": 0.005934932082891464,
      "step": 98220
    },
    {
      "epoch": 27.882486517172865,
      "grad_norm": 0.15285904705524445,
      "learning_rate": 7.212971898949759e-05,
      "loss": 0.015312400460243226,
      "step": 98230
    },
    {
      "epoch": 27.885325007096224,
      "grad_norm": 0.06120777502655983,
      "learning_rate": 7.212688049957423e-05,
      "loss": 0.01133604347705841,
      "step": 98240
    },
    {
      "epoch": 27.888163497019587,
      "grad_norm": 0.14342893660068512,
      "learning_rate": 7.212404200965087e-05,
      "loss": 0.008290559798479081,
      "step": 98250
    },
    {
      "epoch": 27.891001986942946,
      "grad_norm": 0.7989756464958191,
      "learning_rate": 7.212120351972752e-05,
      "loss": 0.009576807916164397,
      "step": 98260
    },
    {
      "epoch": 27.893840476866306,
      "grad_norm": 4.805694103240967,
      "learning_rate": 7.211836502980414e-05,
      "loss": 0.009475824981927871,
      "step": 98270
    },
    {
      "epoch": 27.89667896678967,
      "grad_norm": 3.4726603031158447,
      "learning_rate": 7.211552653988078e-05,
      "loss": 0.013142059743404388,
      "step": 98280
    },
    {
      "epoch": 27.899517456713028,
      "grad_norm": 6.406253337860107,
      "learning_rate": 7.211268804995743e-05,
      "loss": 0.017322225868701933,
      "step": 98290
    },
    {
      "epoch": 27.90235594663639,
      "grad_norm": 0.1765398532152176,
      "learning_rate": 7.210984956003407e-05,
      "loss": 0.01699725687503815,
      "step": 98300
    },
    {
      "epoch": 27.90519443655975,
      "grad_norm": 1.9625648260116577,
      "learning_rate": 7.210701107011071e-05,
      "loss": 0.00621083602309227,
      "step": 98310
    },
    {
      "epoch": 27.90803292648311,
      "grad_norm": 2.850149631500244,
      "learning_rate": 7.210417258018735e-05,
      "loss": 0.010471602529287338,
      "step": 98320
    },
    {
      "epoch": 27.910871416406472,
      "grad_norm": 9.402115821838379,
      "learning_rate": 7.210133409026398e-05,
      "loss": 0.020138588547706605,
      "step": 98330
    },
    {
      "epoch": 27.91370990632983,
      "grad_norm": 0.6400034427642822,
      "learning_rate": 7.209849560034062e-05,
      "loss": 0.011958037316799165,
      "step": 98340
    },
    {
      "epoch": 27.916548396253194,
      "grad_norm": 0.5266924500465393,
      "learning_rate": 7.209565711041726e-05,
      "loss": 0.005979631841182709,
      "step": 98350
    },
    {
      "epoch": 27.919386886176554,
      "grad_norm": 0.10253424197435379,
      "learning_rate": 7.20928186204939e-05,
      "loss": 0.022770293056964874,
      "step": 98360
    },
    {
      "epoch": 27.922225376099917,
      "grad_norm": 11.751627922058105,
      "learning_rate": 7.208998013057054e-05,
      "loss": 0.01326298713684082,
      "step": 98370
    },
    {
      "epoch": 27.925063866023276,
      "grad_norm": 12.277976989746094,
      "learning_rate": 7.208714164064718e-05,
      "loss": 0.018603825569152833,
      "step": 98380
    },
    {
      "epoch": 27.927902355946635,
      "grad_norm": 0.29820170998573303,
      "learning_rate": 7.208430315072383e-05,
      "loss": 0.009783060103654862,
      "step": 98390
    },
    {
      "epoch": 27.930740845869998,
      "grad_norm": 0.40491244196891785,
      "learning_rate": 7.208146466080045e-05,
      "loss": 0.00326567143201828,
      "step": 98400
    },
    {
      "epoch": 27.933579335793358,
      "grad_norm": 0.34837955236434937,
      "learning_rate": 7.20786261708771e-05,
      "loss": 0.006888265907764435,
      "step": 98410
    },
    {
      "epoch": 27.93641782571672,
      "grad_norm": 0.22316290438175201,
      "learning_rate": 7.207578768095374e-05,
      "loss": 0.008432324230670928,
      "step": 98420
    },
    {
      "epoch": 27.93925631564008,
      "grad_norm": 11.993837356567383,
      "learning_rate": 7.207294919103036e-05,
      "loss": 0.007068382948637009,
      "step": 98430
    },
    {
      "epoch": 27.94209480556344,
      "grad_norm": 0.780241072177887,
      "learning_rate": 7.207011070110702e-05,
      "loss": 0.001531766913831234,
      "step": 98440
    },
    {
      "epoch": 27.944933295486802,
      "grad_norm": 0.7169052958488464,
      "learning_rate": 7.206727221118366e-05,
      "loss": 0.004088509827852249,
      "step": 98450
    },
    {
      "epoch": 27.94777178541016,
      "grad_norm": 9.375650405883789,
      "learning_rate": 7.206443372126029e-05,
      "loss": 0.009386953711509705,
      "step": 98460
    },
    {
      "epoch": 27.950610275333524,
      "grad_norm": 0.7772210836410522,
      "learning_rate": 7.206159523133693e-05,
      "loss": 0.004189445823431015,
      "step": 98470
    },
    {
      "epoch": 27.953448765256883,
      "grad_norm": 1.6636050939559937,
      "learning_rate": 7.205875674141357e-05,
      "loss": 0.0035268016159534453,
      "step": 98480
    },
    {
      "epoch": 27.956287255180243,
      "grad_norm": 3.341454029083252,
      "learning_rate": 7.205591825149021e-05,
      "loss": 0.009532211720943451,
      "step": 98490
    },
    {
      "epoch": 27.959125745103606,
      "grad_norm": 0.30353203415870667,
      "learning_rate": 7.205307976156685e-05,
      "loss": 0.009427633136510849,
      "step": 98500
    },
    {
      "epoch": 27.959125745103606,
      "eval_accuracy": 0.9717682965600559,
      "eval_loss": 0.09503769874572754,
      "eval_runtime": 32.6519,
      "eval_samples_per_second": 481.657,
      "eval_steps_per_second": 7.534,
      "step": 98500
    },
    {
      "epoch": 27.961964235026965,
      "grad_norm": 12.016179084777832,
      "learning_rate": 7.20502412716435e-05,
      "loss": 0.006822566688060761,
      "step": 98510
    },
    {
      "epoch": 27.964802724950328,
      "grad_norm": 17.09442901611328,
      "learning_rate": 7.204740278172014e-05,
      "loss": 0.012226901203393935,
      "step": 98520
    },
    {
      "epoch": 27.967641214873687,
      "grad_norm": 6.03265380859375,
      "learning_rate": 7.204456429179676e-05,
      "loss": 0.0039820164442062374,
      "step": 98530
    },
    {
      "epoch": 27.970479704797047,
      "grad_norm": 0.15899983048439026,
      "learning_rate": 7.20417258018734e-05,
      "loss": 0.004237930104136467,
      "step": 98540
    },
    {
      "epoch": 27.97331819472041,
      "grad_norm": 5.050915718078613,
      "learning_rate": 7.203888731195005e-05,
      "loss": 0.005622537806630135,
      "step": 98550
    },
    {
      "epoch": 27.97615668464377,
      "grad_norm": 12.573478698730469,
      "learning_rate": 7.203604882202668e-05,
      "loss": 0.009363728761672973,
      "step": 98560
    },
    {
      "epoch": 27.97899517456713,
      "grad_norm": 2.065699577331543,
      "learning_rate": 7.203321033210333e-05,
      "loss": 0.013362047076225281,
      "step": 98570
    },
    {
      "epoch": 27.98183366449049,
      "grad_norm": 1.4128072261810303,
      "learning_rate": 7.203037184217997e-05,
      "loss": 0.011735916137695312,
      "step": 98580
    },
    {
      "epoch": 27.98467215441385,
      "grad_norm": 0.9049136638641357,
      "learning_rate": 7.20275333522566e-05,
      "loss": 0.008742931485176086,
      "step": 98590
    },
    {
      "epoch": 27.987510644337213,
      "grad_norm": 5.3398118019104,
      "learning_rate": 7.202469486233324e-05,
      "loss": 0.012394602596759795,
      "step": 98600
    },
    {
      "epoch": 27.990349134260573,
      "grad_norm": 0.850437581539154,
      "learning_rate": 7.202185637240988e-05,
      "loss": 0.013493858277797699,
      "step": 98610
    },
    {
      "epoch": 27.993187624183935,
      "grad_norm": 0.22714051604270935,
      "learning_rate": 7.201901788248652e-05,
      "loss": 0.011869265139102936,
      "step": 98620
    },
    {
      "epoch": 27.996026114107295,
      "grad_norm": 1.0199172496795654,
      "learning_rate": 7.201617939256315e-05,
      "loss": 0.013123774528503418,
      "step": 98630
    },
    {
      "epoch": 27.998864604030654,
      "grad_norm": 12.334456443786621,
      "learning_rate": 7.20133409026398e-05,
      "loss": 0.017952007055282593,
      "step": 98640
    },
    {
      "epoch": 28.001703093954017,
      "grad_norm": 7.467791557312012,
      "learning_rate": 7.201050241271643e-05,
      "loss": 0.006950414925813675,
      "step": 98650
    },
    {
      "epoch": 28.004541583877376,
      "grad_norm": 0.29806914925575256,
      "learning_rate": 7.200766392279308e-05,
      "loss": 0.0026668280363082884,
      "step": 98660
    },
    {
      "epoch": 28.00738007380074,
      "grad_norm": 8.481799125671387,
      "learning_rate": 7.200482543286972e-05,
      "loss": 0.007975976169109344,
      "step": 98670
    },
    {
      "epoch": 28.0102185637241,
      "grad_norm": 1.0360645055770874,
      "learning_rate": 7.200198694294636e-05,
      "loss": 0.003578939288854599,
      "step": 98680
    },
    {
      "epoch": 28.013057053647458,
      "grad_norm": 0.4537346363067627,
      "learning_rate": 7.199914845302299e-05,
      "loss": 0.01181703433394432,
      "step": 98690
    },
    {
      "epoch": 28.01589554357082,
      "grad_norm": 4.873838424682617,
      "learning_rate": 7.199630996309964e-05,
      "loss": 0.0031110072508454324,
      "step": 98700
    },
    {
      "epoch": 28.01873403349418,
      "grad_norm": 0.26027458906173706,
      "learning_rate": 7.199347147317628e-05,
      "loss": 0.0170804962515831,
      "step": 98710
    },
    {
      "epoch": 28.021572523417543,
      "grad_norm": 4.413666248321533,
      "learning_rate": 7.199063298325291e-05,
      "loss": 0.0014163525775074959,
      "step": 98720
    },
    {
      "epoch": 28.024411013340902,
      "grad_norm": 1.3233695030212402,
      "learning_rate": 7.198779449332955e-05,
      "loss": 0.007451024651527405,
      "step": 98730
    },
    {
      "epoch": 28.02724950326426,
      "grad_norm": 2.9688849449157715,
      "learning_rate": 7.198495600340619e-05,
      "loss": 0.006008873134851456,
      "step": 98740
    },
    {
      "epoch": 28.030087993187625,
      "grad_norm": 0.539840817451477,
      "learning_rate": 7.198211751348282e-05,
      "loss": 0.009261567890644074,
      "step": 98750
    },
    {
      "epoch": 28.032926483110984,
      "grad_norm": 0.17760953307151794,
      "learning_rate": 7.197927902355946e-05,
      "loss": 0.010997213423252106,
      "step": 98760
    },
    {
      "epoch": 28.035764973034347,
      "grad_norm": 1.8217352628707886,
      "learning_rate": 7.197644053363612e-05,
      "loss": 0.004322588071227073,
      "step": 98770
    },
    {
      "epoch": 28.038603462957706,
      "grad_norm": 9.33586311340332,
      "learning_rate": 7.197360204371274e-05,
      "loss": 0.004479963704943657,
      "step": 98780
    },
    {
      "epoch": 28.04144195288107,
      "grad_norm": 0.9295791387557983,
      "learning_rate": 7.197076355378939e-05,
      "loss": 0.00568314716219902,
      "step": 98790
    },
    {
      "epoch": 28.04428044280443,
      "grad_norm": 15.011871337890625,
      "learning_rate": 7.196792506386603e-05,
      "loss": 0.016332793235778808,
      "step": 98800
    },
    {
      "epoch": 28.047118932727788,
      "grad_norm": 2.3577892780303955,
      "learning_rate": 7.196508657394267e-05,
      "loss": 0.001977499388158321,
      "step": 98810
    },
    {
      "epoch": 28.04995742265115,
      "grad_norm": 1.93413507938385,
      "learning_rate": 7.19622480840193e-05,
      "loss": 0.004534290730953216,
      "step": 98820
    },
    {
      "epoch": 28.05279591257451,
      "grad_norm": 2.9221365451812744,
      "learning_rate": 7.195940959409594e-05,
      "loss": 0.0015928322449326516,
      "step": 98830
    },
    {
      "epoch": 28.055634402497873,
      "grad_norm": 1.1011254787445068,
      "learning_rate": 7.195657110417259e-05,
      "loss": 0.004823616147041321,
      "step": 98840
    },
    {
      "epoch": 28.058472892421232,
      "grad_norm": 0.6268956661224365,
      "learning_rate": 7.195373261424922e-05,
      "loss": 0.0031440861523151396,
      "step": 98850
    },
    {
      "epoch": 28.06131138234459,
      "grad_norm": 0.30382683873176575,
      "learning_rate": 7.195089412432586e-05,
      "loss": 0.009927687793970108,
      "step": 98860
    },
    {
      "epoch": 28.064149872267954,
      "grad_norm": 14.53017807006836,
      "learning_rate": 7.19480556344025e-05,
      "loss": 0.012180811911821365,
      "step": 98870
    },
    {
      "epoch": 28.066988362191314,
      "grad_norm": 2.0771942138671875,
      "learning_rate": 7.194521714447913e-05,
      "loss": 0.013554766774177551,
      "step": 98880
    },
    {
      "epoch": 28.069826852114677,
      "grad_norm": 2.0917203426361084,
      "learning_rate": 7.194237865455577e-05,
      "loss": 0.0029291097074747086,
      "step": 98890
    },
    {
      "epoch": 28.072665342038036,
      "grad_norm": 6.1794233322143555,
      "learning_rate": 7.193954016463243e-05,
      "loss": 0.005445221066474914,
      "step": 98900
    },
    {
      "epoch": 28.075503831961395,
      "grad_norm": 2.51206111907959,
      "learning_rate": 7.193670167470906e-05,
      "loss": 0.0028542133048176767,
      "step": 98910
    },
    {
      "epoch": 28.078342321884758,
      "grad_norm": 0.158002108335495,
      "learning_rate": 7.19338631847857e-05,
      "loss": 0.005274620652198791,
      "step": 98920
    },
    {
      "epoch": 28.081180811808117,
      "grad_norm": 1.1941492557525635,
      "learning_rate": 7.193102469486234e-05,
      "loss": 0.0024107675999403,
      "step": 98930
    },
    {
      "epoch": 28.08401930173148,
      "grad_norm": 0.46413490176200867,
      "learning_rate": 7.192818620493898e-05,
      "loss": 0.01038382202386856,
      "step": 98940
    },
    {
      "epoch": 28.08685779165484,
      "grad_norm": 0.3940821886062622,
      "learning_rate": 7.192534771501561e-05,
      "loss": 0.0020679386332631113,
      "step": 98950
    },
    {
      "epoch": 28.0896962815782,
      "grad_norm": 0.22385403513908386,
      "learning_rate": 7.192250922509225e-05,
      "loss": 0.005021730810403824,
      "step": 98960
    },
    {
      "epoch": 28.092534771501562,
      "grad_norm": 11.810433387756348,
      "learning_rate": 7.19196707351689e-05,
      "loss": 0.03277445435523987,
      "step": 98970
    },
    {
      "epoch": 28.09537326142492,
      "grad_norm": 2.3139777183532715,
      "learning_rate": 7.191683224524553e-05,
      "loss": 0.011056835949420928,
      "step": 98980
    },
    {
      "epoch": 28.098211751348284,
      "grad_norm": 3.5697226524353027,
      "learning_rate": 7.191399375532217e-05,
      "loss": 0.008641229569911956,
      "step": 98990
    },
    {
      "epoch": 28.101050241271643,
      "grad_norm": 0.3516702651977539,
      "learning_rate": 7.191115526539881e-05,
      "loss": 0.004821983352303505,
      "step": 99000
    },
    {
      "epoch": 28.101050241271643,
      "eval_accuracy": 0.9721498060660011,
      "eval_loss": 0.09106661379337311,
      "eval_runtime": 31.6125,
      "eval_samples_per_second": 497.492,
      "eval_steps_per_second": 7.782,
      "step": 99000
    },
    {
      "epoch": 28.103888731195003,
      "grad_norm": 0.166743665933609,
      "learning_rate": 7.190831677547544e-05,
      "loss": 0.00967477709054947,
      "step": 99010
    },
    {
      "epoch": 28.106727221118366,
      "grad_norm": 0.21311303973197937,
      "learning_rate": 7.190547828555208e-05,
      "loss": 0.0070309355854988095,
      "step": 99020
    },
    {
      "epoch": 28.109565711041725,
      "grad_norm": 0.09932377189397812,
      "learning_rate": 7.190263979562873e-05,
      "loss": 0.004808880388736725,
      "step": 99030
    },
    {
      "epoch": 28.112404200965088,
      "grad_norm": 0.29097533226013184,
      "learning_rate": 7.189980130570537e-05,
      "loss": 0.010361646115779877,
      "step": 99040
    },
    {
      "epoch": 28.115242690888447,
      "grad_norm": 0.4040045440196991,
      "learning_rate": 7.189696281578201e-05,
      "loss": 0.009962106496095658,
      "step": 99050
    },
    {
      "epoch": 28.118081180811807,
      "grad_norm": 6.014442443847656,
      "learning_rate": 7.189412432585865e-05,
      "loss": 0.005878487229347229,
      "step": 99060
    },
    {
      "epoch": 28.12091967073517,
      "grad_norm": 0.35094133019447327,
      "learning_rate": 7.189128583593529e-05,
      "loss": 0.013683578372001648,
      "step": 99070
    },
    {
      "epoch": 28.12375816065853,
      "grad_norm": 8.667387962341309,
      "learning_rate": 7.188844734601192e-05,
      "loss": 0.005853508412837982,
      "step": 99080
    },
    {
      "epoch": 28.12659665058189,
      "grad_norm": 11.186172485351562,
      "learning_rate": 7.188560885608856e-05,
      "loss": 0.013310211896896362,
      "step": 99090
    },
    {
      "epoch": 28.12943514050525,
      "grad_norm": 0.38267460465431213,
      "learning_rate": 7.188277036616521e-05,
      "loss": 0.0037995431572198867,
      "step": 99100
    },
    {
      "epoch": 28.13227363042861,
      "grad_norm": 0.9870344400405884,
      "learning_rate": 7.187993187624184e-05,
      "loss": 0.0035861700773239134,
      "step": 99110
    },
    {
      "epoch": 28.135112120351973,
      "grad_norm": 11.831226348876953,
      "learning_rate": 7.187709338631848e-05,
      "loss": 0.01008450835943222,
      "step": 99120
    },
    {
      "epoch": 28.137950610275333,
      "grad_norm": 4.381542205810547,
      "learning_rate": 7.187425489639513e-05,
      "loss": 0.0028457075357437136,
      "step": 99130
    },
    {
      "epoch": 28.140789100198695,
      "grad_norm": 0.5518834590911865,
      "learning_rate": 7.187141640647175e-05,
      "loss": 0.0023403670638799667,
      "step": 99140
    },
    {
      "epoch": 28.143627590122055,
      "grad_norm": 2.9439587593078613,
      "learning_rate": 7.18685779165484e-05,
      "loss": 0.005348050594329834,
      "step": 99150
    },
    {
      "epoch": 28.146466080045414,
      "grad_norm": 0.191433385014534,
      "learning_rate": 7.186573942662504e-05,
      "loss": 0.005311688035726547,
      "step": 99160
    },
    {
      "epoch": 28.149304569968777,
      "grad_norm": 0.3984471261501312,
      "learning_rate": 7.186290093670168e-05,
      "loss": 0.004995758458971977,
      "step": 99170
    },
    {
      "epoch": 28.152143059892136,
      "grad_norm": 1.2335752248764038,
      "learning_rate": 7.186006244677832e-05,
      "loss": 0.010822852700948715,
      "step": 99180
    },
    {
      "epoch": 28.1549815498155,
      "grad_norm": 2.2628772258758545,
      "learning_rate": 7.185722395685496e-05,
      "loss": 0.016571898758411408,
      "step": 99190
    },
    {
      "epoch": 28.15782003973886,
      "grad_norm": 1.8707209825515747,
      "learning_rate": 7.18543854669316e-05,
      "loss": 0.0039032064378261566,
      "step": 99200
    },
    {
      "epoch": 28.16065852966222,
      "grad_norm": 0.0559735931456089,
      "learning_rate": 7.185154697700823e-05,
      "loss": 0.0021983252838253973,
      "step": 99210
    },
    {
      "epoch": 28.16349701958558,
      "grad_norm": 2.437621593475342,
      "learning_rate": 7.184870848708487e-05,
      "loss": 0.0038735304027795792,
      "step": 99220
    },
    {
      "epoch": 28.16633550950894,
      "grad_norm": 0.7910727262496948,
      "learning_rate": 7.184586999716153e-05,
      "loss": 0.0058883953839540485,
      "step": 99230
    },
    {
      "epoch": 28.169173999432303,
      "grad_norm": 4.698671817779541,
      "learning_rate": 7.184303150723815e-05,
      "loss": 0.00824366882443428,
      "step": 99240
    },
    {
      "epoch": 28.172012489355662,
      "grad_norm": 0.28635090589523315,
      "learning_rate": 7.18401930173148e-05,
      "loss": 0.00510806143283844,
      "step": 99250
    },
    {
      "epoch": 28.174850979279025,
      "grad_norm": 0.18305616080760956,
      "learning_rate": 7.183735452739144e-05,
      "loss": 0.0059801720082759855,
      "step": 99260
    },
    {
      "epoch": 28.177689469202384,
      "grad_norm": 0.2710353434085846,
      "learning_rate": 7.183451603746806e-05,
      "loss": 0.016687440872192382,
      "step": 99270
    },
    {
      "epoch": 28.180527959125744,
      "grad_norm": 0.15624389052391052,
      "learning_rate": 7.18316775475447e-05,
      "loss": 0.0055293112993240355,
      "step": 99280
    },
    {
      "epoch": 28.183366449049107,
      "grad_norm": 0.9305402040481567,
      "learning_rate": 7.182883905762135e-05,
      "loss": 0.0020359382033348083,
      "step": 99290
    },
    {
      "epoch": 28.186204938972466,
      "grad_norm": 5.156400680541992,
      "learning_rate": 7.182600056769799e-05,
      "loss": 0.006112484633922577,
      "step": 99300
    },
    {
      "epoch": 28.18904342889583,
      "grad_norm": 0.15358850359916687,
      "learning_rate": 7.182316207777463e-05,
      "loss": 0.004187551885843277,
      "step": 99310
    },
    {
      "epoch": 28.19188191881919,
      "grad_norm": 1.0000665187835693,
      "learning_rate": 7.182032358785127e-05,
      "loss": 0.010005500912666321,
      "step": 99320
    },
    {
      "epoch": 28.194720408742548,
      "grad_norm": 0.31868642568588257,
      "learning_rate": 7.181748509792791e-05,
      "loss": 0.004903925210237503,
      "step": 99330
    },
    {
      "epoch": 28.19755889866591,
      "grad_norm": 7.087325096130371,
      "learning_rate": 7.181464660800454e-05,
      "loss": 0.013305026292800903,
      "step": 99340
    },
    {
      "epoch": 28.20039738858927,
      "grad_norm": 0.7735273838043213,
      "learning_rate": 7.181180811808118e-05,
      "loss": 0.00845407098531723,
      "step": 99350
    },
    {
      "epoch": 28.203235878512633,
      "grad_norm": Infinity,
      "learning_rate": 7.180896962815782e-05,
      "loss": 0.02649623155593872,
      "step": 99360
    },
    {
      "epoch": 28.206074368435992,
      "grad_norm": 2.1063737869262695,
      "learning_rate": 7.18064149872268e-05,
      "loss": 0.019975988566875456,
      "step": 99370
    },
    {
      "epoch": 28.20891285835935,
      "grad_norm": 1.5339776277542114,
      "learning_rate": 7.180357649730344e-05,
      "loss": 0.00294656902551651,
      "step": 99380
    },
    {
      "epoch": 28.211751348282714,
      "grad_norm": 0.7404527068138123,
      "learning_rate": 7.180073800738007e-05,
      "loss": 0.008500242233276367,
      "step": 99390
    },
    {
      "epoch": 28.214589838206074,
      "grad_norm": 9.017537117004395,
      "learning_rate": 7.179789951745671e-05,
      "loss": 0.003944528102874756,
      "step": 99400
    },
    {
      "epoch": 28.217428328129436,
      "grad_norm": 0.5409374237060547,
      "learning_rate": 7.179506102753335e-05,
      "loss": 0.003008117713034153,
      "step": 99410
    },
    {
      "epoch": 28.220266818052796,
      "grad_norm": 1.012901782989502,
      "learning_rate": 7.179222253761e-05,
      "loss": 0.0022559275850653647,
      "step": 99420
    },
    {
      "epoch": 28.223105307976155,
      "grad_norm": 0.2942754328250885,
      "learning_rate": 7.178938404768664e-05,
      "loss": 0.004658424854278564,
      "step": 99430
    },
    {
      "epoch": 28.225943797899518,
      "grad_norm": 3.034888744354248,
      "learning_rate": 7.178654555776328e-05,
      "loss": 0.0037041105329990386,
      "step": 99440
    },
    {
      "epoch": 28.228782287822877,
      "grad_norm": 1.0000689029693604,
      "learning_rate": 7.17837070678399e-05,
      "loss": 0.007320758700370788,
      "step": 99450
    },
    {
      "epoch": 28.23162077774624,
      "grad_norm": 0.6151852607727051,
      "learning_rate": 7.178086857791655e-05,
      "loss": 0.008309341967105865,
      "step": 99460
    },
    {
      "epoch": 28.2344592676696,
      "grad_norm": 8.539571762084961,
      "learning_rate": 7.177803008799319e-05,
      "loss": 0.007549205422401428,
      "step": 99470
    },
    {
      "epoch": 28.23729775759296,
      "grad_norm": 0.3434697985649109,
      "learning_rate": 7.177519159806983e-05,
      "loss": 0.014373625814914703,
      "step": 99480
    },
    {
      "epoch": 28.240136247516322,
      "grad_norm": 0.06811416894197464,
      "learning_rate": 7.177235310814647e-05,
      "loss": 0.004527425020933151,
      "step": 99490
    },
    {
      "epoch": 28.24297473743968,
      "grad_norm": 0.18530352413654327,
      "learning_rate": 7.176951461822311e-05,
      "loss": 0.010271275043487548,
      "step": 99500
    },
    {
      "epoch": 28.24297473743968,
      "eval_accuracy": 0.9710688624658231,
      "eval_loss": 0.10043025016784668,
      "eval_runtime": 32.0465,
      "eval_samples_per_second": 490.756,
      "eval_steps_per_second": 7.676,
      "step": 99500
    },
    {
      "epoch": 28.245813227363044,
      "grad_norm": 7.870345592498779,
      "learning_rate": 7.176667612829974e-05,
      "loss": 0.0062620677053928375,
      "step": 99510
    },
    {
      "epoch": 28.248651717286403,
      "grad_norm": 2.1217336654663086,
      "learning_rate": 7.176383763837638e-05,
      "loss": 0.013517190515995026,
      "step": 99520
    },
    {
      "epoch": 28.251490207209763,
      "grad_norm": 0.18019378185272217,
      "learning_rate": 7.176099914845302e-05,
      "loss": 0.02984822392463684,
      "step": 99530
    },
    {
      "epoch": 28.254328697133126,
      "grad_norm": 1.0927499532699585,
      "learning_rate": 7.175816065852967e-05,
      "loss": 0.007988478243350982,
      "step": 99540
    },
    {
      "epoch": 28.257167187056485,
      "grad_norm": 0.55823814868927,
      "learning_rate": 7.17553221686063e-05,
      "loss": 0.022663332521915436,
      "step": 99550
    },
    {
      "epoch": 28.260005676979848,
      "grad_norm": 0.4943069815635681,
      "learning_rate": 7.175248367868295e-05,
      "loss": 0.0180228590965271,
      "step": 99560
    },
    {
      "epoch": 28.262844166903207,
      "grad_norm": 0.5438591241836548,
      "learning_rate": 7.174964518875959e-05,
      "loss": 0.016398058831691743,
      "step": 99570
    },
    {
      "epoch": 28.26568265682657,
      "grad_norm": 0.8588427305221558,
      "learning_rate": 7.174680669883622e-05,
      "loss": 0.013099882006645202,
      "step": 99580
    },
    {
      "epoch": 28.26852114674993,
      "grad_norm": 0.2103145867586136,
      "learning_rate": 7.174396820891286e-05,
      "loss": 0.0022102640941739082,
      "step": 99590
    },
    {
      "epoch": 28.27135963667329,
      "grad_norm": 1.1357195377349854,
      "learning_rate": 7.17411297189895e-05,
      "loss": 0.014051295816898346,
      "step": 99600
    },
    {
      "epoch": 28.27419812659665,
      "grad_norm": 0.8924902677536011,
      "learning_rate": 7.173829122906614e-05,
      "loss": 0.011151428520679473,
      "step": 99610
    },
    {
      "epoch": 28.27703661652001,
      "grad_norm": 0.5075741410255432,
      "learning_rate": 7.173545273914278e-05,
      "loss": 0.01110089272260666,
      "step": 99620
    },
    {
      "epoch": 28.279875106443374,
      "grad_norm": 13.592966079711914,
      "learning_rate": 7.173261424921942e-05,
      "loss": 0.009353438019752502,
      "step": 99630
    },
    {
      "epoch": 28.282713596366733,
      "grad_norm": 0.902879536151886,
      "learning_rate": 7.172977575929605e-05,
      "loss": 0.013001786172389984,
      "step": 99640
    },
    {
      "epoch": 28.285552086290092,
      "grad_norm": 0.33300724625587463,
      "learning_rate": 7.17269372693727e-05,
      "loss": 0.005142996460199356,
      "step": 99650
    },
    {
      "epoch": 28.288390576213455,
      "grad_norm": 0.56903076171875,
      "learning_rate": 7.172409877944933e-05,
      "loss": 0.006196601688861847,
      "step": 99660
    },
    {
      "epoch": 28.291229066136815,
      "grad_norm": 1.6884076595306396,
      "learning_rate": 7.172126028952598e-05,
      "loss": 0.0052748024463653564,
      "step": 99670
    },
    {
      "epoch": 28.294067556060178,
      "grad_norm": 1.3580198287963867,
      "learning_rate": 7.171842179960262e-05,
      "loss": 0.005803303048014641,
      "step": 99680
    },
    {
      "epoch": 28.296906045983537,
      "grad_norm": 1.6739133596420288,
      "learning_rate": 7.171558330967926e-05,
      "loss": 0.007761512696743011,
      "step": 99690
    },
    {
      "epoch": 28.299744535906896,
      "grad_norm": 0.12523601949214935,
      "learning_rate": 7.17127448197559e-05,
      "loss": 0.004343170672655106,
      "step": 99700
    },
    {
      "epoch": 28.30258302583026,
      "grad_norm": 10.531148910522461,
      "learning_rate": 7.170990632983253e-05,
      "loss": 0.010171770304441451,
      "step": 99710
    },
    {
      "epoch": 28.30542151575362,
      "grad_norm": 0.6256738901138306,
      "learning_rate": 7.170706783990917e-05,
      "loss": 0.006293675303459168,
      "step": 99720
    },
    {
      "epoch": 28.30826000567698,
      "grad_norm": 10.115216255187988,
      "learning_rate": 7.170422934998581e-05,
      "loss": 0.00881604328751564,
      "step": 99730
    },
    {
      "epoch": 28.31109849560034,
      "grad_norm": 4.375977516174316,
      "learning_rate": 7.170139086006244e-05,
      "loss": 0.0037077315151691436,
      "step": 99740
    },
    {
      "epoch": 28.3139369855237,
      "grad_norm": 0.08654285222291946,
      "learning_rate": 7.16985523701391e-05,
      "loss": 0.00287113543599844,
      "step": 99750
    },
    {
      "epoch": 28.316775475447063,
      "grad_norm": 0.05073104426264763,
      "learning_rate": 7.169571388021573e-05,
      "loss": 0.0011185450479388238,
      "step": 99760
    },
    {
      "epoch": 28.319613965370422,
      "grad_norm": 0.3021485209465027,
      "learning_rate": 7.169287539029236e-05,
      "loss": 0.0036738071590662,
      "step": 99770
    },
    {
      "epoch": 28.322452455293785,
      "grad_norm": 0.6410467624664307,
      "learning_rate": 7.1690036900369e-05,
      "loss": 0.004172143712639809,
      "step": 99780
    },
    {
      "epoch": 28.325290945217144,
      "grad_norm": 0.4152567684650421,
      "learning_rate": 7.168719841044565e-05,
      "loss": 0.01196725368499756,
      "step": 99790
    },
    {
      "epoch": 28.328129435140504,
      "grad_norm": 0.11621975898742676,
      "learning_rate": 7.168435992052229e-05,
      "loss": 0.00799572914838791,
      "step": 99800
    },
    {
      "epoch": 28.330967925063867,
      "grad_norm": 2.2739481925964355,
      "learning_rate": 7.168152143059893e-05,
      "loss": 0.007562326639890671,
      "step": 99810
    },
    {
      "epoch": 28.333806414987226,
      "grad_norm": 0.6035052537918091,
      "learning_rate": 7.167868294067557e-05,
      "loss": 0.00640815794467926,
      "step": 99820
    },
    {
      "epoch": 28.33664490491059,
      "grad_norm": 0.5572497248649597,
      "learning_rate": 7.167584445075221e-05,
      "loss": 0.0018093831837177276,
      "step": 99830
    },
    {
      "epoch": 28.339483394833948,
      "grad_norm": 0.46112316846847534,
      "learning_rate": 7.167300596082884e-05,
      "loss": 0.006089796125888824,
      "step": 99840
    },
    {
      "epoch": 28.342321884757308,
      "grad_norm": 0.11471764743328094,
      "learning_rate": 7.167016747090548e-05,
      "loss": 0.016142004728317262,
      "step": 99850
    },
    {
      "epoch": 28.34516037468067,
      "grad_norm": 1.2781332731246948,
      "learning_rate": 7.166732898098212e-05,
      "loss": 0.02902599275112152,
      "step": 99860
    },
    {
      "epoch": 28.34799886460403,
      "grad_norm": 0.7988767623901367,
      "learning_rate": 7.166449049105875e-05,
      "loss": 0.0028774447739124296,
      "step": 99870
    },
    {
      "epoch": 28.350837354527393,
      "grad_norm": 1.424643635749817,
      "learning_rate": 7.16616520011354e-05,
      "loss": 0.00243720356374979,
      "step": 99880
    },
    {
      "epoch": 28.353675844450752,
      "grad_norm": 3.3160929679870605,
      "learning_rate": 7.165881351121205e-05,
      "loss": 0.008519315719604492,
      "step": 99890
    },
    {
      "epoch": 28.35651433437411,
      "grad_norm": 2.2206480503082275,
      "learning_rate": 7.165597502128867e-05,
      "loss": 0.007845304906368256,
      "step": 99900
    },
    {
      "epoch": 28.359352824297474,
      "grad_norm": 1.5007569789886475,
      "learning_rate": 7.165313653136531e-05,
      "loss": 0.019638608396053314,
      "step": 99910
    },
    {
      "epoch": 28.362191314220834,
      "grad_norm": 2.6874520778656006,
      "learning_rate": 7.165029804144196e-05,
      "loss": 0.01536380797624588,
      "step": 99920
    },
    {
      "epoch": 28.365029804144196,
      "grad_norm": 0.7885878682136536,
      "learning_rate": 7.16474595515186e-05,
      "loss": 0.014301693439483643,
      "step": 99930
    },
    {
      "epoch": 28.367868294067556,
      "grad_norm": 4.780697345733643,
      "learning_rate": 7.164462106159523e-05,
      "loss": 0.004816725850105286,
      "step": 99940
    },
    {
      "epoch": 28.37070678399092,
      "grad_norm": 3.338611602783203,
      "learning_rate": 7.164178257167188e-05,
      "loss": 0.012926584482192994,
      "step": 99950
    },
    {
      "epoch": 28.373545273914278,
      "grad_norm": 0.826906681060791,
      "learning_rate": 7.163894408174852e-05,
      "loss": 0.0029450468719005585,
      "step": 99960
    },
    {
      "epoch": 28.376383763837637,
      "grad_norm": 1.2988272905349731,
      "learning_rate": 7.163610559182515e-05,
      "loss": 0.012390592694282531,
      "step": 99970
    },
    {
      "epoch": 28.379222253761,
      "grad_norm": 12.215862274169922,
      "learning_rate": 7.163326710190179e-05,
      "loss": 0.010583670437335968,
      "step": 99980
    },
    {
      "epoch": 28.38206074368436,
      "grad_norm": 0.13780027627944946,
      "learning_rate": 7.163042861197843e-05,
      "loss": 0.006357983499765396,
      "step": 99990
    },
    {
      "epoch": 28.384899233607722,
      "grad_norm": 2.2430219650268555,
      "learning_rate": 7.162759012205506e-05,
      "loss": 0.004751493036746979,
      "step": 100000
    },
    {
      "epoch": 28.384899233607722,
      "eval_accuracy": 0.9714503719717683,
      "eval_loss": 0.09463127702474594,
      "eval_runtime": 31.9562,
      "eval_samples_per_second": 492.142,
      "eval_steps_per_second": 7.698,
      "step": 100000
    },
    {
      "epoch": 28.38773772353108,
      "grad_norm": 14.236083984375,
      "learning_rate": 7.162475163213172e-05,
      "loss": 0.008964965492486954,
      "step": 100010
    },
    {
      "epoch": 28.39057621345444,
      "grad_norm": 0.1571747064590454,
      "learning_rate": 7.162191314220836e-05,
      "loss": 0.006165198236703873,
      "step": 100020
    },
    {
      "epoch": 28.393414703377804,
      "grad_norm": 1.0151523351669312,
      "learning_rate": 7.161907465228498e-05,
      "loss": 0.004084819555282592,
      "step": 100030
    },
    {
      "epoch": 28.396253193301163,
      "grad_norm": 0.3392554521560669,
      "learning_rate": 7.161623616236163e-05,
      "loss": 0.0028458144515752792,
      "step": 100040
    },
    {
      "epoch": 28.399091683224526,
      "grad_norm": 0.07869723439216614,
      "learning_rate": 7.161339767243827e-05,
      "loss": 0.019164933264255522,
      "step": 100050
    },
    {
      "epoch": 28.401930173147885,
      "grad_norm": 0.5327182412147522,
      "learning_rate": 7.161055918251491e-05,
      "loss": 0.00611410140991211,
      "step": 100060
    },
    {
      "epoch": 28.404768663071245,
      "grad_norm": 2.274426221847534,
      "learning_rate": 7.160772069259154e-05,
      "loss": 0.006041222810745239,
      "step": 100070
    },
    {
      "epoch": 28.407607152994608,
      "grad_norm": 1.4661033153533936,
      "learning_rate": 7.160488220266819e-05,
      "loss": 0.0033572964370250703,
      "step": 100080
    },
    {
      "epoch": 28.410445642917967,
      "grad_norm": 0.10061443597078323,
      "learning_rate": 7.160204371274483e-05,
      "loss": 0.017109432816505434,
      "step": 100090
    },
    {
      "epoch": 28.41328413284133,
      "grad_norm": 0.09822703152894974,
      "learning_rate": 7.159920522282146e-05,
      "loss": 0.005338317155838013,
      "step": 100100
    },
    {
      "epoch": 28.41612262276469,
      "grad_norm": 0.4414235055446625,
      "learning_rate": 7.15963667328981e-05,
      "loss": 0.008624079823493957,
      "step": 100110
    },
    {
      "epoch": 28.41896111268805,
      "grad_norm": 0.32682645320892334,
      "learning_rate": 7.159352824297474e-05,
      "loss": 0.01851790100336075,
      "step": 100120
    },
    {
      "epoch": 28.42179960261141,
      "grad_norm": 1.7668160200119019,
      "learning_rate": 7.159068975305137e-05,
      "loss": 0.007768867909908295,
      "step": 100130
    },
    {
      "epoch": 28.42463809253477,
      "grad_norm": 0.9402163028717041,
      "learning_rate": 7.158785126312801e-05,
      "loss": 0.007426076382398605,
      "step": 100140
    },
    {
      "epoch": 28.427476582458134,
      "grad_norm": 6.239004611968994,
      "learning_rate": 7.158501277320467e-05,
      "loss": 0.009530752897262573,
      "step": 100150
    },
    {
      "epoch": 28.430315072381493,
      "grad_norm": 0.8270184993743896,
      "learning_rate": 7.15821742832813e-05,
      "loss": 0.003387078642845154,
      "step": 100160
    },
    {
      "epoch": 28.433153562304852,
      "grad_norm": 0.18614846467971802,
      "learning_rate": 7.157933579335794e-05,
      "loss": 0.006466030329465866,
      "step": 100170
    },
    {
      "epoch": 28.435992052228215,
      "grad_norm": 2.7785985469818115,
      "learning_rate": 7.157649730343458e-05,
      "loss": 0.008332522958517075,
      "step": 100180
    },
    {
      "epoch": 28.438830542151575,
      "grad_norm": 2.2295541763305664,
      "learning_rate": 7.157365881351122e-05,
      "loss": 0.009047508984804154,
      "step": 100190
    },
    {
      "epoch": 28.441669032074937,
      "grad_norm": 1.593537449836731,
      "learning_rate": 7.157082032358785e-05,
      "loss": 0.005991016328334808,
      "step": 100200
    },
    {
      "epoch": 28.444507521998297,
      "grad_norm": 1.1984152793884277,
      "learning_rate": 7.15679818336645e-05,
      "loss": 0.005390486493706703,
      "step": 100210
    },
    {
      "epoch": 28.447346011921656,
      "grad_norm": 6.124396324157715,
      "learning_rate": 7.156514334374114e-05,
      "loss": 0.013970617949962617,
      "step": 100220
    },
    {
      "epoch": 28.45018450184502,
      "grad_norm": 0.9215695858001709,
      "learning_rate": 7.156230485381777e-05,
      "loss": 0.008484168350696564,
      "step": 100230
    },
    {
      "epoch": 28.45302299176838,
      "grad_norm": 1.5122841596603394,
      "learning_rate": 7.155946636389441e-05,
      "loss": 0.007346770912408829,
      "step": 100240
    },
    {
      "epoch": 28.45586148169174,
      "grad_norm": 1.0301827192306519,
      "learning_rate": 7.155662787397105e-05,
      "loss": 0.007182473689317704,
      "step": 100250
    },
    {
      "epoch": 28.4586999716151,
      "grad_norm": 9.288808822631836,
      "learning_rate": 7.155378938404768e-05,
      "loss": 0.007451946288347245,
      "step": 100260
    },
    {
      "epoch": 28.46153846153846,
      "grad_norm": 15.230838775634766,
      "learning_rate": 7.155095089412432e-05,
      "loss": 0.0064343571662902836,
      "step": 100270
    },
    {
      "epoch": 28.464376951461823,
      "grad_norm": 4.403883457183838,
      "learning_rate": 7.154811240420098e-05,
      "loss": 0.01571103185415268,
      "step": 100280
    },
    {
      "epoch": 28.467215441385182,
      "grad_norm": 4.18380069732666,
      "learning_rate": 7.15452739142776e-05,
      "loss": 0.010106173902750015,
      "step": 100290
    },
    {
      "epoch": 28.470053931308545,
      "grad_norm": 0.986223042011261,
      "learning_rate": 7.154243542435425e-05,
      "loss": 0.01157679408788681,
      "step": 100300
    },
    {
      "epoch": 28.472892421231904,
      "grad_norm": 0.552253782749176,
      "learning_rate": 7.153959693443089e-05,
      "loss": 0.019702553749084473,
      "step": 100310
    },
    {
      "epoch": 28.475730911155264,
      "grad_norm": 1.2180366516113281,
      "learning_rate": 7.153675844450753e-05,
      "loss": 0.009517708420753479,
      "step": 100320
    },
    {
      "epoch": 28.478569401078627,
      "grad_norm": 1.1200079917907715,
      "learning_rate": 7.153391995458416e-05,
      "loss": 0.00806499570608139,
      "step": 100330
    },
    {
      "epoch": 28.481407891001986,
      "grad_norm": 0.4424768388271332,
      "learning_rate": 7.15310814646608e-05,
      "loss": 0.004847056791186332,
      "step": 100340
    },
    {
      "epoch": 28.48424638092535,
      "grad_norm": 10.234271049499512,
      "learning_rate": 7.152824297473744e-05,
      "loss": 0.017241856455802916,
      "step": 100350
    },
    {
      "epoch": 28.487084870848708,
      "grad_norm": 17.503644943237305,
      "learning_rate": 7.152540448481408e-05,
      "loss": 0.020249982178211213,
      "step": 100360
    },
    {
      "epoch": 28.48992336077207,
      "grad_norm": 0.2745189666748047,
      "learning_rate": 7.152256599489072e-05,
      "loss": 0.008146333694458007,
      "step": 100370
    },
    {
      "epoch": 28.49276185069543,
      "grad_norm": 0.2709268629550934,
      "learning_rate": 7.151972750496736e-05,
      "loss": 0.00816885232925415,
      "step": 100380
    },
    {
      "epoch": 28.49560034061879,
      "grad_norm": 8.198881149291992,
      "learning_rate": 7.151688901504399e-05,
      "loss": 0.006383834779262543,
      "step": 100390
    },
    {
      "epoch": 28.498438830542153,
      "grad_norm": 0.42209938168525696,
      "learning_rate": 7.151405052512063e-05,
      "loss": 0.013410773873329163,
      "step": 100400
    },
    {
      "epoch": 28.501277320465512,
      "grad_norm": 0.7425680756568909,
      "learning_rate": 7.151121203519729e-05,
      "loss": 0.009715896099805832,
      "step": 100410
    },
    {
      "epoch": 28.504115810388875,
      "grad_norm": 4.963202476501465,
      "learning_rate": 7.150837354527392e-05,
      "loss": 0.00965944305062294,
      "step": 100420
    },
    {
      "epoch": 28.506954300312234,
      "grad_norm": 1.4698773622512817,
      "learning_rate": 7.150553505535056e-05,
      "loss": 0.003851177915930748,
      "step": 100430
    },
    {
      "epoch": 28.509792790235593,
      "grad_norm": 2.54487681388855,
      "learning_rate": 7.15026965654272e-05,
      "loss": 0.004846449196338654,
      "step": 100440
    },
    {
      "epoch": 28.512631280158956,
      "grad_norm": 0.319017618894577,
      "learning_rate": 7.149985807550383e-05,
      "loss": 0.005213583633303642,
      "step": 100450
    },
    {
      "epoch": 28.515469770082316,
      "grad_norm": 0.5422877669334412,
      "learning_rate": 7.149701958558047e-05,
      "loss": 0.02071237862110138,
      "step": 100460
    },
    {
      "epoch": 28.51830826000568,
      "grad_norm": 0.8717231154441833,
      "learning_rate": 7.149418109565711e-05,
      "loss": 0.020321384072303772,
      "step": 100470
    },
    {
      "epoch": 28.521146749929038,
      "grad_norm": 0.016573797911405563,
      "learning_rate": 7.149134260573375e-05,
      "loss": 0.014075252413749694,
      "step": 100480
    },
    {
      "epoch": 28.523985239852397,
      "grad_norm": 0.07129339128732681,
      "learning_rate": 7.148850411581039e-05,
      "loss": 0.012800510227680206,
      "step": 100490
    },
    {
      "epoch": 28.52682372977576,
      "grad_norm": 0.4059860408306122,
      "learning_rate": 7.148566562588703e-05,
      "loss": 0.010989009588956832,
      "step": 100500
    },
    {
      "epoch": 28.52682372977576,
      "eval_accuracy": 0.9732307496661792,
      "eval_loss": 0.08732210099697113,
      "eval_runtime": 31.7492,
      "eval_samples_per_second": 495.351,
      "eval_steps_per_second": 7.748,
      "step": 100500
    },
    {
      "epoch": 28.52966221969912,
      "grad_norm": 19.60007667541504,
      "learning_rate": 7.148282713596368e-05,
      "loss": 0.019346103072166443,
      "step": 100510
    },
    {
      "epoch": 28.532500709622482,
      "grad_norm": 1.2773470878601074,
      "learning_rate": 7.14799886460403e-05,
      "loss": 0.005837064608931542,
      "step": 100520
    },
    {
      "epoch": 28.53533919954584,
      "grad_norm": 7.744444847106934,
      "learning_rate": 7.147715015611694e-05,
      "loss": 0.005224072188138962,
      "step": 100530
    },
    {
      "epoch": 28.5381776894692,
      "grad_norm": 12.79071044921875,
      "learning_rate": 7.147431166619359e-05,
      "loss": 0.018161511421203612,
      "step": 100540
    },
    {
      "epoch": 28.541016179392564,
      "grad_norm": 0.3449995517730713,
      "learning_rate": 7.147147317627023e-05,
      "loss": 0.0075508542358875275,
      "step": 100550
    },
    {
      "epoch": 28.543854669315923,
      "grad_norm": 0.908300518989563,
      "learning_rate": 7.146863468634687e-05,
      "loss": 0.004600193724036217,
      "step": 100560
    },
    {
      "epoch": 28.546693159239286,
      "grad_norm": 14.161280632019043,
      "learning_rate": 7.146579619642351e-05,
      "loss": 0.01685826778411865,
      "step": 100570
    },
    {
      "epoch": 28.549531649162645,
      "grad_norm": 0.5036726593971252,
      "learning_rate": 7.146295770650014e-05,
      "loss": 0.009659422188997268,
      "step": 100580
    },
    {
      "epoch": 28.552370139086005,
      "grad_norm": 3.546938419342041,
      "learning_rate": 7.146011921657678e-05,
      "loss": 0.016599640250205994,
      "step": 100590
    },
    {
      "epoch": 28.555208629009368,
      "grad_norm": 4.862947463989258,
      "learning_rate": 7.145728072665342e-05,
      "loss": 0.01475844979286194,
      "step": 100600
    },
    {
      "epoch": 28.558047118932727,
      "grad_norm": 2.6555538177490234,
      "learning_rate": 7.145444223673006e-05,
      "loss": 0.01906416714191437,
      "step": 100610
    },
    {
      "epoch": 28.56088560885609,
      "grad_norm": 0.3337307274341583,
      "learning_rate": 7.14516037468067e-05,
      "loss": 0.003660879284143448,
      "step": 100620
    },
    {
      "epoch": 28.56372409877945,
      "grad_norm": 0.8337148427963257,
      "learning_rate": 7.144876525688334e-05,
      "loss": 0.0020680509507656096,
      "step": 100630
    },
    {
      "epoch": 28.56656258870281,
      "grad_norm": 5.964834690093994,
      "learning_rate": 7.144592676695999e-05,
      "loss": 0.00949869304895401,
      "step": 100640
    },
    {
      "epoch": 28.56940107862617,
      "grad_norm": 1.5884510278701782,
      "learning_rate": 7.144308827703661e-05,
      "loss": 0.004462628811597824,
      "step": 100650
    },
    {
      "epoch": 28.57223956854953,
      "grad_norm": 0.7103903293609619,
      "learning_rate": 7.144024978711326e-05,
      "loss": 0.0022290127351880075,
      "step": 100660
    },
    {
      "epoch": 28.575078058472894,
      "grad_norm": 9.035968780517578,
      "learning_rate": 7.14374112971899e-05,
      "loss": 0.00630137026309967,
      "step": 100670
    },
    {
      "epoch": 28.577916548396253,
      "grad_norm": 7.934122562408447,
      "learning_rate": 7.143457280726654e-05,
      "loss": 0.008252602070569992,
      "step": 100680
    },
    {
      "epoch": 28.580755038319612,
      "grad_norm": 7.2589616775512695,
      "learning_rate": 7.143173431734318e-05,
      "loss": 0.012956231832504272,
      "step": 100690
    },
    {
      "epoch": 28.583593528242975,
      "grad_norm": 5.236428260803223,
      "learning_rate": 7.142889582741982e-05,
      "loss": 0.007449489831924438,
      "step": 100700
    },
    {
      "epoch": 28.586432018166335,
      "grad_norm": 0.2123892903327942,
      "learning_rate": 7.142605733749645e-05,
      "loss": 0.01028297021985054,
      "step": 100710
    },
    {
      "epoch": 28.589270508089697,
      "grad_norm": 9.255849838256836,
      "learning_rate": 7.142321884757309e-05,
      "loss": 0.021558667719364166,
      "step": 100720
    },
    {
      "epoch": 28.592108998013057,
      "grad_norm": 0.5491239428520203,
      "learning_rate": 7.142038035764973e-05,
      "loss": 0.007042942941188813,
      "step": 100730
    },
    {
      "epoch": 28.594947487936416,
      "grad_norm": 5.788693904876709,
      "learning_rate": 7.141754186772637e-05,
      "loss": 0.005673841387033462,
      "step": 100740
    },
    {
      "epoch": 28.59778597785978,
      "grad_norm": 7.073646545410156,
      "learning_rate": 7.141470337780301e-05,
      "loss": 0.02442290186882019,
      "step": 100750
    },
    {
      "epoch": 28.60062446778314,
      "grad_norm": 0.08729052543640137,
      "learning_rate": 7.141186488787966e-05,
      "loss": 0.011055664718151092,
      "step": 100760
    },
    {
      "epoch": 28.6034629577065,
      "grad_norm": 6.922945022583008,
      "learning_rate": 7.14090263979563e-05,
      "loss": 0.017723338305950166,
      "step": 100770
    },
    {
      "epoch": 28.60630144762986,
      "grad_norm": 1.4177234172821045,
      "learning_rate": 7.140618790803292e-05,
      "loss": 0.0036864906549453737,
      "step": 100780
    },
    {
      "epoch": 28.609139937553223,
      "grad_norm": 5.284659385681152,
      "learning_rate": 7.140334941810957e-05,
      "loss": 0.007224171608686447,
      "step": 100790
    },
    {
      "epoch": 28.611978427476583,
      "grad_norm": 1.7650071382522583,
      "learning_rate": 7.140051092818621e-05,
      "loss": 0.0037026748061180113,
      "step": 100800
    },
    {
      "epoch": 28.614816917399942,
      "grad_norm": 2.2681374549865723,
      "learning_rate": 7.139767243826285e-05,
      "loss": 0.011683833599090577,
      "step": 100810
    },
    {
      "epoch": 28.617655407323305,
      "grad_norm": 7.36200475692749,
      "learning_rate": 7.139483394833949e-05,
      "loss": 0.015427130460739135,
      "step": 100820
    },
    {
      "epoch": 28.620493897246664,
      "grad_norm": 4.682276248931885,
      "learning_rate": 7.139199545841613e-05,
      "loss": 0.011912651360034943,
      "step": 100830
    },
    {
      "epoch": 28.623332387170027,
      "grad_norm": 0.2080886960029602,
      "learning_rate": 7.138915696849276e-05,
      "loss": 0.011899788677692414,
      "step": 100840
    },
    {
      "epoch": 28.626170877093386,
      "grad_norm": 1.6055643558502197,
      "learning_rate": 7.13863184785694e-05,
      "loss": 0.021594719588756563,
      "step": 100850
    },
    {
      "epoch": 28.629009367016746,
      "grad_norm": 0.9615247249603271,
      "learning_rate": 7.138347998864604e-05,
      "loss": 0.014049062132835388,
      "step": 100860
    },
    {
      "epoch": 28.63184785694011,
      "grad_norm": 0.36892643570899963,
      "learning_rate": 7.138064149872268e-05,
      "loss": 0.005364476144313813,
      "step": 100870
    },
    {
      "epoch": 28.634686346863468,
      "grad_norm": 8.04882526397705,
      "learning_rate": 7.137780300879932e-05,
      "loss": 0.010923314839601517,
      "step": 100880
    },
    {
      "epoch": 28.63752483678683,
      "grad_norm": 0.1690046489238739,
      "learning_rate": 7.137496451887597e-05,
      "loss": 0.01091085970401764,
      "step": 100890
    },
    {
      "epoch": 28.64036332671019,
      "grad_norm": 0.24973492324352264,
      "learning_rate": 7.137212602895261e-05,
      "loss": 0.007949941605329514,
      "step": 100900
    },
    {
      "epoch": 28.64320181663355,
      "grad_norm": 0.97822105884552,
      "learning_rate": 7.136928753902924e-05,
      "loss": 0.01410357654094696,
      "step": 100910
    },
    {
      "epoch": 28.646040306556912,
      "grad_norm": 3.3417561054229736,
      "learning_rate": 7.136644904910588e-05,
      "loss": 0.015219548344612121,
      "step": 100920
    },
    {
      "epoch": 28.648878796480272,
      "grad_norm": 8.121927261352539,
      "learning_rate": 7.136361055918252e-05,
      "loss": 0.016392844915390014,
      "step": 100930
    },
    {
      "epoch": 28.651717286403635,
      "grad_norm": 8.13321304321289,
      "learning_rate": 7.136077206925916e-05,
      "loss": 0.017541772127151488,
      "step": 100940
    },
    {
      "epoch": 28.654555776326994,
      "grad_norm": 1.1033810377120972,
      "learning_rate": 7.13579335793358e-05,
      "loss": 0.015376982092857362,
      "step": 100950
    },
    {
      "epoch": 28.657394266250353,
      "grad_norm": 0.8146872520446777,
      "learning_rate": 7.135509508941244e-05,
      "loss": 0.013777221739292144,
      "step": 100960
    },
    {
      "epoch": 28.660232756173716,
      "grad_norm": 0.2390243113040924,
      "learning_rate": 7.135225659948907e-05,
      "loss": 0.005009932443499565,
      "step": 100970
    },
    {
      "epoch": 28.663071246097076,
      "grad_norm": 10.41771411895752,
      "learning_rate": 7.134941810956571e-05,
      "loss": 0.03552972376346588,
      "step": 100980
    },
    {
      "epoch": 28.66590973602044,
      "grad_norm": 3.416131019592285,
      "learning_rate": 7.134657961964235e-05,
      "loss": 0.00991942435503006,
      "step": 100990
    },
    {
      "epoch": 28.668748225943798,
      "grad_norm": 0.5241124629974365,
      "learning_rate": 7.1343741129719e-05,
      "loss": 0.01166784092783928,
      "step": 101000
    },
    {
      "epoch": 28.668748225943798,
      "eval_accuracy": 0.9717047116423985,
      "eval_loss": 0.09747762233018875,
      "eval_runtime": 31.9656,
      "eval_samples_per_second": 491.997,
      "eval_steps_per_second": 7.696,
      "step": 101000
    },
    {
      "epoch": 28.671586715867157,
      "grad_norm": 0.1518595665693283,
      "learning_rate": 7.134090263979564e-05,
      "loss": 0.007452286034822464,
      "step": 101010
    },
    {
      "epoch": 28.67442520579052,
      "grad_norm": 12.517583847045898,
      "learning_rate": 7.133806414987228e-05,
      "loss": 0.01999170035123825,
      "step": 101020
    },
    {
      "epoch": 28.67726369571388,
      "grad_norm": 4.50225305557251,
      "learning_rate": 7.133522565994892e-05,
      "loss": 0.01329212188720703,
      "step": 101030
    },
    {
      "epoch": 28.680102185637242,
      "grad_norm": 0.20579178631305695,
      "learning_rate": 7.133238717002555e-05,
      "loss": 0.010350432991981507,
      "step": 101040
    },
    {
      "epoch": 28.6829406755606,
      "grad_norm": 2.6723058223724365,
      "learning_rate": 7.132954868010219e-05,
      "loss": 0.0059473149478435515,
      "step": 101050
    },
    {
      "epoch": 28.68577916548396,
      "grad_norm": 2.880376100540161,
      "learning_rate": 7.132671019017883e-05,
      "loss": 0.007254064828157425,
      "step": 101060
    },
    {
      "epoch": 28.688617655407324,
      "grad_norm": 2.1201629638671875,
      "learning_rate": 7.132387170025546e-05,
      "loss": 0.004968840256333351,
      "step": 101070
    },
    {
      "epoch": 28.691456145330683,
      "grad_norm": 4.568824291229248,
      "learning_rate": 7.132103321033211e-05,
      "loss": 0.008316630125045776,
      "step": 101080
    },
    {
      "epoch": 28.694294635254046,
      "grad_norm": 0.46482187509536743,
      "learning_rate": 7.131819472040875e-05,
      "loss": 0.009304873645305634,
      "step": 101090
    },
    {
      "epoch": 28.697133125177405,
      "grad_norm": 1.7081698179244995,
      "learning_rate": 7.131535623048538e-05,
      "loss": 0.003351518139243126,
      "step": 101100
    },
    {
      "epoch": 28.699971615100765,
      "grad_norm": 0.5771971940994263,
      "learning_rate": 7.131251774056202e-05,
      "loss": 0.004367371648550033,
      "step": 101110
    },
    {
      "epoch": 28.702810105024128,
      "grad_norm": 2.7055492401123047,
      "learning_rate": 7.130967925063866e-05,
      "loss": 0.003940864652395249,
      "step": 101120
    },
    {
      "epoch": 28.705648594947487,
      "grad_norm": 0.07461835443973541,
      "learning_rate": 7.13068407607153e-05,
      "loss": 0.007275979220867157,
      "step": 101130
    },
    {
      "epoch": 28.70848708487085,
      "grad_norm": 0.7802384495735168,
      "learning_rate": 7.130400227079195e-05,
      "loss": 0.005133271589875222,
      "step": 101140
    },
    {
      "epoch": 28.71132557479421,
      "grad_norm": 6.766273498535156,
      "learning_rate": 7.130116378086859e-05,
      "loss": 0.010099205374717712,
      "step": 101150
    },
    {
      "epoch": 28.714164064717572,
      "grad_norm": 0.2780701518058777,
      "learning_rate": 7.129832529094523e-05,
      "loss": 0.0015162268653512,
      "step": 101160
    },
    {
      "epoch": 28.71700255464093,
      "grad_norm": 0.09015641361474991,
      "learning_rate": 7.129548680102186e-05,
      "loss": 0.00640840083360672,
      "step": 101170
    },
    {
      "epoch": 28.71984104456429,
      "grad_norm": 0.17100796103477478,
      "learning_rate": 7.12926483110985e-05,
      "loss": 0.005519610643386841,
      "step": 101180
    },
    {
      "epoch": 28.722679534487654,
      "grad_norm": 3.5195822715759277,
      "learning_rate": 7.128980982117514e-05,
      "loss": 0.005591012537479401,
      "step": 101190
    },
    {
      "epoch": 28.725518024411013,
      "grad_norm": 11.963899612426758,
      "learning_rate": 7.128697133125177e-05,
      "loss": 0.006942339241504669,
      "step": 101200
    },
    {
      "epoch": 28.728356514334376,
      "grad_norm": 1.07952082157135,
      "learning_rate": 7.128413284132842e-05,
      "loss": 0.005664400011301041,
      "step": 101210
    },
    {
      "epoch": 28.731195004257735,
      "grad_norm": 6.971432209014893,
      "learning_rate": 7.128129435140506e-05,
      "loss": 0.004131811484694481,
      "step": 101220
    },
    {
      "epoch": 28.734033494181094,
      "grad_norm": 5.83760929107666,
      "learning_rate": 7.127845586148169e-05,
      "loss": 0.008218763768672943,
      "step": 101230
    },
    {
      "epoch": 28.736871984104457,
      "grad_norm": 6.275095462799072,
      "learning_rate": 7.127561737155833e-05,
      "loss": 0.008962002396583558,
      "step": 101240
    },
    {
      "epoch": 28.739710474027817,
      "grad_norm": 11.295071601867676,
      "learning_rate": 7.127277888163497e-05,
      "loss": 0.01842151880264282,
      "step": 101250
    },
    {
      "epoch": 28.74254896395118,
      "grad_norm": 1.6877329349517822,
      "learning_rate": 7.126994039171162e-05,
      "loss": 0.006018535047769546,
      "step": 101260
    },
    {
      "epoch": 28.74538745387454,
      "grad_norm": 7.773751735687256,
      "learning_rate": 7.126710190178824e-05,
      "loss": 0.02105031907558441,
      "step": 101270
    },
    {
      "epoch": 28.748225943797898,
      "grad_norm": 1.135111927986145,
      "learning_rate": 7.12642634118649e-05,
      "loss": 0.008589871972799302,
      "step": 101280
    },
    {
      "epoch": 28.75106443372126,
      "grad_norm": 1.116197109222412,
      "learning_rate": 7.126142492194153e-05,
      "loss": 0.0016592826694250106,
      "step": 101290
    },
    {
      "epoch": 28.75390292364462,
      "grad_norm": 1.7515496015548706,
      "learning_rate": 7.125858643201817e-05,
      "loss": 0.006704209744930268,
      "step": 101300
    },
    {
      "epoch": 28.756741413567983,
      "grad_norm": 0.12796032428741455,
      "learning_rate": 7.125574794209481e-05,
      "loss": 0.0037404831498861315,
      "step": 101310
    },
    {
      "epoch": 28.759579903491343,
      "grad_norm": 1.580637812614441,
      "learning_rate": 7.125290945217145e-05,
      "loss": 0.004011278599500656,
      "step": 101320
    },
    {
      "epoch": 28.762418393414702,
      "grad_norm": 0.2218305468559265,
      "learning_rate": 7.125007096224808e-05,
      "loss": 0.015618444979190826,
      "step": 101330
    },
    {
      "epoch": 28.765256883338065,
      "grad_norm": 0.4241727590560913,
      "learning_rate": 7.124723247232473e-05,
      "loss": 0.010917829722166062,
      "step": 101340
    },
    {
      "epoch": 28.768095373261424,
      "grad_norm": 12.797643661499023,
      "learning_rate": 7.124439398240137e-05,
      "loss": 0.017390364408493043,
      "step": 101350
    },
    {
      "epoch": 28.770933863184787,
      "grad_norm": 0.7979776263237,
      "learning_rate": 7.1241555492478e-05,
      "loss": 0.006009407341480255,
      "step": 101360
    },
    {
      "epoch": 28.773772353108146,
      "grad_norm": 5.209594249725342,
      "learning_rate": 7.123871700255464e-05,
      "loss": 0.006633139401674271,
      "step": 101370
    },
    {
      "epoch": 28.776610843031506,
      "grad_norm": 0.2976085841655731,
      "learning_rate": 7.123587851263129e-05,
      "loss": 0.02513929009437561,
      "step": 101380
    },
    {
      "epoch": 28.77944933295487,
      "grad_norm": 0.3071852922439575,
      "learning_rate": 7.123304002270791e-05,
      "loss": 0.011713600158691407,
      "step": 101390
    },
    {
      "epoch": 28.782287822878228,
      "grad_norm": 0.08558647334575653,
      "learning_rate": 7.123020153278455e-05,
      "loss": 0.010322259366512298,
      "step": 101400
    },
    {
      "epoch": 28.78512631280159,
      "grad_norm": 0.4721450209617615,
      "learning_rate": 7.122736304286121e-05,
      "loss": 0.009536734223365784,
      "step": 101410
    },
    {
      "epoch": 28.78796480272495,
      "grad_norm": 1.1148145198822021,
      "learning_rate": 7.122452455293784e-05,
      "loss": 0.013054947555065154,
      "step": 101420
    },
    {
      "epoch": 28.79080329264831,
      "grad_norm": 7.7433576583862305,
      "learning_rate": 7.122168606301448e-05,
      "loss": 0.01272178441286087,
      "step": 101430
    },
    {
      "epoch": 28.793641782571672,
      "grad_norm": 1.0266673564910889,
      "learning_rate": 7.121884757309112e-05,
      "loss": 0.009355069696903228,
      "step": 101440
    },
    {
      "epoch": 28.79648027249503,
      "grad_norm": 1.1132599115371704,
      "learning_rate": 7.121600908316776e-05,
      "loss": 0.010987074673175811,
      "step": 101450
    },
    {
      "epoch": 28.799318762418395,
      "grad_norm": 1.3414323329925537,
      "learning_rate": 7.121317059324439e-05,
      "loss": 0.01369931697845459,
      "step": 101460
    },
    {
      "epoch": 28.802157252341754,
      "grad_norm": 4.0530266761779785,
      "learning_rate": 7.121033210332103e-05,
      "loss": 0.005238620936870575,
      "step": 101470
    },
    {
      "epoch": 28.804995742265113,
      "grad_norm": 0.662348210811615,
      "learning_rate": 7.120749361339769e-05,
      "loss": 0.014664453268051148,
      "step": 101480
    },
    {
      "epoch": 28.807834232188476,
      "grad_norm": 1.7385921478271484,
      "learning_rate": 7.120465512347431e-05,
      "loss": 0.002470959722995758,
      "step": 101490
    },
    {
      "epoch": 28.810672722111835,
      "grad_norm": 1.5477558374404907,
      "learning_rate": 7.120181663355095e-05,
      "loss": 0.022483333945274353,
      "step": 101500
    },
    {
      "epoch": 28.810672722111835,
      "eval_accuracy": 0.9665543333121384,
      "eval_loss": 0.11084696650505066,
      "eval_runtime": 32.0512,
      "eval_samples_per_second": 490.683,
      "eval_steps_per_second": 7.675,
      "step": 101500
    },
    {
      "epoch": 28.8135112120352,
      "grad_norm": 11.907801628112793,
      "learning_rate": 7.11989781436276e-05,
      "loss": 0.02277732938528061,
      "step": 101510
    },
    {
      "epoch": 28.816349701958558,
      "grad_norm": 0.2694256007671356,
      "learning_rate": 7.119613965370422e-05,
      "loss": 0.0041971862316131595,
      "step": 101520
    },
    {
      "epoch": 28.81918819188192,
      "grad_norm": 1.632904291152954,
      "learning_rate": 7.119330116378087e-05,
      "loss": 0.0043401304632425305,
      "step": 101530
    },
    {
      "epoch": 28.82202668180528,
      "grad_norm": 3.6642508506774902,
      "learning_rate": 7.119046267385752e-05,
      "loss": 0.007130427658557892,
      "step": 101540
    },
    {
      "epoch": 28.82486517172864,
      "grad_norm": 1.3708314895629883,
      "learning_rate": 7.118762418393415e-05,
      "loss": 0.007728991657495498,
      "step": 101550
    },
    {
      "epoch": 28.827703661652002,
      "grad_norm": 3.4043214321136475,
      "learning_rate": 7.118478569401079e-05,
      "loss": 0.00688321590423584,
      "step": 101560
    },
    {
      "epoch": 28.83054215157536,
      "grad_norm": 3.7715492248535156,
      "learning_rate": 7.118194720408743e-05,
      "loss": 0.014467497169971467,
      "step": 101570
    },
    {
      "epoch": 28.833380641498724,
      "grad_norm": 1.9979052543640137,
      "learning_rate": 7.117910871416407e-05,
      "loss": 0.02484259903430939,
      "step": 101580
    },
    {
      "epoch": 28.836219131422084,
      "grad_norm": 1.8113259077072144,
      "learning_rate": 7.11762702242407e-05,
      "loss": 0.008798358589410782,
      "step": 101590
    },
    {
      "epoch": 28.839057621345443,
      "grad_norm": 0.44838765263557434,
      "learning_rate": 7.117343173431734e-05,
      "loss": 0.00958983451128006,
      "step": 101600
    },
    {
      "epoch": 28.841896111268806,
      "grad_norm": 9.561324119567871,
      "learning_rate": 7.1170593244394e-05,
      "loss": 0.015247395634651184,
      "step": 101610
    },
    {
      "epoch": 28.844734601192165,
      "grad_norm": 8.192377090454102,
      "learning_rate": 7.116775475447062e-05,
      "loss": 0.01535978764295578,
      "step": 101620
    },
    {
      "epoch": 28.847573091115528,
      "grad_norm": 1.6025258302688599,
      "learning_rate": 7.116491626454727e-05,
      "loss": 0.013985240459442138,
      "step": 101630
    },
    {
      "epoch": 28.850411581038887,
      "grad_norm": 2.9541168212890625,
      "learning_rate": 7.116207777462391e-05,
      "loss": 0.004245470836758613,
      "step": 101640
    },
    {
      "epoch": 28.853250070962247,
      "grad_norm": 4.92551851272583,
      "learning_rate": 7.115923928470053e-05,
      "loss": 0.016747429966926575,
      "step": 101650
    },
    {
      "epoch": 28.85608856088561,
      "grad_norm": 0.7903095483779907,
      "learning_rate": 7.115640079477718e-05,
      "loss": 0.010327613353729248,
      "step": 101660
    },
    {
      "epoch": 28.85892705080897,
      "grad_norm": 6.856456756591797,
      "learning_rate": 7.115356230485382e-05,
      "loss": 0.01973172426223755,
      "step": 101670
    },
    {
      "epoch": 28.861765540732332,
      "grad_norm": 6.61758279800415,
      "learning_rate": 7.115072381493046e-05,
      "loss": 0.010151723772287369,
      "step": 101680
    },
    {
      "epoch": 28.86460403065569,
      "grad_norm": 0.915427029132843,
      "learning_rate": 7.11478853250071e-05,
      "loss": 0.001871420256793499,
      "step": 101690
    },
    {
      "epoch": 28.86744252057905,
      "grad_norm": 2.4556338787078857,
      "learning_rate": 7.114504683508374e-05,
      "loss": 0.02142704576253891,
      "step": 101700
    },
    {
      "epoch": 28.870281010502413,
      "grad_norm": 0.20605270564556122,
      "learning_rate": 7.114220834516038e-05,
      "loss": 0.013202217221260071,
      "step": 101710
    },
    {
      "epoch": 28.873119500425773,
      "grad_norm": 20.19638442993164,
      "learning_rate": 7.113936985523701e-05,
      "loss": 0.03857223391532898,
      "step": 101720
    },
    {
      "epoch": 28.875957990349136,
      "grad_norm": 2.921515941619873,
      "learning_rate": 7.113653136531365e-05,
      "loss": 0.0032826699316501617,
      "step": 101730
    },
    {
      "epoch": 28.878796480272495,
      "grad_norm": 0.9001482129096985,
      "learning_rate": 7.113369287539031e-05,
      "loss": 0.003273313492536545,
      "step": 101740
    },
    {
      "epoch": 28.881634970195854,
      "grad_norm": 4.514135360717773,
      "learning_rate": 7.113085438546693e-05,
      "loss": 0.013035798072814941,
      "step": 101750
    },
    {
      "epoch": 28.884473460119217,
      "grad_norm": 0.3059605062007904,
      "learning_rate": 7.112801589554358e-05,
      "loss": 0.01014857441186905,
      "step": 101760
    },
    {
      "epoch": 28.887311950042577,
      "grad_norm": 0.38479700684547424,
      "learning_rate": 7.112517740562022e-05,
      "loss": 0.018496225774288177,
      "step": 101770
    },
    {
      "epoch": 28.89015043996594,
      "grad_norm": 0.8560890555381775,
      "learning_rate": 7.112233891569685e-05,
      "loss": 0.004734523594379425,
      "step": 101780
    },
    {
      "epoch": 28.8929889298893,
      "grad_norm": 2.4990475177764893,
      "learning_rate": 7.111950042577349e-05,
      "loss": 0.01229851096868515,
      "step": 101790
    },
    {
      "epoch": 28.895827419812658,
      "grad_norm": 6.35207462310791,
      "learning_rate": 7.111666193585013e-05,
      "loss": 0.016225507855415343,
      "step": 101800
    },
    {
      "epoch": 28.89866590973602,
      "grad_norm": 0.111829474568367,
      "learning_rate": 7.111410729491911e-05,
      "loss": 0.0226533904671669,
      "step": 101810
    },
    {
      "epoch": 28.90150439965938,
      "grad_norm": 3.295138120651245,
      "learning_rate": 7.111126880499575e-05,
      "loss": 0.011834307760000228,
      "step": 101820
    },
    {
      "epoch": 28.904342889582743,
      "grad_norm": 0.06354343891143799,
      "learning_rate": 7.110843031507238e-05,
      "loss": 0.016262929141521453,
      "step": 101830
    },
    {
      "epoch": 28.907181379506103,
      "grad_norm": 5.150362014770508,
      "learning_rate": 7.110559182514902e-05,
      "loss": 0.0048189818859100345,
      "step": 101840
    },
    {
      "epoch": 28.910019869429462,
      "grad_norm": 3.729504108428955,
      "learning_rate": 7.110275333522566e-05,
      "loss": 0.009907587617635726,
      "step": 101850
    },
    {
      "epoch": 28.912858359352825,
      "grad_norm": 0.05235784873366356,
      "learning_rate": 7.10999148453023e-05,
      "loss": 0.006338720023632049,
      "step": 101860
    },
    {
      "epoch": 28.915696849276184,
      "grad_norm": 1.9165979623794556,
      "learning_rate": 7.109707635537894e-05,
      "loss": 0.0049846600741147995,
      "step": 101870
    },
    {
      "epoch": 28.918535339199547,
      "grad_norm": 0.6494324803352356,
      "learning_rate": 7.109423786545558e-05,
      "loss": 0.008502402901649475,
      "step": 101880
    },
    {
      "epoch": 28.921373829122906,
      "grad_norm": 2.0666415691375732,
      "learning_rate": 7.109139937553223e-05,
      "loss": 0.00657552108168602,
      "step": 101890
    },
    {
      "epoch": 28.92421231904627,
      "grad_norm": 5.774878025054932,
      "learning_rate": 7.108856088560885e-05,
      "loss": 0.015758527815341948,
      "step": 101900
    },
    {
      "epoch": 28.92705080896963,
      "grad_norm": 11.554808616638184,
      "learning_rate": 7.10857223956855e-05,
      "loss": 0.004877807945013047,
      "step": 101910
    },
    {
      "epoch": 28.929889298892988,
      "grad_norm": 0.165033757686615,
      "learning_rate": 7.108288390576215e-05,
      "loss": 0.007636648416519165,
      "step": 101920
    },
    {
      "epoch": 28.93272778881635,
      "grad_norm": 0.917658805847168,
      "learning_rate": 7.108004541583878e-05,
      "loss": 0.004572440683841705,
      "step": 101930
    },
    {
      "epoch": 28.93556627873971,
      "grad_norm": 0.5313619375228882,
      "learning_rate": 7.107720692591542e-05,
      "loss": 0.010018911957740784,
      "step": 101940
    },
    {
      "epoch": 28.93840476866307,
      "grad_norm": 0.20553848147392273,
      "learning_rate": 7.107436843599206e-05,
      "loss": 0.005402557551860809,
      "step": 101950
    },
    {
      "epoch": 28.941243258586432,
      "grad_norm": 0.5075179934501648,
      "learning_rate": 7.107152994606869e-05,
      "loss": 0.015724650025367735,
      "step": 101960
    },
    {
      "epoch": 28.94408174850979,
      "grad_norm": 1.000664472579956,
      "learning_rate": 7.106869145614533e-05,
      "loss": 0.010566221922636032,
      "step": 101970
    },
    {
      "epoch": 28.946920238433155,
      "grad_norm": 5.444941520690918,
      "learning_rate": 7.106585296622197e-05,
      "loss": 0.0061668016016483305,
      "step": 101980
    },
    {
      "epoch": 28.949758728356514,
      "grad_norm": 3.700824737548828,
      "learning_rate": 7.106301447629861e-05,
      "loss": 0.0048742715269327165,
      "step": 101990
    },
    {
      "epoch": 28.952597218279877,
      "grad_norm": 0.37488502264022827,
      "learning_rate": 7.106017598637525e-05,
      "loss": 0.014347702264785767,
      "step": 102000
    },
    {
      "epoch": 28.952597218279877,
      "eval_accuracy": 0.9671901824887137,
      "eval_loss": 0.11473508179187775,
      "eval_runtime": 32.3327,
      "eval_samples_per_second": 486.411,
      "eval_steps_per_second": 7.608,
      "step": 102000
    },
    {
      "epoch": 28.955435708203236,
      "grad_norm": 2.416783094406128,
      "learning_rate": 7.10573374964519e-05,
      "loss": 0.010539916902780532,
      "step": 102010
    },
    {
      "epoch": 28.958274198126595,
      "grad_norm": 1.0425323247909546,
      "learning_rate": 7.105449900652854e-05,
      "loss": 0.012587831914424896,
      "step": 102020
    },
    {
      "epoch": 28.96111268804996,
      "grad_norm": 0.219413623213768,
      "learning_rate": 7.105166051660516e-05,
      "loss": 0.005184582248330116,
      "step": 102030
    },
    {
      "epoch": 28.963951177973318,
      "grad_norm": 0.34706273674964905,
      "learning_rate": 7.10488220266818e-05,
      "loss": 0.007255873084068299,
      "step": 102040
    },
    {
      "epoch": 28.96678966789668,
      "grad_norm": 0.772415280342102,
      "learning_rate": 7.104598353675845e-05,
      "loss": 0.023017299175262452,
      "step": 102050
    },
    {
      "epoch": 28.96962815782004,
      "grad_norm": 0.8125987648963928,
      "learning_rate": 7.104314504683509e-05,
      "loss": 0.0028581274673342705,
      "step": 102060
    },
    {
      "epoch": 28.9724666477434,
      "grad_norm": 0.9142124056816101,
      "learning_rate": 7.104030655691173e-05,
      "loss": 0.007784959673881531,
      "step": 102070
    },
    {
      "epoch": 28.975305137666762,
      "grad_norm": 3.547302484512329,
      "learning_rate": 7.103746806698837e-05,
      "loss": 0.0034934565424919128,
      "step": 102080
    },
    {
      "epoch": 28.97814362759012,
      "grad_norm": 3.9685609340667725,
      "learning_rate": 7.1034629577065e-05,
      "loss": 0.007599274069070816,
      "step": 102090
    },
    {
      "epoch": 28.980982117513484,
      "grad_norm": 1.5260392427444458,
      "learning_rate": 7.103179108714164e-05,
      "loss": 0.010606472939252853,
      "step": 102100
    },
    {
      "epoch": 28.983820607436844,
      "grad_norm": 1.6723148822784424,
      "learning_rate": 7.102895259721828e-05,
      "loss": 0.007351754605770111,
      "step": 102110
    },
    {
      "epoch": 28.986659097360203,
      "grad_norm": 5.7239484786987305,
      "learning_rate": 7.102611410729492e-05,
      "loss": 0.005101994425058365,
      "step": 102120
    },
    {
      "epoch": 28.989497587283566,
      "grad_norm": 0.31368595361709595,
      "learning_rate": 7.102327561737156e-05,
      "loss": 0.006621608138084411,
      "step": 102130
    },
    {
      "epoch": 28.992336077206925,
      "grad_norm": 1.766554355621338,
      "learning_rate": 7.10204371274482e-05,
      "loss": 0.014098387956619263,
      "step": 102140
    },
    {
      "epoch": 28.995174567130288,
      "grad_norm": 0.29521605372428894,
      "learning_rate": 7.101759863752485e-05,
      "loss": 0.008170235157012939,
      "step": 102150
    },
    {
      "epoch": 28.998013057053647,
      "grad_norm": 0.7069188952445984,
      "learning_rate": 7.101476014760147e-05,
      "loss": 0.008200228959321976,
      "step": 102160
    },
    {
      "epoch": 29.000851546977007,
      "grad_norm": 0.9918931126594543,
      "learning_rate": 7.101192165767812e-05,
      "loss": 0.0060219407081604,
      "step": 102170
    },
    {
      "epoch": 29.00369003690037,
      "grad_norm": 0.28363242745399475,
      "learning_rate": 7.100908316775476e-05,
      "loss": 0.010178916901350022,
      "step": 102180
    },
    {
      "epoch": 29.00652852682373,
      "grad_norm": 0.2931966185569763,
      "learning_rate": 7.10062446778314e-05,
      "loss": 0.007821349799633026,
      "step": 102190
    },
    {
      "epoch": 29.009367016747092,
      "grad_norm": 0.5726573467254639,
      "learning_rate": 7.100340618790804e-05,
      "loss": 0.0022911567240953444,
      "step": 102200
    },
    {
      "epoch": 29.01220550667045,
      "grad_norm": 0.4696698784828186,
      "learning_rate": 7.100056769798468e-05,
      "loss": 0.0012418501079082488,
      "step": 102210
    },
    {
      "epoch": 29.01504399659381,
      "grad_norm": 8.253231048583984,
      "learning_rate": 7.099772920806131e-05,
      "loss": 0.006936902552843094,
      "step": 102220
    },
    {
      "epoch": 29.017882486517173,
      "grad_norm": 0.22573336958885193,
      "learning_rate": 7.099489071813795e-05,
      "loss": 0.0032431773841381075,
      "step": 102230
    },
    {
      "epoch": 29.020720976440533,
      "grad_norm": 1.0605664253234863,
      "learning_rate": 7.099205222821459e-05,
      "loss": 0.007231764495372772,
      "step": 102240
    },
    {
      "epoch": 29.023559466363896,
      "grad_norm": 0.14189422130584717,
      "learning_rate": 7.098921373829123e-05,
      "loss": 0.0010597817599773407,
      "step": 102250
    },
    {
      "epoch": 29.026397956287255,
      "grad_norm": 0.19582737982273102,
      "learning_rate": 7.098637524836788e-05,
      "loss": 0.0017875434830784797,
      "step": 102260
    },
    {
      "epoch": 29.029236446210614,
      "grad_norm": 0.40438878536224365,
      "learning_rate": 7.098353675844452e-05,
      "loss": 0.0010043013840913772,
      "step": 102270
    },
    {
      "epoch": 29.032074936133977,
      "grad_norm": 9.799649238586426,
      "learning_rate": 7.098069826852114e-05,
      "loss": 0.006715791672468186,
      "step": 102280
    },
    {
      "epoch": 29.034913426057336,
      "grad_norm": 0.307248055934906,
      "learning_rate": 7.097785977859779e-05,
      "loss": 0.0010018324479460715,
      "step": 102290
    },
    {
      "epoch": 29.0377519159807,
      "grad_norm": 9.453421592712402,
      "learning_rate": 7.097502128867443e-05,
      "loss": 0.00893569439649582,
      "step": 102300
    },
    {
      "epoch": 29.04059040590406,
      "grad_norm": 0.27662575244903564,
      "learning_rate": 7.097218279875107e-05,
      "loss": 0.003125188872218132,
      "step": 102310
    },
    {
      "epoch": 29.043428895827418,
      "grad_norm": 2.2073044776916504,
      "learning_rate": 7.096934430882771e-05,
      "loss": 0.009253701567649842,
      "step": 102320
    },
    {
      "epoch": 29.04626738575078,
      "grad_norm": 0.08398854732513428,
      "learning_rate": 7.096650581890435e-05,
      "loss": 0.0016285819932818414,
      "step": 102330
    },
    {
      "epoch": 29.04910587567414,
      "grad_norm": 0.9051011204719543,
      "learning_rate": 7.096366732898099e-05,
      "loss": 0.017110762000083924,
      "step": 102340
    },
    {
      "epoch": 29.051944365597503,
      "grad_norm": 2.4019415378570557,
      "learning_rate": 7.096082883905762e-05,
      "loss": 0.019776442646980287,
      "step": 102350
    },
    {
      "epoch": 29.054782855520862,
      "grad_norm": 5.2361860275268555,
      "learning_rate": 7.095799034913426e-05,
      "loss": 0.007220838963985443,
      "step": 102360
    },
    {
      "epoch": 29.057621345444225,
      "grad_norm": 1.2093380689620972,
      "learning_rate": 7.09551518592109e-05,
      "loss": 0.003241689130663872,
      "step": 102370
    },
    {
      "epoch": 29.060459835367585,
      "grad_norm": 2.4927449226379395,
      "learning_rate": 7.095231336928753e-05,
      "loss": 0.004962768405675888,
      "step": 102380
    },
    {
      "epoch": 29.063298325290944,
      "grad_norm": 0.18167811632156372,
      "learning_rate": 7.094947487936419e-05,
      "loss": 0.016752813756465913,
      "step": 102390
    },
    {
      "epoch": 29.066136815214307,
      "grad_norm": 1.8891583681106567,
      "learning_rate": 7.094663638944083e-05,
      "loss": 0.01641814112663269,
      "step": 102400
    },
    {
      "epoch": 29.068975305137666,
      "grad_norm": 5.791284084320068,
      "learning_rate": 7.094379789951746e-05,
      "loss": 0.00697208121418953,
      "step": 102410
    },
    {
      "epoch": 29.07181379506103,
      "grad_norm": 13.99333667755127,
      "learning_rate": 7.09409594095941e-05,
      "loss": 0.022768333554267883,
      "step": 102420
    },
    {
      "epoch": 29.07465228498439,
      "grad_norm": 0.8489956259727478,
      "learning_rate": 7.093812091967074e-05,
      "loss": 0.024726402759552003,
      "step": 102430
    },
    {
      "epoch": 29.077490774907748,
      "grad_norm": 7.231731414794922,
      "learning_rate": 7.093528242974738e-05,
      "loss": 0.03776673674583435,
      "step": 102440
    },
    {
      "epoch": 29.08032926483111,
      "grad_norm": 11.298152923583984,
      "learning_rate": 7.093244393982401e-05,
      "loss": 0.03368097543716431,
      "step": 102450
    },
    {
      "epoch": 29.08316775475447,
      "grad_norm": 3.5031540393829346,
      "learning_rate": 7.092960544990066e-05,
      "loss": 0.005102306604385376,
      "step": 102460
    },
    {
      "epoch": 29.086006244677833,
      "grad_norm": 0.5963215827941895,
      "learning_rate": 7.09267669599773e-05,
      "loss": 0.004594159126281738,
      "step": 102470
    },
    {
      "epoch": 29.088844734601192,
      "grad_norm": 9.248827934265137,
      "learning_rate": 7.092392847005393e-05,
      "loss": 0.014360295236110687,
      "step": 102480
    },
    {
      "epoch": 29.09168322452455,
      "grad_norm": 0.13988201320171356,
      "learning_rate": 7.092108998013057e-05,
      "loss": 0.0067124530673027035,
      "step": 102490
    },
    {
      "epoch": 29.094521714447914,
      "grad_norm": 0.7014368176460266,
      "learning_rate": 7.091825149020721e-05,
      "loss": 0.003118043951690197,
      "step": 102500
    },
    {
      "epoch": 29.094521714447914,
      "eval_accuracy": 0.9705601831245628,
      "eval_loss": 0.09572522342205048,
      "eval_runtime": 32.2884,
      "eval_samples_per_second": 487.079,
      "eval_steps_per_second": 7.619,
      "step": 102500
    },
    {
      "epoch": 29.097360204371274,
      "grad_norm": 0.26777759194374084,
      "learning_rate": 7.091541300028384e-05,
      "loss": 0.00574740469455719,
      "step": 102510
    },
    {
      "epoch": 29.100198694294637,
      "grad_norm": 0.31693804264068604,
      "learning_rate": 7.09125745103605e-05,
      "loss": 0.004652995616197586,
      "step": 102520
    },
    {
      "epoch": 29.103037184217996,
      "grad_norm": 0.5102849006652832,
      "learning_rate": 7.090973602043714e-05,
      "loss": 0.004267404228448868,
      "step": 102530
    },
    {
      "epoch": 29.105875674141355,
      "grad_norm": 0.8930804133415222,
      "learning_rate": 7.090689753051377e-05,
      "loss": 0.013785937428474426,
      "step": 102540
    },
    {
      "epoch": 29.10871416406472,
      "grad_norm": 0.16777321696281433,
      "learning_rate": 7.090405904059041e-05,
      "loss": 0.015562847256660461,
      "step": 102550
    },
    {
      "epoch": 29.111552653988078,
      "grad_norm": 6.0847086906433105,
      "learning_rate": 7.090122055066705e-05,
      "loss": 0.004124090075492859,
      "step": 102560
    },
    {
      "epoch": 29.11439114391144,
      "grad_norm": 0.2268490344285965,
      "learning_rate": 7.089838206074369e-05,
      "loss": 0.005310627818107605,
      "step": 102570
    },
    {
      "epoch": 29.1172296338348,
      "grad_norm": 3.9104433059692383,
      "learning_rate": 7.089554357082032e-05,
      "loss": 0.004322489351034164,
      "step": 102580
    },
    {
      "epoch": 29.12006812375816,
      "grad_norm": 0.33938488364219666,
      "learning_rate": 7.089270508089697e-05,
      "loss": 0.0035197902470827103,
      "step": 102590
    },
    {
      "epoch": 29.122906613681522,
      "grad_norm": 0.05653811991214752,
      "learning_rate": 7.088986659097361e-05,
      "loss": 0.015138925611972808,
      "step": 102600
    },
    {
      "epoch": 29.12574510360488,
      "grad_norm": 0.1376039683818817,
      "learning_rate": 7.088702810105024e-05,
      "loss": 0.006042524427175522,
      "step": 102610
    },
    {
      "epoch": 29.128583593528244,
      "grad_norm": 0.22564101219177246,
      "learning_rate": 7.088418961112688e-05,
      "loss": 0.008459549397230148,
      "step": 102620
    },
    {
      "epoch": 29.131422083451604,
      "grad_norm": 13.43925666809082,
      "learning_rate": 7.088135112120352e-05,
      "loss": 0.012538793683052062,
      "step": 102630
    },
    {
      "epoch": 29.134260573374963,
      "grad_norm": 8.326543807983398,
      "learning_rate": 7.087851263128015e-05,
      "loss": 0.0104373499751091,
      "step": 102640
    },
    {
      "epoch": 29.137099063298326,
      "grad_norm": 0.09716346114873886,
      "learning_rate": 7.087567414135681e-05,
      "loss": 0.009150402992963791,
      "step": 102650
    },
    {
      "epoch": 29.139937553221685,
      "grad_norm": 4.272698879241943,
      "learning_rate": 7.087283565143345e-05,
      "loss": 0.004462076723575592,
      "step": 102660
    },
    {
      "epoch": 29.142776043145048,
      "grad_norm": 0.21730031073093414,
      "learning_rate": 7.086999716151008e-05,
      "loss": 0.007064923644065857,
      "step": 102670
    },
    {
      "epoch": 29.145614533068407,
      "grad_norm": 9.249872207641602,
      "learning_rate": 7.086715867158672e-05,
      "loss": 0.015672376751899718,
      "step": 102680
    },
    {
      "epoch": 29.148453022991767,
      "grad_norm": 7.5102033615112305,
      "learning_rate": 7.086432018166336e-05,
      "loss": 0.004604329168796539,
      "step": 102690
    },
    {
      "epoch": 29.15129151291513,
      "grad_norm": 0.715934157371521,
      "learning_rate": 7.086148169174e-05,
      "loss": 0.01090754047036171,
      "step": 102700
    },
    {
      "epoch": 29.15413000283849,
      "grad_norm": 2.458418607711792,
      "learning_rate": 7.085864320181663e-05,
      "loss": 0.008639335632324219,
      "step": 102710
    },
    {
      "epoch": 29.15696849276185,
      "grad_norm": 11.255325317382812,
      "learning_rate": 7.085580471189328e-05,
      "loss": 0.014961728453636169,
      "step": 102720
    },
    {
      "epoch": 29.15980698268521,
      "grad_norm": 2.3089215755462646,
      "learning_rate": 7.085296622196992e-05,
      "loss": 0.019840715825557707,
      "step": 102730
    },
    {
      "epoch": 29.162645472608574,
      "grad_norm": 1.013698697090149,
      "learning_rate": 7.085012773204655e-05,
      "loss": 0.002451638504862785,
      "step": 102740
    },
    {
      "epoch": 29.165483962531933,
      "grad_norm": 5.578476905822754,
      "learning_rate": 7.08472892421232e-05,
      "loss": 0.0067212596535682675,
      "step": 102750
    },
    {
      "epoch": 29.168322452455293,
      "grad_norm": 0.8668773174285889,
      "learning_rate": 7.084445075219984e-05,
      "loss": 0.015506090223789215,
      "step": 102760
    },
    {
      "epoch": 29.171160942378656,
      "grad_norm": 0.264741986989975,
      "learning_rate": 7.084161226227646e-05,
      "loss": 0.0049165584146976474,
      "step": 102770
    },
    {
      "epoch": 29.173999432302015,
      "grad_norm": 0.295549213886261,
      "learning_rate": 7.08387737723531e-05,
      "loss": 0.006409379839897156,
      "step": 102780
    },
    {
      "epoch": 29.176837922225378,
      "grad_norm": 0.716139554977417,
      "learning_rate": 7.083593528242976e-05,
      "loss": 0.0056939773261547085,
      "step": 102790
    },
    {
      "epoch": 29.179676412148737,
      "grad_norm": 0.7113966345787048,
      "learning_rate": 7.083309679250639e-05,
      "loss": 0.010922443121671677,
      "step": 102800
    },
    {
      "epoch": 29.182514902072096,
      "grad_norm": 7.742880821228027,
      "learning_rate": 7.083025830258303e-05,
      "loss": 0.0054384000599384304,
      "step": 102810
    },
    {
      "epoch": 29.18535339199546,
      "grad_norm": 0.19212348759174347,
      "learning_rate": 7.082741981265967e-05,
      "loss": 0.003705166280269623,
      "step": 102820
    },
    {
      "epoch": 29.18819188191882,
      "grad_norm": 0.15441913902759552,
      "learning_rate": 7.082458132273631e-05,
      "loss": 0.0011392118409276008,
      "step": 102830
    },
    {
      "epoch": 29.19103037184218,
      "grad_norm": 0.08693257719278336,
      "learning_rate": 7.082174283281294e-05,
      "loss": 0.0065846554934978485,
      "step": 102840
    },
    {
      "epoch": 29.19386886176554,
      "grad_norm": 0.5288121700286865,
      "learning_rate": 7.08189043428896e-05,
      "loss": 0.0033392634242773054,
      "step": 102850
    },
    {
      "epoch": 29.1967073516889,
      "grad_norm": 0.5403031706809998,
      "learning_rate": 7.081606585296624e-05,
      "loss": 0.003260399401187897,
      "step": 102860
    },
    {
      "epoch": 29.199545841612263,
      "grad_norm": 0.14243869483470917,
      "learning_rate": 7.081322736304286e-05,
      "loss": 0.012542164325714112,
      "step": 102870
    },
    {
      "epoch": 29.202384331535622,
      "grad_norm": 0.8105551600456238,
      "learning_rate": 7.08103888731195e-05,
      "loss": 0.003414628654718399,
      "step": 102880
    },
    {
      "epoch": 29.205222821458985,
      "grad_norm": 3.517125368118286,
      "learning_rate": 7.080755038319615e-05,
      "loss": 0.0031416021287441255,
      "step": 102890
    },
    {
      "epoch": 29.208061311382345,
      "grad_norm": 0.46577849984169006,
      "learning_rate": 7.080471189327277e-05,
      "loss": 0.014421993494033813,
      "step": 102900
    },
    {
      "epoch": 29.210899801305704,
      "grad_norm": 14.920608520507812,
      "learning_rate": 7.080187340334942e-05,
      "loss": 0.006376342475414276,
      "step": 102910
    },
    {
      "epoch": 29.213738291229067,
      "grad_norm": 0.6095016598701477,
      "learning_rate": 7.079903491342607e-05,
      "loss": 0.004641420394182205,
      "step": 102920
    },
    {
      "epoch": 29.216576781152426,
      "grad_norm": 0.8011102676391602,
      "learning_rate": 7.07961964235027e-05,
      "loss": 0.0020334839820861815,
      "step": 102930
    },
    {
      "epoch": 29.21941527107579,
      "grad_norm": 1.759316086769104,
      "learning_rate": 7.079335793357934e-05,
      "loss": 0.0031461603939533234,
      "step": 102940
    },
    {
      "epoch": 29.22225376099915,
      "grad_norm": 1.7915518283843994,
      "learning_rate": 7.079051944365598e-05,
      "loss": 0.0028503136709332466,
      "step": 102950
    },
    {
      "epoch": 29.225092250922508,
      "grad_norm": 0.14384134113788605,
      "learning_rate": 7.078768095373262e-05,
      "loss": 0.002985171414911747,
      "step": 102960
    },
    {
      "epoch": 29.22793074084587,
      "grad_norm": 1.150887370109558,
      "learning_rate": 7.078484246380925e-05,
      "loss": 0.02281262129545212,
      "step": 102970
    },
    {
      "epoch": 29.23076923076923,
      "grad_norm": 0.2520335912704468,
      "learning_rate": 7.078200397388589e-05,
      "loss": 0.00878232643008232,
      "step": 102980
    },
    {
      "epoch": 29.233607720692593,
      "grad_norm": 0.08101112395524979,
      "learning_rate": 7.077916548396253e-05,
      "loss": 0.006850749254226685,
      "step": 102990
    },
    {
      "epoch": 29.236446210615952,
      "grad_norm": 0.7112416625022888,
      "learning_rate": 7.077632699403917e-05,
      "loss": 0.01065071001648903,
      "step": 103000
    },
    {
      "epoch": 29.236446210615952,
      "eval_accuracy": 0.9721498060660011,
      "eval_loss": 0.09248412400484085,
      "eval_runtime": 31.51,
      "eval_samples_per_second": 499.112,
      "eval_steps_per_second": 7.807,
      "step": 103000
    },
    {
      "epoch": 29.23928470053931,
      "grad_norm": 0.5933818221092224,
      "learning_rate": 7.077348850411582e-05,
      "loss": 0.007601629197597504,
      "step": 103010
    },
    {
      "epoch": 29.242123190462674,
      "grad_norm": 1.4413172006607056,
      "learning_rate": 7.077065001419246e-05,
      "loss": 0.0027877679094672203,
      "step": 103020
    },
    {
      "epoch": 29.244961680386034,
      "grad_norm": 3.930833339691162,
      "learning_rate": 7.076781152426908e-05,
      "loss": 0.01581648588180542,
      "step": 103030
    },
    {
      "epoch": 29.247800170309397,
      "grad_norm": 1.752811312675476,
      "learning_rate": 7.076497303434573e-05,
      "loss": 0.008138402551412582,
      "step": 103040
    },
    {
      "epoch": 29.250638660232756,
      "grad_norm": 4.996903896331787,
      "learning_rate": 7.076213454442238e-05,
      "loss": 0.009723015129566193,
      "step": 103050
    },
    {
      "epoch": 29.253477150156115,
      "grad_norm": 2.8598484992980957,
      "learning_rate": 7.075929605449901e-05,
      "loss": 0.0066393807530403135,
      "step": 103060
    },
    {
      "epoch": 29.256315640079478,
      "grad_norm": 2.6949291229248047,
      "learning_rate": 7.075645756457565e-05,
      "loss": 0.006281520426273346,
      "step": 103070
    },
    {
      "epoch": 29.259154130002837,
      "grad_norm": 0.902625560760498,
      "learning_rate": 7.075361907465229e-05,
      "loss": 0.006649890542030334,
      "step": 103080
    },
    {
      "epoch": 29.2619926199262,
      "grad_norm": 2.1878669261932373,
      "learning_rate": 7.075078058472892e-05,
      "loss": 0.01337079256772995,
      "step": 103090
    },
    {
      "epoch": 29.26483110984956,
      "grad_norm": 5.5928874015808105,
      "learning_rate": 7.074794209480556e-05,
      "loss": 0.013811884820461274,
      "step": 103100
    },
    {
      "epoch": 29.267669599772923,
      "grad_norm": 1.51506507396698,
      "learning_rate": 7.07451036048822e-05,
      "loss": 0.004733823239803314,
      "step": 103110
    },
    {
      "epoch": 29.270508089696282,
      "grad_norm": 0.5004401803016663,
      "learning_rate": 7.074226511495884e-05,
      "loss": 0.0018916735425591469,
      "step": 103120
    },
    {
      "epoch": 29.27334657961964,
      "grad_norm": 0.3076229989528656,
      "learning_rate": 7.073942662503548e-05,
      "loss": 0.00203386265784502,
      "step": 103130
    },
    {
      "epoch": 29.276185069543004,
      "grad_norm": 0.6887105107307434,
      "learning_rate": 7.073658813511213e-05,
      "loss": 0.005257030203938484,
      "step": 103140
    },
    {
      "epoch": 29.279023559466363,
      "grad_norm": 0.15275152027606964,
      "learning_rate": 7.073374964518877e-05,
      "loss": 0.006458427757024765,
      "step": 103150
    },
    {
      "epoch": 29.281862049389726,
      "grad_norm": 1.6810303926467896,
      "learning_rate": 7.07309111552654e-05,
      "loss": 0.005936475098133087,
      "step": 103160
    },
    {
      "epoch": 29.284700539313086,
      "grad_norm": 0.28178727626800537,
      "learning_rate": 7.072807266534204e-05,
      "loss": 0.0045090243220329285,
      "step": 103170
    },
    {
      "epoch": 29.287539029236445,
      "grad_norm": 4.965667724609375,
      "learning_rate": 7.072523417541868e-05,
      "loss": 0.007151518762111664,
      "step": 103180
    },
    {
      "epoch": 29.290377519159808,
      "grad_norm": 2.882051944732666,
      "learning_rate": 7.072239568549532e-05,
      "loss": 0.004828662425279617,
      "step": 103190
    },
    {
      "epoch": 29.293216009083167,
      "grad_norm": 2.061706781387329,
      "learning_rate": 7.071955719557196e-05,
      "loss": 0.003097831830382347,
      "step": 103200
    },
    {
      "epoch": 29.29605449900653,
      "grad_norm": 2.1179275512695312,
      "learning_rate": 7.07167187056486e-05,
      "loss": 0.0065132036805152895,
      "step": 103210
    },
    {
      "epoch": 29.29889298892989,
      "grad_norm": 0.9824199080467224,
      "learning_rate": 7.071388021572523e-05,
      "loss": 0.0025570323690772057,
      "step": 103220
    },
    {
      "epoch": 29.30173147885325,
      "grad_norm": 0.4234238266944885,
      "learning_rate": 7.071104172580187e-05,
      "loss": 0.006880425661802292,
      "step": 103230
    },
    {
      "epoch": 29.30456996877661,
      "grad_norm": 1.14798903465271,
      "learning_rate": 7.070820323587851e-05,
      "loss": 0.013638554513454438,
      "step": 103240
    },
    {
      "epoch": 29.30740845869997,
      "grad_norm": 10.852538108825684,
      "learning_rate": 7.070536474595515e-05,
      "loss": 0.00906771793961525,
      "step": 103250
    },
    {
      "epoch": 29.310246948623334,
      "grad_norm": 3.0900282859802246,
      "learning_rate": 7.07025262560318e-05,
      "loss": 0.0029064876958727836,
      "step": 103260
    },
    {
      "epoch": 29.313085438546693,
      "grad_norm": 0.25750496983528137,
      "learning_rate": 7.069968776610844e-05,
      "loss": 0.022493073344230653,
      "step": 103270
    },
    {
      "epoch": 29.315923928470053,
      "grad_norm": 0.046320654451847076,
      "learning_rate": 7.069684927618508e-05,
      "loss": 0.007278059422969818,
      "step": 103280
    },
    {
      "epoch": 29.318762418393415,
      "grad_norm": 6.1847686767578125,
      "learning_rate": 7.06940107862617e-05,
      "loss": 0.015126746892929078,
      "step": 103290
    },
    {
      "epoch": 29.321600908316775,
      "grad_norm": 0.7371032238006592,
      "learning_rate": 7.069117229633835e-05,
      "loss": 0.008068636059761047,
      "step": 103300
    },
    {
      "epoch": 29.324439398240138,
      "grad_norm": 0.09260711818933487,
      "learning_rate": 7.068833380641499e-05,
      "loss": 0.005408328026533127,
      "step": 103310
    },
    {
      "epoch": 29.327277888163497,
      "grad_norm": 0.9270020127296448,
      "learning_rate": 7.068549531649163e-05,
      "loss": 0.007327474653720856,
      "step": 103320
    },
    {
      "epoch": 29.330116378086856,
      "grad_norm": 0.679845929145813,
      "learning_rate": 7.068265682656827e-05,
      "loss": 0.0036801610141992567,
      "step": 103330
    },
    {
      "epoch": 29.33295486801022,
      "grad_norm": 4.866535663604736,
      "learning_rate": 7.067981833664491e-05,
      "loss": 0.004996693879365921,
      "step": 103340
    },
    {
      "epoch": 29.33579335793358,
      "grad_norm": 0.5973851680755615,
      "learning_rate": 7.067697984672154e-05,
      "loss": 0.010375338047742844,
      "step": 103350
    },
    {
      "epoch": 29.33863184785694,
      "grad_norm": 0.4207924008369446,
      "learning_rate": 7.067414135679818e-05,
      "loss": 0.0074117623269557955,
      "step": 103360
    },
    {
      "epoch": 29.3414703377803,
      "grad_norm": 2.033496141433716,
      "learning_rate": 7.067130286687482e-05,
      "loss": 0.0037599004805088042,
      "step": 103370
    },
    {
      "epoch": 29.34430882770366,
      "grad_norm": 0.33192357420921326,
      "learning_rate": 7.066846437695147e-05,
      "loss": 0.004513479769229889,
      "step": 103380
    },
    {
      "epoch": 29.347147317627023,
      "grad_norm": 10.976330757141113,
      "learning_rate": 7.06656258870281e-05,
      "loss": 0.006702899187803269,
      "step": 103390
    },
    {
      "epoch": 29.349985807550382,
      "grad_norm": 0.4565625488758087,
      "learning_rate": 7.066278739710475e-05,
      "loss": 0.0025810420513153074,
      "step": 103400
    },
    {
      "epoch": 29.352824297473745,
      "grad_norm": 1.2145016193389893,
      "learning_rate": 7.065994890718139e-05,
      "loss": 0.003390244394540787,
      "step": 103410
    },
    {
      "epoch": 29.355662787397105,
      "grad_norm": 1.3472965955734253,
      "learning_rate": 7.065711041725802e-05,
      "loss": 0.009109169244766235,
      "step": 103420
    },
    {
      "epoch": 29.358501277320464,
      "grad_norm": 2.320267677307129,
      "learning_rate": 7.065427192733466e-05,
      "loss": 0.012889544665813445,
      "step": 103430
    },
    {
      "epoch": 29.361339767243827,
      "grad_norm": 1.5763037204742432,
      "learning_rate": 7.06514334374113e-05,
      "loss": 0.011335603892803192,
      "step": 103440
    },
    {
      "epoch": 29.364178257167186,
      "grad_norm": 0.46581053733825684,
      "learning_rate": 7.064859494748794e-05,
      "loss": 0.011664146184921264,
      "step": 103450
    },
    {
      "epoch": 29.36701674709055,
      "grad_norm": 0.3456219732761383,
      "learning_rate": 7.064575645756458e-05,
      "loss": 0.0025394342839717863,
      "step": 103460
    },
    {
      "epoch": 29.36985523701391,
      "grad_norm": 0.09108962118625641,
      "learning_rate": 7.064291796764122e-05,
      "loss": 0.009935784339904784,
      "step": 103470
    },
    {
      "epoch": 29.372693726937268,
      "grad_norm": 3.5361814498901367,
      "learning_rate": 7.064007947771785e-05,
      "loss": 0.009024163335561752,
      "step": 103480
    },
    {
      "epoch": 29.37553221686063,
      "grad_norm": 0.19263404607772827,
      "learning_rate": 7.063724098779449e-05,
      "loss": 0.011487459391355514,
      "step": 103490
    },
    {
      "epoch": 29.37837070678399,
      "grad_norm": 0.8193291425704956,
      "learning_rate": 7.063440249787113e-05,
      "loss": 0.0118890643119812,
      "step": 103500
    },
    {
      "epoch": 29.37837070678399,
      "eval_accuracy": 0.9710052775481656,
      "eval_loss": 0.10009734332561493,
      "eval_runtime": 32.1347,
      "eval_samples_per_second": 489.408,
      "eval_steps_per_second": 7.655,
      "step": 103500
    },
    {
      "epoch": 29.381209196707353,
      "grad_norm": 6.362823963165283,
      "learning_rate": 7.063156400794778e-05,
      "loss": 0.006814927607774734,
      "step": 103510
    },
    {
      "epoch": 29.384047686630712,
      "grad_norm": 1.7562648057937622,
      "learning_rate": 7.062872551802442e-05,
      "loss": 0.005839845538139344,
      "step": 103520
    },
    {
      "epoch": 29.386886176554075,
      "grad_norm": 0.15425477921962738,
      "learning_rate": 7.062588702810106e-05,
      "loss": 0.004590987414121628,
      "step": 103530
    },
    {
      "epoch": 29.389724666477434,
      "grad_norm": 0.7263097167015076,
      "learning_rate": 7.06230485381777e-05,
      "loss": 0.010863494873046876,
      "step": 103540
    },
    {
      "epoch": 29.392563156400794,
      "grad_norm": 0.7492151260375977,
      "learning_rate": 7.062021004825433e-05,
      "loss": 0.007868213206529617,
      "step": 103550
    },
    {
      "epoch": 29.395401646324157,
      "grad_norm": 12.324872016906738,
      "learning_rate": 7.061737155833097e-05,
      "loss": 0.012037272751331329,
      "step": 103560
    },
    {
      "epoch": 29.398240136247516,
      "grad_norm": 2.4293723106384277,
      "learning_rate": 7.061453306840761e-05,
      "loss": 0.01658853143453598,
      "step": 103570
    },
    {
      "epoch": 29.40107862617088,
      "grad_norm": 0.14804920554161072,
      "learning_rate": 7.061169457848424e-05,
      "loss": 0.009303011000156403,
      "step": 103580
    },
    {
      "epoch": 29.403917116094238,
      "grad_norm": 0.9840238094329834,
      "learning_rate": 7.06088560885609e-05,
      "loss": 0.004961763694882393,
      "step": 103590
    },
    {
      "epoch": 29.406755606017597,
      "grad_norm": 11.881843566894531,
      "learning_rate": 7.060601759863753e-05,
      "loss": 0.0165410652756691,
      "step": 103600
    },
    {
      "epoch": 29.40959409594096,
      "grad_norm": 0.881999135017395,
      "learning_rate": 7.060317910871416e-05,
      "loss": 0.004866021871566773,
      "step": 103610
    },
    {
      "epoch": 29.41243258586432,
      "grad_norm": 1.2781749963760376,
      "learning_rate": 7.06003406187908e-05,
      "loss": 0.005559352785348892,
      "step": 103620
    },
    {
      "epoch": 29.415271075787683,
      "grad_norm": 0.5381531119346619,
      "learning_rate": 7.059750212886745e-05,
      "loss": 0.018800212442874907,
      "step": 103630
    },
    {
      "epoch": 29.418109565711042,
      "grad_norm": 0.5681717395782471,
      "learning_rate": 7.059466363894409e-05,
      "loss": 0.009892986714839935,
      "step": 103640
    },
    {
      "epoch": 29.4209480556344,
      "grad_norm": 1.7043876647949219,
      "learning_rate": 7.059182514902073e-05,
      "loss": 0.004879165068268776,
      "step": 103650
    },
    {
      "epoch": 29.423786545557764,
      "grad_norm": 0.06515082716941833,
      "learning_rate": 7.058898665909737e-05,
      "loss": 0.006353899836540222,
      "step": 103660
    },
    {
      "epoch": 29.426625035481123,
      "grad_norm": 9.811002731323242,
      "learning_rate": 7.058614816917401e-05,
      "loss": 0.008224469423294068,
      "step": 103670
    },
    {
      "epoch": 29.429463525404486,
      "grad_norm": 0.1130213513970375,
      "learning_rate": 7.058330967925064e-05,
      "loss": 0.0040574237704277035,
      "step": 103680
    },
    {
      "epoch": 29.432302015327846,
      "grad_norm": 0.4912196397781372,
      "learning_rate": 7.058047118932728e-05,
      "loss": 0.00849318653345108,
      "step": 103690
    },
    {
      "epoch": 29.435140505251205,
      "grad_norm": 0.9292280077934265,
      "learning_rate": 7.057763269940392e-05,
      "loss": 0.0053595930337905886,
      "step": 103700
    },
    {
      "epoch": 29.437978995174568,
      "grad_norm": 0.028805045410990715,
      "learning_rate": 7.057479420948055e-05,
      "loss": 0.004971916973590851,
      "step": 103710
    },
    {
      "epoch": 29.440817485097927,
      "grad_norm": 4.949158668518066,
      "learning_rate": 7.05719557195572e-05,
      "loss": 0.010285651683807373,
      "step": 103720
    },
    {
      "epoch": 29.44365597502129,
      "grad_norm": 2.8312160968780518,
      "learning_rate": 7.056911722963385e-05,
      "loss": 0.01690831035375595,
      "step": 103730
    },
    {
      "epoch": 29.44649446494465,
      "grad_norm": 1.025990605354309,
      "learning_rate": 7.056627873971047e-05,
      "loss": 0.005609037727117539,
      "step": 103740
    },
    {
      "epoch": 29.44933295486801,
      "grad_norm": 0.23950712382793427,
      "learning_rate": 7.056344024978711e-05,
      "loss": 0.0020457122474908827,
      "step": 103750
    },
    {
      "epoch": 29.45217144479137,
      "grad_norm": 0.6270509362220764,
      "learning_rate": 7.056060175986376e-05,
      "loss": 0.005263065919280052,
      "step": 103760
    },
    {
      "epoch": 29.45500993471473,
      "grad_norm": 18.475805282592773,
      "learning_rate": 7.05577632699404e-05,
      "loss": 0.03969619572162628,
      "step": 103770
    },
    {
      "epoch": 29.457848424638094,
      "grad_norm": 1.099635362625122,
      "learning_rate": 7.055492478001704e-05,
      "loss": 0.03034897744655609,
      "step": 103780
    },
    {
      "epoch": 29.460686914561453,
      "grad_norm": 12.22894287109375,
      "learning_rate": 7.055208629009368e-05,
      "loss": 0.02304292768239975,
      "step": 103790
    },
    {
      "epoch": 29.463525404484812,
      "grad_norm": 1.35223388671875,
      "learning_rate": 7.054924780017032e-05,
      "loss": 0.01624814122915268,
      "step": 103800
    },
    {
      "epoch": 29.466363894408175,
      "grad_norm": 1.2336528301239014,
      "learning_rate": 7.054640931024695e-05,
      "loss": 0.005838661640882492,
      "step": 103810
    },
    {
      "epoch": 29.469202384331535,
      "grad_norm": 5.567005157470703,
      "learning_rate": 7.054357082032359e-05,
      "loss": 0.011862041056156158,
      "step": 103820
    },
    {
      "epoch": 29.472040874254898,
      "grad_norm": 12.38755989074707,
      "learning_rate": 7.054073233040023e-05,
      "loss": 0.005309949815273285,
      "step": 103830
    },
    {
      "epoch": 29.474879364178257,
      "grad_norm": 0.29447826743125916,
      "learning_rate": 7.053789384047686e-05,
      "loss": 0.001903078705072403,
      "step": 103840
    },
    {
      "epoch": 29.477717854101616,
      "grad_norm": 0.08942021429538727,
      "learning_rate": 7.053505535055351e-05,
      "loss": 0.00615214891731739,
      "step": 103850
    },
    {
      "epoch": 29.48055634402498,
      "grad_norm": 0.8542690277099609,
      "learning_rate": 7.053221686063016e-05,
      "loss": 0.012230338156223297,
      "step": 103860
    },
    {
      "epoch": 29.48339483394834,
      "grad_norm": 15.261199951171875,
      "learning_rate": 7.052937837070678e-05,
      "loss": 0.006397219002246856,
      "step": 103870
    },
    {
      "epoch": 29.4862333238717,
      "grad_norm": 1.796958088874817,
      "learning_rate": 7.052653988078343e-05,
      "loss": 0.007944531738758087,
      "step": 103880
    },
    {
      "epoch": 29.48907181379506,
      "grad_norm": 0.5633137226104736,
      "learning_rate": 7.052370139086007e-05,
      "loss": 0.002471035346388817,
      "step": 103890
    },
    {
      "epoch": 29.49191030371842,
      "grad_norm": 0.9292938113212585,
      "learning_rate": 7.052086290093671e-05,
      "loss": 0.013493633270263672,
      "step": 103900
    },
    {
      "epoch": 29.494748793641783,
      "grad_norm": 1.0036776065826416,
      "learning_rate": 7.051802441101334e-05,
      "loss": 0.007279355823993683,
      "step": 103910
    },
    {
      "epoch": 29.497587283565142,
      "grad_norm": 1.0998727083206177,
      "learning_rate": 7.051518592108999e-05,
      "loss": 0.009480812400579453,
      "step": 103920
    },
    {
      "epoch": 29.500425773488505,
      "grad_norm": 0.9005405306816101,
      "learning_rate": 7.051234743116662e-05,
      "loss": 0.009021732211112975,
      "step": 103930
    },
    {
      "epoch": 29.503264263411864,
      "grad_norm": 0.09481550753116608,
      "learning_rate": 7.050950894124326e-05,
      "loss": 0.009876537322998046,
      "step": 103940
    },
    {
      "epoch": 29.506102753335227,
      "grad_norm": 0.43748241662979126,
      "learning_rate": 7.05066704513199e-05,
      "loss": 0.006156052649021149,
      "step": 103950
    },
    {
      "epoch": 29.508941243258587,
      "grad_norm": 2.789482831954956,
      "learning_rate": 7.050383196139654e-05,
      "loss": 0.005762055516242981,
      "step": 103960
    },
    {
      "epoch": 29.511779733181946,
      "grad_norm": 0.4689555764198303,
      "learning_rate": 7.050099347147317e-05,
      "loss": 0.006935521215200424,
      "step": 103970
    },
    {
      "epoch": 29.51461822310531,
      "grad_norm": 2.5231363773345947,
      "learning_rate": 7.049815498154983e-05,
      "loss": 0.0020012497901916505,
      "step": 103980
    },
    {
      "epoch": 29.51745671302867,
      "grad_norm": 0.8395127058029175,
      "learning_rate": 7.049531649162647e-05,
      "loss": 0.004673022404313087,
      "step": 103990
    },
    {
      "epoch": 29.52029520295203,
      "grad_norm": 0.774398148059845,
      "learning_rate": 7.04924780017031e-05,
      "loss": 0.0021633140742778776,
      "step": 104000
    },
    {
      "epoch": 29.52029520295203,
      "eval_accuracy": 0.9720226362306861,
      "eval_loss": 0.09336931258440018,
      "eval_runtime": 32.6696,
      "eval_samples_per_second": 481.395,
      "eval_steps_per_second": 7.53,
      "step": 104000
    },
    {
      "epoch": 29.52313369287539,
      "grad_norm": 0.10275911539793015,
      "learning_rate": 7.048963951177974e-05,
      "loss": 0.0031437806785106657,
      "step": 104010
    },
    {
      "epoch": 29.52597218279875,
      "grad_norm": 0.6999396681785583,
      "learning_rate": 7.048680102185638e-05,
      "loss": 0.007735978811979294,
      "step": 104020
    },
    {
      "epoch": 29.528810672722113,
      "grad_norm": 0.47529640793800354,
      "learning_rate": 7.0483962531933e-05,
      "loss": 0.0032308816909790037,
      "step": 104030
    },
    {
      "epoch": 29.531649162645472,
      "grad_norm": 0.5819925665855408,
      "learning_rate": 7.048112404200965e-05,
      "loss": 0.008918210864067078,
      "step": 104040
    },
    {
      "epoch": 29.534487652568835,
      "grad_norm": 9.107738494873047,
      "learning_rate": 7.04782855520863e-05,
      "loss": 0.015941126644611357,
      "step": 104050
    },
    {
      "epoch": 29.537326142492194,
      "grad_norm": 8.273719787597656,
      "learning_rate": 7.047544706216293e-05,
      "loss": 0.011851529031991959,
      "step": 104060
    },
    {
      "epoch": 29.540164632415554,
      "grad_norm": 0.35645440220832825,
      "learning_rate": 7.047260857223957e-05,
      "loss": 0.011782968789339066,
      "step": 104070
    },
    {
      "epoch": 29.543003122338916,
      "grad_norm": 1.1897735595703125,
      "learning_rate": 7.046977008231621e-05,
      "loss": 0.012898813188076019,
      "step": 104080
    },
    {
      "epoch": 29.545841612262276,
      "grad_norm": 1.441650390625,
      "learning_rate": 7.046721544138518e-05,
      "loss": 0.00846182107925415,
      "step": 104090
    },
    {
      "epoch": 29.54868010218564,
      "grad_norm": 10.010531425476074,
      "learning_rate": 7.046437695146183e-05,
      "loss": 0.011055965721607209,
      "step": 104100
    },
    {
      "epoch": 29.551518592108998,
      "grad_norm": 0.5085257291793823,
      "learning_rate": 7.046153846153846e-05,
      "loss": 0.012912529706954955,
      "step": 104110
    },
    {
      "epoch": 29.554357082032357,
      "grad_norm": 0.1733914464712143,
      "learning_rate": 7.04586999716151e-05,
      "loss": 0.006072824448347091,
      "step": 104120
    },
    {
      "epoch": 29.55719557195572,
      "grad_norm": 8.602968215942383,
      "learning_rate": 7.045586148169174e-05,
      "loss": 0.007088814675807953,
      "step": 104130
    },
    {
      "epoch": 29.56003406187908,
      "grad_norm": 0.20394794642925262,
      "learning_rate": 7.045302299176839e-05,
      "loss": 0.007820235937833786,
      "step": 104140
    },
    {
      "epoch": 29.562872551802442,
      "grad_norm": 0.2930407226085663,
      "learning_rate": 7.045018450184501e-05,
      "loss": 0.02364852875471115,
      "step": 104150
    },
    {
      "epoch": 29.565711041725802,
      "grad_norm": 0.12931327521800995,
      "learning_rate": 7.044734601192165e-05,
      "loss": 0.0038503125309944154,
      "step": 104160
    },
    {
      "epoch": 29.56854953164916,
      "grad_norm": 2.476289987564087,
      "learning_rate": 7.044450752199831e-05,
      "loss": 0.0060837332159280775,
      "step": 104170
    },
    {
      "epoch": 29.571388021572524,
      "grad_norm": 1.8453221321105957,
      "learning_rate": 7.044166903207494e-05,
      "loss": 0.009723147749900818,
      "step": 104180
    },
    {
      "epoch": 29.574226511495883,
      "grad_norm": 0.4030969738960266,
      "learning_rate": 7.043883054215158e-05,
      "loss": 0.021486718952655793,
      "step": 104190
    },
    {
      "epoch": 29.577065001419246,
      "grad_norm": 2.210861921310425,
      "learning_rate": 7.043599205222822e-05,
      "loss": 0.0045363172888755795,
      "step": 104200
    },
    {
      "epoch": 29.579903491342606,
      "grad_norm": 0.9037960171699524,
      "learning_rate": 7.043315356230485e-05,
      "loss": 0.01242876723408699,
      "step": 104210
    },
    {
      "epoch": 29.582741981265965,
      "grad_norm": 0.7394652962684631,
      "learning_rate": 7.043031507238149e-05,
      "loss": 0.0073909670114517215,
      "step": 104220
    },
    {
      "epoch": 29.585580471189328,
      "grad_norm": 0.026567431166768074,
      "learning_rate": 7.042747658245814e-05,
      "loss": 0.0064466923475265505,
      "step": 104230
    },
    {
      "epoch": 29.588418961112687,
      "grad_norm": 3.084294319152832,
      "learning_rate": 7.042463809253477e-05,
      "loss": 0.005016718432307244,
      "step": 104240
    },
    {
      "epoch": 29.59125745103605,
      "grad_norm": 6.4906229972839355,
      "learning_rate": 7.042179960261141e-05,
      "loss": 0.015404963493347168,
      "step": 104250
    },
    {
      "epoch": 29.59409594095941,
      "grad_norm": 12.337773323059082,
      "learning_rate": 7.041896111268805e-05,
      "loss": 0.017842201888561247,
      "step": 104260
    },
    {
      "epoch": 29.59693443088277,
      "grad_norm": 0.3310418725013733,
      "learning_rate": 7.04161226227647e-05,
      "loss": 0.004060607403516769,
      "step": 104270
    },
    {
      "epoch": 29.59977292080613,
      "grad_norm": 0.49323394894599915,
      "learning_rate": 7.041328413284132e-05,
      "loss": 0.018381719291210175,
      "step": 104280
    },
    {
      "epoch": 29.60261141072949,
      "grad_norm": 0.13789761066436768,
      "learning_rate": 7.041044564291797e-05,
      "loss": 0.004696869850158691,
      "step": 104290
    },
    {
      "epoch": 29.605449900652854,
      "grad_norm": 0.1203417032957077,
      "learning_rate": 7.040760715299462e-05,
      "loss": 0.0034710045903921126,
      "step": 104300
    },
    {
      "epoch": 29.608288390576213,
      "grad_norm": 2.6133992671966553,
      "learning_rate": 7.040476866307125e-05,
      "loss": 0.005593504756689072,
      "step": 104310
    },
    {
      "epoch": 29.611126880499576,
      "grad_norm": 0.10324645787477493,
      "learning_rate": 7.040193017314789e-05,
      "loss": 0.0038324929773807526,
      "step": 104320
    },
    {
      "epoch": 29.613965370422935,
      "grad_norm": 5.347020626068115,
      "learning_rate": 7.039909168322453e-05,
      "loss": 0.004506857320666313,
      "step": 104330
    },
    {
      "epoch": 29.616803860346295,
      "grad_norm": 0.5524942278862,
      "learning_rate": 7.039625319330116e-05,
      "loss": 0.006664811819791794,
      "step": 104340
    },
    {
      "epoch": 29.619642350269658,
      "grad_norm": 9.748819351196289,
      "learning_rate": 7.03934147033778e-05,
      "loss": 0.010068503767251968,
      "step": 104350
    },
    {
      "epoch": 29.622480840193017,
      "grad_norm": 1.6839975118637085,
      "learning_rate": 7.039057621345446e-05,
      "loss": 0.005738165974617004,
      "step": 104360
    },
    {
      "epoch": 29.62531933011638,
      "grad_norm": 0.795344889163971,
      "learning_rate": 7.038773772353108e-05,
      "loss": 0.006378365308046341,
      "step": 104370
    },
    {
      "epoch": 29.62815782003974,
      "grad_norm": 0.8398072123527527,
      "learning_rate": 7.038489923360772e-05,
      "loss": 0.01780163198709488,
      "step": 104380
    },
    {
      "epoch": 29.6309963099631,
      "grad_norm": 0.21247723698616028,
      "learning_rate": 7.038206074368437e-05,
      "loss": 0.02032453417778015,
      "step": 104390
    },
    {
      "epoch": 29.63383479988646,
      "grad_norm": 0.08152678608894348,
      "learning_rate": 7.037922225376101e-05,
      "loss": 0.0023228060454130173,
      "step": 104400
    },
    {
      "epoch": 29.63667328980982,
      "grad_norm": 0.2128494381904602,
      "learning_rate": 7.037638376383763e-05,
      "loss": 0.004381305724382401,
      "step": 104410
    },
    {
      "epoch": 29.639511779733184,
      "grad_norm": 1.3016115427017212,
      "learning_rate": 7.037354527391428e-05,
      "loss": 0.015048427879810334,
      "step": 104420
    },
    {
      "epoch": 29.642350269656543,
      "grad_norm": 1.2149841785430908,
      "learning_rate": 7.037070678399093e-05,
      "loss": 0.009348060935735703,
      "step": 104430
    },
    {
      "epoch": 29.645188759579902,
      "grad_norm": 8.566803932189941,
      "learning_rate": 7.036786829406756e-05,
      "loss": 0.005944855511188507,
      "step": 104440
    },
    {
      "epoch": 29.648027249503265,
      "grad_norm": 2.476390838623047,
      "learning_rate": 7.03650298041442e-05,
      "loss": 0.002931314334273338,
      "step": 104450
    },
    {
      "epoch": 29.650865739426624,
      "grad_norm": 19.095195770263672,
      "learning_rate": 7.036219131422084e-05,
      "loss": 0.02069008946418762,
      "step": 104460
    },
    {
      "epoch": 29.653704229349987,
      "grad_norm": 8.009967803955078,
      "learning_rate": 7.035935282429747e-05,
      "loss": 0.0062904275953769686,
      "step": 104470
    },
    {
      "epoch": 29.656542719273347,
      "grad_norm": 0.07890402525663376,
      "learning_rate": 7.035651433437411e-05,
      "loss": 0.007290425151586533,
      "step": 104480
    },
    {
      "epoch": 29.659381209196706,
      "grad_norm": 11.004659652709961,
      "learning_rate": 7.035367584445075e-05,
      "loss": 0.005387300997972489,
      "step": 104490
    },
    {
      "epoch": 29.66221969912007,
      "grad_norm": 3.862241268157959,
      "learning_rate": 7.03508373545274e-05,
      "loss": 0.0100385382771492,
      "step": 104500
    },
    {
      "epoch": 29.66221969912007,
      "eval_accuracy": 0.9682711260888918,
      "eval_loss": 0.10713754594326019,
      "eval_runtime": 31.8517,
      "eval_samples_per_second": 493.756,
      "eval_steps_per_second": 7.723,
      "step": 104500
    },
    {
      "epoch": 29.665058189043428,
      "grad_norm": 1.5849838256835938,
      "learning_rate": 7.034799886460404e-05,
      "loss": 0.011078003048896789,
      "step": 104510
    },
    {
      "epoch": 29.66789667896679,
      "grad_norm": 0.14465877413749695,
      "learning_rate": 7.034516037468068e-05,
      "loss": 0.00287940688431263,
      "step": 104520
    },
    {
      "epoch": 29.67073516889015,
      "grad_norm": 15.591758728027344,
      "learning_rate": 7.034232188475732e-05,
      "loss": 0.015643219649791717,
      "step": 104530
    },
    {
      "epoch": 29.67357365881351,
      "grad_norm": 0.7701877951622009,
      "learning_rate": 7.033948339483395e-05,
      "loss": 0.011083225905895232,
      "step": 104540
    },
    {
      "epoch": 29.676412148736873,
      "grad_norm": 1.5144898891448975,
      "learning_rate": 7.033664490491059e-05,
      "loss": 0.02409253120422363,
      "step": 104550
    },
    {
      "epoch": 29.679250638660232,
      "grad_norm": 0.06606651097536087,
      "learning_rate": 7.033380641498724e-05,
      "loss": 0.005826719850301742,
      "step": 104560
    },
    {
      "epoch": 29.682089128583595,
      "grad_norm": 0.249341681599617,
      "learning_rate": 7.033096792506387e-05,
      "loss": 0.0038238178938627245,
      "step": 104570
    },
    {
      "epoch": 29.684927618506954,
      "grad_norm": 10.029897689819336,
      "learning_rate": 7.032812943514051e-05,
      "loss": 0.005377654731273651,
      "step": 104580
    },
    {
      "epoch": 29.687766108430313,
      "grad_norm": 8.690874099731445,
      "learning_rate": 7.032529094521715e-05,
      "loss": 0.008487799018621445,
      "step": 104590
    },
    {
      "epoch": 29.690604598353676,
      "grad_norm": 1.082100510597229,
      "learning_rate": 7.032245245529378e-05,
      "loss": 0.00240826103836298,
      "step": 104600
    },
    {
      "epoch": 29.693443088277036,
      "grad_norm": 4.237621307373047,
      "learning_rate": 7.031961396537042e-05,
      "loss": 0.00365162156522274,
      "step": 104610
    },
    {
      "epoch": 29.6962815782004,
      "grad_norm": 5.967831611633301,
      "learning_rate": 7.031677547544706e-05,
      "loss": 0.021129617094993593,
      "step": 104620
    },
    {
      "epoch": 29.699120068123758,
      "grad_norm": 0.21645548939704895,
      "learning_rate": 7.03139369855237e-05,
      "loss": 0.00919828712940216,
      "step": 104630
    },
    {
      "epoch": 29.701958558047117,
      "grad_norm": 10.868873596191406,
      "learning_rate": 7.031109849560035e-05,
      "loss": 0.01018485650420189,
      "step": 104640
    },
    {
      "epoch": 29.70479704797048,
      "grad_norm": 0.10665369033813477,
      "learning_rate": 7.030826000567699e-05,
      "loss": 0.006110059097409248,
      "step": 104650
    },
    {
      "epoch": 29.70763553789384,
      "grad_norm": 4.778646469116211,
      "learning_rate": 7.030542151575363e-05,
      "loss": 0.014721235632896424,
      "step": 104660
    },
    {
      "epoch": 29.710474027817202,
      "grad_norm": 0.9279530644416809,
      "learning_rate": 7.030258302583026e-05,
      "loss": 0.0019113093614578248,
      "step": 104670
    },
    {
      "epoch": 29.71331251774056,
      "grad_norm": 0.3460279405117035,
      "learning_rate": 7.02997445359069e-05,
      "loss": 0.004975598677992821,
      "step": 104680
    },
    {
      "epoch": 29.716151007663925,
      "grad_norm": 6.227685928344727,
      "learning_rate": 7.029690604598354e-05,
      "loss": 0.024590209126472473,
      "step": 104690
    },
    {
      "epoch": 29.718989497587284,
      "grad_norm": 3.5708725452423096,
      "learning_rate": 7.029406755606018e-05,
      "loss": 0.011556456983089446,
      "step": 104700
    },
    {
      "epoch": 29.721827987510643,
      "grad_norm": 0.42200541496276855,
      "learning_rate": 7.029122906613682e-05,
      "loss": 0.008850054442882537,
      "step": 104710
    },
    {
      "epoch": 29.724666477434006,
      "grad_norm": 1.3981338739395142,
      "learning_rate": 7.028839057621346e-05,
      "loss": 0.01377348005771637,
      "step": 104720
    },
    {
      "epoch": 29.727504967357365,
      "grad_norm": 1.5834133625030518,
      "learning_rate": 7.028555208629009e-05,
      "loss": 0.00329880490899086,
      "step": 104730
    },
    {
      "epoch": 29.73034345728073,
      "grad_norm": 0.38446736335754395,
      "learning_rate": 7.028271359636673e-05,
      "loss": 0.012954989075660705,
      "step": 104740
    },
    {
      "epoch": 29.733181947204088,
      "grad_norm": 1.2588497400283813,
      "learning_rate": 7.027987510644337e-05,
      "loss": 0.008833683282136916,
      "step": 104750
    },
    {
      "epoch": 29.736020437127447,
      "grad_norm": 0.9540572762489319,
      "learning_rate": 7.027703661652002e-05,
      "loss": 0.001432366855442524,
      "step": 104760
    },
    {
      "epoch": 29.73885892705081,
      "grad_norm": 2.3396036624908447,
      "learning_rate": 7.027419812659666e-05,
      "loss": 0.003808160871267319,
      "step": 104770
    },
    {
      "epoch": 29.74169741697417,
      "grad_norm": 0.591968297958374,
      "learning_rate": 7.02713596366733e-05,
      "loss": 0.008841280639171601,
      "step": 104780
    },
    {
      "epoch": 29.744535906897532,
      "grad_norm": 5.853080749511719,
      "learning_rate": 7.026852114674994e-05,
      "loss": 0.011106960475444794,
      "step": 104790
    },
    {
      "epoch": 29.74737439682089,
      "grad_norm": 1.0360151529312134,
      "learning_rate": 7.026568265682657e-05,
      "loss": 0.0021541936323046683,
      "step": 104800
    },
    {
      "epoch": 29.75021288674425,
      "grad_norm": 0.989202618598938,
      "learning_rate": 7.026284416690321e-05,
      "loss": 0.0026784462854266165,
      "step": 104810
    },
    {
      "epoch": 29.753051376667614,
      "grad_norm": 4.244090557098389,
      "learning_rate": 7.026000567697985e-05,
      "loss": 0.004958838596940041,
      "step": 104820
    },
    {
      "epoch": 29.755889866590973,
      "grad_norm": 8.935042381286621,
      "learning_rate": 7.025716718705649e-05,
      "loss": 0.007089726626873016,
      "step": 104830
    },
    {
      "epoch": 29.758728356514336,
      "grad_norm": 0.5610134601593018,
      "learning_rate": 7.025432869713313e-05,
      "loss": 0.013608497381210328,
      "step": 104840
    },
    {
      "epoch": 29.761566846437695,
      "grad_norm": 10.513693809509277,
      "learning_rate": 7.025149020720977e-05,
      "loss": 0.0074524469673633575,
      "step": 104850
    },
    {
      "epoch": 29.764405336361055,
      "grad_norm": 0.41039466857910156,
      "learning_rate": 7.02486517172864e-05,
      "loss": 0.00746232271194458,
      "step": 104860
    },
    {
      "epoch": 29.767243826284417,
      "grad_norm": 0.17791564762592316,
      "learning_rate": 7.024581322736304e-05,
      "loss": 0.004781383275985718,
      "step": 104870
    },
    {
      "epoch": 29.770082316207777,
      "grad_norm": 11.260369300842285,
      "learning_rate": 7.024297473743968e-05,
      "loss": 0.009041262418031692,
      "step": 104880
    },
    {
      "epoch": 29.77292080613114,
      "grad_norm": 1.9183902740478516,
      "learning_rate": 7.024013624751633e-05,
      "loss": 0.006306622177362442,
      "step": 104890
    },
    {
      "epoch": 29.7757592960545,
      "grad_norm": 1.988551378250122,
      "learning_rate": 7.023729775759297e-05,
      "loss": 0.009582685679197312,
      "step": 104900
    },
    {
      "epoch": 29.77859778597786,
      "grad_norm": 1.1689237356185913,
      "learning_rate": 7.023445926766961e-05,
      "loss": 0.004791789874434471,
      "step": 104910
    },
    {
      "epoch": 29.78143627590122,
      "grad_norm": 6.808248519897461,
      "learning_rate": 7.023162077774624e-05,
      "loss": 0.006833506375551223,
      "step": 104920
    },
    {
      "epoch": 29.78427476582458,
      "grad_norm": 5.847177505493164,
      "learning_rate": 7.022878228782288e-05,
      "loss": 0.022859284281730653,
      "step": 104930
    },
    {
      "epoch": 29.787113255747943,
      "grad_norm": 1.2152276039123535,
      "learning_rate": 7.022594379789952e-05,
      "loss": 0.004604285210371017,
      "step": 104940
    },
    {
      "epoch": 29.789951745671303,
      "grad_norm": 12.555948257446289,
      "learning_rate": 7.022310530797616e-05,
      "loss": 0.007945778965950012,
      "step": 104950
    },
    {
      "epoch": 29.792790235594662,
      "grad_norm": 0.548850417137146,
      "learning_rate": 7.02202668180528e-05,
      "loss": 0.015255066752433776,
      "step": 104960
    },
    {
      "epoch": 29.795628725518025,
      "grad_norm": 9.073225021362305,
      "learning_rate": 7.021742832812944e-05,
      "loss": 0.02067209780216217,
      "step": 104970
    },
    {
      "epoch": 29.798467215441384,
      "grad_norm": 0.09555802494287491,
      "learning_rate": 7.021458983820608e-05,
      "loss": 0.01641386151313782,
      "step": 104980
    },
    {
      "epoch": 29.801305705364747,
      "grad_norm": 0.27824223041534424,
      "learning_rate": 7.021175134828271e-05,
      "loss": 0.003710051625967026,
      "step": 104990
    },
    {
      "epoch": 29.804144195288107,
      "grad_norm": 0.20276014506816864,
      "learning_rate": 7.020891285835935e-05,
      "loss": 0.002675805240869522,
      "step": 105000
    },
    {
      "epoch": 29.804144195288107,
      "eval_accuracy": 0.9730399949132066,
      "eval_loss": 0.09709598869085312,
      "eval_runtime": 31.8749,
      "eval_samples_per_second": 493.397,
      "eval_steps_per_second": 7.718,
      "step": 105000
    },
    {
      "epoch": 29.806982685211466,
      "grad_norm": 0.12756562232971191,
      "learning_rate": 7.0206074368436e-05,
      "loss": 0.013611838221549988,
      "step": 105010
    },
    {
      "epoch": 29.80982117513483,
      "grad_norm": 2.987877130508423,
      "learning_rate": 7.020323587851262e-05,
      "loss": 0.006055672466754913,
      "step": 105020
    },
    {
      "epoch": 29.812659665058188,
      "grad_norm": 0.5880629420280457,
      "learning_rate": 7.020039738858928e-05,
      "loss": 0.00497266948223114,
      "step": 105030
    },
    {
      "epoch": 29.81549815498155,
      "grad_norm": 0.9611206650733948,
      "learning_rate": 7.019755889866592e-05,
      "loss": 0.008511928468942642,
      "step": 105040
    },
    {
      "epoch": 29.81833664490491,
      "grad_norm": 0.722910463809967,
      "learning_rate": 7.019472040874255e-05,
      "loss": 0.008684001863002777,
      "step": 105050
    },
    {
      "epoch": 29.82117513482827,
      "grad_norm": 3.6950082778930664,
      "learning_rate": 7.019188191881919e-05,
      "loss": 0.008162318170070649,
      "step": 105060
    },
    {
      "epoch": 29.824013624751633,
      "grad_norm": 1.219530463218689,
      "learning_rate": 7.018904342889583e-05,
      "loss": 0.009760137647390366,
      "step": 105070
    },
    {
      "epoch": 29.826852114674992,
      "grad_norm": 12.354999542236328,
      "learning_rate": 7.018620493897247e-05,
      "loss": 0.00713539719581604,
      "step": 105080
    },
    {
      "epoch": 29.829690604598355,
      "grad_norm": 0.24815727770328522,
      "learning_rate": 7.01833664490491e-05,
      "loss": 0.008691404759883881,
      "step": 105090
    },
    {
      "epoch": 29.832529094521714,
      "grad_norm": 1.3925886154174805,
      "learning_rate": 7.018052795912575e-05,
      "loss": 0.004267697036266327,
      "step": 105100
    },
    {
      "epoch": 29.835367584445073,
      "grad_norm": 1.9502575397491455,
      "learning_rate": 7.01776894692024e-05,
      "loss": 0.005763031169772148,
      "step": 105110
    },
    {
      "epoch": 29.838206074368436,
      "grad_norm": 0.5452384948730469,
      "learning_rate": 7.017485097927902e-05,
      "loss": 0.014007450640201568,
      "step": 105120
    },
    {
      "epoch": 29.841044564291796,
      "grad_norm": 1.8186901807785034,
      "learning_rate": 7.017201248935566e-05,
      "loss": 0.011604809761047363,
      "step": 105130
    },
    {
      "epoch": 29.84388305421516,
      "grad_norm": 1.2743494510650635,
      "learning_rate": 7.01691739994323e-05,
      "loss": 0.021057575941085815,
      "step": 105140
    },
    {
      "epoch": 29.846721544138518,
      "grad_norm": 0.2516765892505646,
      "learning_rate": 7.016633550950893e-05,
      "loss": 0.013565397262573243,
      "step": 105150
    },
    {
      "epoch": 29.84956003406188,
      "grad_norm": 0.30280566215515137,
      "learning_rate": 7.016349701958559e-05,
      "loss": 0.018691384792327882,
      "step": 105160
    },
    {
      "epoch": 29.85239852398524,
      "grad_norm": 1.5659123659133911,
      "learning_rate": 7.016065852966223e-05,
      "loss": 0.017920032143592834,
      "step": 105170
    },
    {
      "epoch": 29.8552370139086,
      "grad_norm": 0.8984304070472717,
      "learning_rate": 7.015782003973886e-05,
      "loss": 0.010006338357925415,
      "step": 105180
    },
    {
      "epoch": 29.858075503831962,
      "grad_norm": 5.645437717437744,
      "learning_rate": 7.01549815498155e-05,
      "loss": 0.004256200045347214,
      "step": 105190
    },
    {
      "epoch": 29.86091399375532,
      "grad_norm": 0.836696982383728,
      "learning_rate": 7.015214305989214e-05,
      "loss": 0.006336534023284912,
      "step": 105200
    },
    {
      "epoch": 29.863752483678685,
      "grad_norm": 0.309998095035553,
      "learning_rate": 7.014930456996878e-05,
      "loss": 0.006125971674919128,
      "step": 105210
    },
    {
      "epoch": 29.866590973602044,
      "grad_norm": 1.6069484949111938,
      "learning_rate": 7.014646608004541e-05,
      "loss": 0.028770118951797485,
      "step": 105220
    },
    {
      "epoch": 29.869429463525403,
      "grad_norm": 0.4023089110851288,
      "learning_rate": 7.014362759012206e-05,
      "loss": 0.0036197923123836516,
      "step": 105230
    },
    {
      "epoch": 29.872267953448766,
      "grad_norm": 3.956683874130249,
      "learning_rate": 7.01407891001987e-05,
      "loss": 0.007113201916217804,
      "step": 105240
    },
    {
      "epoch": 29.875106443372125,
      "grad_norm": 0.1666334718465805,
      "learning_rate": 7.013795061027533e-05,
      "loss": 0.011134114861488343,
      "step": 105250
    },
    {
      "epoch": 29.87794493329549,
      "grad_norm": 3.045342206954956,
      "learning_rate": 7.013511212035198e-05,
      "loss": 0.014592207968235016,
      "step": 105260
    },
    {
      "epoch": 29.880783423218848,
      "grad_norm": 4.169530868530273,
      "learning_rate": 7.013227363042862e-05,
      "loss": 0.005603998899459839,
      "step": 105270
    },
    {
      "epoch": 29.883621913142207,
      "grad_norm": 1.0987203121185303,
      "learning_rate": 7.012943514050524e-05,
      "loss": 0.011438168585300446,
      "step": 105280
    },
    {
      "epoch": 29.88646040306557,
      "grad_norm": 8.558287620544434,
      "learning_rate": 7.012659665058189e-05,
      "loss": 0.0048771630972623825,
      "step": 105290
    },
    {
      "epoch": 29.88929889298893,
      "grad_norm": 0.1910807192325592,
      "learning_rate": 7.012375816065854e-05,
      "loss": 0.007158765196800232,
      "step": 105300
    },
    {
      "epoch": 29.892137382912292,
      "grad_norm": 12.604031562805176,
      "learning_rate": 7.012091967073517e-05,
      "loss": 0.02035667896270752,
      "step": 105310
    },
    {
      "epoch": 29.89497587283565,
      "grad_norm": 0.136419415473938,
      "learning_rate": 7.011808118081181e-05,
      "loss": 0.006935358792543411,
      "step": 105320
    },
    {
      "epoch": 29.89781436275901,
      "grad_norm": 3.6841089725494385,
      "learning_rate": 7.011524269088845e-05,
      "loss": 0.012233049422502518,
      "step": 105330
    },
    {
      "epoch": 29.900652852682374,
      "grad_norm": 5.590911865234375,
      "learning_rate": 7.011240420096509e-05,
      "loss": 0.007571963220834732,
      "step": 105340
    },
    {
      "epoch": 29.903491342605733,
      "grad_norm": 1.7996746301651,
      "learning_rate": 7.010956571104172e-05,
      "loss": 0.0033944204449653627,
      "step": 105350
    },
    {
      "epoch": 29.906329832529096,
      "grad_norm": 0.05662470683455467,
      "learning_rate": 7.010672722111838e-05,
      "loss": 0.008465706557035445,
      "step": 105360
    },
    {
      "epoch": 29.909168322452455,
      "grad_norm": 4.030693054199219,
      "learning_rate": 7.010388873119502e-05,
      "loss": 0.0069080710411071776,
      "step": 105370
    },
    {
      "epoch": 29.912006812375814,
      "grad_norm": 16.568635940551758,
      "learning_rate": 7.010105024127164e-05,
      "loss": 0.026711615920066833,
      "step": 105380
    },
    {
      "epoch": 29.914845302299177,
      "grad_norm": 3.9256691932678223,
      "learning_rate": 7.009821175134829e-05,
      "loss": 0.013809213042259216,
      "step": 105390
    },
    {
      "epoch": 29.917683792222537,
      "grad_norm": 0.5532540082931519,
      "learning_rate": 7.009537326142493e-05,
      "loss": 0.007177532464265823,
      "step": 105400
    },
    {
      "epoch": 29.9205222821459,
      "grad_norm": 0.753618597984314,
      "learning_rate": 7.009253477150156e-05,
      "loss": 0.0048213407397270204,
      "step": 105410
    },
    {
      "epoch": 29.92336077206926,
      "grad_norm": 0.06940331310033798,
      "learning_rate": 7.00896962815782e-05,
      "loss": 0.0029305074363946914,
      "step": 105420
    },
    {
      "epoch": 29.92619926199262,
      "grad_norm": 5.825138568878174,
      "learning_rate": 7.008685779165485e-05,
      "loss": 0.011952961981296539,
      "step": 105430
    },
    {
      "epoch": 29.92903775191598,
      "grad_norm": 0.7064264416694641,
      "learning_rate": 7.008401930173148e-05,
      "loss": 0.0056862208992242815,
      "step": 105440
    },
    {
      "epoch": 29.93187624183934,
      "grad_norm": 0.05675726756453514,
      "learning_rate": 7.008118081180812e-05,
      "loss": 0.004222647100687027,
      "step": 105450
    },
    {
      "epoch": 29.934714731762703,
      "grad_norm": 5.888077259063721,
      "learning_rate": 7.007834232188476e-05,
      "loss": 0.005757405608892441,
      "step": 105460
    },
    {
      "epoch": 29.937553221686063,
      "grad_norm": 9.987832069396973,
      "learning_rate": 7.00755038319614e-05,
      "loss": 0.010487261414527892,
      "step": 105470
    },
    {
      "epoch": 29.940391711609422,
      "grad_norm": 0.5974326133728027,
      "learning_rate": 7.007266534203803e-05,
      "loss": 0.0030100906267762186,
      "step": 105480
    },
    {
      "epoch": 29.943230201532785,
      "grad_norm": 6.271804332733154,
      "learning_rate": 7.006982685211469e-05,
      "loss": 0.004383539408445358,
      "step": 105490
    },
    {
      "epoch": 29.946068691456144,
      "grad_norm": 0.05634259805083275,
      "learning_rate": 7.006698836219133e-05,
      "loss": 0.003014201112091541,
      "step": 105500
    },
    {
      "epoch": 29.946068691456144,
      "eval_accuracy": 0.9731671647485216,
      "eval_loss": 0.086981400847435,
      "eval_runtime": 32.0241,
      "eval_samples_per_second": 491.1,
      "eval_steps_per_second": 7.682,
      "step": 105500
    },
    {
      "epoch": 29.948907181379507,
      "grad_norm": 6.278529644012451,
      "learning_rate": 7.006414987226796e-05,
      "loss": 0.003655734285712242,
      "step": 105510
    },
    {
      "epoch": 29.951745671302866,
      "grad_norm": 2.368325710296631,
      "learning_rate": 7.00613113823446e-05,
      "loss": 0.01915167421102524,
      "step": 105520
    },
    {
      "epoch": 29.95458416122623,
      "grad_norm": 0.7736899852752686,
      "learning_rate": 7.005847289242124e-05,
      "loss": 0.015487337112426757,
      "step": 105530
    },
    {
      "epoch": 29.95742265114959,
      "grad_norm": 7.630155563354492,
      "learning_rate": 7.005563440249787e-05,
      "loss": 0.004491440951824188,
      "step": 105540
    },
    {
      "epoch": 29.960261141072948,
      "grad_norm": 1.8255661725997925,
      "learning_rate": 7.005279591257451e-05,
      "loss": 0.004574146866798401,
      "step": 105550
    },
    {
      "epoch": 29.96309963099631,
      "grad_norm": 14.693599700927734,
      "learning_rate": 7.004995742265116e-05,
      "loss": 0.016626988351345063,
      "step": 105560
    },
    {
      "epoch": 29.96593812091967,
      "grad_norm": 0.3312087953090668,
      "learning_rate": 7.004711893272779e-05,
      "loss": 0.0026948297396302222,
      "step": 105570
    },
    {
      "epoch": 29.968776610843033,
      "grad_norm": 5.3750319480896,
      "learning_rate": 7.004428044280443e-05,
      "loss": 0.00822022631764412,
      "step": 105580
    },
    {
      "epoch": 29.971615100766392,
      "grad_norm": 1.4358543157577515,
      "learning_rate": 7.004144195288107e-05,
      "loss": 0.01368415355682373,
      "step": 105590
    },
    {
      "epoch": 29.974453590689752,
      "grad_norm": 0.16050013899803162,
      "learning_rate": 7.003860346295771e-05,
      "loss": 0.007451897859573365,
      "step": 105600
    },
    {
      "epoch": 29.977292080613115,
      "grad_norm": 10.538488388061523,
      "learning_rate": 7.003576497303434e-05,
      "loss": 0.00491114854812622,
      "step": 105610
    },
    {
      "epoch": 29.980130570536474,
      "grad_norm": 6.043573379516602,
      "learning_rate": 7.003292648311098e-05,
      "loss": 0.004744678735733032,
      "step": 105620
    },
    {
      "epoch": 29.982969060459837,
      "grad_norm": 0.26893484592437744,
      "learning_rate": 7.003008799318764e-05,
      "loss": 0.013608326017856599,
      "step": 105630
    },
    {
      "epoch": 29.985807550383196,
      "grad_norm": 0.4437483549118042,
      "learning_rate": 7.002724950326427e-05,
      "loss": 0.027003341913223268,
      "step": 105640
    },
    {
      "epoch": 29.988646040306556,
      "grad_norm": 1.630883812904358,
      "learning_rate": 7.002441101334091e-05,
      "loss": 0.01748095899820328,
      "step": 105650
    },
    {
      "epoch": 29.99148453022992,
      "grad_norm": 1.0774277448654175,
      "learning_rate": 7.002157252341755e-05,
      "loss": 0.013404801487922668,
      "step": 105660
    },
    {
      "epoch": 29.994323020153278,
      "grad_norm": 10.68116283416748,
      "learning_rate": 7.001873403349418e-05,
      "loss": 0.007102926820516586,
      "step": 105670
    },
    {
      "epoch": 29.99716151007664,
      "grad_norm": 14.528780937194824,
      "learning_rate": 7.001589554357082e-05,
      "loss": 0.009707969427108765,
      "step": 105680
    },
    {
      "epoch": 30.0,
      "grad_norm": 0.1280197948217392,
      "learning_rate": 7.001305705364747e-05,
      "loss": 0.0049740664660930635,
      "step": 105690
    },
    {
      "epoch": 30.00283848992336,
      "grad_norm": 2.396388292312622,
      "learning_rate": 7.00102185637241e-05,
      "loss": 0.003515026718378067,
      "step": 105700
    },
    {
      "epoch": 30.005676979846722,
      "grad_norm": 0.1878899484872818,
      "learning_rate": 7.000738007380074e-05,
      "loss": 0.0049481231719255446,
      "step": 105710
    },
    {
      "epoch": 30.00851546977008,
      "grad_norm": 1.3411624431610107,
      "learning_rate": 7.000454158387738e-05,
      "loss": 0.016996750235557558,
      "step": 105720
    },
    {
      "epoch": 30.011353959693444,
      "grad_norm": 1.184022068977356,
      "learning_rate": 7.000170309395403e-05,
      "loss": 0.008787652850151062,
      "step": 105730
    },
    {
      "epoch": 30.014192449616804,
      "grad_norm": 1.0792043209075928,
      "learning_rate": 6.999886460403065e-05,
      "loss": 0.0060198113322258,
      "step": 105740
    },
    {
      "epoch": 30.017030939540163,
      "grad_norm": 0.22205938398838043,
      "learning_rate": 6.99960261141073e-05,
      "loss": 0.007967585325241089,
      "step": 105750
    },
    {
      "epoch": 30.019869429463526,
      "grad_norm": 1.0031933784484863,
      "learning_rate": 6.999318762418394e-05,
      "loss": 0.0023270612582564354,
      "step": 105760
    },
    {
      "epoch": 30.022707919386885,
      "grad_norm": 1.9731190204620361,
      "learning_rate": 6.999034913426058e-05,
      "loss": 0.006109318137168885,
      "step": 105770
    },
    {
      "epoch": 30.025546409310248,
      "grad_norm": 0.1088927835226059,
      "learning_rate": 6.998751064433722e-05,
      "loss": 0.005002518370747566,
      "step": 105780
    },
    {
      "epoch": 30.028384899233608,
      "grad_norm": 2.598339319229126,
      "learning_rate": 6.998467215441386e-05,
      "loss": 0.008690781891345978,
      "step": 105790
    },
    {
      "epoch": 30.031223389156967,
      "grad_norm": 0.371311217546463,
      "learning_rate": 6.998183366449049e-05,
      "loss": 0.00646035298705101,
      "step": 105800
    },
    {
      "epoch": 30.03406187908033,
      "grad_norm": 1.5478878021240234,
      "learning_rate": 6.997899517456713e-05,
      "loss": 0.0044115521013736725,
      "step": 105810
    },
    {
      "epoch": 30.03690036900369,
      "grad_norm": 0.035102955996990204,
      "learning_rate": 6.997615668464377e-05,
      "loss": 0.004960698634386062,
      "step": 105820
    },
    {
      "epoch": 30.039738858927052,
      "grad_norm": 11.09414291381836,
      "learning_rate": 6.997331819472041e-05,
      "loss": 0.01140763759613037,
      "step": 105830
    },
    {
      "epoch": 30.04257734885041,
      "grad_norm": 1.037131905555725,
      "learning_rate": 6.997047970479705e-05,
      "loss": 0.0031506657600402833,
      "step": 105840
    },
    {
      "epoch": 30.04541583877377,
      "grad_norm": 0.20512822270393372,
      "learning_rate": 6.99676412148737e-05,
      "loss": 0.00498289093375206,
      "step": 105850
    },
    {
      "epoch": 30.048254328697134,
      "grad_norm": 0.9733025431632996,
      "learning_rate": 6.996480272495032e-05,
      "loss": 0.006488683819770813,
      "step": 105860
    },
    {
      "epoch": 30.051092818620493,
      "grad_norm": 3.6543848514556885,
      "learning_rate": 6.996196423502696e-05,
      "loss": 0.011569606512784958,
      "step": 105870
    },
    {
      "epoch": 30.053931308543856,
      "grad_norm": 1.5402642488479614,
      "learning_rate": 6.99591257451036e-05,
      "loss": 0.0052514314651489254,
      "step": 105880
    },
    {
      "epoch": 30.056769798467215,
      "grad_norm": 0.5969558954238892,
      "learning_rate": 6.995628725518025e-05,
      "loss": 0.004439448192715645,
      "step": 105890
    },
    {
      "epoch": 30.059608288390578,
      "grad_norm": 2.308130979537964,
      "learning_rate": 6.995344876525689e-05,
      "loss": 0.009626329690217972,
      "step": 105900
    },
    {
      "epoch": 30.062446778313937,
      "grad_norm": 4.41294527053833,
      "learning_rate": 6.995061027533353e-05,
      "loss": 0.017955288290977478,
      "step": 105910
    },
    {
      "epoch": 30.065285268237297,
      "grad_norm": 0.8770581483840942,
      "learning_rate": 6.994777178541017e-05,
      "loss": 0.010635729134082793,
      "step": 105920
    },
    {
      "epoch": 30.06812375816066,
      "grad_norm": 0.449901282787323,
      "learning_rate": 6.99449332954868e-05,
      "loss": 0.0016405627131462098,
      "step": 105930
    },
    {
      "epoch": 30.07096224808402,
      "grad_norm": 8.81567096710205,
      "learning_rate": 6.994209480556344e-05,
      "loss": 0.002677132375538349,
      "step": 105940
    },
    {
      "epoch": 30.07380073800738,
      "grad_norm": 0.08543285727500916,
      "learning_rate": 6.993925631564008e-05,
      "loss": 0.008950378000736236,
      "step": 105950
    },
    {
      "epoch": 30.07663922793074,
      "grad_norm": 0.2524518370628357,
      "learning_rate": 6.993641782571672e-05,
      "loss": 0.002813626453280449,
      "step": 105960
    },
    {
      "epoch": 30.0794777178541,
      "grad_norm": 0.47767505049705505,
      "learning_rate": 6.993357933579336e-05,
      "loss": 0.0023371132090687754,
      "step": 105970
    },
    {
      "epoch": 30.082316207777463,
      "grad_norm": 0.7387546300888062,
      "learning_rate": 6.993074084587e-05,
      "loss": 0.009980396181344987,
      "step": 105980
    },
    {
      "epoch": 30.085154697700823,
      "grad_norm": 0.49040329456329346,
      "learning_rate": 6.992790235594663e-05,
      "loss": 0.023836809396743774,
      "step": 105990
    },
    {
      "epoch": 30.087993187624186,
      "grad_norm": 0.2203972041606903,
      "learning_rate": 6.992506386602327e-05,
      "loss": 0.0026527363806962967,
      "step": 106000
    },
    {
      "epoch": 30.087993187624186,
      "eval_accuracy": 0.972722070324919,
      "eval_loss": 0.0928550735116005,
      "eval_runtime": 31.2649,
      "eval_samples_per_second": 503.025,
      "eval_steps_per_second": 7.868,
      "step": 106000
    },
    {
      "epoch": 30.090831677547545,
      "grad_norm": 3.8761022090911865,
      "learning_rate": 6.992222537609992e-05,
      "loss": 0.006042758002877235,
      "step": 106010
    },
    {
      "epoch": 30.093670167470904,
      "grad_norm": 0.7080287337303162,
      "learning_rate": 6.991938688617656e-05,
      "loss": 0.00552593469619751,
      "step": 106020
    },
    {
      "epoch": 30.096508657394267,
      "grad_norm": 2.840165376663208,
      "learning_rate": 6.99165483962532e-05,
      "loss": 0.019345279037952422,
      "step": 106030
    },
    {
      "epoch": 30.099347147317626,
      "grad_norm": 1.2198984622955322,
      "learning_rate": 6.991370990632984e-05,
      "loss": 0.0035472460091114043,
      "step": 106040
    },
    {
      "epoch": 30.10218563724099,
      "grad_norm": 2.3342795372009277,
      "learning_rate": 6.991087141640648e-05,
      "loss": 0.005064839497208595,
      "step": 106050
    },
    {
      "epoch": 30.10502412716435,
      "grad_norm": 1.139198660850525,
      "learning_rate": 6.990803292648311e-05,
      "loss": 0.003509926795959473,
      "step": 106060
    },
    {
      "epoch": 30.107862617087708,
      "grad_norm": 4.902596473693848,
      "learning_rate": 6.990519443655975e-05,
      "loss": 0.004718238115310669,
      "step": 106070
    },
    {
      "epoch": 30.11070110701107,
      "grad_norm": 0.13578669726848602,
      "learning_rate": 6.990235594663639e-05,
      "loss": 0.004302560538053513,
      "step": 106080
    },
    {
      "epoch": 30.11353959693443,
      "grad_norm": 1.3112000226974487,
      "learning_rate": 6.989951745671303e-05,
      "loss": 0.004853302240371704,
      "step": 106090
    },
    {
      "epoch": 30.116378086857793,
      "grad_norm": 0.3177511394023895,
      "learning_rate": 6.989667896678967e-05,
      "loss": 0.011783809959888458,
      "step": 106100
    },
    {
      "epoch": 30.119216576781152,
      "grad_norm": 0.5152670741081238,
      "learning_rate": 6.989384047686632e-05,
      "loss": 0.010212457180023194,
      "step": 106110
    },
    {
      "epoch": 30.12205506670451,
      "grad_norm": 0.783198356628418,
      "learning_rate": 6.989100198694294e-05,
      "loss": 0.005921649187803269,
      "step": 106120
    },
    {
      "epoch": 30.124893556627875,
      "grad_norm": 0.011008935049176216,
      "learning_rate": 6.988816349701959e-05,
      "loss": 0.007097382843494415,
      "step": 106130
    },
    {
      "epoch": 30.127732046551234,
      "grad_norm": 1.8573511838912964,
      "learning_rate": 6.988532500709623e-05,
      "loss": 0.00810546949505806,
      "step": 106140
    },
    {
      "epoch": 30.130570536474597,
      "grad_norm": 1.0891584157943726,
      "learning_rate": 6.988248651717287e-05,
      "loss": 0.003433910757303238,
      "step": 106150
    },
    {
      "epoch": 30.133409026397956,
      "grad_norm": 0.09283721446990967,
      "learning_rate": 6.987964802724951e-05,
      "loss": 0.008321655541658401,
      "step": 106160
    },
    {
      "epoch": 30.136247516321315,
      "grad_norm": 3.6078662872314453,
      "learning_rate": 6.987680953732615e-05,
      "loss": 0.0042728234082460405,
      "step": 106170
    },
    {
      "epoch": 30.13908600624468,
      "grad_norm": 0.5276238918304443,
      "learning_rate": 6.987397104740279e-05,
      "loss": 0.008118095993995666,
      "step": 106180
    },
    {
      "epoch": 30.141924496168038,
      "grad_norm": 4.638086795806885,
      "learning_rate": 6.987113255747942e-05,
      "loss": 0.009959150850772858,
      "step": 106190
    },
    {
      "epoch": 30.1447629860914,
      "grad_norm": 0.8666144609451294,
      "learning_rate": 6.986829406755606e-05,
      "loss": 0.0019546801224350928,
      "step": 106200
    },
    {
      "epoch": 30.14760147601476,
      "grad_norm": 1.9906501770019531,
      "learning_rate": 6.98654555776327e-05,
      "loss": 0.007720711827278137,
      "step": 106210
    },
    {
      "epoch": 30.15043996593812,
      "grad_norm": 0.15926460921764374,
      "learning_rate": 6.986261708770933e-05,
      "loss": 0.0020163768902421,
      "step": 106220
    },
    {
      "epoch": 30.153278455861482,
      "grad_norm": 0.05704344063997269,
      "learning_rate": 6.985977859778599e-05,
      "loss": 0.0011629033833742142,
      "step": 106230
    },
    {
      "epoch": 30.15611694578484,
      "grad_norm": 0.4602694511413574,
      "learning_rate": 6.985694010786263e-05,
      "loss": 0.004268961399793625,
      "step": 106240
    },
    {
      "epoch": 30.158955435708204,
      "grad_norm": 15.870415687561035,
      "learning_rate": 6.985410161793925e-05,
      "loss": 0.006577381491661071,
      "step": 106250
    },
    {
      "epoch": 30.161793925631564,
      "grad_norm": 5.653326988220215,
      "learning_rate": 6.98512631280159e-05,
      "loss": 0.012542919814586639,
      "step": 106260
    },
    {
      "epoch": 30.164632415554923,
      "grad_norm": 0.23508703708648682,
      "learning_rate": 6.984842463809254e-05,
      "loss": 0.016380329430103303,
      "step": 106270
    },
    {
      "epoch": 30.167470905478286,
      "grad_norm": 10.603766441345215,
      "learning_rate": 6.984558614816918e-05,
      "loss": 0.01786281615495682,
      "step": 106280
    },
    {
      "epoch": 30.170309395401645,
      "grad_norm": 0.1227574422955513,
      "learning_rate": 6.984274765824582e-05,
      "loss": 0.00489969439804554,
      "step": 106290
    },
    {
      "epoch": 30.173147885325008,
      "grad_norm": 3.113767147064209,
      "learning_rate": 6.983990916832246e-05,
      "loss": 0.001922149769961834,
      "step": 106300
    },
    {
      "epoch": 30.175986375248367,
      "grad_norm": 0.8641319274902344,
      "learning_rate": 6.98370706783991e-05,
      "loss": 0.008935289829969406,
      "step": 106310
    },
    {
      "epoch": 30.17882486517173,
      "grad_norm": 1.948310136795044,
      "learning_rate": 6.983423218847573e-05,
      "loss": 0.005337618291378021,
      "step": 106320
    },
    {
      "epoch": 30.18166335509509,
      "grad_norm": 4.1576247215271,
      "learning_rate": 6.983139369855237e-05,
      "loss": 0.005285832658410072,
      "step": 106330
    },
    {
      "epoch": 30.18450184501845,
      "grad_norm": 0.24573393166065216,
      "learning_rate": 6.982855520862901e-05,
      "loss": 0.004684130847454071,
      "step": 106340
    },
    {
      "epoch": 30.187340334941812,
      "grad_norm": 11.290939331054688,
      "learning_rate": 6.982571671870564e-05,
      "loss": 0.017245198786258697,
      "step": 106350
    },
    {
      "epoch": 30.19017882486517,
      "grad_norm": 0.8905802965164185,
      "learning_rate": 6.98228782287823e-05,
      "loss": 0.003093671053647995,
      "step": 106360
    },
    {
      "epoch": 30.193017314788534,
      "grad_norm": 0.2338159680366516,
      "learning_rate": 6.982003973885894e-05,
      "loss": 0.008088739216327667,
      "step": 106370
    },
    {
      "epoch": 30.195855804711893,
      "grad_norm": 0.6568982005119324,
      "learning_rate": 6.981720124893557e-05,
      "loss": 0.003516152501106262,
      "step": 106380
    },
    {
      "epoch": 30.198694294635253,
      "grad_norm": 0.2358493208885193,
      "learning_rate": 6.981436275901221e-05,
      "loss": 0.0026024546474218368,
      "step": 106390
    },
    {
      "epoch": 30.201532784558616,
      "grad_norm": 2.402280330657959,
      "learning_rate": 6.981152426908885e-05,
      "loss": 0.00999762788414955,
      "step": 106400
    },
    {
      "epoch": 30.204371274481975,
      "grad_norm": 0.05988219380378723,
      "learning_rate": 6.980868577916549e-05,
      "loss": 0.009985704720020295,
      "step": 106410
    },
    {
      "epoch": 30.207209764405338,
      "grad_norm": 6.06828498840332,
      "learning_rate": 6.980584728924212e-05,
      "loss": 0.01427840143442154,
      "step": 106420
    },
    {
      "epoch": 30.210048254328697,
      "grad_norm": 4.138000965118408,
      "learning_rate": 6.980300879931877e-05,
      "loss": 0.0058008283376693726,
      "step": 106430
    },
    {
      "epoch": 30.212886744252057,
      "grad_norm": 0.35986533761024475,
      "learning_rate": 6.980017030939541e-05,
      "loss": 0.009734107553958893,
      "step": 106440
    },
    {
      "epoch": 30.21572523417542,
      "grad_norm": 0.9744632840156555,
      "learning_rate": 6.979733181947204e-05,
      "loss": 0.005061643570661545,
      "step": 106450
    },
    {
      "epoch": 30.21856372409878,
      "grad_norm": 0.09392717480659485,
      "learning_rate": 6.979449332954868e-05,
      "loss": 0.004758042097091675,
      "step": 106460
    },
    {
      "epoch": 30.22140221402214,
      "grad_norm": 3.729412078857422,
      "learning_rate": 6.979165483962532e-05,
      "loss": 0.0031662628054618834,
      "step": 106470
    },
    {
      "epoch": 30.2242407039455,
      "grad_norm": 6.4889302253723145,
      "learning_rate": 6.978881634970195e-05,
      "loss": 0.010526859760284423,
      "step": 106480
    },
    {
      "epoch": 30.22707919386886,
      "grad_norm": 0.206144779920578,
      "learning_rate": 6.978597785977861e-05,
      "loss": 0.0030444245785474777,
      "step": 106490
    },
    {
      "epoch": 30.229917683792223,
      "grad_norm": 0.44263148307800293,
      "learning_rate": 6.978313936985525e-05,
      "loss": 0.0020123934373259544,
      "step": 106500
    },
    {
      "epoch": 30.229917683792223,
      "eval_accuracy": 0.9743116932663572,
      "eval_loss": 0.09001249074935913,
      "eval_runtime": 31.7375,
      "eval_samples_per_second": 495.534,
      "eval_steps_per_second": 7.751,
      "step": 106500
    },
    {
      "epoch": 30.232756173715583,
      "grad_norm": 1.2730681896209717,
      "learning_rate": 6.978030087993188e-05,
      "loss": 0.007269562780857086,
      "step": 106510
    },
    {
      "epoch": 30.235594663638945,
      "grad_norm": 1.1230195760726929,
      "learning_rate": 6.977746239000852e-05,
      "loss": 0.0027793947607278823,
      "step": 106520
    },
    {
      "epoch": 30.238433153562305,
      "grad_norm": 9.606831550598145,
      "learning_rate": 6.977462390008516e-05,
      "loss": 0.01779661625623703,
      "step": 106530
    },
    {
      "epoch": 30.241271643485664,
      "grad_norm": 0.4312793016433716,
      "learning_rate": 6.97717854101618e-05,
      "loss": 0.003860405087471008,
      "step": 106540
    },
    {
      "epoch": 30.244110133409027,
      "grad_norm": 0.9846163988113403,
      "learning_rate": 6.976894692023843e-05,
      "loss": 0.012633754312992096,
      "step": 106550
    },
    {
      "epoch": 30.246948623332386,
      "grad_norm": 0.9642807245254517,
      "learning_rate": 6.976610843031508e-05,
      "loss": 0.009551146626472473,
      "step": 106560
    },
    {
      "epoch": 30.24978711325575,
      "grad_norm": 6.0422868728637695,
      "learning_rate": 6.976326994039171e-05,
      "loss": 0.010451701283454896,
      "step": 106570
    },
    {
      "epoch": 30.25262560317911,
      "grad_norm": 1.267256498336792,
      "learning_rate": 6.976043145046835e-05,
      "loss": 0.011924322694540024,
      "step": 106580
    },
    {
      "epoch": 30.255464093102468,
      "grad_norm": 0.03753344714641571,
      "learning_rate": 6.9757592960545e-05,
      "loss": 0.004644815623760223,
      "step": 106590
    },
    {
      "epoch": 30.25830258302583,
      "grad_norm": 0.05368761345744133,
      "learning_rate": 6.975475447062164e-05,
      "loss": 0.008215417712926864,
      "step": 106600
    },
    {
      "epoch": 30.26114107294919,
      "grad_norm": 0.12343554943799973,
      "learning_rate": 6.975191598069826e-05,
      "loss": 0.010541865229606628,
      "step": 106610
    },
    {
      "epoch": 30.263979562872553,
      "grad_norm": 0.11323529481887817,
      "learning_rate": 6.97490774907749e-05,
      "loss": 0.009044089913368225,
      "step": 106620
    },
    {
      "epoch": 30.266818052795912,
      "grad_norm": 0.32372793555259705,
      "learning_rate": 6.974623900085156e-05,
      "loss": 0.004682055860757828,
      "step": 106630
    },
    {
      "epoch": 30.26965654271927,
      "grad_norm": 0.4746091067790985,
      "learning_rate": 6.974340051092819e-05,
      "loss": 0.00407271534204483,
      "step": 106640
    },
    {
      "epoch": 30.272495032642635,
      "grad_norm": 0.26943066716194153,
      "learning_rate": 6.974056202100483e-05,
      "loss": 0.0013801522552967071,
      "step": 106650
    },
    {
      "epoch": 30.275333522565994,
      "grad_norm": 21.439119338989258,
      "learning_rate": 6.973772353108147e-05,
      "loss": 0.015241830050945282,
      "step": 106660
    },
    {
      "epoch": 30.278172012489357,
      "grad_norm": 1.1437351703643799,
      "learning_rate": 6.973488504115811e-05,
      "loss": 0.008684326708316804,
      "step": 106670
    },
    {
      "epoch": 30.281010502412716,
      "grad_norm": 0.7628901600837708,
      "learning_rate": 6.973204655123474e-05,
      "loss": 0.005086052417755127,
      "step": 106680
    },
    {
      "epoch": 30.283848992336075,
      "grad_norm": 0.06304983049631119,
      "learning_rate": 6.97292080613114e-05,
      "loss": 0.0066182568669319155,
      "step": 106690
    },
    {
      "epoch": 30.28668748225944,
      "grad_norm": 0.5394207835197449,
      "learning_rate": 6.972636957138802e-05,
      "loss": 0.008395110815763473,
      "step": 106700
    },
    {
      "epoch": 30.289525972182798,
      "grad_norm": 6.3293328285217285,
      "learning_rate": 6.972353108146466e-05,
      "loss": 0.007367582619190216,
      "step": 106710
    },
    {
      "epoch": 30.29236446210616,
      "grad_norm": 0.8390308618545532,
      "learning_rate": 6.97206925915413e-05,
      "loss": 0.01210056096315384,
      "step": 106720
    },
    {
      "epoch": 30.29520295202952,
      "grad_norm": 13.101326942443848,
      "learning_rate": 6.971785410161795e-05,
      "loss": 0.008845705538988113,
      "step": 106730
    },
    {
      "epoch": 30.298041441952883,
      "grad_norm": 3.6768155097961426,
      "learning_rate": 6.971501561169457e-05,
      "loss": 0.008067777752876282,
      "step": 106740
    },
    {
      "epoch": 30.300879931876242,
      "grad_norm": 0.10059312731027603,
      "learning_rate": 6.971217712177122e-05,
      "loss": 0.002944110706448555,
      "step": 106750
    },
    {
      "epoch": 30.3037184217996,
      "grad_norm": 4.790853023529053,
      "learning_rate": 6.970933863184787e-05,
      "loss": 0.0054957568645477295,
      "step": 106760
    },
    {
      "epoch": 30.306556911722964,
      "grad_norm": 3.790133237838745,
      "learning_rate": 6.97065001419245e-05,
      "loss": 0.008353428542613983,
      "step": 106770
    },
    {
      "epoch": 30.309395401646324,
      "grad_norm": 0.8537731766700745,
      "learning_rate": 6.970366165200114e-05,
      "loss": 0.02075006812810898,
      "step": 106780
    },
    {
      "epoch": 30.312233891569687,
      "grad_norm": 0.6488693356513977,
      "learning_rate": 6.970082316207778e-05,
      "loss": 0.00684410035610199,
      "step": 106790
    },
    {
      "epoch": 30.315072381493046,
      "grad_norm": 14.015950202941895,
      "learning_rate": 6.969798467215441e-05,
      "loss": 0.014066335558891297,
      "step": 106800
    },
    {
      "epoch": 30.317910871416405,
      "grad_norm": 1.2011818885803223,
      "learning_rate": 6.969514618223105e-05,
      "loss": 0.0031739674508571626,
      "step": 106810
    },
    {
      "epoch": 30.320749361339768,
      "grad_norm": 0.09802905470132828,
      "learning_rate": 6.96923076923077e-05,
      "loss": 0.009791883826255798,
      "step": 106820
    },
    {
      "epoch": 30.323587851263127,
      "grad_norm": 0.366552472114563,
      "learning_rate": 6.968946920238433e-05,
      "loss": 0.005262854322791099,
      "step": 106830
    },
    {
      "epoch": 30.32642634118649,
      "grad_norm": 0.14670521020889282,
      "learning_rate": 6.968663071246097e-05,
      "loss": 0.0035857915878295898,
      "step": 106840
    },
    {
      "epoch": 30.32926483110985,
      "grad_norm": 1.46879243850708,
      "learning_rate": 6.968379222253762e-05,
      "loss": 0.002786737121641636,
      "step": 106850
    },
    {
      "epoch": 30.33210332103321,
      "grad_norm": 14.968185424804688,
      "learning_rate": 6.968095373261426e-05,
      "loss": 0.014249712228775024,
      "step": 106860
    },
    {
      "epoch": 30.334941810956572,
      "grad_norm": 15.365283012390137,
      "learning_rate": 6.967811524269088e-05,
      "loss": 0.013446538150310517,
      "step": 106870
    },
    {
      "epoch": 30.33778030087993,
      "grad_norm": 3.4720499515533447,
      "learning_rate": 6.967527675276753e-05,
      "loss": 0.0047527171671390535,
      "step": 106880
    },
    {
      "epoch": 30.340618790803294,
      "grad_norm": 14.858259201049805,
      "learning_rate": 6.967243826284418e-05,
      "loss": 0.009643489867448807,
      "step": 106890
    },
    {
      "epoch": 30.343457280726653,
      "grad_norm": 0.49035152792930603,
      "learning_rate": 6.966959977292081e-05,
      "loss": 0.01388126313686371,
      "step": 106900
    },
    {
      "epoch": 30.346295770650013,
      "grad_norm": 0.1488294154405594,
      "learning_rate": 6.966676128299745e-05,
      "loss": 0.003382474184036255,
      "step": 106910
    },
    {
      "epoch": 30.349134260573376,
      "grad_norm": 0.8586639165878296,
      "learning_rate": 6.966392279307409e-05,
      "loss": 0.0041621245443820955,
      "step": 106920
    },
    {
      "epoch": 30.351972750496735,
      "grad_norm": 0.21107231080532074,
      "learning_rate": 6.966108430315072e-05,
      "loss": 0.0048301398754119875,
      "step": 106930
    },
    {
      "epoch": 30.354811240420098,
      "grad_norm": 0.08656536042690277,
      "learning_rate": 6.965824581322736e-05,
      "loss": 0.008052317053079605,
      "step": 106940
    },
    {
      "epoch": 30.357649730343457,
      "grad_norm": 0.718498706817627,
      "learning_rate": 6.9655407323304e-05,
      "loss": 0.006357775628566742,
      "step": 106950
    },
    {
      "epoch": 30.360488220266816,
      "grad_norm": 1.1139857769012451,
      "learning_rate": 6.965256883338064e-05,
      "loss": 0.006194550544023514,
      "step": 106960
    },
    {
      "epoch": 30.36332671019018,
      "grad_norm": 0.044830795377492905,
      "learning_rate": 6.964973034345728e-05,
      "loss": 0.025629836320877075,
      "step": 106970
    },
    {
      "epoch": 30.36616520011354,
      "grad_norm": 2.819146156311035,
      "learning_rate": 6.964689185353393e-05,
      "loss": 0.007767321914434433,
      "step": 106980
    },
    {
      "epoch": 30.3690036900369,
      "grad_norm": 2.2762677669525146,
      "learning_rate": 6.964405336361057e-05,
      "loss": 0.006254389137029648,
      "step": 106990
    },
    {
      "epoch": 30.37184217996026,
      "grad_norm": 0.17382320761680603,
      "learning_rate": 6.96412148736872e-05,
      "loss": 0.0013229331001639367,
      "step": 107000
    },
    {
      "epoch": 30.37184217996026,
      "eval_accuracy": 0.9684618808418644,
      "eval_loss": 0.10519951581954956,
      "eval_runtime": 31.7689,
      "eval_samples_per_second": 495.043,
      "eval_steps_per_second": 7.743,
      "step": 107000
    },
    {
      "epoch": 30.37468066988362,
      "grad_norm": 1.1010828018188477,
      "learning_rate": 6.963837638376384e-05,
      "loss": 0.002672521583735943,
      "step": 107010
    },
    {
      "epoch": 30.377519159806983,
      "grad_norm": 0.24397417902946472,
      "learning_rate": 6.963553789384049e-05,
      "loss": 0.011796991527080535,
      "step": 107020
    },
    {
      "epoch": 30.380357649730342,
      "grad_norm": 2.6973917484283447,
      "learning_rate": 6.963269940391712e-05,
      "loss": 0.005751124024391175,
      "step": 107030
    },
    {
      "epoch": 30.383196139653705,
      "grad_norm": 0.49739399552345276,
      "learning_rate": 6.962986091399376e-05,
      "loss": 0.008355898410081863,
      "step": 107040
    },
    {
      "epoch": 30.386034629577065,
      "grad_norm": 0.17412716150283813,
      "learning_rate": 6.96270224240704e-05,
      "loss": 0.006954370439052582,
      "step": 107050
    },
    {
      "epoch": 30.388873119500424,
      "grad_norm": 0.30756106972694397,
      "learning_rate": 6.962418393414703e-05,
      "loss": 0.006160261109471321,
      "step": 107060
    },
    {
      "epoch": 30.391711609423787,
      "grad_norm": 0.26673129200935364,
      "learning_rate": 6.962134544422367e-05,
      "loss": 0.005115195363759995,
      "step": 107070
    },
    {
      "epoch": 30.394550099347146,
      "grad_norm": 5.067689895629883,
      "learning_rate": 6.961850695430031e-05,
      "loss": 0.005962491407990456,
      "step": 107080
    },
    {
      "epoch": 30.39738858927051,
      "grad_norm": 2.64968204498291,
      "learning_rate": 6.961566846437695e-05,
      "loss": 0.003581015020608902,
      "step": 107090
    },
    {
      "epoch": 30.40022707919387,
      "grad_norm": 3.2901649475097656,
      "learning_rate": 6.96128299744536e-05,
      "loss": 0.008079561591148376,
      "step": 107100
    },
    {
      "epoch": 30.40306556911723,
      "grad_norm": 11.252418518066406,
      "learning_rate": 6.960999148453024e-05,
      "loss": 0.007207413762807846,
      "step": 107110
    },
    {
      "epoch": 30.40590405904059,
      "grad_norm": 0.093897745013237,
      "learning_rate": 6.960715299460688e-05,
      "loss": 0.0017349829897284508,
      "step": 107120
    },
    {
      "epoch": 30.40874254896395,
      "grad_norm": 0.22159616649150848,
      "learning_rate": 6.96043145046835e-05,
      "loss": 0.006933601945638657,
      "step": 107130
    },
    {
      "epoch": 30.411581038887313,
      "grad_norm": 4.26894998550415,
      "learning_rate": 6.960147601476015e-05,
      "loss": 0.012655366957187653,
      "step": 107140
    },
    {
      "epoch": 30.414419528810672,
      "grad_norm": 0.9774757623672485,
      "learning_rate": 6.959863752483679e-05,
      "loss": 0.010772609710693359,
      "step": 107150
    },
    {
      "epoch": 30.417258018734035,
      "grad_norm": 0.05760965496301651,
      "learning_rate": 6.959579903491343e-05,
      "loss": 0.00941249281167984,
      "step": 107160
    },
    {
      "epoch": 30.420096508657394,
      "grad_norm": 0.2402833253145218,
      "learning_rate": 6.959296054499007e-05,
      "loss": 0.004643053561449051,
      "step": 107170
    },
    {
      "epoch": 30.422934998580754,
      "grad_norm": 0.2679772675037384,
      "learning_rate": 6.959012205506671e-05,
      "loss": 0.004228454828262329,
      "step": 107180
    },
    {
      "epoch": 30.425773488504117,
      "grad_norm": 0.06389496475458145,
      "learning_rate": 6.958728356514334e-05,
      "loss": 0.009680062532424927,
      "step": 107190
    },
    {
      "epoch": 30.428611978427476,
      "grad_norm": 0.20385506749153137,
      "learning_rate": 6.958444507521998e-05,
      "loss": 0.003939318656921387,
      "step": 107200
    },
    {
      "epoch": 30.43145046835084,
      "grad_norm": 0.24842417240142822,
      "learning_rate": 6.958160658529662e-05,
      "loss": 0.00550590455532074,
      "step": 107210
    },
    {
      "epoch": 30.434288958274198,
      "grad_norm": 12.713326454162598,
      "learning_rate": 6.957876809537326e-05,
      "loss": 0.012340693920850753,
      "step": 107220
    },
    {
      "epoch": 30.437127448197558,
      "grad_norm": 4.94619083404541,
      "learning_rate": 6.95759296054499e-05,
      "loss": 0.005766192823648453,
      "step": 107230
    },
    {
      "epoch": 30.43996593812092,
      "grad_norm": 0.2717863917350769,
      "learning_rate": 6.957309111552655e-05,
      "loss": 0.006558281928300857,
      "step": 107240
    },
    {
      "epoch": 30.44280442804428,
      "grad_norm": 1.4751020669937134,
      "learning_rate": 6.957025262560319e-05,
      "loss": 0.0044717960059642795,
      "step": 107250
    },
    {
      "epoch": 30.445642917967643,
      "grad_norm": 0.2662867307662964,
      "learning_rate": 6.956741413567982e-05,
      "loss": 0.004062313959002495,
      "step": 107260
    },
    {
      "epoch": 30.448481407891002,
      "grad_norm": 7.888032913208008,
      "learning_rate": 6.956457564575646e-05,
      "loss": 0.009430086612701416,
      "step": 107270
    },
    {
      "epoch": 30.45131989781436,
      "grad_norm": 0.8077722787857056,
      "learning_rate": 6.95617371558331e-05,
      "loss": 0.003736356645822525,
      "step": 107280
    },
    {
      "epoch": 30.454158387737724,
      "grad_norm": 0.28179091215133667,
      "learning_rate": 6.955889866590974e-05,
      "loss": 0.004337352514266968,
      "step": 107290
    },
    {
      "epoch": 30.456996877661084,
      "grad_norm": 0.07351366430521011,
      "learning_rate": 6.955606017598638e-05,
      "loss": 0.007281792908906936,
      "step": 107300
    },
    {
      "epoch": 30.459835367584446,
      "grad_norm": 0.7152825593948364,
      "learning_rate": 6.955322168606302e-05,
      "loss": 0.011105398088693619,
      "step": 107310
    },
    {
      "epoch": 30.462673857507806,
      "grad_norm": 8.032296180725098,
      "learning_rate": 6.955038319613965e-05,
      "loss": 0.006558387726545334,
      "step": 107320
    },
    {
      "epoch": 30.465512347431165,
      "grad_norm": 0.055718906223773956,
      "learning_rate": 6.954754470621629e-05,
      "loss": 0.007521789520978928,
      "step": 107330
    },
    {
      "epoch": 30.468350837354528,
      "grad_norm": 1.203129768371582,
      "learning_rate": 6.954470621629293e-05,
      "loss": 0.004936205223202705,
      "step": 107340
    },
    {
      "epoch": 30.471189327277887,
      "grad_norm": 0.5501689910888672,
      "learning_rate": 6.954186772636958e-05,
      "loss": 0.002939385548233986,
      "step": 107350
    },
    {
      "epoch": 30.47402781720125,
      "grad_norm": 0.2961748242378235,
      "learning_rate": 6.953902923644622e-05,
      "loss": 0.006236843764781952,
      "step": 107360
    },
    {
      "epoch": 30.47686630712461,
      "grad_norm": 1.6446514129638672,
      "learning_rate": 6.953619074652286e-05,
      "loss": 0.004973116517066956,
      "step": 107370
    },
    {
      "epoch": 30.47970479704797,
      "grad_norm": 9.791794776916504,
      "learning_rate": 6.95333522565995e-05,
      "loss": 0.009722109884023666,
      "step": 107380
    },
    {
      "epoch": 30.48254328697133,
      "grad_norm": 9.704020500183105,
      "learning_rate": 6.953051376667613e-05,
      "loss": 0.002977362275123596,
      "step": 107390
    },
    {
      "epoch": 30.48538177689469,
      "grad_norm": 6.575239658355713,
      "learning_rate": 6.952767527675277e-05,
      "loss": 0.0028316739946603773,
      "step": 107400
    },
    {
      "epoch": 30.488220266818054,
      "grad_norm": 0.3652830719947815,
      "learning_rate": 6.952483678682941e-05,
      "loss": 0.002178196795284748,
      "step": 107410
    },
    {
      "epoch": 30.491058756741413,
      "grad_norm": 1.559021234512329,
      "learning_rate": 6.952199829690605e-05,
      "loss": 0.004749777540564537,
      "step": 107420
    },
    {
      "epoch": 30.493897246664773,
      "grad_norm": 0.36734530329704285,
      "learning_rate": 6.951915980698269e-05,
      "loss": 0.029553133249282836,
      "step": 107430
    },
    {
      "epoch": 30.496735736588136,
      "grad_norm": 1.6299757957458496,
      "learning_rate": 6.951632131705933e-05,
      "loss": 0.023069895803928375,
      "step": 107440
    },
    {
      "epoch": 30.499574226511495,
      "grad_norm": 2.6239864826202393,
      "learning_rate": 6.951348282713596e-05,
      "loss": 0.014397625625133515,
      "step": 107450
    },
    {
      "epoch": 30.502412716434858,
      "grad_norm": 3.03676176071167,
      "learning_rate": 6.95106443372126e-05,
      "loss": 0.01890326291322708,
      "step": 107460
    },
    {
      "epoch": 30.505251206358217,
      "grad_norm": 16.67452049255371,
      "learning_rate": 6.950780584728925e-05,
      "loss": 0.011852525174617767,
      "step": 107470
    },
    {
      "epoch": 30.50808969628158,
      "grad_norm": 0.45365962386131287,
      "learning_rate": 6.950496735736589e-05,
      "loss": 0.01840576231479645,
      "step": 107480
    },
    {
      "epoch": 30.51092818620494,
      "grad_norm": 0.7891889810562134,
      "learning_rate": 6.950241271643487e-05,
      "loss": 0.014406821131706238,
      "step": 107490
    },
    {
      "epoch": 30.5137666761283,
      "grad_norm": 0.30091020464897156,
      "learning_rate": 6.94995742265115e-05,
      "loss": 0.008165588229894638,
      "step": 107500
    },
    {
      "epoch": 30.5137666761283,
      "eval_accuracy": 0.972976409995549,
      "eval_loss": 0.09441757947206497,
      "eval_runtime": 31.896,
      "eval_samples_per_second": 493.071,
      "eval_steps_per_second": 7.713,
      "step": 107500
    },
    {
      "epoch": 30.51660516605166,
      "grad_norm": 0.29898929595947266,
      "learning_rate": 6.949673573658814e-05,
      "loss": 0.006231321394443512,
      "step": 107510
    },
    {
      "epoch": 30.51944365597502,
      "grad_norm": 13.883907318115234,
      "learning_rate": 6.949389724666478e-05,
      "loss": 0.02149525284767151,
      "step": 107520
    },
    {
      "epoch": 30.522282145898384,
      "grad_norm": 1.4167956113815308,
      "learning_rate": 6.949105875674142e-05,
      "loss": 0.010841847956180572,
      "step": 107530
    },
    {
      "epoch": 30.525120635821743,
      "grad_norm": 0.19293639063835144,
      "learning_rate": 6.948822026681806e-05,
      "loss": 0.03395317494869232,
      "step": 107540
    },
    {
      "epoch": 30.527959125745102,
      "grad_norm": 4.602264881134033,
      "learning_rate": 6.94853817768947e-05,
      "loss": 0.027455607056617738,
      "step": 107550
    },
    {
      "epoch": 30.530797615668465,
      "grad_norm": 2.570601463317871,
      "learning_rate": 6.948254328697133e-05,
      "loss": 0.00917329341173172,
      "step": 107560
    },
    {
      "epoch": 30.533636105591825,
      "grad_norm": 9.291815757751465,
      "learning_rate": 6.947970479704797e-05,
      "loss": 0.016403791308403016,
      "step": 107570
    },
    {
      "epoch": 30.536474595515188,
      "grad_norm": 17.26911163330078,
      "learning_rate": 6.947686630712461e-05,
      "loss": 0.019516144692897797,
      "step": 107580
    },
    {
      "epoch": 30.539313085438547,
      "grad_norm": 0.5098200440406799,
      "learning_rate": 6.947402781720125e-05,
      "loss": 0.01340814381837845,
      "step": 107590
    },
    {
      "epoch": 30.542151575361906,
      "grad_norm": 2.5358641147613525,
      "learning_rate": 6.94711893272779e-05,
      "loss": 0.014878882467746735,
      "step": 107600
    },
    {
      "epoch": 30.54499006528527,
      "grad_norm": 2.153550386428833,
      "learning_rate": 6.946835083735454e-05,
      "loss": 0.005201190710067749,
      "step": 107610
    },
    {
      "epoch": 30.54782855520863,
      "grad_norm": 8.61658000946045,
      "learning_rate": 6.946551234743118e-05,
      "loss": 0.008176913857460022,
      "step": 107620
    },
    {
      "epoch": 30.55066704513199,
      "grad_norm": 0.9009307026863098,
      "learning_rate": 6.94626738575078e-05,
      "loss": 0.007943210005760194,
      "step": 107630
    },
    {
      "epoch": 30.55350553505535,
      "grad_norm": 0.7712855339050293,
      "learning_rate": 6.945983536758445e-05,
      "loss": 0.006427846848964691,
      "step": 107640
    },
    {
      "epoch": 30.55634402497871,
      "grad_norm": 1.9485176801681519,
      "learning_rate": 6.945699687766109e-05,
      "loss": 0.00943184643983841,
      "step": 107650
    },
    {
      "epoch": 30.559182514902073,
      "grad_norm": 0.13528601825237274,
      "learning_rate": 6.945415838773772e-05,
      "loss": 0.00398874245584011,
      "step": 107660
    },
    {
      "epoch": 30.562021004825432,
      "grad_norm": 4.162069797515869,
      "learning_rate": 6.945131989781437e-05,
      "loss": 0.0052512742578983305,
      "step": 107670
    },
    {
      "epoch": 30.564859494748795,
      "grad_norm": 0.6233922839164734,
      "learning_rate": 6.944848140789101e-05,
      "loss": 0.0026118399575352667,
      "step": 107680
    },
    {
      "epoch": 30.567697984672154,
      "grad_norm": 0.3950548470020294,
      "learning_rate": 6.944564291796764e-05,
      "loss": 0.01069374978542328,
      "step": 107690
    },
    {
      "epoch": 30.570536474595514,
      "grad_norm": 2.6571319103240967,
      "learning_rate": 6.944280442804428e-05,
      "loss": 0.00518101379275322,
      "step": 107700
    },
    {
      "epoch": 30.573374964518877,
      "grad_norm": 0.1842406690120697,
      "learning_rate": 6.943996593812092e-05,
      "loss": 0.006627582758665085,
      "step": 107710
    },
    {
      "epoch": 30.576213454442236,
      "grad_norm": 0.6428417563438416,
      "learning_rate": 6.943712744819756e-05,
      "loss": 0.004073525220155716,
      "step": 107720
    },
    {
      "epoch": 30.5790519443656,
      "grad_norm": 1.0376739501953125,
      "learning_rate": 6.943428895827419e-05,
      "loss": 0.007203148305416107,
      "step": 107730
    },
    {
      "epoch": 30.581890434288958,
      "grad_norm": 0.8839228749275208,
      "learning_rate": 6.943145046835085e-05,
      "loss": 0.0031544677913188934,
      "step": 107740
    },
    {
      "epoch": 30.584728924212317,
      "grad_norm": 0.24821701645851135,
      "learning_rate": 6.942861197842749e-05,
      "loss": 0.010313296318054199,
      "step": 107750
    },
    {
      "epoch": 30.58756741413568,
      "grad_norm": 0.30817911028862,
      "learning_rate": 6.942577348850412e-05,
      "loss": 0.00305517241358757,
      "step": 107760
    },
    {
      "epoch": 30.59040590405904,
      "grad_norm": 0.1983669251203537,
      "learning_rate": 6.942293499858076e-05,
      "loss": 0.0030417123809456824,
      "step": 107770
    },
    {
      "epoch": 30.593244393982403,
      "grad_norm": 0.060916002839803696,
      "learning_rate": 6.94200965086574e-05,
      "loss": 0.006653342396020889,
      "step": 107780
    },
    {
      "epoch": 30.596082883905762,
      "grad_norm": 9.371432304382324,
      "learning_rate": 6.941725801873403e-05,
      "loss": 0.011831273138523103,
      "step": 107790
    },
    {
      "epoch": 30.59892137382912,
      "grad_norm": 2.3389103412628174,
      "learning_rate": 6.941441952881068e-05,
      "loss": 0.012470263242721557,
      "step": 107800
    },
    {
      "epoch": 30.601759863752484,
      "grad_norm": 1.4401503801345825,
      "learning_rate": 6.941158103888732e-05,
      "loss": 0.006735991686582565,
      "step": 107810
    },
    {
      "epoch": 30.604598353675843,
      "grad_norm": 5.884941577911377,
      "learning_rate": 6.940874254896395e-05,
      "loss": 0.01686566472053528,
      "step": 107820
    },
    {
      "epoch": 30.607436843599206,
      "grad_norm": 0.4924392104148865,
      "learning_rate": 6.940590405904059e-05,
      "loss": 0.013892973959445953,
      "step": 107830
    },
    {
      "epoch": 30.610275333522566,
      "grad_norm": 4.052254676818848,
      "learning_rate": 6.940306556911723e-05,
      "loss": 0.01976458877325058,
      "step": 107840
    },
    {
      "epoch": 30.61311382344593,
      "grad_norm": 0.05058027431368828,
      "learning_rate": 6.940022707919387e-05,
      "loss": 0.003322318196296692,
      "step": 107850
    },
    {
      "epoch": 30.615952313369288,
      "grad_norm": 11.225318908691406,
      "learning_rate": 6.93973885892705e-05,
      "loss": 0.016716179251670838,
      "step": 107860
    },
    {
      "epoch": 30.618790803292647,
      "grad_norm": 3.4344892501831055,
      "learning_rate": 6.939455009934716e-05,
      "loss": 0.028201863169670105,
      "step": 107870
    },
    {
      "epoch": 30.62162929321601,
      "grad_norm": 0.6744145750999451,
      "learning_rate": 6.93917116094238e-05,
      "loss": 0.009844663739204406,
      "step": 107880
    },
    {
      "epoch": 30.62446778313937,
      "grad_norm": 0.9063137173652649,
      "learning_rate": 6.938887311950043e-05,
      "loss": 0.007519689947366714,
      "step": 107890
    },
    {
      "epoch": 30.627306273062732,
      "grad_norm": 0.0644942969083786,
      "learning_rate": 6.938603462957707e-05,
      "loss": 0.0030784472823143006,
      "step": 107900
    },
    {
      "epoch": 30.63014476298609,
      "grad_norm": 0.17795416712760925,
      "learning_rate": 6.938319613965371e-05,
      "loss": 0.00314871184527874,
      "step": 107910
    },
    {
      "epoch": 30.63298325290945,
      "grad_norm": 2.768763780593872,
      "learning_rate": 6.938035764973034e-05,
      "loss": 0.01097598671913147,
      "step": 107920
    },
    {
      "epoch": 30.635821742832814,
      "grad_norm": 1.2334949970245361,
      "learning_rate": 6.937751915980698e-05,
      "loss": 0.006962595880031586,
      "step": 107930
    },
    {
      "epoch": 30.638660232756173,
      "grad_norm": 2.8110365867614746,
      "learning_rate": 6.937468066988363e-05,
      "loss": 0.00794026106595993,
      "step": 107940
    },
    {
      "epoch": 30.641498722679536,
      "grad_norm": 3.5465445518493652,
      "learning_rate": 6.937184217996026e-05,
      "loss": 0.008417610824108124,
      "step": 107950
    },
    {
      "epoch": 30.644337212602895,
      "grad_norm": 0.07554007321596146,
      "learning_rate": 6.93690036900369e-05,
      "loss": 0.009914899617433548,
      "step": 107960
    },
    {
      "epoch": 30.647175702526255,
      "grad_norm": 0.6295055747032166,
      "learning_rate": 6.936616520011354e-05,
      "loss": 0.0012453913688659668,
      "step": 107970
    },
    {
      "epoch": 30.650014192449618,
      "grad_norm": 8.498252868652344,
      "learning_rate": 6.936332671019019e-05,
      "loss": 0.00815141424536705,
      "step": 107980
    },
    {
      "epoch": 30.652852682372977,
      "grad_norm": 0.23362283408641815,
      "learning_rate": 6.936048822026681e-05,
      "loss": 0.0049035467207431795,
      "step": 107990
    },
    {
      "epoch": 30.65569117229634,
      "grad_norm": 1.8322539329528809,
      "learning_rate": 6.935764973034347e-05,
      "loss": 0.002581584453582764,
      "step": 108000
    },
    {
      "epoch": 30.65569117229634,
      "eval_accuracy": 0.9760284860431105,
      "eval_loss": 0.08486516773700714,
      "eval_runtime": 32.7396,
      "eval_samples_per_second": 480.366,
      "eval_steps_per_second": 7.514,
      "step": 108000
    },
    {
      "epoch": 30.6585296622197,
      "grad_norm": 0.3404947519302368,
      "learning_rate": 6.935481124042011e-05,
      "loss": 0.010818032175302505,
      "step": 108010
    },
    {
      "epoch": 30.66136815214306,
      "grad_norm": 0.24520830810070038,
      "learning_rate": 6.935197275049674e-05,
      "loss": 0.00346316397190094,
      "step": 108020
    },
    {
      "epoch": 30.66420664206642,
      "grad_norm": 0.7958977818489075,
      "learning_rate": 6.934913426057338e-05,
      "loss": 0.007261732965707779,
      "step": 108030
    },
    {
      "epoch": 30.66704513198978,
      "grad_norm": 9.968762397766113,
      "learning_rate": 6.934629577065002e-05,
      "loss": 0.007016778737306595,
      "step": 108040
    },
    {
      "epoch": 30.669883621913144,
      "grad_norm": 2.3097550868988037,
      "learning_rate": 6.934345728072665e-05,
      "loss": 0.006440217792987824,
      "step": 108050
    },
    {
      "epoch": 30.672722111836503,
      "grad_norm": 0.1297004222869873,
      "learning_rate": 6.934061879080329e-05,
      "loss": 0.005982833728194237,
      "step": 108060
    },
    {
      "epoch": 30.675560601759862,
      "grad_norm": 1.8034029006958008,
      "learning_rate": 6.933778030087994e-05,
      "loss": 0.0028621893376111983,
      "step": 108070
    },
    {
      "epoch": 30.678399091683225,
      "grad_norm": 7.865789890289307,
      "learning_rate": 6.933494181095657e-05,
      "loss": 0.005302838236093521,
      "step": 108080
    },
    {
      "epoch": 30.681237581606585,
      "grad_norm": 4.873446941375732,
      "learning_rate": 6.933210332103321e-05,
      "loss": 0.007929010689258576,
      "step": 108090
    },
    {
      "epoch": 30.684076071529947,
      "grad_norm": 0.32345637679100037,
      "learning_rate": 6.932926483110985e-05,
      "loss": 0.006076682358980179,
      "step": 108100
    },
    {
      "epoch": 30.686914561453307,
      "grad_norm": 9.707159042358398,
      "learning_rate": 6.93264263411865e-05,
      "loss": 0.009933063387870788,
      "step": 108110
    },
    {
      "epoch": 30.689753051376666,
      "grad_norm": 0.2627275586128235,
      "learning_rate": 6.932358785126312e-05,
      "loss": 0.003286503255367279,
      "step": 108120
    },
    {
      "epoch": 30.69259154130003,
      "grad_norm": 0.389260858297348,
      "learning_rate": 6.932074936133977e-05,
      "loss": 0.03746757209300995,
      "step": 108130
    },
    {
      "epoch": 30.69543003122339,
      "grad_norm": 1.2788050174713135,
      "learning_rate": 6.931791087141642e-05,
      "loss": 0.0080003023147583,
      "step": 108140
    },
    {
      "epoch": 30.69826852114675,
      "grad_norm": 0.8984013199806213,
      "learning_rate": 6.931507238149305e-05,
      "loss": 0.010008556395769119,
      "step": 108150
    },
    {
      "epoch": 30.70110701107011,
      "grad_norm": 5.575248718261719,
      "learning_rate": 6.931223389156969e-05,
      "loss": 0.009293365478515624,
      "step": 108160
    },
    {
      "epoch": 30.70394550099347,
      "grad_norm": 6.0118584632873535,
      "learning_rate": 6.930939540164633e-05,
      "loss": 0.005042307451367379,
      "step": 108170
    },
    {
      "epoch": 30.706783990916833,
      "grad_norm": 0.4735325574874878,
      "learning_rate": 6.930655691172296e-05,
      "loss": 0.011386523395776749,
      "step": 108180
    },
    {
      "epoch": 30.709622480840192,
      "grad_norm": 15.680233001708984,
      "learning_rate": 6.93037184217996e-05,
      "loss": 0.017141708731651308,
      "step": 108190
    },
    {
      "epoch": 30.712460970763555,
      "grad_norm": 1.390152096748352,
      "learning_rate": 6.930087993187625e-05,
      "loss": 0.012368039041757584,
      "step": 108200
    },
    {
      "epoch": 30.715299460686914,
      "grad_norm": 0.25055795907974243,
      "learning_rate": 6.929804144195288e-05,
      "loss": 0.007332011312246323,
      "step": 108210
    },
    {
      "epoch": 30.718137950610274,
      "grad_norm": 0.04950626194477081,
      "learning_rate": 6.929520295202952e-05,
      "loss": 0.0061017446219921116,
      "step": 108220
    },
    {
      "epoch": 30.720976440533637,
      "grad_norm": 7.89745569229126,
      "learning_rate": 6.929236446210617e-05,
      "loss": 0.010081503540277481,
      "step": 108230
    },
    {
      "epoch": 30.723814930456996,
      "grad_norm": 15.37328052520752,
      "learning_rate": 6.928952597218281e-05,
      "loss": 0.012599964439868928,
      "step": 108240
    },
    {
      "epoch": 30.72665342038036,
      "grad_norm": 0.20166586339473724,
      "learning_rate": 6.928668748225943e-05,
      "loss": 0.0031342938542366027,
      "step": 108250
    },
    {
      "epoch": 30.729491910303718,
      "grad_norm": 4.87388277053833,
      "learning_rate": 6.928384899233608e-05,
      "loss": 0.005464524775743484,
      "step": 108260
    },
    {
      "epoch": 30.732330400227077,
      "grad_norm": 0.41176560521125793,
      "learning_rate": 6.928101050241273e-05,
      "loss": 0.004394324496388435,
      "step": 108270
    },
    {
      "epoch": 30.73516889015044,
      "grad_norm": 1.5845898389816284,
      "learning_rate": 6.927817201248936e-05,
      "loss": 0.016941487789154053,
      "step": 108280
    },
    {
      "epoch": 30.7380073800738,
      "grad_norm": 3.9480695724487305,
      "learning_rate": 6.9275333522566e-05,
      "loss": 0.004984812065958977,
      "step": 108290
    },
    {
      "epoch": 30.740845869997163,
      "grad_norm": 0.4992816746234894,
      "learning_rate": 6.927249503264264e-05,
      "loss": 0.006943334639072418,
      "step": 108300
    },
    {
      "epoch": 30.743684359920522,
      "grad_norm": 0.29689881205558777,
      "learning_rate": 6.926965654271927e-05,
      "loss": 0.01232173889875412,
      "step": 108310
    },
    {
      "epoch": 30.746522849843885,
      "grad_norm": 0.48550698161125183,
      "learning_rate": 6.926681805279591e-05,
      "loss": 0.0021306611597537994,
      "step": 108320
    },
    {
      "epoch": 30.749361339767244,
      "grad_norm": 1.0176149606704712,
      "learning_rate": 6.926397956287255e-05,
      "loss": 0.005108620971441269,
      "step": 108330
    },
    {
      "epoch": 30.752199829690603,
      "grad_norm": 0.7577868700027466,
      "learning_rate": 6.92611410729492e-05,
      "loss": 0.004457224905490875,
      "step": 108340
    },
    {
      "epoch": 30.755038319613966,
      "grad_norm": 0.5519736409187317,
      "learning_rate": 6.925830258302583e-05,
      "loss": 0.0028335295617580416,
      "step": 108350
    },
    {
      "epoch": 30.757876809537326,
      "grad_norm": 0.5229926705360413,
      "learning_rate": 6.925546409310248e-05,
      "loss": 0.0112568698823452,
      "step": 108360
    },
    {
      "epoch": 30.76071529946069,
      "grad_norm": 1.1270365715026855,
      "learning_rate": 6.925262560317912e-05,
      "loss": 0.0031514562666416167,
      "step": 108370
    },
    {
      "epoch": 30.763553789384048,
      "grad_norm": 0.4757707715034485,
      "learning_rate": 6.924978711325575e-05,
      "loss": 0.006855834275484085,
      "step": 108380
    },
    {
      "epoch": 30.766392279307407,
      "grad_norm": 4.540040016174316,
      "learning_rate": 6.924694862333239e-05,
      "loss": 0.003883615881204605,
      "step": 108390
    },
    {
      "epoch": 30.76923076923077,
      "grad_norm": 10.909433364868164,
      "learning_rate": 6.924411013340903e-05,
      "loss": 0.011317004263401032,
      "step": 108400
    },
    {
      "epoch": 30.77206925915413,
      "grad_norm": 1.481785535812378,
      "learning_rate": 6.924127164348567e-05,
      "loss": 0.007198366522789002,
      "step": 108410
    },
    {
      "epoch": 30.774907749077492,
      "grad_norm": 7.005059242248535,
      "learning_rate": 6.923843315356231e-05,
      "loss": 0.01086823120713234,
      "step": 108420
    },
    {
      "epoch": 30.77774623900085,
      "grad_norm": 6.113331317901611,
      "learning_rate": 6.923559466363895e-05,
      "loss": 0.009817898273468018,
      "step": 108430
    },
    {
      "epoch": 30.78058472892421,
      "grad_norm": 6.23679780960083,
      "learning_rate": 6.923275617371558e-05,
      "loss": 0.0035656318068504333,
      "step": 108440
    },
    {
      "epoch": 30.783423218847574,
      "grad_norm": 0.08819983154535294,
      "learning_rate": 6.922991768379222e-05,
      "loss": 0.013261735439300537,
      "step": 108450
    },
    {
      "epoch": 30.786261708770933,
      "grad_norm": 0.16571711003780365,
      "learning_rate": 6.922707919386886e-05,
      "loss": 0.0022543082013726234,
      "step": 108460
    },
    {
      "epoch": 30.789100198694296,
      "grad_norm": 0.45750248432159424,
      "learning_rate": 6.92242407039455e-05,
      "loss": 0.009078514575958253,
      "step": 108470
    },
    {
      "epoch": 30.791938688617655,
      "grad_norm": 0.2267298698425293,
      "learning_rate": 6.922140221402215e-05,
      "loss": 0.0027978764846920967,
      "step": 108480
    },
    {
      "epoch": 30.794777178541015,
      "grad_norm": 0.5764914155006409,
      "learning_rate": 6.921856372409879e-05,
      "loss": 0.0023072164505720137,
      "step": 108490
    },
    {
      "epoch": 30.797615668464378,
      "grad_norm": 0.5392889976501465,
      "learning_rate": 6.921572523417541e-05,
      "loss": 0.0019822876900434496,
      "step": 108500
    },
    {
      "epoch": 30.797615668464378,
      "eval_accuracy": 0.9725949004896038,
      "eval_loss": 0.09143594652414322,
      "eval_runtime": 31.8455,
      "eval_samples_per_second": 493.853,
      "eval_steps_per_second": 7.725,
      "step": 108500
    },
    {
      "epoch": 30.800454158387737,
      "grad_norm": 0.3037811517715454,
      "learning_rate": 6.921288674425206e-05,
      "loss": 0.004638592898845673,
      "step": 108510
    },
    {
      "epoch": 30.8032926483111,
      "grad_norm": 0.3628709614276886,
      "learning_rate": 6.92100482543287e-05,
      "loss": 0.007847178727388382,
      "step": 108520
    },
    {
      "epoch": 30.80613113823446,
      "grad_norm": 0.5309145450592041,
      "learning_rate": 6.920720976440534e-05,
      "loss": 0.007410767674446106,
      "step": 108530
    },
    {
      "epoch": 30.80896962815782,
      "grad_norm": 12.710382461547852,
      "learning_rate": 6.920437127448198e-05,
      "loss": 0.007016655802726745,
      "step": 108540
    },
    {
      "epoch": 30.81180811808118,
      "grad_norm": 0.19307665526866913,
      "learning_rate": 6.920153278455862e-05,
      "loss": 0.003414461761713028,
      "step": 108550
    },
    {
      "epoch": 30.81464660800454,
      "grad_norm": 6.276671409606934,
      "learning_rate": 6.919869429463526e-05,
      "loss": 0.007214243710041046,
      "step": 108560
    },
    {
      "epoch": 30.817485097927904,
      "grad_norm": 0.0478070005774498,
      "learning_rate": 6.919585580471189e-05,
      "loss": 0.006894369423389435,
      "step": 108570
    },
    {
      "epoch": 30.820323587851263,
      "grad_norm": 0.6948897242546082,
      "learning_rate": 6.919301731478853e-05,
      "loss": 0.012487723678350448,
      "step": 108580
    },
    {
      "epoch": 30.823162077774622,
      "grad_norm": 0.15150880813598633,
      "learning_rate": 6.919017882486517e-05,
      "loss": 0.003312503919005394,
      "step": 108590
    },
    {
      "epoch": 30.826000567697985,
      "grad_norm": 0.2102944403886795,
      "learning_rate": 6.918734033494182e-05,
      "loss": 0.009315693378448486,
      "step": 108600
    },
    {
      "epoch": 30.828839057621344,
      "grad_norm": 5.4872894287109375,
      "learning_rate": 6.918450184501846e-05,
      "loss": 0.012835185229778289,
      "step": 108610
    },
    {
      "epoch": 30.831677547544707,
      "grad_norm": 3.5935654640197754,
      "learning_rate": 6.91816633550951e-05,
      "loss": 0.007239504158496857,
      "step": 108620
    },
    {
      "epoch": 30.834516037468067,
      "grad_norm": 2.220975637435913,
      "learning_rate": 6.917882486517173e-05,
      "loss": 0.003723222389817238,
      "step": 108630
    },
    {
      "epoch": 30.837354527391426,
      "grad_norm": 1.1570467948913574,
      "learning_rate": 6.917598637524837e-05,
      "loss": 0.009163733571767807,
      "step": 108640
    },
    {
      "epoch": 30.84019301731479,
      "grad_norm": 5.570822238922119,
      "learning_rate": 6.917314788532501e-05,
      "loss": 0.005310343578457832,
      "step": 108650
    },
    {
      "epoch": 30.84303150723815,
      "grad_norm": 1.2894381284713745,
      "learning_rate": 6.917030939540165e-05,
      "loss": 0.011832579970359802,
      "step": 108660
    },
    {
      "epoch": 30.84586999716151,
      "grad_norm": 0.827042281627655,
      "learning_rate": 6.916747090547829e-05,
      "loss": 0.013782040774822235,
      "step": 108670
    },
    {
      "epoch": 30.84870848708487,
      "grad_norm": 3.7277839183807373,
      "learning_rate": 6.916463241555493e-05,
      "loss": 0.015850119292736053,
      "step": 108680
    },
    {
      "epoch": 30.851546977008233,
      "grad_norm": 0.08223532885313034,
      "learning_rate": 6.916179392563157e-05,
      "loss": 0.015784841775894166,
      "step": 108690
    },
    {
      "epoch": 30.854385466931593,
      "grad_norm": 6.215513229370117,
      "learning_rate": 6.91589554357082e-05,
      "loss": 0.011094256490468978,
      "step": 108700
    },
    {
      "epoch": 30.857223956854952,
      "grad_norm": 6.886241912841797,
      "learning_rate": 6.915611694578484e-05,
      "loss": 0.00689724013209343,
      "step": 108710
    },
    {
      "epoch": 30.860062446778315,
      "grad_norm": 1.544342279434204,
      "learning_rate": 6.915327845586148e-05,
      "loss": 0.003053635358810425,
      "step": 108720
    },
    {
      "epoch": 30.862900936701674,
      "grad_norm": 1.0471025705337524,
      "learning_rate": 6.915043996593813e-05,
      "loss": 0.008847501873970032,
      "step": 108730
    },
    {
      "epoch": 30.865739426625037,
      "grad_norm": 0.2514213025569916,
      "learning_rate": 6.914760147601477e-05,
      "loss": 0.0037189237773418427,
      "step": 108740
    },
    {
      "epoch": 30.868577916548396,
      "grad_norm": 0.21771681308746338,
      "learning_rate": 6.914476298609141e-05,
      "loss": 0.0018671870231628418,
      "step": 108750
    },
    {
      "epoch": 30.871416406471756,
      "grad_norm": 0.23573359847068787,
      "learning_rate": 6.914192449616804e-05,
      "loss": 0.00984133556485176,
      "step": 108760
    },
    {
      "epoch": 30.87425489639512,
      "grad_norm": 0.32714954018592834,
      "learning_rate": 6.913908600624468e-05,
      "loss": 0.005342333018779755,
      "step": 108770
    },
    {
      "epoch": 30.877093386318478,
      "grad_norm": 1.5743167400360107,
      "learning_rate": 6.913624751632132e-05,
      "loss": 0.008735018968582153,
      "step": 108780
    },
    {
      "epoch": 30.87993187624184,
      "grad_norm": 3.8770406246185303,
      "learning_rate": 6.913340902639796e-05,
      "loss": 0.0037653017789125443,
      "step": 108790
    },
    {
      "epoch": 30.8827703661652,
      "grad_norm": 0.1317831128835678,
      "learning_rate": 6.91305705364746e-05,
      "loss": 0.005857710540294647,
      "step": 108800
    },
    {
      "epoch": 30.88560885608856,
      "grad_norm": 0.2634664475917816,
      "learning_rate": 6.912773204655124e-05,
      "loss": 0.0039584323763847355,
      "step": 108810
    },
    {
      "epoch": 30.888447346011922,
      "grad_norm": 0.5718883275985718,
      "learning_rate": 6.912489355662788e-05,
      "loss": 0.004155462980270386,
      "step": 108820
    },
    {
      "epoch": 30.89128583593528,
      "grad_norm": 3.588263988494873,
      "learning_rate": 6.912205506670451e-05,
      "loss": 0.007106095552444458,
      "step": 108830
    },
    {
      "epoch": 30.894124325858645,
      "grad_norm": 0.23640671372413635,
      "learning_rate": 6.911921657678115e-05,
      "loss": 0.010854107141494752,
      "step": 108840
    },
    {
      "epoch": 30.896962815782004,
      "grad_norm": 0.3888101875782013,
      "learning_rate": 6.91163780868578e-05,
      "loss": 0.01080252230167389,
      "step": 108850
    },
    {
      "epoch": 30.899801305705363,
      "grad_norm": 2.3882203102111816,
      "learning_rate": 6.911353959693442e-05,
      "loss": 0.003141210228204727,
      "step": 108860
    },
    {
      "epoch": 30.902639795628726,
      "grad_norm": 6.844480514526367,
      "learning_rate": 6.911070110701108e-05,
      "loss": 0.0035825975239276884,
      "step": 108870
    },
    {
      "epoch": 30.905478285552086,
      "grad_norm": 0.12960387766361237,
      "learning_rate": 6.910786261708772e-05,
      "loss": 0.003118118643760681,
      "step": 108880
    },
    {
      "epoch": 30.90831677547545,
      "grad_norm": 7.557692527770996,
      "learning_rate": 6.910502412716435e-05,
      "loss": 0.009565930813550949,
      "step": 108890
    },
    {
      "epoch": 30.911155265398808,
      "grad_norm": 0.8376349806785583,
      "learning_rate": 6.910218563724099e-05,
      "loss": 0.004120534658432007,
      "step": 108900
    },
    {
      "epoch": 30.913993755322167,
      "grad_norm": 12.440242767333984,
      "learning_rate": 6.909934714731763e-05,
      "loss": 0.006873039156198501,
      "step": 108910
    },
    {
      "epoch": 30.91683224524553,
      "grad_norm": 0.580194890499115,
      "learning_rate": 6.909650865739427e-05,
      "loss": 0.0025897204875946047,
      "step": 108920
    },
    {
      "epoch": 30.91967073516889,
      "grad_norm": 1.3408043384552002,
      "learning_rate": 6.909367016747091e-05,
      "loss": 0.002481282874941826,
      "step": 108930
    },
    {
      "epoch": 30.922509225092252,
      "grad_norm": 3.358201265335083,
      "learning_rate": 6.909083167754755e-05,
      "loss": 0.0063421323895454405,
      "step": 108940
    },
    {
      "epoch": 30.92534771501561,
      "grad_norm": 4.337606906890869,
      "learning_rate": 6.90879931876242e-05,
      "loss": 0.005390232801437378,
      "step": 108950
    },
    {
      "epoch": 30.92818620493897,
      "grad_norm": 5.306088924407959,
      "learning_rate": 6.908515469770082e-05,
      "loss": 0.006689278036355972,
      "step": 108960
    },
    {
      "epoch": 30.931024694862334,
      "grad_norm": 1.0455058813095093,
      "learning_rate": 6.908231620777746e-05,
      "loss": 0.013076059520244598,
      "step": 108970
    },
    {
      "epoch": 30.933863184785693,
      "grad_norm": 8.647638320922852,
      "learning_rate": 6.90794777178541e-05,
      "loss": 0.01877009719610214,
      "step": 108980
    },
    {
      "epoch": 30.936701674709056,
      "grad_norm": 1.3454164266586304,
      "learning_rate": 6.907663922793073e-05,
      "loss": 0.013234955072402955,
      "step": 108990
    },
    {
      "epoch": 30.939540164632415,
      "grad_norm": 9.878752708435059,
      "learning_rate": 6.907380073800739e-05,
      "loss": 0.006241678074002266,
      "step": 109000
    },
    {
      "epoch": 30.939540164632415,
      "eval_accuracy": 0.9729128250778916,
      "eval_loss": 0.09103207290172577,
      "eval_runtime": 31.5111,
      "eval_samples_per_second": 499.094,
      "eval_steps_per_second": 7.807,
      "step": 109000
    },
    {
      "epoch": 30.942378654555775,
      "grad_norm": 2.9619123935699463,
      "learning_rate": 6.907096224808403e-05,
      "loss": 0.010336744785308837,
      "step": 109010
    },
    {
      "epoch": 30.945217144479138,
      "grad_norm": 8.400907516479492,
      "learning_rate": 6.906812375816066e-05,
      "loss": 0.018994046747684477,
      "step": 109020
    },
    {
      "epoch": 30.948055634402497,
      "grad_norm": 1.5803632736206055,
      "learning_rate": 6.90652852682373e-05,
      "loss": 0.006797703355550766,
      "step": 109030
    },
    {
      "epoch": 30.95089412432586,
      "grad_norm": 0.19018879532814026,
      "learning_rate": 6.906244677831394e-05,
      "loss": 0.001114131696522236,
      "step": 109040
    },
    {
      "epoch": 30.95373261424922,
      "grad_norm": 1.1932650804519653,
      "learning_rate": 6.905960828839058e-05,
      "loss": 0.005487564206123352,
      "step": 109050
    },
    {
      "epoch": 30.956571104172582,
      "grad_norm": 0.868994414806366,
      "learning_rate": 6.905676979846721e-05,
      "loss": 0.006269472092390061,
      "step": 109060
    },
    {
      "epoch": 30.95940959409594,
      "grad_norm": 0.10979997366666794,
      "learning_rate": 6.905393130854386e-05,
      "loss": 0.002353961765766144,
      "step": 109070
    },
    {
      "epoch": 30.9622480840193,
      "grad_norm": 1.49248206615448,
      "learning_rate": 6.90510928186205e-05,
      "loss": 0.003878706693649292,
      "step": 109080
    },
    {
      "epoch": 30.965086573942664,
      "grad_norm": 9.79456901550293,
      "learning_rate": 6.904825432869713e-05,
      "loss": 0.011083433032035827,
      "step": 109090
    },
    {
      "epoch": 30.967925063866023,
      "grad_norm": 6.354611396789551,
      "learning_rate": 6.904541583877378e-05,
      "loss": 0.012200254946947098,
      "step": 109100
    },
    {
      "epoch": 30.970763553789386,
      "grad_norm": 5.914480209350586,
      "learning_rate": 6.904257734885042e-05,
      "loss": 0.01154486984014511,
      "step": 109110
    },
    {
      "epoch": 30.973602043712745,
      "grad_norm": 4.286909580230713,
      "learning_rate": 6.903973885892704e-05,
      "loss": 0.024362482130527496,
      "step": 109120
    },
    {
      "epoch": 30.976440533636104,
      "grad_norm": 11.273080825805664,
      "learning_rate": 6.90369003690037e-05,
      "loss": 0.007235788553953171,
      "step": 109130
    },
    {
      "epoch": 30.979279023559467,
      "grad_norm": 0.5949984192848206,
      "learning_rate": 6.903406187908034e-05,
      "loss": 0.0036271050572395324,
      "step": 109140
    },
    {
      "epoch": 30.982117513482827,
      "grad_norm": 1.39650297164917,
      "learning_rate": 6.903122338915697e-05,
      "loss": 0.007727381587028503,
      "step": 109150
    },
    {
      "epoch": 30.98495600340619,
      "grad_norm": 1.249563455581665,
      "learning_rate": 6.902838489923361e-05,
      "loss": 0.0018823854625225067,
      "step": 109160
    },
    {
      "epoch": 30.98779449332955,
      "grad_norm": 0.5204665660858154,
      "learning_rate": 6.902554640931025e-05,
      "loss": 0.005594105646014214,
      "step": 109170
    },
    {
      "epoch": 30.990632983252908,
      "grad_norm": 13.691636085510254,
      "learning_rate": 6.902270791938689e-05,
      "loss": 0.006807877123355866,
      "step": 109180
    },
    {
      "epoch": 30.99347147317627,
      "grad_norm": 1.05091392993927,
      "learning_rate": 6.901986942946352e-05,
      "loss": 0.0050909101963043215,
      "step": 109190
    },
    {
      "epoch": 30.99630996309963,
      "grad_norm": 0.4603828489780426,
      "learning_rate": 6.901703093954018e-05,
      "loss": 0.0030404597520828245,
      "step": 109200
    },
    {
      "epoch": 30.999148453022993,
      "grad_norm": 4.102952480316162,
      "learning_rate": 6.901419244961682e-05,
      "loss": 0.012967592477798462,
      "step": 109210
    },
    {
      "epoch": 31.001986942946353,
      "grad_norm": 0.3826521635055542,
      "learning_rate": 6.901135395969344e-05,
      "loss": 0.012312310189008713,
      "step": 109220
    },
    {
      "epoch": 31.004825432869712,
      "grad_norm": 0.3739717900753021,
      "learning_rate": 6.900851546977009e-05,
      "loss": 0.014077697694301606,
      "step": 109230
    },
    {
      "epoch": 31.007663922793075,
      "grad_norm": 3.453364133834839,
      "learning_rate": 6.900567697984673e-05,
      "loss": 0.011640869081020355,
      "step": 109240
    },
    {
      "epoch": 31.010502412716434,
      "grad_norm": 3.8308191299438477,
      "learning_rate": 6.900283848992336e-05,
      "loss": 0.007118683308362961,
      "step": 109250
    },
    {
      "epoch": 31.013340902639797,
      "grad_norm": 6.8070807456970215,
      "learning_rate": 6.9e-05,
      "loss": 0.011011771857738495,
      "step": 109260
    },
    {
      "epoch": 31.016179392563156,
      "grad_norm": 4.504818916320801,
      "learning_rate": 6.899716151007665e-05,
      "loss": 0.005058494210243225,
      "step": 109270
    },
    {
      "epoch": 31.019017882486516,
      "grad_norm": 10.797273635864258,
      "learning_rate": 6.899432302015328e-05,
      "loss": 0.010744699090719224,
      "step": 109280
    },
    {
      "epoch": 31.02185637240988,
      "grad_norm": 0.6540567278862,
      "learning_rate": 6.899148453022992e-05,
      "loss": 0.004901528358459473,
      "step": 109290
    },
    {
      "epoch": 31.024694862333238,
      "grad_norm": 0.7021021842956543,
      "learning_rate": 6.898864604030656e-05,
      "loss": 0.010893645882606506,
      "step": 109300
    },
    {
      "epoch": 31.0275333522566,
      "grad_norm": 2.056149482727051,
      "learning_rate": 6.89858075503832e-05,
      "loss": 0.0032764200121164324,
      "step": 109310
    },
    {
      "epoch": 31.03037184217996,
      "grad_norm": 1.4120187759399414,
      "learning_rate": 6.898296906045983e-05,
      "loss": 0.009088733047246934,
      "step": 109320
    },
    {
      "epoch": 31.03321033210332,
      "grad_norm": 0.5974423289299011,
      "learning_rate": 6.898013057053649e-05,
      "loss": 0.007695899903774261,
      "step": 109330
    },
    {
      "epoch": 31.036048822026682,
      "grad_norm": 2.229387044906616,
      "learning_rate": 6.897729208061311e-05,
      "loss": 0.0016472173854708671,
      "step": 109340
    },
    {
      "epoch": 31.03888731195004,
      "grad_norm": 0.854745090007782,
      "learning_rate": 6.897445359068976e-05,
      "loss": 0.00120619498193264,
      "step": 109350
    },
    {
      "epoch": 31.041725801873405,
      "grad_norm": 0.1090780571103096,
      "learning_rate": 6.89716151007664e-05,
      "loss": 0.004297414794564247,
      "step": 109360
    },
    {
      "epoch": 31.044564291796764,
      "grad_norm": 1.188860535621643,
      "learning_rate": 6.896877661084304e-05,
      "loss": 0.004518118128180504,
      "step": 109370
    },
    {
      "epoch": 31.047402781720123,
      "grad_norm": 0.9042288064956665,
      "learning_rate": 6.896593812091967e-05,
      "loss": 0.002831179276108742,
      "step": 109380
    },
    {
      "epoch": 31.050241271643486,
      "grad_norm": 1.4692059755325317,
      "learning_rate": 6.896309963099631e-05,
      "loss": 0.0025312095880508424,
      "step": 109390
    },
    {
      "epoch": 31.053079761566845,
      "grad_norm": 0.167169988155365,
      "learning_rate": 6.896026114107296e-05,
      "loss": 0.005077884346246719,
      "step": 109400
    },
    {
      "epoch": 31.05591825149021,
      "grad_norm": 4.306193828582764,
      "learning_rate": 6.895742265114959e-05,
      "loss": 0.007326644659042358,
      "step": 109410
    },
    {
      "epoch": 31.058756741413568,
      "grad_norm": 2.896209716796875,
      "learning_rate": 6.895458416122623e-05,
      "loss": 0.00741128996014595,
      "step": 109420
    },
    {
      "epoch": 31.061595231336927,
      "grad_norm": 0.09312529116868973,
      "learning_rate": 6.895174567130287e-05,
      "loss": 0.0058695215731859205,
      "step": 109430
    },
    {
      "epoch": 31.06443372126029,
      "grad_norm": 0.24507619440555573,
      "learning_rate": 6.89489071813795e-05,
      "loss": 0.0023714618757367132,
      "step": 109440
    },
    {
      "epoch": 31.06727221118365,
      "grad_norm": 1.8298733234405518,
      "learning_rate": 6.894606869145614e-05,
      "loss": 0.004635128378868103,
      "step": 109450
    },
    {
      "epoch": 31.070110701107012,
      "grad_norm": 4.069555282592773,
      "learning_rate": 6.894323020153278e-05,
      "loss": 0.00393756628036499,
      "step": 109460
    },
    {
      "epoch": 31.07294919103037,
      "grad_norm": 4.072540283203125,
      "learning_rate": 6.894039171160942e-05,
      "loss": 0.004617356136441231,
      "step": 109470
    },
    {
      "epoch": 31.075787680953734,
      "grad_norm": 0.8224989175796509,
      "learning_rate": 6.893755322168607e-05,
      "loss": 0.003340653330087662,
      "step": 109480
    },
    {
      "epoch": 31.078626170877094,
      "grad_norm": 0.061782754957675934,
      "learning_rate": 6.893471473176271e-05,
      "loss": 0.0017155501991510391,
      "step": 109490
    },
    {
      "epoch": 31.081464660800453,
      "grad_norm": 1.8758052587509155,
      "learning_rate": 6.893187624183935e-05,
      "loss": 0.006179077550768852,
      "step": 109500
    },
    {
      "epoch": 31.081464660800453,
      "eval_accuracy": 0.9671265975710561,
      "eval_loss": 0.11372263729572296,
      "eval_runtime": 31.2977,
      "eval_samples_per_second": 502.497,
      "eval_steps_per_second": 7.86,
      "step": 109500
    },
    {
      "epoch": 31.084303150723816,
      "grad_norm": 0.04453524947166443,
      "learning_rate": 6.892903775191598e-05,
      "loss": 0.004898843914270401,
      "step": 109510
    },
    {
      "epoch": 31.087141640647175,
      "grad_norm": 0.5262728929519653,
      "learning_rate": 6.892619926199262e-05,
      "loss": 0.0020381802693009376,
      "step": 109520
    },
    {
      "epoch": 31.089980130570538,
      "grad_norm": 0.6505150198936462,
      "learning_rate": 6.892336077206927e-05,
      "loss": 0.003230230510234833,
      "step": 109530
    },
    {
      "epoch": 31.092818620493897,
      "grad_norm": 8.726950645446777,
      "learning_rate": 6.89205222821459e-05,
      "loss": 0.004477530345320701,
      "step": 109540
    },
    {
      "epoch": 31.095657110417257,
      "grad_norm": 6.587113857269287,
      "learning_rate": 6.891768379222254e-05,
      "loss": 0.0033368490636348723,
      "step": 109550
    },
    {
      "epoch": 31.09849560034062,
      "grad_norm": 3.4111487865448,
      "learning_rate": 6.891484530229918e-05,
      "loss": 0.004448067769408226,
      "step": 109560
    },
    {
      "epoch": 31.10133409026398,
      "grad_norm": 9.678326606750488,
      "learning_rate": 6.891200681237581e-05,
      "loss": 0.0057912968099117276,
      "step": 109570
    },
    {
      "epoch": 31.104172580187342,
      "grad_norm": 0.32832619547843933,
      "learning_rate": 6.890916832245245e-05,
      "loss": 0.015910378098487853,
      "step": 109580
    },
    {
      "epoch": 31.1070110701107,
      "grad_norm": 0.5947364568710327,
      "learning_rate": 6.89063298325291e-05,
      "loss": 0.003208599239587784,
      "step": 109590
    },
    {
      "epoch": 31.10984956003406,
      "grad_norm": 0.36792829632759094,
      "learning_rate": 6.890349134260574e-05,
      "loss": 0.00662335455417633,
      "step": 109600
    },
    {
      "epoch": 31.112688049957423,
      "grad_norm": 0.19571690261363983,
      "learning_rate": 6.890065285268238e-05,
      "loss": 0.0040141865611076355,
      "step": 109610
    },
    {
      "epoch": 31.115526539880783,
      "grad_norm": 0.11512776464223862,
      "learning_rate": 6.889781436275902e-05,
      "loss": 0.014331360161304475,
      "step": 109620
    },
    {
      "epoch": 31.118365029804146,
      "grad_norm": 10.446988105773926,
      "learning_rate": 6.889497587283566e-05,
      "loss": 0.01079077273607254,
      "step": 109630
    },
    {
      "epoch": 31.121203519727505,
      "grad_norm": 1.0412211418151855,
      "learning_rate": 6.889213738291229e-05,
      "loss": 0.001989450864493847,
      "step": 109640
    },
    {
      "epoch": 31.124042009650864,
      "grad_norm": 0.4010404646396637,
      "learning_rate": 6.888929889298893e-05,
      "loss": 0.002150782383978367,
      "step": 109650
    },
    {
      "epoch": 31.126880499574227,
      "grad_norm": 0.5337924361228943,
      "learning_rate": 6.888646040306557e-05,
      "loss": 0.0021382132545113564,
      "step": 109660
    },
    {
      "epoch": 31.129718989497587,
      "grad_norm": 0.34064602851867676,
      "learning_rate": 6.888362191314221e-05,
      "loss": 0.003113337233662605,
      "step": 109670
    },
    {
      "epoch": 31.13255747942095,
      "grad_norm": 0.26821035146713257,
      "learning_rate": 6.888078342321885e-05,
      "loss": 0.0013156134635210037,
      "step": 109680
    },
    {
      "epoch": 31.13539596934431,
      "grad_norm": 0.3603932559490204,
      "learning_rate": 6.88779449332955e-05,
      "loss": 0.013469606637954712,
      "step": 109690
    },
    {
      "epoch": 31.138234459267668,
      "grad_norm": 0.2603108584880829,
      "learning_rate": 6.887510644337212e-05,
      "loss": 0.0018426725640892983,
      "step": 109700
    },
    {
      "epoch": 31.14107294919103,
      "grad_norm": 0.03381152078509331,
      "learning_rate": 6.887226795344876e-05,
      "loss": 0.0016591455787420273,
      "step": 109710
    },
    {
      "epoch": 31.14391143911439,
      "grad_norm": 1.1416629552841187,
      "learning_rate": 6.88694294635254e-05,
      "loss": 0.001323341391980648,
      "step": 109720
    },
    {
      "epoch": 31.146749929037753,
      "grad_norm": 0.07575463503599167,
      "learning_rate": 6.886659097360205e-05,
      "loss": 0.004193737357854843,
      "step": 109730
    },
    {
      "epoch": 31.149588418961113,
      "grad_norm": 1.0716884136199951,
      "learning_rate": 6.886375248367869e-05,
      "loss": 0.00312422476708889,
      "step": 109740
    },
    {
      "epoch": 31.152426908884472,
      "grad_norm": 13.743085861206055,
      "learning_rate": 6.886091399375533e-05,
      "loss": 0.009309540688991546,
      "step": 109750
    },
    {
      "epoch": 31.155265398807835,
      "grad_norm": 0.9184790253639221,
      "learning_rate": 6.885807550383197e-05,
      "loss": 0.007169279456138611,
      "step": 109760
    },
    {
      "epoch": 31.158103888731194,
      "grad_norm": 0.03757747262716293,
      "learning_rate": 6.88552370139086e-05,
      "loss": 0.011610660701990128,
      "step": 109770
    },
    {
      "epoch": 31.160942378654557,
      "grad_norm": 1.7496674060821533,
      "learning_rate": 6.885239852398524e-05,
      "loss": 0.007120468467473984,
      "step": 109780
    },
    {
      "epoch": 31.163780868577916,
      "grad_norm": 0.11229296773672104,
      "learning_rate": 6.884956003406188e-05,
      "loss": 0.003583366423845291,
      "step": 109790
    },
    {
      "epoch": 31.166619358501276,
      "grad_norm": 1.0766123533248901,
      "learning_rate": 6.884672154413852e-05,
      "loss": 0.006882242858409882,
      "step": 109800
    },
    {
      "epoch": 31.16945784842464,
      "grad_norm": 0.09572646021842957,
      "learning_rate": 6.884388305421516e-05,
      "loss": 0.0033465396612882615,
      "step": 109810
    },
    {
      "epoch": 31.172296338347998,
      "grad_norm": 0.6458649635314941,
      "learning_rate": 6.88410445642918e-05,
      "loss": 0.004496787860989571,
      "step": 109820
    },
    {
      "epoch": 31.17513482827136,
      "grad_norm": 0.2917235791683197,
      "learning_rate": 6.883820607436843e-05,
      "loss": 0.0022122312337160112,
      "step": 109830
    },
    {
      "epoch": 31.17797331819472,
      "grad_norm": 1.432818055152893,
      "learning_rate": 6.883536758444507e-05,
      "loss": 0.014394460618495942,
      "step": 109840
    },
    {
      "epoch": 31.18081180811808,
      "grad_norm": 0.40243399143218994,
      "learning_rate": 6.883252909452172e-05,
      "loss": 0.017813853919506073,
      "step": 109850
    },
    {
      "epoch": 31.183650298041442,
      "grad_norm": 0.10484733432531357,
      "learning_rate": 6.882969060459836e-05,
      "loss": 0.009522389620542526,
      "step": 109860
    },
    {
      "epoch": 31.1864887879648,
      "grad_norm": 1.1457058191299438,
      "learning_rate": 6.8826852114675e-05,
      "loss": 0.003856068104505539,
      "step": 109870
    },
    {
      "epoch": 31.189327277888165,
      "grad_norm": 0.12661395967006683,
      "learning_rate": 6.882401362475164e-05,
      "loss": 0.011374379694461822,
      "step": 109880
    },
    {
      "epoch": 31.192165767811524,
      "grad_norm": 1.457999348640442,
      "learning_rate": 6.882117513482828e-05,
      "loss": 0.00343543142080307,
      "step": 109890
    },
    {
      "epoch": 31.195004257734887,
      "grad_norm": 1.8819317817687988,
      "learning_rate": 6.881833664490491e-05,
      "loss": 0.006346265971660614,
      "step": 109900
    },
    {
      "epoch": 31.197842747658246,
      "grad_norm": 15.678508758544922,
      "learning_rate": 6.881549815498155e-05,
      "loss": 0.008815167099237442,
      "step": 109910
    },
    {
      "epoch": 31.200681237581605,
      "grad_norm": 0.033606600016355515,
      "learning_rate": 6.881265966505819e-05,
      "loss": 0.0013654526323080064,
      "step": 109920
    },
    {
      "epoch": 31.20351972750497,
      "grad_norm": 0.2993066608905792,
      "learning_rate": 6.880982117513483e-05,
      "loss": 0.004777495563030243,
      "step": 109930
    },
    {
      "epoch": 31.206358217428328,
      "grad_norm": 0.47858014702796936,
      "learning_rate": 6.880698268521147e-05,
      "loss": 0.01005203053355217,
      "step": 109940
    },
    {
      "epoch": 31.20919670735169,
      "grad_norm": 6.438141345977783,
      "learning_rate": 6.880414419528812e-05,
      "loss": 0.003955466300249099,
      "step": 109950
    },
    {
      "epoch": 31.21203519727505,
      "grad_norm": 0.46179136633872986,
      "learning_rate": 6.880130570536474e-05,
      "loss": 0.003509771078824997,
      "step": 109960
    },
    {
      "epoch": 31.21487368719841,
      "grad_norm": 1.218802571296692,
      "learning_rate": 6.879846721544139e-05,
      "loss": 0.00413697436451912,
      "step": 109970
    },
    {
      "epoch": 31.217712177121772,
      "grad_norm": 0.936011016368866,
      "learning_rate": 6.879562872551803e-05,
      "loss": 0.003762988746166229,
      "step": 109980
    },
    {
      "epoch": 31.22055066704513,
      "grad_norm": 0.21286432445049286,
      "learning_rate": 6.879279023559467e-05,
      "loss": 0.0033331245183944704,
      "step": 109990
    },
    {
      "epoch": 31.223389156968494,
      "grad_norm": 5.036059856414795,
      "learning_rate": 6.878995174567131e-05,
      "loss": 0.004612860083580017,
      "step": 110000
    },
    {
      "epoch": 31.223389156968494,
      "eval_accuracy": 0.9720862211483436,
      "eval_loss": 0.09537944197654724,
      "eval_runtime": 32.1478,
      "eval_samples_per_second": 489.209,
      "eval_steps_per_second": 7.652,
      "step": 110000
    },
    {
      "epoch": 31.226227646891854,
      "grad_norm": 0.325810045003891,
      "learning_rate": 6.878711325574795e-05,
      "loss": 0.005835617333650589,
      "step": 110010
    },
    {
      "epoch": 31.229066136815213,
      "grad_norm": 3.817789316177368,
      "learning_rate": 6.878427476582459e-05,
      "loss": 0.006984712183475494,
      "step": 110020
    },
    {
      "epoch": 31.231904626738576,
      "grad_norm": 0.7725151181221008,
      "learning_rate": 6.878143627590122e-05,
      "loss": 0.00346316397190094,
      "step": 110030
    },
    {
      "epoch": 31.234743116661935,
      "grad_norm": 3.954335927963257,
      "learning_rate": 6.877859778597786e-05,
      "loss": 0.0041181530803442,
      "step": 110040
    },
    {
      "epoch": 31.237581606585298,
      "grad_norm": 2.2062954902648926,
      "learning_rate": 6.87757592960545e-05,
      "loss": 0.011963410675525666,
      "step": 110050
    },
    {
      "epoch": 31.240420096508657,
      "grad_norm": 0.5785346031188965,
      "learning_rate": 6.877292080613114e-05,
      "loss": 0.007881715893745422,
      "step": 110060
    },
    {
      "epoch": 31.243258586432017,
      "grad_norm": 1.0986262559890747,
      "learning_rate": 6.877008231620779e-05,
      "loss": 0.004345748573541641,
      "step": 110070
    },
    {
      "epoch": 31.24609707635538,
      "grad_norm": 0.11184180527925491,
      "learning_rate": 6.876724382628443e-05,
      "loss": 0.0021884990856051445,
      "step": 110080
    },
    {
      "epoch": 31.24893556627874,
      "grad_norm": 0.5900527834892273,
      "learning_rate": 6.876440533636105e-05,
      "loss": 0.00928773283958435,
      "step": 110090
    },
    {
      "epoch": 31.251774056202102,
      "grad_norm": 0.3124460279941559,
      "learning_rate": 6.87615668464377e-05,
      "loss": 0.00806880071759224,
      "step": 110100
    },
    {
      "epoch": 31.25461254612546,
      "grad_norm": 0.2854596674442291,
      "learning_rate": 6.875872835651434e-05,
      "loss": 0.021215544641017915,
      "step": 110110
    },
    {
      "epoch": 31.25745103604882,
      "grad_norm": 13.262415885925293,
      "learning_rate": 6.875588986659098e-05,
      "loss": 0.007200486958026886,
      "step": 110120
    },
    {
      "epoch": 31.260289525972183,
      "grad_norm": 0.34954404830932617,
      "learning_rate": 6.875305137666762e-05,
      "loss": 0.01924653798341751,
      "step": 110130
    },
    {
      "epoch": 31.263128015895543,
      "grad_norm": 0.680980920791626,
      "learning_rate": 6.875021288674426e-05,
      "loss": 0.0025742601603269575,
      "step": 110140
    },
    {
      "epoch": 31.265966505818906,
      "grad_norm": 0.5135800242424011,
      "learning_rate": 6.87473743968209e-05,
      "loss": 0.011292635649442672,
      "step": 110150
    },
    {
      "epoch": 31.268804995742265,
      "grad_norm": 0.3259054720401764,
      "learning_rate": 6.874453590689753e-05,
      "loss": 0.0019381167367100716,
      "step": 110160
    },
    {
      "epoch": 31.271643485665624,
      "grad_norm": 0.1391202211380005,
      "learning_rate": 6.874169741697417e-05,
      "loss": 0.0026612916961312296,
      "step": 110170
    },
    {
      "epoch": 31.274481975588987,
      "grad_norm": 0.08063234388828278,
      "learning_rate": 6.873885892705081e-05,
      "loss": 0.00681048184633255,
      "step": 110180
    },
    {
      "epoch": 31.277320465512346,
      "grad_norm": 4.031476974487305,
      "learning_rate": 6.873602043712744e-05,
      "loss": 0.004949242621660232,
      "step": 110190
    },
    {
      "epoch": 31.28015895543571,
      "grad_norm": 0.2691288888454437,
      "learning_rate": 6.87331819472041e-05,
      "loss": 0.010847274214029312,
      "step": 110200
    },
    {
      "epoch": 31.28299744535907,
      "grad_norm": 1.322691559791565,
      "learning_rate": 6.873034345728074e-05,
      "loss": 0.005416613817214966,
      "step": 110210
    },
    {
      "epoch": 31.285835935282428,
      "grad_norm": 0.07893897593021393,
      "learning_rate": 6.872750496735737e-05,
      "loss": 0.0030096346512436868,
      "step": 110220
    },
    {
      "epoch": 31.28867442520579,
      "grad_norm": 1.8029356002807617,
      "learning_rate": 6.872495032642635e-05,
      "loss": 0.01636136770248413,
      "step": 110230
    },
    {
      "epoch": 31.29151291512915,
      "grad_norm": 0.17369554936885834,
      "learning_rate": 6.872211183650299e-05,
      "loss": 0.002129828371107578,
      "step": 110240
    },
    {
      "epoch": 31.294351405052513,
      "grad_norm": 0.037399448454380035,
      "learning_rate": 6.871927334657963e-05,
      "loss": 0.011778134852647781,
      "step": 110250
    },
    {
      "epoch": 31.297189894975872,
      "grad_norm": 1.994122862815857,
      "learning_rate": 6.871643485665627e-05,
      "loss": 0.0034459367394447326,
      "step": 110260
    },
    {
      "epoch": 31.300028384899235,
      "grad_norm": 0.6011873483657837,
      "learning_rate": 6.87135963667329e-05,
      "loss": 0.0038243308663368225,
      "step": 110270
    },
    {
      "epoch": 31.302866874822595,
      "grad_norm": 3.11600399017334,
      "learning_rate": 6.871075787680954e-05,
      "loss": 0.009884706884622573,
      "step": 110280
    },
    {
      "epoch": 31.305705364745954,
      "grad_norm": 8.012303352355957,
      "learning_rate": 6.870791938688618e-05,
      "loss": 0.007906264811754226,
      "step": 110290
    },
    {
      "epoch": 31.308543854669317,
      "grad_norm": 1.3132743835449219,
      "learning_rate": 6.870508089696281e-05,
      "loss": 0.005196478217840195,
      "step": 110300
    },
    {
      "epoch": 31.311382344592676,
      "grad_norm": 1.0238229036331177,
      "learning_rate": 6.870224240703946e-05,
      "loss": 0.015683819353580476,
      "step": 110310
    },
    {
      "epoch": 31.31422083451604,
      "grad_norm": 0.6103878617286682,
      "learning_rate": 6.86994039171161e-05,
      "loss": 0.00277481023222208,
      "step": 110320
    },
    {
      "epoch": 31.3170593244394,
      "grad_norm": 0.2669052183628082,
      "learning_rate": 6.869656542719273e-05,
      "loss": 0.0042536847293376924,
      "step": 110330
    },
    {
      "epoch": 31.319897814362758,
      "grad_norm": 9.841928482055664,
      "learning_rate": 6.869372693726937e-05,
      "loss": 0.008023180067539215,
      "step": 110340
    },
    {
      "epoch": 31.32273630428612,
      "grad_norm": 8.319967269897461,
      "learning_rate": 6.869088844734601e-05,
      "loss": 0.0038462065160274504,
      "step": 110350
    },
    {
      "epoch": 31.32557479420948,
      "grad_norm": 3.77957820892334,
      "learning_rate": 6.868804995742266e-05,
      "loss": 0.01015644371509552,
      "step": 110360
    },
    {
      "epoch": 31.328413284132843,
      "grad_norm": 0.8644422888755798,
      "learning_rate": 6.868521146749928e-05,
      "loss": 0.006607858836650849,
      "step": 110370
    },
    {
      "epoch": 31.331251774056202,
      "grad_norm": 1.9416636228561401,
      "learning_rate": 6.868237297757594e-05,
      "loss": 0.002058306522667408,
      "step": 110380
    },
    {
      "epoch": 31.33409026397956,
      "grad_norm": 2.39371919631958,
      "learning_rate": 6.867953448765258e-05,
      "loss": 0.008337047696113587,
      "step": 110390
    },
    {
      "epoch": 31.336928753902924,
      "grad_norm": 0.9490315914154053,
      "learning_rate": 6.867669599772921e-05,
      "loss": 0.003460328280925751,
      "step": 110400
    },
    {
      "epoch": 31.339767243826284,
      "grad_norm": 1.3871650695800781,
      "learning_rate": 6.867385750780585e-05,
      "loss": 0.008450895547866821,
      "step": 110410
    },
    {
      "epoch": 31.342605733749647,
      "grad_norm": 7.065817356109619,
      "learning_rate": 6.867101901788249e-05,
      "loss": 0.009674198925495148,
      "step": 110420
    },
    {
      "epoch": 31.345444223673006,
      "grad_norm": 5.321462631225586,
      "learning_rate": 6.866818052795912e-05,
      "loss": 0.0072591789066791534,
      "step": 110430
    },
    {
      "epoch": 31.348282713596365,
      "grad_norm": 0.13979879021644592,
      "learning_rate": 6.866534203803577e-05,
      "loss": 0.009426911175251008,
      "step": 110440
    },
    {
      "epoch": 31.351121203519728,
      "grad_norm": 5.368879795074463,
      "learning_rate": 6.866250354811241e-05,
      "loss": 0.005856205895543099,
      "step": 110450
    },
    {
      "epoch": 31.353959693443088,
      "grad_norm": 2.286088466644287,
      "learning_rate": 6.865966505818904e-05,
      "loss": 0.002498668059706688,
      "step": 110460
    },
    {
      "epoch": 31.35679818336645,
      "grad_norm": 0.03411642089486122,
      "learning_rate": 6.865682656826568e-05,
      "loss": 0.009229302406311035,
      "step": 110470
    },
    {
      "epoch": 31.35963667328981,
      "grad_norm": 0.13410453498363495,
      "learning_rate": 6.865398807834233e-05,
      "loss": 0.003095555305480957,
      "step": 110480
    },
    {
      "epoch": 31.36247516321317,
      "grad_norm": 8.528481483459473,
      "learning_rate": 6.865114958841897e-05,
      "loss": 0.00674191266298294,
      "step": 110490
    },
    {
      "epoch": 31.365313653136532,
      "grad_norm": 1.5840367078781128,
      "learning_rate": 6.86483110984956e-05,
      "loss": 0.002378770150244236,
      "step": 110500
    },
    {
      "epoch": 31.365313653136532,
      "eval_accuracy": 0.9734215044191518,
      "eval_loss": 0.09068916738033295,
      "eval_runtime": 31.8541,
      "eval_samples_per_second": 493.719,
      "eval_steps_per_second": 7.723,
      "step": 110500
    },
    {
      "epoch": 31.36815214305989,
      "grad_norm": 0.19217880070209503,
      "learning_rate": 6.864547260857225e-05,
      "loss": 0.005883803218603134,
      "step": 110510
    },
    {
      "epoch": 31.370990632983254,
      "grad_norm": 0.2941473126411438,
      "learning_rate": 6.864263411864889e-05,
      "loss": 0.010930477827787399,
      "step": 110520
    },
    {
      "epoch": 31.373829122906614,
      "grad_norm": 0.07004305720329285,
      "learning_rate": 6.863979562872552e-05,
      "loss": 0.005409552156925202,
      "step": 110530
    },
    {
      "epoch": 31.376667612829973,
      "grad_norm": 2.602637529373169,
      "learning_rate": 6.863695713880216e-05,
      "loss": 0.0034328415989875794,
      "step": 110540
    },
    {
      "epoch": 31.379506102753336,
      "grad_norm": 4.1790571212768555,
      "learning_rate": 6.86341186488788e-05,
      "loss": 0.006360481679439545,
      "step": 110550
    },
    {
      "epoch": 31.382344592676695,
      "grad_norm": 12.8060884475708,
      "learning_rate": 6.863128015895543e-05,
      "loss": 0.012725891172885894,
      "step": 110560
    },
    {
      "epoch": 31.385183082600058,
      "grad_norm": 0.6453924775123596,
      "learning_rate": 6.862844166903207e-05,
      "loss": 0.0034523710608482363,
      "step": 110570
    },
    {
      "epoch": 31.388021572523417,
      "grad_norm": 11.52908706665039,
      "learning_rate": 6.862560317910873e-05,
      "loss": 0.01105356365442276,
      "step": 110580
    },
    {
      "epoch": 31.390860062446777,
      "grad_norm": 3.1964778900146484,
      "learning_rate": 6.862276468918535e-05,
      "loss": 0.014997807145118714,
      "step": 110590
    },
    {
      "epoch": 31.39369855237014,
      "grad_norm": 12.168212890625,
      "learning_rate": 6.8619926199262e-05,
      "loss": 0.004727430269122123,
      "step": 110600
    },
    {
      "epoch": 31.3965370422935,
      "grad_norm": 4.784825801849365,
      "learning_rate": 6.861708770933864e-05,
      "loss": 0.011191074550151826,
      "step": 110610
    },
    {
      "epoch": 31.39937553221686,
      "grad_norm": 0.1228862777352333,
      "learning_rate": 6.861424921941528e-05,
      "loss": 0.003466445952653885,
      "step": 110620
    },
    {
      "epoch": 31.40221402214022,
      "grad_norm": 0.1359306424856186,
      "learning_rate": 6.86114107294919e-05,
      "loss": 0.00432380847632885,
      "step": 110630
    },
    {
      "epoch": 31.405052512063584,
      "grad_norm": 0.5755993723869324,
      "learning_rate": 6.860857223956856e-05,
      "loss": 0.0037147924304008484,
      "step": 110640
    },
    {
      "epoch": 31.407891001986943,
      "grad_norm": 0.23398731648921967,
      "learning_rate": 6.86057337496452e-05,
      "loss": 0.006961977481842041,
      "step": 110650
    },
    {
      "epoch": 31.410729491910303,
      "grad_norm": 3.57582426071167,
      "learning_rate": 6.860289525972183e-05,
      "loss": 0.0070944100618362425,
      "step": 110660
    },
    {
      "epoch": 31.413567981833665,
      "grad_norm": 3.537637948989868,
      "learning_rate": 6.860005676979847e-05,
      "loss": 0.0019071191549301148,
      "step": 110670
    },
    {
      "epoch": 31.416406471757025,
      "grad_norm": 0.09797897934913635,
      "learning_rate": 6.859721827987511e-05,
      "loss": 0.011883698403835297,
      "step": 110680
    },
    {
      "epoch": 31.419244961680388,
      "grad_norm": 0.2095937728881836,
      "learning_rate": 6.859437978995174e-05,
      "loss": 0.021143296360969545,
      "step": 110690
    },
    {
      "epoch": 31.422083451603747,
      "grad_norm": 2.3575212955474854,
      "learning_rate": 6.859154130002838e-05,
      "loss": 0.008230999112129211,
      "step": 110700
    },
    {
      "epoch": 31.424921941527106,
      "grad_norm": 0.20458194613456726,
      "learning_rate": 6.858870281010504e-05,
      "loss": 0.00540386289358139,
      "step": 110710
    },
    {
      "epoch": 31.42776043145047,
      "grad_norm": 1.0072506666183472,
      "learning_rate": 6.858586432018166e-05,
      "loss": 0.007096894085407257,
      "step": 110720
    },
    {
      "epoch": 31.43059892137383,
      "grad_norm": 0.15148283541202545,
      "learning_rate": 6.85830258302583e-05,
      "loss": 0.0016185205429792404,
      "step": 110730
    },
    {
      "epoch": 31.43343741129719,
      "grad_norm": 1.6251716613769531,
      "learning_rate": 6.858018734033495e-05,
      "loss": 0.002851945534348488,
      "step": 110740
    },
    {
      "epoch": 31.43627590122055,
      "grad_norm": 0.309197336435318,
      "learning_rate": 6.857734885041159e-05,
      "loss": 0.003495413064956665,
      "step": 110750
    },
    {
      "epoch": 31.43911439114391,
      "grad_norm": 6.319087505340576,
      "learning_rate": 6.857451036048822e-05,
      "loss": 0.005122091993689537,
      "step": 110760
    },
    {
      "epoch": 31.441952881067273,
      "grad_norm": 0.07136861979961395,
      "learning_rate": 6.857167187056486e-05,
      "loss": 0.00905277132987976,
      "step": 110770
    },
    {
      "epoch": 31.444791370990632,
      "grad_norm": 0.3060901463031769,
      "learning_rate": 6.856883338064151e-05,
      "loss": 0.005736605823040008,
      "step": 110780
    },
    {
      "epoch": 31.447629860913995,
      "grad_norm": 1.0271434783935547,
      "learning_rate": 6.856599489071814e-05,
      "loss": 0.005364743247628212,
      "step": 110790
    },
    {
      "epoch": 31.450468350837355,
      "grad_norm": 0.17331670224666595,
      "learning_rate": 6.856315640079478e-05,
      "loss": 0.0030073340982198714,
      "step": 110800
    },
    {
      "epoch": 31.453306840760714,
      "grad_norm": 0.28742823004722595,
      "learning_rate": 6.856031791087142e-05,
      "loss": 0.008888522535562516,
      "step": 110810
    },
    {
      "epoch": 31.456145330684077,
      "grad_norm": 2.2763500213623047,
      "learning_rate": 6.855747942094805e-05,
      "loss": 0.0019772512838244437,
      "step": 110820
    },
    {
      "epoch": 31.458983820607436,
      "grad_norm": 0.2442139834165573,
      "learning_rate": 6.855464093102469e-05,
      "loss": 0.00967795103788376,
      "step": 110830
    },
    {
      "epoch": 31.4618223105308,
      "grad_norm": 0.14647503197193146,
      "learning_rate": 6.855180244110135e-05,
      "loss": 0.0020475735887885093,
      "step": 110840
    },
    {
      "epoch": 31.46466080045416,
      "grad_norm": 0.0963652953505516,
      "learning_rate": 6.854896395117798e-05,
      "loss": 0.004030054435133934,
      "step": 110850
    },
    {
      "epoch": 31.467499290377518,
      "grad_norm": 0.39934444427490234,
      "learning_rate": 6.854612546125462e-05,
      "loss": 0.01303512454032898,
      "step": 110860
    },
    {
      "epoch": 31.47033778030088,
      "grad_norm": 0.16033871471881866,
      "learning_rate": 6.854328697133126e-05,
      "loss": 0.005937670916318893,
      "step": 110870
    },
    {
      "epoch": 31.47317627022424,
      "grad_norm": 0.41917967796325684,
      "learning_rate": 6.85404484814079e-05,
      "loss": 0.007911767065525054,
      "step": 110880
    },
    {
      "epoch": 31.476014760147603,
      "grad_norm": 0.2379283457994461,
      "learning_rate": 6.853760999148453e-05,
      "loss": 0.01058848351240158,
      "step": 110890
    },
    {
      "epoch": 31.478853250070962,
      "grad_norm": 0.4411291778087616,
      "learning_rate": 6.853477150156117e-05,
      "loss": 0.015578216314315796,
      "step": 110900
    },
    {
      "epoch": 31.48169173999432,
      "grad_norm": 0.8294801712036133,
      "learning_rate": 6.853193301163782e-05,
      "loss": 0.003521782532334328,
      "step": 110910
    },
    {
      "epoch": 31.484530229917684,
      "grad_norm": 2.874532461166382,
      "learning_rate": 6.852909452171445e-05,
      "loss": 0.002393849566578865,
      "step": 110920
    },
    {
      "epoch": 31.487368719841044,
      "grad_norm": 1.7254900932312012,
      "learning_rate": 6.852625603179109e-05,
      "loss": 0.008758755028247833,
      "step": 110930
    },
    {
      "epoch": 31.490207209764407,
      "grad_norm": 0.591719925403595,
      "learning_rate": 6.852341754186773e-05,
      "loss": 0.004954873025417328,
      "step": 110940
    },
    {
      "epoch": 31.493045699687766,
      "grad_norm": 0.4086301028728485,
      "learning_rate": 6.852057905194436e-05,
      "loss": 0.01292019784450531,
      "step": 110950
    },
    {
      "epoch": 31.495884189611125,
      "grad_norm": 0.6835862398147583,
      "learning_rate": 6.8517740562021e-05,
      "loss": 0.0055124431848526,
      "step": 110960
    },
    {
      "epoch": 31.498722679534488,
      "grad_norm": 12.664715766906738,
      "learning_rate": 6.851490207209764e-05,
      "loss": 0.012074268609285354,
      "step": 110970
    },
    {
      "epoch": 31.501561169457847,
      "grad_norm": 0.9274953007698059,
      "learning_rate": 6.851206358217429e-05,
      "loss": 0.003194229304790497,
      "step": 110980
    },
    {
      "epoch": 31.50439965938121,
      "grad_norm": 1.6582834720611572,
      "learning_rate": 6.850922509225093e-05,
      "loss": 0.005818652361631394,
      "step": 110990
    },
    {
      "epoch": 31.50723814930457,
      "grad_norm": 1.101373314857483,
      "learning_rate": 6.850638660232757e-05,
      "loss": 0.006683534383773804,
      "step": 111000
    },
    {
      "epoch": 31.50723814930457,
      "eval_accuracy": 0.9708781077128505,
      "eval_loss": 0.09602735191583633,
      "eval_runtime": 32.6859,
      "eval_samples_per_second": 481.156,
      "eval_steps_per_second": 7.526,
      "step": 111000
    },
    {
      "epoch": 31.510076639227933,
      "grad_norm": 11.351529121398926,
      "learning_rate": 6.850354811240421e-05,
      "loss": 0.008687594532966613,
      "step": 111010
    },
    {
      "epoch": 31.512915129151292,
      "grad_norm": 4.659715175628662,
      "learning_rate": 6.850070962248084e-05,
      "loss": 0.009848679602146148,
      "step": 111020
    },
    {
      "epoch": 31.51575361907465,
      "grad_norm": 3.9261043071746826,
      "learning_rate": 6.849787113255748e-05,
      "loss": 0.01351223886013031,
      "step": 111030
    },
    {
      "epoch": 31.518592108998014,
      "grad_norm": 0.31158801913261414,
      "learning_rate": 6.849503264263412e-05,
      "loss": 0.0028676217421889304,
      "step": 111040
    },
    {
      "epoch": 31.521430598921373,
      "grad_norm": 0.8351271152496338,
      "learning_rate": 6.849219415271076e-05,
      "loss": 0.005151634663343429,
      "step": 111050
    },
    {
      "epoch": 31.524269088844733,
      "grad_norm": 0.262214720249176,
      "learning_rate": 6.84893556627874e-05,
      "loss": 0.004819277673959732,
      "step": 111060
    },
    {
      "epoch": 31.527107578768096,
      "grad_norm": 0.8365206122398376,
      "learning_rate": 6.848651717286404e-05,
      "loss": 0.0012492701411247254,
      "step": 111070
    },
    {
      "epoch": 31.529946068691455,
      "grad_norm": 1.2457735538482666,
      "learning_rate": 6.848367868294067e-05,
      "loss": 0.016850826144218446,
      "step": 111080
    },
    {
      "epoch": 31.532784558614818,
      "grad_norm": 3.509990930557251,
      "learning_rate": 6.848084019301731e-05,
      "loss": 0.02714107632637024,
      "step": 111090
    },
    {
      "epoch": 31.535623048538177,
      "grad_norm": 1.167598843574524,
      "learning_rate": 6.847800170309396e-05,
      "loss": 0.008013924956321717,
      "step": 111100
    },
    {
      "epoch": 31.53846153846154,
      "grad_norm": 0.9673077464103699,
      "learning_rate": 6.84751632131706e-05,
      "loss": 0.019352704286575317,
      "step": 111110
    },
    {
      "epoch": 31.5413000283849,
      "grad_norm": 23.924570083618164,
      "learning_rate": 6.847232472324724e-05,
      "loss": 0.04818193912506104,
      "step": 111120
    },
    {
      "epoch": 31.54413851830826,
      "grad_norm": 0.1178336888551712,
      "learning_rate": 6.846948623332388e-05,
      "loss": 0.032082986831665036,
      "step": 111130
    },
    {
      "epoch": 31.54697700823162,
      "grad_norm": 6.005351543426514,
      "learning_rate": 6.846664774340051e-05,
      "loss": 0.01379796713590622,
      "step": 111140
    },
    {
      "epoch": 31.54981549815498,
      "grad_norm": 1.2460291385650635,
      "learning_rate": 6.846380925347715e-05,
      "loss": 0.0038434986025094988,
      "step": 111150
    },
    {
      "epoch": 31.552653988078344,
      "grad_norm": 0.07101533561944962,
      "learning_rate": 6.846097076355379e-05,
      "loss": 0.005717720836400986,
      "step": 111160
    },
    {
      "epoch": 31.555492478001703,
      "grad_norm": 8.381739616394043,
      "learning_rate": 6.845813227363043e-05,
      "loss": 0.005305105075240135,
      "step": 111170
    },
    {
      "epoch": 31.558330967925063,
      "grad_norm": 0.8707379102706909,
      "learning_rate": 6.845529378370707e-05,
      "loss": 0.002006872557103634,
      "step": 111180
    },
    {
      "epoch": 31.561169457848425,
      "grad_norm": 1.032707929611206,
      "learning_rate": 6.845245529378371e-05,
      "loss": 0.003454416245222092,
      "step": 111190
    },
    {
      "epoch": 31.564007947771785,
      "grad_norm": 0.5950360894203186,
      "learning_rate": 6.844961680386036e-05,
      "loss": 0.005815703794360161,
      "step": 111200
    },
    {
      "epoch": 31.566846437695148,
      "grad_norm": 1.1969664096832275,
      "learning_rate": 6.844677831393698e-05,
      "loss": 0.011549285054206848,
      "step": 111210
    },
    {
      "epoch": 31.569684927618507,
      "grad_norm": 0.086768239736557,
      "learning_rate": 6.844393982401362e-05,
      "loss": 0.0018267950043082237,
      "step": 111220
    },
    {
      "epoch": 31.572523417541866,
      "grad_norm": 0.7640368938446045,
      "learning_rate": 6.844110133409027e-05,
      "loss": 0.0064444899559021,
      "step": 111230
    },
    {
      "epoch": 31.57536190746523,
      "grad_norm": 1.149147629737854,
      "learning_rate": 6.843826284416691e-05,
      "loss": 0.003337462246417999,
      "step": 111240
    },
    {
      "epoch": 31.57820039738859,
      "grad_norm": 0.5072917342185974,
      "learning_rate": 6.843542435424355e-05,
      "loss": 0.009381873905658722,
      "step": 111250
    },
    {
      "epoch": 31.58103888731195,
      "grad_norm": 0.4425654113292694,
      "learning_rate": 6.843258586432019e-05,
      "loss": 0.004287391155958176,
      "step": 111260
    },
    {
      "epoch": 31.58387737723531,
      "grad_norm": 1.9969398975372314,
      "learning_rate": 6.842974737439682e-05,
      "loss": 0.00957765281200409,
      "step": 111270
    },
    {
      "epoch": 31.58671586715867,
      "grad_norm": 3.201249599456787,
      "learning_rate": 6.842690888447346e-05,
      "loss": 0.012544181942939759,
      "step": 111280
    },
    {
      "epoch": 31.589554357082033,
      "grad_norm": 8.698814392089844,
      "learning_rate": 6.84240703945501e-05,
      "loss": 0.004697917774319649,
      "step": 111290
    },
    {
      "epoch": 31.592392847005392,
      "grad_norm": 0.2324749082326889,
      "learning_rate": 6.842123190462674e-05,
      "loss": 0.0021651582792401314,
      "step": 111300
    },
    {
      "epoch": 31.595231336928755,
      "grad_norm": 0.34886401891708374,
      "learning_rate": 6.841839341470338e-05,
      "loss": 0.006894345581531525,
      "step": 111310
    },
    {
      "epoch": 31.598069826852115,
      "grad_norm": 0.1326005756855011,
      "learning_rate": 6.841555492478002e-05,
      "loss": 0.01654348522424698,
      "step": 111320
    },
    {
      "epoch": 31.600908316775474,
      "grad_norm": 0.32292553782463074,
      "learning_rate": 6.841271643485667e-05,
      "loss": 0.004765563830733299,
      "step": 111330
    },
    {
      "epoch": 31.603746806698837,
      "grad_norm": 9.727119445800781,
      "learning_rate": 6.84098779449333e-05,
      "loss": 0.023321005702018737,
      "step": 111340
    },
    {
      "epoch": 31.606585296622196,
      "grad_norm": 0.8241628408432007,
      "learning_rate": 6.840703945500994e-05,
      "loss": 0.005406643822789192,
      "step": 111350
    },
    {
      "epoch": 31.60942378654556,
      "grad_norm": 0.11491557955741882,
      "learning_rate": 6.840420096508658e-05,
      "loss": 0.014735694229602813,
      "step": 111360
    },
    {
      "epoch": 31.61226227646892,
      "grad_norm": 0.5627094507217407,
      "learning_rate": 6.84013624751632e-05,
      "loss": 0.006907147169113159,
      "step": 111370
    },
    {
      "epoch": 31.615100766392278,
      "grad_norm": 0.4521954357624054,
      "learning_rate": 6.839852398523986e-05,
      "loss": 0.014480642974376678,
      "step": 111380
    },
    {
      "epoch": 31.61793925631564,
      "grad_norm": 2.4065756797790527,
      "learning_rate": 6.83956854953165e-05,
      "loss": 0.0038803480565547943,
      "step": 111390
    },
    {
      "epoch": 31.620777746239,
      "grad_norm": 0.04546709358692169,
      "learning_rate": 6.839284700539313e-05,
      "loss": 0.0016823584213852883,
      "step": 111400
    },
    {
      "epoch": 31.623616236162363,
      "grad_norm": 5.215712547302246,
      "learning_rate": 6.839000851546977e-05,
      "loss": 0.004511076956987381,
      "step": 111410
    },
    {
      "epoch": 31.626454726085722,
      "grad_norm": 1.4145536422729492,
      "learning_rate": 6.838717002554641e-05,
      "loss": 0.014722645282745361,
      "step": 111420
    },
    {
      "epoch": 31.62929321600908,
      "grad_norm": 0.3794224262237549,
      "learning_rate": 6.838433153562305e-05,
      "loss": 0.025443363189697265,
      "step": 111430
    },
    {
      "epoch": 31.632131705932444,
      "grad_norm": 4.338743686676025,
      "learning_rate": 6.83814930456997e-05,
      "loss": 0.006875064969062805,
      "step": 111440
    },
    {
      "epoch": 31.634970195855804,
      "grad_norm": 1.1352838277816772,
      "learning_rate": 6.837865455577634e-05,
      "loss": 0.00943584144115448,
      "step": 111450
    },
    {
      "epoch": 31.637808685779166,
      "grad_norm": 4.876721382141113,
      "learning_rate": 6.837581606585298e-05,
      "loss": 0.001181977242231369,
      "step": 111460
    },
    {
      "epoch": 31.640647175702526,
      "grad_norm": 0.3526777923107147,
      "learning_rate": 6.83729775759296e-05,
      "loss": 0.0032115235924720766,
      "step": 111470
    },
    {
      "epoch": 31.64348566562589,
      "grad_norm": 14.518491744995117,
      "learning_rate": 6.837013908600625e-05,
      "loss": 0.008310648798942565,
      "step": 111480
    },
    {
      "epoch": 31.646324155549248,
      "grad_norm": 0.31408146023750305,
      "learning_rate": 6.836730059608289e-05,
      "loss": 0.014019936323165894,
      "step": 111490
    },
    {
      "epoch": 31.649162645472607,
      "grad_norm": 2.747941732406616,
      "learning_rate": 6.836446210615952e-05,
      "loss": 0.004994688928127289,
      "step": 111500
    },
    {
      "epoch": 31.649162645472607,
      "eval_accuracy": 0.972849240160234,
      "eval_loss": 0.09299984574317932,
      "eval_runtime": 31.7372,
      "eval_samples_per_second": 495.539,
      "eval_steps_per_second": 7.751,
      "step": 111500
    },
    {
      "epoch": 31.65200113539597,
      "grad_norm": 0.1669388860464096,
      "learning_rate": 6.836162361623617e-05,
      "loss": 0.0029633624479174616,
      "step": 111510
    },
    {
      "epoch": 31.65483962531933,
      "grad_norm": 2.906898260116577,
      "learning_rate": 6.835878512631281e-05,
      "loss": 0.00947681963443756,
      "step": 111520
    },
    {
      "epoch": 31.657678115242692,
      "grad_norm": 0.23977884650230408,
      "learning_rate": 6.835594663638944e-05,
      "loss": 0.002341122180223465,
      "step": 111530
    },
    {
      "epoch": 31.660516605166052,
      "grad_norm": 0.05543351173400879,
      "learning_rate": 6.835310814646608e-05,
      "loss": 0.005055475980043411,
      "step": 111540
    },
    {
      "epoch": 31.66335509508941,
      "grad_norm": 6.8705878257751465,
      "learning_rate": 6.835026965654272e-05,
      "loss": 0.006296662986278534,
      "step": 111550
    },
    {
      "epoch": 31.666193585012774,
      "grad_norm": 0.07508984953165054,
      "learning_rate": 6.834743116661936e-05,
      "loss": 0.009136031568050384,
      "step": 111560
    },
    {
      "epoch": 31.669032074936133,
      "grad_norm": 0.4821889102458954,
      "learning_rate": 6.8344592676696e-05,
      "loss": 0.00289219431579113,
      "step": 111570
    },
    {
      "epoch": 31.671870564859496,
      "grad_norm": 0.5724067091941833,
      "learning_rate": 6.834175418677265e-05,
      "loss": 0.010458598285913468,
      "step": 111580
    },
    {
      "epoch": 31.674709054782856,
      "grad_norm": 1.17878258228302,
      "learning_rate": 6.833891569684929e-05,
      "loss": 0.007089054584503174,
      "step": 111590
    },
    {
      "epoch": 31.677547544706215,
      "grad_norm": 6.668631553649902,
      "learning_rate": 6.833607720692592e-05,
      "loss": 0.004987912625074387,
      "step": 111600
    },
    {
      "epoch": 31.680386034629578,
      "grad_norm": 0.12468308955430984,
      "learning_rate": 6.833323871700256e-05,
      "loss": 0.0031174197793006897,
      "step": 111610
    },
    {
      "epoch": 31.683224524552937,
      "grad_norm": 5.912275314331055,
      "learning_rate": 6.83304002270792e-05,
      "loss": 0.007123376429080963,
      "step": 111620
    },
    {
      "epoch": 31.6860630144763,
      "grad_norm": 0.08171559125185013,
      "learning_rate": 6.832756173715583e-05,
      "loss": 0.014561636745929718,
      "step": 111630
    },
    {
      "epoch": 31.68890150439966,
      "grad_norm": 0.5651805400848389,
      "learning_rate": 6.832472324723248e-05,
      "loss": 0.009940367192029953,
      "step": 111640
    },
    {
      "epoch": 31.69173999432302,
      "grad_norm": 0.20194415748119354,
      "learning_rate": 6.832188475730912e-05,
      "loss": 0.0039100654423236845,
      "step": 111650
    },
    {
      "epoch": 31.69457848424638,
      "grad_norm": 0.21964016556739807,
      "learning_rate": 6.831904626738575e-05,
      "loss": 0.015068444609642028,
      "step": 111660
    },
    {
      "epoch": 31.69741697416974,
      "grad_norm": 0.029672732576727867,
      "learning_rate": 6.831620777746239e-05,
      "loss": 0.00910898819565773,
      "step": 111670
    },
    {
      "epoch": 31.700255464093104,
      "grad_norm": 0.8006603717803955,
      "learning_rate": 6.831336928753903e-05,
      "loss": 0.004151589050889015,
      "step": 111680
    },
    {
      "epoch": 31.703093954016463,
      "grad_norm": 0.4289233982563019,
      "learning_rate": 6.831053079761567e-05,
      "loss": 0.007662437856197357,
      "step": 111690
    },
    {
      "epoch": 31.705932443939822,
      "grad_norm": 0.19920805096626282,
      "learning_rate": 6.83076923076923e-05,
      "loss": 0.011945778876543045,
      "step": 111700
    },
    {
      "epoch": 31.708770933863185,
      "grad_norm": 0.16350744664669037,
      "learning_rate": 6.830485381776896e-05,
      "loss": 0.0032811239361763,
      "step": 111710
    },
    {
      "epoch": 31.711609423786545,
      "grad_norm": 0.7621529698371887,
      "learning_rate": 6.83020153278456e-05,
      "loss": 0.0025921674445271493,
      "step": 111720
    },
    {
      "epoch": 31.714447913709908,
      "grad_norm": 11.359150886535645,
      "learning_rate": 6.829917683792223e-05,
      "loss": 0.006132229417562485,
      "step": 111730
    },
    {
      "epoch": 31.717286403633267,
      "grad_norm": 11.966462135314941,
      "learning_rate": 6.829633834799887e-05,
      "loss": 0.011749979853630067,
      "step": 111740
    },
    {
      "epoch": 31.720124893556626,
      "grad_norm": 0.20635463297367096,
      "learning_rate": 6.829349985807551e-05,
      "loss": 0.011074859648942947,
      "step": 111750
    },
    {
      "epoch": 31.72296338347999,
      "grad_norm": 2.355894088745117,
      "learning_rate": 6.829066136815214e-05,
      "loss": 0.003608880192041397,
      "step": 111760
    },
    {
      "epoch": 31.72580187340335,
      "grad_norm": 0.19789737462997437,
      "learning_rate": 6.828782287822879e-05,
      "loss": 0.006971706449985504,
      "step": 111770
    },
    {
      "epoch": 31.72864036332671,
      "grad_norm": 22.51043128967285,
      "learning_rate": 6.828498438830543e-05,
      "loss": 0.023025889694690705,
      "step": 111780
    },
    {
      "epoch": 31.73147885325007,
      "grad_norm": 0.44341930747032166,
      "learning_rate": 6.828214589838206e-05,
      "loss": 0.010177663713693618,
      "step": 111790
    },
    {
      "epoch": 31.73431734317343,
      "grad_norm": 0.13608318567276,
      "learning_rate": 6.82793074084587e-05,
      "loss": 0.008347529172897338,
      "step": 111800
    },
    {
      "epoch": 31.737155833096793,
      "grad_norm": 4.800083160400391,
      "learning_rate": 6.827646891853534e-05,
      "loss": 0.010078371316194535,
      "step": 111810
    },
    {
      "epoch": 31.739994323020152,
      "grad_norm": 0.4727610647678375,
      "learning_rate": 6.827363042861199e-05,
      "loss": 0.009345553815364838,
      "step": 111820
    },
    {
      "epoch": 31.742832812943515,
      "grad_norm": 1.2807178497314453,
      "learning_rate": 6.827079193868861e-05,
      "loss": 0.002239334024488926,
      "step": 111830
    },
    {
      "epoch": 31.745671302866874,
      "grad_norm": 0.31881776452064514,
      "learning_rate": 6.826795344876527e-05,
      "loss": 0.005826468765735626,
      "step": 111840
    },
    {
      "epoch": 31.748509792790237,
      "grad_norm": 0.3899812400341034,
      "learning_rate": 6.826511495884191e-05,
      "loss": 0.00906587541103363,
      "step": 111850
    },
    {
      "epoch": 31.751348282713597,
      "grad_norm": 1.1504793167114258,
      "learning_rate": 6.826227646891854e-05,
      "loss": 0.004249905422329903,
      "step": 111860
    },
    {
      "epoch": 31.754186772636956,
      "grad_norm": 6.78041410446167,
      "learning_rate": 6.825943797899518e-05,
      "loss": 0.0038106441497802734,
      "step": 111870
    },
    {
      "epoch": 31.75702526256032,
      "grad_norm": 1.5143126249313354,
      "learning_rate": 6.825659948907182e-05,
      "loss": 0.010604381561279297,
      "step": 111880
    },
    {
      "epoch": 31.759863752483678,
      "grad_norm": 7.184882640838623,
      "learning_rate": 6.825376099914845e-05,
      "loss": 0.0044593319296836855,
      "step": 111890
    },
    {
      "epoch": 31.76270224240704,
      "grad_norm": 4.225615978240967,
      "learning_rate": 6.825092250922509e-05,
      "loss": 0.0053154781460762026,
      "step": 111900
    },
    {
      "epoch": 31.7655407323304,
      "grad_norm": 6.64150857925415,
      "learning_rate": 6.824808401930174e-05,
      "loss": 0.004802253097295761,
      "step": 111910
    },
    {
      "epoch": 31.76837922225376,
      "grad_norm": 9.94938850402832,
      "learning_rate": 6.824524552937837e-05,
      "loss": 0.012674580514430999,
      "step": 111920
    },
    {
      "epoch": 31.771217712177123,
      "grad_norm": 14.90323543548584,
      "learning_rate": 6.824240703945501e-05,
      "loss": 0.02048947513103485,
      "step": 111930
    },
    {
      "epoch": 31.774056202100482,
      "grad_norm": 0.5821764469146729,
      "learning_rate": 6.823956854953165e-05,
      "loss": 0.013197378814220428,
      "step": 111940
    },
    {
      "epoch": 31.776894692023845,
      "grad_norm": 0.5471833944320679,
      "learning_rate": 6.82367300596083e-05,
      "loss": 0.018742209672927855,
      "step": 111950
    },
    {
      "epoch": 31.779733181947204,
      "grad_norm": 18.066133499145508,
      "learning_rate": 6.823389156968492e-05,
      "loss": 0.016490170359611513,
      "step": 111960
    },
    {
      "epoch": 31.782571671870564,
      "grad_norm": 0.054250482469797134,
      "learning_rate": 6.823105307976158e-05,
      "loss": 0.005137863382697105,
      "step": 111970
    },
    {
      "epoch": 31.785410161793926,
      "grad_norm": 0.026516495272517204,
      "learning_rate": 6.82282145898382e-05,
      "loss": 0.01019509881734848,
      "step": 111980
    },
    {
      "epoch": 31.788248651717286,
      "grad_norm": 0.22025428712368011,
      "learning_rate": 6.822537609991485e-05,
      "loss": 0.0022360729053616524,
      "step": 111990
    },
    {
      "epoch": 31.79108714164065,
      "grad_norm": 0.33896440267562866,
      "learning_rate": 6.822253760999149e-05,
      "loss": 0.008224617689847946,
      "step": 112000
    },
    {
      "epoch": 31.79108714164065,
      "eval_accuracy": 0.9695428244420423,
      "eval_loss": 0.10078894346952438,
      "eval_runtime": 31.5115,
      "eval_samples_per_second": 499.088,
      "eval_steps_per_second": 7.807,
      "step": 112000
    },
    {
      "epoch": 31.793925631564008,
      "grad_norm": 0.2273392528295517,
      "learning_rate": 6.821969912006813e-05,
      "loss": 0.005847939848899841,
      "step": 112010
    },
    {
      "epoch": 31.796764121487367,
      "grad_norm": 0.12986287474632263,
      "learning_rate": 6.821686063014476e-05,
      "loss": 0.008841508626937866,
      "step": 112020
    },
    {
      "epoch": 31.79960261141073,
      "grad_norm": 11.993029594421387,
      "learning_rate": 6.82140221402214e-05,
      "loss": 0.007772567868232727,
      "step": 112030
    },
    {
      "epoch": 31.80244110133409,
      "grad_norm": 0.20143119990825653,
      "learning_rate": 6.821118365029805e-05,
      "loss": 0.003965846076607704,
      "step": 112040
    },
    {
      "epoch": 31.805279591257452,
      "grad_norm": 0.16670043766498566,
      "learning_rate": 6.820834516037468e-05,
      "loss": 0.0041697613894939424,
      "step": 112050
    },
    {
      "epoch": 31.80811808118081,
      "grad_norm": 0.1777687817811966,
      "learning_rate": 6.820550667045132e-05,
      "loss": 0.0049547705799341205,
      "step": 112060
    },
    {
      "epoch": 31.81095657110417,
      "grad_norm": 0.5711122751235962,
      "learning_rate": 6.820266818052797e-05,
      "loss": 0.00910588949918747,
      "step": 112070
    },
    {
      "epoch": 31.813795061027534,
      "grad_norm": 4.641970634460449,
      "learning_rate": 6.819982969060459e-05,
      "loss": 0.0032428544014692306,
      "step": 112080
    },
    {
      "epoch": 31.816633550950893,
      "grad_norm": 4.070696830749512,
      "learning_rate": 6.819699120068123e-05,
      "loss": 0.0025896903127431868,
      "step": 112090
    },
    {
      "epoch": 31.819472040874256,
      "grad_norm": 14.867886543273926,
      "learning_rate": 6.819415271075788e-05,
      "loss": 0.017729231715202333,
      "step": 112100
    },
    {
      "epoch": 31.822310530797616,
      "grad_norm": 0.16451802849769592,
      "learning_rate": 6.819131422083452e-05,
      "loss": 0.020681332051753997,
      "step": 112110
    },
    {
      "epoch": 31.825149020720975,
      "grad_norm": 0.20276722311973572,
      "learning_rate": 6.818847573091116e-05,
      "loss": 0.0071222282946109775,
      "step": 112120
    },
    {
      "epoch": 31.827987510644338,
      "grad_norm": 0.12054027616977692,
      "learning_rate": 6.81856372409878e-05,
      "loss": 0.0025136858224868775,
      "step": 112130
    },
    {
      "epoch": 31.830826000567697,
      "grad_norm": 2.381458282470703,
      "learning_rate": 6.818279875106444e-05,
      "loss": 0.016549375653266907,
      "step": 112140
    },
    {
      "epoch": 31.83366449049106,
      "grad_norm": 0.2634775638580322,
      "learning_rate": 6.817996026114107e-05,
      "loss": 0.003928965330123902,
      "step": 112150
    },
    {
      "epoch": 31.83650298041442,
      "grad_norm": 0.09217740595340729,
      "learning_rate": 6.817712177121771e-05,
      "loss": 0.0027442745864391326,
      "step": 112160
    },
    {
      "epoch": 31.83934147033778,
      "grad_norm": 7.691119194030762,
      "learning_rate": 6.817428328129437e-05,
      "loss": 0.01693858504295349,
      "step": 112170
    },
    {
      "epoch": 31.84217996026114,
      "grad_norm": 0.05064951628446579,
      "learning_rate": 6.8171444791371e-05,
      "loss": 0.0014888802543282509,
      "step": 112180
    },
    {
      "epoch": 31.8450184501845,
      "grad_norm": 0.3167444169521332,
      "learning_rate": 6.816860630144763e-05,
      "loss": 0.010639182478189468,
      "step": 112190
    },
    {
      "epoch": 31.847856940107864,
      "grad_norm": 3.1098062992095947,
      "learning_rate": 6.816576781152428e-05,
      "loss": 0.008044886589050292,
      "step": 112200
    },
    {
      "epoch": 31.850695430031223,
      "grad_norm": 3.930600881576538,
      "learning_rate": 6.81629293216009e-05,
      "loss": 0.003921622782945633,
      "step": 112210
    },
    {
      "epoch": 31.853533919954586,
      "grad_norm": 0.3429722785949707,
      "learning_rate": 6.816009083167755e-05,
      "loss": 0.010262078046798706,
      "step": 112220
    },
    {
      "epoch": 31.856372409877945,
      "grad_norm": 1.0070666074752808,
      "learning_rate": 6.815725234175419e-05,
      "loss": 0.007831378281116486,
      "step": 112230
    },
    {
      "epoch": 31.859210899801305,
      "grad_norm": 0.9816868305206299,
      "learning_rate": 6.815441385183083e-05,
      "loss": 0.005563776940107346,
      "step": 112240
    },
    {
      "epoch": 31.862049389724667,
      "grad_norm": 0.7258110642433167,
      "learning_rate": 6.815157536190747e-05,
      "loss": 0.005391286313533783,
      "step": 112250
    },
    {
      "epoch": 31.864887879648027,
      "grad_norm": 0.11904098838567734,
      "learning_rate": 6.814873687198411e-05,
      "loss": 0.006604784727096557,
      "step": 112260
    },
    {
      "epoch": 31.86772636957139,
      "grad_norm": 1.7833142280578613,
      "learning_rate": 6.814589838206075e-05,
      "loss": 0.009094425290822983,
      "step": 112270
    },
    {
      "epoch": 31.87056485949475,
      "grad_norm": 0.11489982157945633,
      "learning_rate": 6.814305989213738e-05,
      "loss": 0.005363580584526062,
      "step": 112280
    },
    {
      "epoch": 31.87340334941811,
      "grad_norm": 3.3595969676971436,
      "learning_rate": 6.814022140221402e-05,
      "loss": 0.00427178293466568,
      "step": 112290
    },
    {
      "epoch": 31.87624183934147,
      "grad_norm": 7.13802433013916,
      "learning_rate": 6.813738291229066e-05,
      "loss": 0.008593113720417022,
      "step": 112300
    },
    {
      "epoch": 31.87908032926483,
      "grad_norm": 1.7315257787704468,
      "learning_rate": 6.81345444223673e-05,
      "loss": 0.004069085046648979,
      "step": 112310
    },
    {
      "epoch": 31.881918819188193,
      "grad_norm": 0.6548401117324829,
      "learning_rate": 6.813170593244395e-05,
      "loss": 0.012579810619354249,
      "step": 112320
    },
    {
      "epoch": 31.884757309111553,
      "grad_norm": 0.7378200888633728,
      "learning_rate": 6.812886744252059e-05,
      "loss": 0.014098866283893586,
      "step": 112330
    },
    {
      "epoch": 31.887595799034912,
      "grad_norm": 1.0156601667404175,
      "learning_rate": 6.812631280158955e-05,
      "loss": 0.04399470686912536,
      "step": 112340
    },
    {
      "epoch": 31.890434288958275,
      "grad_norm": 0.3062853217124939,
      "learning_rate": 6.812347431166621e-05,
      "loss": 0.01922585517168045,
      "step": 112350
    },
    {
      "epoch": 31.893272778881634,
      "grad_norm": 1.0315234661102295,
      "learning_rate": 6.812063582174284e-05,
      "loss": 0.010519689321517945,
      "step": 112360
    },
    {
      "epoch": 31.896111268804997,
      "grad_norm": 11.50934886932373,
      "learning_rate": 6.811779733181948e-05,
      "loss": 0.009597477316856385,
      "step": 112370
    },
    {
      "epoch": 31.898949758728357,
      "grad_norm": 0.14066126942634583,
      "learning_rate": 6.811495884189612e-05,
      "loss": 0.0028650419786572455,
      "step": 112380
    },
    {
      "epoch": 31.901788248651716,
      "grad_norm": 11.789419174194336,
      "learning_rate": 6.811212035197275e-05,
      "loss": 0.004871566221117974,
      "step": 112390
    },
    {
      "epoch": 31.90462673857508,
      "grad_norm": 3.155224561691284,
      "learning_rate": 6.810928186204939e-05,
      "loss": 0.006400573253631592,
      "step": 112400
    },
    {
      "epoch": 31.907465228498438,
      "grad_norm": 3.1055116653442383,
      "learning_rate": 6.810644337212603e-05,
      "loss": 0.005016482621431351,
      "step": 112410
    },
    {
      "epoch": 31.9103037184218,
      "grad_norm": 0.9981109499931335,
      "learning_rate": 6.810360488220267e-05,
      "loss": 0.011814363300800323,
      "step": 112420
    },
    {
      "epoch": 31.91314220834516,
      "grad_norm": 3.2621138095855713,
      "learning_rate": 6.810076639227931e-05,
      "loss": 0.003862859308719635,
      "step": 112430
    },
    {
      "epoch": 31.91598069826852,
      "grad_norm": 0.027387892827391624,
      "learning_rate": 6.809792790235595e-05,
      "loss": 0.005342646315693855,
      "step": 112440
    },
    {
      "epoch": 31.918819188191883,
      "grad_norm": 1.1023670434951782,
      "learning_rate": 6.80950894124326e-05,
      "loss": 0.002160440571606159,
      "step": 112450
    },
    {
      "epoch": 31.921657678115242,
      "grad_norm": 0.1597622036933899,
      "learning_rate": 6.809225092250922e-05,
      "loss": 0.0019233804196119308,
      "step": 112460
    },
    {
      "epoch": 31.924496168038605,
      "grad_norm": 7.0468034744262695,
      "learning_rate": 6.808941243258586e-05,
      "loss": 0.008442067354917527,
      "step": 112470
    },
    {
      "epoch": 31.927334657961964,
      "grad_norm": 0.3132229745388031,
      "learning_rate": 6.80865739426625e-05,
      "loss": 0.016583122313022614,
      "step": 112480
    },
    {
      "epoch": 31.930173147885323,
      "grad_norm": 0.4543894827365875,
      "learning_rate": 6.808373545273915e-05,
      "loss": 0.007372811436653137,
      "step": 112490
    },
    {
      "epoch": 31.933011637808686,
      "grad_norm": 0.06805482506752014,
      "learning_rate": 6.808089696281579e-05,
      "loss": 0.0041646726429462435,
      "step": 112500
    },
    {
      "epoch": 31.933011637808686,
      "eval_accuracy": 0.9738030139250969,
      "eval_loss": 0.08944963663816452,
      "eval_runtime": 31.5907,
      "eval_samples_per_second": 497.836,
      "eval_steps_per_second": 7.787,
      "step": 112500
    },
    {
      "epoch": 31.935850127732046,
      "grad_norm": 1.4102498292922974,
      "learning_rate": 6.807805847289243e-05,
      "loss": 0.006246492266654968,
      "step": 112510
    },
    {
      "epoch": 31.93868861765541,
      "grad_norm": 6.589541912078857,
      "learning_rate": 6.807521998296906e-05,
      "loss": 0.011974203586578368,
      "step": 112520
    },
    {
      "epoch": 31.941527107578768,
      "grad_norm": 0.6699870824813843,
      "learning_rate": 6.80723814930457e-05,
      "loss": 0.0017713833600282668,
      "step": 112530
    },
    {
      "epoch": 31.944365597502127,
      "grad_norm": 0.26982733607292175,
      "learning_rate": 6.806954300312234e-05,
      "loss": 0.004998511075973511,
      "step": 112540
    },
    {
      "epoch": 31.94720408742549,
      "grad_norm": 0.07027362287044525,
      "learning_rate": 6.806670451319898e-05,
      "loss": 0.019851690530776976,
      "step": 112550
    },
    {
      "epoch": 31.95004257734885,
      "grad_norm": 0.21667777001857758,
      "learning_rate": 6.806386602327562e-05,
      "loss": 0.01027294248342514,
      "step": 112560
    },
    {
      "epoch": 31.952881067272212,
      "grad_norm": 0.40778666734695435,
      "learning_rate": 6.806102753335226e-05,
      "loss": 0.006953151524066925,
      "step": 112570
    },
    {
      "epoch": 31.95571955719557,
      "grad_norm": 0.1079963818192482,
      "learning_rate": 6.80581890434289e-05,
      "loss": 0.013990335166454315,
      "step": 112580
    },
    {
      "epoch": 31.958558047118935,
      "grad_norm": 6.210877418518066,
      "learning_rate": 6.805535055350553e-05,
      "loss": 0.008871030807495118,
      "step": 112590
    },
    {
      "epoch": 31.961396537042294,
      "grad_norm": 0.3015681505203247,
      "learning_rate": 6.805251206358217e-05,
      "loss": 0.004505900293588638,
      "step": 112600
    },
    {
      "epoch": 31.964235026965653,
      "grad_norm": 6.5876007080078125,
      "learning_rate": 6.804967357365882e-05,
      "loss": 0.013760749995708466,
      "step": 112610
    },
    {
      "epoch": 31.967073516889016,
      "grad_norm": 22.009775161743164,
      "learning_rate": 6.804683508373546e-05,
      "loss": 0.029219388961791992,
      "step": 112620
    },
    {
      "epoch": 31.969912006812375,
      "grad_norm": 0.18849162757396698,
      "learning_rate": 6.80439965938121e-05,
      "loss": 0.0029139140620827674,
      "step": 112630
    },
    {
      "epoch": 31.972750496735735,
      "grad_norm": 0.8211190104484558,
      "learning_rate": 6.804115810388874e-05,
      "loss": 0.011607404053211211,
      "step": 112640
    },
    {
      "epoch": 31.975588986659098,
      "grad_norm": 1.1625237464904785,
      "learning_rate": 6.803831961396537e-05,
      "loss": 0.019713284075260164,
      "step": 112650
    },
    {
      "epoch": 31.978427476582457,
      "grad_norm": 8.339205741882324,
      "learning_rate": 6.803548112404201e-05,
      "loss": 0.006902317702770233,
      "step": 112660
    },
    {
      "epoch": 31.98126596650582,
      "grad_norm": 1.7514382600784302,
      "learning_rate": 6.803264263411865e-05,
      "loss": 0.002683527022600174,
      "step": 112670
    },
    {
      "epoch": 31.98410445642918,
      "grad_norm": 0.6259246468544006,
      "learning_rate": 6.802980414419529e-05,
      "loss": 0.005297352373600006,
      "step": 112680
    },
    {
      "epoch": 31.986942946352542,
      "grad_norm": 0.25685709714889526,
      "learning_rate": 6.802696565427193e-05,
      "loss": 0.0030623583123087885,
      "step": 112690
    },
    {
      "epoch": 31.9897814362759,
      "grad_norm": 0.7202678918838501,
      "learning_rate": 6.802412716434857e-05,
      "loss": 0.0024009345099329947,
      "step": 112700
    },
    {
      "epoch": 31.99261992619926,
      "grad_norm": 0.1389661431312561,
      "learning_rate": 6.802128867442522e-05,
      "loss": 0.007260927557945251,
      "step": 112710
    },
    {
      "epoch": 31.995458416122624,
      "grad_norm": 0.7495160102844238,
      "learning_rate": 6.801845018450184e-05,
      "loss": 0.003939095884561539,
      "step": 112720
    },
    {
      "epoch": 31.998296906045983,
      "grad_norm": 1.1752415895462036,
      "learning_rate": 6.801561169457849e-05,
      "loss": 0.004041685163974762,
      "step": 112730
    },
    {
      "epoch": 32.00113539596934,
      "grad_norm": 0.341753751039505,
      "learning_rate": 6.801277320465513e-05,
      "loss": 0.004040905460715294,
      "step": 112740
    },
    {
      "epoch": 32.003973885892705,
      "grad_norm": 0.38411256670951843,
      "learning_rate": 6.800993471473177e-05,
      "loss": 0.011023762077093125,
      "step": 112750
    },
    {
      "epoch": 32.00681237581607,
      "grad_norm": 0.9074758887290955,
      "learning_rate": 6.800709622480841e-05,
      "loss": 0.004296112060546875,
      "step": 112760
    },
    {
      "epoch": 32.009650865739424,
      "grad_norm": 5.970540523529053,
      "learning_rate": 6.800425773488505e-05,
      "loss": 0.0031841062009334564,
      "step": 112770
    },
    {
      "epoch": 32.01248935566279,
      "grad_norm": 0.2918536067008972,
      "learning_rate": 6.800141924496168e-05,
      "loss": 0.009583040326833724,
      "step": 112780
    },
    {
      "epoch": 32.01532784558615,
      "grad_norm": 2.315399646759033,
      "learning_rate": 6.799858075503832e-05,
      "loss": 0.0017800677567720413,
      "step": 112790
    },
    {
      "epoch": 32.018166335509505,
      "grad_norm": 0.33088240027427673,
      "learning_rate": 6.799574226511496e-05,
      "loss": 0.0007648751139640808,
      "step": 112800
    },
    {
      "epoch": 32.02100482543287,
      "grad_norm": 0.8620145916938782,
      "learning_rate": 6.79929037751916e-05,
      "loss": 0.004749047011137009,
      "step": 112810
    },
    {
      "epoch": 32.02384331535623,
      "grad_norm": 0.5967131853103638,
      "learning_rate": 6.799006528526824e-05,
      "loss": 0.011634477227926255,
      "step": 112820
    },
    {
      "epoch": 32.026681805279594,
      "grad_norm": 1.3156980276107788,
      "learning_rate": 6.798722679534489e-05,
      "loss": 0.0015296123921871186,
      "step": 112830
    },
    {
      "epoch": 32.02952029520295,
      "grad_norm": 9.440923690795898,
      "learning_rate": 6.798438830542153e-05,
      "loss": 0.010443995893001556,
      "step": 112840
    },
    {
      "epoch": 32.03235878512631,
      "grad_norm": 1.3170238733291626,
      "learning_rate": 6.798154981549815e-05,
      "loss": 0.005109821259975433,
      "step": 112850
    },
    {
      "epoch": 32.035197275049676,
      "grad_norm": 0.15561915934085846,
      "learning_rate": 6.79787113255748e-05,
      "loss": 0.017511712014675142,
      "step": 112860
    },
    {
      "epoch": 32.03803576497303,
      "grad_norm": 0.0910177007317543,
      "learning_rate": 6.797587283565144e-05,
      "loss": 0.016135452687740325,
      "step": 112870
    },
    {
      "epoch": 32.040874254896394,
      "grad_norm": 0.21093660593032837,
      "learning_rate": 6.797303434572807e-05,
      "loss": 0.0010104564949870109,
      "step": 112880
    },
    {
      "epoch": 32.04371274481976,
      "grad_norm": 0.17900234460830688,
      "learning_rate": 6.797019585580472e-05,
      "loss": 0.007267052680253983,
      "step": 112890
    },
    {
      "epoch": 32.04655123474312,
      "grad_norm": 0.11418800801038742,
      "learning_rate": 6.796735736588136e-05,
      "loss": 0.003210850805044174,
      "step": 112900
    },
    {
      "epoch": 32.049389724666476,
      "grad_norm": 0.5047315359115601,
      "learning_rate": 6.796451887595799e-05,
      "loss": 0.0027661284431815146,
      "step": 112910
    },
    {
      "epoch": 32.05222821458984,
      "grad_norm": 0.1105075255036354,
      "learning_rate": 6.796168038603463e-05,
      "loss": 0.0017645357176661492,
      "step": 112920
    },
    {
      "epoch": 32.0550667045132,
      "grad_norm": 0.5283951759338379,
      "learning_rate": 6.795884189611127e-05,
      "loss": 0.008849997818470002,
      "step": 112930
    },
    {
      "epoch": 32.05790519443656,
      "grad_norm": 0.139421284198761,
      "learning_rate": 6.795600340618791e-05,
      "loss": 0.003753601759672165,
      "step": 112940
    },
    {
      "epoch": 32.06074368435992,
      "grad_norm": 1.4927922487258911,
      "learning_rate": 6.795316491626456e-05,
      "loss": 0.005814217031002045,
      "step": 112950
    },
    {
      "epoch": 32.06358217428328,
      "grad_norm": 0.27753469347953796,
      "learning_rate": 6.79503264263412e-05,
      "loss": 0.002770618535578251,
      "step": 112960
    },
    {
      "epoch": 32.06642066420664,
      "grad_norm": 0.4666532576084137,
      "learning_rate": 6.794748793641782e-05,
      "loss": 0.00104095209389925,
      "step": 112970
    },
    {
      "epoch": 32.06925915413,
      "grad_norm": 0.01714489981532097,
      "learning_rate": 6.794464944649447e-05,
      "loss": 0.004307209700345993,
      "step": 112980
    },
    {
      "epoch": 32.072097644053365,
      "grad_norm": 0.5552462339401245,
      "learning_rate": 6.794181095657111e-05,
      "loss": 0.0042713616043329235,
      "step": 112990
    },
    {
      "epoch": 32.07493613397673,
      "grad_norm": 0.058778196573257446,
      "learning_rate": 6.793897246664775e-05,
      "loss": 0.002907343581318855,
      "step": 113000
    },
    {
      "epoch": 32.07493613397673,
      "eval_accuracy": 0.9753290519488778,
      "eval_loss": 0.08318768441677094,
      "eval_runtime": 32.0473,
      "eval_samples_per_second": 490.744,
      "eval_steps_per_second": 7.676,
      "step": 113000
    },
    {
      "epoch": 32.07777462390008,
      "grad_norm": 0.14211684465408325,
      "learning_rate": 6.793613397672438e-05,
      "loss": 0.0030759992077946663,
      "step": 113010
    },
    {
      "epoch": 32.080613113823446,
      "grad_norm": 0.8389070630073547,
      "learning_rate": 6.793329548680103e-05,
      "loss": 0.006130092591047287,
      "step": 113020
    },
    {
      "epoch": 32.08345160374681,
      "grad_norm": 4.882470607757568,
      "learning_rate": 6.793045699687767e-05,
      "loss": 0.006557777523994446,
      "step": 113030
    },
    {
      "epoch": 32.086290093670165,
      "grad_norm": 4.013255596160889,
      "learning_rate": 6.79276185069543e-05,
      "loss": 0.001519666612148285,
      "step": 113040
    },
    {
      "epoch": 32.08912858359353,
      "grad_norm": 12.725340843200684,
      "learning_rate": 6.792478001703094e-05,
      "loss": 0.0051808558404445645,
      "step": 113050
    },
    {
      "epoch": 32.09196707351689,
      "grad_norm": 0.04267686977982521,
      "learning_rate": 6.792194152710758e-05,
      "loss": 0.009982483088970184,
      "step": 113060
    },
    {
      "epoch": 32.09480556344025,
      "grad_norm": 11.574191093444824,
      "learning_rate": 6.791910303718421e-05,
      "loss": 0.015403810143470763,
      "step": 113070
    },
    {
      "epoch": 32.09764405336361,
      "grad_norm": 0.1928398460149765,
      "learning_rate": 6.791626454726085e-05,
      "loss": 0.0013574745506048202,
      "step": 113080
    },
    {
      "epoch": 32.10048254328697,
      "grad_norm": 0.1670929193496704,
      "learning_rate": 6.791342605733751e-05,
      "loss": 0.0019810723140835764,
      "step": 113090
    },
    {
      "epoch": 32.103321033210335,
      "grad_norm": 8.155051231384277,
      "learning_rate": 6.791058756741414e-05,
      "loss": 0.0050445936620235445,
      "step": 113100
    },
    {
      "epoch": 32.10615952313369,
      "grad_norm": 0.05738574638962746,
      "learning_rate": 6.790774907749078e-05,
      "loss": 0.002214076742529869,
      "step": 113110
    },
    {
      "epoch": 32.108998013057054,
      "grad_norm": 3.105058431625366,
      "learning_rate": 6.790491058756742e-05,
      "loss": 0.0025078589096665382,
      "step": 113120
    },
    {
      "epoch": 32.11183650298042,
      "grad_norm": 0.04810801520943642,
      "learning_rate": 6.790207209764406e-05,
      "loss": 0.029672807455062865,
      "step": 113130
    },
    {
      "epoch": 32.11467499290377,
      "grad_norm": 1.1031101942062378,
      "learning_rate": 6.789923360772069e-05,
      "loss": 0.008333415538072587,
      "step": 113140
    },
    {
      "epoch": 32.117513482827135,
      "grad_norm": 0.23200233280658722,
      "learning_rate": 6.789639511779734e-05,
      "loss": 0.004605308175086975,
      "step": 113150
    },
    {
      "epoch": 32.1203519727505,
      "grad_norm": 0.8474830985069275,
      "learning_rate": 6.789355662787398e-05,
      "loss": 0.009357067197561264,
      "step": 113160
    },
    {
      "epoch": 32.123190462673854,
      "grad_norm": 5.159415245056152,
      "learning_rate": 6.789071813795061e-05,
      "loss": 0.010340078175067902,
      "step": 113170
    },
    {
      "epoch": 32.12602895259722,
      "grad_norm": 0.28161850571632385,
      "learning_rate": 6.788787964802725e-05,
      "loss": 0.0026564829051494597,
      "step": 113180
    },
    {
      "epoch": 32.12886744252058,
      "grad_norm": 0.056018196046352386,
      "learning_rate": 6.78850411581039e-05,
      "loss": 0.009786094725131988,
      "step": 113190
    },
    {
      "epoch": 32.13170593244394,
      "grad_norm": 0.08952254056930542,
      "learning_rate": 6.788220266818052e-05,
      "loss": 0.0024977266788482667,
      "step": 113200
    },
    {
      "epoch": 32.1345444223673,
      "grad_norm": 0.09973842650651932,
      "learning_rate": 6.787936417825716e-05,
      "loss": 0.0027458056807518005,
      "step": 113210
    },
    {
      "epoch": 32.13738291229066,
      "grad_norm": 0.24031929671764374,
      "learning_rate": 6.787652568833382e-05,
      "loss": 0.009127199649810791,
      "step": 113220
    },
    {
      "epoch": 32.140221402214024,
      "grad_norm": 1.9957661628723145,
      "learning_rate": 6.787368719841045e-05,
      "loss": 0.004055127501487732,
      "step": 113230
    },
    {
      "epoch": 32.14305989213738,
      "grad_norm": 0.3266041576862335,
      "learning_rate": 6.787084870848709e-05,
      "loss": 0.0052783235907554625,
      "step": 113240
    },
    {
      "epoch": 32.14589838206074,
      "grad_norm": 0.3612093925476074,
      "learning_rate": 6.786801021856373e-05,
      "loss": 0.007082802057266235,
      "step": 113250
    },
    {
      "epoch": 32.148736871984106,
      "grad_norm": 0.1516474038362503,
      "learning_rate": 6.786517172864037e-05,
      "loss": 0.005345091223716736,
      "step": 113260
    },
    {
      "epoch": 32.15157536190747,
      "grad_norm": 0.09937619417905807,
      "learning_rate": 6.7862333238717e-05,
      "loss": 0.00119515098631382,
      "step": 113270
    },
    {
      "epoch": 32.154413851830824,
      "grad_norm": 13.256715774536133,
      "learning_rate": 6.785949474879365e-05,
      "loss": 0.0075756900012493135,
      "step": 113280
    },
    {
      "epoch": 32.15725234175419,
      "grad_norm": 0.06190642714500427,
      "learning_rate": 6.78566562588703e-05,
      "loss": 0.0021891776472330094,
      "step": 113290
    },
    {
      "epoch": 32.16009083167755,
      "grad_norm": 0.2855495512485504,
      "learning_rate": 6.785381776894692e-05,
      "loss": 0.003480631858110428,
      "step": 113300
    },
    {
      "epoch": 32.162929321600906,
      "grad_norm": 1.2464630603790283,
      "learning_rate": 6.785097927902356e-05,
      "loss": 0.0023585380986332893,
      "step": 113310
    },
    {
      "epoch": 32.16576781152427,
      "grad_norm": 0.13999396562576294,
      "learning_rate": 6.78481407891002e-05,
      "loss": 0.003312782943248749,
      "step": 113320
    },
    {
      "epoch": 32.16860630144763,
      "grad_norm": 4.94569206237793,
      "learning_rate": 6.784530229917683e-05,
      "loss": 0.0034679710865020753,
      "step": 113330
    },
    {
      "epoch": 32.17144479137099,
      "grad_norm": 0.38071539998054504,
      "learning_rate": 6.784246380925347e-05,
      "loss": 0.0010952912271022797,
      "step": 113340
    },
    {
      "epoch": 32.17428328129435,
      "grad_norm": 1.7886611223220825,
      "learning_rate": 6.783962531933013e-05,
      "loss": 0.003755585104227066,
      "step": 113350
    },
    {
      "epoch": 32.17712177121771,
      "grad_norm": 0.37936854362487793,
      "learning_rate": 6.783678682940676e-05,
      "loss": 0.0019212637096643447,
      "step": 113360
    },
    {
      "epoch": 32.179960261141076,
      "grad_norm": 12.635538101196289,
      "learning_rate": 6.78339483394834e-05,
      "loss": 0.014818759262561798,
      "step": 113370
    },
    {
      "epoch": 32.18279875106443,
      "grad_norm": 6.771005153656006,
      "learning_rate": 6.783110984956004e-05,
      "loss": 0.004397585242986679,
      "step": 113380
    },
    {
      "epoch": 32.185637240987795,
      "grad_norm": 2.8157174587249756,
      "learning_rate": 6.782827135963668e-05,
      "loss": 0.011247241497039795,
      "step": 113390
    },
    {
      "epoch": 32.18847573091116,
      "grad_norm": 0.02333211526274681,
      "learning_rate": 6.782543286971331e-05,
      "loss": 0.002285885252058506,
      "step": 113400
    },
    {
      "epoch": 32.19131422083451,
      "grad_norm": 0.7998051047325134,
      "learning_rate": 6.782259437978995e-05,
      "loss": 0.004526317864656448,
      "step": 113410
    },
    {
      "epoch": 32.194152710757876,
      "grad_norm": 0.11939138919115067,
      "learning_rate": 6.78197558898666e-05,
      "loss": 0.003458837419748306,
      "step": 113420
    },
    {
      "epoch": 32.19699120068124,
      "grad_norm": 0.254188597202301,
      "learning_rate": 6.781691739994323e-05,
      "loss": 0.0029236553236842155,
      "step": 113430
    },
    {
      "epoch": 32.199829690604595,
      "grad_norm": 0.388068825006485,
      "learning_rate": 6.781407891001987e-05,
      "loss": 0.00623173788189888,
      "step": 113440
    },
    {
      "epoch": 32.20266818052796,
      "grad_norm": 1.9414633512496948,
      "learning_rate": 6.781124042009652e-05,
      "loss": 0.020403265953063965,
      "step": 113450
    },
    {
      "epoch": 32.20550667045132,
      "grad_norm": 0.13548274338245392,
      "learning_rate": 6.780840193017314e-05,
      "loss": 0.003528503328561783,
      "step": 113460
    },
    {
      "epoch": 32.208345160374684,
      "grad_norm": 0.645809531211853,
      "learning_rate": 6.780556344024978e-05,
      "loss": 0.016390702128410338,
      "step": 113470
    },
    {
      "epoch": 32.21118365029804,
      "grad_norm": 1.702184796333313,
      "learning_rate": 6.780272495032644e-05,
      "loss": 0.014328551292419434,
      "step": 113480
    },
    {
      "epoch": 32.2140221402214,
      "grad_norm": 1.6160175800323486,
      "learning_rate": 6.779988646040307e-05,
      "loss": 0.020063447952270507,
      "step": 113490
    },
    {
      "epoch": 32.216860630144765,
      "grad_norm": 7.184590816497803,
      "learning_rate": 6.779704797047971e-05,
      "loss": 0.00843912735581398,
      "step": 113500
    },
    {
      "epoch": 32.216860630144765,
      "eval_accuracy": 0.9726584854072614,
      "eval_loss": 0.09012199938297272,
      "eval_runtime": 32.3481,
      "eval_samples_per_second": 486.18,
      "eval_steps_per_second": 7.605,
      "step": 113500
    },
    {
      "epoch": 32.21969912006812,
      "grad_norm": 0.3943791687488556,
      "learning_rate": 6.779420948055635e-05,
      "loss": 0.006679734587669373,
      "step": 113510
    },
    {
      "epoch": 32.222537609991484,
      "grad_norm": 3.206735372543335,
      "learning_rate": 6.779137099063299e-05,
      "loss": 0.0063702456653118135,
      "step": 113520
    },
    {
      "epoch": 32.22537609991485,
      "grad_norm": 0.4102219045162201,
      "learning_rate": 6.778853250070962e-05,
      "loss": 0.005461166054010392,
      "step": 113530
    },
    {
      "epoch": 32.2282145898382,
      "grad_norm": 11.555947303771973,
      "learning_rate": 6.778569401078626e-05,
      "loss": 0.021569731831550597,
      "step": 113540
    },
    {
      "epoch": 32.231053079761566,
      "grad_norm": 1.0789655447006226,
      "learning_rate": 6.778285552086292e-05,
      "loss": 0.0069807559251785275,
      "step": 113550
    },
    {
      "epoch": 32.23389156968493,
      "grad_norm": 2.709413528442383,
      "learning_rate": 6.778001703093954e-05,
      "loss": 0.012616276741027832,
      "step": 113560
    },
    {
      "epoch": 32.23673005960829,
      "grad_norm": 0.2750871479511261,
      "learning_rate": 6.777717854101618e-05,
      "loss": 0.00706607848405838,
      "step": 113570
    },
    {
      "epoch": 32.23956854953165,
      "grad_norm": 0.15584616363048553,
      "learning_rate": 6.777434005109283e-05,
      "loss": 0.00997084304690361,
      "step": 113580
    },
    {
      "epoch": 32.24240703945501,
      "grad_norm": 0.5570263266563416,
      "learning_rate": 6.777150156116945e-05,
      "loss": 0.0024737365543842315,
      "step": 113590
    },
    {
      "epoch": 32.24524552937837,
      "grad_norm": 0.10775743424892426,
      "learning_rate": 6.77686630712461e-05,
      "loss": 0.0042981825768947605,
      "step": 113600
    },
    {
      "epoch": 32.24808401930173,
      "grad_norm": 0.38575270771980286,
      "learning_rate": 6.776582458132274e-05,
      "loss": 0.0026128383353352545,
      "step": 113610
    },
    {
      "epoch": 32.25092250922509,
      "grad_norm": 0.18029536306858063,
      "learning_rate": 6.776298609139938e-05,
      "loss": 0.003873720020055771,
      "step": 113620
    },
    {
      "epoch": 32.253760999148454,
      "grad_norm": 0.3262554705142975,
      "learning_rate": 6.776014760147602e-05,
      "loss": 0.006739795207977295,
      "step": 113630
    },
    {
      "epoch": 32.25659948907182,
      "grad_norm": 0.26409804821014404,
      "learning_rate": 6.775730911155266e-05,
      "loss": 0.004262235015630722,
      "step": 113640
    },
    {
      "epoch": 32.25943797899517,
      "grad_norm": 0.3425769805908203,
      "learning_rate": 6.77544706216293e-05,
      "loss": 0.002111529745161533,
      "step": 113650
    },
    {
      "epoch": 32.262276468918536,
      "grad_norm": 0.72054123878479,
      "learning_rate": 6.775163213170593e-05,
      "loss": 0.00101242084056139,
      "step": 113660
    },
    {
      "epoch": 32.2651149588419,
      "grad_norm": 7.504349231719971,
      "learning_rate": 6.774879364178257e-05,
      "loss": 0.0037106871604919434,
      "step": 113670
    },
    {
      "epoch": 32.267953448765255,
      "grad_norm": 9.559056282043457,
      "learning_rate": 6.774595515185921e-05,
      "loss": 0.010646243393421174,
      "step": 113680
    },
    {
      "epoch": 32.27079193868862,
      "grad_norm": 0.05254020914435387,
      "learning_rate": 6.774311666193585e-05,
      "loss": 0.0009725546464323997,
      "step": 113690
    },
    {
      "epoch": 32.27363042861198,
      "grad_norm": 1.4245951175689697,
      "learning_rate": 6.77402781720125e-05,
      "loss": 0.004171088710427285,
      "step": 113700
    },
    {
      "epoch": 32.276468918535336,
      "grad_norm": 0.15658673644065857,
      "learning_rate": 6.773743968208914e-05,
      "loss": 0.00855831429362297,
      "step": 113710
    },
    {
      "epoch": 32.2793074084587,
      "grad_norm": 1.3264540433883667,
      "learning_rate": 6.773460119216576e-05,
      "loss": 0.006788210570812225,
      "step": 113720
    },
    {
      "epoch": 32.28214589838206,
      "grad_norm": 0.03787371888756752,
      "learning_rate": 6.77317627022424e-05,
      "loss": 0.0025403501465916635,
      "step": 113730
    },
    {
      "epoch": 32.284984388305425,
      "grad_norm": 0.7597050070762634,
      "learning_rate": 6.772892421231905e-05,
      "loss": 0.007415363192558288,
      "step": 113740
    },
    {
      "epoch": 32.28782287822878,
      "grad_norm": 1.1544818878173828,
      "learning_rate": 6.772608572239569e-05,
      "loss": 0.0034408807754516602,
      "step": 113750
    },
    {
      "epoch": 32.29066136815214,
      "grad_norm": 0.3979548215866089,
      "learning_rate": 6.772324723247233e-05,
      "loss": 0.002588382549583912,
      "step": 113760
    },
    {
      "epoch": 32.293499858075506,
      "grad_norm": 0.025236234068870544,
      "learning_rate": 6.772040874254897e-05,
      "loss": 0.00349077433347702,
      "step": 113770
    },
    {
      "epoch": 32.29633834799886,
      "grad_norm": 0.3230131268501282,
      "learning_rate": 6.77175702526256e-05,
      "loss": 0.0201138973236084,
      "step": 113780
    },
    {
      "epoch": 32.299176837922225,
      "grad_norm": 2.1914327144622803,
      "learning_rate": 6.771473176270224e-05,
      "loss": 0.006201978772878647,
      "step": 113790
    },
    {
      "epoch": 32.30201532784559,
      "grad_norm": 0.03451480716466904,
      "learning_rate": 6.771189327277888e-05,
      "loss": 0.005251342803239823,
      "step": 113800
    },
    {
      "epoch": 32.304853817768944,
      "grad_norm": 1.743640422821045,
      "learning_rate": 6.770905478285552e-05,
      "loss": 0.002060825377702713,
      "step": 113810
    },
    {
      "epoch": 32.30769230769231,
      "grad_norm": 0.22964555025100708,
      "learning_rate": 6.770621629293216e-05,
      "loss": 0.0022746950387954714,
      "step": 113820
    },
    {
      "epoch": 32.31053079761567,
      "grad_norm": 2.10322904586792,
      "learning_rate": 6.77033778030088e-05,
      "loss": 0.007051681727170944,
      "step": 113830
    },
    {
      "epoch": 32.31336928753903,
      "grad_norm": 0.04390684515237808,
      "learning_rate": 6.770053931308545e-05,
      "loss": 0.008618693053722381,
      "step": 113840
    },
    {
      "epoch": 32.31620777746239,
      "grad_norm": 16.890708923339844,
      "learning_rate": 6.769770082316208e-05,
      "loss": 0.006460443139076233,
      "step": 113850
    },
    {
      "epoch": 32.31904626738575,
      "grad_norm": 1.251128077507019,
      "learning_rate": 6.769486233323872e-05,
      "loss": 0.012202472239732743,
      "step": 113860
    },
    {
      "epoch": 32.321884757309114,
      "grad_norm": 1.1209325790405273,
      "learning_rate": 6.769202384331536e-05,
      "loss": 0.0037514366209506988,
      "step": 113870
    },
    {
      "epoch": 32.32472324723247,
      "grad_norm": 2.015658378601074,
      "learning_rate": 6.7689185353392e-05,
      "loss": 0.011603143811225892,
      "step": 113880
    },
    {
      "epoch": 32.32756173715583,
      "grad_norm": 0.68536776304245,
      "learning_rate": 6.768634686346864e-05,
      "loss": 0.0025602078065276148,
      "step": 113890
    },
    {
      "epoch": 32.330400227079195,
      "grad_norm": 0.606306254863739,
      "learning_rate": 6.768350837354528e-05,
      "loss": 0.0037806253880262376,
      "step": 113900
    },
    {
      "epoch": 32.33323871700255,
      "grad_norm": 0.476772665977478,
      "learning_rate": 6.768066988362191e-05,
      "loss": 0.0032348908483982087,
      "step": 113910
    },
    {
      "epoch": 32.336077206925914,
      "grad_norm": 0.708996057510376,
      "learning_rate": 6.767783139369855e-05,
      "loss": 0.010649625211954117,
      "step": 113920
    },
    {
      "epoch": 32.33891569684928,
      "grad_norm": 2.5708374977111816,
      "learning_rate": 6.767499290377519e-05,
      "loss": 0.0012338876724243164,
      "step": 113930
    },
    {
      "epoch": 32.34175418677264,
      "grad_norm": 6.922987461090088,
      "learning_rate": 6.767215441385183e-05,
      "loss": 0.0027254000306129456,
      "step": 113940
    },
    {
      "epoch": 32.344592676695996,
      "grad_norm": 0.010990005917847157,
      "learning_rate": 6.766931592392848e-05,
      "loss": 0.004330401495099068,
      "step": 113950
    },
    {
      "epoch": 32.34743116661936,
      "grad_norm": 3.2982280254364014,
      "learning_rate": 6.766647743400512e-05,
      "loss": 0.009295592457056046,
      "step": 113960
    },
    {
      "epoch": 32.35026965654272,
      "grad_norm": 2.494295358657837,
      "learning_rate": 6.766363894408176e-05,
      "loss": 0.006447308510541916,
      "step": 113970
    },
    {
      "epoch": 32.35310814646608,
      "grad_norm": 4.8559746742248535,
      "learning_rate": 6.766080045415839e-05,
      "loss": 0.008917318284511566,
      "step": 113980
    },
    {
      "epoch": 32.35594663638944,
      "grad_norm": 2.0082144737243652,
      "learning_rate": 6.765796196423503e-05,
      "loss": 0.002145201340317726,
      "step": 113990
    },
    {
      "epoch": 32.3587851263128,
      "grad_norm": 6.820044040679932,
      "learning_rate": 6.765512347431167e-05,
      "loss": 0.011418785899877548,
      "step": 114000
    },
    {
      "epoch": 32.3587851263128,
      "eval_accuracy": 0.9685890506771794,
      "eval_loss": 0.10716046392917633,
      "eval_runtime": 32.3516,
      "eval_samples_per_second": 486.127,
      "eval_steps_per_second": 7.604,
      "step": 114000
    },
    {
      "epoch": 32.36162361623616,
      "grad_norm": 0.42937737703323364,
      "learning_rate": 6.76522849843883e-05,
      "loss": 0.007951658964157105,
      "step": 114010
    },
    {
      "epoch": 32.36446210615952,
      "grad_norm": 0.15281030535697937,
      "learning_rate": 6.764944649446495e-05,
      "loss": 0.0021665066480636595,
      "step": 114020
    },
    {
      "epoch": 32.367300596082885,
      "grad_norm": 0.5121237635612488,
      "learning_rate": 6.764660800454159e-05,
      "loss": 0.003668125718832016,
      "step": 114030
    },
    {
      "epoch": 32.37013908600625,
      "grad_norm": 0.6893860697746277,
      "learning_rate": 6.764376951461822e-05,
      "loss": 0.005641918629407883,
      "step": 114040
    },
    {
      "epoch": 32.3729775759296,
      "grad_norm": 0.22708742320537567,
      "learning_rate": 6.764093102469486e-05,
      "loss": 0.003956592082977295,
      "step": 114050
    },
    {
      "epoch": 32.375816065852966,
      "grad_norm": 0.43383705615997314,
      "learning_rate": 6.76380925347715e-05,
      "loss": 0.004662372171878815,
      "step": 114060
    },
    {
      "epoch": 32.37865455577633,
      "grad_norm": 3.3631432056427,
      "learning_rate": 6.763525404484815e-05,
      "loss": 0.004196884483098984,
      "step": 114070
    },
    {
      "epoch": 32.381493045699685,
      "grad_norm": 6.487922668457031,
      "learning_rate": 6.763241555492479e-05,
      "loss": 0.008331590890884399,
      "step": 114080
    },
    {
      "epoch": 32.38433153562305,
      "grad_norm": 0.16430975496768951,
      "learning_rate": 6.762957706500143e-05,
      "loss": 0.0068503163754940035,
      "step": 114090
    },
    {
      "epoch": 32.38717002554641,
      "grad_norm": 0.17624656856060028,
      "learning_rate": 6.762673857507807e-05,
      "loss": 0.0027605874463915827,
      "step": 114100
    },
    {
      "epoch": 32.39000851546977,
      "grad_norm": 1.004976511001587,
      "learning_rate": 6.76239000851547e-05,
      "loss": 0.003381132334470749,
      "step": 114110
    },
    {
      "epoch": 32.39284700539313,
      "grad_norm": 5.970203399658203,
      "learning_rate": 6.762106159523134e-05,
      "loss": 0.007661319524049759,
      "step": 114120
    },
    {
      "epoch": 32.39568549531649,
      "grad_norm": 5.56792688369751,
      "learning_rate": 6.761822310530798e-05,
      "loss": 0.00980997532606125,
      "step": 114130
    },
    {
      "epoch": 32.398523985239855,
      "grad_norm": 2.284970283508301,
      "learning_rate": 6.761538461538461e-05,
      "loss": 0.0037452824413776397,
      "step": 114140
    },
    {
      "epoch": 32.40136247516321,
      "grad_norm": 0.7185450792312622,
      "learning_rate": 6.761254612546126e-05,
      "loss": 0.0057276800274848935,
      "step": 114150
    },
    {
      "epoch": 32.404200965086574,
      "grad_norm": 0.7018328309059143,
      "learning_rate": 6.76097076355379e-05,
      "loss": 0.0018034325912594795,
      "step": 114160
    },
    {
      "epoch": 32.40703945500994,
      "grad_norm": 1.0469950437545776,
      "learning_rate": 6.760686914561453e-05,
      "loss": 0.003090457245707512,
      "step": 114170
    },
    {
      "epoch": 32.40987794493329,
      "grad_norm": 7.388956546783447,
      "learning_rate": 6.760403065569117e-05,
      "loss": 0.01225554421544075,
      "step": 114180
    },
    {
      "epoch": 32.412716434856655,
      "grad_norm": 3.4771299362182617,
      "learning_rate": 6.760119216576781e-05,
      "loss": 0.009931269288063049,
      "step": 114190
    },
    {
      "epoch": 32.41555492478002,
      "grad_norm": 5.119248867034912,
      "learning_rate": 6.759835367584446e-05,
      "loss": 0.012873387336730957,
      "step": 114200
    },
    {
      "epoch": 32.41839341470338,
      "grad_norm": 3.1303441524505615,
      "learning_rate": 6.759551518592108e-05,
      "loss": 0.00588199570775032,
      "step": 114210
    },
    {
      "epoch": 32.42123190462674,
      "grad_norm": 3.0082404613494873,
      "learning_rate": 6.759267669599774e-05,
      "loss": 0.010008987784385682,
      "step": 114220
    },
    {
      "epoch": 32.4240703945501,
      "grad_norm": 6.552779197692871,
      "learning_rate": 6.758983820607438e-05,
      "loss": 0.011698020249605178,
      "step": 114230
    },
    {
      "epoch": 32.42690888447346,
      "grad_norm": 0.15180553495883942,
      "learning_rate": 6.758699971615101e-05,
      "loss": 0.009099248051643371,
      "step": 114240
    },
    {
      "epoch": 32.42974737439682,
      "grad_norm": 2.1065170764923096,
      "learning_rate": 6.758416122622765e-05,
      "loss": 0.02688334584236145,
      "step": 114250
    },
    {
      "epoch": 32.43258586432018,
      "grad_norm": 0.3048502504825592,
      "learning_rate": 6.758132273630429e-05,
      "loss": 0.010374457389116288,
      "step": 114260
    },
    {
      "epoch": 32.435424354243544,
      "grad_norm": 0.6098058819770813,
      "learning_rate": 6.757848424638092e-05,
      "loss": 0.00654008835554123,
      "step": 114270
    },
    {
      "epoch": 32.4382628441669,
      "grad_norm": 0.9462328553199768,
      "learning_rate": 6.757564575645757e-05,
      "loss": 0.009632043540477753,
      "step": 114280
    },
    {
      "epoch": 32.44110133409026,
      "grad_norm": 0.11408546566963196,
      "learning_rate": 6.757280726653421e-05,
      "loss": 0.007204797863960266,
      "step": 114290
    },
    {
      "epoch": 32.443939824013626,
      "grad_norm": 1.948219895362854,
      "learning_rate": 6.756996877661084e-05,
      "loss": 0.00823022872209549,
      "step": 114300
    },
    {
      "epoch": 32.44677831393699,
      "grad_norm": 0.8632134199142456,
      "learning_rate": 6.756713028668748e-05,
      "loss": 0.0020861439406871796,
      "step": 114310
    },
    {
      "epoch": 32.449616803860344,
      "grad_norm": 4.6393561363220215,
      "learning_rate": 6.756429179676413e-05,
      "loss": 0.009569059312343597,
      "step": 114320
    },
    {
      "epoch": 32.45245529378371,
      "grad_norm": 0.14281955361366272,
      "learning_rate": 6.756145330684077e-05,
      "loss": 0.011338746547698975,
      "step": 114330
    },
    {
      "epoch": 32.45529378370707,
      "grad_norm": 0.2974180579185486,
      "learning_rate": 6.75586148169174e-05,
      "loss": 0.01003139391541481,
      "step": 114340
    },
    {
      "epoch": 32.458132273630426,
      "grad_norm": 0.28766438364982605,
      "learning_rate": 6.755577632699405e-05,
      "loss": 0.007242113351821899,
      "step": 114350
    },
    {
      "epoch": 32.46097076355379,
      "grad_norm": 9.838591575622559,
      "learning_rate": 6.755293783707069e-05,
      "loss": 0.016401273012161256,
      "step": 114360
    },
    {
      "epoch": 32.46380925347715,
      "grad_norm": 0.8696520328521729,
      "learning_rate": 6.755009934714732e-05,
      "loss": 0.006739495694637299,
      "step": 114370
    },
    {
      "epoch": 32.46664774340051,
      "grad_norm": 11.485713005065918,
      "learning_rate": 6.754726085722396e-05,
      "loss": 0.010914240032434464,
      "step": 114380
    },
    {
      "epoch": 32.46948623332387,
      "grad_norm": 0.9926250576972961,
      "learning_rate": 6.75444223673006e-05,
      "loss": 0.004315894842147827,
      "step": 114390
    },
    {
      "epoch": 32.47232472324723,
      "grad_norm": 6.109653472900391,
      "learning_rate": 6.754158387737723e-05,
      "loss": 0.006377974152565002,
      "step": 114400
    },
    {
      "epoch": 32.475163213170596,
      "grad_norm": 1.0385361909866333,
      "learning_rate": 6.753874538745387e-05,
      "loss": 0.001586764119565487,
      "step": 114410
    },
    {
      "epoch": 32.47800170309395,
      "grad_norm": 0.43349531292915344,
      "learning_rate": 6.753590689753053e-05,
      "loss": 0.002714262530207634,
      "step": 114420
    },
    {
      "epoch": 32.480840193017315,
      "grad_norm": 6.788949966430664,
      "learning_rate": 6.753306840760715e-05,
      "loss": 0.007755088806152344,
      "step": 114430
    },
    {
      "epoch": 32.48367868294068,
      "grad_norm": 0.26468878984451294,
      "learning_rate": 6.75302299176838e-05,
      "loss": 0.01067410334944725,
      "step": 114440
    },
    {
      "epoch": 32.48651717286403,
      "grad_norm": 1.767830491065979,
      "learning_rate": 6.752739142776044e-05,
      "loss": 0.011746762692928315,
      "step": 114450
    },
    {
      "epoch": 32.489355662787396,
      "grad_norm": 6.436022758483887,
      "learning_rate": 6.752483678682942e-05,
      "loss": 0.016062068939208984,
      "step": 114460
    },
    {
      "epoch": 32.49219415271076,
      "grad_norm": 0.5355373024940491,
      "learning_rate": 6.752199829690606e-05,
      "loss": 0.005138899385929108,
      "step": 114470
    },
    {
      "epoch": 32.49503264263412,
      "grad_norm": 14.387542724609375,
      "learning_rate": 6.751915980698269e-05,
      "loss": 0.006621388345956802,
      "step": 114480
    },
    {
      "epoch": 32.49787113255748,
      "grad_norm": 5.61583137512207,
      "learning_rate": 6.751632131705933e-05,
      "loss": 0.007416114956140518,
      "step": 114490
    },
    {
      "epoch": 32.50070962248084,
      "grad_norm": 12.762537956237793,
      "learning_rate": 6.751348282713597e-05,
      "loss": 0.009079305082559585,
      "step": 114500
    },
    {
      "epoch": 32.50070962248084,
      "eval_accuracy": 0.9683347110065492,
      "eval_loss": 0.11194168031215668,
      "eval_runtime": 31.9034,
      "eval_samples_per_second": 492.956,
      "eval_steps_per_second": 7.711,
      "step": 114500
    },
    {
      "epoch": 32.503548112404204,
      "grad_norm": 0.8387620449066162,
      "learning_rate": 6.751064433721261e-05,
      "loss": 0.010013447701931,
      "step": 114510
    },
    {
      "epoch": 32.50638660232756,
      "grad_norm": 1.4259146451950073,
      "learning_rate": 6.750780584728924e-05,
      "loss": 0.003548775985836983,
      "step": 114520
    },
    {
      "epoch": 32.50922509225092,
      "grad_norm": 1.381527304649353,
      "learning_rate": 6.750496735736589e-05,
      "loss": 0.0032875917851924895,
      "step": 114530
    },
    {
      "epoch": 32.512063582174285,
      "grad_norm": 0.05731292441487312,
      "learning_rate": 6.750212886744253e-05,
      "loss": 0.0030773110687732697,
      "step": 114540
    },
    {
      "epoch": 32.51490207209764,
      "grad_norm": 1.166197657585144,
      "learning_rate": 6.749929037751916e-05,
      "loss": 0.003911464288830757,
      "step": 114550
    },
    {
      "epoch": 32.517740562021004,
      "grad_norm": 0.865229070186615,
      "learning_rate": 6.74964518875958e-05,
      "loss": 0.009720176458358765,
      "step": 114560
    },
    {
      "epoch": 32.52057905194437,
      "grad_norm": 1.2768666744232178,
      "learning_rate": 6.749361339767244e-05,
      "loss": 0.005226878821849823,
      "step": 114570
    },
    {
      "epoch": 32.52341754186773,
      "grad_norm": 10.134637832641602,
      "learning_rate": 6.749077490774907e-05,
      "loss": 0.004003226011991501,
      "step": 114580
    },
    {
      "epoch": 32.526256031791085,
      "grad_norm": 0.021010393276810646,
      "learning_rate": 6.748793641782571e-05,
      "loss": 0.004367262870073318,
      "step": 114590
    },
    {
      "epoch": 32.52909452171445,
      "grad_norm": 1.3082411289215088,
      "learning_rate": 6.748509792790237e-05,
      "loss": 0.0022121984511613846,
      "step": 114600
    },
    {
      "epoch": 32.53193301163781,
      "grad_norm": 0.13424289226531982,
      "learning_rate": 6.7482259437979e-05,
      "loss": 0.004100373387336731,
      "step": 114610
    },
    {
      "epoch": 32.53477150156117,
      "grad_norm": 0.02352616749703884,
      "learning_rate": 6.747942094805564e-05,
      "loss": 0.0024976471439003946,
      "step": 114620
    },
    {
      "epoch": 32.53760999148453,
      "grad_norm": 0.36953461170196533,
      "learning_rate": 6.747658245813228e-05,
      "loss": 0.0067359328269958494,
      "step": 114630
    },
    {
      "epoch": 32.54044848140789,
      "grad_norm": 1.9611560106277466,
      "learning_rate": 6.747374396820892e-05,
      "loss": 0.002916085720062256,
      "step": 114640
    },
    {
      "epoch": 32.54328697133125,
      "grad_norm": 0.06334231048822403,
      "learning_rate": 6.747090547828555e-05,
      "loss": 0.004664639756083488,
      "step": 114650
    },
    {
      "epoch": 32.54612546125461,
      "grad_norm": 0.27794936299324036,
      "learning_rate": 6.74680669883622e-05,
      "loss": 0.004776120185852051,
      "step": 114660
    },
    {
      "epoch": 32.548963951177974,
      "grad_norm": 0.05011090636253357,
      "learning_rate": 6.746522849843883e-05,
      "loss": 0.017741817235946655,
      "step": 114670
    },
    {
      "epoch": 32.55180244110134,
      "grad_norm": 9.946207046508789,
      "learning_rate": 6.746239000851547e-05,
      "loss": 0.0071158245205879215,
      "step": 114680
    },
    {
      "epoch": 32.55464093102469,
      "grad_norm": 11.498275756835938,
      "learning_rate": 6.745955151859211e-05,
      "loss": 0.011961480975151062,
      "step": 114690
    },
    {
      "epoch": 32.557479420948056,
      "grad_norm": 0.5114032030105591,
      "learning_rate": 6.745671302866875e-05,
      "loss": 0.007185105234384537,
      "step": 114700
    },
    {
      "epoch": 32.56031791087142,
      "grad_norm": 3.362053871154785,
      "learning_rate": 6.745387453874538e-05,
      "loss": 0.0054019328206777574,
      "step": 114710
    },
    {
      "epoch": 32.563156400794774,
      "grad_norm": 0.14989155530929565,
      "learning_rate": 6.745103604882202e-05,
      "loss": 0.002203916572034359,
      "step": 114720
    },
    {
      "epoch": 32.56599489071814,
      "grad_norm": 2.6054012775421143,
      "learning_rate": 6.744819755889868e-05,
      "loss": 0.003519074618816376,
      "step": 114730
    },
    {
      "epoch": 32.5688333806415,
      "grad_norm": 0.688409149646759,
      "learning_rate": 6.74453590689753e-05,
      "loss": 0.0169049933552742,
      "step": 114740
    },
    {
      "epoch": 32.571671870564856,
      "grad_norm": 1.8891355991363525,
      "learning_rate": 6.744252057905195e-05,
      "loss": 0.008291397243738174,
      "step": 114750
    },
    {
      "epoch": 32.57451036048822,
      "grad_norm": 0.9401333332061768,
      "learning_rate": 6.743968208912859e-05,
      "loss": 0.010648560523986817,
      "step": 114760
    },
    {
      "epoch": 32.57734885041158,
      "grad_norm": 0.44594964385032654,
      "learning_rate": 6.743684359920522e-05,
      "loss": 0.006029950082302093,
      "step": 114770
    },
    {
      "epoch": 32.580187340334945,
      "grad_norm": 0.15582551062107086,
      "learning_rate": 6.743400510928186e-05,
      "loss": 0.0011174697428941727,
      "step": 114780
    },
    {
      "epoch": 32.5830258302583,
      "grad_norm": 14.366189002990723,
      "learning_rate": 6.74311666193585e-05,
      "loss": 0.017259415984153748,
      "step": 114790
    },
    {
      "epoch": 32.58586432018166,
      "grad_norm": 6.311891555786133,
      "learning_rate": 6.742832812943514e-05,
      "loss": 0.0034424431622028353,
      "step": 114800
    },
    {
      "epoch": 32.588702810105026,
      "grad_norm": 0.5163989663124084,
      "learning_rate": 6.742548963951178e-05,
      "loss": 0.0036771148443222046,
      "step": 114810
    },
    {
      "epoch": 32.59154130002838,
      "grad_norm": 0.13861115276813507,
      "learning_rate": 6.742265114958842e-05,
      "loss": 0.0036024361848831176,
      "step": 114820
    },
    {
      "epoch": 32.594379789951745,
      "grad_norm": 0.10339943319559097,
      "learning_rate": 6.741981265966507e-05,
      "loss": 0.0018397554755210876,
      "step": 114830
    },
    {
      "epoch": 32.59721827987511,
      "grad_norm": 0.09476244449615479,
      "learning_rate": 6.74169741697417e-05,
      "loss": 0.006398230791091919,
      "step": 114840
    },
    {
      "epoch": 32.60005676979847,
      "grad_norm": 5.224700450897217,
      "learning_rate": 6.741413567981833e-05,
      "loss": 0.0051482245326042175,
      "step": 114850
    },
    {
      "epoch": 32.60289525972183,
      "grad_norm": 0.10674237459897995,
      "learning_rate": 6.741129718989499e-05,
      "loss": 0.007873215526342393,
      "step": 114860
    },
    {
      "epoch": 32.60573374964519,
      "grad_norm": 0.29810312390327454,
      "learning_rate": 6.740845869997162e-05,
      "loss": 0.00433870404958725,
      "step": 114870
    },
    {
      "epoch": 32.60857223956855,
      "grad_norm": 0.07812470197677612,
      "learning_rate": 6.740562021004826e-05,
      "loss": 0.0114394411444664,
      "step": 114880
    },
    {
      "epoch": 32.61141072949191,
      "grad_norm": 0.3613995611667633,
      "learning_rate": 6.74027817201249e-05,
      "loss": 0.006175015866756439,
      "step": 114890
    },
    {
      "epoch": 32.61424921941527,
      "grad_norm": 2.0568666458129883,
      "learning_rate": 6.739994323020153e-05,
      "loss": 0.0044817298650741575,
      "step": 114900
    },
    {
      "epoch": 32.617087709338634,
      "grad_norm": 10.966506958007812,
      "learning_rate": 6.739710474027817e-05,
      "loss": 0.010058432817459106,
      "step": 114910
    },
    {
      "epoch": 32.61992619926199,
      "grad_norm": 0.2765500247478485,
      "learning_rate": 6.739426625035481e-05,
      "loss": 0.006062079221010208,
      "step": 114920
    },
    {
      "epoch": 32.62276468918535,
      "grad_norm": 0.9741296172142029,
      "learning_rate": 6.739142776043145e-05,
      "loss": 0.013068430125713348,
      "step": 114930
    },
    {
      "epoch": 32.625603179108715,
      "grad_norm": 0.11648418009281158,
      "learning_rate": 6.73885892705081e-05,
      "loss": 0.003900745138525963,
      "step": 114940
    },
    {
      "epoch": 32.62844166903208,
      "grad_norm": 0.2820674479007721,
      "learning_rate": 6.738575078058473e-05,
      "loss": 0.020217898488044738,
      "step": 114950
    },
    {
      "epoch": 32.631280158955434,
      "grad_norm": 1.2293736934661865,
      "learning_rate": 6.738291229066138e-05,
      "loss": 0.008869744092226028,
      "step": 114960
    },
    {
      "epoch": 32.6341186488788,
      "grad_norm": 1.0221351385116577,
      "learning_rate": 6.7380073800738e-05,
      "loss": 0.002987317927181721,
      "step": 114970
    },
    {
      "epoch": 32.63695713880216,
      "grad_norm": 0.21123717725276947,
      "learning_rate": 6.737723531081465e-05,
      "loss": 0.006773201376199722,
      "step": 114980
    },
    {
      "epoch": 32.639795628725516,
      "grad_norm": 0.2600741386413574,
      "learning_rate": 6.73743968208913e-05,
      "loss": 0.013052636384963989,
      "step": 114990
    },
    {
      "epoch": 32.64263411864888,
      "grad_norm": 5.860227584838867,
      "learning_rate": 6.737155833096793e-05,
      "loss": 0.007457506656646728,
      "step": 115000
    },
    {
      "epoch": 32.64263411864888,
      "eval_accuracy": 0.9722133909836587,
      "eval_loss": 0.09420241415500641,
      "eval_runtime": 31.7041,
      "eval_samples_per_second": 496.056,
      "eval_steps_per_second": 7.759,
      "step": 115000
    },
    {
      "epoch": 32.64547260857224,
      "grad_norm": 0.8755062222480774,
      "learning_rate": 6.736871984104457e-05,
      "loss": 0.0052278783172369,
      "step": 115010
    },
    {
      "epoch": 32.6483110984956,
      "grad_norm": 3.6057851314544678,
      "learning_rate": 6.736588135112121e-05,
      "loss": 0.0036074697971343996,
      "step": 115020
    },
    {
      "epoch": 32.65114958841896,
      "grad_norm": 1.4727593660354614,
      "learning_rate": 6.736304286119784e-05,
      "loss": 0.0014422724023461341,
      "step": 115030
    },
    {
      "epoch": 32.65398807834232,
      "grad_norm": 0.27728572487831116,
      "learning_rate": 6.736020437127448e-05,
      "loss": 0.003662591427564621,
      "step": 115040
    },
    {
      "epoch": 32.656826568265686,
      "grad_norm": 7.0233540534973145,
      "learning_rate": 6.735736588135112e-05,
      "loss": 0.00503544993698597,
      "step": 115050
    },
    {
      "epoch": 32.65966505818904,
      "grad_norm": 0.052828531712293625,
      "learning_rate": 6.735452739142776e-05,
      "loss": 0.005121733248233795,
      "step": 115060
    },
    {
      "epoch": 32.662503548112404,
      "grad_norm": 1.258283019065857,
      "learning_rate": 6.73516889015044e-05,
      "loss": 0.0035679638385772705,
      "step": 115070
    },
    {
      "epoch": 32.66534203803577,
      "grad_norm": 1.0674129724502563,
      "learning_rate": 6.734885041158105e-05,
      "loss": 0.0040055826306343075,
      "step": 115080
    },
    {
      "epoch": 32.66818052795912,
      "grad_norm": 5.228094577789307,
      "learning_rate": 6.734601192165769e-05,
      "loss": 0.0068355418741703035,
      "step": 115090
    },
    {
      "epoch": 32.671019017882486,
      "grad_norm": 0.4123722314834595,
      "learning_rate": 6.734317343173431e-05,
      "loss": 0.0017310673370957374,
      "step": 115100
    },
    {
      "epoch": 32.67385750780585,
      "grad_norm": 2.8289482593536377,
      "learning_rate": 6.734033494181096e-05,
      "loss": 0.00764349028468132,
      "step": 115110
    },
    {
      "epoch": 32.676695997729205,
      "grad_norm": 5.455272197723389,
      "learning_rate": 6.73374964518876e-05,
      "loss": 0.005287186056375503,
      "step": 115120
    },
    {
      "epoch": 32.67953448765257,
      "grad_norm": 1.1194788217544556,
      "learning_rate": 6.733465796196424e-05,
      "loss": 0.003764507174491882,
      "step": 115130
    },
    {
      "epoch": 32.68237297757593,
      "grad_norm": 0.3061029016971588,
      "learning_rate": 6.733181947204088e-05,
      "loss": 0.002386040985584259,
      "step": 115140
    },
    {
      "epoch": 32.68521146749929,
      "grad_norm": 16.222301483154297,
      "learning_rate": 6.732898098211752e-05,
      "loss": 0.013604512810707093,
      "step": 115150
    },
    {
      "epoch": 32.68804995742265,
      "grad_norm": 0.13470429182052612,
      "learning_rate": 6.732614249219415e-05,
      "loss": 0.004993880167603493,
      "step": 115160
    },
    {
      "epoch": 32.69088844734601,
      "grad_norm": 0.17951203882694244,
      "learning_rate": 6.732330400227079e-05,
      "loss": 0.0012356458231806756,
      "step": 115170
    },
    {
      "epoch": 32.693726937269375,
      "grad_norm": 15.069907188415527,
      "learning_rate": 6.732046551234743e-05,
      "loss": 0.0108691968023777,
      "step": 115180
    },
    {
      "epoch": 32.69656542719273,
      "grad_norm": 1.161422610282898,
      "learning_rate": 6.731762702242407e-05,
      "loss": 0.0027802960947155954,
      "step": 115190
    },
    {
      "epoch": 32.69940391711609,
      "grad_norm": 10.603035926818848,
      "learning_rate": 6.731478853250072e-05,
      "loss": 0.01007988303899765,
      "step": 115200
    },
    {
      "epoch": 32.702242407039456,
      "grad_norm": 0.06016096845269203,
      "learning_rate": 6.731195004257736e-05,
      "loss": 0.006822508573532104,
      "step": 115210
    },
    {
      "epoch": 32.70508089696281,
      "grad_norm": 0.155114084482193,
      "learning_rate": 6.7309111552654e-05,
      "loss": 0.0016975319012999535,
      "step": 115220
    },
    {
      "epoch": 32.707919386886175,
      "grad_norm": 0.5317871570587158,
      "learning_rate": 6.730627306273063e-05,
      "loss": 0.010537860542535782,
      "step": 115230
    },
    {
      "epoch": 32.71075787680954,
      "grad_norm": 0.2896820306777954,
      "learning_rate": 6.730343457280727e-05,
      "loss": 0.004714943468570709,
      "step": 115240
    },
    {
      "epoch": 32.7135963667329,
      "grad_norm": 1.5274219512939453,
      "learning_rate": 6.730059608288391e-05,
      "loss": 0.004906470328569413,
      "step": 115250
    },
    {
      "epoch": 32.71643485665626,
      "grad_norm": 3.5981786251068115,
      "learning_rate": 6.729775759296055e-05,
      "loss": 0.006560913473367691,
      "step": 115260
    },
    {
      "epoch": 32.71927334657962,
      "grad_norm": 0.1862967163324356,
      "learning_rate": 6.729491910303719e-05,
      "loss": 0.015175075829029083,
      "step": 115270
    },
    {
      "epoch": 32.72211183650298,
      "grad_norm": 0.7134433388710022,
      "learning_rate": 6.729208061311383e-05,
      "loss": 0.007485075294971466,
      "step": 115280
    },
    {
      "epoch": 32.72495032642634,
      "grad_norm": 6.883807182312012,
      "learning_rate": 6.728924212319046e-05,
      "loss": 0.009930762648582458,
      "step": 115290
    },
    {
      "epoch": 32.7277888163497,
      "grad_norm": 0.060246542096138,
      "learning_rate": 6.72864036332671e-05,
      "loss": 0.01544618308544159,
      "step": 115300
    },
    {
      "epoch": 32.730627306273064,
      "grad_norm": 0.44245436787605286,
      "learning_rate": 6.728356514334374e-05,
      "loss": 0.011322715133428574,
      "step": 115310
    },
    {
      "epoch": 32.73346579619643,
      "grad_norm": 9.175596237182617,
      "learning_rate": 6.728072665342038e-05,
      "loss": 0.003295111656188965,
      "step": 115320
    },
    {
      "epoch": 32.73630428611978,
      "grad_norm": 0.7547280788421631,
      "learning_rate": 6.727788816349703e-05,
      "loss": 0.008293887227773666,
      "step": 115330
    },
    {
      "epoch": 32.739142776043145,
      "grad_norm": 1.4769798517227173,
      "learning_rate": 6.727504967357367e-05,
      "loss": 0.0032275639474391936,
      "step": 115340
    },
    {
      "epoch": 32.74198126596651,
      "grad_norm": 2.8032193183898926,
      "learning_rate": 6.727221118365031e-05,
      "loss": 0.006897037476301193,
      "step": 115350
    },
    {
      "epoch": 32.744819755889864,
      "grad_norm": 3.883713960647583,
      "learning_rate": 6.726937269372694e-05,
      "loss": 0.01650850772857666,
      "step": 115360
    },
    {
      "epoch": 32.74765824581323,
      "grad_norm": 1.6761583089828491,
      "learning_rate": 6.726653420380358e-05,
      "loss": 0.0033865243196487425,
      "step": 115370
    },
    {
      "epoch": 32.75049673573659,
      "grad_norm": 0.7773210406303406,
      "learning_rate": 6.726369571388022e-05,
      "loss": 0.007043173909187317,
      "step": 115380
    },
    {
      "epoch": 32.753335225659946,
      "grad_norm": 4.391389846801758,
      "learning_rate": 6.726085722395686e-05,
      "loss": 0.005350909382104874,
      "step": 115390
    },
    {
      "epoch": 32.75617371558331,
      "grad_norm": 0.4465000331401825,
      "learning_rate": 6.72580187340335e-05,
      "loss": 0.0065145701169967655,
      "step": 115400
    },
    {
      "epoch": 32.75901220550667,
      "grad_norm": 7.537452697753906,
      "learning_rate": 6.725518024411014e-05,
      "loss": 0.0343333750963211,
      "step": 115410
    },
    {
      "epoch": 32.761850695430034,
      "grad_norm": 0.2141539305448532,
      "learning_rate": 6.725234175418677e-05,
      "loss": 0.010860981047153473,
      "step": 115420
    },
    {
      "epoch": 32.76468918535339,
      "grad_norm": 3.0888445377349854,
      "learning_rate": 6.724950326426341e-05,
      "loss": 0.006370636820793152,
      "step": 115430
    },
    {
      "epoch": 32.76752767527675,
      "grad_norm": 0.05383579060435295,
      "learning_rate": 6.724666477434005e-05,
      "loss": 0.008997134864330292,
      "step": 115440
    },
    {
      "epoch": 32.770366165200116,
      "grad_norm": 0.975656270980835,
      "learning_rate": 6.72438262844167e-05,
      "loss": 0.004427803680300713,
      "step": 115450
    },
    {
      "epoch": 32.77320465512347,
      "grad_norm": 0.31610989570617676,
      "learning_rate": 6.724098779449334e-05,
      "loss": 0.004494031518697738,
      "step": 115460
    },
    {
      "epoch": 32.776043145046835,
      "grad_norm": 0.03594844043254852,
      "learning_rate": 6.723814930456998e-05,
      "loss": 0.007528479397296906,
      "step": 115470
    },
    {
      "epoch": 32.7788816349702,
      "grad_norm": 0.27360719442367554,
      "learning_rate": 6.723531081464662e-05,
      "loss": 0.008582120388746261,
      "step": 115480
    },
    {
      "epoch": 32.78172012489355,
      "grad_norm": 2.762314796447754,
      "learning_rate": 6.723247232472325e-05,
      "loss": 0.02116602659225464,
      "step": 115490
    },
    {
      "epoch": 32.784558614816916,
      "grad_norm": 0.6268345713615417,
      "learning_rate": 6.722963383479989e-05,
      "loss": 0.0033383250236511232,
      "step": 115500
    },
    {
      "epoch": 32.784558614816916,
      "eval_accuracy": 0.9647103707000699,
      "eval_loss": 0.12168284505605698,
      "eval_runtime": 31.7737,
      "eval_samples_per_second": 494.969,
      "eval_steps_per_second": 7.742,
      "step": 115500
    },
    {
      "epoch": 32.78739710474028,
      "grad_norm": 1.470258116722107,
      "learning_rate": 6.722679534487653e-05,
      "loss": 0.009023544937372207,
      "step": 115510
    },
    {
      "epoch": 32.79023559466364,
      "grad_norm": 0.1403089314699173,
      "learning_rate": 6.722395685495316e-05,
      "loss": 0.005251587182283401,
      "step": 115520
    },
    {
      "epoch": 32.793074084587,
      "grad_norm": 0.5549132227897644,
      "learning_rate": 6.722111836502981e-05,
      "loss": 0.0010290227830410004,
      "step": 115530
    },
    {
      "epoch": 32.79591257451036,
      "grad_norm": 0.252884179353714,
      "learning_rate": 6.721827987510645e-05,
      "loss": 0.005143879354000092,
      "step": 115540
    },
    {
      "epoch": 32.79875106443372,
      "grad_norm": 2.968977451324463,
      "learning_rate": 6.721544138518308e-05,
      "loss": 0.0060463428497314455,
      "step": 115550
    },
    {
      "epoch": 32.80158955435708,
      "grad_norm": 1.8864237070083618,
      "learning_rate": 6.721260289525972e-05,
      "loss": 0.005672759190201759,
      "step": 115560
    },
    {
      "epoch": 32.80442804428044,
      "grad_norm": 2.8491411209106445,
      "learning_rate": 6.720976440533636e-05,
      "loss": 0.0023312605917453766,
      "step": 115570
    },
    {
      "epoch": 32.807266534203805,
      "grad_norm": 0.12990666925907135,
      "learning_rate": 6.7206925915413e-05,
      "loss": 0.00189936812967062,
      "step": 115580
    },
    {
      "epoch": 32.81010502412717,
      "grad_norm": 0.3765394985675812,
      "learning_rate": 6.720408742548965e-05,
      "loss": 0.005508154630661011,
      "step": 115590
    },
    {
      "epoch": 32.812943514050524,
      "grad_norm": 0.41737282276153564,
      "learning_rate": 6.720124893556629e-05,
      "loss": 0.00699404776096344,
      "step": 115600
    },
    {
      "epoch": 32.81578200397389,
      "grad_norm": 0.5483384132385254,
      "learning_rate": 6.719841044564292e-05,
      "loss": 0.009424223750829696,
      "step": 115610
    },
    {
      "epoch": 32.81862049389725,
      "grad_norm": 0.21697883307933807,
      "learning_rate": 6.719557195571956e-05,
      "loss": 0.003758159279823303,
      "step": 115620
    },
    {
      "epoch": 32.821458983820605,
      "grad_norm": 0.18853676319122314,
      "learning_rate": 6.71927334657962e-05,
      "loss": 0.0030297825112938883,
      "step": 115630
    },
    {
      "epoch": 32.82429747374397,
      "grad_norm": 0.42127612233161926,
      "learning_rate": 6.718989497587284e-05,
      "loss": 0.009587431699037552,
      "step": 115640
    },
    {
      "epoch": 32.82713596366733,
      "grad_norm": 1.9128867387771606,
      "learning_rate": 6.718705648594947e-05,
      "loss": 0.007943371683359146,
      "step": 115650
    },
    {
      "epoch": 32.82997445359069,
      "grad_norm": 1.639733076095581,
      "learning_rate": 6.718421799602612e-05,
      "loss": 0.014045782387256622,
      "step": 115660
    },
    {
      "epoch": 32.83281294351405,
      "grad_norm": 0.5386577844619751,
      "learning_rate": 6.718137950610276e-05,
      "loss": 0.003357705473899841,
      "step": 115670
    },
    {
      "epoch": 32.83565143343741,
      "grad_norm": 1.4894598722457886,
      "learning_rate": 6.717854101617939e-05,
      "loss": 0.005198423564434051,
      "step": 115680
    },
    {
      "epoch": 32.838489923360775,
      "grad_norm": 0.9534806609153748,
      "learning_rate": 6.717570252625603e-05,
      "loss": 0.013977539539337159,
      "step": 115690
    },
    {
      "epoch": 32.84132841328413,
      "grad_norm": 0.915469765663147,
      "learning_rate": 6.717286403633268e-05,
      "loss": 0.008945496380329132,
      "step": 115700
    },
    {
      "epoch": 32.844166903207494,
      "grad_norm": 0.45460212230682373,
      "learning_rate": 6.71700255464093e-05,
      "loss": 0.010270661860704421,
      "step": 115710
    },
    {
      "epoch": 32.84700539313086,
      "grad_norm": 11.189266204833984,
      "learning_rate": 6.716718705648594e-05,
      "loss": 0.02140667140483856,
      "step": 115720
    },
    {
      "epoch": 32.84984388305421,
      "grad_norm": 0.3372981548309326,
      "learning_rate": 6.71643485665626e-05,
      "loss": 0.02159825712442398,
      "step": 115730
    },
    {
      "epoch": 32.852682372977576,
      "grad_norm": 0.22836704552173615,
      "learning_rate": 6.716151007663923e-05,
      "loss": 0.009089440852403641,
      "step": 115740
    },
    {
      "epoch": 32.85552086290094,
      "grad_norm": 0.3199300169944763,
      "learning_rate": 6.715867158671587e-05,
      "loss": 0.013751201331615448,
      "step": 115750
    },
    {
      "epoch": 32.858359352824294,
      "grad_norm": 0.32413825392723083,
      "learning_rate": 6.715583309679251e-05,
      "loss": 0.012880058586597442,
      "step": 115760
    },
    {
      "epoch": 32.86119784274766,
      "grad_norm": 10.195226669311523,
      "learning_rate": 6.715299460686915e-05,
      "loss": 0.0071233630180358885,
      "step": 115770
    },
    {
      "epoch": 32.86403633267102,
      "grad_norm": 15.308408737182617,
      "learning_rate": 6.715015611694578e-05,
      "loss": 0.009597312659025192,
      "step": 115780
    },
    {
      "epoch": 32.86687482259438,
      "grad_norm": 1.6286441087722778,
      "learning_rate": 6.714731762702243e-05,
      "loss": 0.004611574113368988,
      "step": 115790
    },
    {
      "epoch": 32.86971331251774,
      "grad_norm": 1.9988278150558472,
      "learning_rate": 6.714447913709908e-05,
      "loss": 0.02878168523311615,
      "step": 115800
    },
    {
      "epoch": 32.8725518024411,
      "grad_norm": 0.016223937273025513,
      "learning_rate": 6.71416406471757e-05,
      "loss": 0.017253577709197998,
      "step": 115810
    },
    {
      "epoch": 32.875390292364465,
      "grad_norm": 0.15361864864826202,
      "learning_rate": 6.713880215725234e-05,
      "loss": 0.005740371346473694,
      "step": 115820
    },
    {
      "epoch": 32.87822878228782,
      "grad_norm": 0.810506284236908,
      "learning_rate": 6.713596366732899e-05,
      "loss": 0.00804566890001297,
      "step": 115830
    },
    {
      "epoch": 32.88106727221118,
      "grad_norm": 12.61858081817627,
      "learning_rate": 6.713312517740561e-05,
      "loss": 0.030058106780052184,
      "step": 115840
    },
    {
      "epoch": 32.883905762134546,
      "grad_norm": 11.883477210998535,
      "learning_rate": 6.713028668748226e-05,
      "loss": 0.0060219340026378635,
      "step": 115850
    },
    {
      "epoch": 32.8867442520579,
      "grad_norm": 9.442068099975586,
      "learning_rate": 6.712744819755891e-05,
      "loss": 0.007366572320461273,
      "step": 115860
    },
    {
      "epoch": 32.889582741981265,
      "grad_norm": 3.7921175956726074,
      "learning_rate": 6.712460970763554e-05,
      "loss": 0.0037923097610473634,
      "step": 115870
    },
    {
      "epoch": 32.89242123190463,
      "grad_norm": 2.9303460121154785,
      "learning_rate": 6.712177121771218e-05,
      "loss": 0.005532452464103698,
      "step": 115880
    },
    {
      "epoch": 32.89525972182799,
      "grad_norm": 0.18757256865501404,
      "learning_rate": 6.711893272778882e-05,
      "loss": 0.004501210898160935,
      "step": 115890
    },
    {
      "epoch": 32.898098211751346,
      "grad_norm": 5.959294319152832,
      "learning_rate": 6.711609423786546e-05,
      "loss": 0.011784429103136063,
      "step": 115900
    },
    {
      "epoch": 32.90093670167471,
      "grad_norm": 0.16381728649139404,
      "learning_rate": 6.711325574794209e-05,
      "loss": 0.004614159464836121,
      "step": 115910
    },
    {
      "epoch": 32.90377519159807,
      "grad_norm": 14.170340538024902,
      "learning_rate": 6.711041725801873e-05,
      "loss": 0.010023683309555054,
      "step": 115920
    },
    {
      "epoch": 32.90661368152143,
      "grad_norm": 0.35233521461486816,
      "learning_rate": 6.710757876809539e-05,
      "loss": 0.00986594781279564,
      "step": 115930
    },
    {
      "epoch": 32.90945217144479,
      "grad_norm": 9.273916244506836,
      "learning_rate": 6.710474027817201e-05,
      "loss": 0.010228903591632843,
      "step": 115940
    },
    {
      "epoch": 32.912290661368154,
      "grad_norm": 0.0840216651558876,
      "learning_rate": 6.710190178824866e-05,
      "loss": 0.01614183038473129,
      "step": 115950
    },
    {
      "epoch": 32.91512915129151,
      "grad_norm": 0.9385865330696106,
      "learning_rate": 6.70990632983253e-05,
      "loss": 0.0143500417470932,
      "step": 115960
    },
    {
      "epoch": 32.91796764121487,
      "grad_norm": 0.17552220821380615,
      "learning_rate": 6.709622480840192e-05,
      "loss": 0.0020470311865210533,
      "step": 115970
    },
    {
      "epoch": 32.920806131138235,
      "grad_norm": 0.12615561485290527,
      "learning_rate": 6.709338631847857e-05,
      "loss": 0.01490943431854248,
      "step": 115980
    },
    {
      "epoch": 32.9236446210616,
      "grad_norm": 3.789140224456787,
      "learning_rate": 6.709054782855522e-05,
      "loss": 0.005563593283295631,
      "step": 115990
    },
    {
      "epoch": 32.926483110984954,
      "grad_norm": 3.951472520828247,
      "learning_rate": 6.708770933863185e-05,
      "loss": 0.01912008970975876,
      "step": 116000
    },
    {
      "epoch": 32.926483110984954,
      "eval_accuracy": 0.9711324473834806,
      "eval_loss": 0.10250042378902435,
      "eval_runtime": 31.8767,
      "eval_samples_per_second": 493.369,
      "eval_steps_per_second": 7.717,
      "step": 116000
    },
    {
      "epoch": 32.92932160090832,
      "grad_norm": 10.958779335021973,
      "learning_rate": 6.708487084870849e-05,
      "loss": 0.03415845632553101,
      "step": 116010
    },
    {
      "epoch": 32.93216009083168,
      "grad_norm": 8.922451972961426,
      "learning_rate": 6.708203235878513e-05,
      "loss": 0.017027565836906434,
      "step": 116020
    },
    {
      "epoch": 32.934998580755035,
      "grad_norm": 1.252122402191162,
      "learning_rate": 6.707919386886177e-05,
      "loss": 0.013306140899658203,
      "step": 116030
    },
    {
      "epoch": 32.9378370706784,
      "grad_norm": 0.48863062262535095,
      "learning_rate": 6.70763553789384e-05,
      "loss": 0.006216781213879585,
      "step": 116040
    },
    {
      "epoch": 32.94067556060176,
      "grad_norm": 1.121009111404419,
      "learning_rate": 6.707351688901504e-05,
      "loss": 0.010345569252967835,
      "step": 116050
    },
    {
      "epoch": 32.943514050525124,
      "grad_norm": 7.219755172729492,
      "learning_rate": 6.70706783990917e-05,
      "loss": 0.011841758340597152,
      "step": 116060
    },
    {
      "epoch": 32.94635254044848,
      "grad_norm": 0.11633042246103287,
      "learning_rate": 6.706783990916832e-05,
      "loss": 0.0032011404633522033,
      "step": 116070
    },
    {
      "epoch": 32.94919103037184,
      "grad_norm": 0.7642565965652466,
      "learning_rate": 6.706500141924497e-05,
      "loss": 0.007887057960033417,
      "step": 116080
    },
    {
      "epoch": 32.952029520295206,
      "grad_norm": 0.4195108711719513,
      "learning_rate": 6.706216292932161e-05,
      "loss": 0.003325618803501129,
      "step": 116090
    },
    {
      "epoch": 32.95486801021856,
      "grad_norm": 0.1415892243385315,
      "learning_rate": 6.705932443939824e-05,
      "loss": 0.010518977791070938,
      "step": 116100
    },
    {
      "epoch": 32.957706500141924,
      "grad_norm": 0.05278736725449562,
      "learning_rate": 6.705648594947488e-05,
      "loss": 0.0027638593688607215,
      "step": 116110
    },
    {
      "epoch": 32.96054499006529,
      "grad_norm": 0.5976791381835938,
      "learning_rate": 6.705364745955152e-05,
      "loss": 0.0037149250507354735,
      "step": 116120
    },
    {
      "epoch": 32.96338347998864,
      "grad_norm": 0.13159441947937012,
      "learning_rate": 6.705080896962816e-05,
      "loss": 0.01016145497560501,
      "step": 116130
    },
    {
      "epoch": 32.966221969912006,
      "grad_norm": 0.9723941683769226,
      "learning_rate": 6.70479704797048e-05,
      "loss": 0.0015779921784996986,
      "step": 116140
    },
    {
      "epoch": 32.96906045983537,
      "grad_norm": 9.23215103149414,
      "learning_rate": 6.704513198978144e-05,
      "loss": 0.00601348951458931,
      "step": 116150
    },
    {
      "epoch": 32.97189894975873,
      "grad_norm": 0.2786255478858948,
      "learning_rate": 6.704229349985808e-05,
      "loss": 0.0034077394753694533,
      "step": 116160
    },
    {
      "epoch": 32.97473743968209,
      "grad_norm": 1.6322020292282104,
      "learning_rate": 6.703945500993471e-05,
      "loss": 0.004880034178495407,
      "step": 116170
    },
    {
      "epoch": 32.97757592960545,
      "grad_norm": 0.08954855054616928,
      "learning_rate": 6.703661652001135e-05,
      "loss": 0.003108261153101921,
      "step": 116180
    },
    {
      "epoch": 32.98041441952881,
      "grad_norm": 1.7511281967163086,
      "learning_rate": 6.703377803008801e-05,
      "loss": 0.0126246839761734,
      "step": 116190
    },
    {
      "epoch": 32.98325290945217,
      "grad_norm": 0.8042444586753845,
      "learning_rate": 6.703093954016464e-05,
      "loss": 0.0025573816150426865,
      "step": 116200
    },
    {
      "epoch": 32.98609139937553,
      "grad_norm": 0.40912291407585144,
      "learning_rate": 6.702810105024128e-05,
      "loss": 0.007548514008522034,
      "step": 116210
    },
    {
      "epoch": 32.988929889298895,
      "grad_norm": 1.2805516719818115,
      "learning_rate": 6.702526256031792e-05,
      "loss": 0.024196968972682954,
      "step": 116220
    },
    {
      "epoch": 32.99176837922225,
      "grad_norm": 0.6784161329269409,
      "learning_rate": 6.702242407039455e-05,
      "loss": 0.009925782680511475,
      "step": 116230
    },
    {
      "epoch": 32.99460686914561,
      "grad_norm": 6.0436625480651855,
      "learning_rate": 6.701958558047119e-05,
      "loss": 0.009577143192291259,
      "step": 116240
    },
    {
      "epoch": 32.997445359068976,
      "grad_norm": 9.224055290222168,
      "learning_rate": 6.701674709054783e-05,
      "loss": 0.006043463200330734,
      "step": 116250
    },
    {
      "epoch": 33.00028384899234,
      "grad_norm": 12.260200500488281,
      "learning_rate": 6.701390860062447e-05,
      "loss": 0.009040240198373795,
      "step": 116260
    },
    {
      "epoch": 33.003122338915695,
      "grad_norm": 0.10167325288057327,
      "learning_rate": 6.701107011070111e-05,
      "loss": 0.0020274695008993147,
      "step": 116270
    },
    {
      "epoch": 33.00596082883906,
      "grad_norm": 0.9623202085494995,
      "learning_rate": 6.700823162077775e-05,
      "loss": 0.008140146732330322,
      "step": 116280
    },
    {
      "epoch": 33.00879931876242,
      "grad_norm": 0.6246654987335205,
      "learning_rate": 6.70053931308544e-05,
      "loss": 0.010078358650207519,
      "step": 116290
    },
    {
      "epoch": 33.01163780868578,
      "grad_norm": 0.2700178027153015,
      "learning_rate": 6.700255464093102e-05,
      "loss": 0.010890664160251617,
      "step": 116300
    },
    {
      "epoch": 33.01447629860914,
      "grad_norm": 2.104504346847534,
      "learning_rate": 6.699971615100766e-05,
      "loss": 0.002321568131446838,
      "step": 116310
    },
    {
      "epoch": 33.0173147885325,
      "grad_norm": 7.427857398986816,
      "learning_rate": 6.699687766108432e-05,
      "loss": 0.0026299754157662393,
      "step": 116320
    },
    {
      "epoch": 33.02015327845586,
      "grad_norm": 0.5689986348152161,
      "learning_rate": 6.699403917116095e-05,
      "loss": 0.001200598292052746,
      "step": 116330
    },
    {
      "epoch": 33.02299176837922,
      "grad_norm": 0.6372933983802795,
      "learning_rate": 6.699120068123759e-05,
      "loss": 0.0012028954923152923,
      "step": 116340
    },
    {
      "epoch": 33.025830258302584,
      "grad_norm": 1.522961974143982,
      "learning_rate": 6.698836219131423e-05,
      "loss": 0.005192311853170395,
      "step": 116350
    },
    {
      "epoch": 33.02866874822595,
      "grad_norm": 0.04061300307512283,
      "learning_rate": 6.698552370139086e-05,
      "loss": 0.0055526383221149445,
      "step": 116360
    },
    {
      "epoch": 33.0315072381493,
      "grad_norm": 4.503653526306152,
      "learning_rate": 6.69826852114675e-05,
      "loss": 0.006962700188159943,
      "step": 116370
    },
    {
      "epoch": 33.034345728072665,
      "grad_norm": 0.3460148870944977,
      "learning_rate": 6.697984672154414e-05,
      "loss": 0.006532601267099381,
      "step": 116380
    },
    {
      "epoch": 33.03718421799603,
      "grad_norm": 0.06929676234722137,
      "learning_rate": 6.697700823162078e-05,
      "loss": 0.0016426945105195046,
      "step": 116390
    },
    {
      "epoch": 33.040022707919384,
      "grad_norm": 9.749085426330566,
      "learning_rate": 6.697416974169742e-05,
      "loss": 0.00736527219414711,
      "step": 116400
    },
    {
      "epoch": 33.04286119784275,
      "grad_norm": 2.1877694129943848,
      "learning_rate": 6.697133125177406e-05,
      "loss": 0.009657806903123855,
      "step": 116410
    },
    {
      "epoch": 33.04569968776611,
      "grad_norm": 1.5757001638412476,
      "learning_rate": 6.69684927618507e-05,
      "loss": 0.01277148723602295,
      "step": 116420
    },
    {
      "epoch": 33.04853817768947,
      "grad_norm": 0.596792459487915,
      "learning_rate": 6.696565427192733e-05,
      "loss": 0.011281078308820724,
      "step": 116430
    },
    {
      "epoch": 33.05137666761283,
      "grad_norm": 0.9761044979095459,
      "learning_rate": 6.696281578200397e-05,
      "loss": 0.002975606918334961,
      "step": 116440
    },
    {
      "epoch": 33.05421515753619,
      "grad_norm": 1.9367868900299072,
      "learning_rate": 6.695997729208062e-05,
      "loss": 0.004025884717702865,
      "step": 116450
    },
    {
      "epoch": 33.057053647459554,
      "grad_norm": 1.3975588083267212,
      "learning_rate": 6.695713880215726e-05,
      "loss": 0.005501432716846466,
      "step": 116460
    },
    {
      "epoch": 33.05989213738291,
      "grad_norm": 3.3480329513549805,
      "learning_rate": 6.69543003122339e-05,
      "loss": 0.01589973121881485,
      "step": 116470
    },
    {
      "epoch": 33.06273062730627,
      "grad_norm": 0.6771692633628845,
      "learning_rate": 6.695146182231054e-05,
      "loss": 0.007020026445388794,
      "step": 116480
    },
    {
      "epoch": 33.065569117229636,
      "grad_norm": 0.5640496015548706,
      "learning_rate": 6.694862333238717e-05,
      "loss": 0.01740611791610718,
      "step": 116490
    },
    {
      "epoch": 33.06840760715299,
      "grad_norm": 0.7360278367996216,
      "learning_rate": 6.694578484246381e-05,
      "loss": 0.002274811826646328,
      "step": 116500
    },
    {
      "epoch": 33.06840760715299,
      "eval_accuracy": 0.9696699942773574,
      "eval_loss": 0.10040533542633057,
      "eval_runtime": 32.0197,
      "eval_samples_per_second": 491.166,
      "eval_steps_per_second": 7.683,
      "step": 116500
    },
    {
      "epoch": 33.071246097076354,
      "grad_norm": 0.07331196218729019,
      "learning_rate": 6.694294635254045e-05,
      "loss": 0.0054709076881408695,
      "step": 116510
    },
    {
      "epoch": 33.07408458699972,
      "grad_norm": 0.5650719404220581,
      "learning_rate": 6.694010786261709e-05,
      "loss": 0.002010716684162617,
      "step": 116520
    },
    {
      "epoch": 33.07692307692308,
      "grad_norm": 0.06461764127016068,
      "learning_rate": 6.693726937269373e-05,
      "loss": 0.0028999261558055878,
      "step": 116530
    },
    {
      "epoch": 33.079761566846436,
      "grad_norm": 4.413096904754639,
      "learning_rate": 6.693443088277037e-05,
      "loss": 0.029185721278190614,
      "step": 116540
    },
    {
      "epoch": 33.0826000567698,
      "grad_norm": 2.3148601055145264,
      "learning_rate": 6.6931592392847e-05,
      "loss": 0.008036541938781738,
      "step": 116550
    },
    {
      "epoch": 33.08543854669316,
      "grad_norm": 0.23697713017463684,
      "learning_rate": 6.692875390292364e-05,
      "loss": 0.0027852648869156837,
      "step": 116560
    },
    {
      "epoch": 33.08827703661652,
      "grad_norm": 10.562323570251465,
      "learning_rate": 6.692591541300029e-05,
      "loss": 0.0027911074459552763,
      "step": 116570
    },
    {
      "epoch": 33.09111552653988,
      "grad_norm": 0.9765073657035828,
      "learning_rate": 6.692307692307693e-05,
      "loss": 0.009827610105276108,
      "step": 116580
    },
    {
      "epoch": 33.09395401646324,
      "grad_norm": 0.493289053440094,
      "learning_rate": 6.692023843315357e-05,
      "loss": 0.029821884632110596,
      "step": 116590
    },
    {
      "epoch": 33.0967925063866,
      "grad_norm": 3.4773964881896973,
      "learning_rate": 6.691739994323021e-05,
      "loss": 0.0027792293578386305,
      "step": 116600
    },
    {
      "epoch": 33.09963099630996,
      "grad_norm": 3.2410078048706055,
      "learning_rate": 6.691456145330685e-05,
      "loss": 0.005583294481039047,
      "step": 116610
    },
    {
      "epoch": 33.102469486233325,
      "grad_norm": 0.11262920498847961,
      "learning_rate": 6.691172296338348e-05,
      "loss": 0.006002588570117951,
      "step": 116620
    },
    {
      "epoch": 33.10530797615669,
      "grad_norm": 0.17572687566280365,
      "learning_rate": 6.690888447346012e-05,
      "loss": 0.008459733426570892,
      "step": 116630
    },
    {
      "epoch": 33.10814646608004,
      "grad_norm": 0.7166721224784851,
      "learning_rate": 6.690604598353676e-05,
      "loss": 0.011703053861856461,
      "step": 116640
    },
    {
      "epoch": 33.110984956003406,
      "grad_norm": 2.6370303630828857,
      "learning_rate": 6.690320749361339e-05,
      "loss": 0.007249914109706879,
      "step": 116650
    },
    {
      "epoch": 33.11382344592677,
      "grad_norm": 12.838744163513184,
      "learning_rate": 6.690065285268238e-05,
      "loss": 0.023855648934841156,
      "step": 116660
    },
    {
      "epoch": 33.116661935850125,
      "grad_norm": 2.6808087825775146,
      "learning_rate": 6.689781436275901e-05,
      "loss": 0.002728269062936306,
      "step": 116670
    },
    {
      "epoch": 33.11950042577349,
      "grad_norm": 1.0343464612960815,
      "learning_rate": 6.689497587283565e-05,
      "loss": 0.008923836797475816,
      "step": 116680
    },
    {
      "epoch": 33.12233891569685,
      "grad_norm": 6.45535135269165,
      "learning_rate": 6.689213738291229e-05,
      "loss": 0.008817353844642639,
      "step": 116690
    },
    {
      "epoch": 33.12517740562021,
      "grad_norm": 0.08838874846696854,
      "learning_rate": 6.688929889298893e-05,
      "loss": 0.008095046877861023,
      "step": 116700
    },
    {
      "epoch": 33.12801589554357,
      "grad_norm": 0.2174573540687561,
      "learning_rate": 6.688646040306558e-05,
      "loss": 0.0015282340347766877,
      "step": 116710
    },
    {
      "epoch": 33.13085438546693,
      "grad_norm": 7.259507179260254,
      "learning_rate": 6.688362191314222e-05,
      "loss": 0.003509655222296715,
      "step": 116720
    },
    {
      "epoch": 33.133692875390295,
      "grad_norm": 0.03537416085600853,
      "learning_rate": 6.688078342321885e-05,
      "loss": 0.0032778024673461915,
      "step": 116730
    },
    {
      "epoch": 33.13653136531365,
      "grad_norm": 0.6632115840911865,
      "learning_rate": 6.687794493329549e-05,
      "loss": 0.0017957322299480439,
      "step": 116740
    },
    {
      "epoch": 33.139369855237014,
      "grad_norm": 0.3281362056732178,
      "learning_rate": 6.687510644337213e-05,
      "loss": 0.005974175781011582,
      "step": 116750
    },
    {
      "epoch": 33.14220834516038,
      "grad_norm": 0.11487170308828354,
      "learning_rate": 6.687226795344877e-05,
      "loss": 0.0028112603351473807,
      "step": 116760
    },
    {
      "epoch": 33.14504683508373,
      "grad_norm": 2.022303581237793,
      "learning_rate": 6.686942946352541e-05,
      "loss": 0.0074214845895767215,
      "step": 116770
    },
    {
      "epoch": 33.147885325007096,
      "grad_norm": 0.12305603176355362,
      "learning_rate": 6.686659097360205e-05,
      "loss": 0.005791975557804108,
      "step": 116780
    },
    {
      "epoch": 33.15072381493046,
      "grad_norm": 0.21567302942276,
      "learning_rate": 6.68637524836787e-05,
      "loss": 0.0017539359629154204,
      "step": 116790
    },
    {
      "epoch": 33.15356230485382,
      "grad_norm": 0.10450593382120132,
      "learning_rate": 6.686091399375532e-05,
      "loss": 0.0034534059464931487,
      "step": 116800
    },
    {
      "epoch": 33.15640079477718,
      "grad_norm": 0.20200063288211823,
      "learning_rate": 6.685807550383196e-05,
      "loss": 0.003671243041753769,
      "step": 116810
    },
    {
      "epoch": 33.15923928470054,
      "grad_norm": 0.20154725015163422,
      "learning_rate": 6.68552370139086e-05,
      "loss": 0.005336741358041764,
      "step": 116820
    },
    {
      "epoch": 33.1620777746239,
      "grad_norm": 3.94284987449646,
      "learning_rate": 6.685239852398523e-05,
      "loss": 0.007192402333021164,
      "step": 116830
    },
    {
      "epoch": 33.16491626454726,
      "grad_norm": 0.7578408718109131,
      "learning_rate": 6.684956003406189e-05,
      "loss": 0.0037848081439733504,
      "step": 116840
    },
    {
      "epoch": 33.16775475447062,
      "grad_norm": 1.526747226715088,
      "learning_rate": 6.684672154413853e-05,
      "loss": 0.0022581979632377625,
      "step": 116850
    },
    {
      "epoch": 33.170593244393984,
      "grad_norm": 0.3071438670158386,
      "learning_rate": 6.684388305421516e-05,
      "loss": 0.0032968707382678986,
      "step": 116860
    },
    {
      "epoch": 33.17343173431734,
      "grad_norm": 12.856816291809082,
      "learning_rate": 6.68410445642918e-05,
      "loss": 0.009627282619476318,
      "step": 116870
    },
    {
      "epoch": 33.1762702242407,
      "grad_norm": 0.07208612561225891,
      "learning_rate": 6.683820607436844e-05,
      "loss": 0.0008730346336960793,
      "step": 116880
    },
    {
      "epoch": 33.179108714164066,
      "grad_norm": 0.8568087816238403,
      "learning_rate": 6.683536758444508e-05,
      "loss": 0.005911839753389358,
      "step": 116890
    },
    {
      "epoch": 33.18194720408743,
      "grad_norm": 0.3375151753425598,
      "learning_rate": 6.683252909452172e-05,
      "loss": 0.0071736820042133335,
      "step": 116900
    },
    {
      "epoch": 33.184785694010785,
      "grad_norm": 0.8520231246948242,
      "learning_rate": 6.682969060459836e-05,
      "loss": 0.0014691008254885674,
      "step": 116910
    },
    {
      "epoch": 33.18762418393415,
      "grad_norm": 0.49842408299446106,
      "learning_rate": 6.6826852114675e-05,
      "loss": 0.004839428514242172,
      "step": 116920
    },
    {
      "epoch": 33.19046267385751,
      "grad_norm": 8.875041007995605,
      "learning_rate": 6.682401362475163e-05,
      "loss": 0.012935957312583924,
      "step": 116930
    },
    {
      "epoch": 33.193301163780866,
      "grad_norm": 0.18533369898796082,
      "learning_rate": 6.682117513482827e-05,
      "loss": 0.002854626253247261,
      "step": 116940
    },
    {
      "epoch": 33.19613965370423,
      "grad_norm": 0.6381564140319824,
      "learning_rate": 6.681833664490491e-05,
      "loss": 0.005965876579284668,
      "step": 116950
    },
    {
      "epoch": 33.19897814362759,
      "grad_norm": 0.7147642374038696,
      "learning_rate": 6.681549815498154e-05,
      "loss": 0.004762191697955132,
      "step": 116960
    },
    {
      "epoch": 33.20181663355095,
      "grad_norm": 3.5744431018829346,
      "learning_rate": 6.68126596650582e-05,
      "loss": 0.0021830499172210693,
      "step": 116970
    },
    {
      "epoch": 33.20465512347431,
      "grad_norm": 1.0354992151260376,
      "learning_rate": 6.680982117513484e-05,
      "loss": 0.003933040797710419,
      "step": 116980
    },
    {
      "epoch": 33.20749361339767,
      "grad_norm": 5.887685775756836,
      "learning_rate": 6.680698268521147e-05,
      "loss": 0.009693623334169389,
      "step": 116990
    },
    {
      "epoch": 33.210332103321036,
      "grad_norm": 0.5320745706558228,
      "learning_rate": 6.680414419528811e-05,
      "loss": 0.006465865671634674,
      "step": 117000
    },
    {
      "epoch": 33.210332103321036,
      "eval_accuracy": 0.9697335791950149,
      "eval_loss": 0.10549739748239517,
      "eval_runtime": 31.8203,
      "eval_samples_per_second": 494.244,
      "eval_steps_per_second": 7.731,
      "step": 117000
    },
    {
      "epoch": 33.21317059324439,
      "grad_norm": 15.010702133178711,
      "learning_rate": 6.680130570536475e-05,
      "loss": 0.011933570355176925,
      "step": 117010
    },
    {
      "epoch": 33.216009083167755,
      "grad_norm": 0.10318238288164139,
      "learning_rate": 6.679846721544139e-05,
      "loss": 0.002451440691947937,
      "step": 117020
    },
    {
      "epoch": 33.21884757309112,
      "grad_norm": 0.4122534394264221,
      "learning_rate": 6.679562872551802e-05,
      "loss": 0.005117706209421158,
      "step": 117030
    },
    {
      "epoch": 33.221686063014474,
      "grad_norm": 0.19807782769203186,
      "learning_rate": 6.679279023559467e-05,
      "loss": 0.0018423140048980714,
      "step": 117040
    },
    {
      "epoch": 33.22452455293784,
      "grad_norm": 3.906538963317871,
      "learning_rate": 6.678995174567131e-05,
      "loss": 0.0032202389091253282,
      "step": 117050
    },
    {
      "epoch": 33.2273630428612,
      "grad_norm": 3.8841419219970703,
      "learning_rate": 6.678711325574794e-05,
      "loss": 0.0033941879868507386,
      "step": 117060
    },
    {
      "epoch": 33.230201532784555,
      "grad_norm": 1.8240785598754883,
      "learning_rate": 6.678427476582458e-05,
      "loss": 0.011822700500488281,
      "step": 117070
    },
    {
      "epoch": 33.23304002270792,
      "grad_norm": 3.9706709384918213,
      "learning_rate": 6.678143627590123e-05,
      "loss": 0.007514472305774689,
      "step": 117080
    },
    {
      "epoch": 33.23587851263128,
      "grad_norm": 3.4607510566711426,
      "learning_rate": 6.677859778597785e-05,
      "loss": 0.011683598905801774,
      "step": 117090
    },
    {
      "epoch": 33.238717002554644,
      "grad_norm": 8.42945671081543,
      "learning_rate": 6.677575929605451e-05,
      "loss": 0.004024039581418037,
      "step": 117100
    },
    {
      "epoch": 33.241555492478,
      "grad_norm": 0.8852787613868713,
      "learning_rate": 6.677292080613115e-05,
      "loss": 0.013066606223583221,
      "step": 117110
    },
    {
      "epoch": 33.24439398240136,
      "grad_norm": 14.635392189025879,
      "learning_rate": 6.677008231620778e-05,
      "loss": 0.00991058349609375,
      "step": 117120
    },
    {
      "epoch": 33.247232472324725,
      "grad_norm": 0.1147436648607254,
      "learning_rate": 6.676724382628442e-05,
      "loss": 0.003976458311080932,
      "step": 117130
    },
    {
      "epoch": 33.25007096224808,
      "grad_norm": 0.5611915588378906,
      "learning_rate": 6.676440533636106e-05,
      "loss": 0.007162712514400482,
      "step": 117140
    },
    {
      "epoch": 33.252909452171444,
      "grad_norm": 2.005566358566284,
      "learning_rate": 6.67615668464377e-05,
      "loss": 0.0027954330667853355,
      "step": 117150
    },
    {
      "epoch": 33.25574794209481,
      "grad_norm": 4.083441734313965,
      "learning_rate": 6.675872835651433e-05,
      "loss": 0.0035261183977127073,
      "step": 117160
    },
    {
      "epoch": 33.25858643201816,
      "grad_norm": 2.79569673538208,
      "learning_rate": 6.675588986659098e-05,
      "loss": 0.004823222756385803,
      "step": 117170
    },
    {
      "epoch": 33.261424921941526,
      "grad_norm": 0.2593329846858978,
      "learning_rate": 6.675305137666763e-05,
      "loss": 0.005334918200969696,
      "step": 117180
    },
    {
      "epoch": 33.26426341186489,
      "grad_norm": 0.19562114775180817,
      "learning_rate": 6.675021288674425e-05,
      "loss": 0.002492145448923111,
      "step": 117190
    },
    {
      "epoch": 33.26710190178825,
      "grad_norm": 0.03985077515244484,
      "learning_rate": 6.67473743968209e-05,
      "loss": 0.003584921360015869,
      "step": 117200
    },
    {
      "epoch": 33.26994039171161,
      "grad_norm": 0.5491010546684265,
      "learning_rate": 6.674453590689754e-05,
      "loss": 0.008333267271518707,
      "step": 117210
    },
    {
      "epoch": 33.27277888163497,
      "grad_norm": 1.1943808794021606,
      "learning_rate": 6.674169741697416e-05,
      "loss": 0.004083572328090668,
      "step": 117220
    },
    {
      "epoch": 33.27561737155833,
      "grad_norm": 0.14809031784534454,
      "learning_rate": 6.67388589270508e-05,
      "loss": 0.001823628693819046,
      "step": 117230
    },
    {
      "epoch": 33.27845586148169,
      "grad_norm": 1.3379476070404053,
      "learning_rate": 6.673602043712746e-05,
      "loss": 0.010470181703567505,
      "step": 117240
    },
    {
      "epoch": 33.28129435140505,
      "grad_norm": 0.4275990128517151,
      "learning_rate": 6.673318194720409e-05,
      "loss": 0.0018964186310768127,
      "step": 117250
    },
    {
      "epoch": 33.284132841328415,
      "grad_norm": 0.5102622509002686,
      "learning_rate": 6.673034345728073e-05,
      "loss": 0.009090592712163925,
      "step": 117260
    },
    {
      "epoch": 33.28697133125178,
      "grad_norm": 1.5648858547210693,
      "learning_rate": 6.672750496735737e-05,
      "loss": 0.0017020264640450478,
      "step": 117270
    },
    {
      "epoch": 33.28980982117513,
      "grad_norm": 1.519100308418274,
      "learning_rate": 6.672466647743401e-05,
      "loss": 0.0022423505783081056,
      "step": 117280
    },
    {
      "epoch": 33.292648311098496,
      "grad_norm": 1.1540566682815552,
      "learning_rate": 6.672182798751064e-05,
      "loss": 0.0035326041281223295,
      "step": 117290
    },
    {
      "epoch": 33.29548680102186,
      "grad_norm": 9.145088195800781,
      "learning_rate": 6.67189894975873e-05,
      "loss": 0.007082366943359375,
      "step": 117300
    },
    {
      "epoch": 33.298325290945215,
      "grad_norm": 2.6998770236968994,
      "learning_rate": 6.671615100766392e-05,
      "loss": 0.006271549314260483,
      "step": 117310
    },
    {
      "epoch": 33.30116378086858,
      "grad_norm": 0.6100093722343445,
      "learning_rate": 6.671331251774056e-05,
      "loss": 0.004267510771751404,
      "step": 117320
    },
    {
      "epoch": 33.30400227079194,
      "grad_norm": 0.9361395835876465,
      "learning_rate": 6.67104740278172e-05,
      "loss": 0.003236216306686401,
      "step": 117330
    },
    {
      "epoch": 33.306840760715296,
      "grad_norm": 0.11130796372890472,
      "learning_rate": 6.670763553789385e-05,
      "loss": 0.0078055053949356076,
      "step": 117340
    },
    {
      "epoch": 33.30967925063866,
      "grad_norm": 0.9566770792007446,
      "learning_rate": 6.670479704797047e-05,
      "loss": 0.003107859753072262,
      "step": 117350
    },
    {
      "epoch": 33.31251774056202,
      "grad_norm": 0.27331191301345825,
      "learning_rate": 6.670195855804712e-05,
      "loss": 0.004045741260051727,
      "step": 117360
    },
    {
      "epoch": 33.315356230485385,
      "grad_norm": 3.211246967315674,
      "learning_rate": 6.669912006812377e-05,
      "loss": 0.006707516312599182,
      "step": 117370
    },
    {
      "epoch": 33.31819472040874,
      "grad_norm": 1.721397876739502,
      "learning_rate": 6.66962815782004e-05,
      "loss": 0.004508708417415619,
      "step": 117380
    },
    {
      "epoch": 33.321033210332104,
      "grad_norm": 5.4185051918029785,
      "learning_rate": 6.669344308827704e-05,
      "loss": 0.010296161472797393,
      "step": 117390
    },
    {
      "epoch": 33.32387170025547,
      "grad_norm": 0.3519374132156372,
      "learning_rate": 6.669060459835368e-05,
      "loss": 0.0030946435406804087,
      "step": 117400
    },
    {
      "epoch": 33.32671019017882,
      "grad_norm": 0.49655184149742126,
      "learning_rate": 6.668776610843031e-05,
      "loss": 0.003404499590396881,
      "step": 117410
    },
    {
      "epoch": 33.329548680102185,
      "grad_norm": 0.4788326323032379,
      "learning_rate": 6.668492761850695e-05,
      "loss": 0.004749372228980064,
      "step": 117420
    },
    {
      "epoch": 33.33238717002555,
      "grad_norm": 0.4188593626022339,
      "learning_rate": 6.668208912858359e-05,
      "loss": 0.003109121322631836,
      "step": 117430
    },
    {
      "epoch": 33.335225659948904,
      "grad_norm": 0.3335094153881073,
      "learning_rate": 6.667925063866023e-05,
      "loss": 0.0029158854857087137,
      "step": 117440
    },
    {
      "epoch": 33.33806414987227,
      "grad_norm": 0.9115496873855591,
      "learning_rate": 6.667641214873688e-05,
      "loss": 0.006710460782051087,
      "step": 117450
    },
    {
      "epoch": 33.34090263979563,
      "grad_norm": 0.8134279251098633,
      "learning_rate": 6.667357365881352e-05,
      "loss": 0.010479222983121872,
      "step": 117460
    },
    {
      "epoch": 33.34374112971899,
      "grad_norm": 0.3062744140625,
      "learning_rate": 6.667073516889016e-05,
      "loss": 0.003768964856863022,
      "step": 117470
    },
    {
      "epoch": 33.34657961964235,
      "grad_norm": 0.04725195840001106,
      "learning_rate": 6.666789667896679e-05,
      "loss": 0.005931961536407471,
      "step": 117480
    },
    {
      "epoch": 33.34941810956571,
      "grad_norm": 0.6409868001937866,
      "learning_rate": 6.666505818904343e-05,
      "loss": 0.007691122591495514,
      "step": 117490
    },
    {
      "epoch": 33.352256599489074,
      "grad_norm": 0.13280260562896729,
      "learning_rate": 6.666221969912008e-05,
      "loss": 0.004103472828865052,
      "step": 117500
    },
    {
      "epoch": 33.352256599489074,
      "eval_accuracy": 0.9710052775481656,
      "eval_loss": 0.10176429152488708,
      "eval_runtime": 32.0871,
      "eval_samples_per_second": 490.134,
      "eval_steps_per_second": 7.667,
      "step": 117500
    },
    {
      "epoch": 33.35509508941243,
      "grad_norm": 6.041482448577881,
      "learning_rate": 6.665938120919671e-05,
      "loss": 0.004424911737442016,
      "step": 117510
    },
    {
      "epoch": 33.35793357933579,
      "grad_norm": 1.8242093324661255,
      "learning_rate": 6.665654271927335e-05,
      "loss": 0.0030678030103445052,
      "step": 117520
    },
    {
      "epoch": 33.360772069259156,
      "grad_norm": 0.25828006863594055,
      "learning_rate": 6.665370422934999e-05,
      "loss": 0.002407431975007057,
      "step": 117530
    },
    {
      "epoch": 33.36361055918251,
      "grad_norm": 2.303197145462036,
      "learning_rate": 6.665086573942662e-05,
      "loss": 0.0020633935928344727,
      "step": 117540
    },
    {
      "epoch": 33.366449049105874,
      "grad_norm": 0.10785436630249023,
      "learning_rate": 6.664802724950326e-05,
      "loss": 0.005741871893405914,
      "step": 117550
    },
    {
      "epoch": 33.36928753902924,
      "grad_norm": 0.14172351360321045,
      "learning_rate": 6.66451887595799e-05,
      "loss": 0.01993151903152466,
      "step": 117560
    },
    {
      "epoch": 33.3721260289526,
      "grad_norm": 6.424083709716797,
      "learning_rate": 6.664235026965654e-05,
      "loss": 0.005172615498304367,
      "step": 117570
    },
    {
      "epoch": 33.374964518875956,
      "grad_norm": 0.3155151903629303,
      "learning_rate": 6.663951177973319e-05,
      "loss": 0.002070758119225502,
      "step": 117580
    },
    {
      "epoch": 33.37780300879932,
      "grad_norm": 0.1375153660774231,
      "learning_rate": 6.663667328980983e-05,
      "loss": 0.0025551144033670427,
      "step": 117590
    },
    {
      "epoch": 33.38064149872268,
      "grad_norm": 0.4031994938850403,
      "learning_rate": 6.663383479988647e-05,
      "loss": 0.0011833008378744126,
      "step": 117600
    },
    {
      "epoch": 33.38347998864604,
      "grad_norm": 4.744685649871826,
      "learning_rate": 6.66309963099631e-05,
      "loss": 0.012042104452848434,
      "step": 117610
    },
    {
      "epoch": 33.3863184785694,
      "grad_norm": 2.998222827911377,
      "learning_rate": 6.662815782003974e-05,
      "loss": 0.0020521258935332297,
      "step": 117620
    },
    {
      "epoch": 33.38915696849276,
      "grad_norm": 0.7382822036743164,
      "learning_rate": 6.662531933011638e-05,
      "loss": 0.0011817242950201035,
      "step": 117630
    },
    {
      "epoch": 33.391995458416126,
      "grad_norm": 9.048192024230957,
      "learning_rate": 6.662248084019302e-05,
      "loss": 0.004943189769983291,
      "step": 117640
    },
    {
      "epoch": 33.39483394833948,
      "grad_norm": 0.1631862223148346,
      "learning_rate": 6.661964235026966e-05,
      "loss": 0.0077832967042922975,
      "step": 117650
    },
    {
      "epoch": 33.397672438262845,
      "grad_norm": 0.20605161786079407,
      "learning_rate": 6.66168038603463e-05,
      "loss": 0.0020784838125109673,
      "step": 117660
    },
    {
      "epoch": 33.40051092818621,
      "grad_norm": 0.12529434263706207,
      "learning_rate": 6.661396537042293e-05,
      "loss": 0.002091728150844574,
      "step": 117670
    },
    {
      "epoch": 33.40334941810956,
      "grad_norm": 0.054162587970495224,
      "learning_rate": 6.661112688049957e-05,
      "loss": 0.0035916931927204134,
      "step": 117680
    },
    {
      "epoch": 33.406187908032926,
      "grad_norm": 1.2742531299591064,
      "learning_rate": 6.660828839057621e-05,
      "loss": 0.009231005609035493,
      "step": 117690
    },
    {
      "epoch": 33.40902639795629,
      "grad_norm": 7.961148262023926,
      "learning_rate": 6.660544990065286e-05,
      "loss": 0.0023948408663272856,
      "step": 117700
    },
    {
      "epoch": 33.411864887879645,
      "grad_norm": 0.05892060697078705,
      "learning_rate": 6.66026114107295e-05,
      "loss": 0.006335830688476563,
      "step": 117710
    },
    {
      "epoch": 33.41470337780301,
      "grad_norm": 0.3514561057090759,
      "learning_rate": 6.659977292080614e-05,
      "loss": 0.007222630828619003,
      "step": 117720
    },
    {
      "epoch": 33.41754186772637,
      "grad_norm": 17.219820022583008,
      "learning_rate": 6.659693443088278e-05,
      "loss": 0.0086212158203125,
      "step": 117730
    },
    {
      "epoch": 33.420380357649734,
      "grad_norm": 0.617728590965271,
      "learning_rate": 6.659409594095941e-05,
      "loss": 0.01009007915854454,
      "step": 117740
    },
    {
      "epoch": 33.42321884757309,
      "grad_norm": 1.9427403211593628,
      "learning_rate": 6.659125745103605e-05,
      "loss": 0.0019489798694849014,
      "step": 117750
    },
    {
      "epoch": 33.42605733749645,
      "grad_norm": 1.0313529968261719,
      "learning_rate": 6.658841896111269e-05,
      "loss": 0.005363673716783523,
      "step": 117760
    },
    {
      "epoch": 33.428895827419815,
      "grad_norm": 0.3089061677455902,
      "learning_rate": 6.658558047118933e-05,
      "loss": 0.0030334291979670523,
      "step": 117770
    },
    {
      "epoch": 33.43173431734317,
      "grad_norm": 0.05528602749109268,
      "learning_rate": 6.658274198126597e-05,
      "loss": 0.0025103989988565444,
      "step": 117780
    },
    {
      "epoch": 33.434572807266534,
      "grad_norm": 0.3590381145477295,
      "learning_rate": 6.657990349134261e-05,
      "loss": 0.001472625322639942,
      "step": 117790
    },
    {
      "epoch": 33.4374112971899,
      "grad_norm": 0.6082022786140442,
      "learning_rate": 6.657706500141924e-05,
      "loss": 0.008881958574056626,
      "step": 117800
    },
    {
      "epoch": 33.44024978711325,
      "grad_norm": 11.770957946777344,
      "learning_rate": 6.657422651149588e-05,
      "loss": 0.008058828115463258,
      "step": 117810
    },
    {
      "epoch": 33.443088277036615,
      "grad_norm": 0.6515213847160339,
      "learning_rate": 6.657138802157252e-05,
      "loss": 0.0040625520050525665,
      "step": 117820
    },
    {
      "epoch": 33.44592676695998,
      "grad_norm": 0.10093910992145538,
      "learning_rate": 6.656854953164917e-05,
      "loss": 0.0039804231375455855,
      "step": 117830
    },
    {
      "epoch": 33.44876525688334,
      "grad_norm": 0.03526574373245239,
      "learning_rate": 6.656571104172581e-05,
      "loss": 0.0024762894958257676,
      "step": 117840
    },
    {
      "epoch": 33.4516037468067,
      "grad_norm": 0.11977417021989822,
      "learning_rate": 6.656287255180245e-05,
      "loss": 0.0009741321206092835,
      "step": 117850
    },
    {
      "epoch": 33.45444223673006,
      "grad_norm": 0.03153037652373314,
      "learning_rate": 6.656003406187909e-05,
      "loss": 0.0016812663525342941,
      "step": 117860
    },
    {
      "epoch": 33.45728072665342,
      "grad_norm": 0.21912778913974762,
      "learning_rate": 6.655719557195572e-05,
      "loss": 0.0011515764519572258,
      "step": 117870
    },
    {
      "epoch": 33.46011921657678,
      "grad_norm": 0.1310998797416687,
      "learning_rate": 6.655435708203236e-05,
      "loss": 0.001064511202275753,
      "step": 117880
    },
    {
      "epoch": 33.46295770650014,
      "grad_norm": 0.9500657320022583,
      "learning_rate": 6.6551518592109e-05,
      "loss": 0.0011550221592187882,
      "step": 117890
    },
    {
      "epoch": 33.465796196423504,
      "grad_norm": 1.2433335781097412,
      "learning_rate": 6.654868010218564e-05,
      "loss": 0.0013618255034089088,
      "step": 117900
    },
    {
      "epoch": 33.46863468634686,
      "grad_norm": 0.625828206539154,
      "learning_rate": 6.654584161226228e-05,
      "loss": 0.0035163335502147674,
      "step": 117910
    },
    {
      "epoch": 33.47147317627022,
      "grad_norm": 0.40807151794433594,
      "learning_rate": 6.654300312233892e-05,
      "loss": 0.0018518872559070588,
      "step": 117920
    },
    {
      "epoch": 33.474311666193586,
      "grad_norm": 0.37297049164772034,
      "learning_rate": 6.654016463241555e-05,
      "loss": 0.0020586153492331504,
      "step": 117930
    },
    {
      "epoch": 33.47715015611695,
      "grad_norm": 0.11709805577993393,
      "learning_rate": 6.65373261424922e-05,
      "loss": 0.0014977343380451202,
      "step": 117940
    },
    {
      "epoch": 33.479988646040304,
      "grad_norm": 0.23101064562797546,
      "learning_rate": 6.653448765256884e-05,
      "loss": 0.0036140423268079757,
      "step": 117950
    },
    {
      "epoch": 33.48282713596367,
      "grad_norm": 0.026506168767809868,
      "learning_rate": 6.653164916264548e-05,
      "loss": 0.001249462366104126,
      "step": 117960
    },
    {
      "epoch": 33.48566562588703,
      "grad_norm": 5.453563213348389,
      "learning_rate": 6.652881067272212e-05,
      "loss": 0.0033817552030086517,
      "step": 117970
    },
    {
      "epoch": 33.488504115810386,
      "grad_norm": 0.9004629850387573,
      "learning_rate": 6.652597218279876e-05,
      "loss": 0.005653075873851776,
      "step": 117980
    },
    {
      "epoch": 33.49134260573375,
      "grad_norm": 0.3424185812473297,
      "learning_rate": 6.65231336928754e-05,
      "loss": 0.010584371536970139,
      "step": 117990
    },
    {
      "epoch": 33.49418109565711,
      "grad_norm": 0.1876843422651291,
      "learning_rate": 6.652029520295203e-05,
      "loss": 0.0024077557027339935,
      "step": 118000
    },
    {
      "epoch": 33.49418109565711,
      "eval_accuracy": 0.9739301837604121,
      "eval_loss": 0.09370450675487518,
      "eval_runtime": 32.4486,
      "eval_samples_per_second": 484.674,
      "eval_steps_per_second": 7.581,
      "step": 118000
    },
    {
      "epoch": 33.497019585580475,
      "grad_norm": 16.735023498535156,
      "learning_rate": 6.651745671302867e-05,
      "loss": 0.008442622423171998,
      "step": 118010
    },
    {
      "epoch": 33.49985807550383,
      "grad_norm": 0.6759421825408936,
      "learning_rate": 6.651461822310531e-05,
      "loss": 0.0022421222180128097,
      "step": 118020
    },
    {
      "epoch": 33.50269656542719,
      "grad_norm": 16.317548751831055,
      "learning_rate": 6.651177973318195e-05,
      "loss": 0.011362408101558686,
      "step": 118030
    },
    {
      "epoch": 33.505535055350556,
      "grad_norm": 0.19789162278175354,
      "learning_rate": 6.65089412432586e-05,
      "loss": 0.0050257094204425815,
      "step": 118040
    },
    {
      "epoch": 33.50837354527391,
      "grad_norm": 8.306166648864746,
      "learning_rate": 6.650610275333524e-05,
      "loss": 0.007889412343502045,
      "step": 118050
    },
    {
      "epoch": 33.511212035197275,
      "grad_norm": 1.6560711860656738,
      "learning_rate": 6.650326426341186e-05,
      "loss": 0.008737242221832276,
      "step": 118060
    },
    {
      "epoch": 33.51405052512064,
      "grad_norm": 0.582359254360199,
      "learning_rate": 6.65004257734885e-05,
      "loss": 0.0024565406143665313,
      "step": 118070
    },
    {
      "epoch": 33.51688901504399,
      "grad_norm": 0.4189947545528412,
      "learning_rate": 6.649758728356515e-05,
      "loss": 0.002820005640387535,
      "step": 118080
    },
    {
      "epoch": 33.519727504967356,
      "grad_norm": 0.03143097832798958,
      "learning_rate": 6.649474879364179e-05,
      "loss": 0.004976770654320717,
      "step": 118090
    },
    {
      "epoch": 33.52256599489072,
      "grad_norm": 1.8940445184707642,
      "learning_rate": 6.649191030371843e-05,
      "loss": 0.005611953884363174,
      "step": 118100
    },
    {
      "epoch": 33.52540448481408,
      "grad_norm": 2.091912269592285,
      "learning_rate": 6.648907181379507e-05,
      "loss": 0.012508454918861388,
      "step": 118110
    },
    {
      "epoch": 33.52824297473744,
      "grad_norm": 0.15885968506336212,
      "learning_rate": 6.648623332387171e-05,
      "loss": 0.002666310220956802,
      "step": 118120
    },
    {
      "epoch": 33.5310814646608,
      "grad_norm": 3.1097934246063232,
      "learning_rate": 6.648339483394834e-05,
      "loss": 0.006251376867294311,
      "step": 118130
    },
    {
      "epoch": 33.533919954584164,
      "grad_norm": 1.7463575601577759,
      "learning_rate": 6.648055634402498e-05,
      "loss": 0.007226001471281052,
      "step": 118140
    },
    {
      "epoch": 33.53675844450752,
      "grad_norm": 0.28277215361595154,
      "learning_rate": 6.647771785410162e-05,
      "loss": 0.006041520088911056,
      "step": 118150
    },
    {
      "epoch": 33.53959693443088,
      "grad_norm": 1.0614129304885864,
      "learning_rate": 6.647487936417825e-05,
      "loss": 0.010642955452203751,
      "step": 118160
    },
    {
      "epoch": 33.542435424354245,
      "grad_norm": 0.38257908821105957,
      "learning_rate": 6.64720408742549e-05,
      "loss": 0.002994239330291748,
      "step": 118170
    },
    {
      "epoch": 33.5452739142776,
      "grad_norm": 0.17365016043186188,
      "learning_rate": 6.646920238433155e-05,
      "loss": 0.006804358959197998,
      "step": 118180
    },
    {
      "epoch": 33.548112404200964,
      "grad_norm": 0.09420165419578552,
      "learning_rate": 6.646636389440817e-05,
      "loss": 0.014821115136146545,
      "step": 118190
    },
    {
      "epoch": 33.55095089412433,
      "grad_norm": 3.804135322570801,
      "learning_rate": 6.646352540448482e-05,
      "loss": 0.007811338454484939,
      "step": 118200
    },
    {
      "epoch": 33.55378938404769,
      "grad_norm": 0.16372185945510864,
      "learning_rate": 6.646068691456146e-05,
      "loss": 0.0031401701271533966,
      "step": 118210
    },
    {
      "epoch": 33.556627873971046,
      "grad_norm": 1.3102390766143799,
      "learning_rate": 6.64578484246381e-05,
      "loss": 0.001990114897489548,
      "step": 118220
    },
    {
      "epoch": 33.55946636389441,
      "grad_norm": 0.27745521068573,
      "learning_rate": 6.645500993471474e-05,
      "loss": 0.006152790039777756,
      "step": 118230
    },
    {
      "epoch": 33.56230485381777,
      "grad_norm": 0.13625353574752808,
      "learning_rate": 6.645217144479138e-05,
      "loss": 0.00917484685778618,
      "step": 118240
    },
    {
      "epoch": 33.56514334374113,
      "grad_norm": 0.15715394914150238,
      "learning_rate": 6.644933295486801e-05,
      "loss": 0.007474235445261002,
      "step": 118250
    },
    {
      "epoch": 33.56798183366449,
      "grad_norm": 4.922328948974609,
      "learning_rate": 6.644649446494465e-05,
      "loss": 0.004115850478410721,
      "step": 118260
    },
    {
      "epoch": 33.57082032358785,
      "grad_norm": 0.2671124339103699,
      "learning_rate": 6.644365597502129e-05,
      "loss": 0.004395883902907372,
      "step": 118270
    },
    {
      "epoch": 33.57365881351121,
      "grad_norm": 1.0523269176483154,
      "learning_rate": 6.644081748509793e-05,
      "loss": 0.009388954937458038,
      "step": 118280
    },
    {
      "epoch": 33.57649730343457,
      "grad_norm": 0.6893566250801086,
      "learning_rate": 6.643797899517456e-05,
      "loss": 0.018699949979782103,
      "step": 118290
    },
    {
      "epoch": 33.579335793357934,
      "grad_norm": 0.3954017162322998,
      "learning_rate": 6.643514050525122e-05,
      "loss": 0.005369166284799576,
      "step": 118300
    },
    {
      "epoch": 33.5821742832813,
      "grad_norm": 1.7790608406066895,
      "learning_rate": 6.643230201532786e-05,
      "loss": 0.0035857439041137697,
      "step": 118310
    },
    {
      "epoch": 33.58501277320465,
      "grad_norm": 0.8237112760543823,
      "learning_rate": 6.642946352540448e-05,
      "loss": 0.00260560717433691,
      "step": 118320
    },
    {
      "epoch": 33.587851263128016,
      "grad_norm": 1.2745623588562012,
      "learning_rate": 6.642662503548113e-05,
      "loss": 0.009308111667633057,
      "step": 118330
    },
    {
      "epoch": 33.59068975305138,
      "grad_norm": 2.8505353927612305,
      "learning_rate": 6.642378654555777e-05,
      "loss": 0.0034400660544633864,
      "step": 118340
    },
    {
      "epoch": 33.593528242974735,
      "grad_norm": 0.832823634147644,
      "learning_rate": 6.64209480556344e-05,
      "loss": 0.0026300268247723578,
      "step": 118350
    },
    {
      "epoch": 33.5963667328981,
      "grad_norm": 0.11357098072767258,
      "learning_rate": 6.641810956571104e-05,
      "loss": 0.004665753990411759,
      "step": 118360
    },
    {
      "epoch": 33.59920522282146,
      "grad_norm": 0.12819187343120575,
      "learning_rate": 6.641527107578769e-05,
      "loss": 0.004100415855646133,
      "step": 118370
    },
    {
      "epoch": 33.602043712744816,
      "grad_norm": 1.0581845045089722,
      "learning_rate": 6.641243258586432e-05,
      "loss": 0.013720394670963287,
      "step": 118380
    },
    {
      "epoch": 33.60488220266818,
      "grad_norm": 9.020030975341797,
      "learning_rate": 6.640959409594096e-05,
      "loss": 0.009853778034448623,
      "step": 118390
    },
    {
      "epoch": 33.60772069259154,
      "grad_norm": 2.7118237018585205,
      "learning_rate": 6.64067556060176e-05,
      "loss": 0.00771016925573349,
      "step": 118400
    },
    {
      "epoch": 33.610559182514905,
      "grad_norm": 0.40723446011543274,
      "learning_rate": 6.640391711609424e-05,
      "loss": 0.006423964351415634,
      "step": 118410
    },
    {
      "epoch": 33.61339767243826,
      "grad_norm": 12.010626792907715,
      "learning_rate": 6.640107862617087e-05,
      "loss": 0.013506373763084412,
      "step": 118420
    },
    {
      "epoch": 33.61623616236162,
      "grad_norm": 18.09345817565918,
      "learning_rate": 6.639824013624753e-05,
      "loss": 0.017544372379779814,
      "step": 118430
    },
    {
      "epoch": 33.619074652284986,
      "grad_norm": 2.6834728717803955,
      "learning_rate": 6.639540164632417e-05,
      "loss": 0.005507006123661995,
      "step": 118440
    },
    {
      "epoch": 33.62191314220834,
      "grad_norm": 0.1859956681728363,
      "learning_rate": 6.63925631564008e-05,
      "loss": 0.004598772898316383,
      "step": 118450
    },
    {
      "epoch": 33.624751632131705,
      "grad_norm": 0.677299439907074,
      "learning_rate": 6.638972466647744e-05,
      "loss": 0.002621745876967907,
      "step": 118460
    },
    {
      "epoch": 33.62759012205507,
      "grad_norm": 0.10251724720001221,
      "learning_rate": 6.638688617655408e-05,
      "loss": 0.003544922173023224,
      "step": 118470
    },
    {
      "epoch": 33.63042861197843,
      "grad_norm": 2.2537879943847656,
      "learning_rate": 6.63840476866307e-05,
      "loss": 0.0033119872212409975,
      "step": 118480
    },
    {
      "epoch": 33.63326710190179,
      "grad_norm": 14.366512298583984,
      "learning_rate": 6.638120919670735e-05,
      "loss": 0.0173773393034935,
      "step": 118490
    },
    {
      "epoch": 33.63610559182515,
      "grad_norm": 0.27313777804374695,
      "learning_rate": 6.6378370706784e-05,
      "loss": 0.004499942064285278,
      "step": 118500
    },
    {
      "epoch": 33.63610559182515,
      "eval_accuracy": 0.9694156546067273,
      "eval_loss": 0.10657570511102676,
      "eval_runtime": 31.9221,
      "eval_samples_per_second": 492.667,
      "eval_steps_per_second": 7.706,
      "step": 118500
    },
    {
      "epoch": 33.63894408174851,
      "grad_norm": 12.45316219329834,
      "learning_rate": 6.637553221686063e-05,
      "loss": 0.010004299879074096,
      "step": 118510
    },
    {
      "epoch": 33.64178257167187,
      "grad_norm": 4.8703694343566895,
      "learning_rate": 6.637269372693727e-05,
      "loss": 0.01895466595888138,
      "step": 118520
    },
    {
      "epoch": 33.64462106159523,
      "grad_norm": 0.43803897500038147,
      "learning_rate": 6.636985523701391e-05,
      "loss": 0.008851355314254761,
      "step": 118530
    },
    {
      "epoch": 33.647459551518594,
      "grad_norm": 0.45062896609306335,
      "learning_rate": 6.636701674709055e-05,
      "loss": 0.005634177848696709,
      "step": 118540
    },
    {
      "epoch": 33.65029804144195,
      "grad_norm": 7.604182720184326,
      "learning_rate": 6.636417825716718e-05,
      "loss": 0.006014469265937805,
      "step": 118550
    },
    {
      "epoch": 33.65313653136531,
      "grad_norm": 1.0684056282043457,
      "learning_rate": 6.636133976724382e-05,
      "loss": 0.005597289651632309,
      "step": 118560
    },
    {
      "epoch": 33.655975021288675,
      "grad_norm": 0.15650734305381775,
      "learning_rate": 6.635850127732048e-05,
      "loss": 0.003872975707054138,
      "step": 118570
    },
    {
      "epoch": 33.65881351121204,
      "grad_norm": 0.13986270129680634,
      "learning_rate": 6.63556627873971e-05,
      "loss": 0.01006293147802353,
      "step": 118580
    },
    {
      "epoch": 33.661652001135394,
      "grad_norm": 0.5802822709083557,
      "learning_rate": 6.635282429747375e-05,
      "loss": 0.005593271553516388,
      "step": 118590
    },
    {
      "epoch": 33.66449049105876,
      "grad_norm": 0.9265126585960388,
      "learning_rate": 6.634998580755039e-05,
      "loss": 0.0027913374826312063,
      "step": 118600
    },
    {
      "epoch": 33.66732898098212,
      "grad_norm": 0.06432979553937912,
      "learning_rate": 6.634714731762702e-05,
      "loss": 0.001954894885420799,
      "step": 118610
    },
    {
      "epoch": 33.670167470905476,
      "grad_norm": 0.18897230923175812,
      "learning_rate": 6.634430882770366e-05,
      "loss": 0.0052628666162490845,
      "step": 118620
    },
    {
      "epoch": 33.67300596082884,
      "grad_norm": 10.748712539672852,
      "learning_rate": 6.634147033778031e-05,
      "loss": 0.014139921963214874,
      "step": 118630
    },
    {
      "epoch": 33.6758444507522,
      "grad_norm": 1.3574153184890747,
      "learning_rate": 6.633863184785694e-05,
      "loss": 0.0062481328845024105,
      "step": 118640
    },
    {
      "epoch": 33.67868294067556,
      "grad_norm": 0.28493428230285645,
      "learning_rate": 6.633579335793358e-05,
      "loss": 0.002008277177810669,
      "step": 118650
    },
    {
      "epoch": 33.68152143059892,
      "grad_norm": 1.6867533922195435,
      "learning_rate": 6.633295486801022e-05,
      "loss": 0.01278042197227478,
      "step": 118660
    },
    {
      "epoch": 33.68435992052228,
      "grad_norm": 0.05934680253267288,
      "learning_rate": 6.633011637808687e-05,
      "loss": 0.011395807564258575,
      "step": 118670
    },
    {
      "epoch": 33.687198410445646,
      "grad_norm": 4.227171421051025,
      "learning_rate": 6.632727788816349e-05,
      "loss": 0.0017517570406198502,
      "step": 118680
    },
    {
      "epoch": 33.690036900369,
      "grad_norm": 0.5052345395088196,
      "learning_rate": 6.632443939824013e-05,
      "loss": 0.003521227091550827,
      "step": 118690
    },
    {
      "epoch": 33.692875390292365,
      "grad_norm": 1.6888450384140015,
      "learning_rate": 6.632160090831679e-05,
      "loss": 0.005626156181097031,
      "step": 118700
    },
    {
      "epoch": 33.69571388021573,
      "grad_norm": 0.39632701873779297,
      "learning_rate": 6.631876241839342e-05,
      "loss": 0.005465247482061386,
      "step": 118710
    },
    {
      "epoch": 33.69855237013908,
      "grad_norm": 9.108137130737305,
      "learning_rate": 6.631592392847006e-05,
      "loss": 0.0037058554589748383,
      "step": 118720
    },
    {
      "epoch": 33.701390860062446,
      "grad_norm": 1.7768782377243042,
      "learning_rate": 6.63130854385467e-05,
      "loss": 0.00430828109383583,
      "step": 118730
    },
    {
      "epoch": 33.70422934998581,
      "grad_norm": 1.021230936050415,
      "learning_rate": 6.631024694862333e-05,
      "loss": 0.00667378231883049,
      "step": 118740
    },
    {
      "epoch": 33.70706783990917,
      "grad_norm": 1.0527336597442627,
      "learning_rate": 6.630740845869997e-05,
      "loss": 0.004770369455218315,
      "step": 118750
    },
    {
      "epoch": 33.70990632983253,
      "grad_norm": 2.049328565597534,
      "learning_rate": 6.630456996877661e-05,
      "loss": 0.0014734882861375809,
      "step": 118760
    },
    {
      "epoch": 33.71274481975589,
      "grad_norm": 0.03193238750100136,
      "learning_rate": 6.630173147885325e-05,
      "loss": 0.0069622650742530824,
      "step": 118770
    },
    {
      "epoch": 33.71558330967925,
      "grad_norm": 0.3328356146812439,
      "learning_rate": 6.62988929889299e-05,
      "loss": 0.0036469902843236923,
      "step": 118780
    },
    {
      "epoch": 33.71842179960261,
      "grad_norm": 0.5575516819953918,
      "learning_rate": 6.629605449900653e-05,
      "loss": 0.009485018998384475,
      "step": 118790
    },
    {
      "epoch": 33.72126028952597,
      "grad_norm": 0.4404263198375702,
      "learning_rate": 6.629321600908318e-05,
      "loss": 0.006475710123777389,
      "step": 118800
    },
    {
      "epoch": 33.724098779449335,
      "grad_norm": 0.07204845547676086,
      "learning_rate": 6.62903775191598e-05,
      "loss": 0.0013757141306996346,
      "step": 118810
    },
    {
      "epoch": 33.72693726937269,
      "grad_norm": 0.10146921873092651,
      "learning_rate": 6.628753902923645e-05,
      "loss": 0.0020020967349410057,
      "step": 118820
    },
    {
      "epoch": 33.729775759296054,
      "grad_norm": 0.33432987332344055,
      "learning_rate": 6.62847005393131e-05,
      "loss": 0.012701784074306489,
      "step": 118830
    },
    {
      "epoch": 33.73261424921942,
      "grad_norm": 0.6578109264373779,
      "learning_rate": 6.628186204938973e-05,
      "loss": 0.004331035912036896,
      "step": 118840
    },
    {
      "epoch": 33.73545273914278,
      "grad_norm": 1.9578584432601929,
      "learning_rate": 6.627902355946637e-05,
      "loss": 0.002381997928023338,
      "step": 118850
    },
    {
      "epoch": 33.738291229066135,
      "grad_norm": 0.22290807962417603,
      "learning_rate": 6.627618506954301e-05,
      "loss": 0.013844422996044159,
      "step": 118860
    },
    {
      "epoch": 33.7411297189895,
      "grad_norm": 11.913777351379395,
      "learning_rate": 6.627334657961964e-05,
      "loss": 0.010574301332235336,
      "step": 118870
    },
    {
      "epoch": 33.74396820891286,
      "grad_norm": 0.3002444803714752,
      "learning_rate": 6.627050808969628e-05,
      "loss": 0.014105919003486633,
      "step": 118880
    },
    {
      "epoch": 33.74680669883622,
      "grad_norm": 0.7923224568367004,
      "learning_rate": 6.626766959977292e-05,
      "loss": 0.002935725264251232,
      "step": 118890
    },
    {
      "epoch": 33.74964518875958,
      "grad_norm": 0.2138952612876892,
      "learning_rate": 6.626483110984956e-05,
      "loss": 0.007966580986976623,
      "step": 118900
    },
    {
      "epoch": 33.75248367868294,
      "grad_norm": 0.1423509120941162,
      "learning_rate": 6.62619926199262e-05,
      "loss": 0.00687478557229042,
      "step": 118910
    },
    {
      "epoch": 33.7553221686063,
      "grad_norm": 0.32538729906082153,
      "learning_rate": 6.625915413000285e-05,
      "loss": 0.007221842557191849,
      "step": 118920
    },
    {
      "epoch": 33.75816065852966,
      "grad_norm": 2.1803457736968994,
      "learning_rate": 6.625631564007949e-05,
      "loss": 0.002268337085843086,
      "step": 118930
    },
    {
      "epoch": 33.760999148453024,
      "grad_norm": 11.882611274719238,
      "learning_rate": 6.625347715015611e-05,
      "loss": 0.008549454063177109,
      "step": 118940
    },
    {
      "epoch": 33.76383763837639,
      "grad_norm": 0.8159015774726868,
      "learning_rate": 6.625063866023276e-05,
      "loss": 0.012360304594039917,
      "step": 118950
    },
    {
      "epoch": 33.76667612829974,
      "grad_norm": 4.055695056915283,
      "learning_rate": 6.62478001703094e-05,
      "loss": 0.0035634774714708327,
      "step": 118960
    },
    {
      "epoch": 33.769514618223106,
      "grad_norm": 0.03216114640235901,
      "learning_rate": 6.624496168038604e-05,
      "loss": 0.005543794855475426,
      "step": 118970
    },
    {
      "epoch": 33.77235310814647,
      "grad_norm": 0.5488452911376953,
      "learning_rate": 6.624212319046268e-05,
      "loss": 0.0016342954710125922,
      "step": 118980
    },
    {
      "epoch": 33.775191598069824,
      "grad_norm": 1.044049620628357,
      "learning_rate": 6.623928470053932e-05,
      "loss": 0.007719455659389496,
      "step": 118990
    },
    {
      "epoch": 33.77803008799319,
      "grad_norm": 0.8266379833221436,
      "learning_rate": 6.623644621061595e-05,
      "loss": 0.005917473509907722,
      "step": 119000
    },
    {
      "epoch": 33.77803008799319,
      "eval_accuracy": 0.972722070324919,
      "eval_loss": 0.09347350895404816,
      "eval_runtime": 31.7361,
      "eval_samples_per_second": 495.555,
      "eval_steps_per_second": 7.751,
      "step": 119000
    },
    {
      "epoch": 33.78086857791655,
      "grad_norm": 0.18504337966442108,
      "learning_rate": 6.623360772069259e-05,
      "loss": 0.0025294085964560507,
      "step": 119010
    },
    {
      "epoch": 33.783707067839906,
      "grad_norm": 1.9773956537246704,
      "learning_rate": 6.623076923076923e-05,
      "loss": 0.0055483274161815645,
      "step": 119020
    },
    {
      "epoch": 33.78654555776327,
      "grad_norm": 19.135757446289062,
      "learning_rate": 6.622793074084587e-05,
      "loss": 0.021419247984886168,
      "step": 119030
    },
    {
      "epoch": 33.78938404768663,
      "grad_norm": 0.40320274233818054,
      "learning_rate": 6.622509225092251e-05,
      "loss": 0.005116710811853409,
      "step": 119040
    },
    {
      "epoch": 33.792222537609995,
      "grad_norm": 14.738720893859863,
      "learning_rate": 6.622253760999148e-05,
      "loss": 0.0250186026096344,
      "step": 119050
    },
    {
      "epoch": 33.79506102753335,
      "grad_norm": 0.658054769039154,
      "learning_rate": 6.621969912006812e-05,
      "loss": 0.005745350569486618,
      "step": 119060
    },
    {
      "epoch": 33.79789951745671,
      "grad_norm": 7.9984025955200195,
      "learning_rate": 6.621686063014476e-05,
      "loss": 0.008312442153692246,
      "step": 119070
    },
    {
      "epoch": 33.800738007380076,
      "grad_norm": 4.527327537536621,
      "learning_rate": 6.62140221402214e-05,
      "loss": 0.004733985289931297,
      "step": 119080
    },
    {
      "epoch": 33.80357649730343,
      "grad_norm": 0.683903694152832,
      "learning_rate": 6.621118365029805e-05,
      "loss": 0.008793150633573532,
      "step": 119090
    },
    {
      "epoch": 33.806414987226795,
      "grad_norm": 0.10046201944351196,
      "learning_rate": 6.620834516037469e-05,
      "loss": 0.005145049095153809,
      "step": 119100
    },
    {
      "epoch": 33.80925347715016,
      "grad_norm": 0.527502179145813,
      "learning_rate": 6.620550667045133e-05,
      "loss": 0.003981425613164902,
      "step": 119110
    },
    {
      "epoch": 33.81209196707351,
      "grad_norm": 1.7586145401000977,
      "learning_rate": 6.620266818052796e-05,
      "loss": 0.03190743923187256,
      "step": 119120
    },
    {
      "epoch": 33.814930456996876,
      "grad_norm": 0.8558820486068726,
      "learning_rate": 6.61998296906046e-05,
      "loss": 0.0036613203585147856,
      "step": 119130
    },
    {
      "epoch": 33.81776894692024,
      "grad_norm": 5.289999961853027,
      "learning_rate": 6.619699120068124e-05,
      "loss": 0.004878237470984459,
      "step": 119140
    },
    {
      "epoch": 33.8206074368436,
      "grad_norm": 2.6638681888580322,
      "learning_rate": 6.619415271075788e-05,
      "loss": 0.002939035929739475,
      "step": 119150
    },
    {
      "epoch": 33.82344592676696,
      "grad_norm": 0.48380157351493835,
      "learning_rate": 6.619131422083452e-05,
      "loss": 0.013601365685462951,
      "step": 119160
    },
    {
      "epoch": 33.82628441669032,
      "grad_norm": 0.13613036274909973,
      "learning_rate": 6.618847573091116e-05,
      "loss": 0.006524347513914108,
      "step": 119170
    },
    {
      "epoch": 33.829122906613684,
      "grad_norm": 0.08632665872573853,
      "learning_rate": 6.618563724098779e-05,
      "loss": 0.0023039234802126884,
      "step": 119180
    },
    {
      "epoch": 33.83196139653704,
      "grad_norm": 3.8785178661346436,
      "learning_rate": 6.618279875106443e-05,
      "loss": 0.006764629483222961,
      "step": 119190
    },
    {
      "epoch": 33.8347998864604,
      "grad_norm": 7.321354389190674,
      "learning_rate": 6.617996026114107e-05,
      "loss": 0.005146923661231995,
      "step": 119200
    },
    {
      "epoch": 33.837638376383765,
      "grad_norm": 1.948907732963562,
      "learning_rate": 6.617712177121772e-05,
      "loss": 0.0028041077777743338,
      "step": 119210
    },
    {
      "epoch": 33.84047686630713,
      "grad_norm": 0.5973936915397644,
      "learning_rate": 6.617428328129436e-05,
      "loss": 0.011538489907979965,
      "step": 119220
    },
    {
      "epoch": 33.843315356230484,
      "grad_norm": 1.4393985271453857,
      "learning_rate": 6.6171444791371e-05,
      "loss": 0.0024961814284324645,
      "step": 119230
    },
    {
      "epoch": 33.84615384615385,
      "grad_norm": 0.06161246448755264,
      "learning_rate": 6.616860630144763e-05,
      "loss": 0.006900164484977722,
      "step": 119240
    },
    {
      "epoch": 33.84899233607721,
      "grad_norm": 0.19868294894695282,
      "learning_rate": 6.616576781152427e-05,
      "loss": 0.0014829928055405616,
      "step": 119250
    },
    {
      "epoch": 33.851830826000565,
      "grad_norm": 1.5332965850830078,
      "learning_rate": 6.616292932160091e-05,
      "loss": 0.005622328072786331,
      "step": 119260
    },
    {
      "epoch": 33.85466931592393,
      "grad_norm": 0.18568873405456543,
      "learning_rate": 6.616009083167755e-05,
      "loss": 0.0013919040560722352,
      "step": 119270
    },
    {
      "epoch": 33.85750780584729,
      "grad_norm": 2.5446200370788574,
      "learning_rate": 6.615725234175419e-05,
      "loss": 0.0068504571914672855,
      "step": 119280
    },
    {
      "epoch": 33.86034629577065,
      "grad_norm": 16.650548934936523,
      "learning_rate": 6.615441385183083e-05,
      "loss": 0.009473662078380584,
      "step": 119290
    },
    {
      "epoch": 33.86318478569401,
      "grad_norm": 0.08913446217775345,
      "learning_rate": 6.615157536190747e-05,
      "loss": 0.0028274305164813994,
      "step": 119300
    },
    {
      "epoch": 33.86602327561737,
      "grad_norm": 0.8502799868583679,
      "learning_rate": 6.61487368719841e-05,
      "loss": 0.002146740071475506,
      "step": 119310
    },
    {
      "epoch": 33.868861765540736,
      "grad_norm": 0.9019044041633606,
      "learning_rate": 6.614589838206074e-05,
      "loss": 0.004552269354462624,
      "step": 119320
    },
    {
      "epoch": 33.87170025546409,
      "grad_norm": 0.37760862708091736,
      "learning_rate": 6.614305989213739e-05,
      "loss": 0.005337106063961983,
      "step": 119330
    },
    {
      "epoch": 33.874538745387454,
      "grad_norm": 9.183419227600098,
      "learning_rate": 6.614022140221401e-05,
      "loss": 0.008210556954145432,
      "step": 119340
    },
    {
      "epoch": 33.87737723531082,
      "grad_norm": 6.854710102081299,
      "learning_rate": 6.613738291229067e-05,
      "loss": 0.005192628130316734,
      "step": 119350
    },
    {
      "epoch": 33.88021572523417,
      "grad_norm": 0.9912996888160706,
      "learning_rate": 6.613454442236731e-05,
      "loss": 0.008374271541833877,
      "step": 119360
    },
    {
      "epoch": 33.883054215157536,
      "grad_norm": 0.19534367322921753,
      "learning_rate": 6.613170593244394e-05,
      "loss": 0.005766856670379639,
      "step": 119370
    },
    {
      "epoch": 33.8858927050809,
      "grad_norm": 4.808982849121094,
      "learning_rate": 6.612886744252058e-05,
      "loss": 0.005156569182872772,
      "step": 119380
    },
    {
      "epoch": 33.888731195004254,
      "grad_norm": 1.2250205278396606,
      "learning_rate": 6.612602895259722e-05,
      "loss": 0.016035692393779756,
      "step": 119390
    },
    {
      "epoch": 33.89156968492762,
      "grad_norm": 1.4834147691726685,
      "learning_rate": 6.612319046267386e-05,
      "loss": 0.004677487164735794,
      "step": 119400
    },
    {
      "epoch": 33.89440817485098,
      "grad_norm": 0.7850981950759888,
      "learning_rate": 6.61203519727505e-05,
      "loss": 0.0022174160927534103,
      "step": 119410
    },
    {
      "epoch": 33.89724666477434,
      "grad_norm": 0.38531237840652466,
      "learning_rate": 6.611751348282714e-05,
      "loss": 0.004461555182933808,
      "step": 119420
    },
    {
      "epoch": 33.9000851546977,
      "grad_norm": 0.4589572548866272,
      "learning_rate": 6.611467499290379e-05,
      "loss": 0.005551864206790924,
      "step": 119430
    },
    {
      "epoch": 33.90292364462106,
      "grad_norm": 0.008059073239564896,
      "learning_rate": 6.611183650298041e-05,
      "loss": 0.006022947281599045,
      "step": 119440
    },
    {
      "epoch": 33.905762134544425,
      "grad_norm": 0.009944310411810875,
      "learning_rate": 6.610899801305705e-05,
      "loss": 0.006749437004327774,
      "step": 119450
    },
    {
      "epoch": 33.90860062446778,
      "grad_norm": 0.11665124446153641,
      "learning_rate": 6.61061595231337e-05,
      "loss": 0.00559355691075325,
      "step": 119460
    },
    {
      "epoch": 33.91143911439114,
      "grad_norm": 1.4448776245117188,
      "learning_rate": 6.610332103321032e-05,
      "loss": 0.002122136205434799,
      "step": 119470
    },
    {
      "epoch": 33.914277604314506,
      "grad_norm": 0.32620781660079956,
      "learning_rate": 6.610048254328698e-05,
      "loss": 0.006646633148193359,
      "step": 119480
    },
    {
      "epoch": 33.91711609423786,
      "grad_norm": 0.5517118573188782,
      "learning_rate": 6.609764405336362e-05,
      "loss": 0.003561890125274658,
      "step": 119490
    },
    {
      "epoch": 33.919954584161225,
      "grad_norm": 0.19459865987300873,
      "learning_rate": 6.609480556344025e-05,
      "loss": 0.002037677727639675,
      "step": 119500
    },
    {
      "epoch": 33.919954584161225,
      "eval_accuracy": 0.9712596172187957,
      "eval_loss": 0.0952950045466423,
      "eval_runtime": 32.0591,
      "eval_samples_per_second": 490.562,
      "eval_steps_per_second": 7.673,
      "step": 119500
    },
    {
      "epoch": 33.92279307408459,
      "grad_norm": 0.42899176478385925,
      "learning_rate": 6.609196707351689e-05,
      "loss": 0.002759654074907303,
      "step": 119510
    },
    {
      "epoch": 33.92563156400795,
      "grad_norm": 0.42460140585899353,
      "learning_rate": 6.608912858359353e-05,
      "loss": 0.007476992160081864,
      "step": 119520
    },
    {
      "epoch": 33.928470053931306,
      "grad_norm": 1.0898528099060059,
      "learning_rate": 6.608629009367017e-05,
      "loss": 0.002463330328464508,
      "step": 119530
    },
    {
      "epoch": 33.93130854385467,
      "grad_norm": 0.07702265679836273,
      "learning_rate": 6.60834516037468e-05,
      "loss": 0.017772521078586578,
      "step": 119540
    },
    {
      "epoch": 33.93414703377803,
      "grad_norm": 8.78197956085205,
      "learning_rate": 6.608061311382346e-05,
      "loss": 0.01873738467693329,
      "step": 119550
    },
    {
      "epoch": 33.93698552370139,
      "grad_norm": 0.35545623302459717,
      "learning_rate": 6.60777746239001e-05,
      "loss": 0.0028546275570988655,
      "step": 119560
    },
    {
      "epoch": 33.93982401362475,
      "grad_norm": 0.6335259675979614,
      "learning_rate": 6.607493613397672e-05,
      "loss": 0.006526433676481247,
      "step": 119570
    },
    {
      "epoch": 33.942662503548114,
      "grad_norm": 2.91133713722229,
      "learning_rate": 6.607209764405337e-05,
      "loss": 0.004704262316226959,
      "step": 119580
    },
    {
      "epoch": 33.94550099347148,
      "grad_norm": 10.96039867401123,
      "learning_rate": 6.606925915413001e-05,
      "loss": 0.012880182266235352,
      "step": 119590
    },
    {
      "epoch": 33.94833948339483,
      "grad_norm": 12.452750205993652,
      "learning_rate": 6.606642066420663e-05,
      "loss": 0.006967640668153763,
      "step": 119600
    },
    {
      "epoch": 33.951177973318195,
      "grad_norm": 3.2905380725860596,
      "learning_rate": 6.606358217428329e-05,
      "loss": 0.0017825430259108543,
      "step": 119610
    },
    {
      "epoch": 33.95401646324156,
      "grad_norm": 0.3956787884235382,
      "learning_rate": 6.606074368435993e-05,
      "loss": 0.0058490529656410216,
      "step": 119620
    },
    {
      "epoch": 33.956854953164914,
      "grad_norm": 3.344552516937256,
      "learning_rate": 6.605790519443656e-05,
      "loss": 0.003191586583852768,
      "step": 119630
    },
    {
      "epoch": 33.95969344308828,
      "grad_norm": 0.6010175943374634,
      "learning_rate": 6.60550667045132e-05,
      "loss": 0.002622613124549389,
      "step": 119640
    },
    {
      "epoch": 33.96253193301164,
      "grad_norm": 0.24956364929676056,
      "learning_rate": 6.605222821458984e-05,
      "loss": 0.0021810749545693397,
      "step": 119650
    },
    {
      "epoch": 33.965370422934996,
      "grad_norm": 0.5087161660194397,
      "learning_rate": 6.604938972466648e-05,
      "loss": 0.0039704151451587675,
      "step": 119660
    },
    {
      "epoch": 33.96820891285836,
      "grad_norm": 0.9862024784088135,
      "learning_rate": 6.604655123474311e-05,
      "loss": 0.004416496679186821,
      "step": 119670
    },
    {
      "epoch": 33.97104740278172,
      "grad_norm": 4.125974178314209,
      "learning_rate": 6.604371274481977e-05,
      "loss": 0.004498625174164772,
      "step": 119680
    },
    {
      "epoch": 33.973885892705084,
      "grad_norm": 0.3017573654651642,
      "learning_rate": 6.604087425489641e-05,
      "loss": 0.003621348738670349,
      "step": 119690
    },
    {
      "epoch": 33.97672438262844,
      "grad_norm": 0.2410620003938675,
      "learning_rate": 6.603803576497304e-05,
      "loss": 0.004757682606577873,
      "step": 119700
    },
    {
      "epoch": 33.9795628725518,
      "grad_norm": 0.6310582756996155,
      "learning_rate": 6.603519727504968e-05,
      "loss": 0.004366743564605713,
      "step": 119710
    },
    {
      "epoch": 33.982401362475166,
      "grad_norm": 0.9471608400344849,
      "learning_rate": 6.603235878512632e-05,
      "loss": 0.006595972925424576,
      "step": 119720
    },
    {
      "epoch": 33.98523985239852,
      "grad_norm": 14.939306259155273,
      "learning_rate": 6.602952029520295e-05,
      "loss": 0.01201133131980896,
      "step": 119730
    },
    {
      "epoch": 33.988078342321884,
      "grad_norm": 2.4232983589172363,
      "learning_rate": 6.60266818052796e-05,
      "loss": 0.0035892192274332045,
      "step": 119740
    },
    {
      "epoch": 33.99091683224525,
      "grad_norm": 4.69612979888916,
      "learning_rate": 6.602384331535624e-05,
      "loss": 0.007551954686641693,
      "step": 119750
    },
    {
      "epoch": 33.9937553221686,
      "grad_norm": 0.8327898979187012,
      "learning_rate": 6.602100482543287e-05,
      "loss": 0.003397147357463837,
      "step": 119760
    },
    {
      "epoch": 33.996593812091966,
      "grad_norm": 0.8515961170196533,
      "learning_rate": 6.601816633550951e-05,
      "loss": 0.002370903082191944,
      "step": 119770
    },
    {
      "epoch": 33.99943230201533,
      "grad_norm": 0.24727791547775269,
      "learning_rate": 6.601532784558615e-05,
      "loss": 0.004083870723843574,
      "step": 119780
    },
    {
      "epoch": 34.00227079193869,
      "grad_norm": 1.560390830039978,
      "learning_rate": 6.60124893556628e-05,
      "loss": 0.0021931326016783716,
      "step": 119790
    },
    {
      "epoch": 34.00510928186205,
      "grad_norm": 0.23149053752422333,
      "learning_rate": 6.600965086573942e-05,
      "loss": 0.002705189771950245,
      "step": 119800
    },
    {
      "epoch": 34.00794777178541,
      "grad_norm": 0.4312277138233185,
      "learning_rate": 6.600681237581608e-05,
      "loss": 0.001137382909655571,
      "step": 119810
    },
    {
      "epoch": 34.01078626170877,
      "grad_norm": 8.387903213500977,
      "learning_rate": 6.600397388589272e-05,
      "loss": 0.0077542029321193695,
      "step": 119820
    },
    {
      "epoch": 34.01362475163213,
      "grad_norm": 0.804916501045227,
      "learning_rate": 6.600113539596935e-05,
      "loss": 0.004466918483376503,
      "step": 119830
    },
    {
      "epoch": 34.01646324155549,
      "grad_norm": 0.31707876920700073,
      "learning_rate": 6.599829690604599e-05,
      "loss": 0.002220351994037628,
      "step": 119840
    },
    {
      "epoch": 34.019301731478855,
      "grad_norm": 0.2828819453716278,
      "learning_rate": 6.599545841612263e-05,
      "loss": 0.0023949654772877694,
      "step": 119850
    },
    {
      "epoch": 34.02214022140221,
      "grad_norm": 0.3264423906803131,
      "learning_rate": 6.599261992619926e-05,
      "loss": 0.0023763000965118406,
      "step": 119860
    },
    {
      "epoch": 34.02497871132557,
      "grad_norm": 3.518228530883789,
      "learning_rate": 6.59897814362759e-05,
      "loss": 0.008187740296125411,
      "step": 119870
    },
    {
      "epoch": 34.027817201248936,
      "grad_norm": 0.411467045545578,
      "learning_rate": 6.598694294635255e-05,
      "loss": 0.011641748249530792,
      "step": 119880
    },
    {
      "epoch": 34.0306556911723,
      "grad_norm": 0.6357895135879517,
      "learning_rate": 6.598410445642918e-05,
      "loss": 0.014119769632816314,
      "step": 119890
    },
    {
      "epoch": 34.033494181095655,
      "grad_norm": 1.389652967453003,
      "learning_rate": 6.598126596650582e-05,
      "loss": 0.003645886480808258,
      "step": 119900
    },
    {
      "epoch": 34.03633267101902,
      "grad_norm": 0.0479893833398819,
      "learning_rate": 6.597842747658246e-05,
      "loss": 0.0020871927961707117,
      "step": 119910
    },
    {
      "epoch": 34.03917116094238,
      "grad_norm": 3.1221442222595215,
      "learning_rate": 6.59755889866591e-05,
      "loss": 0.003246936947107315,
      "step": 119920
    },
    {
      "epoch": 34.04200965086574,
      "grad_norm": 1.9789921045303345,
      "learning_rate": 6.597275049673573e-05,
      "loss": 0.011044969409704208,
      "step": 119930
    },
    {
      "epoch": 34.0448481407891,
      "grad_norm": 0.4218142032623291,
      "learning_rate": 6.596991200681239e-05,
      "loss": 0.00381898432970047,
      "step": 119940
    },
    {
      "epoch": 34.04768663071246,
      "grad_norm": 0.8461046814918518,
      "learning_rate": 6.596707351688902e-05,
      "loss": 0.0024044355377554893,
      "step": 119950
    },
    {
      "epoch": 34.050525120635825,
      "grad_norm": 0.15074428915977478,
      "learning_rate": 6.596423502696566e-05,
      "loss": 0.0025313768535852434,
      "step": 119960
    },
    {
      "epoch": 34.05336361055918,
      "grad_norm": 0.37584373354911804,
      "learning_rate": 6.59613965370423e-05,
      "loss": 0.003507961332798004,
      "step": 119970
    },
    {
      "epoch": 34.056202100482544,
      "grad_norm": 0.22898602485656738,
      "learning_rate": 6.595855804711894e-05,
      "loss": 0.00391046330332756,
      "step": 119980
    },
    {
      "epoch": 34.05904059040591,
      "grad_norm": 0.2502862513065338,
      "learning_rate": 6.595571955719557e-05,
      "loss": 0.008375391364097595,
      "step": 119990
    },
    {
      "epoch": 34.06187908032926,
      "grad_norm": 0.04171254485845566,
      "learning_rate": 6.595288106727221e-05,
      "loss": 0.002264373190701008,
      "step": 120000
    },
    {
      "epoch": 34.06187908032926,
      "eval_accuracy": 0.97501112736059,
      "eval_loss": 0.09171520918607712,
      "eval_runtime": 31.7329,
      "eval_samples_per_second": 495.606,
      "eval_steps_per_second": 7.752,
      "step": 120000
    },
    {
      "epoch": 34.064717570252625,
      "grad_norm": 0.12566813826560974,
      "learning_rate": 6.595004257734886e-05,
      "loss": 0.008542881906032562,
      "step": 120010
    },
    {
      "epoch": 34.06755606017599,
      "grad_norm": 3.2479872703552246,
      "learning_rate": 6.594720408742549e-05,
      "loss": 0.00406017079949379,
      "step": 120020
    },
    {
      "epoch": 34.070394550099344,
      "grad_norm": 0.08720032125711441,
      "learning_rate": 6.594436559750213e-05,
      "loss": 0.001068960689008236,
      "step": 120030
    },
    {
      "epoch": 34.07323304002271,
      "grad_norm": 0.14192959666252136,
      "learning_rate": 6.594152710757877e-05,
      "loss": 0.005984793603420258,
      "step": 120040
    },
    {
      "epoch": 34.07607152994607,
      "grad_norm": 1.612888216972351,
      "learning_rate": 6.593868861765542e-05,
      "loss": 0.0025998856872320175,
      "step": 120050
    },
    {
      "epoch": 34.07891001986943,
      "grad_norm": 0.8385291695594788,
      "learning_rate": 6.593585012773204e-05,
      "loss": 0.001395496353507042,
      "step": 120060
    },
    {
      "epoch": 34.08174850979279,
      "grad_norm": 0.4187453091144562,
      "learning_rate": 6.593301163780868e-05,
      "loss": 0.0017596635967493056,
      "step": 120070
    },
    {
      "epoch": 34.08458699971615,
      "grad_norm": 0.7616415023803711,
      "learning_rate": 6.593017314788533e-05,
      "loss": 0.0006042171269655227,
      "step": 120080
    },
    {
      "epoch": 34.087425489639514,
      "grad_norm": 0.3409002125263214,
      "learning_rate": 6.592733465796197e-05,
      "loss": 0.0041004780679941176,
      "step": 120090
    },
    {
      "epoch": 34.09026397956287,
      "grad_norm": 2.6917452812194824,
      "learning_rate": 6.592449616803861e-05,
      "loss": 0.002159907855093479,
      "step": 120100
    },
    {
      "epoch": 34.09310246948623,
      "grad_norm": 1.0723531246185303,
      "learning_rate": 6.592165767811525e-05,
      "loss": 0.007290668785572052,
      "step": 120110
    },
    {
      "epoch": 34.095940959409596,
      "grad_norm": 17.104623794555664,
      "learning_rate": 6.591881918819188e-05,
      "loss": 0.008852966874837876,
      "step": 120120
    },
    {
      "epoch": 34.09877944933295,
      "grad_norm": 2.0907278060913086,
      "learning_rate": 6.591598069826852e-05,
      "loss": 0.0036791734397411346,
      "step": 120130
    },
    {
      "epoch": 34.101617939256315,
      "grad_norm": 0.28257763385772705,
      "learning_rate": 6.591314220834517e-05,
      "loss": 0.0014428889378905296,
      "step": 120140
    },
    {
      "epoch": 34.10445642917968,
      "grad_norm": 11.50788402557373,
      "learning_rate": 6.59103037184218e-05,
      "loss": 0.021514533460140227,
      "step": 120150
    },
    {
      "epoch": 34.10729491910304,
      "grad_norm": 0.06060031056404114,
      "learning_rate": 6.590746522849844e-05,
      "loss": 0.004983774572610855,
      "step": 120160
    },
    {
      "epoch": 34.110133409026396,
      "grad_norm": 0.09006636589765549,
      "learning_rate": 6.590462673857508e-05,
      "loss": 0.00188368558883667,
      "step": 120170
    },
    {
      "epoch": 34.11297189894976,
      "grad_norm": 0.08525079488754272,
      "learning_rate": 6.590178824865171e-05,
      "loss": 0.014009492099285125,
      "step": 120180
    },
    {
      "epoch": 34.11581038887312,
      "grad_norm": 1.4263910055160522,
      "learning_rate": 6.589894975872835e-05,
      "loss": 0.0017824728041887284,
      "step": 120190
    },
    {
      "epoch": 34.11864887879648,
      "grad_norm": 13.674447059631348,
      "learning_rate": 6.5896111268805e-05,
      "loss": 0.015447154641151428,
      "step": 120200
    },
    {
      "epoch": 34.12148736871984,
      "grad_norm": 0.20295418798923492,
      "learning_rate": 6.589327277888164e-05,
      "loss": 0.0009308716282248497,
      "step": 120210
    },
    {
      "epoch": 34.1243258586432,
      "grad_norm": 0.3147740066051483,
      "learning_rate": 6.589043428895828e-05,
      "loss": 0.006511497497558594,
      "step": 120220
    },
    {
      "epoch": 34.12716434856656,
      "grad_norm": 0.0220764372497797,
      "learning_rate": 6.588759579903492e-05,
      "loss": 0.0010855557397007943,
      "step": 120230
    },
    {
      "epoch": 34.13000283848992,
      "grad_norm": 1.274138331413269,
      "learning_rate": 6.588475730911156e-05,
      "loss": 0.0011312618851661682,
      "step": 120240
    },
    {
      "epoch": 34.132841328413285,
      "grad_norm": 6.901713848114014,
      "learning_rate": 6.588191881918819e-05,
      "loss": 0.008790388703346252,
      "step": 120250
    },
    {
      "epoch": 34.13567981833665,
      "grad_norm": 0.783248245716095,
      "learning_rate": 6.587908032926483e-05,
      "loss": 0.001980581507086754,
      "step": 120260
    },
    {
      "epoch": 34.138518308260004,
      "grad_norm": 5.808366298675537,
      "learning_rate": 6.587624183934147e-05,
      "loss": 0.009575360268354417,
      "step": 120270
    },
    {
      "epoch": 34.14135679818337,
      "grad_norm": 0.7367664575576782,
      "learning_rate": 6.587340334941811e-05,
      "loss": 0.003168274462223053,
      "step": 120280
    },
    {
      "epoch": 34.14419528810673,
      "grad_norm": 0.3028112053871155,
      "learning_rate": 6.587056485949475e-05,
      "loss": 0.004181566834449768,
      "step": 120290
    },
    {
      "epoch": 34.147033778030085,
      "grad_norm": 2.263658285140991,
      "learning_rate": 6.58677263695714e-05,
      "loss": 0.0023328032344579696,
      "step": 120300
    },
    {
      "epoch": 34.14987226795345,
      "grad_norm": 0.3305845856666565,
      "learning_rate": 6.586488787964802e-05,
      "loss": 0.00887860804796219,
      "step": 120310
    },
    {
      "epoch": 34.15271075787681,
      "grad_norm": 0.46553438901901245,
      "learning_rate": 6.586204938972466e-05,
      "loss": 0.0010540813207626342,
      "step": 120320
    },
    {
      "epoch": 34.15554924780017,
      "grad_norm": 2.120094060897827,
      "learning_rate": 6.58592108998013e-05,
      "loss": 0.007890991866588593,
      "step": 120330
    },
    {
      "epoch": 34.15838773772353,
      "grad_norm": 0.057511698454618454,
      "learning_rate": 6.585637240987795e-05,
      "loss": 0.0018104396760463715,
      "step": 120340
    },
    {
      "epoch": 34.16122622764689,
      "grad_norm": 0.4922485947608948,
      "learning_rate": 6.585353391995459e-05,
      "loss": 0.0028358491137623785,
      "step": 120350
    },
    {
      "epoch": 34.164064717570255,
      "grad_norm": 0.14623531699180603,
      "learning_rate": 6.585069543003123e-05,
      "loss": 0.0010311707854270935,
      "step": 120360
    },
    {
      "epoch": 34.16690320749361,
      "grad_norm": 1.3655275106430054,
      "learning_rate": 6.584785694010787e-05,
      "loss": 0.007381750643253327,
      "step": 120370
    },
    {
      "epoch": 34.169741697416974,
      "grad_norm": 3.024090051651001,
      "learning_rate": 6.58450184501845e-05,
      "loss": 0.005179420113563538,
      "step": 120380
    },
    {
      "epoch": 34.17258018734034,
      "grad_norm": 0.28700101375579834,
      "learning_rate": 6.584217996026114e-05,
      "loss": 0.0009599108248949051,
      "step": 120390
    },
    {
      "epoch": 34.17541867726369,
      "grad_norm": 9.125275611877441,
      "learning_rate": 6.583934147033778e-05,
      "loss": 0.0028434379026293756,
      "step": 120400
    },
    {
      "epoch": 34.178257167187056,
      "grad_norm": 0.12669123709201813,
      "learning_rate": 6.583650298041442e-05,
      "loss": 0.006512732803821563,
      "step": 120410
    },
    {
      "epoch": 34.18109565711042,
      "grad_norm": 0.018964238464832306,
      "learning_rate": 6.583366449049106e-05,
      "loss": 0.013804316520690918,
      "step": 120420
    },
    {
      "epoch": 34.18393414703378,
      "grad_norm": 5.092179298400879,
      "learning_rate": 6.58308260005677e-05,
      "loss": 0.006323505192995071,
      "step": 120430
    },
    {
      "epoch": 34.18677263695714,
      "grad_norm": 0.09260304272174835,
      "learning_rate": 6.582798751064433e-05,
      "loss": 0.004588256776332855,
      "step": 120440
    },
    {
      "epoch": 34.1896111268805,
      "grad_norm": 0.4579649269580841,
      "learning_rate": 6.582514902072098e-05,
      "loss": 0.0031266190111637116,
      "step": 120450
    },
    {
      "epoch": 34.19244961680386,
      "grad_norm": 1.9222780466079712,
      "learning_rate": 6.582231053079762e-05,
      "loss": 0.007798251509666443,
      "step": 120460
    },
    {
      "epoch": 34.19528810672722,
      "grad_norm": 0.7869184613227844,
      "learning_rate": 6.581947204087426e-05,
      "loss": 0.005959853902459145,
      "step": 120470
    },
    {
      "epoch": 34.19812659665058,
      "grad_norm": 2.2712228298187256,
      "learning_rate": 6.58166335509509e-05,
      "loss": 0.0022471331059932707,
      "step": 120480
    },
    {
      "epoch": 34.200965086573945,
      "grad_norm": 0.1493186205625534,
      "learning_rate": 6.581379506102754e-05,
      "loss": 0.015672385692596436,
      "step": 120490
    },
    {
      "epoch": 34.2038035764973,
      "grad_norm": 0.18912315368652344,
      "learning_rate": 6.581095657110418e-05,
      "loss": 0.0024687185883522033,
      "step": 120500
    },
    {
      "epoch": 34.2038035764973,
      "eval_accuracy": 0.9753926368665352,
      "eval_loss": 0.08686452358961105,
      "eval_runtime": 32.1024,
      "eval_samples_per_second": 489.9,
      "eval_steps_per_second": 7.663,
      "step": 120500
    },
    {
      "epoch": 34.20664206642066,
      "grad_norm": 2.927769660949707,
      "learning_rate": 6.580811808118081e-05,
      "loss": 0.0013673042878508568,
      "step": 120510
    },
    {
      "epoch": 34.209480556344026,
      "grad_norm": 0.27282968163490295,
      "learning_rate": 6.580527959125745e-05,
      "loss": 0.007888178527355193,
      "step": 120520
    },
    {
      "epoch": 34.21231904626739,
      "grad_norm": 0.11811738461256027,
      "learning_rate": 6.580244110133409e-05,
      "loss": 0.0040873602032661435,
      "step": 120530
    },
    {
      "epoch": 34.215157536190745,
      "grad_norm": 1.1339547634124756,
      "learning_rate": 6.579960261141073e-05,
      "loss": 0.014372356235980988,
      "step": 120540
    },
    {
      "epoch": 34.21799602611411,
      "grad_norm": 0.15421226620674133,
      "learning_rate": 6.579676412148738e-05,
      "loss": 0.00357535220682621,
      "step": 120550
    },
    {
      "epoch": 34.22083451603747,
      "grad_norm": 0.8814896941184998,
      "learning_rate": 6.579392563156402e-05,
      "loss": 0.01395888477563858,
      "step": 120560
    },
    {
      "epoch": 34.223673005960826,
      "grad_norm": 0.6410775184631348,
      "learning_rate": 6.579108714164064e-05,
      "loss": 0.002811104618012905,
      "step": 120570
    },
    {
      "epoch": 34.22651149588419,
      "grad_norm": 0.21123194694519043,
      "learning_rate": 6.578824865171729e-05,
      "loss": 0.0010377543047070503,
      "step": 120580
    },
    {
      "epoch": 34.22934998580755,
      "grad_norm": 0.23363442718982697,
      "learning_rate": 6.578541016179393e-05,
      "loss": 0.004100901260972023,
      "step": 120590
    },
    {
      "epoch": 34.23218847573091,
      "grad_norm": 0.1753082275390625,
      "learning_rate": 6.578257167187057e-05,
      "loss": 0.004530787467956543,
      "step": 120600
    },
    {
      "epoch": 34.23502696565427,
      "grad_norm": 9.359720230102539,
      "learning_rate": 6.577973318194721e-05,
      "loss": 0.01224035769701004,
      "step": 120610
    },
    {
      "epoch": 34.237865455577634,
      "grad_norm": 0.8083338141441345,
      "learning_rate": 6.577689469202385e-05,
      "loss": 0.009027906507253648,
      "step": 120620
    },
    {
      "epoch": 34.240703945501,
      "grad_norm": 9.791426658630371,
      "learning_rate": 6.577405620210049e-05,
      "loss": 0.0037582565099000933,
      "step": 120630
    },
    {
      "epoch": 34.24354243542435,
      "grad_norm": 0.20598530769348145,
      "learning_rate": 6.577121771217712e-05,
      "loss": 0.01041233241558075,
      "step": 120640
    },
    {
      "epoch": 34.246380925347715,
      "grad_norm": 0.8702098727226257,
      "learning_rate": 6.576837922225376e-05,
      "loss": 0.003046915866434574,
      "step": 120650
    },
    {
      "epoch": 34.24921941527108,
      "grad_norm": 0.41778644919395447,
      "learning_rate": 6.57655407323304e-05,
      "loss": 0.00889195129275322,
      "step": 120660
    },
    {
      "epoch": 34.252057905194434,
      "grad_norm": 0.037582848221063614,
      "learning_rate": 6.576270224240703e-05,
      "loss": 0.006374266743659973,
      "step": 120670
    },
    {
      "epoch": 34.2548963951178,
      "grad_norm": 1.0132782459259033,
      "learning_rate": 6.575986375248369e-05,
      "loss": 0.003450435400009155,
      "step": 120680
    },
    {
      "epoch": 34.25773488504116,
      "grad_norm": 0.14376351237297058,
      "learning_rate": 6.575702526256033e-05,
      "loss": 0.0028980176895856856,
      "step": 120690
    },
    {
      "epoch": 34.260573374964515,
      "grad_norm": 0.8079062700271606,
      "learning_rate": 6.575418677263696e-05,
      "loss": 0.015229898691177367,
      "step": 120700
    },
    {
      "epoch": 34.26341186488788,
      "grad_norm": 0.8749398589134216,
      "learning_rate": 6.57513482827136e-05,
      "loss": 0.007276272773742676,
      "step": 120710
    },
    {
      "epoch": 34.26625035481124,
      "grad_norm": 1.4335858821868896,
      "learning_rate": 6.574850979279024e-05,
      "loss": 0.0052808687090873715,
      "step": 120720
    },
    {
      "epoch": 34.269088844734604,
      "grad_norm": 1.8931385278701782,
      "learning_rate": 6.574567130286688e-05,
      "loss": 0.019981157779693604,
      "step": 120730
    },
    {
      "epoch": 34.27192733465796,
      "grad_norm": 0.4695318341255188,
      "learning_rate": 6.574283281294352e-05,
      "loss": 0.008774477988481522,
      "step": 120740
    },
    {
      "epoch": 34.27476582458132,
      "grad_norm": 0.08131548017263412,
      "learning_rate": 6.573999432302016e-05,
      "loss": 0.0059345878660678865,
      "step": 120750
    },
    {
      "epoch": 34.277604314504686,
      "grad_norm": 3.9901278018951416,
      "learning_rate": 6.57371558330968e-05,
      "loss": 0.003952443972229958,
      "step": 120760
    },
    {
      "epoch": 34.28044280442804,
      "grad_norm": 17.154191970825195,
      "learning_rate": 6.573431734317343e-05,
      "loss": 0.007435213774442673,
      "step": 120770
    },
    {
      "epoch": 34.283281294351404,
      "grad_norm": 0.6555381417274475,
      "learning_rate": 6.573147885325007e-05,
      "loss": 0.002556019090116024,
      "step": 120780
    },
    {
      "epoch": 34.28611978427477,
      "grad_norm": 0.37856563925743103,
      "learning_rate": 6.572864036332671e-05,
      "loss": 0.005672322213649749,
      "step": 120790
    },
    {
      "epoch": 34.28895827419813,
      "grad_norm": 0.2495548576116562,
      "learning_rate": 6.572580187340334e-05,
      "loss": 0.0071455128490924835,
      "step": 120800
    },
    {
      "epoch": 34.291796764121486,
      "grad_norm": 0.08386978507041931,
      "learning_rate": 6.572296338348e-05,
      "loss": 0.000952913612127304,
      "step": 120810
    },
    {
      "epoch": 34.29463525404485,
      "grad_norm": 0.1970139592885971,
      "learning_rate": 6.572012489355664e-05,
      "loss": 0.0030345387756824495,
      "step": 120820
    },
    {
      "epoch": 34.29747374396821,
      "grad_norm": 0.6047740578651428,
      "learning_rate": 6.571728640363327e-05,
      "loss": 0.0037004269659519197,
      "step": 120830
    },
    {
      "epoch": 34.30031223389157,
      "grad_norm": 0.25278300046920776,
      "learning_rate": 6.571444791370991e-05,
      "loss": 0.0041191868484020235,
      "step": 120840
    },
    {
      "epoch": 34.30315072381493,
      "grad_norm": 2.0211734771728516,
      "learning_rate": 6.571160942378655e-05,
      "loss": 0.004446744173765183,
      "step": 120850
    },
    {
      "epoch": 34.30598921373829,
      "grad_norm": 0.31450673937797546,
      "learning_rate": 6.570877093386319e-05,
      "loss": 0.0034073799848556518,
      "step": 120860
    },
    {
      "epoch": 34.30882770366165,
      "grad_norm": 0.10980676114559174,
      "learning_rate": 6.570593244393982e-05,
      "loss": 0.0017848830670118333,
      "step": 120870
    },
    {
      "epoch": 34.31166619358501,
      "grad_norm": 0.32644227147102356,
      "learning_rate": 6.570309395401647e-05,
      "loss": 0.004441918432712555,
      "step": 120880
    },
    {
      "epoch": 34.314504683508375,
      "grad_norm": 17.252668380737305,
      "learning_rate": 6.57002554640931e-05,
      "loss": 0.02058267742395401,
      "step": 120890
    },
    {
      "epoch": 34.31734317343174,
      "grad_norm": 0.3855568766593933,
      "learning_rate": 6.569741697416974e-05,
      "loss": 0.009337116777896882,
      "step": 120900
    },
    {
      "epoch": 34.32018166335509,
      "grad_norm": 5.581953525543213,
      "learning_rate": 6.569457848424638e-05,
      "loss": 0.007589308917522431,
      "step": 120910
    },
    {
      "epoch": 34.323020153278456,
      "grad_norm": 0.056428126990795135,
      "learning_rate": 6.569173999432303e-05,
      "loss": 0.004325633868575096,
      "step": 120920
    },
    {
      "epoch": 34.32585864320182,
      "grad_norm": 0.21463581919670105,
      "learning_rate": 6.568890150439965e-05,
      "loss": 0.005055137723684311,
      "step": 120930
    },
    {
      "epoch": 34.328697133125175,
      "grad_norm": 0.41009339690208435,
      "learning_rate": 6.568606301447631e-05,
      "loss": 0.004133815318346024,
      "step": 120940
    },
    {
      "epoch": 34.33153562304854,
      "grad_norm": 0.7506110668182373,
      "learning_rate": 6.568322452455295e-05,
      "loss": 0.006041218340396881,
      "step": 120950
    },
    {
      "epoch": 34.3343741129719,
      "grad_norm": 0.02781783416867256,
      "learning_rate": 6.568038603462958e-05,
      "loss": 0.0034258291125297546,
      "step": 120960
    },
    {
      "epoch": 34.33721260289526,
      "grad_norm": 1.2344199419021606,
      "learning_rate": 6.567754754470622e-05,
      "loss": 0.017416100203990936,
      "step": 120970
    },
    {
      "epoch": 34.34005109281862,
      "grad_norm": 0.4481377601623535,
      "learning_rate": 6.567470905478286e-05,
      "loss": 0.005416756868362427,
      "step": 120980
    },
    {
      "epoch": 34.34288958274198,
      "grad_norm": 8.962422370910645,
      "learning_rate": 6.567187056485949e-05,
      "loss": 0.0033637385815382,
      "step": 120990
    },
    {
      "epoch": 34.345728072665345,
      "grad_norm": 1.0761725902557373,
      "learning_rate": 6.566903207493613e-05,
      "loss": 0.010275451838970185,
      "step": 121000
    },
    {
      "epoch": 34.345728072665345,
      "eval_accuracy": 0.9714503719717683,
      "eval_loss": 0.09643729776144028,
      "eval_runtime": 32.301,
      "eval_samples_per_second": 486.888,
      "eval_steps_per_second": 7.616,
      "step": 121000
    },
    {
      "epoch": 34.3485665625887,
      "grad_norm": 3.3196048736572266,
      "learning_rate": 6.566619358501278e-05,
      "loss": 0.014252856373786926,
      "step": 121010
    },
    {
      "epoch": 34.351405052512064,
      "grad_norm": 0.10674888640642166,
      "learning_rate": 6.566335509508941e-05,
      "loss": 0.009489893168210983,
      "step": 121020
    },
    {
      "epoch": 34.35424354243543,
      "grad_norm": 12.795866966247559,
      "learning_rate": 6.566051660516605e-05,
      "loss": 0.01366342306137085,
      "step": 121030
    },
    {
      "epoch": 34.35708203235878,
      "grad_norm": 0.35392606258392334,
      "learning_rate": 6.56576781152427e-05,
      "loss": 0.002171451784670353,
      "step": 121040
    },
    {
      "epoch": 34.359920522282145,
      "grad_norm": 0.11029020696878433,
      "learning_rate": 6.565483962531934e-05,
      "loss": 0.010325114428997039,
      "step": 121050
    },
    {
      "epoch": 34.36275901220551,
      "grad_norm": 0.17874668538570404,
      "learning_rate": 6.565200113539596e-05,
      "loss": 0.010382829606533051,
      "step": 121060
    },
    {
      "epoch": 34.365597502128864,
      "grad_norm": 0.06169803813099861,
      "learning_rate": 6.564916264547262e-05,
      "loss": 0.002348296530544758,
      "step": 121070
    },
    {
      "epoch": 34.36843599205223,
      "grad_norm": 0.09570597112178802,
      "learning_rate": 6.564632415554926e-05,
      "loss": 0.007990719377994537,
      "step": 121080
    },
    {
      "epoch": 34.37127448197559,
      "grad_norm": 4.681507587432861,
      "learning_rate": 6.564348566562589e-05,
      "loss": 0.009579497575759887,
      "step": 121090
    },
    {
      "epoch": 34.37411297189895,
      "grad_norm": 5.019866943359375,
      "learning_rate": 6.564064717570253e-05,
      "loss": 0.005193861573934555,
      "step": 121100
    },
    {
      "epoch": 34.37695146182231,
      "grad_norm": 0.050225336104631424,
      "learning_rate": 6.563780868577917e-05,
      "loss": 0.008365646004676819,
      "step": 121110
    },
    {
      "epoch": 34.37978995174567,
      "grad_norm": 0.5290006399154663,
      "learning_rate": 6.56349701958558e-05,
      "loss": 0.005773923546075821,
      "step": 121120
    },
    {
      "epoch": 34.382628441669034,
      "grad_norm": 9.007638931274414,
      "learning_rate": 6.563213170593244e-05,
      "loss": 0.0076716005802154544,
      "step": 121130
    },
    {
      "epoch": 34.38546693159239,
      "grad_norm": 10.064278602600098,
      "learning_rate": 6.56292932160091e-05,
      "loss": 0.006720879673957824,
      "step": 121140
    },
    {
      "epoch": 34.38830542151575,
      "grad_norm": 5.260037899017334,
      "learning_rate": 6.562645472608572e-05,
      "loss": 0.002526211738586426,
      "step": 121150
    },
    {
      "epoch": 34.391143911439116,
      "grad_norm": 0.14219024777412415,
      "learning_rate": 6.562361623616236e-05,
      "loss": 0.00200135949999094,
      "step": 121160
    },
    {
      "epoch": 34.39398240136248,
      "grad_norm": 0.059102702885866165,
      "learning_rate": 6.5620777746239e-05,
      "loss": 0.001701488345861435,
      "step": 121170
    },
    {
      "epoch": 34.396820891285834,
      "grad_norm": 2.133500099182129,
      "learning_rate": 6.561793925631565e-05,
      "loss": 0.001616235449910164,
      "step": 121180
    },
    {
      "epoch": 34.3996593812092,
      "grad_norm": 3.6466965675354004,
      "learning_rate": 6.561510076639227e-05,
      "loss": 0.0028897901996970178,
      "step": 121190
    },
    {
      "epoch": 34.40249787113256,
      "grad_norm": 0.9116111993789673,
      "learning_rate": 6.561226227646892e-05,
      "loss": 0.010547797381877898,
      "step": 121200
    },
    {
      "epoch": 34.405336361055916,
      "grad_norm": 0.3015027642250061,
      "learning_rate": 6.560942378654557e-05,
      "loss": 0.00779331848025322,
      "step": 121210
    },
    {
      "epoch": 34.40817485097928,
      "grad_norm": 0.6132565140724182,
      "learning_rate": 6.56065852966222e-05,
      "loss": 0.0037642888724803924,
      "step": 121220
    },
    {
      "epoch": 34.41101334090264,
      "grad_norm": 8.30250072479248,
      "learning_rate": 6.560374680669884e-05,
      "loss": 0.01046159565448761,
      "step": 121230
    },
    {
      "epoch": 34.413851830826,
      "grad_norm": 0.05436651036143303,
      "learning_rate": 6.560090831677548e-05,
      "loss": 0.008311940729618073,
      "step": 121240
    },
    {
      "epoch": 34.41669032074936,
      "grad_norm": 0.21833334863185883,
      "learning_rate": 6.559806982685211e-05,
      "loss": 0.002599102072417736,
      "step": 121250
    },
    {
      "epoch": 34.41952881067272,
      "grad_norm": 3.46553373336792,
      "learning_rate": 6.559523133692875e-05,
      "loss": 0.0048821233212947845,
      "step": 121260
    },
    {
      "epoch": 34.422367300596086,
      "grad_norm": 0.17556701600551605,
      "learning_rate": 6.55923928470054e-05,
      "loss": 0.0015192406252026557,
      "step": 121270
    },
    {
      "epoch": 34.42520579051944,
      "grad_norm": 1.5152485370635986,
      "learning_rate": 6.558955435708203e-05,
      "loss": 0.0024162139743566513,
      "step": 121280
    },
    {
      "epoch": 34.428044280442805,
      "grad_norm": 0.16614964604377747,
      "learning_rate": 6.558671586715867e-05,
      "loss": 0.003900546580553055,
      "step": 121290
    },
    {
      "epoch": 34.43088277036617,
      "grad_norm": 0.1593964695930481,
      "learning_rate": 6.558387737723532e-05,
      "loss": 0.009399491548538207,
      "step": 121300
    },
    {
      "epoch": 34.43372126028952,
      "grad_norm": 0.5776618719100952,
      "learning_rate": 6.558103888731196e-05,
      "loss": 0.0017070671543478965,
      "step": 121310
    },
    {
      "epoch": 34.436559750212886,
      "grad_norm": 0.10168550163507462,
      "learning_rate": 6.557820039738859e-05,
      "loss": 0.014530080556869506,
      "step": 121320
    },
    {
      "epoch": 34.43939824013625,
      "grad_norm": 0.027144042775034904,
      "learning_rate": 6.557536190746523e-05,
      "loss": 0.002441391907632351,
      "step": 121330
    },
    {
      "epoch": 34.442236730059605,
      "grad_norm": 3.639847755432129,
      "learning_rate": 6.557252341754188e-05,
      "loss": 0.009512382745742797,
      "step": 121340
    },
    {
      "epoch": 34.44507521998297,
      "grad_norm": 1.7980903387069702,
      "learning_rate": 6.556968492761851e-05,
      "loss": 0.014887143671512604,
      "step": 121350
    },
    {
      "epoch": 34.44791370990633,
      "grad_norm": 0.637545108795166,
      "learning_rate": 6.556684643769515e-05,
      "loss": 0.0011794516816735269,
      "step": 121360
    },
    {
      "epoch": 34.450752199829694,
      "grad_norm": 0.6996497511863708,
      "learning_rate": 6.556400794777179e-05,
      "loss": 0.004240016639232636,
      "step": 121370
    },
    {
      "epoch": 34.45359068975305,
      "grad_norm": 0.428958922624588,
      "learning_rate": 6.556116945784842e-05,
      "loss": 0.002185925841331482,
      "step": 121380
    },
    {
      "epoch": 34.45642917967641,
      "grad_norm": 2.982541561126709,
      "learning_rate": 6.555833096792506e-05,
      "loss": 0.0016149669885635377,
      "step": 121390
    },
    {
      "epoch": 34.459267669599775,
      "grad_norm": 15.298772811889648,
      "learning_rate": 6.55554924780017e-05,
      "loss": 0.007458152621984482,
      "step": 121400
    },
    {
      "epoch": 34.46210615952313,
      "grad_norm": 0.24099060893058777,
      "learning_rate": 6.555265398807834e-05,
      "loss": 0.001996976509690285,
      "step": 121410
    },
    {
      "epoch": 34.464944649446494,
      "grad_norm": 0.44171348214149475,
      "learning_rate": 6.554981549815499e-05,
      "loss": 0.003290959447622299,
      "step": 121420
    },
    {
      "epoch": 34.46778313936986,
      "grad_norm": 0.19654907286167145,
      "learning_rate": 6.554697700823163e-05,
      "loss": 0.01628538519144058,
      "step": 121430
    },
    {
      "epoch": 34.47062162929321,
      "grad_norm": 0.10233055055141449,
      "learning_rate": 6.554413851830827e-05,
      "loss": 0.004568058997392654,
      "step": 121440
    },
    {
      "epoch": 34.473460119216575,
      "grad_norm": 0.11782721430063248,
      "learning_rate": 6.55413000283849e-05,
      "loss": 0.0028743093833327295,
      "step": 121450
    },
    {
      "epoch": 34.47629860913994,
      "grad_norm": 0.24863453209400177,
      "learning_rate": 6.553846153846154e-05,
      "loss": 0.0036957345902919768,
      "step": 121460
    },
    {
      "epoch": 34.4791370990633,
      "grad_norm": 8.27115249633789,
      "learning_rate": 6.553562304853819e-05,
      "loss": 0.003077494539320469,
      "step": 121470
    },
    {
      "epoch": 34.48197558898666,
      "grad_norm": 0.6603391766548157,
      "learning_rate": 6.553278455861482e-05,
      "loss": 0.010180327296257018,
      "step": 121480
    },
    {
      "epoch": 34.48481407891002,
      "grad_norm": 1.912031888961792,
      "learning_rate": 6.552994606869146e-05,
      "loss": 0.003331070765852928,
      "step": 121490
    },
    {
      "epoch": 34.48765256883338,
      "grad_norm": 0.8125287294387817,
      "learning_rate": 6.55271075787681e-05,
      "loss": 0.005793993920087814,
      "step": 121500
    },
    {
      "epoch": 34.48765256883338,
      "eval_accuracy": 0.9713232021364532,
      "eval_loss": 0.09893113374710083,
      "eval_runtime": 32.1117,
      "eval_samples_per_second": 489.759,
      "eval_steps_per_second": 7.661,
      "step": 121500
    },
    {
      "epoch": 34.49049105875674,
      "grad_norm": 0.4688956141471863,
      "learning_rate": 6.552426908884473e-05,
      "loss": 0.005483905225992203,
      "step": 121510
    },
    {
      "epoch": 34.4933295486801,
      "grad_norm": 4.4383087158203125,
      "learning_rate": 6.552143059892137e-05,
      "loss": 0.005467564612627029,
      "step": 121520
    },
    {
      "epoch": 34.496168038603464,
      "grad_norm": 0.3712189197540283,
      "learning_rate": 6.551859210899801e-05,
      "loss": 0.008706380426883698,
      "step": 121530
    },
    {
      "epoch": 34.49900652852683,
      "grad_norm": 0.15487441420555115,
      "learning_rate": 6.551575361907465e-05,
      "loss": 0.007328922301530838,
      "step": 121540
    },
    {
      "epoch": 34.50184501845018,
      "grad_norm": 0.30117157101631165,
      "learning_rate": 6.55129151291513e-05,
      "loss": 0.006180235743522644,
      "step": 121550
    },
    {
      "epoch": 34.504683508373546,
      "grad_norm": 0.16331055760383606,
      "learning_rate": 6.551007663922794e-05,
      "loss": 0.004076334461569786,
      "step": 121560
    },
    {
      "epoch": 34.50752199829691,
      "grad_norm": 9.480448722839355,
      "learning_rate": 6.550723814930458e-05,
      "loss": 0.010015219449996948,
      "step": 121570
    },
    {
      "epoch": 34.510360488220265,
      "grad_norm": 0.07059287279844284,
      "learning_rate": 6.550439965938121e-05,
      "loss": 0.004824448376893997,
      "step": 121580
    },
    {
      "epoch": 34.51319897814363,
      "grad_norm": 1.363734483718872,
      "learning_rate": 6.550156116945785e-05,
      "loss": 0.0052822701632976535,
      "step": 121590
    },
    {
      "epoch": 34.51603746806699,
      "grad_norm": 26.711833953857422,
      "learning_rate": 6.549872267953449e-05,
      "loss": 0.012368667870759964,
      "step": 121600
    },
    {
      "epoch": 34.518875957990346,
      "grad_norm": 15.351613998413086,
      "learning_rate": 6.549588418961113e-05,
      "loss": 0.012082774937152863,
      "step": 121610
    },
    {
      "epoch": 34.52171444791371,
      "grad_norm": 1.0990090370178223,
      "learning_rate": 6.549304569968777e-05,
      "loss": 0.0031744129955768585,
      "step": 121620
    },
    {
      "epoch": 34.52455293783707,
      "grad_norm": 2.9962265491485596,
      "learning_rate": 6.549020720976441e-05,
      "loss": 0.003898720443248749,
      "step": 121630
    },
    {
      "epoch": 34.527391427760435,
      "grad_norm": 0.008796045556664467,
      "learning_rate": 6.548736871984104e-05,
      "loss": 0.012318804860115051,
      "step": 121640
    },
    {
      "epoch": 34.53022991768379,
      "grad_norm": 2.5744450092315674,
      "learning_rate": 6.548453022991768e-05,
      "loss": 0.007464762777090073,
      "step": 121650
    },
    {
      "epoch": 34.53306840760715,
      "grad_norm": 4.2871904373168945,
      "learning_rate": 6.548169173999432e-05,
      "loss": 0.0058799199759960175,
      "step": 121660
    },
    {
      "epoch": 34.535906897530516,
      "grad_norm": 1.9031122922897339,
      "learning_rate": 6.547885325007097e-05,
      "loss": 0.010009279847145081,
      "step": 121670
    },
    {
      "epoch": 34.53874538745387,
      "grad_norm": 2.049429416656494,
      "learning_rate": 6.547601476014761e-05,
      "loss": 0.00995267480611801,
      "step": 121680
    },
    {
      "epoch": 34.541583877377235,
      "grad_norm": 2.6767995357513428,
      "learning_rate": 6.547317627022425e-05,
      "loss": 0.003879908472299576,
      "step": 121690
    },
    {
      "epoch": 34.5444223673006,
      "grad_norm": 3.3350377082824707,
      "learning_rate": 6.547033778030089e-05,
      "loss": 0.004507974535226822,
      "step": 121700
    },
    {
      "epoch": 34.547260857223954,
      "grad_norm": 0.21623603999614716,
      "learning_rate": 6.546749929037752e-05,
      "loss": 0.003957278281450272,
      "step": 121710
    },
    {
      "epoch": 34.55009934714732,
      "grad_norm": 0.571140706539154,
      "learning_rate": 6.546466080045416e-05,
      "loss": 0.00429561510682106,
      "step": 121720
    },
    {
      "epoch": 34.55293783707068,
      "grad_norm": 1.5757650136947632,
      "learning_rate": 6.54618223105308e-05,
      "loss": 0.004278173297643661,
      "step": 121730
    },
    {
      "epoch": 34.55577632699404,
      "grad_norm": 0.3858022391796112,
      "learning_rate": 6.545898382060744e-05,
      "loss": 0.003992342948913574,
      "step": 121740
    },
    {
      "epoch": 34.5586148169174,
      "grad_norm": 0.06836457550525665,
      "learning_rate": 6.545614533068408e-05,
      "loss": 0.004588346928358078,
      "step": 121750
    },
    {
      "epoch": 34.56145330684076,
      "grad_norm": 2.7228176593780518,
      "learning_rate": 6.545330684076072e-05,
      "loss": 0.00953756421804428,
      "step": 121760
    },
    {
      "epoch": 34.564291796764124,
      "grad_norm": 12.314213752746582,
      "learning_rate": 6.545046835083735e-05,
      "loss": 0.006161943078041077,
      "step": 121770
    },
    {
      "epoch": 34.56713028668748,
      "grad_norm": 2.7650938034057617,
      "learning_rate": 6.5447629860914e-05,
      "loss": 0.01035730168223381,
      "step": 121780
    },
    {
      "epoch": 34.56996877661084,
      "grad_norm": 0.4451490342617035,
      "learning_rate": 6.544479137099064e-05,
      "loss": 0.009587416052818298,
      "step": 121790
    },
    {
      "epoch": 34.572807266534205,
      "grad_norm": 9.189227104187012,
      "learning_rate": 6.544195288106728e-05,
      "loss": 0.006797210872173309,
      "step": 121800
    },
    {
      "epoch": 34.57564575645756,
      "grad_norm": 5.982112884521484,
      "learning_rate": 6.543911439114392e-05,
      "loss": 0.005052819848060608,
      "step": 121810
    },
    {
      "epoch": 34.578484246380924,
      "grad_norm": 4.866293430328369,
      "learning_rate": 6.543627590122056e-05,
      "loss": 0.004836326092481613,
      "step": 121820
    },
    {
      "epoch": 34.58132273630429,
      "grad_norm": 1.0301543474197388,
      "learning_rate": 6.543343741129719e-05,
      "loss": 0.0048585444688797,
      "step": 121830
    },
    {
      "epoch": 34.58416122622765,
      "grad_norm": 8.754075050354004,
      "learning_rate": 6.543059892137383e-05,
      "loss": 0.008437854051589967,
      "step": 121840
    },
    {
      "epoch": 34.586999716151006,
      "grad_norm": 0.020057469606399536,
      "learning_rate": 6.542776043145047e-05,
      "loss": 0.007250663638114929,
      "step": 121850
    },
    {
      "epoch": 34.58983820607437,
      "grad_norm": 2.0008208751678467,
      "learning_rate": 6.542492194152711e-05,
      "loss": 0.018709914386272432,
      "step": 121860
    },
    {
      "epoch": 34.59267669599773,
      "grad_norm": 0.41609352827072144,
      "learning_rate": 6.542208345160375e-05,
      "loss": 0.002455700933933258,
      "step": 121870
    },
    {
      "epoch": 34.59551518592109,
      "grad_norm": 0.6780348420143127,
      "learning_rate": 6.54192449616804e-05,
      "loss": 0.002100434899330139,
      "step": 121880
    },
    {
      "epoch": 34.59835367584445,
      "grad_norm": 0.28256723284721375,
      "learning_rate": 6.541640647175704e-05,
      "loss": 0.008593764156103134,
      "step": 121890
    },
    {
      "epoch": 34.60119216576781,
      "grad_norm": 12.566946029663086,
      "learning_rate": 6.541356798183366e-05,
      "loss": 0.011824636161327362,
      "step": 121900
    },
    {
      "epoch": 34.604030655691176,
      "grad_norm": 0.7104384899139404,
      "learning_rate": 6.54107294919103e-05,
      "loss": 0.006410237401723862,
      "step": 121910
    },
    {
      "epoch": 34.60686914561453,
      "grad_norm": 0.7847989201545715,
      "learning_rate": 6.540789100198695e-05,
      "loss": 0.002907312475144863,
      "step": 121920
    },
    {
      "epoch": 34.609707635537895,
      "grad_norm": 0.4069633185863495,
      "learning_rate": 6.540505251206357e-05,
      "loss": 0.005336733162403106,
      "step": 121930
    },
    {
      "epoch": 34.61254612546126,
      "grad_norm": 0.8873326182365417,
      "learning_rate": 6.540221402214023e-05,
      "loss": 0.003549569100141525,
      "step": 121940
    },
    {
      "epoch": 34.61538461538461,
      "grad_norm": 0.24483853578567505,
      "learning_rate": 6.539937553221687e-05,
      "loss": 0.0016845526173710824,
      "step": 121950
    },
    {
      "epoch": 34.618223105307976,
      "grad_norm": 1.4285119771957397,
      "learning_rate": 6.53965370422935e-05,
      "loss": 0.002040030062198639,
      "step": 121960
    },
    {
      "epoch": 34.62106159523134,
      "grad_norm": 0.49573495984077454,
      "learning_rate": 6.539369855237014e-05,
      "loss": 0.005358568951487541,
      "step": 121970
    },
    {
      "epoch": 34.623900085154695,
      "grad_norm": 0.16889546811580658,
      "learning_rate": 6.539086006244678e-05,
      "loss": 0.001979339122772217,
      "step": 121980
    },
    {
      "epoch": 34.62673857507806,
      "grad_norm": 0.19205884635448456,
      "learning_rate": 6.538802157252342e-05,
      "loss": 0.00336933434009552,
      "step": 121990
    },
    {
      "epoch": 34.62957706500142,
      "grad_norm": 1.6046061515808105,
      "learning_rate": 6.538518308260005e-05,
      "loss": 0.004387809336185456,
      "step": 122000
    },
    {
      "epoch": 34.62957706500142,
      "eval_accuracy": 0.9740573535957271,
      "eval_loss": 0.08764299005270004,
      "eval_runtime": 32.2426,
      "eval_samples_per_second": 487.771,
      "eval_steps_per_second": 7.63,
      "step": 122000
    },
    {
      "epoch": 34.63241555492478,
      "grad_norm": 0.08497347682714462,
      "learning_rate": 6.53823445926767e-05,
      "loss": 0.002357286959886551,
      "step": 122010
    },
    {
      "epoch": 34.63525404484814,
      "grad_norm": 0.41470205783843994,
      "learning_rate": 6.537950610275335e-05,
      "loss": 0.0036878764629364015,
      "step": 122020
    },
    {
      "epoch": 34.6380925347715,
      "grad_norm": 0.4962767958641052,
      "learning_rate": 6.537666761282997e-05,
      "loss": 0.008641448616981507,
      "step": 122030
    },
    {
      "epoch": 34.640931024694865,
      "grad_norm": 2.749258518218994,
      "learning_rate": 6.537382912290662e-05,
      "loss": 0.00379687137901783,
      "step": 122040
    },
    {
      "epoch": 34.64376951461822,
      "grad_norm": 1.0091066360473633,
      "learning_rate": 6.537099063298326e-05,
      "loss": 0.002864320017397404,
      "step": 122050
    },
    {
      "epoch": 34.646608004541584,
      "grad_norm": 0.385221928358078,
      "learning_rate": 6.536815214305988e-05,
      "loss": 0.004537220299243927,
      "step": 122060
    },
    {
      "epoch": 34.64944649446495,
      "grad_norm": 3.0260488986968994,
      "learning_rate": 6.536531365313654e-05,
      "loss": 0.006437567621469497,
      "step": 122070
    },
    {
      "epoch": 34.6522849843883,
      "grad_norm": 4.358546257019043,
      "learning_rate": 6.536247516321318e-05,
      "loss": 0.004372327029705048,
      "step": 122080
    },
    {
      "epoch": 34.655123474311665,
      "grad_norm": 1.408868432044983,
      "learning_rate": 6.535963667328981e-05,
      "loss": 0.010747304558753968,
      "step": 122090
    },
    {
      "epoch": 34.65796196423503,
      "grad_norm": 0.19077467918395996,
      "learning_rate": 6.535679818336645e-05,
      "loss": 0.011761073768138886,
      "step": 122100
    },
    {
      "epoch": 34.66080045415839,
      "grad_norm": 0.2147844433784485,
      "learning_rate": 6.535395969344309e-05,
      "loss": 0.0072051376104354855,
      "step": 122110
    },
    {
      "epoch": 34.66363894408175,
      "grad_norm": 0.6149144172668457,
      "learning_rate": 6.535112120351973e-05,
      "loss": 0.0058714568614959715,
      "step": 122120
    },
    {
      "epoch": 34.66647743400511,
      "grad_norm": 0.11503326892852783,
      "learning_rate": 6.534828271359636e-05,
      "loss": 0.0031664468348026275,
      "step": 122130
    },
    {
      "epoch": 34.66931592392847,
      "grad_norm": 0.7054546475410461,
      "learning_rate": 6.534544422367302e-05,
      "loss": 0.002845547907054424,
      "step": 122140
    },
    {
      "epoch": 34.67215441385183,
      "grad_norm": 0.1965297907590866,
      "learning_rate": 6.534260573374966e-05,
      "loss": 0.0036231093108654022,
      "step": 122150
    },
    {
      "epoch": 34.67499290377519,
      "grad_norm": 11.49445629119873,
      "learning_rate": 6.533976724382628e-05,
      "loss": 0.003418346494436264,
      "step": 122160
    },
    {
      "epoch": 34.677831393698554,
      "grad_norm": 9.815134048461914,
      "learning_rate": 6.533692875390293e-05,
      "loss": 0.014207223057746887,
      "step": 122170
    },
    {
      "epoch": 34.68066988362191,
      "grad_norm": 1.6379940509796143,
      "learning_rate": 6.533409026397957e-05,
      "loss": 0.004850973933935165,
      "step": 122180
    },
    {
      "epoch": 34.68350837354527,
      "grad_norm": 0.38227379322052,
      "learning_rate": 6.53312517740562e-05,
      "loss": 0.003502725064754486,
      "step": 122190
    },
    {
      "epoch": 34.686346863468636,
      "grad_norm": 0.032944899052381516,
      "learning_rate": 6.532841328413285e-05,
      "loss": 0.001232980191707611,
      "step": 122200
    },
    {
      "epoch": 34.689185353392,
      "grad_norm": 1.1862001419067383,
      "learning_rate": 6.532557479420949e-05,
      "loss": 0.006770524382591248,
      "step": 122210
    },
    {
      "epoch": 34.692023843315354,
      "grad_norm": 1.5211012363433838,
      "learning_rate": 6.532273630428612e-05,
      "loss": 0.003845862299203873,
      "step": 122220
    },
    {
      "epoch": 34.69486233323872,
      "grad_norm": 0.07238578796386719,
      "learning_rate": 6.531989781436276e-05,
      "loss": 0.003885899484157562,
      "step": 122230
    },
    {
      "epoch": 34.69770082316208,
      "grad_norm": 0.0045026689767837524,
      "learning_rate": 6.53170593244394e-05,
      "loss": 0.0027515754103660584,
      "step": 122240
    },
    {
      "epoch": 34.700539313085436,
      "grad_norm": 0.07014446705579758,
      "learning_rate": 6.531422083451604e-05,
      "loss": 0.005787523835897446,
      "step": 122250
    },
    {
      "epoch": 34.7033778030088,
      "grad_norm": 9.372686386108398,
      "learning_rate": 6.531138234459267e-05,
      "loss": 0.008054859936237335,
      "step": 122260
    },
    {
      "epoch": 34.70621629293216,
      "grad_norm": 9.588811874389648,
      "learning_rate": 6.530854385466933e-05,
      "loss": 0.004153379052877426,
      "step": 122270
    },
    {
      "epoch": 34.70905478285552,
      "grad_norm": 0.18317072093486786,
      "learning_rate": 6.530570536474597e-05,
      "loss": 0.007642856240272522,
      "step": 122280
    },
    {
      "epoch": 34.71189327277888,
      "grad_norm": 1.1680307388305664,
      "learning_rate": 6.53028668748226e-05,
      "loss": 0.005600081384181976,
      "step": 122290
    },
    {
      "epoch": 34.71473176270224,
      "grad_norm": 0.47032883763313293,
      "learning_rate": 6.530002838489924e-05,
      "loss": 0.0019518153741955758,
      "step": 122300
    },
    {
      "epoch": 34.717570252625606,
      "grad_norm": 0.17898622155189514,
      "learning_rate": 6.529718989497588e-05,
      "loss": 0.004440861940383911,
      "step": 122310
    },
    {
      "epoch": 34.72040874254896,
      "grad_norm": 0.1917894184589386,
      "learning_rate": 6.52943514050525e-05,
      "loss": 0.006739561259746551,
      "step": 122320
    },
    {
      "epoch": 34.723247232472325,
      "grad_norm": 0.13850022852420807,
      "learning_rate": 6.529151291512915e-05,
      "loss": 0.010078959912061692,
      "step": 122330
    },
    {
      "epoch": 34.72608572239569,
      "grad_norm": 0.04218889772891998,
      "learning_rate": 6.52886744252058e-05,
      "loss": 0.010856794565916062,
      "step": 122340
    },
    {
      "epoch": 34.72892421231904,
      "grad_norm": 3.785734176635742,
      "learning_rate": 6.528583593528243e-05,
      "loss": 0.004300510510802269,
      "step": 122350
    },
    {
      "epoch": 34.731762702242406,
      "grad_norm": 1.9344936609268188,
      "learning_rate": 6.528299744535907e-05,
      "loss": 0.004588557034730911,
      "step": 122360
    },
    {
      "epoch": 34.73460119216577,
      "grad_norm": 0.8153248429298401,
      "learning_rate": 6.528015895543571e-05,
      "loss": 0.003909099847078323,
      "step": 122370
    },
    {
      "epoch": 34.73743968208913,
      "grad_norm": 2.579099178314209,
      "learning_rate": 6.527732046551235e-05,
      "loss": 0.00475742518901825,
      "step": 122380
    },
    {
      "epoch": 34.74027817201249,
      "grad_norm": 0.02742755599319935,
      "learning_rate": 6.527448197558898e-05,
      "loss": 0.0064716808497905735,
      "step": 122390
    },
    {
      "epoch": 34.74311666193585,
      "grad_norm": 5.555974960327148,
      "learning_rate": 6.527164348566564e-05,
      "loss": 0.008911204338073731,
      "step": 122400
    },
    {
      "epoch": 34.745955151859214,
      "grad_norm": 0.40832000970840454,
      "learning_rate": 6.526880499574228e-05,
      "loss": 0.006559894233942032,
      "step": 122410
    },
    {
      "epoch": 34.74879364178257,
      "grad_norm": 0.07574240863323212,
      "learning_rate": 6.52659665058189e-05,
      "loss": 0.007863300293684006,
      "step": 122420
    },
    {
      "epoch": 34.75163213170593,
      "grad_norm": 28.71633529663086,
      "learning_rate": 6.526312801589555e-05,
      "loss": 0.02056107670068741,
      "step": 122430
    },
    {
      "epoch": 34.754470621629295,
      "grad_norm": 0.052262160927057266,
      "learning_rate": 6.526028952597219e-05,
      "loss": 0.00662291944026947,
      "step": 122440
    },
    {
      "epoch": 34.75730911155265,
      "grad_norm": 0.5298376679420471,
      "learning_rate": 6.525745103604882e-05,
      "loss": 0.010420435667037964,
      "step": 122450
    },
    {
      "epoch": 34.760147601476014,
      "grad_norm": 0.2313220500946045,
      "learning_rate": 6.525461254612546e-05,
      "loss": 0.008924816548824311,
      "step": 122460
    },
    {
      "epoch": 34.76298609139938,
      "grad_norm": 0.2534920871257782,
      "learning_rate": 6.525177405620211e-05,
      "loss": 0.0008092386648058892,
      "step": 122470
    },
    {
      "epoch": 34.76582458132274,
      "grad_norm": 0.7692059278488159,
      "learning_rate": 6.524893556627874e-05,
      "loss": 0.004016993939876557,
      "step": 122480
    },
    {
      "epoch": 34.768663071246095,
      "grad_norm": 1.4794586896896362,
      "learning_rate": 6.524609707635538e-05,
      "loss": 0.012088869512081147,
      "step": 122490
    },
    {
      "epoch": 34.77150156116946,
      "grad_norm": 6.549243927001953,
      "learning_rate": 6.524325858643202e-05,
      "loss": 0.0027276264503598214,
      "step": 122500
    },
    {
      "epoch": 34.77150156116946,
      "eval_accuracy": 0.9693520696890697,
      "eval_loss": 0.10795033723115921,
      "eval_runtime": 31.7387,
      "eval_samples_per_second": 495.515,
      "eval_steps_per_second": 7.751,
      "step": 122500
    },
    {
      "epoch": 34.77434005109282,
      "grad_norm": 0.8010633587837219,
      "learning_rate": 6.524042009650867e-05,
      "loss": 0.006794467568397522,
      "step": 122510
    },
    {
      "epoch": 34.77717854101618,
      "grad_norm": 0.5210892558097839,
      "learning_rate": 6.523758160658529e-05,
      "loss": 0.002064274065196514,
      "step": 122520
    },
    {
      "epoch": 34.78001703093954,
      "grad_norm": 0.38127318024635315,
      "learning_rate": 6.523474311666193e-05,
      "loss": 0.006865138560533524,
      "step": 122530
    },
    {
      "epoch": 34.7828555208629,
      "grad_norm": 0.23486526310443878,
      "learning_rate": 6.523190462673859e-05,
      "loss": 0.005589088797569275,
      "step": 122540
    },
    {
      "epoch": 34.78569401078626,
      "grad_norm": 0.14956527948379517,
      "learning_rate": 6.522906613681522e-05,
      "loss": 0.008493191003799439,
      "step": 122550
    },
    {
      "epoch": 34.78853250070962,
      "grad_norm": 8.225197792053223,
      "learning_rate": 6.522622764689186e-05,
      "loss": 0.009365185350179672,
      "step": 122560
    },
    {
      "epoch": 34.791370990632984,
      "grad_norm": 0.6843498349189758,
      "learning_rate": 6.52233891569685e-05,
      "loss": 0.003148317337036133,
      "step": 122570
    },
    {
      "epoch": 34.79420948055635,
      "grad_norm": 0.058947645127773285,
      "learning_rate": 6.522055066704513e-05,
      "loss": 0.006049583852291107,
      "step": 122580
    },
    {
      "epoch": 34.7970479704797,
      "grad_norm": 0.07899049669504166,
      "learning_rate": 6.521771217712177e-05,
      "loss": 0.00302484966814518,
      "step": 122590
    },
    {
      "epoch": 34.799886460403066,
      "grad_norm": 0.4935496151447296,
      "learning_rate": 6.521487368719842e-05,
      "loss": 0.006386324763298035,
      "step": 122600
    },
    {
      "epoch": 34.80272495032643,
      "grad_norm": 3.368997097015381,
      "learning_rate": 6.521203519727505e-05,
      "loss": 0.0053942278027534485,
      "step": 122610
    },
    {
      "epoch": 34.805563440249784,
      "grad_norm": 0.04686836898326874,
      "learning_rate": 6.520919670735169e-05,
      "loss": 0.013360641896724701,
      "step": 122620
    },
    {
      "epoch": 34.80840193017315,
      "grad_norm": 0.2947852909564972,
      "learning_rate": 6.520635821742833e-05,
      "loss": 0.006451833248138428,
      "step": 122630
    },
    {
      "epoch": 34.81124042009651,
      "grad_norm": 4.3377861976623535,
      "learning_rate": 6.520351972750498e-05,
      "loss": 0.012702731788158417,
      "step": 122640
    },
    {
      "epoch": 34.814078910019866,
      "grad_norm": 0.4195486605167389,
      "learning_rate": 6.52006812375816e-05,
      "loss": 0.0025812866166234016,
      "step": 122650
    },
    {
      "epoch": 34.81691739994323,
      "grad_norm": 0.5187996625900269,
      "learning_rate": 6.519784274765824e-05,
      "loss": 0.0030008174479007723,
      "step": 122660
    },
    {
      "epoch": 34.81975588986659,
      "grad_norm": 0.20222580432891846,
      "learning_rate": 6.519500425773489e-05,
      "loss": 0.007525639235973358,
      "step": 122670
    },
    {
      "epoch": 34.822594379789955,
      "grad_norm": 2.2353641986846924,
      "learning_rate": 6.519216576781153e-05,
      "loss": 0.0022400256246328356,
      "step": 122680
    },
    {
      "epoch": 34.82543286971331,
      "grad_norm": 0.03016316331923008,
      "learning_rate": 6.518932727788817e-05,
      "loss": 0.0030402921140193937,
      "step": 122690
    },
    {
      "epoch": 34.82827135963667,
      "grad_norm": 0.7809737324714661,
      "learning_rate": 6.518648878796481e-05,
      "loss": 0.0027246598154306413,
      "step": 122700
    },
    {
      "epoch": 34.831109849560036,
      "grad_norm": 0.3580397665500641,
      "learning_rate": 6.518365029804144e-05,
      "loss": 0.0029597712680697443,
      "step": 122710
    },
    {
      "epoch": 34.83394833948339,
      "grad_norm": 1.078988790512085,
      "learning_rate": 6.518081180811808e-05,
      "loss": 0.008649495244026185,
      "step": 122720
    },
    {
      "epoch": 34.836786829406755,
      "grad_norm": 2.592554807662964,
      "learning_rate": 6.517797331819472e-05,
      "loss": 0.00869375467300415,
      "step": 122730
    },
    {
      "epoch": 34.83962531933012,
      "grad_norm": 1.322304129600525,
      "learning_rate": 6.517513482827136e-05,
      "loss": 0.0024959685280919073,
      "step": 122740
    },
    {
      "epoch": 34.84246380925348,
      "grad_norm": 13.843086242675781,
      "learning_rate": 6.5172296338348e-05,
      "loss": 0.009198613464832306,
      "step": 122750
    },
    {
      "epoch": 34.845302299176836,
      "grad_norm": 1.5036993026733398,
      "learning_rate": 6.516945784842465e-05,
      "loss": 0.006504300981760025,
      "step": 122760
    },
    {
      "epoch": 34.8481407891002,
      "grad_norm": 1.190007209777832,
      "learning_rate": 6.516661935850127e-05,
      "loss": 0.004325412586331368,
      "step": 122770
    },
    {
      "epoch": 34.85097927902356,
      "grad_norm": 4.4854230880737305,
      "learning_rate": 6.516378086857791e-05,
      "loss": 0.007730226218700409,
      "step": 122780
    },
    {
      "epoch": 34.85381776894692,
      "grad_norm": 0.6568772792816162,
      "learning_rate": 6.516094237865456e-05,
      "loss": 0.005169256776571274,
      "step": 122790
    },
    {
      "epoch": 34.85665625887028,
      "grad_norm": 0.49595290422439575,
      "learning_rate": 6.51581038887312e-05,
      "loss": 0.004600518941879272,
      "step": 122800
    },
    {
      "epoch": 34.859494748793644,
      "grad_norm": 1.6020830869674683,
      "learning_rate": 6.515526539880784e-05,
      "loss": 0.005419518053531647,
      "step": 122810
    },
    {
      "epoch": 34.862333238717,
      "grad_norm": 9.591131210327148,
      "learning_rate": 6.515242690888448e-05,
      "loss": 0.008065285533666611,
      "step": 122820
    },
    {
      "epoch": 34.86517172864036,
      "grad_norm": 5.1198248863220215,
      "learning_rate": 6.514958841896112e-05,
      "loss": 0.0038939177989959716,
      "step": 122830
    },
    {
      "epoch": 34.868010218563725,
      "grad_norm": 0.11730759590864182,
      "learning_rate": 6.514674992903775e-05,
      "loss": 0.00441715270280838,
      "step": 122840
    },
    {
      "epoch": 34.87084870848709,
      "grad_norm": 0.3771185576915741,
      "learning_rate": 6.514391143911439e-05,
      "loss": 0.010656940191984177,
      "step": 122850
    },
    {
      "epoch": 34.873687198410444,
      "grad_norm": 0.5060539245605469,
      "learning_rate": 6.514107294919103e-05,
      "loss": 0.007317967712879181,
      "step": 122860
    },
    {
      "epoch": 34.87652568833381,
      "grad_norm": 1.1179192066192627,
      "learning_rate": 6.513823445926767e-05,
      "loss": 0.0019321886822581292,
      "step": 122870
    },
    {
      "epoch": 34.87936417825717,
      "grad_norm": 0.1301707625389099,
      "learning_rate": 6.513539596934431e-05,
      "loss": 0.002627173252403736,
      "step": 122880
    },
    {
      "epoch": 34.882202668180526,
      "grad_norm": 0.4482869505882263,
      "learning_rate": 6.513255747942096e-05,
      "loss": 0.0059134714305400845,
      "step": 122890
    },
    {
      "epoch": 34.88504115810389,
      "grad_norm": 1.2510905265808105,
      "learning_rate": 6.512971898949758e-05,
      "loss": 0.004336058348417282,
      "step": 122900
    },
    {
      "epoch": 34.88787964802725,
      "grad_norm": 0.43234360218048096,
      "learning_rate": 6.512688049957423e-05,
      "loss": 0.0222098246216774,
      "step": 122910
    },
    {
      "epoch": 34.89071813795061,
      "grad_norm": 6.758641242980957,
      "learning_rate": 6.512404200965087e-05,
      "loss": 0.004389127343893051,
      "step": 122920
    },
    {
      "epoch": 34.89355662787397,
      "grad_norm": 0.5900225043296814,
      "learning_rate": 6.512120351972751e-05,
      "loss": 0.014180150628089905,
      "step": 122930
    },
    {
      "epoch": 34.89639511779733,
      "grad_norm": 2.0587778091430664,
      "learning_rate": 6.511836502980415e-05,
      "loss": 0.004373796284198761,
      "step": 122940
    },
    {
      "epoch": 34.899233607720696,
      "grad_norm": 2.4004077911376953,
      "learning_rate": 6.511552653988079e-05,
      "loss": 0.007106940448284149,
      "step": 122950
    },
    {
      "epoch": 34.90207209764405,
      "grad_norm": 3.9666500091552734,
      "learning_rate": 6.511268804995743e-05,
      "loss": 0.00499482974410057,
      "step": 122960
    },
    {
      "epoch": 34.904910587567414,
      "grad_norm": 5.166602611541748,
      "learning_rate": 6.510984956003406e-05,
      "loss": 0.0032051924616098406,
      "step": 122970
    },
    {
      "epoch": 34.90774907749078,
      "grad_norm": 2.019221544265747,
      "learning_rate": 6.51070110701107e-05,
      "loss": 0.014046119153499603,
      "step": 122980
    },
    {
      "epoch": 34.91058756741413,
      "grad_norm": 0.0988413617014885,
      "learning_rate": 6.510417258018734e-05,
      "loss": 0.006494200229644776,
      "step": 122990
    },
    {
      "epoch": 34.913426057337496,
      "grad_norm": 0.32847774028778076,
      "learning_rate": 6.510133409026398e-05,
      "loss": 0.004254234954714775,
      "step": 123000
    },
    {
      "epoch": 34.913426057337496,
      "eval_accuracy": 0.9715775418070833,
      "eval_loss": 0.0980750173330307,
      "eval_runtime": 31.5336,
      "eval_samples_per_second": 498.739,
      "eval_steps_per_second": 7.801,
      "step": 123000
    },
    {
      "epoch": 34.91626454726086,
      "grad_norm": 1.9581429958343506,
      "learning_rate": 6.509849560034063e-05,
      "loss": 0.006026730313897133,
      "step": 123010
    },
    {
      "epoch": 34.919103037184215,
      "grad_norm": 0.0716891884803772,
      "learning_rate": 6.509565711041727e-05,
      "loss": 0.004169018566608429,
      "step": 123020
    },
    {
      "epoch": 34.92194152710758,
      "grad_norm": 7.145716667175293,
      "learning_rate": 6.50928186204939e-05,
      "loss": 0.009784446656703949,
      "step": 123030
    },
    {
      "epoch": 34.92478001703094,
      "grad_norm": 6.575927734375,
      "learning_rate": 6.508998013057054e-05,
      "loss": 0.004447620362043381,
      "step": 123040
    },
    {
      "epoch": 34.9276185069543,
      "grad_norm": 5.477135181427002,
      "learning_rate": 6.508714164064718e-05,
      "loss": 0.007135018706321716,
      "step": 123050
    },
    {
      "epoch": 34.93045699687766,
      "grad_norm": 2.1018199920654297,
      "learning_rate": 6.508458699971616e-05,
      "loss": 0.008049090206623078,
      "step": 123060
    },
    {
      "epoch": 34.93329548680102,
      "grad_norm": 0.4277038872241974,
      "learning_rate": 6.50817485097928e-05,
      "loss": 0.014344774186611176,
      "step": 123070
    },
    {
      "epoch": 34.936133976724385,
      "grad_norm": 13.465224266052246,
      "learning_rate": 6.507891001986943e-05,
      "loss": 0.0125269815325737,
      "step": 123080
    },
    {
      "epoch": 34.93897246664774,
      "grad_norm": 19.660667419433594,
      "learning_rate": 6.507607152994607e-05,
      "loss": 0.01384229063987732,
      "step": 123090
    },
    {
      "epoch": 34.9418109565711,
      "grad_norm": 0.11826115846633911,
      "learning_rate": 6.507323304002271e-05,
      "loss": 0.01463172435760498,
      "step": 123100
    },
    {
      "epoch": 34.944649446494466,
      "grad_norm": 4.548632621765137,
      "learning_rate": 6.507039455009935e-05,
      "loss": 0.005859664082527161,
      "step": 123110
    },
    {
      "epoch": 34.94748793641783,
      "grad_norm": 4.475899696350098,
      "learning_rate": 6.506755606017599e-05,
      "loss": 0.028899195790290832,
      "step": 123120
    },
    {
      "epoch": 34.950326426341185,
      "grad_norm": 0.1496761292219162,
      "learning_rate": 6.506471757025263e-05,
      "loss": 0.0035422831773757935,
      "step": 123130
    },
    {
      "epoch": 34.95316491626455,
      "grad_norm": 0.6479604244232178,
      "learning_rate": 6.506187908032927e-05,
      "loss": 0.011058783531188965,
      "step": 123140
    },
    {
      "epoch": 34.95600340618791,
      "grad_norm": 0.12235426157712936,
      "learning_rate": 6.50590405904059e-05,
      "loss": 0.003103846870362759,
      "step": 123150
    },
    {
      "epoch": 34.95884189611127,
      "grad_norm": 1.573138952255249,
      "learning_rate": 6.505620210048254e-05,
      "loss": 0.0033440306782722474,
      "step": 123160
    },
    {
      "epoch": 34.96168038603463,
      "grad_norm": 0.3734928369522095,
      "learning_rate": 6.505336361055919e-05,
      "loss": 0.010607029497623443,
      "step": 123170
    },
    {
      "epoch": 34.96451887595799,
      "grad_norm": 1.0418274402618408,
      "learning_rate": 6.505052512063583e-05,
      "loss": 0.005559181794524193,
      "step": 123180
    },
    {
      "epoch": 34.96735736588135,
      "grad_norm": 6.6771440505981445,
      "learning_rate": 6.504768663071247e-05,
      "loss": 0.005438496544957161,
      "step": 123190
    },
    {
      "epoch": 34.97019585580471,
      "grad_norm": 0.11661142855882645,
      "learning_rate": 6.504484814078911e-05,
      "loss": 0.009551359713077546,
      "step": 123200
    },
    {
      "epoch": 34.973034345728074,
      "grad_norm": 1.4923397302627563,
      "learning_rate": 6.504200965086574e-05,
      "loss": 0.004081368446350098,
      "step": 123210
    },
    {
      "epoch": 34.97587283565144,
      "grad_norm": 1.7484431266784668,
      "learning_rate": 6.503917116094238e-05,
      "loss": 0.0041174527257680895,
      "step": 123220
    },
    {
      "epoch": 34.97871132557479,
      "grad_norm": 3.9637866020202637,
      "learning_rate": 6.503633267101902e-05,
      "loss": 0.0025552863255143164,
      "step": 123230
    },
    {
      "epoch": 34.981549815498155,
      "grad_norm": 0.6464236378669739,
      "learning_rate": 6.503349418109566e-05,
      "loss": 0.0028840348124504088,
      "step": 123240
    },
    {
      "epoch": 34.98438830542152,
      "grad_norm": 0.12997175753116608,
      "learning_rate": 6.50306556911723e-05,
      "loss": 0.0010608045384287834,
      "step": 123250
    },
    {
      "epoch": 34.987226795344874,
      "grad_norm": 0.026326360180974007,
      "learning_rate": 6.502781720124894e-05,
      "loss": 0.008442394435405731,
      "step": 123260
    },
    {
      "epoch": 34.99006528526824,
      "grad_norm": 0.8273661732673645,
      "learning_rate": 6.502497871132559e-05,
      "loss": 0.007401278614997864,
      "step": 123270
    },
    {
      "epoch": 34.9929037751916,
      "grad_norm": 16.157241821289062,
      "learning_rate": 6.502214022140221e-05,
      "loss": 0.01034441515803337,
      "step": 123280
    },
    {
      "epoch": 34.995742265114956,
      "grad_norm": 0.9919429421424866,
      "learning_rate": 6.501930173147885e-05,
      "loss": 0.00947769433259964,
      "step": 123290
    },
    {
      "epoch": 34.99858075503832,
      "grad_norm": 4.441378116607666,
      "learning_rate": 6.50164632415555e-05,
      "loss": 0.0032621417194604874,
      "step": 123300
    },
    {
      "epoch": 35.00141924496168,
      "grad_norm": 0.28039756417274475,
      "learning_rate": 6.501362475163212e-05,
      "loss": 0.00488453134894371,
      "step": 123310
    },
    {
      "epoch": 35.004257734885044,
      "grad_norm": 1.2004787921905518,
      "learning_rate": 6.501078626170878e-05,
      "loss": 0.0033238157629966736,
      "step": 123320
    },
    {
      "epoch": 35.0070962248084,
      "grad_norm": 10.741399765014648,
      "learning_rate": 6.500794777178542e-05,
      "loss": 0.007548578083515167,
      "step": 123330
    },
    {
      "epoch": 35.00993471473176,
      "grad_norm": 0.3778873383998871,
      "learning_rate": 6.500510928186205e-05,
      "loss": 0.0014742210507392884,
      "step": 123340
    },
    {
      "epoch": 35.012773204655126,
      "grad_norm": 2.465121269226074,
      "learning_rate": 6.500227079193869e-05,
      "loss": 0.00376603826880455,
      "step": 123350
    },
    {
      "epoch": 35.01561169457848,
      "grad_norm": 0.22278501093387604,
      "learning_rate": 6.499943230201533e-05,
      "loss": 0.0022728117182850838,
      "step": 123360
    },
    {
      "epoch": 35.018450184501845,
      "grad_norm": 0.1620955616235733,
      "learning_rate": 6.499659381209197e-05,
      "loss": 0.003637935221195221,
      "step": 123370
    },
    {
      "epoch": 35.02128867442521,
      "grad_norm": 1.1723861694335938,
      "learning_rate": 6.499375532216861e-05,
      "loss": 0.00351867750287056,
      "step": 123380
    },
    {
      "epoch": 35.02412716434856,
      "grad_norm": 0.044418927282094955,
      "learning_rate": 6.499091683224525e-05,
      "loss": 0.0013112232089042663,
      "step": 123390
    },
    {
      "epoch": 35.026965654271926,
      "grad_norm": 0.2748942971229553,
      "learning_rate": 6.49880783423219e-05,
      "loss": 0.008447934687137604,
      "step": 123400
    },
    {
      "epoch": 35.02980414419529,
      "grad_norm": 0.6357675194740295,
      "learning_rate": 6.498523985239852e-05,
      "loss": 0.005553607642650604,
      "step": 123410
    },
    {
      "epoch": 35.03264263411865,
      "grad_norm": 0.07412460446357727,
      "learning_rate": 6.498240136247517e-05,
      "loss": 0.004420771449804306,
      "step": 123420
    },
    {
      "epoch": 35.03548112404201,
      "grad_norm": 0.019539849832654,
      "learning_rate": 6.497956287255181e-05,
      "loss": 0.0026613591238856316,
      "step": 123430
    },
    {
      "epoch": 35.03831961396537,
      "grad_norm": 0.5748937129974365,
      "learning_rate": 6.497672438262843e-05,
      "loss": 0.0058439701795578005,
      "step": 123440
    },
    {
      "epoch": 35.04115810388873,
      "grad_norm": 0.1638011485338211,
      "learning_rate": 6.497388589270509e-05,
      "loss": 0.010530485212802887,
      "step": 123450
    },
    {
      "epoch": 35.04399659381209,
      "grad_norm": 0.041668184101581573,
      "learning_rate": 6.497104740278173e-05,
      "loss": 0.001187422126531601,
      "step": 123460
    },
    {
      "epoch": 35.04683508373545,
      "grad_norm": 0.7090847492218018,
      "learning_rate": 6.496820891285836e-05,
      "loss": 0.0016980024054646492,
      "step": 123470
    },
    {
      "epoch": 35.049673573658815,
      "grad_norm": 0.13345836102962494,
      "learning_rate": 6.4965370422935e-05,
      "loss": 0.0009114157408475875,
      "step": 123480
    },
    {
      "epoch": 35.05251206358217,
      "grad_norm": 1.2167409658432007,
      "learning_rate": 6.496253193301164e-05,
      "loss": 0.002050544321537018,
      "step": 123490
    },
    {
      "epoch": 35.055350553505534,
      "grad_norm": 0.042073171585798264,
      "learning_rate": 6.495969344308828e-05,
      "loss": 0.0042908497154712675,
      "step": 123500
    },
    {
      "epoch": 35.055350553505534,
      "eval_accuracy": 0.9746296178546449,
      "eval_loss": 0.08595505356788635,
      "eval_runtime": 31.7227,
      "eval_samples_per_second": 495.764,
      "eval_steps_per_second": 7.755,
      "step": 123500
    },
    {
      "epoch": 35.0581890434289,
      "grad_norm": 1.1798826456069946,
      "learning_rate": 6.495685495316491e-05,
      "loss": 0.003264184296131134,
      "step": 123510
    },
    {
      "epoch": 35.06102753335226,
      "grad_norm": 14.037063598632812,
      "learning_rate": 6.495401646324157e-05,
      "loss": 0.0070368461310863495,
      "step": 123520
    },
    {
      "epoch": 35.063866023275615,
      "grad_norm": 1.8519291877746582,
      "learning_rate": 6.495117797331821e-05,
      "loss": 0.0022337231785058975,
      "step": 123530
    },
    {
      "epoch": 35.06670451319898,
      "grad_norm": 0.13651880621910095,
      "learning_rate": 6.494833948339483e-05,
      "loss": 0.004648395627737045,
      "step": 123540
    },
    {
      "epoch": 35.06954300312234,
      "grad_norm": 0.5002447366714478,
      "learning_rate": 6.494550099347148e-05,
      "loss": 0.0021394338458776475,
      "step": 123550
    },
    {
      "epoch": 35.0723814930457,
      "grad_norm": 0.7359775900840759,
      "learning_rate": 6.494266250354812e-05,
      "loss": 0.002437192015349865,
      "step": 123560
    },
    {
      "epoch": 35.07521998296906,
      "grad_norm": 1.0284568071365356,
      "learning_rate": 6.493982401362475e-05,
      "loss": 0.0009309140965342522,
      "step": 123570
    },
    {
      "epoch": 35.07805847289242,
      "grad_norm": 1.1425306797027588,
      "learning_rate": 6.49369855237014e-05,
      "loss": 0.003125985711812973,
      "step": 123580
    },
    {
      "epoch": 35.080896962815785,
      "grad_norm": 0.47948789596557617,
      "learning_rate": 6.493414703377804e-05,
      "loss": 0.0025972478091716767,
      "step": 123590
    },
    {
      "epoch": 35.08373545273914,
      "grad_norm": 0.11085069924592972,
      "learning_rate": 6.493130854385467e-05,
      "loss": 0.004784049838781357,
      "step": 123600
    },
    {
      "epoch": 35.086573942662504,
      "grad_norm": 0.07507847994565964,
      "learning_rate": 6.492847005393131e-05,
      "loss": 0.007839050143957138,
      "step": 123610
    },
    {
      "epoch": 35.08941243258587,
      "grad_norm": 0.7462412714958191,
      "learning_rate": 6.492563156400795e-05,
      "loss": 0.002127721719443798,
      "step": 123620
    },
    {
      "epoch": 35.09225092250922,
      "grad_norm": 0.07783081382513046,
      "learning_rate": 6.49227930740846e-05,
      "loss": 0.0026726726442575454,
      "step": 123630
    },
    {
      "epoch": 35.095089412432586,
      "grad_norm": 0.38809579610824585,
      "learning_rate": 6.491995458416122e-05,
      "loss": 0.006369081139564514,
      "step": 123640
    },
    {
      "epoch": 35.09792790235595,
      "grad_norm": 0.2409714311361313,
      "learning_rate": 6.491711609423788e-05,
      "loss": 0.014126959443092345,
      "step": 123650
    },
    {
      "epoch": 35.100766392279304,
      "grad_norm": 0.7101395130157471,
      "learning_rate": 6.49142776043145e-05,
      "loss": 0.003222520649433136,
      "step": 123660
    },
    {
      "epoch": 35.10360488220267,
      "grad_norm": 3.327662944793701,
      "learning_rate": 6.491143911439115e-05,
      "loss": 0.004929027706384659,
      "step": 123670
    },
    {
      "epoch": 35.10644337212603,
      "grad_norm": 0.11978889256715775,
      "learning_rate": 6.490860062446779e-05,
      "loss": 0.0018755283206701278,
      "step": 123680
    },
    {
      "epoch": 35.10928186204939,
      "grad_norm": 5.899048805236816,
      "learning_rate": 6.490576213454443e-05,
      "loss": 0.006101794540882111,
      "step": 123690
    },
    {
      "epoch": 35.11212035197275,
      "grad_norm": 0.2056846022605896,
      "learning_rate": 6.490292364462106e-05,
      "loss": 0.005246323347091675,
      "step": 123700
    },
    {
      "epoch": 35.11495884189611,
      "grad_norm": 0.9035454988479614,
      "learning_rate": 6.49000851546977e-05,
      "loss": 0.0018155500292778015,
      "step": 123710
    },
    {
      "epoch": 35.117797331819474,
      "grad_norm": 0.03675409406423569,
      "learning_rate": 6.489724666477435e-05,
      "loss": 0.001511766202747822,
      "step": 123720
    },
    {
      "epoch": 35.12063582174283,
      "grad_norm": 0.6537216901779175,
      "learning_rate": 6.489440817485098e-05,
      "loss": 0.0027268130332231523,
      "step": 123730
    },
    {
      "epoch": 35.12347431166619,
      "grad_norm": 0.016659442335367203,
      "learning_rate": 6.489156968492762e-05,
      "loss": 0.002885117568075657,
      "step": 123740
    },
    {
      "epoch": 35.126312801589556,
      "grad_norm": 5.474729061126709,
      "learning_rate": 6.488873119500426e-05,
      "loss": 0.0033930309116840364,
      "step": 123750
    },
    {
      "epoch": 35.12915129151291,
      "grad_norm": 0.2748189866542816,
      "learning_rate": 6.488589270508089e-05,
      "loss": 0.010052946954965591,
      "step": 123760
    },
    {
      "epoch": 35.131989781436275,
      "grad_norm": 0.18555425107479095,
      "learning_rate": 6.488305421515753e-05,
      "loss": 0.0016905482858419417,
      "step": 123770
    },
    {
      "epoch": 35.13482827135964,
      "grad_norm": 0.11990313977003098,
      "learning_rate": 6.488021572523419e-05,
      "loss": 0.0018359489738941192,
      "step": 123780
    },
    {
      "epoch": 35.137666761283,
      "grad_norm": 0.0719369575381279,
      "learning_rate": 6.487737723531081e-05,
      "loss": 0.0013567242771387101,
      "step": 123790
    },
    {
      "epoch": 35.140505251206356,
      "grad_norm": 0.28665465116500854,
      "learning_rate": 6.487453874538746e-05,
      "loss": 0.0009961199015378953,
      "step": 123800
    },
    {
      "epoch": 35.14334374112972,
      "grad_norm": 0.4773728847503662,
      "learning_rate": 6.48717002554641e-05,
      "loss": 0.0012740563601255418,
      "step": 123810
    },
    {
      "epoch": 35.14618223105308,
      "grad_norm": 0.7105950713157654,
      "learning_rate": 6.486886176554074e-05,
      "loss": 0.0010430144146084785,
      "step": 123820
    },
    {
      "epoch": 35.14902072097644,
      "grad_norm": 0.1763571947813034,
      "learning_rate": 6.486602327561737e-05,
      "loss": 0.004257035255432129,
      "step": 123830
    },
    {
      "epoch": 35.1518592108998,
      "grad_norm": 0.6583411693572998,
      "learning_rate": 6.486318478569401e-05,
      "loss": 0.0017475824803113937,
      "step": 123840
    },
    {
      "epoch": 35.154697700823164,
      "grad_norm": 0.24023713171482086,
      "learning_rate": 6.486034629577066e-05,
      "loss": 0.002329159528017044,
      "step": 123850
    },
    {
      "epoch": 35.15753619074652,
      "grad_norm": 0.03802018240094185,
      "learning_rate": 6.485750780584729e-05,
      "loss": 0.003923746570944786,
      "step": 123860
    },
    {
      "epoch": 35.16037468066988,
      "grad_norm": 0.09055881202220917,
      "learning_rate": 6.485466931592393e-05,
      "loss": 0.010653698444366455,
      "step": 123870
    },
    {
      "epoch": 35.163213170593245,
      "grad_norm": 0.02531651221215725,
      "learning_rate": 6.485183082600057e-05,
      "loss": 0.0019262006506323814,
      "step": 123880
    },
    {
      "epoch": 35.16605166051661,
      "grad_norm": 2.1929142475128174,
      "learning_rate": 6.48489923360772e-05,
      "loss": 0.005838973075151443,
      "step": 123890
    },
    {
      "epoch": 35.168890150439964,
      "grad_norm": 0.10594841092824936,
      "learning_rate": 6.484615384615384e-05,
      "loss": 0.0043249309062957765,
      "step": 123900
    },
    {
      "epoch": 35.17172864036333,
      "grad_norm": 0.11679910868406296,
      "learning_rate": 6.48433153562305e-05,
      "loss": 0.0023911599069833755,
      "step": 123910
    },
    {
      "epoch": 35.17456713028669,
      "grad_norm": 0.20320293307304382,
      "learning_rate": 6.484047686630713e-05,
      "loss": 0.015488362312316895,
      "step": 123920
    },
    {
      "epoch": 35.177405620210045,
      "grad_norm": 0.3133673071861267,
      "learning_rate": 6.483763837638377e-05,
      "loss": 0.0070457324385643,
      "step": 123930
    },
    {
      "epoch": 35.18024411013341,
      "grad_norm": 0.07562874257564545,
      "learning_rate": 6.483479988646041e-05,
      "loss": 0.0022812724113464356,
      "step": 123940
    },
    {
      "epoch": 35.18308260005677,
      "grad_norm": 0.07803227752447128,
      "learning_rate": 6.483196139653705e-05,
      "loss": 0.004335168749094009,
      "step": 123950
    },
    {
      "epoch": 35.185921089980134,
      "grad_norm": 0.5244848728179932,
      "learning_rate": 6.482912290661368e-05,
      "loss": 0.001566123776137829,
      "step": 123960
    },
    {
      "epoch": 35.18875957990349,
      "grad_norm": 1.4270581007003784,
      "learning_rate": 6.482628441669032e-05,
      "loss": 0.002507426403462887,
      "step": 123970
    },
    {
      "epoch": 35.19159806982685,
      "grad_norm": 0.5445083975791931,
      "learning_rate": 6.482344592676697e-05,
      "loss": 0.0021049004048109055,
      "step": 123980
    },
    {
      "epoch": 35.194436559750216,
      "grad_norm": 0.04573143646121025,
      "learning_rate": 6.48206074368436e-05,
      "loss": 0.004109510779380798,
      "step": 123990
    },
    {
      "epoch": 35.19727504967357,
      "grad_norm": 0.3981334865093231,
      "learning_rate": 6.481776894692024e-05,
      "loss": 0.0154224693775177,
      "step": 124000
    },
    {
      "epoch": 35.19727504967357,
      "eval_accuracy": 0.9736122591721244,
      "eval_loss": 0.09443560987710953,
      "eval_runtime": 32.6858,
      "eval_samples_per_second": 481.157,
      "eval_steps_per_second": 7.526,
      "step": 124000
    },
    {
      "epoch": 35.200113539596934,
      "grad_norm": 0.7593827843666077,
      "learning_rate": 6.481493045699688e-05,
      "loss": 0.005224959924817085,
      "step": 124010
    },
    {
      "epoch": 35.2029520295203,
      "grad_norm": 3.7218992710113525,
      "learning_rate": 6.481209196707351e-05,
      "loss": 0.0020976241677999495,
      "step": 124020
    },
    {
      "epoch": 35.20579051944365,
      "grad_norm": 0.028922753408551216,
      "learning_rate": 6.480925347715015e-05,
      "loss": 0.005799264460802078,
      "step": 124030
    },
    {
      "epoch": 35.208629009367016,
      "grad_norm": 1.5619020462036133,
      "learning_rate": 6.48064149872268e-05,
      "loss": 0.0034693576395511626,
      "step": 124040
    },
    {
      "epoch": 35.21146749929038,
      "grad_norm": 0.1801603138446808,
      "learning_rate": 6.480357649730344e-05,
      "loss": 0.00175578985363245,
      "step": 124050
    },
    {
      "epoch": 35.21430598921374,
      "grad_norm": 1.5496667623519897,
      "learning_rate": 6.480073800738008e-05,
      "loss": 0.0032393354922533035,
      "step": 124060
    },
    {
      "epoch": 35.2171444791371,
      "grad_norm": 11.993553161621094,
      "learning_rate": 6.479789951745672e-05,
      "loss": 0.007441597431898117,
      "step": 124070
    },
    {
      "epoch": 35.21998296906046,
      "grad_norm": 0.22900712490081787,
      "learning_rate": 6.479506102753336e-05,
      "loss": 0.008034928888082504,
      "step": 124080
    },
    {
      "epoch": 35.22282145898382,
      "grad_norm": 0.8359187245368958,
      "learning_rate": 6.479222253760999e-05,
      "loss": 0.012940695881843567,
      "step": 124090
    },
    {
      "epoch": 35.22565994890718,
      "grad_norm": 0.23055890202522278,
      "learning_rate": 6.478938404768663e-05,
      "loss": 0.001070609875023365,
      "step": 124100
    },
    {
      "epoch": 35.22849843883054,
      "grad_norm": 10.563587188720703,
      "learning_rate": 6.478654555776328e-05,
      "loss": 0.011516466736793518,
      "step": 124110
    },
    {
      "epoch": 35.231336928753905,
      "grad_norm": 0.04512307420372963,
      "learning_rate": 6.478370706783991e-05,
      "loss": 0.0006166955456137658,
      "step": 124120
    },
    {
      "epoch": 35.23417541867726,
      "grad_norm": 0.11999720335006714,
      "learning_rate": 6.478086857791655e-05,
      "loss": 0.0038831207901239397,
      "step": 124130
    },
    {
      "epoch": 35.23701390860062,
      "grad_norm": 0.26278963685035706,
      "learning_rate": 6.47780300879932e-05,
      "loss": 0.00224434956908226,
      "step": 124140
    },
    {
      "epoch": 35.239852398523986,
      "grad_norm": 0.1841161847114563,
      "learning_rate": 6.477519159806982e-05,
      "loss": 0.00695534348487854,
      "step": 124150
    },
    {
      "epoch": 35.24269088844735,
      "grad_norm": 0.3225573003292084,
      "learning_rate": 6.477235310814646e-05,
      "loss": 0.0033076439052820207,
      "step": 124160
    },
    {
      "epoch": 35.245529378370705,
      "grad_norm": 0.49394699931144714,
      "learning_rate": 6.47695146182231e-05,
      "loss": 0.0024590885266661646,
      "step": 124170
    },
    {
      "epoch": 35.24836786829407,
      "grad_norm": 0.3010150194168091,
      "learning_rate": 6.476667612829975e-05,
      "loss": 0.005991066619753837,
      "step": 124180
    },
    {
      "epoch": 35.25120635821743,
      "grad_norm": 3.922555446624756,
      "learning_rate": 6.476383763837639e-05,
      "loss": 0.004172184318304062,
      "step": 124190
    },
    {
      "epoch": 35.254044848140786,
      "grad_norm": 18.703628540039062,
      "learning_rate": 6.476099914845303e-05,
      "loss": 0.013839299976825713,
      "step": 124200
    },
    {
      "epoch": 35.25688333806415,
      "grad_norm": 0.04546620324254036,
      "learning_rate": 6.475816065852967e-05,
      "loss": 0.001329273171722889,
      "step": 124210
    },
    {
      "epoch": 35.25972182798751,
      "grad_norm": 0.16649018228054047,
      "learning_rate": 6.47553221686063e-05,
      "loss": 0.0022044172510504723,
      "step": 124220
    },
    {
      "epoch": 35.26256031791087,
      "grad_norm": 0.22203882038593292,
      "learning_rate": 6.475248367868294e-05,
      "loss": 0.003903895989060402,
      "step": 124230
    },
    {
      "epoch": 35.26539880783423,
      "grad_norm": 1.7335249185562134,
      "learning_rate": 6.474964518875958e-05,
      "loss": 0.008229069411754608,
      "step": 124240
    },
    {
      "epoch": 35.268237297757594,
      "grad_norm": 12.773519515991211,
      "learning_rate": 6.474680669883622e-05,
      "loss": 0.009718339145183563,
      "step": 124250
    },
    {
      "epoch": 35.27107578768096,
      "grad_norm": 0.32537272572517395,
      "learning_rate": 6.474396820891286e-05,
      "loss": 0.0024151701480150225,
      "step": 124260
    },
    {
      "epoch": 35.27391427760431,
      "grad_norm": 0.29036396741867065,
      "learning_rate": 6.47411297189895e-05,
      "loss": 0.002510245516896248,
      "step": 124270
    },
    {
      "epoch": 35.276752767527675,
      "grad_norm": 9.233658790588379,
      "learning_rate": 6.473829122906613e-05,
      "loss": 0.008023773133754731,
      "step": 124280
    },
    {
      "epoch": 35.27959125745104,
      "grad_norm": 0.6115303039550781,
      "learning_rate": 6.473545273914278e-05,
      "loss": 0.0014342835173010826,
      "step": 124290
    },
    {
      "epoch": 35.282429747374394,
      "grad_norm": 0.2653685212135315,
      "learning_rate": 6.473261424921942e-05,
      "loss": 0.0017992928624153137,
      "step": 124300
    },
    {
      "epoch": 35.28526823729776,
      "grad_norm": 0.07249756902456284,
      "learning_rate": 6.472977575929606e-05,
      "loss": 0.0012476861476898194,
      "step": 124310
    },
    {
      "epoch": 35.28810672722112,
      "grad_norm": 0.041601844131946564,
      "learning_rate": 6.47269372693727e-05,
      "loss": 0.007989827543497086,
      "step": 124320
    },
    {
      "epoch": 35.29094521714448,
      "grad_norm": 0.07185827940702438,
      "learning_rate": 6.472409877944934e-05,
      "loss": 0.0005686616525053978,
      "step": 124330
    },
    {
      "epoch": 35.29378370706784,
      "grad_norm": 2.94441556930542,
      "learning_rate": 6.472126028952598e-05,
      "loss": 0.001534784398972988,
      "step": 124340
    },
    {
      "epoch": 35.2966221969912,
      "grad_norm": 1.1852378845214844,
      "learning_rate": 6.471842179960261e-05,
      "loss": 0.0017348313704133034,
      "step": 124350
    },
    {
      "epoch": 35.299460686914564,
      "grad_norm": 4.328054904937744,
      "learning_rate": 6.471586715867159e-05,
      "loss": 0.012050793319940568,
      "step": 124360
    },
    {
      "epoch": 35.30229917683792,
      "grad_norm": 0.7507839798927307,
      "learning_rate": 6.471302866874823e-05,
      "loss": 0.00557251125574112,
      "step": 124370
    },
    {
      "epoch": 35.30513766676128,
      "grad_norm": 1.5571794509887695,
      "learning_rate": 6.471019017882487e-05,
      "loss": 0.0037160709500312803,
      "step": 124380
    },
    {
      "epoch": 35.307976156684646,
      "grad_norm": 0.6430335640907288,
      "learning_rate": 6.470735168890151e-05,
      "loss": 0.003933653980493545,
      "step": 124390
    },
    {
      "epoch": 35.310814646608,
      "grad_norm": 0.3176755905151367,
      "learning_rate": 6.470451319897814e-05,
      "loss": 0.007839932292699813,
      "step": 124400
    },
    {
      "epoch": 35.313653136531364,
      "grad_norm": 0.6366283893585205,
      "learning_rate": 6.470167470905478e-05,
      "loss": 0.004763571172952652,
      "step": 124410
    },
    {
      "epoch": 35.31649162645473,
      "grad_norm": 0.08230401575565338,
      "learning_rate": 6.469883621913142e-05,
      "loss": 0.012808883190155029,
      "step": 124420
    },
    {
      "epoch": 35.31933011637809,
      "grad_norm": 0.08792869001626968,
      "learning_rate": 6.469599772920807e-05,
      "loss": 0.00757981464266777,
      "step": 124430
    },
    {
      "epoch": 35.322168606301446,
      "grad_norm": 0.13389721512794495,
      "learning_rate": 6.469315923928471e-05,
      "loss": 0.0026062030345201494,
      "step": 124440
    },
    {
      "epoch": 35.32500709622481,
      "grad_norm": 0.17631679773330688,
      "learning_rate": 6.469032074936135e-05,
      "loss": 0.0029279107227921488,
      "step": 124450
    },
    {
      "epoch": 35.32784558614817,
      "grad_norm": 0.11230606585741043,
      "learning_rate": 6.468748225943798e-05,
      "loss": 0.003203073889017105,
      "step": 124460
    },
    {
      "epoch": 35.33068407607153,
      "grad_norm": 5.077963352203369,
      "learning_rate": 6.468464376951462e-05,
      "loss": 0.0040066268295049666,
      "step": 124470
    },
    {
      "epoch": 35.33352256599489,
      "grad_norm": 0.2952515184879303,
      "learning_rate": 6.468180527959126e-05,
      "loss": 0.006278103590011597,
      "step": 124480
    },
    {
      "epoch": 35.33636105591825,
      "grad_norm": 0.6553015112876892,
      "learning_rate": 6.46789667896679e-05,
      "loss": 0.006164129823446274,
      "step": 124490
    },
    {
      "epoch": 35.33919954584161,
      "grad_norm": 0.05470352619886398,
      "learning_rate": 6.467612829974454e-05,
      "loss": 0.003238702192902565,
      "step": 124500
    },
    {
      "epoch": 35.33919954584161,
      "eval_accuracy": 0.9672537674063713,
      "eval_loss": 0.11323834955692291,
      "eval_runtime": 31.7161,
      "eval_samples_per_second": 495.867,
      "eval_steps_per_second": 7.756,
      "step": 124500
    },
    {
      "epoch": 35.34203803576497,
      "grad_norm": 0.2669536769390106,
      "learning_rate": 6.467328980982118e-05,
      "loss": 0.0028995968401432036,
      "step": 124510
    },
    {
      "epoch": 35.344876525688335,
      "grad_norm": 0.1884731650352478,
      "learning_rate": 6.467045131989781e-05,
      "loss": 0.010411982238292695,
      "step": 124520
    },
    {
      "epoch": 35.3477150156117,
      "grad_norm": 5.714053153991699,
      "learning_rate": 6.466761282997445e-05,
      "loss": 0.0029901187866926194,
      "step": 124530
    },
    {
      "epoch": 35.35055350553505,
      "grad_norm": 0.018948949873447418,
      "learning_rate": 6.46647743400511e-05,
      "loss": 0.002273883670568466,
      "step": 124540
    },
    {
      "epoch": 35.353391995458416,
      "grad_norm": 0.5497522354125977,
      "learning_rate": 6.466193585012774e-05,
      "loss": 0.004661240428686142,
      "step": 124550
    },
    {
      "epoch": 35.35623048538178,
      "grad_norm": 0.2104068547487259,
      "learning_rate": 6.465909736020438e-05,
      "loss": 0.002843768149614334,
      "step": 124560
    },
    {
      "epoch": 35.359068975305135,
      "grad_norm": 0.35174691677093506,
      "learning_rate": 6.465625887028102e-05,
      "loss": 0.0036916732788085938,
      "step": 124570
    },
    {
      "epoch": 35.3619074652285,
      "grad_norm": 1.3930494785308838,
      "learning_rate": 6.465342038035766e-05,
      "loss": 0.009654825925827027,
      "step": 124580
    },
    {
      "epoch": 35.36474595515186,
      "grad_norm": 0.2108253836631775,
      "learning_rate": 6.465058189043429e-05,
      "loss": 0.014609508216381073,
      "step": 124590
    },
    {
      "epoch": 35.36758444507522,
      "grad_norm": 3.552393674850464,
      "learning_rate": 6.464774340051093e-05,
      "loss": 0.015583406388759612,
      "step": 124600
    },
    {
      "epoch": 35.37042293499858,
      "grad_norm": 0.09530099481344223,
      "learning_rate": 6.464490491058757e-05,
      "loss": 0.006617793440818786,
      "step": 124610
    },
    {
      "epoch": 35.37326142492194,
      "grad_norm": 0.19090984761714935,
      "learning_rate": 6.46420664206642e-05,
      "loss": 0.0013797318562865258,
      "step": 124620
    },
    {
      "epoch": 35.376099914845305,
      "grad_norm": 0.30811187624931335,
      "learning_rate": 6.463922793074085e-05,
      "loss": 0.007688663899898529,
      "step": 124630
    },
    {
      "epoch": 35.37893840476866,
      "grad_norm": 1.1452854871749878,
      "learning_rate": 6.46363894408175e-05,
      "loss": 0.0016369694843888284,
      "step": 124640
    },
    {
      "epoch": 35.381776894692024,
      "grad_norm": 0.29348257184028625,
      "learning_rate": 6.463355095089412e-05,
      "loss": 0.002300905808806419,
      "step": 124650
    },
    {
      "epoch": 35.38461538461539,
      "grad_norm": 3.268918991088867,
      "learning_rate": 6.463071246097076e-05,
      "loss": 0.007275217771530151,
      "step": 124660
    },
    {
      "epoch": 35.38745387453874,
      "grad_norm": 0.20252858102321625,
      "learning_rate": 6.46278739710474e-05,
      "loss": 0.005437327921390534,
      "step": 124670
    },
    {
      "epoch": 35.390292364462105,
      "grad_norm": 0.0661584809422493,
      "learning_rate": 6.462503548112405e-05,
      "loss": 0.0055758640170097355,
      "step": 124680
    },
    {
      "epoch": 35.39313085438547,
      "grad_norm": 1.109292984008789,
      "learning_rate": 6.462219699120069e-05,
      "loss": 0.012798486649990082,
      "step": 124690
    },
    {
      "epoch": 35.395969344308824,
      "grad_norm": 0.048671137541532516,
      "learning_rate": 6.461935850127733e-05,
      "loss": 0.006836451590061188,
      "step": 124700
    },
    {
      "epoch": 35.39880783423219,
      "grad_norm": 0.5001567006111145,
      "learning_rate": 6.461652001135397e-05,
      "loss": 0.004541724920272827,
      "step": 124710
    },
    {
      "epoch": 35.40164632415555,
      "grad_norm": 0.02809389866888523,
      "learning_rate": 6.46136815214306e-05,
      "loss": 0.004792189970612526,
      "step": 124720
    },
    {
      "epoch": 35.40448481407891,
      "grad_norm": 0.08299035578966141,
      "learning_rate": 6.461084303150724e-05,
      "loss": 0.005103607475757599,
      "step": 124730
    },
    {
      "epoch": 35.40732330400227,
      "grad_norm": 1.3819055557250977,
      "learning_rate": 6.460800454158388e-05,
      "loss": 0.012244104593992233,
      "step": 124740
    },
    {
      "epoch": 35.41016179392563,
      "grad_norm": 0.6637468338012695,
      "learning_rate": 6.460516605166051e-05,
      "loss": 0.012076373398303985,
      "step": 124750
    },
    {
      "epoch": 35.413000283848994,
      "grad_norm": 0.6656967401504517,
      "learning_rate": 6.460232756173716e-05,
      "loss": 0.028901314735412596,
      "step": 124760
    },
    {
      "epoch": 35.41583877377235,
      "grad_norm": 0.024717943742871284,
      "learning_rate": 6.45994890718138e-05,
      "loss": 0.001671173982322216,
      "step": 124770
    },
    {
      "epoch": 35.41867726369571,
      "grad_norm": 3.7956604957580566,
      "learning_rate": 6.459665058189043e-05,
      "loss": 0.0027819059789180754,
      "step": 124780
    },
    {
      "epoch": 35.421515753619076,
      "grad_norm": 0.9437547922134399,
      "learning_rate": 6.459381209196707e-05,
      "loss": 0.003428303450345993,
      "step": 124790
    },
    {
      "epoch": 35.42435424354244,
      "grad_norm": 0.9660364389419556,
      "learning_rate": 6.459097360204372e-05,
      "loss": 0.007007385045289994,
      "step": 124800
    },
    {
      "epoch": 35.427192733465795,
      "grad_norm": 7.1223955154418945,
      "learning_rate": 6.458813511212036e-05,
      "loss": 0.0050876297056674956,
      "step": 124810
    },
    {
      "epoch": 35.43003122338916,
      "grad_norm": 4.550117015838623,
      "learning_rate": 6.458529662219698e-05,
      "loss": 0.006623812019824982,
      "step": 124820
    },
    {
      "epoch": 35.43286971331252,
      "grad_norm": 0.4335205852985382,
      "learning_rate": 6.458245813227364e-05,
      "loss": 0.003278358280658722,
      "step": 124830
    },
    {
      "epoch": 35.435708203235876,
      "grad_norm": 0.025949157774448395,
      "learning_rate": 6.457961964235028e-05,
      "loss": 0.006486078351736068,
      "step": 124840
    },
    {
      "epoch": 35.43854669315924,
      "grad_norm": 6.395306587219238,
      "learning_rate": 6.457678115242691e-05,
      "loss": 0.014768750965595245,
      "step": 124850
    },
    {
      "epoch": 35.4413851830826,
      "grad_norm": 3.8973159790039062,
      "learning_rate": 6.457394266250355e-05,
      "loss": 0.0062584981322288515,
      "step": 124860
    },
    {
      "epoch": 35.44422367300596,
      "grad_norm": 1.3165786266326904,
      "learning_rate": 6.457110417258019e-05,
      "loss": 0.010468196123838425,
      "step": 124870
    },
    {
      "epoch": 35.44706216292932,
      "grad_norm": 0.1820409744977951,
      "learning_rate": 6.456826568265682e-05,
      "loss": 0.022558340430259706,
      "step": 124880
    },
    {
      "epoch": 35.44990065285268,
      "grad_norm": 0.5113018751144409,
      "learning_rate": 6.456542719273347e-05,
      "loss": 0.003203253448009491,
      "step": 124890
    },
    {
      "epoch": 35.452739142776046,
      "grad_norm": 0.032387614250183105,
      "learning_rate": 6.456258870281012e-05,
      "loss": 0.002933865785598755,
      "step": 124900
    },
    {
      "epoch": 35.4555776326994,
      "grad_norm": 0.34201836585998535,
      "learning_rate": 6.455975021288674e-05,
      "loss": 0.006065483391284943,
      "step": 124910
    },
    {
      "epoch": 35.458416122622765,
      "grad_norm": 2.2335569858551025,
      "learning_rate": 6.455691172296338e-05,
      "loss": 0.008176913857460022,
      "step": 124920
    },
    {
      "epoch": 35.46125461254613,
      "grad_norm": 8.967089653015137,
      "learning_rate": 6.455407323304003e-05,
      "loss": 0.005386056005954742,
      "step": 124930
    },
    {
      "epoch": 35.464093102469484,
      "grad_norm": 13.451428413391113,
      "learning_rate": 6.455123474311667e-05,
      "loss": 0.009882912039756775,
      "step": 124940
    },
    {
      "epoch": 35.46693159239285,
      "grad_norm": 2.963137149810791,
      "learning_rate": 6.45483962531933e-05,
      "loss": 0.014495460689067841,
      "step": 124950
    },
    {
      "epoch": 35.46977008231621,
      "grad_norm": 0.19372975826263428,
      "learning_rate": 6.454555776326995e-05,
      "loss": 0.007631847262382507,
      "step": 124960
    },
    {
      "epoch": 35.472608572239565,
      "grad_norm": 12.685774803161621,
      "learning_rate": 6.454271927334659e-05,
      "loss": 0.01924920529127121,
      "step": 124970
    },
    {
      "epoch": 35.47544706216293,
      "grad_norm": 5.221495628356934,
      "learning_rate": 6.453988078342322e-05,
      "loss": 0.009094873815774918,
      "step": 124980
    },
    {
      "epoch": 35.47828555208629,
      "grad_norm": 0.3292238116264343,
      "learning_rate": 6.453704229349986e-05,
      "loss": 0.009680644422769547,
      "step": 124990
    },
    {
      "epoch": 35.481124042009654,
      "grad_norm": 0.09861037135124207,
      "learning_rate": 6.45342038035765e-05,
      "loss": 0.003667459636926651,
      "step": 125000
    },
    {
      "epoch": 35.481124042009654,
      "eval_accuracy": 0.9736122591721244,
      "eval_loss": 0.09246641397476196,
      "eval_runtime": 32.1676,
      "eval_samples_per_second": 488.908,
      "eval_steps_per_second": 7.647,
      "step": 125000
    },
    {
      "epoch": 35.48396253193301,
      "grad_norm": 0.35291433334350586,
      "learning_rate": 6.453136531365313e-05,
      "loss": 0.01767518073320389,
      "step": 125010
    },
    {
      "epoch": 35.48680102185637,
      "grad_norm": 0.6744071245193481,
      "learning_rate": 6.452852682372977e-05,
      "loss": 0.003070857934653759,
      "step": 125020
    },
    {
      "epoch": 35.489639511779735,
      "grad_norm": 3.7759416103363037,
      "learning_rate": 6.452568833380643e-05,
      "loss": 0.0032647754997015,
      "step": 125030
    },
    {
      "epoch": 35.49247800170309,
      "grad_norm": 0.10367639362812042,
      "learning_rate": 6.452284984388305e-05,
      "loss": 0.003720211237668991,
      "step": 125040
    },
    {
      "epoch": 35.495316491626454,
      "grad_norm": 0.506496250629425,
      "learning_rate": 6.45200113539597e-05,
      "loss": 0.010861226171255112,
      "step": 125050
    },
    {
      "epoch": 35.49815498154982,
      "grad_norm": 0.14185407757759094,
      "learning_rate": 6.451717286403634e-05,
      "loss": 0.0069375157356262205,
      "step": 125060
    },
    {
      "epoch": 35.50099347147318,
      "grad_norm": 0.16966380178928375,
      "learning_rate": 6.451433437411298e-05,
      "loss": 0.0066543005406856535,
      "step": 125070
    },
    {
      "epoch": 35.503831961396536,
      "grad_norm": 0.15068873763084412,
      "learning_rate": 6.45114958841896e-05,
      "loss": 0.0008128028362989426,
      "step": 125080
    },
    {
      "epoch": 35.5066704513199,
      "grad_norm": 0.4455798864364624,
      "learning_rate": 6.450865739426626e-05,
      "loss": 0.0015239287167787552,
      "step": 125090
    },
    {
      "epoch": 35.50950894124326,
      "grad_norm": 0.9672191143035889,
      "learning_rate": 6.45058189043429e-05,
      "loss": 0.015783116221427917,
      "step": 125100
    },
    {
      "epoch": 35.51234743116662,
      "grad_norm": 4.735830783843994,
      "learning_rate": 6.450298041441953e-05,
      "loss": 0.010145072638988496,
      "step": 125110
    },
    {
      "epoch": 35.51518592108998,
      "grad_norm": 0.3422497510910034,
      "learning_rate": 6.450014192449617e-05,
      "loss": 0.013720910251140594,
      "step": 125120
    },
    {
      "epoch": 35.51802441101334,
      "grad_norm": 0.12498065084218979,
      "learning_rate": 6.449730343457281e-05,
      "loss": 0.002984499931335449,
      "step": 125130
    },
    {
      "epoch": 35.5208629009367,
      "grad_norm": 0.5811486840248108,
      "learning_rate": 6.449446494464944e-05,
      "loss": 0.0034018225967884065,
      "step": 125140
    },
    {
      "epoch": 35.52370139086006,
      "grad_norm": 0.480955570936203,
      "learning_rate": 6.449162645472608e-05,
      "loss": 0.008165840804576874,
      "step": 125150
    },
    {
      "epoch": 35.526539880783425,
      "grad_norm": 2.1032793521881104,
      "learning_rate": 6.448878796480274e-05,
      "loss": 0.005098474770784378,
      "step": 125160
    },
    {
      "epoch": 35.52937837070679,
      "grad_norm": 0.7500012516975403,
      "learning_rate": 6.448594947487937e-05,
      "loss": 0.002728801034390926,
      "step": 125170
    },
    {
      "epoch": 35.53221686063014,
      "grad_norm": 0.1109260618686676,
      "learning_rate": 6.4483110984956e-05,
      "loss": 0.011433902382850646,
      "step": 125180
    },
    {
      "epoch": 35.535055350553506,
      "grad_norm": 0.8421873450279236,
      "learning_rate": 6.448027249503265e-05,
      "loss": 0.010049564391374588,
      "step": 125190
    },
    {
      "epoch": 35.53789384047687,
      "grad_norm": 5.967371463775635,
      "learning_rate": 6.447743400510929e-05,
      "loss": 0.0035402797162532806,
      "step": 125200
    },
    {
      "epoch": 35.540732330400225,
      "grad_norm": 0.2739831805229187,
      "learning_rate": 6.447459551518592e-05,
      "loss": 0.003483092039823532,
      "step": 125210
    },
    {
      "epoch": 35.54357082032359,
      "grad_norm": 0.308971107006073,
      "learning_rate": 6.447175702526256e-05,
      "loss": 0.008164165914058686,
      "step": 125220
    },
    {
      "epoch": 35.54640931024695,
      "grad_norm": 0.10572206228971481,
      "learning_rate": 6.446891853533921e-05,
      "loss": 0.003227214515209198,
      "step": 125230
    },
    {
      "epoch": 35.549247800170306,
      "grad_norm": 1.0289645195007324,
      "learning_rate": 6.446608004541584e-05,
      "loss": 0.0036446630954742433,
      "step": 125240
    },
    {
      "epoch": 35.55208629009367,
      "grad_norm": 0.1717691421508789,
      "learning_rate": 6.446324155549248e-05,
      "loss": 0.002933488413691521,
      "step": 125250
    },
    {
      "epoch": 35.55492478001703,
      "grad_norm": 0.03167391195893288,
      "learning_rate": 6.446040306556912e-05,
      "loss": 0.005520378798246383,
      "step": 125260
    },
    {
      "epoch": 35.557763269940395,
      "grad_norm": 0.2930864095687866,
      "learning_rate": 6.445756457564575e-05,
      "loss": 0.000985959731042385,
      "step": 125270
    },
    {
      "epoch": 35.56060175986375,
      "grad_norm": 0.1712752878665924,
      "learning_rate": 6.445472608572239e-05,
      "loss": 0.0023083217442035676,
      "step": 125280
    },
    {
      "epoch": 35.563440249787114,
      "grad_norm": 2.1108486652374268,
      "learning_rate": 6.445188759579905e-05,
      "loss": 0.01200648695230484,
      "step": 125290
    },
    {
      "epoch": 35.56627873971048,
      "grad_norm": 0.371765673160553,
      "learning_rate": 6.444904910587568e-05,
      "loss": 0.005719281733036041,
      "step": 125300
    },
    {
      "epoch": 35.56911722963383,
      "grad_norm": 0.29231005907058716,
      "learning_rate": 6.444621061595232e-05,
      "loss": 0.0013308390974998475,
      "step": 125310
    },
    {
      "epoch": 35.571955719557195,
      "grad_norm": 0.05381132289767265,
      "learning_rate": 6.444337212602896e-05,
      "loss": 0.011894132941961288,
      "step": 125320
    },
    {
      "epoch": 35.57479420948056,
      "grad_norm": 0.15411996841430664,
      "learning_rate": 6.44405336361056e-05,
      "loss": 0.011256928741931915,
      "step": 125330
    },
    {
      "epoch": 35.577632699403914,
      "grad_norm": 5.019742012023926,
      "learning_rate": 6.443769514618223e-05,
      "loss": 0.009076657891273498,
      "step": 125340
    },
    {
      "epoch": 35.58047118932728,
      "grad_norm": 0.7339893579483032,
      "learning_rate": 6.443485665625887e-05,
      "loss": 0.00530652366578579,
      "step": 125350
    },
    {
      "epoch": 35.58330967925064,
      "grad_norm": 1.5370879173278809,
      "learning_rate": 6.443201816633551e-05,
      "loss": 0.0019224384799599648,
      "step": 125360
    },
    {
      "epoch": 35.586148169174,
      "grad_norm": 7.998415946960449,
      "learning_rate": 6.442917967641215e-05,
      "loss": 0.00514814630150795,
      "step": 125370
    },
    {
      "epoch": 35.58898665909736,
      "grad_norm": 7.727628707885742,
      "learning_rate": 6.44263411864888e-05,
      "loss": 0.005549687147140503,
      "step": 125380
    },
    {
      "epoch": 35.59182514902072,
      "grad_norm": 2.280776023864746,
      "learning_rate": 6.442350269656543e-05,
      "loss": 0.010034140944480897,
      "step": 125390
    },
    {
      "epoch": 35.594663638944084,
      "grad_norm": 9.786527633666992,
      "learning_rate": 6.442066420664206e-05,
      "loss": 0.014894461631774903,
      "step": 125400
    },
    {
      "epoch": 35.59750212886744,
      "grad_norm": 2.7524898052215576,
      "learning_rate": 6.44178257167187e-05,
      "loss": 0.0026578644290566446,
      "step": 125410
    },
    {
      "epoch": 35.6003406187908,
      "grad_norm": 14.630492210388184,
      "learning_rate": 6.441498722679535e-05,
      "loss": 0.007752618938684464,
      "step": 125420
    },
    {
      "epoch": 35.603179108714166,
      "grad_norm": 1.4213225841522217,
      "learning_rate": 6.441214873687199e-05,
      "loss": 0.014157319068908691,
      "step": 125430
    },
    {
      "epoch": 35.60601759863752,
      "grad_norm": 0.6881102919578552,
      "learning_rate": 6.440931024694863e-05,
      "loss": 0.005177746713161469,
      "step": 125440
    },
    {
      "epoch": 35.608856088560884,
      "grad_norm": 21.151691436767578,
      "learning_rate": 6.440647175702527e-05,
      "loss": 0.008037223666906356,
      "step": 125450
    },
    {
      "epoch": 35.61169457848425,
      "grad_norm": 1.5915447473526,
      "learning_rate": 6.44036332671019e-05,
      "loss": 0.007538111507892608,
      "step": 125460
    },
    {
      "epoch": 35.61453306840761,
      "grad_norm": 1.09759521484375,
      "learning_rate": 6.440079477717854e-05,
      "loss": 0.006154146045446396,
      "step": 125470
    },
    {
      "epoch": 35.617371558330966,
      "grad_norm": 0.8544142246246338,
      "learning_rate": 6.439795628725518e-05,
      "loss": 0.006271190196275711,
      "step": 125480
    },
    {
      "epoch": 35.62021004825433,
      "grad_norm": 0.6008771657943726,
      "learning_rate": 6.439511779733182e-05,
      "loss": 0.012101072818040848,
      "step": 125490
    },
    {
      "epoch": 35.62304853817769,
      "grad_norm": 9.811751365661621,
      "learning_rate": 6.439227930740846e-05,
      "loss": 0.006930852681398392,
      "step": 125500
    },
    {
      "epoch": 35.62304853817769,
      "eval_accuracy": 0.9694156546067273,
      "eval_loss": 0.11055191606283188,
      "eval_runtime": 31.695,
      "eval_samples_per_second": 496.197,
      "eval_steps_per_second": 7.761,
      "step": 125500
    },
    {
      "epoch": 35.62588702810105,
      "grad_norm": 0.7095000147819519,
      "learning_rate": 6.43894408174851e-05,
      "loss": 0.00657695084810257,
      "step": 125510
    },
    {
      "epoch": 35.62872551802441,
      "grad_norm": 1.8637582063674927,
      "learning_rate": 6.438660232756175e-05,
      "loss": 0.023449543118476867,
      "step": 125520
    },
    {
      "epoch": 35.63156400794777,
      "grad_norm": 0.3539074957370758,
      "learning_rate": 6.438376383763837e-05,
      "loss": 0.009968521445989609,
      "step": 125530
    },
    {
      "epoch": 35.634402497871136,
      "grad_norm": 0.253376305103302,
      "learning_rate": 6.438092534771501e-05,
      "loss": 0.002223910391330719,
      "step": 125540
    },
    {
      "epoch": 35.63724098779449,
      "grad_norm": 0.6831166744232178,
      "learning_rate": 6.437808685779166e-05,
      "loss": 0.0034550823271274566,
      "step": 125550
    },
    {
      "epoch": 35.640079477717855,
      "grad_norm": 0.0465189591050148,
      "learning_rate": 6.43752483678683e-05,
      "loss": 0.012002493441104888,
      "step": 125560
    },
    {
      "epoch": 35.64291796764122,
      "grad_norm": 0.03205685690045357,
      "learning_rate": 6.437240987794494e-05,
      "loss": 0.003970462456345558,
      "step": 125570
    },
    {
      "epoch": 35.64575645756457,
      "grad_norm": 1.2504947185516357,
      "learning_rate": 6.436957138802158e-05,
      "loss": 0.0035271897912025453,
      "step": 125580
    },
    {
      "epoch": 35.648594947487936,
      "grad_norm": 0.5116320848464966,
      "learning_rate": 6.436673289809821e-05,
      "loss": 0.0025082001462578773,
      "step": 125590
    },
    {
      "epoch": 35.6514334374113,
      "grad_norm": 0.22356906533241272,
      "learning_rate": 6.436389440817485e-05,
      "loss": 0.005789599567651749,
      "step": 125600
    },
    {
      "epoch": 35.654271927334655,
      "grad_norm": 0.161332905292511,
      "learning_rate": 6.436105591825149e-05,
      "loss": 0.0014455650001764298,
      "step": 125610
    },
    {
      "epoch": 35.65711041725802,
      "grad_norm": 0.19064517319202423,
      "learning_rate": 6.435821742832813e-05,
      "loss": 0.0017866864800453186,
      "step": 125620
    },
    {
      "epoch": 35.65994890718138,
      "grad_norm": 0.052568469196558,
      "learning_rate": 6.435537893840477e-05,
      "loss": 0.0048654571175575255,
      "step": 125630
    },
    {
      "epoch": 35.66278739710474,
      "grad_norm": 1.3232976198196411,
      "learning_rate": 6.435254044848141e-05,
      "loss": 0.001846633478999138,
      "step": 125640
    },
    {
      "epoch": 35.6656258870281,
      "grad_norm": 4.506834030151367,
      "learning_rate": 6.434970195855806e-05,
      "loss": 0.005079007148742676,
      "step": 125650
    },
    {
      "epoch": 35.66846437695146,
      "grad_norm": 0.17795799672603607,
      "learning_rate": 6.434686346863468e-05,
      "loss": 0.003507944941520691,
      "step": 125660
    },
    {
      "epoch": 35.671302866874825,
      "grad_norm": 5.064487934112549,
      "learning_rate": 6.434402497871133e-05,
      "loss": 0.0053322933614254,
      "step": 125670
    },
    {
      "epoch": 35.67414135679818,
      "grad_norm": 0.5494018197059631,
      "learning_rate": 6.434118648878797e-05,
      "loss": 0.0027816742658615112,
      "step": 125680
    },
    {
      "epoch": 35.676979846721544,
      "grad_norm": 0.3183806538581848,
      "learning_rate": 6.433834799886461e-05,
      "loss": 0.003345290943980217,
      "step": 125690
    },
    {
      "epoch": 35.67981833664491,
      "grad_norm": 0.848863422870636,
      "learning_rate": 6.433550950894125e-05,
      "loss": 0.007506284862756729,
      "step": 125700
    },
    {
      "epoch": 35.68265682656826,
      "grad_norm": 0.24562405049800873,
      "learning_rate": 6.433267101901789e-05,
      "loss": 0.0064976766705513,
      "step": 125710
    },
    {
      "epoch": 35.685495316491625,
      "grad_norm": 0.11981731653213501,
      "learning_rate": 6.432983252909452e-05,
      "loss": 0.0031975656747817993,
      "step": 125720
    },
    {
      "epoch": 35.68833380641499,
      "grad_norm": 0.7863903045654297,
      "learning_rate": 6.432699403917116e-05,
      "loss": 0.0029272908344864846,
      "step": 125730
    },
    {
      "epoch": 35.69117229633835,
      "grad_norm": 7.400240421295166,
      "learning_rate": 6.43241555492478e-05,
      "loss": 0.01401362419128418,
      "step": 125740
    },
    {
      "epoch": 35.69401078626171,
      "grad_norm": 0.49115607142448425,
      "learning_rate": 6.432131705932444e-05,
      "loss": 0.0011231260374188422,
      "step": 125750
    },
    {
      "epoch": 35.69684927618507,
      "grad_norm": 0.0890878215432167,
      "learning_rate": 6.431847856940108e-05,
      "loss": 0.001536744087934494,
      "step": 125760
    },
    {
      "epoch": 35.69968776610843,
      "grad_norm": 0.2775839567184448,
      "learning_rate": 6.431564007947773e-05,
      "loss": 0.005969899520277977,
      "step": 125770
    },
    {
      "epoch": 35.70252625603179,
      "grad_norm": 4.4697771072387695,
      "learning_rate": 6.431280158955437e-05,
      "loss": 0.012081263959407807,
      "step": 125780
    },
    {
      "epoch": 35.70536474595515,
      "grad_norm": 6.609982490539551,
      "learning_rate": 6.4309963099631e-05,
      "loss": 0.0028816327452659608,
      "step": 125790
    },
    {
      "epoch": 35.708203235878514,
      "grad_norm": 3.862739324569702,
      "learning_rate": 6.430712460970764e-05,
      "loss": 0.004419993609189987,
      "step": 125800
    },
    {
      "epoch": 35.71104172580187,
      "grad_norm": 0.19788038730621338,
      "learning_rate": 6.430428611978428e-05,
      "loss": 0.007346046715974807,
      "step": 125810
    },
    {
      "epoch": 35.71388021572523,
      "grad_norm": 1.3134608268737793,
      "learning_rate": 6.430144762986092e-05,
      "loss": 0.017233341932296753,
      "step": 125820
    },
    {
      "epoch": 35.716718705648596,
      "grad_norm": 0.24054394662380219,
      "learning_rate": 6.429860913993756e-05,
      "loss": 0.0023723315447568894,
      "step": 125830
    },
    {
      "epoch": 35.71955719557196,
      "grad_norm": 1.1830635070800781,
      "learning_rate": 6.42957706500142e-05,
      "loss": 0.008591021597385406,
      "step": 125840
    },
    {
      "epoch": 35.722395685495314,
      "grad_norm": 0.3654405176639557,
      "learning_rate": 6.429293216009083e-05,
      "loss": 0.004867617040872574,
      "step": 125850
    },
    {
      "epoch": 35.72523417541868,
      "grad_norm": 7.728053569793701,
      "learning_rate": 6.429009367016747e-05,
      "loss": 0.0101021908223629,
      "step": 125860
    },
    {
      "epoch": 35.72807266534204,
      "grad_norm": 0.08115047216415405,
      "learning_rate": 6.428725518024411e-05,
      "loss": 0.004279579967260361,
      "step": 125870
    },
    {
      "epoch": 35.730911155265396,
      "grad_norm": 0.019951509311795235,
      "learning_rate": 6.428441669032075e-05,
      "loss": 0.007779677212238312,
      "step": 125880
    },
    {
      "epoch": 35.73374964518876,
      "grad_norm": 0.07166038453578949,
      "learning_rate": 6.42815782003974e-05,
      "loss": 0.007013005018234253,
      "step": 125890
    },
    {
      "epoch": 35.73658813511212,
      "grad_norm": 4.195037841796875,
      "learning_rate": 6.427873971047404e-05,
      "loss": 0.014965492486953735,
      "step": 125900
    },
    {
      "epoch": 35.739426625035485,
      "grad_norm": 13.27398681640625,
      "learning_rate": 6.427590122055068e-05,
      "loss": 0.008800321817398071,
      "step": 125910
    },
    {
      "epoch": 35.74226511495884,
      "grad_norm": 0.03212512657046318,
      "learning_rate": 6.42730627306273e-05,
      "loss": 0.021278107166290285,
      "step": 125920
    },
    {
      "epoch": 35.7451036048822,
      "grad_norm": 0.21651475131511688,
      "learning_rate": 6.427022424070395e-05,
      "loss": 0.018256112933158875,
      "step": 125930
    },
    {
      "epoch": 35.747942094805566,
      "grad_norm": 11.894150733947754,
      "learning_rate": 6.426738575078059e-05,
      "loss": 0.0115379199385643,
      "step": 125940
    },
    {
      "epoch": 35.75078058472892,
      "grad_norm": 1.8476530313491821,
      "learning_rate": 6.426454726085722e-05,
      "loss": 0.012787336111068725,
      "step": 125950
    },
    {
      "epoch": 35.753619074652285,
      "grad_norm": 2.3486788272857666,
      "learning_rate": 6.426170877093387e-05,
      "loss": 0.007745042443275452,
      "step": 125960
    },
    {
      "epoch": 35.75645756457565,
      "grad_norm": 0.5078985095024109,
      "learning_rate": 6.425887028101051e-05,
      "loss": 0.003848437964916229,
      "step": 125970
    },
    {
      "epoch": 35.759296054499,
      "grad_norm": 8.703825950622559,
      "learning_rate": 6.425603179108714e-05,
      "loss": 0.0069452449679374695,
      "step": 125980
    },
    {
      "epoch": 35.762134544422366,
      "grad_norm": 0.6124324202537537,
      "learning_rate": 6.425319330116378e-05,
      "loss": 0.003283681720495224,
      "step": 125990
    },
    {
      "epoch": 35.76497303434573,
      "grad_norm": 11.745828628540039,
      "learning_rate": 6.425035481124042e-05,
      "loss": 0.007817354798316956,
      "step": 126000
    },
    {
      "epoch": 35.76497303434573,
      "eval_accuracy": 0.9734850893368093,
      "eval_loss": 0.08920043706893921,
      "eval_runtime": 32.3174,
      "eval_samples_per_second": 486.642,
      "eval_steps_per_second": 7.612,
      "step": 126000
    },
    {
      "epoch": 35.76781152426909,
      "grad_norm": 2.3494677543640137,
      "learning_rate": 6.424751632131706e-05,
      "loss": 0.008326833695173263,
      "step": 126010
    },
    {
      "epoch": 35.77065001419245,
      "grad_norm": 0.1206546351313591,
      "learning_rate": 6.42446778313937e-05,
      "loss": 0.002287722937762737,
      "step": 126020
    },
    {
      "epoch": 35.77348850411581,
      "grad_norm": 0.47574958205223083,
      "learning_rate": 6.424183934147035e-05,
      "loss": 0.00491691455245018,
      "step": 126030
    },
    {
      "epoch": 35.776326994039174,
      "grad_norm": 0.15117399394512177,
      "learning_rate": 6.423900085154699e-05,
      "loss": 0.002325243316590786,
      "step": 126040
    },
    {
      "epoch": 35.77916548396253,
      "grad_norm": 0.5581682920455933,
      "learning_rate": 6.423616236162362e-05,
      "loss": 0.004037883877754211,
      "step": 126050
    },
    {
      "epoch": 35.78200397388589,
      "grad_norm": 0.791415810585022,
      "learning_rate": 6.423332387170026e-05,
      "loss": 0.0028008786961436273,
      "step": 126060
    },
    {
      "epoch": 35.784842463809255,
      "grad_norm": 0.15140633285045624,
      "learning_rate": 6.42304853817769e-05,
      "loss": 0.005774249881505966,
      "step": 126070
    },
    {
      "epoch": 35.78768095373261,
      "grad_norm": 0.44931066036224365,
      "learning_rate": 6.422764689185353e-05,
      "loss": 0.00166766457259655,
      "step": 126080
    },
    {
      "epoch": 35.790519443655974,
      "grad_norm": 0.6648796200752258,
      "learning_rate": 6.422480840193018e-05,
      "loss": 0.001513228565454483,
      "step": 126090
    },
    {
      "epoch": 35.79335793357934,
      "grad_norm": 5.194144248962402,
      "learning_rate": 6.422196991200682e-05,
      "loss": 0.003154616430401802,
      "step": 126100
    },
    {
      "epoch": 35.7961964235027,
      "grad_norm": 2.333834409713745,
      "learning_rate": 6.421913142208345e-05,
      "loss": 0.011298684030771255,
      "step": 126110
    },
    {
      "epoch": 35.799034913426055,
      "grad_norm": 0.6022517085075378,
      "learning_rate": 6.421629293216009e-05,
      "loss": 0.006935279071331024,
      "step": 126120
    },
    {
      "epoch": 35.80187340334942,
      "grad_norm": 4.017784118652344,
      "learning_rate": 6.421345444223673e-05,
      "loss": 0.010142439603805542,
      "step": 126130
    },
    {
      "epoch": 35.80471189327278,
      "grad_norm": 0.036865461617708206,
      "learning_rate": 6.421061595231338e-05,
      "loss": 0.007009932398796081,
      "step": 126140
    },
    {
      "epoch": 35.80755038319614,
      "grad_norm": 0.4198978841304779,
      "learning_rate": 6.420777746239e-05,
      "loss": 0.002792798914015293,
      "step": 126150
    },
    {
      "epoch": 35.8103888731195,
      "grad_norm": 4.522526264190674,
      "learning_rate": 6.420493897246666e-05,
      "loss": 0.006036895141005516,
      "step": 126160
    },
    {
      "epoch": 35.81322736304286,
      "grad_norm": 0.0803339034318924,
      "learning_rate": 6.42021004825433e-05,
      "loss": 0.0018586892634630202,
      "step": 126170
    },
    {
      "epoch": 35.81606585296622,
      "grad_norm": 0.15588295459747314,
      "learning_rate": 6.419926199261993e-05,
      "loss": 0.003602844476699829,
      "step": 126180
    },
    {
      "epoch": 35.81890434288958,
      "grad_norm": 0.5945289134979248,
      "learning_rate": 6.419642350269657e-05,
      "loss": 0.005804742127656937,
      "step": 126190
    },
    {
      "epoch": 35.821742832812944,
      "grad_norm": 0.4530585706233978,
      "learning_rate": 6.419358501277321e-05,
      "loss": 0.0030143633484840393,
      "step": 126200
    },
    {
      "epoch": 35.82458132273631,
      "grad_norm": 0.060854967683553696,
      "learning_rate": 6.419074652284984e-05,
      "loss": 0.010567860305309295,
      "step": 126210
    },
    {
      "epoch": 35.82741981265966,
      "grad_norm": 0.41671624779701233,
      "learning_rate": 6.418790803292649e-05,
      "loss": 0.005266121774911881,
      "step": 126220
    },
    {
      "epoch": 35.830258302583026,
      "grad_norm": 0.2279660999774933,
      "learning_rate": 6.418506954300313e-05,
      "loss": 0.007976356893777847,
      "step": 126230
    },
    {
      "epoch": 35.83309679250639,
      "grad_norm": 0.05304163321852684,
      "learning_rate": 6.418223105307976e-05,
      "loss": 0.00886397585272789,
      "step": 126240
    },
    {
      "epoch": 35.835935282429745,
      "grad_norm": 0.10024485737085342,
      "learning_rate": 6.41793925631564e-05,
      "loss": 0.005539890751242638,
      "step": 126250
    },
    {
      "epoch": 35.83877377235311,
      "grad_norm": 10.355852127075195,
      "learning_rate": 6.417655407323304e-05,
      "loss": 0.009502045810222626,
      "step": 126260
    },
    {
      "epoch": 35.84161226227647,
      "grad_norm": 2.6603891849517822,
      "learning_rate": 6.417371558330969e-05,
      "loss": 0.001631772518157959,
      "step": 126270
    },
    {
      "epoch": 35.84445075219983,
      "grad_norm": 1.4265003204345703,
      "learning_rate": 6.417087709338631e-05,
      "loss": 0.013410523533821106,
      "step": 126280
    },
    {
      "epoch": 35.84728924212319,
      "grad_norm": 0.29218432307243347,
      "learning_rate": 6.416803860346297e-05,
      "loss": 0.0052363332360982895,
      "step": 126290
    },
    {
      "epoch": 35.85012773204655,
      "grad_norm": 0.15043163299560547,
      "learning_rate": 6.41652001135396e-05,
      "loss": 0.0010573316365480423,
      "step": 126300
    },
    {
      "epoch": 35.852966221969915,
      "grad_norm": 19.424463272094727,
      "learning_rate": 6.416236162361624e-05,
      "loss": 0.01594058722257614,
      "step": 126310
    },
    {
      "epoch": 35.85580471189327,
      "grad_norm": 0.3940888047218323,
      "learning_rate": 6.415952313369288e-05,
      "loss": 0.011520786583423615,
      "step": 126320
    },
    {
      "epoch": 35.85864320181663,
      "grad_norm": 0.31801101565361023,
      "learning_rate": 6.415668464376952e-05,
      "loss": 0.001605990156531334,
      "step": 126330
    },
    {
      "epoch": 35.861481691739996,
      "grad_norm": 0.5530164837837219,
      "learning_rate": 6.415384615384615e-05,
      "loss": 0.004565146565437317,
      "step": 126340
    },
    {
      "epoch": 35.86432018166335,
      "grad_norm": 0.14780156314373016,
      "learning_rate": 6.415100766392279e-05,
      "loss": 0.0013087816536426545,
      "step": 126350
    },
    {
      "epoch": 35.867158671586715,
      "grad_norm": 1.1247615814208984,
      "learning_rate": 6.414816917399944e-05,
      "loss": 0.006440658867359161,
      "step": 126360
    },
    {
      "epoch": 35.86999716151008,
      "grad_norm": 6.3111114501953125,
      "learning_rate": 6.414533068407607e-05,
      "loss": 0.005968622118234635,
      "step": 126370
    },
    {
      "epoch": 35.87283565143344,
      "grad_norm": 2.370638608932495,
      "learning_rate": 6.414249219415271e-05,
      "loss": 0.01015380620956421,
      "step": 126380
    },
    {
      "epoch": 35.8756741413568,
      "grad_norm": 0.6105343699455261,
      "learning_rate": 6.413965370422936e-05,
      "loss": 0.0028698213398456575,
      "step": 126390
    },
    {
      "epoch": 35.87851263128016,
      "grad_norm": 11.494352340698242,
      "learning_rate": 6.413681521430598e-05,
      "loss": 0.008557707071304321,
      "step": 126400
    },
    {
      "epoch": 35.88135112120352,
      "grad_norm": 0.4002112150192261,
      "learning_rate": 6.413397672438262e-05,
      "loss": 0.002713472954928875,
      "step": 126410
    },
    {
      "epoch": 35.88418961112688,
      "grad_norm": 6.1648736000061035,
      "learning_rate": 6.413113823445928e-05,
      "loss": 0.0023005463182926176,
      "step": 126420
    },
    {
      "epoch": 35.88702810105024,
      "grad_norm": 0.5191795229911804,
      "learning_rate": 6.412829974453591e-05,
      "loss": 0.004546607285737992,
      "step": 126430
    },
    {
      "epoch": 35.889866590973604,
      "grad_norm": 0.1352374255657196,
      "learning_rate": 6.412546125461255e-05,
      "loss": 0.011713664233684539,
      "step": 126440
    },
    {
      "epoch": 35.89270508089696,
      "grad_norm": 3.385709047317505,
      "learning_rate": 6.412262276468919e-05,
      "loss": 0.006901535391807556,
      "step": 126450
    },
    {
      "epoch": 35.89554357082032,
      "grad_norm": 1.5677236318588257,
      "learning_rate": 6.411978427476583e-05,
      "loss": 0.007343468815088272,
      "step": 126460
    },
    {
      "epoch": 35.898382060743685,
      "grad_norm": 0.2934154272079468,
      "learning_rate": 6.411694578484246e-05,
      "loss": 0.005378042161464691,
      "step": 126470
    },
    {
      "epoch": 35.90122055066705,
      "grad_norm": 0.5809357166290283,
      "learning_rate": 6.41141072949191e-05,
      "loss": 0.004280347377061844,
      "step": 126480
    },
    {
      "epoch": 35.904059040590404,
      "grad_norm": 0.8176398277282715,
      "learning_rate": 6.411126880499576e-05,
      "loss": 0.003025081381201744,
      "step": 126490
    },
    {
      "epoch": 35.90689753051377,
      "grad_norm": 0.032708168029785156,
      "learning_rate": 6.410843031507238e-05,
      "loss": 0.0015533359721302987,
      "step": 126500
    },
    {
      "epoch": 35.90689753051377,
      "eval_accuracy": 0.9737394290074395,
      "eval_loss": 0.08901378512382507,
      "eval_runtime": 32.1781,
      "eval_samples_per_second": 488.749,
      "eval_steps_per_second": 7.645,
      "step": 126500
    },
    {
      "epoch": 35.90973602043713,
      "grad_norm": 0.736684262752533,
      "learning_rate": 6.410559182514902e-05,
      "loss": 0.0011504879221320152,
      "step": 126510
    },
    {
      "epoch": 35.912574510360486,
      "grad_norm": 0.07730340212583542,
      "learning_rate": 6.410275333522567e-05,
      "loss": 0.002045847661793232,
      "step": 126520
    },
    {
      "epoch": 35.91541300028385,
      "grad_norm": 0.728770911693573,
      "learning_rate": 6.40999148453023e-05,
      "loss": 0.003300675377249718,
      "step": 126530
    },
    {
      "epoch": 35.91825149020721,
      "grad_norm": 3.564445972442627,
      "learning_rate": 6.409707635537894e-05,
      "loss": 0.0019577568396925925,
      "step": 126540
    },
    {
      "epoch": 35.92108998013057,
      "grad_norm": 3.454472303390503,
      "learning_rate": 6.409423786545558e-05,
      "loss": 0.0023244909942150118,
      "step": 126550
    },
    {
      "epoch": 35.92392847005393,
      "grad_norm": 9.89323616027832,
      "learning_rate": 6.409139937553222e-05,
      "loss": 0.004548443108797073,
      "step": 126560
    },
    {
      "epoch": 35.92676695997729,
      "grad_norm": 1.1458797454833984,
      "learning_rate": 6.408856088560886e-05,
      "loss": 0.014385245740413666,
      "step": 126570
    },
    {
      "epoch": 35.929605449900656,
      "grad_norm": 0.5685465931892395,
      "learning_rate": 6.40857223956855e-05,
      "loss": 0.003281041979789734,
      "step": 126580
    },
    {
      "epoch": 35.93244393982401,
      "grad_norm": 0.2190071940422058,
      "learning_rate": 6.408288390576214e-05,
      "loss": 0.0005115475505590439,
      "step": 126590
    },
    {
      "epoch": 35.935282429747375,
      "grad_norm": 0.2857154905796051,
      "learning_rate": 6.408004541583877e-05,
      "loss": 0.0019325120374560356,
      "step": 126600
    },
    {
      "epoch": 35.93812091967074,
      "grad_norm": 14.508387565612793,
      "learning_rate": 6.407720692591541e-05,
      "loss": 0.005317140370607376,
      "step": 126610
    },
    {
      "epoch": 35.94095940959409,
      "grad_norm": 3.4920389652252197,
      "learning_rate": 6.407436843599207e-05,
      "loss": 0.0015887487679719925,
      "step": 126620
    },
    {
      "epoch": 35.943797899517456,
      "grad_norm": 0.12757577002048492,
      "learning_rate": 6.40715299460687e-05,
      "loss": 0.005288310348987579,
      "step": 126630
    },
    {
      "epoch": 35.94663638944082,
      "grad_norm": 1.0794126987457275,
      "learning_rate": 6.406869145614534e-05,
      "loss": 0.0079939104616642,
      "step": 126640
    },
    {
      "epoch": 35.949474879364175,
      "grad_norm": 0.49123790860176086,
      "learning_rate": 6.406585296622198e-05,
      "loss": 0.007430511713027954,
      "step": 126650
    },
    {
      "epoch": 35.95231336928754,
      "grad_norm": 0.15124057233333588,
      "learning_rate": 6.40630144762986e-05,
      "loss": 0.003562916815280914,
      "step": 126660
    },
    {
      "epoch": 35.9551518592109,
      "grad_norm": 2.425823450088501,
      "learning_rate": 6.406017598637525e-05,
      "loss": 0.005152301862835884,
      "step": 126670
    },
    {
      "epoch": 35.95799034913426,
      "grad_norm": 0.26592516899108887,
      "learning_rate": 6.405733749645189e-05,
      "loss": 0.0022968359291553496,
      "step": 126680
    },
    {
      "epoch": 35.96082883905762,
      "grad_norm": 0.5596168041229248,
      "learning_rate": 6.405449900652853e-05,
      "loss": 0.007101870328187943,
      "step": 126690
    },
    {
      "epoch": 35.96366732898098,
      "grad_norm": 0.7218759059906006,
      "learning_rate": 6.405166051660517e-05,
      "loss": 0.0033067218959331513,
      "step": 126700
    },
    {
      "epoch": 35.966505818904345,
      "grad_norm": 0.9604300856590271,
      "learning_rate": 6.404882202668181e-05,
      "loss": 0.0033266060054302216,
      "step": 126710
    },
    {
      "epoch": 35.9693443088277,
      "grad_norm": 0.6491814851760864,
      "learning_rate": 6.404598353675845e-05,
      "loss": 0.006630474328994751,
      "step": 126720
    },
    {
      "epoch": 35.972182798751064,
      "grad_norm": 0.28935250639915466,
      "learning_rate": 6.404314504683508e-05,
      "loss": 0.0036221802234649657,
      "step": 126730
    },
    {
      "epoch": 35.97502128867443,
      "grad_norm": 5.347591400146484,
      "learning_rate": 6.404030655691172e-05,
      "loss": 0.004163079708814621,
      "step": 126740
    },
    {
      "epoch": 35.97785977859779,
      "grad_norm": 0.634170413017273,
      "learning_rate": 6.403746806698836e-05,
      "loss": 0.006770875304937363,
      "step": 126750
    },
    {
      "epoch": 35.980698268521145,
      "grad_norm": 0.5171111822128296,
      "learning_rate": 6.4034629577065e-05,
      "loss": 0.007034525275230408,
      "step": 126760
    },
    {
      "epoch": 35.98353675844451,
      "grad_norm": 0.5424671173095703,
      "learning_rate": 6.403179108714165e-05,
      "loss": 0.01683698445558548,
      "step": 126770
    },
    {
      "epoch": 35.98637524836787,
      "grad_norm": 12.746862411499023,
      "learning_rate": 6.402895259721829e-05,
      "loss": 0.011363303661346436,
      "step": 126780
    },
    {
      "epoch": 35.98921373829123,
      "grad_norm": 4.254042625427246,
      "learning_rate": 6.402611410729492e-05,
      "loss": 0.007727722078561783,
      "step": 126790
    },
    {
      "epoch": 35.99205222821459,
      "grad_norm": 0.5924074053764343,
      "learning_rate": 6.402327561737156e-05,
      "loss": 0.00403762012720108,
      "step": 126800
    },
    {
      "epoch": 35.99489071813795,
      "grad_norm": 0.04714516922831535,
      "learning_rate": 6.40204371274482e-05,
      "loss": 0.003626175969839096,
      "step": 126810
    },
    {
      "epoch": 35.99772920806131,
      "grad_norm": 1.6308715343475342,
      "learning_rate": 6.401759863752484e-05,
      "loss": 0.004816995561122894,
      "step": 126820
    },
    {
      "epoch": 36.00056769798467,
      "grad_norm": 0.7485119104385376,
      "learning_rate": 6.401476014760148e-05,
      "loss": 0.007126141339540482,
      "step": 126830
    },
    {
      "epoch": 36.003406187908034,
      "grad_norm": 0.2731200158596039,
      "learning_rate": 6.401192165767812e-05,
      "loss": 0.005330976843833923,
      "step": 126840
    },
    {
      "epoch": 36.0062446778314,
      "grad_norm": 7.831894874572754,
      "learning_rate": 6.400908316775476e-05,
      "loss": 0.007614757120609284,
      "step": 126850
    },
    {
      "epoch": 36.00908316775475,
      "grad_norm": 0.6452323794364929,
      "learning_rate": 6.400624467783139e-05,
      "loss": 0.014581537246704102,
      "step": 126860
    },
    {
      "epoch": 36.011921657678116,
      "grad_norm": 9.264128684997559,
      "learning_rate": 6.400340618790803e-05,
      "loss": 0.016015347838401795,
      "step": 126870
    },
    {
      "epoch": 36.01476014760148,
      "grad_norm": 0.07857528328895569,
      "learning_rate": 6.400056769798467e-05,
      "loss": 0.0005820931866765023,
      "step": 126880
    },
    {
      "epoch": 36.017598637524834,
      "grad_norm": 0.2249826341867447,
      "learning_rate": 6.399772920806132e-05,
      "loss": 0.005755890905857086,
      "step": 126890
    },
    {
      "epoch": 36.0204371274482,
      "grad_norm": 0.06895659863948822,
      "learning_rate": 6.399489071813796e-05,
      "loss": 0.008543316274881363,
      "step": 126900
    },
    {
      "epoch": 36.02327561737156,
      "grad_norm": 3.3606367111206055,
      "learning_rate": 6.39920522282146e-05,
      "loss": 0.00845354124903679,
      "step": 126910
    },
    {
      "epoch": 36.026114107294916,
      "grad_norm": 0.03915700316429138,
      "learning_rate": 6.398921373829123e-05,
      "loss": 0.003287498652935028,
      "step": 126920
    },
    {
      "epoch": 36.02895259721828,
      "grad_norm": 12.868856430053711,
      "learning_rate": 6.398637524836787e-05,
      "loss": 0.020087313652038575,
      "step": 126930
    },
    {
      "epoch": 36.03179108714164,
      "grad_norm": 0.17498674988746643,
      "learning_rate": 6.398353675844451e-05,
      "loss": 0.00724179744720459,
      "step": 126940
    },
    {
      "epoch": 36.034629577065004,
      "grad_norm": 0.6085094809532166,
      "learning_rate": 6.398069826852115e-05,
      "loss": 0.023564092814922333,
      "step": 126950
    },
    {
      "epoch": 36.03746806698836,
      "grad_norm": 2.0589208602905273,
      "learning_rate": 6.397785977859779e-05,
      "loss": 0.005112854391336441,
      "step": 126960
    },
    {
      "epoch": 36.04030655691172,
      "grad_norm": 4.473414421081543,
      "learning_rate": 6.397530513766676e-05,
      "loss": 0.028100308775901795,
      "step": 126970
    },
    {
      "epoch": 36.043145046835086,
      "grad_norm": 0.36405524611473083,
      "learning_rate": 6.39724666477434e-05,
      "loss": 0.01251579523086548,
      "step": 126980
    },
    {
      "epoch": 36.04598353675844,
      "grad_norm": 5.049113750457764,
      "learning_rate": 6.396962815782004e-05,
      "loss": 0.004196828603744507,
      "step": 126990
    },
    {
      "epoch": 36.048822026681805,
      "grad_norm": 0.3501921594142914,
      "learning_rate": 6.396678966789668e-05,
      "loss": 0.0025804173201322556,
      "step": 127000
    },
    {
      "epoch": 36.048822026681805,
      "eval_accuracy": 0.9729128250778916,
      "eval_loss": 0.09345213323831558,
      "eval_runtime": 31.7659,
      "eval_samples_per_second": 495.091,
      "eval_steps_per_second": 7.744,
      "step": 127000
    },
    {
      "epoch": 36.05166051660517,
      "grad_norm": 0.41528764367103577,
      "learning_rate": 6.396395117797332e-05,
      "loss": 0.0034525208175182343,
      "step": 127010
    },
    {
      "epoch": 36.05449900652852,
      "grad_norm": 1.0354595184326172,
      "learning_rate": 6.396111268804996e-05,
      "loss": 0.01257609874010086,
      "step": 127020
    },
    {
      "epoch": 36.057337496451886,
      "grad_norm": 4.490610599517822,
      "learning_rate": 6.39582741981266e-05,
      "loss": 0.00984746590256691,
      "step": 127030
    },
    {
      "epoch": 36.06017598637525,
      "grad_norm": 3.345027208328247,
      "learning_rate": 6.395543570820323e-05,
      "loss": 0.0022512603551149367,
      "step": 127040
    },
    {
      "epoch": 36.06301447629861,
      "grad_norm": 1.1294705867767334,
      "learning_rate": 6.395259721827988e-05,
      "loss": 0.0022397279739379884,
      "step": 127050
    },
    {
      "epoch": 36.06585296622197,
      "grad_norm": 0.031205175444483757,
      "learning_rate": 6.394975872835652e-05,
      "loss": 0.0008827924728393554,
      "step": 127060
    },
    {
      "epoch": 36.06869145614533,
      "grad_norm": 1.3361119031906128,
      "learning_rate": 6.394692023843316e-05,
      "loss": 0.002103321999311447,
      "step": 127070
    },
    {
      "epoch": 36.071529946068694,
      "grad_norm": 0.0718621164560318,
      "learning_rate": 6.39440817485098e-05,
      "loss": 0.003419715166091919,
      "step": 127080
    },
    {
      "epoch": 36.07436843599205,
      "grad_norm": 0.07301519811153412,
      "learning_rate": 6.394124325858644e-05,
      "loss": 0.010206998139619828,
      "step": 127090
    },
    {
      "epoch": 36.07720692591541,
      "grad_norm": 0.4432388246059418,
      "learning_rate": 6.393840476866307e-05,
      "loss": 0.016574613749980927,
      "step": 127100
    },
    {
      "epoch": 36.080045415838775,
      "grad_norm": 7.471394062042236,
      "learning_rate": 6.393556627873971e-05,
      "loss": 0.00944097265601158,
      "step": 127110
    },
    {
      "epoch": 36.08288390576214,
      "grad_norm": 1.7806057929992676,
      "learning_rate": 6.393272778881635e-05,
      "loss": 0.011175611615180969,
      "step": 127120
    },
    {
      "epoch": 36.085722395685494,
      "grad_norm": 0.045527562499046326,
      "learning_rate": 6.392988929889299e-05,
      "loss": 0.0006334753707051277,
      "step": 127130
    },
    {
      "epoch": 36.08856088560886,
      "grad_norm": 0.26164624094963074,
      "learning_rate": 6.392705080896963e-05,
      "loss": 0.0032220132648944854,
      "step": 127140
    },
    {
      "epoch": 36.09139937553222,
      "grad_norm": 0.5625957250595093,
      "learning_rate": 6.392421231904628e-05,
      "loss": 0.0038161464035511017,
      "step": 127150
    },
    {
      "epoch": 36.094237865455575,
      "grad_norm": 2.360600471496582,
      "learning_rate": 6.39213738291229e-05,
      "loss": 0.001096723973751068,
      "step": 127160
    },
    {
      "epoch": 36.09707635537894,
      "grad_norm": 0.0593549944460392,
      "learning_rate": 6.391853533919954e-05,
      "loss": 0.000880332849919796,
      "step": 127170
    },
    {
      "epoch": 36.0999148453023,
      "grad_norm": 3.7415199279785156,
      "learning_rate": 6.391569684927619e-05,
      "loss": 0.0011992288753390312,
      "step": 127180
    },
    {
      "epoch": 36.10275333522566,
      "grad_norm": 0.5388703346252441,
      "learning_rate": 6.391285835935283e-05,
      "loss": 0.0010394833981990815,
      "step": 127190
    },
    {
      "epoch": 36.10559182514902,
      "grad_norm": 6.951554298400879,
      "learning_rate": 6.391001986942947e-05,
      "loss": 0.00624723955988884,
      "step": 127200
    },
    {
      "epoch": 36.10843031507238,
      "grad_norm": 5.447987079620361,
      "learning_rate": 6.390718137950611e-05,
      "loss": 0.001547386683523655,
      "step": 127210
    },
    {
      "epoch": 36.111268804995746,
      "grad_norm": 1.3525289297103882,
      "learning_rate": 6.390434288958275e-05,
      "loss": 0.0023940665647387504,
      "step": 127220
    },
    {
      "epoch": 36.1141072949191,
      "grad_norm": 4.19992733001709,
      "learning_rate": 6.390150439965938e-05,
      "loss": 0.0039614934474229814,
      "step": 127230
    },
    {
      "epoch": 36.116945784842464,
      "grad_norm": 0.027442842721939087,
      "learning_rate": 6.389866590973602e-05,
      "loss": 0.001865350641310215,
      "step": 127240
    },
    {
      "epoch": 36.11978427476583,
      "grad_norm": 0.061473265290260315,
      "learning_rate": 6.389582741981266e-05,
      "loss": 0.01696058511734009,
      "step": 127250
    },
    {
      "epoch": 36.12262276468918,
      "grad_norm": 1.4688719511032104,
      "learning_rate": 6.389298892988929e-05,
      "loss": 0.003261295333504677,
      "step": 127260
    },
    {
      "epoch": 36.125461254612546,
      "grad_norm": 0.08504585176706314,
      "learning_rate": 6.389015043996595e-05,
      "loss": 0.007860887050628661,
      "step": 127270
    },
    {
      "epoch": 36.12829974453591,
      "grad_norm": 0.36169302463531494,
      "learning_rate": 6.388731195004259e-05,
      "loss": 0.009373527765274049,
      "step": 127280
    },
    {
      "epoch": 36.131138234459264,
      "grad_norm": 0.6977614760398865,
      "learning_rate": 6.388447346011921e-05,
      "loss": 0.0014830667525529862,
      "step": 127290
    },
    {
      "epoch": 36.13397672438263,
      "grad_norm": 17.43887710571289,
      "learning_rate": 6.388163497019586e-05,
      "loss": 0.007896265387535096,
      "step": 127300
    },
    {
      "epoch": 36.13681521430599,
      "grad_norm": 0.4687079191207886,
      "learning_rate": 6.38787964802725e-05,
      "loss": 0.009588398039340973,
      "step": 127310
    },
    {
      "epoch": 36.13965370422935,
      "grad_norm": 0.42098483443260193,
      "learning_rate": 6.387595799034914e-05,
      "loss": 0.0017126014456152916,
      "step": 127320
    },
    {
      "epoch": 36.14249219415271,
      "grad_norm": 0.8760297298431396,
      "learning_rate": 6.387311950042578e-05,
      "loss": 0.0017194518819451331,
      "step": 127330
    },
    {
      "epoch": 36.14533068407607,
      "grad_norm": 0.14376872777938843,
      "learning_rate": 6.387028101050242e-05,
      "loss": 0.00064933430403471,
      "step": 127340
    },
    {
      "epoch": 36.148169173999435,
      "grad_norm": 0.07841896265745163,
      "learning_rate": 6.386744252057906e-05,
      "loss": 0.0008113643154501915,
      "step": 127350
    },
    {
      "epoch": 36.15100766392279,
      "grad_norm": 0.7485700249671936,
      "learning_rate": 6.386460403065569e-05,
      "loss": 0.008631116896867751,
      "step": 127360
    },
    {
      "epoch": 36.15384615384615,
      "grad_norm": 0.13008366525173187,
      "learning_rate": 6.386176554073233e-05,
      "loss": 0.0023050157353281973,
      "step": 127370
    },
    {
      "epoch": 36.156684643769516,
      "grad_norm": 0.1320813149213791,
      "learning_rate": 6.385892705080897e-05,
      "loss": 0.0016317794099450112,
      "step": 127380
    },
    {
      "epoch": 36.15952313369287,
      "grad_norm": 1.6797621250152588,
      "learning_rate": 6.38560885608856e-05,
      "loss": 0.002636714465916157,
      "step": 127390
    },
    {
      "epoch": 36.162361623616235,
      "grad_norm": 0.061095159500837326,
      "learning_rate": 6.385325007096226e-05,
      "loss": 0.0011680079624056816,
      "step": 127400
    },
    {
      "epoch": 36.1652001135396,
      "grad_norm": 1.6298158168792725,
      "learning_rate": 6.38504115810389e-05,
      "loss": 0.011523217707872391,
      "step": 127410
    },
    {
      "epoch": 36.16803860346296,
      "grad_norm": 0.5967862606048584,
      "learning_rate": 6.384757309111553e-05,
      "loss": 0.0023915402591228483,
      "step": 127420
    },
    {
      "epoch": 36.170877093386316,
      "grad_norm": 0.1472238153219223,
      "learning_rate": 6.384473460119217e-05,
      "loss": 0.003700447082519531,
      "step": 127430
    },
    {
      "epoch": 36.17371558330968,
      "grad_norm": 0.8949035406112671,
      "learning_rate": 6.384189611126881e-05,
      "loss": 0.0012044599279761314,
      "step": 127440
    },
    {
      "epoch": 36.17655407323304,
      "grad_norm": 0.5069131851196289,
      "learning_rate": 6.383905762134545e-05,
      "loss": 0.002039975859224796,
      "step": 127450
    },
    {
      "epoch": 36.1793925631564,
      "grad_norm": 0.1094423159956932,
      "learning_rate": 6.383621913142208e-05,
      "loss": 0.004139226675033569,
      "step": 127460
    },
    {
      "epoch": 36.18223105307976,
      "grad_norm": 2.3406026363372803,
      "learning_rate": 6.383338064149873e-05,
      "loss": 0.012016795575618744,
      "step": 127470
    },
    {
      "epoch": 36.185069543003124,
      "grad_norm": 1.3236818313598633,
      "learning_rate": 6.383054215157537e-05,
      "loss": 0.00615473985671997,
      "step": 127480
    },
    {
      "epoch": 36.18790803292649,
      "grad_norm": 0.37161555886268616,
      "learning_rate": 6.3827703661652e-05,
      "loss": 0.00452272966504097,
      "step": 127490
    },
    {
      "epoch": 36.19074652284984,
      "grad_norm": 2.7792532444000244,
      "learning_rate": 6.382486517172864e-05,
      "loss": 0.002688353881239891,
      "step": 127500
    },
    {
      "epoch": 36.19074652284984,
      "eval_accuracy": 0.9731671647485216,
      "eval_loss": 0.09406692534685135,
      "eval_runtime": 31.9699,
      "eval_samples_per_second": 491.932,
      "eval_steps_per_second": 7.695,
      "step": 127500
    },
    {
      "epoch": 36.193585012773205,
      "grad_norm": 0.19656357169151306,
      "learning_rate": 6.382202668180528e-05,
      "loss": 0.0017799204215407372,
      "step": 127510
    },
    {
      "epoch": 36.19642350269657,
      "grad_norm": 16.63015365600586,
      "learning_rate": 6.381918819188191e-05,
      "loss": 0.006682169437408447,
      "step": 127520
    },
    {
      "epoch": 36.199261992619924,
      "grad_norm": 0.07477105408906937,
      "learning_rate": 6.381634970195857e-05,
      "loss": 0.007875749468803405,
      "step": 127530
    },
    {
      "epoch": 36.20210048254329,
      "grad_norm": 1.3011082410812378,
      "learning_rate": 6.381351121203521e-05,
      "loss": 0.018955110013484953,
      "step": 127540
    },
    {
      "epoch": 36.20493897246665,
      "grad_norm": 2.0958359241485596,
      "learning_rate": 6.381067272211184e-05,
      "loss": 0.004847618192434311,
      "step": 127550
    },
    {
      "epoch": 36.207777462390005,
      "grad_norm": 0.5526466369628906,
      "learning_rate": 6.380783423218848e-05,
      "loss": 0.002208054251968861,
      "step": 127560
    },
    {
      "epoch": 36.21061595231337,
      "grad_norm": 5.690003871917725,
      "learning_rate": 6.380499574226512e-05,
      "loss": 0.0014346564188599587,
      "step": 127570
    },
    {
      "epoch": 36.21345444223673,
      "grad_norm": 2.031998872756958,
      "learning_rate": 6.380215725234176e-05,
      "loss": 0.009736639261245728,
      "step": 127580
    },
    {
      "epoch": 36.216292932160094,
      "grad_norm": 4.549363136291504,
      "learning_rate": 6.379931876241839e-05,
      "loss": 0.007268596440553665,
      "step": 127590
    },
    {
      "epoch": 36.21913142208345,
      "grad_norm": 0.4121505320072174,
      "learning_rate": 6.379648027249504e-05,
      "loss": 0.003952358290553093,
      "step": 127600
    },
    {
      "epoch": 36.22196991200681,
      "grad_norm": 1.6879030466079712,
      "learning_rate": 6.379364178257168e-05,
      "loss": 0.009512005746364594,
      "step": 127610
    },
    {
      "epoch": 36.224808401930176,
      "grad_norm": 0.20126743614673615,
      "learning_rate": 6.379080329264831e-05,
      "loss": 0.0018951848149299621,
      "step": 127620
    },
    {
      "epoch": 36.22764689185353,
      "grad_norm": 0.6883359551429749,
      "learning_rate": 6.378796480272495e-05,
      "loss": 0.001357901468873024,
      "step": 127630
    },
    {
      "epoch": 36.230485381776894,
      "grad_norm": 2.239004135131836,
      "learning_rate": 6.37851263128016e-05,
      "loss": 0.00416656956076622,
      "step": 127640
    },
    {
      "epoch": 36.23332387170026,
      "grad_norm": 0.1031096801161766,
      "learning_rate": 6.378228782287822e-05,
      "loss": 0.003200557455420494,
      "step": 127650
    },
    {
      "epoch": 36.23616236162361,
      "grad_norm": 0.046665020287036896,
      "learning_rate": 6.377944933295486e-05,
      "loss": 0.008613959699869157,
      "step": 127660
    },
    {
      "epoch": 36.239000851546976,
      "grad_norm": 0.3083481788635254,
      "learning_rate": 6.377661084303152e-05,
      "loss": 0.003672643005847931,
      "step": 127670
    },
    {
      "epoch": 36.24183934147034,
      "grad_norm": 0.15197885036468506,
      "learning_rate": 6.377377235310815e-05,
      "loss": 0.014231361448764801,
      "step": 127680
    },
    {
      "epoch": 36.2446778313937,
      "grad_norm": 1.635524034500122,
      "learning_rate": 6.377093386318479e-05,
      "loss": 0.004113158956170082,
      "step": 127690
    },
    {
      "epoch": 36.24751632131706,
      "grad_norm": 7.497893333435059,
      "learning_rate": 6.376809537326143e-05,
      "loss": 0.0060790546238422396,
      "step": 127700
    },
    {
      "epoch": 36.25035481124042,
      "grad_norm": 0.03468652814626694,
      "learning_rate": 6.376525688333807e-05,
      "loss": 0.002390434592962265,
      "step": 127710
    },
    {
      "epoch": 36.25319330116378,
      "grad_norm": 10.676828384399414,
      "learning_rate": 6.37624183934147e-05,
      "loss": 0.004556234925985336,
      "step": 127720
    },
    {
      "epoch": 36.25603179108714,
      "grad_norm": 0.039073631167411804,
      "learning_rate": 6.375957990349135e-05,
      "loss": 0.0015852043405175209,
      "step": 127730
    },
    {
      "epoch": 36.2588702810105,
      "grad_norm": 0.042142949998378754,
      "learning_rate": 6.3756741413568e-05,
      "loss": 0.0017862040549516678,
      "step": 127740
    },
    {
      "epoch": 36.261708770933865,
      "grad_norm": 0.044744137674570084,
      "learning_rate": 6.375390292364462e-05,
      "loss": 0.002254301495850086,
      "step": 127750
    },
    {
      "epoch": 36.26454726085722,
      "grad_norm": 0.09166830033063889,
      "learning_rate": 6.375106443372126e-05,
      "loss": 0.004956484586000442,
      "step": 127760
    },
    {
      "epoch": 36.26738575078058,
      "grad_norm": 0.048585277050733566,
      "learning_rate": 6.37482259437979e-05,
      "loss": 0.0030336380004882812,
      "step": 127770
    },
    {
      "epoch": 36.270224240703946,
      "grad_norm": 0.986164391040802,
      "learning_rate": 6.374538745387453e-05,
      "loss": 0.002197781763970852,
      "step": 127780
    },
    {
      "epoch": 36.27306273062731,
      "grad_norm": 0.23840592801570892,
      "learning_rate": 6.374254896395117e-05,
      "loss": 0.0013004055246710778,
      "step": 127790
    },
    {
      "epoch": 36.275901220550665,
      "grad_norm": 3.966578483581543,
      "learning_rate": 6.373971047402783e-05,
      "loss": 0.0036557894200086594,
      "step": 127800
    },
    {
      "epoch": 36.27873971047403,
      "grad_norm": 1.0389047861099243,
      "learning_rate": 6.373687198410446e-05,
      "loss": 0.003493938222527504,
      "step": 127810
    },
    {
      "epoch": 36.28157820039739,
      "grad_norm": 0.13905692100524902,
      "learning_rate": 6.37340334941811e-05,
      "loss": 0.002410623989999294,
      "step": 127820
    },
    {
      "epoch": 36.28441669032075,
      "grad_norm": 0.1657152622938156,
      "learning_rate": 6.373119500425774e-05,
      "loss": 0.001114712655544281,
      "step": 127830
    },
    {
      "epoch": 36.28725518024411,
      "grad_norm": 0.09387261420488358,
      "learning_rate": 6.372835651433438e-05,
      "loss": 0.007324092090129852,
      "step": 127840
    },
    {
      "epoch": 36.29009367016747,
      "grad_norm": 0.060805726796388626,
      "learning_rate": 6.372551802441101e-05,
      "loss": 0.003158874437212944,
      "step": 127850
    },
    {
      "epoch": 36.29293216009083,
      "grad_norm": 0.34216102957725525,
      "learning_rate": 6.372267953448765e-05,
      "loss": 0.0020167721435427667,
      "step": 127860
    },
    {
      "epoch": 36.29577065001419,
      "grad_norm": 0.10189714282751083,
      "learning_rate": 6.37198410445643e-05,
      "loss": 0.00444999523460865,
      "step": 127870
    },
    {
      "epoch": 36.298609139937554,
      "grad_norm": 0.19940786063671112,
      "learning_rate": 6.371700255464093e-05,
      "loss": 0.0025221148505806925,
      "step": 127880
    },
    {
      "epoch": 36.30144762986092,
      "grad_norm": 0.19623884558677673,
      "learning_rate": 6.371416406471757e-05,
      "loss": 0.007103340327739715,
      "step": 127890
    },
    {
      "epoch": 36.30428611978427,
      "grad_norm": 0.10340545326471329,
      "learning_rate": 6.371132557479422e-05,
      "loss": 0.0008633177727460861,
      "step": 127900
    },
    {
      "epoch": 36.307124609707635,
      "grad_norm": 1.7378860712051392,
      "learning_rate": 6.370848708487084e-05,
      "loss": 0.0050704386085271835,
      "step": 127910
    },
    {
      "epoch": 36.309963099631,
      "grad_norm": 0.1937931478023529,
      "learning_rate": 6.370564859494749e-05,
      "loss": 0.002620670571923256,
      "step": 127920
    },
    {
      "epoch": 36.312801589554354,
      "grad_norm": 2.300102472305298,
      "learning_rate": 6.370281010502414e-05,
      "loss": 0.0028615819290280344,
      "step": 127930
    },
    {
      "epoch": 36.31564007947772,
      "grad_norm": 0.09869740903377533,
      "learning_rate": 6.369997161510077e-05,
      "loss": 0.005170597136020661,
      "step": 127940
    },
    {
      "epoch": 36.31847856940108,
      "grad_norm": 0.1035519540309906,
      "learning_rate": 6.369713312517741e-05,
      "loss": 0.0022327609360218046,
      "step": 127950
    },
    {
      "epoch": 36.32131705932444,
      "grad_norm": 0.04852621629834175,
      "learning_rate": 6.369429463525405e-05,
      "loss": 0.0018946433439850808,
      "step": 127960
    },
    {
      "epoch": 36.3241555492478,
      "grad_norm": 0.06586398929357529,
      "learning_rate": 6.369145614533069e-05,
      "loss": 0.0036411739885807036,
      "step": 127970
    },
    {
      "epoch": 36.32699403917116,
      "grad_norm": 0.17175254225730896,
      "learning_rate": 6.368861765540732e-05,
      "loss": 0.0009551232680678368,
      "step": 127980
    },
    {
      "epoch": 36.329832529094524,
      "grad_norm": 0.05443062260746956,
      "learning_rate": 6.368577916548396e-05,
      "loss": 0.003121565096080303,
      "step": 127990
    },
    {
      "epoch": 36.33267101901788,
      "grad_norm": 0.24511653184890747,
      "learning_rate": 6.36829406755606e-05,
      "loss": 0.00266999751329422,
      "step": 128000
    },
    {
      "epoch": 36.33267101901788,
      "eval_accuracy": 0.9730399949132066,
      "eval_loss": 0.08970324695110321,
      "eval_runtime": 31.8666,
      "eval_samples_per_second": 493.527,
      "eval_steps_per_second": 7.72,
      "step": 128000
    },
    {
      "epoch": 36.33550950894124,
      "grad_norm": 13.086302757263184,
      "learning_rate": 6.368010218563724e-05,
      "loss": 0.004339936375617981,
      "step": 128010
    },
    {
      "epoch": 36.338347998864606,
      "grad_norm": 0.25706109404563904,
      "learning_rate": 6.367726369571389e-05,
      "loss": 0.0017901195213198661,
      "step": 128020
    },
    {
      "epoch": 36.34118648878796,
      "grad_norm": 0.17140182852745056,
      "learning_rate": 6.367442520579053e-05,
      "loss": 0.0032442830502986907,
      "step": 128030
    },
    {
      "epoch": 36.344024978711325,
      "grad_norm": 0.17065833508968353,
      "learning_rate": 6.367158671586715e-05,
      "loss": 0.0031328584998846056,
      "step": 128040
    },
    {
      "epoch": 36.34686346863469,
      "grad_norm": 0.33550211787223816,
      "learning_rate": 6.36687482259438e-05,
      "loss": 0.011456573754549027,
      "step": 128050
    },
    {
      "epoch": 36.34970195855805,
      "grad_norm": 0.5969600677490234,
      "learning_rate": 6.366590973602044e-05,
      "loss": 0.004975881427526474,
      "step": 128060
    },
    {
      "epoch": 36.352540448481406,
      "grad_norm": 2.8020832538604736,
      "learning_rate": 6.366307124609708e-05,
      "loss": 0.00581776537001133,
      "step": 128070
    },
    {
      "epoch": 36.35537893840477,
      "grad_norm": 0.19912339746952057,
      "learning_rate": 6.366023275617372e-05,
      "loss": 0.0022149460390210153,
      "step": 128080
    },
    {
      "epoch": 36.35821742832813,
      "grad_norm": 1.316356897354126,
      "learning_rate": 6.365739426625036e-05,
      "loss": 0.0019860155880451202,
      "step": 128090
    },
    {
      "epoch": 36.36105591825149,
      "grad_norm": 0.10104569047689438,
      "learning_rate": 6.365455577632699e-05,
      "loss": 0.0036674156785011292,
      "step": 128100
    },
    {
      "epoch": 36.36389440817485,
      "grad_norm": 0.07789231091737747,
      "learning_rate": 6.365171728640363e-05,
      "loss": 0.0013231365010142326,
      "step": 128110
    },
    {
      "epoch": 36.36673289809821,
      "grad_norm": 1.383571982383728,
      "learning_rate": 6.364887879648027e-05,
      "loss": 0.0022407697513699533,
      "step": 128120
    },
    {
      "epoch": 36.36957138802157,
      "grad_norm": 7.491725444793701,
      "learning_rate": 6.364604030655691e-05,
      "loss": 0.0041404839605093,
      "step": 128130
    },
    {
      "epoch": 36.37240987794493,
      "grad_norm": 13.068432807922363,
      "learning_rate": 6.364320181663355e-05,
      "loss": 0.0044719275087118145,
      "step": 128140
    },
    {
      "epoch": 36.375248367868295,
      "grad_norm": 6.452056884765625,
      "learning_rate": 6.36403633267102e-05,
      "loss": 0.004169797152280807,
      "step": 128150
    },
    {
      "epoch": 36.37808685779166,
      "grad_norm": 0.021738294512033463,
      "learning_rate": 6.363752483678684e-05,
      "loss": 0.0056602489203214645,
      "step": 128160
    },
    {
      "epoch": 36.380925347715014,
      "grad_norm": 0.13982366025447845,
      "learning_rate": 6.363468634686347e-05,
      "loss": 0.008925019204616547,
      "step": 128170
    },
    {
      "epoch": 36.38376383763838,
      "grad_norm": 2.598951816558838,
      "learning_rate": 6.363184785694011e-05,
      "loss": 0.0027162447571754456,
      "step": 128180
    },
    {
      "epoch": 36.38660232756174,
      "grad_norm": 0.17163889110088348,
      "learning_rate": 6.362900936701675e-05,
      "loss": 0.003969043493270874,
      "step": 128190
    },
    {
      "epoch": 36.389440817485095,
      "grad_norm": 0.03473769500851631,
      "learning_rate": 6.362617087709339e-05,
      "loss": 0.00586436279118061,
      "step": 128200
    },
    {
      "epoch": 36.39227930740846,
      "grad_norm": 0.13865157961845398,
      "learning_rate": 6.362333238717003e-05,
      "loss": 0.0024255817756056786,
      "step": 128210
    },
    {
      "epoch": 36.39511779733182,
      "grad_norm": 0.3917488753795624,
      "learning_rate": 6.362049389724667e-05,
      "loss": 0.00288480818271637,
      "step": 128220
    },
    {
      "epoch": 36.39795628725518,
      "grad_norm": 0.09312880039215088,
      "learning_rate": 6.36176554073233e-05,
      "loss": 0.00928426831960678,
      "step": 128230
    },
    {
      "epoch": 36.40079477717854,
      "grad_norm": 0.21374282240867615,
      "learning_rate": 6.361481691739994e-05,
      "loss": 0.007262980937957764,
      "step": 128240
    },
    {
      "epoch": 36.4036332671019,
      "grad_norm": 0.11260316520929337,
      "learning_rate": 6.361197842747658e-05,
      "loss": 0.0019806986674666405,
      "step": 128250
    },
    {
      "epoch": 36.406471757025265,
      "grad_norm": 10.372431755065918,
      "learning_rate": 6.360913993755322e-05,
      "loss": 0.01027105525135994,
      "step": 128260
    },
    {
      "epoch": 36.40931024694862,
      "grad_norm": 1.049002766609192,
      "learning_rate": 6.360630144762987e-05,
      "loss": 0.0022628437727689743,
      "step": 128270
    },
    {
      "epoch": 36.412148736871984,
      "grad_norm": 1.1305917501449585,
      "learning_rate": 6.360346295770651e-05,
      "loss": 0.002162879146635532,
      "step": 128280
    },
    {
      "epoch": 36.41498722679535,
      "grad_norm": 0.3266650140285492,
      "learning_rate": 6.360062446778315e-05,
      "loss": 0.002702903561294079,
      "step": 128290
    },
    {
      "epoch": 36.4178257167187,
      "grad_norm": 0.6826332807540894,
      "learning_rate": 6.359778597785978e-05,
      "loss": 0.006277526170015335,
      "step": 128300
    },
    {
      "epoch": 36.420664206642066,
      "grad_norm": 1.6387525796890259,
      "learning_rate": 6.359494748793642e-05,
      "loss": 0.005201103165745735,
      "step": 128310
    },
    {
      "epoch": 36.42350269656543,
      "grad_norm": 6.370752334594727,
      "learning_rate": 6.359210899801306e-05,
      "loss": 0.007139098644256592,
      "step": 128320
    },
    {
      "epoch": 36.42634118648879,
      "grad_norm": 0.5603417754173279,
      "learning_rate": 6.35892705080897e-05,
      "loss": 0.006142812967300415,
      "step": 128330
    },
    {
      "epoch": 36.42917967641215,
      "grad_norm": 1.1552526950836182,
      "learning_rate": 6.358643201816634e-05,
      "loss": 0.002818172238767147,
      "step": 128340
    },
    {
      "epoch": 36.43201816633551,
      "grad_norm": 2.741763114929199,
      "learning_rate": 6.358359352824298e-05,
      "loss": 0.005219575390219689,
      "step": 128350
    },
    {
      "epoch": 36.43485665625887,
      "grad_norm": 0.9224965572357178,
      "learning_rate": 6.358075503831961e-05,
      "loss": 0.003419680893421173,
      "step": 128360
    },
    {
      "epoch": 36.43769514618223,
      "grad_norm": 3.190770387649536,
      "learning_rate": 6.357791654839625e-05,
      "loss": 0.0038978178054094315,
      "step": 128370
    },
    {
      "epoch": 36.44053363610559,
      "grad_norm": 0.05639544874429703,
      "learning_rate": 6.35750780584729e-05,
      "loss": 0.009301922470331191,
      "step": 128380
    },
    {
      "epoch": 36.443372126028954,
      "grad_norm": 9.033286094665527,
      "learning_rate": 6.357223956854954e-05,
      "loss": 0.010313080996274948,
      "step": 128390
    },
    {
      "epoch": 36.44621061595231,
      "grad_norm": 0.0202597938477993,
      "learning_rate": 6.356940107862618e-05,
      "loss": 0.0026626212522387503,
      "step": 128400
    },
    {
      "epoch": 36.44904910587567,
      "grad_norm": 0.4026896357536316,
      "learning_rate": 6.356656258870282e-05,
      "loss": 0.004579829424619675,
      "step": 128410
    },
    {
      "epoch": 36.451887595799036,
      "grad_norm": 6.656915187835693,
      "learning_rate": 6.356372409877946e-05,
      "loss": 0.018889476358890534,
      "step": 128420
    },
    {
      "epoch": 36.4547260857224,
      "grad_norm": 0.5472156405448914,
      "learning_rate": 6.356088560885609e-05,
      "loss": 0.017475323379039766,
      "step": 128430
    },
    {
      "epoch": 36.457564575645755,
      "grad_norm": 0.24752402305603027,
      "learning_rate": 6.355804711893273e-05,
      "loss": 0.005521754175424576,
      "step": 128440
    },
    {
      "epoch": 36.46040306556912,
      "grad_norm": 2.5118281841278076,
      "learning_rate": 6.355520862900937e-05,
      "loss": 0.005071083456277848,
      "step": 128450
    },
    {
      "epoch": 36.46324155549248,
      "grad_norm": 0.0630308985710144,
      "learning_rate": 6.3552370139086e-05,
      "loss": 0.004758365452289581,
      "step": 128460
    },
    {
      "epoch": 36.466080045415836,
      "grad_norm": 0.12117838859558105,
      "learning_rate": 6.354953164916265e-05,
      "loss": 0.008446036279201508,
      "step": 128470
    },
    {
      "epoch": 36.4689185353392,
      "grad_norm": 20.681432723999023,
      "learning_rate": 6.35466931592393e-05,
      "loss": 0.008392959833145142,
      "step": 128480
    },
    {
      "epoch": 36.47175702526256,
      "grad_norm": 12.232525825500488,
      "learning_rate": 6.354385466931592e-05,
      "loss": 0.009375204145908356,
      "step": 128490
    },
    {
      "epoch": 36.47459551518592,
      "grad_norm": 0.1576116383075714,
      "learning_rate": 6.354101617939256e-05,
      "loss": 0.0052847139537334446,
      "step": 128500
    },
    {
      "epoch": 36.47459551518592,
      "eval_accuracy": 0.972849240160234,
      "eval_loss": 0.09215620160102844,
      "eval_runtime": 32.7131,
      "eval_samples_per_second": 480.756,
      "eval_steps_per_second": 7.52,
      "step": 128500
    },
    {
      "epoch": 36.47743400510928,
      "grad_norm": 0.6093406081199646,
      "learning_rate": 6.35381776894692e-05,
      "loss": 0.012586894631385803,
      "step": 128510
    },
    {
      "epoch": 36.480272495032644,
      "grad_norm": 0.14331737160682678,
      "learning_rate": 6.353533919954585e-05,
      "loss": 0.005738149583339691,
      "step": 128520
    },
    {
      "epoch": 36.48311098495601,
      "grad_norm": 2.3704588413238525,
      "learning_rate": 6.353250070962249e-05,
      "loss": 0.01546432077884674,
      "step": 128530
    },
    {
      "epoch": 36.48594947487936,
      "grad_norm": 0.11676281690597534,
      "learning_rate": 6.352966221969913e-05,
      "loss": 0.008744961023330689,
      "step": 128540
    },
    {
      "epoch": 36.488787964802725,
      "grad_norm": 2.250260353088379,
      "learning_rate": 6.352682372977577e-05,
      "loss": 0.004321060329675675,
      "step": 128550
    },
    {
      "epoch": 36.49162645472609,
      "grad_norm": 0.09385251998901367,
      "learning_rate": 6.35239852398524e-05,
      "loss": 0.00914500504732132,
      "step": 128560
    },
    {
      "epoch": 36.494464944649444,
      "grad_norm": 0.29716208577156067,
      "learning_rate": 6.352114674992904e-05,
      "loss": 0.006280108541250229,
      "step": 128570
    },
    {
      "epoch": 36.49730343457281,
      "grad_norm": 0.21534784138202667,
      "learning_rate": 6.351830826000568e-05,
      "loss": 0.005754648894071579,
      "step": 128580
    },
    {
      "epoch": 36.50014192449617,
      "grad_norm": 0.10736148804426193,
      "learning_rate": 6.351546977008231e-05,
      "loss": 0.004745303839445114,
      "step": 128590
    },
    {
      "epoch": 36.502980414419525,
      "grad_norm": 0.3634936511516571,
      "learning_rate": 6.351263128015896e-05,
      "loss": 0.007020667940378189,
      "step": 128600
    },
    {
      "epoch": 36.50581890434289,
      "grad_norm": 4.621170520782471,
      "learning_rate": 6.35097927902356e-05,
      "loss": 0.005616667121648789,
      "step": 128610
    },
    {
      "epoch": 36.50865739426625,
      "grad_norm": 0.48352116346359253,
      "learning_rate": 6.350695430031223e-05,
      "loss": 0.003449523076415062,
      "step": 128620
    },
    {
      "epoch": 36.511495884189614,
      "grad_norm": 1.4400978088378906,
      "learning_rate": 6.350411581038887e-05,
      "loss": 0.015318857133388519,
      "step": 128630
    },
    {
      "epoch": 36.51433437411297,
      "grad_norm": 0.6999803781509399,
      "learning_rate": 6.350127732046552e-05,
      "loss": 0.005331846326589585,
      "step": 128640
    },
    {
      "epoch": 36.51717286403633,
      "grad_norm": 1.2224411964416504,
      "learning_rate": 6.349843883054216e-05,
      "loss": 0.0011890539899468422,
      "step": 128650
    },
    {
      "epoch": 36.520011353959696,
      "grad_norm": 3.0679917335510254,
      "learning_rate": 6.34956003406188e-05,
      "loss": 0.0034409984946250916,
      "step": 128660
    },
    {
      "epoch": 36.52284984388305,
      "grad_norm": 0.5644010901451111,
      "learning_rate": 6.349276185069544e-05,
      "loss": 0.0022447856143116953,
      "step": 128670
    },
    {
      "epoch": 36.525688333806414,
      "grad_norm": 6.425459384918213,
      "learning_rate": 6.348992336077208e-05,
      "loss": 0.0019124269485473633,
      "step": 128680
    },
    {
      "epoch": 36.52852682372978,
      "grad_norm": 0.07166341692209244,
      "learning_rate": 6.348708487084871e-05,
      "loss": 0.0013645041733980178,
      "step": 128690
    },
    {
      "epoch": 36.53136531365314,
      "grad_norm": 7.907047748565674,
      "learning_rate": 6.348424638092535e-05,
      "loss": 0.007405373454093933,
      "step": 128700
    },
    {
      "epoch": 36.534203803576496,
      "grad_norm": 17.879535675048828,
      "learning_rate": 6.348140789100199e-05,
      "loss": 0.017444223165512085,
      "step": 128710
    },
    {
      "epoch": 36.53704229349986,
      "grad_norm": 0.2733509838581085,
      "learning_rate": 6.347856940107862e-05,
      "loss": 0.01838797926902771,
      "step": 128720
    },
    {
      "epoch": 36.53988078342322,
      "grad_norm": 0.6011088490486145,
      "learning_rate": 6.347573091115527e-05,
      "loss": 0.0011929968371987344,
      "step": 128730
    },
    {
      "epoch": 36.54271927334658,
      "grad_norm": 0.311436265707016,
      "learning_rate": 6.347289242123192e-05,
      "loss": 0.0064258076250553135,
      "step": 128740
    },
    {
      "epoch": 36.54555776326994,
      "grad_norm": 0.18849098682403564,
      "learning_rate": 6.347005393130854e-05,
      "loss": 0.0012023285031318664,
      "step": 128750
    },
    {
      "epoch": 36.5483962531933,
      "grad_norm": 0.28044795989990234,
      "learning_rate": 6.346721544138518e-05,
      "loss": 0.001198618859052658,
      "step": 128760
    },
    {
      "epoch": 36.55123474311666,
      "grad_norm": 0.03143656253814697,
      "learning_rate": 6.346437695146183e-05,
      "loss": 0.0015751926228404045,
      "step": 128770
    },
    {
      "epoch": 36.55407323304002,
      "grad_norm": 2.046753406524658,
      "learning_rate": 6.346153846153847e-05,
      "loss": 0.001050523854792118,
      "step": 128780
    },
    {
      "epoch": 36.556911722963385,
      "grad_norm": 0.2764652967453003,
      "learning_rate": 6.34586999716151e-05,
      "loss": 0.010814299434423446,
      "step": 128790
    },
    {
      "epoch": 36.55975021288675,
      "grad_norm": 0.03935953602194786,
      "learning_rate": 6.345586148169175e-05,
      "loss": 0.0011992184445261955,
      "step": 128800
    },
    {
      "epoch": 36.5625887028101,
      "grad_norm": 0.3302364945411682,
      "learning_rate": 6.345302299176839e-05,
      "loss": 0.0026190012693405153,
      "step": 128810
    },
    {
      "epoch": 36.565427192733466,
      "grad_norm": 0.7445336580276489,
      "learning_rate": 6.345018450184502e-05,
      "loss": 0.001105300709605217,
      "step": 128820
    },
    {
      "epoch": 36.56826568265683,
      "grad_norm": 0.13778544962406158,
      "learning_rate": 6.344734601192166e-05,
      "loss": 0.0033204302191734312,
      "step": 128830
    },
    {
      "epoch": 36.571104172580185,
      "grad_norm": 1.4326865673065186,
      "learning_rate": 6.34445075219983e-05,
      "loss": 0.0022096866741776466,
      "step": 128840
    },
    {
      "epoch": 36.57394266250355,
      "grad_norm": 7.33085823059082,
      "learning_rate": 6.344166903207493e-05,
      "loss": 0.0032810993492603303,
      "step": 128850
    },
    {
      "epoch": 36.57678115242691,
      "grad_norm": 0.8473078012466431,
      "learning_rate": 6.343883054215158e-05,
      "loss": 0.002782110869884491,
      "step": 128860
    },
    {
      "epoch": 36.579619642350266,
      "grad_norm": 0.04309206083416939,
      "learning_rate": 6.343599205222823e-05,
      "loss": 0.005980565771460533,
      "step": 128870
    },
    {
      "epoch": 36.58245813227363,
      "grad_norm": 0.10303124040365219,
      "learning_rate": 6.343315356230485e-05,
      "loss": 0.004922766610980034,
      "step": 128880
    },
    {
      "epoch": 36.58529662219699,
      "grad_norm": 0.613318681716919,
      "learning_rate": 6.34303150723815e-05,
      "loss": 0.010484806448221206,
      "step": 128890
    },
    {
      "epoch": 36.588135112120355,
      "grad_norm": 1.4879721403121948,
      "learning_rate": 6.342747658245814e-05,
      "loss": 0.0008864721283316612,
      "step": 128900
    },
    {
      "epoch": 36.59097360204371,
      "grad_norm": 0.014926394447684288,
      "learning_rate": 6.342463809253478e-05,
      "loss": 0.002685549668967724,
      "step": 128910
    },
    {
      "epoch": 36.593812091967074,
      "grad_norm": 0.08097674697637558,
      "learning_rate": 6.34217996026114e-05,
      "loss": 0.009730508178472519,
      "step": 128920
    },
    {
      "epoch": 36.59665058189044,
      "grad_norm": 6.7847161293029785,
      "learning_rate": 6.341896111268806e-05,
      "loss": 0.003635186329483986,
      "step": 128930
    },
    {
      "epoch": 36.59948907181379,
      "grad_norm": 0.748887836933136,
      "learning_rate": 6.341612262276469e-05,
      "loss": 0.020578068494796754,
      "step": 128940
    },
    {
      "epoch": 36.602327561737155,
      "grad_norm": 6.1447434425354,
      "learning_rate": 6.341328413284133e-05,
      "loss": 0.007363961637020111,
      "step": 128950
    },
    {
      "epoch": 36.60516605166052,
      "grad_norm": 0.1851801574230194,
      "learning_rate": 6.341044564291797e-05,
      "loss": 0.008729495853185654,
      "step": 128960
    },
    {
      "epoch": 36.608004541583874,
      "grad_norm": 0.18269306421279907,
      "learning_rate": 6.340760715299461e-05,
      "loss": 0.013525110483169556,
      "step": 128970
    },
    {
      "epoch": 36.61084303150724,
      "grad_norm": 16.460660934448242,
      "learning_rate": 6.340476866307124e-05,
      "loss": 0.01247837245464325,
      "step": 128980
    },
    {
      "epoch": 36.6136815214306,
      "grad_norm": 2.8644165992736816,
      "learning_rate": 6.340193017314788e-05,
      "loss": 0.014566735923290252,
      "step": 128990
    },
    {
      "epoch": 36.61652001135396,
      "grad_norm": 0.42880287766456604,
      "learning_rate": 6.339909168322454e-05,
      "loss": 0.0036171190440654755,
      "step": 129000
    },
    {
      "epoch": 36.61652001135396,
      "eval_accuracy": 0.9737394290074395,
      "eval_loss": 0.09346146136522293,
      "eval_runtime": 32.3784,
      "eval_samples_per_second": 485.725,
      "eval_steps_per_second": 7.598,
      "step": 129000
    },
    {
      "epoch": 36.61935850127732,
      "grad_norm": 2.3125431537628174,
      "learning_rate": 6.339625319330116e-05,
      "loss": 0.005539394170045853,
      "step": 129010
    },
    {
      "epoch": 36.62219699120068,
      "grad_norm": 1.638205885887146,
      "learning_rate": 6.33934147033778e-05,
      "loss": 0.0030544910579919813,
      "step": 129020
    },
    {
      "epoch": 36.625035481124044,
      "grad_norm": 0.20315228402614594,
      "learning_rate": 6.339057621345445e-05,
      "loss": 0.006136573851108551,
      "step": 129030
    },
    {
      "epoch": 36.6278739710474,
      "grad_norm": 4.5584397315979,
      "learning_rate": 6.338773772353108e-05,
      "loss": 0.0066944994032382965,
      "step": 129040
    },
    {
      "epoch": 36.63071246097076,
      "grad_norm": 0.9973592758178711,
      "learning_rate": 6.338489923360772e-05,
      "loss": 0.008231399208307266,
      "step": 129050
    },
    {
      "epoch": 36.633550950894126,
      "grad_norm": 2.0751712322235107,
      "learning_rate": 6.338206074368437e-05,
      "loss": 0.007291973382234573,
      "step": 129060
    },
    {
      "epoch": 36.63638944081748,
      "grad_norm": 1.9103273153305054,
      "learning_rate": 6.3379222253761e-05,
      "loss": 0.01072971075773239,
      "step": 129070
    },
    {
      "epoch": 36.639227930740844,
      "grad_norm": 0.23109903931617737,
      "learning_rate": 6.337638376383764e-05,
      "loss": 0.008249661326408387,
      "step": 129080
    },
    {
      "epoch": 36.64206642066421,
      "grad_norm": 0.6804870963096619,
      "learning_rate": 6.337354527391428e-05,
      "loss": 0.004257912933826447,
      "step": 129090
    },
    {
      "epoch": 36.64490491058757,
      "grad_norm": 1.897863507270813,
      "learning_rate": 6.337070678399092e-05,
      "loss": 0.0033970408141613005,
      "step": 129100
    },
    {
      "epoch": 36.647743400510926,
      "grad_norm": 0.8044957518577576,
      "learning_rate": 6.336786829406755e-05,
      "loss": 0.007437290251255035,
      "step": 129110
    },
    {
      "epoch": 36.65058189043429,
      "grad_norm": 7.828952789306641,
      "learning_rate": 6.336502980414419e-05,
      "loss": 0.0060139715671539305,
      "step": 129120
    },
    {
      "epoch": 36.65342038035765,
      "grad_norm": 4.810834884643555,
      "learning_rate": 6.336219131422085e-05,
      "loss": 0.003734876960515976,
      "step": 129130
    },
    {
      "epoch": 36.65625887028101,
      "grad_norm": 0.24256737530231476,
      "learning_rate": 6.335935282429748e-05,
      "loss": 0.003031609021127224,
      "step": 129140
    },
    {
      "epoch": 36.65909736020437,
      "grad_norm": 1.8751940727233887,
      "learning_rate": 6.335651433437412e-05,
      "loss": 0.0058440230786800385,
      "step": 129150
    },
    {
      "epoch": 36.66193585012773,
      "grad_norm": 0.9842199683189392,
      "learning_rate": 6.335367584445076e-05,
      "loss": 0.0043347708880901335,
      "step": 129160
    },
    {
      "epoch": 36.664774340051096,
      "grad_norm": 4.380272388458252,
      "learning_rate": 6.335083735452739e-05,
      "loss": 0.011398304998874665,
      "step": 129170
    },
    {
      "epoch": 36.66761282997445,
      "grad_norm": 0.07377107441425323,
      "learning_rate": 6.334799886460403e-05,
      "loss": 0.004765710234642029,
      "step": 129180
    },
    {
      "epoch": 36.670451319897815,
      "grad_norm": 0.3610929250717163,
      "learning_rate": 6.334516037468067e-05,
      "loss": 0.006116589903831482,
      "step": 129190
    },
    {
      "epoch": 36.67328980982118,
      "grad_norm": 7.359393119812012,
      "learning_rate": 6.334232188475731e-05,
      "loss": 0.002503790892660618,
      "step": 129200
    },
    {
      "epoch": 36.67612829974453,
      "grad_norm": 0.24853478372097015,
      "learning_rate": 6.333948339483395e-05,
      "loss": 0.003330572694540024,
      "step": 129210
    },
    {
      "epoch": 36.678966789667896,
      "grad_norm": 0.043273068964481354,
      "learning_rate": 6.333664490491059e-05,
      "loss": 0.0007842764258384705,
      "step": 129220
    },
    {
      "epoch": 36.68180527959126,
      "grad_norm": 0.026256702840328217,
      "learning_rate": 6.333380641498723e-05,
      "loss": 0.004211011528968811,
      "step": 129230
    },
    {
      "epoch": 36.684643769514615,
      "grad_norm": 7.325388431549072,
      "learning_rate": 6.333096792506386e-05,
      "loss": 0.013927152752876282,
      "step": 129240
    },
    {
      "epoch": 36.68748225943798,
      "grad_norm": 0.10616528242826462,
      "learning_rate": 6.33281294351405e-05,
      "loss": 0.009010542184114456,
      "step": 129250
    },
    {
      "epoch": 36.69032074936134,
      "grad_norm": 0.27823007106781006,
      "learning_rate": 6.332529094521716e-05,
      "loss": 0.004000075161457062,
      "step": 129260
    },
    {
      "epoch": 36.693159239284704,
      "grad_norm": 0.15269601345062256,
      "learning_rate": 6.332245245529379e-05,
      "loss": 0.007821942120790482,
      "step": 129270
    },
    {
      "epoch": 36.69599772920806,
      "grad_norm": 0.46389326453208923,
      "learning_rate": 6.331961396537043e-05,
      "loss": 0.0022553984075784684,
      "step": 129280
    },
    {
      "epoch": 36.69883621913142,
      "grad_norm": 1.809531331062317,
      "learning_rate": 6.331677547544707e-05,
      "loss": 0.0018227126449346543,
      "step": 129290
    },
    {
      "epoch": 36.701674709054785,
      "grad_norm": 9.893295288085938,
      "learning_rate": 6.33139369855237e-05,
      "loss": 0.004665595665574074,
      "step": 129300
    },
    {
      "epoch": 36.70451319897814,
      "grad_norm": 0.1619918793439865,
      "learning_rate": 6.331109849560034e-05,
      "loss": 0.0009306618943810463,
      "step": 129310
    },
    {
      "epoch": 36.707351688901504,
      "grad_norm": 0.12259563058614731,
      "learning_rate": 6.330826000567698e-05,
      "loss": 0.006989867985248565,
      "step": 129320
    },
    {
      "epoch": 36.71019017882487,
      "grad_norm": 6.396764755249023,
      "learning_rate": 6.330542151575362e-05,
      "loss": 0.0029924996197223663,
      "step": 129330
    },
    {
      "epoch": 36.71302866874822,
      "grad_norm": 0.33126208186149597,
      "learning_rate": 6.330258302583026e-05,
      "loss": 0.0009299600496888161,
      "step": 129340
    },
    {
      "epoch": 36.715867158671585,
      "grad_norm": 0.2216186374425888,
      "learning_rate": 6.32997445359069e-05,
      "loss": 0.0019865959882736208,
      "step": 129350
    },
    {
      "epoch": 36.71870564859495,
      "grad_norm": 0.15415926277637482,
      "learning_rate": 6.329690604598355e-05,
      "loss": 0.0022443262860178947,
      "step": 129360
    },
    {
      "epoch": 36.72154413851831,
      "grad_norm": 1.9715499877929688,
      "learning_rate": 6.329406755606017e-05,
      "loss": 0.0027319744229316713,
      "step": 129370
    },
    {
      "epoch": 36.72438262844167,
      "grad_norm": 2.377908706665039,
      "learning_rate": 6.329122906613681e-05,
      "loss": 0.0027299953624606134,
      "step": 129380
    },
    {
      "epoch": 36.72722111836503,
      "grad_norm": 0.24208156764507294,
      "learning_rate": 6.328839057621346e-05,
      "loss": 0.006608296930789947,
      "step": 129390
    },
    {
      "epoch": 36.73005960828839,
      "grad_norm": 0.010012074373662472,
      "learning_rate": 6.32855520862901e-05,
      "loss": 0.00527685359120369,
      "step": 129400
    },
    {
      "epoch": 36.73289809821175,
      "grad_norm": 0.16252736747264862,
      "learning_rate": 6.328271359636674e-05,
      "loss": 0.005410671234130859,
      "step": 129410
    },
    {
      "epoch": 36.73573658813511,
      "grad_norm": 16.146242141723633,
      "learning_rate": 6.327987510644338e-05,
      "loss": 0.01796700358390808,
      "step": 129420
    },
    {
      "epoch": 36.738575078058474,
      "grad_norm": 14.447863578796387,
      "learning_rate": 6.327703661652001e-05,
      "loss": 0.010531289875507355,
      "step": 129430
    },
    {
      "epoch": 36.74141356798184,
      "grad_norm": 18.59294319152832,
      "learning_rate": 6.327419812659665e-05,
      "loss": 0.006996383517980575,
      "step": 129440
    },
    {
      "epoch": 36.74425205790519,
      "grad_norm": 0.24117887020111084,
      "learning_rate": 6.327135963667329e-05,
      "loss": 0.005534819141030311,
      "step": 129450
    },
    {
      "epoch": 36.747090547828556,
      "grad_norm": 2.1923325061798096,
      "learning_rate": 6.326852114674993e-05,
      "loss": 0.0020588096231222153,
      "step": 129460
    },
    {
      "epoch": 36.74992903775192,
      "grad_norm": 0.23226113617420197,
      "learning_rate": 6.326568265682657e-05,
      "loss": 0.0009400393813848495,
      "step": 129470
    },
    {
      "epoch": 36.752767527675275,
      "grad_norm": 0.8297702074050903,
      "learning_rate": 6.326284416690321e-05,
      "loss": 0.006173324584960937,
      "step": 129480
    },
    {
      "epoch": 36.75560601759864,
      "grad_norm": 2.8535878658294678,
      "learning_rate": 6.326000567697986e-05,
      "loss": 0.004022371023893356,
      "step": 129490
    },
    {
      "epoch": 36.758444507522,
      "grad_norm": 1.3793009519577026,
      "learning_rate": 6.325716718705648e-05,
      "loss": 0.0027482248842716217,
      "step": 129500
    },
    {
      "epoch": 36.758444507522,
      "eval_accuracy": 0.9741209385133847,
      "eval_loss": 0.08918464928865433,
      "eval_runtime": 31.6225,
      "eval_samples_per_second": 497.335,
      "eval_steps_per_second": 7.779,
      "step": 129500
    },
    {
      "epoch": 36.761282997445356,
      "grad_norm": 0.8029513955116272,
      "learning_rate": 6.325432869713313e-05,
      "loss": 0.00414060652256012,
      "step": 129510
    },
    {
      "epoch": 36.76412148736872,
      "grad_norm": 0.8600261807441711,
      "learning_rate": 6.325149020720977e-05,
      "loss": 0.004316479340195656,
      "step": 129520
    },
    {
      "epoch": 36.76695997729208,
      "grad_norm": 19.443056106567383,
      "learning_rate": 6.324865171728641e-05,
      "loss": 0.025584471225738526,
      "step": 129530
    },
    {
      "epoch": 36.769798467215445,
      "grad_norm": 0.18524987995624542,
      "learning_rate": 6.324581322736305e-05,
      "loss": 0.003167290985584259,
      "step": 129540
    },
    {
      "epoch": 36.7726369571388,
      "grad_norm": 0.6329717636108398,
      "learning_rate": 6.324297473743969e-05,
      "loss": 0.006857694685459137,
      "step": 129550
    },
    {
      "epoch": 36.77547544706216,
      "grad_norm": 0.5915277004241943,
      "learning_rate": 6.324013624751632e-05,
      "loss": 0.004746244102716446,
      "step": 129560
    },
    {
      "epoch": 36.778313936985526,
      "grad_norm": 3.187633752822876,
      "learning_rate": 6.323729775759296e-05,
      "loss": 0.012950533628463745,
      "step": 129570
    },
    {
      "epoch": 36.78115242690888,
      "grad_norm": 2.5667121410369873,
      "learning_rate": 6.32344592676696e-05,
      "loss": 0.00438239574432373,
      "step": 129580
    },
    {
      "epoch": 36.783990916832245,
      "grad_norm": 4.938164234161377,
      "learning_rate": 6.323162077774624e-05,
      "loss": 0.017377015948295594,
      "step": 129590
    },
    {
      "epoch": 36.78682940675561,
      "grad_norm": 0.24072721600532532,
      "learning_rate": 6.322878228782288e-05,
      "loss": 0.011724398285150529,
      "step": 129600
    },
    {
      "epoch": 36.789667896678964,
      "grad_norm": 0.3767806589603424,
      "learning_rate": 6.322594379789953e-05,
      "loss": 0.022611044347286224,
      "step": 129610
    },
    {
      "epoch": 36.79250638660233,
      "grad_norm": 0.3283828794956207,
      "learning_rate": 6.322310530797617e-05,
      "loss": 0.0015596816316246986,
      "step": 129620
    },
    {
      "epoch": 36.79534487652569,
      "grad_norm": 0.47462981939315796,
      "learning_rate": 6.32202668180528e-05,
      "loss": 0.009086674451828003,
      "step": 129630
    },
    {
      "epoch": 36.79818336644905,
      "grad_norm": 0.18554306030273438,
      "learning_rate": 6.321742832812944e-05,
      "loss": 0.008681416511535645,
      "step": 129640
    },
    {
      "epoch": 36.80102185637241,
      "grad_norm": 1.013665795326233,
      "learning_rate": 6.321458983820608e-05,
      "loss": 0.0018669061362743377,
      "step": 129650
    },
    {
      "epoch": 36.80386034629577,
      "grad_norm": 0.05997001379728317,
      "learning_rate": 6.321175134828272e-05,
      "loss": 0.00434509888291359,
      "step": 129660
    },
    {
      "epoch": 36.806698836219134,
      "grad_norm": 0.37173959612846375,
      "learning_rate": 6.320891285835936e-05,
      "loss": 0.005068081989884376,
      "step": 129670
    },
    {
      "epoch": 36.80953732614249,
      "grad_norm": 0.19517859816551208,
      "learning_rate": 6.3206074368436e-05,
      "loss": 0.010366223752498627,
      "step": 129680
    },
    {
      "epoch": 36.81237581606585,
      "grad_norm": 0.019192226231098175,
      "learning_rate": 6.320323587851263e-05,
      "loss": 0.0025984589010477066,
      "step": 129690
    },
    {
      "epoch": 36.815214305989215,
      "grad_norm": 0.21634961664676666,
      "learning_rate": 6.320039738858927e-05,
      "loss": 0.003962265327572823,
      "step": 129700
    },
    {
      "epoch": 36.81805279591257,
      "grad_norm": 4.878888130187988,
      "learning_rate": 6.319755889866591e-05,
      "loss": 0.0021356716752052307,
      "step": 129710
    },
    {
      "epoch": 36.820891285835934,
      "grad_norm": 1.1203750371932983,
      "learning_rate": 6.319472040874255e-05,
      "loss": 0.0032168880105018617,
      "step": 129720
    },
    {
      "epoch": 36.8237297757593,
      "grad_norm": 11.216403007507324,
      "learning_rate": 6.31918819188192e-05,
      "loss": 0.018211647868156433,
      "step": 129730
    },
    {
      "epoch": 36.82656826568266,
      "grad_norm": 0.4733777344226837,
      "learning_rate": 6.318904342889584e-05,
      "loss": 0.0008802158758044243,
      "step": 129740
    },
    {
      "epoch": 36.829406755606016,
      "grad_norm": 3.07503080368042,
      "learning_rate": 6.318620493897248e-05,
      "loss": 0.0013648156076669694,
      "step": 129750
    },
    {
      "epoch": 36.83224524552938,
      "grad_norm": 0.6883289217948914,
      "learning_rate": 6.31833664490491e-05,
      "loss": 0.003224961832165718,
      "step": 129760
    },
    {
      "epoch": 36.83508373545274,
      "grad_norm": 2.002540111541748,
      "learning_rate": 6.318081180811809e-05,
      "loss": 0.0060753501951694485,
      "step": 129770
    },
    {
      "epoch": 36.8379222253761,
      "grad_norm": 1.736228346824646,
      "learning_rate": 6.317797331819473e-05,
      "loss": 0.005475520342588425,
      "step": 129780
    },
    {
      "epoch": 36.84076071529946,
      "grad_norm": 0.2607590854167938,
      "learning_rate": 6.317513482827137e-05,
      "loss": 0.010686306655406952,
      "step": 129790
    },
    {
      "epoch": 36.84359920522282,
      "grad_norm": 6.973796367645264,
      "learning_rate": 6.317229633834801e-05,
      "loss": 0.019275836646556854,
      "step": 129800
    },
    {
      "epoch": 36.84643769514618,
      "grad_norm": 0.1851673126220703,
      "learning_rate": 6.316945784842464e-05,
      "loss": 0.020482569932937622,
      "step": 129810
    },
    {
      "epoch": 36.84927618506954,
      "grad_norm": 0.29992789030075073,
      "learning_rate": 6.316661935850128e-05,
      "loss": 0.010313406586647034,
      "step": 129820
    },
    {
      "epoch": 36.852114674992904,
      "grad_norm": 0.8173611760139465,
      "learning_rate": 6.316378086857792e-05,
      "loss": 0.009486261755228043,
      "step": 129830
    },
    {
      "epoch": 36.85495316491627,
      "grad_norm": 1.0202186107635498,
      "learning_rate": 6.316094237865456e-05,
      "loss": 0.0014545038342475892,
      "step": 129840
    },
    {
      "epoch": 36.85779165483962,
      "grad_norm": 2.476170063018799,
      "learning_rate": 6.31581038887312e-05,
      "loss": 0.01063292995095253,
      "step": 129850
    },
    {
      "epoch": 36.860630144762986,
      "grad_norm": 0.5094186663627625,
      "learning_rate": 6.315526539880784e-05,
      "loss": 0.004208836331963539,
      "step": 129860
    },
    {
      "epoch": 36.86346863468635,
      "grad_norm": 0.13013063371181488,
      "learning_rate": 6.315242690888447e-05,
      "loss": 0.003684304654598236,
      "step": 129870
    },
    {
      "epoch": 36.866307124609705,
      "grad_norm": 0.17290210723876953,
      "learning_rate": 6.314958841896111e-05,
      "loss": 0.00754975825548172,
      "step": 129880
    },
    {
      "epoch": 36.86914561453307,
      "grad_norm": 1.6510149240493774,
      "learning_rate": 6.314674992903775e-05,
      "loss": 0.008528110384941102,
      "step": 129890
    },
    {
      "epoch": 36.87198410445643,
      "grad_norm": 4.656164169311523,
      "learning_rate": 6.31439114391144e-05,
      "loss": 0.00805177241563797,
      "step": 129900
    },
    {
      "epoch": 36.87482259437979,
      "grad_norm": 0.12451548129320145,
      "learning_rate": 6.314107294919104e-05,
      "loss": 0.010972409695386886,
      "step": 129910
    },
    {
      "epoch": 36.87766108430315,
      "grad_norm": 0.8624584078788757,
      "learning_rate": 6.313823445926768e-05,
      "loss": 0.015606716275215149,
      "step": 129920
    },
    {
      "epoch": 36.88049957422651,
      "grad_norm": 0.5240159034729004,
      "learning_rate": 6.31353959693443e-05,
      "loss": 0.0037956714630126955,
      "step": 129930
    },
    {
      "epoch": 36.883338064149875,
      "grad_norm": 0.13141736388206482,
      "learning_rate": 6.313255747942095e-05,
      "loss": 0.009870089590549469,
      "step": 129940
    },
    {
      "epoch": 36.88617655407323,
      "grad_norm": 0.42530563473701477,
      "learning_rate": 6.312971898949759e-05,
      "loss": 0.0013246871531009674,
      "step": 129950
    },
    {
      "epoch": 36.889015043996594,
      "grad_norm": 0.3775751292705536,
      "learning_rate": 6.312688049957423e-05,
      "loss": 0.0010652609169483184,
      "step": 129960
    },
    {
      "epoch": 36.89185353391996,
      "grad_norm": 0.2522916793823242,
      "learning_rate": 6.312404200965086e-05,
      "loss": 0.0014710525050759316,
      "step": 129970
    },
    {
      "epoch": 36.89469202384331,
      "grad_norm": 0.052453719079494476,
      "learning_rate": 6.312120351972751e-05,
      "loss": 0.003587979078292847,
      "step": 129980
    },
    {
      "epoch": 36.897530513766675,
      "grad_norm": 2.422349214553833,
      "learning_rate": 6.311836502980415e-05,
      "loss": 0.0016953453421592712,
      "step": 129990
    },
    {
      "epoch": 36.90036900369004,
      "grad_norm": 0.5440014004707336,
      "learning_rate": 6.311552653988078e-05,
      "loss": 0.0029727375134825707,
      "step": 130000
    },
    {
      "epoch": 36.90036900369004,
      "eval_accuracy": 0.9686526355948369,
      "eval_loss": 0.10726909339427948,
      "eval_runtime": 32.1793,
      "eval_samples_per_second": 488.731,
      "eval_steps_per_second": 7.645,
      "step": 130000
    },
    {
      "epoch": 36.9032074936134,
      "grad_norm": 0.03966816887259483,
      "learning_rate": 6.311268804995742e-05,
      "loss": 0.002824271284043789,
      "step": 130010
    },
    {
      "epoch": 36.90604598353676,
      "grad_norm": 4.140873432159424,
      "learning_rate": 6.310984956003407e-05,
      "loss": 0.012478328496217727,
      "step": 130020
    },
    {
      "epoch": 36.90888447346012,
      "grad_norm": 0.2851111590862274,
      "learning_rate": 6.31070110701107e-05,
      "loss": 0.015599347651004791,
      "step": 130030
    },
    {
      "epoch": 36.91172296338348,
      "grad_norm": 1.598029613494873,
      "learning_rate": 6.310417258018735e-05,
      "loss": 0.0022613028064370154,
      "step": 130040
    },
    {
      "epoch": 36.91456145330684,
      "grad_norm": 3.9007551670074463,
      "learning_rate": 6.310133409026399e-05,
      "loss": 0.005545905977487564,
      "step": 130050
    },
    {
      "epoch": 36.9173999432302,
      "grad_norm": 0.20168644189834595,
      "learning_rate": 6.309849560034062e-05,
      "loss": 0.0030916605144739153,
      "step": 130060
    },
    {
      "epoch": 36.920238433153564,
      "grad_norm": 0.7060447931289673,
      "learning_rate": 6.309565711041726e-05,
      "loss": 0.004274709150195122,
      "step": 130070
    },
    {
      "epoch": 36.92307692307692,
      "grad_norm": 12.111396789550781,
      "learning_rate": 6.30928186204939e-05,
      "loss": 0.0052076026797294615,
      "step": 130080
    },
    {
      "epoch": 36.92591541300028,
      "grad_norm": 14.857709884643555,
      "learning_rate": 6.308998013057054e-05,
      "loss": 0.012939468026161194,
      "step": 130090
    },
    {
      "epoch": 36.928753902923646,
      "grad_norm": 0.8208363056182861,
      "learning_rate": 6.308714164064717e-05,
      "loss": 0.0035960406064987184,
      "step": 130100
    },
    {
      "epoch": 36.93159239284701,
      "grad_norm": 10.019550323486328,
      "learning_rate": 6.308430315072382e-05,
      "loss": 0.008321501314640045,
      "step": 130110
    },
    {
      "epoch": 36.934430882770364,
      "grad_norm": 1.8831768035888672,
      "learning_rate": 6.308146466080047e-05,
      "loss": 0.004503098130226135,
      "step": 130120
    },
    {
      "epoch": 36.93726937269373,
      "grad_norm": 2.117215633392334,
      "learning_rate": 6.30786261708771e-05,
      "loss": 0.0022235911339521406,
      "step": 130130
    },
    {
      "epoch": 36.94010786261709,
      "grad_norm": 14.876653671264648,
      "learning_rate": 6.307578768095373e-05,
      "loss": 0.005610499531030655,
      "step": 130140
    },
    {
      "epoch": 36.942946352540446,
      "grad_norm": 0.5633611083030701,
      "learning_rate": 6.307294919103038e-05,
      "loss": 0.004958879947662353,
      "step": 130150
    },
    {
      "epoch": 36.94578484246381,
      "grad_norm": 0.24144604802131653,
      "learning_rate": 6.3070110701107e-05,
      "loss": 0.006815387308597565,
      "step": 130160
    },
    {
      "epoch": 36.94862333238717,
      "grad_norm": 0.05977444350719452,
      "learning_rate": 6.306727221118365e-05,
      "loss": 0.0021437158808112144,
      "step": 130170
    },
    {
      "epoch": 36.95146182231053,
      "grad_norm": 0.5437458753585815,
      "learning_rate": 6.30644337212603e-05,
      "loss": 0.0010036014020442963,
      "step": 130180
    },
    {
      "epoch": 36.95430031223389,
      "grad_norm": 0.2767156958580017,
      "learning_rate": 6.306159523133693e-05,
      "loss": 0.004242165386676789,
      "step": 130190
    },
    {
      "epoch": 36.95713880215725,
      "grad_norm": 0.9214121103286743,
      "learning_rate": 6.305875674141357e-05,
      "loss": 0.011180636286735535,
      "step": 130200
    },
    {
      "epoch": 36.959977292080616,
      "grad_norm": 1.5011546611785889,
      "learning_rate": 6.305591825149021e-05,
      "loss": 0.002765375375747681,
      "step": 130210
    },
    {
      "epoch": 36.96281578200397,
      "grad_norm": 3.544799327850342,
      "learning_rate": 6.305307976156685e-05,
      "loss": 0.0022524427622556686,
      "step": 130220
    },
    {
      "epoch": 36.965654271927335,
      "grad_norm": 6.53395414352417,
      "learning_rate": 6.305024127164348e-05,
      "loss": 0.008183082938194275,
      "step": 130230
    },
    {
      "epoch": 36.9684927618507,
      "grad_norm": 0.36019039154052734,
      "learning_rate": 6.304740278172014e-05,
      "loss": 0.010857483744621277,
      "step": 130240
    },
    {
      "epoch": 36.97133125177405,
      "grad_norm": 1.58556067943573,
      "learning_rate": 6.304456429179678e-05,
      "loss": 0.0022740768268704414,
      "step": 130250
    },
    {
      "epoch": 36.974169741697416,
      "grad_norm": 4.802194118499756,
      "learning_rate": 6.30417258018734e-05,
      "loss": 0.008049385994672776,
      "step": 130260
    },
    {
      "epoch": 36.97700823162078,
      "grad_norm": 11.787601470947266,
      "learning_rate": 6.303888731195005e-05,
      "loss": 0.007207253575325012,
      "step": 130270
    },
    {
      "epoch": 36.97984672154414,
      "grad_norm": 0.09446095675230026,
      "learning_rate": 6.303604882202669e-05,
      "loss": 0.001432538777589798,
      "step": 130280
    },
    {
      "epoch": 36.9826852114675,
      "grad_norm": 0.24842219054698944,
      "learning_rate": 6.303321033210331e-05,
      "loss": 0.0013066751882433892,
      "step": 130290
    },
    {
      "epoch": 36.98552370139086,
      "grad_norm": 0.7556681036949158,
      "learning_rate": 6.303037184217996e-05,
      "loss": 0.0037295877933502196,
      "step": 130300
    },
    {
      "epoch": 36.98836219131422,
      "grad_norm": 0.6380088925361633,
      "learning_rate": 6.302753335225661e-05,
      "loss": 0.00798134133219719,
      "step": 130310
    },
    {
      "epoch": 36.99120068123758,
      "grad_norm": 1.0664875507354736,
      "learning_rate": 6.302469486233324e-05,
      "loss": 0.0020342197269201277,
      "step": 130320
    },
    {
      "epoch": 36.99403917116094,
      "grad_norm": 3.7254788875579834,
      "learning_rate": 6.302185637240988e-05,
      "loss": 0.00406913086771965,
      "step": 130330
    },
    {
      "epoch": 36.996877661084305,
      "grad_norm": 1.4459738731384277,
      "learning_rate": 6.301901788248652e-05,
      "loss": 0.006305234879255295,
      "step": 130340
    },
    {
      "epoch": 36.99971615100766,
      "grad_norm": 0.06888779252767563,
      "learning_rate": 6.301617939256316e-05,
      "loss": 0.0030141718685626983,
      "step": 130350
    },
    {
      "epoch": 37.002554640931024,
      "grad_norm": 0.11672481894493103,
      "learning_rate": 6.301334090263979e-05,
      "loss": 0.0013808565214276313,
      "step": 130360
    },
    {
      "epoch": 37.00539313085439,
      "grad_norm": 0.5524692535400391,
      "learning_rate": 6.301050241271645e-05,
      "loss": 0.014604221284389495,
      "step": 130370
    },
    {
      "epoch": 37.00823162077775,
      "grad_norm": 0.11652400344610214,
      "learning_rate": 6.300766392279309e-05,
      "loss": 0.004034185409545898,
      "step": 130380
    },
    {
      "epoch": 37.011070110701105,
      "grad_norm": 0.07715091854333878,
      "learning_rate": 6.300482543286971e-05,
      "loss": 0.0007389606907963753,
      "step": 130390
    },
    {
      "epoch": 37.01390860062447,
      "grad_norm": 0.17035825550556183,
      "learning_rate": 6.300198694294636e-05,
      "loss": 0.005327817052602768,
      "step": 130400
    },
    {
      "epoch": 37.01674709054783,
      "grad_norm": 0.539364755153656,
      "learning_rate": 6.2999148453023e-05,
      "loss": 0.0017512964084744453,
      "step": 130410
    },
    {
      "epoch": 37.01958558047119,
      "grad_norm": 0.06850375980138779,
      "learning_rate": 6.299630996309963e-05,
      "loss": 0.0015660757198929787,
      "step": 130420
    },
    {
      "epoch": 37.02242407039455,
      "grad_norm": 0.20674671232700348,
      "learning_rate": 6.299347147317627e-05,
      "loss": 0.0029807770624756813,
      "step": 130430
    },
    {
      "epoch": 37.02526256031791,
      "grad_norm": 1.017618179321289,
      "learning_rate": 6.299063298325292e-05,
      "loss": 0.0009050402790307998,
      "step": 130440
    },
    {
      "epoch": 37.02810105024127,
      "grad_norm": 2.5087740421295166,
      "learning_rate": 6.298779449332955e-05,
      "loss": 0.0015734119340777397,
      "step": 130450
    },
    {
      "epoch": 37.03093954016463,
      "grad_norm": 7.713961124420166,
      "learning_rate": 6.298495600340619e-05,
      "loss": 0.002575315348803997,
      "step": 130460
    },
    {
      "epoch": 37.033778030087994,
      "grad_norm": 0.03889383748173714,
      "learning_rate": 6.298211751348283e-05,
      "loss": 0.002600409835577011,
      "step": 130470
    },
    {
      "epoch": 37.03661652001136,
      "grad_norm": 7.656311511993408,
      "learning_rate": 6.297927902355947e-05,
      "loss": 0.005116645619273186,
      "step": 130480
    },
    {
      "epoch": 37.03945500993471,
      "grad_norm": 0.04063142463564873,
      "learning_rate": 6.29764405336361e-05,
      "loss": 0.0005413470789790154,
      "step": 130490
    },
    {
      "epoch": 37.042293499858076,
      "grad_norm": 0.23168174922466278,
      "learning_rate": 6.297360204371274e-05,
      "loss": 0.0012384606525301933,
      "step": 130500
    },
    {
      "epoch": 37.042293499858076,
      "eval_accuracy": 0.9766007503020283,
      "eval_loss": 0.08327371627092361,
      "eval_runtime": 32.3173,
      "eval_samples_per_second": 486.644,
      "eval_steps_per_second": 7.612,
      "step": 130500
    },
    {
      "epoch": 37.04513198978144,
      "grad_norm": 0.24813707172870636,
      "learning_rate": 6.29707635537894e-05,
      "loss": 0.005601443350315094,
      "step": 130510
    },
    {
      "epoch": 37.047970479704794,
      "grad_norm": 5.603160381317139,
      "learning_rate": 6.296792506386603e-05,
      "loss": 0.0025729740038514136,
      "step": 130520
    },
    {
      "epoch": 37.05080896962816,
      "grad_norm": 1.1618857383728027,
      "learning_rate": 6.296508657394267e-05,
      "loss": 0.0021844714879989623,
      "step": 130530
    },
    {
      "epoch": 37.05364745955152,
      "grad_norm": 0.010869505815207958,
      "learning_rate": 6.296224808401931e-05,
      "loss": 0.0009341269731521606,
      "step": 130540
    },
    {
      "epoch": 37.056485949474876,
      "grad_norm": 0.26706835627555847,
      "learning_rate": 6.295940959409594e-05,
      "loss": 0.0025830110535025597,
      "step": 130550
    },
    {
      "epoch": 37.05932443939824,
      "grad_norm": 0.13113319873809814,
      "learning_rate": 6.295657110417258e-05,
      "loss": 0.0023242367431521417,
      "step": 130560
    },
    {
      "epoch": 37.0621629293216,
      "grad_norm": 0.12004057317972183,
      "learning_rate": 6.295373261424923e-05,
      "loss": 0.008387956023216247,
      "step": 130570
    },
    {
      "epoch": 37.065001419244965,
      "grad_norm": 4.2592692375183105,
      "learning_rate": 6.295089412432586e-05,
      "loss": 0.0021336326375603677,
      "step": 130580
    },
    {
      "epoch": 37.06783990916832,
      "grad_norm": 0.03199302777647972,
      "learning_rate": 6.29480556344025e-05,
      "loss": 0.0020484507083892824,
      "step": 130590
    },
    {
      "epoch": 37.07067839909168,
      "grad_norm": 0.16396111249923706,
      "learning_rate": 6.294521714447914e-05,
      "loss": 0.0009353829547762871,
      "step": 130600
    },
    {
      "epoch": 37.073516889015046,
      "grad_norm": 0.20396865904331207,
      "learning_rate": 6.294237865455578e-05,
      "loss": 0.0007122941315174103,
      "step": 130610
    },
    {
      "epoch": 37.0763553789384,
      "grad_norm": 0.14715063571929932,
      "learning_rate": 6.293954016463241e-05,
      "loss": 0.002794131264090538,
      "step": 130620
    },
    {
      "epoch": 37.079193868861765,
      "grad_norm": 0.18097048997879028,
      "learning_rate": 6.293670167470905e-05,
      "loss": 0.0017539788037538528,
      "step": 130630
    },
    {
      "epoch": 37.08203235878513,
      "grad_norm": 0.08686074614524841,
      "learning_rate": 6.29338631847857e-05,
      "loss": 0.0007848981767892838,
      "step": 130640
    },
    {
      "epoch": 37.08487084870849,
      "grad_norm": 1.241327166557312,
      "learning_rate": 6.293102469486234e-05,
      "loss": 0.0019084749743342399,
      "step": 130650
    },
    {
      "epoch": 37.087709338631846,
      "grad_norm": 0.35967665910720825,
      "learning_rate": 6.292818620493898e-05,
      "loss": 0.0025888266041874885,
      "step": 130660
    },
    {
      "epoch": 37.09054782855521,
      "grad_norm": 2.762559652328491,
      "learning_rate": 6.292534771501562e-05,
      "loss": 0.0031750939786434175,
      "step": 130670
    },
    {
      "epoch": 37.09338631847857,
      "grad_norm": 0.22140389680862427,
      "learning_rate": 6.292250922509225e-05,
      "loss": 0.0068203344941139225,
      "step": 130680
    },
    {
      "epoch": 37.09622480840193,
      "grad_norm": 12.226556777954102,
      "learning_rate": 6.291967073516889e-05,
      "loss": 0.005363423377275467,
      "step": 130690
    },
    {
      "epoch": 37.09906329832529,
      "grad_norm": 3.410160541534424,
      "learning_rate": 6.291683224524553e-05,
      "loss": 0.0042905699461698536,
      "step": 130700
    },
    {
      "epoch": 37.101901788248654,
      "grad_norm": 2.1112289428710938,
      "learning_rate": 6.291399375532217e-05,
      "loss": 0.0014471398666501046,
      "step": 130710
    },
    {
      "epoch": 37.10474027817201,
      "grad_norm": 0.09496190398931503,
      "learning_rate": 6.291115526539881e-05,
      "loss": 0.004353401437401772,
      "step": 130720
    },
    {
      "epoch": 37.10757876809537,
      "grad_norm": 0.034049637615680695,
      "learning_rate": 6.290831677547545e-05,
      "loss": 0.005679722875356674,
      "step": 130730
    },
    {
      "epoch": 37.110417258018735,
      "grad_norm": 0.053114213049411774,
      "learning_rate": 6.290547828555208e-05,
      "loss": 0.002184465527534485,
      "step": 130740
    },
    {
      "epoch": 37.1132557479421,
      "grad_norm": 1.4256616830825806,
      "learning_rate": 6.290263979562872e-05,
      "loss": 0.0043985359370708466,
      "step": 130750
    },
    {
      "epoch": 37.116094237865454,
      "grad_norm": 0.19390933215618134,
      "learning_rate": 6.289980130570536e-05,
      "loss": 0.008119472116231919,
      "step": 130760
    },
    {
      "epoch": 37.11893272778882,
      "grad_norm": 0.5172305107116699,
      "learning_rate": 6.2896962815782e-05,
      "loss": 0.004832695424556732,
      "step": 130770
    },
    {
      "epoch": 37.12177121771218,
      "grad_norm": 0.4204552173614502,
      "learning_rate": 6.289412432585865e-05,
      "loss": 0.003250330686569214,
      "step": 130780
    },
    {
      "epoch": 37.124609707635535,
      "grad_norm": 2.3222405910491943,
      "learning_rate": 6.289128583593529e-05,
      "loss": 0.001523638889193535,
      "step": 130790
    },
    {
      "epoch": 37.1274481975589,
      "grad_norm": 0.15209142863750458,
      "learning_rate": 6.288844734601193e-05,
      "loss": 0.0020307138562202453,
      "step": 130800
    },
    {
      "epoch": 37.13028668748226,
      "grad_norm": 0.00636715954169631,
      "learning_rate": 6.288560885608856e-05,
      "loss": 0.001542331837117672,
      "step": 130810
    },
    {
      "epoch": 37.13312517740562,
      "grad_norm": 0.9724351167678833,
      "learning_rate": 6.28827703661652e-05,
      "loss": 0.005102554336190224,
      "step": 130820
    },
    {
      "epoch": 37.13596366732898,
      "grad_norm": 0.2466776818037033,
      "learning_rate": 6.287993187624184e-05,
      "loss": 0.004802029579877853,
      "step": 130830
    },
    {
      "epoch": 37.13880215725234,
      "grad_norm": 0.03119310922920704,
      "learning_rate": 6.287709338631848e-05,
      "loss": 0.0008781157433986664,
      "step": 130840
    },
    {
      "epoch": 37.141640647175706,
      "grad_norm": 0.5078030824661255,
      "learning_rate": 6.287425489639512e-05,
      "loss": 0.003531506285071373,
      "step": 130850
    },
    {
      "epoch": 37.14447913709906,
      "grad_norm": 0.30165284872055054,
      "learning_rate": 6.287141640647176e-05,
      "loss": 0.0018650492653250693,
      "step": 130860
    },
    {
      "epoch": 37.147317627022424,
      "grad_norm": 2.7572038173675537,
      "learning_rate": 6.286857791654839e-05,
      "loss": 0.0033584870398044586,
      "step": 130870
    },
    {
      "epoch": 37.15015611694579,
      "grad_norm": 3.474074363708496,
      "learning_rate": 6.286573942662503e-05,
      "loss": 0.007018668949604035,
      "step": 130880
    },
    {
      "epoch": 37.15299460686914,
      "grad_norm": 0.03878495842218399,
      "learning_rate": 6.286290093670168e-05,
      "loss": 0.0035669267177581786,
      "step": 130890
    },
    {
      "epoch": 37.155833096792506,
      "grad_norm": 28.961444854736328,
      "learning_rate": 6.286006244677832e-05,
      "loss": 0.02030410021543503,
      "step": 130900
    },
    {
      "epoch": 37.15867158671587,
      "grad_norm": 0.2948603928089142,
      "learning_rate": 6.285722395685496e-05,
      "loss": 0.00799187421798706,
      "step": 130910
    },
    {
      "epoch": 37.161510076639225,
      "grad_norm": 0.05488612502813339,
      "learning_rate": 6.28543854669316e-05,
      "loss": 0.007093751430511474,
      "step": 130920
    },
    {
      "epoch": 37.16434856656259,
      "grad_norm": 5.548788070678711,
      "learning_rate": 6.285154697700824e-05,
      "loss": 0.002365946024656296,
      "step": 130930
    },
    {
      "epoch": 37.16718705648595,
      "grad_norm": 0.16475054621696472,
      "learning_rate": 6.284870848708487e-05,
      "loss": 0.0029524678364396097,
      "step": 130940
    },
    {
      "epoch": 37.17002554640931,
      "grad_norm": 11.743864059448242,
      "learning_rate": 6.284586999716151e-05,
      "loss": 0.012582795321941375,
      "step": 130950
    },
    {
      "epoch": 37.17286403633267,
      "grad_norm": 0.1441352814435959,
      "learning_rate": 6.284303150723815e-05,
      "loss": 0.0060310795903205875,
      "step": 130960
    },
    {
      "epoch": 37.17570252625603,
      "grad_norm": 1.0272866487503052,
      "learning_rate": 6.284019301731479e-05,
      "loss": 0.002225098945200443,
      "step": 130970
    },
    {
      "epoch": 37.178541016179395,
      "grad_norm": 2.3722569942474365,
      "learning_rate": 6.283735452739143e-05,
      "loss": 0.005828509852290153,
      "step": 130980
    },
    {
      "epoch": 37.18137950610275,
      "grad_norm": 0.10453873872756958,
      "learning_rate": 6.283451603746808e-05,
      "loss": 0.002018792927265167,
      "step": 130990
    },
    {
      "epoch": 37.18421799602611,
      "grad_norm": 1.2867122888565063,
      "learning_rate": 6.28316775475447e-05,
      "loss": 0.002536172792315483,
      "step": 131000
    },
    {
      "epoch": 37.18421799602611,
      "eval_accuracy": 0.9684618808418644,
      "eval_loss": 0.11624494940042496,
      "eval_runtime": 31.7395,
      "eval_samples_per_second": 495.503,
      "eval_steps_per_second": 7.751,
      "step": 131000
    },
    {
      "epoch": 37.187056485949476,
      "grad_norm": 0.9096093773841858,
      "learning_rate": 6.282883905762134e-05,
      "loss": 0.005689398199319839,
      "step": 131010
    },
    {
      "epoch": 37.18989497587283,
      "grad_norm": 2.2749664783477783,
      "learning_rate": 6.282600056769799e-05,
      "loss": 0.01010909229516983,
      "step": 131020
    },
    {
      "epoch": 37.192733465796195,
      "grad_norm": 0.16351597011089325,
      "learning_rate": 6.282316207777463e-05,
      "loss": 0.0020475007593631746,
      "step": 131030
    },
    {
      "epoch": 37.19557195571956,
      "grad_norm": 11.037381172180176,
      "learning_rate": 6.282032358785127e-05,
      "loss": 0.004963525384664535,
      "step": 131040
    },
    {
      "epoch": 37.19841044564292,
      "grad_norm": 0.036546166986227036,
      "learning_rate": 6.281748509792791e-05,
      "loss": 0.003959363698959351,
      "step": 131050
    },
    {
      "epoch": 37.20124893556628,
      "grad_norm": 0.3034837245941162,
      "learning_rate": 6.281464660800455e-05,
      "loss": 0.0020316850394010546,
      "step": 131060
    },
    {
      "epoch": 37.20408742548964,
      "grad_norm": 0.23153412342071533,
      "learning_rate": 6.281180811808118e-05,
      "loss": 0.0018282707780599595,
      "step": 131070
    },
    {
      "epoch": 37.206925915413,
      "grad_norm": 0.09012892097234726,
      "learning_rate": 6.280896962815782e-05,
      "loss": 0.0038454826921224595,
      "step": 131080
    },
    {
      "epoch": 37.20976440533636,
      "grad_norm": 5.106657981872559,
      "learning_rate": 6.280613113823446e-05,
      "loss": 0.0024235064163804056,
      "step": 131090
    },
    {
      "epoch": 37.21260289525972,
      "grad_norm": 4.090630054473877,
      "learning_rate": 6.280329264831109e-05,
      "loss": 0.0017037678509950638,
      "step": 131100
    },
    {
      "epoch": 37.215441385183084,
      "grad_norm": 0.12549522519111633,
      "learning_rate": 6.280045415838774e-05,
      "loss": 0.004420855641365051,
      "step": 131110
    },
    {
      "epoch": 37.21827987510645,
      "grad_norm": 0.2063904106616974,
      "learning_rate": 6.279761566846439e-05,
      "loss": 0.0009550334885716439,
      "step": 131120
    },
    {
      "epoch": 37.2211183650298,
      "grad_norm": 0.27155205607414246,
      "learning_rate": 6.279477717854101e-05,
      "loss": 0.008056730031967163,
      "step": 131130
    },
    {
      "epoch": 37.223956854953165,
      "grad_norm": 0.04324699193239212,
      "learning_rate": 6.279193868861766e-05,
      "loss": 0.0013077192008495331,
      "step": 131140
    },
    {
      "epoch": 37.22679534487653,
      "grad_norm": 1.5315698385238647,
      "learning_rate": 6.27891001986943e-05,
      "loss": 0.00098564513027668,
      "step": 131150
    },
    {
      "epoch": 37.229633834799884,
      "grad_norm": 0.23244625329971313,
      "learning_rate": 6.278626170877094e-05,
      "loss": 0.0010462000966072083,
      "step": 131160
    },
    {
      "epoch": 37.23247232472325,
      "grad_norm": 0.033028729259967804,
      "learning_rate": 6.278342321884758e-05,
      "loss": 0.003750918060541153,
      "step": 131170
    },
    {
      "epoch": 37.23531081464661,
      "grad_norm": 0.16220255196094513,
      "learning_rate": 6.278058472892422e-05,
      "loss": 0.001012563519179821,
      "step": 131180
    },
    {
      "epoch": 37.238149304569966,
      "grad_norm": 0.14952395856380463,
      "learning_rate": 6.277774623900086e-05,
      "loss": 0.004271767660975456,
      "step": 131190
    },
    {
      "epoch": 37.24098779449333,
      "grad_norm": 8.265573501586914,
      "learning_rate": 6.277490774907749e-05,
      "loss": 0.003409551829099655,
      "step": 131200
    },
    {
      "epoch": 37.24382628441669,
      "grad_norm": 0.04811207577586174,
      "learning_rate": 6.277206925915413e-05,
      "loss": 0.001033814065158367,
      "step": 131210
    },
    {
      "epoch": 37.246664774340054,
      "grad_norm": 0.06130290776491165,
      "learning_rate": 6.276923076923077e-05,
      "loss": 0.004459200054407119,
      "step": 131220
    },
    {
      "epoch": 37.24950326426341,
      "grad_norm": 5.857054233551025,
      "learning_rate": 6.27663922793074e-05,
      "loss": 0.0048430532217025755,
      "step": 131230
    },
    {
      "epoch": 37.25234175418677,
      "grad_norm": 0.901811957359314,
      "learning_rate": 6.276355378938406e-05,
      "loss": 0.010946673154830933,
      "step": 131240
    },
    {
      "epoch": 37.255180244110136,
      "grad_norm": 0.024879150092601776,
      "learning_rate": 6.27607152994607e-05,
      "loss": 0.0023593151941895484,
      "step": 131250
    },
    {
      "epoch": 37.25801873403349,
      "grad_norm": 0.037461698055267334,
      "learning_rate": 6.275787680953732e-05,
      "loss": 0.0031211191788315775,
      "step": 131260
    },
    {
      "epoch": 37.260857223956855,
      "grad_norm": 0.18380141258239746,
      "learning_rate": 6.275503831961397e-05,
      "loss": 0.0008625281974673271,
      "step": 131270
    },
    {
      "epoch": 37.26369571388022,
      "grad_norm": 1.1647462844848633,
      "learning_rate": 6.275219982969061e-05,
      "loss": 0.0018008915707468987,
      "step": 131280
    },
    {
      "epoch": 37.26653420380357,
      "grad_norm": 0.023566801100969315,
      "learning_rate": 6.274936133976725e-05,
      "loss": 0.001033123955130577,
      "step": 131290
    },
    {
      "epoch": 37.269372693726936,
      "grad_norm": 0.2482381910085678,
      "learning_rate": 6.274652284984388e-05,
      "loss": 0.002074230834841728,
      "step": 131300
    },
    {
      "epoch": 37.2722111836503,
      "grad_norm": 1.1313368082046509,
      "learning_rate": 6.274368435992053e-05,
      "loss": 0.001044745184481144,
      "step": 131310
    },
    {
      "epoch": 37.27504967357366,
      "grad_norm": 0.025125112384557724,
      "learning_rate": 6.274084586999717e-05,
      "loss": 0.001435760036110878,
      "step": 131320
    },
    {
      "epoch": 37.27788816349702,
      "grad_norm": 0.06331707537174225,
      "learning_rate": 6.27380073800738e-05,
      "loss": 0.005990580841898918,
      "step": 131330
    },
    {
      "epoch": 37.28072665342038,
      "grad_norm": 0.5997627973556519,
      "learning_rate": 6.273516889015044e-05,
      "loss": 0.0007789181545376778,
      "step": 131340
    },
    {
      "epoch": 37.28356514334374,
      "grad_norm": 0.22059839963912964,
      "learning_rate": 6.273233040022708e-05,
      "loss": 0.0006396165117621422,
      "step": 131350
    },
    {
      "epoch": 37.2864036332671,
      "grad_norm": 0.05978359654545784,
      "learning_rate": 6.272949191030371e-05,
      "loss": 0.0016806956380605697,
      "step": 131360
    },
    {
      "epoch": 37.28924212319046,
      "grad_norm": 0.11193135380744934,
      "learning_rate": 6.272665342038037e-05,
      "loss": 0.001833895407617092,
      "step": 131370
    },
    {
      "epoch": 37.292080613113825,
      "grad_norm": 2.195242404937744,
      "learning_rate": 6.272381493045701e-05,
      "loss": 0.0022886825725436212,
      "step": 131380
    },
    {
      "epoch": 37.29491910303718,
      "grad_norm": 0.3864086866378784,
      "learning_rate": 6.272097644053364e-05,
      "loss": 0.018172965943813325,
      "step": 131390
    },
    {
      "epoch": 37.297757592960544,
      "grad_norm": 0.077408567070961,
      "learning_rate": 6.271813795061028e-05,
      "loss": 0.012025503814220429,
      "step": 131400
    },
    {
      "epoch": 37.30059608288391,
      "grad_norm": 0.10832397639751434,
      "learning_rate": 6.271529946068692e-05,
      "loss": 0.004276793450117111,
      "step": 131410
    },
    {
      "epoch": 37.30343457280727,
      "grad_norm": 0.3043539822101593,
      "learning_rate": 6.271246097076356e-05,
      "loss": 0.009828600287437438,
      "step": 131420
    },
    {
      "epoch": 37.306273062730625,
      "grad_norm": 0.1570448875427246,
      "learning_rate": 6.270962248084019e-05,
      "loss": 0.024268218874931337,
      "step": 131430
    },
    {
      "epoch": 37.30911155265399,
      "grad_norm": 0.11421597003936768,
      "learning_rate": 6.270678399091684e-05,
      "loss": 0.0017327869310975075,
      "step": 131440
    },
    {
      "epoch": 37.31195004257735,
      "grad_norm": 0.1929979771375656,
      "learning_rate": 6.270394550099348e-05,
      "loss": 0.003777717798948288,
      "step": 131450
    },
    {
      "epoch": 37.31478853250071,
      "grad_norm": 0.5578688979148865,
      "learning_rate": 6.270110701107011e-05,
      "loss": 0.004466625303030014,
      "step": 131460
    },
    {
      "epoch": 37.31762702242407,
      "grad_norm": 0.1220560222864151,
      "learning_rate": 6.269826852114675e-05,
      "loss": 0.005531726032495498,
      "step": 131470
    },
    {
      "epoch": 37.32046551234743,
      "grad_norm": 0.7332008481025696,
      "learning_rate": 6.26954300312234e-05,
      "loss": 0.006013903766870499,
      "step": 131480
    },
    {
      "epoch": 37.323304002270795,
      "grad_norm": 0.1988469958305359,
      "learning_rate": 6.269259154130002e-05,
      "loss": 0.0027560507878661156,
      "step": 131490
    },
    {
      "epoch": 37.32614249219415,
      "grad_norm": 1.6989895105361938,
      "learning_rate": 6.268975305137666e-05,
      "loss": 0.0019672904163599014,
      "step": 131500
    },
    {
      "epoch": 37.32614249219415,
      "eval_accuracy": 0.9741209385133847,
      "eval_loss": 0.09308131784200668,
      "eval_runtime": 32.3375,
      "eval_samples_per_second": 486.34,
      "eval_steps_per_second": 7.607,
      "step": 131500
    },
    {
      "epoch": 37.328980982117514,
      "grad_norm": 1.309671401977539,
      "learning_rate": 6.268691456145332e-05,
      "loss": 0.002148168720304966,
      "step": 131510
    },
    {
      "epoch": 37.33181947204088,
      "grad_norm": 0.16804414987564087,
      "learning_rate": 6.268407607152995e-05,
      "loss": 0.006189702451229096,
      "step": 131520
    },
    {
      "epoch": 37.33465796196423,
      "grad_norm": 1.0050872564315796,
      "learning_rate": 6.268123758160659e-05,
      "loss": 0.002596258744597435,
      "step": 131530
    },
    {
      "epoch": 37.337496451887596,
      "grad_norm": 1.9548320770263672,
      "learning_rate": 6.267839909168323e-05,
      "loss": 0.0045948468148708345,
      "step": 131540
    },
    {
      "epoch": 37.34033494181096,
      "grad_norm": 0.19570277631282806,
      "learning_rate": 6.267556060175987e-05,
      "loss": 0.0014979248866438867,
      "step": 131550
    },
    {
      "epoch": 37.343173431734314,
      "grad_norm": 0.22376661002635956,
      "learning_rate": 6.26727221118365e-05,
      "loss": 0.0027515044435858726,
      "step": 131560
    },
    {
      "epoch": 37.34601192165768,
      "grad_norm": 1.8288902044296265,
      "learning_rate": 6.266988362191315e-05,
      "loss": 0.0031682685017585754,
      "step": 131570
    },
    {
      "epoch": 37.34885041158104,
      "grad_norm": 1.6682987213134766,
      "learning_rate": 6.266704513198978e-05,
      "loss": 0.018534722924232482,
      "step": 131580
    },
    {
      "epoch": 37.3516889015044,
      "grad_norm": 0.19539980590343475,
      "learning_rate": 6.266420664206642e-05,
      "loss": 0.005206095427274704,
      "step": 131590
    },
    {
      "epoch": 37.35452739142776,
      "grad_norm": 0.12350120395421982,
      "learning_rate": 6.266136815214306e-05,
      "loss": 0.004144991934299469,
      "step": 131600
    },
    {
      "epoch": 37.35736588135112,
      "grad_norm": 1.4572023153305054,
      "learning_rate": 6.26585296622197e-05,
      "loss": 0.0028891973197460175,
      "step": 131610
    },
    {
      "epoch": 37.360204371274484,
      "grad_norm": 0.15302659571170807,
      "learning_rate": 6.265569117229633e-05,
      "loss": 0.001631498523056507,
      "step": 131620
    },
    {
      "epoch": 37.36304286119784,
      "grad_norm": 0.03710472956299782,
      "learning_rate": 6.265285268237297e-05,
      "loss": 0.007481534779071808,
      "step": 131630
    },
    {
      "epoch": 37.3658813511212,
      "grad_norm": 0.7931613326072693,
      "learning_rate": 6.265001419244963e-05,
      "loss": 0.0028386315330863,
      "step": 131640
    },
    {
      "epoch": 37.368719841044566,
      "grad_norm": 0.24774785339832306,
      "learning_rate": 6.264717570252626e-05,
      "loss": 0.003350292146205902,
      "step": 131650
    },
    {
      "epoch": 37.37155833096792,
      "grad_norm": 4.3190765380859375,
      "learning_rate": 6.26443372126029e-05,
      "loss": 0.0019821641966700555,
      "step": 131660
    },
    {
      "epoch": 37.374396820891285,
      "grad_norm": 0.13862355053424835,
      "learning_rate": 6.264149872267954e-05,
      "loss": 0.017703670263290405,
      "step": 131670
    },
    {
      "epoch": 37.37723531081465,
      "grad_norm": 0.2711879014968872,
      "learning_rate": 6.263866023275617e-05,
      "loss": 0.005402345582842827,
      "step": 131680
    },
    {
      "epoch": 37.38007380073801,
      "grad_norm": 0.2839202284812927,
      "learning_rate": 6.263582174283281e-05,
      "loss": 0.003753827512264252,
      "step": 131690
    },
    {
      "epoch": 37.382912290661366,
      "grad_norm": 2.89367413520813,
      "learning_rate": 6.263298325290946e-05,
      "loss": 0.008917482197284698,
      "step": 131700
    },
    {
      "epoch": 37.38575078058473,
      "grad_norm": 2.1834144592285156,
      "learning_rate": 6.263014476298609e-05,
      "loss": 0.006561017036437989,
      "step": 131710
    },
    {
      "epoch": 37.38858927050809,
      "grad_norm": 0.17815791070461273,
      "learning_rate": 6.262730627306273e-05,
      "loss": 0.00616636760532856,
      "step": 131720
    },
    {
      "epoch": 37.39142776043145,
      "grad_norm": 0.4269798696041107,
      "learning_rate": 6.262446778313937e-05,
      "loss": 0.013853214681148529,
      "step": 131730
    },
    {
      "epoch": 37.39426625035481,
      "grad_norm": 10.929792404174805,
      "learning_rate": 6.262162929321602e-05,
      "loss": 0.009705546498298644,
      "step": 131740
    },
    {
      "epoch": 37.39710474027817,
      "grad_norm": 0.4829752743244171,
      "learning_rate": 6.261879080329264e-05,
      "loss": 0.002416595444083214,
      "step": 131750
    },
    {
      "epoch": 37.39994323020153,
      "grad_norm": 4.949604511260986,
      "learning_rate": 6.261595231336929e-05,
      "loss": 0.006063929200172425,
      "step": 131760
    },
    {
      "epoch": 37.40278172012489,
      "grad_norm": 0.8582398295402527,
      "learning_rate": 6.261311382344594e-05,
      "loss": 0.013048869371414185,
      "step": 131770
    },
    {
      "epoch": 37.405620210048255,
      "grad_norm": 8.469538688659668,
      "learning_rate": 6.261027533352257e-05,
      "loss": 0.006428474932909012,
      "step": 131780
    },
    {
      "epoch": 37.40845869997162,
      "grad_norm": 2.118295192718506,
      "learning_rate": 6.260743684359921e-05,
      "loss": 0.009956207871437073,
      "step": 131790
    },
    {
      "epoch": 37.411297189894974,
      "grad_norm": 0.4764280617237091,
      "learning_rate": 6.260459835367585e-05,
      "loss": 0.005274946987628937,
      "step": 131800
    },
    {
      "epoch": 37.41413567981834,
      "grad_norm": 3.7286934852600098,
      "learning_rate": 6.260175986375248e-05,
      "loss": 0.004202421382069587,
      "step": 131810
    },
    {
      "epoch": 37.4169741697417,
      "grad_norm": 2.1792564392089844,
      "learning_rate": 6.259892137382912e-05,
      "loss": 0.0013472307473421097,
      "step": 131820
    },
    {
      "epoch": 37.419812659665055,
      "grad_norm": 0.14795002341270447,
      "learning_rate": 6.259608288390576e-05,
      "loss": 0.011506377160549164,
      "step": 131830
    },
    {
      "epoch": 37.42265114958842,
      "grad_norm": 1.9586554765701294,
      "learning_rate": 6.25932443939824e-05,
      "loss": 0.004862635582685471,
      "step": 131840
    },
    {
      "epoch": 37.42548963951178,
      "grad_norm": 0.7770957350730896,
      "learning_rate": 6.259040590405904e-05,
      "loss": 0.0034213945269584655,
      "step": 131850
    },
    {
      "epoch": 37.428328129435144,
      "grad_norm": 0.6758635640144348,
      "learning_rate": 6.258756741413569e-05,
      "loss": 0.010343948751688004,
      "step": 131860
    },
    {
      "epoch": 37.4311666193585,
      "grad_norm": 1.428584337234497,
      "learning_rate": 6.258472892421233e-05,
      "loss": 0.010746560245752334,
      "step": 131870
    },
    {
      "epoch": 37.43400510928186,
      "grad_norm": 0.2744644284248352,
      "learning_rate": 6.258189043428895e-05,
      "loss": 0.002670061029493809,
      "step": 131880
    },
    {
      "epoch": 37.436843599205226,
      "grad_norm": 0.11933600902557373,
      "learning_rate": 6.25790519443656e-05,
      "loss": 0.0031556248664855956,
      "step": 131890
    },
    {
      "epoch": 37.43968208912858,
      "grad_norm": 2.065722703933716,
      "learning_rate": 6.257621345444225e-05,
      "loss": 0.0010117342695593833,
      "step": 131900
    },
    {
      "epoch": 37.442520579051944,
      "grad_norm": 15.049694061279297,
      "learning_rate": 6.257337496451888e-05,
      "loss": 0.00541214719414711,
      "step": 131910
    },
    {
      "epoch": 37.44535906897531,
      "grad_norm": 0.15314093232154846,
      "learning_rate": 6.257053647459552e-05,
      "loss": 0.0013120271265506744,
      "step": 131920
    },
    {
      "epoch": 37.44819755889866,
      "grad_norm": 0.596635103225708,
      "learning_rate": 6.256769798467216e-05,
      "loss": 0.0016798578202724458,
      "step": 131930
    },
    {
      "epoch": 37.451036048822026,
      "grad_norm": 0.0935182124376297,
      "learning_rate": 6.256485949474879e-05,
      "loss": 0.0035447575151920318,
      "step": 131940
    },
    {
      "epoch": 37.45387453874539,
      "grad_norm": 6.7385478019714355,
      "learning_rate": 6.256202100482543e-05,
      "loss": 0.0028651636093854904,
      "step": 131950
    },
    {
      "epoch": 37.45671302866875,
      "grad_norm": 0.2648725211620331,
      "learning_rate": 6.255918251490207e-05,
      "loss": 0.003851566091179848,
      "step": 131960
    },
    {
      "epoch": 37.45955151859211,
      "grad_norm": 11.077371597290039,
      "learning_rate": 6.255634402497871e-05,
      "loss": 0.007340320199728012,
      "step": 131970
    },
    {
      "epoch": 37.46239000851547,
      "grad_norm": 0.17729884386062622,
      "learning_rate": 6.255350553505535e-05,
      "loss": 0.0028548333793878555,
      "step": 131980
    },
    {
      "epoch": 37.46522849843883,
      "grad_norm": 0.38628140091896057,
      "learning_rate": 6.2550667045132e-05,
      "loss": 0.004797953739762306,
      "step": 131990
    },
    {
      "epoch": 37.46806698836219,
      "grad_norm": 1.0531806945800781,
      "learning_rate": 6.254782855520864e-05,
      "loss": 0.0011209480464458466,
      "step": 132000
    },
    {
      "epoch": 37.46806698836219,
      "eval_accuracy": 0.9743752781840147,
      "eval_loss": 0.09310868382453918,
      "eval_runtime": 32.362,
      "eval_samples_per_second": 485.972,
      "eval_steps_per_second": 7.602,
      "step": 132000
    },
    {
      "epoch": 37.47090547828555,
      "grad_norm": 0.09871648997068405,
      "learning_rate": 6.254499006528527e-05,
      "loss": 0.009373778849840165,
      "step": 132010
    },
    {
      "epoch": 37.473743968208915,
      "grad_norm": 0.10134980827569962,
      "learning_rate": 6.25421515753619e-05,
      "loss": 0.0011668138206005097,
      "step": 132020
    },
    {
      "epoch": 37.47658245813227,
      "grad_norm": 1.1469206809997559,
      "learning_rate": 6.253931308543855e-05,
      "loss": 0.0013335609808564186,
      "step": 132030
    },
    {
      "epoch": 37.47942094805563,
      "grad_norm": 1.8722736835479736,
      "learning_rate": 6.253647459551519e-05,
      "loss": 0.0031882777810096742,
      "step": 132040
    },
    {
      "epoch": 37.482259437978996,
      "grad_norm": 4.049002647399902,
      "learning_rate": 6.253363610559183e-05,
      "loss": 0.001576954498887062,
      "step": 132050
    },
    {
      "epoch": 37.48509792790236,
      "grad_norm": 3.817969799041748,
      "learning_rate": 6.253079761566847e-05,
      "loss": 0.0037019729614257814,
      "step": 132060
    },
    {
      "epoch": 37.487936417825715,
      "grad_norm": 2.039700746536255,
      "learning_rate": 6.25279591257451e-05,
      "loss": 0.0018210779875516892,
      "step": 132070
    },
    {
      "epoch": 37.49077490774908,
      "grad_norm": 0.34361350536346436,
      "learning_rate": 6.252512063582174e-05,
      "loss": 0.000877358764410019,
      "step": 132080
    },
    {
      "epoch": 37.49361339767244,
      "grad_norm": 0.20258112251758575,
      "learning_rate": 6.252228214589838e-05,
      "loss": 0.014889316260814666,
      "step": 132090
    },
    {
      "epoch": 37.496451887595796,
      "grad_norm": 2.361514091491699,
      "learning_rate": 6.251944365597502e-05,
      "loss": 0.00841614305973053,
      "step": 132100
    },
    {
      "epoch": 37.49929037751916,
      "grad_norm": 1.1773335933685303,
      "learning_rate": 6.251660516605167e-05,
      "loss": 0.003871069103479385,
      "step": 132110
    },
    {
      "epoch": 37.50212886744252,
      "grad_norm": 0.18657879531383514,
      "learning_rate": 6.251376667612831e-05,
      "loss": 0.0026958780363202095,
      "step": 132120
    },
    {
      "epoch": 37.50496735736588,
      "grad_norm": 6.229252338409424,
      "learning_rate": 6.251092818620495e-05,
      "loss": 0.001969832554459572,
      "step": 132130
    },
    {
      "epoch": 37.50780584728924,
      "grad_norm": 0.45797938108444214,
      "learning_rate": 6.250808969628158e-05,
      "loss": 0.0019026946276426316,
      "step": 132140
    },
    {
      "epoch": 37.510644337212604,
      "grad_norm": 0.5916337966918945,
      "learning_rate": 6.250525120635822e-05,
      "loss": 0.007045965641736984,
      "step": 132150
    },
    {
      "epoch": 37.51348282713597,
      "grad_norm": 0.041298478841781616,
      "learning_rate": 6.250241271643486e-05,
      "loss": 0.007796673476696015,
      "step": 132160
    },
    {
      "epoch": 37.51632131705932,
      "grad_norm": 0.1929798722267151,
      "learning_rate": 6.24995742265115e-05,
      "loss": 0.007389642298221588,
      "step": 132170
    },
    {
      "epoch": 37.519159806982685,
      "grad_norm": 0.6311588287353516,
      "learning_rate": 6.249673573658814e-05,
      "loss": 0.004484188556671142,
      "step": 132180
    },
    {
      "epoch": 37.52199829690605,
      "grad_norm": 18.881084442138672,
      "learning_rate": 6.249389724666478e-05,
      "loss": 0.005851928517222405,
      "step": 132190
    },
    {
      "epoch": 37.524836786829404,
      "grad_norm": 2.152219533920288,
      "learning_rate": 6.249105875674141e-05,
      "loss": 0.011918966472148896,
      "step": 132200
    },
    {
      "epoch": 37.52767527675277,
      "grad_norm": 3.280728816986084,
      "learning_rate": 6.248822026681805e-05,
      "loss": 0.007176411151885986,
      "step": 132210
    },
    {
      "epoch": 37.53051376667613,
      "grad_norm": 0.31383973360061646,
      "learning_rate": 6.24853817768947e-05,
      "loss": 0.005173692479729652,
      "step": 132220
    },
    {
      "epoch": 37.533352256599485,
      "grad_norm": 0.20483887195587158,
      "learning_rate": 6.248254328697133e-05,
      "loss": 0.005251907557249069,
      "step": 132230
    },
    {
      "epoch": 37.53619074652285,
      "grad_norm": 0.15790414810180664,
      "learning_rate": 6.247970479704798e-05,
      "loss": 0.0015483779832720756,
      "step": 132240
    },
    {
      "epoch": 37.53902923644621,
      "grad_norm": 1.085664987564087,
      "learning_rate": 6.247686630712462e-05,
      "loss": 0.006709817796945572,
      "step": 132250
    },
    {
      "epoch": 37.541867726369574,
      "grad_norm": 4.931957721710205,
      "learning_rate": 6.247402781720126e-05,
      "loss": 0.0018745260313153267,
      "step": 132260
    },
    {
      "epoch": 37.54470621629293,
      "grad_norm": 0.7235962152481079,
      "learning_rate": 6.247118932727789e-05,
      "loss": 0.015942877531051634,
      "step": 132270
    },
    {
      "epoch": 37.54754470621629,
      "grad_norm": 0.965523362159729,
      "learning_rate": 6.246835083735453e-05,
      "loss": 0.00424588993191719,
      "step": 132280
    },
    {
      "epoch": 37.550383196139656,
      "grad_norm": 0.08497064560651779,
      "learning_rate": 6.246551234743117e-05,
      "loss": 0.001767456904053688,
      "step": 132290
    },
    {
      "epoch": 37.55322168606301,
      "grad_norm": 4.354334831237793,
      "learning_rate": 6.246267385750781e-05,
      "loss": 0.0019675798714160917,
      "step": 132300
    },
    {
      "epoch": 37.556060175986374,
      "grad_norm": 1.51112699508667,
      "learning_rate": 6.245983536758445e-05,
      "loss": 0.0034618955105543137,
      "step": 132310
    },
    {
      "epoch": 37.55889866590974,
      "grad_norm": 2.4551963806152344,
      "learning_rate": 6.24569968776611e-05,
      "loss": 0.004053299129009247,
      "step": 132320
    },
    {
      "epoch": 37.5617371558331,
      "grad_norm": 0.8450458645820618,
      "learning_rate": 6.245415838773772e-05,
      "loss": 0.011071632802486419,
      "step": 132330
    },
    {
      "epoch": 37.564575645756456,
      "grad_norm": 0.1548750400543213,
      "learning_rate": 6.245131989781436e-05,
      "loss": 0.0032876778393983843,
      "step": 132340
    },
    {
      "epoch": 37.56741413567982,
      "grad_norm": 0.12953384220600128,
      "learning_rate": 6.2448481407891e-05,
      "loss": 0.0035818852484226227,
      "step": 132350
    },
    {
      "epoch": 37.57025262560318,
      "grad_norm": 0.18858279287815094,
      "learning_rate": 6.244564291796765e-05,
      "loss": 0.0031421855092048645,
      "step": 132360
    },
    {
      "epoch": 37.57309111552654,
      "grad_norm": 1.0477215051651,
      "learning_rate": 6.244280442804429e-05,
      "loss": 0.0027070142328739165,
      "step": 132370
    },
    {
      "epoch": 37.5759296054499,
      "grad_norm": 5.662510871887207,
      "learning_rate": 6.243996593812093e-05,
      "loss": 0.003234283998608589,
      "step": 132380
    },
    {
      "epoch": 37.57876809537326,
      "grad_norm": 0.20722918212413788,
      "learning_rate": 6.243712744819757e-05,
      "loss": 0.002323152311146259,
      "step": 132390
    },
    {
      "epoch": 37.58160658529662,
      "grad_norm": 0.2880652844905853,
      "learning_rate": 6.24342889582742e-05,
      "loss": 0.012477408349514007,
      "step": 132400
    },
    {
      "epoch": 37.58444507521998,
      "grad_norm": 0.29456761479377747,
      "learning_rate": 6.243145046835084e-05,
      "loss": 0.02442405968904495,
      "step": 132410
    },
    {
      "epoch": 37.587283565143345,
      "grad_norm": 0.1021864265203476,
      "learning_rate": 6.242861197842748e-05,
      "loss": 0.004192773252725601,
      "step": 132420
    },
    {
      "epoch": 37.59012205506671,
      "grad_norm": 0.7154480814933777,
      "learning_rate": 6.242577348850411e-05,
      "loss": 0.002216958999633789,
      "step": 132430
    },
    {
      "epoch": 37.59296054499006,
      "grad_norm": 1.254004716873169,
      "learning_rate": 6.242293499858076e-05,
      "loss": 0.0014498740434646606,
      "step": 132440
    },
    {
      "epoch": 37.595799034913426,
      "grad_norm": 0.41714709997177124,
      "learning_rate": 6.24200965086574e-05,
      "loss": 0.0025337714701890945,
      "step": 132450
    },
    {
      "epoch": 37.59863752483679,
      "grad_norm": 0.6684608459472656,
      "learning_rate": 6.241725801873403e-05,
      "loss": 0.0041950389742851256,
      "step": 132460
    },
    {
      "epoch": 37.601476014760145,
      "grad_norm": 0.4297643303871155,
      "learning_rate": 6.241441952881067e-05,
      "loss": 0.0021754536777734757,
      "step": 132470
    },
    {
      "epoch": 37.60431450468351,
      "grad_norm": 7.331753253936768,
      "learning_rate": 6.241158103888732e-05,
      "loss": 0.008207141607999801,
      "step": 132480
    },
    {
      "epoch": 37.60715299460687,
      "grad_norm": 10.085725784301758,
      "learning_rate": 6.240874254896396e-05,
      "loss": 0.024471521377563477,
      "step": 132490
    },
    {
      "epoch": 37.60999148453023,
      "grad_norm": 0.057595256716012955,
      "learning_rate": 6.24059040590406e-05,
      "loss": 0.008684705942869186,
      "step": 132500
    },
    {
      "epoch": 37.60999148453023,
      "eval_accuracy": 0.9724677306542888,
      "eval_loss": 0.09692180156707764,
      "eval_runtime": 32.112,
      "eval_samples_per_second": 489.754,
      "eval_steps_per_second": 7.661,
      "step": 132500
    },
    {
      "epoch": 37.61282997445359,
      "grad_norm": 0.2997278869152069,
      "learning_rate": 6.240306556911724e-05,
      "loss": 0.005181872844696045,
      "step": 132510
    },
    {
      "epoch": 37.61566846437695,
      "grad_norm": 0.3243466019630432,
      "learning_rate": 6.240022707919387e-05,
      "loss": 0.0038658004254102708,
      "step": 132520
    },
    {
      "epoch": 37.618506954300315,
      "grad_norm": 0.10478179901838303,
      "learning_rate": 6.239738858927051e-05,
      "loss": 0.005461958795785904,
      "step": 132530
    },
    {
      "epoch": 37.62134544422367,
      "grad_norm": 0.567054033279419,
      "learning_rate": 6.239455009934715e-05,
      "loss": 0.006647210568189621,
      "step": 132540
    },
    {
      "epoch": 37.624183934147034,
      "grad_norm": 0.359387069940567,
      "learning_rate": 6.239171160942379e-05,
      "loss": 0.001304195448756218,
      "step": 132550
    },
    {
      "epoch": 37.6270224240704,
      "grad_norm": 2.377549409866333,
      "learning_rate": 6.238887311950042e-05,
      "loss": 0.002070714719593525,
      "step": 132560
    },
    {
      "epoch": 37.62986091399375,
      "grad_norm": 2.464228630065918,
      "learning_rate": 6.238603462957707e-05,
      "loss": 0.0017978737130761146,
      "step": 132570
    },
    {
      "epoch": 37.632699403917115,
      "grad_norm": 0.514973521232605,
      "learning_rate": 6.238319613965372e-05,
      "loss": 0.002048587612807751,
      "step": 132580
    },
    {
      "epoch": 37.63553789384048,
      "grad_norm": 0.5121153593063354,
      "learning_rate": 6.238035764973034e-05,
      "loss": 0.011153070628643036,
      "step": 132590
    },
    {
      "epoch": 37.63837638376384,
      "grad_norm": 1.0856119394302368,
      "learning_rate": 6.237751915980698e-05,
      "loss": 0.016050618886947633,
      "step": 132600
    },
    {
      "epoch": 37.6412148736872,
      "grad_norm": 1.9948545694351196,
      "learning_rate": 6.237468066988363e-05,
      "loss": 0.007011393457651139,
      "step": 132610
    },
    {
      "epoch": 37.64405336361056,
      "grad_norm": 0.4396890103816986,
      "learning_rate": 6.237184217996025e-05,
      "loss": 0.002546379156410694,
      "step": 132620
    },
    {
      "epoch": 37.64689185353392,
      "grad_norm": 7.393730163574219,
      "learning_rate": 6.23690036900369e-05,
      "loss": 0.004048900306224823,
      "step": 132630
    },
    {
      "epoch": 37.64973034345728,
      "grad_norm": 0.06148531287908554,
      "learning_rate": 6.236616520011355e-05,
      "loss": 0.0010700980201363564,
      "step": 132640
    },
    {
      "epoch": 37.65256883338064,
      "grad_norm": 2.0977513790130615,
      "learning_rate": 6.236332671019018e-05,
      "loss": 0.0025764688849449157,
      "step": 132650
    },
    {
      "epoch": 37.655407323304004,
      "grad_norm": 0.14940853416919708,
      "learning_rate": 6.236048822026682e-05,
      "loss": 0.000983455777168274,
      "step": 132660
    },
    {
      "epoch": 37.65824581322736,
      "grad_norm": 0.19542144238948822,
      "learning_rate": 6.235764973034346e-05,
      "loss": 0.008302377909421921,
      "step": 132670
    },
    {
      "epoch": 37.66108430315072,
      "grad_norm": 0.7092940807342529,
      "learning_rate": 6.23548112404201e-05,
      "loss": 0.0015957662835717201,
      "step": 132680
    },
    {
      "epoch": 37.663922793074086,
      "grad_norm": 0.45525434613227844,
      "learning_rate": 6.235197275049673e-05,
      "loss": 0.002026466839015484,
      "step": 132690
    },
    {
      "epoch": 37.66676128299745,
      "grad_norm": 0.09463665634393692,
      "learning_rate": 6.234913426057338e-05,
      "loss": 0.002368542551994324,
      "step": 132700
    },
    {
      "epoch": 37.669599772920805,
      "grad_norm": 1.0793935060501099,
      "learning_rate": 6.234629577065003e-05,
      "loss": 0.002449742890894413,
      "step": 132710
    },
    {
      "epoch": 37.67243826284417,
      "grad_norm": 0.6311933994293213,
      "learning_rate": 6.234345728072665e-05,
      "loss": 0.004752984642982483,
      "step": 132720
    },
    {
      "epoch": 37.67527675276753,
      "grad_norm": 0.11249015480279922,
      "learning_rate": 6.23406187908033e-05,
      "loss": 0.0039205759763717655,
      "step": 132730
    },
    {
      "epoch": 37.678115242690886,
      "grad_norm": 6.136289596557617,
      "learning_rate": 6.233778030087994e-05,
      "loss": 0.01242113783955574,
      "step": 132740
    },
    {
      "epoch": 37.68095373261425,
      "grad_norm": 1.0341088771820068,
      "learning_rate": 6.233494181095656e-05,
      "loss": 0.017194324731826784,
      "step": 132750
    },
    {
      "epoch": 37.68379222253761,
      "grad_norm": 4.73037052154541,
      "learning_rate": 6.23321033210332e-05,
      "loss": 0.003039441630244255,
      "step": 132760
    },
    {
      "epoch": 37.68663071246097,
      "grad_norm": 13.117598533630371,
      "learning_rate": 6.232926483110986e-05,
      "loss": 0.007743534445762634,
      "step": 132770
    },
    {
      "epoch": 37.68946920238433,
      "grad_norm": 0.159669890999794,
      "learning_rate": 6.232642634118649e-05,
      "loss": 0.006506147980690003,
      "step": 132780
    },
    {
      "epoch": 37.69230769230769,
      "grad_norm": 0.028011225163936615,
      "learning_rate": 6.232358785126313e-05,
      "loss": 0.0007077228277921676,
      "step": 132790
    },
    {
      "epoch": 37.695146182231056,
      "grad_norm": 0.35714855790138245,
      "learning_rate": 6.232074936133977e-05,
      "loss": 0.0049905598163604735,
      "step": 132800
    },
    {
      "epoch": 37.69798467215441,
      "grad_norm": 1.3885563611984253,
      "learning_rate": 6.231791087141641e-05,
      "loss": 0.01562163382768631,
      "step": 132810
    },
    {
      "epoch": 37.700823162077775,
      "grad_norm": 5.245763301849365,
      "learning_rate": 6.231507238149304e-05,
      "loss": 0.0016142314299941063,
      "step": 132820
    },
    {
      "epoch": 37.70366165200114,
      "grad_norm": 1.7521334886550903,
      "learning_rate": 6.231223389156968e-05,
      "loss": 0.0013959165662527084,
      "step": 132830
    },
    {
      "epoch": 37.706500141924494,
      "grad_norm": 1.8494197130203247,
      "learning_rate": 6.230939540164634e-05,
      "loss": 0.0030511913821101187,
      "step": 132840
    },
    {
      "epoch": 37.70933863184786,
      "grad_norm": 0.2823350131511688,
      "learning_rate": 6.230655691172296e-05,
      "loss": 0.0056950576603412625,
      "step": 132850
    },
    {
      "epoch": 37.71217712177122,
      "grad_norm": 1.437185525894165,
      "learning_rate": 6.23037184217996e-05,
      "loss": 0.0032662130892276765,
      "step": 132860
    },
    {
      "epoch": 37.715015611694575,
      "grad_norm": 0.02005903795361519,
      "learning_rate": 6.230087993187625e-05,
      "loss": 0.0014844300225377084,
      "step": 132870
    },
    {
      "epoch": 37.71785410161794,
      "grad_norm": 0.16124209761619568,
      "learning_rate": 6.229804144195288e-05,
      "loss": 0.025777983665466308,
      "step": 132880
    },
    {
      "epoch": 37.7206925915413,
      "grad_norm": 2.0263826847076416,
      "learning_rate": 6.229520295202952e-05,
      "loss": 0.005041095614433289,
      "step": 132890
    },
    {
      "epoch": 37.723531081464664,
      "grad_norm": 2.158900737762451,
      "learning_rate": 6.229236446210617e-05,
      "loss": 0.012619401514530181,
      "step": 132900
    },
    {
      "epoch": 37.72636957138802,
      "grad_norm": 11.915014266967773,
      "learning_rate": 6.22895259721828e-05,
      "loss": 0.004944156482815742,
      "step": 132910
    },
    {
      "epoch": 37.72920806131138,
      "grad_norm": 0.14055761694908142,
      "learning_rate": 6.228668748225944e-05,
      "loss": 0.0059983029961586,
      "step": 132920
    },
    {
      "epoch": 37.732046551234745,
      "grad_norm": 0.11732164770364761,
      "learning_rate": 6.228384899233608e-05,
      "loss": 0.0019366150721907617,
      "step": 132930
    },
    {
      "epoch": 37.7348850411581,
      "grad_norm": 0.1532517373561859,
      "learning_rate": 6.228101050241272e-05,
      "loss": 0.004197642207145691,
      "step": 132940
    },
    {
      "epoch": 37.737723531081464,
      "grad_norm": 16.41945457458496,
      "learning_rate": 6.227817201248935e-05,
      "loss": 0.017532306909561157,
      "step": 132950
    },
    {
      "epoch": 37.74056202100483,
      "grad_norm": 0.09246093034744263,
      "learning_rate": 6.227533352256599e-05,
      "loss": 0.004091132059693336,
      "step": 132960
    },
    {
      "epoch": 37.74340051092818,
      "grad_norm": 0.28916335105895996,
      "learning_rate": 6.227249503264265e-05,
      "loss": 0.005170441418886185,
      "step": 132970
    },
    {
      "epoch": 37.746239000851546,
      "grad_norm": 0.37603336572647095,
      "learning_rate": 6.226965654271928e-05,
      "loss": 0.002044243551790714,
      "step": 132980
    },
    {
      "epoch": 37.74907749077491,
      "grad_norm": 8.819252014160156,
      "learning_rate": 6.226681805279592e-05,
      "loss": 0.012336017191410064,
      "step": 132990
    },
    {
      "epoch": 37.75191598069827,
      "grad_norm": 2.41961669921875,
      "learning_rate": 6.226397956287256e-05,
      "loss": 0.003954678773880005,
      "step": 133000
    },
    {
      "epoch": 37.75191598069827,
      "eval_accuracy": 0.9724041457366313,
      "eval_loss": 0.09708116948604584,
      "eval_runtime": 32.9351,
      "eval_samples_per_second": 477.514,
      "eval_steps_per_second": 7.469,
      "step": 133000
    },
    {
      "epoch": 37.75475447062163,
      "grad_norm": 0.12998168170452118,
      "learning_rate": 6.226114107294919e-05,
      "loss": 0.0009527182206511498,
      "step": 133010
    },
    {
      "epoch": 37.75759296054499,
      "grad_norm": 0.222847580909729,
      "learning_rate": 6.225830258302583e-05,
      "loss": 0.0050263553857803345,
      "step": 133020
    },
    {
      "epoch": 37.76043145046835,
      "grad_norm": 0.3734874725341797,
      "learning_rate": 6.225546409310248e-05,
      "loss": 0.0008156495168805122,
      "step": 133030
    },
    {
      "epoch": 37.76326994039171,
      "grad_norm": 0.24316823482513428,
      "learning_rate": 6.225262560317911e-05,
      "loss": 0.001613054983317852,
      "step": 133040
    },
    {
      "epoch": 37.76610843031507,
      "grad_norm": 0.06607042998075485,
      "learning_rate": 6.224978711325575e-05,
      "loss": 0.004872652143239975,
      "step": 133050
    },
    {
      "epoch": 37.768946920238434,
      "grad_norm": 0.31327736377716064,
      "learning_rate": 6.224694862333239e-05,
      "loss": 0.005223363637924194,
      "step": 133060
    },
    {
      "epoch": 37.7717854101618,
      "grad_norm": 2.095780372619629,
      "learning_rate": 6.224411013340903e-05,
      "loss": 0.010399572551250458,
      "step": 133070
    },
    {
      "epoch": 37.77462390008515,
      "grad_norm": 2.0874862670898438,
      "learning_rate": 6.224127164348566e-05,
      "loss": 0.003095918335020542,
      "step": 133080
    },
    {
      "epoch": 37.777462390008516,
      "grad_norm": 0.6543291211128235,
      "learning_rate": 6.22384331535623e-05,
      "loss": 0.017857806384563447,
      "step": 133090
    },
    {
      "epoch": 37.78030087993188,
      "grad_norm": 0.02607010491192341,
      "learning_rate": 6.223559466363896e-05,
      "loss": 0.0015245603397488594,
      "step": 133100
    },
    {
      "epoch": 37.783139369855235,
      "grad_norm": 0.1765156388282776,
      "learning_rate": 6.223275617371559e-05,
      "loss": 0.007004989683628083,
      "step": 133110
    },
    {
      "epoch": 37.7859778597786,
      "grad_norm": 0.40197938680648804,
      "learning_rate": 6.222991768379223e-05,
      "loss": 0.005685955286026001,
      "step": 133120
    },
    {
      "epoch": 37.78881634970196,
      "grad_norm": 2.2790653705596924,
      "learning_rate": 6.222707919386887e-05,
      "loss": 0.0015490902587771416,
      "step": 133130
    },
    {
      "epoch": 37.791654839625316,
      "grad_norm": 1.7385667562484741,
      "learning_rate": 6.22242407039455e-05,
      "loss": 0.010572496056556701,
      "step": 133140
    },
    {
      "epoch": 37.79449332954868,
      "grad_norm": 0.29737934470176697,
      "learning_rate": 6.222140221402214e-05,
      "loss": 0.004896527528762818,
      "step": 133150
    },
    {
      "epoch": 37.79733181947204,
      "grad_norm": 1.0798273086547852,
      "learning_rate": 6.221856372409878e-05,
      "loss": 0.0012504935264587402,
      "step": 133160
    },
    {
      "epoch": 37.800170309395405,
      "grad_norm": 4.986037731170654,
      "learning_rate": 6.221572523417542e-05,
      "loss": 0.0018505590036511422,
      "step": 133170
    },
    {
      "epoch": 37.80300879931876,
      "grad_norm": 0.07906194776296616,
      "learning_rate": 6.221288674425206e-05,
      "loss": 0.006117449700832367,
      "step": 133180
    },
    {
      "epoch": 37.805847289242124,
      "grad_norm": 0.07539534568786621,
      "learning_rate": 6.22100482543287e-05,
      "loss": 0.0040138278156518934,
      "step": 133190
    },
    {
      "epoch": 37.80868577916549,
      "grad_norm": 1.0952357053756714,
      "learning_rate": 6.220720976440534e-05,
      "loss": 0.0031634148210287092,
      "step": 133200
    },
    {
      "epoch": 37.81152426908884,
      "grad_norm": 7.378997802734375,
      "learning_rate": 6.220437127448197e-05,
      "loss": 0.005262964591383934,
      "step": 133210
    },
    {
      "epoch": 37.814362759012205,
      "grad_norm": 0.356638640165329,
      "learning_rate": 6.220153278455861e-05,
      "loss": 0.0005699669942259789,
      "step": 133220
    },
    {
      "epoch": 37.81720124893557,
      "grad_norm": 0.12266595661640167,
      "learning_rate": 6.219869429463527e-05,
      "loss": 0.004715427756309509,
      "step": 133230
    },
    {
      "epoch": 37.820039738858924,
      "grad_norm": 6.195380687713623,
      "learning_rate": 6.21958558047119e-05,
      "loss": 0.00301548857241869,
      "step": 133240
    },
    {
      "epoch": 37.82287822878229,
      "grad_norm": 0.21377554535865784,
      "learning_rate": 6.219301731478854e-05,
      "loss": 0.0037096865475177766,
      "step": 133250
    },
    {
      "epoch": 37.82571671870565,
      "grad_norm": 0.19940659403800964,
      "learning_rate": 6.219017882486518e-05,
      "loss": 0.012438584119081497,
      "step": 133260
    },
    {
      "epoch": 37.82855520862901,
      "grad_norm": 2.8873722553253174,
      "learning_rate": 6.218734033494181e-05,
      "loss": 0.001673341728746891,
      "step": 133270
    },
    {
      "epoch": 37.83139369855237,
      "grad_norm": 1.2639799118041992,
      "learning_rate": 6.218450184501845e-05,
      "loss": 0.0005596622824668885,
      "step": 133280
    },
    {
      "epoch": 37.83423218847573,
      "grad_norm": 0.6432152390480042,
      "learning_rate": 6.218166335509509e-05,
      "loss": 0.001684900000691414,
      "step": 133290
    },
    {
      "epoch": 37.837070678399094,
      "grad_norm": 0.30038389563560486,
      "learning_rate": 6.217882486517173e-05,
      "loss": 0.00300500076264143,
      "step": 133300
    },
    {
      "epoch": 37.83990916832245,
      "grad_norm": 0.21321283280849457,
      "learning_rate": 6.217598637524837e-05,
      "loss": 0.004956291988492012,
      "step": 133310
    },
    {
      "epoch": 37.84274765824581,
      "grad_norm": 0.12209369987249374,
      "learning_rate": 6.217314788532501e-05,
      "loss": 0.003255471959710121,
      "step": 133320
    },
    {
      "epoch": 37.845586148169176,
      "grad_norm": 0.2678591012954712,
      "learning_rate": 6.217030939540166e-05,
      "loss": 0.0021209243685007097,
      "step": 133330
    },
    {
      "epoch": 37.84842463809253,
      "grad_norm": 2.2129557132720947,
      "learning_rate": 6.216747090547828e-05,
      "loss": 0.004342694580554962,
      "step": 133340
    },
    {
      "epoch": 37.851263128015894,
      "grad_norm": 0.9906684160232544,
      "learning_rate": 6.216491626454726e-05,
      "loss": 0.03175324201583862,
      "step": 133350
    },
    {
      "epoch": 37.85410161793926,
      "grad_norm": 6.482047080993652,
      "learning_rate": 6.21620777746239e-05,
      "loss": 0.004478826373815537,
      "step": 133360
    },
    {
      "epoch": 37.85694010786262,
      "grad_norm": 0.8999180793762207,
      "learning_rate": 6.215923928470055e-05,
      "loss": 0.0024129973724484445,
      "step": 133370
    },
    {
      "epoch": 37.859778597785976,
      "grad_norm": 1.0929769277572632,
      "learning_rate": 6.215640079477719e-05,
      "loss": 0.0009130131453275681,
      "step": 133380
    },
    {
      "epoch": 37.86261708770934,
      "grad_norm": 0.9031277894973755,
      "learning_rate": 6.215356230485382e-05,
      "loss": 0.012177640199661255,
      "step": 133390
    },
    {
      "epoch": 37.8654555776327,
      "grad_norm": 1.5737007856369019,
      "learning_rate": 6.215072381493046e-05,
      "loss": 0.004903804883360863,
      "step": 133400
    },
    {
      "epoch": 37.86829406755606,
      "grad_norm": 0.889591634273529,
      "learning_rate": 6.21478853250071e-05,
      "loss": 0.0056223064661026,
      "step": 133410
    },
    {
      "epoch": 37.87113255747942,
      "grad_norm": 0.953218400478363,
      "learning_rate": 6.214504683508374e-05,
      "loss": 0.007409502565860748,
      "step": 133420
    },
    {
      "epoch": 37.87397104740278,
      "grad_norm": 1.053886890411377,
      "learning_rate": 6.214220834516038e-05,
      "loss": 0.00195941012352705,
      "step": 133430
    },
    {
      "epoch": 37.876809537326146,
      "grad_norm": 0.1118682324886322,
      "learning_rate": 6.213936985523702e-05,
      "loss": 0.0025756197050213815,
      "step": 133440
    },
    {
      "epoch": 37.8796480272495,
      "grad_norm": 0.1571532040834427,
      "learning_rate": 6.213653136531365e-05,
      "loss": 0.0010448066517710687,
      "step": 133450
    },
    {
      "epoch": 37.882486517172865,
      "grad_norm": 0.37357431650161743,
      "learning_rate": 6.213369287539029e-05,
      "loss": 0.002458176575601101,
      "step": 133460
    },
    {
      "epoch": 37.88532500709623,
      "grad_norm": 0.4780252277851105,
      "learning_rate": 6.213085438546693e-05,
      "loss": 0.0034547030925750734,
      "step": 133470
    },
    {
      "epoch": 37.88816349701958,
      "grad_norm": 0.059194665402173996,
      "learning_rate": 6.212801589554357e-05,
      "loss": 0.0010642634704709053,
      "step": 133480
    },
    {
      "epoch": 37.891001986942946,
      "grad_norm": 0.06451675295829773,
      "learning_rate": 6.212517740562022e-05,
      "loss": 0.004234927892684937,
      "step": 133490
    },
    {
      "epoch": 37.89384047686631,
      "grad_norm": 0.4888061285018921,
      "learning_rate": 6.212233891569686e-05,
      "loss": 0.031558871269226074,
      "step": 133500
    },
    {
      "epoch": 37.89384047686631,
      "eval_accuracy": 0.9761556558784257,
      "eval_loss": 0.08558957278728485,
      "eval_runtime": 31.9137,
      "eval_samples_per_second": 492.798,
      "eval_steps_per_second": 7.708,
      "step": 133500
    },
    {
      "epoch": 37.896678966789665,
      "grad_norm": 0.2645043432712555,
      "learning_rate": 6.211950042577348e-05,
      "loss": 0.004266504943370819,
      "step": 133510
    },
    {
      "epoch": 37.89951745671303,
      "grad_norm": 1.0531411170959473,
      "learning_rate": 6.211666193585013e-05,
      "loss": 0.0090446799993515,
      "step": 133520
    },
    {
      "epoch": 37.90235594663639,
      "grad_norm": 1.3234773874282837,
      "learning_rate": 6.211382344592677e-05,
      "loss": 0.0029716160148382186,
      "step": 133530
    },
    {
      "epoch": 37.90519443655975,
      "grad_norm": 1.9012441635131836,
      "learning_rate": 6.211098495600341e-05,
      "loss": 0.0021519467234611513,
      "step": 133540
    },
    {
      "epoch": 37.90803292648311,
      "grad_norm": 2.706608295440674,
      "learning_rate": 6.210814646608005e-05,
      "loss": 0.009835876524448395,
      "step": 133550
    },
    {
      "epoch": 37.91087141640647,
      "grad_norm": 3.6343371868133545,
      "learning_rate": 6.210530797615669e-05,
      "loss": 0.00712469220161438,
      "step": 133560
    },
    {
      "epoch": 37.913709906329835,
      "grad_norm": 1.2140740156173706,
      "learning_rate": 6.210246948623333e-05,
      "loss": 0.007036004215478897,
      "step": 133570
    },
    {
      "epoch": 37.91654839625319,
      "grad_norm": 0.5931113958358765,
      "learning_rate": 6.209963099630996e-05,
      "loss": 0.015692949295043945,
      "step": 133580
    },
    {
      "epoch": 37.919386886176554,
      "grad_norm": 1.4090523719787598,
      "learning_rate": 6.20967925063866e-05,
      "loss": 0.00514179989695549,
      "step": 133590
    },
    {
      "epoch": 37.92222537609992,
      "grad_norm": 3.1565916538238525,
      "learning_rate": 6.209395401646324e-05,
      "loss": 0.00160933006554842,
      "step": 133600
    },
    {
      "epoch": 37.92506386602327,
      "grad_norm": 0.09420684725046158,
      "learning_rate": 6.209111552653989e-05,
      "loss": 0.004977817833423615,
      "step": 133610
    },
    {
      "epoch": 37.927902355946635,
      "grad_norm": 0.20956964790821075,
      "learning_rate": 6.208827703661653e-05,
      "loss": 0.015343551337718964,
      "step": 133620
    },
    {
      "epoch": 37.93074084587,
      "grad_norm": 4.91356897354126,
      "learning_rate": 6.208543854669317e-05,
      "loss": 0.0038940757513046265,
      "step": 133630
    },
    {
      "epoch": 37.93357933579336,
      "grad_norm": 1.4136955738067627,
      "learning_rate": 6.20826000567698e-05,
      "loss": 0.0038405947387218477,
      "step": 133640
    },
    {
      "epoch": 37.93641782571672,
      "grad_norm": 0.9950119853019714,
      "learning_rate": 6.207976156684644e-05,
      "loss": 0.007612425833940506,
      "step": 133650
    },
    {
      "epoch": 37.93925631564008,
      "grad_norm": 0.37792518734931946,
      "learning_rate": 6.207692307692308e-05,
      "loss": 0.007347790896892548,
      "step": 133660
    },
    {
      "epoch": 37.94209480556344,
      "grad_norm": 0.08669348061084747,
      "learning_rate": 6.207408458699972e-05,
      "loss": 0.007537807524204254,
      "step": 133670
    },
    {
      "epoch": 37.9449332954868,
      "grad_norm": 1.3634811639785767,
      "learning_rate": 6.207124609707636e-05,
      "loss": 0.002638531103730202,
      "step": 133680
    },
    {
      "epoch": 37.94777178541016,
      "grad_norm": 1.146072268486023,
      "learning_rate": 6.2068407607153e-05,
      "loss": 0.0021408434957265856,
      "step": 133690
    },
    {
      "epoch": 37.950610275333524,
      "grad_norm": 0.2799375057220459,
      "learning_rate": 6.206556911722964e-05,
      "loss": 0.0016608472913503646,
      "step": 133700
    },
    {
      "epoch": 37.95344876525688,
      "grad_norm": 4.991639614105225,
      "learning_rate": 6.206273062730627e-05,
      "loss": 0.0023230288177728655,
      "step": 133710
    },
    {
      "epoch": 37.95628725518024,
      "grad_norm": 0.23622044920921326,
      "learning_rate": 6.205989213738291e-05,
      "loss": 0.010309036821126938,
      "step": 133720
    },
    {
      "epoch": 37.959125745103606,
      "grad_norm": 1.4886629581451416,
      "learning_rate": 6.205705364745955e-05,
      "loss": 0.002765454538166523,
      "step": 133730
    },
    {
      "epoch": 37.96196423502697,
      "grad_norm": 0.10436052083969116,
      "learning_rate": 6.205421515753618e-05,
      "loss": 0.001004292443394661,
      "step": 133740
    },
    {
      "epoch": 37.964802724950324,
      "grad_norm": 0.7403091192245483,
      "learning_rate": 6.205137666761284e-05,
      "loss": 0.0029820667579770086,
      "step": 133750
    },
    {
      "epoch": 37.96764121487369,
      "grad_norm": 4.340145587921143,
      "learning_rate": 6.204853817768948e-05,
      "loss": 0.007359112799167633,
      "step": 133760
    },
    {
      "epoch": 37.97047970479705,
      "grad_norm": 7.15266227722168,
      "learning_rate": 6.20456996877661e-05,
      "loss": 0.00341038778424263,
      "step": 133770
    },
    {
      "epoch": 37.973318194720406,
      "grad_norm": 0.22077172994613647,
      "learning_rate": 6.204286119784275e-05,
      "loss": 0.002966208755970001,
      "step": 133780
    },
    {
      "epoch": 37.97615668464377,
      "grad_norm": 0.2161400318145752,
      "learning_rate": 6.204002270791939e-05,
      "loss": 0.0014428023248910903,
      "step": 133790
    },
    {
      "epoch": 37.97899517456713,
      "grad_norm": 0.13567674160003662,
      "learning_rate": 6.203718421799603e-05,
      "loss": 0.006211702525615692,
      "step": 133800
    },
    {
      "epoch": 37.981833664490495,
      "grad_norm": 0.5223957896232605,
      "learning_rate": 6.203434572807267e-05,
      "loss": 0.0025690630078315734,
      "step": 133810
    },
    {
      "epoch": 37.98467215441385,
      "grad_norm": 3.2517518997192383,
      "learning_rate": 6.203150723814931e-05,
      "loss": 0.0032192133367061615,
      "step": 133820
    },
    {
      "epoch": 37.98751064433721,
      "grad_norm": 0.045168712735176086,
      "learning_rate": 6.202866874822595e-05,
      "loss": 0.0024882715195417406,
      "step": 133830
    },
    {
      "epoch": 37.990349134260576,
      "grad_norm": 0.5610141754150391,
      "learning_rate": 6.202583025830258e-05,
      "loss": 0.00626506432890892,
      "step": 133840
    },
    {
      "epoch": 37.99318762418393,
      "grad_norm": 1.4371424913406372,
      "learning_rate": 6.202299176837922e-05,
      "loss": 0.0032473891973495484,
      "step": 133850
    },
    {
      "epoch": 37.996026114107295,
      "grad_norm": 5.081765651702881,
      "learning_rate": 6.202015327845587e-05,
      "loss": 0.005523406714200973,
      "step": 133860
    },
    {
      "epoch": 37.99886460403066,
      "grad_norm": 9.350892066955566,
      "learning_rate": 6.201731478853249e-05,
      "loss": 0.0034673213958740234,
      "step": 133870
    },
    {
      "epoch": 38.00170309395401,
      "grad_norm": 0.051754266023635864,
      "learning_rate": 6.201447629860915e-05,
      "loss": 0.007218610495328903,
      "step": 133880
    },
    {
      "epoch": 38.004541583877376,
      "grad_norm": 0.7074530720710754,
      "learning_rate": 6.201163780868579e-05,
      "loss": 0.002277693711221218,
      "step": 133890
    },
    {
      "epoch": 38.00738007380074,
      "grad_norm": 0.12943114340305328,
      "learning_rate": 6.200879931876242e-05,
      "loss": 0.0024940351024270056,
      "step": 133900
    },
    {
      "epoch": 38.0102185637241,
      "grad_norm": 0.3673385679721832,
      "learning_rate": 6.200596082883906e-05,
      "loss": 0.009372051060199737,
      "step": 133910
    },
    {
      "epoch": 38.01305705364746,
      "grad_norm": 6.83774471282959,
      "learning_rate": 6.20031223389157e-05,
      "loss": 0.0028739402070641516,
      "step": 133920
    },
    {
      "epoch": 38.01589554357082,
      "grad_norm": 0.26884937286376953,
      "learning_rate": 6.200028384899234e-05,
      "loss": 0.0052588354796171185,
      "step": 133930
    },
    {
      "epoch": 38.018734033494184,
      "grad_norm": 6.7967047691345215,
      "learning_rate": 6.199744535906897e-05,
      "loss": 0.007299830764532089,
      "step": 133940
    },
    {
      "epoch": 38.02157252341754,
      "grad_norm": 0.5923227071762085,
      "learning_rate": 6.199460686914562e-05,
      "loss": 0.0042127460241317746,
      "step": 133950
    },
    {
      "epoch": 38.0244110133409,
      "grad_norm": 0.1493835151195526,
      "learning_rate": 6.199176837922227e-05,
      "loss": 0.001406092755496502,
      "step": 133960
    },
    {
      "epoch": 38.027249503264265,
      "grad_norm": 0.25654828548431396,
      "learning_rate": 6.198892988929889e-05,
      "loss": 0.0037756316363811494,
      "step": 133970
    },
    {
      "epoch": 38.03008799318762,
      "grad_norm": 0.04529784619808197,
      "learning_rate": 6.198609139937553e-05,
      "loss": 0.002508901245892048,
      "step": 133980
    },
    {
      "epoch": 38.032926483110984,
      "grad_norm": 0.19847343862056732,
      "learning_rate": 6.198325290945218e-05,
      "loss": 0.00103923287242651,
      "step": 133990
    },
    {
      "epoch": 38.03576497303435,
      "grad_norm": 0.6736425757408142,
      "learning_rate": 6.19804144195288e-05,
      "loss": 0.001282622665166855,
      "step": 134000
    },
    {
      "epoch": 38.03576497303435,
      "eval_accuracy": 0.9725949004896038,
      "eval_loss": 0.10176767408847809,
      "eval_runtime": 32.262,
      "eval_samples_per_second": 487.478,
      "eval_steps_per_second": 7.625,
      "step": 134000
    },
    {
      "epoch": 38.03860346295771,
      "grad_norm": 0.3281838595867157,
      "learning_rate": 6.197757592960546e-05,
      "loss": 0.0008872263133525848,
      "step": 134010
    },
    {
      "epoch": 38.041441952881065,
      "grad_norm": 0.3041618764400482,
      "learning_rate": 6.19747374396821e-05,
      "loss": 0.0015516072511672975,
      "step": 134020
    },
    {
      "epoch": 38.04428044280443,
      "grad_norm": 0.5932643413543701,
      "learning_rate": 6.197189894975873e-05,
      "loss": 0.0013360675424337388,
      "step": 134030
    },
    {
      "epoch": 38.04711893272779,
      "grad_norm": 0.8423849940299988,
      "learning_rate": 6.196906045983537e-05,
      "loss": 0.00694122314453125,
      "step": 134040
    },
    {
      "epoch": 38.04995742265115,
      "grad_norm": 0.18612664937973022,
      "learning_rate": 6.196622196991201e-05,
      "loss": 0.0027197122573852537,
      "step": 134050
    },
    {
      "epoch": 38.05279591257451,
      "grad_norm": 0.9791676998138428,
      "learning_rate": 6.196338347998865e-05,
      "loss": 0.004170769453048706,
      "step": 134060
    },
    {
      "epoch": 38.05563440249787,
      "grad_norm": 2.237821102142334,
      "learning_rate": 6.196054499006528e-05,
      "loss": 0.002253599651157856,
      "step": 134070
    },
    {
      "epoch": 38.05847289242123,
      "grad_norm": 0.1873857080936432,
      "learning_rate": 6.195770650014193e-05,
      "loss": 0.0007913874462246895,
      "step": 134080
    },
    {
      "epoch": 38.06131138234459,
      "grad_norm": 0.021373817697167397,
      "learning_rate": 6.195486801021858e-05,
      "loss": 0.001890280842781067,
      "step": 134090
    },
    {
      "epoch": 38.064149872267954,
      "grad_norm": 0.10305590182542801,
      "learning_rate": 6.19520295202952e-05,
      "loss": 0.003622611612081528,
      "step": 134100
    },
    {
      "epoch": 38.06698836219132,
      "grad_norm": 7.983122825622559,
      "learning_rate": 6.194919103037185e-05,
      "loss": 0.01153704673051834,
      "step": 134110
    },
    {
      "epoch": 38.06982685211467,
      "grad_norm": 0.015571177937090397,
      "learning_rate": 6.194635254044849e-05,
      "loss": 0.00203020628541708,
      "step": 134120
    },
    {
      "epoch": 38.072665342038036,
      "grad_norm": 0.13727964460849762,
      "learning_rate": 6.194351405052511e-05,
      "loss": 0.00445866659283638,
      "step": 134130
    },
    {
      "epoch": 38.0755038319614,
      "grad_norm": 1.0107402801513672,
      "learning_rate": 6.194067556060176e-05,
      "loss": 0.005461192131042481,
      "step": 134140
    },
    {
      "epoch": 38.078342321884755,
      "grad_norm": 0.05921114236116409,
      "learning_rate": 6.193783707067841e-05,
      "loss": 0.0036296255886554716,
      "step": 134150
    },
    {
      "epoch": 38.08118081180812,
      "grad_norm": 0.2927892804145813,
      "learning_rate": 6.193499858075504e-05,
      "loss": 0.0005807241424918175,
      "step": 134160
    },
    {
      "epoch": 38.08401930173148,
      "grad_norm": 0.0699792206287384,
      "learning_rate": 6.193216009083168e-05,
      "loss": 0.0020666634663939476,
      "step": 134170
    },
    {
      "epoch": 38.086857791654836,
      "grad_norm": 0.07194444537162781,
      "learning_rate": 6.192932160090832e-05,
      "loss": 0.0012159183621406554,
      "step": 134180
    },
    {
      "epoch": 38.0896962815782,
      "grad_norm": 2.000232219696045,
      "learning_rate": 6.192648311098496e-05,
      "loss": 0.0020729703828692436,
      "step": 134190
    },
    {
      "epoch": 38.09253477150156,
      "grad_norm": 0.27747848629951477,
      "learning_rate": 6.192364462106159e-05,
      "loss": 0.0008992934599518776,
      "step": 134200
    },
    {
      "epoch": 38.095373261424925,
      "grad_norm": 0.6077998876571655,
      "learning_rate": 6.192080613113825e-05,
      "loss": 0.00133347287774086,
      "step": 134210
    },
    {
      "epoch": 38.09821175134828,
      "grad_norm": 0.3227548897266388,
      "learning_rate": 6.191796764121487e-05,
      "loss": 0.003961642831563949,
      "step": 134220
    },
    {
      "epoch": 38.10105024127164,
      "grad_norm": 0.05675831064581871,
      "learning_rate": 6.191512915129151e-05,
      "loss": 0.0005013752728700638,
      "step": 134230
    },
    {
      "epoch": 38.103888731195006,
      "grad_norm": 0.4183262586593628,
      "learning_rate": 6.191229066136816e-05,
      "loss": 0.0006160566583275795,
      "step": 134240
    },
    {
      "epoch": 38.10672722111836,
      "grad_norm": 0.6983895301818848,
      "learning_rate": 6.19094521714448e-05,
      "loss": 0.000806819275021553,
      "step": 134250
    },
    {
      "epoch": 38.109565711041725,
      "grad_norm": 0.038608014583587646,
      "learning_rate": 6.190661368152143e-05,
      "loss": 0.000459945946931839,
      "step": 134260
    },
    {
      "epoch": 38.11240420096509,
      "grad_norm": 0.13304409384727478,
      "learning_rate": 6.190377519159807e-05,
      "loss": 0.004243259131908417,
      "step": 134270
    },
    {
      "epoch": 38.11524269088845,
      "grad_norm": 0.06416783481836319,
      "learning_rate": 6.190093670167472e-05,
      "loss": 0.005541495606303215,
      "step": 134280
    },
    {
      "epoch": 38.11808118081181,
      "grad_norm": 0.019597381353378296,
      "learning_rate": 6.189809821175135e-05,
      "loss": 0.01608901172876358,
      "step": 134290
    },
    {
      "epoch": 38.12091967073517,
      "grad_norm": 0.05220317468047142,
      "learning_rate": 6.189525972182799e-05,
      "loss": 0.00225713811814785,
      "step": 134300
    },
    {
      "epoch": 38.12375816065853,
      "grad_norm": 11.762308120727539,
      "learning_rate": 6.189242123190463e-05,
      "loss": 0.010987860709428787,
      "step": 134310
    },
    {
      "epoch": 38.12659665058189,
      "grad_norm": 0.7038143873214722,
      "learning_rate": 6.188958274198127e-05,
      "loss": 0.007453412562608719,
      "step": 134320
    },
    {
      "epoch": 38.12943514050525,
      "grad_norm": 0.1096457988023758,
      "learning_rate": 6.18867442520579e-05,
      "loss": 0.0036039143800735475,
      "step": 134330
    },
    {
      "epoch": 38.132273630428614,
      "grad_norm": 0.11033488810062408,
      "learning_rate": 6.188390576213454e-05,
      "loss": 0.005243409797549248,
      "step": 134340
    },
    {
      "epoch": 38.13511212035197,
      "grad_norm": 0.37281447649002075,
      "learning_rate": 6.188106727221118e-05,
      "loss": 0.0007205948233604432,
      "step": 134350
    },
    {
      "epoch": 38.13795061027533,
      "grad_norm": 0.043452195823192596,
      "learning_rate": 6.187822878228783e-05,
      "loss": 0.000520840473473072,
      "step": 134360
    },
    {
      "epoch": 38.140789100198695,
      "grad_norm": 0.19350898265838623,
      "learning_rate": 6.187539029236447e-05,
      "loss": 0.005626681447029114,
      "step": 134370
    },
    {
      "epoch": 38.14362759012206,
      "grad_norm": 1.6138943433761597,
      "learning_rate": 6.187255180244111e-05,
      "loss": 0.0021019140258431435,
      "step": 134380
    },
    {
      "epoch": 38.146466080045414,
      "grad_norm": 0.5595139265060425,
      "learning_rate": 6.186971331251774e-05,
      "loss": 0.015924552083015443,
      "step": 134390
    },
    {
      "epoch": 38.14930456996878,
      "grad_norm": 0.05853088200092316,
      "learning_rate": 6.186687482259438e-05,
      "loss": 0.004021376371383667,
      "step": 134400
    },
    {
      "epoch": 38.15214305989214,
      "grad_norm": 2.0732181072235107,
      "learning_rate": 6.186403633267103e-05,
      "loss": 0.0045388080179691315,
      "step": 134410
    },
    {
      "epoch": 38.154981549815496,
      "grad_norm": 0.06178136169910431,
      "learning_rate": 6.186119784274766e-05,
      "loss": 0.00310603529214859,
      "step": 134420
    },
    {
      "epoch": 38.15782003973886,
      "grad_norm": 0.07122131437063217,
      "learning_rate": 6.18583593528243e-05,
      "loss": 0.0036707647144794465,
      "step": 134430
    },
    {
      "epoch": 38.16065852966222,
      "grad_norm": 0.13076157867908478,
      "learning_rate": 6.185552086290094e-05,
      "loss": 0.004789078980684281,
      "step": 134440
    },
    {
      "epoch": 38.16349701958558,
      "grad_norm": 0.39343711733818054,
      "learning_rate": 6.185268237297757e-05,
      "loss": 0.0068543747067451475,
      "step": 134450
    },
    {
      "epoch": 38.16633550950894,
      "grad_norm": 11.603266716003418,
      "learning_rate": 6.184984388305421e-05,
      "loss": 0.005468378216028214,
      "step": 134460
    },
    {
      "epoch": 38.1691739994323,
      "grad_norm": 0.32400277256965637,
      "learning_rate": 6.184700539313085e-05,
      "loss": 0.015094992518424988,
      "step": 134470
    },
    {
      "epoch": 38.172012489355666,
      "grad_norm": 0.01852233149111271,
      "learning_rate": 6.18441669032075e-05,
      "loss": 0.000986446626484394,
      "step": 134480
    },
    {
      "epoch": 38.17485097927902,
      "grad_norm": 5.0386505126953125,
      "learning_rate": 6.184132841328414e-05,
      "loss": 0.001744224689900875,
      "step": 134490
    },
    {
      "epoch": 38.177689469202384,
      "grad_norm": 0.08164342492818832,
      "learning_rate": 6.183848992336078e-05,
      "loss": 0.0028729850426316263,
      "step": 134500
    },
    {
      "epoch": 38.177689469202384,
      "eval_accuracy": 0.9746932027723024,
      "eval_loss": 0.09170718491077423,
      "eval_runtime": 31.8122,
      "eval_samples_per_second": 494.371,
      "eval_steps_per_second": 7.733,
      "step": 134500
    },
    {
      "epoch": 38.18052795912575,
      "grad_norm": 0.08953797817230225,
      "learning_rate": 6.183565143343742e-05,
      "loss": 0.0027678653597831724,
      "step": 134510
    },
    {
      "epoch": 38.1833664490491,
      "grad_norm": 0.6210519075393677,
      "learning_rate": 6.183281294351405e-05,
      "loss": 0.0008255355060100555,
      "step": 134520
    },
    {
      "epoch": 38.186204938972466,
      "grad_norm": 0.7011472582817078,
      "learning_rate": 6.182997445359069e-05,
      "loss": 0.0018625002354383468,
      "step": 134530
    },
    {
      "epoch": 38.18904342889583,
      "grad_norm": 0.28650230169296265,
      "learning_rate": 6.182713596366733e-05,
      "loss": 0.0012908091768622399,
      "step": 134540
    },
    {
      "epoch": 38.191881918819185,
      "grad_norm": 1.9494290351867676,
      "learning_rate": 6.182429747374397e-05,
      "loss": 0.0010882403701543808,
      "step": 134550
    },
    {
      "epoch": 38.19472040874255,
      "grad_norm": 4.1134724617004395,
      "learning_rate": 6.182145898382061e-05,
      "loss": 0.003326386585831642,
      "step": 134560
    },
    {
      "epoch": 38.19755889866591,
      "grad_norm": 0.11450708657503128,
      "learning_rate": 6.181862049389725e-05,
      "loss": 0.0017510443925857545,
      "step": 134570
    },
    {
      "epoch": 38.20039738858927,
      "grad_norm": 0.6780282855033875,
      "learning_rate": 6.181578200397388e-05,
      "loss": 0.0007807288318872451,
      "step": 134580
    },
    {
      "epoch": 38.20323587851263,
      "grad_norm": 0.33603140711784363,
      "learning_rate": 6.181294351405052e-05,
      "loss": 0.002268545143306255,
      "step": 134590
    },
    {
      "epoch": 38.20607436843599,
      "grad_norm": 1.5957492589950562,
      "learning_rate": 6.181010502412716e-05,
      "loss": 0.003979887068271637,
      "step": 134600
    },
    {
      "epoch": 38.208912858359355,
      "grad_norm": 1.7548292875289917,
      "learning_rate": 6.18072665342038e-05,
      "loss": 0.004050934687256813,
      "step": 134610
    },
    {
      "epoch": 38.21175134828271,
      "grad_norm": 0.09020157903432846,
      "learning_rate": 6.180442804428045e-05,
      "loss": 0.0010388627648353576,
      "step": 134620
    },
    {
      "epoch": 38.214589838206074,
      "grad_norm": 0.036795664578676224,
      "learning_rate": 6.180158955435709e-05,
      "loss": 0.00508514791727066,
      "step": 134630
    },
    {
      "epoch": 38.21742832812944,
      "grad_norm": 0.4582359790802002,
      "learning_rate": 6.179875106443373e-05,
      "loss": 0.0007269438356161118,
      "step": 134640
    },
    {
      "epoch": 38.2202668180528,
      "grad_norm": 1.890023946762085,
      "learning_rate": 6.179591257451036e-05,
      "loss": 0.005734815821051597,
      "step": 134650
    },
    {
      "epoch": 38.223105307976155,
      "grad_norm": 0.40506550669670105,
      "learning_rate": 6.1793074084587e-05,
      "loss": 0.0010772382840514183,
      "step": 134660
    },
    {
      "epoch": 38.22594379789952,
      "grad_norm": 0.23205557465553284,
      "learning_rate": 6.179023559466364e-05,
      "loss": 0.003627569600939751,
      "step": 134670
    },
    {
      "epoch": 38.22878228782288,
      "grad_norm": 7.9705586433410645,
      "learning_rate": 6.178739710474028e-05,
      "loss": 0.012706337869167328,
      "step": 134680
    },
    {
      "epoch": 38.23162077774624,
      "grad_norm": 0.30798599123954773,
      "learning_rate": 6.178455861481692e-05,
      "loss": 0.0007306167855858803,
      "step": 134690
    },
    {
      "epoch": 38.2344592676696,
      "grad_norm": 14.042913436889648,
      "learning_rate": 6.178172012489356e-05,
      "loss": 0.006286744028329849,
      "step": 134700
    },
    {
      "epoch": 38.23729775759296,
      "grad_norm": 0.6448888778686523,
      "learning_rate": 6.177888163497019e-05,
      "loss": 0.012516550719738007,
      "step": 134710
    },
    {
      "epoch": 38.24013624751632,
      "grad_norm": 0.34193095564842224,
      "learning_rate": 6.177604314504683e-05,
      "loss": 0.0078007087111473085,
      "step": 134720
    },
    {
      "epoch": 38.24297473743968,
      "grad_norm": 0.7417977452278137,
      "learning_rate": 6.177320465512348e-05,
      "loss": 0.006073121726512909,
      "step": 134730
    },
    {
      "epoch": 38.245813227363044,
      "grad_norm": 1.441645860671997,
      "learning_rate": 6.177036616520012e-05,
      "loss": 0.00965147465467453,
      "step": 134740
    },
    {
      "epoch": 38.24865171728641,
      "grad_norm": 3.3172965049743652,
      "learning_rate": 6.176752767527676e-05,
      "loss": 0.0067450180649757385,
      "step": 134750
    },
    {
      "epoch": 38.25149020720976,
      "grad_norm": 0.4145173132419586,
      "learning_rate": 6.17646891853534e-05,
      "loss": 0.005305605381727219,
      "step": 134760
    },
    {
      "epoch": 38.254328697133126,
      "grad_norm": 0.9570725560188293,
      "learning_rate": 6.176185069543004e-05,
      "loss": 0.003194957971572876,
      "step": 134770
    },
    {
      "epoch": 38.25716718705649,
      "grad_norm": 0.12894731760025024,
      "learning_rate": 6.175901220550667e-05,
      "loss": 0.018111911416053773,
      "step": 134780
    },
    {
      "epoch": 38.260005676979844,
      "grad_norm": 0.1613883674144745,
      "learning_rate": 6.175617371558331e-05,
      "loss": 0.005432094633579254,
      "step": 134790
    },
    {
      "epoch": 38.26284416690321,
      "grad_norm": 0.8154504299163818,
      "learning_rate": 6.175333522565995e-05,
      "loss": 0.0064793616533279415,
      "step": 134800
    },
    {
      "epoch": 38.26568265682657,
      "grad_norm": 10.099153518676758,
      "learning_rate": 6.175049673573659e-05,
      "loss": 0.005156424269080162,
      "step": 134810
    },
    {
      "epoch": 38.268521146749926,
      "grad_norm": 4.767364025115967,
      "learning_rate": 6.174765824581323e-05,
      "loss": 0.020709221065044404,
      "step": 134820
    },
    {
      "epoch": 38.27135963667329,
      "grad_norm": 1.3279755115509033,
      "learning_rate": 6.174481975588988e-05,
      "loss": 0.0051625072956085205,
      "step": 134830
    },
    {
      "epoch": 38.27419812659665,
      "grad_norm": 5.663919925689697,
      "learning_rate": 6.17419812659665e-05,
      "loss": 0.004571446776390075,
      "step": 134840
    },
    {
      "epoch": 38.277036616520014,
      "grad_norm": 4.391523361206055,
      "learning_rate": 6.173914277604314e-05,
      "loss": 0.0019650917500257494,
      "step": 134850
    },
    {
      "epoch": 38.27987510644337,
      "grad_norm": 0.0959833413362503,
      "learning_rate": 6.173630428611979e-05,
      "loss": 0.0049162209033966064,
      "step": 134860
    },
    {
      "epoch": 38.28271359636673,
      "grad_norm": 0.12046205997467041,
      "learning_rate": 6.173346579619643e-05,
      "loss": 0.012587885558605193,
      "step": 134870
    },
    {
      "epoch": 38.285552086290096,
      "grad_norm": 10.680061340332031,
      "learning_rate": 6.173062730627307e-05,
      "loss": 0.0034802518784999847,
      "step": 134880
    },
    {
      "epoch": 38.28839057621345,
      "grad_norm": 0.814301609992981,
      "learning_rate": 6.172778881634971e-05,
      "loss": 0.0006178801879286766,
      "step": 134890
    },
    {
      "epoch": 38.291229066136815,
      "grad_norm": 0.19150963425636292,
      "learning_rate": 6.172495032642635e-05,
      "loss": 0.0017397027462720872,
      "step": 134900
    },
    {
      "epoch": 38.29406755606018,
      "grad_norm": 0.04701870679855347,
      "learning_rate": 6.172211183650298e-05,
      "loss": 0.0014583354815840722,
      "step": 134910
    },
    {
      "epoch": 38.29690604598353,
      "grad_norm": 14.860255241394043,
      "learning_rate": 6.171927334657962e-05,
      "loss": 0.008625441789627075,
      "step": 134920
    },
    {
      "epoch": 38.299744535906896,
      "grad_norm": 1.892711877822876,
      "learning_rate": 6.171643485665626e-05,
      "loss": 0.006551828235387802,
      "step": 134930
    },
    {
      "epoch": 38.30258302583026,
      "grad_norm": 8.801395416259766,
      "learning_rate": 6.17135963667329e-05,
      "loss": 0.0031756356358528135,
      "step": 134940
    },
    {
      "epoch": 38.30542151575362,
      "grad_norm": 0.056268416345119476,
      "learning_rate": 6.171075787680954e-05,
      "loss": 0.005384530872106552,
      "step": 134950
    },
    {
      "epoch": 38.30826000567698,
      "grad_norm": 0.11726606637239456,
      "learning_rate": 6.170791938688619e-05,
      "loss": 0.001880672201514244,
      "step": 134960
    },
    {
      "epoch": 38.31109849560034,
      "grad_norm": 14.109869956970215,
      "learning_rate": 6.170508089696281e-05,
      "loss": 0.04095170497894287,
      "step": 134970
    },
    {
      "epoch": 38.3139369855237,
      "grad_norm": 7.809932231903076,
      "learning_rate": 6.170224240703946e-05,
      "loss": 0.004560225084424019,
      "step": 134980
    },
    {
      "epoch": 38.31677547544706,
      "grad_norm": 0.07856812328100204,
      "learning_rate": 6.16994039171161e-05,
      "loss": 0.0033401377499103547,
      "step": 134990
    },
    {
      "epoch": 38.31961396537042,
      "grad_norm": 0.925116240978241,
      "learning_rate": 6.169656542719274e-05,
      "loss": 0.011678503453731537,
      "step": 135000
    },
    {
      "epoch": 38.31961396537042,
      "eval_accuracy": 0.9743752781840147,
      "eval_loss": 0.0941469818353653,
      "eval_runtime": 31.7565,
      "eval_samples_per_second": 495.238,
      "eval_steps_per_second": 7.746,
      "step": 135000
    },
    {
      "epoch": 38.322452455293785,
      "grad_norm": 0.105397529900074,
      "learning_rate": 6.169372693726938e-05,
      "loss": 0.009975657612085343,
      "step": 135010
    },
    {
      "epoch": 38.32529094521715,
      "grad_norm": 0.7006086707115173,
      "learning_rate": 6.169088844734602e-05,
      "loss": 0.0010053662583231926,
      "step": 135020
    },
    {
      "epoch": 38.328129435140504,
      "grad_norm": 0.46642282605171204,
      "learning_rate": 6.168804995742266e-05,
      "loss": 0.0025414397940039636,
      "step": 135030
    },
    {
      "epoch": 38.33096792506387,
      "grad_norm": 0.25644588470458984,
      "learning_rate": 6.168521146749929e-05,
      "loss": 0.003680461645126343,
      "step": 135040
    },
    {
      "epoch": 38.33380641498723,
      "grad_norm": 0.4130198061466217,
      "learning_rate": 6.168237297757593e-05,
      "loss": 0.010602089762687682,
      "step": 135050
    },
    {
      "epoch": 38.336644904910585,
      "grad_norm": 1.493390440940857,
      "learning_rate": 6.167953448765257e-05,
      "loss": 0.0018177447840571404,
      "step": 135060
    },
    {
      "epoch": 38.33948339483395,
      "grad_norm": 0.26527324318885803,
      "learning_rate": 6.16766959977292e-05,
      "loss": 0.00382680743932724,
      "step": 135070
    },
    {
      "epoch": 38.34232188475731,
      "grad_norm": 3.4446351528167725,
      "learning_rate": 6.167385750780586e-05,
      "loss": 0.0018974712118506431,
      "step": 135080
    },
    {
      "epoch": 38.34516037468067,
      "grad_norm": 0.5435878038406372,
      "learning_rate": 6.16710190178825e-05,
      "loss": 0.005870004743337631,
      "step": 135090
    },
    {
      "epoch": 38.34799886460403,
      "grad_norm": 0.056403566151857376,
      "learning_rate": 6.166818052795912e-05,
      "loss": 0.0030626475811004637,
      "step": 135100
    },
    {
      "epoch": 38.35083735452739,
      "grad_norm": 1.920772671699524,
      "learning_rate": 6.166534203803577e-05,
      "loss": 0.013771346211433411,
      "step": 135110
    },
    {
      "epoch": 38.353675844450756,
      "grad_norm": 1.9258034229278564,
      "learning_rate": 6.166250354811241e-05,
      "loss": 0.0019804190844297407,
      "step": 135120
    },
    {
      "epoch": 38.35651433437411,
      "grad_norm": 0.05152112618088722,
      "learning_rate": 6.165966505818905e-05,
      "loss": 0.0018835114315152168,
      "step": 135130
    },
    {
      "epoch": 38.359352824297474,
      "grad_norm": 0.11888402700424194,
      "learning_rate": 6.165682656826569e-05,
      "loss": 0.002815871685743332,
      "step": 135140
    },
    {
      "epoch": 38.36219131422084,
      "grad_norm": 0.9970793128013611,
      "learning_rate": 6.165398807834233e-05,
      "loss": 0.0035604190081357954,
      "step": 135150
    },
    {
      "epoch": 38.36502980414419,
      "grad_norm": 0.24656173586845398,
      "learning_rate": 6.165114958841896e-05,
      "loss": 0.0029092352837324143,
      "step": 135160
    },
    {
      "epoch": 38.367868294067556,
      "grad_norm": 1.7609609365463257,
      "learning_rate": 6.16483110984956e-05,
      "loss": 0.007068584114313126,
      "step": 135170
    },
    {
      "epoch": 38.37070678399092,
      "grad_norm": 0.03300812840461731,
      "learning_rate": 6.164547260857224e-05,
      "loss": 0.004900767281651497,
      "step": 135180
    },
    {
      "epoch": 38.373545273914274,
      "grad_norm": 0.08716969937086105,
      "learning_rate": 6.164263411864888e-05,
      "loss": 0.008443042635917664,
      "step": 135190
    },
    {
      "epoch": 38.37638376383764,
      "grad_norm": 0.12718762457370758,
      "learning_rate": 6.163979562872551e-05,
      "loss": 0.0075580857694149016,
      "step": 135200
    },
    {
      "epoch": 38.379222253761,
      "grad_norm": 1.444724678993225,
      "learning_rate": 6.163695713880217e-05,
      "loss": 0.0019911292940378187,
      "step": 135210
    },
    {
      "epoch": 38.38206074368436,
      "grad_norm": 0.08032118529081345,
      "learning_rate": 6.163411864887881e-05,
      "loss": 0.001308680884540081,
      "step": 135220
    },
    {
      "epoch": 38.38489923360772,
      "grad_norm": 1.1914821863174438,
      "learning_rate": 6.163128015895544e-05,
      "loss": 0.005745340138673782,
      "step": 135230
    },
    {
      "epoch": 38.38773772353108,
      "grad_norm": 0.24814899265766144,
      "learning_rate": 6.162844166903208e-05,
      "loss": 0.0017528858035802842,
      "step": 135240
    },
    {
      "epoch": 38.390576213454445,
      "grad_norm": 4.454397201538086,
      "learning_rate": 6.162560317910872e-05,
      "loss": 0.003952928632497787,
      "step": 135250
    },
    {
      "epoch": 38.3934147033778,
      "grad_norm": 0.06842509657144547,
      "learning_rate": 6.162276468918535e-05,
      "loss": 0.005411344021558762,
      "step": 135260
    },
    {
      "epoch": 38.39625319330116,
      "grad_norm": 3.2981350421905518,
      "learning_rate": 6.161992619926199e-05,
      "loss": 0.002562856115400791,
      "step": 135270
    },
    {
      "epoch": 38.399091683224526,
      "grad_norm": 13.176570892333984,
      "learning_rate": 6.161708770933864e-05,
      "loss": 0.008661288768053055,
      "step": 135280
    },
    {
      "epoch": 38.40193017314788,
      "grad_norm": 0.8192607164382935,
      "learning_rate": 6.161424921941527e-05,
      "loss": 0.007953906059265136,
      "step": 135290
    },
    {
      "epoch": 38.404768663071245,
      "grad_norm": 0.5461180806159973,
      "learning_rate": 6.161141072949191e-05,
      "loss": 0.005651094764471054,
      "step": 135300
    },
    {
      "epoch": 38.40760715299461,
      "grad_norm": 1.5097296237945557,
      "learning_rate": 6.160857223956855e-05,
      "loss": 0.001774081028997898,
      "step": 135310
    },
    {
      "epoch": 38.41044564291797,
      "grad_norm": 0.9418918490409851,
      "learning_rate": 6.16057337496452e-05,
      "loss": 0.006616681814193726,
      "step": 135320
    },
    {
      "epoch": 38.413284132841326,
      "grad_norm": 0.11012096703052521,
      "learning_rate": 6.160289525972182e-05,
      "loss": 0.002159501425921917,
      "step": 135330
    },
    {
      "epoch": 38.41612262276469,
      "grad_norm": 0.8838050365447998,
      "learning_rate": 6.160005676979848e-05,
      "loss": 0.0020592864602804184,
      "step": 135340
    },
    {
      "epoch": 38.41896111268805,
      "grad_norm": 0.08345706015825272,
      "learning_rate": 6.159721827987512e-05,
      "loss": 0.005993042513728142,
      "step": 135350
    },
    {
      "epoch": 38.42179960261141,
      "grad_norm": 0.6298748254776001,
      "learning_rate": 6.159437978995175e-05,
      "loss": 0.014646728336811066,
      "step": 135360
    },
    {
      "epoch": 38.42463809253477,
      "grad_norm": 0.4491177499294281,
      "learning_rate": 6.159154130002839e-05,
      "loss": 0.0010119926184415818,
      "step": 135370
    },
    {
      "epoch": 38.427476582458134,
      "grad_norm": 7.686692714691162,
      "learning_rate": 6.158870281010503e-05,
      "loss": 0.0043272413313388824,
      "step": 135380
    },
    {
      "epoch": 38.43031507238149,
      "grad_norm": 0.6566407680511475,
      "learning_rate": 6.158586432018166e-05,
      "loss": 0.0029710086062550546,
      "step": 135390
    },
    {
      "epoch": 38.43315356230485,
      "grad_norm": 0.031021684408187866,
      "learning_rate": 6.15830258302583e-05,
      "loss": 0.0018673669546842575,
      "step": 135400
    },
    {
      "epoch": 38.435992052228215,
      "grad_norm": 0.09514325112104416,
      "learning_rate": 6.158018734033495e-05,
      "loss": 0.0032774358987808228,
      "step": 135410
    },
    {
      "epoch": 38.43883054215158,
      "grad_norm": 0.050353869795799255,
      "learning_rate": 6.157734885041158e-05,
      "loss": 0.0014400530606508255,
      "step": 135420
    },
    {
      "epoch": 38.441669032074934,
      "grad_norm": 0.2416570633649826,
      "learning_rate": 6.157451036048822e-05,
      "loss": 0.009577415883541107,
      "step": 135430
    },
    {
      "epoch": 38.4445075219983,
      "grad_norm": 7.328619480133057,
      "learning_rate": 6.157167187056486e-05,
      "loss": 0.011363565921783447,
      "step": 135440
    },
    {
      "epoch": 38.44734601192166,
      "grad_norm": 1.306361436843872,
      "learning_rate": 6.15688333806415e-05,
      "loss": 0.008180183917284011,
      "step": 135450
    },
    {
      "epoch": 38.450184501845015,
      "grad_norm": 0.6085950136184692,
      "learning_rate": 6.156599489071813e-05,
      "loss": 0.013829648494720459,
      "step": 135460
    },
    {
      "epoch": 38.45302299176838,
      "grad_norm": 2.418565034866333,
      "learning_rate": 6.156315640079477e-05,
      "loss": 0.0010970816016197204,
      "step": 135470
    },
    {
      "epoch": 38.45586148169174,
      "grad_norm": 0.11012092977762222,
      "learning_rate": 6.156031791087143e-05,
      "loss": 0.001015191338956356,
      "step": 135480
    },
    {
      "epoch": 38.458699971615104,
      "grad_norm": 0.13875535130500793,
      "learning_rate": 6.155747942094806e-05,
      "loss": 0.004551955312490463,
      "step": 135490
    },
    {
      "epoch": 38.46153846153846,
      "grad_norm": 12.91567611694336,
      "learning_rate": 6.15546409310247e-05,
      "loss": 0.007015454769134522,
      "step": 135500
    },
    {
      "epoch": 38.46153846153846,
      "eval_accuracy": 0.9723405608189737,
      "eval_loss": 0.09708716720342636,
      "eval_runtime": 31.9768,
      "eval_samples_per_second": 491.825,
      "eval_steps_per_second": 7.693,
      "step": 135500
    },
    {
      "epoch": 38.46437695146182,
      "grad_norm": 0.1532626748085022,
      "learning_rate": 6.155180244110134e-05,
      "loss": 0.007028988748788834,
      "step": 135510
    },
    {
      "epoch": 38.467215441385186,
      "grad_norm": 0.14090636372566223,
      "learning_rate": 6.154896395117797e-05,
      "loss": 0.0077640779316425325,
      "step": 135520
    },
    {
      "epoch": 38.47005393130854,
      "grad_norm": 0.13461485505104065,
      "learning_rate": 6.154612546125461e-05,
      "loss": 0.016653384268283843,
      "step": 135530
    },
    {
      "epoch": 38.472892421231904,
      "grad_norm": 3.1759281158447266,
      "learning_rate": 6.154328697133126e-05,
      "loss": 0.0027630701661109924,
      "step": 135540
    },
    {
      "epoch": 38.47573091115527,
      "grad_norm": 0.9256454706192017,
      "learning_rate": 6.154044848140789e-05,
      "loss": 0.0050191331654787065,
      "step": 135550
    },
    {
      "epoch": 38.47856940107862,
      "grad_norm": 0.09942079335451126,
      "learning_rate": 6.153760999148453e-05,
      "loss": 0.0005018610507249832,
      "step": 135560
    },
    {
      "epoch": 38.481407891001986,
      "grad_norm": 0.15628722310066223,
      "learning_rate": 6.153477150156117e-05,
      "loss": 0.0029232488945126534,
      "step": 135570
    },
    {
      "epoch": 38.48424638092535,
      "grad_norm": 0.13773216307163239,
      "learning_rate": 6.153193301163782e-05,
      "loss": 0.0012789636850357056,
      "step": 135580
    },
    {
      "epoch": 38.48708487084871,
      "grad_norm": 0.39429011940956116,
      "learning_rate": 6.152909452171444e-05,
      "loss": 0.0008003953844308853,
      "step": 135590
    },
    {
      "epoch": 38.48992336077207,
      "grad_norm": 0.04783034324645996,
      "learning_rate": 6.152625603179108e-05,
      "loss": 0.0008941607549786568,
      "step": 135600
    },
    {
      "epoch": 38.49276185069543,
      "grad_norm": 0.1754629760980606,
      "learning_rate": 6.152341754186774e-05,
      "loss": 0.0018739040940999984,
      "step": 135610
    },
    {
      "epoch": 38.49560034061879,
      "grad_norm": 0.04675730690360069,
      "learning_rate": 6.152057905194437e-05,
      "loss": 0.0011871449649333953,
      "step": 135620
    },
    {
      "epoch": 38.49843883054215,
      "grad_norm": 0.028517620638012886,
      "learning_rate": 6.151774056202101e-05,
      "loss": 0.0008887562900781632,
      "step": 135630
    },
    {
      "epoch": 38.50127732046551,
      "grad_norm": 0.3093649744987488,
      "learning_rate": 6.151490207209765e-05,
      "loss": 0.0036879539489746093,
      "step": 135640
    },
    {
      "epoch": 38.504115810388875,
      "grad_norm": 2.7688379287719727,
      "learning_rate": 6.151206358217428e-05,
      "loss": 0.002188309095799923,
      "step": 135650
    },
    {
      "epoch": 38.50695430031223,
      "grad_norm": 0.022523559629917145,
      "learning_rate": 6.150922509225092e-05,
      "loss": 0.0018501771613955499,
      "step": 135660
    },
    {
      "epoch": 38.50979279023559,
      "grad_norm": 0.02356768399477005,
      "learning_rate": 6.150638660232756e-05,
      "loss": 0.010983145236968994,
      "step": 135670
    },
    {
      "epoch": 38.512631280158956,
      "grad_norm": 2.3858256340026855,
      "learning_rate": 6.15035481124042e-05,
      "loss": 0.0036229632794857027,
      "step": 135680
    },
    {
      "epoch": 38.51546977008232,
      "grad_norm": 0.12981124222278595,
      "learning_rate": 6.150070962248084e-05,
      "loss": 0.0038952045142650604,
      "step": 135690
    },
    {
      "epoch": 38.518308260005675,
      "grad_norm": 0.16601401567459106,
      "learning_rate": 6.149787113255749e-05,
      "loss": 0.0012140143662691116,
      "step": 135700
    },
    {
      "epoch": 38.52114674992904,
      "grad_norm": 0.3520420491695404,
      "learning_rate": 6.149503264263413e-05,
      "loss": 0.0020994456484913827,
      "step": 135710
    },
    {
      "epoch": 38.5239852398524,
      "grad_norm": 2.565340757369995,
      "learning_rate": 6.149219415271075e-05,
      "loss": 0.0010496702045202256,
      "step": 135720
    },
    {
      "epoch": 38.52682372977576,
      "grad_norm": 1.2032439708709717,
      "learning_rate": 6.14893556627874e-05,
      "loss": 0.0026920245960354804,
      "step": 135730
    },
    {
      "epoch": 38.52966221969912,
      "grad_norm": 28.78321075439453,
      "learning_rate": 6.148651717286405e-05,
      "loss": 0.017881177365779877,
      "step": 135740
    },
    {
      "epoch": 38.53250070962248,
      "grad_norm": 0.04024608060717583,
      "learning_rate": 6.148367868294068e-05,
      "loss": 0.00619904138147831,
      "step": 135750
    },
    {
      "epoch": 38.535339199545845,
      "grad_norm": 1.1123851537704468,
      "learning_rate": 6.148084019301732e-05,
      "loss": 0.0021531926468014717,
      "step": 135760
    },
    {
      "epoch": 38.5381776894692,
      "grad_norm": 0.3989243507385254,
      "learning_rate": 6.147800170309396e-05,
      "loss": 0.0010667555034160614,
      "step": 135770
    },
    {
      "epoch": 38.541016179392564,
      "grad_norm": 0.2814774811267853,
      "learning_rate": 6.147516321317059e-05,
      "loss": 0.0010479971766471862,
      "step": 135780
    },
    {
      "epoch": 38.54385466931593,
      "grad_norm": 0.3770263195037842,
      "learning_rate": 6.147232472324723e-05,
      "loss": 0.0008475968614220619,
      "step": 135790
    },
    {
      "epoch": 38.54669315923928,
      "grad_norm": 0.15267719328403473,
      "learning_rate": 6.146948623332387e-05,
      "loss": 0.001021953672170639,
      "step": 135800
    },
    {
      "epoch": 38.549531649162645,
      "grad_norm": 0.18413841724395752,
      "learning_rate": 6.146664774340051e-05,
      "loss": 0.0028862584382295608,
      "step": 135810
    },
    {
      "epoch": 38.55237013908601,
      "grad_norm": 0.7051997780799866,
      "learning_rate": 6.146380925347715e-05,
      "loss": 0.0021086476743221285,
      "step": 135820
    },
    {
      "epoch": 38.555208629009364,
      "grad_norm": 1.5941591262817383,
      "learning_rate": 6.14609707635538e-05,
      "loss": 0.006381133943796158,
      "step": 135830
    },
    {
      "epoch": 38.55804711893273,
      "grad_norm": 0.24233686923980713,
      "learning_rate": 6.145813227363044e-05,
      "loss": 0.0007729744538664818,
      "step": 135840
    },
    {
      "epoch": 38.56088560885609,
      "grad_norm": 0.1543876677751541,
      "learning_rate": 6.145529378370707e-05,
      "loss": 0.002078931592404842,
      "step": 135850
    },
    {
      "epoch": 38.56372409877945,
      "grad_norm": 0.5416765809059143,
      "learning_rate": 6.14524552937837e-05,
      "loss": 0.001722593791782856,
      "step": 135860
    },
    {
      "epoch": 38.56656258870281,
      "grad_norm": 0.35925862193107605,
      "learning_rate": 6.144961680386035e-05,
      "loss": 0.0046719186007976535,
      "step": 135870
    },
    {
      "epoch": 38.56940107862617,
      "grad_norm": 0.0941983312368393,
      "learning_rate": 6.144677831393699e-05,
      "loss": 0.0016037749126553535,
      "step": 135880
    },
    {
      "epoch": 38.572239568549534,
      "grad_norm": 4.017645835876465,
      "learning_rate": 6.144393982401363e-05,
      "loss": 0.0025598552078008653,
      "step": 135890
    },
    {
      "epoch": 38.57507805847289,
      "grad_norm": 0.13510210812091827,
      "learning_rate": 6.144110133409027e-05,
      "loss": 0.0008298946544528007,
      "step": 135900
    },
    {
      "epoch": 38.57791654839625,
      "grad_norm": 0.23703446984291077,
      "learning_rate": 6.14382628441669e-05,
      "loss": 0.002044939436018467,
      "step": 135910
    },
    {
      "epoch": 38.580755038319616,
      "grad_norm": 0.19244232773780823,
      "learning_rate": 6.143542435424354e-05,
      "loss": 0.0028385674580931664,
      "step": 135920
    },
    {
      "epoch": 38.58359352824297,
      "grad_norm": 0.09505518525838852,
      "learning_rate": 6.143258586432018e-05,
      "loss": 0.0015394626185297965,
      "step": 135930
    },
    {
      "epoch": 38.586432018166335,
      "grad_norm": 3.29378342628479,
      "learning_rate": 6.142974737439682e-05,
      "loss": 0.015290457010269164,
      "step": 135940
    },
    {
      "epoch": 38.5892705080897,
      "grad_norm": 0.8285510540008545,
      "learning_rate": 6.142690888447347e-05,
      "loss": 0.0037109963595867157,
      "step": 135950
    },
    {
      "epoch": 38.59210899801306,
      "grad_norm": 3.2156736850738525,
      "learning_rate": 6.14240703945501e-05,
      "loss": 0.001501442864537239,
      "step": 135960
    },
    {
      "epoch": 38.594947487936416,
      "grad_norm": 0.10440534353256226,
      "learning_rate": 6.142123190462675e-05,
      "loss": 0.004285608977079391,
      "step": 135970
    },
    {
      "epoch": 38.59778597785978,
      "grad_norm": 1.7456245422363281,
      "learning_rate": 6.141839341470338e-05,
      "loss": 0.0019434705376625062,
      "step": 135980
    },
    {
      "epoch": 38.60062446778314,
      "grad_norm": 0.05066037178039551,
      "learning_rate": 6.141555492478002e-05,
      "loss": 0.001310076005756855,
      "step": 135990
    },
    {
      "epoch": 38.6034629577065,
      "grad_norm": 0.11218142509460449,
      "learning_rate": 6.141271643485666e-05,
      "loss": 0.0029816806316375734,
      "step": 136000
    },
    {
      "epoch": 38.6034629577065,
      "eval_accuracy": 0.9762192407960831,
      "eval_loss": 0.08278802782297134,
      "eval_runtime": 32.1749,
      "eval_samples_per_second": 488.797,
      "eval_steps_per_second": 7.646,
      "step": 136000
    },
    {
      "epoch": 38.60630144762986,
      "grad_norm": 0.04110505059361458,
      "learning_rate": 6.14098779449333e-05,
      "loss": 0.00704413428902626,
      "step": 136010
    },
    {
      "epoch": 38.60913993755322,
      "grad_norm": 0.20143942534923553,
      "learning_rate": 6.140703945500994e-05,
      "loss": 0.0005810655653476715,
      "step": 136020
    },
    {
      "epoch": 38.61197842747658,
      "grad_norm": 5.58424711227417,
      "learning_rate": 6.140420096508658e-05,
      "loss": 0.009471157193183899,
      "step": 136030
    },
    {
      "epoch": 38.61481691739994,
      "grad_norm": 2.136950969696045,
      "learning_rate": 6.140136247516321e-05,
      "loss": 0.015326546132564544,
      "step": 136040
    },
    {
      "epoch": 38.617655407323305,
      "grad_norm": 0.05483608320355415,
      "learning_rate": 6.139852398523985e-05,
      "loss": 0.002490892820060253,
      "step": 136050
    },
    {
      "epoch": 38.62049389724667,
      "grad_norm": 0.054581888020038605,
      "learning_rate": 6.13956854953165e-05,
      "loss": 0.0021986329928040506,
      "step": 136060
    },
    {
      "epoch": 38.623332387170024,
      "grad_norm": 0.2335558980703354,
      "learning_rate": 6.139284700539313e-05,
      "loss": 0.0022899048402905465,
      "step": 136070
    },
    {
      "epoch": 38.62617087709339,
      "grad_norm": 1.161386489868164,
      "learning_rate": 6.139000851546978e-05,
      "loss": 0.0024105969816446303,
      "step": 136080
    },
    {
      "epoch": 38.62900936701675,
      "grad_norm": 0.3985218405723572,
      "learning_rate": 6.138717002554642e-05,
      "loss": 0.0021485598757863046,
      "step": 136090
    },
    {
      "epoch": 38.631847856940105,
      "grad_norm": 0.39690661430358887,
      "learning_rate": 6.138433153562305e-05,
      "loss": 0.0011351585388183593,
      "step": 136100
    },
    {
      "epoch": 38.63468634686347,
      "grad_norm": 0.9182251691818237,
      "learning_rate": 6.138149304569969e-05,
      "loss": 0.0016373202204704284,
      "step": 136110
    },
    {
      "epoch": 38.63752483678683,
      "grad_norm": 12.3392915725708,
      "learning_rate": 6.137865455577633e-05,
      "loss": 0.011935946345329285,
      "step": 136120
    },
    {
      "epoch": 38.64036332671019,
      "grad_norm": 0.05375846102833748,
      "learning_rate": 6.137581606585297e-05,
      "loss": 0.0032425083220005034,
      "step": 136130
    },
    {
      "epoch": 38.64320181663355,
      "grad_norm": 0.26288774609565735,
      "learning_rate": 6.137297757592961e-05,
      "loss": 0.008097366988658905,
      "step": 136140
    },
    {
      "epoch": 38.64604030655691,
      "grad_norm": 0.059297073632478714,
      "learning_rate": 6.137042293499858e-05,
      "loss": 0.022571422159671783,
      "step": 136150
    },
    {
      "epoch": 38.648878796480275,
      "grad_norm": 1.421044111251831,
      "learning_rate": 6.136758444507522e-05,
      "loss": 0.011364725977182388,
      "step": 136160
    },
    {
      "epoch": 38.65171728640363,
      "grad_norm": 0.16265228390693665,
      "learning_rate": 6.136474595515186e-05,
      "loss": 0.0038663528859615324,
      "step": 136170
    },
    {
      "epoch": 38.654555776326994,
      "grad_norm": 7.549160957336426,
      "learning_rate": 6.13619074652285e-05,
      "loss": 0.0059774771332740785,
      "step": 136180
    },
    {
      "epoch": 38.65739426625036,
      "grad_norm": 0.24436140060424805,
      "learning_rate": 6.135906897530514e-05,
      "loss": 0.006120540201663971,
      "step": 136190
    },
    {
      "epoch": 38.66023275617371,
      "grad_norm": 0.036731258034706116,
      "learning_rate": 6.135623048538178e-05,
      "loss": 0.006040460243821144,
      "step": 136200
    },
    {
      "epoch": 38.663071246097076,
      "grad_norm": 0.08173620700836182,
      "learning_rate": 6.135339199545843e-05,
      "loss": 0.005126922205090523,
      "step": 136210
    },
    {
      "epoch": 38.66590973602044,
      "grad_norm": 1.0930724143981934,
      "learning_rate": 6.135055350553505e-05,
      "loss": 0.001257179118692875,
      "step": 136220
    },
    {
      "epoch": 38.6687482259438,
      "grad_norm": 1.3666458129882812,
      "learning_rate": 6.13477150156117e-05,
      "loss": 0.009613914787769318,
      "step": 136230
    },
    {
      "epoch": 38.67158671586716,
      "grad_norm": 0.29617321491241455,
      "learning_rate": 6.134487652568834e-05,
      "loss": 0.013505506515502929,
      "step": 136240
    },
    {
      "epoch": 38.67442520579052,
      "grad_norm": 0.19346386194229126,
      "learning_rate": 6.134203803576496e-05,
      "loss": 0.006604056805372238,
      "step": 136250
    },
    {
      "epoch": 38.67726369571388,
      "grad_norm": 0.9535866975784302,
      "learning_rate": 6.133919954584162e-05,
      "loss": 0.0025373678654432297,
      "step": 136260
    },
    {
      "epoch": 38.68010218563724,
      "grad_norm": 2.5597715377807617,
      "learning_rate": 6.133636105591826e-05,
      "loss": 0.001963675394654274,
      "step": 136270
    },
    {
      "epoch": 38.6829406755606,
      "grad_norm": 0.18233366310596466,
      "learning_rate": 6.133352256599489e-05,
      "loss": 0.0017403900623321534,
      "step": 136280
    },
    {
      "epoch": 38.685779165483964,
      "grad_norm": 0.3750947117805481,
      "learning_rate": 6.133068407607153e-05,
      "loss": 0.0013620039448142053,
      "step": 136290
    },
    {
      "epoch": 38.68861765540732,
      "grad_norm": 0.03508025035262108,
      "learning_rate": 6.132784558614817e-05,
      "loss": 0.000747206062078476,
      "step": 136300
    },
    {
      "epoch": 38.69145614533068,
      "grad_norm": 9.164497375488281,
      "learning_rate": 6.132500709622481e-05,
      "loss": 0.002782498300075531,
      "step": 136310
    },
    {
      "epoch": 38.694294635254046,
      "grad_norm": 0.5890185832977295,
      "learning_rate": 6.132216860630145e-05,
      "loss": 0.0012815730646252633,
      "step": 136320
    },
    {
      "epoch": 38.69713312517741,
      "grad_norm": 0.29477575421333313,
      "learning_rate": 6.13193301163781e-05,
      "loss": 0.001544569805264473,
      "step": 136330
    },
    {
      "epoch": 38.699971615100765,
      "grad_norm": 0.16373345255851746,
      "learning_rate": 6.131649162645474e-05,
      "loss": 0.003889068216085434,
      "step": 136340
    },
    {
      "epoch": 38.70281010502413,
      "grad_norm": 0.0695229098200798,
      "learning_rate": 6.131365313653136e-05,
      "loss": 0.00486629381775856,
      "step": 136350
    },
    {
      "epoch": 38.70564859494749,
      "grad_norm": 2.305974006652832,
      "learning_rate": 6.1310814646608e-05,
      "loss": 0.0033214457333087923,
      "step": 136360
    },
    {
      "epoch": 38.708487084870846,
      "grad_norm": 0.14764226973056793,
      "learning_rate": 6.130797615668465e-05,
      "loss": 0.004053351655602455,
      "step": 136370
    },
    {
      "epoch": 38.71132557479421,
      "grad_norm": 10.6625394821167,
      "learning_rate": 6.130513766676127e-05,
      "loss": 0.006433658301830292,
      "step": 136380
    },
    {
      "epoch": 38.71416406471757,
      "grad_norm": 3.6864173412323,
      "learning_rate": 6.130229917683793e-05,
      "loss": 0.005073889344930649,
      "step": 136390
    },
    {
      "epoch": 38.71700255464093,
      "grad_norm": 0.9584582448005676,
      "learning_rate": 6.129946068691457e-05,
      "loss": 0.0029225749894976616,
      "step": 136400
    },
    {
      "epoch": 38.71984104456429,
      "grad_norm": 0.2951776087284088,
      "learning_rate": 6.12966221969912e-05,
      "loss": 0.0012059558182954788,
      "step": 136410
    },
    {
      "epoch": 38.72267953448765,
      "grad_norm": 0.6344748139381409,
      "learning_rate": 6.129378370706784e-05,
      "loss": 0.010846298187971115,
      "step": 136420
    },
    {
      "epoch": 38.725518024411016,
      "grad_norm": 0.1910606175661087,
      "learning_rate": 6.129094521714448e-05,
      "loss": 0.0012677567079663276,
      "step": 136430
    },
    {
      "epoch": 38.72835651433437,
      "grad_norm": 6.896583080291748,
      "learning_rate": 6.128810672722112e-05,
      "loss": 0.0036260675638914107,
      "step": 136440
    },
    {
      "epoch": 38.731195004257735,
      "grad_norm": 4.85918664932251,
      "learning_rate": 6.128526823729776e-05,
      "loss": 0.0026777951046824456,
      "step": 136450
    },
    {
      "epoch": 38.7340334941811,
      "grad_norm": 0.04556845501065254,
      "learning_rate": 6.12824297473744e-05,
      "loss": 0.006358266621828079,
      "step": 136460
    },
    {
      "epoch": 38.736871984104454,
      "grad_norm": 0.6142772436141968,
      "learning_rate": 6.127959125745105e-05,
      "loss": 0.002424268238246441,
      "step": 136470
    },
    {
      "epoch": 38.73971047402782,
      "grad_norm": 1.728261113166809,
      "learning_rate": 6.127675276752767e-05,
      "loss": 0.011589571833610535,
      "step": 136480
    },
    {
      "epoch": 38.74254896395118,
      "grad_norm": 0.058594170957803726,
      "learning_rate": 6.127391427760432e-05,
      "loss": 0.005717622488737107,
      "step": 136490
    },
    {
      "epoch": 38.745387453874535,
      "grad_norm": 0.03481850400567055,
      "learning_rate": 6.127107578768096e-05,
      "loss": 0.014588183164596558,
      "step": 136500
    },
    {
      "epoch": 38.745387453874535,
      "eval_accuracy": 0.9734215044191518,
      "eval_loss": 0.09129461646080017,
      "eval_runtime": 31.6834,
      "eval_samples_per_second": 496.379,
      "eval_steps_per_second": 7.764,
      "step": 136500
    },
    {
      "epoch": 38.7482259437979,
      "grad_norm": 0.11015772819519043,
      "learning_rate": 6.126823729775759e-05,
      "loss": 0.0039011433720588682,
      "step": 136510
    },
    {
      "epoch": 38.75106443372126,
      "grad_norm": 12.22962760925293,
      "learning_rate": 6.126539880783424e-05,
      "loss": 0.014401167631149292,
      "step": 136520
    },
    {
      "epoch": 38.753902923644624,
      "grad_norm": 2.4870171546936035,
      "learning_rate": 6.126256031791088e-05,
      "loss": 0.0054236166179180145,
      "step": 136530
    },
    {
      "epoch": 38.75674141356798,
      "grad_norm": 0.3709111511707306,
      "learning_rate": 6.125972182798751e-05,
      "loss": 0.0009145205840468407,
      "step": 136540
    },
    {
      "epoch": 38.75957990349134,
      "grad_norm": 0.06053968518972397,
      "learning_rate": 6.125688333806415e-05,
      "loss": 0.004316484555602074,
      "step": 136550
    },
    {
      "epoch": 38.762418393414706,
      "grad_norm": 10.312936782836914,
      "learning_rate": 6.125404484814079e-05,
      "loss": 0.00904449075460434,
      "step": 136560
    },
    {
      "epoch": 38.76525688333806,
      "grad_norm": 0.06283903121948242,
      "learning_rate": 6.125120635821743e-05,
      "loss": 0.002323652245104313,
      "step": 136570
    },
    {
      "epoch": 38.768095373261424,
      "grad_norm": 0.4236854910850525,
      "learning_rate": 6.124836786829406e-05,
      "loss": 0.004421686381101608,
      "step": 136580
    },
    {
      "epoch": 38.77093386318479,
      "grad_norm": 2.8406593799591064,
      "learning_rate": 6.124552937837072e-05,
      "loss": 0.003348906338214874,
      "step": 136590
    },
    {
      "epoch": 38.77377235310815,
      "grad_norm": 0.9530077576637268,
      "learning_rate": 6.124269088844736e-05,
      "loss": 0.0016872406005859375,
      "step": 136600
    },
    {
      "epoch": 38.776610843031506,
      "grad_norm": 0.20477396249771118,
      "learning_rate": 6.123985239852399e-05,
      "loss": 0.0012434175238013267,
      "step": 136610
    },
    {
      "epoch": 38.77944933295487,
      "grad_norm": 7.7630133628845215,
      "learning_rate": 6.123701390860063e-05,
      "loss": 0.002184251695871353,
      "step": 136620
    },
    {
      "epoch": 38.78228782287823,
      "grad_norm": 0.09650157392024994,
      "learning_rate": 6.123417541867727e-05,
      "loss": 0.0012489393353462218,
      "step": 136630
    },
    {
      "epoch": 38.78512631280159,
      "grad_norm": 0.19266565144062042,
      "learning_rate": 6.12313369287539e-05,
      "loss": 0.0038189269602298737,
      "step": 136640
    },
    {
      "epoch": 38.78796480272495,
      "grad_norm": 0.72478848695755,
      "learning_rate": 6.122849843883055e-05,
      "loss": 0.0020806048065423965,
      "step": 136650
    },
    {
      "epoch": 38.79080329264831,
      "grad_norm": 0.16608668863773346,
      "learning_rate": 6.122565994890719e-05,
      "loss": 0.002335289493203163,
      "step": 136660
    },
    {
      "epoch": 38.79364178257167,
      "grad_norm": 0.03468051180243492,
      "learning_rate": 6.122282145898382e-05,
      "loss": 0.0026918921619653704,
      "step": 136670
    },
    {
      "epoch": 38.79648027249503,
      "grad_norm": 0.2249375581741333,
      "learning_rate": 6.121998296906046e-05,
      "loss": 0.0068877778947353365,
      "step": 136680
    },
    {
      "epoch": 38.799318762418395,
      "grad_norm": 0.19205904006958008,
      "learning_rate": 6.12171444791371e-05,
      "loss": 0.004375426471233368,
      "step": 136690
    },
    {
      "epoch": 38.80215725234176,
      "grad_norm": 1.9620513916015625,
      "learning_rate": 6.121430598921374e-05,
      "loss": 0.002422507479786873,
      "step": 136700
    },
    {
      "epoch": 38.80499574226511,
      "grad_norm": 0.35271015763282776,
      "learning_rate": 6.121146749929037e-05,
      "loss": 0.0030752597376704216,
      "step": 136710
    },
    {
      "epoch": 38.807834232188476,
      "grad_norm": 0.3097192943096161,
      "learning_rate": 6.120862900936703e-05,
      "loss": 0.0034665547311306,
      "step": 136720
    },
    {
      "epoch": 38.81067272211184,
      "grad_norm": 0.17545726895332336,
      "learning_rate": 6.120579051944367e-05,
      "loss": 0.005319380015134811,
      "step": 136730
    },
    {
      "epoch": 38.813511212035195,
      "grad_norm": 0.2879747748374939,
      "learning_rate": 6.12029520295203e-05,
      "loss": 0.008793916553258896,
      "step": 136740
    },
    {
      "epoch": 38.81634970195856,
      "grad_norm": 0.45000073313713074,
      "learning_rate": 6.120011353959694e-05,
      "loss": 0.0033906400203704834,
      "step": 136750
    },
    {
      "epoch": 38.81918819188192,
      "grad_norm": 2.3715920448303223,
      "learning_rate": 6.119727504967358e-05,
      "loss": 0.01173495054244995,
      "step": 136760
    },
    {
      "epoch": 38.822026681805276,
      "grad_norm": 10.430177688598633,
      "learning_rate": 6.119443655975021e-05,
      "loss": 0.008167661726474762,
      "step": 136770
    },
    {
      "epoch": 38.82486517172864,
      "grad_norm": 0.3860531449317932,
      "learning_rate": 6.119159806982685e-05,
      "loss": 0.011427169293165207,
      "step": 136780
    },
    {
      "epoch": 38.827703661652,
      "grad_norm": 0.06467914581298828,
      "learning_rate": 6.11887595799035e-05,
      "loss": 0.010730832815170288,
      "step": 136790
    },
    {
      "epoch": 38.830542151575365,
      "grad_norm": 0.8924474716186523,
      "learning_rate": 6.118592108998013e-05,
      "loss": 0.009281785786151886,
      "step": 136800
    },
    {
      "epoch": 38.83338064149872,
      "grad_norm": 0.23478184640407562,
      "learning_rate": 6.118308260005677e-05,
      "loss": 0.021039161086082458,
      "step": 136810
    },
    {
      "epoch": 38.836219131422084,
      "grad_norm": 0.09967546910047531,
      "learning_rate": 6.118024411013341e-05,
      "loss": 0.018042591214179993,
      "step": 136820
    },
    {
      "epoch": 38.83905762134545,
      "grad_norm": 0.27845102548599243,
      "learning_rate": 6.117740562021006e-05,
      "loss": 0.015605142712593079,
      "step": 136830
    },
    {
      "epoch": 38.8418961112688,
      "grad_norm": 2.200705051422119,
      "learning_rate": 6.117456713028668e-05,
      "loss": 0.012432622909545898,
      "step": 136840
    },
    {
      "epoch": 38.844734601192165,
      "grad_norm": 0.06835105270147324,
      "learning_rate": 6.117172864036334e-05,
      "loss": 0.0038689129054546357,
      "step": 136850
    },
    {
      "epoch": 38.84757309111553,
      "grad_norm": 0.27165040373802185,
      "learning_rate": 6.116889015043998e-05,
      "loss": 0.006652887910604477,
      "step": 136860
    },
    {
      "epoch": 38.850411581038884,
      "grad_norm": 0.9472457766532898,
      "learning_rate": 6.116605166051661e-05,
      "loss": 0.0073095932602882385,
      "step": 136870
    },
    {
      "epoch": 38.85325007096225,
      "grad_norm": 3.0258140563964844,
      "learning_rate": 6.116321317059325e-05,
      "loss": 0.013051986694335938,
      "step": 136880
    },
    {
      "epoch": 38.85608856088561,
      "grad_norm": 0.17335623502731323,
      "learning_rate": 6.116037468066989e-05,
      "loss": 0.0005784519016742706,
      "step": 136890
    },
    {
      "epoch": 38.85892705080897,
      "grad_norm": 5.466840744018555,
      "learning_rate": 6.115753619074652e-05,
      "loss": 0.0043010756373405455,
      "step": 136900
    },
    {
      "epoch": 38.86176554073233,
      "grad_norm": 0.2209157943725586,
      "learning_rate": 6.115469770082316e-05,
      "loss": 0.003944623470306397,
      "step": 136910
    },
    {
      "epoch": 38.86460403065569,
      "grad_norm": 0.21972675621509552,
      "learning_rate": 6.115185921089981e-05,
      "loss": 0.018045033514499664,
      "step": 136920
    },
    {
      "epoch": 38.867442520579054,
      "grad_norm": 1.0783528089523315,
      "learning_rate": 6.114902072097644e-05,
      "loss": 0.017407304048538207,
      "step": 136930
    },
    {
      "epoch": 38.87028101050241,
      "grad_norm": 0.1905258446931839,
      "learning_rate": 6.114618223105308e-05,
      "loss": 0.007618250697851181,
      "step": 136940
    },
    {
      "epoch": 38.87311950042577,
      "grad_norm": 1.7849047183990479,
      "learning_rate": 6.114334374112972e-05,
      "loss": 0.006351820379495621,
      "step": 136950
    },
    {
      "epoch": 38.875957990349136,
      "grad_norm": 0.045467957854270935,
      "learning_rate": 6.114050525120637e-05,
      "loss": 0.0009369080886244774,
      "step": 136960
    },
    {
      "epoch": 38.8787964802725,
      "grad_norm": 1.6275182962417603,
      "learning_rate": 6.1137666761283e-05,
      "loss": 0.0020040601491928102,
      "step": 136970
    },
    {
      "epoch": 38.881634970195854,
      "grad_norm": 0.13557808101177216,
      "learning_rate": 6.113482827135964e-05,
      "loss": 0.014672042429447174,
      "step": 136980
    },
    {
      "epoch": 38.88447346011922,
      "grad_norm": 2.6658971309661865,
      "learning_rate": 6.113198978143628e-05,
      "loss": 0.01650681495666504,
      "step": 136990
    },
    {
      "epoch": 38.88731195004258,
      "grad_norm": 1.719480037689209,
      "learning_rate": 6.112915129151292e-05,
      "loss": 0.004433354735374451,
      "step": 137000
    },
    {
      "epoch": 38.88731195004258,
      "eval_accuracy": 0.9740573535957271,
      "eval_loss": 0.09641898423433304,
      "eval_runtime": 31.7762,
      "eval_samples_per_second": 494.93,
      "eval_steps_per_second": 7.742,
      "step": 137000
    },
    {
      "epoch": 38.890150439965936,
      "grad_norm": 7.064290523529053,
      "learning_rate": 6.112631280158956e-05,
      "loss": 0.009441594779491424,
      "step": 137010
    },
    {
      "epoch": 38.8929889298893,
      "grad_norm": 0.36176249384880066,
      "learning_rate": 6.11234743116662e-05,
      "loss": 0.0051873397082090374,
      "step": 137020
    },
    {
      "epoch": 38.89582741981266,
      "grad_norm": 9.650701522827148,
      "learning_rate": 6.112063582174283e-05,
      "loss": 0.0037537820637226103,
      "step": 137030
    },
    {
      "epoch": 38.89866590973602,
      "grad_norm": 0.9703720211982727,
      "learning_rate": 6.111779733181947e-05,
      "loss": 0.0056422889232635495,
      "step": 137040
    },
    {
      "epoch": 38.90150439965938,
      "grad_norm": 0.5276385545730591,
      "learning_rate": 6.111495884189612e-05,
      "loss": 0.006073486804962158,
      "step": 137050
    },
    {
      "epoch": 38.90434288958274,
      "grad_norm": 0.4081055819988251,
      "learning_rate": 6.111212035197275e-05,
      "loss": 0.0009991828352212905,
      "step": 137060
    },
    {
      "epoch": 38.907181379506106,
      "grad_norm": 0.36674511432647705,
      "learning_rate": 6.11092818620494e-05,
      "loss": 0.001810632459819317,
      "step": 137070
    },
    {
      "epoch": 38.91001986942946,
      "grad_norm": 0.038281768560409546,
      "learning_rate": 6.110644337212604e-05,
      "loss": 0.005989895015954971,
      "step": 137080
    },
    {
      "epoch": 38.912858359352825,
      "grad_norm": 0.1562529057264328,
      "learning_rate": 6.110360488220266e-05,
      "loss": 0.0038223348557949064,
      "step": 137090
    },
    {
      "epoch": 38.91569684927619,
      "grad_norm": 9.216922760009766,
      "learning_rate": 6.11007663922793e-05,
      "loss": 0.00597425177693367,
      "step": 137100
    },
    {
      "epoch": 38.91853533919954,
      "grad_norm": 0.08718281984329224,
      "learning_rate": 6.109792790235595e-05,
      "loss": 0.000975412130355835,
      "step": 137110
    },
    {
      "epoch": 38.921373829122906,
      "grad_norm": 0.22173075377941132,
      "learning_rate": 6.109508941243259e-05,
      "loss": 0.0017149850726127624,
      "step": 137120
    },
    {
      "epoch": 38.92421231904627,
      "grad_norm": 0.46883898973464966,
      "learning_rate": 6.109225092250923e-05,
      "loss": 0.008884665369987488,
      "step": 137130
    },
    {
      "epoch": 38.927050808969625,
      "grad_norm": 0.26036038994789124,
      "learning_rate": 6.108941243258587e-05,
      "loss": 0.004381699115037918,
      "step": 137140
    },
    {
      "epoch": 38.92988929889299,
      "grad_norm": 0.23351457715034485,
      "learning_rate": 6.108657394266251e-05,
      "loss": 0.0016026290133595467,
      "step": 137150
    },
    {
      "epoch": 38.93272778881635,
      "grad_norm": 0.12381860613822937,
      "learning_rate": 6.108373545273914e-05,
      "loss": 0.011141019314527512,
      "step": 137160
    },
    {
      "epoch": 38.935566278739714,
      "grad_norm": 5.0101752281188965,
      "learning_rate": 6.108089696281578e-05,
      "loss": 0.002365664765238762,
      "step": 137170
    },
    {
      "epoch": 38.93840476866307,
      "grad_norm": 0.6688987612724304,
      "learning_rate": 6.107805847289242e-05,
      "loss": 0.0030611764639616014,
      "step": 137180
    },
    {
      "epoch": 38.94124325858643,
      "grad_norm": 8.807551383972168,
      "learning_rate": 6.107521998296906e-05,
      "loss": 0.004492054507136345,
      "step": 137190
    },
    {
      "epoch": 38.944081748509795,
      "grad_norm": 0.2584785223007202,
      "learning_rate": 6.10723814930457e-05,
      "loss": 0.00657244622707367,
      "step": 137200
    },
    {
      "epoch": 38.94692023843315,
      "grad_norm": 16.21637535095215,
      "learning_rate": 6.106954300312235e-05,
      "loss": 0.010976333171129227,
      "step": 137210
    },
    {
      "epoch": 38.949758728356514,
      "grad_norm": 0.09858343750238419,
      "learning_rate": 6.106670451319897e-05,
      "loss": 0.0068354949355125426,
      "step": 137220
    },
    {
      "epoch": 38.95259721827988,
      "grad_norm": 0.296319842338562,
      "learning_rate": 6.106386602327562e-05,
      "loss": 0.005221719294786454,
      "step": 137230
    },
    {
      "epoch": 38.95543570820323,
      "grad_norm": 2.3170037269592285,
      "learning_rate": 6.106102753335226e-05,
      "loss": 0.0049011029303073885,
      "step": 137240
    },
    {
      "epoch": 38.958274198126595,
      "grad_norm": 0.7258913516998291,
      "learning_rate": 6.10581890434289e-05,
      "loss": 0.0026680197566747665,
      "step": 137250
    },
    {
      "epoch": 38.96111268804996,
      "grad_norm": 0.4000585079193115,
      "learning_rate": 6.105535055350554e-05,
      "loss": 0.0014177164062857627,
      "step": 137260
    },
    {
      "epoch": 38.96395117797332,
      "grad_norm": 0.6566436290740967,
      "learning_rate": 6.105251206358218e-05,
      "loss": 0.004951765388250351,
      "step": 137270
    },
    {
      "epoch": 38.96678966789668,
      "grad_norm": 0.5991978645324707,
      "learning_rate": 6.104967357365882e-05,
      "loss": 0.0011709747835993768,
      "step": 137280
    },
    {
      "epoch": 38.96962815782004,
      "grad_norm": 0.634044885635376,
      "learning_rate": 6.104683508373545e-05,
      "loss": 0.003566627949476242,
      "step": 137290
    },
    {
      "epoch": 38.9724666477434,
      "grad_norm": 0.2746218144893646,
      "learning_rate": 6.104399659381209e-05,
      "loss": 0.0034033864736557006,
      "step": 137300
    },
    {
      "epoch": 38.97530513766676,
      "grad_norm": 2.897240161895752,
      "learning_rate": 6.104115810388873e-05,
      "loss": 0.005538882315158844,
      "step": 137310
    },
    {
      "epoch": 38.97814362759012,
      "grad_norm": 0.22855786979198456,
      "learning_rate": 6.103831961396537e-05,
      "loss": 0.004798336327075959,
      "step": 137320
    },
    {
      "epoch": 38.980982117513484,
      "grad_norm": 0.2276849001646042,
      "learning_rate": 6.103548112404202e-05,
      "loss": 0.008087386190891267,
      "step": 137330
    },
    {
      "epoch": 38.98382060743684,
      "grad_norm": 0.25852397084236145,
      "learning_rate": 6.103264263411865e-05,
      "loss": 0.0025078503414988516,
      "step": 137340
    },
    {
      "epoch": 38.9866590973602,
      "grad_norm": 0.5816524624824524,
      "learning_rate": 6.102980414419529e-05,
      "loss": 0.004313325881958008,
      "step": 137350
    },
    {
      "epoch": 38.989497587283566,
      "grad_norm": 0.26520130038261414,
      "learning_rate": 6.1026965654271926e-05,
      "loss": 0.0020789306610822676,
      "step": 137360
    },
    {
      "epoch": 38.99233607720693,
      "grad_norm": 0.4833289086818695,
      "learning_rate": 6.102412716434857e-05,
      "loss": 0.0011721242219209671,
      "step": 137370
    },
    {
      "epoch": 38.995174567130285,
      "grad_norm": 0.5899637341499329,
      "learning_rate": 6.10212886744252e-05,
      "loss": 0.008053042739629746,
      "step": 137380
    },
    {
      "epoch": 38.99801305705365,
      "grad_norm": 0.8573595881462097,
      "learning_rate": 6.101845018450185e-05,
      "loss": 0.0019636169075965882,
      "step": 137390
    },
    {
      "epoch": 39.00085154697701,
      "grad_norm": 0.006774546578526497,
      "learning_rate": 6.101561169457849e-05,
      "loss": 0.0015545954927802085,
      "step": 137400
    },
    {
      "epoch": 39.003690036900366,
      "grad_norm": 0.10945258289575577,
      "learning_rate": 6.1012773204655126e-05,
      "loss": 0.0005326934158802032,
      "step": 137410
    },
    {
      "epoch": 39.00652852682373,
      "grad_norm": 0.0803971067070961,
      "learning_rate": 6.100993471473177e-05,
      "loss": 0.0009184282273054123,
      "step": 137420
    },
    {
      "epoch": 39.00936701674709,
      "grad_norm": 0.11306066811084747,
      "learning_rate": 6.10070962248084e-05,
      "loss": 0.005082246288657189,
      "step": 137430
    },
    {
      "epoch": 39.012205506670455,
      "grad_norm": 0.00984280277043581,
      "learning_rate": 6.100425773488504e-05,
      "loss": 0.0009250149130821228,
      "step": 137440
    },
    {
      "epoch": 39.01504399659381,
      "grad_norm": 0.7278012633323669,
      "learning_rate": 6.100141924496169e-05,
      "loss": 0.00113204438239336,
      "step": 137450
    },
    {
      "epoch": 39.01788248651717,
      "grad_norm": 0.08461342751979828,
      "learning_rate": 6.0998580755038326e-05,
      "loss": 0.0007893219590187073,
      "step": 137460
    },
    {
      "epoch": 39.020720976440536,
      "grad_norm": 0.06453245133161545,
      "learning_rate": 6.099574226511496e-05,
      "loss": 0.0024091882631182672,
      "step": 137470
    },
    {
      "epoch": 39.02355946636389,
      "grad_norm": 0.1434095799922943,
      "learning_rate": 6.09929037751916e-05,
      "loss": 0.0023527691140770913,
      "step": 137480
    },
    {
      "epoch": 39.026397956287255,
      "grad_norm": 0.2553968131542206,
      "learning_rate": 6.099006528526824e-05,
      "loss": 0.0008763695135712623,
      "step": 137490
    },
    {
      "epoch": 39.02923644621062,
      "grad_norm": 2.6161484718322754,
      "learning_rate": 6.098722679534488e-05,
      "loss": 0.0016762204468250275,
      "step": 137500
    },
    {
      "epoch": 39.02923644621062,
      "eval_accuracy": 0.9767279201373434,
      "eval_loss": 0.08128369599580765,
      "eval_runtime": 31.5043,
      "eval_samples_per_second": 499.202,
      "eval_steps_per_second": 7.808,
      "step": 137500
    },
    {
      "epoch": 39.032074936133974,
      "grad_norm": 0.09355601668357849,
      "learning_rate": 6.098438830542151e-05,
      "loss": 0.0012256275862455369,
      "step": 137510
    },
    {
      "epoch": 39.03491342605734,
      "grad_norm": 0.1080615296959877,
      "learning_rate": 6.098154981549816e-05,
      "loss": 0.0006028557196259499,
      "step": 137520
    },
    {
      "epoch": 39.0377519159807,
      "grad_norm": 0.3344254791736603,
      "learning_rate": 6.09787113255748e-05,
      "loss": 0.0006280548870563508,
      "step": 137530
    },
    {
      "epoch": 39.04059040590406,
      "grad_norm": 0.5869518518447876,
      "learning_rate": 6.097587283565144e-05,
      "loss": 0.0009614804759621621,
      "step": 137540
    },
    {
      "epoch": 39.04342889582742,
      "grad_norm": 0.05939723923802376,
      "learning_rate": 6.097303434572808e-05,
      "loss": 0.003757853806018829,
      "step": 137550
    },
    {
      "epoch": 39.04626738575078,
      "grad_norm": 3.4026780128479004,
      "learning_rate": 6.097019585580471e-05,
      "loss": 0.0031615853309631347,
      "step": 137560
    },
    {
      "epoch": 39.049105875674144,
      "grad_norm": 0.3942703306674957,
      "learning_rate": 6.096735736588135e-05,
      "loss": 0.012047493457794189,
      "step": 137570
    },
    {
      "epoch": 39.0519443655975,
      "grad_norm": 17.280351638793945,
      "learning_rate": 6.096451887595799e-05,
      "loss": 0.017882511019706726,
      "step": 137580
    },
    {
      "epoch": 39.05478285552086,
      "grad_norm": 1.0733215808868408,
      "learning_rate": 6.096168038603464e-05,
      "loss": 0.005104930698871612,
      "step": 137590
    },
    {
      "epoch": 39.057621345444225,
      "grad_norm": 3.197697162628174,
      "learning_rate": 6.095884189611127e-05,
      "loss": 0.002362445928156376,
      "step": 137600
    },
    {
      "epoch": 39.06045983536758,
      "grad_norm": 0.17124387621879578,
      "learning_rate": 6.095600340618791e-05,
      "loss": 0.0020661937072873116,
      "step": 137610
    },
    {
      "epoch": 39.063298325290944,
      "grad_norm": 0.1346268355846405,
      "learning_rate": 6.095316491626455e-05,
      "loss": 0.00855952724814415,
      "step": 137620
    },
    {
      "epoch": 39.06613681521431,
      "grad_norm": 0.13660840690135956,
      "learning_rate": 6.095032642634119e-05,
      "loss": 0.005910919979214668,
      "step": 137630
    },
    {
      "epoch": 39.06897530513767,
      "grad_norm": 0.09878029674291611,
      "learning_rate": 6.0947487936417823e-05,
      "loss": 0.0011193074285984039,
      "step": 137640
    },
    {
      "epoch": 39.071813795061026,
      "grad_norm": 0.03971888870000839,
      "learning_rate": 6.094464944649447e-05,
      "loss": 0.0014145774766802788,
      "step": 137650
    },
    {
      "epoch": 39.07465228498439,
      "grad_norm": 0.3018629252910614,
      "learning_rate": 6.094181095657111e-05,
      "loss": 0.011739248037338256,
      "step": 137660
    },
    {
      "epoch": 39.07749077490775,
      "grad_norm": 5.928213119506836,
      "learning_rate": 6.093897246664775e-05,
      "loss": 0.006136106699705124,
      "step": 137670
    },
    {
      "epoch": 39.08032926483111,
      "grad_norm": 9.246639251708984,
      "learning_rate": 6.093613397672438e-05,
      "loss": 0.006184373050928116,
      "step": 137680
    },
    {
      "epoch": 39.08316775475447,
      "grad_norm": 0.455435574054718,
      "learning_rate": 6.0933295486801024e-05,
      "loss": 0.0007635755464434624,
      "step": 137690
    },
    {
      "epoch": 39.08600624467783,
      "grad_norm": 0.10587795078754425,
      "learning_rate": 6.093045699687766e-05,
      "loss": 0.00048276036977767944,
      "step": 137700
    },
    {
      "epoch": 39.08884473460119,
      "grad_norm": 0.03420078009366989,
      "learning_rate": 6.09276185069543e-05,
      "loss": 0.0029030552133917807,
      "step": 137710
    },
    {
      "epoch": 39.09168322452455,
      "grad_norm": 0.2092326134443283,
      "learning_rate": 6.092478001703095e-05,
      "loss": 0.00044975243508815765,
      "step": 137720
    },
    {
      "epoch": 39.094521714447914,
      "grad_norm": 1.6473523378372192,
      "learning_rate": 6.092194152710758e-05,
      "loss": 0.00887780711054802,
      "step": 137730
    },
    {
      "epoch": 39.09736020437128,
      "grad_norm": 0.061115700751543045,
      "learning_rate": 6.0919103037184224e-05,
      "loss": 0.0009986110031604766,
      "step": 137740
    },
    {
      "epoch": 39.10019869429463,
      "grad_norm": 0.20632874965667725,
      "learning_rate": 6.091626454726086e-05,
      "loss": 0.00251407865434885,
      "step": 137750
    },
    {
      "epoch": 39.103037184217996,
      "grad_norm": 0.4865431785583496,
      "learning_rate": 6.09134260573375e-05,
      "loss": 0.0032470479607582094,
      "step": 137760
    },
    {
      "epoch": 39.10587567414136,
      "grad_norm": 0.30130788683891296,
      "learning_rate": 6.0910587567414134e-05,
      "loss": 0.0026256956160068514,
      "step": 137770
    },
    {
      "epoch": 39.108714164064715,
      "grad_norm": 0.8512123227119446,
      "learning_rate": 6.090774907749078e-05,
      "loss": 0.00871809795498848,
      "step": 137780
    },
    {
      "epoch": 39.11155265398808,
      "grad_norm": 0.936905026435852,
      "learning_rate": 6.0904910587567424e-05,
      "loss": 0.006030123680830002,
      "step": 137790
    },
    {
      "epoch": 39.11439114391144,
      "grad_norm": 0.3430178761482239,
      "learning_rate": 6.090207209764406e-05,
      "loss": 0.0007317159324884415,
      "step": 137800
    },
    {
      "epoch": 39.1172296338348,
      "grad_norm": 1.301710844039917,
      "learning_rate": 6.089923360772069e-05,
      "loss": 0.0037686169147491456,
      "step": 137810
    },
    {
      "epoch": 39.12006812375816,
      "grad_norm": 0.8421335816383362,
      "learning_rate": 6.0896395117797334e-05,
      "loss": 0.00961814671754837,
      "step": 137820
    },
    {
      "epoch": 39.12290661368152,
      "grad_norm": 1.4414587020874023,
      "learning_rate": 6.089355662787397e-05,
      "loss": 0.0007542893290519714,
      "step": 137830
    },
    {
      "epoch": 39.125745103604885,
      "grad_norm": 0.08410868048667908,
      "learning_rate": 6.089071813795061e-05,
      "loss": 0.003299996629357338,
      "step": 137840
    },
    {
      "epoch": 39.12858359352824,
      "grad_norm": 0.07982895523309708,
      "learning_rate": 6.088787964802726e-05,
      "loss": 0.0015040313825011252,
      "step": 137850
    },
    {
      "epoch": 39.131422083451604,
      "grad_norm": 0.16509340703487396,
      "learning_rate": 6.088504115810389e-05,
      "loss": 0.0018917864188551902,
      "step": 137860
    },
    {
      "epoch": 39.134260573374966,
      "grad_norm": 0.09333012998104095,
      "learning_rate": 6.0882202668180534e-05,
      "loss": 0.005116412043571472,
      "step": 137870
    },
    {
      "epoch": 39.13709906329832,
      "grad_norm": 0.16603335738182068,
      "learning_rate": 6.087936417825717e-05,
      "loss": 0.001081143505871296,
      "step": 137880
    },
    {
      "epoch": 39.139937553221685,
      "grad_norm": 2.381290912628174,
      "learning_rate": 6.087652568833381e-05,
      "loss": 0.0008555615320801734,
      "step": 137890
    },
    {
      "epoch": 39.14277604314505,
      "grad_norm": 0.10461077839136124,
      "learning_rate": 6.0873687198410445e-05,
      "loss": 0.0013933278620243072,
      "step": 137900
    },
    {
      "epoch": 39.14561453306841,
      "grad_norm": 0.012128238566219807,
      "learning_rate": 6.087084870848708e-05,
      "loss": 0.0016804268583655358,
      "step": 137910
    },
    {
      "epoch": 39.14845302299177,
      "grad_norm": 0.08353256434202194,
      "learning_rate": 6.0868010218563735e-05,
      "loss": 0.0022597117349505424,
      "step": 137920
    },
    {
      "epoch": 39.15129151291513,
      "grad_norm": 0.02323201112449169,
      "learning_rate": 6.086517172864037e-05,
      "loss": 0.00500510036945343,
      "step": 137930
    },
    {
      "epoch": 39.15413000283849,
      "grad_norm": 0.8645755648612976,
      "learning_rate": 6.0862333238717004e-05,
      "loss": 0.0008523259311914444,
      "step": 137940
    },
    {
      "epoch": 39.15696849276185,
      "grad_norm": 10.219121932983398,
      "learning_rate": 6.0859494748793645e-05,
      "loss": 0.004079770669341088,
      "step": 137950
    },
    {
      "epoch": 39.15980698268521,
      "grad_norm": 0.06384527683258057,
      "learning_rate": 6.085665625887028e-05,
      "loss": 0.0008188333362340927,
      "step": 137960
    },
    {
      "epoch": 39.162645472608574,
      "grad_norm": 0.04710692539811134,
      "learning_rate": 6.085381776894692e-05,
      "loss": 0.0014145579189062119,
      "step": 137970
    },
    {
      "epoch": 39.16548396253193,
      "grad_norm": 0.21720804274082184,
      "learning_rate": 6.085097927902357e-05,
      "loss": 0.000653439573943615,
      "step": 137980
    },
    {
      "epoch": 39.16832245245529,
      "grad_norm": 0.3116194009780884,
      "learning_rate": 6.0848140789100204e-05,
      "loss": 0.0036538906395435335,
      "step": 137990
    },
    {
      "epoch": 39.171160942378656,
      "grad_norm": 0.5946470499038696,
      "learning_rate": 6.0845302299176845e-05,
      "loss": 0.0011351684108376503,
      "step": 138000
    },
    {
      "epoch": 39.171160942378656,
      "eval_accuracy": 0.974883957525275,
      "eval_loss": 0.0871768519282341,
      "eval_runtime": 31.6531,
      "eval_samples_per_second": 496.855,
      "eval_steps_per_second": 7.772,
      "step": 138000
    },
    {
      "epoch": 39.17399943230202,
      "grad_norm": 0.023812010884284973,
      "learning_rate": 6.084246380925348e-05,
      "loss": 0.001424885354936123,
      "step": 138010
    },
    {
      "epoch": 39.176837922225374,
      "grad_norm": 1.167183756828308,
      "learning_rate": 6.083962531933012e-05,
      "loss": 0.0016846036538481711,
      "step": 138020
    },
    {
      "epoch": 39.17967641214874,
      "grad_norm": 0.06904592365026474,
      "learning_rate": 6.0836786829406756e-05,
      "loss": 0.0014289602637290954,
      "step": 138030
    },
    {
      "epoch": 39.1825149020721,
      "grad_norm": 0.4332338571548462,
      "learning_rate": 6.083394833948339e-05,
      "loss": 0.0045259203761816025,
      "step": 138040
    },
    {
      "epoch": 39.185353391995456,
      "grad_norm": 0.11826693266630173,
      "learning_rate": 6.083110984956004e-05,
      "loss": 0.000598406046628952,
      "step": 138050
    },
    {
      "epoch": 39.18819188191882,
      "grad_norm": 0.16060800850391388,
      "learning_rate": 6.082827135963668e-05,
      "loss": 0.0015125835314393043,
      "step": 138060
    },
    {
      "epoch": 39.19103037184218,
      "grad_norm": 0.36900851130485535,
      "learning_rate": 6.0825432869713315e-05,
      "loss": 0.001500137336552143,
      "step": 138070
    },
    {
      "epoch": 39.19386886176554,
      "grad_norm": 0.4478924572467804,
      "learning_rate": 6.0822594379789956e-05,
      "loss": 0.004307914525270462,
      "step": 138080
    },
    {
      "epoch": 39.1967073516889,
      "grad_norm": 0.10561967641115189,
      "learning_rate": 6.081975588986659e-05,
      "loss": 0.0012365575879812241,
      "step": 138090
    },
    {
      "epoch": 39.19954584161226,
      "grad_norm": 0.42396971583366394,
      "learning_rate": 6.081691739994323e-05,
      "loss": 0.0037472926080226896,
      "step": 138100
    },
    {
      "epoch": 39.202384331535626,
      "grad_norm": 3.2284858226776123,
      "learning_rate": 6.0814078910019866e-05,
      "loss": 0.004078273475170135,
      "step": 138110
    },
    {
      "epoch": 39.20522282145898,
      "grad_norm": 3.336428642272949,
      "learning_rate": 6.0811240420096515e-05,
      "loss": 0.0034972336143255234,
      "step": 138120
    },
    {
      "epoch": 39.208061311382345,
      "grad_norm": 0.5801212191581726,
      "learning_rate": 6.0808401930173156e-05,
      "loss": 0.003932987153530121,
      "step": 138130
    },
    {
      "epoch": 39.21089980130571,
      "grad_norm": 1.5830198526382446,
      "learning_rate": 6.080556344024979e-05,
      "loss": 0.005855109542608261,
      "step": 138140
    },
    {
      "epoch": 39.21373829122906,
      "grad_norm": 0.1764943152666092,
      "learning_rate": 6.0802724950326425e-05,
      "loss": 0.007715848088264465,
      "step": 138150
    },
    {
      "epoch": 39.216576781152426,
      "grad_norm": 3.043212890625,
      "learning_rate": 6.0799886460403067e-05,
      "loss": 0.0018861185759305953,
      "step": 138160
    },
    {
      "epoch": 39.21941527107579,
      "grad_norm": 16.60807991027832,
      "learning_rate": 6.07970479704797e-05,
      "loss": 0.012019102275371552,
      "step": 138170
    },
    {
      "epoch": 39.22225376099915,
      "grad_norm": 0.0162091925740242,
      "learning_rate": 6.079420948055635e-05,
      "loss": 0.0028994901105761526,
      "step": 138180
    },
    {
      "epoch": 39.22509225092251,
      "grad_norm": 2.6232378482818604,
      "learning_rate": 6.079137099063299e-05,
      "loss": 0.00649256631731987,
      "step": 138190
    },
    {
      "epoch": 39.22793074084587,
      "grad_norm": 0.5794569253921509,
      "learning_rate": 6.0788532500709625e-05,
      "loss": 0.0012230440974235536,
      "step": 138200
    },
    {
      "epoch": 39.23076923076923,
      "grad_norm": 0.153490349650383,
      "learning_rate": 6.078569401078627e-05,
      "loss": 0.0009051540866494179,
      "step": 138210
    },
    {
      "epoch": 39.23360772069259,
      "grad_norm": 1.236511468887329,
      "learning_rate": 6.07828555208629e-05,
      "loss": 0.002259830757975578,
      "step": 138220
    },
    {
      "epoch": 39.23644621061595,
      "grad_norm": 0.0276777483522892,
      "learning_rate": 6.078001703093954e-05,
      "loss": 0.00093733761459589,
      "step": 138230
    },
    {
      "epoch": 39.239284700539315,
      "grad_norm": 0.31755301356315613,
      "learning_rate": 6.077717854101618e-05,
      "loss": 0.0020687647163867952,
      "step": 138240
    },
    {
      "epoch": 39.24212319046267,
      "grad_norm": 0.5143074989318848,
      "learning_rate": 6.0774340051092825e-05,
      "loss": 0.003288868069648743,
      "step": 138250
    },
    {
      "epoch": 39.244961680386034,
      "grad_norm": 0.6853947639465332,
      "learning_rate": 6.077150156116947e-05,
      "loss": 0.0055705726146698,
      "step": 138260
    },
    {
      "epoch": 39.2478001703094,
      "grad_norm": 0.015403758734464645,
      "learning_rate": 6.07686630712461e-05,
      "loss": 0.0014558464288711547,
      "step": 138270
    },
    {
      "epoch": 39.25063866023276,
      "grad_norm": 7.742937088012695,
      "learning_rate": 6.0765824581322736e-05,
      "loss": 0.010842492431402206,
      "step": 138280
    },
    {
      "epoch": 39.253477150156115,
      "grad_norm": 0.04685800150036812,
      "learning_rate": 6.076298609139938e-05,
      "loss": 0.0014076102524995804,
      "step": 138290
    },
    {
      "epoch": 39.25631564007948,
      "grad_norm": 3.1685903072357178,
      "learning_rate": 6.076014760147601e-05,
      "loss": 0.004139552265405655,
      "step": 138300
    },
    {
      "epoch": 39.25915413000284,
      "grad_norm": 0.022206230089068413,
      "learning_rate": 6.075730911155265e-05,
      "loss": 0.0065355576574802395,
      "step": 138310
    },
    {
      "epoch": 39.2619926199262,
      "grad_norm": 20.028385162353516,
      "learning_rate": 6.07544706216293e-05,
      "loss": 0.014773531258106232,
      "step": 138320
    },
    {
      "epoch": 39.26483110984956,
      "grad_norm": 0.039726655930280685,
      "learning_rate": 6.0751632131705936e-05,
      "loss": 0.0036755047738552095,
      "step": 138330
    },
    {
      "epoch": 39.26766959977292,
      "grad_norm": 1.9771883487701416,
      "learning_rate": 6.074879364178258e-05,
      "loss": 0.0033871538937091826,
      "step": 138340
    },
    {
      "epoch": 39.27050808969628,
      "grad_norm": 0.07113926857709885,
      "learning_rate": 6.074595515185921e-05,
      "loss": 0.002391861379146576,
      "step": 138350
    },
    {
      "epoch": 39.27334657961964,
      "grad_norm": 17.838214874267578,
      "learning_rate": 6.0743116661935853e-05,
      "loss": 0.01301446110010147,
      "step": 138360
    },
    {
      "epoch": 39.276185069543004,
      "grad_norm": 1.5915685892105103,
      "learning_rate": 6.074027817201249e-05,
      "loss": 0.0026036594063043594,
      "step": 138370
    },
    {
      "epoch": 39.27902355946637,
      "grad_norm": 0.4081057608127594,
      "learning_rate": 6.0737439682089136e-05,
      "loss": 0.0028338827192783355,
      "step": 138380
    },
    {
      "epoch": 39.28186204938972,
      "grad_norm": 2.8502750396728516,
      "learning_rate": 6.073460119216578e-05,
      "loss": 0.005361150950193405,
      "step": 138390
    },
    {
      "epoch": 39.284700539313086,
      "grad_norm": 0.7020703554153442,
      "learning_rate": 6.073176270224241e-05,
      "loss": 0.0009952127933502198,
      "step": 138400
    },
    {
      "epoch": 39.28753902923645,
      "grad_norm": 0.11265525966882706,
      "learning_rate": 6.072892421231905e-05,
      "loss": 0.0034624986350536347,
      "step": 138410
    },
    {
      "epoch": 39.290377519159804,
      "grad_norm": 0.9328380823135376,
      "learning_rate": 6.072608572239569e-05,
      "loss": 0.005276872217655182,
      "step": 138420
    },
    {
      "epoch": 39.29321600908317,
      "grad_norm": 0.03923322632908821,
      "learning_rate": 6.072324723247232e-05,
      "loss": 0.0022132568061351775,
      "step": 138430
    },
    {
      "epoch": 39.29605449900653,
      "grad_norm": 0.11271873861551285,
      "learning_rate": 6.0720408742548964e-05,
      "loss": 0.0007368421182036399,
      "step": 138440
    },
    {
      "epoch": 39.298892988929886,
      "grad_norm": 0.5709341168403625,
      "learning_rate": 6.0717854101617944e-05,
      "loss": 0.006585609912872314,
      "step": 138450
    },
    {
      "epoch": 39.30173147885325,
      "grad_norm": 1.7319859266281128,
      "learning_rate": 6.071501561169458e-05,
      "loss": 0.0020401835441589355,
      "step": 138460
    },
    {
      "epoch": 39.30456996877661,
      "grad_norm": 0.35285666584968567,
      "learning_rate": 6.071217712177122e-05,
      "loss": 0.0028673205524683,
      "step": 138470
    },
    {
      "epoch": 39.307408458699975,
      "grad_norm": 0.11032412946224213,
      "learning_rate": 6.0709338631847855e-05,
      "loss": 0.005562671273946762,
      "step": 138480
    },
    {
      "epoch": 39.31024694862333,
      "grad_norm": 0.1384722739458084,
      "learning_rate": 6.0706500141924496e-05,
      "loss": 0.011051515489816666,
      "step": 138490
    },
    {
      "epoch": 39.31308543854669,
      "grad_norm": 0.12851504981517792,
      "learning_rate": 6.0703661652001144e-05,
      "loss": 0.0017360903322696686,
      "step": 138500
    },
    {
      "epoch": 39.31308543854669,
      "eval_accuracy": 0.9756469765371654,
      "eval_loss": 0.08536995202302933,
      "eval_runtime": 32.0601,
      "eval_samples_per_second": 490.548,
      "eval_steps_per_second": 7.673,
      "step": 138500
    },
    {
      "epoch": 39.315923928470056,
      "grad_norm": 0.20889928936958313,
      "learning_rate": 6.070082316207778e-05,
      "loss": 0.0011299841105937957,
      "step": 138510
    },
    {
      "epoch": 39.31876241839341,
      "grad_norm": 0.18569187819957733,
      "learning_rate": 6.069798467215442e-05,
      "loss": 0.0051317159086465836,
      "step": 138520
    },
    {
      "epoch": 39.321600908316775,
      "grad_norm": 0.5365309715270996,
      "learning_rate": 6.0695146182231055e-05,
      "loss": 0.0025659797713160513,
      "step": 138530
    },
    {
      "epoch": 39.32443939824014,
      "grad_norm": 0.15547697246074677,
      "learning_rate": 6.0692307692307696e-05,
      "loss": 0.003668425977230072,
      "step": 138540
    },
    {
      "epoch": 39.32727788816349,
      "grad_norm": 0.056269995868206024,
      "learning_rate": 6.068946920238433e-05,
      "loss": 0.0006794262677431107,
      "step": 138550
    },
    {
      "epoch": 39.330116378086856,
      "grad_norm": 0.4943416714668274,
      "learning_rate": 6.068663071246098e-05,
      "loss": 0.0022031737491488455,
      "step": 138560
    },
    {
      "epoch": 39.33295486801022,
      "grad_norm": 0.16368018090724945,
      "learning_rate": 6.0683792222537613e-05,
      "loss": 0.00043320301920175553,
      "step": 138570
    },
    {
      "epoch": 39.33579335793358,
      "grad_norm": 0.06819257140159607,
      "learning_rate": 6.0680953732614255e-05,
      "loss": 0.0007342305034399033,
      "step": 138580
    },
    {
      "epoch": 39.33863184785694,
      "grad_norm": 5.623698711395264,
      "learning_rate": 6.067811524269089e-05,
      "loss": 0.008658371865749359,
      "step": 138590
    },
    {
      "epoch": 39.3414703377803,
      "grad_norm": 0.03333594277501106,
      "learning_rate": 6.067527675276753e-05,
      "loss": 0.006858749687671662,
      "step": 138600
    },
    {
      "epoch": 39.344308827703664,
      "grad_norm": 0.1830771118402481,
      "learning_rate": 6.0672438262844165e-05,
      "loss": 0.0006727039813995362,
      "step": 138610
    },
    {
      "epoch": 39.34714731762702,
      "grad_norm": 0.06880036741495132,
      "learning_rate": 6.066959977292081e-05,
      "loss": 0.0008875031024217605,
      "step": 138620
    },
    {
      "epoch": 39.34998580755038,
      "grad_norm": 0.935390055179596,
      "learning_rate": 6.0666761282997455e-05,
      "loss": 0.004733657836914063,
      "step": 138630
    },
    {
      "epoch": 39.352824297473745,
      "grad_norm": 0.6304066181182861,
      "learning_rate": 6.066392279307409e-05,
      "loss": 0.0005578668788075447,
      "step": 138640
    },
    {
      "epoch": 39.35566278739711,
      "grad_norm": 0.27680617570877075,
      "learning_rate": 6.066108430315073e-05,
      "loss": 0.0006093733012676239,
      "step": 138650
    },
    {
      "epoch": 39.358501277320464,
      "grad_norm": 0.03157861903309822,
      "learning_rate": 6.0658245813227365e-05,
      "loss": 0.0005899008363485336,
      "step": 138660
    },
    {
      "epoch": 39.36133976724383,
      "grad_norm": 1.552221417427063,
      "learning_rate": 6.0655407323304e-05,
      "loss": 0.0014631872996687888,
      "step": 138670
    },
    {
      "epoch": 39.36417825716719,
      "grad_norm": 0.7908819913864136,
      "learning_rate": 6.065256883338064e-05,
      "loss": 0.0007385041564702988,
      "step": 138680
    },
    {
      "epoch": 39.367016747090545,
      "grad_norm": 4.120865345001221,
      "learning_rate": 6.0649730343457276e-05,
      "loss": 0.0030291903764009475,
      "step": 138690
    },
    {
      "epoch": 39.36985523701391,
      "grad_norm": 0.03769755735993385,
      "learning_rate": 6.0646891853533924e-05,
      "loss": 0.011622709035873414,
      "step": 138700
    },
    {
      "epoch": 39.37269372693727,
      "grad_norm": 0.10677117854356766,
      "learning_rate": 6.0644053363610566e-05,
      "loss": 0.004477967321872711,
      "step": 138710
    },
    {
      "epoch": 39.37553221686063,
      "grad_norm": 0.08403872698545456,
      "learning_rate": 6.06412148736872e-05,
      "loss": 0.0025386758148670196,
      "step": 138720
    },
    {
      "epoch": 39.37837070678399,
      "grad_norm": 1.9870363473892212,
      "learning_rate": 6.063837638376384e-05,
      "loss": 0.003840316832065582,
      "step": 138730
    },
    {
      "epoch": 39.38120919670735,
      "grad_norm": 0.14122192561626434,
      "learning_rate": 6.0635537893840476e-05,
      "loss": 0.0011181939393281938,
      "step": 138740
    },
    {
      "epoch": 39.384047686630716,
      "grad_norm": 0.12941376864910126,
      "learning_rate": 6.063269940391712e-05,
      "loss": 0.0021334828808903692,
      "step": 138750
    },
    {
      "epoch": 39.38688617655407,
      "grad_norm": 0.3391941785812378,
      "learning_rate": 6.0629860913993766e-05,
      "loss": 0.0004289405420422554,
      "step": 138760
    },
    {
      "epoch": 39.389724666477434,
      "grad_norm": 0.03545178845524788,
      "learning_rate": 6.06270224240704e-05,
      "loss": 0.002675817534327507,
      "step": 138770
    },
    {
      "epoch": 39.3925631564008,
      "grad_norm": 0.330510675907135,
      "learning_rate": 6.062418393414704e-05,
      "loss": 0.01122249811887741,
      "step": 138780
    },
    {
      "epoch": 39.39540164632415,
      "grad_norm": 0.6019308567047119,
      "learning_rate": 6.0621345444223676e-05,
      "loss": 0.0034800685942173006,
      "step": 138790
    },
    {
      "epoch": 39.398240136247516,
      "grad_norm": 0.13632014393806458,
      "learning_rate": 6.061850695430031e-05,
      "loss": 0.0010219585150480271,
      "step": 138800
    },
    {
      "epoch": 39.40107862617088,
      "grad_norm": 0.44659972190856934,
      "learning_rate": 6.061566846437695e-05,
      "loss": 0.001603710651397705,
      "step": 138810
    },
    {
      "epoch": 39.403917116094235,
      "grad_norm": 0.0122533543035388,
      "learning_rate": 6.061282997445359e-05,
      "loss": 0.001211416907608509,
      "step": 138820
    },
    {
      "epoch": 39.4067556060176,
      "grad_norm": 0.7950981259346008,
      "learning_rate": 6.0609991484530235e-05,
      "loss": 0.008475624769926072,
      "step": 138830
    },
    {
      "epoch": 39.40959409594096,
      "grad_norm": 0.06071178615093231,
      "learning_rate": 6.0607152994606876e-05,
      "loss": 0.0018236760050058364,
      "step": 138840
    },
    {
      "epoch": 39.41243258586432,
      "grad_norm": 0.1182396411895752,
      "learning_rate": 6.060431450468351e-05,
      "loss": 0.0005002036690711975,
      "step": 138850
    },
    {
      "epoch": 39.41527107578768,
      "grad_norm": 0.05120754987001419,
      "learning_rate": 6.060147601476015e-05,
      "loss": 0.0015990445390343666,
      "step": 138860
    },
    {
      "epoch": 39.41810956571104,
      "grad_norm": 1.545422077178955,
      "learning_rate": 6.059863752483679e-05,
      "loss": 0.007767621427774429,
      "step": 138870
    },
    {
      "epoch": 39.420948055634405,
      "grad_norm": 5.893726348876953,
      "learning_rate": 6.059579903491343e-05,
      "loss": 0.012136691808700561,
      "step": 138880
    },
    {
      "epoch": 39.42378654555776,
      "grad_norm": 5.254700183868408,
      "learning_rate": 6.059296054499006e-05,
      "loss": 0.004616284742951393,
      "step": 138890
    },
    {
      "epoch": 39.42662503548112,
      "grad_norm": 5.23870325088501,
      "learning_rate": 6.059012205506671e-05,
      "loss": 0.005939216911792755,
      "step": 138900
    },
    {
      "epoch": 39.429463525404486,
      "grad_norm": 1.1079213619232178,
      "learning_rate": 6.058728356514335e-05,
      "loss": 0.0015039023011922837,
      "step": 138910
    },
    {
      "epoch": 39.43230201532784,
      "grad_norm": 0.04870188236236572,
      "learning_rate": 6.058444507521999e-05,
      "loss": 0.0006049912422895432,
      "step": 138920
    },
    {
      "epoch": 39.435140505251205,
      "grad_norm": 0.37689369916915894,
      "learning_rate": 6.058160658529662e-05,
      "loss": 0.003204350918531418,
      "step": 138930
    },
    {
      "epoch": 39.43797899517457,
      "grad_norm": 13.268476486206055,
      "learning_rate": 6.057876809537326e-05,
      "loss": 0.0030410535633563994,
      "step": 138940
    },
    {
      "epoch": 39.44081748509793,
      "grad_norm": 3.697195053100586,
      "learning_rate": 6.05759296054499e-05,
      "loss": 0.0026168907061219215,
      "step": 138950
    },
    {
      "epoch": 39.44365597502129,
      "grad_norm": 1.7088240385055542,
      "learning_rate": 6.0573091115526546e-05,
      "loss": 0.003200293704867363,
      "step": 138960
    },
    {
      "epoch": 39.44649446494465,
      "grad_norm": 0.10265671461820602,
      "learning_rate": 6.057025262560319e-05,
      "loss": 0.010031526535749435,
      "step": 138970
    },
    {
      "epoch": 39.44933295486801,
      "grad_norm": 1.5349925756454468,
      "learning_rate": 6.056741413567982e-05,
      "loss": 0.004440639168024063,
      "step": 138980
    },
    {
      "epoch": 39.45217144479137,
      "grad_norm": 0.041087791323661804,
      "learning_rate": 6.056457564575646e-05,
      "loss": 0.013201980292797089,
      "step": 138990
    },
    {
      "epoch": 39.45500993471473,
      "grad_norm": 0.24191564321517944,
      "learning_rate": 6.05617371558331e-05,
      "loss": 0.0029610762372612952,
      "step": 139000
    },
    {
      "epoch": 39.45500993471473,
      "eval_accuracy": 0.9732307496661792,
      "eval_loss": 0.09388997405767441,
      "eval_runtime": 32.1935,
      "eval_samples_per_second": 488.515,
      "eval_steps_per_second": 7.641,
      "step": 139000
    },
    {
      "epoch": 39.457848424638094,
      "grad_norm": 0.12238845974206924,
      "learning_rate": 6.055889866590974e-05,
      "loss": 0.004166174679994583,
      "step": 139010
    },
    {
      "epoch": 39.46068691456146,
      "grad_norm": 10.061746597290039,
      "learning_rate": 6.0556060175986374e-05,
      "loss": 0.019758693873882294,
      "step": 139020
    },
    {
      "epoch": 39.46352540448481,
      "grad_norm": 0.1553926318883896,
      "learning_rate": 6.055322168606302e-05,
      "loss": 0.0042705446481704715,
      "step": 139030
    },
    {
      "epoch": 39.466363894408175,
      "grad_norm": 17.36062240600586,
      "learning_rate": 6.0550383196139656e-05,
      "loss": 0.009921664744615555,
      "step": 139040
    },
    {
      "epoch": 39.46920238433154,
      "grad_norm": 0.49641841650009155,
      "learning_rate": 6.05475447062163e-05,
      "loss": 0.0020416708663105964,
      "step": 139050
    },
    {
      "epoch": 39.472040874254894,
      "grad_norm": 0.168116495013237,
      "learning_rate": 6.054470621629293e-05,
      "loss": 0.003057621419429779,
      "step": 139060
    },
    {
      "epoch": 39.47487936417826,
      "grad_norm": 0.07465113699436188,
      "learning_rate": 6.0541867726369574e-05,
      "loss": 0.007385559380054474,
      "step": 139070
    },
    {
      "epoch": 39.47771785410162,
      "grad_norm": 0.6764495968818665,
      "learning_rate": 6.053902923644621e-05,
      "loss": 0.0035275742411613464,
      "step": 139080
    },
    {
      "epoch": 39.480556344024976,
      "grad_norm": 4.340326309204102,
      "learning_rate": 6.053619074652285e-05,
      "loss": 0.008909739553928375,
      "step": 139090
    },
    {
      "epoch": 39.48339483394834,
      "grad_norm": 0.1992167830467224,
      "learning_rate": 6.05333522565995e-05,
      "loss": 0.008172467350959778,
      "step": 139100
    },
    {
      "epoch": 39.4862333238717,
      "grad_norm": 0.3296763300895691,
      "learning_rate": 6.053051376667613e-05,
      "loss": 0.005192354321479797,
      "step": 139110
    },
    {
      "epoch": 39.489071813795064,
      "grad_norm": 10.435248374938965,
      "learning_rate": 6.0527675276752774e-05,
      "loss": 0.018149109184741975,
      "step": 139120
    },
    {
      "epoch": 39.49191030371842,
      "grad_norm": 0.2836419343948364,
      "learning_rate": 6.052483678682941e-05,
      "loss": 0.00365847572684288,
      "step": 139130
    },
    {
      "epoch": 39.49474879364178,
      "grad_norm": 0.15508265793323517,
      "learning_rate": 6.052199829690604e-05,
      "loss": 0.006747350096702576,
      "step": 139140
    },
    {
      "epoch": 39.497587283565146,
      "grad_norm": 2.163086175918579,
      "learning_rate": 6.0519159806982684e-05,
      "loss": 0.0023795176297426225,
      "step": 139150
    },
    {
      "epoch": 39.5004257734885,
      "grad_norm": 1.1789687871932983,
      "learning_rate": 6.051632131705933e-05,
      "loss": 0.007911676168441772,
      "step": 139160
    },
    {
      "epoch": 39.503264263411864,
      "grad_norm": 0.538994312286377,
      "learning_rate": 6.051348282713597e-05,
      "loss": 0.0031716935336589813,
      "step": 139170
    },
    {
      "epoch": 39.50610275333523,
      "grad_norm": 11.735756874084473,
      "learning_rate": 6.051064433721261e-05,
      "loss": 0.012116793543100357,
      "step": 139180
    },
    {
      "epoch": 39.50894124325858,
      "grad_norm": 0.1775570958852768,
      "learning_rate": 6.050780584728924e-05,
      "loss": 0.0034813590347766874,
      "step": 139190
    },
    {
      "epoch": 39.511779733181946,
      "grad_norm": 2.986088275909424,
      "learning_rate": 6.0504967357365884e-05,
      "loss": 0.0022414546459913255,
      "step": 139200
    },
    {
      "epoch": 39.51461822310531,
      "grad_norm": 19.068557739257812,
      "learning_rate": 6.050212886744252e-05,
      "loss": 0.01525673121213913,
      "step": 139210
    },
    {
      "epoch": 39.51745671302867,
      "grad_norm": 0.4115059971809387,
      "learning_rate": 6.049929037751916e-05,
      "loss": 0.0007110729813575745,
      "step": 139220
    },
    {
      "epoch": 39.52029520295203,
      "grad_norm": 0.10299807041883469,
      "learning_rate": 6.049645188759581e-05,
      "loss": 0.0017435448244214058,
      "step": 139230
    },
    {
      "epoch": 39.52313369287539,
      "grad_norm": 0.06878902018070221,
      "learning_rate": 6.049361339767244e-05,
      "loss": 0.005529165267944336,
      "step": 139240
    },
    {
      "epoch": 39.52597218279875,
      "grad_norm": 0.6363896727561951,
      "learning_rate": 6.0490774907749085e-05,
      "loss": 0.0014021214097738265,
      "step": 139250
    },
    {
      "epoch": 39.52881067272211,
      "grad_norm": 2.591054677963257,
      "learning_rate": 6.048793641782572e-05,
      "loss": 0.0031134381890296934,
      "step": 139260
    },
    {
      "epoch": 39.53164916264547,
      "grad_norm": 0.11640580743551254,
      "learning_rate": 6.0485097927902354e-05,
      "loss": 0.0015712512657046318,
      "step": 139270
    },
    {
      "epoch": 39.534487652568835,
      "grad_norm": 0.12123462557792664,
      "learning_rate": 6.0482259437978995e-05,
      "loss": 0.009001583606004716,
      "step": 139280
    },
    {
      "epoch": 39.53732614249219,
      "grad_norm": 14.135578155517578,
      "learning_rate": 6.047942094805564e-05,
      "loss": 0.005289077013731003,
      "step": 139290
    },
    {
      "epoch": 39.540164632415554,
      "grad_norm": 10.1737060546875,
      "learning_rate": 6.047658245813228e-05,
      "loss": 0.0061090394854545595,
      "step": 139300
    },
    {
      "epoch": 39.54300312233892,
      "grad_norm": 0.24329543113708496,
      "learning_rate": 6.047374396820892e-05,
      "loss": 0.001027282141149044,
      "step": 139310
    },
    {
      "epoch": 39.54584161226228,
      "grad_norm": 0.15187223255634308,
      "learning_rate": 6.0470905478285554e-05,
      "loss": 0.0018104445189237595,
      "step": 139320
    },
    {
      "epoch": 39.548680102185635,
      "grad_norm": 0.12122810631990433,
      "learning_rate": 6.0468066988362195e-05,
      "loss": 0.0032040536403656005,
      "step": 139330
    },
    {
      "epoch": 39.551518592109,
      "grad_norm": 0.03264455869793892,
      "learning_rate": 6.046522849843883e-05,
      "loss": 0.0015556132420897484,
      "step": 139340
    },
    {
      "epoch": 39.55435708203236,
      "grad_norm": 1.6046301126480103,
      "learning_rate": 6.046239000851547e-05,
      "loss": 0.001749090850353241,
      "step": 139350
    },
    {
      "epoch": 39.55719557195572,
      "grad_norm": 0.2947319447994232,
      "learning_rate": 6.045955151859212e-05,
      "loss": 0.0008201774209737778,
      "step": 139360
    },
    {
      "epoch": 39.56003406187908,
      "grad_norm": 1.223129391670227,
      "learning_rate": 6.0456713028668754e-05,
      "loss": 0.0019808651879429815,
      "step": 139370
    },
    {
      "epoch": 39.56287255180244,
      "grad_norm": 0.9171944260597229,
      "learning_rate": 6.0453874538745395e-05,
      "loss": 0.0030045343562960626,
      "step": 139380
    },
    {
      "epoch": 39.565711041725805,
      "grad_norm": 0.19484840333461761,
      "learning_rate": 6.045103604882203e-05,
      "loss": 0.001980600319802761,
      "step": 139390
    },
    {
      "epoch": 39.56854953164916,
      "grad_norm": 7.515980243682861,
      "learning_rate": 6.0448197558898665e-05,
      "loss": 0.0042570605874061584,
      "step": 139400
    },
    {
      "epoch": 39.571388021572524,
      "grad_norm": 0.40401598811149597,
      "learning_rate": 6.0445359068975306e-05,
      "loss": 0.00739060714840889,
      "step": 139410
    },
    {
      "epoch": 39.57422651149589,
      "grad_norm": 2.4137284755706787,
      "learning_rate": 6.044252057905194e-05,
      "loss": 0.0021855276077985763,
      "step": 139420
    },
    {
      "epoch": 39.57706500141924,
      "grad_norm": 0.06560931354761124,
      "learning_rate": 6.043968208912859e-05,
      "loss": 0.00098164863884449,
      "step": 139430
    },
    {
      "epoch": 39.579903491342606,
      "grad_norm": 4.309883117675781,
      "learning_rate": 6.043684359920523e-05,
      "loss": 0.00350627563893795,
      "step": 139440
    },
    {
      "epoch": 39.58274198126597,
      "grad_norm": 0.059342317283153534,
      "learning_rate": 6.0434005109281865e-05,
      "loss": 0.00507737547159195,
      "step": 139450
    },
    {
      "epoch": 39.585580471189324,
      "grad_norm": 3.8229546546936035,
      "learning_rate": 6.0431166619358506e-05,
      "loss": 0.0023159828037023545,
      "step": 139460
    },
    {
      "epoch": 39.58841896111269,
      "grad_norm": 0.06483537703752518,
      "learning_rate": 6.042832812943514e-05,
      "loss": 0.00431598462164402,
      "step": 139470
    },
    {
      "epoch": 39.59125745103605,
      "grad_norm": 0.057901959866285324,
      "learning_rate": 6.042548963951178e-05,
      "loss": 0.01255812644958496,
      "step": 139480
    },
    {
      "epoch": 39.59409594095941,
      "grad_norm": 8.8146333694458,
      "learning_rate": 6.042265114958843e-05,
      "loss": 0.007699554413557052,
      "step": 139490
    },
    {
      "epoch": 39.59693443088277,
      "grad_norm": 0.7698335647583008,
      "learning_rate": 6.0419812659665065e-05,
      "loss": 0.0015274036675691604,
      "step": 139500
    },
    {
      "epoch": 39.59693443088277,
      "eval_accuracy": 0.9713867870541107,
      "eval_loss": 0.10272873193025589,
      "eval_runtime": 32.0554,
      "eval_samples_per_second": 490.619,
      "eval_steps_per_second": 7.674,
      "step": 139500
    },
    {
      "epoch": 39.59977292080613,
      "grad_norm": 3.913221836090088,
      "learning_rate": 6.04169741697417e-05,
      "loss": 0.0021123886108398438,
      "step": 139510
    },
    {
      "epoch": 39.602611410729494,
      "grad_norm": 0.12325819581747055,
      "learning_rate": 6.041413567981834e-05,
      "loss": 0.005768696963787079,
      "step": 139520
    },
    {
      "epoch": 39.60544990065285,
      "grad_norm": 0.8495907187461853,
      "learning_rate": 6.0411297189894975e-05,
      "loss": 0.0016661401838064193,
      "step": 139530
    },
    {
      "epoch": 39.60828839057621,
      "grad_norm": 0.13071849942207336,
      "learning_rate": 6.040845869997162e-05,
      "loss": 0.0024972803890705108,
      "step": 139540
    },
    {
      "epoch": 39.611126880499576,
      "grad_norm": 0.41720494627952576,
      "learning_rate": 6.040562021004825e-05,
      "loss": 0.0015149889513850212,
      "step": 139550
    },
    {
      "epoch": 39.61396537042293,
      "grad_norm": 0.01639995165169239,
      "learning_rate": 6.04027817201249e-05,
      "loss": 0.0017783386632800103,
      "step": 139560
    },
    {
      "epoch": 39.616803860346295,
      "grad_norm": 0.15964235365390778,
      "learning_rate": 6.039994323020154e-05,
      "loss": 0.0022891581058502196,
      "step": 139570
    },
    {
      "epoch": 39.61964235026966,
      "grad_norm": 1.8176294565200806,
      "learning_rate": 6.0397104740278175e-05,
      "loss": 0.016858980059623718,
      "step": 139580
    },
    {
      "epoch": 39.62248084019302,
      "grad_norm": 2.083268642425537,
      "learning_rate": 6.039426625035482e-05,
      "loss": 0.0023907598108053206,
      "step": 139590
    },
    {
      "epoch": 39.625319330116376,
      "grad_norm": 0.6479378342628479,
      "learning_rate": 6.039142776043145e-05,
      "loss": 0.0036281563341617586,
      "step": 139600
    },
    {
      "epoch": 39.62815782003974,
      "grad_norm": 7.420990943908691,
      "learning_rate": 6.0388589270508086e-05,
      "loss": 0.007852285355329513,
      "step": 139610
    },
    {
      "epoch": 39.6309963099631,
      "grad_norm": 3.2397515773773193,
      "learning_rate": 6.038575078058473e-05,
      "loss": 0.007737544178962707,
      "step": 139620
    },
    {
      "epoch": 39.63383479988646,
      "grad_norm": 0.9483978152275085,
      "learning_rate": 6.0382912290661376e-05,
      "loss": 0.0075049236416816715,
      "step": 139630
    },
    {
      "epoch": 39.63667328980982,
      "grad_norm": 0.06129680573940277,
      "learning_rate": 6.038007380073801e-05,
      "loss": 0.005858098715543747,
      "step": 139640
    },
    {
      "epoch": 39.63951177973318,
      "grad_norm": 1.0665252208709717,
      "learning_rate": 6.037723531081465e-05,
      "loss": 0.0031964600086212156,
      "step": 139650
    },
    {
      "epoch": 39.64235026965654,
      "grad_norm": 2.998321294784546,
      "learning_rate": 6.0374396820891286e-05,
      "loss": 0.004929675534367561,
      "step": 139660
    },
    {
      "epoch": 39.6451887595799,
      "grad_norm": 1.9591948986053467,
      "learning_rate": 6.037155833096793e-05,
      "loss": 0.0023342397063970564,
      "step": 139670
    },
    {
      "epoch": 39.648027249503265,
      "grad_norm": 0.08371206372976303,
      "learning_rate": 6.036871984104456e-05,
      "loss": 0.008088162541389466,
      "step": 139680
    },
    {
      "epoch": 39.65086573942663,
      "grad_norm": 3.0946784019470215,
      "learning_rate": 6.036588135112121e-05,
      "loss": 0.0017350615933537483,
      "step": 139690
    },
    {
      "epoch": 39.653704229349984,
      "grad_norm": 9.415877342224121,
      "learning_rate": 6.036304286119785e-05,
      "loss": 0.006233911216259003,
      "step": 139700
    },
    {
      "epoch": 39.65654271927335,
      "grad_norm": 0.035978496074676514,
      "learning_rate": 6.0360204371274486e-05,
      "loss": 0.0028254522010684015,
      "step": 139710
    },
    {
      "epoch": 39.65938120919671,
      "grad_norm": 5.757728576660156,
      "learning_rate": 6.035736588135113e-05,
      "loss": 0.004225365072488785,
      "step": 139720
    },
    {
      "epoch": 39.662219699120065,
      "grad_norm": 1.8212720155715942,
      "learning_rate": 6.035452739142776e-05,
      "loss": 0.003016148693859577,
      "step": 139730
    },
    {
      "epoch": 39.66505818904343,
      "grad_norm": 0.06038428097963333,
      "learning_rate": 6.03516889015044e-05,
      "loss": 0.007064750790596009,
      "step": 139740
    },
    {
      "epoch": 39.66789667896679,
      "grad_norm": 1.093817114830017,
      "learning_rate": 6.034885041158104e-05,
      "loss": 0.0026820652186870576,
      "step": 139750
    },
    {
      "epoch": 39.67073516889015,
      "grad_norm": 1.3431864976882935,
      "learning_rate": 6.0346011921657686e-05,
      "loss": 0.004437118396162986,
      "step": 139760
    },
    {
      "epoch": 39.67357365881351,
      "grad_norm": 0.06486725807189941,
      "learning_rate": 6.034317343173432e-05,
      "loss": 0.004336366429924965,
      "step": 139770
    },
    {
      "epoch": 39.67641214873687,
      "grad_norm": 0.02308625355362892,
      "learning_rate": 6.034033494181096e-05,
      "loss": 0.005778943002223968,
      "step": 139780
    },
    {
      "epoch": 39.679250638660235,
      "grad_norm": 0.32858359813690186,
      "learning_rate": 6.03374964518876e-05,
      "loss": 0.002763346582651138,
      "step": 139790
    },
    {
      "epoch": 39.68208912858359,
      "grad_norm": 0.20044772326946259,
      "learning_rate": 6.033465796196424e-05,
      "loss": 0.0023469334468245507,
      "step": 139800
    },
    {
      "epoch": 39.684927618506954,
      "grad_norm": 0.035473112016916275,
      "learning_rate": 6.033181947204087e-05,
      "loss": 0.0018551722168922424,
      "step": 139810
    },
    {
      "epoch": 39.68776610843032,
      "grad_norm": 0.032913029193878174,
      "learning_rate": 6.0328980982117514e-05,
      "loss": 0.006726323068141938,
      "step": 139820
    },
    {
      "epoch": 39.69060459835367,
      "grad_norm": 0.0315186083316803,
      "learning_rate": 6.032614249219416e-05,
      "loss": 0.0021501323208212852,
      "step": 139830
    },
    {
      "epoch": 39.693443088277036,
      "grad_norm": 0.055103130638599396,
      "learning_rate": 6.03233040022708e-05,
      "loss": 0.001127505674958229,
      "step": 139840
    },
    {
      "epoch": 39.6962815782004,
      "grad_norm": 1.3189668655395508,
      "learning_rate": 6.032046551234743e-05,
      "loss": 0.003320043534040451,
      "step": 139850
    },
    {
      "epoch": 39.69912006812376,
      "grad_norm": 1.2478957176208496,
      "learning_rate": 6.031762702242407e-05,
      "loss": 0.0030379649251699448,
      "step": 139860
    },
    {
      "epoch": 39.70195855804712,
      "grad_norm": 0.041675593703985214,
      "learning_rate": 6.031478853250071e-05,
      "loss": 0.0013714443892240524,
      "step": 139870
    },
    {
      "epoch": 39.70479704797048,
      "grad_norm": 0.10999451577663422,
      "learning_rate": 6.031195004257735e-05,
      "loss": 0.002758191153407097,
      "step": 139880
    },
    {
      "epoch": 39.70763553789384,
      "grad_norm": 1.0315507650375366,
      "learning_rate": 6.0309111552654e-05,
      "loss": 0.0023459285497665406,
      "step": 139890
    },
    {
      "epoch": 39.7104740278172,
      "grad_norm": 0.010173357091844082,
      "learning_rate": 6.030627306273063e-05,
      "loss": 0.0033757761120796204,
      "step": 139900
    },
    {
      "epoch": 39.71331251774056,
      "grad_norm": 12.188918113708496,
      "learning_rate": 6.030343457280727e-05,
      "loss": 0.0052807372063398365,
      "step": 139910
    },
    {
      "epoch": 39.716151007663925,
      "grad_norm": 0.020379696041345596,
      "learning_rate": 6.030059608288391e-05,
      "loss": 0.0008846750482916832,
      "step": 139920
    },
    {
      "epoch": 39.71898949758728,
      "grad_norm": 5.913230895996094,
      "learning_rate": 6.029775759296055e-05,
      "loss": 0.006001229956746101,
      "step": 139930
    },
    {
      "epoch": 39.72182798751064,
      "grad_norm": 9.762709617614746,
      "learning_rate": 6.0294919103037184e-05,
      "loss": 0.006464720517396927,
      "step": 139940
    },
    {
      "epoch": 39.724666477434006,
      "grad_norm": 0.3091679811477661,
      "learning_rate": 6.0292080613113825e-05,
      "loss": 0.0035059597343206407,
      "step": 139950
    },
    {
      "epoch": 39.72750496735737,
      "grad_norm": 0.03917648643255234,
      "learning_rate": 6.028924212319047e-05,
      "loss": 0.009939504414796829,
      "step": 139960
    },
    {
      "epoch": 39.730343457280725,
      "grad_norm": 3.0805511474609375,
      "learning_rate": 6.028640363326711e-05,
      "loss": 0.005064672231674195,
      "step": 139970
    },
    {
      "epoch": 39.73318194720409,
      "grad_norm": 0.5887700915336609,
      "learning_rate": 6.028356514334374e-05,
      "loss": 0.0012383176013827323,
      "step": 139980
    },
    {
      "epoch": 39.73602043712745,
      "grad_norm": 0.11344687640666962,
      "learning_rate": 6.0280726653420384e-05,
      "loss": 0.002821212261915207,
      "step": 139990
    },
    {
      "epoch": 39.738858927050806,
      "grad_norm": 0.05641164258122444,
      "learning_rate": 6.027788816349702e-05,
      "loss": 0.0049358226358890535,
      "step": 140000
    },
    {
      "epoch": 39.738858927050806,
      "eval_accuracy": 0.974883957525275,
      "eval_loss": 0.08956696093082428,
      "eval_runtime": 31.793,
      "eval_samples_per_second": 494.668,
      "eval_steps_per_second": 7.738,
      "step": 140000
    },
    {
      "epoch": 39.74169741697417,
      "grad_norm": 1.2411690950393677,
      "learning_rate": 6.027504967357366e-05,
      "loss": 0.003394593298435211,
      "step": 140010
    },
    {
      "epoch": 39.74453590689753,
      "grad_norm": 0.051181625574827194,
      "learning_rate": 6.0272211183650294e-05,
      "loss": 0.0035420209169387817,
      "step": 140020
    },
    {
      "epoch": 39.74737439682089,
      "grad_norm": 4.276035308837891,
      "learning_rate": 6.026937269372694e-05,
      "loss": 0.009138558804988862,
      "step": 140030
    },
    {
      "epoch": 39.75021288674425,
      "grad_norm": 0.19671708345413208,
      "learning_rate": 6.0266534203803584e-05,
      "loss": 0.006965765357017517,
      "step": 140040
    },
    {
      "epoch": 39.753051376667614,
      "grad_norm": 10.033811569213867,
      "learning_rate": 6.026369571388022e-05,
      "loss": 0.0063288785517215725,
      "step": 140050
    },
    {
      "epoch": 39.75588986659098,
      "grad_norm": 0.8704296946525574,
      "learning_rate": 6.026085722395686e-05,
      "loss": 0.00313953161239624,
      "step": 140060
    },
    {
      "epoch": 39.75872835651433,
      "grad_norm": 0.06608618050813675,
      "learning_rate": 6.0258018734033494e-05,
      "loss": 0.0021096281707286835,
      "step": 140070
    },
    {
      "epoch": 39.761566846437695,
      "grad_norm": 2.4122865200042725,
      "learning_rate": 6.025518024411013e-05,
      "loss": 0.005744149163365364,
      "step": 140080
    },
    {
      "epoch": 39.76440533636106,
      "grad_norm": 0.36474162340164185,
      "learning_rate": 6.0252341754186784e-05,
      "loss": 0.0022863009944558144,
      "step": 140090
    },
    {
      "epoch": 39.767243826284414,
      "grad_norm": 0.3941896855831146,
      "learning_rate": 6.024950326426342e-05,
      "loss": 0.004908876866102219,
      "step": 140100
    },
    {
      "epoch": 39.77008231620778,
      "grad_norm": 2.551473379135132,
      "learning_rate": 6.024666477434005e-05,
      "loss": 0.018268853425979614,
      "step": 140110
    },
    {
      "epoch": 39.77292080613114,
      "grad_norm": 0.9494673013687134,
      "learning_rate": 6.0243826284416694e-05,
      "loss": 0.0062050651758909225,
      "step": 140120
    },
    {
      "epoch": 39.7757592960545,
      "grad_norm": 0.06125468760728836,
      "learning_rate": 6.024098779449333e-05,
      "loss": 0.002519560232758522,
      "step": 140130
    },
    {
      "epoch": 39.77859778597786,
      "grad_norm": 0.5896734595298767,
      "learning_rate": 6.023814930456997e-05,
      "loss": 0.003685469925403595,
      "step": 140140
    },
    {
      "epoch": 39.78143627590122,
      "grad_norm": 0.10069131851196289,
      "learning_rate": 6.0235310814646605e-05,
      "loss": 0.0008576733991503716,
      "step": 140150
    },
    {
      "epoch": 39.784274765824584,
      "grad_norm": 0.05000810697674751,
      "learning_rate": 6.023247232472325e-05,
      "loss": 0.007351435720920563,
      "step": 140160
    },
    {
      "epoch": 39.78711325574794,
      "grad_norm": 0.09790735691785812,
      "learning_rate": 6.0229633834799895e-05,
      "loss": 0.0039072509855031965,
      "step": 140170
    },
    {
      "epoch": 39.7899517456713,
      "grad_norm": 5.382041931152344,
      "learning_rate": 6.022679534487653e-05,
      "loss": 0.0069732718169689175,
      "step": 140180
    },
    {
      "epoch": 39.792790235594666,
      "grad_norm": 0.09428535401821136,
      "learning_rate": 6.022395685495317e-05,
      "loss": 0.0019293110817670822,
      "step": 140190
    },
    {
      "epoch": 39.79562872551802,
      "grad_norm": 1.814720869064331,
      "learning_rate": 6.0221118365029805e-05,
      "loss": 0.007151990383863449,
      "step": 140200
    },
    {
      "epoch": 39.798467215441384,
      "grad_norm": 0.6311657428741455,
      "learning_rate": 6.021827987510644e-05,
      "loss": 0.014634586870670319,
      "step": 140210
    },
    {
      "epoch": 39.80130570536475,
      "grad_norm": 0.3648786246776581,
      "learning_rate": 6.021544138518308e-05,
      "loss": 0.008969998359680176,
      "step": 140220
    },
    {
      "epoch": 39.80414419528811,
      "grad_norm": 1.6945114135742188,
      "learning_rate": 6.021260289525973e-05,
      "loss": 0.004368625581264496,
      "step": 140230
    },
    {
      "epoch": 39.806982685211466,
      "grad_norm": 0.49376946687698364,
      "learning_rate": 6.0209764405336364e-05,
      "loss": 0.00094765555113554,
      "step": 140240
    },
    {
      "epoch": 39.80982117513483,
      "grad_norm": 0.027799341827630997,
      "learning_rate": 6.0206925915413005e-05,
      "loss": 0.0003513531759381294,
      "step": 140250
    },
    {
      "epoch": 39.81265966505819,
      "grad_norm": 0.28534236550331116,
      "learning_rate": 6.020408742548964e-05,
      "loss": 0.0050453715026378635,
      "step": 140260
    },
    {
      "epoch": 39.81549815498155,
      "grad_norm": 13.636035919189453,
      "learning_rate": 6.020124893556628e-05,
      "loss": 0.00275521632283926,
      "step": 140270
    },
    {
      "epoch": 39.81833664490491,
      "grad_norm": 0.1367509365081787,
      "learning_rate": 6.0198410445642916e-05,
      "loss": 0.0012105245143175126,
      "step": 140280
    },
    {
      "epoch": 39.82117513482827,
      "grad_norm": 1.6144416332244873,
      "learning_rate": 6.0195571955719564e-05,
      "loss": 0.0010772377252578735,
      "step": 140290
    },
    {
      "epoch": 39.82401362475163,
      "grad_norm": 3.217334032058716,
      "learning_rate": 6.0192733465796205e-05,
      "loss": 0.002232023328542709,
      "step": 140300
    },
    {
      "epoch": 39.82685211467499,
      "grad_norm": 0.11124135553836823,
      "learning_rate": 6.018989497587284e-05,
      "loss": 0.013311173021793365,
      "step": 140310
    },
    {
      "epoch": 39.829690604598355,
      "grad_norm": 1.462989091873169,
      "learning_rate": 6.0187056485949475e-05,
      "loss": 0.004180317372083664,
      "step": 140320
    },
    {
      "epoch": 39.83252909452172,
      "grad_norm": 0.1094597578048706,
      "learning_rate": 6.0184217996026116e-05,
      "loss": 0.006835350394248962,
      "step": 140330
    },
    {
      "epoch": 39.83536758444507,
      "grad_norm": 0.4314813017845154,
      "learning_rate": 6.018137950610275e-05,
      "loss": 0.0024197274819016457,
      "step": 140340
    },
    {
      "epoch": 39.838206074368436,
      "grad_norm": 0.7982630133628845,
      "learning_rate": 6.017854101617939e-05,
      "loss": 0.0026509566232562065,
      "step": 140350
    },
    {
      "epoch": 39.8410445642918,
      "grad_norm": 0.605013906955719,
      "learning_rate": 6.017570252625604e-05,
      "loss": 0.0019555581733584405,
      "step": 140360
    },
    {
      "epoch": 39.843883054215155,
      "grad_norm": 5.26009464263916,
      "learning_rate": 6.0172864036332675e-05,
      "loss": 0.0020079949870705606,
      "step": 140370
    },
    {
      "epoch": 39.84672154413852,
      "grad_norm": 1.0180621147155762,
      "learning_rate": 6.0170025546409316e-05,
      "loss": 0.00843769684433937,
      "step": 140380
    },
    {
      "epoch": 39.84956003406188,
      "grad_norm": 0.3391980826854706,
      "learning_rate": 6.016718705648595e-05,
      "loss": 0.007211767137050629,
      "step": 140390
    },
    {
      "epoch": 39.85239852398524,
      "grad_norm": 0.13601602613925934,
      "learning_rate": 6.016434856656259e-05,
      "loss": 0.0024050241336226463,
      "step": 140400
    },
    {
      "epoch": 39.8552370139086,
      "grad_norm": 0.4353223145008087,
      "learning_rate": 6.0161510076639227e-05,
      "loss": 0.0009262898936867713,
      "step": 140410
    },
    {
      "epoch": 39.85807550383196,
      "grad_norm": 10.732467651367188,
      "learning_rate": 6.015867158671587e-05,
      "loss": 0.007705754041671753,
      "step": 140420
    },
    {
      "epoch": 39.860913993755325,
      "grad_norm": 1.2364544868469238,
      "learning_rate": 6.0155833096792516e-05,
      "loss": 0.0016876265406608581,
      "step": 140430
    },
    {
      "epoch": 39.86375248367868,
      "grad_norm": 0.07032128423452377,
      "learning_rate": 6.015299460686915e-05,
      "loss": 0.0006453802809119225,
      "step": 140440
    },
    {
      "epoch": 39.866590973602044,
      "grad_norm": 0.636163055896759,
      "learning_rate": 6.0150156116945785e-05,
      "loss": 0.0051218122243881226,
      "step": 140450
    },
    {
      "epoch": 39.86942946352541,
      "grad_norm": 0.6675548553466797,
      "learning_rate": 6.014731762702243e-05,
      "loss": 0.0008221242576837539,
      "step": 140460
    },
    {
      "epoch": 39.87226795344876,
      "grad_norm": 10.39493465423584,
      "learning_rate": 6.014447913709906e-05,
      "loss": 0.005090107768774032,
      "step": 140470
    },
    {
      "epoch": 39.875106443372125,
      "grad_norm": 0.10883325338363647,
      "learning_rate": 6.01416406471757e-05,
      "loss": 0.0008152389898896217,
      "step": 140480
    },
    {
      "epoch": 39.87794493329549,
      "grad_norm": 18.141572952270508,
      "learning_rate": 6.013880215725235e-05,
      "loss": 0.011504334211349488,
      "step": 140490
    },
    {
      "epoch": 39.880783423218844,
      "grad_norm": 13.353707313537598,
      "learning_rate": 6.0135963667328985e-05,
      "loss": 0.009438977390527726,
      "step": 140500
    },
    {
      "epoch": 39.880783423218844,
      "eval_accuracy": 0.9767279201373434,
      "eval_loss": 0.08431672304868698,
      "eval_runtime": 32.0394,
      "eval_samples_per_second": 490.864,
      "eval_steps_per_second": 7.678,
      "step": 140500
    },
    {
      "epoch": 39.88362191314221,
      "grad_norm": 0.6881870627403259,
      "learning_rate": 6.013312517740563e-05,
      "loss": 0.0026777882128953935,
      "step": 140510
    },
    {
      "epoch": 39.88646040306557,
      "grad_norm": 0.5353898406028748,
      "learning_rate": 6.013028668748226e-05,
      "loss": 0.007203893363475799,
      "step": 140520
    },
    {
      "epoch": 39.88929889298893,
      "grad_norm": 0.04074109345674515,
      "learning_rate": 6.01274481975589e-05,
      "loss": 0.0034690462052822113,
      "step": 140530
    },
    {
      "epoch": 39.89213738291229,
      "grad_norm": 0.07092523574829102,
      "learning_rate": 6.012460970763554e-05,
      "loss": 0.0026951268315315246,
      "step": 140540
    },
    {
      "epoch": 39.89497587283565,
      "grad_norm": 2.031587839126587,
      "learning_rate": 6.012177121771217e-05,
      "loss": 0.00539964959025383,
      "step": 140550
    },
    {
      "epoch": 39.897814362759014,
      "grad_norm": 2.4877140522003174,
      "learning_rate": 6.011893272778883e-05,
      "loss": 0.001929086074233055,
      "step": 140560
    },
    {
      "epoch": 39.90065285268237,
      "grad_norm": 1.518911361694336,
      "learning_rate": 6.011609423786546e-05,
      "loss": 0.0015271568670868873,
      "step": 140570
    },
    {
      "epoch": 39.90349134260573,
      "grad_norm": 0.06446368247270584,
      "learning_rate": 6.0113255747942096e-05,
      "loss": 0.0006911342963576317,
      "step": 140580
    },
    {
      "epoch": 39.906329832529096,
      "grad_norm": 0.36873260140419006,
      "learning_rate": 6.011041725801874e-05,
      "loss": 0.002900325693190098,
      "step": 140590
    },
    {
      "epoch": 39.90916832245246,
      "grad_norm": 0.02814585343003273,
      "learning_rate": 6.010757876809537e-05,
      "loss": 0.0015156226232647895,
      "step": 140600
    },
    {
      "epoch": 39.912006812375814,
      "grad_norm": 9.48901081085205,
      "learning_rate": 6.0104740278172013e-05,
      "loss": 0.007503252476453781,
      "step": 140610
    },
    {
      "epoch": 39.91484530229918,
      "grad_norm": 0.4717351794242859,
      "learning_rate": 6.010190178824866e-05,
      "loss": 0.00471775159239769,
      "step": 140620
    },
    {
      "epoch": 39.91768379222254,
      "grad_norm": 0.2773877680301666,
      "learning_rate": 6.0099063298325296e-05,
      "loss": 0.0036538586020469664,
      "step": 140630
    },
    {
      "epoch": 39.920522282145896,
      "grad_norm": 0.043775565922260284,
      "learning_rate": 6.009622480840194e-05,
      "loss": 0.000935385748744011,
      "step": 140640
    },
    {
      "epoch": 39.92336077206926,
      "grad_norm": 0.1913825422525406,
      "learning_rate": 6.009338631847857e-05,
      "loss": 0.0012554600834846497,
      "step": 140650
    },
    {
      "epoch": 39.92619926199262,
      "grad_norm": 0.5335685610771179,
      "learning_rate": 6.0090547828555213e-05,
      "loss": 0.006734833121299744,
      "step": 140660
    },
    {
      "epoch": 39.92903775191598,
      "grad_norm": 1.2815186977386475,
      "learning_rate": 6.008770933863185e-05,
      "loss": 0.002618023753166199,
      "step": 140670
    },
    {
      "epoch": 39.93187624183934,
      "grad_norm": 0.039956893771886826,
      "learning_rate": 6.008487084870848e-05,
      "loss": 0.0031925756484270096,
      "step": 140680
    },
    {
      "epoch": 39.9347147317627,
      "grad_norm": 0.2748216390609741,
      "learning_rate": 6.008203235878513e-05,
      "loss": 0.0044944554567337034,
      "step": 140690
    },
    {
      "epoch": 39.937553221686066,
      "grad_norm": 0.08194872736930847,
      "learning_rate": 6.007919386886177e-05,
      "loss": 0.005241891369223595,
      "step": 140700
    },
    {
      "epoch": 39.94039171160942,
      "grad_norm": 0.13116778433322906,
      "learning_rate": 6.007635537893841e-05,
      "loss": 0.008138033747673034,
      "step": 140710
    },
    {
      "epoch": 39.943230201532785,
      "grad_norm": 11.924056053161621,
      "learning_rate": 6.007351688901505e-05,
      "loss": 0.012098485976457596,
      "step": 140720
    },
    {
      "epoch": 39.94606869145615,
      "grad_norm": 0.14996901154518127,
      "learning_rate": 6.007067839909168e-05,
      "loss": 0.000800822488963604,
      "step": 140730
    },
    {
      "epoch": 39.948907181379504,
      "grad_norm": 0.981072723865509,
      "learning_rate": 6.0067839909168324e-05,
      "loss": 0.0014003431424498558,
      "step": 140740
    },
    {
      "epoch": 39.95174567130287,
      "grad_norm": 1.7009402513504028,
      "learning_rate": 6.006500141924496e-05,
      "loss": 0.005894765257835388,
      "step": 140750
    },
    {
      "epoch": 39.95458416122623,
      "grad_norm": 10.593342781066895,
      "learning_rate": 6.006216292932161e-05,
      "loss": 0.0149017333984375,
      "step": 140760
    },
    {
      "epoch": 39.957422651149585,
      "grad_norm": 1.5137261152267456,
      "learning_rate": 6.005932443939825e-05,
      "loss": 0.009461023658514024,
      "step": 140770
    },
    {
      "epoch": 39.96026114107295,
      "grad_norm": 0.6203153729438782,
      "learning_rate": 6.005648594947488e-05,
      "loss": 0.007053401321172714,
      "step": 140780
    },
    {
      "epoch": 39.96309963099631,
      "grad_norm": 0.07132931798696518,
      "learning_rate": 6.005364745955152e-05,
      "loss": 0.002069216966629028,
      "step": 140790
    },
    {
      "epoch": 39.965938120919674,
      "grad_norm": 0.5278233885765076,
      "learning_rate": 6.005080896962816e-05,
      "loss": 0.0014589808881282807,
      "step": 140800
    },
    {
      "epoch": 39.96877661084303,
      "grad_norm": 12.345726013183594,
      "learning_rate": 6.0047970479704793e-05,
      "loss": 0.007135550677776337,
      "step": 140810
    },
    {
      "epoch": 39.97161510076639,
      "grad_norm": 0.5323654413223267,
      "learning_rate": 6.004513198978144e-05,
      "loss": 0.006078997254371643,
      "step": 140820
    },
    {
      "epoch": 39.974453590689755,
      "grad_norm": 0.4918469786643982,
      "learning_rate": 6.004229349985808e-05,
      "loss": 0.00623975470662117,
      "step": 140830
    },
    {
      "epoch": 39.97729208061311,
      "grad_norm": 0.12139986455440521,
      "learning_rate": 6.003945500993472e-05,
      "loss": 0.0022832226008176804,
      "step": 140840
    },
    {
      "epoch": 39.980130570536474,
      "grad_norm": 0.24073602259159088,
      "learning_rate": 6.003661652001136e-05,
      "loss": 0.0028021719306707384,
      "step": 140850
    },
    {
      "epoch": 39.98296906045984,
      "grad_norm": 0.03108965791761875,
      "learning_rate": 6.0033778030087994e-05,
      "loss": 0.014249612390995026,
      "step": 140860
    },
    {
      "epoch": 39.98580755038319,
      "grad_norm": 0.6516008377075195,
      "learning_rate": 6.0030939540164635e-05,
      "loss": 0.012614621222019196,
      "step": 140870
    },
    {
      "epoch": 39.988646040306556,
      "grad_norm": 0.26067066192626953,
      "learning_rate": 6.002810105024127e-05,
      "loss": 0.0032917670905590057,
      "step": 140880
    },
    {
      "epoch": 39.99148453022992,
      "grad_norm": 0.04338408634066582,
      "learning_rate": 6.002526256031792e-05,
      "loss": 0.002953190356492996,
      "step": 140890
    },
    {
      "epoch": 39.99432302015328,
      "grad_norm": 0.06802965700626373,
      "learning_rate": 6.002242407039456e-05,
      "loss": 0.0036380767822265627,
      "step": 140900
    },
    {
      "epoch": 39.99716151007664,
      "grad_norm": 9.522027969360352,
      "learning_rate": 6.0019585580471194e-05,
      "loss": 0.00637759193778038,
      "step": 140910
    },
    {
      "epoch": 40.0,
      "grad_norm": 0.28387269377708435,
      "learning_rate": 6.001674709054783e-05,
      "loss": 0.008763550221920014,
      "step": 140920
    },
    {
      "epoch": 40.00283848992336,
      "grad_norm": 0.8514469861984253,
      "learning_rate": 6.001390860062447e-05,
      "loss": 0.01131848394870758,
      "step": 140930
    },
    {
      "epoch": 40.00567697984672,
      "grad_norm": 4.396357536315918,
      "learning_rate": 6.0011070110701104e-05,
      "loss": 0.005232750624418259,
      "step": 140940
    },
    {
      "epoch": 40.00851546977008,
      "grad_norm": 1.8497380018234253,
      "learning_rate": 6.0008231620777746e-05,
      "loss": 0.0025484314188361167,
      "step": 140950
    },
    {
      "epoch": 40.011353959693444,
      "grad_norm": 0.0686957985162735,
      "learning_rate": 6.0005393130854394e-05,
      "loss": 0.0011997101828455925,
      "step": 140960
    },
    {
      "epoch": 40.01419244961681,
      "grad_norm": 0.09986752271652222,
      "learning_rate": 6.000255464093103e-05,
      "loss": 0.001767115667462349,
      "step": 140970
    },
    {
      "epoch": 40.01703093954016,
      "grad_norm": 0.04710081219673157,
      "learning_rate": 5.999971615100767e-05,
      "loss": 0.0010882586240768433,
      "step": 140980
    },
    {
      "epoch": 40.019869429463526,
      "grad_norm": 6.7900004386901855,
      "learning_rate": 5.9996877661084304e-05,
      "loss": 0.004949909448623657,
      "step": 140990
    },
    {
      "epoch": 40.02270791938689,
      "grad_norm": 3.7599222660064697,
      "learning_rate": 5.9994039171160946e-05,
      "loss": 0.0025281870737671854,
      "step": 141000
    },
    {
      "epoch": 40.02270791938689,
      "eval_accuracy": 0.9753290519488778,
      "eval_loss": 0.08811134099960327,
      "eval_runtime": 31.655,
      "eval_samples_per_second": 496.824,
      "eval_steps_per_second": 7.771,
      "step": 141000
    },
    {
      "epoch": 40.025546409310245,
      "grad_norm": 1.5409983396530151,
      "learning_rate": 5.999120068123758e-05,
      "loss": 0.002070675604045391,
      "step": 141010
    },
    {
      "epoch": 40.02838489923361,
      "grad_norm": 0.9793306589126587,
      "learning_rate": 5.998836219131423e-05,
      "loss": 0.001387464813888073,
      "step": 141020
    },
    {
      "epoch": 40.03122338915697,
      "grad_norm": 0.11398293823003769,
      "learning_rate": 5.998552370139087e-05,
      "loss": 0.001777113974094391,
      "step": 141030
    },
    {
      "epoch": 40.034061879080326,
      "grad_norm": 3.822636842727661,
      "learning_rate": 5.9982685211467504e-05,
      "loss": 0.0023824451491236687,
      "step": 141040
    },
    {
      "epoch": 40.03690036900369,
      "grad_norm": 0.03092256933450699,
      "learning_rate": 5.997984672154414e-05,
      "loss": 0.00811997652053833,
      "step": 141050
    },
    {
      "epoch": 40.03973885892705,
      "grad_norm": 1.8151283264160156,
      "learning_rate": 5.997700823162078e-05,
      "loss": 0.0037443444132804872,
      "step": 141060
    },
    {
      "epoch": 40.042577348850415,
      "grad_norm": 0.0816049575805664,
      "learning_rate": 5.9974169741697415e-05,
      "loss": 0.003933548927307129,
      "step": 141070
    },
    {
      "epoch": 40.04541583877377,
      "grad_norm": 1.2705925703048706,
      "learning_rate": 5.9971331251774056e-05,
      "loss": 0.011086380481719971,
      "step": 141080
    },
    {
      "epoch": 40.04825432869713,
      "grad_norm": 0.4045654833316803,
      "learning_rate": 5.9968492761850705e-05,
      "loss": 0.0018155571073293687,
      "step": 141090
    },
    {
      "epoch": 40.051092818620496,
      "grad_norm": 0.02475903369486332,
      "learning_rate": 5.996565427192734e-05,
      "loss": 0.00466834157705307,
      "step": 141100
    },
    {
      "epoch": 40.05393130854385,
      "grad_norm": 7.658180236816406,
      "learning_rate": 5.996281578200398e-05,
      "loss": 0.007724925130605698,
      "step": 141110
    },
    {
      "epoch": 40.056769798467215,
      "grad_norm": 12.927838325500488,
      "learning_rate": 5.9959977292080615e-05,
      "loss": 0.009171854704618454,
      "step": 141120
    },
    {
      "epoch": 40.05960828839058,
      "grad_norm": Infinity,
      "learning_rate": 5.9957138802157256e-05,
      "loss": 0.014033439755439758,
      "step": 141130
    },
    {
      "epoch": 40.062446778313934,
      "grad_norm": 5.594430446624756,
      "learning_rate": 5.9954584161226236e-05,
      "loss": 0.00665440559387207,
      "step": 141140
    },
    {
      "epoch": 40.0652852682373,
      "grad_norm": 2.498629331588745,
      "learning_rate": 5.995174567130287e-05,
      "loss": 0.003264225646853447,
      "step": 141150
    },
    {
      "epoch": 40.06812375816066,
      "grad_norm": 1.520716905593872,
      "learning_rate": 5.994890718137951e-05,
      "loss": 0.009837794303894042,
      "step": 141160
    },
    {
      "epoch": 40.07096224808402,
      "grad_norm": 0.09265720844268799,
      "learning_rate": 5.994606869145615e-05,
      "loss": 0.0010774070397019386,
      "step": 141170
    },
    {
      "epoch": 40.07380073800738,
      "grad_norm": 7.2474870681762695,
      "learning_rate": 5.994323020153279e-05,
      "loss": 0.0023976681753993035,
      "step": 141180
    },
    {
      "epoch": 40.07663922793074,
      "grad_norm": 0.14696305990219116,
      "learning_rate": 5.994039171160942e-05,
      "loss": 0.00358167439699173,
      "step": 141190
    },
    {
      "epoch": 40.079477717854104,
      "grad_norm": 0.3328649401664734,
      "learning_rate": 5.993755322168607e-05,
      "loss": 0.0019380806013941764,
      "step": 141200
    },
    {
      "epoch": 40.08231620777746,
      "grad_norm": 6.6797990798950195,
      "learning_rate": 5.9934714731762706e-05,
      "loss": 0.0036746986210346223,
      "step": 141210
    },
    {
      "epoch": 40.08515469770082,
      "grad_norm": 1.7957972288131714,
      "learning_rate": 5.993187624183935e-05,
      "loss": 0.002004554681479931,
      "step": 141220
    },
    {
      "epoch": 40.087993187624186,
      "grad_norm": 0.16373403370380402,
      "learning_rate": 5.992903775191598e-05,
      "loss": 0.002941415086388588,
      "step": 141230
    },
    {
      "epoch": 40.09083167754754,
      "grad_norm": 0.03368532657623291,
      "learning_rate": 5.992619926199262e-05,
      "loss": 0.0030576467514038088,
      "step": 141240
    },
    {
      "epoch": 40.093670167470904,
      "grad_norm": 0.062392618507146835,
      "learning_rate": 5.992336077206926e-05,
      "loss": 0.004089279472827912,
      "step": 141250
    },
    {
      "epoch": 40.09650865739427,
      "grad_norm": 0.3097897171974182,
      "learning_rate": 5.99205222821459e-05,
      "loss": 0.0013048615306615829,
      "step": 141260
    },
    {
      "epoch": 40.09934714731763,
      "grad_norm": 0.103785939514637,
      "learning_rate": 5.991768379222255e-05,
      "loss": 0.006063215434551239,
      "step": 141270
    },
    {
      "epoch": 40.102185637240986,
      "grad_norm": 2.2910470962524414,
      "learning_rate": 5.991484530229918e-05,
      "loss": 0.009402237087488174,
      "step": 141280
    },
    {
      "epoch": 40.10502412716435,
      "grad_norm": 0.4142440855503082,
      "learning_rate": 5.991200681237582e-05,
      "loss": 0.002428651973605156,
      "step": 141290
    },
    {
      "epoch": 40.10786261708771,
      "grad_norm": 0.07694739103317261,
      "learning_rate": 5.990916832245246e-05,
      "loss": 0.0013912970200181008,
      "step": 141300
    },
    {
      "epoch": 40.11070110701107,
      "grad_norm": 15.2948579788208,
      "learning_rate": 5.990632983252909e-05,
      "loss": 0.01813354641199112,
      "step": 141310
    },
    {
      "epoch": 40.11353959693443,
      "grad_norm": 0.0976259857416153,
      "learning_rate": 5.9903491342605734e-05,
      "loss": 0.0032697319984436035,
      "step": 141320
    },
    {
      "epoch": 40.11637808685779,
      "grad_norm": 0.23160488903522491,
      "learning_rate": 5.990065285268237e-05,
      "loss": 0.004273773357272148,
      "step": 141330
    },
    {
      "epoch": 40.119216576781156,
      "grad_norm": 3.36407732963562,
      "learning_rate": 5.9897814362759016e-05,
      "loss": 0.0016891650855541229,
      "step": 141340
    },
    {
      "epoch": 40.12205506670451,
      "grad_norm": 0.12114007771015167,
      "learning_rate": 5.989497587283566e-05,
      "loss": 0.0010487152263522147,
      "step": 141350
    },
    {
      "epoch": 40.124893556627875,
      "grad_norm": 0.5600854754447937,
      "learning_rate": 5.989213738291229e-05,
      "loss": 0.0018773723393678665,
      "step": 141360
    },
    {
      "epoch": 40.12773204655124,
      "grad_norm": 2.2968366146087646,
      "learning_rate": 5.9889298892988934e-05,
      "loss": 0.002494002506136894,
      "step": 141370
    },
    {
      "epoch": 40.13057053647459,
      "grad_norm": 0.09448488801717758,
      "learning_rate": 5.988646040306557e-05,
      "loss": 0.0030395571142435075,
      "step": 141380
    },
    {
      "epoch": 40.133409026397956,
      "grad_norm": 2.960134744644165,
      "learning_rate": 5.988362191314221e-05,
      "loss": 0.002326612547039986,
      "step": 141390
    },
    {
      "epoch": 40.13624751632132,
      "grad_norm": 0.20933572947978973,
      "learning_rate": 5.988078342321886e-05,
      "loss": 0.0021002670750021935,
      "step": 141400
    },
    {
      "epoch": 40.139086006244675,
      "grad_norm": 0.07288416475057602,
      "learning_rate": 5.987794493329549e-05,
      "loss": 0.0024751640856266023,
      "step": 141410
    },
    {
      "epoch": 40.14192449616804,
      "grad_norm": 0.3387976288795471,
      "learning_rate": 5.9875106443372134e-05,
      "loss": 0.0009743316099047661,
      "step": 141420
    },
    {
      "epoch": 40.1447629860914,
      "grad_norm": 2.274984121322632,
      "learning_rate": 5.987226795344877e-05,
      "loss": 0.007703641057014465,
      "step": 141430
    },
    {
      "epoch": 40.14760147601476,
      "grad_norm": 0.3951001763343811,
      "learning_rate": 5.98694294635254e-05,
      "loss": 0.0011407285928726196,
      "step": 141440
    },
    {
      "epoch": 40.15043996593812,
      "grad_norm": 0.35357916355133057,
      "learning_rate": 5.9866590973602044e-05,
      "loss": 0.0029900169000029566,
      "step": 141450
    },
    {
      "epoch": 40.15327845586148,
      "grad_norm": 0.04896289482712746,
      "learning_rate": 5.986375248367868e-05,
      "loss": 0.0015193462371826173,
      "step": 141460
    },
    {
      "epoch": 40.156116945784845,
      "grad_norm": 0.541138231754303,
      "learning_rate": 5.986091399375533e-05,
      "loss": 0.0006352784112095833,
      "step": 141470
    },
    {
      "epoch": 40.1589554357082,
      "grad_norm": 1.0357104539871216,
      "learning_rate": 5.985807550383197e-05,
      "loss": 0.008746492862701415,
      "step": 141480
    },
    {
      "epoch": 40.161793925631564,
      "grad_norm": 0.454870343208313,
      "learning_rate": 5.98552370139086e-05,
      "loss": 0.0009992912411689759,
      "step": 141490
    },
    {
      "epoch": 40.16463241555493,
      "grad_norm": 0.28350916504859924,
      "learning_rate": 5.9852398523985245e-05,
      "loss": 0.004488542675971985,
      "step": 141500
    },
    {
      "epoch": 40.16463241555493,
      "eval_accuracy": 0.9769822598079736,
      "eval_loss": 0.07938940078020096,
      "eval_runtime": 31.9266,
      "eval_samples_per_second": 492.598,
      "eval_steps_per_second": 7.705,
      "step": 141500
    },
    {
      "epoch": 40.16747090547828,
      "grad_norm": 0.029779765754938126,
      "learning_rate": 5.984956003406188e-05,
      "loss": 0.000894186832010746,
      "step": 141510
    },
    {
      "epoch": 40.170309395401645,
      "grad_norm": 0.33482983708381653,
      "learning_rate": 5.984672154413852e-05,
      "loss": 0.0012454506009817123,
      "step": 141520
    },
    {
      "epoch": 40.17314788532501,
      "grad_norm": 0.13806049525737762,
      "learning_rate": 5.9843883054215155e-05,
      "loss": 0.0006639633327722549,
      "step": 141530
    },
    {
      "epoch": 40.17598637524837,
      "grad_norm": 0.671042263507843,
      "learning_rate": 5.98410445642918e-05,
      "loss": 0.0007220933213829995,
      "step": 141540
    },
    {
      "epoch": 40.17882486517173,
      "grad_norm": 0.781765878200531,
      "learning_rate": 5.9838206074368445e-05,
      "loss": 0.000602009892463684,
      "step": 141550
    },
    {
      "epoch": 40.18166335509509,
      "grad_norm": 0.01682823896408081,
      "learning_rate": 5.983536758444508e-05,
      "loss": 0.0032154031097888945,
      "step": 141560
    },
    {
      "epoch": 40.18450184501845,
      "grad_norm": 0.4468253254890442,
      "learning_rate": 5.9832529094521714e-05,
      "loss": 0.004222622141242027,
      "step": 141570
    },
    {
      "epoch": 40.18734033494181,
      "grad_norm": 0.03703214228153229,
      "learning_rate": 5.9829690604598355e-05,
      "loss": 0.0003958847373723984,
      "step": 141580
    },
    {
      "epoch": 40.19017882486517,
      "grad_norm": 1.3852514028549194,
      "learning_rate": 5.982685211467499e-05,
      "loss": 0.0016364442184567452,
      "step": 141590
    },
    {
      "epoch": 40.193017314788534,
      "grad_norm": 0.3018795847892761,
      "learning_rate": 5.982401362475164e-05,
      "loss": 0.002281428873538971,
      "step": 141600
    },
    {
      "epoch": 40.19585580471189,
      "grad_norm": 5.125916004180908,
      "learning_rate": 5.982117513482828e-05,
      "loss": 0.01114809736609459,
      "step": 141610
    },
    {
      "epoch": 40.19869429463525,
      "grad_norm": 1.877244234085083,
      "learning_rate": 5.9818336644904914e-05,
      "loss": 0.0007845491170883178,
      "step": 141620
    },
    {
      "epoch": 40.201532784558616,
      "grad_norm": 0.3787298798561096,
      "learning_rate": 5.9815498154981555e-05,
      "loss": 0.0005596024915575981,
      "step": 141630
    },
    {
      "epoch": 40.20437127448198,
      "grad_norm": 0.07247426360845566,
      "learning_rate": 5.981265966505819e-05,
      "loss": 0.004596986621618271,
      "step": 141640
    },
    {
      "epoch": 40.207209764405334,
      "grad_norm": 0.042944468557834625,
      "learning_rate": 5.980982117513483e-05,
      "loss": 0.001783546805381775,
      "step": 141650
    },
    {
      "epoch": 40.2100482543287,
      "grad_norm": 0.4978560209274292,
      "learning_rate": 5.9806982685211466e-05,
      "loss": 0.012850983440876007,
      "step": 141660
    },
    {
      "epoch": 40.21288674425206,
      "grad_norm": 13.367024421691895,
      "learning_rate": 5.9804144195288114e-05,
      "loss": 0.00840640664100647,
      "step": 141670
    },
    {
      "epoch": 40.215725234175416,
      "grad_norm": 0.14834816753864288,
      "learning_rate": 5.980130570536475e-05,
      "loss": 0.0025252092629671095,
      "step": 141680
    },
    {
      "epoch": 40.21856372409878,
      "grad_norm": 0.09729539602994919,
      "learning_rate": 5.979846721544139e-05,
      "loss": 0.0019758220762014387,
      "step": 141690
    },
    {
      "epoch": 40.22140221402214,
      "grad_norm": 20.349626541137695,
      "learning_rate": 5.9795628725518025e-05,
      "loss": 0.01949440687894821,
      "step": 141700
    },
    {
      "epoch": 40.2242407039455,
      "grad_norm": 0.037174321711063385,
      "learning_rate": 5.9792790235594666e-05,
      "loss": 0.001443978026509285,
      "step": 141710
    },
    {
      "epoch": 40.22707919386886,
      "grad_norm": 1.1710386276245117,
      "learning_rate": 5.97899517456713e-05,
      "loss": 0.00927068516612053,
      "step": 141720
    },
    {
      "epoch": 40.22991768379222,
      "grad_norm": 0.09592163562774658,
      "learning_rate": 5.978711325574794e-05,
      "loss": 0.0022571787238121034,
      "step": 141730
    },
    {
      "epoch": 40.232756173715586,
      "grad_norm": 0.17454227805137634,
      "learning_rate": 5.978427476582459e-05,
      "loss": 0.0012124985456466674,
      "step": 141740
    },
    {
      "epoch": 40.23559466363894,
      "grad_norm": 0.08364821970462799,
      "learning_rate": 5.9781436275901225e-05,
      "loss": 0.001541801542043686,
      "step": 141750
    },
    {
      "epoch": 40.238433153562305,
      "grad_norm": 0.0061753592453897,
      "learning_rate": 5.9778597785977866e-05,
      "loss": 0.0006104664877057076,
      "step": 141760
    },
    {
      "epoch": 40.24127164348567,
      "grad_norm": 0.4816244840621948,
      "learning_rate": 5.97757592960545e-05,
      "loss": 0.0006246676668524743,
      "step": 141770
    },
    {
      "epoch": 40.24411013340902,
      "grad_norm": 0.14365136623382568,
      "learning_rate": 5.9772920806131135e-05,
      "loss": 0.0023251485079526903,
      "step": 141780
    },
    {
      "epoch": 40.246948623332386,
      "grad_norm": 0.29194337129592896,
      "learning_rate": 5.977008231620778e-05,
      "loss": 0.00385005921125412,
      "step": 141790
    },
    {
      "epoch": 40.24978711325575,
      "grad_norm": 0.1193542331457138,
      "learning_rate": 5.9767243826284425e-05,
      "loss": 0.004979455843567848,
      "step": 141800
    },
    {
      "epoch": 40.25262560317911,
      "grad_norm": 0.4120159447193146,
      "learning_rate": 5.976440533636106e-05,
      "loss": 0.0018924063071608544,
      "step": 141810
    },
    {
      "epoch": 40.25546409310247,
      "grad_norm": 10.911212921142578,
      "learning_rate": 5.97615668464377e-05,
      "loss": 0.009361256659030915,
      "step": 141820
    },
    {
      "epoch": 40.25830258302583,
      "grad_norm": 18.51230812072754,
      "learning_rate": 5.9758728356514335e-05,
      "loss": 0.00924486294388771,
      "step": 141830
    },
    {
      "epoch": 40.261141072949194,
      "grad_norm": 8.473150253295898,
      "learning_rate": 5.975588986659098e-05,
      "loss": 0.0069797798991203305,
      "step": 141840
    },
    {
      "epoch": 40.26397956287255,
      "grad_norm": 0.31222599744796753,
      "learning_rate": 5.975305137666761e-05,
      "loss": 0.0031681660562753677,
      "step": 141850
    },
    {
      "epoch": 40.26681805279591,
      "grad_norm": 0.06635484099388123,
      "learning_rate": 5.975021288674425e-05,
      "loss": 0.002317035011947155,
      "step": 141860
    },
    {
      "epoch": 40.269656542719275,
      "grad_norm": 14.258094787597656,
      "learning_rate": 5.97473743968209e-05,
      "loss": 0.006613697856664658,
      "step": 141870
    },
    {
      "epoch": 40.27249503264263,
      "grad_norm": 0.02786726877093315,
      "learning_rate": 5.9744535906897536e-05,
      "loss": 0.0020290933549404143,
      "step": 141880
    },
    {
      "epoch": 40.275333522565994,
      "grad_norm": 0.33624267578125,
      "learning_rate": 5.974169741697418e-05,
      "loss": 0.0017712401226162911,
      "step": 141890
    },
    {
      "epoch": 40.27817201248936,
      "grad_norm": 0.7553533315658569,
      "learning_rate": 5.973885892705081e-05,
      "loss": 0.0025712938979268072,
      "step": 141900
    },
    {
      "epoch": 40.28101050241272,
      "grad_norm": 0.2746584117412567,
      "learning_rate": 5.9736020437127446e-05,
      "loss": 0.0008062634617090226,
      "step": 141910
    },
    {
      "epoch": 40.283848992336075,
      "grad_norm": 13.061668395996094,
      "learning_rate": 5.973318194720409e-05,
      "loss": 0.008780070394277573,
      "step": 141920
    },
    {
      "epoch": 40.28668748225944,
      "grad_norm": 22.92772102355957,
      "learning_rate": 5.973034345728072e-05,
      "loss": 0.015335842967033386,
      "step": 141930
    },
    {
      "epoch": 40.2895259721828,
      "grad_norm": 2.1875548362731934,
      "learning_rate": 5.972750496735737e-05,
      "loss": 0.012350553274154663,
      "step": 141940
    },
    {
      "epoch": 40.29236446210616,
      "grad_norm": 0.283885657787323,
      "learning_rate": 5.972466647743401e-05,
      "loss": 0.0013665201142430306,
      "step": 141950
    },
    {
      "epoch": 40.29520295202952,
      "grad_norm": 0.3917623460292816,
      "learning_rate": 5.9721827987510646e-05,
      "loss": 0.000819229707121849,
      "step": 141960
    },
    {
      "epoch": 40.29804144195288,
      "grad_norm": 16.855709075927734,
      "learning_rate": 5.971898949758729e-05,
      "loss": 0.018814995884895325,
      "step": 141970
    },
    {
      "epoch": 40.30087993187624,
      "grad_norm": 0.5393671989440918,
      "learning_rate": 5.971615100766392e-05,
      "loss": 0.002152710407972336,
      "step": 141980
    },
    {
      "epoch": 40.3037184217996,
      "grad_norm": 0.6449386477470398,
      "learning_rate": 5.9713312517740564e-05,
      "loss": 0.002814857475459576,
      "step": 141990
    },
    {
      "epoch": 40.306556911722964,
      "grad_norm": 0.241722971200943,
      "learning_rate": 5.971047402781721e-05,
      "loss": 0.004410463944077491,
      "step": 142000
    },
    {
      "epoch": 40.306556911722964,
      "eval_accuracy": 0.9745660329369873,
      "eval_loss": 0.09066623449325562,
      "eval_runtime": 31.5119,
      "eval_samples_per_second": 499.081,
      "eval_steps_per_second": 7.807,
      "step": 142000
    },
    {
      "epoch": 40.30939540164633,
      "grad_norm": 1.0706580877304077,
      "learning_rate": 5.9707635537893846e-05,
      "loss": 0.007949984073638916,
      "step": 142010
    },
    {
      "epoch": 40.31223389156968,
      "grad_norm": 8.754439353942871,
      "learning_rate": 5.970479704797049e-05,
      "loss": 0.0062861904501914975,
      "step": 142020
    },
    {
      "epoch": 40.315072381493046,
      "grad_norm": 0.10794955492019653,
      "learning_rate": 5.970195855804712e-05,
      "loss": 0.001417907141149044,
      "step": 142030
    },
    {
      "epoch": 40.31791087141641,
      "grad_norm": 3.55732798576355,
      "learning_rate": 5.969912006812376e-05,
      "loss": 0.0011975906789302825,
      "step": 142040
    },
    {
      "epoch": 40.320749361339765,
      "grad_norm": 0.2522185742855072,
      "learning_rate": 5.96962815782004e-05,
      "loss": 0.001125376671552658,
      "step": 142050
    },
    {
      "epoch": 40.32358785126313,
      "grad_norm": 10.290528297424316,
      "learning_rate": 5.969344308827703e-05,
      "loss": 0.006360298395156861,
      "step": 142060
    },
    {
      "epoch": 40.32642634118649,
      "grad_norm": 1.0646207332611084,
      "learning_rate": 5.969060459835368e-05,
      "loss": 0.012793755531311036,
      "step": 142070
    },
    {
      "epoch": 40.329264831109846,
      "grad_norm": 0.09985020011663437,
      "learning_rate": 5.968776610843032e-05,
      "loss": 0.006348150968551636,
      "step": 142080
    },
    {
      "epoch": 40.33210332103321,
      "grad_norm": 0.13184960186481476,
      "learning_rate": 5.968492761850696e-05,
      "loss": 0.010489391535520554,
      "step": 142090
    },
    {
      "epoch": 40.33494181095657,
      "grad_norm": 15.567636489868164,
      "learning_rate": 5.96820891285836e-05,
      "loss": 0.00817161351442337,
      "step": 142100
    },
    {
      "epoch": 40.337780300879935,
      "grad_norm": 0.763323187828064,
      "learning_rate": 5.967925063866023e-05,
      "loss": 0.0026276268064975737,
      "step": 142110
    },
    {
      "epoch": 40.34061879080329,
      "grad_norm": 0.24279308319091797,
      "learning_rate": 5.9676412148736874e-05,
      "loss": 0.010016703605651855,
      "step": 142120
    },
    {
      "epoch": 40.34345728072665,
      "grad_norm": 0.8303472995758057,
      "learning_rate": 5.967357365881351e-05,
      "loss": 0.008593074977397919,
      "step": 142130
    },
    {
      "epoch": 40.346295770650016,
      "grad_norm": 11.335101127624512,
      "learning_rate": 5.967073516889016e-05,
      "loss": 0.006815989315509796,
      "step": 142140
    },
    {
      "epoch": 40.34913426057337,
      "grad_norm": 0.20901642739772797,
      "learning_rate": 5.966789667896679e-05,
      "loss": 0.007770717889070511,
      "step": 142150
    },
    {
      "epoch": 40.351972750496735,
      "grad_norm": 14.806703567504883,
      "learning_rate": 5.966505818904343e-05,
      "loss": 0.007448455691337586,
      "step": 142160
    },
    {
      "epoch": 40.3548112404201,
      "grad_norm": 3.6475741863250732,
      "learning_rate": 5.966221969912007e-05,
      "loss": 0.010018229484558105,
      "step": 142170
    },
    {
      "epoch": 40.35764973034346,
      "grad_norm": 1.5319020748138428,
      "learning_rate": 5.965938120919671e-05,
      "loss": 0.004115867614746094,
      "step": 142180
    },
    {
      "epoch": 40.36048822026682,
      "grad_norm": 0.436472088098526,
      "learning_rate": 5.9656542719273344e-05,
      "loss": 0.004052790999412537,
      "step": 142190
    },
    {
      "epoch": 40.36332671019018,
      "grad_norm": 19.666156768798828,
      "learning_rate": 5.965370422934999e-05,
      "loss": 0.008727312088012695,
      "step": 142200
    },
    {
      "epoch": 40.36616520011354,
      "grad_norm": 0.3000084161758423,
      "learning_rate": 5.965086573942663e-05,
      "loss": 0.0011414213106036187,
      "step": 142210
    },
    {
      "epoch": 40.3690036900369,
      "grad_norm": 0.504807710647583,
      "learning_rate": 5.964802724950327e-05,
      "loss": 0.003113292343914509,
      "step": 142220
    },
    {
      "epoch": 40.37184217996026,
      "grad_norm": 0.4487690329551697,
      "learning_rate": 5.964518875957991e-05,
      "loss": 0.007101792097091675,
      "step": 142230
    },
    {
      "epoch": 40.374680669883624,
      "grad_norm": 0.13410568237304688,
      "learning_rate": 5.9642350269656544e-05,
      "loss": 0.00459064319729805,
      "step": 142240
    },
    {
      "epoch": 40.37751915980698,
      "grad_norm": 0.06903411448001862,
      "learning_rate": 5.963951177973318e-05,
      "loss": 0.005993295460939407,
      "step": 142250
    },
    {
      "epoch": 40.38035764973034,
      "grad_norm": 0.2009345442056656,
      "learning_rate": 5.963667328980982e-05,
      "loss": 0.002538302168250084,
      "step": 142260
    },
    {
      "epoch": 40.383196139653705,
      "grad_norm": 0.08502268046140671,
      "learning_rate": 5.963383479988647e-05,
      "loss": 0.012985649704933166,
      "step": 142270
    },
    {
      "epoch": 40.38603462957707,
      "grad_norm": 0.4799043536186218,
      "learning_rate": 5.96309963099631e-05,
      "loss": 0.0014283055439591407,
      "step": 142280
    },
    {
      "epoch": 40.388873119500424,
      "grad_norm": 6.5437798500061035,
      "learning_rate": 5.9628157820039744e-05,
      "loss": 0.00463481992483139,
      "step": 142290
    },
    {
      "epoch": 40.39171160942379,
      "grad_norm": 6.778109073638916,
      "learning_rate": 5.962531933011638e-05,
      "loss": 0.013238084316253663,
      "step": 142300
    },
    {
      "epoch": 40.39455009934715,
      "grad_norm": 0.1805339902639389,
      "learning_rate": 5.962248084019302e-05,
      "loss": 0.004393396154046059,
      "step": 142310
    },
    {
      "epoch": 40.397388589270506,
      "grad_norm": 3.6597604751586914,
      "learning_rate": 5.9619642350269654e-05,
      "loss": 0.004605831205844879,
      "step": 142320
    },
    {
      "epoch": 40.40022707919387,
      "grad_norm": 0.11904043704271317,
      "learning_rate": 5.96168038603463e-05,
      "loss": 0.003317001089453697,
      "step": 142330
    },
    {
      "epoch": 40.40306556911723,
      "grad_norm": 2.5841944217681885,
      "learning_rate": 5.9613965370422944e-05,
      "loss": 0.0045594900846481325,
      "step": 142340
    },
    {
      "epoch": 40.40590405904059,
      "grad_norm": 0.11179997771978378,
      "learning_rate": 5.961112688049958e-05,
      "loss": 0.008069639652967453,
      "step": 142350
    },
    {
      "epoch": 40.40874254896395,
      "grad_norm": 0.7843412160873413,
      "learning_rate": 5.960828839057622e-05,
      "loss": 0.0018729304894804955,
      "step": 142360
    },
    {
      "epoch": 40.41158103888731,
      "grad_norm": 2.934800863265991,
      "learning_rate": 5.9605449900652854e-05,
      "loss": 0.019526392221450806,
      "step": 142370
    },
    {
      "epoch": 40.414419528810676,
      "grad_norm": 0.5769354701042175,
      "learning_rate": 5.960261141072949e-05,
      "loss": 0.012909942865371704,
      "step": 142380
    },
    {
      "epoch": 40.41725801873403,
      "grad_norm": 0.25136059522628784,
      "learning_rate": 5.959977292080613e-05,
      "loss": 0.009529231488704682,
      "step": 142390
    },
    {
      "epoch": 40.420096508657394,
      "grad_norm": 0.08988066762685776,
      "learning_rate": 5.959693443088278e-05,
      "loss": 0.005256969109177589,
      "step": 142400
    },
    {
      "epoch": 40.42293499858076,
      "grad_norm": 0.7548783421516418,
      "learning_rate": 5.959409594095941e-05,
      "loss": 0.0011375119909644126,
      "step": 142410
    },
    {
      "epoch": 40.42577348850411,
      "grad_norm": 3.813415050506592,
      "learning_rate": 5.9591257451036055e-05,
      "loss": 0.004427648335695267,
      "step": 142420
    },
    {
      "epoch": 40.428611978427476,
      "grad_norm": 0.06868291646242142,
      "learning_rate": 5.958841896111269e-05,
      "loss": 0.0034062966704368593,
      "step": 142430
    },
    {
      "epoch": 40.43145046835084,
      "grad_norm": 1.811394214630127,
      "learning_rate": 5.958558047118933e-05,
      "loss": 0.002454865537583828,
      "step": 142440
    },
    {
      "epoch": 40.434288958274195,
      "grad_norm": 1.7922894954681396,
      "learning_rate": 5.9582741981265965e-05,
      "loss": 0.002000577375292778,
      "step": 142450
    },
    {
      "epoch": 40.43712744819756,
      "grad_norm": 0.5963754057884216,
      "learning_rate": 5.9579903491342606e-05,
      "loss": 0.004449887573719025,
      "step": 142460
    },
    {
      "epoch": 40.43996593812092,
      "grad_norm": 0.10476581752300262,
      "learning_rate": 5.9577065001419255e-05,
      "loss": 0.00043170023709535597,
      "step": 142470
    },
    {
      "epoch": 40.44280442804428,
      "grad_norm": 0.3502050042152405,
      "learning_rate": 5.957422651149589e-05,
      "loss": 0.001779523491859436,
      "step": 142480
    },
    {
      "epoch": 40.44564291796764,
      "grad_norm": 0.03826627880334854,
      "learning_rate": 5.957138802157253e-05,
      "loss": 0.0007922479882836342,
      "step": 142490
    },
    {
      "epoch": 40.448481407891,
      "grad_norm": 4.65405797958374,
      "learning_rate": 5.9568549531649165e-05,
      "loss": 0.0019172223284840584,
      "step": 142500
    },
    {
      "epoch": 40.448481407891,
      "eval_accuracy": 0.9725949004896038,
      "eval_loss": 0.09436613321304321,
      "eval_runtime": 32.2473,
      "eval_samples_per_second": 487.7,
      "eval_steps_per_second": 7.629,
      "step": 142500
    },
    {
      "epoch": 40.451319897814365,
      "grad_norm": 0.7506247758865356,
      "learning_rate": 5.95657110417258e-05,
      "loss": 0.0015973871573805808,
      "step": 142510
    },
    {
      "epoch": 40.45415838773772,
      "grad_norm": 2.9844815731048584,
      "learning_rate": 5.956287255180244e-05,
      "loss": 0.005705595761537552,
      "step": 142520
    },
    {
      "epoch": 40.45699687766108,
      "grad_norm": 1.3179161548614502,
      "learning_rate": 5.956003406187909e-05,
      "loss": 0.003897494077682495,
      "step": 142530
    },
    {
      "epoch": 40.459835367584446,
      "grad_norm": 1.0199309587478638,
      "learning_rate": 5.9557195571955724e-05,
      "loss": 0.007392019033432007,
      "step": 142540
    },
    {
      "epoch": 40.46267385750781,
      "grad_norm": 0.9979227781295776,
      "learning_rate": 5.9554357082032365e-05,
      "loss": 0.0027575120329856873,
      "step": 142550
    },
    {
      "epoch": 40.465512347431165,
      "grad_norm": 1.0810099840164185,
      "learning_rate": 5.9551518592109e-05,
      "loss": 0.0050247333943843845,
      "step": 142560
    },
    {
      "epoch": 40.46835083735453,
      "grad_norm": 0.1577218621969223,
      "learning_rate": 5.954868010218564e-05,
      "loss": 0.002727147191762924,
      "step": 142570
    },
    {
      "epoch": 40.47118932727789,
      "grad_norm": 0.08516820520162582,
      "learning_rate": 5.9545841612262276e-05,
      "loss": 0.004151012748479843,
      "step": 142580
    },
    {
      "epoch": 40.47402781720125,
      "grad_norm": 0.803595244884491,
      "learning_rate": 5.954300312233892e-05,
      "loss": 0.002104494534432888,
      "step": 142590
    },
    {
      "epoch": 40.47686630712461,
      "grad_norm": 0.5441530346870422,
      "learning_rate": 5.9540164632415565e-05,
      "loss": 0.01703992784023285,
      "step": 142600
    },
    {
      "epoch": 40.47970479704797,
      "grad_norm": 0.2445831149816513,
      "learning_rate": 5.95373261424922e-05,
      "loss": 0.0015179546549916267,
      "step": 142610
    },
    {
      "epoch": 40.48254328697133,
      "grad_norm": 0.27345046401023865,
      "learning_rate": 5.9534487652568835e-05,
      "loss": 0.004393500089645386,
      "step": 142620
    },
    {
      "epoch": 40.48538177689469,
      "grad_norm": 3.094712734222412,
      "learning_rate": 5.9531649162645476e-05,
      "loss": 0.004758062213659287,
      "step": 142630
    },
    {
      "epoch": 40.488220266818054,
      "grad_norm": 0.22535870969295502,
      "learning_rate": 5.952881067272211e-05,
      "loss": 0.0016810072585940361,
      "step": 142640
    },
    {
      "epoch": 40.49105875674142,
      "grad_norm": 1.0743430852890015,
      "learning_rate": 5.952597218279875e-05,
      "loss": 0.005776835978031159,
      "step": 142650
    },
    {
      "epoch": 40.49389724666477,
      "grad_norm": 0.6973803639411926,
      "learning_rate": 5.9523133692875387e-05,
      "loss": 0.005281873792409897,
      "step": 142660
    },
    {
      "epoch": 40.496735736588136,
      "grad_norm": 0.02622833661735058,
      "learning_rate": 5.9520295202952035e-05,
      "loss": 0.0017042310908436775,
      "step": 142670
    },
    {
      "epoch": 40.4995742265115,
      "grad_norm": 3.8924121856689453,
      "learning_rate": 5.9517456713028676e-05,
      "loss": 0.009185433387756348,
      "step": 142680
    },
    {
      "epoch": 40.502412716434854,
      "grad_norm": 1.2406048774719238,
      "learning_rate": 5.951461822310531e-05,
      "loss": 0.007903410494327546,
      "step": 142690
    },
    {
      "epoch": 40.50525120635822,
      "grad_norm": 0.7078983187675476,
      "learning_rate": 5.951177973318195e-05,
      "loss": 0.004825123399496078,
      "step": 142700
    },
    {
      "epoch": 40.50808969628158,
      "grad_norm": 0.41203054785728455,
      "learning_rate": 5.950894124325859e-05,
      "loss": 0.010903126001358033,
      "step": 142710
    },
    {
      "epoch": 40.510928186204936,
      "grad_norm": 0.4151240885257721,
      "learning_rate": 5.950610275333522e-05,
      "loss": 0.0050274964421987535,
      "step": 142720
    },
    {
      "epoch": 40.5137666761283,
      "grad_norm": 0.245313361287117,
      "learning_rate": 5.9503264263411876e-05,
      "loss": 0.008467765152454376,
      "step": 142730
    },
    {
      "epoch": 40.51660516605166,
      "grad_norm": 0.15165676176548004,
      "learning_rate": 5.950042577348851e-05,
      "loss": 0.0008470991626381875,
      "step": 142740
    },
    {
      "epoch": 40.519443655975024,
      "grad_norm": 1.4660048484802246,
      "learning_rate": 5.9497587283565145e-05,
      "loss": 0.007968025654554367,
      "step": 142750
    },
    {
      "epoch": 40.52228214589838,
      "grad_norm": 0.2969275116920471,
      "learning_rate": 5.949474879364179e-05,
      "loss": 0.0014995524659752845,
      "step": 142760
    },
    {
      "epoch": 40.52512063582174,
      "grad_norm": 1.2491763830184937,
      "learning_rate": 5.949191030371842e-05,
      "loss": 0.0022022584453225136,
      "step": 142770
    },
    {
      "epoch": 40.527959125745106,
      "grad_norm": 0.3539761006832123,
      "learning_rate": 5.948907181379506e-05,
      "loss": 0.004901013150811195,
      "step": 142780
    },
    {
      "epoch": 40.53079761566846,
      "grad_norm": 0.2932395040988922,
      "learning_rate": 5.94862333238717e-05,
      "loss": 0.0009664127603173256,
      "step": 142790
    },
    {
      "epoch": 40.533636105591825,
      "grad_norm": 0.06991647928953171,
      "learning_rate": 5.9483394833948345e-05,
      "loss": 0.0027611522004008294,
      "step": 142800
    },
    {
      "epoch": 40.53647459551519,
      "grad_norm": 0.04482040926814079,
      "learning_rate": 5.948055634402499e-05,
      "loss": 0.0039312392473220825,
      "step": 142810
    },
    {
      "epoch": 40.53931308543854,
      "grad_norm": 10.150687217712402,
      "learning_rate": 5.947771785410162e-05,
      "loss": 0.005960384756326676,
      "step": 142820
    },
    {
      "epoch": 40.542151575361906,
      "grad_norm": 0.01725679822266102,
      "learning_rate": 5.947487936417826e-05,
      "loss": 0.003008277155458927,
      "step": 142830
    },
    {
      "epoch": 40.54499006528527,
      "grad_norm": 8.811765670776367,
      "learning_rate": 5.94720408742549e-05,
      "loss": 0.008335787057876586,
      "step": 142840
    },
    {
      "epoch": 40.54782855520863,
      "grad_norm": 0.2974540591239929,
      "learning_rate": 5.946920238433153e-05,
      "loss": 0.010828736424446105,
      "step": 142850
    },
    {
      "epoch": 40.55066704513199,
      "grad_norm": 0.5259221792221069,
      "learning_rate": 5.9466363894408173e-05,
      "loss": 0.0023151716217398644,
      "step": 142860
    },
    {
      "epoch": 40.55350553505535,
      "grad_norm": 0.025457030162215233,
      "learning_rate": 5.946352540448482e-05,
      "loss": 0.000605616346001625,
      "step": 142870
    },
    {
      "epoch": 40.55634402497871,
      "grad_norm": 0.6669510006904602,
      "learning_rate": 5.9460686914561456e-05,
      "loss": 0.0019593492150306703,
      "step": 142880
    },
    {
      "epoch": 40.55918251490207,
      "grad_norm": 1.8184717893600464,
      "learning_rate": 5.94578484246381e-05,
      "loss": 0.001833096705377102,
      "step": 142890
    },
    {
      "epoch": 40.56202100482543,
      "grad_norm": 0.07129436731338501,
      "learning_rate": 5.945500993471473e-05,
      "loss": 0.0073227077722549435,
      "step": 142900
    },
    {
      "epoch": 40.564859494748795,
      "grad_norm": 15.05155086517334,
      "learning_rate": 5.9452171444791373e-05,
      "loss": 0.00830138623714447,
      "step": 142910
    },
    {
      "epoch": 40.56769798467215,
      "grad_norm": 0.0586596354842186,
      "learning_rate": 5.944933295486801e-05,
      "loss": 0.0008426051586866379,
      "step": 142920
    },
    {
      "epoch": 40.570536474595514,
      "grad_norm": 0.06396622955799103,
      "learning_rate": 5.9446494464944656e-05,
      "loss": 0.003173793479800224,
      "step": 142930
    },
    {
      "epoch": 40.57337496451888,
      "grad_norm": 0.7168685793876648,
      "learning_rate": 5.94436559750213e-05,
      "loss": 0.0038959700614213943,
      "step": 142940
    },
    {
      "epoch": 40.57621345444224,
      "grad_norm": 0.30210697650909424,
      "learning_rate": 5.944081748509793e-05,
      "loss": 0.0015396114438772201,
      "step": 142950
    },
    {
      "epoch": 40.579051944365595,
      "grad_norm": 1.1743474006652832,
      "learning_rate": 5.9437978995174574e-05,
      "loss": 0.0014950692653656005,
      "step": 142960
    },
    {
      "epoch": 40.58189043428896,
      "grad_norm": 1.883886694908142,
      "learning_rate": 5.943514050525121e-05,
      "loss": 0.000945562869310379,
      "step": 142970
    },
    {
      "epoch": 40.58472892421232,
      "grad_norm": 0.27309176325798035,
      "learning_rate": 5.943230201532784e-05,
      "loss": 0.0006550701335072517,
      "step": 142980
    },
    {
      "epoch": 40.58756741413568,
      "grad_norm": 0.8929429054260254,
      "learning_rate": 5.9429463525404484e-05,
      "loss": 0.010310186445713044,
      "step": 142990
    },
    {
      "epoch": 40.59040590405904,
      "grad_norm": 0.5350882411003113,
      "learning_rate": 5.942662503548113e-05,
      "loss": 0.002197693847119808,
      "step": 143000
    },
    {
      "epoch": 40.59040590405904,
      "eval_accuracy": 0.9751382971959052,
      "eval_loss": 0.09259281307458878,
      "eval_runtime": 31.7159,
      "eval_samples_per_second": 495.872,
      "eval_steps_per_second": 7.756,
      "step": 143000
    },
    {
      "epoch": 40.5932443939824,
      "grad_norm": 0.37987083196640015,
      "learning_rate": 5.942378654555777e-05,
      "loss": 0.00469839945435524,
      "step": 143010
    },
    {
      "epoch": 40.596082883905765,
      "grad_norm": 9.071054458618164,
      "learning_rate": 5.942094805563441e-05,
      "loss": 0.0034980453550815583,
      "step": 143020
    },
    {
      "epoch": 40.59892137382912,
      "grad_norm": 0.37976956367492676,
      "learning_rate": 5.941810956571104e-05,
      "loss": 0.00638132244348526,
      "step": 143030
    },
    {
      "epoch": 40.601759863752484,
      "grad_norm": 0.4288770854473114,
      "learning_rate": 5.9415271075787684e-05,
      "loss": 0.0011041268706321716,
      "step": 143040
    },
    {
      "epoch": 40.60459835367585,
      "grad_norm": 0.32534340023994446,
      "learning_rate": 5.941243258586432e-05,
      "loss": 0.0007978539913892746,
      "step": 143050
    },
    {
      "epoch": 40.6074368435992,
      "grad_norm": 0.9897381663322449,
      "learning_rate": 5.940959409594096e-05,
      "loss": 0.000856412760913372,
      "step": 143060
    },
    {
      "epoch": 40.610275333522566,
      "grad_norm": 0.5289235711097717,
      "learning_rate": 5.940675560601761e-05,
      "loss": 0.0027045389637351037,
      "step": 143070
    },
    {
      "epoch": 40.61311382344593,
      "grad_norm": 0.018197203055024147,
      "learning_rate": 5.940391711609424e-05,
      "loss": 0.002272966876626015,
      "step": 143080
    },
    {
      "epoch": 40.615952313369284,
      "grad_norm": 0.2511475682258606,
      "learning_rate": 5.940107862617088e-05,
      "loss": 0.005710091441869736,
      "step": 143090
    },
    {
      "epoch": 40.61879080329265,
      "grad_norm": 0.25161951780319214,
      "learning_rate": 5.939824013624752e-05,
      "loss": 0.005178947374224663,
      "step": 143100
    },
    {
      "epoch": 40.62162929321601,
      "grad_norm": 0.30159807205200195,
      "learning_rate": 5.9395401646324154e-05,
      "loss": 0.0008851200342178345,
      "step": 143110
    },
    {
      "epoch": 40.62446778313937,
      "grad_norm": 4.385353088378906,
      "learning_rate": 5.9392563156400795e-05,
      "loss": 0.0027853500097990034,
      "step": 143120
    },
    {
      "epoch": 40.62730627306273,
      "grad_norm": 0.12101732194423676,
      "learning_rate": 5.938972466647744e-05,
      "loss": 0.0026125997304916383,
      "step": 143130
    },
    {
      "epoch": 40.63014476298609,
      "grad_norm": 0.7842867970466614,
      "learning_rate": 5.938688617655408e-05,
      "loss": 0.0013510072603821754,
      "step": 143140
    },
    {
      "epoch": 40.632983252909455,
      "grad_norm": 0.029502004384994507,
      "learning_rate": 5.938404768663072e-05,
      "loss": 0.0018030934035778046,
      "step": 143150
    },
    {
      "epoch": 40.63582174283281,
      "grad_norm": 0.04478999599814415,
      "learning_rate": 5.9381209196707354e-05,
      "loss": 0.0013331590220332147,
      "step": 143160
    },
    {
      "epoch": 40.63866023275617,
      "grad_norm": 19.59030532836914,
      "learning_rate": 5.9378370706783995e-05,
      "loss": 0.013360127806663513,
      "step": 143170
    },
    {
      "epoch": 40.641498722679536,
      "grad_norm": 0.6162864565849304,
      "learning_rate": 5.937553221686063e-05,
      "loss": 0.0010724138468503953,
      "step": 143180
    },
    {
      "epoch": 40.64433721260289,
      "grad_norm": 0.12171696871519089,
      "learning_rate": 5.9372693726937264e-05,
      "loss": 0.0011655807495117188,
      "step": 143190
    },
    {
      "epoch": 40.647175702526255,
      "grad_norm": 16.83856201171875,
      "learning_rate": 5.936985523701392e-05,
      "loss": 0.006465972959995269,
      "step": 143200
    },
    {
      "epoch": 40.65001419244962,
      "grad_norm": 8.244622230529785,
      "learning_rate": 5.9367016747090554e-05,
      "loss": 0.004117476567625999,
      "step": 143210
    },
    {
      "epoch": 40.65285268237298,
      "grad_norm": 1.1474430561065674,
      "learning_rate": 5.936417825716719e-05,
      "loss": 0.01116764172911644,
      "step": 143220
    },
    {
      "epoch": 40.655691172296336,
      "grad_norm": 0.09357480704784393,
      "learning_rate": 5.936133976724383e-05,
      "loss": 0.003266451507806778,
      "step": 143230
    },
    {
      "epoch": 40.6585296622197,
      "grad_norm": 0.09653209894895554,
      "learning_rate": 5.9358501277320464e-05,
      "loss": 0.004706088453531265,
      "step": 143240
    },
    {
      "epoch": 40.66136815214306,
      "grad_norm": 0.44743612408638,
      "learning_rate": 5.9355662787397106e-05,
      "loss": 0.0010750528424978256,
      "step": 143250
    },
    {
      "epoch": 40.66420664206642,
      "grad_norm": 2.37954044342041,
      "learning_rate": 5.935282429747374e-05,
      "loss": 0.009068956971168518,
      "step": 143260
    },
    {
      "epoch": 40.66704513198978,
      "grad_norm": 1.0571801662445068,
      "learning_rate": 5.934998580755039e-05,
      "loss": 0.0034874562174081803,
      "step": 143270
    },
    {
      "epoch": 40.669883621913144,
      "grad_norm": 14.328583717346191,
      "learning_rate": 5.934714731762703e-05,
      "loss": 0.007115014642477035,
      "step": 143280
    },
    {
      "epoch": 40.67272211183651,
      "grad_norm": 0.07091622054576874,
      "learning_rate": 5.9344308827703664e-05,
      "loss": 0.0014853779226541518,
      "step": 143290
    },
    {
      "epoch": 40.67556060175986,
      "grad_norm": 23.17186164855957,
      "learning_rate": 5.934175418677264e-05,
      "loss": 0.02061387002468109,
      "step": 143300
    },
    {
      "epoch": 40.678399091683225,
      "grad_norm": 0.5094237327575684,
      "learning_rate": 5.9338915696849286e-05,
      "loss": 0.006845498085021972,
      "step": 143310
    },
    {
      "epoch": 40.68123758160659,
      "grad_norm": 0.054741036146879196,
      "learning_rate": 5.933607720692592e-05,
      "loss": 0.010843484848737716,
      "step": 143320
    },
    {
      "epoch": 40.684076071529944,
      "grad_norm": 0.17377549409866333,
      "learning_rate": 5.933323871700256e-05,
      "loss": 0.006014581024646759,
      "step": 143330
    },
    {
      "epoch": 40.68691456145331,
      "grad_norm": 8.716096878051758,
      "learning_rate": 5.9330400227079196e-05,
      "loss": 0.009864570200443267,
      "step": 143340
    },
    {
      "epoch": 40.68975305137667,
      "grad_norm": 16.881797790527344,
      "learning_rate": 5.932756173715584e-05,
      "loss": 0.007041013985872269,
      "step": 143350
    },
    {
      "epoch": 40.692591541300025,
      "grad_norm": 0.269989550113678,
      "learning_rate": 5.932472324723247e-05,
      "loss": 0.0024568244814872743,
      "step": 143360
    },
    {
      "epoch": 40.69543003122339,
      "grad_norm": 0.391430139541626,
      "learning_rate": 5.932188475730911e-05,
      "loss": 0.0036700524389743803,
      "step": 143370
    },
    {
      "epoch": 40.69826852114675,
      "grad_norm": 6.283731460571289,
      "learning_rate": 5.9319046267385755e-05,
      "loss": 0.0034056290984153747,
      "step": 143380
    },
    {
      "epoch": 40.701107011070114,
      "grad_norm": 0.13294091820716858,
      "learning_rate": 5.9316207777462396e-05,
      "loss": 0.004032780230045318,
      "step": 143390
    },
    {
      "epoch": 40.70394550099347,
      "grad_norm": 4.740020751953125,
      "learning_rate": 5.931336928753903e-05,
      "loss": 0.008626027405261994,
      "step": 143400
    },
    {
      "epoch": 40.70678399091683,
      "grad_norm": 1.1918461322784424,
      "learning_rate": 5.931053079761567e-05,
      "loss": 0.019481292366981505,
      "step": 143410
    },
    {
      "epoch": 40.709622480840196,
      "grad_norm": 0.389034241437912,
      "learning_rate": 5.930769230769231e-05,
      "loss": 0.003907892853021622,
      "step": 143420
    },
    {
      "epoch": 40.71246097076355,
      "grad_norm": 0.27757883071899414,
      "learning_rate": 5.930485381776895e-05,
      "loss": 0.007003211230039596,
      "step": 143430
    },
    {
      "epoch": 40.715299460686914,
      "grad_norm": 0.1613476425409317,
      "learning_rate": 5.930201532784558e-05,
      "loss": 0.010487939417362212,
      "step": 143440
    },
    {
      "epoch": 40.71813795061028,
      "grad_norm": 0.48295992612838745,
      "learning_rate": 5.929917683792223e-05,
      "loss": 0.000825997069478035,
      "step": 143450
    },
    {
      "epoch": 40.72097644053363,
      "grad_norm": 0.0925155058503151,
      "learning_rate": 5.929633834799887e-05,
      "loss": 0.009538304805755616,
      "step": 143460
    },
    {
      "epoch": 40.723814930456996,
      "grad_norm": 1.1203057765960693,
      "learning_rate": 5.929349985807551e-05,
      "loss": 0.004477202147245407,
      "step": 143470
    },
    {
      "epoch": 40.72665342038036,
      "grad_norm": 0.14477191865444183,
      "learning_rate": 5.929066136815214e-05,
      "loss": 0.006990453600883484,
      "step": 143480
    },
    {
      "epoch": 40.72949191030372,
      "grad_norm": 0.19251132011413574,
      "learning_rate": 5.928782287822878e-05,
      "loss": 0.0024141045287251472,
      "step": 143490
    },
    {
      "epoch": 40.73233040022708,
      "grad_norm": 0.47015950083732605,
      "learning_rate": 5.928498438830542e-05,
      "loss": 0.024002668261528016,
      "step": 143500
    },
    {
      "epoch": 40.73233040022708,
      "eval_accuracy": 0.9729128250778916,
      "eval_loss": 0.09465668350458145,
      "eval_runtime": 32.0739,
      "eval_samples_per_second": 490.336,
      "eval_steps_per_second": 7.67,
      "step": 143500
    },
    {
      "epoch": 40.73516889015044,
      "grad_norm": 0.09650905430316925,
      "learning_rate": 5.9282145898382066e-05,
      "loss": 0.0011524861678481103,
      "step": 143510
    },
    {
      "epoch": 40.7380073800738,
      "grad_norm": 0.7886379957199097,
      "learning_rate": 5.927930740845871e-05,
      "loss": 0.0030128732323646545,
      "step": 143520
    },
    {
      "epoch": 40.74084586999716,
      "grad_norm": 1.2624696493148804,
      "learning_rate": 5.927646891853534e-05,
      "loss": 0.004453198611736297,
      "step": 143530
    },
    {
      "epoch": 40.74368435992052,
      "grad_norm": 2.1769087314605713,
      "learning_rate": 5.927363042861198e-05,
      "loss": 0.004307749122381211,
      "step": 143540
    },
    {
      "epoch": 40.746522849843885,
      "grad_norm": 0.29042473435401917,
      "learning_rate": 5.927079193868862e-05,
      "loss": 0.007949556410312652,
      "step": 143550
    },
    {
      "epoch": 40.74936133976724,
      "grad_norm": 0.4124404489994049,
      "learning_rate": 5.926795344876526e-05,
      "loss": 0.00828683227300644,
      "step": 143560
    },
    {
      "epoch": 40.7521998296906,
      "grad_norm": 0.41829487681388855,
      "learning_rate": 5.9265114958841894e-05,
      "loss": 0.004325699433684349,
      "step": 143570
    },
    {
      "epoch": 40.755038319613966,
      "grad_norm": 6.501859664916992,
      "learning_rate": 5.926227646891854e-05,
      "loss": 0.0022047311067581177,
      "step": 143580
    },
    {
      "epoch": 40.75787680953733,
      "grad_norm": 0.3197261393070221,
      "learning_rate": 5.925943797899518e-05,
      "loss": 0.001522185280919075,
      "step": 143590
    },
    {
      "epoch": 40.760715299460685,
      "grad_norm": 0.036058973520994186,
      "learning_rate": 5.925659948907182e-05,
      "loss": 0.004107113182544708,
      "step": 143600
    },
    {
      "epoch": 40.76355378938405,
      "grad_norm": 1.0390346050262451,
      "learning_rate": 5.925376099914845e-05,
      "loss": 0.0022269533947110176,
      "step": 143610
    },
    {
      "epoch": 40.76639227930741,
      "grad_norm": 16.896867752075195,
      "learning_rate": 5.9250922509225094e-05,
      "loss": 0.010248854756355286,
      "step": 143620
    },
    {
      "epoch": 40.76923076923077,
      "grad_norm": 0.5223643183708191,
      "learning_rate": 5.924808401930173e-05,
      "loss": 0.0008883951231837273,
      "step": 143630
    },
    {
      "epoch": 40.77206925915413,
      "grad_norm": 17.652469635009766,
      "learning_rate": 5.924524552937837e-05,
      "loss": 0.013144730031490326,
      "step": 143640
    },
    {
      "epoch": 40.77490774907749,
      "grad_norm": 0.915235698223114,
      "learning_rate": 5.924240703945502e-05,
      "loss": 0.001236843317747116,
      "step": 143650
    },
    {
      "epoch": 40.77774623900085,
      "grad_norm": 0.07645450532436371,
      "learning_rate": 5.923956854953165e-05,
      "loss": 0.003057604283094406,
      "step": 143660
    },
    {
      "epoch": 40.78058472892421,
      "grad_norm": 1.5591310262680054,
      "learning_rate": 5.9236730059608294e-05,
      "loss": 0.006219648569822311,
      "step": 143670
    },
    {
      "epoch": 40.783423218847574,
      "grad_norm": 1.6420620679855347,
      "learning_rate": 5.923389156968493e-05,
      "loss": 0.0022735850885510444,
      "step": 143680
    },
    {
      "epoch": 40.78626170877094,
      "grad_norm": 16.00141143798828,
      "learning_rate": 5.923105307976157e-05,
      "loss": 0.0214938759803772,
      "step": 143690
    },
    {
      "epoch": 40.78910019869429,
      "grad_norm": 0.6556671261787415,
      "learning_rate": 5.9228214589838204e-05,
      "loss": 0.004216181486845017,
      "step": 143700
    },
    {
      "epoch": 40.791938688617655,
      "grad_norm": 1.5717039108276367,
      "learning_rate": 5.922537609991485e-05,
      "loss": 0.0013555359095335008,
      "step": 143710
    },
    {
      "epoch": 40.79477717854102,
      "grad_norm": 0.12915366888046265,
      "learning_rate": 5.9222537609991494e-05,
      "loss": 0.001456248201429844,
      "step": 143720
    },
    {
      "epoch": 40.797615668464374,
      "grad_norm": 0.18152378499507904,
      "learning_rate": 5.921969912006813e-05,
      "loss": 0.0017806293442845344,
      "step": 143730
    },
    {
      "epoch": 40.80045415838774,
      "grad_norm": 0.02659328281879425,
      "learning_rate": 5.921686063014476e-05,
      "loss": 0.0010754212737083435,
      "step": 143740
    },
    {
      "epoch": 40.8032926483111,
      "grad_norm": 0.06944055110216141,
      "learning_rate": 5.9214022140221405e-05,
      "loss": 0.000644514337182045,
      "step": 143750
    },
    {
      "epoch": 40.80613113823446,
      "grad_norm": 0.09241548180580139,
      "learning_rate": 5.921118365029804e-05,
      "loss": 0.0017124557867646216,
      "step": 143760
    },
    {
      "epoch": 40.80896962815782,
      "grad_norm": 6.628686904907227,
      "learning_rate": 5.920834516037468e-05,
      "loss": 0.0020653253421187403,
      "step": 143770
    },
    {
      "epoch": 40.81180811808118,
      "grad_norm": 0.5456545948982239,
      "learning_rate": 5.920550667045133e-05,
      "loss": 0.0012607796117663384,
      "step": 143780
    },
    {
      "epoch": 40.814646608004544,
      "grad_norm": 17.83376693725586,
      "learning_rate": 5.920266818052796e-05,
      "loss": 0.009623174369335175,
      "step": 143790
    },
    {
      "epoch": 40.8174850979279,
      "grad_norm": 0.06481852382421494,
      "learning_rate": 5.9199829690604605e-05,
      "loss": 0.0019452426582574843,
      "step": 143800
    },
    {
      "epoch": 40.82032358785126,
      "grad_norm": 0.07820414751768112,
      "learning_rate": 5.919699120068124e-05,
      "loss": 0.001421624980866909,
      "step": 143810
    },
    {
      "epoch": 40.823162077774626,
      "grad_norm": 0.0670999139547348,
      "learning_rate": 5.919415271075788e-05,
      "loss": 0.001543007232248783,
      "step": 143820
    },
    {
      "epoch": 40.82600056769798,
      "grad_norm": 0.13607297837734222,
      "learning_rate": 5.9191314220834515e-05,
      "loss": 0.0052397135645151135,
      "step": 143830
    },
    {
      "epoch": 40.828839057621344,
      "grad_norm": 0.1956016719341278,
      "learning_rate": 5.918847573091115e-05,
      "loss": 0.0017496682703495026,
      "step": 143840
    },
    {
      "epoch": 40.83167754754471,
      "grad_norm": 5.186715602874756,
      "learning_rate": 5.91856372409878e-05,
      "loss": 0.00398058295249939,
      "step": 143850
    },
    {
      "epoch": 40.83451603746807,
      "grad_norm": 0.18321655690670013,
      "learning_rate": 5.918279875106444e-05,
      "loss": 0.0027581514790654183,
      "step": 143860
    },
    {
      "epoch": 40.837354527391426,
      "grad_norm": 1.7091448307037354,
      "learning_rate": 5.9179960261141074e-05,
      "loss": 0.0009309165179729462,
      "step": 143870
    },
    {
      "epoch": 40.84019301731479,
      "grad_norm": 7.4326653480529785,
      "learning_rate": 5.9177121771217715e-05,
      "loss": 0.004746191948652267,
      "step": 143880
    },
    {
      "epoch": 40.84303150723815,
      "grad_norm": 0.03949809819459915,
      "learning_rate": 5.917428328129435e-05,
      "loss": 0.0022472869604825974,
      "step": 143890
    },
    {
      "epoch": 40.84586999716151,
      "grad_norm": 0.11706339567899704,
      "learning_rate": 5.917144479137099e-05,
      "loss": 0.0025604210793972017,
      "step": 143900
    },
    {
      "epoch": 40.84870848708487,
      "grad_norm": 4.989575386047363,
      "learning_rate": 5.916860630144764e-05,
      "loss": 0.0019094107672572135,
      "step": 143910
    },
    {
      "epoch": 40.85154697700823,
      "grad_norm": 0.0716421976685524,
      "learning_rate": 5.9165767811524274e-05,
      "loss": 0.0038660924881696703,
      "step": 143920
    },
    {
      "epoch": 40.85438546693159,
      "grad_norm": 0.13984334468841553,
      "learning_rate": 5.9162929321600915e-05,
      "loss": 0.00573047436773777,
      "step": 143930
    },
    {
      "epoch": 40.85722395685495,
      "grad_norm": 0.0861579105257988,
      "learning_rate": 5.916009083167755e-05,
      "loss": 0.004090435802936554,
      "step": 143940
    },
    {
      "epoch": 40.860062446778315,
      "grad_norm": 0.12914298474788666,
      "learning_rate": 5.9157252341754185e-05,
      "loss": 0.006060747802257538,
      "step": 143950
    },
    {
      "epoch": 40.86290093670168,
      "grad_norm": 0.5987483859062195,
      "learning_rate": 5.9154413851830826e-05,
      "loss": 0.0017599912360310554,
      "step": 143960
    },
    {
      "epoch": 40.865739426625034,
      "grad_norm": 0.5651119947433472,
      "learning_rate": 5.915157536190746e-05,
      "loss": 0.0009978052228689193,
      "step": 143970
    },
    {
      "epoch": 40.868577916548396,
      "grad_norm": 3.598417043685913,
      "learning_rate": 5.914873687198411e-05,
      "loss": 0.002420127019286156,
      "step": 143980
    },
    {
      "epoch": 40.87141640647176,
      "grad_norm": 0.5995482802391052,
      "learning_rate": 5.914589838206075e-05,
      "loss": 0.002202806994318962,
      "step": 143990
    },
    {
      "epoch": 40.874254896395115,
      "grad_norm": 0.7933487296104431,
      "learning_rate": 5.9143059892137385e-05,
      "loss": 0.0027261780574917792,
      "step": 144000
    },
    {
      "epoch": 40.874254896395115,
      "eval_accuracy": 0.9737394290074395,
      "eval_loss": 0.09021281450986862,
      "eval_runtime": 31.4668,
      "eval_samples_per_second": 499.796,
      "eval_steps_per_second": 7.818,
      "step": 144000
    },
    {
      "epoch": 40.87709338631848,
      "grad_norm": 1.2335987091064453,
      "learning_rate": 5.9140221402214026e-05,
      "loss": 0.004290033876895904,
      "step": 144010
    },
    {
      "epoch": 40.87993187624184,
      "grad_norm": 0.2803191840648651,
      "learning_rate": 5.913738291229066e-05,
      "loss": 0.003197193518280983,
      "step": 144020
    },
    {
      "epoch": 40.8827703661652,
      "grad_norm": 0.6220667958259583,
      "learning_rate": 5.91345444223673e-05,
      "loss": 0.003210219740867615,
      "step": 144030
    },
    {
      "epoch": 40.88560885608856,
      "grad_norm": 19.223304748535156,
      "learning_rate": 5.913170593244395e-05,
      "loss": 0.01699031889438629,
      "step": 144040
    },
    {
      "epoch": 40.88844734601192,
      "grad_norm": 0.05284702032804489,
      "learning_rate": 5.9128867442520585e-05,
      "loss": 0.0030202146619558333,
      "step": 144050
    },
    {
      "epoch": 40.891285835935285,
      "grad_norm": 0.20078560709953308,
      "learning_rate": 5.9126028952597226e-05,
      "loss": 0.002349603921175003,
      "step": 144060
    },
    {
      "epoch": 40.89412432585864,
      "grad_norm": 0.7740545272827148,
      "learning_rate": 5.912319046267386e-05,
      "loss": 0.0013783114030957223,
      "step": 144070
    },
    {
      "epoch": 40.896962815782004,
      "grad_norm": 1.0110328197479248,
      "learning_rate": 5.9120351972750495e-05,
      "loss": 0.0038494981825351717,
      "step": 144080
    },
    {
      "epoch": 40.89980130570537,
      "grad_norm": 0.07562289386987686,
      "learning_rate": 5.911751348282714e-05,
      "loss": 0.0016973508521914483,
      "step": 144090
    },
    {
      "epoch": 40.90263979562872,
      "grad_norm": 0.3942168056964874,
      "learning_rate": 5.911467499290377e-05,
      "loss": 0.0029261182993650437,
      "step": 144100
    },
    {
      "epoch": 40.905478285552086,
      "grad_norm": 0.034911178052425385,
      "learning_rate": 5.911183650298042e-05,
      "loss": 0.0006117725744843483,
      "step": 144110
    },
    {
      "epoch": 40.90831677547545,
      "grad_norm": 0.12756870687007904,
      "learning_rate": 5.910899801305706e-05,
      "loss": 0.0014733750373125075,
      "step": 144120
    },
    {
      "epoch": 40.91115526539881,
      "grad_norm": 13.804348945617676,
      "learning_rate": 5.9106159523133696e-05,
      "loss": 0.006341956555843353,
      "step": 144130
    },
    {
      "epoch": 40.91399375532217,
      "grad_norm": 0.037310294806957245,
      "learning_rate": 5.910332103321034e-05,
      "loss": 0.0033818788826465605,
      "step": 144140
    },
    {
      "epoch": 40.91683224524553,
      "grad_norm": 0.13073444366455078,
      "learning_rate": 5.910048254328697e-05,
      "loss": 0.0013555899262428285,
      "step": 144150
    },
    {
      "epoch": 40.91967073516889,
      "grad_norm": 0.06129813939332962,
      "learning_rate": 5.909764405336361e-05,
      "loss": 0.0014847632497549056,
      "step": 144160
    },
    {
      "epoch": 40.92250922509225,
      "grad_norm": 0.08728554099798203,
      "learning_rate": 5.909480556344025e-05,
      "loss": 0.0020656703040003777,
      "step": 144170
    },
    {
      "epoch": 40.92534771501561,
      "grad_norm": 6.159107208251953,
      "learning_rate": 5.9091967073516896e-05,
      "loss": 0.0031618215143680573,
      "step": 144180
    },
    {
      "epoch": 40.928186204938974,
      "grad_norm": 0.3490411341190338,
      "learning_rate": 5.908912858359354e-05,
      "loss": 0.0013254201039671898,
      "step": 144190
    },
    {
      "epoch": 40.93102469486233,
      "grad_norm": 0.09766576439142227,
      "learning_rate": 5.908629009367017e-05,
      "loss": 0.0020377801731228827,
      "step": 144200
    },
    {
      "epoch": 40.93386318478569,
      "grad_norm": 0.9051247239112854,
      "learning_rate": 5.9083451603746806e-05,
      "loss": 0.00801626518368721,
      "step": 144210
    },
    {
      "epoch": 40.936701674709056,
      "grad_norm": 0.4656331241130829,
      "learning_rate": 5.908061311382345e-05,
      "loss": 0.001072445698082447,
      "step": 144220
    },
    {
      "epoch": 40.93954016463242,
      "grad_norm": 11.59216022491455,
      "learning_rate": 5.907777462390008e-05,
      "loss": 0.007193699479103088,
      "step": 144230
    },
    {
      "epoch": 40.942378654555775,
      "grad_norm": 0.25008267164230347,
      "learning_rate": 5.907493613397673e-05,
      "loss": 0.0019127871841192246,
      "step": 144240
    },
    {
      "epoch": 40.94521714447914,
      "grad_norm": 0.6627295017242432,
      "learning_rate": 5.907209764405337e-05,
      "loss": 0.005394937843084336,
      "step": 144250
    },
    {
      "epoch": 40.9480556344025,
      "grad_norm": 2.252340316772461,
      "learning_rate": 5.9069259154130006e-05,
      "loss": 0.0020184533670544623,
      "step": 144260
    },
    {
      "epoch": 40.950894124325856,
      "grad_norm": 5.645776748657227,
      "learning_rate": 5.906642066420665e-05,
      "loss": 0.0031109314411878588,
      "step": 144270
    },
    {
      "epoch": 40.95373261424922,
      "grad_norm": 0.11262688785791397,
      "learning_rate": 5.906358217428328e-05,
      "loss": 0.002817433327436447,
      "step": 144280
    },
    {
      "epoch": 40.95657110417258,
      "grad_norm": 1.9388688802719116,
      "learning_rate": 5.9060743684359924e-05,
      "loss": 0.00715155228972435,
      "step": 144290
    },
    {
      "epoch": 40.95940959409594,
      "grad_norm": 3.580753803253174,
      "learning_rate": 5.905790519443656e-05,
      "loss": 0.004485449939966202,
      "step": 144300
    },
    {
      "epoch": 40.9622480840193,
      "grad_norm": 0.1425146758556366,
      "learning_rate": 5.9055066704513206e-05,
      "loss": 0.002665933407843113,
      "step": 144310
    },
    {
      "epoch": 40.96508657394266,
      "grad_norm": 0.10770668834447861,
      "learning_rate": 5.905222821458984e-05,
      "loss": 0.003633125871419907,
      "step": 144320
    },
    {
      "epoch": 40.967925063866026,
      "grad_norm": 0.26043370366096497,
      "learning_rate": 5.904938972466648e-05,
      "loss": 0.0013338970020413399,
      "step": 144330
    },
    {
      "epoch": 40.97076355378938,
      "grad_norm": 0.18556566536426544,
      "learning_rate": 5.904655123474312e-05,
      "loss": 0.01361810863018036,
      "step": 144340
    },
    {
      "epoch": 40.973602043712745,
      "grad_norm": 2.348299503326416,
      "learning_rate": 5.904371274481976e-05,
      "loss": 0.01884114444255829,
      "step": 144350
    },
    {
      "epoch": 40.97644053363611,
      "grad_norm": 6.988702774047852,
      "learning_rate": 5.904087425489639e-05,
      "loss": 0.0031696990132331847,
      "step": 144360
    },
    {
      "epoch": 40.979279023559464,
      "grad_norm": 0.3378346562385559,
      "learning_rate": 5.9038035764973034e-05,
      "loss": 0.007236155867576599,
      "step": 144370
    },
    {
      "epoch": 40.98211751348283,
      "grad_norm": 1.0769144296646118,
      "learning_rate": 5.903519727504968e-05,
      "loss": 0.009124004095792771,
      "step": 144380
    },
    {
      "epoch": 40.98495600340619,
      "grad_norm": 0.3540835380554199,
      "learning_rate": 5.903235878512632e-05,
      "loss": 0.0022158462554216383,
      "step": 144390
    },
    {
      "epoch": 40.987794493329545,
      "grad_norm": 0.15500855445861816,
      "learning_rate": 5.902952029520296e-05,
      "loss": 0.001619946025311947,
      "step": 144400
    },
    {
      "epoch": 40.99063298325291,
      "grad_norm": 1.889005184173584,
      "learning_rate": 5.902668180527959e-05,
      "loss": 0.00350106880068779,
      "step": 144410
    },
    {
      "epoch": 40.99347147317627,
      "grad_norm": 0.9555891752243042,
      "learning_rate": 5.902384331535623e-05,
      "loss": 0.0017476914450526237,
      "step": 144420
    },
    {
      "epoch": 40.996309963099634,
      "grad_norm": 0.1884075254201889,
      "learning_rate": 5.902100482543287e-05,
      "loss": 0.006235598027706147,
      "step": 144430
    },
    {
      "epoch": 40.99914845302299,
      "grad_norm": 1.1192851066589355,
      "learning_rate": 5.901816633550952e-05,
      "loss": 0.0019264807924628258,
      "step": 144440
    },
    {
      "epoch": 41.00198694294635,
      "grad_norm": 0.46357545256614685,
      "learning_rate": 5.901532784558615e-05,
      "loss": 0.0009607197716832161,
      "step": 144450
    },
    {
      "epoch": 41.004825432869715,
      "grad_norm": 0.15065976977348328,
      "learning_rate": 5.901248935566279e-05,
      "loss": 0.0011963650584220886,
      "step": 144460
    },
    {
      "epoch": 41.00766392279307,
      "grad_norm": 0.2063741534948349,
      "learning_rate": 5.900965086573943e-05,
      "loss": 0.0011081535369157792,
      "step": 144470
    },
    {
      "epoch": 41.010502412716434,
      "grad_norm": 2.1283223628997803,
      "learning_rate": 5.900681237581607e-05,
      "loss": 0.007787109911441803,
      "step": 144480
    },
    {
      "epoch": 41.0133409026398,
      "grad_norm": 0.027833949774503708,
      "learning_rate": 5.9003973885892704e-05,
      "loss": 0.003902086988091469,
      "step": 144490
    },
    {
      "epoch": 41.01617939256316,
      "grad_norm": 0.1856052279472351,
      "learning_rate": 5.9001135395969345e-05,
      "loss": 0.00402202382683754,
      "step": 144500
    },
    {
      "epoch": 41.01617939256316,
      "eval_accuracy": 0.9760920709607681,
      "eval_loss": 0.0826275423169136,
      "eval_runtime": 31.6204,
      "eval_samples_per_second": 497.368,
      "eval_steps_per_second": 7.78,
      "step": 144500
    },
    {
      "epoch": 41.019017882486516,
      "grad_norm": 0.2354753464460373,
      "learning_rate": 5.899829690604599e-05,
      "loss": 0.0023992141708731652,
      "step": 144510
    },
    {
      "epoch": 41.02185637240988,
      "grad_norm": 0.15165922045707703,
      "learning_rate": 5.899545841612263e-05,
      "loss": 0.004209460318088531,
      "step": 144520
    },
    {
      "epoch": 41.02469486233324,
      "grad_norm": 0.3885304629802704,
      "learning_rate": 5.899261992619927e-05,
      "loss": 0.0012431671842932702,
      "step": 144530
    },
    {
      "epoch": 41.0275333522566,
      "grad_norm": 0.042866360396146774,
      "learning_rate": 5.8989781436275904e-05,
      "loss": 0.004140231758356094,
      "step": 144540
    },
    {
      "epoch": 41.03037184217996,
      "grad_norm": 0.03467678278684616,
      "learning_rate": 5.898694294635254e-05,
      "loss": 0.00139390230178833,
      "step": 144550
    },
    {
      "epoch": 41.03321033210332,
      "grad_norm": 0.38611334562301636,
      "learning_rate": 5.898410445642918e-05,
      "loss": 0.0009662073105573654,
      "step": 144560
    },
    {
      "epoch": 41.03604882202668,
      "grad_norm": 0.3164624273777008,
      "learning_rate": 5.8981265966505814e-05,
      "loss": 0.00044748857617378236,
      "step": 144570
    },
    {
      "epoch": 41.03888731195004,
      "grad_norm": 0.07682400196790695,
      "learning_rate": 5.897842747658246e-05,
      "loss": 0.002089514955878258,
      "step": 144580
    },
    {
      "epoch": 41.041725801873405,
      "grad_norm": 0.41576650738716125,
      "learning_rate": 5.8975588986659104e-05,
      "loss": 0.002344292029738426,
      "step": 144590
    },
    {
      "epoch": 41.04456429179677,
      "grad_norm": 0.11612515896558762,
      "learning_rate": 5.897275049673574e-05,
      "loss": 0.019361433386802674,
      "step": 144600
    },
    {
      "epoch": 41.04740278172012,
      "grad_norm": 0.18328799307346344,
      "learning_rate": 5.896991200681238e-05,
      "loss": 0.006509527564048767,
      "step": 144610
    },
    {
      "epoch": 41.050241271643486,
      "grad_norm": 0.043397288769483566,
      "learning_rate": 5.8967073516889014e-05,
      "loss": 0.00677272081375122,
      "step": 144620
    },
    {
      "epoch": 41.05307976156685,
      "grad_norm": 0.7659888863563538,
      "learning_rate": 5.8964235026965656e-05,
      "loss": 0.0016362672671675683,
      "step": 144630
    },
    {
      "epoch": 41.055918251490205,
      "grad_norm": 0.47028297185897827,
      "learning_rate": 5.8961396537042304e-05,
      "loss": 0.0068118274211883545,
      "step": 144640
    },
    {
      "epoch": 41.05875674141357,
      "grad_norm": 0.030550427734851837,
      "learning_rate": 5.895855804711894e-05,
      "loss": 0.0008579207584261894,
      "step": 144650
    },
    {
      "epoch": 41.06159523133693,
      "grad_norm": 15.132512092590332,
      "learning_rate": 5.895571955719558e-05,
      "loss": 0.01919366419315338,
      "step": 144660
    },
    {
      "epoch": 41.064433721260286,
      "grad_norm": 21.779508590698242,
      "learning_rate": 5.8952881067272215e-05,
      "loss": 0.01040000021457672,
      "step": 144670
    },
    {
      "epoch": 41.06727221118365,
      "grad_norm": 0.026017799973487854,
      "learning_rate": 5.895004257734885e-05,
      "loss": 0.0019133087247610093,
      "step": 144680
    },
    {
      "epoch": 41.07011070110701,
      "grad_norm": 0.6309369206428528,
      "learning_rate": 5.894720408742549e-05,
      "loss": 0.0006341924890875816,
      "step": 144690
    },
    {
      "epoch": 41.072949191030375,
      "grad_norm": 0.012568322941660881,
      "learning_rate": 5.8944365597502125e-05,
      "loss": 0.0006431559100747109,
      "step": 144700
    },
    {
      "epoch": 41.07578768095373,
      "grad_norm": 0.0832361951470375,
      "learning_rate": 5.894152710757877e-05,
      "loss": 0.00044695138931274416,
      "step": 144710
    },
    {
      "epoch": 41.078626170877094,
      "grad_norm": 0.7290443778038025,
      "learning_rate": 5.8938688617655415e-05,
      "loss": 0.0009395722299814224,
      "step": 144720
    },
    {
      "epoch": 41.08146466080046,
      "grad_norm": 0.05872228369116783,
      "learning_rate": 5.893585012773205e-05,
      "loss": 0.0034778788685798645,
      "step": 144730
    },
    {
      "epoch": 41.08430315072381,
      "grad_norm": 0.0184700358659029,
      "learning_rate": 5.893301163780869e-05,
      "loss": 0.0009902486577630042,
      "step": 144740
    },
    {
      "epoch": 41.087141640647175,
      "grad_norm": 2.1326634883880615,
      "learning_rate": 5.8930173147885325e-05,
      "loss": 0.00894305258989334,
      "step": 144750
    },
    {
      "epoch": 41.08998013057054,
      "grad_norm": 0.19816449284553528,
      "learning_rate": 5.8927334657961967e-05,
      "loss": 0.006394730508327484,
      "step": 144760
    },
    {
      "epoch": 41.092818620493894,
      "grad_norm": 0.10723547637462616,
      "learning_rate": 5.89244961680386e-05,
      "loss": 0.009548346698284148,
      "step": 144770
    },
    {
      "epoch": 41.09565711041726,
      "grad_norm": 0.2586078941822052,
      "learning_rate": 5.892165767811525e-05,
      "loss": 0.0029551828280091285,
      "step": 144780
    },
    {
      "epoch": 41.09849560034062,
      "grad_norm": 0.006448034197092056,
      "learning_rate": 5.8918819188191884e-05,
      "loss": 0.0050257608294487,
      "step": 144790
    },
    {
      "epoch": 41.10133409026398,
      "grad_norm": 0.2986944019794464,
      "learning_rate": 5.8915980698268525e-05,
      "loss": 0.008844950050115586,
      "step": 144800
    },
    {
      "epoch": 41.10417258018734,
      "grad_norm": 0.00902221817523241,
      "learning_rate": 5.891314220834516e-05,
      "loss": 0.007380933314561844,
      "step": 144810
    },
    {
      "epoch": 41.1070110701107,
      "grad_norm": 0.11151455342769623,
      "learning_rate": 5.89103037184218e-05,
      "loss": 0.002139158546924591,
      "step": 144820
    },
    {
      "epoch": 41.109849560034064,
      "grad_norm": 1.2141203880310059,
      "learning_rate": 5.8907465228498436e-05,
      "loss": 0.009808076173067093,
      "step": 144830
    },
    {
      "epoch": 41.11268804995742,
      "grad_norm": 7.722042083740234,
      "learning_rate": 5.8904626738575084e-05,
      "loss": 0.025947058200836183,
      "step": 144840
    },
    {
      "epoch": 41.11552653988078,
      "grad_norm": 1.430345892906189,
      "learning_rate": 5.8901788248651725e-05,
      "loss": 0.006334872543811798,
      "step": 144850
    },
    {
      "epoch": 41.118365029804146,
      "grad_norm": 0.153959721326828,
      "learning_rate": 5.889894975872836e-05,
      "loss": 0.0007213087752461434,
      "step": 144860
    },
    {
      "epoch": 41.1212035197275,
      "grad_norm": 0.7330533862113953,
      "learning_rate": 5.8896111268805e-05,
      "loss": 0.0019706562161445618,
      "step": 144870
    },
    {
      "epoch": 41.124042009650864,
      "grad_norm": 0.06862904131412506,
      "learning_rate": 5.8893272778881636e-05,
      "loss": 0.007902867347002029,
      "step": 144880
    },
    {
      "epoch": 41.12688049957423,
      "grad_norm": 14.054582595825195,
      "learning_rate": 5.889043428895827e-05,
      "loss": 0.01762775331735611,
      "step": 144890
    },
    {
      "epoch": 41.12971898949759,
      "grad_norm": 0.5380801558494568,
      "learning_rate": 5.888759579903491e-05,
      "loss": 0.002777547389268875,
      "step": 144900
    },
    {
      "epoch": 41.132557479420946,
      "grad_norm": 0.7241384387016296,
      "learning_rate": 5.888475730911156e-05,
      "loss": 0.0009766379371285439,
      "step": 144910
    },
    {
      "epoch": 41.13539596934431,
      "grad_norm": 0.174872025847435,
      "learning_rate": 5.8881918819188195e-05,
      "loss": 0.0022920602932572365,
      "step": 144920
    },
    {
      "epoch": 41.13823445926767,
      "grad_norm": 0.44176143407821655,
      "learning_rate": 5.8879080329264836e-05,
      "loss": 0.010026943683624268,
      "step": 144930
    },
    {
      "epoch": 41.14107294919103,
      "grad_norm": 9.389435768127441,
      "learning_rate": 5.887624183934147e-05,
      "loss": 0.00444253571331501,
      "step": 144940
    },
    {
      "epoch": 41.14391143911439,
      "grad_norm": 0.1750490367412567,
      "learning_rate": 5.887340334941811e-05,
      "loss": 0.0013035297393798827,
      "step": 144950
    },
    {
      "epoch": 41.14674992903775,
      "grad_norm": 0.43471288681030273,
      "learning_rate": 5.887056485949475e-05,
      "loss": 0.0019206492230296134,
      "step": 144960
    },
    {
      "epoch": 41.149588418961116,
      "grad_norm": 0.39026525616645813,
      "learning_rate": 5.886772636957139e-05,
      "loss": 0.009408944100141526,
      "step": 144970
    },
    {
      "epoch": 41.15242690888447,
      "grad_norm": 0.15557347238063812,
      "learning_rate": 5.8864887879648036e-05,
      "loss": 0.012084005773067475,
      "step": 144980
    },
    {
      "epoch": 41.155265398807835,
      "grad_norm": 0.24035830795764923,
      "learning_rate": 5.886204938972467e-05,
      "loss": 0.0006706902757287025,
      "step": 144990
    },
    {
      "epoch": 41.1581038887312,
      "grad_norm": 0.021844398230314255,
      "learning_rate": 5.885921089980131e-05,
      "loss": 0.0008088205009698868,
      "step": 145000
    },
    {
      "epoch": 41.1581038887312,
      "eval_accuracy": 0.9741845234310421,
      "eval_loss": 0.09154852479696274,
      "eval_runtime": 31.6504,
      "eval_samples_per_second": 496.898,
      "eval_steps_per_second": 7.772,
      "step": 145000
    },
    {
      "epoch": 41.16094237865455,
      "grad_norm": 0.04944935068488121,
      "learning_rate": 5.885637240987795e-05,
      "loss": 0.002171516790986061,
      "step": 145010
    },
    {
      "epoch": 41.163780868577916,
      "grad_norm": 5.444542407989502,
      "learning_rate": 5.885353391995458e-05,
      "loss": 0.0021147722378373145,
      "step": 145020
    },
    {
      "epoch": 41.16661935850128,
      "grad_norm": 0.28274980187416077,
      "learning_rate": 5.885069543003122e-05,
      "loss": 0.0013165883719921112,
      "step": 145030
    },
    {
      "epoch": 41.169457848424635,
      "grad_norm": 1.0897352695465088,
      "learning_rate": 5.884785694010787e-05,
      "loss": 0.0014064969494938851,
      "step": 145040
    },
    {
      "epoch": 41.172296338348,
      "grad_norm": 0.014121860265731812,
      "learning_rate": 5.8845018450184505e-05,
      "loss": 0.0024939673021435738,
      "step": 145050
    },
    {
      "epoch": 41.17513482827136,
      "grad_norm": 0.0752185583114624,
      "learning_rate": 5.884217996026115e-05,
      "loss": 0.0042482312768697735,
      "step": 145060
    },
    {
      "epoch": 41.177973318194724,
      "grad_norm": 0.6332480311393738,
      "learning_rate": 5.883934147033778e-05,
      "loss": 0.0029303392395377157,
      "step": 145070
    },
    {
      "epoch": 41.18081180811808,
      "grad_norm": 0.1931389719247818,
      "learning_rate": 5.883650298041442e-05,
      "loss": 0.00168503075838089,
      "step": 145080
    },
    {
      "epoch": 41.18365029804144,
      "grad_norm": 2.711085081100464,
      "learning_rate": 5.883366449049106e-05,
      "loss": 0.0037287812680006026,
      "step": 145090
    },
    {
      "epoch": 41.186488787964805,
      "grad_norm": 2.196972608566284,
      "learning_rate": 5.88308260005677e-05,
      "loss": 0.0017890317365527153,
      "step": 145100
    },
    {
      "epoch": 41.18932727788816,
      "grad_norm": 0.3981809616088867,
      "learning_rate": 5.882798751064435e-05,
      "loss": 0.010249429941177368,
      "step": 145110
    },
    {
      "epoch": 41.192165767811524,
      "grad_norm": 20.459148406982422,
      "learning_rate": 5.882514902072098e-05,
      "loss": 0.013088300824165344,
      "step": 145120
    },
    {
      "epoch": 41.19500425773489,
      "grad_norm": 0.2179378718137741,
      "learning_rate": 5.882231053079762e-05,
      "loss": 0.0017420345917344092,
      "step": 145130
    },
    {
      "epoch": 41.19784274765824,
      "grad_norm": 0.22374498844146729,
      "learning_rate": 5.881947204087426e-05,
      "loss": 0.0007976058870553971,
      "step": 145140
    },
    {
      "epoch": 41.200681237581605,
      "grad_norm": 0.10189680755138397,
      "learning_rate": 5.881663355095089e-05,
      "loss": 0.0005658766254782677,
      "step": 145150
    },
    {
      "epoch": 41.20351972750497,
      "grad_norm": 1.3007010221481323,
      "learning_rate": 5.8813795061027533e-05,
      "loss": 0.002078183926641941,
      "step": 145160
    },
    {
      "epoch": 41.20635821742833,
      "grad_norm": 0.15351806581020355,
      "learning_rate": 5.881095657110417e-05,
      "loss": 0.0015081293880939485,
      "step": 145170
    },
    {
      "epoch": 41.20919670735169,
      "grad_norm": 12.227166175842285,
      "learning_rate": 5.8808118081180816e-05,
      "loss": 0.0032019950449466705,
      "step": 145180
    },
    {
      "epoch": 41.21203519727505,
      "grad_norm": 0.04119537025690079,
      "learning_rate": 5.880527959125746e-05,
      "loss": 0.0008252900093793869,
      "step": 145190
    },
    {
      "epoch": 41.21487368719841,
      "grad_norm": 0.12294608354568481,
      "learning_rate": 5.880244110133409e-05,
      "loss": 0.0005951128900051117,
      "step": 145200
    },
    {
      "epoch": 41.21771217712177,
      "grad_norm": 0.018754906952381134,
      "learning_rate": 5.8799602611410734e-05,
      "loss": 0.008452561497688294,
      "step": 145210
    },
    {
      "epoch": 41.22055066704513,
      "grad_norm": 0.1233539804816246,
      "learning_rate": 5.879676412148737e-05,
      "loss": 0.0011247556656599044,
      "step": 145220
    },
    {
      "epoch": 41.223389156968494,
      "grad_norm": 0.1056794673204422,
      "learning_rate": 5.879392563156401e-05,
      "loss": 0.0011659959331154824,
      "step": 145230
    },
    {
      "epoch": 41.22622764689185,
      "grad_norm": 0.024493688717484474,
      "learning_rate": 5.879108714164066e-05,
      "loss": 0.002005003020167351,
      "step": 145240
    },
    {
      "epoch": 41.22906613681521,
      "grad_norm": 0.3306806683540344,
      "learning_rate": 5.878824865171729e-05,
      "loss": 0.0010999446734786034,
      "step": 145250
    },
    {
      "epoch": 41.231904626738576,
      "grad_norm": 7.67410135269165,
      "learning_rate": 5.878541016179393e-05,
      "loss": 0.005765287578105927,
      "step": 145260
    },
    {
      "epoch": 41.23474311666194,
      "grad_norm": 0.2873842716217041,
      "learning_rate": 5.878257167187057e-05,
      "loss": 0.008141850680112838,
      "step": 145270
    },
    {
      "epoch": 41.237581606585294,
      "grad_norm": 0.6817256808280945,
      "learning_rate": 5.87797331819472e-05,
      "loss": 0.0006693607196211815,
      "step": 145280
    },
    {
      "epoch": 41.24042009650866,
      "grad_norm": 0.21207401156425476,
      "learning_rate": 5.8776894692023844e-05,
      "loss": 0.005440269410610199,
      "step": 145290
    },
    {
      "epoch": 41.24325858643202,
      "grad_norm": 0.6745044589042664,
      "learning_rate": 5.877405620210048e-05,
      "loss": 0.003842515125870705,
      "step": 145300
    },
    {
      "epoch": 41.246097076355376,
      "grad_norm": 0.0258292555809021,
      "learning_rate": 5.877121771217713e-05,
      "loss": 0.0012391526252031326,
      "step": 145310
    },
    {
      "epoch": 41.24893556627874,
      "grad_norm": 7.156068801879883,
      "learning_rate": 5.876837922225377e-05,
      "loss": 0.008458669483661651,
      "step": 145320
    },
    {
      "epoch": 41.2517740562021,
      "grad_norm": 0.9577639698982239,
      "learning_rate": 5.87655407323304e-05,
      "loss": 0.0007464131340384483,
      "step": 145330
    },
    {
      "epoch": 41.254612546125465,
      "grad_norm": 0.06425510346889496,
      "learning_rate": 5.8762702242407044e-05,
      "loss": 0.008060353249311447,
      "step": 145340
    },
    {
      "epoch": 41.25745103604882,
      "grad_norm": 0.21420449018478394,
      "learning_rate": 5.875986375248368e-05,
      "loss": 0.009614725410938264,
      "step": 145350
    },
    {
      "epoch": 41.26028952597218,
      "grad_norm": 0.5821102261543274,
      "learning_rate": 5.8757025262560314e-05,
      "loss": 0.0005190258845686913,
      "step": 145360
    },
    {
      "epoch": 41.263128015895546,
      "grad_norm": 0.008450837805867195,
      "learning_rate": 5.875418677263697e-05,
      "loss": 0.0007219363003969192,
      "step": 145370
    },
    {
      "epoch": 41.2659665058189,
      "grad_norm": 0.1311962902545929,
      "learning_rate": 5.87513482827136e-05,
      "loss": 0.0009339587762951851,
      "step": 145380
    },
    {
      "epoch": 41.268804995742265,
      "grad_norm": 0.0633564293384552,
      "learning_rate": 5.874850979279024e-05,
      "loss": 0.0006327318027615547,
      "step": 145390
    },
    {
      "epoch": 41.27164348566563,
      "grad_norm": 9.734660148620605,
      "learning_rate": 5.874567130286688e-05,
      "loss": 0.007412242889404297,
      "step": 145400
    },
    {
      "epoch": 41.274481975588984,
      "grad_norm": 0.03324989229440689,
      "learning_rate": 5.8742832812943514e-05,
      "loss": 0.001157754845917225,
      "step": 145410
    },
    {
      "epoch": 41.27732046551235,
      "grad_norm": 0.07201370596885681,
      "learning_rate": 5.8739994323020155e-05,
      "loss": 0.004410015046596527,
      "step": 145420
    },
    {
      "epoch": 41.28015895543571,
      "grad_norm": 0.029034489765763283,
      "learning_rate": 5.873715583309679e-05,
      "loss": 0.001518348976969719,
      "step": 145430
    },
    {
      "epoch": 41.28299744535907,
      "grad_norm": 0.7649355530738831,
      "learning_rate": 5.873431734317344e-05,
      "loss": 0.0031650718301534653,
      "step": 145440
    },
    {
      "epoch": 41.28583593528243,
      "grad_norm": 0.2929200530052185,
      "learning_rate": 5.873147885325008e-05,
      "loss": 0.0032642263919115066,
      "step": 145450
    },
    {
      "epoch": 41.28867442520579,
      "grad_norm": 0.11481241136789322,
      "learning_rate": 5.8728640363326714e-05,
      "loss": 0.003931598365306854,
      "step": 145460
    },
    {
      "epoch": 41.291512915129154,
      "grad_norm": 4.839142799377441,
      "learning_rate": 5.8725801873403355e-05,
      "loss": 0.0013753734529018402,
      "step": 145470
    },
    {
      "epoch": 41.29435140505251,
      "grad_norm": 9.920478820800781,
      "learning_rate": 5.872296338347999e-05,
      "loss": 0.0020329605787992476,
      "step": 145480
    },
    {
      "epoch": 41.29718989497587,
      "grad_norm": 6.25946569442749,
      "learning_rate": 5.8720124893556624e-05,
      "loss": 0.004577324539422989,
      "step": 145490
    },
    {
      "epoch": 41.300028384899235,
      "grad_norm": 0.11472012102603912,
      "learning_rate": 5.8717286403633266e-05,
      "loss": 0.005163762718439102,
      "step": 145500
    },
    {
      "epoch": 41.300028384899235,
      "eval_accuracy": 0.9755198067018503,
      "eval_loss": 0.09338675439357758,
      "eval_runtime": 32.2479,
      "eval_samples_per_second": 487.691,
      "eval_steps_per_second": 7.628,
      "step": 145500
    },
    {
      "epoch": 41.30286687482259,
      "grad_norm": 0.07173950970172882,
      "learning_rate": 5.8714447913709914e-05,
      "loss": 0.010778900235891342,
      "step": 145510
    },
    {
      "epoch": 41.305705364745954,
      "grad_norm": 1.6479077339172363,
      "learning_rate": 5.871160942378655e-05,
      "loss": 0.0010015567764639854,
      "step": 145520
    },
    {
      "epoch": 41.30854385466932,
      "grad_norm": 0.04712911695241928,
      "learning_rate": 5.870877093386319e-05,
      "loss": 0.011802864074707032,
      "step": 145530
    },
    {
      "epoch": 41.31138234459268,
      "grad_norm": 0.3111684322357178,
      "learning_rate": 5.8705932443939824e-05,
      "loss": 0.002516550198197365,
      "step": 145540
    },
    {
      "epoch": 41.314220834516036,
      "grad_norm": 9.086713790893555,
      "learning_rate": 5.8703093954016466e-05,
      "loss": 0.006249541044235229,
      "step": 145550
    },
    {
      "epoch": 41.3170593244394,
      "grad_norm": 0.04649857059121132,
      "learning_rate": 5.87002554640931e-05,
      "loss": 0.0026647591963410377,
      "step": 145560
    },
    {
      "epoch": 41.31989781436276,
      "grad_norm": 1.4562132358551025,
      "learning_rate": 5.869741697416975e-05,
      "loss": 0.0009696928784251214,
      "step": 145570
    },
    {
      "epoch": 41.32273630428612,
      "grad_norm": 4.884853839874268,
      "learning_rate": 5.869457848424639e-05,
      "loss": 0.004829346761107444,
      "step": 145580
    },
    {
      "epoch": 41.32557479420948,
      "grad_norm": 0.13710953295230865,
      "learning_rate": 5.8691739994323025e-05,
      "loss": 0.013420160114765167,
      "step": 145590
    },
    {
      "epoch": 41.32841328413284,
      "grad_norm": 1.368688702583313,
      "learning_rate": 5.8688901504399666e-05,
      "loss": 0.0017476556822657585,
      "step": 145600
    },
    {
      "epoch": 41.3312517740562,
      "grad_norm": 0.7805861234664917,
      "learning_rate": 5.86860630144763e-05,
      "loss": 0.0015975465998053552,
      "step": 145610
    },
    {
      "epoch": 41.33409026397956,
      "grad_norm": 0.08170569688081741,
      "learning_rate": 5.8683224524552935e-05,
      "loss": 0.01065317764878273,
      "step": 145620
    },
    {
      "epoch": 41.336928753902924,
      "grad_norm": 2.0264453887939453,
      "learning_rate": 5.8680386034629576e-05,
      "loss": 0.005311264842748642,
      "step": 145630
    },
    {
      "epoch": 41.33976724382629,
      "grad_norm": 0.6953568458557129,
      "learning_rate": 5.8677547544706225e-05,
      "loss": 0.0026627188548445702,
      "step": 145640
    },
    {
      "epoch": 41.34260573374964,
      "grad_norm": 1.5391521453857422,
      "learning_rate": 5.867470905478286e-05,
      "loss": 0.0055102802813053135,
      "step": 145650
    },
    {
      "epoch": 41.345444223673006,
      "grad_norm": 0.07290791720151901,
      "learning_rate": 5.86718705648595e-05,
      "loss": 0.0025353509932756423,
      "step": 145660
    },
    {
      "epoch": 41.34828271359637,
      "grad_norm": 0.09601853787899017,
      "learning_rate": 5.8669032074936135e-05,
      "loss": 0.0034578032791614532,
      "step": 145670
    },
    {
      "epoch": 41.351121203519725,
      "grad_norm": 1.0158658027648926,
      "learning_rate": 5.8666193585012777e-05,
      "loss": 0.0020593130961060525,
      "step": 145680
    },
    {
      "epoch": 41.35395969344309,
      "grad_norm": 5.684053421020508,
      "learning_rate": 5.866335509508941e-05,
      "loss": 0.004532975703477859,
      "step": 145690
    },
    {
      "epoch": 41.35679818336645,
      "grad_norm": 0.06394979357719421,
      "learning_rate": 5.866051660516605e-05,
      "loss": 0.007678359001874924,
      "step": 145700
    },
    {
      "epoch": 41.35963667328981,
      "grad_norm": 0.7588818073272705,
      "learning_rate": 5.86576781152427e-05,
      "loss": 0.0011769847944378853,
      "step": 145710
    },
    {
      "epoch": 41.36247516321317,
      "grad_norm": 0.534053385257721,
      "learning_rate": 5.8654839625319335e-05,
      "loss": 0.00214748103171587,
      "step": 145720
    },
    {
      "epoch": 41.36531365313653,
      "grad_norm": 0.1367609053850174,
      "learning_rate": 5.865200113539597e-05,
      "loss": 0.0035141266882419584,
      "step": 145730
    },
    {
      "epoch": 41.368152143059895,
      "grad_norm": 0.217321515083313,
      "learning_rate": 5.864916264547261e-05,
      "loss": 0.0026524193584918977,
      "step": 145740
    },
    {
      "epoch": 41.37099063298325,
      "grad_norm": 0.034695882350206375,
      "learning_rate": 5.8646324155549246e-05,
      "loss": 0.007649475336074829,
      "step": 145750
    },
    {
      "epoch": 41.37382912290661,
      "grad_norm": 3.788834810256958,
      "learning_rate": 5.864348566562589e-05,
      "loss": 0.003025222010910511,
      "step": 145760
    },
    {
      "epoch": 41.376667612829976,
      "grad_norm": 1.6785097122192383,
      "learning_rate": 5.8640647175702535e-05,
      "loss": 0.007743267714977265,
      "step": 145770
    },
    {
      "epoch": 41.37950610275333,
      "grad_norm": 0.09142900258302689,
      "learning_rate": 5.863780868577917e-05,
      "loss": 0.0012349803000688553,
      "step": 145780
    },
    {
      "epoch": 41.382344592676695,
      "grad_norm": 0.8324795961380005,
      "learning_rate": 5.863497019585581e-05,
      "loss": 0.003918953984975815,
      "step": 145790
    },
    {
      "epoch": 41.38518308260006,
      "grad_norm": 0.1989109069108963,
      "learning_rate": 5.8632131705932446e-05,
      "loss": 0.00940801203250885,
      "step": 145800
    },
    {
      "epoch": 41.38802157252342,
      "grad_norm": 2.302684783935547,
      "learning_rate": 5.862929321600909e-05,
      "loss": 0.015047499537467956,
      "step": 145810
    },
    {
      "epoch": 41.39086006244678,
      "grad_norm": 2.939088821411133,
      "learning_rate": 5.862645472608572e-05,
      "loss": 0.009790368378162384,
      "step": 145820
    },
    {
      "epoch": 41.39369855237014,
      "grad_norm": 0.38063910603523254,
      "learning_rate": 5.8623616236162357e-05,
      "loss": 0.0014207422733306884,
      "step": 145830
    },
    {
      "epoch": 41.3965370422935,
      "grad_norm": 0.25288718938827515,
      "learning_rate": 5.862077774623901e-05,
      "loss": 0.006240326911211014,
      "step": 145840
    },
    {
      "epoch": 41.39937553221686,
      "grad_norm": 34.06725311279297,
      "learning_rate": 5.8617939256315646e-05,
      "loss": 0.020991221070289612,
      "step": 145850
    },
    {
      "epoch": 41.40221402214022,
      "grad_norm": 0.08132422715425491,
      "learning_rate": 5.861510076639228e-05,
      "loss": 0.013836082816123963,
      "step": 145860
    },
    {
      "epoch": 41.405052512063584,
      "grad_norm": 0.782006561756134,
      "learning_rate": 5.861226227646892e-05,
      "loss": 0.004109282791614532,
      "step": 145870
    },
    {
      "epoch": 41.40789100198694,
      "grad_norm": 0.8225546479225159,
      "learning_rate": 5.860942378654556e-05,
      "loss": 0.0011184751987457276,
      "step": 145880
    },
    {
      "epoch": 41.4107294919103,
      "grad_norm": 0.42617157101631165,
      "learning_rate": 5.86065852966222e-05,
      "loss": 0.0021033253520727157,
      "step": 145890
    },
    {
      "epoch": 41.413567981833665,
      "grad_norm": 0.9703125953674316,
      "learning_rate": 5.860374680669883e-05,
      "loss": 0.0019806453958153723,
      "step": 145900
    },
    {
      "epoch": 41.41640647175703,
      "grad_norm": 0.03499014675617218,
      "learning_rate": 5.860090831677548e-05,
      "loss": 0.010398822277784348,
      "step": 145910
    },
    {
      "epoch": 41.419244961680384,
      "grad_norm": 2.9516637325286865,
      "learning_rate": 5.859806982685212e-05,
      "loss": 0.01679158806800842,
      "step": 145920
    },
    {
      "epoch": 41.42208345160375,
      "grad_norm": 0.27762582898139954,
      "learning_rate": 5.859523133692876e-05,
      "loss": 0.005511163175106049,
      "step": 145930
    },
    {
      "epoch": 41.42492194152711,
      "grad_norm": 0.32349109649658203,
      "learning_rate": 5.85923928470054e-05,
      "loss": 0.0019127661362290382,
      "step": 145940
    },
    {
      "epoch": 41.427760431450466,
      "grad_norm": 0.8312756419181824,
      "learning_rate": 5.858955435708203e-05,
      "loss": 0.0010938579216599464,
      "step": 145950
    },
    {
      "epoch": 41.43059892137383,
      "grad_norm": 0.03454333171248436,
      "learning_rate": 5.858671586715867e-05,
      "loss": 0.001368466205894947,
      "step": 145960
    },
    {
      "epoch": 41.43343741129719,
      "grad_norm": 1.7128342390060425,
      "learning_rate": 5.858387737723532e-05,
      "loss": 0.001388213410973549,
      "step": 145970
    },
    {
      "epoch": 41.43627590122055,
      "grad_norm": 0.6372775435447693,
      "learning_rate": 5.858103888731196e-05,
      "loss": 0.0012451563030481338,
      "step": 145980
    },
    {
      "epoch": 41.43911439114391,
      "grad_norm": 0.5016331076622009,
      "learning_rate": 5.857820039738859e-05,
      "loss": 0.0017490141093730927,
      "step": 145990
    },
    {
      "epoch": 41.44195288106727,
      "grad_norm": 2.7045888900756836,
      "learning_rate": 5.857536190746523e-05,
      "loss": 0.0021271197125315665,
      "step": 146000
    },
    {
      "epoch": 41.44195288106727,
      "eval_accuracy": 0.9755833916195078,
      "eval_loss": 0.08457674086093903,
      "eval_runtime": 32.2163,
      "eval_samples_per_second": 488.169,
      "eval_steps_per_second": 7.636,
      "step": 146000
    },
    {
      "epoch": 41.444791370990636,
      "grad_norm": 0.15012149512767792,
      "learning_rate": 5.857252341754187e-05,
      "loss": 0.00483151376247406,
      "step": 146010
    },
    {
      "epoch": 41.44762986091399,
      "grad_norm": 2.36712908744812,
      "learning_rate": 5.856968492761851e-05,
      "loss": 0.0020745106041431425,
      "step": 146020
    },
    {
      "epoch": 41.450468350837355,
      "grad_norm": 0.8119973540306091,
      "learning_rate": 5.856684643769514e-05,
      "loss": 0.002600039727985859,
      "step": 146030
    },
    {
      "epoch": 41.45330684076072,
      "grad_norm": 1.290583848953247,
      "learning_rate": 5.856400794777179e-05,
      "loss": 0.0017417054623365402,
      "step": 146040
    },
    {
      "epoch": 41.45614533068407,
      "grad_norm": 2.629918336868286,
      "learning_rate": 5.856116945784843e-05,
      "loss": 0.012116938829421997,
      "step": 146050
    },
    {
      "epoch": 41.458983820607436,
      "grad_norm": 0.017288506031036377,
      "learning_rate": 5.855833096792507e-05,
      "loss": 0.0030928194522857667,
      "step": 146060
    },
    {
      "epoch": 41.4618223105308,
      "grad_norm": 0.027421049773693085,
      "learning_rate": 5.855549247800171e-05,
      "loss": 0.003031834028661251,
      "step": 146070
    },
    {
      "epoch": 41.464660800454155,
      "grad_norm": 0.6058752536773682,
      "learning_rate": 5.8552653988078343e-05,
      "loss": 0.008507491648197174,
      "step": 146080
    },
    {
      "epoch": 41.46749929037752,
      "grad_norm": 0.07510282099246979,
      "learning_rate": 5.854981549815498e-05,
      "loss": 0.0008268224075436593,
      "step": 146090
    },
    {
      "epoch": 41.47033778030088,
      "grad_norm": 1.5502657890319824,
      "learning_rate": 5.854697700823162e-05,
      "loss": 0.0015598092228174209,
      "step": 146100
    },
    {
      "epoch": 41.47317627022424,
      "grad_norm": 0.13633672893047333,
      "learning_rate": 5.854413851830827e-05,
      "loss": 0.0009845592081546783,
      "step": 146110
    },
    {
      "epoch": 41.4760147601476,
      "grad_norm": 0.07037965208292007,
      "learning_rate": 5.85413000283849e-05,
      "loss": 0.003234618902206421,
      "step": 146120
    },
    {
      "epoch": 41.47885325007096,
      "grad_norm": 0.558222770690918,
      "learning_rate": 5.8538461538461544e-05,
      "loss": 0.001115960069000721,
      "step": 146130
    },
    {
      "epoch": 41.481691739994325,
      "grad_norm": 1.0038466453552246,
      "learning_rate": 5.853562304853818e-05,
      "loss": 0.0006118463352322578,
      "step": 146140
    },
    {
      "epoch": 41.48453022991768,
      "grad_norm": 2.4307286739349365,
      "learning_rate": 5.853278455861482e-05,
      "loss": 0.0014129804447293281,
      "step": 146150
    },
    {
      "epoch": 41.487368719841044,
      "grad_norm": 0.030612953007221222,
      "learning_rate": 5.8529946068691454e-05,
      "loss": 0.001954994536936283,
      "step": 146160
    },
    {
      "epoch": 41.49020720976441,
      "grad_norm": 1.3565865755081177,
      "learning_rate": 5.85271075787681e-05,
      "loss": 0.006973596662282944,
      "step": 146170
    },
    {
      "epoch": 41.49304569968777,
      "grad_norm": 0.11380746215581894,
      "learning_rate": 5.8524269088844744e-05,
      "loss": 0.001620149053633213,
      "step": 146180
    },
    {
      "epoch": 41.495884189611125,
      "grad_norm": 0.09980122745037079,
      "learning_rate": 5.852143059892138e-05,
      "loss": 0.003427586704492569,
      "step": 146190
    },
    {
      "epoch": 41.49872267953449,
      "grad_norm": 0.3162451982498169,
      "learning_rate": 5.851859210899801e-05,
      "loss": 0.006603902578353882,
      "step": 146200
    },
    {
      "epoch": 41.50156116945785,
      "grad_norm": 0.037389349192380905,
      "learning_rate": 5.8515753619074654e-05,
      "loss": 0.0033258195966482163,
      "step": 146210
    },
    {
      "epoch": 41.50439965938121,
      "grad_norm": 0.0722975879907608,
      "learning_rate": 5.851291512915129e-05,
      "loss": 0.0030925540253520013,
      "step": 146220
    },
    {
      "epoch": 41.50723814930457,
      "grad_norm": 0.941077470779419,
      "learning_rate": 5.851007663922793e-05,
      "loss": 0.003653224930167198,
      "step": 146230
    },
    {
      "epoch": 41.51007663922793,
      "grad_norm": 0.14868809282779694,
      "learning_rate": 5.850723814930458e-05,
      "loss": 0.0018859857693314553,
      "step": 146240
    },
    {
      "epoch": 41.51291512915129,
      "grad_norm": 0.26543790102005005,
      "learning_rate": 5.850439965938121e-05,
      "loss": 0.0016311092302203178,
      "step": 146250
    },
    {
      "epoch": 41.51575361907465,
      "grad_norm": 0.6739418506622314,
      "learning_rate": 5.8501561169457854e-05,
      "loss": 0.002829033508896828,
      "step": 146260
    },
    {
      "epoch": 41.518592108998014,
      "grad_norm": 0.05034260079264641,
      "learning_rate": 5.849872267953449e-05,
      "loss": 0.006151802837848663,
      "step": 146270
    },
    {
      "epoch": 41.52143059892138,
      "grad_norm": 0.18017621338367462,
      "learning_rate": 5.849588418961113e-05,
      "loss": 0.001409577578306198,
      "step": 146280
    },
    {
      "epoch": 41.52426908884473,
      "grad_norm": 0.020883165299892426,
      "learning_rate": 5.8493045699687765e-05,
      "loss": 0.0012053649872541428,
      "step": 146290
    },
    {
      "epoch": 41.527107578768096,
      "grad_norm": 0.027409229427576065,
      "learning_rate": 5.84902072097644e-05,
      "loss": 0.0016215600073337555,
      "step": 146300
    },
    {
      "epoch": 41.52994606869146,
      "grad_norm": 0.2754102945327759,
      "learning_rate": 5.8487368719841054e-05,
      "loss": 0.003755996376276016,
      "step": 146310
    },
    {
      "epoch": 41.532784558614814,
      "grad_norm": 0.04513297602534294,
      "learning_rate": 5.848453022991769e-05,
      "loss": 0.0024851767346262934,
      "step": 146320
    },
    {
      "epoch": 41.53562304853818,
      "grad_norm": 0.10702929645776749,
      "learning_rate": 5.8481691739994324e-05,
      "loss": 0.0011313427239656448,
      "step": 146330
    },
    {
      "epoch": 41.53846153846154,
      "grad_norm": 0.024906206876039505,
      "learning_rate": 5.8478853250070965e-05,
      "loss": 0.0014658907428383827,
      "step": 146340
    },
    {
      "epoch": 41.541300028384896,
      "grad_norm": 0.018419666215777397,
      "learning_rate": 5.84760147601476e-05,
      "loss": 0.0027098411694169043,
      "step": 146350
    },
    {
      "epoch": 41.54413851830826,
      "grad_norm": 0.6486563086509705,
      "learning_rate": 5.847317627022424e-05,
      "loss": 0.006957846134901047,
      "step": 146360
    },
    {
      "epoch": 41.54697700823162,
      "grad_norm": 0.34428128600120544,
      "learning_rate": 5.847033778030089e-05,
      "loss": 0.0010294463485479356,
      "step": 146370
    },
    {
      "epoch": 41.549815498154985,
      "grad_norm": 5.296970367431641,
      "learning_rate": 5.8467499290377524e-05,
      "loss": 0.0021519895642995836,
      "step": 146380
    },
    {
      "epoch": 41.55265398807834,
      "grad_norm": 0.21933986246585846,
      "learning_rate": 5.8464660800454165e-05,
      "loss": 0.0022130172699689867,
      "step": 146390
    },
    {
      "epoch": 41.5554924780017,
      "grad_norm": 0.6236490607261658,
      "learning_rate": 5.84618223105308e-05,
      "loss": 0.004703926295042038,
      "step": 146400
    },
    {
      "epoch": 41.558330967925066,
      "grad_norm": 1.012663722038269,
      "learning_rate": 5.845898382060744e-05,
      "loss": 0.001440688967704773,
      "step": 146410
    },
    {
      "epoch": 41.56116945784842,
      "grad_norm": 0.08192196488380432,
      "learning_rate": 5.8456145330684076e-05,
      "loss": 0.0007352463901042939,
      "step": 146420
    },
    {
      "epoch": 41.564007947771785,
      "grad_norm": 3.1840624809265137,
      "learning_rate": 5.845330684076071e-05,
      "loss": 0.0022382942959666254,
      "step": 146430
    },
    {
      "epoch": 41.56684643769515,
      "grad_norm": 13.333020210266113,
      "learning_rate": 5.8450468350837365e-05,
      "loss": 0.0022849526256322862,
      "step": 146440
    },
    {
      "epoch": 41.56968492761851,
      "grad_norm": 1.0773590803146362,
      "learning_rate": 5.8447629860914e-05,
      "loss": 0.005852021276950836,
      "step": 146450
    },
    {
      "epoch": 41.572523417541866,
      "grad_norm": 3.876779556274414,
      "learning_rate": 5.8444791370990634e-05,
      "loss": 0.00789552479982376,
      "step": 146460
    },
    {
      "epoch": 41.57536190746523,
      "grad_norm": 0.21269336342811584,
      "learning_rate": 5.8441952881067276e-05,
      "loss": 0.0036786511540412904,
      "step": 146470
    },
    {
      "epoch": 41.57820039738859,
      "grad_norm": 0.03821628540754318,
      "learning_rate": 5.843911439114391e-05,
      "loss": 0.0013611899688839913,
      "step": 146480
    },
    {
      "epoch": 41.58103888731195,
      "grad_norm": 0.045921362936496735,
      "learning_rate": 5.843627590122055e-05,
      "loss": 0.00150004792958498,
      "step": 146490
    },
    {
      "epoch": 41.58387737723531,
      "grad_norm": 1.1227059364318848,
      "learning_rate": 5.8433437411297186e-05,
      "loss": 0.0010045304894447326,
      "step": 146500
    },
    {
      "epoch": 41.58387737723531,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.08926868438720703,
      "eval_runtime": 32.4508,
      "eval_samples_per_second": 484.642,
      "eval_steps_per_second": 7.581,
      "step": 146500
    },
    {
      "epoch": 41.586715867158674,
      "grad_norm": 0.06089174374938011,
      "learning_rate": 5.8430598921373834e-05,
      "loss": 0.007467126846313477,
      "step": 146510
    },
    {
      "epoch": 41.58955435708203,
      "grad_norm": 0.19324269890785217,
      "learning_rate": 5.8427760431450476e-05,
      "loss": 0.007337193936109543,
      "step": 146520
    },
    {
      "epoch": 41.59239284700539,
      "grad_norm": 0.8590618371963501,
      "learning_rate": 5.842492194152711e-05,
      "loss": 0.015850162506103514,
      "step": 146530
    },
    {
      "epoch": 41.595231336928755,
      "grad_norm": 0.10166192054748535,
      "learning_rate": 5.842208345160375e-05,
      "loss": 0.0023206565529108047,
      "step": 146540
    },
    {
      "epoch": 41.59806982685212,
      "grad_norm": 0.02678696997463703,
      "learning_rate": 5.8419244961680386e-05,
      "loss": 0.0014745475724339484,
      "step": 146550
    },
    {
      "epoch": 41.600908316775474,
      "grad_norm": 0.15094992518424988,
      "learning_rate": 5.841640647175702e-05,
      "loss": 0.0025883780792355536,
      "step": 146560
    },
    {
      "epoch": 41.60374680669884,
      "grad_norm": 0.1696808785200119,
      "learning_rate": 5.841356798183367e-05,
      "loss": 0.0011387221515178681,
      "step": 146570
    },
    {
      "epoch": 41.6065852966222,
      "grad_norm": 0.1471494436264038,
      "learning_rate": 5.841072949191031e-05,
      "loss": 0.000934501364827156,
      "step": 146580
    },
    {
      "epoch": 41.609423786545555,
      "grad_norm": 0.0824422538280487,
      "learning_rate": 5.8407891001986945e-05,
      "loss": 0.0021930107846856116,
      "step": 146590
    },
    {
      "epoch": 41.61226227646892,
      "grad_norm": 3.8588407039642334,
      "learning_rate": 5.8405052512063587e-05,
      "loss": 0.005647891014814377,
      "step": 146600
    },
    {
      "epoch": 41.61510076639228,
      "grad_norm": 0.5680772662162781,
      "learning_rate": 5.840221402214022e-05,
      "loss": 0.0016940992325544357,
      "step": 146610
    },
    {
      "epoch": 41.61793925631564,
      "grad_norm": 0.15707533061504364,
      "learning_rate": 5.839937553221686e-05,
      "loss": 0.001939178816974163,
      "step": 146620
    },
    {
      "epoch": 41.620777746239,
      "grad_norm": 0.7400066256523132,
      "learning_rate": 5.83965370422935e-05,
      "loss": 0.00835677981376648,
      "step": 146630
    },
    {
      "epoch": 41.62361623616236,
      "grad_norm": 0.18991097807884216,
      "learning_rate": 5.8393698552370145e-05,
      "loss": 0.004536560922861099,
      "step": 146640
    },
    {
      "epoch": 41.626454726085726,
      "grad_norm": 0.8818489909172058,
      "learning_rate": 5.839086006244679e-05,
      "loss": 0.0011925742030143738,
      "step": 146650
    },
    {
      "epoch": 41.62929321600908,
      "grad_norm": 0.5600959658622742,
      "learning_rate": 5.838802157252342e-05,
      "loss": 0.004391982033848762,
      "step": 146660
    },
    {
      "epoch": 41.632131705932444,
      "grad_norm": 1.530693531036377,
      "learning_rate": 5.8385183082600056e-05,
      "loss": 0.004468155652284622,
      "step": 146670
    },
    {
      "epoch": 41.63497019585581,
      "grad_norm": 1.5133119821548462,
      "learning_rate": 5.83823445926767e-05,
      "loss": 0.002784373238682747,
      "step": 146680
    },
    {
      "epoch": 41.63780868577916,
      "grad_norm": 0.01541772298514843,
      "learning_rate": 5.837950610275333e-05,
      "loss": 0.0006218397989869118,
      "step": 146690
    },
    {
      "epoch": 41.640647175702526,
      "grad_norm": 0.1743476837873459,
      "learning_rate": 5.837666761282998e-05,
      "loss": 0.0012261386960744857,
      "step": 146700
    },
    {
      "epoch": 41.64348566562589,
      "grad_norm": 0.04073362052440643,
      "learning_rate": 5.837382912290662e-05,
      "loss": 0.0010330639779567719,
      "step": 146710
    },
    {
      "epoch": 41.646324155549244,
      "grad_norm": 0.21755410730838776,
      "learning_rate": 5.8370990632983256e-05,
      "loss": 0.0016678448766469956,
      "step": 146720
    },
    {
      "epoch": 41.64916264547261,
      "grad_norm": 1.0872743129730225,
      "learning_rate": 5.83681521430599e-05,
      "loss": 0.0011772913858294487,
      "step": 146730
    },
    {
      "epoch": 41.65200113539597,
      "grad_norm": 3.5790693759918213,
      "learning_rate": 5.836531365313653e-05,
      "loss": 0.005790837854146957,
      "step": 146740
    },
    {
      "epoch": 41.65483962531933,
      "grad_norm": 16.24040412902832,
      "learning_rate": 5.836247516321317e-05,
      "loss": 0.010630351305007935,
      "step": 146750
    },
    {
      "epoch": 41.65767811524269,
      "grad_norm": 0.74370276927948,
      "learning_rate": 5.835963667328981e-05,
      "loss": 0.0034596666693687437,
      "step": 146760
    },
    {
      "epoch": 41.66051660516605,
      "grad_norm": 0.079166479408741,
      "learning_rate": 5.8356798183366456e-05,
      "loss": 0.009169553965330124,
      "step": 146770
    },
    {
      "epoch": 41.663355095089415,
      "grad_norm": 0.9231542348861694,
      "learning_rate": 5.83539596934431e-05,
      "loss": 0.001522885262966156,
      "step": 146780
    },
    {
      "epoch": 41.66619358501277,
      "grad_norm": 0.6771502494812012,
      "learning_rate": 5.835112120351973e-05,
      "loss": 0.0025657746940851212,
      "step": 146790
    },
    {
      "epoch": 41.66903207493613,
      "grad_norm": 13.296609878540039,
      "learning_rate": 5.8348282713596367e-05,
      "loss": 0.009668388962745666,
      "step": 146800
    },
    {
      "epoch": 41.671870564859496,
      "grad_norm": 0.15599903464317322,
      "learning_rate": 5.834544422367301e-05,
      "loss": 0.017988750338554384,
      "step": 146810
    },
    {
      "epoch": 41.67470905478285,
      "grad_norm": 0.08326566219329834,
      "learning_rate": 5.834260573374964e-05,
      "loss": 0.004128555208444596,
      "step": 146820
    },
    {
      "epoch": 41.677547544706215,
      "grad_norm": 1.728201150894165,
      "learning_rate": 5.8339767243826284e-05,
      "loss": 0.0039325617253780365,
      "step": 146830
    },
    {
      "epoch": 41.68038603462958,
      "grad_norm": 0.33730950951576233,
      "learning_rate": 5.833692875390293e-05,
      "loss": 0.005415488034486771,
      "step": 146840
    },
    {
      "epoch": 41.68322452455294,
      "grad_norm": 0.17004677653312683,
      "learning_rate": 5.833409026397957e-05,
      "loss": 0.0012051627039909364,
      "step": 146850
    },
    {
      "epoch": 41.6860630144763,
      "grad_norm": 0.032009221613407135,
      "learning_rate": 5.833125177405621e-05,
      "loss": 0.004847260564565659,
      "step": 146860
    },
    {
      "epoch": 41.68890150439966,
      "grad_norm": 0.035009924322366714,
      "learning_rate": 5.832841328413284e-05,
      "loss": 0.003128740191459656,
      "step": 146870
    },
    {
      "epoch": 41.69173999432302,
      "grad_norm": 1.0089366436004639,
      "learning_rate": 5.8325574794209484e-05,
      "loss": 0.0009750666096806527,
      "step": 146880
    },
    {
      "epoch": 41.69457848424638,
      "grad_norm": 0.11992795765399933,
      "learning_rate": 5.832273630428612e-05,
      "loss": 0.003372279927134514,
      "step": 146890
    },
    {
      "epoch": 41.69741697416974,
      "grad_norm": 4.18795919418335,
      "learning_rate": 5.831989781436277e-05,
      "loss": 0.004128644615411759,
      "step": 146900
    },
    {
      "epoch": 41.700255464093104,
      "grad_norm": 0.07851298898458481,
      "learning_rate": 5.831705932443941e-05,
      "loss": 0.005390677601099014,
      "step": 146910
    },
    {
      "epoch": 41.70309395401647,
      "grad_norm": 0.33531221747398376,
      "learning_rate": 5.831422083451604e-05,
      "loss": 0.0010899268090724945,
      "step": 146920
    },
    {
      "epoch": 41.70593244393982,
      "grad_norm": 0.868411660194397,
      "learning_rate": 5.831138234459268e-05,
      "loss": 0.003384992852807045,
      "step": 146930
    },
    {
      "epoch": 41.708770933863185,
      "grad_norm": 0.737566351890564,
      "learning_rate": 5.830854385466932e-05,
      "loss": 0.002056899666786194,
      "step": 146940
    },
    {
      "epoch": 41.71160942378655,
      "grad_norm": 0.03186449408531189,
      "learning_rate": 5.830570536474595e-05,
      "loss": 0.0067067302763462065,
      "step": 146950
    },
    {
      "epoch": 41.714447913709904,
      "grad_norm": 3.8335111141204834,
      "learning_rate": 5.8302866874822595e-05,
      "loss": 0.002956207096576691,
      "step": 146960
    },
    {
      "epoch": 41.71728640363327,
      "grad_norm": 3.4212467670440674,
      "learning_rate": 5.830002838489924e-05,
      "loss": 0.011097586154937744,
      "step": 146970
    },
    {
      "epoch": 41.72012489355663,
      "grad_norm": 3.9659836292266846,
      "learning_rate": 5.829718989497588e-05,
      "loss": 0.015092048048973083,
      "step": 146980
    },
    {
      "epoch": 41.722963383479986,
      "grad_norm": 11.982810974121094,
      "learning_rate": 5.829435140505252e-05,
      "loss": 0.012914462387561798,
      "step": 146990
    },
    {
      "epoch": 41.72580187340335,
      "grad_norm": 0.0876314714550972,
      "learning_rate": 5.8291512915129153e-05,
      "loss": 0.0031192868947982786,
      "step": 147000
    },
    {
      "epoch": 41.72580187340335,
      "eval_accuracy": 0.972976409995549,
      "eval_loss": 0.09423615783452988,
      "eval_runtime": 32.4026,
      "eval_samples_per_second": 485.363,
      "eval_steps_per_second": 7.592,
      "step": 147000
    },
    {
      "epoch": 41.72864036332671,
      "grad_norm": 0.08131708204746246,
      "learning_rate": 5.8288674425205795e-05,
      "loss": 0.003147163242101669,
      "step": 147010
    },
    {
      "epoch": 41.731478853250074,
      "grad_norm": 0.33748137950897217,
      "learning_rate": 5.828583593528243e-05,
      "loss": 0.0013698210939764976,
      "step": 147020
    },
    {
      "epoch": 41.73431734317343,
      "grad_norm": 0.6632359027862549,
      "learning_rate": 5.8282997445359064e-05,
      "loss": 0.0026838269084692003,
      "step": 147030
    },
    {
      "epoch": 41.73715583309679,
      "grad_norm": 0.24051140248775482,
      "learning_rate": 5.828015895543571e-05,
      "loss": 0.001899159513413906,
      "step": 147040
    },
    {
      "epoch": 41.739994323020156,
      "grad_norm": 0.6021842360496521,
      "learning_rate": 5.8277320465512354e-05,
      "loss": 0.013184882700443268,
      "step": 147050
    },
    {
      "epoch": 41.74283281294351,
      "grad_norm": 0.7475828528404236,
      "learning_rate": 5.827448197558899e-05,
      "loss": 0.009713547676801682,
      "step": 147060
    },
    {
      "epoch": 41.745671302866874,
      "grad_norm": 0.5117128491401672,
      "learning_rate": 5.827164348566563e-05,
      "loss": 0.015313288569450379,
      "step": 147070
    },
    {
      "epoch": 41.74850979279024,
      "grad_norm": 0.11252349615097046,
      "learning_rate": 5.8268804995742264e-05,
      "loss": 0.002850516326725483,
      "step": 147080
    },
    {
      "epoch": 41.75134828271359,
      "grad_norm": 0.8572073578834534,
      "learning_rate": 5.8265966505818905e-05,
      "loss": 0.0026715762913227083,
      "step": 147090
    },
    {
      "epoch": 41.754186772636956,
      "grad_norm": 0.11815090477466583,
      "learning_rate": 5.8263128015895554e-05,
      "loss": 0.001083856076002121,
      "step": 147100
    },
    {
      "epoch": 41.75702526256032,
      "grad_norm": 3.042510509490967,
      "learning_rate": 5.826028952597219e-05,
      "loss": 0.0012336423620581626,
      "step": 147110
    },
    {
      "epoch": 41.75986375248368,
      "grad_norm": 0.2670920491218567,
      "learning_rate": 5.825745103604883e-05,
      "loss": 0.0025270368903875353,
      "step": 147120
    },
    {
      "epoch": 41.76270224240704,
      "grad_norm": 0.047469254583120346,
      "learning_rate": 5.8254612546125464e-05,
      "loss": 0.004080427438020706,
      "step": 147130
    },
    {
      "epoch": 41.7655407323304,
      "grad_norm": 0.6935328245162964,
      "learning_rate": 5.82517740562021e-05,
      "loss": 0.004740985110402107,
      "step": 147140
    },
    {
      "epoch": 41.76837922225376,
      "grad_norm": 4.168476581573486,
      "learning_rate": 5.824893556627874e-05,
      "loss": 0.002882787026464939,
      "step": 147150
    },
    {
      "epoch": 41.77121771217712,
      "grad_norm": 0.0887054055929184,
      "learning_rate": 5.8246097076355375e-05,
      "loss": 0.008554118871688842,
      "step": 147160
    },
    {
      "epoch": 41.77405620210048,
      "grad_norm": 0.2847028970718384,
      "learning_rate": 5.824325858643202e-05,
      "loss": 0.0011478766798973083,
      "step": 147170
    },
    {
      "epoch": 41.776894692023845,
      "grad_norm": 0.40821418166160583,
      "learning_rate": 5.8240420096508664e-05,
      "loss": 0.012484822422266006,
      "step": 147180
    },
    {
      "epoch": 41.7797331819472,
      "grad_norm": 0.44162291288375854,
      "learning_rate": 5.82375816065853e-05,
      "loss": 0.0021684063598513604,
      "step": 147190
    },
    {
      "epoch": 41.78257167187056,
      "grad_norm": 0.19811448454856873,
      "learning_rate": 5.823474311666194e-05,
      "loss": 0.01060350239276886,
      "step": 147200
    },
    {
      "epoch": 41.785410161793926,
      "grad_norm": 12.074902534484863,
      "learning_rate": 5.8231904626738575e-05,
      "loss": 0.017477259039878845,
      "step": 147210
    },
    {
      "epoch": 41.78824865171729,
      "grad_norm": 0.6531860828399658,
      "learning_rate": 5.8229066136815216e-05,
      "loss": 0.006938712298870086,
      "step": 147220
    },
    {
      "epoch": 41.791087141640645,
      "grad_norm": 0.047674279659986496,
      "learning_rate": 5.822622764689185e-05,
      "loss": 0.0006639098748564721,
      "step": 147230
    },
    {
      "epoch": 41.79392563156401,
      "grad_norm": 1.0376085042953491,
      "learning_rate": 5.82233891569685e-05,
      "loss": 0.005062695965170861,
      "step": 147240
    },
    {
      "epoch": 41.79676412148737,
      "grad_norm": 0.05884583666920662,
      "learning_rate": 5.822055066704514e-05,
      "loss": 0.002892335504293442,
      "step": 147250
    },
    {
      "epoch": 41.79960261141073,
      "grad_norm": 0.2725791931152344,
      "learning_rate": 5.8217712177121775e-05,
      "loss": 0.0029735064134001734,
      "step": 147260
    },
    {
      "epoch": 41.80244110133409,
      "grad_norm": 0.23383000493049622,
      "learning_rate": 5.821487368719841e-05,
      "loss": 0.002071506716310978,
      "step": 147270
    },
    {
      "epoch": 41.80527959125745,
      "grad_norm": 0.11522078514099121,
      "learning_rate": 5.821203519727505e-05,
      "loss": 0.004184079170227051,
      "step": 147280
    },
    {
      "epoch": 41.808118081180815,
      "grad_norm": 0.07482937723398209,
      "learning_rate": 5.8209196707351686e-05,
      "loss": 0.009014660865068436,
      "step": 147290
    },
    {
      "epoch": 41.81095657110417,
      "grad_norm": 0.04252932593226433,
      "learning_rate": 5.8206358217428334e-05,
      "loss": 0.0013432646170258521,
      "step": 147300
    },
    {
      "epoch": 41.813795061027534,
      "grad_norm": 2.7575271129608154,
      "learning_rate": 5.8203519727504975e-05,
      "loss": 0.0031303130090236664,
      "step": 147310
    },
    {
      "epoch": 41.8166335509509,
      "grad_norm": 0.19590429961681366,
      "learning_rate": 5.820096508657394e-05,
      "loss": 0.01184888780117035,
      "step": 147320
    },
    {
      "epoch": 41.81947204087425,
      "grad_norm": 0.23850500583648682,
      "learning_rate": 5.819812659665058e-05,
      "loss": 0.004408773779869079,
      "step": 147330
    },
    {
      "epoch": 41.822310530797616,
      "grad_norm": 7.02516508102417,
      "learning_rate": 5.819528810672722e-05,
      "loss": 0.010765761137008667,
      "step": 147340
    },
    {
      "epoch": 41.82514902072098,
      "grad_norm": 0.7308682203292847,
      "learning_rate": 5.8192449616803866e-05,
      "loss": 0.008822530508041382,
      "step": 147350
    },
    {
      "epoch": 41.827987510644334,
      "grad_norm": 0.1495169699192047,
      "learning_rate": 5.818961112688051e-05,
      "loss": 0.0010842572897672652,
      "step": 147360
    },
    {
      "epoch": 41.8308260005677,
      "grad_norm": 0.287382036447525,
      "learning_rate": 5.818677263695714e-05,
      "loss": 0.0037784580141305923,
      "step": 147370
    },
    {
      "epoch": 41.83366449049106,
      "grad_norm": 0.020530765876173973,
      "learning_rate": 5.818393414703378e-05,
      "loss": 0.0010303573682904244,
      "step": 147380
    },
    {
      "epoch": 41.83650298041442,
      "grad_norm": 6.735590934753418,
      "learning_rate": 5.818109565711042e-05,
      "loss": 0.005411159992218017,
      "step": 147390
    },
    {
      "epoch": 41.83934147033778,
      "grad_norm": 18.0218563079834,
      "learning_rate": 5.817825716718706e-05,
      "loss": 0.020626306533813477,
      "step": 147400
    },
    {
      "epoch": 41.84217996026114,
      "grad_norm": 0.13333527743816376,
      "learning_rate": 5.8175418677263693e-05,
      "loss": 0.0016832556575536728,
      "step": 147410
    },
    {
      "epoch": 41.845018450184504,
      "grad_norm": 3.54260516166687,
      "learning_rate": 5.817258018734034e-05,
      "loss": 0.005826597288250923,
      "step": 147420
    },
    {
      "epoch": 41.84785694010786,
      "grad_norm": 0.11311717331409454,
      "learning_rate": 5.8169741697416976e-05,
      "loss": 0.0017257191240787507,
      "step": 147430
    },
    {
      "epoch": 41.85069543003122,
      "grad_norm": 0.14873036742210388,
      "learning_rate": 5.816690320749362e-05,
      "loss": 0.014115546643733979,
      "step": 147440
    },
    {
      "epoch": 41.853533919954586,
      "grad_norm": 0.7668888568878174,
      "learning_rate": 5.816406471757025e-05,
      "loss": 0.004536485671997071,
      "step": 147450
    },
    {
      "epoch": 41.85637240987794,
      "grad_norm": 0.2620188891887665,
      "learning_rate": 5.8161226227646894e-05,
      "loss": 0.005364926531910896,
      "step": 147460
    },
    {
      "epoch": 41.859210899801305,
      "grad_norm": 0.5208867788314819,
      "learning_rate": 5.815838773772353e-05,
      "loss": 0.003286202996969223,
      "step": 147470
    },
    {
      "epoch": 41.86204938972467,
      "grad_norm": 0.3200452923774719,
      "learning_rate": 5.8155549247800176e-05,
      "loss": 0.006648755818605423,
      "step": 147480
    },
    {
      "epoch": 41.86488787964803,
      "grad_norm": 7.516880512237549,
      "learning_rate": 5.815271075787682e-05,
      "loss": 0.007744076102972031,
      "step": 147490
    },
    {
      "epoch": 41.867726369571386,
      "grad_norm": 0.26730769872665405,
      "learning_rate": 5.814987226795345e-05,
      "loss": 0.000565139576792717,
      "step": 147500
    },
    {
      "epoch": 41.867726369571386,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.08290290087461472,
      "eval_runtime": 32.6819,
      "eval_samples_per_second": 481.214,
      "eval_steps_per_second": 7.527,
      "step": 147500
    },
    {
      "epoch": 41.87056485949475,
      "grad_norm": 1.3433256149291992,
      "learning_rate": 5.8147033778030094e-05,
      "loss": 0.0023064883425831795,
      "step": 147510
    },
    {
      "epoch": 41.87340334941811,
      "grad_norm": 0.3978353440761566,
      "learning_rate": 5.814419528810673e-05,
      "loss": 0.0038914754986763,
      "step": 147520
    },
    {
      "epoch": 41.87624183934147,
      "grad_norm": 9.57226276397705,
      "learning_rate": 5.814135679818336e-05,
      "loss": 0.012706468999385833,
      "step": 147530
    },
    {
      "epoch": 41.87908032926483,
      "grad_norm": 0.041330594569444656,
      "learning_rate": 5.8138518308260004e-05,
      "loss": 0.0021071359515190125,
      "step": 147540
    },
    {
      "epoch": 41.88191881918819,
      "grad_norm": 5.490805149078369,
      "learning_rate": 5.813567981833665e-05,
      "loss": 0.0053326129913330075,
      "step": 147550
    },
    {
      "epoch": 41.88475730911155,
      "grad_norm": 0.4416806101799011,
      "learning_rate": 5.813284132841329e-05,
      "loss": 0.003304172307252884,
      "step": 147560
    },
    {
      "epoch": 41.88759579903491,
      "grad_norm": 0.06793039292097092,
      "learning_rate": 5.813000283848993e-05,
      "loss": 0.007199781388044358,
      "step": 147570
    },
    {
      "epoch": 41.890434288958275,
      "grad_norm": 0.3708445727825165,
      "learning_rate": 5.812716434856656e-05,
      "loss": 0.008208294957876205,
      "step": 147580
    },
    {
      "epoch": 41.89327277888164,
      "grad_norm": 4.664440155029297,
      "learning_rate": 5.8124325858643204e-05,
      "loss": 0.002201572060585022,
      "step": 147590
    },
    {
      "epoch": 41.896111268804994,
      "grad_norm": 0.04257483780384064,
      "learning_rate": 5.812148736871984e-05,
      "loss": 0.0008619504049420356,
      "step": 147600
    },
    {
      "epoch": 41.89894975872836,
      "grad_norm": 2.6341915130615234,
      "learning_rate": 5.811864887879648e-05,
      "loss": 0.0020460978150367735,
      "step": 147610
    },
    {
      "epoch": 41.90178824865172,
      "grad_norm": 0.07851394265890121,
      "learning_rate": 5.811581038887313e-05,
      "loss": 0.0030582845211029053,
      "step": 147620
    },
    {
      "epoch": 41.904626738575075,
      "grad_norm": 4.1485819816589355,
      "learning_rate": 5.811297189894976e-05,
      "loss": 0.003317224979400635,
      "step": 147630
    },
    {
      "epoch": 41.90746522849844,
      "grad_norm": 1.9572997093200684,
      "learning_rate": 5.8110133409026404e-05,
      "loss": 0.007461913675069809,
      "step": 147640
    },
    {
      "epoch": 41.9103037184218,
      "grad_norm": 1.7077418565750122,
      "learning_rate": 5.810729491910304e-05,
      "loss": 0.005768641829490662,
      "step": 147650
    },
    {
      "epoch": 41.913142208345164,
      "grad_norm": 3.030463457107544,
      "learning_rate": 5.8104456429179674e-05,
      "loss": 0.004807484894990921,
      "step": 147660
    },
    {
      "epoch": 41.91598069826852,
      "grad_norm": 0.41401901841163635,
      "learning_rate": 5.8101617939256315e-05,
      "loss": 0.008545652776956559,
      "step": 147670
    },
    {
      "epoch": 41.91881918819188,
      "grad_norm": 0.0818423181772232,
      "learning_rate": 5.809877944933296e-05,
      "loss": 0.004737967997789383,
      "step": 147680
    },
    {
      "epoch": 41.921657678115245,
      "grad_norm": 0.5672314167022705,
      "learning_rate": 5.80959409594096e-05,
      "loss": 0.009779351949691772,
      "step": 147690
    },
    {
      "epoch": 41.9244961680386,
      "grad_norm": 2.996091842651367,
      "learning_rate": 5.809310246948624e-05,
      "loss": 0.0027373740449547768,
      "step": 147700
    },
    {
      "epoch": 41.927334657961964,
      "grad_norm": 3.0213868618011475,
      "learning_rate": 5.8090263979562874e-05,
      "loss": 0.003720492497086525,
      "step": 147710
    },
    {
      "epoch": 41.93017314788533,
      "grad_norm": 1.7075763940811157,
      "learning_rate": 5.8087425489639515e-05,
      "loss": 0.0017891475930809976,
      "step": 147720
    },
    {
      "epoch": 41.93301163780868,
      "grad_norm": 5.161940574645996,
      "learning_rate": 5.808458699971615e-05,
      "loss": 0.0019787611439824105,
      "step": 147730
    },
    {
      "epoch": 41.935850127732046,
      "grad_norm": 0.11445596069097519,
      "learning_rate": 5.808174850979279e-05,
      "loss": 0.008735276758670807,
      "step": 147740
    },
    {
      "epoch": 41.93868861765541,
      "grad_norm": 0.6976454854011536,
      "learning_rate": 5.807891001986944e-05,
      "loss": 0.0025481412187218664,
      "step": 147750
    },
    {
      "epoch": 41.94152710757877,
      "grad_norm": 0.6270613670349121,
      "learning_rate": 5.8076071529946074e-05,
      "loss": 0.003670267015695572,
      "step": 147760
    },
    {
      "epoch": 41.94436559750213,
      "grad_norm": 4.4429450035095215,
      "learning_rate": 5.8073233040022715e-05,
      "loss": 0.008809293806552886,
      "step": 147770
    },
    {
      "epoch": 41.94720408742549,
      "grad_norm": 1.2670964002609253,
      "learning_rate": 5.807039455009935e-05,
      "loss": 0.00708872526884079,
      "step": 147780
    },
    {
      "epoch": 41.95004257734885,
      "grad_norm": 0.062080785632133484,
      "learning_rate": 5.8067556060175984e-05,
      "loss": 0.005391184613108635,
      "step": 147790
    },
    {
      "epoch": 41.95288106727221,
      "grad_norm": 0.05487651750445366,
      "learning_rate": 5.8064717570252626e-05,
      "loss": 0.002111133560538292,
      "step": 147800
    },
    {
      "epoch": 41.95571955719557,
      "grad_norm": 0.3206935226917267,
      "learning_rate": 5.806187908032926e-05,
      "loss": 0.0010101672261953354,
      "step": 147810
    },
    {
      "epoch": 41.958558047118935,
      "grad_norm": 0.04768664762377739,
      "learning_rate": 5.805904059040591e-05,
      "loss": 0.0011282308027148248,
      "step": 147820
    },
    {
      "epoch": 41.96139653704229,
      "grad_norm": 14.992984771728516,
      "learning_rate": 5.805620210048255e-05,
      "loss": 0.0044780105352401735,
      "step": 147830
    },
    {
      "epoch": 41.96423502696565,
      "grad_norm": 0.9442441463470459,
      "learning_rate": 5.8053363610559185e-05,
      "loss": 0.0007804796099662781,
      "step": 147840
    },
    {
      "epoch": 41.967073516889016,
      "grad_norm": 0.10974982380867004,
      "learning_rate": 5.8050525120635826e-05,
      "loss": 0.0007784070447087287,
      "step": 147850
    },
    {
      "epoch": 41.96991200681238,
      "grad_norm": 0.34406018257141113,
      "learning_rate": 5.804768663071246e-05,
      "loss": 0.0025797396898269652,
      "step": 147860
    },
    {
      "epoch": 41.972750496735735,
      "grad_norm": 0.03177465498447418,
      "learning_rate": 5.80448481407891e-05,
      "loss": 0.00403272807598114,
      "step": 147870
    },
    {
      "epoch": 41.9755889866591,
      "grad_norm": 0.03622500225901604,
      "learning_rate": 5.804200965086575e-05,
      "loss": 0.010548439621925355,
      "step": 147880
    },
    {
      "epoch": 41.97842747658246,
      "grad_norm": 0.034628067165613174,
      "learning_rate": 5.8039171160942385e-05,
      "loss": 0.0008944230154156685,
      "step": 147890
    },
    {
      "epoch": 41.981265966505816,
      "grad_norm": 0.20137465000152588,
      "learning_rate": 5.803633267101902e-05,
      "loss": 0.0024202339351177215,
      "step": 147900
    },
    {
      "epoch": 41.98410445642918,
      "grad_norm": 1.4294395446777344,
      "learning_rate": 5.803349418109566e-05,
      "loss": 0.007550927251577378,
      "step": 147910
    },
    {
      "epoch": 41.98694294635254,
      "grad_norm": 4.568727970123291,
      "learning_rate": 5.8030655691172295e-05,
      "loss": 0.004716400429606438,
      "step": 147920
    },
    {
      "epoch": 41.9897814362759,
      "grad_norm": 0.061145152896642685,
      "learning_rate": 5.8027817201248937e-05,
      "loss": 0.0014157941564917564,
      "step": 147930
    },
    {
      "epoch": 41.99261992619926,
      "grad_norm": 0.06142732873558998,
      "learning_rate": 5.802497871132557e-05,
      "loss": 0.009902379661798476,
      "step": 147940
    },
    {
      "epoch": 41.995458416122624,
      "grad_norm": 0.016658145934343338,
      "learning_rate": 5.802214022140222e-05,
      "loss": 0.005079532414674759,
      "step": 147950
    },
    {
      "epoch": 41.99829690604599,
      "grad_norm": 0.11378389596939087,
      "learning_rate": 5.801930173147886e-05,
      "loss": 0.0011573072522878646,
      "step": 147960
    },
    {
      "epoch": 42.00113539596934,
      "grad_norm": 10.588242530822754,
      "learning_rate": 5.8016463241555495e-05,
      "loss": 0.002886012941598892,
      "step": 147970
    },
    {
      "epoch": 42.003973885892705,
      "grad_norm": 0.22206243872642517,
      "learning_rate": 5.801362475163214e-05,
      "loss": 0.0018182601779699326,
      "step": 147980
    },
    {
      "epoch": 42.00681237581607,
      "grad_norm": 0.05613575875759125,
      "learning_rate": 5.801078626170877e-05,
      "loss": 0.0010436916723847388,
      "step": 147990
    },
    {
      "epoch": 42.009650865739424,
      "grad_norm": 0.167624369263649,
      "learning_rate": 5.8007947771785406e-05,
      "loss": 0.003317735716700554,
      "step": 148000
    },
    {
      "epoch": 42.009650865739424,
      "eval_accuracy": 0.974883957525275,
      "eval_loss": 0.09121189266443253,
      "eval_runtime": 31.741,
      "eval_samples_per_second": 495.48,
      "eval_steps_per_second": 7.75,
      "step": 148000
    },
    {
      "epoch": 42.01248935566279,
      "grad_norm": 5.4926347732543945,
      "learning_rate": 5.800510928186205e-05,
      "loss": 0.0019615299999713896,
      "step": 148010
    },
    {
      "epoch": 42.01532784558615,
      "grad_norm": 5.0397467613220215,
      "learning_rate": 5.8002270791938695e-05,
      "loss": 0.003204669803380966,
      "step": 148020
    },
    {
      "epoch": 42.018166335509505,
      "grad_norm": 0.20917385816574097,
      "learning_rate": 5.799943230201533e-05,
      "loss": 0.0020218903198838236,
      "step": 148030
    },
    {
      "epoch": 42.02100482543287,
      "grad_norm": 0.01600181870162487,
      "learning_rate": 5.799659381209197e-05,
      "loss": 0.002963532507419586,
      "step": 148040
    },
    {
      "epoch": 42.02384331535623,
      "grad_norm": 0.07229290902614594,
      "learning_rate": 5.7993755322168606e-05,
      "loss": 0.0038999214768409727,
      "step": 148050
    },
    {
      "epoch": 42.026681805279594,
      "grad_norm": 0.9045941233634949,
      "learning_rate": 5.799091683224525e-05,
      "loss": 0.0007863270118832588,
      "step": 148060
    },
    {
      "epoch": 42.02952029520295,
      "grad_norm": 0.12249844521284103,
      "learning_rate": 5.798807834232188e-05,
      "loss": 0.00138645451515913,
      "step": 148070
    },
    {
      "epoch": 42.03235878512631,
      "grad_norm": 0.03424160182476044,
      "learning_rate": 5.798523985239853e-05,
      "loss": 0.004904663190245628,
      "step": 148080
    },
    {
      "epoch": 42.035197275049676,
      "grad_norm": 0.40684759616851807,
      "learning_rate": 5.798240136247517e-05,
      "loss": 0.0022047240287065505,
      "step": 148090
    },
    {
      "epoch": 42.03803576497303,
      "grad_norm": 0.14067214727401733,
      "learning_rate": 5.7979562872551806e-05,
      "loss": 0.004714854434132576,
      "step": 148100
    },
    {
      "epoch": 42.040874254896394,
      "grad_norm": 4.484437942504883,
      "learning_rate": 5.797672438262845e-05,
      "loss": 0.003201736509799957,
      "step": 148110
    },
    {
      "epoch": 42.04371274481976,
      "grad_norm": 0.16209560632705688,
      "learning_rate": 5.797388589270508e-05,
      "loss": 0.0010487783700227737,
      "step": 148120
    },
    {
      "epoch": 42.04655123474312,
      "grad_norm": 0.01927056722342968,
      "learning_rate": 5.797104740278172e-05,
      "loss": 0.0006518881767988205,
      "step": 148130
    },
    {
      "epoch": 42.049389724666476,
      "grad_norm": 0.5525190830230713,
      "learning_rate": 5.796820891285836e-05,
      "loss": 0.004366601258516312,
      "step": 148140
    },
    {
      "epoch": 42.05222821458984,
      "grad_norm": 0.0631304606795311,
      "learning_rate": 5.7965370422935006e-05,
      "loss": 0.0025670694187283514,
      "step": 148150
    },
    {
      "epoch": 42.0550667045132,
      "grad_norm": 1.1734358072280884,
      "learning_rate": 5.796253193301164e-05,
      "loss": 0.005134715884923935,
      "step": 148160
    },
    {
      "epoch": 42.05790519443656,
      "grad_norm": 2.5967254638671875,
      "learning_rate": 5.795969344308828e-05,
      "loss": 0.004553484916687012,
      "step": 148170
    },
    {
      "epoch": 42.06074368435992,
      "grad_norm": 0.06469224393367767,
      "learning_rate": 5.795685495316492e-05,
      "loss": 0.0013846702873706817,
      "step": 148180
    },
    {
      "epoch": 42.06358217428328,
      "grad_norm": 0.07613036036491394,
      "learning_rate": 5.795401646324156e-05,
      "loss": 0.0019763268530368803,
      "step": 148190
    },
    {
      "epoch": 42.06642066420664,
      "grad_norm": 4.161798000335693,
      "learning_rate": 5.795117797331819e-05,
      "loss": 0.0025999821722507478,
      "step": 148200
    },
    {
      "epoch": 42.06925915413,
      "grad_norm": 0.09685018658638,
      "learning_rate": 5.7948339483394834e-05,
      "loss": 0.003942451626062393,
      "step": 148210
    },
    {
      "epoch": 42.072097644053365,
      "grad_norm": 0.13936004042625427,
      "learning_rate": 5.794550099347148e-05,
      "loss": 0.002917243354022503,
      "step": 148220
    },
    {
      "epoch": 42.07493613397673,
      "grad_norm": 0.09791680425405502,
      "learning_rate": 5.794266250354812e-05,
      "loss": 0.0016802605241537094,
      "step": 148230
    },
    {
      "epoch": 42.07777462390008,
      "grad_norm": 0.16870345175266266,
      "learning_rate": 5.793982401362476e-05,
      "loss": 0.00816027894616127,
      "step": 148240
    },
    {
      "epoch": 42.080613113823446,
      "grad_norm": 1.217581033706665,
      "learning_rate": 5.793698552370139e-05,
      "loss": 0.0022849004715681078,
      "step": 148250
    },
    {
      "epoch": 42.08345160374681,
      "grad_norm": 0.06864102929830551,
      "learning_rate": 5.793414703377803e-05,
      "loss": 0.0018223153427243232,
      "step": 148260
    },
    {
      "epoch": 42.086290093670165,
      "grad_norm": 0.252546489238739,
      "learning_rate": 5.793130854385467e-05,
      "loss": 0.0015053559094667436,
      "step": 148270
    },
    {
      "epoch": 42.08912858359353,
      "grad_norm": 0.006149183958768845,
      "learning_rate": 5.792847005393132e-05,
      "loss": 0.0008971454575657844,
      "step": 148280
    },
    {
      "epoch": 42.09196707351689,
      "grad_norm": 0.1221720352768898,
      "learning_rate": 5.792563156400795e-05,
      "loss": 0.0007475227117538452,
      "step": 148290
    },
    {
      "epoch": 42.09480556344025,
      "grad_norm": 1.5749468803405762,
      "learning_rate": 5.792279307408459e-05,
      "loss": 0.0032606054097414016,
      "step": 148300
    },
    {
      "epoch": 42.09764405336361,
      "grad_norm": 0.5113258957862854,
      "learning_rate": 5.791995458416123e-05,
      "loss": 0.009348827600479125,
      "step": 148310
    },
    {
      "epoch": 42.10048254328697,
      "grad_norm": 0.14693041145801544,
      "learning_rate": 5.791711609423787e-05,
      "loss": 0.0007658191025257111,
      "step": 148320
    },
    {
      "epoch": 42.103321033210335,
      "grad_norm": 0.8705531358718872,
      "learning_rate": 5.7914277604314503e-05,
      "loss": 0.005033469200134278,
      "step": 148330
    },
    {
      "epoch": 42.10615952313369,
      "grad_norm": 0.11414027959108353,
      "learning_rate": 5.7911439114391145e-05,
      "loss": 0.0024085933342576026,
      "step": 148340
    },
    {
      "epoch": 42.108998013057054,
      "grad_norm": 0.21188539266586304,
      "learning_rate": 5.790860062446779e-05,
      "loss": 0.001377679780125618,
      "step": 148350
    },
    {
      "epoch": 42.11183650298042,
      "grad_norm": 0.5982687473297119,
      "learning_rate": 5.790576213454443e-05,
      "loss": 0.0022996239364147185,
      "step": 148360
    },
    {
      "epoch": 42.11467499290377,
      "grad_norm": 8.45866870880127,
      "learning_rate": 5.790292364462106e-05,
      "loss": 0.006784467399120331,
      "step": 148370
    },
    {
      "epoch": 42.117513482827135,
      "grad_norm": 0.1945953071117401,
      "learning_rate": 5.7900085154697704e-05,
      "loss": 0.0046535424888134004,
      "step": 148380
    },
    {
      "epoch": 42.1203519727505,
      "grad_norm": 1.637442946434021,
      "learning_rate": 5.789724666477434e-05,
      "loss": 0.0013331301510334016,
      "step": 148390
    },
    {
      "epoch": 42.123190462673854,
      "grad_norm": 0.034490346908569336,
      "learning_rate": 5.789440817485098e-05,
      "loss": 0.0010250095278024674,
      "step": 148400
    },
    {
      "epoch": 42.12602895259722,
      "grad_norm": 0.16968096792697906,
      "learning_rate": 5.789156968492763e-05,
      "loss": 0.0010447947308421135,
      "step": 148410
    },
    {
      "epoch": 42.12886744252058,
      "grad_norm": 0.012737770564854145,
      "learning_rate": 5.788873119500426e-05,
      "loss": 0.0008038455620408058,
      "step": 148420
    },
    {
      "epoch": 42.13170593244394,
      "grad_norm": 0.522477924823761,
      "learning_rate": 5.7885892705080904e-05,
      "loss": 0.0010614242404699326,
      "step": 148430
    },
    {
      "epoch": 42.1345444223673,
      "grad_norm": 0.09368687123060226,
      "learning_rate": 5.788305421515754e-05,
      "loss": 0.017077405750751496,
      "step": 148440
    },
    {
      "epoch": 42.13738291229066,
      "grad_norm": 0.23763950169086456,
      "learning_rate": 5.788021572523418e-05,
      "loss": 0.0013159574940800668,
      "step": 148450
    },
    {
      "epoch": 42.140221402214024,
      "grad_norm": 0.023820364847779274,
      "learning_rate": 5.7877377235310814e-05,
      "loss": 0.0026591250672936438,
      "step": 148460
    },
    {
      "epoch": 42.14305989213738,
      "grad_norm": 0.09343444555997849,
      "learning_rate": 5.787453874538745e-05,
      "loss": 0.0009782956913113594,
      "step": 148470
    },
    {
      "epoch": 42.14589838206074,
      "grad_norm": 3.2588257789611816,
      "learning_rate": 5.7871700255464104e-05,
      "loss": 0.002280403859913349,
      "step": 148480
    },
    {
      "epoch": 42.148736871984106,
      "grad_norm": 9.96806526184082,
      "learning_rate": 5.786886176554074e-05,
      "loss": 0.008225031197071075,
      "step": 148490
    },
    {
      "epoch": 42.15157536190747,
      "grad_norm": 0.048505425453186035,
      "learning_rate": 5.786602327561737e-05,
      "loss": 0.000473700650036335,
      "step": 148500
    },
    {
      "epoch": 42.15157536190747,
      "eval_accuracy": 0.975837731290138,
      "eval_loss": 0.08783867955207825,
      "eval_runtime": 32.0658,
      "eval_samples_per_second": 490.461,
      "eval_steps_per_second": 7.672,
      "step": 148500
    },
    {
      "epoch": 42.154413851830824,
      "grad_norm": 0.02808949165046215,
      "learning_rate": 5.7863184785694014e-05,
      "loss": 0.00038874801248311996,
      "step": 148510
    },
    {
      "epoch": 42.15725234175419,
      "grad_norm": 0.12483259290456772,
      "learning_rate": 5.786034629577065e-05,
      "loss": 0.0015161557123064995,
      "step": 148520
    },
    {
      "epoch": 42.16009083167755,
      "grad_norm": 0.06408767402172089,
      "learning_rate": 5.785750780584729e-05,
      "loss": 0.0007976977154612541,
      "step": 148530
    },
    {
      "epoch": 42.162929321600906,
      "grad_norm": 0.18530237674713135,
      "learning_rate": 5.7854669315923925e-05,
      "loss": 0.0013058725744485855,
      "step": 148540
    },
    {
      "epoch": 42.16576781152427,
      "grad_norm": 0.23273755609989166,
      "learning_rate": 5.785183082600057e-05,
      "loss": 0.0030816461890935897,
      "step": 148550
    },
    {
      "epoch": 42.16860630144763,
      "grad_norm": 1.2946865558624268,
      "learning_rate": 5.7848992336077214e-05,
      "loss": 0.0039606474339962,
      "step": 148560
    },
    {
      "epoch": 42.17144479137099,
      "grad_norm": 0.20483577251434326,
      "learning_rate": 5.784615384615385e-05,
      "loss": 0.001383386179804802,
      "step": 148570
    },
    {
      "epoch": 42.17428328129435,
      "grad_norm": 0.9227098822593689,
      "learning_rate": 5.784331535623049e-05,
      "loss": 0.0020934173837304114,
      "step": 148580
    },
    {
      "epoch": 42.17712177121771,
      "grad_norm": 19.145475387573242,
      "learning_rate": 5.7840476866307125e-05,
      "loss": 0.005884546041488648,
      "step": 148590
    },
    {
      "epoch": 42.179960261141076,
      "grad_norm": 0.8967348337173462,
      "learning_rate": 5.783763837638376e-05,
      "loss": 0.0015535371378064156,
      "step": 148600
    },
    {
      "epoch": 42.18279875106443,
      "grad_norm": 2.5110926628112793,
      "learning_rate": 5.7834799886460415e-05,
      "loss": 0.002668975479900837,
      "step": 148610
    },
    {
      "epoch": 42.185637240987795,
      "grad_norm": 0.17472758889198303,
      "learning_rate": 5.783196139653705e-05,
      "loss": 0.002374446578323841,
      "step": 148620
    },
    {
      "epoch": 42.18847573091116,
      "grad_norm": 0.21558703482151031,
      "learning_rate": 5.7829122906613684e-05,
      "loss": 0.0027756338939070703,
      "step": 148630
    },
    {
      "epoch": 42.19131422083451,
      "grad_norm": 0.5122671723365784,
      "learning_rate": 5.7826284416690325e-05,
      "loss": 0.0006461162120103836,
      "step": 148640
    },
    {
      "epoch": 42.194152710757876,
      "grad_norm": 0.1394033581018448,
      "learning_rate": 5.782344592676696e-05,
      "loss": 0.001151873730123043,
      "step": 148650
    },
    {
      "epoch": 42.19699120068124,
      "grad_norm": 7.446272373199463,
      "learning_rate": 5.78206074368436e-05,
      "loss": 0.006092076003551483,
      "step": 148660
    },
    {
      "epoch": 42.199829690604595,
      "grad_norm": 5.922327518463135,
      "learning_rate": 5.7817768946920236e-05,
      "loss": 0.003573869913816452,
      "step": 148670
    },
    {
      "epoch": 42.20266818052796,
      "grad_norm": 9.825607299804688,
      "learning_rate": 5.7814930456996884e-05,
      "loss": 0.008025586605072021,
      "step": 148680
    },
    {
      "epoch": 42.20550667045132,
      "grad_norm": 0.8440441489219666,
      "learning_rate": 5.7812091967073525e-05,
      "loss": 0.0035113126039505007,
      "step": 148690
    },
    {
      "epoch": 42.208345160374684,
      "grad_norm": 0.029959499835968018,
      "learning_rate": 5.780925347715016e-05,
      "loss": 0.0037767499685287475,
      "step": 148700
    },
    {
      "epoch": 42.21118365029804,
      "grad_norm": 1.2642771005630493,
      "learning_rate": 5.78064149872268e-05,
      "loss": 0.000986599363386631,
      "step": 148710
    },
    {
      "epoch": 42.2140221402214,
      "grad_norm": 2.8234050273895264,
      "learning_rate": 5.7803576497303436e-05,
      "loss": 0.00487886555492878,
      "step": 148720
    },
    {
      "epoch": 42.216860630144765,
      "grad_norm": 0.03957867994904518,
      "learning_rate": 5.780073800738007e-05,
      "loss": 0.006504786014556885,
      "step": 148730
    },
    {
      "epoch": 42.21969912006812,
      "grad_norm": 0.427886962890625,
      "learning_rate": 5.779789951745671e-05,
      "loss": 0.004809509590268135,
      "step": 148740
    },
    {
      "epoch": 42.222537609991484,
      "grad_norm": 0.7828105092048645,
      "learning_rate": 5.779506102753336e-05,
      "loss": 0.004443696141242981,
      "step": 148750
    },
    {
      "epoch": 42.22537609991485,
      "grad_norm": 0.1608571708202362,
      "learning_rate": 5.7792222537609994e-05,
      "loss": 0.0009651016443967819,
      "step": 148760
    },
    {
      "epoch": 42.2282145898382,
      "grad_norm": 0.06307506561279297,
      "learning_rate": 5.7789384047686636e-05,
      "loss": 0.0010277647525072098,
      "step": 148770
    },
    {
      "epoch": 42.231053079761566,
      "grad_norm": 0.650701642036438,
      "learning_rate": 5.778654555776327e-05,
      "loss": 0.0035609032958745955,
      "step": 148780
    },
    {
      "epoch": 42.23389156968493,
      "grad_norm": 0.16893111169338226,
      "learning_rate": 5.778370706783991e-05,
      "loss": 0.0019431231543421746,
      "step": 148790
    },
    {
      "epoch": 42.23673005960829,
      "grad_norm": 3.2165372371673584,
      "learning_rate": 5.7780868577916546e-05,
      "loss": 0.007633565366268158,
      "step": 148800
    },
    {
      "epoch": 42.23956854953165,
      "grad_norm": 15.186038970947266,
      "learning_rate": 5.7778030087993195e-05,
      "loss": 0.004476289451122284,
      "step": 148810
    },
    {
      "epoch": 42.24240703945501,
      "grad_norm": 0.12604351341724396,
      "learning_rate": 5.7775191598069836e-05,
      "loss": 0.00441405177116394,
      "step": 148820
    },
    {
      "epoch": 42.24524552937837,
      "grad_norm": 0.07461309432983398,
      "learning_rate": 5.777235310814647e-05,
      "loss": 0.0020631251856684683,
      "step": 148830
    },
    {
      "epoch": 42.24808401930173,
      "grad_norm": 0.2549339532852173,
      "learning_rate": 5.7769514618223105e-05,
      "loss": 0.002142311818897724,
      "step": 148840
    },
    {
      "epoch": 42.25092250922509,
      "grad_norm": 0.19857609272003174,
      "learning_rate": 5.7766676128299747e-05,
      "loss": 0.0020988911390304567,
      "step": 148850
    },
    {
      "epoch": 42.253760999148454,
      "grad_norm": 3.5835609436035156,
      "learning_rate": 5.776383763837638e-05,
      "loss": 0.004204310476779938,
      "step": 148860
    },
    {
      "epoch": 42.25659948907182,
      "grad_norm": 0.9174870252609253,
      "learning_rate": 5.776099914845302e-05,
      "loss": 0.0028860537335276605,
      "step": 148870
    },
    {
      "epoch": 42.25943797899517,
      "grad_norm": 0.7222262620925903,
      "learning_rate": 5.775816065852967e-05,
      "loss": 0.0025588532909750937,
      "step": 148880
    },
    {
      "epoch": 42.262276468918536,
      "grad_norm": 0.4497431516647339,
      "learning_rate": 5.7755322168606305e-05,
      "loss": 0.0009511243551969528,
      "step": 148890
    },
    {
      "epoch": 42.2651149588419,
      "grad_norm": 0.19622796773910522,
      "learning_rate": 5.775248367868295e-05,
      "loss": 0.0010238831862807273,
      "step": 148900
    },
    {
      "epoch": 42.267953448765255,
      "grad_norm": 0.05724845081567764,
      "learning_rate": 5.774964518875958e-05,
      "loss": 0.004121816903352738,
      "step": 148910
    },
    {
      "epoch": 42.27079193868862,
      "grad_norm": 0.6709813475608826,
      "learning_rate": 5.774680669883622e-05,
      "loss": 0.009487597644329071,
      "step": 148920
    },
    {
      "epoch": 42.27363042861198,
      "grad_norm": 4.7786712646484375,
      "learning_rate": 5.774396820891286e-05,
      "loss": 0.007756906002759934,
      "step": 148930
    },
    {
      "epoch": 42.276468918535336,
      "grad_norm": 1.7194970846176147,
      "learning_rate": 5.774112971898949e-05,
      "loss": 0.005990612506866455,
      "step": 148940
    },
    {
      "epoch": 42.2793074084587,
      "grad_norm": 0.8138684630393982,
      "learning_rate": 5.773829122906615e-05,
      "loss": 0.003585950285196304,
      "step": 148950
    },
    {
      "epoch": 42.28214589838206,
      "grad_norm": 0.024683451279997826,
      "learning_rate": 5.773545273914278e-05,
      "loss": 0.00307918693870306,
      "step": 148960
    },
    {
      "epoch": 42.284984388305425,
      "grad_norm": 0.06129902973771095,
      "learning_rate": 5.7732614249219416e-05,
      "loss": 0.0009402599185705185,
      "step": 148970
    },
    {
      "epoch": 42.28782287822878,
      "grad_norm": 0.28746408224105835,
      "learning_rate": 5.772977575929606e-05,
      "loss": 0.007334737479686737,
      "step": 148980
    },
    {
      "epoch": 42.29066136815214,
      "grad_norm": 0.262344628572464,
      "learning_rate": 5.772693726937269e-05,
      "loss": 0.005848598480224609,
      "step": 148990
    },
    {
      "epoch": 42.293499858075506,
      "grad_norm": 0.3501192629337311,
      "learning_rate": 5.772409877944933e-05,
      "loss": 0.0011028772220015525,
      "step": 149000
    },
    {
      "epoch": 42.293499858075506,
      "eval_accuracy": 0.9762828257137407,
      "eval_loss": 0.08106570690870285,
      "eval_runtime": 33.8397,
      "eval_samples_per_second": 464.75,
      "eval_steps_per_second": 7.27,
      "step": 149000
    },
    {
      "epoch": 42.29633834799886,
      "grad_norm": 1.8995906114578247,
      "learning_rate": 5.772126028952598e-05,
      "loss": 0.0007258778437972069,
      "step": 149010
    },
    {
      "epoch": 42.299176837922225,
      "grad_norm": 1.8555222749710083,
      "learning_rate": 5.7718421799602616e-05,
      "loss": 0.0010192519053816794,
      "step": 149020
    },
    {
      "epoch": 42.30201532784559,
      "grad_norm": 1.5609384775161743,
      "learning_rate": 5.771586715867159e-05,
      "loss": 0.004942493140697479,
      "step": 149030
    },
    {
      "epoch": 42.304853817768944,
      "grad_norm": 6.585638523101807,
      "learning_rate": 5.7713028668748224e-05,
      "loss": 0.0029562147334218024,
      "step": 149040
    },
    {
      "epoch": 42.30769230769231,
      "grad_norm": 0.07576531171798706,
      "learning_rate": 5.7710190178824865e-05,
      "loss": 0.0008905645459890366,
      "step": 149050
    },
    {
      "epoch": 42.31053079761567,
      "grad_norm": 4.283070087432861,
      "learning_rate": 5.770735168890151e-05,
      "loss": 0.00501292496919632,
      "step": 149060
    },
    {
      "epoch": 42.31336928753903,
      "grad_norm": 0.49264270067214966,
      "learning_rate": 5.770451319897815e-05,
      "loss": 0.0007249932736158371,
      "step": 149070
    },
    {
      "epoch": 42.31620777746239,
      "grad_norm": 0.05113029107451439,
      "learning_rate": 5.770167470905479e-05,
      "loss": 0.004156504943966866,
      "step": 149080
    },
    {
      "epoch": 42.31904626738575,
      "grad_norm": 9.798721313476562,
      "learning_rate": 5.7698836219131424e-05,
      "loss": 0.003379834443330765,
      "step": 149090
    },
    {
      "epoch": 42.321884757309114,
      "grad_norm": 14.896623611450195,
      "learning_rate": 5.7695997729208065e-05,
      "loss": 0.005084022134542465,
      "step": 149100
    },
    {
      "epoch": 42.32472324723247,
      "grad_norm": 0.18293990194797516,
      "learning_rate": 5.76931592392847e-05,
      "loss": 0.0008425790816545487,
      "step": 149110
    },
    {
      "epoch": 42.32756173715583,
      "grad_norm": 0.1553041636943817,
      "learning_rate": 5.7690320749361334e-05,
      "loss": 0.0030866006389260294,
      "step": 149120
    },
    {
      "epoch": 42.330400227079195,
      "grad_norm": 0.11500255018472672,
      "learning_rate": 5.768748225943799e-05,
      "loss": 0.0015179915353655815,
      "step": 149130
    },
    {
      "epoch": 42.33323871700255,
      "grad_norm": 0.3087509572505951,
      "learning_rate": 5.7684643769514624e-05,
      "loss": 0.0034201137721538545,
      "step": 149140
    },
    {
      "epoch": 42.336077206925914,
      "grad_norm": 0.45559585094451904,
      "learning_rate": 5.768180527959126e-05,
      "loss": 0.0027576066553592683,
      "step": 149150
    },
    {
      "epoch": 42.33891569684928,
      "grad_norm": 0.18280380964279175,
      "learning_rate": 5.76789667896679e-05,
      "loss": 0.001021822914481163,
      "step": 149160
    },
    {
      "epoch": 42.34175418677264,
      "grad_norm": 0.04084552451968193,
      "learning_rate": 5.7676128299744535e-05,
      "loss": 0.0037411779165267946,
      "step": 149170
    },
    {
      "epoch": 42.344592676695996,
      "grad_norm": 0.19830608367919922,
      "learning_rate": 5.7673289809821176e-05,
      "loss": 0.006974570453166962,
      "step": 149180
    },
    {
      "epoch": 42.34743116661936,
      "grad_norm": 0.027047844603657722,
      "learning_rate": 5.7670451319897824e-05,
      "loss": 0.002917725592851639,
      "step": 149190
    },
    {
      "epoch": 42.35026965654272,
      "grad_norm": 0.038492824882268906,
      "learning_rate": 5.766761282997446e-05,
      "loss": 0.0018636330962181092,
      "step": 149200
    },
    {
      "epoch": 42.35310814646608,
      "grad_norm": 1.1602259874343872,
      "learning_rate": 5.76647743400511e-05,
      "loss": 0.0006949957460165024,
      "step": 149210
    },
    {
      "epoch": 42.35594663638944,
      "grad_norm": 0.71578049659729,
      "learning_rate": 5.7661935850127735e-05,
      "loss": 0.003599078580737114,
      "step": 149220
    },
    {
      "epoch": 42.3587851263128,
      "grad_norm": 0.07188554853200912,
      "learning_rate": 5.7659097360204376e-05,
      "loss": 0.0013271400704979897,
      "step": 149230
    },
    {
      "epoch": 42.36162361623616,
      "grad_norm": 0.08176995068788528,
      "learning_rate": 5.765625887028101e-05,
      "loss": 0.002788268029689789,
      "step": 149240
    },
    {
      "epoch": 42.36446210615952,
      "grad_norm": 0.20479156076908112,
      "learning_rate": 5.7653420380357645e-05,
      "loss": 0.005922697857022286,
      "step": 149250
    },
    {
      "epoch": 42.367300596082885,
      "grad_norm": 0.03780616447329521,
      "learning_rate": 5.765058189043429e-05,
      "loss": 0.0005988942459225655,
      "step": 149260
    },
    {
      "epoch": 42.37013908600625,
      "grad_norm": 0.044834546744823456,
      "learning_rate": 5.7647743400510935e-05,
      "loss": 0.0007186215370893478,
      "step": 149270
    },
    {
      "epoch": 42.3729775759296,
      "grad_norm": 0.365369588136673,
      "learning_rate": 5.764490491058757e-05,
      "loss": 0.0010452115908265113,
      "step": 149280
    },
    {
      "epoch": 42.375816065852966,
      "grad_norm": 0.9095590710639954,
      "learning_rate": 5.764206642066421e-05,
      "loss": 0.0015563469380140305,
      "step": 149290
    },
    {
      "epoch": 42.37865455577633,
      "grad_norm": 0.15202443301677704,
      "learning_rate": 5.7639227930740845e-05,
      "loss": 0.003445339947938919,
      "step": 149300
    },
    {
      "epoch": 42.381493045699685,
      "grad_norm": 0.41718608140945435,
      "learning_rate": 5.763638944081749e-05,
      "loss": 0.0013163993135094643,
      "step": 149310
    },
    {
      "epoch": 42.38433153562305,
      "grad_norm": 0.046813055872917175,
      "learning_rate": 5.763355095089412e-05,
      "loss": 0.0007233038544654846,
      "step": 149320
    },
    {
      "epoch": 42.38717002554641,
      "grad_norm": 0.0664995014667511,
      "learning_rate": 5.763071246097077e-05,
      "loss": 0.0009822048246860503,
      "step": 149330
    },
    {
      "epoch": 42.39000851546977,
      "grad_norm": 3.7311480045318604,
      "learning_rate": 5.762787397104741e-05,
      "loss": 0.00882490649819374,
      "step": 149340
    },
    {
      "epoch": 42.39284700539313,
      "grad_norm": 2.3043625354766846,
      "learning_rate": 5.7625035481124045e-05,
      "loss": 0.0015574658289551734,
      "step": 149350
    },
    {
      "epoch": 42.39568549531649,
      "grad_norm": 0.6846745014190674,
      "learning_rate": 5.762219699120068e-05,
      "loss": 0.0026110684499144553,
      "step": 149360
    },
    {
      "epoch": 42.398523985239855,
      "grad_norm": 0.061022449284791946,
      "learning_rate": 5.761935850127732e-05,
      "loss": 0.0007597656920552253,
      "step": 149370
    },
    {
      "epoch": 42.40136247516321,
      "grad_norm": 0.16794158518314362,
      "learning_rate": 5.7616520011353956e-05,
      "loss": 0.0011022666469216347,
      "step": 149380
    },
    {
      "epoch": 42.404200965086574,
      "grad_norm": 0.013561318628489971,
      "learning_rate": 5.7613681521430604e-05,
      "loss": 0.002890249527990818,
      "step": 149390
    },
    {
      "epoch": 42.40703945500994,
      "grad_norm": 1.4052187204360962,
      "learning_rate": 5.7610843031507246e-05,
      "loss": 0.0014119157567620277,
      "step": 149400
    },
    {
      "epoch": 42.40987794493329,
      "grad_norm": 0.48844704031944275,
      "learning_rate": 5.760800454158388e-05,
      "loss": 0.0015898007899522782,
      "step": 149410
    },
    {
      "epoch": 42.412716434856655,
      "grad_norm": 0.1818181872367859,
      "learning_rate": 5.760516605166052e-05,
      "loss": 0.0011587318032979966,
      "step": 149420
    },
    {
      "epoch": 42.41555492478002,
      "grad_norm": 0.013466368429362774,
      "learning_rate": 5.7602327561737156e-05,
      "loss": 0.0006939476355910301,
      "step": 149430
    },
    {
      "epoch": 42.41839341470338,
      "grad_norm": 0.266481876373291,
      "learning_rate": 5.75994890718138e-05,
      "loss": 0.0009322015568614006,
      "step": 149440
    },
    {
      "epoch": 42.42123190462674,
      "grad_norm": 0.08928917348384857,
      "learning_rate": 5.759665058189043e-05,
      "loss": 0.0011673543602228164,
      "step": 149450
    },
    {
      "epoch": 42.4240703945501,
      "grad_norm": 0.0835476890206337,
      "learning_rate": 5.759381209196708e-05,
      "loss": 0.0018524067476391793,
      "step": 149460
    },
    {
      "epoch": 42.42690888447346,
      "grad_norm": 0.07462149858474731,
      "learning_rate": 5.759097360204372e-05,
      "loss": 0.0008162502199411392,
      "step": 149470
    },
    {
      "epoch": 42.42974737439682,
      "grad_norm": 0.1084233745932579,
      "learning_rate": 5.7588135112120356e-05,
      "loss": 0.0011632420122623444,
      "step": 149480
    },
    {
      "epoch": 42.43258586432018,
      "grad_norm": 0.16020548343658447,
      "learning_rate": 5.758529662219699e-05,
      "loss": 0.0006304403766989708,
      "step": 149490
    },
    {
      "epoch": 42.435424354243544,
      "grad_norm": 0.025915855541825294,
      "learning_rate": 5.758245813227363e-05,
      "loss": 0.000703650526702404,
      "step": 149500
    },
    {
      "epoch": 42.435424354243544,
      "eval_accuracy": 0.9786990525847269,
      "eval_loss": 0.07884743809700012,
      "eval_runtime": 32.481,
      "eval_samples_per_second": 484.19,
      "eval_steps_per_second": 7.574,
      "step": 149500
    },
    {
      "epoch": 42.4382628441669,
      "grad_norm": 0.03452825918793678,
      "learning_rate": 5.757961964235027e-05,
      "loss": 0.0020655402913689613,
      "step": 149510
    },
    {
      "epoch": 42.44110133409026,
      "grad_norm": 2.167642116546631,
      "learning_rate": 5.757678115242691e-05,
      "loss": 0.003572450578212738,
      "step": 149520
    },
    {
      "epoch": 42.443939824013626,
      "grad_norm": 0.11725883185863495,
      "learning_rate": 5.7573942662503556e-05,
      "loss": 0.0007973648607730865,
      "step": 149530
    },
    {
      "epoch": 42.44677831393699,
      "grad_norm": 0.240860253572464,
      "learning_rate": 5.757110417258019e-05,
      "loss": 0.0015113582834601403,
      "step": 149540
    },
    {
      "epoch": 42.449616803860344,
      "grad_norm": 0.03504910692572594,
      "learning_rate": 5.756826568265683e-05,
      "loss": 0.0011814264580607414,
      "step": 149550
    },
    {
      "epoch": 42.45245529378371,
      "grad_norm": 1.6692795753479004,
      "learning_rate": 5.756542719273347e-05,
      "loss": 0.0020489107817411424,
      "step": 149560
    },
    {
      "epoch": 42.45529378370707,
      "grad_norm": 1.7897883653640747,
      "learning_rate": 5.756258870281011e-05,
      "loss": 0.006054595857858658,
      "step": 149570
    },
    {
      "epoch": 42.458132273630426,
      "grad_norm": 0.07541073113679886,
      "learning_rate": 5.755975021288674e-05,
      "loss": 0.009281496703624725,
      "step": 149580
    },
    {
      "epoch": 42.46097076355379,
      "grad_norm": 0.0854191705584526,
      "learning_rate": 5.755691172296339e-05,
      "loss": 0.005772685259580612,
      "step": 149590
    },
    {
      "epoch": 42.46380925347715,
      "grad_norm": 0.29860568046569824,
      "learning_rate": 5.755407323304003e-05,
      "loss": 0.01062425822019577,
      "step": 149600
    },
    {
      "epoch": 42.46664774340051,
      "grad_norm": 0.0560116171836853,
      "learning_rate": 5.755123474311667e-05,
      "loss": 0.006598441302776337,
      "step": 149610
    },
    {
      "epoch": 42.46948623332387,
      "grad_norm": 6.561726093292236,
      "learning_rate": 5.75483962531933e-05,
      "loss": 0.005295054987072945,
      "step": 149620
    },
    {
      "epoch": 42.47232472324723,
      "grad_norm": 0.17114068567752838,
      "learning_rate": 5.754555776326994e-05,
      "loss": 0.006758679449558258,
      "step": 149630
    },
    {
      "epoch": 42.475163213170596,
      "grad_norm": 0.5224401354789734,
      "learning_rate": 5.754271927334658e-05,
      "loss": 0.007378491014242173,
      "step": 149640
    },
    {
      "epoch": 42.47800170309395,
      "grad_norm": 2.0284016132354736,
      "learning_rate": 5.753988078342322e-05,
      "loss": 0.001416935957968235,
      "step": 149650
    },
    {
      "epoch": 42.480840193017315,
      "grad_norm": 1.306514024734497,
      "learning_rate": 5.753704229349987e-05,
      "loss": 0.002282816916704178,
      "step": 149660
    },
    {
      "epoch": 42.48367868294068,
      "grad_norm": 18.141342163085938,
      "learning_rate": 5.75342038035765e-05,
      "loss": 0.010499419271945953,
      "step": 149670
    },
    {
      "epoch": 42.48651717286403,
      "grad_norm": 0.9916077256202698,
      "learning_rate": 5.753136531365314e-05,
      "loss": 0.00396006777882576,
      "step": 149680
    },
    {
      "epoch": 42.489355662787396,
      "grad_norm": 1.541317105293274,
      "learning_rate": 5.752852682372978e-05,
      "loss": 0.006942304223775864,
      "step": 149690
    },
    {
      "epoch": 42.49219415271076,
      "grad_norm": 0.05948227271437645,
      "learning_rate": 5.752568833380642e-05,
      "loss": 0.0006588274613022804,
      "step": 149700
    },
    {
      "epoch": 42.49503264263412,
      "grad_norm": 0.27741312980651855,
      "learning_rate": 5.7522849843883054e-05,
      "loss": 0.00031501520425081253,
      "step": 149710
    },
    {
      "epoch": 42.49787113255748,
      "grad_norm": 7.813920497894287,
      "learning_rate": 5.752001135395969e-05,
      "loss": 0.007956470549106597,
      "step": 149720
    },
    {
      "epoch": 42.50070962248084,
      "grad_norm": 1.3294410705566406,
      "learning_rate": 5.7517172864036336e-05,
      "loss": 0.0014052055776119232,
      "step": 149730
    },
    {
      "epoch": 42.503548112404204,
      "grad_norm": 0.13576719164848328,
      "learning_rate": 5.751433437411298e-05,
      "loss": 0.004079211503267288,
      "step": 149740
    },
    {
      "epoch": 42.50638660232756,
      "grad_norm": 1.4054192304611206,
      "learning_rate": 5.751149588418961e-05,
      "loss": 0.004032426699995995,
      "step": 149750
    },
    {
      "epoch": 42.50922509225092,
      "grad_norm": 0.37282854318618774,
      "learning_rate": 5.7508657394266254e-05,
      "loss": 0.0011860737577080727,
      "step": 149760
    },
    {
      "epoch": 42.512063582174285,
      "grad_norm": 0.33757326006889343,
      "learning_rate": 5.750581890434289e-05,
      "loss": 0.0034485340118408205,
      "step": 149770
    },
    {
      "epoch": 42.51490207209764,
      "grad_norm": 6.6259870529174805,
      "learning_rate": 5.750298041441953e-05,
      "loss": 0.00277415718883276,
      "step": 149780
    },
    {
      "epoch": 42.517740562021004,
      "grad_norm": 0.14565500617027283,
      "learning_rate": 5.750014192449618e-05,
      "loss": 0.001903793402016163,
      "step": 149790
    },
    {
      "epoch": 42.52057905194437,
      "grad_norm": 0.20912469923496246,
      "learning_rate": 5.749730343457281e-05,
      "loss": 0.0032007738947868345,
      "step": 149800
    },
    {
      "epoch": 42.52341754186773,
      "grad_norm": 0.1471225917339325,
      "learning_rate": 5.7494464944649454e-05,
      "loss": 0.0014086702838540077,
      "step": 149810
    },
    {
      "epoch": 42.526256031791085,
      "grad_norm": 0.08750477433204651,
      "learning_rate": 5.749162645472609e-05,
      "loss": 0.007166118174791336,
      "step": 149820
    },
    {
      "epoch": 42.52909452171445,
      "grad_norm": 0.20422011613845825,
      "learning_rate": 5.748878796480272e-05,
      "loss": 0.00940827578306198,
      "step": 149830
    },
    {
      "epoch": 42.53193301163781,
      "grad_norm": 0.20312394201755524,
      "learning_rate": 5.7485949474879364e-05,
      "loss": 0.0023881040513515474,
      "step": 149840
    },
    {
      "epoch": 42.53477150156117,
      "grad_norm": 2.5897226333618164,
      "learning_rate": 5.7483110984956e-05,
      "loss": 0.001211434043943882,
      "step": 149850
    },
    {
      "epoch": 42.53760999148453,
      "grad_norm": 0.012378913350403309,
      "learning_rate": 5.748027249503265e-05,
      "loss": 0.0029115533456206323,
      "step": 149860
    },
    {
      "epoch": 42.54044848140789,
      "grad_norm": 0.17284952104091644,
      "learning_rate": 5.747743400510929e-05,
      "loss": 0.007810847461223602,
      "step": 149870
    },
    {
      "epoch": 42.54328697133125,
      "grad_norm": 4.997737407684326,
      "learning_rate": 5.747459551518592e-05,
      "loss": 0.0017485266551375389,
      "step": 149880
    },
    {
      "epoch": 42.54612546125461,
      "grad_norm": 0.11255983263254166,
      "learning_rate": 5.7471757025262564e-05,
      "loss": 0.0005200183019042015,
      "step": 149890
    },
    {
      "epoch": 42.548963951177974,
      "grad_norm": 0.08184505254030228,
      "learning_rate": 5.74689185353392e-05,
      "loss": 0.0038756825029850005,
      "step": 149900
    },
    {
      "epoch": 42.55180244110134,
      "grad_norm": 0.3472974896430969,
      "learning_rate": 5.746608004541584e-05,
      "loss": 0.0008942432701587677,
      "step": 149910
    },
    {
      "epoch": 42.55464093102469,
      "grad_norm": 0.1257934719324112,
      "learning_rate": 5.7463241555492475e-05,
      "loss": 0.009531313925981522,
      "step": 149920
    },
    {
      "epoch": 42.557479420948056,
      "grad_norm": 0.24506719410419464,
      "learning_rate": 5.746040306556912e-05,
      "loss": 0.0013905897736549377,
      "step": 149930
    },
    {
      "epoch": 42.56031791087142,
      "grad_norm": 1.3813766241073608,
      "learning_rate": 5.7457564575645765e-05,
      "loss": 0.0020813683047890665,
      "step": 149940
    },
    {
      "epoch": 42.563156400794774,
      "grad_norm": 0.283517986536026,
      "learning_rate": 5.74547260857224e-05,
      "loss": 0.0022549683228135107,
      "step": 149950
    },
    {
      "epoch": 42.56599489071814,
      "grad_norm": 0.18578699231147766,
      "learning_rate": 5.7451887595799034e-05,
      "loss": 0.0062153637409210205,
      "step": 149960
    },
    {
      "epoch": 42.5688333806415,
      "grad_norm": 0.3587241768836975,
      "learning_rate": 5.7449049105875675e-05,
      "loss": 0.012299595773220063,
      "step": 149970
    },
    {
      "epoch": 42.571671870564856,
      "grad_norm": 0.07118772715330124,
      "learning_rate": 5.744621061595231e-05,
      "loss": 0.002682354301214218,
      "step": 149980
    },
    {
      "epoch": 42.57451036048822,
      "grad_norm": 8.508957862854004,
      "learning_rate": 5.744337212602896e-05,
      "loss": 0.015873529016971588,
      "step": 149990
    },
    {
      "epoch": 42.57734885041158,
      "grad_norm": 0.6336022019386292,
      "learning_rate": 5.74405336361056e-05,
      "loss": 0.006864231079816818,
      "step": 150000
    },
    {
      "epoch": 42.57734885041158,
      "eval_accuracy": 0.9753290519488778,
      "eval_loss": 0.09033339470624924,
      "eval_runtime": 32.0947,
      "eval_samples_per_second": 490.019,
      "eval_steps_per_second": 7.665,
      "step": 150000
    },
    {
      "epoch": 42.580187340334945,
      "grad_norm": 0.3210078775882721,
      "learning_rate": 5.7437695146182234e-05,
      "loss": 0.0005479916930198669,
      "step": 150010
    },
    {
      "epoch": 42.5830258302583,
      "grad_norm": 0.12213665246963501,
      "learning_rate": 5.7434856656258875e-05,
      "loss": 0.002155088447034359,
      "step": 150020
    },
    {
      "epoch": 42.58586432018166,
      "grad_norm": 0.07583698630332947,
      "learning_rate": 5.743201816633551e-05,
      "loss": 0.0016990404576063157,
      "step": 150030
    },
    {
      "epoch": 42.588702810105026,
      "grad_norm": 0.045593857765197754,
      "learning_rate": 5.742917967641215e-05,
      "loss": 0.0010635141283273698,
      "step": 150040
    },
    {
      "epoch": 42.59154130002838,
      "grad_norm": 0.012724583968520164,
      "learning_rate": 5.7426341186488786e-05,
      "loss": 0.0017008380964398384,
      "step": 150050
    },
    {
      "epoch": 42.594379789951745,
      "grad_norm": 0.01829829439520836,
      "learning_rate": 5.7423502696565434e-05,
      "loss": 0.0013929100707173348,
      "step": 150060
    },
    {
      "epoch": 42.59721827987511,
      "grad_norm": 0.19252076745033264,
      "learning_rate": 5.742066420664207e-05,
      "loss": 0.0010853147134184838,
      "step": 150070
    },
    {
      "epoch": 42.60005676979847,
      "grad_norm": 0.20845939218997955,
      "learning_rate": 5.741782571671871e-05,
      "loss": 0.0044466637074947355,
      "step": 150080
    },
    {
      "epoch": 42.60289525972183,
      "grad_norm": 0.08697447180747986,
      "learning_rate": 5.7414987226795345e-05,
      "loss": 0.00136849507689476,
      "step": 150090
    },
    {
      "epoch": 42.60573374964519,
      "grad_norm": 0.30494001507759094,
      "learning_rate": 5.7412148736871986e-05,
      "loss": 0.00639970526099205,
      "step": 150100
    },
    {
      "epoch": 42.60857223956855,
      "grad_norm": 0.14865972101688385,
      "learning_rate": 5.740931024694862e-05,
      "loss": 0.01313437670469284,
      "step": 150110
    },
    {
      "epoch": 42.61141072949191,
      "grad_norm": 1.243951678276062,
      "learning_rate": 5.740647175702527e-05,
      "loss": 0.008478510379791259,
      "step": 150120
    },
    {
      "epoch": 42.61424921941527,
      "grad_norm": 0.16383923590183258,
      "learning_rate": 5.740363326710191e-05,
      "loss": 0.0009566636756062508,
      "step": 150130
    },
    {
      "epoch": 42.617087709338634,
      "grad_norm": 0.15019963681697845,
      "learning_rate": 5.7400794777178545e-05,
      "loss": 0.004200580716133118,
      "step": 150140
    },
    {
      "epoch": 42.61992619926199,
      "grad_norm": 0.644383430480957,
      "learning_rate": 5.7397956287255186e-05,
      "loss": 0.0029541900381445884,
      "step": 150150
    },
    {
      "epoch": 42.62276468918535,
      "grad_norm": 0.0352880097925663,
      "learning_rate": 5.739511779733182e-05,
      "loss": 0.0013392327353358268,
      "step": 150160
    },
    {
      "epoch": 42.625603179108715,
      "grad_norm": 0.3577956557273865,
      "learning_rate": 5.739227930740846e-05,
      "loss": 0.007454320788383484,
      "step": 150170
    },
    {
      "epoch": 42.62844166903208,
      "grad_norm": 0.45475128293037415,
      "learning_rate": 5.7389440817485097e-05,
      "loss": 0.001341792568564415,
      "step": 150180
    },
    {
      "epoch": 42.631280158955434,
      "grad_norm": 0.7097508311271667,
      "learning_rate": 5.7386602327561745e-05,
      "loss": 0.0022515781223773957,
      "step": 150190
    },
    {
      "epoch": 42.6341186488788,
      "grad_norm": 0.07850458472967148,
      "learning_rate": 5.738376383763838e-05,
      "loss": 0.0009548047557473183,
      "step": 150200
    },
    {
      "epoch": 42.63695713880216,
      "grad_norm": 0.03114534541964531,
      "learning_rate": 5.738092534771502e-05,
      "loss": 0.0025249719619750977,
      "step": 150210
    },
    {
      "epoch": 42.639795628725516,
      "grad_norm": 2.694411516189575,
      "learning_rate": 5.7378086857791655e-05,
      "loss": 0.001954171434044838,
      "step": 150220
    },
    {
      "epoch": 42.64263411864888,
      "grad_norm": 0.32755333185195923,
      "learning_rate": 5.73752483678683e-05,
      "loss": 0.0014858070760965347,
      "step": 150230
    },
    {
      "epoch": 42.64547260857224,
      "grad_norm": 0.21322500705718994,
      "learning_rate": 5.737240987794493e-05,
      "loss": 0.0011690650135278702,
      "step": 150240
    },
    {
      "epoch": 42.6483110984956,
      "grad_norm": 0.05923840403556824,
      "learning_rate": 5.736957138802157e-05,
      "loss": 0.001018458418548107,
      "step": 150250
    },
    {
      "epoch": 42.65114958841896,
      "grad_norm": 3.1690590381622314,
      "learning_rate": 5.736673289809822e-05,
      "loss": 0.0017630696296691895,
      "step": 150260
    },
    {
      "epoch": 42.65398807834232,
      "grad_norm": 0.5497409701347351,
      "learning_rate": 5.7363894408174855e-05,
      "loss": 0.0017364714294672012,
      "step": 150270
    },
    {
      "epoch": 42.656826568265686,
      "grad_norm": 0.15725433826446533,
      "learning_rate": 5.73610559182515e-05,
      "loss": 0.0007915511727333069,
      "step": 150280
    },
    {
      "epoch": 42.65966505818904,
      "grad_norm": 0.1343006193637848,
      "learning_rate": 5.735821742832813e-05,
      "loss": 0.0009281590580940246,
      "step": 150290
    },
    {
      "epoch": 42.662503548112404,
      "grad_norm": 2.792534112930298,
      "learning_rate": 5.7355378938404766e-05,
      "loss": 0.0007065312936902046,
      "step": 150300
    },
    {
      "epoch": 42.66534203803577,
      "grad_norm": 0.5772119164466858,
      "learning_rate": 5.735254044848141e-05,
      "loss": 0.0074326939880847934,
      "step": 150310
    },
    {
      "epoch": 42.66818052795912,
      "grad_norm": 3.087285280227661,
      "learning_rate": 5.7349701958558055e-05,
      "loss": 0.0037145815789699555,
      "step": 150320
    },
    {
      "epoch": 42.671019017882486,
      "grad_norm": 0.14270351827144623,
      "learning_rate": 5.734686346863469e-05,
      "loss": 0.005557875335216522,
      "step": 150330
    },
    {
      "epoch": 42.67385750780585,
      "grad_norm": 0.413318395614624,
      "learning_rate": 5.734402497871133e-05,
      "loss": 0.0016404284164309503,
      "step": 150340
    },
    {
      "epoch": 42.676695997729205,
      "grad_norm": 0.2032013237476349,
      "learning_rate": 5.7341186488787966e-05,
      "loss": 0.0016983469948172569,
      "step": 150350
    },
    {
      "epoch": 42.67953448765257,
      "grad_norm": 1.5514073371887207,
      "learning_rate": 5.733834799886461e-05,
      "loss": 0.0027741821482777595,
      "step": 150360
    },
    {
      "epoch": 42.68237297757593,
      "grad_norm": 4.071971893310547,
      "learning_rate": 5.733550950894124e-05,
      "loss": 0.0010571483522653579,
      "step": 150370
    },
    {
      "epoch": 42.68521146749929,
      "grad_norm": 0.16359339654445648,
      "learning_rate": 5.733267101901788e-05,
      "loss": 0.0008526347577571869,
      "step": 150380
    },
    {
      "epoch": 42.68804995742265,
      "grad_norm": 6.108848571777344,
      "learning_rate": 5.732983252909453e-05,
      "loss": 0.0035112112760543822,
      "step": 150390
    },
    {
      "epoch": 42.69088844734601,
      "grad_norm": 1.1128220558166504,
      "learning_rate": 5.7326994039171166e-05,
      "loss": 0.0015738485381007195,
      "step": 150400
    },
    {
      "epoch": 42.693726937269375,
      "grad_norm": 2.535426378250122,
      "learning_rate": 5.732415554924781e-05,
      "loss": 0.003577391058206558,
      "step": 150410
    },
    {
      "epoch": 42.69656542719273,
      "grad_norm": 0.21612073481082916,
      "learning_rate": 5.732131705932444e-05,
      "loss": 0.0022890446707606316,
      "step": 150420
    },
    {
      "epoch": 42.69940391711609,
      "grad_norm": 0.7193753123283386,
      "learning_rate": 5.731847856940108e-05,
      "loss": 0.0017169078812003136,
      "step": 150430
    },
    {
      "epoch": 42.702242407039456,
      "grad_norm": 3.320112943649292,
      "learning_rate": 5.731564007947772e-05,
      "loss": 0.0021342338994145395,
      "step": 150440
    },
    {
      "epoch": 42.70508089696281,
      "grad_norm": 1.8155882358551025,
      "learning_rate": 5.731280158955435e-05,
      "loss": 0.0017885973677039147,
      "step": 150450
    },
    {
      "epoch": 42.707919386886175,
      "grad_norm": 0.8427953720092773,
      "learning_rate": 5.7309963099631e-05,
      "loss": 0.004203182086348534,
      "step": 150460
    },
    {
      "epoch": 42.71075787680954,
      "grad_norm": 0.27758195996284485,
      "learning_rate": 5.730712460970764e-05,
      "loss": 0.004022154211997986,
      "step": 150470
    },
    {
      "epoch": 42.7135963667329,
      "grad_norm": 12.116328239440918,
      "learning_rate": 5.730428611978428e-05,
      "loss": 0.003638196736574173,
      "step": 150480
    },
    {
      "epoch": 42.71643485665626,
      "grad_norm": 0.4715937674045563,
      "learning_rate": 5.730144762986092e-05,
      "loss": 0.0016477832570672036,
      "step": 150490
    },
    {
      "epoch": 42.71927334657962,
      "grad_norm": 0.6735902428627014,
      "learning_rate": 5.729860913993755e-05,
      "loss": 0.003999575227499008,
      "step": 150500
    },
    {
      "epoch": 42.71927334657962,
      "eval_accuracy": 0.9752018821135626,
      "eval_loss": 0.09470957517623901,
      "eval_runtime": 31.8729,
      "eval_samples_per_second": 493.429,
      "eval_steps_per_second": 7.718,
      "step": 150500
    },
    {
      "epoch": 42.72211183650298,
      "grad_norm": 0.1094418317079544,
      "learning_rate": 5.7295770650014194e-05,
      "loss": 0.012765003740787506,
      "step": 150510
    },
    {
      "epoch": 42.72495032642634,
      "grad_norm": 0.18066728115081787,
      "learning_rate": 5.729293216009084e-05,
      "loss": 0.0011145541444420814,
      "step": 150520
    },
    {
      "epoch": 42.7277888163497,
      "grad_norm": 0.04154348373413086,
      "learning_rate": 5.729009367016748e-05,
      "loss": 0.0031032875180244447,
      "step": 150530
    },
    {
      "epoch": 42.730627306273064,
      "grad_norm": 1.1184370517730713,
      "learning_rate": 5.728725518024411e-05,
      "loss": 0.0032140046358108522,
      "step": 150540
    },
    {
      "epoch": 42.73346579619643,
      "grad_norm": 1.3468046188354492,
      "learning_rate": 5.728441669032075e-05,
      "loss": 0.0019706591963768007,
      "step": 150550
    },
    {
      "epoch": 42.73630428611978,
      "grad_norm": 0.034683480858802795,
      "learning_rate": 5.728157820039739e-05,
      "loss": 0.0010897317901253701,
      "step": 150560
    },
    {
      "epoch": 42.739142776043145,
      "grad_norm": 0.03907227888703346,
      "learning_rate": 5.727873971047403e-05,
      "loss": 0.0004331288859248161,
      "step": 150570
    },
    {
      "epoch": 42.74198126596651,
      "grad_norm": 4.75922155380249,
      "learning_rate": 5.7275901220550663e-05,
      "loss": 0.012294979393482208,
      "step": 150580
    },
    {
      "epoch": 42.744819755889864,
      "grad_norm": 0.07869251817464828,
      "learning_rate": 5.727306273062731e-05,
      "loss": 0.0014568734914064407,
      "step": 150590
    },
    {
      "epoch": 42.74765824581323,
      "grad_norm": 0.0386231429874897,
      "learning_rate": 5.727022424070395e-05,
      "loss": 0.0008156327530741692,
      "step": 150600
    },
    {
      "epoch": 42.75049673573659,
      "grad_norm": 0.7213430404663086,
      "learning_rate": 5.726738575078059e-05,
      "loss": 0.001945074461400509,
      "step": 150610
    },
    {
      "epoch": 42.753335225659946,
      "grad_norm": 17.490074157714844,
      "learning_rate": 5.726454726085723e-05,
      "loss": 0.00801035389304161,
      "step": 150620
    },
    {
      "epoch": 42.75617371558331,
      "grad_norm": 0.049720119684934616,
      "learning_rate": 5.7261708770933864e-05,
      "loss": 0.009720776975154877,
      "step": 150630
    },
    {
      "epoch": 42.75901220550667,
      "grad_norm": 6.460530757904053,
      "learning_rate": 5.7258870281010505e-05,
      "loss": 0.00235158484429121,
      "step": 150640
    },
    {
      "epoch": 42.761850695430034,
      "grad_norm": 0.12171939760446548,
      "learning_rate": 5.725603179108714e-05,
      "loss": 0.0057321939617395404,
      "step": 150650
    },
    {
      "epoch": 42.76468918535339,
      "grad_norm": 1.604061484336853,
      "learning_rate": 5.725319330116379e-05,
      "loss": 0.0013187101110816001,
      "step": 150660
    },
    {
      "epoch": 42.76752767527675,
      "grad_norm": 0.10278558731079102,
      "learning_rate": 5.725035481124042e-05,
      "loss": 0.0035054072737693786,
      "step": 150670
    },
    {
      "epoch": 42.770366165200116,
      "grad_norm": 0.358737975358963,
      "learning_rate": 5.7247516321317064e-05,
      "loss": 0.004805418476462364,
      "step": 150680
    },
    {
      "epoch": 42.77320465512347,
      "grad_norm": 8.513472557067871,
      "learning_rate": 5.72446778313937e-05,
      "loss": 0.0061926361173391346,
      "step": 150690
    },
    {
      "epoch": 42.776043145046835,
      "grad_norm": 1.5776017904281616,
      "learning_rate": 5.724183934147034e-05,
      "loss": 0.005824965983629226,
      "step": 150700
    },
    {
      "epoch": 42.7788816349702,
      "grad_norm": 1.6449989080429077,
      "learning_rate": 5.7239000851546974e-05,
      "loss": 0.0040501363575458525,
      "step": 150710
    },
    {
      "epoch": 42.78172012489355,
      "grad_norm": 1.2675914764404297,
      "learning_rate": 5.723616236162362e-05,
      "loss": 0.0009574921801686287,
      "step": 150720
    },
    {
      "epoch": 42.784558614816916,
      "grad_norm": 0.08177550882101059,
      "learning_rate": 5.7233323871700264e-05,
      "loss": 0.0078052222728729245,
      "step": 150730
    },
    {
      "epoch": 42.78739710474028,
      "grad_norm": 0.12202019989490509,
      "learning_rate": 5.72304853817769e-05,
      "loss": 0.0019467292353510857,
      "step": 150740
    },
    {
      "epoch": 42.79023559466364,
      "grad_norm": 0.7652762532234192,
      "learning_rate": 5.722764689185354e-05,
      "loss": 0.015189316868782044,
      "step": 150750
    },
    {
      "epoch": 42.793074084587,
      "grad_norm": 1.027755856513977,
      "learning_rate": 5.7224808401930174e-05,
      "loss": 0.0024920105934143066,
      "step": 150760
    },
    {
      "epoch": 42.79591257451036,
      "grad_norm": 0.9882604479789734,
      "learning_rate": 5.722196991200681e-05,
      "loss": 0.012038751691579818,
      "step": 150770
    },
    {
      "epoch": 42.79875106443372,
      "grad_norm": 10.733354568481445,
      "learning_rate": 5.721913142208345e-05,
      "loss": 0.024357083439826965,
      "step": 150780
    },
    {
      "epoch": 42.80158955435708,
      "grad_norm": 0.34257903695106506,
      "learning_rate": 5.72162929321601e-05,
      "loss": 0.012957312166690826,
      "step": 150790
    },
    {
      "epoch": 42.80442804428044,
      "grad_norm": 0.46132591366767883,
      "learning_rate": 5.721345444223673e-05,
      "loss": 0.006321241706609726,
      "step": 150800
    },
    {
      "epoch": 42.807266534203805,
      "grad_norm": 0.35794907808303833,
      "learning_rate": 5.7210615952313374e-05,
      "loss": 0.0012566056102514267,
      "step": 150810
    },
    {
      "epoch": 42.81010502412717,
      "grad_norm": 0.31972241401672363,
      "learning_rate": 5.720777746239001e-05,
      "loss": 0.0013749055564403535,
      "step": 150820
    },
    {
      "epoch": 42.812943514050524,
      "grad_norm": 0.7825968265533447,
      "learning_rate": 5.720493897246665e-05,
      "loss": 0.0016559494659304618,
      "step": 150830
    },
    {
      "epoch": 42.81578200397389,
      "grad_norm": 0.058416660875082016,
      "learning_rate": 5.7202100482543285e-05,
      "loss": 0.009011832624673843,
      "step": 150840
    },
    {
      "epoch": 42.81862049389725,
      "grad_norm": 10.921631813049316,
      "learning_rate": 5.7199261992619926e-05,
      "loss": 0.002360529825091362,
      "step": 150850
    },
    {
      "epoch": 42.821458983820605,
      "grad_norm": 0.14741648733615875,
      "learning_rate": 5.7196423502696575e-05,
      "loss": 0.0036463100463151933,
      "step": 150860
    },
    {
      "epoch": 42.82429747374397,
      "grad_norm": 11.011202812194824,
      "learning_rate": 5.719358501277321e-05,
      "loss": 0.003688524663448334,
      "step": 150870
    },
    {
      "epoch": 42.82713596366733,
      "grad_norm": 11.639580726623535,
      "learning_rate": 5.719074652284985e-05,
      "loss": 0.0025018757209181786,
      "step": 150880
    },
    {
      "epoch": 42.82997445359069,
      "grad_norm": 1.2539687156677246,
      "learning_rate": 5.7187908032926485e-05,
      "loss": 0.0008635060861706733,
      "step": 150890
    },
    {
      "epoch": 42.83281294351405,
      "grad_norm": 0.31407153606414795,
      "learning_rate": 5.718506954300312e-05,
      "loss": 0.004352733492851257,
      "step": 150900
    },
    {
      "epoch": 42.83565143343741,
      "grad_norm": 0.06852952390909195,
      "learning_rate": 5.718223105307976e-05,
      "loss": 0.0017024286091327668,
      "step": 150910
    },
    {
      "epoch": 42.838489923360775,
      "grad_norm": 12.521977424621582,
      "learning_rate": 5.717939256315641e-05,
      "loss": 0.0035060569643974304,
      "step": 150920
    },
    {
      "epoch": 42.84132841328413,
      "grad_norm": 0.6829904913902283,
      "learning_rate": 5.7176554073233044e-05,
      "loss": 0.0024214066565036775,
      "step": 150930
    },
    {
      "epoch": 42.844166903207494,
      "grad_norm": 0.10814104229211807,
      "learning_rate": 5.7173715583309685e-05,
      "loss": 0.00471544899046421,
      "step": 150940
    },
    {
      "epoch": 42.84700539313086,
      "grad_norm": 2.7155392169952393,
      "learning_rate": 5.717087709338632e-05,
      "loss": 0.007820738852024079,
      "step": 150950
    },
    {
      "epoch": 42.84984388305421,
      "grad_norm": 1.3649444580078125,
      "learning_rate": 5.716803860346296e-05,
      "loss": 0.0033947251737117766,
      "step": 150960
    },
    {
      "epoch": 42.852682372977576,
      "grad_norm": 0.07427273690700531,
      "learning_rate": 5.7165200113539596e-05,
      "loss": 0.002572057209908962,
      "step": 150970
    },
    {
      "epoch": 42.85552086290094,
      "grad_norm": 1.8192777633666992,
      "learning_rate": 5.716236162361624e-05,
      "loss": 0.006395651400089264,
      "step": 150980
    },
    {
      "epoch": 42.858359352824294,
      "grad_norm": 0.9430938363075256,
      "learning_rate": 5.7159523133692885e-05,
      "loss": 0.003343820944428444,
      "step": 150990
    },
    {
      "epoch": 42.86119784274766,
      "grad_norm": 1.028078556060791,
      "learning_rate": 5.715668464376952e-05,
      "loss": 0.001557229645550251,
      "step": 151000
    },
    {
      "epoch": 42.86119784274766,
      "eval_accuracy": 0.9722133909836587,
      "eval_loss": 0.10417237132787704,
      "eval_runtime": 31.6951,
      "eval_samples_per_second": 496.197,
      "eval_steps_per_second": 7.761,
      "step": 151000
    },
    {
      "epoch": 42.86403633267102,
      "grad_norm": 0.6586612462997437,
      "learning_rate": 5.7153846153846154e-05,
      "loss": 0.0047128133475780485,
      "step": 151010
    },
    {
      "epoch": 42.86687482259438,
      "grad_norm": 4.868896961212158,
      "learning_rate": 5.7151007663922796e-05,
      "loss": 0.0025026831775903703,
      "step": 151020
    },
    {
      "epoch": 42.86971331251774,
      "grad_norm": 3.928346872329712,
      "learning_rate": 5.714816917399943e-05,
      "loss": 0.0031259231269359588,
      "step": 151030
    },
    {
      "epoch": 42.8725518024411,
      "grad_norm": 0.04308502748608589,
      "learning_rate": 5.714533068407607e-05,
      "loss": 0.0005053285509347916,
      "step": 151040
    },
    {
      "epoch": 42.875390292364465,
      "grad_norm": 0.033393073827028275,
      "learning_rate": 5.7142492194152706e-05,
      "loss": 0.004916827380657196,
      "step": 151050
    },
    {
      "epoch": 42.87822878228782,
      "grad_norm": 0.1288062483072281,
      "learning_rate": 5.7139653704229355e-05,
      "loss": 0.000999310053884983,
      "step": 151060
    },
    {
      "epoch": 42.88106727221118,
      "grad_norm": 0.020401980727910995,
      "learning_rate": 5.7136815214305996e-05,
      "loss": 0.00369303897023201,
      "step": 151070
    },
    {
      "epoch": 42.883905762134546,
      "grad_norm": 0.49744778871536255,
      "learning_rate": 5.713397672438263e-05,
      "loss": 0.005060420930385589,
      "step": 151080
    },
    {
      "epoch": 42.8867442520579,
      "grad_norm": 0.28286388516426086,
      "learning_rate": 5.713113823445927e-05,
      "loss": 0.002316769398748875,
      "step": 151090
    },
    {
      "epoch": 42.889582741981265,
      "grad_norm": 0.4040142893791199,
      "learning_rate": 5.7128299744535907e-05,
      "loss": 0.0017907632514834403,
      "step": 151100
    },
    {
      "epoch": 42.89242123190463,
      "grad_norm": 8.743243217468262,
      "learning_rate": 5.712546125461254e-05,
      "loss": 0.005298207327723503,
      "step": 151110
    },
    {
      "epoch": 42.89525972182799,
      "grad_norm": 0.48302075266838074,
      "learning_rate": 5.7122622764689196e-05,
      "loss": 0.003266366571187973,
      "step": 151120
    },
    {
      "epoch": 42.898098211751346,
      "grad_norm": 0.40087440609931946,
      "learning_rate": 5.711978427476583e-05,
      "loss": 0.00104418583214283,
      "step": 151130
    },
    {
      "epoch": 42.90093670167471,
      "grad_norm": 0.7117441892623901,
      "learning_rate": 5.7116945784842465e-05,
      "loss": 0.00856827050447464,
      "step": 151140
    },
    {
      "epoch": 42.90377519159807,
      "grad_norm": 0.9934229850769043,
      "learning_rate": 5.7114107294919107e-05,
      "loss": 0.0017387747764587402,
      "step": 151150
    },
    {
      "epoch": 42.90661368152143,
      "grad_norm": 0.20940840244293213,
      "learning_rate": 5.711126880499574e-05,
      "loss": 0.004924660176038742,
      "step": 151160
    },
    {
      "epoch": 42.90945217144479,
      "grad_norm": 2.7715225219726562,
      "learning_rate": 5.710843031507238e-05,
      "loss": 0.0010767662897706032,
      "step": 151170
    },
    {
      "epoch": 42.912290661368154,
      "grad_norm": 0.8113084435462952,
      "learning_rate": 5.710559182514902e-05,
      "loss": 0.0011831255629658698,
      "step": 151180
    },
    {
      "epoch": 42.91512915129151,
      "grad_norm": 0.140447735786438,
      "learning_rate": 5.7102753335225665e-05,
      "loss": 0.005987203493714332,
      "step": 151190
    },
    {
      "epoch": 42.91796764121487,
      "grad_norm": 0.8365229964256287,
      "learning_rate": 5.709991484530231e-05,
      "loss": 0.0024534158408641817,
      "step": 151200
    },
    {
      "epoch": 42.920806131138235,
      "grad_norm": 0.20525787770748138,
      "learning_rate": 5.709707635537894e-05,
      "loss": 0.0019706249237060546,
      "step": 151210
    },
    {
      "epoch": 42.9236446210616,
      "grad_norm": 0.922799289226532,
      "learning_rate": 5.709423786545558e-05,
      "loss": 0.0012364571914076806,
      "step": 151220
    },
    {
      "epoch": 42.926483110984954,
      "grad_norm": 0.5661875605583191,
      "learning_rate": 5.709139937553222e-05,
      "loss": 0.00219468642026186,
      "step": 151230
    },
    {
      "epoch": 42.92932160090832,
      "grad_norm": 0.14702697098255157,
      "learning_rate": 5.708856088560885e-05,
      "loss": 0.00042118672281503676,
      "step": 151240
    },
    {
      "epoch": 42.93216009083168,
      "grad_norm": 4.3587846755981445,
      "learning_rate": 5.708572239568549e-05,
      "loss": 0.007931852340698242,
      "step": 151250
    },
    {
      "epoch": 42.934998580755035,
      "grad_norm": 0.2341473400592804,
      "learning_rate": 5.708288390576214e-05,
      "loss": 0.012785941362380981,
      "step": 151260
    },
    {
      "epoch": 42.9378370706784,
      "grad_norm": 0.7541271448135376,
      "learning_rate": 5.7080045415838776e-05,
      "loss": 0.003152751922607422,
      "step": 151270
    },
    {
      "epoch": 42.94067556060176,
      "grad_norm": 0.042412642389535904,
      "learning_rate": 5.707720692591542e-05,
      "loss": 0.0014407865703105927,
      "step": 151280
    },
    {
      "epoch": 42.943514050525124,
      "grad_norm": 1.1052659749984741,
      "learning_rate": 5.707436843599205e-05,
      "loss": 0.006433963775634766,
      "step": 151290
    },
    {
      "epoch": 42.94635254044848,
      "grad_norm": 0.497271329164505,
      "learning_rate": 5.707152994606869e-05,
      "loss": 0.0021620284765958785,
      "step": 151300
    },
    {
      "epoch": 42.94919103037184,
      "grad_norm": 1.0877059698104858,
      "learning_rate": 5.706869145614533e-05,
      "loss": 0.0011028502136468886,
      "step": 151310
    },
    {
      "epoch": 42.952029520295206,
      "grad_norm": 0.01732019893825054,
      "learning_rate": 5.7065852966221976e-05,
      "loss": 0.0009060354903340339,
      "step": 151320
    },
    {
      "epoch": 42.95486801021856,
      "grad_norm": 0.38991209864616394,
      "learning_rate": 5.706301447629862e-05,
      "loss": 0.0012246167287230491,
      "step": 151330
    },
    {
      "epoch": 42.957706500141924,
      "grad_norm": 1.1046416759490967,
      "learning_rate": 5.706017598637525e-05,
      "loss": 0.001381961815059185,
      "step": 151340
    },
    {
      "epoch": 42.96054499006529,
      "grad_norm": 1.6200536489486694,
      "learning_rate": 5.7057337496451893e-05,
      "loss": 0.0009480997920036316,
      "step": 151350
    },
    {
      "epoch": 42.96338347998864,
      "grad_norm": 0.43610674142837524,
      "learning_rate": 5.705449900652853e-05,
      "loss": 0.0005201894789934158,
      "step": 151360
    },
    {
      "epoch": 42.966221969912006,
      "grad_norm": 2.9958713054656982,
      "learning_rate": 5.705166051660516e-05,
      "loss": 0.004076436161994934,
      "step": 151370
    },
    {
      "epoch": 42.96906045983537,
      "grad_norm": 0.07383344322443008,
      "learning_rate": 5.7048822026681804e-05,
      "loss": 0.004621308296918869,
      "step": 151380
    },
    {
      "epoch": 42.97189894975873,
      "grad_norm": 0.1340925693511963,
      "learning_rate": 5.704598353675845e-05,
      "loss": 0.011564357578754425,
      "step": 151390
    },
    {
      "epoch": 42.97473743968209,
      "grad_norm": 1.1192890405654907,
      "learning_rate": 5.704314504683509e-05,
      "loss": 0.028056687116622923,
      "step": 151400
    },
    {
      "epoch": 42.97757592960545,
      "grad_norm": 0.548682451248169,
      "learning_rate": 5.704030655691173e-05,
      "loss": 0.005934283137321472,
      "step": 151410
    },
    {
      "epoch": 42.98041441952881,
      "grad_norm": 0.28245776891708374,
      "learning_rate": 5.703746806698836e-05,
      "loss": 0.005049395561218262,
      "step": 151420
    },
    {
      "epoch": 42.98325290945217,
      "grad_norm": 0.7276083827018738,
      "learning_rate": 5.7034629577065004e-05,
      "loss": 0.004849177971482277,
      "step": 151430
    },
    {
      "epoch": 42.98609139937553,
      "grad_norm": 0.018039369955658913,
      "learning_rate": 5.703179108714164e-05,
      "loss": 0.0021641574800014495,
      "step": 151440
    },
    {
      "epoch": 42.988929889298895,
      "grad_norm": 0.6865327954292297,
      "learning_rate": 5.702895259721829e-05,
      "loss": 0.004317700490355491,
      "step": 151450
    },
    {
      "epoch": 42.99176837922225,
      "grad_norm": 0.07889662683010101,
      "learning_rate": 5.702611410729493e-05,
      "loss": 0.0008823204785585404,
      "step": 151460
    },
    {
      "epoch": 42.99460686914561,
      "grad_norm": 0.14053185284137726,
      "learning_rate": 5.702327561737156e-05,
      "loss": 0.0097741961479187,
      "step": 151470
    },
    {
      "epoch": 42.997445359068976,
      "grad_norm": 0.3924703299999237,
      "learning_rate": 5.70204371274482e-05,
      "loss": 0.016547949612140657,
      "step": 151480
    },
    {
      "epoch": 43.00028384899234,
      "grad_norm": 0.03356565162539482,
      "learning_rate": 5.701788248651717e-05,
      "loss": 0.005016471818089485,
      "step": 151490
    },
    {
      "epoch": 43.003122338915695,
      "grad_norm": 0.2132159173488617,
      "learning_rate": 5.701504399659382e-05,
      "loss": 0.00340183712542057,
      "step": 151500
    },
    {
      "epoch": 43.003122338915695,
      "eval_accuracy": 0.9760920709607681,
      "eval_loss": 0.08960431069135666,
      "eval_runtime": 31.7302,
      "eval_samples_per_second": 495.648,
      "eval_steps_per_second": 7.753,
      "step": 151500
    },
    {
      "epoch": 43.00596082883906,
      "grad_norm": 0.07106353342533112,
      "learning_rate": 5.701220550667046e-05,
      "loss": 0.00279498677700758,
      "step": 151510
    },
    {
      "epoch": 43.00879931876242,
      "grad_norm": 0.027225233614444733,
      "learning_rate": 5.7009367016747095e-05,
      "loss": 0.0008680166676640511,
      "step": 151520
    },
    {
      "epoch": 43.01163780868578,
      "grad_norm": 0.24606360495090485,
      "learning_rate": 5.700652852682373e-05,
      "loss": 0.0014991562813520432,
      "step": 151530
    },
    {
      "epoch": 43.01447629860914,
      "grad_norm": 0.9538046717643738,
      "learning_rate": 5.700369003690037e-05,
      "loss": 0.0037996582686901093,
      "step": 151540
    },
    {
      "epoch": 43.0173147885325,
      "grad_norm": 7.158541202545166,
      "learning_rate": 5.7000851546977005e-05,
      "loss": 0.004549169540405273,
      "step": 151550
    },
    {
      "epoch": 43.02015327845586,
      "grad_norm": 0.0938863456249237,
      "learning_rate": 5.699801305705365e-05,
      "loss": 0.0009634293615818024,
      "step": 151560
    },
    {
      "epoch": 43.02299176837922,
      "grad_norm": 0.6828452944755554,
      "learning_rate": 5.6995174567130295e-05,
      "loss": 0.0008728491142392158,
      "step": 151570
    },
    {
      "epoch": 43.025830258302584,
      "grad_norm": 0.28954392671585083,
      "learning_rate": 5.699233607720693e-05,
      "loss": 0.0006391836330294609,
      "step": 151580
    },
    {
      "epoch": 43.02866874822595,
      "grad_norm": 7.125875949859619,
      "learning_rate": 5.698949758728357e-05,
      "loss": 0.0023804975673556326,
      "step": 151590
    },
    {
      "epoch": 43.0315072381493,
      "grad_norm": 0.01594804972410202,
      "learning_rate": 5.6986659097360205e-05,
      "loss": 0.00042392611503601075,
      "step": 151600
    },
    {
      "epoch": 43.034345728072665,
      "grad_norm": 0.08081094175577164,
      "learning_rate": 5.698382060743685e-05,
      "loss": 0.00041200965642929077,
      "step": 151610
    },
    {
      "epoch": 43.03718421799603,
      "grad_norm": 4.560952663421631,
      "learning_rate": 5.698098211751348e-05,
      "loss": 0.00442599281668663,
      "step": 151620
    },
    {
      "epoch": 43.040022707919384,
      "grad_norm": 0.04551665857434273,
      "learning_rate": 5.6978143627590116e-05,
      "loss": 0.0030796565115451813,
      "step": 151630
    },
    {
      "epoch": 43.04286119784275,
      "grad_norm": 0.22345243394374847,
      "learning_rate": 5.697530513766677e-05,
      "loss": 0.000508817471563816,
      "step": 151640
    },
    {
      "epoch": 43.04569968776611,
      "grad_norm": 0.09498488903045654,
      "learning_rate": 5.6972466647743405e-05,
      "loss": 0.0006739828735589981,
      "step": 151650
    },
    {
      "epoch": 43.04853817768947,
      "grad_norm": 0.01621210016310215,
      "learning_rate": 5.696962815782004e-05,
      "loss": 0.0006063472479581833,
      "step": 151660
    },
    {
      "epoch": 43.05137666761283,
      "grad_norm": 0.10096614062786102,
      "learning_rate": 5.696678966789668e-05,
      "loss": 0.003948544710874557,
      "step": 151670
    },
    {
      "epoch": 43.05421515753619,
      "grad_norm": 0.7879394292831421,
      "learning_rate": 5.6963951177973316e-05,
      "loss": 0.008247603476047517,
      "step": 151680
    },
    {
      "epoch": 43.057053647459554,
      "grad_norm": 0.710813581943512,
      "learning_rate": 5.696111268804996e-05,
      "loss": 0.00279113557189703,
      "step": 151690
    },
    {
      "epoch": 43.05989213738291,
      "grad_norm": 3.666478395462036,
      "learning_rate": 5.6958274198126606e-05,
      "loss": 0.0014192681759595872,
      "step": 151700
    },
    {
      "epoch": 43.06273062730627,
      "grad_norm": 0.4814639985561371,
      "learning_rate": 5.695543570820324e-05,
      "loss": 0.00549580454826355,
      "step": 151710
    },
    {
      "epoch": 43.065569117229636,
      "grad_norm": 0.05331253260374069,
      "learning_rate": 5.695259721827988e-05,
      "loss": 0.006542760878801346,
      "step": 151720
    },
    {
      "epoch": 43.06840760715299,
      "grad_norm": 0.4315067231655121,
      "learning_rate": 5.6949758728356516e-05,
      "loss": 0.0015864398330450058,
      "step": 151730
    },
    {
      "epoch": 43.071246097076354,
      "grad_norm": 0.2586740255355835,
      "learning_rate": 5.694692023843316e-05,
      "loss": 0.0007991410791873931,
      "step": 151740
    },
    {
      "epoch": 43.07408458699972,
      "grad_norm": 0.6832228899002075,
      "learning_rate": 5.694408174850979e-05,
      "loss": 0.006490696966648102,
      "step": 151750
    },
    {
      "epoch": 43.07692307692308,
      "grad_norm": 0.3536789119243622,
      "learning_rate": 5.694124325858643e-05,
      "loss": 0.0008261259645223618,
      "step": 151760
    },
    {
      "epoch": 43.079761566846436,
      "grad_norm": 12.853556632995605,
      "learning_rate": 5.693840476866308e-05,
      "loss": 0.0036619093269109728,
      "step": 151770
    },
    {
      "epoch": 43.0826000567698,
      "grad_norm": 0.08694365620613098,
      "learning_rate": 5.6935566278739716e-05,
      "loss": 0.0014129746705293655,
      "step": 151780
    },
    {
      "epoch": 43.08543854669316,
      "grad_norm": 13.35361099243164,
      "learning_rate": 5.693272778881635e-05,
      "loss": 0.006362512707710266,
      "step": 151790
    },
    {
      "epoch": 43.08827703661652,
      "grad_norm": 0.07861967384815216,
      "learning_rate": 5.692988929889299e-05,
      "loss": 0.005175063014030456,
      "step": 151800
    },
    {
      "epoch": 43.09111552653988,
      "grad_norm": 0.13682618737220764,
      "learning_rate": 5.692705080896963e-05,
      "loss": 0.001588154025375843,
      "step": 151810
    },
    {
      "epoch": 43.09395401646324,
      "grad_norm": 0.07961110770702362,
      "learning_rate": 5.692421231904627e-05,
      "loss": 0.008250161260366439,
      "step": 151820
    },
    {
      "epoch": 43.0967925063866,
      "grad_norm": 0.10837347060441971,
      "learning_rate": 5.6921373829122916e-05,
      "loss": 0.0005296060815453529,
      "step": 151830
    },
    {
      "epoch": 43.09963099630996,
      "grad_norm": 0.0552312433719635,
      "learning_rate": 5.691853533919955e-05,
      "loss": 0.0008237535133957863,
      "step": 151840
    },
    {
      "epoch": 43.102469486233325,
      "grad_norm": 0.12459081411361694,
      "learning_rate": 5.691569684927619e-05,
      "loss": 0.0017899278551340102,
      "step": 151850
    },
    {
      "epoch": 43.10530797615669,
      "grad_norm": 0.14180979132652283,
      "learning_rate": 5.691285835935283e-05,
      "loss": 0.0005112074315547944,
      "step": 151860
    },
    {
      "epoch": 43.10814646608004,
      "grad_norm": 0.10183339565992355,
      "learning_rate": 5.691001986942947e-05,
      "loss": 0.00047392938286066054,
      "step": 151870
    },
    {
      "epoch": 43.110984956003406,
      "grad_norm": 0.03730219975113869,
      "learning_rate": 5.69071813795061e-05,
      "loss": 0.0017279006540775299,
      "step": 151880
    },
    {
      "epoch": 43.11382344592677,
      "grad_norm": 0.031032048165798187,
      "learning_rate": 5.690434288958274e-05,
      "loss": 0.00019947458058595658,
      "step": 151890
    },
    {
      "epoch": 43.116661935850125,
      "grad_norm": 0.07724631577730179,
      "learning_rate": 5.6901504399659386e-05,
      "loss": 0.0005313152447342873,
      "step": 151900
    },
    {
      "epoch": 43.11950042577349,
      "grad_norm": 0.5991203188896179,
      "learning_rate": 5.689866590973603e-05,
      "loss": 0.0011872129514813423,
      "step": 151910
    },
    {
      "epoch": 43.12233891569685,
      "grad_norm": 0.027738042175769806,
      "learning_rate": 5.689582741981266e-05,
      "loss": 0.005116463452577591,
      "step": 151920
    },
    {
      "epoch": 43.12517740562021,
      "grad_norm": 0.03408701345324516,
      "learning_rate": 5.68929889298893e-05,
      "loss": 0.0005747660994529724,
      "step": 151930
    },
    {
      "epoch": 43.12801589554357,
      "grad_norm": 4.866060733795166,
      "learning_rate": 5.689015043996594e-05,
      "loss": 0.0018263310194015503,
      "step": 151940
    },
    {
      "epoch": 43.13085438546693,
      "grad_norm": 0.172432541847229,
      "learning_rate": 5.688731195004258e-05,
      "loss": 0.0006231596693396568,
      "step": 151950
    },
    {
      "epoch": 43.133692875390295,
      "grad_norm": 0.8032639622688293,
      "learning_rate": 5.6884473460119214e-05,
      "loss": 0.001492762193083763,
      "step": 151960
    },
    {
      "epoch": 43.13653136531365,
      "grad_norm": 0.5606970191001892,
      "learning_rate": 5.688163497019586e-05,
      "loss": 0.0009769955649971962,
      "step": 151970
    },
    {
      "epoch": 43.139369855237014,
      "grad_norm": 0.03375739976763725,
      "learning_rate": 5.68787964802725e-05,
      "loss": 0.0031879890710115434,
      "step": 151980
    },
    {
      "epoch": 43.14220834516038,
      "grad_norm": 0.23029004037380219,
      "learning_rate": 5.687595799034914e-05,
      "loss": 0.004812368005514145,
      "step": 151990
    },
    {
      "epoch": 43.14504683508373,
      "grad_norm": 0.1062697172164917,
      "learning_rate": 5.687311950042577e-05,
      "loss": 0.0015996862202882767,
      "step": 152000
    },
    {
      "epoch": 43.14504683508373,
      "eval_accuracy": 0.972849240160234,
      "eval_loss": 0.09538374096155167,
      "eval_runtime": 32.0064,
      "eval_samples_per_second": 491.371,
      "eval_steps_per_second": 7.686,
      "step": 152000
    },
    {
      "epoch": 43.147885325007096,
      "grad_norm": 0.08808532357215881,
      "learning_rate": 5.6870281010502414e-05,
      "loss": 0.0010589540004730224,
      "step": 152010
    },
    {
      "epoch": 43.15072381493046,
      "grad_norm": 0.03220302239060402,
      "learning_rate": 5.686744252057905e-05,
      "loss": 0.0013726312667131425,
      "step": 152020
    },
    {
      "epoch": 43.15356230485382,
      "grad_norm": 0.15309451520442963,
      "learning_rate": 5.6864604030655696e-05,
      "loss": 0.0019029887393116952,
      "step": 152030
    },
    {
      "epoch": 43.15640079477718,
      "grad_norm": 0.40568867325782776,
      "learning_rate": 5.686176554073234e-05,
      "loss": 0.002128172852098942,
      "step": 152040
    },
    {
      "epoch": 43.15923928470054,
      "grad_norm": 0.2462402731180191,
      "learning_rate": 5.685892705080897e-05,
      "loss": 0.003699745610356331,
      "step": 152050
    },
    {
      "epoch": 43.1620777746239,
      "grad_norm": 0.0401785671710968,
      "learning_rate": 5.6856088560885614e-05,
      "loss": 0.0003518974408507347,
      "step": 152060
    },
    {
      "epoch": 43.16491626454726,
      "grad_norm": 1.8910232782363892,
      "learning_rate": 5.685325007096225e-05,
      "loss": 0.0012026021257042884,
      "step": 152070
    },
    {
      "epoch": 43.16775475447062,
      "grad_norm": 0.08961507678031921,
      "learning_rate": 5.685041158103889e-05,
      "loss": 0.0006802882999181747,
      "step": 152080
    },
    {
      "epoch": 43.170593244393984,
      "grad_norm": 2.1218137741088867,
      "learning_rate": 5.6847573091115524e-05,
      "loss": 0.0014594115316867828,
      "step": 152090
    },
    {
      "epoch": 43.17343173431734,
      "grad_norm": 0.34050852060317993,
      "learning_rate": 5.684473460119217e-05,
      "loss": 0.0009594541043043137,
      "step": 152100
    },
    {
      "epoch": 43.1762702242407,
      "grad_norm": 0.06501282751560211,
      "learning_rate": 5.6841896111268814e-05,
      "loss": 0.002069770358502865,
      "step": 152110
    },
    {
      "epoch": 43.179108714164066,
      "grad_norm": 5.9305949211120605,
      "learning_rate": 5.683905762134545e-05,
      "loss": 0.002483154460787773,
      "step": 152120
    },
    {
      "epoch": 43.18194720408743,
      "grad_norm": 1.2521289587020874,
      "learning_rate": 5.683621913142208e-05,
      "loss": 0.018283867835998537,
      "step": 152130
    },
    {
      "epoch": 43.184785694010785,
      "grad_norm": 1.1412521600723267,
      "learning_rate": 5.6833380641498724e-05,
      "loss": 0.0013024747371673585,
      "step": 152140
    },
    {
      "epoch": 43.18762418393415,
      "grad_norm": 4.002958297729492,
      "learning_rate": 5.683054215157536e-05,
      "loss": 0.004960101097822189,
      "step": 152150
    },
    {
      "epoch": 43.19046267385751,
      "grad_norm": 6.770664215087891,
      "learning_rate": 5.6827703661652e-05,
      "loss": 0.003757569193840027,
      "step": 152160
    },
    {
      "epoch": 43.193301163780866,
      "grad_norm": 0.5399791598320007,
      "learning_rate": 5.682486517172865e-05,
      "loss": 0.002188009396195412,
      "step": 152170
    },
    {
      "epoch": 43.19613965370423,
      "grad_norm": 1.0973259210586548,
      "learning_rate": 5.682202668180528e-05,
      "loss": 0.0016393065452575684,
      "step": 152180
    },
    {
      "epoch": 43.19897814362759,
      "grad_norm": 0.5125619769096375,
      "learning_rate": 5.6819188191881925e-05,
      "loss": 0.0006142444908618927,
      "step": 152190
    },
    {
      "epoch": 43.20181663355095,
      "grad_norm": 0.2593333125114441,
      "learning_rate": 5.681634970195856e-05,
      "loss": 0.011070017516613007,
      "step": 152200
    },
    {
      "epoch": 43.20465512347431,
      "grad_norm": 0.46329739689826965,
      "learning_rate": 5.68135112120352e-05,
      "loss": 0.0012663915753364564,
      "step": 152210
    },
    {
      "epoch": 43.20749361339767,
      "grad_norm": 0.21926258504390717,
      "learning_rate": 5.6810672722111835e-05,
      "loss": 0.003219062089920044,
      "step": 152220
    },
    {
      "epoch": 43.210332103321036,
      "grad_norm": 0.053675517439842224,
      "learning_rate": 5.680783423218848e-05,
      "loss": 0.0050923660397529606,
      "step": 152230
    },
    {
      "epoch": 43.21317059324439,
      "grad_norm": 0.04515977203845978,
      "learning_rate": 5.6804995742265125e-05,
      "loss": 0.011871426552534103,
      "step": 152240
    },
    {
      "epoch": 43.216009083167755,
      "grad_norm": 0.016196435317397118,
      "learning_rate": 5.680215725234176e-05,
      "loss": 0.001237376034259796,
      "step": 152250
    },
    {
      "epoch": 43.21884757309112,
      "grad_norm": 0.21483221650123596,
      "learning_rate": 5.6799318762418394e-05,
      "loss": 0.003402319550514221,
      "step": 152260
    },
    {
      "epoch": 43.221686063014474,
      "grad_norm": 2.240077495574951,
      "learning_rate": 5.6796480272495035e-05,
      "loss": 0.0026287468150258063,
      "step": 152270
    },
    {
      "epoch": 43.22452455293784,
      "grad_norm": 0.21066312491893768,
      "learning_rate": 5.679364178257167e-05,
      "loss": 0.0008545840159058571,
      "step": 152280
    },
    {
      "epoch": 43.2273630428612,
      "grad_norm": 0.1933826506137848,
      "learning_rate": 5.679080329264831e-05,
      "loss": 0.0016031524166464806,
      "step": 152290
    },
    {
      "epoch": 43.230201532784555,
      "grad_norm": 0.7001194953918457,
      "learning_rate": 5.678796480272496e-05,
      "loss": 0.010772044211626053,
      "step": 152300
    },
    {
      "epoch": 43.23304002270792,
      "grad_norm": 0.11431702226400375,
      "learning_rate": 5.6785126312801594e-05,
      "loss": 0.014571861922740936,
      "step": 152310
    },
    {
      "epoch": 43.23587851263128,
      "grad_norm": 0.2644888460636139,
      "learning_rate": 5.6782287822878235e-05,
      "loss": 0.0034686725586652754,
      "step": 152320
    },
    {
      "epoch": 43.238717002554644,
      "grad_norm": 12.104090690612793,
      "learning_rate": 5.677944933295487e-05,
      "loss": 0.0033518128097057343,
      "step": 152330
    },
    {
      "epoch": 43.241555492478,
      "grad_norm": 0.16671255230903625,
      "learning_rate": 5.677661084303151e-05,
      "loss": 0.001333530992269516,
      "step": 152340
    },
    {
      "epoch": 43.24439398240136,
      "grad_norm": 0.08636177331209183,
      "learning_rate": 5.6773772353108146e-05,
      "loss": 0.004307601600885391,
      "step": 152350
    },
    {
      "epoch": 43.247232472324725,
      "grad_norm": 0.13912631571292877,
      "learning_rate": 5.677093386318478e-05,
      "loss": 0.0027288509532809257,
      "step": 152360
    },
    {
      "epoch": 43.25007096224808,
      "grad_norm": 14.268440246582031,
      "learning_rate": 5.676809537326143e-05,
      "loss": 0.00548018142580986,
      "step": 152370
    },
    {
      "epoch": 43.252909452171444,
      "grad_norm": 0.11489082127809525,
      "learning_rate": 5.676525688333807e-05,
      "loss": 0.0015092611312866211,
      "step": 152380
    },
    {
      "epoch": 43.25574794209481,
      "grad_norm": 0.13625004887580872,
      "learning_rate": 5.6762418393414705e-05,
      "loss": 0.004902688413858413,
      "step": 152390
    },
    {
      "epoch": 43.25858643201816,
      "grad_norm": 0.4013904631137848,
      "learning_rate": 5.6759579903491346e-05,
      "loss": 0.0045593492686748505,
      "step": 152400
    },
    {
      "epoch": 43.261424921941526,
      "grad_norm": 0.13857543468475342,
      "learning_rate": 5.675674141356798e-05,
      "loss": 0.0010303063318133354,
      "step": 152410
    },
    {
      "epoch": 43.26426341186489,
      "grad_norm": 0.029036447405815125,
      "learning_rate": 5.675390292364462e-05,
      "loss": 0.0005287723615765572,
      "step": 152420
    },
    {
      "epoch": 43.26710190178825,
      "grad_norm": 0.019244540482759476,
      "learning_rate": 5.675106443372127e-05,
      "loss": 0.0010225621983408928,
      "step": 152430
    },
    {
      "epoch": 43.26994039171161,
      "grad_norm": 0.07808752357959747,
      "learning_rate": 5.6748225943797905e-05,
      "loss": 0.0012760989367961884,
      "step": 152440
    },
    {
      "epoch": 43.27277888163497,
      "grad_norm": 1.2281979322433472,
      "learning_rate": 5.6745387453874546e-05,
      "loss": 0.005228409916162491,
      "step": 152450
    },
    {
      "epoch": 43.27561737155833,
      "grad_norm": 0.09017087519168854,
      "learning_rate": 5.674254896395118e-05,
      "loss": 0.006734376400709152,
      "step": 152460
    },
    {
      "epoch": 43.27845586148169,
      "grad_norm": 0.07635440677404404,
      "learning_rate": 5.6739710474027815e-05,
      "loss": 0.003857327625155449,
      "step": 152470
    },
    {
      "epoch": 43.28129435140505,
      "grad_norm": 0.16468802094459534,
      "learning_rate": 5.673687198410446e-05,
      "loss": 0.0010821793228387832,
      "step": 152480
    },
    {
      "epoch": 43.284132841328415,
      "grad_norm": 2.002392053604126,
      "learning_rate": 5.673403349418109e-05,
      "loss": 0.0011932019144296647,
      "step": 152490
    },
    {
      "epoch": 43.28697133125178,
      "grad_norm": 1.8941805362701416,
      "learning_rate": 5.673119500425774e-05,
      "loss": 0.005160294473171234,
      "step": 152500
    },
    {
      "epoch": 43.28697133125178,
      "eval_accuracy": 0.9726584854072614,
      "eval_loss": 0.10405123978853226,
      "eval_runtime": 31.9196,
      "eval_samples_per_second": 492.706,
      "eval_steps_per_second": 7.707,
      "step": 152500
    },
    {
      "epoch": 43.28980982117513,
      "grad_norm": 14.314725875854492,
      "learning_rate": 5.672835651433438e-05,
      "loss": 0.010491292178630828,
      "step": 152510
    },
    {
      "epoch": 43.292648311098496,
      "grad_norm": 0.1541241705417633,
      "learning_rate": 5.6725518024411015e-05,
      "loss": 0.01071455106139183,
      "step": 152520
    },
    {
      "epoch": 43.29548680102186,
      "grad_norm": 0.5924636721611023,
      "learning_rate": 5.672267953448766e-05,
      "loss": 0.011588039994239806,
      "step": 152530
    },
    {
      "epoch": 43.298325290945215,
      "grad_norm": 0.11296749860048294,
      "learning_rate": 5.671984104456429e-05,
      "loss": 0.0007361149415373802,
      "step": 152540
    },
    {
      "epoch": 43.30116378086858,
      "grad_norm": 0.07744500041007996,
      "learning_rate": 5.671700255464093e-05,
      "loss": 0.0013620154932141304,
      "step": 152550
    },
    {
      "epoch": 43.30400227079194,
      "grad_norm": 0.22644060850143433,
      "learning_rate": 5.671416406471757e-05,
      "loss": 0.002292870730161667,
      "step": 152560
    },
    {
      "epoch": 43.306840760715296,
      "grad_norm": 7.088444232940674,
      "learning_rate": 5.6711325574794215e-05,
      "loss": 0.004728332161903381,
      "step": 152570
    },
    {
      "epoch": 43.30967925063866,
      "grad_norm": 0.06695357710123062,
      "learning_rate": 5.670848708487086e-05,
      "loss": 0.0009051904082298279,
      "step": 152580
    },
    {
      "epoch": 43.31251774056202,
      "grad_norm": 19.764301300048828,
      "learning_rate": 5.670564859494749e-05,
      "loss": 0.012910376489162444,
      "step": 152590
    },
    {
      "epoch": 43.315356230485385,
      "grad_norm": 19.718048095703125,
      "learning_rate": 5.6702810105024126e-05,
      "loss": 0.010928825289011002,
      "step": 152600
    },
    {
      "epoch": 43.31819472040874,
      "grad_norm": 1.0078260898590088,
      "learning_rate": 5.669997161510077e-05,
      "loss": 0.007726716995239258,
      "step": 152610
    },
    {
      "epoch": 43.321033210332104,
      "grad_norm": 0.3209133744239807,
      "learning_rate": 5.66971331251774e-05,
      "loss": 0.015328006446361541,
      "step": 152620
    },
    {
      "epoch": 43.32387170025547,
      "grad_norm": 0.12341020256280899,
      "learning_rate": 5.669429463525405e-05,
      "loss": 0.0026687515899538996,
      "step": 152630
    },
    {
      "epoch": 43.32671019017882,
      "grad_norm": 1.8654800653457642,
      "learning_rate": 5.669145614533069e-05,
      "loss": 0.003351179510354996,
      "step": 152640
    },
    {
      "epoch": 43.329548680102185,
      "grad_norm": 0.21532008051872253,
      "learning_rate": 5.6688617655407326e-05,
      "loss": 0.0014159915968775748,
      "step": 152650
    },
    {
      "epoch": 43.33238717002555,
      "grad_norm": 0.03760688006877899,
      "learning_rate": 5.668577916548397e-05,
      "loss": 0.0011512348428368568,
      "step": 152660
    },
    {
      "epoch": 43.335225659948904,
      "grad_norm": 0.3844260573387146,
      "learning_rate": 5.66829406755606e-05,
      "loss": 0.0010544518008828163,
      "step": 152670
    },
    {
      "epoch": 43.33806414987227,
      "grad_norm": 0.2905004024505615,
      "learning_rate": 5.6680102185637243e-05,
      "loss": 0.003919313848018646,
      "step": 152680
    },
    {
      "epoch": 43.34090263979563,
      "grad_norm": 9.121471405029297,
      "learning_rate": 5.667726369571388e-05,
      "loss": 0.005170821771025657,
      "step": 152690
    },
    {
      "epoch": 43.34374112971899,
      "grad_norm": 6.7965240478515625,
      "learning_rate": 5.6674425205790526e-05,
      "loss": 0.002664828673005104,
      "step": 152700
    },
    {
      "epoch": 43.34657961964235,
      "grad_norm": 0.19793333113193512,
      "learning_rate": 5.667158671586717e-05,
      "loss": 0.0025952840223908424,
      "step": 152710
    },
    {
      "epoch": 43.34941810956571,
      "grad_norm": 0.07345113158226013,
      "learning_rate": 5.66687482259438e-05,
      "loss": 0.002797003835439682,
      "step": 152720
    },
    {
      "epoch": 43.352256599489074,
      "grad_norm": 1.87361741065979,
      "learning_rate": 5.666590973602044e-05,
      "loss": 0.002520333044230938,
      "step": 152730
    },
    {
      "epoch": 43.35509508941243,
      "grad_norm": 0.07221171259880066,
      "learning_rate": 5.666307124609708e-05,
      "loss": 0.0005156448110938072,
      "step": 152740
    },
    {
      "epoch": 43.35793357933579,
      "grad_norm": 0.06739336252212524,
      "learning_rate": 5.666023275617371e-05,
      "loss": 0.0004887999966740608,
      "step": 152750
    },
    {
      "epoch": 43.360772069259156,
      "grad_norm": 5.223864555358887,
      "learning_rate": 5.6657394266250354e-05,
      "loss": 0.002117645926773548,
      "step": 152760
    },
    {
      "epoch": 43.36361055918251,
      "grad_norm": 0.016317103058099747,
      "learning_rate": 5.6654555776327e-05,
      "loss": 0.0007661398500204087,
      "step": 152770
    },
    {
      "epoch": 43.366449049105874,
      "grad_norm": 13.086366653442383,
      "learning_rate": 5.665171728640364e-05,
      "loss": 0.004624612629413605,
      "step": 152780
    },
    {
      "epoch": 43.36928753902924,
      "grad_norm": 0.2688022255897522,
      "learning_rate": 5.664887879648028e-05,
      "loss": 0.004203634709119797,
      "step": 152790
    },
    {
      "epoch": 43.3721260289526,
      "grad_norm": 1.2943884134292603,
      "learning_rate": 5.664604030655691e-05,
      "loss": 0.0012246865779161453,
      "step": 152800
    },
    {
      "epoch": 43.374964518875956,
      "grad_norm": 0.2189081907272339,
      "learning_rate": 5.6643201816633554e-05,
      "loss": 0.001220511645078659,
      "step": 152810
    },
    {
      "epoch": 43.37780300879932,
      "grad_norm": 4.960872650146484,
      "learning_rate": 5.664036332671019e-05,
      "loss": 0.0021679319441318514,
      "step": 152820
    },
    {
      "epoch": 43.38064149872268,
      "grad_norm": 0.41459572315216064,
      "learning_rate": 5.663752483678684e-05,
      "loss": 0.001709306053817272,
      "step": 152830
    },
    {
      "epoch": 43.38347998864604,
      "grad_norm": 2.584622383117676,
      "learning_rate": 5.663468634686347e-05,
      "loss": 0.0014330632984638214,
      "step": 152840
    },
    {
      "epoch": 43.3863184785694,
      "grad_norm": 0.9256609082221985,
      "learning_rate": 5.663184785694011e-05,
      "loss": 0.0019030198454856873,
      "step": 152850
    },
    {
      "epoch": 43.38915696849276,
      "grad_norm": 0.9857291579246521,
      "learning_rate": 5.662900936701675e-05,
      "loss": 0.002566688321530819,
      "step": 152860
    },
    {
      "epoch": 43.391995458416126,
      "grad_norm": 0.2584203779697418,
      "learning_rate": 5.662617087709339e-05,
      "loss": 0.00041969213634729385,
      "step": 152870
    },
    {
      "epoch": 43.39483394833948,
      "grad_norm": 1.7201082706451416,
      "learning_rate": 5.6623332387170024e-05,
      "loss": 0.0028254957869648933,
      "step": 152880
    },
    {
      "epoch": 43.397672438262845,
      "grad_norm": 0.7246617674827576,
      "learning_rate": 5.6620493897246665e-05,
      "loss": 0.0013693271204829216,
      "step": 152890
    },
    {
      "epoch": 43.40051092818621,
      "grad_norm": 0.24480748176574707,
      "learning_rate": 5.661765540732331e-05,
      "loss": 0.002864978089928627,
      "step": 152900
    },
    {
      "epoch": 43.40334941810956,
      "grad_norm": 0.11099433898925781,
      "learning_rate": 5.661481691739995e-05,
      "loss": 0.0009727595373988151,
      "step": 152910
    },
    {
      "epoch": 43.406187908032926,
      "grad_norm": 0.009493552148342133,
      "learning_rate": 5.661197842747659e-05,
      "loss": 0.0004898803308606148,
      "step": 152920
    },
    {
      "epoch": 43.40902639795629,
      "grad_norm": 0.7696323394775391,
      "learning_rate": 5.6609139937553224e-05,
      "loss": 0.0018105385825037956,
      "step": 152930
    },
    {
      "epoch": 43.411864887879645,
      "grad_norm": 10.798267364501953,
      "learning_rate": 5.660630144762986e-05,
      "loss": 0.007064898312091827,
      "step": 152940
    },
    {
      "epoch": 43.41470337780301,
      "grad_norm": 1.795678973197937,
      "learning_rate": 5.66034629577065e-05,
      "loss": 0.007675459980964661,
      "step": 152950
    },
    {
      "epoch": 43.41754186772637,
      "grad_norm": 0.4045662581920624,
      "learning_rate": 5.6600624467783134e-05,
      "loss": 0.0015947656705975533,
      "step": 152960
    },
    {
      "epoch": 43.420380357649734,
      "grad_norm": 0.3066219091415405,
      "learning_rate": 5.659778597785978e-05,
      "loss": 0.002019953913986683,
      "step": 152970
    },
    {
      "epoch": 43.42321884757309,
      "grad_norm": 0.12634728848934174,
      "learning_rate": 5.6594947487936424e-05,
      "loss": 0.008224523067474366,
      "step": 152980
    },
    {
      "epoch": 43.42605733749645,
      "grad_norm": 1.460869312286377,
      "learning_rate": 5.659210899801306e-05,
      "loss": 0.0020893769338726997,
      "step": 152990
    },
    {
      "epoch": 43.428895827419815,
      "grad_norm": 3.644944667816162,
      "learning_rate": 5.65892705080897e-05,
      "loss": 0.004289470613002777,
      "step": 153000
    },
    {
      "epoch": 43.428895827419815,
      "eval_accuracy": 0.9748203726076174,
      "eval_loss": 0.09331263601779938,
      "eval_runtime": 32.6677,
      "eval_samples_per_second": 481.423,
      "eval_steps_per_second": 7.53,
      "step": 153000
    },
    {
      "epoch": 43.43173431734317,
      "grad_norm": 1.3801361322402954,
      "learning_rate": 5.6586432018166334e-05,
      "loss": 0.0017812473699450493,
      "step": 153010
    },
    {
      "epoch": 43.434572807266534,
      "grad_norm": 0.2670861482620239,
      "learning_rate": 5.6583593528242976e-05,
      "loss": 0.0013030862435698509,
      "step": 153020
    },
    {
      "epoch": 43.4374112971899,
      "grad_norm": 0.09452912956476212,
      "learning_rate": 5.6580755038319624e-05,
      "loss": 0.0018481139093637466,
      "step": 153030
    },
    {
      "epoch": 43.44024978711325,
      "grad_norm": 0.10628576576709747,
      "learning_rate": 5.657791654839626e-05,
      "loss": 0.0008358459919691086,
      "step": 153040
    },
    {
      "epoch": 43.443088277036615,
      "grad_norm": 2.2888851165771484,
      "learning_rate": 5.65750780584729e-05,
      "loss": 0.00723518431186676,
      "step": 153050
    },
    {
      "epoch": 43.44592676695998,
      "grad_norm": 2.4254162311553955,
      "learning_rate": 5.6572239568549534e-05,
      "loss": 0.0029633793979883193,
      "step": 153060
    },
    {
      "epoch": 43.44876525688334,
      "grad_norm": 2.268907070159912,
      "learning_rate": 5.656940107862617e-05,
      "loss": 0.0010126948356628418,
      "step": 153070
    },
    {
      "epoch": 43.4516037468067,
      "grad_norm": 0.8506951332092285,
      "learning_rate": 5.656656258870281e-05,
      "loss": 0.001731904223561287,
      "step": 153080
    },
    {
      "epoch": 43.45444223673006,
      "grad_norm": 0.07425431907176971,
      "learning_rate": 5.6563724098779445e-05,
      "loss": 0.006238695606589318,
      "step": 153090
    },
    {
      "epoch": 43.45728072665342,
      "grad_norm": 0.38330450654029846,
      "learning_rate": 5.656088560885609e-05,
      "loss": 0.00559648871421814,
      "step": 153100
    },
    {
      "epoch": 43.46011921657678,
      "grad_norm": 0.09684891253709793,
      "learning_rate": 5.6558047118932735e-05,
      "loss": 0.007665575295686722,
      "step": 153110
    },
    {
      "epoch": 43.46295770650014,
      "grad_norm": 0.097519151866436,
      "learning_rate": 5.655520862900937e-05,
      "loss": 0.0023943042382597924,
      "step": 153120
    },
    {
      "epoch": 43.465796196423504,
      "grad_norm": 0.19495947659015656,
      "learning_rate": 5.655237013908601e-05,
      "loss": 0.0033308058977127076,
      "step": 153130
    },
    {
      "epoch": 43.46863468634686,
      "grad_norm": 0.25330281257629395,
      "learning_rate": 5.6549531649162645e-05,
      "loss": 0.003457040339708328,
      "step": 153140
    },
    {
      "epoch": 43.47147317627022,
      "grad_norm": 0.231252521276474,
      "learning_rate": 5.6546693159239286e-05,
      "loss": 0.0021714739501476287,
      "step": 153150
    },
    {
      "epoch": 43.474311666193586,
      "grad_norm": 0.7858866453170776,
      "learning_rate": 5.6543854669315935e-05,
      "loss": 0.0011362943798303605,
      "step": 153160
    },
    {
      "epoch": 43.47715015611695,
      "grad_norm": 3.0450685024261475,
      "learning_rate": 5.654101617939257e-05,
      "loss": 0.0025546543300151827,
      "step": 153170
    },
    {
      "epoch": 43.479988646040304,
      "grad_norm": 0.08516174554824829,
      "learning_rate": 5.653817768946921e-05,
      "loss": 0.0009009996429085732,
      "step": 153180
    },
    {
      "epoch": 43.48282713596367,
      "grad_norm": 0.07210567593574524,
      "learning_rate": 5.6535339199545845e-05,
      "loss": 0.0018630864098668099,
      "step": 153190
    },
    {
      "epoch": 43.48566562588703,
      "grad_norm": 0.8571279048919678,
      "learning_rate": 5.653250070962248e-05,
      "loss": 0.002955697290599346,
      "step": 153200
    },
    {
      "epoch": 43.488504115810386,
      "grad_norm": 0.061951376497745514,
      "learning_rate": 5.652966221969912e-05,
      "loss": 0.004855797439813614,
      "step": 153210
    },
    {
      "epoch": 43.49134260573375,
      "grad_norm": 1.0942081212997437,
      "learning_rate": 5.6526823729775756e-05,
      "loss": 0.0029168806970119475,
      "step": 153220
    },
    {
      "epoch": 43.49418109565711,
      "grad_norm": 0.6955099105834961,
      "learning_rate": 5.6523985239852404e-05,
      "loss": 0.0016587171703577042,
      "step": 153230
    },
    {
      "epoch": 43.497019585580475,
      "grad_norm": 0.7128551602363586,
      "learning_rate": 5.6521146749929045e-05,
      "loss": 0.006736001372337342,
      "step": 153240
    },
    {
      "epoch": 43.49985807550383,
      "grad_norm": 10.458624839782715,
      "learning_rate": 5.651830826000568e-05,
      "loss": 0.009339608252048492,
      "step": 153250
    },
    {
      "epoch": 43.50269656542719,
      "grad_norm": 1.3465211391448975,
      "learning_rate": 5.651546977008232e-05,
      "loss": 0.002453538216650486,
      "step": 153260
    },
    {
      "epoch": 43.505535055350556,
      "grad_norm": 0.5278250575065613,
      "learning_rate": 5.6512631280158956e-05,
      "loss": 0.008037395775318146,
      "step": 153270
    },
    {
      "epoch": 43.50837354527391,
      "grad_norm": 0.08596598356962204,
      "learning_rate": 5.65097927902356e-05,
      "loss": 0.0030168814584612845,
      "step": 153280
    },
    {
      "epoch": 43.511212035197275,
      "grad_norm": 0.8708239793777466,
      "learning_rate": 5.650695430031223e-05,
      "loss": 0.004456581920385361,
      "step": 153290
    },
    {
      "epoch": 43.51405052512064,
      "grad_norm": 13.819242477416992,
      "learning_rate": 5.650411581038888e-05,
      "loss": 0.005208434164524078,
      "step": 153300
    },
    {
      "epoch": 43.51688901504399,
      "grad_norm": 0.02657460607588291,
      "learning_rate": 5.6501277320465515e-05,
      "loss": 0.0012954723089933396,
      "step": 153310
    },
    {
      "epoch": 43.519727504967356,
      "grad_norm": 0.7239187955856323,
      "learning_rate": 5.6498438830542156e-05,
      "loss": 0.00208011195063591,
      "step": 153320
    },
    {
      "epoch": 43.52256599489072,
      "grad_norm": 0.051288191229104996,
      "learning_rate": 5.649560034061879e-05,
      "loss": 0.0009571239352226257,
      "step": 153330
    },
    {
      "epoch": 43.52540448481408,
      "grad_norm": 0.3386920392513275,
      "learning_rate": 5.649276185069543e-05,
      "loss": 0.0017803581431508064,
      "step": 153340
    },
    {
      "epoch": 43.52824297473744,
      "grad_norm": 0.7638600468635559,
      "learning_rate": 5.6489923360772067e-05,
      "loss": 0.0008437344804406166,
      "step": 153350
    },
    {
      "epoch": 43.5310814646608,
      "grad_norm": 0.07631951570510864,
      "learning_rate": 5.6487084870848715e-05,
      "loss": 0.0015773527324199677,
      "step": 153360
    },
    {
      "epoch": 43.533919954584164,
      "grad_norm": 6.410444736480713,
      "learning_rate": 5.6484246380925356e-05,
      "loss": 0.006498938798904419,
      "step": 153370
    },
    {
      "epoch": 43.53675844450752,
      "grad_norm": 0.14903263747692108,
      "learning_rate": 5.648140789100199e-05,
      "loss": 0.004223138839006424,
      "step": 153380
    },
    {
      "epoch": 43.53959693443088,
      "grad_norm": 0.006421033293008804,
      "learning_rate": 5.647856940107863e-05,
      "loss": 0.0031513508409261705,
      "step": 153390
    },
    {
      "epoch": 43.542435424354245,
      "grad_norm": 0.07998734712600708,
      "learning_rate": 5.6475730911155267e-05,
      "loss": 0.007575667649507523,
      "step": 153400
    },
    {
      "epoch": 43.5452739142776,
      "grad_norm": 0.10729268193244934,
      "learning_rate": 5.64728924212319e-05,
      "loss": 0.0045166205614805225,
      "step": 153410
    },
    {
      "epoch": 43.548112404200964,
      "grad_norm": 6.214387893676758,
      "learning_rate": 5.647005393130854e-05,
      "loss": 0.0028106285259127616,
      "step": 153420
    },
    {
      "epoch": 43.55095089412433,
      "grad_norm": 0.10489952564239502,
      "learning_rate": 5.646721544138519e-05,
      "loss": 0.0031630709767341615,
      "step": 153430
    },
    {
      "epoch": 43.55378938404769,
      "grad_norm": 0.19084492325782776,
      "learning_rate": 5.6464376951461825e-05,
      "loss": 0.00226014107465744,
      "step": 153440
    },
    {
      "epoch": 43.556627873971046,
      "grad_norm": 1.1417911052703857,
      "learning_rate": 5.646153846153847e-05,
      "loss": 0.0030605683103203775,
      "step": 153450
    },
    {
      "epoch": 43.55946636389441,
      "grad_norm": 0.014101510867476463,
      "learning_rate": 5.64586999716151e-05,
      "loss": 0.0016963986679911613,
      "step": 153460
    },
    {
      "epoch": 43.56230485381777,
      "grad_norm": 0.8174677491188049,
      "learning_rate": 5.645586148169174e-05,
      "loss": 0.0024310821667313577,
      "step": 153470
    },
    {
      "epoch": 43.56514334374113,
      "grad_norm": 13.392679214477539,
      "learning_rate": 5.645302299176838e-05,
      "loss": 0.004158905148506165,
      "step": 153480
    },
    {
      "epoch": 43.56798183366449,
      "grad_norm": 1.612534999847412,
      "learning_rate": 5.645018450184502e-05,
      "loss": 0.0016244685277342797,
      "step": 153490
    },
    {
      "epoch": 43.57082032358785,
      "grad_norm": 0.1790313720703125,
      "learning_rate": 5.644734601192167e-05,
      "loss": 0.0010216221213340758,
      "step": 153500
    },
    {
      "epoch": 43.57082032358785,
      "eval_accuracy": 0.9717682965600559,
      "eval_loss": 0.09897135943174362,
      "eval_runtime": 32.1903,
      "eval_samples_per_second": 488.563,
      "eval_steps_per_second": 7.642,
      "step": 153500
    },
    {
      "epoch": 43.57365881351121,
      "grad_norm": 0.21555952727794647,
      "learning_rate": 5.64445075219983e-05,
      "loss": 0.003413374722003937,
      "step": 153510
    },
    {
      "epoch": 43.57649730343457,
      "grad_norm": 1.1163606643676758,
      "learning_rate": 5.644166903207494e-05,
      "loss": 0.0015559513121843338,
      "step": 153520
    },
    {
      "epoch": 43.579335793357934,
      "grad_norm": 1.9895554780960083,
      "learning_rate": 5.643883054215158e-05,
      "loss": 0.001264207437634468,
      "step": 153530
    },
    {
      "epoch": 43.5821742832813,
      "grad_norm": 0.2352142333984375,
      "learning_rate": 5.643599205222821e-05,
      "loss": 0.005587928369641304,
      "step": 153540
    },
    {
      "epoch": 43.58501277320465,
      "grad_norm": 0.08359673619270325,
      "learning_rate": 5.643315356230485e-05,
      "loss": 0.002614370733499527,
      "step": 153550
    },
    {
      "epoch": 43.587851263128016,
      "grad_norm": 0.04170574992895126,
      "learning_rate": 5.64303150723815e-05,
      "loss": 0.0072138309478759766,
      "step": 153560
    },
    {
      "epoch": 43.59068975305138,
      "grad_norm": 0.3711927533149719,
      "learning_rate": 5.6427476582458136e-05,
      "loss": 0.0020316669717431067,
      "step": 153570
    },
    {
      "epoch": 43.593528242974735,
      "grad_norm": 5.076942443847656,
      "learning_rate": 5.642463809253478e-05,
      "loss": 0.011276464164257049,
      "step": 153580
    },
    {
      "epoch": 43.5963667328981,
      "grad_norm": 1.1341822147369385,
      "learning_rate": 5.642179960261141e-05,
      "loss": 0.002077527344226837,
      "step": 153590
    },
    {
      "epoch": 43.59920522282146,
      "grad_norm": 4.92354679107666,
      "learning_rate": 5.6418961112688053e-05,
      "loss": 0.011387376487255097,
      "step": 153600
    },
    {
      "epoch": 43.602043712744816,
      "grad_norm": 16.01750373840332,
      "learning_rate": 5.641612262276469e-05,
      "loss": 0.013058266043663025,
      "step": 153610
    },
    {
      "epoch": 43.60488220266818,
      "grad_norm": 0.6884289383888245,
      "learning_rate": 5.641328413284133e-05,
      "loss": 0.0031808063387870787,
      "step": 153620
    },
    {
      "epoch": 43.60772069259154,
      "grad_norm": 0.2355767786502838,
      "learning_rate": 5.641044564291798e-05,
      "loss": 0.0008702097460627556,
      "step": 153630
    },
    {
      "epoch": 43.610559182514905,
      "grad_norm": 0.3145964741706848,
      "learning_rate": 5.640760715299461e-05,
      "loss": 0.00044127143919467924,
      "step": 153640
    },
    {
      "epoch": 43.61339767243826,
      "grad_norm": 0.06981642544269562,
      "learning_rate": 5.6404768663071254e-05,
      "loss": 0.0018135149031877518,
      "step": 153650
    },
    {
      "epoch": 43.61623616236162,
      "grad_norm": 1.7476050853729248,
      "learning_rate": 5.640193017314789e-05,
      "loss": 0.0010967563837766647,
      "step": 153660
    },
    {
      "epoch": 43.619074652284986,
      "grad_norm": 0.2768338918685913,
      "learning_rate": 5.639909168322452e-05,
      "loss": 0.004553082585334778,
      "step": 153670
    },
    {
      "epoch": 43.62191314220834,
      "grad_norm": 0.20510859787464142,
      "learning_rate": 5.6396253193301164e-05,
      "loss": 0.0032284677028656004,
      "step": 153680
    },
    {
      "epoch": 43.624751632131705,
      "grad_norm": 0.8298428058624268,
      "learning_rate": 5.63934147033778e-05,
      "loss": 0.0019706910476088523,
      "step": 153690
    },
    {
      "epoch": 43.62759012205507,
      "grad_norm": 0.14169205725193024,
      "learning_rate": 5.639057621345445e-05,
      "loss": 0.011586420238018036,
      "step": 153700
    },
    {
      "epoch": 43.63042861197843,
      "grad_norm": 0.23942674696445465,
      "learning_rate": 5.638773772353109e-05,
      "loss": 0.0013110648840665817,
      "step": 153710
    },
    {
      "epoch": 43.63326710190179,
      "grad_norm": 0.18888431787490845,
      "learning_rate": 5.6385183082600055e-05,
      "loss": 0.007014836370944977,
      "step": 153720
    },
    {
      "epoch": 43.63610559182515,
      "grad_norm": 0.21401743590831757,
      "learning_rate": 5.6382344592676696e-05,
      "loss": 0.0016837948933243752,
      "step": 153730
    },
    {
      "epoch": 43.63894408174851,
      "grad_norm": 0.6137626767158508,
      "learning_rate": 5.6379506102753344e-05,
      "loss": 0.0006144683808088303,
      "step": 153740
    },
    {
      "epoch": 43.64178257167187,
      "grad_norm": 0.12881550192832947,
      "learning_rate": 5.637666761282998e-05,
      "loss": 0.0006083633750677108,
      "step": 153750
    },
    {
      "epoch": 43.64462106159523,
      "grad_norm": 0.20826640725135803,
      "learning_rate": 5.637382912290662e-05,
      "loss": 0.0009685803204774856,
      "step": 153760
    },
    {
      "epoch": 43.647459551518594,
      "grad_norm": 2.3518834114074707,
      "learning_rate": 5.6370990632983255e-05,
      "loss": 0.0010279864072799683,
      "step": 153770
    },
    {
      "epoch": 43.65029804144195,
      "grad_norm": 0.7419127821922302,
      "learning_rate": 5.6368152143059896e-05,
      "loss": 0.0006368599832057952,
      "step": 153780
    },
    {
      "epoch": 43.65313653136531,
      "grad_norm": 0.05684041604399681,
      "learning_rate": 5.636531365313653e-05,
      "loss": 0.002681879699230194,
      "step": 153790
    },
    {
      "epoch": 43.655975021288675,
      "grad_norm": 0.5694453120231628,
      "learning_rate": 5.6362475163213165e-05,
      "loss": 0.002405887469649315,
      "step": 153800
    },
    {
      "epoch": 43.65881351121204,
      "grad_norm": 0.1537306308746338,
      "learning_rate": 5.635963667328982e-05,
      "loss": 0.0014763515442609786,
      "step": 153810
    },
    {
      "epoch": 43.661652001135394,
      "grad_norm": 0.6849051713943481,
      "learning_rate": 5.6356798183366455e-05,
      "loss": 0.0009526051580905914,
      "step": 153820
    },
    {
      "epoch": 43.66449049105876,
      "grad_norm": 0.29667145013809204,
      "learning_rate": 5.635395969344309e-05,
      "loss": 0.0013875419273972512,
      "step": 153830
    },
    {
      "epoch": 43.66732898098212,
      "grad_norm": 0.021146979182958603,
      "learning_rate": 5.635112120351973e-05,
      "loss": 0.0011237954720854759,
      "step": 153840
    },
    {
      "epoch": 43.670167470905476,
      "grad_norm": 0.029883261770009995,
      "learning_rate": 5.6348282713596365e-05,
      "loss": 0.0004882743582129478,
      "step": 153850
    },
    {
      "epoch": 43.67300596082884,
      "grad_norm": 0.11274713277816772,
      "learning_rate": 5.634544422367301e-05,
      "loss": 0.0009260876104235649,
      "step": 153860
    },
    {
      "epoch": 43.6758444507522,
      "grad_norm": 0.20485463738441467,
      "learning_rate": 5.634260573374964e-05,
      "loss": 0.0005394136533141136,
      "step": 153870
    },
    {
      "epoch": 43.67868294067556,
      "grad_norm": 0.05315103754401207,
      "learning_rate": 5.633976724382629e-05,
      "loss": 0.005065117031335831,
      "step": 153880
    },
    {
      "epoch": 43.68152143059892,
      "grad_norm": 0.09237052500247955,
      "learning_rate": 5.633692875390293e-05,
      "loss": 0.0011650104075670241,
      "step": 153890
    },
    {
      "epoch": 43.68435992052228,
      "grad_norm": 0.3734491467475891,
      "learning_rate": 5.6334090263979565e-05,
      "loss": 0.011818775534629821,
      "step": 153900
    },
    {
      "epoch": 43.687198410445646,
      "grad_norm": 0.3581596612930298,
      "learning_rate": 5.633125177405621e-05,
      "loss": 0.001146983541548252,
      "step": 153910
    },
    {
      "epoch": 43.690036900369,
      "grad_norm": 0.2726362943649292,
      "learning_rate": 5.632841328413284e-05,
      "loss": 0.0009339885786175728,
      "step": 153920
    },
    {
      "epoch": 43.692875390292365,
      "grad_norm": 0.025112181901931763,
      "learning_rate": 5.6325574794209476e-05,
      "loss": 0.0005887394770979881,
      "step": 153930
    },
    {
      "epoch": 43.69571388021573,
      "grad_norm": 0.21182525157928467,
      "learning_rate": 5.632273630428613e-05,
      "loss": 0.0014316171407699585,
      "step": 153940
    },
    {
      "epoch": 43.69855237013908,
      "grad_norm": 0.1607232242822647,
      "learning_rate": 5.6319897814362766e-05,
      "loss": 0.0008581899106502533,
      "step": 153950
    },
    {
      "epoch": 43.701390860062446,
      "grad_norm": 0.07222142070531845,
      "learning_rate": 5.63170593244394e-05,
      "loss": 0.0032442010939121245,
      "step": 153960
    },
    {
      "epoch": 43.70422934998581,
      "grad_norm": 0.8958426117897034,
      "learning_rate": 5.631422083451604e-05,
      "loss": 0.0066508248448371885,
      "step": 153970
    },
    {
      "epoch": 43.70706783990917,
      "grad_norm": 0.46692803502082825,
      "learning_rate": 5.6311382344592676e-05,
      "loss": 0.0005263220518827438,
      "step": 153980
    },
    {
      "epoch": 43.70990632983253,
      "grad_norm": 1.412104845046997,
      "learning_rate": 5.630854385466932e-05,
      "loss": 0.0006815085187554359,
      "step": 153990
    },
    {
      "epoch": 43.71274481975589,
      "grad_norm": 0.26946479082107544,
      "learning_rate": 5.630570536474595e-05,
      "loss": 0.0027316099032759665,
      "step": 154000
    },
    {
      "epoch": 43.71274481975589,
      "eval_accuracy": 0.9765371653843709,
      "eval_loss": 0.07975849509239197,
      "eval_runtime": 31.7448,
      "eval_samples_per_second": 495.42,
      "eval_steps_per_second": 7.749,
      "step": 154000
    },
    {
      "epoch": 43.71558330967925,
      "grad_norm": 0.5882258415222168,
      "learning_rate": 5.63028668748226e-05,
      "loss": 0.001388123817741871,
      "step": 154010
    },
    {
      "epoch": 43.71842179960261,
      "grad_norm": 2.0372314453125,
      "learning_rate": 5.630002838489924e-05,
      "loss": 0.0013981403782963753,
      "step": 154020
    },
    {
      "epoch": 43.72126028952597,
      "grad_norm": 0.13697589933872223,
      "learning_rate": 5.6297189894975876e-05,
      "loss": 0.0010839227586984634,
      "step": 154030
    },
    {
      "epoch": 43.724098779449335,
      "grad_norm": 0.03263700753450394,
      "learning_rate": 5.629435140505252e-05,
      "loss": 0.0033743176609277725,
      "step": 154040
    },
    {
      "epoch": 43.72693726937269,
      "grad_norm": 0.1429240107536316,
      "learning_rate": 5.629151291512915e-05,
      "loss": 0.0039602693170309065,
      "step": 154050
    },
    {
      "epoch": 43.729775759296054,
      "grad_norm": 0.05665097013115883,
      "learning_rate": 5.628867442520579e-05,
      "loss": 0.0010849833488464355,
      "step": 154060
    },
    {
      "epoch": 43.73261424921942,
      "grad_norm": 2.1330153942108154,
      "learning_rate": 5.628583593528243e-05,
      "loss": 0.00393885001540184,
      "step": 154070
    },
    {
      "epoch": 43.73545273914278,
      "grad_norm": 2.0124130249023438,
      "learning_rate": 5.6282997445359076e-05,
      "loss": 0.001319502666592598,
      "step": 154080
    },
    {
      "epoch": 43.738291229066135,
      "grad_norm": 0.789679765701294,
      "learning_rate": 5.628015895543571e-05,
      "loss": 0.0027764173224568366,
      "step": 154090
    },
    {
      "epoch": 43.7411297189895,
      "grad_norm": 0.1983095407485962,
      "learning_rate": 5.627732046551235e-05,
      "loss": 0.0013423211872577666,
      "step": 154100
    },
    {
      "epoch": 43.74396820891286,
      "grad_norm": 0.13133090734481812,
      "learning_rate": 5.627448197558899e-05,
      "loss": 0.0008127989247441291,
      "step": 154110
    },
    {
      "epoch": 43.74680669883622,
      "grad_norm": 0.7549835443496704,
      "learning_rate": 5.627164348566563e-05,
      "loss": 0.0022030359134078024,
      "step": 154120
    },
    {
      "epoch": 43.74964518875958,
      "grad_norm": 0.08254050463438034,
      "learning_rate": 5.626880499574226e-05,
      "loss": 0.0007023654878139495,
      "step": 154130
    },
    {
      "epoch": 43.75248367868294,
      "grad_norm": 6.490147590637207,
      "learning_rate": 5.626596650581891e-05,
      "loss": 0.008414805680513383,
      "step": 154140
    },
    {
      "epoch": 43.7553221686063,
      "grad_norm": 0.9044446349143982,
      "learning_rate": 5.626312801589555e-05,
      "loss": 0.0013672027736902237,
      "step": 154150
    },
    {
      "epoch": 43.75816065852966,
      "grad_norm": 0.01440686360001564,
      "learning_rate": 5.626028952597219e-05,
      "loss": 0.0005390513688325882,
      "step": 154160
    },
    {
      "epoch": 43.760999148453024,
      "grad_norm": 0.3232276737689972,
      "learning_rate": 5.625745103604882e-05,
      "loss": 0.005749291926622391,
      "step": 154170
    },
    {
      "epoch": 43.76383763837639,
      "grad_norm": 0.21527157723903656,
      "learning_rate": 5.625461254612546e-05,
      "loss": 0.0009574376046657563,
      "step": 154180
    },
    {
      "epoch": 43.76667612829974,
      "grad_norm": 0.14485757052898407,
      "learning_rate": 5.62517740562021e-05,
      "loss": 0.005866488814353943,
      "step": 154190
    },
    {
      "epoch": 43.769514618223106,
      "grad_norm": 0.6496561169624329,
      "learning_rate": 5.624893556627874e-05,
      "loss": 0.0015949683263897895,
      "step": 154200
    },
    {
      "epoch": 43.77235310814647,
      "grad_norm": 0.9769965410232544,
      "learning_rate": 5.624609707635539e-05,
      "loss": 0.00897739827632904,
      "step": 154210
    },
    {
      "epoch": 43.775191598069824,
      "grad_norm": 0.56364506483078,
      "learning_rate": 5.624325858643202e-05,
      "loss": 0.009181302785873414,
      "step": 154220
    },
    {
      "epoch": 43.77803008799319,
      "grad_norm": 0.22408020496368408,
      "learning_rate": 5.624042009650866e-05,
      "loss": 0.006026215106248856,
      "step": 154230
    },
    {
      "epoch": 43.78086857791655,
      "grad_norm": 0.4522612392902374,
      "learning_rate": 5.62375816065853e-05,
      "loss": 0.0012129290029406548,
      "step": 154240
    },
    {
      "epoch": 43.783707067839906,
      "grad_norm": 2.799694538116455,
      "learning_rate": 5.623474311666194e-05,
      "loss": 0.0012379903346300125,
      "step": 154250
    },
    {
      "epoch": 43.78654555776327,
      "grad_norm": 0.813376247882843,
      "learning_rate": 5.6231904626738574e-05,
      "loss": 0.00205894373357296,
      "step": 154260
    },
    {
      "epoch": 43.78938404768663,
      "grad_norm": 0.1333845555782318,
      "learning_rate": 5.622906613681521e-05,
      "loss": 0.002887822687625885,
      "step": 154270
    },
    {
      "epoch": 43.792222537609995,
      "grad_norm": 3.3997066020965576,
      "learning_rate": 5.622622764689186e-05,
      "loss": 0.005990870296955109,
      "step": 154280
    },
    {
      "epoch": 43.79506102753335,
      "grad_norm": 6.343020439147949,
      "learning_rate": 5.62233891569685e-05,
      "loss": 0.004266654327511788,
      "step": 154290
    },
    {
      "epoch": 43.79789951745671,
      "grad_norm": 0.46669575572013855,
      "learning_rate": 5.622055066704513e-05,
      "loss": 0.0039650309830904,
      "step": 154300
    },
    {
      "epoch": 43.800738007380076,
      "grad_norm": 14.155376434326172,
      "learning_rate": 5.6217712177121774e-05,
      "loss": 0.005905056744813919,
      "step": 154310
    },
    {
      "epoch": 43.80357649730343,
      "grad_norm": 6.257410049438477,
      "learning_rate": 5.621487368719841e-05,
      "loss": 0.007551933079957962,
      "step": 154320
    },
    {
      "epoch": 43.806414987226795,
      "grad_norm": 2.8909084796905518,
      "learning_rate": 5.621203519727505e-05,
      "loss": 0.001943877898156643,
      "step": 154330
    },
    {
      "epoch": 43.80925347715016,
      "grad_norm": 0.03649016469717026,
      "learning_rate": 5.62091967073517e-05,
      "loss": 0.002396298572421074,
      "step": 154340
    },
    {
      "epoch": 43.81209196707351,
      "grad_norm": 0.11190072447061539,
      "learning_rate": 5.620635821742833e-05,
      "loss": 0.0014627480879426003,
      "step": 154350
    },
    {
      "epoch": 43.814930456996876,
      "grad_norm": 0.03898775577545166,
      "learning_rate": 5.6203519727504974e-05,
      "loss": 0.0144993856549263,
      "step": 154360
    },
    {
      "epoch": 43.81776894692024,
      "grad_norm": 1.8075590133666992,
      "learning_rate": 5.620068123758161e-05,
      "loss": 0.0010585809126496316,
      "step": 154370
    },
    {
      "epoch": 43.8206074368436,
      "grad_norm": 0.26776123046875,
      "learning_rate": 5.619784274765825e-05,
      "loss": 0.002392314560711384,
      "step": 154380
    },
    {
      "epoch": 43.82344592676696,
      "grad_norm": 0.06966307759284973,
      "learning_rate": 5.6195004257734884e-05,
      "loss": 0.003548109531402588,
      "step": 154390
    },
    {
      "epoch": 43.82628441669032,
      "grad_norm": 0.15655361115932465,
      "learning_rate": 5.619216576781152e-05,
      "loss": 0.0030846957117319105,
      "step": 154400
    },
    {
      "epoch": 43.829122906613684,
      "grad_norm": 4.437291145324707,
      "learning_rate": 5.6189327277888174e-05,
      "loss": 0.002964920736849308,
      "step": 154410
    },
    {
      "epoch": 43.83196139653704,
      "grad_norm": 0.5085890889167786,
      "learning_rate": 5.618648878796481e-05,
      "loss": 0.0007456650957465172,
      "step": 154420
    },
    {
      "epoch": 43.8347998864604,
      "grad_norm": 0.10757845640182495,
      "learning_rate": 5.618365029804144e-05,
      "loss": 0.01442403644323349,
      "step": 154430
    },
    {
      "epoch": 43.837638376383765,
      "grad_norm": 0.28071504831314087,
      "learning_rate": 5.6180811808118085e-05,
      "loss": 0.001076718233525753,
      "step": 154440
    },
    {
      "epoch": 43.84047686630713,
      "grad_norm": 0.07889541983604431,
      "learning_rate": 5.617797331819472e-05,
      "loss": 0.003766489028930664,
      "step": 154450
    },
    {
      "epoch": 43.843315356230484,
      "grad_norm": 0.015603361651301384,
      "learning_rate": 5.617513482827136e-05,
      "loss": 0.0013997703790664673,
      "step": 154460
    },
    {
      "epoch": 43.84615384615385,
      "grad_norm": 0.1579984724521637,
      "learning_rate": 5.6172296338347995e-05,
      "loss": 0.0014079922810196876,
      "step": 154470
    },
    {
      "epoch": 43.84899233607721,
      "grad_norm": 1.1978918313980103,
      "learning_rate": 5.616945784842464e-05,
      "loss": 0.0014616191387176515,
      "step": 154480
    },
    {
      "epoch": 43.851830826000565,
      "grad_norm": 12.10133171081543,
      "learning_rate": 5.6166619358501285e-05,
      "loss": 0.007242640107870102,
      "step": 154490
    },
    {
      "epoch": 43.85466931592393,
      "grad_norm": 1.0381139516830444,
      "learning_rate": 5.616378086857792e-05,
      "loss": 0.003041612543165684,
      "step": 154500
    },
    {
      "epoch": 43.85466931592393,
      "eval_accuracy": 0.972849240160234,
      "eval_loss": 0.0968237966299057,
      "eval_runtime": 31.9377,
      "eval_samples_per_second": 492.427,
      "eval_steps_per_second": 7.702,
      "step": 154500
    },
    {
      "epoch": 43.85750780584729,
      "grad_norm": 1.2096731662750244,
      "learning_rate": 5.616094237865456e-05,
      "loss": 0.0045863516628742215,
      "step": 154510
    },
    {
      "epoch": 43.86034629577065,
      "grad_norm": 0.10764015465974808,
      "learning_rate": 5.6158103888731195e-05,
      "loss": 0.018301483988761903,
      "step": 154520
    },
    {
      "epoch": 43.86318478569401,
      "grad_norm": 0.039228178560733795,
      "learning_rate": 5.615526539880783e-05,
      "loss": 0.005936878174543381,
      "step": 154530
    },
    {
      "epoch": 43.86602327561737,
      "grad_norm": 0.17235271632671356,
      "learning_rate": 5.615242690888448e-05,
      "loss": 0.0011363167315721512,
      "step": 154540
    },
    {
      "epoch": 43.868861765540736,
      "grad_norm": 0.08166135102510452,
      "learning_rate": 5.614958841896112e-05,
      "loss": 0.002928302809596062,
      "step": 154550
    },
    {
      "epoch": 43.87170025546409,
      "grad_norm": 0.02972782775759697,
      "learning_rate": 5.6146749929037754e-05,
      "loss": 0.0018457956612110138,
      "step": 154560
    },
    {
      "epoch": 43.874538745387454,
      "grad_norm": 1.2956883907318115,
      "learning_rate": 5.6143911439114395e-05,
      "loss": 0.013591650128364562,
      "step": 154570
    },
    {
      "epoch": 43.87737723531082,
      "grad_norm": 0.15672563016414642,
      "learning_rate": 5.614107294919103e-05,
      "loss": 0.0035447567701339723,
      "step": 154580
    },
    {
      "epoch": 43.88021572523417,
      "grad_norm": 2.1472482681274414,
      "learning_rate": 5.613823445926767e-05,
      "loss": 0.0016180604696273803,
      "step": 154590
    },
    {
      "epoch": 43.883054215157536,
      "grad_norm": 0.06237005814909935,
      "learning_rate": 5.6135395969344306e-05,
      "loss": 0.001287197135388851,
      "step": 154600
    },
    {
      "epoch": 43.8858927050809,
      "grad_norm": 0.21725216507911682,
      "learning_rate": 5.6132557479420954e-05,
      "loss": 0.0022531339898705484,
      "step": 154610
    },
    {
      "epoch": 43.888731195004254,
      "grad_norm": 0.05493052676320076,
      "learning_rate": 5.6129718989497595e-05,
      "loss": 0.0009204939007759095,
      "step": 154620
    },
    {
      "epoch": 43.89156968492762,
      "grad_norm": 0.5109702348709106,
      "learning_rate": 5.612688049957423e-05,
      "loss": 0.0012773189693689347,
      "step": 154630
    },
    {
      "epoch": 43.89440817485098,
      "grad_norm": 0.14308279752731323,
      "learning_rate": 5.6124042009650865e-05,
      "loss": 0.0013164093717932702,
      "step": 154640
    },
    {
      "epoch": 43.89724666477434,
      "grad_norm": 0.05201289430260658,
      "learning_rate": 5.6121203519727506e-05,
      "loss": 0.0006537100300192833,
      "step": 154650
    },
    {
      "epoch": 43.9000851546977,
      "grad_norm": 0.057875387370586395,
      "learning_rate": 5.611836502980414e-05,
      "loss": 0.0017512915655970573,
      "step": 154660
    },
    {
      "epoch": 43.90292364462106,
      "grad_norm": 0.06067872792482376,
      "learning_rate": 5.611552653988078e-05,
      "loss": 0.002324228733778,
      "step": 154670
    },
    {
      "epoch": 43.905762134544425,
      "grad_norm": 0.03390556946396828,
      "learning_rate": 5.611268804995743e-05,
      "loss": 0.001713894121348858,
      "step": 154680
    },
    {
      "epoch": 43.90860062446778,
      "grad_norm": 6.078786373138428,
      "learning_rate": 5.6109849560034065e-05,
      "loss": 0.01019090786576271,
      "step": 154690
    },
    {
      "epoch": 43.91143911439114,
      "grad_norm": 0.533894419670105,
      "learning_rate": 5.6107011070110706e-05,
      "loss": 0.0009850585833191873,
      "step": 154700
    },
    {
      "epoch": 43.914277604314506,
      "grad_norm": 0.22418056428432465,
      "learning_rate": 5.610417258018734e-05,
      "loss": 0.01278441995382309,
      "step": 154710
    },
    {
      "epoch": 43.91711609423786,
      "grad_norm": 0.1395626664161682,
      "learning_rate": 5.610133409026398e-05,
      "loss": 0.005101402848958969,
      "step": 154720
    },
    {
      "epoch": 43.919954584161225,
      "grad_norm": 0.25007742643356323,
      "learning_rate": 5.609849560034062e-05,
      "loss": 0.004803206771612167,
      "step": 154730
    },
    {
      "epoch": 43.92279307408459,
      "grad_norm": 0.612004280090332,
      "learning_rate": 5.6095657110417265e-05,
      "loss": 0.0010332951322197914,
      "step": 154740
    },
    {
      "epoch": 43.92563156400795,
      "grad_norm": 0.7500231862068176,
      "learning_rate": 5.6092818620493906e-05,
      "loss": 0.0009252235293388366,
      "step": 154750
    },
    {
      "epoch": 43.928470053931306,
      "grad_norm": 2.3525748252868652,
      "learning_rate": 5.608998013057054e-05,
      "loss": 0.0020056005567312242,
      "step": 154760
    },
    {
      "epoch": 43.93130854385467,
      "grad_norm": 0.9214540719985962,
      "learning_rate": 5.6087141640647175e-05,
      "loss": 0.0026182567700743673,
      "step": 154770
    },
    {
      "epoch": 43.93414703377803,
      "grad_norm": 0.10013450682163239,
      "learning_rate": 5.608430315072382e-05,
      "loss": 0.0005221609026193619,
      "step": 154780
    },
    {
      "epoch": 43.93698552370139,
      "grad_norm": 1.0548710823059082,
      "learning_rate": 5.608146466080045e-05,
      "loss": 0.003946314379572868,
      "step": 154790
    },
    {
      "epoch": 43.93982401362475,
      "grad_norm": 3.3023335933685303,
      "learning_rate": 5.607862617087709e-05,
      "loss": 0.002786969766020775,
      "step": 154800
    },
    {
      "epoch": 43.942662503548114,
      "grad_norm": 0.07353068888187408,
      "learning_rate": 5.607578768095374e-05,
      "loss": 0.0038492873311042784,
      "step": 154810
    },
    {
      "epoch": 43.94550099347148,
      "grad_norm": 0.05260717496275902,
      "learning_rate": 5.6072949191030375e-05,
      "loss": 0.007866273820400237,
      "step": 154820
    },
    {
      "epoch": 43.94833948339483,
      "grad_norm": 0.03648190572857857,
      "learning_rate": 5.607011070110702e-05,
      "loss": 0.0012519409880042075,
      "step": 154830
    },
    {
      "epoch": 43.951177973318195,
      "grad_norm": 0.10745439678430557,
      "learning_rate": 5.606727221118365e-05,
      "loss": 0.00225951187312603,
      "step": 154840
    },
    {
      "epoch": 43.95401646324156,
      "grad_norm": 1.1740463972091675,
      "learning_rate": 5.606443372126029e-05,
      "loss": 0.0034063123166561127,
      "step": 154850
    },
    {
      "epoch": 43.956854953164914,
      "grad_norm": 0.44358083605766296,
      "learning_rate": 5.606159523133693e-05,
      "loss": 0.0015564899891614913,
      "step": 154860
    },
    {
      "epoch": 43.95969344308828,
      "grad_norm": 0.10875990241765976,
      "learning_rate": 5.6058756741413576e-05,
      "loss": 0.001439131610095501,
      "step": 154870
    },
    {
      "epoch": 43.96253193301164,
      "grad_norm": 0.03178190067410469,
      "learning_rate": 5.605591825149022e-05,
      "loss": 0.007920160144567489,
      "step": 154880
    },
    {
      "epoch": 43.965370422934996,
      "grad_norm": 0.21149788796901703,
      "learning_rate": 5.605307976156685e-05,
      "loss": 0.0020446011796593664,
      "step": 154890
    },
    {
      "epoch": 43.96820891285836,
      "grad_norm": 16.05747413635254,
      "learning_rate": 5.6050241271643486e-05,
      "loss": 0.0057833768427371975,
      "step": 154900
    },
    {
      "epoch": 43.97104740278172,
      "grad_norm": 0.8648762702941895,
      "learning_rate": 5.604740278172013e-05,
      "loss": 0.0012465499341487885,
      "step": 154910
    },
    {
      "epoch": 43.973885892705084,
      "grad_norm": 3.130760669708252,
      "learning_rate": 5.604456429179676e-05,
      "loss": 0.004151615127921104,
      "step": 154920
    },
    {
      "epoch": 43.97672438262844,
      "grad_norm": 0.10990035533905029,
      "learning_rate": 5.6041725801873403e-05,
      "loss": 0.0015458039939403533,
      "step": 154930
    },
    {
      "epoch": 43.9795628725518,
      "grad_norm": 2.220709800720215,
      "learning_rate": 5.603888731195005e-05,
      "loss": 0.0024961942806839945,
      "step": 154940
    },
    {
      "epoch": 43.982401362475166,
      "grad_norm": 0.50340735912323,
      "learning_rate": 5.6036048822026686e-05,
      "loss": 0.0042699694633483885,
      "step": 154950
    },
    {
      "epoch": 43.98523985239852,
      "grad_norm": 0.25851163268089294,
      "learning_rate": 5.603321033210333e-05,
      "loss": 0.0031494498252868652,
      "step": 154960
    },
    {
      "epoch": 43.988078342321884,
      "grad_norm": 0.03720707446336746,
      "learning_rate": 5.603037184217996e-05,
      "loss": 0.0015873134136199952,
      "step": 154970
    },
    {
      "epoch": 43.99091683224525,
      "grad_norm": 0.629663348197937,
      "learning_rate": 5.6027533352256604e-05,
      "loss": 0.0012056361883878708,
      "step": 154980
    },
    {
      "epoch": 43.9937553221686,
      "grad_norm": 0.4239121377468109,
      "learning_rate": 5.602469486233324e-05,
      "loss": 0.004947818070650101,
      "step": 154990
    },
    {
      "epoch": 43.996593812091966,
      "grad_norm": 1.2359132766723633,
      "learning_rate": 5.602185637240987e-05,
      "loss": 0.001002364605665207,
      "step": 155000
    },
    {
      "epoch": 43.996593812091966,
      "eval_accuracy": 0.9745660329369873,
      "eval_loss": 0.0955958217382431,
      "eval_runtime": 32.062,
      "eval_samples_per_second": 490.519,
      "eval_steps_per_second": 7.673,
      "step": 155000
    },
    {
      "epoch": 43.99943230201533,
      "grad_norm": 0.12081151455640793,
      "learning_rate": 5.601901788248652e-05,
      "loss": 0.0013408733531832695,
      "step": 155010
    },
    {
      "epoch": 44.00227079193869,
      "grad_norm": 4.688199043273926,
      "learning_rate": 5.601617939256316e-05,
      "loss": 0.0015946285799145698,
      "step": 155020
    },
    {
      "epoch": 44.00510928186205,
      "grad_norm": 1.5189108848571777,
      "learning_rate": 5.60133409026398e-05,
      "loss": 0.002266383357346058,
      "step": 155030
    },
    {
      "epoch": 44.00794777178541,
      "grad_norm": 0.7865810394287109,
      "learning_rate": 5.601050241271644e-05,
      "loss": 0.002522528916597366,
      "step": 155040
    },
    {
      "epoch": 44.01078626170877,
      "grad_norm": 1.2058391571044922,
      "learning_rate": 5.600766392279307e-05,
      "loss": 0.002017972059547901,
      "step": 155050
    },
    {
      "epoch": 44.01362475163213,
      "grad_norm": 0.1725837141275406,
      "learning_rate": 5.6004825432869714e-05,
      "loss": 0.0013281796127557755,
      "step": 155060
    },
    {
      "epoch": 44.01646324155549,
      "grad_norm": 0.21224147081375122,
      "learning_rate": 5.600198694294636e-05,
      "loss": 0.0020997019484639166,
      "step": 155070
    },
    {
      "epoch": 44.019301731478855,
      "grad_norm": 0.27076128125190735,
      "learning_rate": 5.5999148453023e-05,
      "loss": 0.0013346156105399133,
      "step": 155080
    },
    {
      "epoch": 44.02214022140221,
      "grad_norm": 0.06251348555088043,
      "learning_rate": 5.599630996309964e-05,
      "loss": 0.0029946783557534217,
      "step": 155090
    },
    {
      "epoch": 44.02497871132557,
      "grad_norm": 0.010691402480006218,
      "learning_rate": 5.599347147317627e-05,
      "loss": 0.00544789507985115,
      "step": 155100
    },
    {
      "epoch": 44.027817201248936,
      "grad_norm": 0.4243180453777313,
      "learning_rate": 5.599063298325291e-05,
      "loss": 0.004329235851764679,
      "step": 155110
    },
    {
      "epoch": 44.0306556911723,
      "grad_norm": 0.1557484269142151,
      "learning_rate": 5.598779449332955e-05,
      "loss": 0.0009197456762194634,
      "step": 155120
    },
    {
      "epoch": 44.033494181095655,
      "grad_norm": 0.1867402046918869,
      "learning_rate": 5.5984956003406184e-05,
      "loss": 0.004693562164902687,
      "step": 155130
    },
    {
      "epoch": 44.03633267101902,
      "grad_norm": 0.08925186097621918,
      "learning_rate": 5.598211751348283e-05,
      "loss": 0.0023587899282574653,
      "step": 155140
    },
    {
      "epoch": 44.03917116094238,
      "grad_norm": 0.6825480461120605,
      "learning_rate": 5.597927902355947e-05,
      "loss": 0.00494980663061142,
      "step": 155150
    },
    {
      "epoch": 44.04200965086574,
      "grad_norm": 0.06050508841872215,
      "learning_rate": 5.597644053363611e-05,
      "loss": 0.005510495975613594,
      "step": 155160
    },
    {
      "epoch": 44.0448481407891,
      "grad_norm": 0.09389100223779678,
      "learning_rate": 5.597360204371275e-05,
      "loss": 0.0005796948447823524,
      "step": 155170
    },
    {
      "epoch": 44.04768663071246,
      "grad_norm": 0.13872447609901428,
      "learning_rate": 5.5970763553789384e-05,
      "loss": 0.001945292204618454,
      "step": 155180
    },
    {
      "epoch": 44.050525120635825,
      "grad_norm": 1.585614800453186,
      "learning_rate": 5.5967925063866025e-05,
      "loss": 0.0033866841346025466,
      "step": 155190
    },
    {
      "epoch": 44.05336361055918,
      "grad_norm": 2.0271971225738525,
      "learning_rate": 5.596508657394266e-05,
      "loss": 0.0005810413509607315,
      "step": 155200
    },
    {
      "epoch": 44.056202100482544,
      "grad_norm": 0.6688646078109741,
      "learning_rate": 5.596224808401931e-05,
      "loss": 0.004142041504383087,
      "step": 155210
    },
    {
      "epoch": 44.05904059040591,
      "grad_norm": 1.002747893333435,
      "learning_rate": 5.595940959409595e-05,
      "loss": 0.002308500185608864,
      "step": 155220
    },
    {
      "epoch": 44.06187908032926,
      "grad_norm": 0.08227542787790298,
      "learning_rate": 5.5956571104172584e-05,
      "loss": 0.0006013959646224976,
      "step": 155230
    },
    {
      "epoch": 44.064717570252625,
      "grad_norm": 0.007276277057826519,
      "learning_rate": 5.595373261424922e-05,
      "loss": 0.0010032646358013153,
      "step": 155240
    },
    {
      "epoch": 44.06755606017599,
      "grad_norm": 0.045482609421014786,
      "learning_rate": 5.595089412432586e-05,
      "loss": 0.0028629707172513007,
      "step": 155250
    },
    {
      "epoch": 44.070394550099344,
      "grad_norm": 0.08305247128009796,
      "learning_rate": 5.5948055634402494e-05,
      "loss": 0.0004930403083562851,
      "step": 155260
    },
    {
      "epoch": 44.07323304002271,
      "grad_norm": 0.13856832683086395,
      "learning_rate": 5.594521714447914e-05,
      "loss": 0.003337107226252556,
      "step": 155270
    },
    {
      "epoch": 44.07607152994607,
      "grad_norm": 0.07452182471752167,
      "learning_rate": 5.5942378654555784e-05,
      "loss": 0.0006238771602511406,
      "step": 155280
    },
    {
      "epoch": 44.07891001986943,
      "grad_norm": 2.890653610229492,
      "learning_rate": 5.593954016463242e-05,
      "loss": 0.0013073312118649482,
      "step": 155290
    },
    {
      "epoch": 44.08174850979279,
      "grad_norm": 0.14537155628204346,
      "learning_rate": 5.593670167470906e-05,
      "loss": 0.0005009999498724937,
      "step": 155300
    },
    {
      "epoch": 44.08458699971615,
      "grad_norm": 0.14274157583713531,
      "learning_rate": 5.5933863184785694e-05,
      "loss": 0.000645207054913044,
      "step": 155310
    },
    {
      "epoch": 44.087425489639514,
      "grad_norm": 0.028981607407331467,
      "learning_rate": 5.5931024694862336e-05,
      "loss": 0.012816362082958221,
      "step": 155320
    },
    {
      "epoch": 44.09026397956287,
      "grad_norm": 0.29338452219963074,
      "learning_rate": 5.592818620493897e-05,
      "loss": 0.0023719411343336104,
      "step": 155330
    },
    {
      "epoch": 44.09310246948623,
      "grad_norm": 0.1496899127960205,
      "learning_rate": 5.592534771501562e-05,
      "loss": 0.00653311014175415,
      "step": 155340
    },
    {
      "epoch": 44.095940959409596,
      "grad_norm": 1.0653738975524902,
      "learning_rate": 5.592250922509226e-05,
      "loss": 0.0006035493686795234,
      "step": 155350
    },
    {
      "epoch": 44.09877944933295,
      "grad_norm": 5.344523906707764,
      "learning_rate": 5.5919670735168894e-05,
      "loss": 0.0025301782414317133,
      "step": 155360
    },
    {
      "epoch": 44.101617939256315,
      "grad_norm": 0.9273077845573425,
      "learning_rate": 5.591683224524553e-05,
      "loss": 0.0004893086850643158,
      "step": 155370
    },
    {
      "epoch": 44.10445642917968,
      "grad_norm": 0.018016008660197258,
      "learning_rate": 5.591399375532217e-05,
      "loss": 0.0004981115460395813,
      "step": 155380
    },
    {
      "epoch": 44.10729491910304,
      "grad_norm": 0.21669256687164307,
      "learning_rate": 5.5911155265398805e-05,
      "loss": 0.0007338380441069603,
      "step": 155390
    },
    {
      "epoch": 44.110133409026396,
      "grad_norm": 0.019063038751482964,
      "learning_rate": 5.5908316775475446e-05,
      "loss": 0.0010421304032206535,
      "step": 155400
    },
    {
      "epoch": 44.11297189894976,
      "grad_norm": 0.17975348234176636,
      "learning_rate": 5.5905478285552095e-05,
      "loss": 0.0004953864961862564,
      "step": 155410
    },
    {
      "epoch": 44.11581038887312,
      "grad_norm": 0.8710076808929443,
      "learning_rate": 5.590263979562873e-05,
      "loss": 0.0012292493134737014,
      "step": 155420
    },
    {
      "epoch": 44.11864887879648,
      "grad_norm": 0.6368151903152466,
      "learning_rate": 5.589980130570537e-05,
      "loss": 0.004834149032831192,
      "step": 155430
    },
    {
      "epoch": 44.12148736871984,
      "grad_norm": 16.330114364624023,
      "learning_rate": 5.5896962815782005e-05,
      "loss": 0.017424367368221283,
      "step": 155440
    },
    {
      "epoch": 44.1243258586432,
      "grad_norm": 2.5074987411499023,
      "learning_rate": 5.5894124325858647e-05,
      "loss": 0.0013069190084934235,
      "step": 155450
    },
    {
      "epoch": 44.12716434856656,
      "grad_norm": 0.04622608423233032,
      "learning_rate": 5.589128583593528e-05,
      "loss": 0.0004179947078227997,
      "step": 155460
    },
    {
      "epoch": 44.13000283848992,
      "grad_norm": 0.9279323220252991,
      "learning_rate": 5.588844734601193e-05,
      "loss": 0.0008724264800548554,
      "step": 155470
    },
    {
      "epoch": 44.132841328413285,
      "grad_norm": 1.57437002658844,
      "learning_rate": 5.5885608856088564e-05,
      "loss": 0.001136513240635395,
      "step": 155480
    },
    {
      "epoch": 44.13567981833665,
      "grad_norm": 0.744023859500885,
      "learning_rate": 5.5882770366165205e-05,
      "loss": 0.0014183659106492997,
      "step": 155490
    },
    {
      "epoch": 44.138518308260004,
      "grad_norm": 0.02020115777850151,
      "learning_rate": 5.587993187624184e-05,
      "loss": 0.0006978582590818405,
      "step": 155500
    },
    {
      "epoch": 44.138518308260004,
      "eval_accuracy": 0.9761556558784257,
      "eval_loss": 0.08622970432043076,
      "eval_runtime": 31.8332,
      "eval_samples_per_second": 494.044,
      "eval_steps_per_second": 7.728,
      "step": 155500
    },
    {
      "epoch": 44.14135679818337,
      "grad_norm": 0.6633445620536804,
      "learning_rate": 5.587709338631848e-05,
      "loss": 0.0008507579565048218,
      "step": 155510
    },
    {
      "epoch": 44.14419528810673,
      "grad_norm": 0.07336318492889404,
      "learning_rate": 5.5874254896395116e-05,
      "loss": 0.005176513642072678,
      "step": 155520
    },
    {
      "epoch": 44.147033778030085,
      "grad_norm": 5.161157608032227,
      "learning_rate": 5.587141640647176e-05,
      "loss": 0.00277931559830904,
      "step": 155530
    },
    {
      "epoch": 44.14987226795345,
      "grad_norm": 6.645428657531738,
      "learning_rate": 5.5868577916548405e-05,
      "loss": 0.00671527236700058,
      "step": 155540
    },
    {
      "epoch": 44.15271075787681,
      "grad_norm": 0.3229260742664337,
      "learning_rate": 5.586573942662504e-05,
      "loss": 0.0010661270469427108,
      "step": 155550
    },
    {
      "epoch": 44.15554924780017,
      "grad_norm": 3.415201425552368,
      "learning_rate": 5.586290093670168e-05,
      "loss": 0.0021433599293231966,
      "step": 155560
    },
    {
      "epoch": 44.15838773772353,
      "grad_norm": 0.0876779854297638,
      "learning_rate": 5.5860062446778316e-05,
      "loss": 0.012350483983755111,
      "step": 155570
    },
    {
      "epoch": 44.16122622764689,
      "grad_norm": 0.23315846920013428,
      "learning_rate": 5.585722395685495e-05,
      "loss": 0.0018002785742282867,
      "step": 155580
    },
    {
      "epoch": 44.164064717570255,
      "grad_norm": 0.07376910746097565,
      "learning_rate": 5.585438546693159e-05,
      "loss": 0.012228453159332275,
      "step": 155590
    },
    {
      "epoch": 44.16690320749361,
      "grad_norm": 0.08329633623361588,
      "learning_rate": 5.5851546977008227e-05,
      "loss": 0.000987526960670948,
      "step": 155600
    },
    {
      "epoch": 44.169741697416974,
      "grad_norm": 0.21942900121212006,
      "learning_rate": 5.5848708487084875e-05,
      "loss": 0.0008349103853106498,
      "step": 155610
    },
    {
      "epoch": 44.17258018734034,
      "grad_norm": 0.05533232167363167,
      "learning_rate": 5.5845869997161516e-05,
      "loss": 0.0007051410153508187,
      "step": 155620
    },
    {
      "epoch": 44.17541867726369,
      "grad_norm": 0.16942307353019714,
      "learning_rate": 5.584303150723815e-05,
      "loss": 0.006307969987392426,
      "step": 155630
    },
    {
      "epoch": 44.178257167187056,
      "grad_norm": 1.175797700881958,
      "learning_rate": 5.584019301731479e-05,
      "loss": 0.0035138830542564394,
      "step": 155640
    },
    {
      "epoch": 44.18109565711042,
      "grad_norm": 14.757887840270996,
      "learning_rate": 5.5837354527391427e-05,
      "loss": 0.022019030153751375,
      "step": 155650
    },
    {
      "epoch": 44.18393414703378,
      "grad_norm": 0.5485493540763855,
      "learning_rate": 5.583451603746807e-05,
      "loss": 0.015479469299316406,
      "step": 155660
    },
    {
      "epoch": 44.18677263695714,
      "grad_norm": 1.075817584991455,
      "learning_rate": 5.5831677547544716e-05,
      "loss": 0.023088744282722472,
      "step": 155670
    },
    {
      "epoch": 44.1896111268805,
      "grad_norm": 0.04194627329707146,
      "learning_rate": 5.582883905762135e-05,
      "loss": 0.001118011213839054,
      "step": 155680
    },
    {
      "epoch": 44.19244961680386,
      "grad_norm": 0.02505800686776638,
      "learning_rate": 5.582600056769799e-05,
      "loss": 0.0059205576777458194,
      "step": 155690
    },
    {
      "epoch": 44.19528810672722,
      "grad_norm": 0.3731585741043091,
      "learning_rate": 5.582316207777463e-05,
      "loss": 0.003644236922264099,
      "step": 155700
    },
    {
      "epoch": 44.19812659665058,
      "grad_norm": 0.05007988214492798,
      "learning_rate": 5.582032358785126e-05,
      "loss": 0.0004676518961787224,
      "step": 155710
    },
    {
      "epoch": 44.200965086573945,
      "grad_norm": 0.052320413291454315,
      "learning_rate": 5.58174850979279e-05,
      "loss": 0.0016772165894508363,
      "step": 155720
    },
    {
      "epoch": 44.2038035764973,
      "grad_norm": 0.13691850006580353,
      "learning_rate": 5.581464660800454e-05,
      "loss": 0.007365772128105163,
      "step": 155730
    },
    {
      "epoch": 44.20664206642066,
      "grad_norm": 0.23982998728752136,
      "learning_rate": 5.5811808118081185e-05,
      "loss": 0.004035079479217529,
      "step": 155740
    },
    {
      "epoch": 44.209480556344026,
      "grad_norm": Infinity,
      "learning_rate": 5.580896962815783e-05,
      "loss": 0.005990353226661682,
      "step": 155750
    },
    {
      "epoch": 44.21231904626739,
      "grad_norm": 1.0479826927185059,
      "learning_rate": 5.580641498722679e-05,
      "loss": 0.005594062805175781,
      "step": 155760
    },
    {
      "epoch": 44.215157536190745,
      "grad_norm": 0.11009394377470016,
      "learning_rate": 5.5803576497303435e-05,
      "loss": 0.004703034460544586,
      "step": 155770
    },
    {
      "epoch": 44.21799602611411,
      "grad_norm": 0.17580120265483856,
      "learning_rate": 5.580073800738007e-05,
      "loss": 0.0016249192878603936,
      "step": 155780
    },
    {
      "epoch": 44.22083451603747,
      "grad_norm": 0.027019785717129707,
      "learning_rate": 5.579789951745672e-05,
      "loss": 0.0014676047489047051,
      "step": 155790
    },
    {
      "epoch": 44.223673005960826,
      "grad_norm": 0.14550164341926575,
      "learning_rate": 5.579506102753336e-05,
      "loss": 0.014735449850559235,
      "step": 155800
    },
    {
      "epoch": 44.22651149588419,
      "grad_norm": 0.27819570899009705,
      "learning_rate": 5.579222253760999e-05,
      "loss": 0.0011222263798117637,
      "step": 155810
    },
    {
      "epoch": 44.22934998580755,
      "grad_norm": 0.05686723068356514,
      "learning_rate": 5.5789384047686635e-05,
      "loss": 0.0006251854822039605,
      "step": 155820
    },
    {
      "epoch": 44.23218847573091,
      "grad_norm": 1.6666109561920166,
      "learning_rate": 5.578654555776327e-05,
      "loss": 0.003666987642645836,
      "step": 155830
    },
    {
      "epoch": 44.23502696565427,
      "grad_norm": 0.08091799914836884,
      "learning_rate": 5.578370706783991e-05,
      "loss": 0.0009525692090392113,
      "step": 155840
    },
    {
      "epoch": 44.237865455577634,
      "grad_norm": 0.20077741146087646,
      "learning_rate": 5.578086857791656e-05,
      "loss": 0.0007513668388128281,
      "step": 155850
    },
    {
      "epoch": 44.240703945501,
      "grad_norm": 0.3532382845878601,
      "learning_rate": 5.5778030087993193e-05,
      "loss": 0.0012952215969562531,
      "step": 155860
    },
    {
      "epoch": 44.24354243542435,
      "grad_norm": 1.4566701650619507,
      "learning_rate": 5.5775191598069835e-05,
      "loss": 0.00212986133992672,
      "step": 155870
    },
    {
      "epoch": 44.246380925347715,
      "grad_norm": 2.030355453491211,
      "learning_rate": 5.577235310814647e-05,
      "loss": 0.008475768566131591,
      "step": 155880
    },
    {
      "epoch": 44.24921941527108,
      "grad_norm": 0.2468125969171524,
      "learning_rate": 5.5769514618223104e-05,
      "loss": 0.001193469762802124,
      "step": 155890
    },
    {
      "epoch": 44.252057905194434,
      "grad_norm": 5.4765849113464355,
      "learning_rate": 5.5766676128299745e-05,
      "loss": 0.006750188767910004,
      "step": 155900
    },
    {
      "epoch": 44.2548963951178,
      "grad_norm": 0.23432430624961853,
      "learning_rate": 5.576383763837638e-05,
      "loss": 0.001567850448191166,
      "step": 155910
    },
    {
      "epoch": 44.25773488504116,
      "grad_norm": 0.1899024248123169,
      "learning_rate": 5.576099914845303e-05,
      "loss": 0.00036251526325941085,
      "step": 155920
    },
    {
      "epoch": 44.260573374964515,
      "grad_norm": 0.19278600811958313,
      "learning_rate": 5.575816065852967e-05,
      "loss": 0.0036965854465961455,
      "step": 155930
    },
    {
      "epoch": 44.26341186488788,
      "grad_norm": 1.2772260904312134,
      "learning_rate": 5.5755322168606304e-05,
      "loss": 0.000999061018228531,
      "step": 155940
    },
    {
      "epoch": 44.26625035481124,
      "grad_norm": 1.1016117334365845,
      "learning_rate": 5.5752483678682945e-05,
      "loss": 0.00040059220045804975,
      "step": 155950
    },
    {
      "epoch": 44.269088844734604,
      "grad_norm": 0.21109764277935028,
      "learning_rate": 5.574964518875958e-05,
      "loss": 0.0005548624321818352,
      "step": 155960
    },
    {
      "epoch": 44.27192733465796,
      "grad_norm": 0.0556270033121109,
      "learning_rate": 5.574680669883622e-05,
      "loss": 0.0008237482979893684,
      "step": 155970
    },
    {
      "epoch": 44.27476582458132,
      "grad_norm": 0.17667454481124878,
      "learning_rate": 5.5743968208912856e-05,
      "loss": 0.003913568705320359,
      "step": 155980
    },
    {
      "epoch": 44.277604314504686,
      "grad_norm": 0.8863526582717896,
      "learning_rate": 5.5741129718989504e-05,
      "loss": 0.0010313203558325768,
      "step": 155990
    },
    {
      "epoch": 44.28044280442804,
      "grad_norm": 0.047872792929410934,
      "learning_rate": 5.573829122906614e-05,
      "loss": 0.007978184521198273,
      "step": 156000
    },
    {
      "epoch": 44.28044280442804,
      "eval_accuracy": 0.9725949004896038,
      "eval_loss": 0.09618312120437622,
      "eval_runtime": 32.0601,
      "eval_samples_per_second": 490.547,
      "eval_steps_per_second": 7.673,
      "step": 156000
    },
    {
      "epoch": 44.283281294351404,
      "grad_norm": 8.187911987304688,
      "learning_rate": 5.573545273914278e-05,
      "loss": 0.0035466670989990233,
      "step": 156010
    },
    {
      "epoch": 44.28611978427477,
      "grad_norm": 1.4802619218826294,
      "learning_rate": 5.5732614249219415e-05,
      "loss": 0.0011472728103399277,
      "step": 156020
    },
    {
      "epoch": 44.28895827419813,
      "grad_norm": 6.284488201141357,
      "learning_rate": 5.5729775759296056e-05,
      "loss": 0.003243030607700348,
      "step": 156030
    },
    {
      "epoch": 44.291796764121486,
      "grad_norm": 0.009590772911906242,
      "learning_rate": 5.572693726937269e-05,
      "loss": 0.0037116315215826035,
      "step": 156040
    },
    {
      "epoch": 44.29463525404485,
      "grad_norm": 0.03841324895620346,
      "learning_rate": 5.572409877944934e-05,
      "loss": 0.001790476031601429,
      "step": 156050
    },
    {
      "epoch": 44.29747374396821,
      "grad_norm": 0.12985964119434357,
      "learning_rate": 5.572126028952598e-05,
      "loss": 0.0009689955040812492,
      "step": 156060
    },
    {
      "epoch": 44.30031223389157,
      "grad_norm": 0.1811363846063614,
      "learning_rate": 5.5718421799602615e-05,
      "loss": 0.0054326742887496945,
      "step": 156070
    },
    {
      "epoch": 44.30315072381493,
      "grad_norm": 14.500101089477539,
      "learning_rate": 5.5715583309679256e-05,
      "loss": 0.009616748988628387,
      "step": 156080
    },
    {
      "epoch": 44.30598921373829,
      "grad_norm": 19.557910919189453,
      "learning_rate": 5.571274481975589e-05,
      "loss": 0.014311686158180237,
      "step": 156090
    },
    {
      "epoch": 44.30882770366165,
      "grad_norm": 0.08027077466249466,
      "learning_rate": 5.5709906329832525e-05,
      "loss": 0.0009311344474554062,
      "step": 156100
    },
    {
      "epoch": 44.31166619358501,
      "grad_norm": 3.8744192123413086,
      "learning_rate": 5.570706783990917e-05,
      "loss": 0.002939853072166443,
      "step": 156110
    },
    {
      "epoch": 44.314504683508375,
      "grad_norm": 0.10265488922595978,
      "learning_rate": 5.5704229349985815e-05,
      "loss": 0.0017543647438287735,
      "step": 156120
    },
    {
      "epoch": 44.31734317343174,
      "grad_norm": 0.7896320223808289,
      "learning_rate": 5.570139086006245e-05,
      "loss": 0.0042994610965251924,
      "step": 156130
    },
    {
      "epoch": 44.32018166335509,
      "grad_norm": 1.4950884580612183,
      "learning_rate": 5.569855237013909e-05,
      "loss": 0.005641095712780953,
      "step": 156140
    },
    {
      "epoch": 44.323020153278456,
      "grad_norm": 0.0156754981726408,
      "learning_rate": 5.5695713880215725e-05,
      "loss": 0.0016691450029611588,
      "step": 156150
    },
    {
      "epoch": 44.32585864320182,
      "grad_norm": 0.4329633116722107,
      "learning_rate": 5.569287539029237e-05,
      "loss": 0.0005211511626839638,
      "step": 156160
    },
    {
      "epoch": 44.328697133125175,
      "grad_norm": 0.2618117928504944,
      "learning_rate": 5.5690036900369e-05,
      "loss": 0.0018415918573737144,
      "step": 156170
    },
    {
      "epoch": 44.33153562304854,
      "grad_norm": 0.21483588218688965,
      "learning_rate": 5.568719841044564e-05,
      "loss": 0.00474206693470478,
      "step": 156180
    },
    {
      "epoch": 44.3343741129719,
      "grad_norm": 0.14301232993602753,
      "learning_rate": 5.568435992052229e-05,
      "loss": 0.0010020313784480095,
      "step": 156190
    },
    {
      "epoch": 44.33721260289526,
      "grad_norm": 0.21054522693157196,
      "learning_rate": 5.5681521430598926e-05,
      "loss": 0.0062213890254497525,
      "step": 156200
    },
    {
      "epoch": 44.34005109281862,
      "grad_norm": 0.11314497888088226,
      "learning_rate": 5.567868294067557e-05,
      "loss": 0.0023398233577609064,
      "step": 156210
    },
    {
      "epoch": 44.34288958274198,
      "grad_norm": 1.2806332111358643,
      "learning_rate": 5.56758444507522e-05,
      "loss": 0.004950839653611183,
      "step": 156220
    },
    {
      "epoch": 44.345728072665345,
      "grad_norm": 1.388614296913147,
      "learning_rate": 5.5673005960828836e-05,
      "loss": 0.002655931003391743,
      "step": 156230
    },
    {
      "epoch": 44.3485665625887,
      "grad_norm": 0.05348653346300125,
      "learning_rate": 5.567016747090548e-05,
      "loss": 0.0009570807218551636,
      "step": 156240
    },
    {
      "epoch": 44.351405052512064,
      "grad_norm": 0.28925183415412903,
      "learning_rate": 5.5667328980982126e-05,
      "loss": 0.004495839029550553,
      "step": 156250
    },
    {
      "epoch": 44.35424354243543,
      "grad_norm": 0.5233582258224487,
      "learning_rate": 5.566449049105876e-05,
      "loss": 0.001182178221642971,
      "step": 156260
    },
    {
      "epoch": 44.35708203235878,
      "grad_norm": 0.5457421541213989,
      "learning_rate": 5.56616520011354e-05,
      "loss": 0.000948018953204155,
      "step": 156270
    },
    {
      "epoch": 44.359920522282145,
      "grad_norm": 0.1991112381219864,
      "learning_rate": 5.5658813511212036e-05,
      "loss": 0.006093402951955795,
      "step": 156280
    },
    {
      "epoch": 44.36275901220551,
      "grad_norm": 0.13211938738822937,
      "learning_rate": 5.565597502128868e-05,
      "loss": 0.0007587602362036705,
      "step": 156290
    },
    {
      "epoch": 44.365597502128864,
      "grad_norm": 0.014945046044886112,
      "learning_rate": 5.565313653136531e-05,
      "loss": 0.0016382332891225816,
      "step": 156300
    },
    {
      "epoch": 44.36843599205223,
      "grad_norm": 0.2198319137096405,
      "learning_rate": 5.5650298041441954e-05,
      "loss": 0.0027062106877565385,
      "step": 156310
    },
    {
      "epoch": 44.37127448197559,
      "grad_norm": 0.14761118590831757,
      "learning_rate": 5.56474595515186e-05,
      "loss": 0.0017314694821834564,
      "step": 156320
    },
    {
      "epoch": 44.37411297189895,
      "grad_norm": 0.3220134675502777,
      "learning_rate": 5.5644621061595236e-05,
      "loss": 0.0005900824442505836,
      "step": 156330
    },
    {
      "epoch": 44.37695146182231,
      "grad_norm": 2.4303300380706787,
      "learning_rate": 5.564178257167188e-05,
      "loss": 0.0010822070762515068,
      "step": 156340
    },
    {
      "epoch": 44.37978995174567,
      "grad_norm": 0.6298781037330627,
      "learning_rate": 5.563894408174851e-05,
      "loss": 0.0008630633354187011,
      "step": 156350
    },
    {
      "epoch": 44.382628441669034,
      "grad_norm": 0.04808414354920387,
      "learning_rate": 5.563610559182515e-05,
      "loss": 0.011258452385663986,
      "step": 156360
    },
    {
      "epoch": 44.38546693159239,
      "grad_norm": 0.06292856484651566,
      "learning_rate": 5.563326710190179e-05,
      "loss": 0.0007125638425350189,
      "step": 156370
    },
    {
      "epoch": 44.38830542151575,
      "grad_norm": 0.07413753122091293,
      "learning_rate": 5.563042861197842e-05,
      "loss": 0.0014533979818224906,
      "step": 156380
    },
    {
      "epoch": 44.391143911439116,
      "grad_norm": 0.13524796068668365,
      "learning_rate": 5.562759012205507e-05,
      "loss": 0.0012958969920873642,
      "step": 156390
    },
    {
      "epoch": 44.39398240136248,
      "grad_norm": 0.15299531817436218,
      "learning_rate": 5.562475163213171e-05,
      "loss": 0.0003364142030477524,
      "step": 156400
    },
    {
      "epoch": 44.396820891285834,
      "grad_norm": 0.29562321305274963,
      "learning_rate": 5.562191314220835e-05,
      "loss": 0.0009649790823459625,
      "step": 156410
    },
    {
      "epoch": 44.3996593812092,
      "grad_norm": 1.9424642324447632,
      "learning_rate": 5.561907465228499e-05,
      "loss": 0.0019111409783363342,
      "step": 156420
    },
    {
      "epoch": 44.40249787113256,
      "grad_norm": 0.030079824849963188,
      "learning_rate": 5.561623616236162e-05,
      "loss": 0.003973577171564102,
      "step": 156430
    },
    {
      "epoch": 44.405336361055916,
      "grad_norm": 1.194277286529541,
      "learning_rate": 5.5613397672438264e-05,
      "loss": 0.0012745341286063195,
      "step": 156440
    },
    {
      "epoch": 44.40817485097928,
      "grad_norm": 0.27833881974220276,
      "learning_rate": 5.561055918251491e-05,
      "loss": 0.002846476808190346,
      "step": 156450
    },
    {
      "epoch": 44.41101334090264,
      "grad_norm": 0.03315252438187599,
      "learning_rate": 5.560772069259155e-05,
      "loss": 0.004304178059101105,
      "step": 156460
    },
    {
      "epoch": 44.413851830826,
      "grad_norm": 0.44952377676963806,
      "learning_rate": 5.560488220266818e-05,
      "loss": 0.0010804397985339164,
      "step": 156470
    },
    {
      "epoch": 44.41669032074936,
      "grad_norm": 7.029206275939941,
      "learning_rate": 5.560204371274482e-05,
      "loss": 0.004529815167188644,
      "step": 156480
    },
    {
      "epoch": 44.41952881067272,
      "grad_norm": 0.6543444991111755,
      "learning_rate": 5.559920522282146e-05,
      "loss": 0.000869227759540081,
      "step": 156490
    },
    {
      "epoch": 44.422367300596086,
      "grad_norm": 0.3735924959182739,
      "learning_rate": 5.55963667328981e-05,
      "loss": 0.0007621526718139648,
      "step": 156500
    },
    {
      "epoch": 44.422367300596086,
      "eval_accuracy": 0.9753926368665352,
      "eval_loss": 0.08934874832630157,
      "eval_runtime": 32.9707,
      "eval_samples_per_second": 476.999,
      "eval_steps_per_second": 7.461,
      "step": 156500
    },
    {
      "epoch": 44.42520579051944,
      "grad_norm": 0.16405533254146576,
      "learning_rate": 5.5593528242974734e-05,
      "loss": 0.003479709103703499,
      "step": 156510
    },
    {
      "epoch": 44.428044280442805,
      "grad_norm": 0.0796213299036026,
      "learning_rate": 5.559068975305138e-05,
      "loss": 0.008253187686204911,
      "step": 156520
    },
    {
      "epoch": 44.43088277036617,
      "grad_norm": 19.184030532836914,
      "learning_rate": 5.558785126312802e-05,
      "loss": 0.00849815011024475,
      "step": 156530
    },
    {
      "epoch": 44.43372126028952,
      "grad_norm": 0.02729126811027527,
      "learning_rate": 5.558501277320466e-05,
      "loss": 0.005532644689083099,
      "step": 156540
    },
    {
      "epoch": 44.436559750212886,
      "grad_norm": 0.08558712154626846,
      "learning_rate": 5.55821742832813e-05,
      "loss": 0.003780781850218773,
      "step": 156550
    },
    {
      "epoch": 44.43939824013625,
      "grad_norm": 0.0852358415722847,
      "learning_rate": 5.5579335793357934e-05,
      "loss": 0.0030274834483861924,
      "step": 156560
    },
    {
      "epoch": 44.442236730059605,
      "grad_norm": 0.032969292253255844,
      "learning_rate": 5.557649730343457e-05,
      "loss": 0.003079797141253948,
      "step": 156570
    },
    {
      "epoch": 44.44507521998297,
      "grad_norm": 0.05022619664669037,
      "learning_rate": 5.557365881351122e-05,
      "loss": 0.001191606931388378,
      "step": 156580
    },
    {
      "epoch": 44.44791370990633,
      "grad_norm": 0.053763579577207565,
      "learning_rate": 5.557082032358786e-05,
      "loss": 0.0009511610493063927,
      "step": 156590
    },
    {
      "epoch": 44.450752199829694,
      "grad_norm": 0.8627061247825623,
      "learning_rate": 5.556798183366449e-05,
      "loss": 0.003018593601882458,
      "step": 156600
    },
    {
      "epoch": 44.45359068975305,
      "grad_norm": 4.448215961456299,
      "learning_rate": 5.5565143343741134e-05,
      "loss": 0.008427640050649643,
      "step": 156610
    },
    {
      "epoch": 44.45642917967641,
      "grad_norm": 1.170464277267456,
      "learning_rate": 5.556230485381777e-05,
      "loss": 0.004199491068720818,
      "step": 156620
    },
    {
      "epoch": 44.459267669599775,
      "grad_norm": 0.0877622589468956,
      "learning_rate": 5.555946636389441e-05,
      "loss": 0.0013260142877697945,
      "step": 156630
    },
    {
      "epoch": 44.46210615952313,
      "grad_norm": 0.39314037561416626,
      "learning_rate": 5.5556627873971044e-05,
      "loss": 0.004947011917829513,
      "step": 156640
    },
    {
      "epoch": 44.464944649446494,
      "grad_norm": 0.22980397939682007,
      "learning_rate": 5.555378938404769e-05,
      "loss": 0.005471432954072953,
      "step": 156650
    },
    {
      "epoch": 44.46778313936986,
      "grad_norm": 0.2401466816663742,
      "learning_rate": 5.5550950894124334e-05,
      "loss": 0.002054726146161556,
      "step": 156660
    },
    {
      "epoch": 44.47062162929321,
      "grad_norm": 0.058545272797346115,
      "learning_rate": 5.554811240420097e-05,
      "loss": 0.0003682799637317657,
      "step": 156670
    },
    {
      "epoch": 44.473460119216575,
      "grad_norm": 0.18302831053733826,
      "learning_rate": 5.554527391427761e-05,
      "loss": 0.0005778366699814797,
      "step": 156680
    },
    {
      "epoch": 44.47629860913994,
      "grad_norm": 0.5530268549919128,
      "learning_rate": 5.5542435424354245e-05,
      "loss": 0.0011473154649138451,
      "step": 156690
    },
    {
      "epoch": 44.4791370990633,
      "grad_norm": 0.11177114397287369,
      "learning_rate": 5.553959693443088e-05,
      "loss": 0.0008618583902716637,
      "step": 156700
    },
    {
      "epoch": 44.48197558898666,
      "grad_norm": 2.068586587905884,
      "learning_rate": 5.553675844450752e-05,
      "loss": 0.0035226158797740936,
      "step": 156710
    },
    {
      "epoch": 44.48481407891002,
      "grad_norm": 16.873788833618164,
      "learning_rate": 5.553391995458417e-05,
      "loss": 0.012060369551181793,
      "step": 156720
    },
    {
      "epoch": 44.48765256883338,
      "grad_norm": 6.657575607299805,
      "learning_rate": 5.55310814646608e-05,
      "loss": 0.0025987731292843818,
      "step": 156730
    },
    {
      "epoch": 44.49049105875674,
      "grad_norm": 0.3594381511211395,
      "learning_rate": 5.5528242974737445e-05,
      "loss": 0.004024265706539154,
      "step": 156740
    },
    {
      "epoch": 44.4933295486801,
      "grad_norm": 1.5028173923492432,
      "learning_rate": 5.552540448481408e-05,
      "loss": 0.0045437000691890715,
      "step": 156750
    },
    {
      "epoch": 44.496168038603464,
      "grad_norm": 5.458350658416748,
      "learning_rate": 5.552256599489072e-05,
      "loss": 0.002493935637176037,
      "step": 156760
    },
    {
      "epoch": 44.49900652852683,
      "grad_norm": 0.19793866574764252,
      "learning_rate": 5.5519727504967355e-05,
      "loss": 0.009273440390825272,
      "step": 156770
    },
    {
      "epoch": 44.50184501845018,
      "grad_norm": 4.402702331542969,
      "learning_rate": 5.5516889015044e-05,
      "loss": 0.005135349929332733,
      "step": 156780
    },
    {
      "epoch": 44.504683508373546,
      "grad_norm": 2.735562324523926,
      "learning_rate": 5.5514050525120645e-05,
      "loss": 0.004181587696075439,
      "step": 156790
    },
    {
      "epoch": 44.50752199829691,
      "grad_norm": 0.13431775569915771,
      "learning_rate": 5.551121203519728e-05,
      "loss": 0.0009526653215289116,
      "step": 156800
    },
    {
      "epoch": 44.510360488220265,
      "grad_norm": 0.040909476578235626,
      "learning_rate": 5.5508373545273914e-05,
      "loss": 0.002434661239385605,
      "step": 156810
    },
    {
      "epoch": 44.51319897814363,
      "grad_norm": 1.6300687789916992,
      "learning_rate": 5.5505535055350555e-05,
      "loss": 0.008047007769346238,
      "step": 156820
    },
    {
      "epoch": 44.51603746806699,
      "grad_norm": 1.2510794401168823,
      "learning_rate": 5.550269656542719e-05,
      "loss": 0.0012539243325591087,
      "step": 156830
    },
    {
      "epoch": 44.518875957990346,
      "grad_norm": 2.2361247539520264,
      "learning_rate": 5.549985807550383e-05,
      "loss": 0.0011626319959759713,
      "step": 156840
    },
    {
      "epoch": 44.52171444791371,
      "grad_norm": 1.9043679237365723,
      "learning_rate": 5.549701958558048e-05,
      "loss": 0.0007231362164020538,
      "step": 156850
    },
    {
      "epoch": 44.52455293783707,
      "grad_norm": 0.19966337084770203,
      "learning_rate": 5.5494181095657114e-05,
      "loss": 0.0008175697177648545,
      "step": 156860
    },
    {
      "epoch": 44.527391427760435,
      "grad_norm": 0.4779660999774933,
      "learning_rate": 5.5491342605733755e-05,
      "loss": 0.0009424742311239243,
      "step": 156870
    },
    {
      "epoch": 44.53022991768379,
      "grad_norm": 0.47506165504455566,
      "learning_rate": 5.548850411581039e-05,
      "loss": 0.00657263770699501,
      "step": 156880
    },
    {
      "epoch": 44.53306840760715,
      "grad_norm": 0.19432635605335236,
      "learning_rate": 5.548566562588703e-05,
      "loss": 0.003304903954267502,
      "step": 156890
    },
    {
      "epoch": 44.535906897530516,
      "grad_norm": 17.39341926574707,
      "learning_rate": 5.5482827135963666e-05,
      "loss": 0.003694375976920128,
      "step": 156900
    },
    {
      "epoch": 44.53874538745387,
      "grad_norm": 0.04237337037920952,
      "learning_rate": 5.547998864604031e-05,
      "loss": 0.0011420505121350288,
      "step": 156910
    },
    {
      "epoch": 44.541583877377235,
      "grad_norm": 0.039938390254974365,
      "learning_rate": 5.5477150156116955e-05,
      "loss": 0.001709469221532345,
      "step": 156920
    },
    {
      "epoch": 44.5444223673006,
      "grad_norm": 0.02060115337371826,
      "learning_rate": 5.547431166619359e-05,
      "loss": 0.0006421186029911041,
      "step": 156930
    },
    {
      "epoch": 44.547260857223954,
      "grad_norm": 0.027185415849089622,
      "learning_rate": 5.5471473176270225e-05,
      "loss": 0.0010054798796772957,
      "step": 156940
    },
    {
      "epoch": 44.55009934714732,
      "grad_norm": 0.09638462215662003,
      "learning_rate": 5.5468634686346866e-05,
      "loss": 0.005101707205176354,
      "step": 156950
    },
    {
      "epoch": 44.55293783707068,
      "grad_norm": 0.16471844911575317,
      "learning_rate": 5.54657961964235e-05,
      "loss": 0.0036167684942483903,
      "step": 156960
    },
    {
      "epoch": 44.55577632699404,
      "grad_norm": 0.6293802261352539,
      "learning_rate": 5.546295770650014e-05,
      "loss": 0.010815633088350296,
      "step": 156970
    },
    {
      "epoch": 44.5586148169174,
      "grad_norm": 0.2345420867204666,
      "learning_rate": 5.546011921657679e-05,
      "loss": 0.001975184865295887,
      "step": 156980
    },
    {
      "epoch": 44.56145330684076,
      "grad_norm": 0.11501927673816681,
      "learning_rate": 5.5457280726653425e-05,
      "loss": 0.004590024799108505,
      "step": 156990
    },
    {
      "epoch": 44.564291796764124,
      "grad_norm": 0.22351092100143433,
      "learning_rate": 5.5454442236730066e-05,
      "loss": 0.004032894968986511,
      "step": 157000
    },
    {
      "epoch": 44.564291796764124,
      "eval_accuracy": 0.9736758440897819,
      "eval_loss": 0.09944281727075577,
      "eval_runtime": 32.2713,
      "eval_samples_per_second": 487.337,
      "eval_steps_per_second": 7.623,
      "step": 157000
    },
    {
      "epoch": 44.56713028668748,
      "grad_norm": 0.3598601818084717,
      "learning_rate": 5.54516037468067e-05,
      "loss": 0.0010118138045072556,
      "step": 157010
    },
    {
      "epoch": 44.56996877661084,
      "grad_norm": 0.0883999690413475,
      "learning_rate": 5.544876525688334e-05,
      "loss": 0.0013120384886860848,
      "step": 157020
    },
    {
      "epoch": 44.572807266534205,
      "grad_norm": 0.11062635481357574,
      "learning_rate": 5.544592676695998e-05,
      "loss": 0.0014809662476181983,
      "step": 157030
    },
    {
      "epoch": 44.57564575645756,
      "grad_norm": 4.825936794281006,
      "learning_rate": 5.544308827703661e-05,
      "loss": 0.0015304675325751305,
      "step": 157040
    },
    {
      "epoch": 44.578484246380924,
      "grad_norm": 1.0261945724487305,
      "learning_rate": 5.5440249787113266e-05,
      "loss": 0.001288410648703575,
      "step": 157050
    },
    {
      "epoch": 44.58132273630429,
      "grad_norm": 1.421907663345337,
      "learning_rate": 5.54374112971899e-05,
      "loss": 0.0020915113389492037,
      "step": 157060
    },
    {
      "epoch": 44.58416122622765,
      "grad_norm": 0.6488006711006165,
      "learning_rate": 5.5434572807266535e-05,
      "loss": 0.0036552704870700834,
      "step": 157070
    },
    {
      "epoch": 44.586999716151006,
      "grad_norm": 0.19210556149482727,
      "learning_rate": 5.543173431734318e-05,
      "loss": 0.0008521279320120812,
      "step": 157080
    },
    {
      "epoch": 44.58983820607437,
      "grad_norm": 0.8077877163887024,
      "learning_rate": 5.542889582741981e-05,
      "loss": 0.0020171890035271646,
      "step": 157090
    },
    {
      "epoch": 44.59267669599773,
      "grad_norm": 0.21412065625190735,
      "learning_rate": 5.542605733749645e-05,
      "loss": 0.0013764137402176857,
      "step": 157100
    },
    {
      "epoch": 44.59551518592109,
      "grad_norm": 3.524261236190796,
      "learning_rate": 5.542321884757309e-05,
      "loss": 0.006814248114824295,
      "step": 157110
    },
    {
      "epoch": 44.59835367584445,
      "grad_norm": 0.1553005874156952,
      "learning_rate": 5.5420380357649736e-05,
      "loss": 0.001669175736606121,
      "step": 157120
    },
    {
      "epoch": 44.60119216576781,
      "grad_norm": 0.27482202649116516,
      "learning_rate": 5.541754186772638e-05,
      "loss": 0.003913236036896706,
      "step": 157130
    },
    {
      "epoch": 44.604030655691176,
      "grad_norm": 0.02628505788743496,
      "learning_rate": 5.541470337780301e-05,
      "loss": 0.006351328641176224,
      "step": 157140
    },
    {
      "epoch": 44.60686914561453,
      "grad_norm": 0.0274568572640419,
      "learning_rate": 5.541186488787965e-05,
      "loss": 0.005317531526088715,
      "step": 157150
    },
    {
      "epoch": 44.609707635537895,
      "grad_norm": 0.053229253739118576,
      "learning_rate": 5.540902639795629e-05,
      "loss": 0.002090226300060749,
      "step": 157160
    },
    {
      "epoch": 44.61254612546126,
      "grad_norm": 7.986204147338867,
      "learning_rate": 5.540618790803292e-05,
      "loss": 0.0033102646470069886,
      "step": 157170
    },
    {
      "epoch": 44.61538461538461,
      "grad_norm": 0.46482500433921814,
      "learning_rate": 5.540334941810957e-05,
      "loss": 0.009142915904521941,
      "step": 157180
    },
    {
      "epoch": 44.618223105307976,
      "grad_norm": 2.049476146697998,
      "learning_rate": 5.540051092818621e-05,
      "loss": 0.0022250888869166374,
      "step": 157190
    },
    {
      "epoch": 44.62106159523134,
      "grad_norm": 1.8629913330078125,
      "learning_rate": 5.5397672438262846e-05,
      "loss": 0.0019881924614310264,
      "step": 157200
    },
    {
      "epoch": 44.623900085154695,
      "grad_norm": 0.06500549614429474,
      "learning_rate": 5.539483394833949e-05,
      "loss": 0.002598787099123001,
      "step": 157210
    },
    {
      "epoch": 44.62673857507806,
      "grad_norm": 0.8193596005439758,
      "learning_rate": 5.539199545841612e-05,
      "loss": 0.012364324927330018,
      "step": 157220
    },
    {
      "epoch": 44.62957706500142,
      "grad_norm": 0.020166484639048576,
      "learning_rate": 5.5389156968492764e-05,
      "loss": 0.003661905229091644,
      "step": 157230
    },
    {
      "epoch": 44.63241555492478,
      "grad_norm": 2.322923183441162,
      "learning_rate": 5.53863184785694e-05,
      "loss": 0.0022890416905283926,
      "step": 157240
    },
    {
      "epoch": 44.63525404484814,
      "grad_norm": 0.14427955448627472,
      "learning_rate": 5.5383479988646046e-05,
      "loss": 0.004511619359254837,
      "step": 157250
    },
    {
      "epoch": 44.6380925347715,
      "grad_norm": 0.12158554047346115,
      "learning_rate": 5.538064149872269e-05,
      "loss": 0.00046363212168216706,
      "step": 157260
    },
    {
      "epoch": 44.640931024694865,
      "grad_norm": 0.6672930121421814,
      "learning_rate": 5.537780300879932e-05,
      "loss": 0.004486248642206192,
      "step": 157270
    },
    {
      "epoch": 44.64376951461822,
      "grad_norm": 0.4299502968788147,
      "learning_rate": 5.537496451887596e-05,
      "loss": 0.00474189892411232,
      "step": 157280
    },
    {
      "epoch": 44.646608004541584,
      "grad_norm": 0.1382371485233307,
      "learning_rate": 5.53721260289526e-05,
      "loss": 0.001899905689060688,
      "step": 157290
    },
    {
      "epoch": 44.64944649446495,
      "grad_norm": 0.447896808385849,
      "learning_rate": 5.536928753902923e-05,
      "loss": 0.010579816997051239,
      "step": 157300
    },
    {
      "epoch": 44.6522849843883,
      "grad_norm": 0.5387517213821411,
      "learning_rate": 5.5366449049105874e-05,
      "loss": 0.0007550682872533798,
      "step": 157310
    },
    {
      "epoch": 44.655123474311665,
      "grad_norm": 1.9794776439666748,
      "learning_rate": 5.536361055918252e-05,
      "loss": 0.0013253185898065567,
      "step": 157320
    },
    {
      "epoch": 44.65796196423503,
      "grad_norm": 0.6754081845283508,
      "learning_rate": 5.536077206925916e-05,
      "loss": 0.0031715922057628633,
      "step": 157330
    },
    {
      "epoch": 44.66080045415839,
      "grad_norm": 0.13467875123023987,
      "learning_rate": 5.53579335793358e-05,
      "loss": 0.0006900811567902565,
      "step": 157340
    },
    {
      "epoch": 44.66363894408175,
      "grad_norm": 0.02628202736377716,
      "learning_rate": 5.535509508941243e-05,
      "loss": 0.003623931109905243,
      "step": 157350
    },
    {
      "epoch": 44.66647743400511,
      "grad_norm": 0.08537986874580383,
      "learning_rate": 5.5352256599489074e-05,
      "loss": 0.0027131034061312676,
      "step": 157360
    },
    {
      "epoch": 44.66931592392847,
      "grad_norm": 0.10028999298810959,
      "learning_rate": 5.534941810956571e-05,
      "loss": 0.0012365680187940598,
      "step": 157370
    },
    {
      "epoch": 44.67215441385183,
      "grad_norm": 0.043753731995821,
      "learning_rate": 5.534657961964236e-05,
      "loss": 0.0054683715105056764,
      "step": 157380
    },
    {
      "epoch": 44.67499290377519,
      "grad_norm": 13.11633586883545,
      "learning_rate": 5.5343741129719e-05,
      "loss": 0.00984930396080017,
      "step": 157390
    },
    {
      "epoch": 44.677831393698554,
      "grad_norm": 2.39262318611145,
      "learning_rate": 5.534090263979563e-05,
      "loss": 0.0043279752135276794,
      "step": 157400
    },
    {
      "epoch": 44.68066988362191,
      "grad_norm": 5.638765335083008,
      "learning_rate": 5.533806414987227e-05,
      "loss": 0.015101422369480134,
      "step": 157410
    },
    {
      "epoch": 44.68350837354527,
      "grad_norm": 1.4931830167770386,
      "learning_rate": 5.533522565994891e-05,
      "loss": 0.0011003555729985238,
      "step": 157420
    },
    {
      "epoch": 44.686346863468636,
      "grad_norm": 1.04764986038208,
      "learning_rate": 5.5332387170025544e-05,
      "loss": 0.0030232537537813187,
      "step": 157430
    },
    {
      "epoch": 44.689185353392,
      "grad_norm": 0.7580337524414062,
      "learning_rate": 5.5329548680102185e-05,
      "loss": 0.0013986112549901008,
      "step": 157440
    },
    {
      "epoch": 44.692023843315354,
      "grad_norm": 0.152394637465477,
      "learning_rate": 5.532671019017883e-05,
      "loss": 0.003720354661345482,
      "step": 157450
    },
    {
      "epoch": 44.69486233323872,
      "grad_norm": 2.5823490619659424,
      "learning_rate": 5.532387170025547e-05,
      "loss": 0.008514960110187531,
      "step": 157460
    },
    {
      "epoch": 44.69770082316208,
      "grad_norm": 0.0679573267698288,
      "learning_rate": 5.532103321033211e-05,
      "loss": 0.0012488575652241708,
      "step": 157470
    },
    {
      "epoch": 44.700539313085436,
      "grad_norm": 0.09576544165611267,
      "learning_rate": 5.5318194720408744e-05,
      "loss": 0.0026581190526485444,
      "step": 157480
    },
    {
      "epoch": 44.7033778030088,
      "grad_norm": 0.04460605978965759,
      "learning_rate": 5.5315356230485385e-05,
      "loss": 0.0015082472935318946,
      "step": 157490
    },
    {
      "epoch": 44.70621629293216,
      "grad_norm": 0.1593327820301056,
      "learning_rate": 5.531251774056202e-05,
      "loss": 0.0024950157850980757,
      "step": 157500
    },
    {
      "epoch": 44.70621629293216,
      "eval_accuracy": 0.972976409995549,
      "eval_loss": 0.10400718450546265,
      "eval_runtime": 32.28,
      "eval_samples_per_second": 487.206,
      "eval_steps_per_second": 7.621,
      "step": 157500
    },
    {
      "epoch": 44.70905478285552,
      "grad_norm": 0.07861676812171936,
      "learning_rate": 5.5309679250638654e-05,
      "loss": 0.004844729602336883,
      "step": 157510
    },
    {
      "epoch": 44.71189327277888,
      "grad_norm": 0.11340316385030746,
      "learning_rate": 5.530684076071531e-05,
      "loss": 0.001508037932217121,
      "step": 157520
    },
    {
      "epoch": 44.71473176270224,
      "grad_norm": 1.2561005353927612,
      "learning_rate": 5.5304002270791944e-05,
      "loss": 0.0034207697957754136,
      "step": 157530
    },
    {
      "epoch": 44.717570252625606,
      "grad_norm": 1.3330212831497192,
      "learning_rate": 5.530116378086858e-05,
      "loss": 0.0008071361109614373,
      "step": 157540
    },
    {
      "epoch": 44.72040874254896,
      "grad_norm": 0.05329032614827156,
      "learning_rate": 5.529832529094522e-05,
      "loss": 0.00037527326494455335,
      "step": 157550
    },
    {
      "epoch": 44.723247232472325,
      "grad_norm": 0.04077674821019173,
      "learning_rate": 5.5295486801021854e-05,
      "loss": 0.0013462277129292488,
      "step": 157560
    },
    {
      "epoch": 44.72608572239569,
      "grad_norm": 0.18929848074913025,
      "learning_rate": 5.5292648311098496e-05,
      "loss": 0.002169429138302803,
      "step": 157570
    },
    {
      "epoch": 44.72892421231904,
      "grad_norm": 0.19470995664596558,
      "learning_rate": 5.5289809821175144e-05,
      "loss": 0.001561649888753891,
      "step": 157580
    },
    {
      "epoch": 44.731762702242406,
      "grad_norm": 0.5087531805038452,
      "learning_rate": 5.528697133125178e-05,
      "loss": 0.0035420402884483336,
      "step": 157590
    },
    {
      "epoch": 44.73460119216577,
      "grad_norm": 10.44636344909668,
      "learning_rate": 5.528413284132842e-05,
      "loss": 0.0034440584480762483,
      "step": 157600
    },
    {
      "epoch": 44.73743968208913,
      "grad_norm": 0.058147307485342026,
      "learning_rate": 5.5281294351405054e-05,
      "loss": 0.0006294047459959983,
      "step": 157610
    },
    {
      "epoch": 44.74027817201249,
      "grad_norm": 0.9640230536460876,
      "learning_rate": 5.5278455861481696e-05,
      "loss": 0.0008123567327857018,
      "step": 157620
    },
    {
      "epoch": 44.74311666193585,
      "grad_norm": 0.33940911293029785,
      "learning_rate": 5.527561737155833e-05,
      "loss": 0.001241137646138668,
      "step": 157630
    },
    {
      "epoch": 44.745955151859214,
      "grad_norm": 0.0839107409119606,
      "learning_rate": 5.5272778881634965e-05,
      "loss": 0.007611919939517975,
      "step": 157640
    },
    {
      "epoch": 44.74879364178257,
      "grad_norm": 0.48125582933425903,
      "learning_rate": 5.526994039171161e-05,
      "loss": 0.0010891607031226157,
      "step": 157650
    },
    {
      "epoch": 44.75163213170593,
      "grad_norm": 0.027461860328912735,
      "learning_rate": 5.5267101901788255e-05,
      "loss": 0.0008801359683275223,
      "step": 157660
    },
    {
      "epoch": 44.754470621629295,
      "grad_norm": 0.06395852565765381,
      "learning_rate": 5.526426341186489e-05,
      "loss": 0.00028124842792749406,
      "step": 157670
    },
    {
      "epoch": 44.75730911155265,
      "grad_norm": 0.03503318876028061,
      "learning_rate": 5.526142492194153e-05,
      "loss": 0.0009743746370077134,
      "step": 157680
    },
    {
      "epoch": 44.760147601476014,
      "grad_norm": 0.5716536045074463,
      "learning_rate": 5.5258586432018165e-05,
      "loss": 0.0009931830689311027,
      "step": 157690
    },
    {
      "epoch": 44.76298609139938,
      "grad_norm": 0.03745920583605766,
      "learning_rate": 5.5255747942094807e-05,
      "loss": 0.0009652700275182724,
      "step": 157700
    },
    {
      "epoch": 44.76582458132274,
      "grad_norm": 0.2622953951358795,
      "learning_rate": 5.5252909452171455e-05,
      "loss": 0.00228087417781353,
      "step": 157710
    },
    {
      "epoch": 44.768663071246095,
      "grad_norm": 0.055658817291259766,
      "learning_rate": 5.525007096224809e-05,
      "loss": 0.0042261935770511625,
      "step": 157720
    },
    {
      "epoch": 44.77150156116946,
      "grad_norm": 0.04795948415994644,
      "learning_rate": 5.524723247232473e-05,
      "loss": 0.0022229550406336783,
      "step": 157730
    },
    {
      "epoch": 44.77434005109282,
      "grad_norm": 0.05885598063468933,
      "learning_rate": 5.5244393982401365e-05,
      "loss": 0.0038751911371946333,
      "step": 157740
    },
    {
      "epoch": 44.77717854101618,
      "grad_norm": 0.2620353400707245,
      "learning_rate": 5.5241555492478e-05,
      "loss": 0.00743177980184555,
      "step": 157750
    },
    {
      "epoch": 44.78001703093954,
      "grad_norm": 0.30183443427085876,
      "learning_rate": 5.523871700255464e-05,
      "loss": 0.0017947282642126084,
      "step": 157760
    },
    {
      "epoch": 44.7828555208629,
      "grad_norm": 0.8360267877578735,
      "learning_rate": 5.5235878512631276e-05,
      "loss": 0.004555998742580414,
      "step": 157770
    },
    {
      "epoch": 44.78569401078626,
      "grad_norm": 1.1649465560913086,
      "learning_rate": 5.5233040022707924e-05,
      "loss": 0.006046444922685623,
      "step": 157780
    },
    {
      "epoch": 44.78853250070962,
      "grad_norm": 10.252805709838867,
      "learning_rate": 5.5230201532784565e-05,
      "loss": 0.0029329689219594,
      "step": 157790
    },
    {
      "epoch": 44.791370990632984,
      "grad_norm": 2.4858438968658447,
      "learning_rate": 5.52273630428612e-05,
      "loss": 0.0016216130927205086,
      "step": 157800
    },
    {
      "epoch": 44.79420948055635,
      "grad_norm": 0.12734635174274445,
      "learning_rate": 5.522452455293784e-05,
      "loss": 0.01193375512957573,
      "step": 157810
    },
    {
      "epoch": 44.7970479704797,
      "grad_norm": 0.10532432794570923,
      "learning_rate": 5.5221686063014476e-05,
      "loss": 0.0007083123549818992,
      "step": 157820
    },
    {
      "epoch": 44.799886460403066,
      "grad_norm": 0.0994689092040062,
      "learning_rate": 5.521884757309112e-05,
      "loss": 0.0012152742594480515,
      "step": 157830
    },
    {
      "epoch": 44.80272495032643,
      "grad_norm": 0.02299094758927822,
      "learning_rate": 5.521600908316775e-05,
      "loss": 0.0008349131792783737,
      "step": 157840
    },
    {
      "epoch": 44.805563440249784,
      "grad_norm": 0.214923694729805,
      "learning_rate": 5.52131705932444e-05,
      "loss": 0.0034560278058052064,
      "step": 157850
    },
    {
      "epoch": 44.80840193017315,
      "grad_norm": 1.6340503692626953,
      "learning_rate": 5.521033210332104e-05,
      "loss": 0.005717422813177109,
      "step": 157860
    },
    {
      "epoch": 44.81124042009651,
      "grad_norm": 0.25019407272338867,
      "learning_rate": 5.5207493613397676e-05,
      "loss": 0.0007282810285687447,
      "step": 157870
    },
    {
      "epoch": 44.814078910019866,
      "grad_norm": 12.4509859085083,
      "learning_rate": 5.520465512347431e-05,
      "loss": 0.0034079603850841523,
      "step": 157880
    },
    {
      "epoch": 44.81691739994323,
      "grad_norm": 4.388509750366211,
      "learning_rate": 5.520181663355095e-05,
      "loss": 0.005642292648553848,
      "step": 157890
    },
    {
      "epoch": 44.81975588986659,
      "grad_norm": 0.4250636696815491,
      "learning_rate": 5.5198978143627587e-05,
      "loss": 0.006513231992721557,
      "step": 157900
    },
    {
      "epoch": 44.822594379789955,
      "grad_norm": 2.885422468185425,
      "learning_rate": 5.5196139653704235e-05,
      "loss": 0.017210948467254638,
      "step": 157910
    },
    {
      "epoch": 44.82543286971331,
      "grad_norm": 0.007075225003063679,
      "learning_rate": 5.5193301163780876e-05,
      "loss": 0.007407717406749725,
      "step": 157920
    },
    {
      "epoch": 44.82827135963667,
      "grad_norm": 15.851521492004395,
      "learning_rate": 5.519046267385751e-05,
      "loss": 0.007784406840801239,
      "step": 157930
    },
    {
      "epoch": 44.831109849560036,
      "grad_norm": 0.10416088253259659,
      "learning_rate": 5.518762418393415e-05,
      "loss": 0.0037881582975387573,
      "step": 157940
    },
    {
      "epoch": 44.83394833948339,
      "grad_norm": 0.9782705903053284,
      "learning_rate": 5.518478569401079e-05,
      "loss": 0.0033412180840969087,
      "step": 157950
    },
    {
      "epoch": 44.836786829406755,
      "grad_norm": 3.805529832839966,
      "learning_rate": 5.518194720408743e-05,
      "loss": 0.002774055115878582,
      "step": 157960
    },
    {
      "epoch": 44.83962531933012,
      "grad_norm": 10.517626762390137,
      "learning_rate": 5.517910871416406e-05,
      "loss": 0.0051384449005126955,
      "step": 157970
    },
    {
      "epoch": 44.84246380925348,
      "grad_norm": 0.4152916371822357,
      "learning_rate": 5.517627022424071e-05,
      "loss": 0.000879182480275631,
      "step": 157980
    },
    {
      "epoch": 44.845302299176836,
      "grad_norm": 0.04031709209084511,
      "learning_rate": 5.517343173431735e-05,
      "loss": 0.010970105230808259,
      "step": 157990
    },
    {
      "epoch": 44.8481407891002,
      "grad_norm": 1.6094872951507568,
      "learning_rate": 5.517059324439399e-05,
      "loss": 0.0008959716185927391,
      "step": 158000
    },
    {
      "epoch": 44.8481407891002,
      "eval_accuracy": 0.9748203726076174,
      "eval_loss": 0.08915136009454727,
      "eval_runtime": 32.0232,
      "eval_samples_per_second": 491.112,
      "eval_steps_per_second": 7.682,
      "step": 158000
    },
    {
      "epoch": 44.85097927902356,
      "grad_norm": 0.3079792261123657,
      "learning_rate": 5.516775475447062e-05,
      "loss": 0.00936158150434494,
      "step": 158010
    },
    {
      "epoch": 44.85381776894692,
      "grad_norm": 0.08267982304096222,
      "learning_rate": 5.516491626454726e-05,
      "loss": 0.0007191915065050125,
      "step": 158020
    },
    {
      "epoch": 44.85665625887028,
      "grad_norm": 1.1622956991195679,
      "learning_rate": 5.51620777746239e-05,
      "loss": 0.0011630786582827568,
      "step": 158030
    },
    {
      "epoch": 44.859494748793644,
      "grad_norm": 0.25272417068481445,
      "learning_rate": 5.515923928470054e-05,
      "loss": 0.002562100812792778,
      "step": 158040
    },
    {
      "epoch": 44.862333238717,
      "grad_norm": 0.599229633808136,
      "learning_rate": 5.515640079477719e-05,
      "loss": 0.0006453372538089752,
      "step": 158050
    },
    {
      "epoch": 44.86517172864036,
      "grad_norm": 0.8068385720252991,
      "learning_rate": 5.515356230485382e-05,
      "loss": 0.004246723651885986,
      "step": 158060
    },
    {
      "epoch": 44.868010218563725,
      "grad_norm": 1.126819133758545,
      "learning_rate": 5.515072381493046e-05,
      "loss": 0.000901707261800766,
      "step": 158070
    },
    {
      "epoch": 44.87084870848709,
      "grad_norm": 0.11892633140087128,
      "learning_rate": 5.51478853250071e-05,
      "loss": 0.001006169244647026,
      "step": 158080
    },
    {
      "epoch": 44.873687198410444,
      "grad_norm": 0.028892138972878456,
      "learning_rate": 5.514504683508374e-05,
      "loss": 0.0023022569715976716,
      "step": 158090
    },
    {
      "epoch": 44.87652568833381,
      "grad_norm": 0.3775464594364166,
      "learning_rate": 5.5142208345160373e-05,
      "loss": 0.004564998298883438,
      "step": 158100
    },
    {
      "epoch": 44.87936417825717,
      "grad_norm": 0.09537631273269653,
      "learning_rate": 5.513936985523702e-05,
      "loss": 0.0006770772859454155,
      "step": 158110
    },
    {
      "epoch": 44.882202668180526,
      "grad_norm": 0.13109567761421204,
      "learning_rate": 5.5136531365313656e-05,
      "loss": 0.0017221098765730857,
      "step": 158120
    },
    {
      "epoch": 44.88504115810389,
      "grad_norm": 0.069832943379879,
      "learning_rate": 5.51336928753903e-05,
      "loss": 0.0010123152285814285,
      "step": 158130
    },
    {
      "epoch": 44.88787964802725,
      "grad_norm": 0.07793183624744415,
      "learning_rate": 5.513085438546693e-05,
      "loss": 0.000627504475414753,
      "step": 158140
    },
    {
      "epoch": 44.89071813795061,
      "grad_norm": 1.6821398735046387,
      "learning_rate": 5.5128015895543574e-05,
      "loss": 0.0018203791230916976,
      "step": 158150
    },
    {
      "epoch": 44.89355662787397,
      "grad_norm": 0.17002630233764648,
      "learning_rate": 5.512517740562021e-05,
      "loss": 0.0011295968666672706,
      "step": 158160
    },
    {
      "epoch": 44.89639511779733,
      "grad_norm": 0.1596800684928894,
      "learning_rate": 5.512233891569685e-05,
      "loss": 0.00046041086316108706,
      "step": 158170
    },
    {
      "epoch": 44.899233607720696,
      "grad_norm": 0.07032977044582367,
      "learning_rate": 5.51195004257735e-05,
      "loss": 0.0019820904359221458,
      "step": 158180
    },
    {
      "epoch": 44.90207209764405,
      "grad_norm": 0.061592888087034225,
      "learning_rate": 5.511666193585013e-05,
      "loss": 0.0005488898605108261,
      "step": 158190
    },
    {
      "epoch": 44.904910587567414,
      "grad_norm": 0.08529050648212433,
      "learning_rate": 5.5113823445926774e-05,
      "loss": 0.0008605325594544411,
      "step": 158200
    },
    {
      "epoch": 44.90774907749078,
      "grad_norm": 0.082225501537323,
      "learning_rate": 5.511098495600341e-05,
      "loss": 0.0005826020613312721,
      "step": 158210
    },
    {
      "epoch": 44.91058756741413,
      "grad_norm": 0.011607063002884388,
      "learning_rate": 5.510814646608004e-05,
      "loss": 0.0008266100659966469,
      "step": 158220
    },
    {
      "epoch": 44.913426057337496,
      "grad_norm": 5.987351894378662,
      "learning_rate": 5.5105307976156684e-05,
      "loss": 0.001955020986497402,
      "step": 158230
    },
    {
      "epoch": 44.91626454726086,
      "grad_norm": 0.046347372233867645,
      "learning_rate": 5.510246948623332e-05,
      "loss": 0.0015226827934384346,
      "step": 158240
    },
    {
      "epoch": 44.919103037184215,
      "grad_norm": 0.2993282377719879,
      "learning_rate": 5.509963099630997e-05,
      "loss": 0.0010629979893565177,
      "step": 158250
    },
    {
      "epoch": 44.92194152710758,
      "grad_norm": 0.07779286801815033,
      "learning_rate": 5.509679250638661e-05,
      "loss": 0.0006899459287524224,
      "step": 158260
    },
    {
      "epoch": 44.92478001703094,
      "grad_norm": 0.19331727921962738,
      "learning_rate": 5.509395401646324e-05,
      "loss": 0.0005407802760601044,
      "step": 158270
    },
    {
      "epoch": 44.9276185069543,
      "grad_norm": 0.16289174556732178,
      "learning_rate": 5.5091115526539884e-05,
      "loss": 0.0012634962797164917,
      "step": 158280
    },
    {
      "epoch": 44.93045699687766,
      "grad_norm": 0.16641496121883392,
      "learning_rate": 5.508827703661652e-05,
      "loss": 0.0025248169898986817,
      "step": 158290
    },
    {
      "epoch": 44.93329548680102,
      "grad_norm": 0.0467965193092823,
      "learning_rate": 5.508543854669316e-05,
      "loss": 0.0012122685089707374,
      "step": 158300
    },
    {
      "epoch": 44.936133976724385,
      "grad_norm": 0.1455937623977661,
      "learning_rate": 5.508260005676981e-05,
      "loss": 0.001590898260474205,
      "step": 158310
    },
    {
      "epoch": 44.93897246664774,
      "grad_norm": 2.0370683670043945,
      "learning_rate": 5.507976156684644e-05,
      "loss": 0.004656960070133209,
      "step": 158320
    },
    {
      "epoch": 44.9418109565711,
      "grad_norm": 2.585007667541504,
      "learning_rate": 5.5076923076923084e-05,
      "loss": 0.00512169897556305,
      "step": 158330
    },
    {
      "epoch": 44.944649446494466,
      "grad_norm": 0.9029721021652222,
      "learning_rate": 5.507408458699972e-05,
      "loss": 0.0007221681997179985,
      "step": 158340
    },
    {
      "epoch": 44.94748793641783,
      "grad_norm": 0.4834648370742798,
      "learning_rate": 5.5071246097076354e-05,
      "loss": 0.0022313324734568596,
      "step": 158350
    },
    {
      "epoch": 44.950326426341185,
      "grad_norm": 0.16262923181056976,
      "learning_rate": 5.5068407607152995e-05,
      "loss": 0.003935231268405915,
      "step": 158360
    },
    {
      "epoch": 44.95316491626455,
      "grad_norm": 2.9060633182525635,
      "learning_rate": 5.506556911722963e-05,
      "loss": 0.005235027521848679,
      "step": 158370
    },
    {
      "epoch": 44.95600340618791,
      "grad_norm": 0.5851252675056458,
      "learning_rate": 5.506273062730628e-05,
      "loss": 0.0008652579039335251,
      "step": 158380
    },
    {
      "epoch": 44.95884189611127,
      "grad_norm": 0.10988026857376099,
      "learning_rate": 5.505989213738292e-05,
      "loss": 0.007165488600730896,
      "step": 158390
    },
    {
      "epoch": 44.96168038603463,
      "grad_norm": 0.080855593085289,
      "learning_rate": 5.5057053647459554e-05,
      "loss": 0.0013157708570361137,
      "step": 158400
    },
    {
      "epoch": 44.96451887595799,
      "grad_norm": 1.039370059967041,
      "learning_rate": 5.5054215157536195e-05,
      "loss": 0.015054267644882203,
      "step": 158410
    },
    {
      "epoch": 44.96735736588135,
      "grad_norm": 0.9792690277099609,
      "learning_rate": 5.505137666761283e-05,
      "loss": 0.007184583693742752,
      "step": 158420
    },
    {
      "epoch": 44.97019585580471,
      "grad_norm": 0.21361017227172852,
      "learning_rate": 5.504853817768947e-05,
      "loss": 0.007206868380308151,
      "step": 158430
    },
    {
      "epoch": 44.973034345728074,
      "grad_norm": 3.469264507293701,
      "learning_rate": 5.5045699687766106e-05,
      "loss": 0.004988017305731773,
      "step": 158440
    },
    {
      "epoch": 44.97587283565144,
      "grad_norm": 0.06772604584693909,
      "learning_rate": 5.5042861197842754e-05,
      "loss": 0.000697045587003231,
      "step": 158450
    },
    {
      "epoch": 44.97871132557479,
      "grad_norm": 0.10127173364162445,
      "learning_rate": 5.5040022707919395e-05,
      "loss": 0.0003275051712989807,
      "step": 158460
    },
    {
      "epoch": 44.981549815498155,
      "grad_norm": 0.19566664099693298,
      "learning_rate": 5.503718421799603e-05,
      "loss": 0.001727229543030262,
      "step": 158470
    },
    {
      "epoch": 44.98438830542152,
      "grad_norm": 0.12345828860998154,
      "learning_rate": 5.5034345728072664e-05,
      "loss": 0.0004245707765221596,
      "step": 158480
    },
    {
      "epoch": 44.987226795344874,
      "grad_norm": 0.045536983758211136,
      "learning_rate": 5.5031507238149306e-05,
      "loss": 0.0016574151813983917,
      "step": 158490
    },
    {
      "epoch": 44.99006528526824,
      "grad_norm": 1.9903802871704102,
      "learning_rate": 5.502866874822594e-05,
      "loss": 0.0023794559761881827,
      "step": 158500
    },
    {
      "epoch": 44.99006528526824,
      "eval_accuracy": 0.9745660329369873,
      "eval_loss": 0.09136876463890076,
      "eval_runtime": 31.8606,
      "eval_samples_per_second": 493.619,
      "eval_steps_per_second": 7.721,
      "step": 158500
    },
    {
      "epoch": 44.9929037751916,
      "grad_norm": 0.04272126778960228,
      "learning_rate": 5.502583025830259e-05,
      "loss": 0.0008862776681780815,
      "step": 158510
    },
    {
      "epoch": 44.995742265114956,
      "grad_norm": 0.8683083653450012,
      "learning_rate": 5.502299176837923e-05,
      "loss": 0.005330255255103111,
      "step": 158520
    },
    {
      "epoch": 44.99858075503832,
      "grad_norm": 0.060387711971998215,
      "learning_rate": 5.5020153278455864e-05,
      "loss": 0.001135408878326416,
      "step": 158530
    },
    {
      "epoch": 45.00141924496168,
      "grad_norm": 0.25611069798469543,
      "learning_rate": 5.5017314788532506e-05,
      "loss": 0.006997285783290863,
      "step": 158540
    },
    {
      "epoch": 45.004257734885044,
      "grad_norm": 0.023752298206090927,
      "learning_rate": 5.501447629860914e-05,
      "loss": 0.002735184133052826,
      "step": 158550
    },
    {
      "epoch": 45.0070962248084,
      "grad_norm": 0.012567800469696522,
      "learning_rate": 5.501163780868578e-05,
      "loss": 0.0006977530196309089,
      "step": 158560
    },
    {
      "epoch": 45.00993471473176,
      "grad_norm": 0.1428321897983551,
      "learning_rate": 5.5008799318762416e-05,
      "loss": 0.00831996276974678,
      "step": 158570
    },
    {
      "epoch": 45.012773204655126,
      "grad_norm": 0.35385504364967346,
      "learning_rate": 5.5005960828839065e-05,
      "loss": 0.002828647755086422,
      "step": 158580
    },
    {
      "epoch": 45.01561169457848,
      "grad_norm": 0.08775291591882706,
      "learning_rate": 5.50031223389157e-05,
      "loss": 0.0012270327657461166,
      "step": 158590
    },
    {
      "epoch": 45.018450184501845,
      "grad_norm": 0.11947706341743469,
      "learning_rate": 5.500028384899234e-05,
      "loss": 0.00974058136343956,
      "step": 158600
    },
    {
      "epoch": 45.02128867442521,
      "grad_norm": 5.909399032592773,
      "learning_rate": 5.4997445359068975e-05,
      "loss": 0.004210643097758293,
      "step": 158610
    },
    {
      "epoch": 45.02412716434856,
      "grad_norm": Infinity,
      "learning_rate": 5.4994606869145617e-05,
      "loss": 0.004809434339404106,
      "step": 158620
    },
    {
      "epoch": 45.026965654271926,
      "grad_norm": 1.364916443824768,
      "learning_rate": 5.4992052228214596e-05,
      "loss": 0.002781349606812,
      "step": 158630
    },
    {
      "epoch": 45.02980414419529,
      "grad_norm": 4.440274715423584,
      "learning_rate": 5.498921373829123e-05,
      "loss": 0.002674400433897972,
      "step": 158640
    },
    {
      "epoch": 45.03264263411865,
      "grad_norm": 1.0811306238174438,
      "learning_rate": 5.498637524836787e-05,
      "loss": 0.0009710287675261497,
      "step": 158650
    },
    {
      "epoch": 45.03548112404201,
      "grad_norm": 0.13827070593833923,
      "learning_rate": 5.498353675844451e-05,
      "loss": 0.0006848923861980438,
      "step": 158660
    },
    {
      "epoch": 45.03831961396537,
      "grad_norm": 0.16593563556671143,
      "learning_rate": 5.498069826852115e-05,
      "loss": 0.0021458635106682777,
      "step": 158670
    },
    {
      "epoch": 45.04115810388873,
      "grad_norm": 0.13837239146232605,
      "learning_rate": 5.497785977859778e-05,
      "loss": 0.0011216219514608382,
      "step": 158680
    },
    {
      "epoch": 45.04399659381209,
      "grad_norm": 0.26147106289863586,
      "learning_rate": 5.497502128867443e-05,
      "loss": 0.007856369018554688,
      "step": 158690
    },
    {
      "epoch": 45.04683508373545,
      "grad_norm": 0.3573833405971527,
      "learning_rate": 5.497218279875107e-05,
      "loss": 0.0011279525235295296,
      "step": 158700
    },
    {
      "epoch": 45.049673573658815,
      "grad_norm": 0.31289637088775635,
      "learning_rate": 5.496934430882771e-05,
      "loss": 0.002954649366438389,
      "step": 158710
    },
    {
      "epoch": 45.05251206358217,
      "grad_norm": 0.02787191979587078,
      "learning_rate": 5.496650581890435e-05,
      "loss": 0.0023475375026464464,
      "step": 158720
    },
    {
      "epoch": 45.055350553505534,
      "grad_norm": 9.228250503540039,
      "learning_rate": 5.496366732898098e-05,
      "loss": 0.006078017503023147,
      "step": 158730
    },
    {
      "epoch": 45.0581890434289,
      "grad_norm": 0.2058650553226471,
      "learning_rate": 5.496082883905762e-05,
      "loss": 0.0012133600190281867,
      "step": 158740
    },
    {
      "epoch": 45.06102753335226,
      "grad_norm": 0.02194630540907383,
      "learning_rate": 5.495799034913426e-05,
      "loss": 0.003739486634731293,
      "step": 158750
    },
    {
      "epoch": 45.063866023275615,
      "grad_norm": 3.892699956893921,
      "learning_rate": 5.495515185921091e-05,
      "loss": 0.002123177982866764,
      "step": 158760
    },
    {
      "epoch": 45.06670451319898,
      "grad_norm": 6.4801177978515625,
      "learning_rate": 5.495231336928754e-05,
      "loss": 0.001968359388411045,
      "step": 158770
    },
    {
      "epoch": 45.06954300312234,
      "grad_norm": 0.29151490330696106,
      "learning_rate": 5.494947487936418e-05,
      "loss": 0.0038392387330532076,
      "step": 158780
    },
    {
      "epoch": 45.0723814930457,
      "grad_norm": 0.03870563581585884,
      "learning_rate": 5.494663638944082e-05,
      "loss": 0.005506311729550362,
      "step": 158790
    },
    {
      "epoch": 45.07521998296906,
      "grad_norm": 0.06824429333209991,
      "learning_rate": 5.494379789951746e-05,
      "loss": 0.0007027361541986465,
      "step": 158800
    },
    {
      "epoch": 45.07805847289242,
      "grad_norm": 0.05247722193598747,
      "learning_rate": 5.4940959409594094e-05,
      "loss": 0.014284053444862365,
      "step": 158810
    },
    {
      "epoch": 45.080896962815785,
      "grad_norm": 12.008231163024902,
      "learning_rate": 5.4938120919670735e-05,
      "loss": 0.0030184928327798843,
      "step": 158820
    },
    {
      "epoch": 45.08373545273914,
      "grad_norm": 0.0152906384319067,
      "learning_rate": 5.493528242974738e-05,
      "loss": 0.0033500224351882936,
      "step": 158830
    },
    {
      "epoch": 45.086573942662504,
      "grad_norm": 0.6462904810905457,
      "learning_rate": 5.493244393982402e-05,
      "loss": 0.0014458335936069489,
      "step": 158840
    },
    {
      "epoch": 45.08941243258587,
      "grad_norm": 0.020498517900705338,
      "learning_rate": 5.492960544990066e-05,
      "loss": 0.005823376774787903,
      "step": 158850
    },
    {
      "epoch": 45.09225092250922,
      "grad_norm": 1.6525417566299438,
      "learning_rate": 5.4926766959977294e-05,
      "loss": 0.001875150017440319,
      "step": 158860
    },
    {
      "epoch": 45.095089412432586,
      "grad_norm": 15.990608215332031,
      "learning_rate": 5.492392847005393e-05,
      "loss": 0.013257858157157899,
      "step": 158870
    },
    {
      "epoch": 45.09792790235595,
      "grad_norm": 1.0381603240966797,
      "learning_rate": 5.492108998013057e-05,
      "loss": 0.0030856169760227203,
      "step": 158880
    },
    {
      "epoch": 45.100766392279304,
      "grad_norm": 14.545747756958008,
      "learning_rate": 5.491825149020722e-05,
      "loss": 0.010082698613405227,
      "step": 158890
    },
    {
      "epoch": 45.10360488220267,
      "grad_norm": 0.15826590359210968,
      "learning_rate": 5.491541300028385e-05,
      "loss": 0.000534714013338089,
      "step": 158900
    },
    {
      "epoch": 45.10644337212603,
      "grad_norm": 2.7404963970184326,
      "learning_rate": 5.4912574510360494e-05,
      "loss": 0.0007791947573423386,
      "step": 158910
    },
    {
      "epoch": 45.10928186204939,
      "grad_norm": 0.6924488544464111,
      "learning_rate": 5.490973602043713e-05,
      "loss": 0.0006656453013420105,
      "step": 158920
    },
    {
      "epoch": 45.11212035197275,
      "grad_norm": 0.05508305877447128,
      "learning_rate": 5.490689753051377e-05,
      "loss": 0.002227383479475975,
      "step": 158930
    },
    {
      "epoch": 45.11495884189611,
      "grad_norm": 0.014868048951029778,
      "learning_rate": 5.4904059040590405e-05,
      "loss": 0.0007092274725437164,
      "step": 158940
    },
    {
      "epoch": 45.117797331819474,
      "grad_norm": 0.14755873382091522,
      "learning_rate": 5.4901220550667046e-05,
      "loss": 0.0010996513068675996,
      "step": 158950
    },
    {
      "epoch": 45.12063582174283,
      "grad_norm": 0.08222337812185287,
      "learning_rate": 5.4898382060743694e-05,
      "loss": 0.001336578093469143,
      "step": 158960
    },
    {
      "epoch": 45.12347431166619,
      "grad_norm": 0.09603218734264374,
      "learning_rate": 5.489554357082033e-05,
      "loss": 0.000968967005610466,
      "step": 158970
    },
    {
      "epoch": 45.126312801589556,
      "grad_norm": 7.601822376251221,
      "learning_rate": 5.489270508089697e-05,
      "loss": 0.001595294661819935,
      "step": 158980
    },
    {
      "epoch": 45.12915129151291,
      "grad_norm": 0.008883538655936718,
      "learning_rate": 5.4889866590973605e-05,
      "loss": 0.0028997644782066344,
      "step": 158990
    },
    {
      "epoch": 45.131989781436275,
      "grad_norm": 0.2879927158355713,
      "learning_rate": 5.488702810105024e-05,
      "loss": 0.0030341632664203644,
      "step": 159000
    },
    {
      "epoch": 45.131989781436275,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.09030680358409882,
      "eval_runtime": 32.3828,
      "eval_samples_per_second": 485.66,
      "eval_steps_per_second": 7.597,
      "step": 159000
    },
    {
      "epoch": 45.13482827135964,
      "grad_norm": 7.417815685272217,
      "learning_rate": 5.488418961112688e-05,
      "loss": 0.0016166044399142266,
      "step": 159010
    },
    {
      "epoch": 45.137666761283,
      "grad_norm": 4.801940441131592,
      "learning_rate": 5.4881351121203515e-05,
      "loss": 0.0023504666984081267,
      "step": 159020
    },
    {
      "epoch": 45.140505251206356,
      "grad_norm": 1.0077598094940186,
      "learning_rate": 5.487851263128016e-05,
      "loss": 0.000833679549396038,
      "step": 159030
    },
    {
      "epoch": 45.14334374112972,
      "grad_norm": 1.24357271194458,
      "learning_rate": 5.4875674141356805e-05,
      "loss": 0.001734432578086853,
      "step": 159040
    },
    {
      "epoch": 45.14618223105308,
      "grad_norm": 0.19335401058197021,
      "learning_rate": 5.487283565143344e-05,
      "loss": 0.007602711021900177,
      "step": 159050
    },
    {
      "epoch": 45.14902072097644,
      "grad_norm": 0.04252135381102562,
      "learning_rate": 5.486999716151008e-05,
      "loss": 0.0013842761516571044,
      "step": 159060
    },
    {
      "epoch": 45.1518592108998,
      "grad_norm": 0.5231683850288391,
      "learning_rate": 5.4867158671586715e-05,
      "loss": 0.003377676010131836,
      "step": 159070
    },
    {
      "epoch": 45.154697700823164,
      "grad_norm": 0.04143055900931358,
      "learning_rate": 5.486432018166336e-05,
      "loss": 0.0005700143054127693,
      "step": 159080
    },
    {
      "epoch": 45.15753619074652,
      "grad_norm": 0.0542815737426281,
      "learning_rate": 5.4861481691740005e-05,
      "loss": 0.00034311339259147645,
      "step": 159090
    },
    {
      "epoch": 45.16037468066988,
      "grad_norm": 0.03018852323293686,
      "learning_rate": 5.485864320181664e-05,
      "loss": 0.0005938779562711715,
      "step": 159100
    },
    {
      "epoch": 45.163213170593245,
      "grad_norm": 0.543493390083313,
      "learning_rate": 5.4855804711893274e-05,
      "loss": 0.0012732425704598427,
      "step": 159110
    },
    {
      "epoch": 45.16605166051661,
      "grad_norm": 2.836855173110962,
      "learning_rate": 5.4852966221969915e-05,
      "loss": 0.0009161999449133873,
      "step": 159120
    },
    {
      "epoch": 45.168890150439964,
      "grad_norm": 0.06751839816570282,
      "learning_rate": 5.485012773204655e-05,
      "loss": 0.0008341988548636436,
      "step": 159130
    },
    {
      "epoch": 45.17172864036333,
      "grad_norm": 0.06446419656276703,
      "learning_rate": 5.484728924212319e-05,
      "loss": 0.0013551948592066764,
      "step": 159140
    },
    {
      "epoch": 45.17456713028669,
      "grad_norm": 0.15203124284744263,
      "learning_rate": 5.4844450752199826e-05,
      "loss": 0.0006399227306246758,
      "step": 159150
    },
    {
      "epoch": 45.177405620210045,
      "grad_norm": 0.24372035264968872,
      "learning_rate": 5.4841612262276474e-05,
      "loss": 0.0010714277625083924,
      "step": 159160
    },
    {
      "epoch": 45.18024411013341,
      "grad_norm": 2.453103542327881,
      "learning_rate": 5.4838773772353115e-05,
      "loss": 0.0009769123047590257,
      "step": 159170
    },
    {
      "epoch": 45.18308260005677,
      "grad_norm": 6.146474361419678,
      "learning_rate": 5.483593528242975e-05,
      "loss": 0.008214885741472245,
      "step": 159180
    },
    {
      "epoch": 45.185921089980134,
      "grad_norm": 2.4516446590423584,
      "learning_rate": 5.483309679250639e-05,
      "loss": 0.0070764578878879545,
      "step": 159190
    },
    {
      "epoch": 45.18875957990349,
      "grad_norm": 0.14871080219745636,
      "learning_rate": 5.4830258302583026e-05,
      "loss": 0.022754067182540895,
      "step": 159200
    },
    {
      "epoch": 45.19159806982685,
      "grad_norm": 0.13743652403354645,
      "learning_rate": 5.482741981265966e-05,
      "loss": 0.003486846014857292,
      "step": 159210
    },
    {
      "epoch": 45.194436559750216,
      "grad_norm": 0.303010493516922,
      "learning_rate": 5.48245813227363e-05,
      "loss": 0.0032066166400909423,
      "step": 159220
    },
    {
      "epoch": 45.19727504967357,
      "grad_norm": 0.9396266937255859,
      "learning_rate": 5.482174283281295e-05,
      "loss": 0.003225202113389969,
      "step": 159230
    },
    {
      "epoch": 45.200113539596934,
      "grad_norm": 0.1093941256403923,
      "learning_rate": 5.4818904342889585e-05,
      "loss": 0.0003883266821503639,
      "step": 159240
    },
    {
      "epoch": 45.2029520295203,
      "grad_norm": 0.205396369099617,
      "learning_rate": 5.4816065852966226e-05,
      "loss": 0.0008455120027065277,
      "step": 159250
    },
    {
      "epoch": 45.20579051944365,
      "grad_norm": 0.9547229409217834,
      "learning_rate": 5.481322736304286e-05,
      "loss": 0.0006294973194599151,
      "step": 159260
    },
    {
      "epoch": 45.208629009367016,
      "grad_norm": 0.44635310769081116,
      "learning_rate": 5.48103888731195e-05,
      "loss": 0.000569654256105423,
      "step": 159270
    },
    {
      "epoch": 45.21146749929038,
      "grad_norm": 0.17590799927711487,
      "learning_rate": 5.480755038319614e-05,
      "loss": 0.00045420415699481964,
      "step": 159280
    },
    {
      "epoch": 45.21430598921374,
      "grad_norm": 0.22263941168785095,
      "learning_rate": 5.4804711893272785e-05,
      "loss": 0.0020836981013417242,
      "step": 159290
    },
    {
      "epoch": 45.2171444791371,
      "grad_norm": 0.06841296702623367,
      "learning_rate": 5.4801873403349426e-05,
      "loss": 0.0012127941474318505,
      "step": 159300
    },
    {
      "epoch": 45.21998296906046,
      "grad_norm": 0.029327819123864174,
      "learning_rate": 5.479903491342606e-05,
      "loss": 0.00021926909685134887,
      "step": 159310
    },
    {
      "epoch": 45.22282145898382,
      "grad_norm": 0.014533000998198986,
      "learning_rate": 5.47961964235027e-05,
      "loss": 0.0007028359919786453,
      "step": 159320
    },
    {
      "epoch": 45.22565994890718,
      "grad_norm": 0.15540358424186707,
      "learning_rate": 5.479335793357934e-05,
      "loss": 0.0007843375205993652,
      "step": 159330
    },
    {
      "epoch": 45.22849843883054,
      "grad_norm": 0.6212250590324402,
      "learning_rate": 5.479051944365597e-05,
      "loss": 0.0020840223878622055,
      "step": 159340
    },
    {
      "epoch": 45.231336928753905,
      "grad_norm": 0.7989044189453125,
      "learning_rate": 5.478768095373261e-05,
      "loss": 0.000829460471868515,
      "step": 159350
    },
    {
      "epoch": 45.23417541867726,
      "grad_norm": 0.35492047667503357,
      "learning_rate": 5.478484246380926e-05,
      "loss": 0.0034455753862857818,
      "step": 159360
    },
    {
      "epoch": 45.23701390860062,
      "grad_norm": 0.13733632862567902,
      "learning_rate": 5.4782003973885896e-05,
      "loss": 0.000853135995566845,
      "step": 159370
    },
    {
      "epoch": 45.239852398523986,
      "grad_norm": 8.629515647888184,
      "learning_rate": 5.477916548396254e-05,
      "loss": 0.007530410587787628,
      "step": 159380
    },
    {
      "epoch": 45.24269088844735,
      "grad_norm": 0.2308046519756317,
      "learning_rate": 5.477632699403917e-05,
      "loss": 0.005640815943479538,
      "step": 159390
    },
    {
      "epoch": 45.245529378370705,
      "grad_norm": 0.09870049357414246,
      "learning_rate": 5.477348850411581e-05,
      "loss": 0.004666077718138695,
      "step": 159400
    },
    {
      "epoch": 45.24836786829407,
      "grad_norm": 0.05703624337911606,
      "learning_rate": 5.477065001419245e-05,
      "loss": 0.0029806135222315787,
      "step": 159410
    },
    {
      "epoch": 45.25120635821743,
      "grad_norm": 0.07162238657474518,
      "learning_rate": 5.4767811524269096e-05,
      "loss": 0.004788489639759063,
      "step": 159420
    },
    {
      "epoch": 45.254044848140786,
      "grad_norm": 5.795346736907959,
      "learning_rate": 5.476497303434574e-05,
      "loss": 0.0038419514894485474,
      "step": 159430
    },
    {
      "epoch": 45.25688333806415,
      "grad_norm": 4.59437370300293,
      "learning_rate": 5.476213454442237e-05,
      "loss": 0.0024315565824508667,
      "step": 159440
    },
    {
      "epoch": 45.25972182798751,
      "grad_norm": 0.03767161816358566,
      "learning_rate": 5.475929605449901e-05,
      "loss": 0.0021321330219507217,
      "step": 159450
    },
    {
      "epoch": 45.26256031791087,
      "grad_norm": 0.23922036588191986,
      "learning_rate": 5.475645756457565e-05,
      "loss": 0.0019891796633601188,
      "step": 159460
    },
    {
      "epoch": 45.26539880783423,
      "grad_norm": 0.021695628762245178,
      "learning_rate": 5.475361907465228e-05,
      "loss": 0.002495933882892132,
      "step": 159470
    },
    {
      "epoch": 45.268237297757594,
      "grad_norm": 8.530279159545898,
      "learning_rate": 5.4750780584728924e-05,
      "loss": 0.0025252345949411392,
      "step": 159480
    },
    {
      "epoch": 45.27107578768096,
      "grad_norm": 0.1892513483762741,
      "learning_rate": 5.474794209480557e-05,
      "loss": 0.004962759092450142,
      "step": 159490
    },
    {
      "epoch": 45.27391427760431,
      "grad_norm": 0.023996317759156227,
      "learning_rate": 5.4745103604882206e-05,
      "loss": 0.002082580886781216,
      "step": 159500
    },
    {
      "epoch": 45.27391427760431,
      "eval_accuracy": 0.9755833916195078,
      "eval_loss": 0.09203548729419708,
      "eval_runtime": 32.1218,
      "eval_samples_per_second": 489.604,
      "eval_steps_per_second": 7.658,
      "step": 159500
    },
    {
      "epoch": 45.276752767527675,
      "grad_norm": 0.027179552242159843,
      "learning_rate": 5.474226511495885e-05,
      "loss": 0.0035132870078086855,
      "step": 159510
    },
    {
      "epoch": 45.27959125745104,
      "grad_norm": 0.893366277217865,
      "learning_rate": 5.473942662503548e-05,
      "loss": 0.0012043889611959458,
      "step": 159520
    },
    {
      "epoch": 45.282429747374394,
      "grad_norm": 0.8222613334655762,
      "learning_rate": 5.4736588135112124e-05,
      "loss": 0.003142710030078888,
      "step": 159530
    },
    {
      "epoch": 45.28526823729776,
      "grad_norm": 0.9906347990036011,
      "learning_rate": 5.473374964518876e-05,
      "loss": 0.001996843330562115,
      "step": 159540
    },
    {
      "epoch": 45.28810672722112,
      "grad_norm": 0.6114441156387329,
      "learning_rate": 5.47309111552654e-05,
      "loss": 0.0016777245327830315,
      "step": 159550
    },
    {
      "epoch": 45.29094521714448,
      "grad_norm": 0.4522089660167694,
      "learning_rate": 5.472807266534205e-05,
      "loss": 0.003216322511434555,
      "step": 159560
    },
    {
      "epoch": 45.29378370706784,
      "grad_norm": 0.4906410276889801,
      "learning_rate": 5.472523417541868e-05,
      "loss": 0.0005746565759181977,
      "step": 159570
    },
    {
      "epoch": 45.2966221969912,
      "grad_norm": 0.15105196833610535,
      "learning_rate": 5.472239568549532e-05,
      "loss": 0.0016492562368512153,
      "step": 159580
    },
    {
      "epoch": 45.299460686914564,
      "grad_norm": 0.11974382400512695,
      "learning_rate": 5.471955719557196e-05,
      "loss": 0.0005740797147154808,
      "step": 159590
    },
    {
      "epoch": 45.30229917683792,
      "grad_norm": 0.6734367609024048,
      "learning_rate": 5.471671870564859e-05,
      "loss": 0.002903760597109795,
      "step": 159600
    },
    {
      "epoch": 45.30513766676128,
      "grad_norm": 8.564242362976074,
      "learning_rate": 5.4713880215725234e-05,
      "loss": 0.007018131017684936,
      "step": 159610
    },
    {
      "epoch": 45.307976156684646,
      "grad_norm": 0.6227974891662598,
      "learning_rate": 5.471104172580188e-05,
      "loss": 0.0009871799498796462,
      "step": 159620
    },
    {
      "epoch": 45.310814646608,
      "grad_norm": 0.04845915362238884,
      "learning_rate": 5.470820323587852e-05,
      "loss": 0.0020764444023370745,
      "step": 159630
    },
    {
      "epoch": 45.313653136531364,
      "grad_norm": 11.776887893676758,
      "learning_rate": 5.470536474595516e-05,
      "loss": 0.005326366052031517,
      "step": 159640
    },
    {
      "epoch": 45.31649162645473,
      "grad_norm": 0.05242522433400154,
      "learning_rate": 5.470252625603179e-05,
      "loss": 0.0018631642684340476,
      "step": 159650
    },
    {
      "epoch": 45.31933011637809,
      "grad_norm": 1.9630978107452393,
      "learning_rate": 5.4699687766108434e-05,
      "loss": 0.0009176427498459816,
      "step": 159660
    },
    {
      "epoch": 45.322168606301446,
      "grad_norm": 0.03910171985626221,
      "learning_rate": 5.469684927618507e-05,
      "loss": 0.0008452320471405983,
      "step": 159670
    },
    {
      "epoch": 45.32500709622481,
      "grad_norm": 0.15117180347442627,
      "learning_rate": 5.4694010786261704e-05,
      "loss": 0.008197706192731857,
      "step": 159680
    },
    {
      "epoch": 45.32784558614817,
      "grad_norm": 4.397069931030273,
      "learning_rate": 5.469117229633836e-05,
      "loss": 0.0032413676381111145,
      "step": 159690
    },
    {
      "epoch": 45.33068407607153,
      "grad_norm": 0.5860365629196167,
      "learning_rate": 5.468833380641499e-05,
      "loss": 0.0007270824164152146,
      "step": 159700
    },
    {
      "epoch": 45.33352256599489,
      "grad_norm": 0.6500213742256165,
      "learning_rate": 5.468549531649163e-05,
      "loss": 0.008513080328702927,
      "step": 159710
    },
    {
      "epoch": 45.33636105591825,
      "grad_norm": 1.2633423805236816,
      "learning_rate": 5.468265682656827e-05,
      "loss": 0.0035127781331539152,
      "step": 159720
    },
    {
      "epoch": 45.33919954584161,
      "grad_norm": 0.019626816734671593,
      "learning_rate": 5.4679818336644904e-05,
      "loss": 0.0005236783996224404,
      "step": 159730
    },
    {
      "epoch": 45.34203803576497,
      "grad_norm": 0.6326139569282532,
      "learning_rate": 5.4676979846721545e-05,
      "loss": 0.0007827391847968102,
      "step": 159740
    },
    {
      "epoch": 45.344876525688335,
      "grad_norm": 0.0590711310505867,
      "learning_rate": 5.467414135679818e-05,
      "loss": 0.0004486672580242157,
      "step": 159750
    },
    {
      "epoch": 45.3477150156117,
      "grad_norm": 0.17331865429878235,
      "learning_rate": 5.467130286687483e-05,
      "loss": 0.0013717750087380409,
      "step": 159760
    },
    {
      "epoch": 45.35055350553505,
      "grad_norm": 0.06820020079612732,
      "learning_rate": 5.466846437695147e-05,
      "loss": 0.00045934971421957016,
      "step": 159770
    },
    {
      "epoch": 45.353391995458416,
      "grad_norm": 0.39945074915885925,
      "learning_rate": 5.4665625887028104e-05,
      "loss": 0.0003818601369857788,
      "step": 159780
    },
    {
      "epoch": 45.35623048538178,
      "grad_norm": 0.12620282173156738,
      "learning_rate": 5.4662787397104745e-05,
      "loss": 0.001192593015730381,
      "step": 159790
    },
    {
      "epoch": 45.359068975305135,
      "grad_norm": 0.43111440539360046,
      "learning_rate": 5.465994890718138e-05,
      "loss": 0.00087503083050251,
      "step": 159800
    },
    {
      "epoch": 45.3619074652285,
      "grad_norm": 0.24580995738506317,
      "learning_rate": 5.4657110417258014e-05,
      "loss": 0.0008986582979559898,
      "step": 159810
    },
    {
      "epoch": 45.36474595515186,
      "grad_norm": 0.8788180351257324,
      "learning_rate": 5.465427192733467e-05,
      "loss": 0.000782516784965992,
      "step": 159820
    },
    {
      "epoch": 45.36758444507522,
      "grad_norm": 1.376896619796753,
      "learning_rate": 5.4651433437411304e-05,
      "loss": 0.004692263901233673,
      "step": 159830
    },
    {
      "epoch": 45.37042293499858,
      "grad_norm": 1.479958176612854,
      "learning_rate": 5.464859494748794e-05,
      "loss": 0.0067971058189868925,
      "step": 159840
    },
    {
      "epoch": 45.37326142492194,
      "grad_norm": 0.37699586153030396,
      "learning_rate": 5.464575645756458e-05,
      "loss": 0.007466115802526474,
      "step": 159850
    },
    {
      "epoch": 45.376099914845305,
      "grad_norm": 0.29239821434020996,
      "learning_rate": 5.4642917967641214e-05,
      "loss": 0.0010301649570465087,
      "step": 159860
    },
    {
      "epoch": 45.37893840476866,
      "grad_norm": 0.08531893789768219,
      "learning_rate": 5.4640079477717856e-05,
      "loss": 0.006689010560512543,
      "step": 159870
    },
    {
      "epoch": 45.381776894692024,
      "grad_norm": 0.42887499928474426,
      "learning_rate": 5.463724098779449e-05,
      "loss": 0.0028180574998259546,
      "step": 159880
    },
    {
      "epoch": 45.38461538461539,
      "grad_norm": 0.036703355610370636,
      "learning_rate": 5.463440249787114e-05,
      "loss": 0.0038086168467998506,
      "step": 159890
    },
    {
      "epoch": 45.38745387453874,
      "grad_norm": 0.047074250876903534,
      "learning_rate": 5.463156400794778e-05,
      "loss": 0.0005731752142310142,
      "step": 159900
    },
    {
      "epoch": 45.390292364462105,
      "grad_norm": 0.028384489938616753,
      "learning_rate": 5.4628725518024415e-05,
      "loss": 0.0017218556255102157,
      "step": 159910
    },
    {
      "epoch": 45.39313085438547,
      "grad_norm": 0.010840391740202904,
      "learning_rate": 5.4625887028101056e-05,
      "loss": 0.0014307018369436263,
      "step": 159920
    },
    {
      "epoch": 45.395969344308824,
      "grad_norm": 3.2325572967529297,
      "learning_rate": 5.462304853817769e-05,
      "loss": 0.0027173059061169624,
      "step": 159930
    },
    {
      "epoch": 45.39880783423219,
      "grad_norm": 0.4755809009075165,
      "learning_rate": 5.4620210048254325e-05,
      "loss": 0.0003724571317434311,
      "step": 159940
    },
    {
      "epoch": 45.40164632415555,
      "grad_norm": 0.0686555951833725,
      "learning_rate": 5.4617371558330967e-05,
      "loss": 0.001898372545838356,
      "step": 159950
    },
    {
      "epoch": 45.40448481407891,
      "grad_norm": 0.38064396381378174,
      "learning_rate": 5.4614533068407615e-05,
      "loss": 0.007128436118364334,
      "step": 159960
    },
    {
      "epoch": 45.40732330400227,
      "grad_norm": 13.691572189331055,
      "learning_rate": 5.461169457848425e-05,
      "loss": 0.01318945586681366,
      "step": 159970
    },
    {
      "epoch": 45.41016179392563,
      "grad_norm": 0.013193273916840553,
      "learning_rate": 5.460885608856089e-05,
      "loss": 0.0010002166032791137,
      "step": 159980
    },
    {
      "epoch": 45.413000283848994,
      "grad_norm": 0.37461698055267334,
      "learning_rate": 5.4606017598637525e-05,
      "loss": 0.0003531694412231445,
      "step": 159990
    },
    {
      "epoch": 45.41583877377235,
      "grad_norm": 0.11720555275678635,
      "learning_rate": 5.460317910871417e-05,
      "loss": 0.0019867658615112306,
      "step": 160000
    },
    {
      "epoch": 45.41583877377235,
      "eval_accuracy": 0.974883957525275,
      "eval_loss": 0.0900978073477745,
      "eval_runtime": 32.6263,
      "eval_samples_per_second": 482.035,
      "eval_steps_per_second": 7.54,
      "step": 160000
    },
    {
      "epoch": 45.41867726369571,
      "grad_norm": 4.958388805389404,
      "learning_rate": 5.46003406187908e-05,
      "loss": 0.0009833529591560363,
      "step": 160010
    },
    {
      "epoch": 45.421515753619076,
      "grad_norm": 0.06956835836172104,
      "learning_rate": 5.459750212886745e-05,
      "loss": 0.0013391496613621712,
      "step": 160020
    },
    {
      "epoch": 45.42435424354244,
      "grad_norm": 0.026643194258213043,
      "learning_rate": 5.459466363894409e-05,
      "loss": 0.0011911196634173393,
      "step": 160030
    },
    {
      "epoch": 45.427192733465795,
      "grad_norm": 0.2273787409067154,
      "learning_rate": 5.4591825149020725e-05,
      "loss": 0.0006981289014220237,
      "step": 160040
    },
    {
      "epoch": 45.43003122338916,
      "grad_norm": 14.426826477050781,
      "learning_rate": 5.458898665909736e-05,
      "loss": 0.0055053897202014925,
      "step": 160050
    },
    {
      "epoch": 45.43286971331252,
      "grad_norm": 0.431980699300766,
      "learning_rate": 5.4586148169174e-05,
      "loss": 0.0007277240976691246,
      "step": 160060
    },
    {
      "epoch": 45.435708203235876,
      "grad_norm": 1.786981463432312,
      "learning_rate": 5.4583309679250636e-05,
      "loss": 0.0007163923233747483,
      "step": 160070
    },
    {
      "epoch": 45.43854669315924,
      "grad_norm": 0.17560170590877533,
      "learning_rate": 5.458047118932728e-05,
      "loss": 0.00046657361090183256,
      "step": 160080
    },
    {
      "epoch": 45.4413851830826,
      "grad_norm": 0.023844320327043533,
      "learning_rate": 5.4577632699403925e-05,
      "loss": 0.0002781439572572708,
      "step": 160090
    },
    {
      "epoch": 45.44422367300596,
      "grad_norm": 0.010763479396700859,
      "learning_rate": 5.457479420948056e-05,
      "loss": 0.000533713772892952,
      "step": 160100
    },
    {
      "epoch": 45.44706216292932,
      "grad_norm": 0.025526750832796097,
      "learning_rate": 5.45719557195572e-05,
      "loss": 0.0005674446001648903,
      "step": 160110
    },
    {
      "epoch": 45.44990065285268,
      "grad_norm": 6.410455226898193,
      "learning_rate": 5.4569117229633836e-05,
      "loss": 0.0033612500876188276,
      "step": 160120
    },
    {
      "epoch": 45.452739142776046,
      "grad_norm": 0.08697986602783203,
      "learning_rate": 5.456627873971048e-05,
      "loss": 0.003753416985273361,
      "step": 160130
    },
    {
      "epoch": 45.4555776326994,
      "grad_norm": 7.346334457397461,
      "learning_rate": 5.456344024978711e-05,
      "loss": 0.0026715217158198357,
      "step": 160140
    },
    {
      "epoch": 45.458416122622765,
      "grad_norm": 0.8403939604759216,
      "learning_rate": 5.4560601759863747e-05,
      "loss": 0.0032804422080516815,
      "step": 160150
    },
    {
      "epoch": 45.46125461254613,
      "grad_norm": 0.1929531842470169,
      "learning_rate": 5.45577632699404e-05,
      "loss": 0.0012107642367482185,
      "step": 160160
    },
    {
      "epoch": 45.464093102469484,
      "grad_norm": 1.530828833580017,
      "learning_rate": 5.4554924780017036e-05,
      "loss": 0.0019105054438114167,
      "step": 160170
    },
    {
      "epoch": 45.46693159239285,
      "grad_norm": 1.5098488330841064,
      "learning_rate": 5.455208629009367e-05,
      "loss": 0.002033945731818676,
      "step": 160180
    },
    {
      "epoch": 45.46977008231621,
      "grad_norm": 16.700885772705078,
      "learning_rate": 5.454924780017031e-05,
      "loss": 0.006096748635172844,
      "step": 160190
    },
    {
      "epoch": 45.472608572239565,
      "grad_norm": 0.8432513475418091,
      "learning_rate": 5.454640931024695e-05,
      "loss": 0.005097533762454987,
      "step": 160200
    },
    {
      "epoch": 45.47544706216293,
      "grad_norm": 0.8789168000221252,
      "learning_rate": 5.454357082032359e-05,
      "loss": 0.005188009887933731,
      "step": 160210
    },
    {
      "epoch": 45.47828555208629,
      "grad_norm": 0.4936543405056,
      "learning_rate": 5.4540732330400236e-05,
      "loss": 0.0010072994977235795,
      "step": 160220
    },
    {
      "epoch": 45.481124042009654,
      "grad_norm": 0.048039380460977554,
      "learning_rate": 5.453789384047687e-05,
      "loss": 0.004002508521080017,
      "step": 160230
    },
    {
      "epoch": 45.48396253193301,
      "grad_norm": 5.892132759094238,
      "learning_rate": 5.453505535055351e-05,
      "loss": 0.005026974529027939,
      "step": 160240
    },
    {
      "epoch": 45.48680102185637,
      "grad_norm": 0.048962634056806564,
      "learning_rate": 5.453221686063015e-05,
      "loss": 0.0012001559138298036,
      "step": 160250
    },
    {
      "epoch": 45.489639511779735,
      "grad_norm": 0.15728425979614258,
      "learning_rate": 5.452937837070679e-05,
      "loss": 0.0015126109123229981,
      "step": 160260
    },
    {
      "epoch": 45.49247800170309,
      "grad_norm": 0.11251980811357498,
      "learning_rate": 5.452653988078342e-05,
      "loss": 0.0026271834969520567,
      "step": 160270
    },
    {
      "epoch": 45.495316491626454,
      "grad_norm": 1.234209656715393,
      "learning_rate": 5.452370139086006e-05,
      "loss": 0.0013548396527767181,
      "step": 160280
    },
    {
      "epoch": 45.49815498154982,
      "grad_norm": 0.22638452053070068,
      "learning_rate": 5.4520862900936706e-05,
      "loss": 0.0005816129967570304,
      "step": 160290
    },
    {
      "epoch": 45.50099347147318,
      "grad_norm": 4.075749397277832,
      "learning_rate": 5.451802441101335e-05,
      "loss": 0.0029568765312433243,
      "step": 160300
    },
    {
      "epoch": 45.503831961396536,
      "grad_norm": 0.01485262718051672,
      "learning_rate": 5.451518592108998e-05,
      "loss": 0.0004826672375202179,
      "step": 160310
    },
    {
      "epoch": 45.5066704513199,
      "grad_norm": 0.04394745081663132,
      "learning_rate": 5.451234743116662e-05,
      "loss": 0.0036881044507026672,
      "step": 160320
    },
    {
      "epoch": 45.50950894124326,
      "grad_norm": 0.07913976162672043,
      "learning_rate": 5.450950894124326e-05,
      "loss": 0.003663523495197296,
      "step": 160330
    },
    {
      "epoch": 45.51234743116662,
      "grad_norm": 0.2938038408756256,
      "learning_rate": 5.45066704513199e-05,
      "loss": 0.0007372915744781495,
      "step": 160340
    },
    {
      "epoch": 45.51518592108998,
      "grad_norm": 0.2071990668773651,
      "learning_rate": 5.4503831961396533e-05,
      "loss": 0.004440349712967873,
      "step": 160350
    },
    {
      "epoch": 45.51802441101334,
      "grad_norm": 1.4186187982559204,
      "learning_rate": 5.450099347147318e-05,
      "loss": 0.00868437960743904,
      "step": 160360
    },
    {
      "epoch": 45.5208629009367,
      "grad_norm": 0.02248881570994854,
      "learning_rate": 5.449815498154982e-05,
      "loss": 0.001902194507420063,
      "step": 160370
    },
    {
      "epoch": 45.52370139086006,
      "grad_norm": 0.536666750907898,
      "learning_rate": 5.449531649162646e-05,
      "loss": 0.0017030799761414527,
      "step": 160380
    },
    {
      "epoch": 45.526539880783425,
      "grad_norm": 0.7999932169914246,
      "learning_rate": 5.44924780017031e-05,
      "loss": 0.008185773342847823,
      "step": 160390
    },
    {
      "epoch": 45.52937837070679,
      "grad_norm": 0.053994931280612946,
      "learning_rate": 5.4489639511779734e-05,
      "loss": 0.004301007837057114,
      "step": 160400
    },
    {
      "epoch": 45.53221686063014,
      "grad_norm": 0.11849386990070343,
      "learning_rate": 5.448680102185637e-05,
      "loss": 0.007906029373407364,
      "step": 160410
    },
    {
      "epoch": 45.535055350553506,
      "grad_norm": 0.08292537182569504,
      "learning_rate": 5.4483962531933016e-05,
      "loss": 0.013807235658168793,
      "step": 160420
    },
    {
      "epoch": 45.53789384047687,
      "grad_norm": 0.18553201854228973,
      "learning_rate": 5.448112404200966e-05,
      "loss": 0.0031265586614608763,
      "step": 160430
    },
    {
      "epoch": 45.540732330400225,
      "grad_norm": 7.04033899307251,
      "learning_rate": 5.447828555208629e-05,
      "loss": 0.010655637830495834,
      "step": 160440
    },
    {
      "epoch": 45.54357082032359,
      "grad_norm": 0.03732825815677643,
      "learning_rate": 5.4475447062162934e-05,
      "loss": 0.004703976958990097,
      "step": 160450
    },
    {
      "epoch": 45.54640931024695,
      "grad_norm": 0.04775502160191536,
      "learning_rate": 5.447260857223957e-05,
      "loss": 0.012050092220306396,
      "step": 160460
    },
    {
      "epoch": 45.549247800170306,
      "grad_norm": 0.3837871849536896,
      "learning_rate": 5.446977008231621e-05,
      "loss": 0.004815547168254853,
      "step": 160470
    },
    {
      "epoch": 45.55208629009367,
      "grad_norm": 0.1937353014945984,
      "learning_rate": 5.4466931592392844e-05,
      "loss": 0.001340416818857193,
      "step": 160480
    },
    {
      "epoch": 45.55492478001703,
      "grad_norm": 0.09393924474716187,
      "learning_rate": 5.446409310246949e-05,
      "loss": 0.004818068444728851,
      "step": 160490
    },
    {
      "epoch": 45.557763269940395,
      "grad_norm": 0.2552574574947357,
      "learning_rate": 5.4461254612546134e-05,
      "loss": 0.003326507657766342,
      "step": 160500
    },
    {
      "epoch": 45.557763269940395,
      "eval_accuracy": 0.9755833916195078,
      "eval_loss": 0.08832064270973206,
      "eval_runtime": 32.4717,
      "eval_samples_per_second": 484.329,
      "eval_steps_per_second": 7.576,
      "step": 160500
    },
    {
      "epoch": 45.56060175986375,
      "grad_norm": 0.15256895124912262,
      "learning_rate": 5.445841612262277e-05,
      "loss": 0.004682324081659317,
      "step": 160510
    },
    {
      "epoch": 45.563440249787114,
      "grad_norm": 1.1362718343734741,
      "learning_rate": 5.44555776326994e-05,
      "loss": 0.006822779029607773,
      "step": 160520
    },
    {
      "epoch": 45.56627873971048,
      "grad_norm": 0.19589383900165558,
      "learning_rate": 5.4452739142776044e-05,
      "loss": 0.003481447696685791,
      "step": 160530
    },
    {
      "epoch": 45.56911722963383,
      "grad_norm": 2.7340099811553955,
      "learning_rate": 5.444990065285268e-05,
      "loss": 0.0034966353327035905,
      "step": 160540
    },
    {
      "epoch": 45.571955719557195,
      "grad_norm": 0.1126856580376625,
      "learning_rate": 5.444706216292932e-05,
      "loss": 0.012968908250331878,
      "step": 160550
    },
    {
      "epoch": 45.57479420948056,
      "grad_norm": 0.2477794587612152,
      "learning_rate": 5.444422367300597e-05,
      "loss": 0.0015185700729489326,
      "step": 160560
    },
    {
      "epoch": 45.577632699403914,
      "grad_norm": 0.16882750391960144,
      "learning_rate": 5.44413851830826e-05,
      "loss": 0.004273098334670067,
      "step": 160570
    },
    {
      "epoch": 45.58047118932728,
      "grad_norm": 0.14559344947338104,
      "learning_rate": 5.4438546693159244e-05,
      "loss": 0.0012252137064933776,
      "step": 160580
    },
    {
      "epoch": 45.58330967925064,
      "grad_norm": 0.043425749987363815,
      "learning_rate": 5.443570820323588e-05,
      "loss": 0.0019860824570059775,
      "step": 160590
    },
    {
      "epoch": 45.586148169174,
      "grad_norm": 0.03831684961915016,
      "learning_rate": 5.443286971331252e-05,
      "loss": 0.0009444382041692734,
      "step": 160600
    },
    {
      "epoch": 45.58898665909736,
      "grad_norm": 0.7685577869415283,
      "learning_rate": 5.4430031223389155e-05,
      "loss": 0.01020767241716385,
      "step": 160610
    },
    {
      "epoch": 45.59182514902072,
      "grad_norm": 0.2840322256088257,
      "learning_rate": 5.44271927334658e-05,
      "loss": 0.0070201151072978975,
      "step": 160620
    },
    {
      "epoch": 45.594663638944084,
      "grad_norm": 2.4745922088623047,
      "learning_rate": 5.4424354243542444e-05,
      "loss": 0.003092491067945957,
      "step": 160630
    },
    {
      "epoch": 45.59750212886744,
      "grad_norm": 5.53104829788208,
      "learning_rate": 5.442151575361908e-05,
      "loss": 0.011387206614017487,
      "step": 160640
    },
    {
      "epoch": 45.6003406187908,
      "grad_norm": 0.024465475231409073,
      "learning_rate": 5.4418677263695714e-05,
      "loss": 0.0013364035636186599,
      "step": 160650
    },
    {
      "epoch": 45.603179108714166,
      "grad_norm": 0.01895463839173317,
      "learning_rate": 5.4415838773772355e-05,
      "loss": 0.0007305957376956939,
      "step": 160660
    },
    {
      "epoch": 45.60601759863752,
      "grad_norm": 0.32076385617256165,
      "learning_rate": 5.441300028384899e-05,
      "loss": 0.002248106338083744,
      "step": 160670
    },
    {
      "epoch": 45.608856088560884,
      "grad_norm": 3.1270592212677,
      "learning_rate": 5.441016179392563e-05,
      "loss": 0.0029866548255085947,
      "step": 160680
    },
    {
      "epoch": 45.61169457848425,
      "grad_norm": 0.11881868541240692,
      "learning_rate": 5.440732330400228e-05,
      "loss": 0.006914189457893372,
      "step": 160690
    },
    {
      "epoch": 45.61453306840761,
      "grad_norm": 0.057857874780893326,
      "learning_rate": 5.4404484814078914e-05,
      "loss": 0.004651518911123276,
      "step": 160700
    },
    {
      "epoch": 45.617371558330966,
      "grad_norm": 0.28851884603500366,
      "learning_rate": 5.4401646324155555e-05,
      "loss": 0.017529812455177308,
      "step": 160710
    },
    {
      "epoch": 45.62021004825433,
      "grad_norm": 0.1574196070432663,
      "learning_rate": 5.439880783423219e-05,
      "loss": 0.002354937419295311,
      "step": 160720
    },
    {
      "epoch": 45.62304853817769,
      "grad_norm": 1.5710221529006958,
      "learning_rate": 5.439596934430883e-05,
      "loss": 0.0017747893929481507,
      "step": 160730
    },
    {
      "epoch": 45.62588702810105,
      "grad_norm": 0.05999365076422691,
      "learning_rate": 5.4393130854385466e-05,
      "loss": 0.003612013906240463,
      "step": 160740
    },
    {
      "epoch": 45.62872551802441,
      "grad_norm": 0.3694084584712982,
      "learning_rate": 5.4390292364462114e-05,
      "loss": 0.0051765207201242445,
      "step": 160750
    },
    {
      "epoch": 45.63156400794777,
      "grad_norm": 1.15484619140625,
      "learning_rate": 5.438745387453875e-05,
      "loss": 0.0035917483270168304,
      "step": 160760
    },
    {
      "epoch": 45.634402497871136,
      "grad_norm": 0.17776988446712494,
      "learning_rate": 5.438461538461539e-05,
      "loss": 0.0012994332239031792,
      "step": 160770
    },
    {
      "epoch": 45.63724098779449,
      "grad_norm": 0.09033899754285812,
      "learning_rate": 5.4381776894692024e-05,
      "loss": 0.0009247826412320137,
      "step": 160780
    },
    {
      "epoch": 45.640079477717855,
      "grad_norm": 0.47340404987335205,
      "learning_rate": 5.4378938404768666e-05,
      "loss": 0.0009023766964673996,
      "step": 160790
    },
    {
      "epoch": 45.64291796764122,
      "grad_norm": 0.4041638970375061,
      "learning_rate": 5.43760999148453e-05,
      "loss": 0.0017378553748130798,
      "step": 160800
    },
    {
      "epoch": 45.64575645756457,
      "grad_norm": 14.27238655090332,
      "learning_rate": 5.437326142492194e-05,
      "loss": 0.0035444900393486025,
      "step": 160810
    },
    {
      "epoch": 45.648594947487936,
      "grad_norm": 3.5852174758911133,
      "learning_rate": 5.437042293499859e-05,
      "loss": 0.0025195956230163576,
      "step": 160820
    },
    {
      "epoch": 45.6514334374113,
      "grad_norm": 0.16901333630084991,
      "learning_rate": 5.4367584445075225e-05,
      "loss": 0.006023672223091125,
      "step": 160830
    },
    {
      "epoch": 45.654271927334655,
      "grad_norm": 0.060844410210847855,
      "learning_rate": 5.4364745955151866e-05,
      "loss": 0.006948278099298477,
      "step": 160840
    },
    {
      "epoch": 45.65711041725802,
      "grad_norm": 0.11191053688526154,
      "learning_rate": 5.43619074652285e-05,
      "loss": 0.0009334005415439605,
      "step": 160850
    },
    {
      "epoch": 45.65994890718138,
      "grad_norm": 1.9554001092910767,
      "learning_rate": 5.435906897530514e-05,
      "loss": 0.0008851522579789162,
      "step": 160860
    },
    {
      "epoch": 45.66278739710474,
      "grad_norm": 0.06833608448505402,
      "learning_rate": 5.4356230485381777e-05,
      "loss": 0.005166731029748917,
      "step": 160870
    },
    {
      "epoch": 45.6656258870281,
      "grad_norm": 2.002039670944214,
      "learning_rate": 5.435339199545841e-05,
      "loss": 0.0012178577482700348,
      "step": 160880
    },
    {
      "epoch": 45.66846437695146,
      "grad_norm": 0.5515984892845154,
      "learning_rate": 5.435055350553506e-05,
      "loss": 0.0012314150109887124,
      "step": 160890
    },
    {
      "epoch": 45.671302866874825,
      "grad_norm": 1.4120569229125977,
      "learning_rate": 5.43477150156117e-05,
      "loss": 0.0019286371767520904,
      "step": 160900
    },
    {
      "epoch": 45.67414135679818,
      "grad_norm": 1.3146798610687256,
      "learning_rate": 5.4344876525688335e-05,
      "loss": 0.001442255638539791,
      "step": 160910
    },
    {
      "epoch": 45.676979846721544,
      "grad_norm": 11.858302116394043,
      "learning_rate": 5.4342038035764977e-05,
      "loss": 0.002761480025947094,
      "step": 160920
    },
    {
      "epoch": 45.67981833664491,
      "grad_norm": 0.33902010321617126,
      "learning_rate": 5.433919954584161e-05,
      "loss": 0.0006219649687409401,
      "step": 160930
    },
    {
      "epoch": 45.68265682656826,
      "grad_norm": 0.08200369030237198,
      "learning_rate": 5.433636105591825e-05,
      "loss": 0.0021788103505969046,
      "step": 160940
    },
    {
      "epoch": 45.685495316491625,
      "grad_norm": 0.8130740523338318,
      "learning_rate": 5.43335225659949e-05,
      "loss": 0.0027270039543509485,
      "step": 160950
    },
    {
      "epoch": 45.68833380641499,
      "grad_norm": 0.34162166714668274,
      "learning_rate": 5.4330684076071535e-05,
      "loss": 0.0019761096686124803,
      "step": 160960
    },
    {
      "epoch": 45.69117229633835,
      "grad_norm": 0.047219403088092804,
      "learning_rate": 5.432784558614818e-05,
      "loss": 0.004996160045266151,
      "step": 160970
    },
    {
      "epoch": 45.69401078626171,
      "grad_norm": 0.2030765861272812,
      "learning_rate": 5.432500709622481e-05,
      "loss": 0.0004779621958732605,
      "step": 160980
    },
    {
      "epoch": 45.69684927618507,
      "grad_norm": 0.08220353722572327,
      "learning_rate": 5.4322168606301446e-05,
      "loss": 0.001334654539823532,
      "step": 160990
    },
    {
      "epoch": 45.69968776610843,
      "grad_norm": 0.1498045176267624,
      "learning_rate": 5.431933011637809e-05,
      "loss": 0.006664988398551941,
      "step": 161000
    },
    {
      "epoch": 45.69968776610843,
      "eval_accuracy": 0.9773637693139188,
      "eval_loss": 0.08327616751194,
      "eval_runtime": 32.7296,
      "eval_samples_per_second": 480.513,
      "eval_steps_per_second": 7.516,
      "step": 161000
    },
    {
      "epoch": 45.70252625603179,
      "grad_norm": 8.07216739654541,
      "learning_rate": 5.431649162645472e-05,
      "loss": 0.0029264120385050774,
      "step": 161010
    },
    {
      "epoch": 45.70536474595515,
      "grad_norm": 10.56723690032959,
      "learning_rate": 5.431365313653137e-05,
      "loss": 0.020343662798404695,
      "step": 161020
    },
    {
      "epoch": 45.708203235878514,
      "grad_norm": 0.24377812445163727,
      "learning_rate": 5.431081464660801e-05,
      "loss": 0.002720019780099392,
      "step": 161030
    },
    {
      "epoch": 45.71104172580187,
      "grad_norm": 0.38839972019195557,
      "learning_rate": 5.4307976156684646e-05,
      "loss": 0.004410373046994209,
      "step": 161040
    },
    {
      "epoch": 45.71388021572523,
      "grad_norm": 11.797942161560059,
      "learning_rate": 5.430513766676129e-05,
      "loss": 0.012972970306873322,
      "step": 161050
    },
    {
      "epoch": 45.716718705648596,
      "grad_norm": 0.23124346137046814,
      "learning_rate": 5.430229917683792e-05,
      "loss": 0.008590377867221832,
      "step": 161060
    },
    {
      "epoch": 45.71955719557196,
      "grad_norm": 0.02862338349223137,
      "learning_rate": 5.429946068691456e-05,
      "loss": 0.0035096026957035066,
      "step": 161070
    },
    {
      "epoch": 45.722395685495314,
      "grad_norm": 0.1332978457212448,
      "learning_rate": 5.42966221969912e-05,
      "loss": 0.008227867633104324,
      "step": 161080
    },
    {
      "epoch": 45.72523417541868,
      "grad_norm": 0.09351450204849243,
      "learning_rate": 5.4293783707067846e-05,
      "loss": 0.003157670050859451,
      "step": 161090
    },
    {
      "epoch": 45.72807266534204,
      "grad_norm": 0.04391924664378166,
      "learning_rate": 5.429094521714449e-05,
      "loss": 0.002350090630352497,
      "step": 161100
    },
    {
      "epoch": 45.730911155265396,
      "grad_norm": 0.7595681548118591,
      "learning_rate": 5.428810672722112e-05,
      "loss": 0.000853271409869194,
      "step": 161110
    },
    {
      "epoch": 45.73374964518876,
      "grad_norm": 0.016838688403367996,
      "learning_rate": 5.428526823729776e-05,
      "loss": 0.0007471306249499321,
      "step": 161120
    },
    {
      "epoch": 45.73658813511212,
      "grad_norm": 0.715798556804657,
      "learning_rate": 5.42824297473744e-05,
      "loss": 0.0015186576172709465,
      "step": 161130
    },
    {
      "epoch": 45.739426625035485,
      "grad_norm": 0.015432768501341343,
      "learning_rate": 5.427959125745103e-05,
      "loss": 0.0006268056109547615,
      "step": 161140
    },
    {
      "epoch": 45.74226511495884,
      "grad_norm": 5.377730369567871,
      "learning_rate": 5.427675276752768e-05,
      "loss": 0.00495859682559967,
      "step": 161150
    },
    {
      "epoch": 45.7451036048822,
      "grad_norm": 0.10866178572177887,
      "learning_rate": 5.427391427760432e-05,
      "loss": 0.0012252561748027802,
      "step": 161160
    },
    {
      "epoch": 45.747942094805566,
      "grad_norm": 0.13667306303977966,
      "learning_rate": 5.427107578768096e-05,
      "loss": 0.0005437683314085007,
      "step": 161170
    },
    {
      "epoch": 45.75078058472892,
      "grad_norm": 0.04786505177617073,
      "learning_rate": 5.42682372977576e-05,
      "loss": 0.005026413500308991,
      "step": 161180
    },
    {
      "epoch": 45.753619074652285,
      "grad_norm": 0.40971362590789795,
      "learning_rate": 5.426539880783423e-05,
      "loss": 0.0017123386263847352,
      "step": 161190
    },
    {
      "epoch": 45.75645756457565,
      "grad_norm": 0.037988048046827316,
      "learning_rate": 5.4262560317910874e-05,
      "loss": 0.0012234851717948913,
      "step": 161200
    },
    {
      "epoch": 45.759296054499,
      "grad_norm": 0.4397001564502716,
      "learning_rate": 5.425972182798751e-05,
      "loss": 0.001748490147292614,
      "step": 161210
    },
    {
      "epoch": 45.762134544422366,
      "grad_norm": 0.042097996920347214,
      "learning_rate": 5.425688333806416e-05,
      "loss": 0.0038283582776784896,
      "step": 161220
    },
    {
      "epoch": 45.76497303434573,
      "grad_norm": 0.7592105865478516,
      "learning_rate": 5.425404484814079e-05,
      "loss": 0.0008447635918855667,
      "step": 161230
    },
    {
      "epoch": 45.76781152426909,
      "grad_norm": 0.36097195744514465,
      "learning_rate": 5.425120635821743e-05,
      "loss": 0.001230390928685665,
      "step": 161240
    },
    {
      "epoch": 45.77065001419245,
      "grad_norm": 2.0582427978515625,
      "learning_rate": 5.424836786829407e-05,
      "loss": 0.0010907815769314766,
      "step": 161250
    },
    {
      "epoch": 45.77348850411581,
      "grad_norm": 11.278580665588379,
      "learning_rate": 5.424552937837071e-05,
      "loss": 0.004879453778266906,
      "step": 161260
    },
    {
      "epoch": 45.776326994039174,
      "grad_norm": 0.3125346004962921,
      "learning_rate": 5.4242690888447343e-05,
      "loss": 0.00045435577630996703,
      "step": 161270
    },
    {
      "epoch": 45.77916548396253,
      "grad_norm": 0.5384764075279236,
      "learning_rate": 5.4239852398523985e-05,
      "loss": 0.0020776621997356413,
      "step": 161280
    },
    {
      "epoch": 45.78200397388589,
      "grad_norm": 0.1224488690495491,
      "learning_rate": 5.423701390860063e-05,
      "loss": 0.001577579416334629,
      "step": 161290
    },
    {
      "epoch": 45.784842463809255,
      "grad_norm": 0.08446390181779861,
      "learning_rate": 5.423417541867727e-05,
      "loss": 0.000662391446530819,
      "step": 161300
    },
    {
      "epoch": 45.78768095373261,
      "grad_norm": 1.3894742727279663,
      "learning_rate": 5.423133692875391e-05,
      "loss": 0.010084594786167144,
      "step": 161310
    },
    {
      "epoch": 45.790519443655974,
      "grad_norm": 0.3065516948699951,
      "learning_rate": 5.4228498438830543e-05,
      "loss": 0.0021734993904829027,
      "step": 161320
    },
    {
      "epoch": 45.79335793357934,
      "grad_norm": 0.25830909609794617,
      "learning_rate": 5.4225659948907185e-05,
      "loss": 0.0012144390493631363,
      "step": 161330
    },
    {
      "epoch": 45.7961964235027,
      "grad_norm": 0.23656773567199707,
      "learning_rate": 5.422282145898382e-05,
      "loss": 0.0013937612995505333,
      "step": 161340
    },
    {
      "epoch": 45.799034913426055,
      "grad_norm": 0.17942753434181213,
      "learning_rate": 5.421998296906047e-05,
      "loss": 0.002073255926370621,
      "step": 161350
    },
    {
      "epoch": 45.80187340334942,
      "grad_norm": 0.07943885028362274,
      "learning_rate": 5.42171444791371e-05,
      "loss": 0.0021154118701815607,
      "step": 161360
    },
    {
      "epoch": 45.80471189327278,
      "grad_norm": 0.5212387442588806,
      "learning_rate": 5.4214305989213744e-05,
      "loss": 0.004924054071307182,
      "step": 161370
    },
    {
      "epoch": 45.80755038319614,
      "grad_norm": 0.026930060237646103,
      "learning_rate": 5.421146749929038e-05,
      "loss": 0.0008995577692985535,
      "step": 161380
    },
    {
      "epoch": 45.8103888731195,
      "grad_norm": 1.582486867904663,
      "learning_rate": 5.420862900936702e-05,
      "loss": 0.000835118442773819,
      "step": 161390
    },
    {
      "epoch": 45.81322736304286,
      "grad_norm": 7.061641693115234,
      "learning_rate": 5.4205790519443654e-05,
      "loss": 0.004739048331975937,
      "step": 161400
    },
    {
      "epoch": 45.81606585296622,
      "grad_norm": 0.0528242252767086,
      "learning_rate": 5.4202952029520296e-05,
      "loss": 0.00348777137696743,
      "step": 161410
    },
    {
      "epoch": 45.81890434288958,
      "grad_norm": 0.4669622778892517,
      "learning_rate": 5.4200113539596944e-05,
      "loss": 0.0009508721530437469,
      "step": 161420
    },
    {
      "epoch": 45.821742832812944,
      "grad_norm": 0.10133606195449829,
      "learning_rate": 5.419727504967358e-05,
      "loss": 0.0009340867400169373,
      "step": 161430
    },
    {
      "epoch": 45.82458132273631,
      "grad_norm": 0.10479370504617691,
      "learning_rate": 5.419443655975022e-05,
      "loss": 0.00103929340839386,
      "step": 161440
    },
    {
      "epoch": 45.82741981265966,
      "grad_norm": 0.4021565020084381,
      "learning_rate": 5.4191598069826854e-05,
      "loss": 0.0011530278250575065,
      "step": 161450
    },
    {
      "epoch": 45.830258302583026,
      "grad_norm": 0.2476356327533722,
      "learning_rate": 5.418875957990349e-05,
      "loss": 0.0013172492384910584,
      "step": 161460
    },
    {
      "epoch": 45.83309679250639,
      "grad_norm": 0.10578019171953201,
      "learning_rate": 5.418592108998013e-05,
      "loss": 0.0006564496085047722,
      "step": 161470
    },
    {
      "epoch": 45.835935282429745,
      "grad_norm": 2.070984125137329,
      "learning_rate": 5.4183082600056765e-05,
      "loss": 0.0017098024487495421,
      "step": 161480
    },
    {
      "epoch": 45.83877377235311,
      "grad_norm": 0.710365891456604,
      "learning_rate": 5.418024411013341e-05,
      "loss": 0.0031216679140925407,
      "step": 161490
    },
    {
      "epoch": 45.84161226227647,
      "grad_norm": 0.02391446754336357,
      "learning_rate": 5.4177405620210054e-05,
      "loss": 0.0007541656494140625,
      "step": 161500
    },
    {
      "epoch": 45.84161226227647,
      "eval_accuracy": 0.9771730145609462,
      "eval_loss": 0.08208510279655457,
      "eval_runtime": 32.4833,
      "eval_samples_per_second": 484.157,
      "eval_steps_per_second": 7.573,
      "step": 161500
    },
    {
      "epoch": 45.84445075219983,
      "grad_norm": 0.12032294273376465,
      "learning_rate": 5.417456713028669e-05,
      "loss": 0.0010130520910024642,
      "step": 161510
    },
    {
      "epoch": 45.84728924212319,
      "grad_norm": 0.037861891090869904,
      "learning_rate": 5.417172864036333e-05,
      "loss": 0.0032112617045640945,
      "step": 161520
    },
    {
      "epoch": 45.85012773204655,
      "grad_norm": 0.6204045414924622,
      "learning_rate": 5.4168890150439965e-05,
      "loss": 0.0014088433235883713,
      "step": 161530
    },
    {
      "epoch": 45.852966221969915,
      "grad_norm": 0.11994437873363495,
      "learning_rate": 5.4166051660516606e-05,
      "loss": 0.005461747571825981,
      "step": 161540
    },
    {
      "epoch": 45.85580471189327,
      "grad_norm": 0.0325290709733963,
      "learning_rate": 5.4163213170593254e-05,
      "loss": 0.004705595597624779,
      "step": 161550
    },
    {
      "epoch": 45.85864320181663,
      "grad_norm": 0.04252807796001434,
      "learning_rate": 5.416037468066989e-05,
      "loss": 0.0005035419017076492,
      "step": 161560
    },
    {
      "epoch": 45.861481691739996,
      "grad_norm": 0.22419458627700806,
      "learning_rate": 5.415753619074653e-05,
      "loss": 0.000624791719019413,
      "step": 161570
    },
    {
      "epoch": 45.86432018166335,
      "grad_norm": 0.1020413190126419,
      "learning_rate": 5.4154697700823165e-05,
      "loss": 0.000715498998761177,
      "step": 161580
    },
    {
      "epoch": 45.867158671586715,
      "grad_norm": 0.032956041395664215,
      "learning_rate": 5.41518592108998e-05,
      "loss": 0.002926127426326275,
      "step": 161590
    },
    {
      "epoch": 45.86999716151008,
      "grad_norm": 4.767463684082031,
      "learning_rate": 5.414902072097644e-05,
      "loss": 0.0011207493022084235,
      "step": 161600
    },
    {
      "epoch": 45.87283565143344,
      "grad_norm": 1.5675450563430786,
      "learning_rate": 5.4146182231053076e-05,
      "loss": 0.005903331190347671,
      "step": 161610
    },
    {
      "epoch": 45.8756741413568,
      "grad_norm": 0.2801133394241333,
      "learning_rate": 5.4143343741129724e-05,
      "loss": 0.007808262109756469,
      "step": 161620
    },
    {
      "epoch": 45.87851263128016,
      "grad_norm": 11.746882438659668,
      "learning_rate": 5.4140505251206365e-05,
      "loss": 0.007883566617965698,
      "step": 161630
    },
    {
      "epoch": 45.88135112120352,
      "grad_norm": 2.616649866104126,
      "learning_rate": 5.4137666761283e-05,
      "loss": 0.0033796489238739013,
      "step": 161640
    },
    {
      "epoch": 45.88418961112688,
      "grad_norm": 0.5332644581794739,
      "learning_rate": 5.413482827135964e-05,
      "loss": 0.003921709582209587,
      "step": 161650
    },
    {
      "epoch": 45.88702810105024,
      "grad_norm": 0.054670125246047974,
      "learning_rate": 5.4131989781436276e-05,
      "loss": 0.0034120339900255204,
      "step": 161660
    },
    {
      "epoch": 45.889866590973604,
      "grad_norm": 0.0413186177611351,
      "learning_rate": 5.412915129151292e-05,
      "loss": 0.005305983126163483,
      "step": 161670
    },
    {
      "epoch": 45.89270508089696,
      "grad_norm": 0.08059436827898026,
      "learning_rate": 5.412631280158955e-05,
      "loss": 0.0027314046397805213,
      "step": 161680
    },
    {
      "epoch": 45.89554357082032,
      "grad_norm": 0.3735695779323578,
      "learning_rate": 5.41234743116662e-05,
      "loss": 0.0011419204995036126,
      "step": 161690
    },
    {
      "epoch": 45.898382060743685,
      "grad_norm": 2.8354036808013916,
      "learning_rate": 5.4120635821742834e-05,
      "loss": 0.004120286554098129,
      "step": 161700
    },
    {
      "epoch": 45.90122055066705,
      "grad_norm": 1.4753535985946655,
      "learning_rate": 5.4117797331819476e-05,
      "loss": 0.0028513403609395026,
      "step": 161710
    },
    {
      "epoch": 45.904059040590404,
      "grad_norm": 0.14362750947475433,
      "learning_rate": 5.411495884189611e-05,
      "loss": 0.0010727571323513984,
      "step": 161720
    },
    {
      "epoch": 45.90689753051377,
      "grad_norm": 0.09154773503541946,
      "learning_rate": 5.411212035197275e-05,
      "loss": 0.0012993168085813523,
      "step": 161730
    },
    {
      "epoch": 45.90973602043713,
      "grad_norm": 0.23596924543380737,
      "learning_rate": 5.4109281862049386e-05,
      "loss": 0.0007534706965088844,
      "step": 161740
    },
    {
      "epoch": 45.912574510360486,
      "grad_norm": 3.5349955558776855,
      "learning_rate": 5.4106443372126035e-05,
      "loss": 0.0033326372504234314,
      "step": 161750
    },
    {
      "epoch": 45.91541300028385,
      "grad_norm": 1.8685142993927002,
      "learning_rate": 5.4103604882202676e-05,
      "loss": 0.006194271147251129,
      "step": 161760
    },
    {
      "epoch": 45.91825149020721,
      "grad_norm": 0.045612432062625885,
      "learning_rate": 5.410076639227931e-05,
      "loss": 0.0038339927792549132,
      "step": 161770
    },
    {
      "epoch": 45.92108998013057,
      "grad_norm": 2.440669059753418,
      "learning_rate": 5.409792790235595e-05,
      "loss": 0.014538729190826416,
      "step": 161780
    },
    {
      "epoch": 45.92392847005393,
      "grad_norm": 0.712125301361084,
      "learning_rate": 5.4095089412432586e-05,
      "loss": 0.0017345525324344636,
      "step": 161790
    },
    {
      "epoch": 45.92676695997729,
      "grad_norm": 1.2181425094604492,
      "learning_rate": 5.409225092250922e-05,
      "loss": 0.0027525674551725386,
      "step": 161800
    },
    {
      "epoch": 45.929605449900656,
      "grad_norm": 0.42354992032051086,
      "learning_rate": 5.408941243258586e-05,
      "loss": 0.0016804732382297515,
      "step": 161810
    },
    {
      "epoch": 45.93244393982401,
      "grad_norm": 0.5312386155128479,
      "learning_rate": 5.408657394266251e-05,
      "loss": 0.0012051908299326897,
      "step": 161820
    },
    {
      "epoch": 45.935282429747375,
      "grad_norm": 1.04977548122406,
      "learning_rate": 5.4083735452739145e-05,
      "loss": 0.0024868519976735115,
      "step": 161830
    },
    {
      "epoch": 45.93812091967074,
      "grad_norm": 0.11321166902780533,
      "learning_rate": 5.4080896962815787e-05,
      "loss": 0.0007630808278918267,
      "step": 161840
    },
    {
      "epoch": 45.94095940959409,
      "grad_norm": 0.3283163905143738,
      "learning_rate": 5.407805847289242e-05,
      "loss": 0.0074503995478153225,
      "step": 161850
    },
    {
      "epoch": 45.943797899517456,
      "grad_norm": 0.03881268948316574,
      "learning_rate": 5.407521998296906e-05,
      "loss": 0.006832519173622131,
      "step": 161860
    },
    {
      "epoch": 45.94663638944082,
      "grad_norm": 0.03810558840632439,
      "learning_rate": 5.40723814930457e-05,
      "loss": 0.010600626468658447,
      "step": 161870
    },
    {
      "epoch": 45.949474879364175,
      "grad_norm": 0.20908771455287933,
      "learning_rate": 5.406954300312234e-05,
      "loss": 0.00047017689794301985,
      "step": 161880
    },
    {
      "epoch": 45.95231336928754,
      "grad_norm": 0.23305509984493256,
      "learning_rate": 5.406670451319899e-05,
      "loss": 0.0033332813531160355,
      "step": 161890
    },
    {
      "epoch": 45.9551518592109,
      "grad_norm": 0.9041756391525269,
      "learning_rate": 5.406386602327562e-05,
      "loss": 0.002411099337041378,
      "step": 161900
    },
    {
      "epoch": 45.95799034913426,
      "grad_norm": 0.3472764194011688,
      "learning_rate": 5.406102753335226e-05,
      "loss": 0.0038720719516277314,
      "step": 161910
    },
    {
      "epoch": 45.96082883905762,
      "grad_norm": 1.885081171989441,
      "learning_rate": 5.40581890434289e-05,
      "loss": 0.0008585911244153976,
      "step": 161920
    },
    {
      "epoch": 45.96366732898098,
      "grad_norm": 4.354450702667236,
      "learning_rate": 5.405535055350553e-05,
      "loss": 0.0013964829966425895,
      "step": 161930
    },
    {
      "epoch": 45.966505818904345,
      "grad_norm": 1.279211163520813,
      "learning_rate": 5.405251206358217e-05,
      "loss": 0.012046570330858231,
      "step": 161940
    },
    {
      "epoch": 45.9693443088277,
      "grad_norm": 0.2382349967956543,
      "learning_rate": 5.404967357365882e-05,
      "loss": 0.0011683886870741845,
      "step": 161950
    },
    {
      "epoch": 45.972182798751064,
      "grad_norm": 1.485707402229309,
      "learning_rate": 5.4046835083735456e-05,
      "loss": 0.003632603958249092,
      "step": 161960
    },
    {
      "epoch": 45.97502128867443,
      "grad_norm": 0.26036351919174194,
      "learning_rate": 5.40439965938121e-05,
      "loss": 0.00047646872699260714,
      "step": 161970
    },
    {
      "epoch": 45.97785977859779,
      "grad_norm": 0.13081778585910797,
      "learning_rate": 5.404115810388873e-05,
      "loss": 0.005368375778198242,
      "step": 161980
    },
    {
      "epoch": 45.980698268521145,
      "grad_norm": 3.8794097900390625,
      "learning_rate": 5.403831961396537e-05,
      "loss": 0.0022597135975956917,
      "step": 161990
    },
    {
      "epoch": 45.98353675844451,
      "grad_norm": 0.036129508167505264,
      "learning_rate": 5.403548112404201e-05,
      "loss": 0.001491733081638813,
      "step": 162000
    },
    {
      "epoch": 45.98353675844451,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.08672595024108887,
      "eval_runtime": 32.7045,
      "eval_samples_per_second": 480.882,
      "eval_steps_per_second": 7.522,
      "step": 162000
    },
    {
      "epoch": 45.98637524836787,
      "grad_norm": 0.560282826423645,
      "learning_rate": 5.403264263411865e-05,
      "loss": 0.0043994951993227005,
      "step": 162010
    },
    {
      "epoch": 45.98921373829123,
      "grad_norm": 0.32815879583358765,
      "learning_rate": 5.40298041441953e-05,
      "loss": 0.0065226837992668155,
      "step": 162020
    },
    {
      "epoch": 45.99205222821459,
      "grad_norm": 0.04081922397017479,
      "learning_rate": 5.402696565427193e-05,
      "loss": 0.0009791692718863487,
      "step": 162030
    },
    {
      "epoch": 45.99489071813795,
      "grad_norm": 0.38631582260131836,
      "learning_rate": 5.4024127164348573e-05,
      "loss": 0.0017499243840575218,
      "step": 162040
    },
    {
      "epoch": 45.99772920806131,
      "grad_norm": 0.8775612711906433,
      "learning_rate": 5.402128867442521e-05,
      "loss": 0.0029018744826316833,
      "step": 162050
    },
    {
      "epoch": 46.00056769798467,
      "grad_norm": 0.0812491625547409,
      "learning_rate": 5.401845018450184e-05,
      "loss": 0.0011390838772058487,
      "step": 162060
    },
    {
      "epoch": 46.003406187908034,
      "grad_norm": 0.040921665728092194,
      "learning_rate": 5.4015611694578484e-05,
      "loss": 0.0009308991953730583,
      "step": 162070
    },
    {
      "epoch": 46.0062446778314,
      "grad_norm": 0.036484867334365845,
      "learning_rate": 5.401277320465513e-05,
      "loss": 0.0006519539281725883,
      "step": 162080
    },
    {
      "epoch": 46.00908316775475,
      "grad_norm": 0.6142832040786743,
      "learning_rate": 5.400993471473177e-05,
      "loss": 0.0008300293236970901,
      "step": 162090
    },
    {
      "epoch": 46.011921657678116,
      "grad_norm": 0.04029611498117447,
      "learning_rate": 5.400709622480841e-05,
      "loss": 0.001254924200475216,
      "step": 162100
    },
    {
      "epoch": 46.01476014760148,
      "grad_norm": 0.027345988899469376,
      "learning_rate": 5.400425773488504e-05,
      "loss": 0.0015933027490973473,
      "step": 162110
    },
    {
      "epoch": 46.017598637524834,
      "grad_norm": 1.4587512016296387,
      "learning_rate": 5.4001419244961684e-05,
      "loss": 0.0022168003022670748,
      "step": 162120
    },
    {
      "epoch": 46.0204371274482,
      "grad_norm": 0.22474893927574158,
      "learning_rate": 5.399858075503832e-05,
      "loss": 0.00043445434421300887,
      "step": 162130
    },
    {
      "epoch": 46.02327561737156,
      "grad_norm": 0.04749462381005287,
      "learning_rate": 5.399574226511496e-05,
      "loss": 0.0011852780357003213,
      "step": 162140
    },
    {
      "epoch": 46.026114107294916,
      "grad_norm": 0.07627009600400925,
      "learning_rate": 5.399290377519161e-05,
      "loss": 0.0017149774357676506,
      "step": 162150
    },
    {
      "epoch": 46.02895259721828,
      "grad_norm": 3.3753859996795654,
      "learning_rate": 5.399006528526824e-05,
      "loss": 0.0013398149982094764,
      "step": 162160
    },
    {
      "epoch": 46.03179108714164,
      "grad_norm": 0.3647576868534088,
      "learning_rate": 5.398722679534488e-05,
      "loss": 0.00043796300888061525,
      "step": 162170
    },
    {
      "epoch": 46.034629577065004,
      "grad_norm": 0.06689789146184921,
      "learning_rate": 5.398438830542152e-05,
      "loss": 0.0007028438150882721,
      "step": 162180
    },
    {
      "epoch": 46.03746806698836,
      "grad_norm": 4.7765421867370605,
      "learning_rate": 5.398154981549815e-05,
      "loss": 0.001455855555832386,
      "step": 162190
    },
    {
      "epoch": 46.04030655691172,
      "grad_norm": 1.0843403339385986,
      "learning_rate": 5.3978711325574795e-05,
      "loss": 0.0006022730842232705,
      "step": 162200
    },
    {
      "epoch": 46.043145046835086,
      "grad_norm": 0.044870927929878235,
      "learning_rate": 5.397587283565143e-05,
      "loss": 0.0016254452988505364,
      "step": 162210
    },
    {
      "epoch": 46.04598353675844,
      "grad_norm": 0.7910535931587219,
      "learning_rate": 5.397303434572808e-05,
      "loss": 0.0007785676047205925,
      "step": 162220
    },
    {
      "epoch": 46.048822026681805,
      "grad_norm": 0.4219972491264343,
      "learning_rate": 5.397019585580472e-05,
      "loss": 0.0022998496890068052,
      "step": 162230
    },
    {
      "epoch": 46.05166051660517,
      "grad_norm": 0.7925904989242554,
      "learning_rate": 5.3967357365881353e-05,
      "loss": 0.001314728520810604,
      "step": 162240
    },
    {
      "epoch": 46.05449900652852,
      "grad_norm": 0.12839733064174652,
      "learning_rate": 5.3964518875957995e-05,
      "loss": 0.0023147203028202057,
      "step": 162250
    },
    {
      "epoch": 46.057337496451886,
      "grad_norm": 0.07025919109582901,
      "learning_rate": 5.396168038603463e-05,
      "loss": 0.0011056412011384964,
      "step": 162260
    },
    {
      "epoch": 46.06017598637525,
      "grad_norm": 2.206904888153076,
      "learning_rate": 5.3958841896111264e-05,
      "loss": 0.0029628485441207885,
      "step": 162270
    },
    {
      "epoch": 46.06301447629861,
      "grad_norm": 0.4191048741340637,
      "learning_rate": 5.395600340618792e-05,
      "loss": 0.0017273362725973129,
      "step": 162280
    },
    {
      "epoch": 46.06585296622197,
      "grad_norm": 0.10081643611192703,
      "learning_rate": 5.3953164916264554e-05,
      "loss": 0.0011167537420988082,
      "step": 162290
    },
    {
      "epoch": 46.06869145614533,
      "grad_norm": 0.32089418172836304,
      "learning_rate": 5.395032642634119e-05,
      "loss": 0.00539734959602356,
      "step": 162300
    },
    {
      "epoch": 46.071529946068694,
      "grad_norm": 1.1832629442214966,
      "learning_rate": 5.394748793641783e-05,
      "loss": 0.002499717473983765,
      "step": 162310
    },
    {
      "epoch": 46.07436843599205,
      "grad_norm": 0.19669432938098907,
      "learning_rate": 5.3944649446494464e-05,
      "loss": 0.002529860846698284,
      "step": 162320
    },
    {
      "epoch": 46.07720692591541,
      "grad_norm": 0.034073006361722946,
      "learning_rate": 5.3941810956571106e-05,
      "loss": 0.0012178221717476846,
      "step": 162330
    },
    {
      "epoch": 46.080045415838775,
      "grad_norm": 0.5585459470748901,
      "learning_rate": 5.393897246664774e-05,
      "loss": 0.0019234787672758103,
      "step": 162340
    },
    {
      "epoch": 46.08288390576214,
      "grad_norm": 0.05181216448545456,
      "learning_rate": 5.393613397672439e-05,
      "loss": 0.0014020778238773347,
      "step": 162350
    },
    {
      "epoch": 46.085722395685494,
      "grad_norm": 0.17448733747005463,
      "learning_rate": 5.393329548680103e-05,
      "loss": 0.006489067524671555,
      "step": 162360
    },
    {
      "epoch": 46.08856088560886,
      "grad_norm": 0.07429579645395279,
      "learning_rate": 5.3930456996877664e-05,
      "loss": 0.000692986324429512,
      "step": 162370
    },
    {
      "epoch": 46.09139937553222,
      "grad_norm": 3.005459785461426,
      "learning_rate": 5.3927618506954306e-05,
      "loss": 0.002151169814169407,
      "step": 162380
    },
    {
      "epoch": 46.094237865455575,
      "grad_norm": 1.0353405475616455,
      "learning_rate": 5.392478001703094e-05,
      "loss": 0.005647667497396469,
      "step": 162390
    },
    {
      "epoch": 46.09707635537894,
      "grad_norm": 0.02238048054277897,
      "learning_rate": 5.3921941527107575e-05,
      "loss": 0.0006479542702436447,
      "step": 162400
    },
    {
      "epoch": 46.0999148453023,
      "grad_norm": 0.2964235544204712,
      "learning_rate": 5.3919103037184216e-05,
      "loss": 0.0009610552340745925,
      "step": 162410
    },
    {
      "epoch": 46.10275333522566,
      "grad_norm": 0.017714111134409904,
      "learning_rate": 5.3916264547260864e-05,
      "loss": 0.0003964593634009361,
      "step": 162420
    },
    {
      "epoch": 46.10559182514902,
      "grad_norm": 0.3485843539237976,
      "learning_rate": 5.39134260573375e-05,
      "loss": 0.0004403118044137955,
      "step": 162430
    },
    {
      "epoch": 46.10843031507238,
      "grad_norm": 0.2335071712732315,
      "learning_rate": 5.391058756741414e-05,
      "loss": 0.0017133953049778938,
      "step": 162440
    },
    {
      "epoch": 46.111268804995746,
      "grad_norm": 0.13199400901794434,
      "learning_rate": 5.3907749077490775e-05,
      "loss": 0.0008099187165498733,
      "step": 162450
    },
    {
      "epoch": 46.1141072949191,
      "grad_norm": 0.02528182603418827,
      "learning_rate": 5.3904910587567416e-05,
      "loss": 0.008167947828769683,
      "step": 162460
    },
    {
      "epoch": 46.116945784842464,
      "grad_norm": 2.1413111686706543,
      "learning_rate": 5.390207209764405e-05,
      "loss": 0.002173840068280697,
      "step": 162470
    },
    {
      "epoch": 46.11978427476583,
      "grad_norm": 22.540597915649414,
      "learning_rate": 5.38992336077207e-05,
      "loss": 0.005780110508203507,
      "step": 162480
    },
    {
      "epoch": 46.12262276468918,
      "grad_norm": 0.03292245417833328,
      "learning_rate": 5.389639511779734e-05,
      "loss": 0.0012766674160957336,
      "step": 162490
    },
    {
      "epoch": 46.125461254612546,
      "grad_norm": 0.025272870436310768,
      "learning_rate": 5.3893556627873975e-05,
      "loss": 0.005342083424329758,
      "step": 162500
    },
    {
      "epoch": 46.125461254612546,
      "eval_accuracy": 0.9769822598079736,
      "eval_loss": 0.08267591893672943,
      "eval_runtime": 31.8567,
      "eval_samples_per_second": 493.679,
      "eval_steps_per_second": 7.722,
      "step": 162500
    },
    {
      "epoch": 46.12829974453591,
      "grad_norm": 0.362273633480072,
      "learning_rate": 5.3890718137950616e-05,
      "loss": 0.0008373919874429703,
      "step": 162510
    },
    {
      "epoch": 46.131138234459264,
      "grad_norm": 0.03289893642067909,
      "learning_rate": 5.388787964802725e-05,
      "loss": 0.0003962647169828415,
      "step": 162520
    },
    {
      "epoch": 46.13397672438263,
      "grad_norm": 0.1362258642911911,
      "learning_rate": 5.3885041158103886e-05,
      "loss": 0.000758872926235199,
      "step": 162530
    },
    {
      "epoch": 46.13681521430599,
      "grad_norm": 0.08325715363025665,
      "learning_rate": 5.388220266818053e-05,
      "loss": 0.0005447294563055039,
      "step": 162540
    },
    {
      "epoch": 46.13965370422935,
      "grad_norm": 0.0715063065290451,
      "learning_rate": 5.3879364178257175e-05,
      "loss": 0.002281661704182625,
      "step": 162550
    },
    {
      "epoch": 46.14249219415271,
      "grad_norm": 4.7066426277160645,
      "learning_rate": 5.387652568833381e-05,
      "loss": 0.0012405378744006156,
      "step": 162560
    },
    {
      "epoch": 46.14533068407607,
      "grad_norm": 0.3484973609447479,
      "learning_rate": 5.387368719841045e-05,
      "loss": 0.0006002500653266907,
      "step": 162570
    },
    {
      "epoch": 46.148169173999435,
      "grad_norm": 0.054399099200963974,
      "learning_rate": 5.3870848708487086e-05,
      "loss": 0.0005867704749107361,
      "step": 162580
    },
    {
      "epoch": 46.15100766392279,
      "grad_norm": 0.3551778197288513,
      "learning_rate": 5.386801021856373e-05,
      "loss": 0.0009035732597112656,
      "step": 162590
    },
    {
      "epoch": 46.15384615384615,
      "grad_norm": 0.3016417324542999,
      "learning_rate": 5.386517172864036e-05,
      "loss": 0.004234226047992706,
      "step": 162600
    },
    {
      "epoch": 46.156684643769516,
      "grad_norm": 0.09140706807374954,
      "learning_rate": 5.3862333238717e-05,
      "loss": 0.0005207752808928489,
      "step": 162610
    },
    {
      "epoch": 46.15952313369287,
      "grad_norm": 0.04676120728254318,
      "learning_rate": 5.385949474879365e-05,
      "loss": 0.0007847731932997704,
      "step": 162620
    },
    {
      "epoch": 46.162361623616235,
      "grad_norm": 4.9255475997924805,
      "learning_rate": 5.3856656258870286e-05,
      "loss": 0.002993122488260269,
      "step": 162630
    },
    {
      "epoch": 46.1652001135396,
      "grad_norm": 2.1443607807159424,
      "learning_rate": 5.385381776894692e-05,
      "loss": 0.001250469870865345,
      "step": 162640
    },
    {
      "epoch": 46.16803860346296,
      "grad_norm": 0.23584608733654022,
      "learning_rate": 5.3851263128015894e-05,
      "loss": 0.011904342472553254,
      "step": 162650
    },
    {
      "epoch": 46.170877093386316,
      "grad_norm": 0.05330846086144447,
      "learning_rate": 5.384842463809254e-05,
      "loss": 0.0007619190961122513,
      "step": 162660
    },
    {
      "epoch": 46.17371558330968,
      "grad_norm": 0.07755720615386963,
      "learning_rate": 5.384558614816918e-05,
      "loss": 0.0005170943215489388,
      "step": 162670
    },
    {
      "epoch": 46.17655407323304,
      "grad_norm": 0.013385762460529804,
      "learning_rate": 5.384274765824582e-05,
      "loss": 0.001262364722788334,
      "step": 162680
    },
    {
      "epoch": 46.1793925631564,
      "grad_norm": 1.475870966911316,
      "learning_rate": 5.383990916832245e-05,
      "loss": 0.001679704524576664,
      "step": 162690
    },
    {
      "epoch": 46.18223105307976,
      "grad_norm": 0.09247991442680359,
      "learning_rate": 5.3837070678399094e-05,
      "loss": 0.0008517038077116013,
      "step": 162700
    },
    {
      "epoch": 46.185069543003124,
      "grad_norm": 0.25815045833587646,
      "learning_rate": 5.383423218847573e-05,
      "loss": 0.0023378446698188783,
      "step": 162710
    },
    {
      "epoch": 46.18790803292649,
      "grad_norm": 0.04754171147942543,
      "learning_rate": 5.383139369855237e-05,
      "loss": 0.0009436378255486489,
      "step": 162720
    },
    {
      "epoch": 46.19074652284984,
      "grad_norm": 0.05192510038614273,
      "learning_rate": 5.382855520862902e-05,
      "loss": 0.0006022121757268906,
      "step": 162730
    },
    {
      "epoch": 46.193585012773205,
      "grad_norm": 0.11065036803483963,
      "learning_rate": 5.382571671870565e-05,
      "loss": 0.0002622058615088463,
      "step": 162740
    },
    {
      "epoch": 46.19642350269657,
      "grad_norm": 0.021045805886387825,
      "learning_rate": 5.3822878228782294e-05,
      "loss": 0.00031739603728055956,
      "step": 162750
    },
    {
      "epoch": 46.199261992619924,
      "grad_norm": 0.44834938645362854,
      "learning_rate": 5.382003973885893e-05,
      "loss": 0.0005587384104728699,
      "step": 162760
    },
    {
      "epoch": 46.20210048254329,
      "grad_norm": 0.05369019880890846,
      "learning_rate": 5.381720124893557e-05,
      "loss": 0.0004134180024266243,
      "step": 162770
    },
    {
      "epoch": 46.20493897246665,
      "grad_norm": 0.19729112088680267,
      "learning_rate": 5.3814362759012204e-05,
      "loss": 0.0016839876770973206,
      "step": 162780
    },
    {
      "epoch": 46.207777462390005,
      "grad_norm": 0.15583227574825287,
      "learning_rate": 5.381152426908884e-05,
      "loss": 0.0017824217677116394,
      "step": 162790
    },
    {
      "epoch": 46.21061595231337,
      "grad_norm": 0.10630655288696289,
      "learning_rate": 5.3808685779165494e-05,
      "loss": 0.001859956979751587,
      "step": 162800
    },
    {
      "epoch": 46.21345444223673,
      "grad_norm": 0.13975103199481964,
      "learning_rate": 5.380584728924213e-05,
      "loss": 0.0002676393836736679,
      "step": 162810
    },
    {
      "epoch": 46.216292932160094,
      "grad_norm": 3.1000845432281494,
      "learning_rate": 5.380300879931876e-05,
      "loss": 0.002632913738489151,
      "step": 162820
    },
    {
      "epoch": 46.21913142208345,
      "grad_norm": 0.0686105266213417,
      "learning_rate": 5.3800170309395404e-05,
      "loss": 0.011178734898567199,
      "step": 162830
    },
    {
      "epoch": 46.22196991200681,
      "grad_norm": 0.09672816097736359,
      "learning_rate": 5.379733181947204e-05,
      "loss": 0.00491722822189331,
      "step": 162840
    },
    {
      "epoch": 46.224808401930176,
      "grad_norm": 0.9965134263038635,
      "learning_rate": 5.379449332954868e-05,
      "loss": 0.0023222699761390685,
      "step": 162850
    },
    {
      "epoch": 46.22764689185353,
      "grad_norm": 0.04138294234871864,
      "learning_rate": 5.379165483962533e-05,
      "loss": 0.0013759551569819451,
      "step": 162860
    },
    {
      "epoch": 46.230485381776894,
      "grad_norm": 1.2666488885879517,
      "learning_rate": 5.378881634970196e-05,
      "loss": 0.0005918083712458611,
      "step": 162870
    },
    {
      "epoch": 46.23332387170026,
      "grad_norm": 0.2824135720729828,
      "learning_rate": 5.3785977859778604e-05,
      "loss": 0.005900303274393082,
      "step": 162880
    },
    {
      "epoch": 46.23616236162361,
      "grad_norm": 1.0564559698104858,
      "learning_rate": 5.378342321884757e-05,
      "loss": 0.015002001821994782,
      "step": 162890
    },
    {
      "epoch": 46.239000851546976,
      "grad_norm": 0.09444060176610947,
      "learning_rate": 5.378058472892421e-05,
      "loss": 0.015026721358299255,
      "step": 162900
    },
    {
      "epoch": 46.24183934147034,
      "grad_norm": 1.216881275177002,
      "learning_rate": 5.377774623900086e-05,
      "loss": 0.0018469871953129768,
      "step": 162910
    },
    {
      "epoch": 46.2446778313937,
      "grad_norm": 0.06251490861177444,
      "learning_rate": 5.3774907749077495e-05,
      "loss": 0.0011286428198218346,
      "step": 162920
    },
    {
      "epoch": 46.24751632131706,
      "grad_norm": 7.733246803283691,
      "learning_rate": 5.3772069259154136e-05,
      "loss": 0.0019750483334064484,
      "step": 162930
    },
    {
      "epoch": 46.25035481124042,
      "grad_norm": 0.9402521848678589,
      "learning_rate": 5.376923076923077e-05,
      "loss": 0.0015726329758763314,
      "step": 162940
    },
    {
      "epoch": 46.25319330116378,
      "grad_norm": 0.464796245098114,
      "learning_rate": 5.376639227930741e-05,
      "loss": 0.0016161747276782989,
      "step": 162950
    },
    {
      "epoch": 46.25603179108714,
      "grad_norm": 2.556934356689453,
      "learning_rate": 5.376355378938405e-05,
      "loss": 0.0008942753076553345,
      "step": 162960
    },
    {
      "epoch": 46.2588702810105,
      "grad_norm": 2.0480096340179443,
      "learning_rate": 5.376071529946068e-05,
      "loss": 0.0023515654727816583,
      "step": 162970
    },
    {
      "epoch": 46.261708770933865,
      "grad_norm": 0.5556602478027344,
      "learning_rate": 5.375787680953733e-05,
      "loss": 0.0008019698783755302,
      "step": 162980
    },
    {
      "epoch": 46.26454726085722,
      "grad_norm": 0.1041545420885086,
      "learning_rate": 5.375503831961397e-05,
      "loss": 0.002244109660387039,
      "step": 162990
    },
    {
      "epoch": 46.26738575078058,
      "grad_norm": 0.537129819393158,
      "learning_rate": 5.3752199829690606e-05,
      "loss": 0.0029065165668725967,
      "step": 163000
    },
    {
      "epoch": 46.26738575078058,
      "eval_accuracy": 0.9783811279964393,
      "eval_loss": 0.07968826591968536,
      "eval_runtime": 31.8078,
      "eval_samples_per_second": 494.438,
      "eval_steps_per_second": 7.734,
      "step": 163000
    },
    {
      "epoch": 46.270224240703946,
      "grad_norm": 0.03095131181180477,
      "learning_rate": 5.374936133976725e-05,
      "loss": 0.0015403581783175468,
      "step": 163010
    },
    {
      "epoch": 46.27306273062731,
      "grad_norm": 0.07068592309951782,
      "learning_rate": 5.374652284984388e-05,
      "loss": 0.006108860298991203,
      "step": 163020
    },
    {
      "epoch": 46.275901220550665,
      "grad_norm": 0.10401742160320282,
      "learning_rate": 5.374368435992052e-05,
      "loss": 0.000548681989312172,
      "step": 163030
    },
    {
      "epoch": 46.27873971047403,
      "grad_norm": 0.027929771691560745,
      "learning_rate": 5.374084586999717e-05,
      "loss": 0.0009438948705792427,
      "step": 163040
    },
    {
      "epoch": 46.28157820039739,
      "grad_norm": 0.021428057923913002,
      "learning_rate": 5.3738007380073806e-05,
      "loss": 0.0013277532532811164,
      "step": 163050
    },
    {
      "epoch": 46.28441669032075,
      "grad_norm": 1.715158462524414,
      "learning_rate": 5.373516889015045e-05,
      "loss": 0.0009073581546545029,
      "step": 163060
    },
    {
      "epoch": 46.28725518024411,
      "grad_norm": 0.10850457847118378,
      "learning_rate": 5.373233040022708e-05,
      "loss": 0.0006232853978872299,
      "step": 163070
    },
    {
      "epoch": 46.29009367016747,
      "grad_norm": 0.15604525804519653,
      "learning_rate": 5.372949191030372e-05,
      "loss": 0.0010053314268589019,
      "step": 163080
    },
    {
      "epoch": 46.29293216009083,
      "grad_norm": 0.043750420212745667,
      "learning_rate": 5.372665342038036e-05,
      "loss": 0.0021579721942543983,
      "step": 163090
    },
    {
      "epoch": 46.29577065001419,
      "grad_norm": 0.018516860902309418,
      "learning_rate": 5.372381493045699e-05,
      "loss": 0.0006486643105745316,
      "step": 163100
    },
    {
      "epoch": 46.298609139937554,
      "grad_norm": 0.03702171519398689,
      "learning_rate": 5.372097644053364e-05,
      "loss": 0.0017677929252386093,
      "step": 163110
    },
    {
      "epoch": 46.30144762986092,
      "grad_norm": 0.5449747443199158,
      "learning_rate": 5.371813795061028e-05,
      "loss": 0.009665670990943908,
      "step": 163120
    },
    {
      "epoch": 46.30428611978427,
      "grad_norm": 1.1285896301269531,
      "learning_rate": 5.3715299460686916e-05,
      "loss": 0.00674062967300415,
      "step": 163130
    },
    {
      "epoch": 46.307124609707635,
      "grad_norm": 3.202997922897339,
      "learning_rate": 5.371246097076356e-05,
      "loss": 0.0023164553567767143,
      "step": 163140
    },
    {
      "epoch": 46.309963099631,
      "grad_norm": 0.017277976498007774,
      "learning_rate": 5.370962248084019e-05,
      "loss": 0.0006163859739899636,
      "step": 163150
    },
    {
      "epoch": 46.312801589554354,
      "grad_norm": 0.290584921836853,
      "learning_rate": 5.3706783990916834e-05,
      "loss": 0.00032588671892881395,
      "step": 163160
    },
    {
      "epoch": 46.31564007947772,
      "grad_norm": 0.07470045983791351,
      "learning_rate": 5.370394550099347e-05,
      "loss": 0.0010370852425694466,
      "step": 163170
    },
    {
      "epoch": 46.31847856940108,
      "grad_norm": 2.802698850631714,
      "learning_rate": 5.3701107011070117e-05,
      "loss": 0.0010930132120847702,
      "step": 163180
    },
    {
      "epoch": 46.32131705932444,
      "grad_norm": 0.18260355293750763,
      "learning_rate": 5.369826852114676e-05,
      "loss": 0.0005731813609600068,
      "step": 163190
    },
    {
      "epoch": 46.3241555492478,
      "grad_norm": 0.029815582558512688,
      "learning_rate": 5.369543003122339e-05,
      "loss": 0.0006133759394288063,
      "step": 163200
    },
    {
      "epoch": 46.32699403917116,
      "grad_norm": 0.18213213980197906,
      "learning_rate": 5.369259154130003e-05,
      "loss": 0.004297978058457374,
      "step": 163210
    },
    {
      "epoch": 46.329832529094524,
      "grad_norm": 0.08083415031433105,
      "learning_rate": 5.368975305137667e-05,
      "loss": 0.00445735901594162,
      "step": 163220
    },
    {
      "epoch": 46.33267101901788,
      "grad_norm": 0.04082489758729935,
      "learning_rate": 5.36869145614533e-05,
      "loss": 0.0039620637893676754,
      "step": 163230
    },
    {
      "epoch": 46.33550950894124,
      "grad_norm": 0.44436439871788025,
      "learning_rate": 5.368407607152995e-05,
      "loss": 0.002559477835893631,
      "step": 163240
    },
    {
      "epoch": 46.338347998864606,
      "grad_norm": 0.08053095638751984,
      "learning_rate": 5.368123758160659e-05,
      "loss": 0.0014695804566144943,
      "step": 163250
    },
    {
      "epoch": 46.34118648878796,
      "grad_norm": 1.4194035530090332,
      "learning_rate": 5.367839909168323e-05,
      "loss": 0.0059247203171253204,
      "step": 163260
    },
    {
      "epoch": 46.344024978711325,
      "grad_norm": 0.31859317421913147,
      "learning_rate": 5.367556060175987e-05,
      "loss": 0.006473459303379059,
      "step": 163270
    },
    {
      "epoch": 46.34686346863469,
      "grad_norm": 0.11668837070465088,
      "learning_rate": 5.36727221118365e-05,
      "loss": 0.0009562671184539795,
      "step": 163280
    },
    {
      "epoch": 46.34970195855805,
      "grad_norm": 0.509774386882782,
      "learning_rate": 5.3669883621913145e-05,
      "loss": 0.0011636849492788314,
      "step": 163290
    },
    {
      "epoch": 46.352540448481406,
      "grad_norm": 0.010050732642412186,
      "learning_rate": 5.366704513198978e-05,
      "loss": 0.0008292518556118012,
      "step": 163300
    },
    {
      "epoch": 46.35537893840477,
      "grad_norm": 0.3488582372665405,
      "learning_rate": 5.366420664206643e-05,
      "loss": 0.0004210282117128372,
      "step": 163310
    },
    {
      "epoch": 46.35821742832813,
      "grad_norm": 0.06465422362089157,
      "learning_rate": 5.366136815214307e-05,
      "loss": 0.004685015231370926,
      "step": 163320
    },
    {
      "epoch": 46.36105591825149,
      "grad_norm": 19.52214813232422,
      "learning_rate": 5.36585296622197e-05,
      "loss": 0.01181247979402542,
      "step": 163330
    },
    {
      "epoch": 46.36389440817485,
      "grad_norm": 0.03908555209636688,
      "learning_rate": 5.365569117229634e-05,
      "loss": 0.00047593656927347183,
      "step": 163340
    },
    {
      "epoch": 46.36673289809821,
      "grad_norm": 0.3469253480434418,
      "learning_rate": 5.365285268237298e-05,
      "loss": 0.003521581366658211,
      "step": 163350
    },
    {
      "epoch": 46.36957138802157,
      "grad_norm": 0.30431291460990906,
      "learning_rate": 5.3650014192449614e-05,
      "loss": 0.0016184480860829352,
      "step": 163360
    },
    {
      "epoch": 46.37240987794493,
      "grad_norm": 0.15744605660438538,
      "learning_rate": 5.3647175702526255e-05,
      "loss": 0.001879645511507988,
      "step": 163370
    },
    {
      "epoch": 46.375248367868295,
      "grad_norm": 0.34713560342788696,
      "learning_rate": 5.36443372126029e-05,
      "loss": 0.0005812043324112892,
      "step": 163380
    },
    {
      "epoch": 46.37808685779166,
      "grad_norm": 11.435036659240723,
      "learning_rate": 5.364149872267954e-05,
      "loss": 0.003355357050895691,
      "step": 163390
    },
    {
      "epoch": 46.380925347715014,
      "grad_norm": 0.15355592966079712,
      "learning_rate": 5.363866023275618e-05,
      "loss": 0.006638573855161667,
      "step": 163400
    },
    {
      "epoch": 46.38376383763838,
      "grad_norm": 0.030175425112247467,
      "learning_rate": 5.3635821742832814e-05,
      "loss": 0.0026629120111465454,
      "step": 163410
    },
    {
      "epoch": 46.38660232756174,
      "grad_norm": 0.12945377826690674,
      "learning_rate": 5.3632983252909455e-05,
      "loss": 0.011742984503507614,
      "step": 163420
    },
    {
      "epoch": 46.389440817485095,
      "grad_norm": 19.04966163635254,
      "learning_rate": 5.363014476298609e-05,
      "loss": 0.004785959422588348,
      "step": 163430
    },
    {
      "epoch": 46.39227930740846,
      "grad_norm": 0.2328379899263382,
      "learning_rate": 5.362730627306274e-05,
      "loss": 0.0027263203635811807,
      "step": 163440
    },
    {
      "epoch": 46.39511779733182,
      "grad_norm": 0.23741237819194794,
      "learning_rate": 5.362446778313937e-05,
      "loss": 0.00951758623123169,
      "step": 163450
    },
    {
      "epoch": 46.39795628725518,
      "grad_norm": 0.42345044016838074,
      "learning_rate": 5.3621629293216014e-05,
      "loss": 0.003709961473941803,
      "step": 163460
    },
    {
      "epoch": 46.40079477717854,
      "grad_norm": 2.548549175262451,
      "learning_rate": 5.361879080329265e-05,
      "loss": 0.0042207270860672,
      "step": 163470
    },
    {
      "epoch": 46.4036332671019,
      "grad_norm": 0.3839523196220398,
      "learning_rate": 5.361595231336929e-05,
      "loss": 0.014558973908424377,
      "step": 163480
    },
    {
      "epoch": 46.406471757025265,
      "grad_norm": 10.432478904724121,
      "learning_rate": 5.3613113823445925e-05,
      "loss": 0.014939486980438232,
      "step": 163490
    },
    {
      "epoch": 46.40931024694862,
      "grad_norm": 15.05435848236084,
      "learning_rate": 5.3610275333522566e-05,
      "loss": 0.01800643652677536,
      "step": 163500
    },
    {
      "epoch": 46.40931024694862,
      "eval_accuracy": 0.9713867870541107,
      "eval_loss": 0.11181638389825821,
      "eval_runtime": 32.6356,
      "eval_samples_per_second": 481.897,
      "eval_steps_per_second": 7.538,
      "step": 163500
    },
    {
      "epoch": 46.412148736871984,
      "grad_norm": 0.3410666286945343,
      "learning_rate": 5.3607436843599214e-05,
      "loss": 0.0025023983791470526,
      "step": 163510
    },
    {
      "epoch": 46.41498722679535,
      "grad_norm": 0.027016954496502876,
      "learning_rate": 5.360459835367585e-05,
      "loss": 0.006730303168296814,
      "step": 163520
    },
    {
      "epoch": 46.4178257167187,
      "grad_norm": 0.08643855899572372,
      "learning_rate": 5.360175986375249e-05,
      "loss": 0.005042828619480133,
      "step": 163530
    },
    {
      "epoch": 46.420664206642066,
      "grad_norm": 2.9451241493225098,
      "learning_rate": 5.3598921373829125e-05,
      "loss": 0.006969895213842392,
      "step": 163540
    },
    {
      "epoch": 46.42350269656543,
      "grad_norm": 1.2815172672271729,
      "learning_rate": 5.3596082883905766e-05,
      "loss": 0.0017025817185640335,
      "step": 163550
    },
    {
      "epoch": 46.42634118648879,
      "grad_norm": 0.1149652898311615,
      "learning_rate": 5.35932443939824e-05,
      "loss": 0.0007763613015413284,
      "step": 163560
    },
    {
      "epoch": 46.42917967641215,
      "grad_norm": 0.11142764985561371,
      "learning_rate": 5.3590405904059035e-05,
      "loss": 0.0011886388063430786,
      "step": 163570
    },
    {
      "epoch": 46.43201816633551,
      "grad_norm": 5.675490379333496,
      "learning_rate": 5.3587567414135683e-05,
      "loss": 0.006672661006450653,
      "step": 163580
    },
    {
      "epoch": 46.43485665625887,
      "grad_norm": 4.5757575035095215,
      "learning_rate": 5.3584728924212325e-05,
      "loss": 0.009459036588668823,
      "step": 163590
    },
    {
      "epoch": 46.43769514618223,
      "grad_norm": 14.7116117477417,
      "learning_rate": 5.358189043428896e-05,
      "loss": 0.013829952478408814,
      "step": 163600
    },
    {
      "epoch": 46.44053363610559,
      "grad_norm": 8.242639541625977,
      "learning_rate": 5.35790519443656e-05,
      "loss": 0.003092874214053154,
      "step": 163610
    },
    {
      "epoch": 46.443372126028954,
      "grad_norm": 0.279328316450119,
      "learning_rate": 5.3576213454442235e-05,
      "loss": 0.00081463772803545,
      "step": 163620
    },
    {
      "epoch": 46.44621061595231,
      "grad_norm": 1.7517021894454956,
      "learning_rate": 5.357337496451888e-05,
      "loss": 0.003731635957956314,
      "step": 163630
    },
    {
      "epoch": 46.44904910587567,
      "grad_norm": 0.1343434602022171,
      "learning_rate": 5.3570536474595525e-05,
      "loss": 0.001091160997748375,
      "step": 163640
    },
    {
      "epoch": 46.451887595799036,
      "grad_norm": 0.10400504618883133,
      "learning_rate": 5.356769798467216e-05,
      "loss": 0.0029919516295194627,
      "step": 163650
    },
    {
      "epoch": 46.4547260857224,
      "grad_norm": 0.03137136250734329,
      "learning_rate": 5.35648594947488e-05,
      "loss": 0.005298639088869095,
      "step": 163660
    },
    {
      "epoch": 46.457564575645755,
      "grad_norm": 0.14458556473255157,
      "learning_rate": 5.3562021004825435e-05,
      "loss": 0.0039064474403858185,
      "step": 163670
    },
    {
      "epoch": 46.46040306556912,
      "grad_norm": 0.31425708532333374,
      "learning_rate": 5.355918251490207e-05,
      "loss": 0.004829246550798416,
      "step": 163680
    },
    {
      "epoch": 46.46324155549248,
      "grad_norm": 0.7644392848014832,
      "learning_rate": 5.355634402497871e-05,
      "loss": 0.003656395524740219,
      "step": 163690
    },
    {
      "epoch": 46.466080045415836,
      "grad_norm": 0.842171847820282,
      "learning_rate": 5.3553505535055346e-05,
      "loss": 0.015391390025615691,
      "step": 163700
    },
    {
      "epoch": 46.4689185353392,
      "grad_norm": 5.345112323760986,
      "learning_rate": 5.3550667045131994e-05,
      "loss": 0.009842721372842788,
      "step": 163710
    },
    {
      "epoch": 46.47175702526256,
      "grad_norm": 0.7716372609138489,
      "learning_rate": 5.3547828555208636e-05,
      "loss": 0.002174505777657032,
      "step": 163720
    },
    {
      "epoch": 46.47459551518592,
      "grad_norm": 2.6899378299713135,
      "learning_rate": 5.354499006528527e-05,
      "loss": 0.012052251398563385,
      "step": 163730
    },
    {
      "epoch": 46.47743400510928,
      "grad_norm": 0.8180320858955383,
      "learning_rate": 5.354215157536191e-05,
      "loss": 0.0018224529922008514,
      "step": 163740
    },
    {
      "epoch": 46.480272495032644,
      "grad_norm": 0.09394227713346481,
      "learning_rate": 5.3539313085438546e-05,
      "loss": 0.0018266221508383752,
      "step": 163750
    },
    {
      "epoch": 46.48311098495601,
      "grad_norm": 0.5276504158973694,
      "learning_rate": 5.353647459551519e-05,
      "loss": 0.0030203510075807573,
      "step": 163760
    },
    {
      "epoch": 46.48594947487936,
      "grad_norm": 0.046393219381570816,
      "learning_rate": 5.353363610559182e-05,
      "loss": 0.0018759462982416153,
      "step": 163770
    },
    {
      "epoch": 46.488787964802725,
      "grad_norm": 0.24885918200016022,
      "learning_rate": 5.353079761566847e-05,
      "loss": 0.004050688445568084,
      "step": 163780
    },
    {
      "epoch": 46.49162645472609,
      "grad_norm": 0.9437452554702759,
      "learning_rate": 5.352795912574511e-05,
      "loss": 0.0016717275604605675,
      "step": 163790
    },
    {
      "epoch": 46.494464944649444,
      "grad_norm": 0.5857776403427124,
      "learning_rate": 5.3525120635821746e-05,
      "loss": 0.0010287323966622353,
      "step": 163800
    },
    {
      "epoch": 46.49730343457281,
      "grad_norm": 0.07450392097234726,
      "learning_rate": 5.352228214589838e-05,
      "loss": 0.017902839183807372,
      "step": 163810
    },
    {
      "epoch": 46.50014192449617,
      "grad_norm": 4.893764972686768,
      "learning_rate": 5.351944365597502e-05,
      "loss": 0.002689189091324806,
      "step": 163820
    },
    {
      "epoch": 46.502980414419525,
      "grad_norm": 0.6912021040916443,
      "learning_rate": 5.351660516605166e-05,
      "loss": 0.0005746757611632347,
      "step": 163830
    },
    {
      "epoch": 46.50581890434289,
      "grad_norm": 2.7230052947998047,
      "learning_rate": 5.3513766676128305e-05,
      "loss": 0.0014847451820969582,
      "step": 163840
    },
    {
      "epoch": 46.50865739426625,
      "grad_norm": 0.22354038059711456,
      "learning_rate": 5.3510928186204946e-05,
      "loss": 0.0010698890313506127,
      "step": 163850
    },
    {
      "epoch": 46.511495884189614,
      "grad_norm": 4.0255513191223145,
      "learning_rate": 5.350808969628158e-05,
      "loss": 0.0024413086473941805,
      "step": 163860
    },
    {
      "epoch": 46.51433437411297,
      "grad_norm": 0.04058212414383888,
      "learning_rate": 5.350525120635822e-05,
      "loss": 0.004021137580275535,
      "step": 163870
    },
    {
      "epoch": 46.51717286403633,
      "grad_norm": 0.2017292082309723,
      "learning_rate": 5.350241271643486e-05,
      "loss": 0.0007887685671448708,
      "step": 163880
    },
    {
      "epoch": 46.520011353959696,
      "grad_norm": 0.8076528310775757,
      "learning_rate": 5.34995742265115e-05,
      "loss": 0.001407521776854992,
      "step": 163890
    },
    {
      "epoch": 46.52284984388305,
      "grad_norm": 0.5083214044570923,
      "learning_rate": 5.349673573658813e-05,
      "loss": 0.0022427033632993697,
      "step": 163900
    },
    {
      "epoch": 46.525688333806414,
      "grad_norm": 1.9543944597244263,
      "learning_rate": 5.349389724666478e-05,
      "loss": 0.0015397829934954642,
      "step": 163910
    },
    {
      "epoch": 46.52852682372978,
      "grad_norm": 0.08514902740716934,
      "learning_rate": 5.3491058756741416e-05,
      "loss": 0.0005365850403904915,
      "step": 163920
    },
    {
      "epoch": 46.53136531365314,
      "grad_norm": 0.2308827042579651,
      "learning_rate": 5.348822026681806e-05,
      "loss": 0.00125583503395319,
      "step": 163930
    },
    {
      "epoch": 46.534203803576496,
      "grad_norm": 0.056419603526592255,
      "learning_rate": 5.348538177689469e-05,
      "loss": 0.003990138694643974,
      "step": 163940
    },
    {
      "epoch": 46.53704229349986,
      "grad_norm": 0.689139187335968,
      "learning_rate": 5.348254328697133e-05,
      "loss": 0.0010517885908484458,
      "step": 163950
    },
    {
      "epoch": 46.53988078342322,
      "grad_norm": 8.854610443115234,
      "learning_rate": 5.347970479704797e-05,
      "loss": 0.0038975954055786135,
      "step": 163960
    },
    {
      "epoch": 46.54271927334658,
      "grad_norm": 0.07472768425941467,
      "learning_rate": 5.347686630712461e-05,
      "loss": 0.0012585034593939782,
      "step": 163970
    },
    {
      "epoch": 46.54555776326994,
      "grad_norm": 0.1801338791847229,
      "learning_rate": 5.347402781720126e-05,
      "loss": 0.01476655900478363,
      "step": 163980
    },
    {
      "epoch": 46.5483962531933,
      "grad_norm": 0.3647598922252655,
      "learning_rate": 5.347118932727789e-05,
      "loss": 0.0034795574843883514,
      "step": 163990
    },
    {
      "epoch": 46.55123474311666,
      "grad_norm": 20.6460018157959,
      "learning_rate": 5.346835083735453e-05,
      "loss": 0.0075932078063488005,
      "step": 164000
    },
    {
      "epoch": 46.55123474311666,
      "eval_accuracy": 0.9720862211483436,
      "eval_loss": 0.09905950725078583,
      "eval_runtime": 33.0849,
      "eval_samples_per_second": 475.353,
      "eval_steps_per_second": 7.435,
      "step": 164000
    },
    {
      "epoch": 46.55407323304002,
      "grad_norm": 15.626503944396973,
      "learning_rate": 5.346551234743117e-05,
      "loss": 0.004081713408231736,
      "step": 164010
    },
    {
      "epoch": 46.556911722963385,
      "grad_norm": 0.44443973898887634,
      "learning_rate": 5.34626738575078e-05,
      "loss": 0.002146083302795887,
      "step": 164020
    },
    {
      "epoch": 46.55975021288675,
      "grad_norm": 0.24075619876384735,
      "learning_rate": 5.3459835367584444e-05,
      "loss": 0.004924489557743073,
      "step": 164030
    },
    {
      "epoch": 46.5625887028101,
      "grad_norm": 0.043185025453567505,
      "learning_rate": 5.345699687766109e-05,
      "loss": 0.00041001234203577044,
      "step": 164040
    },
    {
      "epoch": 46.565427192733466,
      "grad_norm": 0.2908719778060913,
      "learning_rate": 5.3454158387737726e-05,
      "loss": 0.0038782693445682526,
      "step": 164050
    },
    {
      "epoch": 46.56826568265683,
      "grad_norm": 0.32294660806655884,
      "learning_rate": 5.345131989781437e-05,
      "loss": 0.0010839059948921203,
      "step": 164060
    },
    {
      "epoch": 46.571104172580185,
      "grad_norm": 1.0365335941314697,
      "learning_rate": 5.3448481407891e-05,
      "loss": 0.0026176033541560175,
      "step": 164070
    },
    {
      "epoch": 46.57394266250355,
      "grad_norm": 18.542984008789062,
      "learning_rate": 5.3445642917967644e-05,
      "loss": 0.018523530662059785,
      "step": 164080
    },
    {
      "epoch": 46.57678115242691,
      "grad_norm": 0.12884201109409332,
      "learning_rate": 5.344280442804428e-05,
      "loss": 0.0008348416537046433,
      "step": 164090
    },
    {
      "epoch": 46.579619642350266,
      "grad_norm": 0.21558894217014313,
      "learning_rate": 5.343996593812092e-05,
      "loss": 0.0011558257043361663,
      "step": 164100
    },
    {
      "epoch": 46.58245813227363,
      "grad_norm": 1.712548017501831,
      "learning_rate": 5.343712744819757e-05,
      "loss": 0.0010267585515975953,
      "step": 164110
    },
    {
      "epoch": 46.58529662219699,
      "grad_norm": 1.4514437913894653,
      "learning_rate": 5.34342889582742e-05,
      "loss": 0.003362574800848961,
      "step": 164120
    },
    {
      "epoch": 46.588135112120355,
      "grad_norm": 3.3931257724761963,
      "learning_rate": 5.3431450468350844e-05,
      "loss": 0.0011229487136006355,
      "step": 164130
    },
    {
      "epoch": 46.59097360204371,
      "grad_norm": 0.011247451417148113,
      "learning_rate": 5.342861197842748e-05,
      "loss": 0.0004738014191389084,
      "step": 164140
    },
    {
      "epoch": 46.593812091967074,
      "grad_norm": 0.6923418045043945,
      "learning_rate": 5.342577348850411e-05,
      "loss": 0.002503395453095436,
      "step": 164150
    },
    {
      "epoch": 46.59665058189044,
      "grad_norm": 0.5976182818412781,
      "learning_rate": 5.3422934998580754e-05,
      "loss": 0.0006764981895685196,
      "step": 164160
    },
    {
      "epoch": 46.59948907181379,
      "grad_norm": 1.8433115482330322,
      "learning_rate": 5.34200965086574e-05,
      "loss": 0.0014811884611845016,
      "step": 164170
    },
    {
      "epoch": 46.602327561737155,
      "grad_norm": 0.07827173918485641,
      "learning_rate": 5.341725801873404e-05,
      "loss": 0.0009638853371143341,
      "step": 164180
    },
    {
      "epoch": 46.60516605166052,
      "grad_norm": 1.1586744785308838,
      "learning_rate": 5.341441952881068e-05,
      "loss": 0.0013540873304009437,
      "step": 164190
    },
    {
      "epoch": 46.608004541583874,
      "grad_norm": 10.735965728759766,
      "learning_rate": 5.341158103888731e-05,
      "loss": 0.0033691756427288054,
      "step": 164200
    },
    {
      "epoch": 46.61084303150724,
      "grad_norm": 0.15648287534713745,
      "learning_rate": 5.3408742548963955e-05,
      "loss": 0.004374609887599945,
      "step": 164210
    },
    {
      "epoch": 46.6136815214306,
      "grad_norm": 0.17912697792053223,
      "learning_rate": 5.340590405904059e-05,
      "loss": 0.005459198355674743,
      "step": 164220
    },
    {
      "epoch": 46.61652001135396,
      "grad_norm": 1.06454598903656,
      "learning_rate": 5.340306556911723e-05,
      "loss": 0.000906112976372242,
      "step": 164230
    },
    {
      "epoch": 46.61935850127732,
      "grad_norm": 0.04322906956076622,
      "learning_rate": 5.340022707919388e-05,
      "loss": 0.005678822100162506,
      "step": 164240
    },
    {
      "epoch": 46.62219699120068,
      "grad_norm": 4.784919738769531,
      "learning_rate": 5.339738858927051e-05,
      "loss": 0.0025967231020331384,
      "step": 164250
    },
    {
      "epoch": 46.625035481124044,
      "grad_norm": 0.6090016961097717,
      "learning_rate": 5.3394550099347155e-05,
      "loss": 0.0023083219304680823,
      "step": 164260
    },
    {
      "epoch": 46.6278739710474,
      "grad_norm": 0.07309899479150772,
      "learning_rate": 5.339171160942379e-05,
      "loss": 0.0012869982048869133,
      "step": 164270
    },
    {
      "epoch": 46.63071246097076,
      "grad_norm": 13.747698783874512,
      "learning_rate": 5.3388873119500424e-05,
      "loss": 0.003963427990674973,
      "step": 164280
    },
    {
      "epoch": 46.633550950894126,
      "grad_norm": 0.43263405561447144,
      "learning_rate": 5.3386034629577065e-05,
      "loss": 0.000798116996884346,
      "step": 164290
    },
    {
      "epoch": 46.63638944081748,
      "grad_norm": 0.04291701316833496,
      "learning_rate": 5.33831961396537e-05,
      "loss": 0.002957405708730221,
      "step": 164300
    },
    {
      "epoch": 46.639227930740844,
      "grad_norm": 0.3192860782146454,
      "learning_rate": 5.338035764973035e-05,
      "loss": 0.0012529529631137848,
      "step": 164310
    },
    {
      "epoch": 46.64206642066421,
      "grad_norm": 4.044543743133545,
      "learning_rate": 5.337751915980699e-05,
      "loss": 0.008240612596273423,
      "step": 164320
    },
    {
      "epoch": 46.64490491058757,
      "grad_norm": 0.056108467280864716,
      "learning_rate": 5.3374680669883624e-05,
      "loss": 0.009401299059391022,
      "step": 164330
    },
    {
      "epoch": 46.647743400510926,
      "grad_norm": 0.03932775557041168,
      "learning_rate": 5.3371842179960265e-05,
      "loss": 0.0019448256120085717,
      "step": 164340
    },
    {
      "epoch": 46.65058189043429,
      "grad_norm": 5.225566864013672,
      "learning_rate": 5.33690036900369e-05,
      "loss": 0.0057066213339567184,
      "step": 164350
    },
    {
      "epoch": 46.65342038035765,
      "grad_norm": 0.18549293279647827,
      "learning_rate": 5.336616520011354e-05,
      "loss": 0.0005222808569669724,
      "step": 164360
    },
    {
      "epoch": 46.65625887028101,
      "grad_norm": 3.60960054397583,
      "learning_rate": 5.336332671019019e-05,
      "loss": 0.004709808900952339,
      "step": 164370
    },
    {
      "epoch": 46.65909736020437,
      "grad_norm": 0.08255498856306076,
      "learning_rate": 5.3360488220266824e-05,
      "loss": 0.0035531751811504362,
      "step": 164380
    },
    {
      "epoch": 46.66193585012773,
      "grad_norm": 0.0329025499522686,
      "learning_rate": 5.335764973034346e-05,
      "loss": 0.002903095260262489,
      "step": 164390
    },
    {
      "epoch": 46.664774340051096,
      "grad_norm": 0.4546467959880829,
      "learning_rate": 5.33548112404201e-05,
      "loss": 0.0012167034670710563,
      "step": 164400
    },
    {
      "epoch": 46.66761282997445,
      "grad_norm": 0.17145900428295135,
      "learning_rate": 5.3351972750496735e-05,
      "loss": 0.0014900915324687959,
      "step": 164410
    },
    {
      "epoch": 46.670451319897815,
      "grad_norm": 0.04574727639555931,
      "learning_rate": 5.3349134260573376e-05,
      "loss": 0.0003438059240579605,
      "step": 164420
    },
    {
      "epoch": 46.67328980982118,
      "grad_norm": 0.017299553379416466,
      "learning_rate": 5.334629577065001e-05,
      "loss": 0.00023807957768440247,
      "step": 164430
    },
    {
      "epoch": 46.67612829974453,
      "grad_norm": 0.0297741387039423,
      "learning_rate": 5.334345728072666e-05,
      "loss": 0.0004010375589132309,
      "step": 164440
    },
    {
      "epoch": 46.678966789667896,
      "grad_norm": 0.10284103453159332,
      "learning_rate": 5.33406187908033e-05,
      "loss": 0.0018596142530441285,
      "step": 164450
    },
    {
      "epoch": 46.68180527959126,
      "grad_norm": 0.10028381645679474,
      "learning_rate": 5.3337780300879935e-05,
      "loss": 0.0006998443976044654,
      "step": 164460
    },
    {
      "epoch": 46.684643769514615,
      "grad_norm": 0.1154487356543541,
      "learning_rate": 5.3334941810956576e-05,
      "loss": 0.000403880700469017,
      "step": 164470
    },
    {
      "epoch": 46.68748225943798,
      "grad_norm": 0.13004150986671448,
      "learning_rate": 5.333210332103321e-05,
      "loss": 0.001231623813509941,
      "step": 164480
    },
    {
      "epoch": 46.69032074936134,
      "grad_norm": 0.22551992535591125,
      "learning_rate": 5.3329264831109845e-05,
      "loss": 0.0038978207856416704,
      "step": 164490
    },
    {
      "epoch": 46.693159239284704,
      "grad_norm": 0.005688790697604418,
      "learning_rate": 5.332642634118649e-05,
      "loss": 0.0018117938190698624,
      "step": 164500
    },
    {
      "epoch": 46.693159239284704,
      "eval_accuracy": 0.9773001843962612,
      "eval_loss": 0.08091847598552704,
      "eval_runtime": 32.637,
      "eval_samples_per_second": 481.877,
      "eval_steps_per_second": 7.537,
      "step": 164500
    },
    {
      "epoch": 46.69599772920806,
      "grad_norm": 0.1395488828420639,
      "learning_rate": 5.3323587851263135e-05,
      "loss": 0.0003485584631562233,
      "step": 164510
    },
    {
      "epoch": 46.69883621913142,
      "grad_norm": 0.15540945529937744,
      "learning_rate": 5.332074936133977e-05,
      "loss": 0.000623115710914135,
      "step": 164520
    },
    {
      "epoch": 46.701674709054785,
      "grad_norm": 0.6340265274047852,
      "learning_rate": 5.331791087141641e-05,
      "loss": 0.0016589660197496414,
      "step": 164530
    },
    {
      "epoch": 46.70451319897814,
      "grad_norm": 0.0849124863743782,
      "learning_rate": 5.3315072381493045e-05,
      "loss": 0.0036720722913742066,
      "step": 164540
    },
    {
      "epoch": 46.707351688901504,
      "grad_norm": 0.024355188012123108,
      "learning_rate": 5.331223389156969e-05,
      "loss": 0.000666770339012146,
      "step": 164550
    },
    {
      "epoch": 46.71019017882487,
      "grad_norm": 0.18630103766918182,
      "learning_rate": 5.330939540164632e-05,
      "loss": 0.0010414257645606994,
      "step": 164560
    },
    {
      "epoch": 46.71302866874822,
      "grad_norm": 0.016069777309894562,
      "learning_rate": 5.330655691172297e-05,
      "loss": 0.0012613140046596528,
      "step": 164570
    },
    {
      "epoch": 46.715867158671585,
      "grad_norm": 0.33277541399002075,
      "learning_rate": 5.330371842179961e-05,
      "loss": 0.0005870601162314415,
      "step": 164580
    },
    {
      "epoch": 46.71870564859495,
      "grad_norm": 0.2891557812690735,
      "learning_rate": 5.3300879931876245e-05,
      "loss": 0.0014350913465023042,
      "step": 164590
    },
    {
      "epoch": 46.72154413851831,
      "grad_norm": 0.150354266166687,
      "learning_rate": 5.329804144195289e-05,
      "loss": 0.0062258407473564144,
      "step": 164600
    },
    {
      "epoch": 46.72438262844167,
      "grad_norm": 0.043504498898983,
      "learning_rate": 5.329520295202952e-05,
      "loss": 0.0022360702976584436,
      "step": 164610
    },
    {
      "epoch": 46.72722111836503,
      "grad_norm": 0.6093674898147583,
      "learning_rate": 5.3292364462106156e-05,
      "loss": 0.0012466447427868844,
      "step": 164620
    },
    {
      "epoch": 46.73005960828839,
      "grad_norm": 1.6744266748428345,
      "learning_rate": 5.32895259721828e-05,
      "loss": 0.0014390576630830764,
      "step": 164630
    },
    {
      "epoch": 46.73289809821175,
      "grad_norm": 6.413366794586182,
      "learning_rate": 5.3286687482259446e-05,
      "loss": 0.012248972058296203,
      "step": 164640
    },
    {
      "epoch": 46.73573658813511,
      "grad_norm": 0.3209405839443207,
      "learning_rate": 5.328384899233608e-05,
      "loss": 0.0032893434166908265,
      "step": 164650
    },
    {
      "epoch": 46.738575078058474,
      "grad_norm": 0.20171353220939636,
      "learning_rate": 5.328101050241272e-05,
      "loss": 0.004455301910638809,
      "step": 164660
    },
    {
      "epoch": 46.74141356798184,
      "grad_norm": 0.2415858954191208,
      "learning_rate": 5.3278172012489356e-05,
      "loss": 0.002097412571310997,
      "step": 164670
    },
    {
      "epoch": 46.74425205790519,
      "grad_norm": 4.924857139587402,
      "learning_rate": 5.3275333522566e-05,
      "loss": 0.00393880233168602,
      "step": 164680
    },
    {
      "epoch": 46.747090547828556,
      "grad_norm": 0.23658417165279388,
      "learning_rate": 5.327249503264263e-05,
      "loss": 0.0024287857115268706,
      "step": 164690
    },
    {
      "epoch": 46.74992903775192,
      "grad_norm": 0.33361124992370605,
      "learning_rate": 5.3269656542719273e-05,
      "loss": 0.0012161394581198693,
      "step": 164700
    },
    {
      "epoch": 46.752767527675275,
      "grad_norm": 0.022411111742258072,
      "learning_rate": 5.326681805279592e-05,
      "loss": 0.0011892318725585938,
      "step": 164710
    },
    {
      "epoch": 46.75560601759864,
      "grad_norm": 0.10893318802118301,
      "learning_rate": 5.3263979562872556e-05,
      "loss": 0.007594121247529983,
      "step": 164720
    },
    {
      "epoch": 46.758444507522,
      "grad_norm": 0.04642578586935997,
      "learning_rate": 5.32611410729492e-05,
      "loss": 0.000713401846587658,
      "step": 164730
    },
    {
      "epoch": 46.761282997445356,
      "grad_norm": 0.07465706765651703,
      "learning_rate": 5.325830258302583e-05,
      "loss": 0.002715184725821018,
      "step": 164740
    },
    {
      "epoch": 46.76412148736872,
      "grad_norm": 0.10463391989469528,
      "learning_rate": 5.325546409310247e-05,
      "loss": 0.0007320273667573929,
      "step": 164750
    },
    {
      "epoch": 46.76695997729208,
      "grad_norm": 0.07132776826620102,
      "learning_rate": 5.325262560317911e-05,
      "loss": 0.0007017970085144043,
      "step": 164760
    },
    {
      "epoch": 46.769798467215445,
      "grad_norm": 0.8641230463981628,
      "learning_rate": 5.3249787113255756e-05,
      "loss": 0.0004623936489224434,
      "step": 164770
    },
    {
      "epoch": 46.7726369571388,
      "grad_norm": 13.238468170166016,
      "learning_rate": 5.324694862333239e-05,
      "loss": 0.002437843568623066,
      "step": 164780
    },
    {
      "epoch": 46.77547544706216,
      "grad_norm": 0.04031866788864136,
      "learning_rate": 5.324411013340903e-05,
      "loss": 0.0026259573176503183,
      "step": 164790
    },
    {
      "epoch": 46.778313936985526,
      "grad_norm": 0.23712824285030365,
      "learning_rate": 5.324127164348567e-05,
      "loss": 0.001870042458176613,
      "step": 164800
    },
    {
      "epoch": 46.78115242690888,
      "grad_norm": 0.018666600808501244,
      "learning_rate": 5.323843315356231e-05,
      "loss": 0.0005648083984851837,
      "step": 164810
    },
    {
      "epoch": 46.783990916832245,
      "grad_norm": 0.12453712522983551,
      "learning_rate": 5.323559466363894e-05,
      "loss": 0.011680512875318527,
      "step": 164820
    },
    {
      "epoch": 46.78682940675561,
      "grad_norm": 0.39277002215385437,
      "learning_rate": 5.3232756173715584e-05,
      "loss": 0.0030176300555467606,
      "step": 164830
    },
    {
      "epoch": 46.789667896678964,
      "grad_norm": 0.21137875318527222,
      "learning_rate": 5.322991768379223e-05,
      "loss": 0.0030567942187190054,
      "step": 164840
    },
    {
      "epoch": 46.79250638660233,
      "grad_norm": 0.8106091022491455,
      "learning_rate": 5.322707919386887e-05,
      "loss": 0.00247596874833107,
      "step": 164850
    },
    {
      "epoch": 46.79534487652569,
      "grad_norm": 4.175459384918213,
      "learning_rate": 5.32242407039455e-05,
      "loss": 0.008488701283931732,
      "step": 164860
    },
    {
      "epoch": 46.79818336644905,
      "grad_norm": 0.5828638076782227,
      "learning_rate": 5.322140221402214e-05,
      "loss": 0.001956317760050297,
      "step": 164870
    },
    {
      "epoch": 46.80102185637241,
      "grad_norm": 0.05157916992902756,
      "learning_rate": 5.321856372409878e-05,
      "loss": 0.0005281424149870872,
      "step": 164880
    },
    {
      "epoch": 46.80386034629577,
      "grad_norm": 0.06228640303015709,
      "learning_rate": 5.321572523417542e-05,
      "loss": 0.0008106857538223266,
      "step": 164890
    },
    {
      "epoch": 46.806698836219134,
      "grad_norm": 0.13045260310173035,
      "learning_rate": 5.3212886744252054e-05,
      "loss": 0.0008294342085719108,
      "step": 164900
    },
    {
      "epoch": 46.80953732614249,
      "grad_norm": 0.013395009562373161,
      "learning_rate": 5.32100482543287e-05,
      "loss": 0.005664100125432015,
      "step": 164910
    },
    {
      "epoch": 46.81237581606585,
      "grad_norm": 0.06896400451660156,
      "learning_rate": 5.320720976440534e-05,
      "loss": 0.0004715407267212868,
      "step": 164920
    },
    {
      "epoch": 46.815214305989215,
      "grad_norm": 0.8006630539894104,
      "learning_rate": 5.320437127448198e-05,
      "loss": 0.004637760668992996,
      "step": 164930
    },
    {
      "epoch": 46.81805279591257,
      "grad_norm": 0.1953216791152954,
      "learning_rate": 5.320153278455862e-05,
      "loss": 0.0015667589381337167,
      "step": 164940
    },
    {
      "epoch": 46.820891285835934,
      "grad_norm": 0.45883411169052124,
      "learning_rate": 5.3198694294635254e-05,
      "loss": 0.0017401978373527526,
      "step": 164950
    },
    {
      "epoch": 46.8237297757593,
      "grad_norm": 0.18919387459754944,
      "learning_rate": 5.319585580471189e-05,
      "loss": 0.0006857763975858689,
      "step": 164960
    },
    {
      "epoch": 46.82656826568266,
      "grad_norm": 0.7087239027023315,
      "learning_rate": 5.319301731478854e-05,
      "loss": 0.0009487476199865341,
      "step": 164970
    },
    {
      "epoch": 46.829406755606016,
      "grad_norm": 0.10814406722784042,
      "learning_rate": 5.319017882486518e-05,
      "loss": 0.00575784370303154,
      "step": 164980
    },
    {
      "epoch": 46.83224524552938,
      "grad_norm": 0.016473762691020966,
      "learning_rate": 5.318734033494181e-05,
      "loss": 0.002390122599899769,
      "step": 164990
    },
    {
      "epoch": 46.83508373545274,
      "grad_norm": 0.04235539957880974,
      "learning_rate": 5.3184501845018454e-05,
      "loss": 0.00070070531219244,
      "step": 165000
    },
    {
      "epoch": 46.83508373545274,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.08768175542354584,
      "eval_runtime": 32.5308,
      "eval_samples_per_second": 483.45,
      "eval_steps_per_second": 7.562,
      "step": 165000
    },
    {
      "epoch": 46.8379222253761,
      "grad_norm": 0.05907853692770004,
      "learning_rate": 5.318166335509509e-05,
      "loss": 0.0013150973245501518,
      "step": 165010
    },
    {
      "epoch": 46.84076071529946,
      "grad_norm": 0.06244242563843727,
      "learning_rate": 5.317882486517173e-05,
      "loss": 0.00194683987647295,
      "step": 165020
    },
    {
      "epoch": 46.84359920522282,
      "grad_norm": 0.042970847338438034,
      "learning_rate": 5.3175986375248364e-05,
      "loss": 0.0011794626712799071,
      "step": 165030
    },
    {
      "epoch": 46.84643769514618,
      "grad_norm": 0.12341023236513138,
      "learning_rate": 5.317314788532501e-05,
      "loss": 0.0035207368433475494,
      "step": 165040
    },
    {
      "epoch": 46.84927618506954,
      "grad_norm": 1.291841745376587,
      "learning_rate": 5.3170309395401654e-05,
      "loss": 0.0012865677475929261,
      "step": 165050
    },
    {
      "epoch": 46.852114674992904,
      "grad_norm": 6.701313495635986,
      "learning_rate": 5.316747090547829e-05,
      "loss": 0.005393410101532936,
      "step": 165060
    },
    {
      "epoch": 46.85495316491627,
      "grad_norm": 0.21338629722595215,
      "learning_rate": 5.316463241555493e-05,
      "loss": 0.0015699679031968116,
      "step": 165070
    },
    {
      "epoch": 46.85779165483962,
      "grad_norm": 0.08192148059606552,
      "learning_rate": 5.3161793925631564e-05,
      "loss": 0.0016354961320757867,
      "step": 165080
    },
    {
      "epoch": 46.860630144762986,
      "grad_norm": 0.051396988332271576,
      "learning_rate": 5.31589554357082e-05,
      "loss": 0.0030400993302464483,
      "step": 165090
    },
    {
      "epoch": 46.86346863468635,
      "grad_norm": 0.16242475807666779,
      "learning_rate": 5.315611694578484e-05,
      "loss": 0.0008845651522278785,
      "step": 165100
    },
    {
      "epoch": 46.866307124609705,
      "grad_norm": 0.41115468740463257,
      "learning_rate": 5.315327845586149e-05,
      "loss": 0.0019312728196382522,
      "step": 165110
    },
    {
      "epoch": 46.86914561453307,
      "grad_norm": 0.9477577209472656,
      "learning_rate": 5.315043996593812e-05,
      "loss": 0.0011264583095908164,
      "step": 165120
    },
    {
      "epoch": 46.87198410445643,
      "grad_norm": 0.026373257860541344,
      "learning_rate": 5.3147601476014764e-05,
      "loss": 0.0011991256847977637,
      "step": 165130
    },
    {
      "epoch": 46.87482259437979,
      "grad_norm": 16.66499900817871,
      "learning_rate": 5.31447629860914e-05,
      "loss": 0.007589232921600342,
      "step": 165140
    },
    {
      "epoch": 46.87766108430315,
      "grad_norm": 0.41995543241500854,
      "learning_rate": 5.314192449616804e-05,
      "loss": 0.001023968867957592,
      "step": 165150
    },
    {
      "epoch": 46.88049957422651,
      "grad_norm": 0.7740252017974854,
      "learning_rate": 5.3139086006244675e-05,
      "loss": 0.00249041635543108,
      "step": 165160
    },
    {
      "epoch": 46.883338064149875,
      "grad_norm": 0.07213396579027176,
      "learning_rate": 5.313624751632132e-05,
      "loss": 0.000631587952375412,
      "step": 165170
    },
    {
      "epoch": 46.88617655407323,
      "grad_norm": 0.14581839740276337,
      "learning_rate": 5.3133409026397965e-05,
      "loss": 0.0009664466604590416,
      "step": 165180
    },
    {
      "epoch": 46.889015043996594,
      "grad_norm": 9.587495803833008,
      "learning_rate": 5.31305705364746e-05,
      "loss": 0.0036456815898418425,
      "step": 165190
    },
    {
      "epoch": 46.89185353391996,
      "grad_norm": 1.0758684873580933,
      "learning_rate": 5.312773204655124e-05,
      "loss": 0.003959966450929641,
      "step": 165200
    },
    {
      "epoch": 46.89469202384331,
      "grad_norm": 0.6814945340156555,
      "learning_rate": 5.3124893556627875e-05,
      "loss": 0.0008657170459628105,
      "step": 165210
    },
    {
      "epoch": 46.897530513766675,
      "grad_norm": 0.02936691604554653,
      "learning_rate": 5.312205506670451e-05,
      "loss": 0.0010898422449827194,
      "step": 165220
    },
    {
      "epoch": 46.90036900369004,
      "grad_norm": 0.2673785388469696,
      "learning_rate": 5.311921657678115e-05,
      "loss": 0.0010003086179494857,
      "step": 165230
    },
    {
      "epoch": 46.9032074936134,
      "grad_norm": 0.7853323221206665,
      "learning_rate": 5.31163780868578e-05,
      "loss": 0.0013068223372101784,
      "step": 165240
    },
    {
      "epoch": 46.90604598353676,
      "grad_norm": 0.18180415034294128,
      "learning_rate": 5.3113539596934434e-05,
      "loss": 0.0009606335312128067,
      "step": 165250
    },
    {
      "epoch": 46.90888447346012,
      "grad_norm": 0.07592189311981201,
      "learning_rate": 5.3110701107011075e-05,
      "loss": 0.003295877203345299,
      "step": 165260
    },
    {
      "epoch": 46.91172296338348,
      "grad_norm": 0.07576707005500793,
      "learning_rate": 5.310786261708771e-05,
      "loss": 0.0006053812801837921,
      "step": 165270
    },
    {
      "epoch": 46.91456145330684,
      "grad_norm": 0.4915252923965454,
      "learning_rate": 5.310502412716435e-05,
      "loss": 0.00052510816603899,
      "step": 165280
    },
    {
      "epoch": 46.9173999432302,
      "grad_norm": 0.28922924399375916,
      "learning_rate": 5.3102185637240986e-05,
      "loss": 0.000751427561044693,
      "step": 165290
    },
    {
      "epoch": 46.920238433153564,
      "grad_norm": 0.04297211393713951,
      "learning_rate": 5.309934714731763e-05,
      "loss": 0.0085616834461689,
      "step": 165300
    },
    {
      "epoch": 46.92307692307692,
      "grad_norm": 2.4428343772888184,
      "learning_rate": 5.3096508657394275e-05,
      "loss": 0.007534244656562805,
      "step": 165310
    },
    {
      "epoch": 46.92591541300028,
      "grad_norm": 9.895331382751465,
      "learning_rate": 5.309367016747091e-05,
      "loss": 0.0017750348895788192,
      "step": 165320
    },
    {
      "epoch": 46.928753902923646,
      "grad_norm": 18.46493911743164,
      "learning_rate": 5.3090831677547545e-05,
      "loss": 0.01760469377040863,
      "step": 165330
    },
    {
      "epoch": 46.93159239284701,
      "grad_norm": 0.45224106311798096,
      "learning_rate": 5.3087993187624186e-05,
      "loss": 0.0005628550425171852,
      "step": 165340
    },
    {
      "epoch": 46.934430882770364,
      "grad_norm": 16.781593322753906,
      "learning_rate": 5.308515469770082e-05,
      "loss": 0.006248362362384796,
      "step": 165350
    },
    {
      "epoch": 46.93726937269373,
      "grad_norm": 0.406368225812912,
      "learning_rate": 5.308231620777746e-05,
      "loss": 0.008037784695625305,
      "step": 165360
    },
    {
      "epoch": 46.94010786261709,
      "grad_norm": 0.043875083327293396,
      "learning_rate": 5.307947771785411e-05,
      "loss": 0.0033710118383169173,
      "step": 165370
    },
    {
      "epoch": 46.942946352540446,
      "grad_norm": 4.744696617126465,
      "learning_rate": 5.3076639227930745e-05,
      "loss": 0.00232350118458271,
      "step": 165380
    },
    {
      "epoch": 46.94578484246381,
      "grad_norm": 0.13025854527950287,
      "learning_rate": 5.3073800738007386e-05,
      "loss": 0.005455337464809418,
      "step": 165390
    },
    {
      "epoch": 46.94862333238717,
      "grad_norm": 21.67111587524414,
      "learning_rate": 5.307096224808402e-05,
      "loss": 0.0073852457106113436,
      "step": 165400
    },
    {
      "epoch": 46.95146182231053,
      "grad_norm": 0.08563057333230972,
      "learning_rate": 5.306812375816066e-05,
      "loss": 0.0005698682740330696,
      "step": 165410
    },
    {
      "epoch": 46.95430031223389,
      "grad_norm": 12.591188430786133,
      "learning_rate": 5.3065285268237297e-05,
      "loss": 0.005222213640809059,
      "step": 165420
    },
    {
      "epoch": 46.95713880215725,
      "grad_norm": 0.058262646198272705,
      "learning_rate": 5.306244677831393e-05,
      "loss": 0.0011304263025522232,
      "step": 165430
    },
    {
      "epoch": 46.959977292080616,
      "grad_norm": 0.03133535757660866,
      "learning_rate": 5.3059608288390586e-05,
      "loss": 0.014902885258197784,
      "step": 165440
    },
    {
      "epoch": 46.96281578200397,
      "grad_norm": 2.620091438293457,
      "learning_rate": 5.305676979846722e-05,
      "loss": 0.001906036213040352,
      "step": 165450
    },
    {
      "epoch": 46.965654271927335,
      "grad_norm": 0.6794065833091736,
      "learning_rate": 5.3053931308543855e-05,
      "loss": 0.007320637255907059,
      "step": 165460
    },
    {
      "epoch": 46.9684927618507,
      "grad_norm": 9.20463752746582,
      "learning_rate": 5.30510928186205e-05,
      "loss": 0.005141761898994446,
      "step": 165470
    },
    {
      "epoch": 46.97133125177405,
      "grad_norm": 11.017587661743164,
      "learning_rate": 5.304825432869713e-05,
      "loss": 0.0037460081279277802,
      "step": 165480
    },
    {
      "epoch": 46.974169741697416,
      "grad_norm": 2.173290491104126,
      "learning_rate": 5.304541583877377e-05,
      "loss": 0.0010919300839304924,
      "step": 165490
    },
    {
      "epoch": 46.97700823162078,
      "grad_norm": 0.10547366738319397,
      "learning_rate": 5.304257734885042e-05,
      "loss": 0.0012940242886543274,
      "step": 165500
    },
    {
      "epoch": 46.97700823162078,
      "eval_accuracy": 0.9753926368665352,
      "eval_loss": 0.08706554025411606,
      "eval_runtime": 32.7415,
      "eval_samples_per_second": 480.339,
      "eval_steps_per_second": 7.513,
      "step": 165500
    },
    {
      "epoch": 46.97984672154414,
      "grad_norm": 0.35043612122535706,
      "learning_rate": 5.3039738858927055e-05,
      "loss": 0.0014933021739125252,
      "step": 165510
    },
    {
      "epoch": 46.9826852114675,
      "grad_norm": 0.022090183570981026,
      "learning_rate": 5.30369003690037e-05,
      "loss": 0.0019676128402352335,
      "step": 165520
    },
    {
      "epoch": 46.98552370139086,
      "grad_norm": 0.173744797706604,
      "learning_rate": 5.303406187908033e-05,
      "loss": 0.005157839506864548,
      "step": 165530
    },
    {
      "epoch": 46.98836219131422,
      "grad_norm": 0.06979185342788696,
      "learning_rate": 5.303122338915697e-05,
      "loss": 0.01020594835281372,
      "step": 165540
    },
    {
      "epoch": 46.99120068123758,
      "grad_norm": 0.5719161629676819,
      "learning_rate": 5.302838489923361e-05,
      "loss": 0.0029745982959866525,
      "step": 165550
    },
    {
      "epoch": 46.99403917116094,
      "grad_norm": 0.026198362931609154,
      "learning_rate": 5.302554640931024e-05,
      "loss": 0.0036046609282493593,
      "step": 165560
    },
    {
      "epoch": 46.996877661084305,
      "grad_norm": 0.1037336215376854,
      "learning_rate": 5.30227079193869e-05,
      "loss": 0.0013667086139321327,
      "step": 165570
    },
    {
      "epoch": 46.99971615100766,
      "grad_norm": 0.05398159846663475,
      "learning_rate": 5.301986942946353e-05,
      "loss": 0.0020139461383223535,
      "step": 165580
    },
    {
      "epoch": 47.002554640931024,
      "grad_norm": 0.7803266048431396,
      "learning_rate": 5.3017030939540166e-05,
      "loss": 0.000467654038220644,
      "step": 165590
    },
    {
      "epoch": 47.00539313085439,
      "grad_norm": 0.07606060802936554,
      "learning_rate": 5.301419244961681e-05,
      "loss": 0.00870707631111145,
      "step": 165600
    },
    {
      "epoch": 47.00823162077775,
      "grad_norm": 0.29465097188949585,
      "learning_rate": 5.301135395969344e-05,
      "loss": 0.0005928380414843559,
      "step": 165610
    },
    {
      "epoch": 47.011070110701105,
      "grad_norm": 0.029950475320219994,
      "learning_rate": 5.3008515469770083e-05,
      "loss": 0.0055006884038448335,
      "step": 165620
    },
    {
      "epoch": 47.01390860062447,
      "grad_norm": 0.02078227698802948,
      "learning_rate": 5.300567697984672e-05,
      "loss": 0.0034839794039726256,
      "step": 165630
    },
    {
      "epoch": 47.01674709054783,
      "grad_norm": 0.1554749459028244,
      "learning_rate": 5.3002838489923366e-05,
      "loss": 0.0033458419144153594,
      "step": 165640
    },
    {
      "epoch": 47.01958558047119,
      "grad_norm": 0.24589680135250092,
      "learning_rate": 5.300000000000001e-05,
      "loss": 0.002386242151260376,
      "step": 165650
    },
    {
      "epoch": 47.02242407039455,
      "grad_norm": 0.03632420673966408,
      "learning_rate": 5.299716151007664e-05,
      "loss": 0.006929624080657959,
      "step": 165660
    },
    {
      "epoch": 47.02526256031791,
      "grad_norm": 0.04455699026584625,
      "learning_rate": 5.2994323020153284e-05,
      "loss": 0.004643542319536209,
      "step": 165670
    },
    {
      "epoch": 47.02810105024127,
      "grad_norm": 1.5836797952651978,
      "learning_rate": 5.299148453022992e-05,
      "loss": 0.000981689803302288,
      "step": 165680
    },
    {
      "epoch": 47.03093954016463,
      "grad_norm": 0.5580620169639587,
      "learning_rate": 5.298864604030655e-05,
      "loss": 0.0017681455239653588,
      "step": 165690
    },
    {
      "epoch": 47.033778030087994,
      "grad_norm": 10.556929588317871,
      "learning_rate": 5.29858075503832e-05,
      "loss": 0.008701587468385697,
      "step": 165700
    },
    {
      "epoch": 47.03661652001136,
      "grad_norm": 0.32172122597694397,
      "learning_rate": 5.298296906045984e-05,
      "loss": 0.00590515248477459,
      "step": 165710
    },
    {
      "epoch": 47.03945500993471,
      "grad_norm": 0.0861821398139,
      "learning_rate": 5.298013057053648e-05,
      "loss": 0.0009194945916533471,
      "step": 165720
    },
    {
      "epoch": 47.042293499858076,
      "grad_norm": 0.03437894582748413,
      "learning_rate": 5.297729208061312e-05,
      "loss": 0.010029523074626923,
      "step": 165730
    },
    {
      "epoch": 47.04513198978144,
      "grad_norm": 0.05261431261897087,
      "learning_rate": 5.297445359068975e-05,
      "loss": 0.0018523328006267547,
      "step": 165740
    },
    {
      "epoch": 47.047970479704794,
      "grad_norm": 0.28681281208992004,
      "learning_rate": 5.2971615100766394e-05,
      "loss": 0.0004867391660809517,
      "step": 165750
    },
    {
      "epoch": 47.05080896962816,
      "grad_norm": 1.0440691709518433,
      "learning_rate": 5.296877661084303e-05,
      "loss": 0.0010393798351287843,
      "step": 165760
    },
    {
      "epoch": 47.05364745955152,
      "grad_norm": 0.6126929521560669,
      "learning_rate": 5.296593812091968e-05,
      "loss": 0.0012228472158312797,
      "step": 165770
    },
    {
      "epoch": 47.056485949474876,
      "grad_norm": 2.4870898723602295,
      "learning_rate": 5.296309963099632e-05,
      "loss": 0.0006401024758815766,
      "step": 165780
    },
    {
      "epoch": 47.05932443939824,
      "grad_norm": 0.44979774951934814,
      "learning_rate": 5.296026114107295e-05,
      "loss": 0.0011499036103487016,
      "step": 165790
    },
    {
      "epoch": 47.0621629293216,
      "grad_norm": 2.9465341567993164,
      "learning_rate": 5.295742265114959e-05,
      "loss": 0.002703012339770794,
      "step": 165800
    },
    {
      "epoch": 47.065001419244965,
      "grad_norm": 0.06846540421247482,
      "learning_rate": 5.295458416122623e-05,
      "loss": 0.0028541328385472296,
      "step": 165810
    },
    {
      "epoch": 47.06783990916832,
      "grad_norm": 7.641231060028076,
      "learning_rate": 5.2951745671302863e-05,
      "loss": 0.012261182069778442,
      "step": 165820
    },
    {
      "epoch": 47.07067839909168,
      "grad_norm": 0.24076054990291595,
      "learning_rate": 5.2948907181379505e-05,
      "loss": 0.0006500562652945518,
      "step": 165830
    },
    {
      "epoch": 47.073516889015046,
      "grad_norm": 0.2534443736076355,
      "learning_rate": 5.294606869145615e-05,
      "loss": 0.00388994887471199,
      "step": 165840
    },
    {
      "epoch": 47.0763553789384,
      "grad_norm": 0.02348289266228676,
      "learning_rate": 5.294323020153279e-05,
      "loss": 0.0010604171082377433,
      "step": 165850
    },
    {
      "epoch": 47.079193868861765,
      "grad_norm": 0.045930467545986176,
      "learning_rate": 5.294039171160943e-05,
      "loss": 0.003213449195027351,
      "step": 165860
    },
    {
      "epoch": 47.08203235878513,
      "grad_norm": 0.15712912380695343,
      "learning_rate": 5.2937553221686064e-05,
      "loss": 0.0009224269539117813,
      "step": 165870
    },
    {
      "epoch": 47.08487084870849,
      "grad_norm": 0.08416629582643509,
      "learning_rate": 5.2934714731762705e-05,
      "loss": 0.0004641583189368248,
      "step": 165880
    },
    {
      "epoch": 47.087709338631846,
      "grad_norm": 0.3400707244873047,
      "learning_rate": 5.293187624183934e-05,
      "loss": 0.0010037094354629517,
      "step": 165890
    },
    {
      "epoch": 47.09054782855521,
      "grad_norm": 0.4553772509098053,
      "learning_rate": 5.292903775191599e-05,
      "loss": 0.006027178093791008,
      "step": 165900
    },
    {
      "epoch": 47.09338631847857,
      "grad_norm": 0.09822464734315872,
      "learning_rate": 5.292619926199263e-05,
      "loss": 0.00034078452736139295,
      "step": 165910
    },
    {
      "epoch": 47.09622480840193,
      "grad_norm": 0.2871064841747284,
      "learning_rate": 5.2923360772069264e-05,
      "loss": 0.0007214108482003212,
      "step": 165920
    },
    {
      "epoch": 47.09906329832529,
      "grad_norm": 0.04511956125497818,
      "learning_rate": 5.29205222821459e-05,
      "loss": 0.000470147468149662,
      "step": 165930
    },
    {
      "epoch": 47.101901788248654,
      "grad_norm": 0.02885298989713192,
      "learning_rate": 5.291768379222254e-05,
      "loss": 0.00041520241647958757,
      "step": 165940
    },
    {
      "epoch": 47.10474027817201,
      "grad_norm": 0.011977088637650013,
      "learning_rate": 5.2914845302299174e-05,
      "loss": 0.0011746549978852272,
      "step": 165950
    },
    {
      "epoch": 47.10757876809537,
      "grad_norm": 0.4777931272983551,
      "learning_rate": 5.2912006812375816e-05,
      "loss": 0.0006107177585363388,
      "step": 165960
    },
    {
      "epoch": 47.110417258018735,
      "grad_norm": 0.0485212542116642,
      "learning_rate": 5.2909168322452464e-05,
      "loss": 0.0011476477608084678,
      "step": 165970
    },
    {
      "epoch": 47.1132557479421,
      "grad_norm": 0.09256909787654877,
      "learning_rate": 5.29063298325291e-05,
      "loss": 0.0006046649068593979,
      "step": 165980
    },
    {
      "epoch": 47.116094237865454,
      "grad_norm": 0.009375392459332943,
      "learning_rate": 5.290349134260574e-05,
      "loss": 0.00045146569609642027,
      "step": 165990
    },
    {
      "epoch": 47.11893272778882,
      "grad_norm": 0.0645822212100029,
      "learning_rate": 5.2900652852682374e-05,
      "loss": 0.004844057559967041,
      "step": 166000
    },
    {
      "epoch": 47.11893272778882,
      "eval_accuracy": 0.9774273542315762,
      "eval_loss": 0.0815049335360527,
      "eval_runtime": 32.477,
      "eval_samples_per_second": 484.251,
      "eval_steps_per_second": 7.575,
      "step": 166000
    },
    {
      "epoch": 47.12177121771218,
      "grad_norm": 0.05515605956315994,
      "learning_rate": 5.2897814362759016e-05,
      "loss": 0.0005520680919289589,
      "step": 166010
    },
    {
      "epoch": 47.124609707635535,
      "grad_norm": 0.1961761862039566,
      "learning_rate": 5.289497587283565e-05,
      "loss": 0.0013574207201600074,
      "step": 166020
    },
    {
      "epoch": 47.1274481975589,
      "grad_norm": 0.08087031543254852,
      "learning_rate": 5.2892137382912285e-05,
      "loss": 0.0004737488925457001,
      "step": 166030
    },
    {
      "epoch": 47.13028668748226,
      "grad_norm": 0.045700203627347946,
      "learning_rate": 5.288929889298894e-05,
      "loss": 0.0015544636175036431,
      "step": 166040
    },
    {
      "epoch": 47.13312517740562,
      "grad_norm": 0.02284938097000122,
      "learning_rate": 5.2886460403065574e-05,
      "loss": 0.0004111355170607567,
      "step": 166050
    },
    {
      "epoch": 47.13596366732898,
      "grad_norm": 0.009912445209920406,
      "learning_rate": 5.288362191314221e-05,
      "loss": 0.000979447178542614,
      "step": 166060
    },
    {
      "epoch": 47.13880215725234,
      "grad_norm": 0.11229164153337479,
      "learning_rate": 5.288078342321885e-05,
      "loss": 0.002558627538383007,
      "step": 166070
    },
    {
      "epoch": 47.141640647175706,
      "grad_norm": 0.1649119108915329,
      "learning_rate": 5.2877944933295485e-05,
      "loss": 0.0022402111440896987,
      "step": 166080
    },
    {
      "epoch": 47.14447913709906,
      "grad_norm": 0.4111264944076538,
      "learning_rate": 5.2875106443372126e-05,
      "loss": 0.0012208811938762664,
      "step": 166090
    },
    {
      "epoch": 47.147317627022424,
      "grad_norm": 0.7173179984092712,
      "learning_rate": 5.2872267953448775e-05,
      "loss": 0.0033980872482061385,
      "step": 166100
    },
    {
      "epoch": 47.15015611694579,
      "grad_norm": 0.014923032373189926,
      "learning_rate": 5.286942946352541e-05,
      "loss": 0.0033872973173856733,
      "step": 166110
    },
    {
      "epoch": 47.15299460686914,
      "grad_norm": 0.07523858547210693,
      "learning_rate": 5.286659097360205e-05,
      "loss": 0.0019164461642503738,
      "step": 166120
    },
    {
      "epoch": 47.155833096792506,
      "grad_norm": 0.6826074719429016,
      "learning_rate": 5.2863752483678685e-05,
      "loss": 0.0015787174925208092,
      "step": 166130
    },
    {
      "epoch": 47.15867158671587,
      "grad_norm": 11.463093757629395,
      "learning_rate": 5.2860913993755326e-05,
      "loss": 0.00977133959531784,
      "step": 166140
    },
    {
      "epoch": 47.161510076639225,
      "grad_norm": 0.1495233178138733,
      "learning_rate": 5.285807550383196e-05,
      "loss": 0.014733101427555084,
      "step": 166150
    },
    {
      "epoch": 47.16434856656259,
      "grad_norm": 0.1507890671491623,
      "learning_rate": 5.2855237013908596e-05,
      "loss": 0.007516796886920929,
      "step": 166160
    },
    {
      "epoch": 47.16718705648595,
      "grad_norm": 3.0286548137664795,
      "learning_rate": 5.2852398523985244e-05,
      "loss": 0.0013044342398643495,
      "step": 166170
    },
    {
      "epoch": 47.17002554640931,
      "grad_norm": 0.7042717337608337,
      "learning_rate": 5.2849560034061885e-05,
      "loss": 0.0006618889048695565,
      "step": 166180
    },
    {
      "epoch": 47.17286403633267,
      "grad_norm": 0.12900876998901367,
      "learning_rate": 5.284672154413852e-05,
      "loss": 0.0012743659317493439,
      "step": 166190
    },
    {
      "epoch": 47.17570252625603,
      "grad_norm": 2.6079959869384766,
      "learning_rate": 5.284388305421516e-05,
      "loss": 0.0010534578934311867,
      "step": 166200
    },
    {
      "epoch": 47.178541016179395,
      "grad_norm": 0.03617927432060242,
      "learning_rate": 5.2841044564291796e-05,
      "loss": 0.011403732001781464,
      "step": 166210
    },
    {
      "epoch": 47.18137950610275,
      "grad_norm": 3.6006979942321777,
      "learning_rate": 5.283820607436844e-05,
      "loss": 0.004512877762317657,
      "step": 166220
    },
    {
      "epoch": 47.18421799602611,
      "grad_norm": 1.7622861862182617,
      "learning_rate": 5.283536758444507e-05,
      "loss": 0.00175212100148201,
      "step": 166230
    },
    {
      "epoch": 47.187056485949476,
      "grad_norm": 1.1863501071929932,
      "learning_rate": 5.283252909452172e-05,
      "loss": 0.009258320927619934,
      "step": 166240
    },
    {
      "epoch": 47.18989497587283,
      "grad_norm": 0.2306436002254486,
      "learning_rate": 5.282969060459836e-05,
      "loss": 0.0019975405186414717,
      "step": 166250
    },
    {
      "epoch": 47.192733465796195,
      "grad_norm": 0.9713077545166016,
      "learning_rate": 5.2826852114674996e-05,
      "loss": 0.0034524496644735337,
      "step": 166260
    },
    {
      "epoch": 47.19557195571956,
      "grad_norm": 0.21769149601459503,
      "learning_rate": 5.282401362475163e-05,
      "loss": 0.000939839705824852,
      "step": 166270
    },
    {
      "epoch": 47.19841044564292,
      "grad_norm": 3.201869249343872,
      "learning_rate": 5.282117513482827e-05,
      "loss": 0.002042139321565628,
      "step": 166280
    },
    {
      "epoch": 47.20124893556628,
      "grad_norm": 0.0459790974855423,
      "learning_rate": 5.2818336644904906e-05,
      "loss": 0.008180337399244309,
      "step": 166290
    },
    {
      "epoch": 47.20408742548964,
      "grad_norm": 0.24088405072689056,
      "learning_rate": 5.2815498154981555e-05,
      "loss": 0.001287614554166794,
      "step": 166300
    },
    {
      "epoch": 47.206925915413,
      "grad_norm": 0.018882904201745987,
      "learning_rate": 5.2812659665058196e-05,
      "loss": 0.002986639179289341,
      "step": 166310
    },
    {
      "epoch": 47.20976440533636,
      "grad_norm": 0.08961990475654602,
      "learning_rate": 5.280982117513483e-05,
      "loss": 0.0024691296741366385,
      "step": 166320
    },
    {
      "epoch": 47.21260289525972,
      "grad_norm": 0.33202579617500305,
      "learning_rate": 5.280698268521147e-05,
      "loss": 0.00163177028298378,
      "step": 166330
    },
    {
      "epoch": 47.215441385183084,
      "grad_norm": 0.057728614658117294,
      "learning_rate": 5.2804144195288107e-05,
      "loss": 0.0005844833329319953,
      "step": 166340
    },
    {
      "epoch": 47.21827987510645,
      "grad_norm": 0.17069363594055176,
      "learning_rate": 5.280130570536475e-05,
      "loss": 0.0024190155789256098,
      "step": 166350
    },
    {
      "epoch": 47.2211183650298,
      "grad_norm": 0.613150417804718,
      "learning_rate": 5.279846721544138e-05,
      "loss": 0.002414032816886902,
      "step": 166360
    },
    {
      "epoch": 47.223956854953165,
      "grad_norm": 0.16664165258407593,
      "learning_rate": 5.279562872551803e-05,
      "loss": 0.002566816471517086,
      "step": 166370
    },
    {
      "epoch": 47.22679534487653,
      "grad_norm": 0.11754713207483292,
      "learning_rate": 5.279279023559467e-05,
      "loss": 0.007681527733802795,
      "step": 166380
    },
    {
      "epoch": 47.229633834799884,
      "grad_norm": 7.869407653808594,
      "learning_rate": 5.278995174567131e-05,
      "loss": 0.002509535290300846,
      "step": 166390
    },
    {
      "epoch": 47.23247232472325,
      "grad_norm": 0.03814363107085228,
      "learning_rate": 5.278711325574794e-05,
      "loss": 0.0014111274853348733,
      "step": 166400
    },
    {
      "epoch": 47.23531081464661,
      "grad_norm": 0.32036903500556946,
      "learning_rate": 5.278427476582458e-05,
      "loss": 0.0025612499564886095,
      "step": 166410
    },
    {
      "epoch": 47.238149304569966,
      "grad_norm": 1.2629543542861938,
      "learning_rate": 5.278143627590122e-05,
      "loss": 0.002174685336649418,
      "step": 166420
    },
    {
      "epoch": 47.24098779449333,
      "grad_norm": 7.14158296585083,
      "learning_rate": 5.277859778597786e-05,
      "loss": 0.0019754907116293907,
      "step": 166430
    },
    {
      "epoch": 47.24382628441669,
      "grad_norm": 0.7406134605407715,
      "learning_rate": 5.277575929605451e-05,
      "loss": 0.0006751446053385735,
      "step": 166440
    },
    {
      "epoch": 47.246664774340054,
      "grad_norm": 0.05451953411102295,
      "learning_rate": 5.277292080613114e-05,
      "loss": 0.005029942467808723,
      "step": 166450
    },
    {
      "epoch": 47.24950326426341,
      "grad_norm": 0.5551103949546814,
      "learning_rate": 5.277008231620778e-05,
      "loss": 0.0017206134274601937,
      "step": 166460
    },
    {
      "epoch": 47.25234175418677,
      "grad_norm": 0.025522135198116302,
      "learning_rate": 5.276724382628442e-05,
      "loss": 0.0007320387288928032,
      "step": 166470
    },
    {
      "epoch": 47.255180244110136,
      "grad_norm": 1.0864540338516235,
      "learning_rate": 5.276440533636106e-05,
      "loss": 0.0006345083937048912,
      "step": 166480
    },
    {
      "epoch": 47.25801873403349,
      "grad_norm": 1.630776286125183,
      "learning_rate": 5.276156684643769e-05,
      "loss": 0.0022073639556765556,
      "step": 166490
    },
    {
      "epoch": 47.260857223956855,
      "grad_norm": 0.1615860015153885,
      "learning_rate": 5.275872835651434e-05,
      "loss": 0.001271134801208973,
      "step": 166500
    },
    {
      "epoch": 47.260857223956855,
      "eval_accuracy": 0.9731035798308642,
      "eval_loss": 0.09798971563577652,
      "eval_runtime": 32.3479,
      "eval_samples_per_second": 486.184,
      "eval_steps_per_second": 7.605,
      "step": 166500
    },
    {
      "epoch": 47.26369571388022,
      "grad_norm": 6.387979507446289,
      "learning_rate": 5.275588986659098e-05,
      "loss": 0.0245142787694931,
      "step": 166510
    },
    {
      "epoch": 47.26653420380357,
      "grad_norm": 1.0076404809951782,
      "learning_rate": 5.275305137666762e-05,
      "loss": 0.0019852887839078904,
      "step": 166520
    },
    {
      "epoch": 47.269372693726936,
      "grad_norm": 0.1098475456237793,
      "learning_rate": 5.275021288674425e-05,
      "loss": 0.0047672510147094725,
      "step": 166530
    },
    {
      "epoch": 47.2722111836503,
      "grad_norm": 0.6762949824333191,
      "learning_rate": 5.2747374396820893e-05,
      "loss": 0.0014756165444850921,
      "step": 166540
    },
    {
      "epoch": 47.27504967357366,
      "grad_norm": 0.10568742454051971,
      "learning_rate": 5.274453590689753e-05,
      "loss": 0.007734911143779754,
      "step": 166550
    },
    {
      "epoch": 47.27788816349702,
      "grad_norm": 0.2346913367509842,
      "learning_rate": 5.274169741697417e-05,
      "loss": 0.001791946031153202,
      "step": 166560
    },
    {
      "epoch": 47.28072665342038,
      "grad_norm": 0.1376802921295166,
      "learning_rate": 5.273885892705082e-05,
      "loss": 0.003552095964550972,
      "step": 166570
    },
    {
      "epoch": 47.28356514334374,
      "grad_norm": 0.4562109410762787,
      "learning_rate": 5.273602043712745e-05,
      "loss": 0.0028885962441563607,
      "step": 166580
    },
    {
      "epoch": 47.2864036332671,
      "grad_norm": 2.786576509475708,
      "learning_rate": 5.2733181947204093e-05,
      "loss": 0.0011640481650829315,
      "step": 166590
    },
    {
      "epoch": 47.28924212319046,
      "grad_norm": 0.5501305460929871,
      "learning_rate": 5.273034345728073e-05,
      "loss": 0.0067717239260673525,
      "step": 166600
    },
    {
      "epoch": 47.292080613113825,
      "grad_norm": 0.04443579912185669,
      "learning_rate": 5.272750496735737e-05,
      "loss": 0.0014348624274134636,
      "step": 166610
    },
    {
      "epoch": 47.29491910303718,
      "grad_norm": 0.6530584096908569,
      "learning_rate": 5.2724666477434004e-05,
      "loss": 0.00040276926010847093,
      "step": 166620
    },
    {
      "epoch": 47.297757592960544,
      "grad_norm": 2.501415729522705,
      "learning_rate": 5.272182798751064e-05,
      "loss": 0.0007356952875852585,
      "step": 166630
    },
    {
      "epoch": 47.30059608288391,
      "grad_norm": 7.387036323547363,
      "learning_rate": 5.271898949758729e-05,
      "loss": 0.0021438641473650933,
      "step": 166640
    },
    {
      "epoch": 47.30343457280727,
      "grad_norm": 0.08627983927726746,
      "learning_rate": 5.271615100766393e-05,
      "loss": 0.0011327279731631278,
      "step": 166650
    },
    {
      "epoch": 47.306273062730625,
      "grad_norm": 6.514647006988525,
      "learning_rate": 5.271331251774056e-05,
      "loss": 0.0016768177971243857,
      "step": 166660
    },
    {
      "epoch": 47.30911155265399,
      "grad_norm": 0.4373135566711426,
      "learning_rate": 5.2710474027817204e-05,
      "loss": 0.0006712866947054863,
      "step": 166670
    },
    {
      "epoch": 47.31195004257735,
      "grad_norm": 0.023332899436354637,
      "learning_rate": 5.270763553789384e-05,
      "loss": 0.002923675626516342,
      "step": 166680
    },
    {
      "epoch": 47.31478853250071,
      "grad_norm": 1.7862659692764282,
      "learning_rate": 5.270479704797048e-05,
      "loss": 0.010661233961582185,
      "step": 166690
    },
    {
      "epoch": 47.31762702242407,
      "grad_norm": 0.8820787668228149,
      "learning_rate": 5.270195855804713e-05,
      "loss": 0.0009172948077321053,
      "step": 166700
    },
    {
      "epoch": 47.32046551234743,
      "grad_norm": 0.9722907543182373,
      "learning_rate": 5.269912006812376e-05,
      "loss": 0.0009654708206653595,
      "step": 166710
    },
    {
      "epoch": 47.323304002270795,
      "grad_norm": 4.111182689666748,
      "learning_rate": 5.2696281578200404e-05,
      "loss": 0.00997893065214157,
      "step": 166720
    },
    {
      "epoch": 47.32614249219415,
      "grad_norm": 2.895232915878296,
      "learning_rate": 5.269344308827704e-05,
      "loss": 0.0013338617980480194,
      "step": 166730
    },
    {
      "epoch": 47.328980982117514,
      "grad_norm": 0.9905418753623962,
      "learning_rate": 5.2690604598353673e-05,
      "loss": 0.002423496171832085,
      "step": 166740
    },
    {
      "epoch": 47.33181947204088,
      "grad_norm": 0.2844701409339905,
      "learning_rate": 5.2687766108430315e-05,
      "loss": 0.0046717777848243715,
      "step": 166750
    },
    {
      "epoch": 47.33465796196423,
      "grad_norm": 0.034229766577482224,
      "learning_rate": 5.268492761850695e-05,
      "loss": 0.000941377878189087,
      "step": 166760
    },
    {
      "epoch": 47.337496451887596,
      "grad_norm": 0.16480103135108948,
      "learning_rate": 5.26820891285836e-05,
      "loss": 0.007043737918138504,
      "step": 166770
    },
    {
      "epoch": 47.34033494181096,
      "grad_norm": 0.052474845200777054,
      "learning_rate": 5.267925063866024e-05,
      "loss": 0.002284706011414528,
      "step": 166780
    },
    {
      "epoch": 47.343173431734314,
      "grad_norm": 0.022756602615118027,
      "learning_rate": 5.2676412148736874e-05,
      "loss": 0.00041964594274759295,
      "step": 166790
    },
    {
      "epoch": 47.34601192165768,
      "grad_norm": 0.08518175035715103,
      "learning_rate": 5.2673573658813515e-05,
      "loss": 0.00022263973951339722,
      "step": 166800
    },
    {
      "epoch": 47.34885041158104,
      "grad_norm": 0.68224036693573,
      "learning_rate": 5.267073516889015e-05,
      "loss": 0.005037920176982879,
      "step": 166810
    },
    {
      "epoch": 47.3516889015044,
      "grad_norm": 0.0067217121832072735,
      "learning_rate": 5.266789667896679e-05,
      "loss": 0.0004578862339258194,
      "step": 166820
    },
    {
      "epoch": 47.35452739142776,
      "grad_norm": 0.016836734488606453,
      "learning_rate": 5.266505818904344e-05,
      "loss": 0.0023566238582134248,
      "step": 166830
    },
    {
      "epoch": 47.35736588135112,
      "grad_norm": 0.4242056608200073,
      "learning_rate": 5.2662219699120074e-05,
      "loss": 0.01018211543560028,
      "step": 166840
    },
    {
      "epoch": 47.360204371274484,
      "grad_norm": 0.4246620237827301,
      "learning_rate": 5.2659381209196715e-05,
      "loss": 0.0006241114810109138,
      "step": 166850
    },
    {
      "epoch": 47.36304286119784,
      "grad_norm": 0.5629755258560181,
      "learning_rate": 5.265654271927335e-05,
      "loss": 0.0011145751923322678,
      "step": 166860
    },
    {
      "epoch": 47.3658813511212,
      "grad_norm": 0.17603833973407745,
      "learning_rate": 5.2653704229349984e-05,
      "loss": 0.0015360020101070405,
      "step": 166870
    },
    {
      "epoch": 47.368719841044566,
      "grad_norm": 0.12736016511917114,
      "learning_rate": 5.2650865739426626e-05,
      "loss": 0.000812043622136116,
      "step": 166880
    },
    {
      "epoch": 47.37155833096792,
      "grad_norm": 16.477394104003906,
      "learning_rate": 5.264802724950326e-05,
      "loss": 0.004146634414792061,
      "step": 166890
    },
    {
      "epoch": 47.374396820891285,
      "grad_norm": 0.057493966072797775,
      "learning_rate": 5.264518875957991e-05,
      "loss": 0.000408392958343029,
      "step": 166900
    },
    {
      "epoch": 47.37723531081465,
      "grad_norm": 0.21680188179016113,
      "learning_rate": 5.264235026965655e-05,
      "loss": 0.0022554801777005195,
      "step": 166910
    },
    {
      "epoch": 47.38007380073801,
      "grad_norm": 0.2876989543437958,
      "learning_rate": 5.2639511779733184e-05,
      "loss": 0.0005332782864570617,
      "step": 166920
    },
    {
      "epoch": 47.382912290661366,
      "grad_norm": 1.8345699310302734,
      "learning_rate": 5.2636673289809826e-05,
      "loss": 0.0006250036880373955,
      "step": 166930
    },
    {
      "epoch": 47.38575078058473,
      "grad_norm": 0.32941848039627075,
      "learning_rate": 5.263411864887879e-05,
      "loss": 0.0045709840953350065,
      "step": 166940
    },
    {
      "epoch": 47.38858927050809,
      "grad_norm": 1.108206868171692,
      "learning_rate": 5.263128015895544e-05,
      "loss": 0.002474643290042877,
      "step": 166950
    },
    {
      "epoch": 47.39142776043145,
      "grad_norm": 0.026148110628128052,
      "learning_rate": 5.262844166903208e-05,
      "loss": 0.0009128781035542488,
      "step": 166960
    },
    {
      "epoch": 47.39426625035481,
      "grad_norm": 1.4909217357635498,
      "learning_rate": 5.2625603179108716e-05,
      "loss": 0.0008707690984010696,
      "step": 166970
    },
    {
      "epoch": 47.39710474027817,
      "grad_norm": 0.10078208148479462,
      "learning_rate": 5.262276468918536e-05,
      "loss": 0.0038687162101268767,
      "step": 166980
    },
    {
      "epoch": 47.39994323020153,
      "grad_norm": 0.08496785163879395,
      "learning_rate": 5.261992619926199e-05,
      "loss": 0.003842758387327194,
      "step": 166990
    },
    {
      "epoch": 47.40278172012489,
      "grad_norm": 0.09671993553638458,
      "learning_rate": 5.2617087709338634e-05,
      "loss": 0.0009970778599381446,
      "step": 167000
    },
    {
      "epoch": 47.40278172012489,
      "eval_accuracy": 0.9780632034081516,
      "eval_loss": 0.07823197543621063,
      "eval_runtime": 32.3913,
      "eval_samples_per_second": 485.531,
      "eval_steps_per_second": 7.595,
      "step": 167000
    },
    {
      "epoch": 47.405620210048255,
      "grad_norm": 0.7634956240653992,
      "learning_rate": 5.261424921941527e-05,
      "loss": 0.0007111424580216408,
      "step": 167010
    },
    {
      "epoch": 47.40845869997162,
      "grad_norm": 3.878443956375122,
      "learning_rate": 5.2611410729491916e-05,
      "loss": 0.003624328598380089,
      "step": 167020
    },
    {
      "epoch": 47.411297189894974,
      "grad_norm": 0.12696479260921478,
      "learning_rate": 5.260857223956856e-05,
      "loss": 0.003395608067512512,
      "step": 167030
    },
    {
      "epoch": 47.41413567981834,
      "grad_norm": 0.2264775037765503,
      "learning_rate": 5.260573374964519e-05,
      "loss": 0.001472318172454834,
      "step": 167040
    },
    {
      "epoch": 47.4169741697417,
      "grad_norm": 0.4618097245693207,
      "learning_rate": 5.260289525972183e-05,
      "loss": 0.0015398543328046799,
      "step": 167050
    },
    {
      "epoch": 47.419812659665055,
      "grad_norm": 0.7450781464576721,
      "learning_rate": 5.260005676979847e-05,
      "loss": 0.001631312258541584,
      "step": 167060
    },
    {
      "epoch": 47.42265114958842,
      "grad_norm": 0.10932065546512604,
      "learning_rate": 5.25972182798751e-05,
      "loss": 0.0033196743577718736,
      "step": 167070
    },
    {
      "epoch": 47.42548963951178,
      "grad_norm": 0.1365431249141693,
      "learning_rate": 5.259437978995175e-05,
      "loss": 0.0022184979170560837,
      "step": 167080
    },
    {
      "epoch": 47.428328129435144,
      "grad_norm": 0.14188408851623535,
      "learning_rate": 5.259154130002839e-05,
      "loss": 0.0033716864883899687,
      "step": 167090
    },
    {
      "epoch": 47.4311666193585,
      "grad_norm": 0.07446951419115067,
      "learning_rate": 5.258870281010503e-05,
      "loss": 0.0005040997639298439,
      "step": 167100
    },
    {
      "epoch": 47.43400510928186,
      "grad_norm": 0.15657943487167358,
      "learning_rate": 5.258586432018167e-05,
      "loss": 0.0006972922012209893,
      "step": 167110
    },
    {
      "epoch": 47.436843599205226,
      "grad_norm": 0.07956122606992722,
      "learning_rate": 5.25830258302583e-05,
      "loss": 0.0031499117612838745,
      "step": 167120
    },
    {
      "epoch": 47.43968208912858,
      "grad_norm": 0.5864604711532593,
      "learning_rate": 5.2580187340334944e-05,
      "loss": 0.0006696803495287896,
      "step": 167130
    },
    {
      "epoch": 47.442520579051944,
      "grad_norm": 0.05199510604143143,
      "learning_rate": 5.257734885041158e-05,
      "loss": 0.00045557301491498946,
      "step": 167140
    },
    {
      "epoch": 47.44535906897531,
      "grad_norm": 1.0746148824691772,
      "learning_rate": 5.257451036048823e-05,
      "loss": 0.0006883697584271431,
      "step": 167150
    },
    {
      "epoch": 47.44819755889866,
      "grad_norm": 0.007694663945585489,
      "learning_rate": 5.257167187056486e-05,
      "loss": 0.00048616491258144376,
      "step": 167160
    },
    {
      "epoch": 47.451036048822026,
      "grad_norm": 0.2054125964641571,
      "learning_rate": 5.25688333806415e-05,
      "loss": 0.00030087288469076156,
      "step": 167170
    },
    {
      "epoch": 47.45387453874539,
      "grad_norm": 0.23879587650299072,
      "learning_rate": 5.256599489071814e-05,
      "loss": 0.0003641139715909958,
      "step": 167180
    },
    {
      "epoch": 47.45671302866875,
      "grad_norm": 0.10503577440977097,
      "learning_rate": 5.256315640079478e-05,
      "loss": 0.001046133041381836,
      "step": 167190
    },
    {
      "epoch": 47.45955151859211,
      "grad_norm": 0.2110137939453125,
      "learning_rate": 5.2560317910871414e-05,
      "loss": 0.0010547686368227005,
      "step": 167200
    },
    {
      "epoch": 47.46239000851547,
      "grad_norm": 0.7565997242927551,
      "learning_rate": 5.255747942094806e-05,
      "loss": 0.0018064182251691819,
      "step": 167210
    },
    {
      "epoch": 47.46522849843883,
      "grad_norm": 0.048939298838377,
      "learning_rate": 5.25546409310247e-05,
      "loss": 0.0009963680058717728,
      "step": 167220
    },
    {
      "epoch": 47.46806698836219,
      "grad_norm": 0.12362924218177795,
      "learning_rate": 5.255180244110134e-05,
      "loss": 0.0016524802893400192,
      "step": 167230
    },
    {
      "epoch": 47.47090547828555,
      "grad_norm": 0.05248033255338669,
      "learning_rate": 5.254896395117798e-05,
      "loss": 0.0037679560482501984,
      "step": 167240
    },
    {
      "epoch": 47.473743968208915,
      "grad_norm": 0.34650927782058716,
      "learning_rate": 5.2546125461254614e-05,
      "loss": 0.00043793749064207076,
      "step": 167250
    },
    {
      "epoch": 47.47658245813227,
      "grad_norm": 0.7219845056533813,
      "learning_rate": 5.254328697133125e-05,
      "loss": 0.0063042744994163515,
      "step": 167260
    },
    {
      "epoch": 47.47942094805563,
      "grad_norm": 0.0650295838713646,
      "learning_rate": 5.254044848140789e-05,
      "loss": 0.0011741900816559792,
      "step": 167270
    },
    {
      "epoch": 47.482259437978996,
      "grad_norm": 0.21481147408485413,
      "learning_rate": 5.253760999148454e-05,
      "loss": 0.001363087072968483,
      "step": 167280
    },
    {
      "epoch": 47.48509792790236,
      "grad_norm": 0.19698317348957062,
      "learning_rate": 5.253477150156117e-05,
      "loss": 0.005024828761816025,
      "step": 167290
    },
    {
      "epoch": 47.487936417825715,
      "grad_norm": 0.31931695342063904,
      "learning_rate": 5.2531933011637814e-05,
      "loss": 0.0033177703619003294,
      "step": 167300
    },
    {
      "epoch": 47.49077490774908,
      "grad_norm": 2.6678307056427,
      "learning_rate": 5.252909452171445e-05,
      "loss": 0.005327405780553818,
      "step": 167310
    },
    {
      "epoch": 47.49361339767244,
      "grad_norm": 3.221245288848877,
      "learning_rate": 5.252625603179109e-05,
      "loss": 0.002242329344153404,
      "step": 167320
    },
    {
      "epoch": 47.496451887595796,
      "grad_norm": 0.07731877267360687,
      "learning_rate": 5.2523417541867724e-05,
      "loss": 0.0011165473610162736,
      "step": 167330
    },
    {
      "epoch": 47.49929037751916,
      "grad_norm": 0.1471671760082245,
      "learning_rate": 5.2520579051944366e-05,
      "loss": 0.0007773425430059433,
      "step": 167340
    },
    {
      "epoch": 47.50212886744252,
      "grad_norm": 0.31566593050956726,
      "learning_rate": 5.2517740562021014e-05,
      "loss": 0.001333497278392315,
      "step": 167350
    },
    {
      "epoch": 47.50496735736588,
      "grad_norm": 0.20246922969818115,
      "learning_rate": 5.251490207209765e-05,
      "loss": 0.0033161580562591554,
      "step": 167360
    },
    {
      "epoch": 47.50780584728924,
      "grad_norm": 0.1592196822166443,
      "learning_rate": 5.251206358217429e-05,
      "loss": 0.003666132688522339,
      "step": 167370
    },
    {
      "epoch": 47.510644337212604,
      "grad_norm": 0.045315347611904144,
      "learning_rate": 5.2509225092250924e-05,
      "loss": 0.009665761888027192,
      "step": 167380
    },
    {
      "epoch": 47.51348282713597,
      "grad_norm": 0.028198417276144028,
      "learning_rate": 5.250638660232756e-05,
      "loss": 0.0010032622143626214,
      "step": 167390
    },
    {
      "epoch": 47.51632131705932,
      "grad_norm": 18.67765235900879,
      "learning_rate": 5.25035481124042e-05,
      "loss": 0.009872537106275558,
      "step": 167400
    },
    {
      "epoch": 47.519159806982685,
      "grad_norm": 0.10111254453659058,
      "learning_rate": 5.250070962248085e-05,
      "loss": 0.0007732365280389786,
      "step": 167410
    },
    {
      "epoch": 47.52199829690605,
      "grad_norm": 1.6557422876358032,
      "learning_rate": 5.249787113255748e-05,
      "loss": 0.0020810745656490324,
      "step": 167420
    },
    {
      "epoch": 47.524836786829404,
      "grad_norm": 0.1193055659532547,
      "learning_rate": 5.2495032642634125e-05,
      "loss": 0.0017474636435508728,
      "step": 167430
    },
    {
      "epoch": 47.52767527675277,
      "grad_norm": 0.15565578639507294,
      "learning_rate": 5.249219415271076e-05,
      "loss": 0.0027186179533600805,
      "step": 167440
    },
    {
      "epoch": 47.53051376667613,
      "grad_norm": 0.8586099147796631,
      "learning_rate": 5.24893556627874e-05,
      "loss": 0.0034389063715934754,
      "step": 167450
    },
    {
      "epoch": 47.533352256599485,
      "grad_norm": 3.602588415145874,
      "learning_rate": 5.2486517172864035e-05,
      "loss": 0.004334207624197006,
      "step": 167460
    },
    {
      "epoch": 47.53619074652285,
      "grad_norm": 0.6245102882385254,
      "learning_rate": 5.2483678682940677e-05,
      "loss": 0.0024789247661828996,
      "step": 167470
    },
    {
      "epoch": 47.53902923644621,
      "grad_norm": 0.03193337097764015,
      "learning_rate": 5.2480840193017325e-05,
      "loss": 0.0022315897047519686,
      "step": 167480
    },
    {
      "epoch": 47.541867726369574,
      "grad_norm": 0.7624863386154175,
      "learning_rate": 5.247800170309396e-05,
      "loss": 0.00043298304080963135,
      "step": 167490
    },
    {
      "epoch": 47.54470621629293,
      "grad_norm": 0.08291812986135483,
      "learning_rate": 5.2475163213170594e-05,
      "loss": 0.007259470224380493,
      "step": 167500
    },
    {
      "epoch": 47.54470621629293,
      "eval_accuracy": 0.9761556558784257,
      "eval_loss": 0.08670477569103241,
      "eval_runtime": 32.1733,
      "eval_samples_per_second": 488.822,
      "eval_steps_per_second": 7.646,
      "step": 167500
    },
    {
      "epoch": 47.54754470621629,
      "grad_norm": 0.23990261554718018,
      "learning_rate": 5.2472324723247235e-05,
      "loss": 0.0005029913038015366,
      "step": 167510
    },
    {
      "epoch": 47.550383196139656,
      "grad_norm": 0.04974672198295593,
      "learning_rate": 5.246948623332387e-05,
      "loss": 0.0020073408260941504,
      "step": 167520
    },
    {
      "epoch": 47.55322168606301,
      "grad_norm": 0.14446726441383362,
      "learning_rate": 5.246664774340051e-05,
      "loss": 0.003113732486963272,
      "step": 167530
    },
    {
      "epoch": 47.556060175986374,
      "grad_norm": 0.1731187403202057,
      "learning_rate": 5.2463809253477146e-05,
      "loss": 0.0013014884665608407,
      "step": 167540
    },
    {
      "epoch": 47.55889866590974,
      "grad_norm": 0.9669291377067566,
      "learning_rate": 5.2460970763553794e-05,
      "loss": 0.0069547370076179504,
      "step": 167550
    },
    {
      "epoch": 47.5617371558331,
      "grad_norm": 0.2419898808002472,
      "learning_rate": 5.2458132273630435e-05,
      "loss": 0.0009422214701771736,
      "step": 167560
    },
    {
      "epoch": 47.564575645756456,
      "grad_norm": 0.01126257423311472,
      "learning_rate": 5.245529378370707e-05,
      "loss": 0.0006084395572543144,
      "step": 167570
    },
    {
      "epoch": 47.56741413567982,
      "grad_norm": 0.09846566617488861,
      "learning_rate": 5.245245529378371e-05,
      "loss": 0.012254256010055541,
      "step": 167580
    },
    {
      "epoch": 47.57025262560318,
      "grad_norm": 0.3159501850605011,
      "learning_rate": 5.2449616803860346e-05,
      "loss": 0.00090787373483181,
      "step": 167590
    },
    {
      "epoch": 47.57309111552654,
      "grad_norm": 10.154810905456543,
      "learning_rate": 5.244677831393699e-05,
      "loss": 0.0022215677425265314,
      "step": 167600
    },
    {
      "epoch": 47.5759296054499,
      "grad_norm": 1.2654898166656494,
      "learning_rate": 5.2443939824013635e-05,
      "loss": 0.0015854045748710632,
      "step": 167610
    },
    {
      "epoch": 47.57876809537326,
      "grad_norm": 1.8486334085464478,
      "learning_rate": 5.244110133409027e-05,
      "loss": 0.006145539879798889,
      "step": 167620
    },
    {
      "epoch": 47.58160658529662,
      "grad_norm": 3.197897434234619,
      "learning_rate": 5.2438262844166905e-05,
      "loss": 0.0019372975453734397,
      "step": 167630
    },
    {
      "epoch": 47.58444507521998,
      "grad_norm": 0.05604947358369827,
      "learning_rate": 5.2435424354243546e-05,
      "loss": 0.0009148495271801948,
      "step": 167640
    },
    {
      "epoch": 47.587283565143345,
      "grad_norm": 0.025449348613619804,
      "learning_rate": 5.243258586432018e-05,
      "loss": 0.0012341614812612534,
      "step": 167650
    },
    {
      "epoch": 47.59012205506671,
      "grad_norm": 0.0940377339720726,
      "learning_rate": 5.242974737439682e-05,
      "loss": 0.001336769200861454,
      "step": 167660
    },
    {
      "epoch": 47.59296054499006,
      "grad_norm": 0.39495372772216797,
      "learning_rate": 5.2426908884473457e-05,
      "loss": 0.0012563586235046387,
      "step": 167670
    },
    {
      "epoch": 47.595799034913426,
      "grad_norm": 0.5860205888748169,
      "learning_rate": 5.2424070394550105e-05,
      "loss": 0.0007381979376077652,
      "step": 167680
    },
    {
      "epoch": 47.59863752483679,
      "grad_norm": 0.046888090670108795,
      "learning_rate": 5.2421231904626746e-05,
      "loss": 0.0012173319235444068,
      "step": 167690
    },
    {
      "epoch": 47.601476014760145,
      "grad_norm": 0.02788430266082287,
      "learning_rate": 5.241839341470338e-05,
      "loss": 0.0017235185950994492,
      "step": 167700
    },
    {
      "epoch": 47.60431450468351,
      "grad_norm": 0.16087184846401215,
      "learning_rate": 5.241555492478002e-05,
      "loss": 0.013849596679210662,
      "step": 167710
    },
    {
      "epoch": 47.60715299460687,
      "grad_norm": 0.045793380588293076,
      "learning_rate": 5.241271643485666e-05,
      "loss": 0.011382525414228439,
      "step": 167720
    },
    {
      "epoch": 47.60999148453023,
      "grad_norm": 16.612462997436523,
      "learning_rate": 5.240987794493329e-05,
      "loss": 0.015231046080589294,
      "step": 167730
    },
    {
      "epoch": 47.61282997445359,
      "grad_norm": 0.14068502187728882,
      "learning_rate": 5.240703945500993e-05,
      "loss": 0.0008405828848481178,
      "step": 167740
    },
    {
      "epoch": 47.61566846437695,
      "grad_norm": 6.106640338897705,
      "learning_rate": 5.240420096508658e-05,
      "loss": 0.0014650883153080941,
      "step": 167750
    },
    {
      "epoch": 47.618506954300315,
      "grad_norm": 0.624684751033783,
      "learning_rate": 5.2401362475163215e-05,
      "loss": 0.0007203679531812667,
      "step": 167760
    },
    {
      "epoch": 47.62134544422367,
      "grad_norm": 0.10901641100645065,
      "learning_rate": 5.239852398523986e-05,
      "loss": 0.0009631568565964699,
      "step": 167770
    },
    {
      "epoch": 47.624183934147034,
      "grad_norm": 7.982468605041504,
      "learning_rate": 5.239568549531649e-05,
      "loss": 0.003823201358318329,
      "step": 167780
    },
    {
      "epoch": 47.6270224240704,
      "grad_norm": 0.06653553992509842,
      "learning_rate": 5.239284700539313e-05,
      "loss": 0.017026133835315704,
      "step": 167790
    },
    {
      "epoch": 47.62986091399375,
      "grad_norm": 0.10966093838214874,
      "learning_rate": 5.239000851546977e-05,
      "loss": 0.0029721369966864584,
      "step": 167800
    },
    {
      "epoch": 47.632699403917115,
      "grad_norm": 0.02199809066951275,
      "learning_rate": 5.2387170025546416e-05,
      "loss": 0.01081976741552353,
      "step": 167810
    },
    {
      "epoch": 47.63553789384048,
      "grad_norm": 2.0907764434814453,
      "learning_rate": 5.238433153562306e-05,
      "loss": 0.005662202462553978,
      "step": 167820
    },
    {
      "epoch": 47.63837638376384,
      "grad_norm": 3.259350538253784,
      "learning_rate": 5.238149304569969e-05,
      "loss": 0.01005442589521408,
      "step": 167830
    },
    {
      "epoch": 47.6412148736872,
      "grad_norm": 0.15806244313716888,
      "learning_rate": 5.237865455577633e-05,
      "loss": 0.0011510159820318222,
      "step": 167840
    },
    {
      "epoch": 47.64405336361056,
      "grad_norm": 0.5399385690689087,
      "learning_rate": 5.237581606585297e-05,
      "loss": 0.0056302003562450405,
      "step": 167850
    },
    {
      "epoch": 47.64689185353392,
      "grad_norm": 1.4807854890823364,
      "learning_rate": 5.23729775759296e-05,
      "loss": 0.0017160283401608466,
      "step": 167860
    },
    {
      "epoch": 47.64973034345728,
      "grad_norm": 0.19436568021774292,
      "learning_rate": 5.2370139086006243e-05,
      "loss": 0.001097014546394348,
      "step": 167870
    },
    {
      "epoch": 47.65256883338064,
      "grad_norm": 0.1556328386068344,
      "learning_rate": 5.236730059608289e-05,
      "loss": 0.006869234889745712,
      "step": 167880
    },
    {
      "epoch": 47.655407323304004,
      "grad_norm": 1.1442910432815552,
      "learning_rate": 5.2364462106159526e-05,
      "loss": 0.0029334781691432,
      "step": 167890
    },
    {
      "epoch": 47.65824581322736,
      "grad_norm": 0.3498494625091553,
      "learning_rate": 5.236162361623617e-05,
      "loss": 0.0010056469589471817,
      "step": 167900
    },
    {
      "epoch": 47.66108430315072,
      "grad_norm": 0.03680235147476196,
      "learning_rate": 5.23587851263128e-05,
      "loss": 0.0005800355225801468,
      "step": 167910
    },
    {
      "epoch": 47.663922793074086,
      "grad_norm": 0.16412007808685303,
      "learning_rate": 5.2355946636389444e-05,
      "loss": 0.0003994446247816086,
      "step": 167920
    },
    {
      "epoch": 47.66676128299745,
      "grad_norm": 0.011415796354413033,
      "learning_rate": 5.235310814646608e-05,
      "loss": 0.0005628641694784165,
      "step": 167930
    },
    {
      "epoch": 47.669599772920805,
      "grad_norm": 0.21112807095050812,
      "learning_rate": 5.235026965654272e-05,
      "loss": 0.006454319506883621,
      "step": 167940
    },
    {
      "epoch": 47.67243826284417,
      "grad_norm": 0.6781848073005676,
      "learning_rate": 5.234743116661937e-05,
      "loss": 0.007819818705320359,
      "step": 167950
    },
    {
      "epoch": 47.67527675276753,
      "grad_norm": 1.879334807395935,
      "learning_rate": 5.2344592676696e-05,
      "loss": 0.0011822026222944259,
      "step": 167960
    },
    {
      "epoch": 47.678115242690886,
      "grad_norm": 0.05251937359571457,
      "learning_rate": 5.234175418677264e-05,
      "loss": 0.0031770091503858567,
      "step": 167970
    },
    {
      "epoch": 47.68095373261425,
      "grad_norm": 1.402992606163025,
      "learning_rate": 5.233891569684928e-05,
      "loss": 0.0007762029767036438,
      "step": 167980
    },
    {
      "epoch": 47.68379222253761,
      "grad_norm": 1.045709490776062,
      "learning_rate": 5.233607720692591e-05,
      "loss": 0.0005439357832074165,
      "step": 167990
    },
    {
      "epoch": 47.68663071246097,
      "grad_norm": 0.02092956192791462,
      "learning_rate": 5.2333238717002554e-05,
      "loss": 0.002361937612295151,
      "step": 168000
    },
    {
      "epoch": 47.68663071246097,
      "eval_accuracy": 0.9783175430787817,
      "eval_loss": 0.07708636671304703,
      "eval_runtime": 32.529,
      "eval_samples_per_second": 483.476,
      "eval_steps_per_second": 7.562,
      "step": 168000
    },
    {
      "epoch": 47.68946920238433,
      "grad_norm": 0.3399779200553894,
      "learning_rate": 5.23304002270792e-05,
      "loss": 0.00028940942138433456,
      "step": 168010
    },
    {
      "epoch": 47.69230769230769,
      "grad_norm": 0.020810747519135475,
      "learning_rate": 5.232756173715584e-05,
      "loss": 0.0006418092176318169,
      "step": 168020
    },
    {
      "epoch": 47.695146182231056,
      "grad_norm": 0.03103160485625267,
      "learning_rate": 5.232472324723248e-05,
      "loss": 0.000648314505815506,
      "step": 168030
    },
    {
      "epoch": 47.69798467215441,
      "grad_norm": 0.030476439744234085,
      "learning_rate": 5.232188475730911e-05,
      "loss": 0.00026870854198932646,
      "step": 168040
    },
    {
      "epoch": 47.700823162077775,
      "grad_norm": 5.814699649810791,
      "learning_rate": 5.2319046267385754e-05,
      "loss": 0.0016576172783970833,
      "step": 168050
    },
    {
      "epoch": 47.70366165200114,
      "grad_norm": 0.09303194284439087,
      "learning_rate": 5.231620777746239e-05,
      "loss": 0.0017091328278183936,
      "step": 168060
    },
    {
      "epoch": 47.706500141924494,
      "grad_norm": 12.274508476257324,
      "learning_rate": 5.231336928753903e-05,
      "loss": 0.004623696953058243,
      "step": 168070
    },
    {
      "epoch": 47.70933863184786,
      "grad_norm": 0.24436743557453156,
      "learning_rate": 5.231053079761568e-05,
      "loss": 0.004368770122528076,
      "step": 168080
    },
    {
      "epoch": 47.71217712177122,
      "grad_norm": 6.048383712768555,
      "learning_rate": 5.230769230769231e-05,
      "loss": 0.0047577288001775745,
      "step": 168090
    },
    {
      "epoch": 47.715015611694575,
      "grad_norm": 0.010169700719416142,
      "learning_rate": 5.230485381776895e-05,
      "loss": 0.006274067610502243,
      "step": 168100
    },
    {
      "epoch": 47.71785410161794,
      "grad_norm": 1.0150330066680908,
      "learning_rate": 5.230201532784559e-05,
      "loss": 0.0009244069457054138,
      "step": 168110
    },
    {
      "epoch": 47.7206925915413,
      "grad_norm": 0.14437630772590637,
      "learning_rate": 5.2299176837922224e-05,
      "loss": 0.0009477993473410607,
      "step": 168120
    },
    {
      "epoch": 47.723531081464664,
      "grad_norm": 11.315572738647461,
      "learning_rate": 5.229662219699121e-05,
      "loss": 0.01503133773803711,
      "step": 168130
    },
    {
      "epoch": 47.72636957138802,
      "grad_norm": 0.1693817675113678,
      "learning_rate": 5.2293783707067845e-05,
      "loss": 0.001133992150425911,
      "step": 168140
    },
    {
      "epoch": 47.72920806131138,
      "grad_norm": 16.55990219116211,
      "learning_rate": 5.229094521714448e-05,
      "loss": 0.00455244705080986,
      "step": 168150
    },
    {
      "epoch": 47.732046551234745,
      "grad_norm": 0.017984464764595032,
      "learning_rate": 5.228810672722112e-05,
      "loss": 0.0014146020635962485,
      "step": 168160
    },
    {
      "epoch": 47.7348850411581,
      "grad_norm": 0.028887253254652023,
      "learning_rate": 5.2285268237297755e-05,
      "loss": 0.0017999103292822838,
      "step": 168170
    },
    {
      "epoch": 47.737723531081464,
      "grad_norm": 0.09977608919143677,
      "learning_rate": 5.22824297473744e-05,
      "loss": 0.0006079711019992828,
      "step": 168180
    },
    {
      "epoch": 47.74056202100483,
      "grad_norm": 0.15819256007671356,
      "learning_rate": 5.2279591257451045e-05,
      "loss": 0.002408851683139801,
      "step": 168190
    },
    {
      "epoch": 47.74340051092818,
      "grad_norm": 0.05671108141541481,
      "learning_rate": 5.227675276752768e-05,
      "loss": 0.001308099552989006,
      "step": 168200
    },
    {
      "epoch": 47.746239000851546,
      "grad_norm": 0.2843754291534424,
      "learning_rate": 5.227391427760432e-05,
      "loss": 0.007566286623477936,
      "step": 168210
    },
    {
      "epoch": 47.74907749077491,
      "grad_norm": 0.20854462683200836,
      "learning_rate": 5.2271075787680956e-05,
      "loss": 0.0013257438316941262,
      "step": 168220
    },
    {
      "epoch": 47.75191598069827,
      "grad_norm": 0.7090956568717957,
      "learning_rate": 5.22682372977576e-05,
      "loss": 0.005258341878652572,
      "step": 168230
    },
    {
      "epoch": 47.75475447062163,
      "grad_norm": 0.012205834500491619,
      "learning_rate": 5.226539880783423e-05,
      "loss": 0.0010363118723034858,
      "step": 168240
    },
    {
      "epoch": 47.75759296054499,
      "grad_norm": 0.01964578591287136,
      "learning_rate": 5.2262560317910866e-05,
      "loss": 0.0006534801796078682,
      "step": 168250
    },
    {
      "epoch": 47.76043145046835,
      "grad_norm": 0.04396224021911621,
      "learning_rate": 5.225972182798752e-05,
      "loss": 0.0006730172783136368,
      "step": 168260
    },
    {
      "epoch": 47.76326994039171,
      "grad_norm": 2.2706873416900635,
      "learning_rate": 5.2256883338064156e-05,
      "loss": 0.000860372930765152,
      "step": 168270
    },
    {
      "epoch": 47.76610843031507,
      "grad_norm": 0.499252587556839,
      "learning_rate": 5.225404484814079e-05,
      "loss": 0.0005581287667155265,
      "step": 168280
    },
    {
      "epoch": 47.768946920238434,
      "grad_norm": 10.355218887329102,
      "learning_rate": 5.225120635821743e-05,
      "loss": 0.003390992432832718,
      "step": 168290
    },
    {
      "epoch": 47.7717854101618,
      "grad_norm": 0.3579513430595398,
      "learning_rate": 5.2248367868294066e-05,
      "loss": 0.002184603177011013,
      "step": 168300
    },
    {
      "epoch": 47.77462390008515,
      "grad_norm": 0.030845995992422104,
      "learning_rate": 5.224552937837071e-05,
      "loss": 0.0020654793828725816,
      "step": 168310
    },
    {
      "epoch": 47.777462390008516,
      "grad_norm": 0.2472856044769287,
      "learning_rate": 5.224269088844734e-05,
      "loss": 0.0015220526605844499,
      "step": 168320
    },
    {
      "epoch": 47.78030087993188,
      "grad_norm": 0.1597907692193985,
      "learning_rate": 5.223985239852399e-05,
      "loss": 0.005758265405893326,
      "step": 168330
    },
    {
      "epoch": 47.783139369855235,
      "grad_norm": 0.032328344881534576,
      "learning_rate": 5.223701390860063e-05,
      "loss": 0.005417917668819427,
      "step": 168340
    },
    {
      "epoch": 47.7859778597786,
      "grad_norm": 0.2601768374443054,
      "learning_rate": 5.2234175418677266e-05,
      "loss": 0.0005021145567297936,
      "step": 168350
    },
    {
      "epoch": 47.78881634970196,
      "grad_norm": 0.043230459094047546,
      "learning_rate": 5.223133692875391e-05,
      "loss": 0.00043454207479953766,
      "step": 168360
    },
    {
      "epoch": 47.791654839625316,
      "grad_norm": 0.06415538489818573,
      "learning_rate": 5.222849843883054e-05,
      "loss": 0.000359492190182209,
      "step": 168370
    },
    {
      "epoch": 47.79449332954868,
      "grad_norm": 0.13824020326137543,
      "learning_rate": 5.222565994890718e-05,
      "loss": 0.00033246222883462904,
      "step": 168380
    },
    {
      "epoch": 47.79733181947204,
      "grad_norm": 0.01076455321162939,
      "learning_rate": 5.2222821458983825e-05,
      "loss": 0.003440087288618088,
      "step": 168390
    },
    {
      "epoch": 47.800170309395405,
      "grad_norm": 0.4593620002269745,
      "learning_rate": 5.2219982969060466e-05,
      "loss": 0.000251116044819355,
      "step": 168400
    },
    {
      "epoch": 47.80300879931876,
      "grad_norm": 0.11444472521543503,
      "learning_rate": 5.22171444791371e-05,
      "loss": 0.0006476037204265594,
      "step": 168410
    },
    {
      "epoch": 47.805847289242124,
      "grad_norm": 0.0401892364025116,
      "learning_rate": 5.221430598921374e-05,
      "loss": 0.00023717451840639115,
      "step": 168420
    },
    {
      "epoch": 47.80868577916549,
      "grad_norm": 0.07347897440195084,
      "learning_rate": 5.221146749929038e-05,
      "loss": 0.0036431849002838135,
      "step": 168430
    },
    {
      "epoch": 47.81152426908884,
      "grad_norm": 2.086364507675171,
      "learning_rate": 5.220862900936702e-05,
      "loss": 0.0055583551526069645,
      "step": 168440
    },
    {
      "epoch": 47.814362759012205,
      "grad_norm": 0.22393475472927094,
      "learning_rate": 5.220579051944365e-05,
      "loss": 0.0014174820855259896,
      "step": 168450
    },
    {
      "epoch": 47.81720124893557,
      "grad_norm": 3.2646610736846924,
      "learning_rate": 5.22029520295203e-05,
      "loss": 0.003411164879798889,
      "step": 168460
    },
    {
      "epoch": 47.820039738858924,
      "grad_norm": 0.07020312547683716,
      "learning_rate": 5.220011353959694e-05,
      "loss": 0.000320008397102356,
      "step": 168470
    },
    {
      "epoch": 47.82287822878229,
      "grad_norm": 0.5748687982559204,
      "learning_rate": 5.219727504967358e-05,
      "loss": 0.002785246819257736,
      "step": 168480
    },
    {
      "epoch": 47.82571671870565,
      "grad_norm": 0.030405137687921524,
      "learning_rate": 5.219443655975021e-05,
      "loss": 0.002555319294333458,
      "step": 168490
    },
    {
      "epoch": 47.82855520862901,
      "grad_norm": 0.6916009187698364,
      "learning_rate": 5.219159806982685e-05,
      "loss": 0.0007593544200062752,
      "step": 168500
    },
    {
      "epoch": 47.82855520862901,
      "eval_accuracy": 0.9743116932663572,
      "eval_loss": 0.09616226702928543,
      "eval_runtime": 32.4671,
      "eval_samples_per_second": 484.398,
      "eval_steps_per_second": 7.577,
      "step": 168500
    },
    {
      "epoch": 47.83139369855237,
      "grad_norm": 0.19176864624023438,
      "learning_rate": 5.218875957990349e-05,
      "loss": 0.0015144472941756248,
      "step": 168510
    },
    {
      "epoch": 47.83423218847573,
      "grad_norm": 0.05753355100750923,
      "learning_rate": 5.218592108998013e-05,
      "loss": 0.004366567730903626,
      "step": 168520
    },
    {
      "epoch": 47.837070678399094,
      "grad_norm": 0.2718294858932495,
      "learning_rate": 5.218308260005678e-05,
      "loss": 0.0005094885826110839,
      "step": 168530
    },
    {
      "epoch": 47.83990916832245,
      "grad_norm": 0.16654092073440552,
      "learning_rate": 5.218024411013341e-05,
      "loss": 0.008084045350551605,
      "step": 168540
    },
    {
      "epoch": 47.84274765824581,
      "grad_norm": 1.7645411491394043,
      "learning_rate": 5.217740562021005e-05,
      "loss": 0.0013424735516309738,
      "step": 168550
    },
    {
      "epoch": 47.845586148169176,
      "grad_norm": 0.1354674994945526,
      "learning_rate": 5.217456713028669e-05,
      "loss": 0.0007955929264426232,
      "step": 168560
    },
    {
      "epoch": 47.84842463809253,
      "grad_norm": 0.44384124875068665,
      "learning_rate": 5.217172864036333e-05,
      "loss": 0.0030870558694005013,
      "step": 168570
    },
    {
      "epoch": 47.851263128015894,
      "grad_norm": 0.2225676029920578,
      "learning_rate": 5.2168890150439964e-05,
      "loss": 0.001619594544172287,
      "step": 168580
    },
    {
      "epoch": 47.85410161793926,
      "grad_norm": 0.09802650660276413,
      "learning_rate": 5.216605166051661e-05,
      "loss": 0.0011863468214869499,
      "step": 168590
    },
    {
      "epoch": 47.85694010786262,
      "grad_norm": 2.1131603717803955,
      "learning_rate": 5.216321317059325e-05,
      "loss": 0.005621275305747986,
      "step": 168600
    },
    {
      "epoch": 47.859778597785976,
      "grad_norm": 8.648219108581543,
      "learning_rate": 5.216037468066989e-05,
      "loss": 0.012919063866138458,
      "step": 168610
    },
    {
      "epoch": 47.86261708770934,
      "grad_norm": 0.16001147031784058,
      "learning_rate": 5.215753619074652e-05,
      "loss": 0.004187991470098495,
      "step": 168620
    },
    {
      "epoch": 47.8654555776327,
      "grad_norm": 0.6787115931510925,
      "learning_rate": 5.2154697700823164e-05,
      "loss": 0.007874543964862823,
      "step": 168630
    },
    {
      "epoch": 47.86829406755606,
      "grad_norm": 13.137056350708008,
      "learning_rate": 5.21518592108998e-05,
      "loss": 0.010380356013774872,
      "step": 168640
    },
    {
      "epoch": 47.87113255747942,
      "grad_norm": 0.4771811068058014,
      "learning_rate": 5.214902072097644e-05,
      "loss": 0.002875529415905476,
      "step": 168650
    },
    {
      "epoch": 47.87397104740278,
      "grad_norm": 0.2702544629573822,
      "learning_rate": 5.214618223105309e-05,
      "loss": 0.0008950099349021911,
      "step": 168660
    },
    {
      "epoch": 47.876809537326146,
      "grad_norm": 4.477654933929443,
      "learning_rate": 5.214334374112972e-05,
      "loss": 0.0025077000260353088,
      "step": 168670
    },
    {
      "epoch": 47.8796480272495,
      "grad_norm": 0.18927837908267975,
      "learning_rate": 5.2140505251206364e-05,
      "loss": 0.00776558592915535,
      "step": 168680
    },
    {
      "epoch": 47.882486517172865,
      "grad_norm": 0.42188459634780884,
      "learning_rate": 5.2137666761283e-05,
      "loss": 0.013357076048851012,
      "step": 168690
    },
    {
      "epoch": 47.88532500709623,
      "grad_norm": 0.09594525396823883,
      "learning_rate": 5.213482827135964e-05,
      "loss": 0.0006707321852445602,
      "step": 168700
    },
    {
      "epoch": 47.88816349701958,
      "grad_norm": 0.3196571469306946,
      "learning_rate": 5.2131989781436275e-05,
      "loss": 0.0007226757705211639,
      "step": 168710
    },
    {
      "epoch": 47.891001986942946,
      "grad_norm": 0.3872777819633484,
      "learning_rate": 5.212915129151291e-05,
      "loss": 0.0008358694612979889,
      "step": 168720
    },
    {
      "epoch": 47.89384047686631,
      "grad_norm": 0.20607343316078186,
      "learning_rate": 5.2126312801589564e-05,
      "loss": 0.0006414493545889854,
      "step": 168730
    },
    {
      "epoch": 47.896678966789665,
      "grad_norm": 1.1029314994812012,
      "learning_rate": 5.21234743116662e-05,
      "loss": 0.00043768975883722304,
      "step": 168740
    },
    {
      "epoch": 47.89951745671303,
      "grad_norm": 0.05939270183444023,
      "learning_rate": 5.212063582174283e-05,
      "loss": 0.001099812053143978,
      "step": 168750
    },
    {
      "epoch": 47.90235594663639,
      "grad_norm": 0.5977734327316284,
      "learning_rate": 5.2117797331819475e-05,
      "loss": 0.0009144533425569535,
      "step": 168760
    },
    {
      "epoch": 47.90519443655975,
      "grad_norm": 0.2970137894153595,
      "learning_rate": 5.211495884189611e-05,
      "loss": 0.0012066274881362914,
      "step": 168770
    },
    {
      "epoch": 47.90803292648311,
      "grad_norm": 0.028434520587325096,
      "learning_rate": 5.211212035197275e-05,
      "loss": 0.0005499390885233879,
      "step": 168780
    },
    {
      "epoch": 47.91087141640647,
      "grad_norm": 0.2620388865470886,
      "learning_rate": 5.21092818620494e-05,
      "loss": 0.0009521033614873886,
      "step": 168790
    },
    {
      "epoch": 47.913709906329835,
      "grad_norm": 0.12066516280174255,
      "learning_rate": 5.210644337212603e-05,
      "loss": 0.0028778892010450363,
      "step": 168800
    },
    {
      "epoch": 47.91654839625319,
      "grad_norm": 0.051235031336545944,
      "learning_rate": 5.2103604882202675e-05,
      "loss": 0.0011702662333846093,
      "step": 168810
    },
    {
      "epoch": 47.919386886176554,
      "grad_norm": 0.11312884837388992,
      "learning_rate": 5.210076639227931e-05,
      "loss": 0.01661757528781891,
      "step": 168820
    },
    {
      "epoch": 47.92222537609992,
      "grad_norm": 0.1583506315946579,
      "learning_rate": 5.209792790235595e-05,
      "loss": 0.0075116358697414395,
      "step": 168830
    },
    {
      "epoch": 47.92506386602327,
      "grad_norm": 0.16481782495975494,
      "learning_rate": 5.2095089412432585e-05,
      "loss": 0.0015135286375880241,
      "step": 168840
    },
    {
      "epoch": 47.927902355946635,
      "grad_norm": 0.33268028497695923,
      "learning_rate": 5.209225092250922e-05,
      "loss": 0.0020930180326104164,
      "step": 168850
    },
    {
      "epoch": 47.93074084587,
      "grad_norm": 0.0243054311722517,
      "learning_rate": 5.208941243258587e-05,
      "loss": 0.003306695073843002,
      "step": 168860
    },
    {
      "epoch": 47.93357933579336,
      "grad_norm": 3.1700685024261475,
      "learning_rate": 5.208657394266251e-05,
      "loss": 0.0020365085452795028,
      "step": 168870
    },
    {
      "epoch": 47.93641782571672,
      "grad_norm": 0.037230346351861954,
      "learning_rate": 5.2083735452739144e-05,
      "loss": 0.0010077066719532013,
      "step": 168880
    },
    {
      "epoch": 47.93925631564008,
      "grad_norm": 0.05465046688914299,
      "learning_rate": 5.2080896962815785e-05,
      "loss": 0.003841538354754448,
      "step": 168890
    },
    {
      "epoch": 47.94209480556344,
      "grad_norm": 1.094699501991272,
      "learning_rate": 5.207805847289242e-05,
      "loss": 0.0013385560363531114,
      "step": 168900
    },
    {
      "epoch": 47.9449332954868,
      "grad_norm": 0.8456916809082031,
      "learning_rate": 5.207521998296906e-05,
      "loss": 0.001687139645218849,
      "step": 168910
    },
    {
      "epoch": 47.94777178541016,
      "grad_norm": 0.7954090237617493,
      "learning_rate": 5.207238149304571e-05,
      "loss": 0.0013142066076397897,
      "step": 168920
    },
    {
      "epoch": 47.950610275333524,
      "grad_norm": 0.4194909930229187,
      "learning_rate": 5.2069543003122344e-05,
      "loss": 0.0054662637412548065,
      "step": 168930
    },
    {
      "epoch": 47.95344876525688,
      "grad_norm": 0.3298946022987366,
      "learning_rate": 5.2066704513198985e-05,
      "loss": 0.0028723591938614847,
      "step": 168940
    },
    {
      "epoch": 47.95628725518024,
      "grad_norm": 0.07811582088470459,
      "learning_rate": 5.206386602327562e-05,
      "loss": 0.0013351984322071075,
      "step": 168950
    },
    {
      "epoch": 47.959125745103606,
      "grad_norm": 0.049127377569675446,
      "learning_rate": 5.2061027533352255e-05,
      "loss": 0.003604147583246231,
      "step": 168960
    },
    {
      "epoch": 47.96196423502697,
      "grad_norm": 6.8462724685668945,
      "learning_rate": 5.2058189043428896e-05,
      "loss": 0.0033238433301448823,
      "step": 168970
    },
    {
      "epoch": 47.964802724950324,
      "grad_norm": 0.06216711923480034,
      "learning_rate": 5.205535055350553e-05,
      "loss": 0.0019249454140663146,
      "step": 168980
    },
    {
      "epoch": 47.96764121487369,
      "grad_norm": 0.15977944433689117,
      "learning_rate": 5.205251206358218e-05,
      "loss": 0.002648826874792576,
      "step": 168990
    },
    {
      "epoch": 47.97047970479705,
      "grad_norm": 0.7573155760765076,
      "learning_rate": 5.204967357365882e-05,
      "loss": 0.0034080419689416887,
      "step": 169000
    },
    {
      "epoch": 47.97047970479705,
      "eval_accuracy": 0.9743752781840147,
      "eval_loss": 0.09073596447706223,
      "eval_runtime": 32.4432,
      "eval_samples_per_second": 484.756,
      "eval_steps_per_second": 7.582,
      "step": 169000
    },
    {
      "epoch": 47.973318194720406,
      "grad_norm": 0.48568931221961975,
      "learning_rate": 5.2046835083735455e-05,
      "loss": 0.002006968855857849,
      "step": 169010
    },
    {
      "epoch": 47.97615668464377,
      "grad_norm": 0.05867546796798706,
      "learning_rate": 5.2043996593812096e-05,
      "loss": 0.0007272759452462196,
      "step": 169020
    },
    {
      "epoch": 47.97899517456713,
      "grad_norm": 0.030273746699094772,
      "learning_rate": 5.204115810388873e-05,
      "loss": 0.0006490414962172508,
      "step": 169030
    },
    {
      "epoch": 47.981833664490495,
      "grad_norm": 0.11249876767396927,
      "learning_rate": 5.203831961396537e-05,
      "loss": 0.0006806440651416779,
      "step": 169040
    },
    {
      "epoch": 47.98467215441385,
      "grad_norm": 0.08497555553913116,
      "learning_rate": 5.203548112404201e-05,
      "loss": 0.005417253822088242,
      "step": 169050
    },
    {
      "epoch": 47.98751064433721,
      "grad_norm": 0.8877143263816833,
      "learning_rate": 5.2032642634118655e-05,
      "loss": 0.0015332011505961418,
      "step": 169060
    },
    {
      "epoch": 47.990349134260576,
      "grad_norm": 0.28416627645492554,
      "learning_rate": 5.2029804144195296e-05,
      "loss": 0.004342702031135559,
      "step": 169070
    },
    {
      "epoch": 47.99318762418393,
      "grad_norm": 0.04870780184864998,
      "learning_rate": 5.202696565427193e-05,
      "loss": 0.0015967017039656638,
      "step": 169080
    },
    {
      "epoch": 47.996026114107295,
      "grad_norm": 0.4140115976333618,
      "learning_rate": 5.2024127164348565e-05,
      "loss": 0.0004907928407192231,
      "step": 169090
    },
    {
      "epoch": 47.99886460403066,
      "grad_norm": 1.8854641914367676,
      "learning_rate": 5.202128867442521e-05,
      "loss": 0.0015880800783634186,
      "step": 169100
    },
    {
      "epoch": 48.00170309395401,
      "grad_norm": 1.3292404413223267,
      "learning_rate": 5.201845018450184e-05,
      "loss": 0.008714805543422698,
      "step": 169110
    },
    {
      "epoch": 48.004541583877376,
      "grad_norm": 0.45640259981155396,
      "learning_rate": 5.201561169457849e-05,
      "loss": 0.0004358574748039246,
      "step": 169120
    },
    {
      "epoch": 48.00738007380074,
      "grad_norm": 0.20175772905349731,
      "learning_rate": 5.201277320465513e-05,
      "loss": 0.000773577019572258,
      "step": 169130
    },
    {
      "epoch": 48.0102185637241,
      "grad_norm": 0.16190128028392792,
      "learning_rate": 5.2009934714731766e-05,
      "loss": 0.0019964657723903654,
      "step": 169140
    },
    {
      "epoch": 48.01305705364746,
      "grad_norm": 0.015249173156917095,
      "learning_rate": 5.200709622480841e-05,
      "loss": 0.006442061811685562,
      "step": 169150
    },
    {
      "epoch": 48.01589554357082,
      "grad_norm": 0.9231210350990295,
      "learning_rate": 5.200425773488504e-05,
      "loss": 0.0018255233764648438,
      "step": 169160
    },
    {
      "epoch": 48.018734033494184,
      "grad_norm": 0.03372298181056976,
      "learning_rate": 5.200141924496168e-05,
      "loss": 0.0012127842754125595,
      "step": 169170
    },
    {
      "epoch": 48.02157252341754,
      "grad_norm": 1.0563827753067017,
      "learning_rate": 5.199858075503832e-05,
      "loss": 0.0007660938426852226,
      "step": 169180
    },
    {
      "epoch": 48.0244110133409,
      "grad_norm": 7.174070358276367,
      "learning_rate": 5.1995742265114966e-05,
      "loss": 0.003965583071112633,
      "step": 169190
    },
    {
      "epoch": 48.027249503264265,
      "grad_norm": 0.040483344346284866,
      "learning_rate": 5.199290377519161e-05,
      "loss": 0.002101879194378853,
      "step": 169200
    },
    {
      "epoch": 48.03008799318762,
      "grad_norm": 0.14513573050498962,
      "learning_rate": 5.199006528526824e-05,
      "loss": 0.0010218260809779166,
      "step": 169210
    },
    {
      "epoch": 48.032926483110984,
      "grad_norm": 0.08491690456867218,
      "learning_rate": 5.1987226795344876e-05,
      "loss": 0.0019350145012140274,
      "step": 169220
    },
    {
      "epoch": 48.03576497303435,
      "grad_norm": 0.5712893009185791,
      "learning_rate": 5.198438830542152e-05,
      "loss": 0.004747232049703598,
      "step": 169230
    },
    {
      "epoch": 48.03860346295771,
      "grad_norm": 0.2694304585456848,
      "learning_rate": 5.198154981549815e-05,
      "loss": 0.0013046547770500182,
      "step": 169240
    },
    {
      "epoch": 48.041441952881065,
      "grad_norm": 0.005628513637930155,
      "learning_rate": 5.1978711325574794e-05,
      "loss": 0.004674214869737625,
      "step": 169250
    },
    {
      "epoch": 48.04428044280443,
      "grad_norm": 0.03180970996618271,
      "learning_rate": 5.197587283565144e-05,
      "loss": 0.0007027646526694298,
      "step": 169260
    },
    {
      "epoch": 48.04711893272779,
      "grad_norm": 0.02389702759683132,
      "learning_rate": 5.1973034345728076e-05,
      "loss": 0.0015586500987410546,
      "step": 169270
    },
    {
      "epoch": 48.04995742265115,
      "grad_norm": 0.39796867966651917,
      "learning_rate": 5.197019585580472e-05,
      "loss": 0.0013189222663640976,
      "step": 169280
    },
    {
      "epoch": 48.05279591257451,
      "grad_norm": 0.365920752286911,
      "learning_rate": 5.196735736588135e-05,
      "loss": 0.001648235134780407,
      "step": 169290
    },
    {
      "epoch": 48.05563440249787,
      "grad_norm": 0.856377363204956,
      "learning_rate": 5.1964518875957994e-05,
      "loss": 0.0018580781295895576,
      "step": 169300
    },
    {
      "epoch": 48.05847289242123,
      "grad_norm": 0.05779247730970383,
      "learning_rate": 5.196168038603463e-05,
      "loss": 0.002620731666684151,
      "step": 169310
    },
    {
      "epoch": 48.06131138234459,
      "grad_norm": 0.1380361169576645,
      "learning_rate": 5.1958841896111276e-05,
      "loss": 0.0025946395471692085,
      "step": 169320
    },
    {
      "epoch": 48.064149872267954,
      "grad_norm": 15.447072982788086,
      "learning_rate": 5.195600340618791e-05,
      "loss": 0.004500969499349594,
      "step": 169330
    },
    {
      "epoch": 48.06698836219132,
      "grad_norm": 0.1305493265390396,
      "learning_rate": 5.195316491626455e-05,
      "loss": 0.0011416334658861161,
      "step": 169340
    },
    {
      "epoch": 48.06982685211467,
      "grad_norm": 0.08362076431512833,
      "learning_rate": 5.195032642634119e-05,
      "loss": 0.008530428260564804,
      "step": 169350
    },
    {
      "epoch": 48.072665342038036,
      "grad_norm": 0.3601750433444977,
      "learning_rate": 5.194748793641783e-05,
      "loss": 0.00043891221284866335,
      "step": 169360
    },
    {
      "epoch": 48.0755038319614,
      "grad_norm": 8.761248588562012,
      "learning_rate": 5.194464944649446e-05,
      "loss": 0.0077661886811256405,
      "step": 169370
    },
    {
      "epoch": 48.078342321884755,
      "grad_norm": 0.0535542257130146,
      "learning_rate": 5.1941810956571104e-05,
      "loss": 0.0008526010438799858,
      "step": 169380
    },
    {
      "epoch": 48.08118081180812,
      "grad_norm": 0.030601203441619873,
      "learning_rate": 5.193897246664775e-05,
      "loss": 0.0011587688699364663,
      "step": 169390
    },
    {
      "epoch": 48.08401930173148,
      "grad_norm": 0.3816034197807312,
      "learning_rate": 5.193613397672439e-05,
      "loss": 0.0017326164990663529,
      "step": 169400
    },
    {
      "epoch": 48.086857791654836,
      "grad_norm": 0.01729806885123253,
      "learning_rate": 5.193329548680103e-05,
      "loss": 0.0002913212403655052,
      "step": 169410
    },
    {
      "epoch": 48.0896962815782,
      "grad_norm": 0.443890243768692,
      "learning_rate": 5.193045699687766e-05,
      "loss": 0.0007170522585511208,
      "step": 169420
    },
    {
      "epoch": 48.09253477150156,
      "grad_norm": 0.17501603066921234,
      "learning_rate": 5.19276185069543e-05,
      "loss": 0.00034133903682231903,
      "step": 169430
    },
    {
      "epoch": 48.095373261424925,
      "grad_norm": 0.01665317267179489,
      "learning_rate": 5.192478001703094e-05,
      "loss": 0.001759827882051468,
      "step": 169440
    },
    {
      "epoch": 48.09821175134828,
      "grad_norm": 0.12499500811100006,
      "learning_rate": 5.1921941527107574e-05,
      "loss": 0.0005489757284522056,
      "step": 169450
    },
    {
      "epoch": 48.10105024127164,
      "grad_norm": 0.014039736241102219,
      "learning_rate": 5.191910303718422e-05,
      "loss": 0.0004667326807975769,
      "step": 169460
    },
    {
      "epoch": 48.103888731195006,
      "grad_norm": 0.06626798957586288,
      "learning_rate": 5.191626454726086e-05,
      "loss": 0.0006232718005776405,
      "step": 169470
    },
    {
      "epoch": 48.10672722111836,
      "grad_norm": 0.12584063410758972,
      "learning_rate": 5.19134260573375e-05,
      "loss": 0.0003408443182706833,
      "step": 169480
    },
    {
      "epoch": 48.109565711041725,
      "grad_norm": 0.44045889377593994,
      "learning_rate": 5.191058756741414e-05,
      "loss": 0.000580497831106186,
      "step": 169490
    },
    {
      "epoch": 48.11240420096509,
      "grad_norm": 0.012688093818724155,
      "learning_rate": 5.1907749077490774e-05,
      "loss": 0.0005011571571230889,
      "step": 169500
    },
    {
      "epoch": 48.11240420096509,
      "eval_accuracy": 0.9771730145609462,
      "eval_loss": 0.08051182329654694,
      "eval_runtime": 32.8314,
      "eval_samples_per_second": 479.023,
      "eval_steps_per_second": 7.493,
      "step": 169500
    },
    {
      "epoch": 48.11524269088845,
      "grad_norm": 0.13200466334819794,
      "learning_rate": 5.1904910587567415e-05,
      "loss": 0.0029042115435004233,
      "step": 169510
    },
    {
      "epoch": 48.11808118081181,
      "grad_norm": 0.28291431069374084,
      "learning_rate": 5.190207209764406e-05,
      "loss": 0.0011116934940218926,
      "step": 169520
    },
    {
      "epoch": 48.12091967073517,
      "grad_norm": 0.015270712785422802,
      "learning_rate": 5.18992336077207e-05,
      "loss": 0.00032479409128427505,
      "step": 169530
    },
    {
      "epoch": 48.12375816065853,
      "grad_norm": 0.08360501378774643,
      "learning_rate": 5.189639511779734e-05,
      "loss": 0.0005689503625035286,
      "step": 169540
    },
    {
      "epoch": 48.12659665058189,
      "grad_norm": 0.24155980348587036,
      "learning_rate": 5.1893556627873974e-05,
      "loss": 0.00032717250287532805,
      "step": 169550
    },
    {
      "epoch": 48.12943514050525,
      "grad_norm": 0.07552524656057358,
      "learning_rate": 5.189071813795061e-05,
      "loss": 0.00027983933687210084,
      "step": 169560
    },
    {
      "epoch": 48.132273630428614,
      "grad_norm": 0.009576866403222084,
      "learning_rate": 5.188787964802725e-05,
      "loss": 0.0002992328256368637,
      "step": 169570
    },
    {
      "epoch": 48.13511212035197,
      "grad_norm": 0.054039116948843,
      "learning_rate": 5.1885041158103884e-05,
      "loss": 0.0003409136086702347,
      "step": 169580
    },
    {
      "epoch": 48.13795061027533,
      "grad_norm": 0.020957382395863533,
      "learning_rate": 5.188220266818053e-05,
      "loss": 0.00027796104550361635,
      "step": 169590
    },
    {
      "epoch": 48.140789100198695,
      "grad_norm": 0.03670799359679222,
      "learning_rate": 5.1879364178257174e-05,
      "loss": 0.00024406313896179198,
      "step": 169600
    },
    {
      "epoch": 48.14362759012206,
      "grad_norm": 0.038834430277347565,
      "learning_rate": 5.187652568833381e-05,
      "loss": 0.0005086598917841912,
      "step": 169610
    },
    {
      "epoch": 48.146466080045414,
      "grad_norm": 0.052877455949783325,
      "learning_rate": 5.187368719841045e-05,
      "loss": 0.00035694371908903124,
      "step": 169620
    },
    {
      "epoch": 48.14930456996878,
      "grad_norm": 0.03896389901638031,
      "learning_rate": 5.1870848708487084e-05,
      "loss": 0.0002942768856883049,
      "step": 169630
    },
    {
      "epoch": 48.15214305989214,
      "grad_norm": 7.123621940612793,
      "learning_rate": 5.1868010218563726e-05,
      "loss": 0.0020161252468824387,
      "step": 169640
    },
    {
      "epoch": 48.154981549815496,
      "grad_norm": 0.04535391926765442,
      "learning_rate": 5.186517172864036e-05,
      "loss": 0.0004251953214406967,
      "step": 169650
    },
    {
      "epoch": 48.15782003973886,
      "grad_norm": 0.11210761964321136,
      "learning_rate": 5.186233323871701e-05,
      "loss": 0.0008195187896490097,
      "step": 169660
    },
    {
      "epoch": 48.16065852966222,
      "grad_norm": 0.04501883685588837,
      "learning_rate": 5.185949474879365e-05,
      "loss": 0.0003616882488131523,
      "step": 169670
    },
    {
      "epoch": 48.16349701958558,
      "grad_norm": 0.010544823482632637,
      "learning_rate": 5.1856656258870285e-05,
      "loss": 0.0002958007156848907,
      "step": 169680
    },
    {
      "epoch": 48.16633550950894,
      "grad_norm": 0.00409243069589138,
      "learning_rate": 5.185381776894692e-05,
      "loss": 0.00030428580939769747,
      "step": 169690
    },
    {
      "epoch": 48.1691739994323,
      "grad_norm": 0.16142898797988892,
      "learning_rate": 5.185097927902356e-05,
      "loss": 0.0026078324764966963,
      "step": 169700
    },
    {
      "epoch": 48.172012489355666,
      "grad_norm": 0.041417717933654785,
      "learning_rate": 5.1848140789100195e-05,
      "loss": 0.0023034652695059776,
      "step": 169710
    },
    {
      "epoch": 48.17485097927902,
      "grad_norm": 0.06607963889837265,
      "learning_rate": 5.184530229917684e-05,
      "loss": 0.005011995136737823,
      "step": 169720
    },
    {
      "epoch": 48.177689469202384,
      "grad_norm": 0.08289404213428497,
      "learning_rate": 5.1842463809253485e-05,
      "loss": 0.005770009011030197,
      "step": 169730
    },
    {
      "epoch": 48.18052795912575,
      "grad_norm": 0.23443347215652466,
      "learning_rate": 5.183962531933012e-05,
      "loss": 0.0032154083251953124,
      "step": 169740
    },
    {
      "epoch": 48.1833664490491,
      "grad_norm": 0.2570763826370239,
      "learning_rate": 5.183678682940676e-05,
      "loss": 0.0017732597887516021,
      "step": 169750
    },
    {
      "epoch": 48.186204938972466,
      "grad_norm": 0.041140682995319366,
      "learning_rate": 5.1833948339483395e-05,
      "loss": 0.00029384605586528777,
      "step": 169760
    },
    {
      "epoch": 48.18904342889583,
      "grad_norm": 0.030419262126088142,
      "learning_rate": 5.183110984956004e-05,
      "loss": 0.0003978312015533447,
      "step": 169770
    },
    {
      "epoch": 48.191881918819185,
      "grad_norm": 0.5438762307167053,
      "learning_rate": 5.182827135963667e-05,
      "loss": 0.0003368331119418144,
      "step": 169780
    },
    {
      "epoch": 48.19472040874255,
      "grad_norm": 0.03378736227750778,
      "learning_rate": 5.182543286971332e-05,
      "loss": 0.0005394203588366508,
      "step": 169790
    },
    {
      "epoch": 48.19755889866591,
      "grad_norm": 0.02825208380818367,
      "learning_rate": 5.1822594379789954e-05,
      "loss": 0.0002612173557281494,
      "step": 169800
    },
    {
      "epoch": 48.20039738858927,
      "grad_norm": 0.449723482131958,
      "learning_rate": 5.1819755889866595e-05,
      "loss": 0.0017918827012181281,
      "step": 169810
    },
    {
      "epoch": 48.20323587851263,
      "grad_norm": 6.63954496383667,
      "learning_rate": 5.181691739994323e-05,
      "loss": 0.002589108794927597,
      "step": 169820
    },
    {
      "epoch": 48.20607436843599,
      "grad_norm": 0.009792627766728401,
      "learning_rate": 5.181407891001987e-05,
      "loss": 0.0001951834186911583,
      "step": 169830
    },
    {
      "epoch": 48.208912858359355,
      "grad_norm": 2.8621339797973633,
      "learning_rate": 5.1811240420096506e-05,
      "loss": 0.0012693621218204498,
      "step": 169840
    },
    {
      "epoch": 48.21175134828271,
      "grad_norm": 0.06013490632176399,
      "learning_rate": 5.180840193017315e-05,
      "loss": 0.0004048362374305725,
      "step": 169850
    },
    {
      "epoch": 48.214589838206074,
      "grad_norm": 0.13003575801849365,
      "learning_rate": 5.1805563440249795e-05,
      "loss": 0.012642888724803925,
      "step": 169860
    },
    {
      "epoch": 48.21742832812944,
      "grad_norm": 0.6260200142860413,
      "learning_rate": 5.180272495032643e-05,
      "loss": 0.0004621027037501335,
      "step": 169870
    },
    {
      "epoch": 48.2202668180528,
      "grad_norm": 0.04754795879125595,
      "learning_rate": 5.179988646040307e-05,
      "loss": 0.0005402198061347008,
      "step": 169880
    },
    {
      "epoch": 48.223105307976155,
      "grad_norm": 0.06795411556959152,
      "learning_rate": 5.1797047970479706e-05,
      "loss": 0.00043850503861904145,
      "step": 169890
    },
    {
      "epoch": 48.22594379789952,
      "grad_norm": 0.007030848879367113,
      "learning_rate": 5.179420948055634e-05,
      "loss": 0.00211766492575407,
      "step": 169900
    },
    {
      "epoch": 48.22878228782288,
      "grad_norm": 0.4730381369590759,
      "learning_rate": 5.179137099063298e-05,
      "loss": 0.0004621380940079689,
      "step": 169910
    },
    {
      "epoch": 48.23162077774624,
      "grad_norm": 0.11463402956724167,
      "learning_rate": 5.178853250070963e-05,
      "loss": 0.001957203075289726,
      "step": 169920
    },
    {
      "epoch": 48.2344592676696,
      "grad_norm": 0.6039824485778809,
      "learning_rate": 5.1785694010786265e-05,
      "loss": 0.000527418963611126,
      "step": 169930
    },
    {
      "epoch": 48.23729775759296,
      "grad_norm": 0.08987002819776535,
      "learning_rate": 5.1782855520862906e-05,
      "loss": 0.001402302272617817,
      "step": 169940
    },
    {
      "epoch": 48.24013624751632,
      "grad_norm": 0.07461860775947571,
      "learning_rate": 5.178001703093954e-05,
      "loss": 0.0011251326650381088,
      "step": 169950
    },
    {
      "epoch": 48.24297473743968,
      "grad_norm": 0.23247699439525604,
      "learning_rate": 5.177717854101618e-05,
      "loss": 0.0009967634454369546,
      "step": 169960
    },
    {
      "epoch": 48.245813227363044,
      "grad_norm": 0.3495427668094635,
      "learning_rate": 5.177434005109282e-05,
      "loss": 0.0011664878576993943,
      "step": 169970
    },
    {
      "epoch": 48.24865171728641,
      "grad_norm": 0.23903661966323853,
      "learning_rate": 5.177150156116946e-05,
      "loss": 0.0020857520401477815,
      "step": 169980
    },
    {
      "epoch": 48.25149020720976,
      "grad_norm": 0.35860276222229004,
      "learning_rate": 5.1768663071246106e-05,
      "loss": 0.0015297045931220055,
      "step": 169990
    },
    {
      "epoch": 48.254328697133126,
      "grad_norm": 0.013609755784273148,
      "learning_rate": 5.176582458132274e-05,
      "loss": 0.0011699140071868897,
      "step": 170000
    },
    {
      "epoch": 48.254328697133126,
      "eval_accuracy": 0.9751382971959052,
      "eval_loss": 0.0916023850440979,
      "eval_runtime": 32.5129,
      "eval_samples_per_second": 483.716,
      "eval_steps_per_second": 7.566,
      "step": 170000
    },
    {
      "epoch": 48.25716718705649,
      "grad_norm": 0.37539881467819214,
      "learning_rate": 5.176298609139938e-05,
      "loss": 0.0008285157382488251,
      "step": 170010
    },
    {
      "epoch": 48.260005676979844,
      "grad_norm": 0.010953563265502453,
      "learning_rate": 5.176014760147602e-05,
      "loss": 0.0034502238035202027,
      "step": 170020
    },
    {
      "epoch": 48.26284416690321,
      "grad_norm": 6.054699897766113,
      "learning_rate": 5.175730911155265e-05,
      "loss": 0.0027752116322517393,
      "step": 170030
    },
    {
      "epoch": 48.26568265682657,
      "grad_norm": 0.24358613789081573,
      "learning_rate": 5.175447062162929e-05,
      "loss": 0.00709536075592041,
      "step": 170040
    },
    {
      "epoch": 48.268521146749926,
      "grad_norm": 0.036644309759140015,
      "learning_rate": 5.175163213170593e-05,
      "loss": 0.0018010647967457772,
      "step": 170050
    },
    {
      "epoch": 48.27135963667329,
      "grad_norm": 0.6279668211936951,
      "learning_rate": 5.1748793641782576e-05,
      "loss": 0.0055326387286186215,
      "step": 170060
    },
    {
      "epoch": 48.27419812659665,
      "grad_norm": 0.18827930092811584,
      "learning_rate": 5.174595515185922e-05,
      "loss": 0.0009545262902975083,
      "step": 170070
    },
    {
      "epoch": 48.277036616520014,
      "grad_norm": 0.16071078181266785,
      "learning_rate": 5.174311666193585e-05,
      "loss": 0.003045186214148998,
      "step": 170080
    },
    {
      "epoch": 48.27987510644337,
      "grad_norm": 0.052226122468709946,
      "learning_rate": 5.174027817201249e-05,
      "loss": 0.0010632585734128952,
      "step": 170090
    },
    {
      "epoch": 48.28271359636673,
      "grad_norm": 1.552156686782837,
      "learning_rate": 5.173743968208913e-05,
      "loss": 0.001186223328113556,
      "step": 170100
    },
    {
      "epoch": 48.285552086290096,
      "grad_norm": 0.4222797155380249,
      "learning_rate": 5.173460119216577e-05,
      "loss": 0.0063815474510192875,
      "step": 170110
    },
    {
      "epoch": 48.28839057621345,
      "grad_norm": 1.296903371810913,
      "learning_rate": 5.173176270224242e-05,
      "loss": 0.0023703442886471747,
      "step": 170120
    },
    {
      "epoch": 48.291229066136815,
      "grad_norm": 1.901888132095337,
      "learning_rate": 5.172892421231905e-05,
      "loss": 0.0011137500405311584,
      "step": 170130
    },
    {
      "epoch": 48.29406755606018,
      "grad_norm": 0.20559951663017273,
      "learning_rate": 5.172608572239569e-05,
      "loss": 0.002310704067349434,
      "step": 170140
    },
    {
      "epoch": 48.29690604598353,
      "grad_norm": 0.05210656672716141,
      "learning_rate": 5.172324723247233e-05,
      "loss": 0.0009772177785634995,
      "step": 170150
    },
    {
      "epoch": 48.299744535906896,
      "grad_norm": 0.1821916550397873,
      "learning_rate": 5.172040874254896e-05,
      "loss": 0.00040080733597278593,
      "step": 170160
    },
    {
      "epoch": 48.30258302583026,
      "grad_norm": 0.7238579988479614,
      "learning_rate": 5.1717570252625604e-05,
      "loss": 0.0011339543387293815,
      "step": 170170
    },
    {
      "epoch": 48.30542151575362,
      "grad_norm": 0.14310337603092194,
      "learning_rate": 5.171473176270224e-05,
      "loss": 0.0009576469659805298,
      "step": 170180
    },
    {
      "epoch": 48.30826000567698,
      "grad_norm": 0.038881540298461914,
      "learning_rate": 5.1711893272778886e-05,
      "loss": 0.0015356259420514107,
      "step": 170190
    },
    {
      "epoch": 48.31109849560034,
      "grad_norm": 4.312447547912598,
      "learning_rate": 5.170905478285553e-05,
      "loss": 0.0018091056495904922,
      "step": 170200
    },
    {
      "epoch": 48.3139369855237,
      "grad_norm": 0.06684769690036774,
      "learning_rate": 5.170621629293216e-05,
      "loss": 0.0010862916707992554,
      "step": 170210
    },
    {
      "epoch": 48.31677547544706,
      "grad_norm": 0.41808241605758667,
      "learning_rate": 5.1703377803008804e-05,
      "loss": 0.005118612200021744,
      "step": 170220
    },
    {
      "epoch": 48.31961396537042,
      "grad_norm": 4.225318431854248,
      "learning_rate": 5.170053931308544e-05,
      "loss": 0.001921781338751316,
      "step": 170230
    },
    {
      "epoch": 48.322452455293785,
      "grad_norm": 0.36002129316329956,
      "learning_rate": 5.169770082316208e-05,
      "loss": 0.004115990549325943,
      "step": 170240
    },
    {
      "epoch": 48.32529094521715,
      "grad_norm": 0.19535623490810394,
      "learning_rate": 5.169486233323873e-05,
      "loss": 0.007371518015861511,
      "step": 170250
    },
    {
      "epoch": 48.328129435140504,
      "grad_norm": 0.6150732636451721,
      "learning_rate": 5.169202384331536e-05,
      "loss": 0.009201030433177947,
      "step": 170260
    },
    {
      "epoch": 48.33096792506387,
      "grad_norm": 0.22400610148906708,
      "learning_rate": 5.1689185353392e-05,
      "loss": 0.0021664125844836233,
      "step": 170270
    },
    {
      "epoch": 48.33380641498723,
      "grad_norm": 0.9035622477531433,
      "learning_rate": 5.168634686346864e-05,
      "loss": 0.002152431383728981,
      "step": 170280
    },
    {
      "epoch": 48.336644904910585,
      "grad_norm": 6.398060321807861,
      "learning_rate": 5.168350837354527e-05,
      "loss": 0.0017887163907289506,
      "step": 170290
    },
    {
      "epoch": 48.33948339483395,
      "grad_norm": 0.05517261102795601,
      "learning_rate": 5.1680669883621914e-05,
      "loss": 0.006261145323514938,
      "step": 170300
    },
    {
      "epoch": 48.34232188475731,
      "grad_norm": 0.041457630693912506,
      "learning_rate": 5.167783139369855e-05,
      "loss": 0.0031805258244276048,
      "step": 170310
    },
    {
      "epoch": 48.34516037468067,
      "grad_norm": 0.07920007407665253,
      "learning_rate": 5.16749929037752e-05,
      "loss": 0.011030313372612,
      "step": 170320
    },
    {
      "epoch": 48.34799886460403,
      "grad_norm": 0.29872041940689087,
      "learning_rate": 5.167215441385184e-05,
      "loss": 0.014321193099021912,
      "step": 170330
    },
    {
      "epoch": 48.35083735452739,
      "grad_norm": 0.149387925863266,
      "learning_rate": 5.166931592392847e-05,
      "loss": 0.0015392230823636055,
      "step": 170340
    },
    {
      "epoch": 48.353675844450756,
      "grad_norm": 0.03618776053190231,
      "learning_rate": 5.1666477434005114e-05,
      "loss": 0.002508210577070713,
      "step": 170350
    },
    {
      "epoch": 48.35651433437411,
      "grad_norm": 0.7413767576217651,
      "learning_rate": 5.166363894408175e-05,
      "loss": 0.0013350460678339005,
      "step": 170360
    },
    {
      "epoch": 48.359352824297474,
      "grad_norm": 3.5192720890045166,
      "learning_rate": 5.1660800454158384e-05,
      "loss": 0.00800531953573227,
      "step": 170370
    },
    {
      "epoch": 48.36219131422084,
      "grad_norm": 0.2785535454750061,
      "learning_rate": 5.1657961964235025e-05,
      "loss": 0.0011600932106375695,
      "step": 170380
    },
    {
      "epoch": 48.36502980414419,
      "grad_norm": 0.5922547578811646,
      "learning_rate": 5.165512347431167e-05,
      "loss": 0.0019769491627812386,
      "step": 170390
    },
    {
      "epoch": 48.367868294067556,
      "grad_norm": 0.2807304561138153,
      "learning_rate": 5.165228498438831e-05,
      "loss": 0.0013354629278182983,
      "step": 170400
    },
    {
      "epoch": 48.37070678399092,
      "grad_norm": 0.09057430177927017,
      "learning_rate": 5.164944649446495e-05,
      "loss": 0.0003337142989039421,
      "step": 170410
    },
    {
      "epoch": 48.373545273914274,
      "grad_norm": 0.15213540196418762,
      "learning_rate": 5.1646608004541584e-05,
      "loss": 0.01547018438577652,
      "step": 170420
    },
    {
      "epoch": 48.37638376383764,
      "grad_norm": 0.008683275431394577,
      "learning_rate": 5.1643769514618225e-05,
      "loss": 0.0011819789186120033,
      "step": 170430
    },
    {
      "epoch": 48.379222253761,
      "grad_norm": 0.0502949059009552,
      "learning_rate": 5.164093102469486e-05,
      "loss": 0.0005090538412332535,
      "step": 170440
    },
    {
      "epoch": 48.38206074368436,
      "grad_norm": 0.6799803972244263,
      "learning_rate": 5.163809253477151e-05,
      "loss": 0.004370549321174621,
      "step": 170450
    },
    {
      "epoch": 48.38489923360772,
      "grad_norm": 11.3541898727417,
      "learning_rate": 5.163525404484815e-05,
      "loss": 0.004545418918132782,
      "step": 170460
    },
    {
      "epoch": 48.38773772353108,
      "grad_norm": 0.3891279995441437,
      "learning_rate": 5.1632415554924784e-05,
      "loss": 0.0007805190980434418,
      "step": 170470
    },
    {
      "epoch": 48.390576213454445,
      "grad_norm": 0.05284084379673004,
      "learning_rate": 5.1629577065001425e-05,
      "loss": 0.0009583435952663421,
      "step": 170480
    },
    {
      "epoch": 48.3934147033778,
      "grad_norm": 0.08011160045862198,
      "learning_rate": 5.162673857507806e-05,
      "loss": 0.00040725190192461016,
      "step": 170490
    },
    {
      "epoch": 48.39625319330116,
      "grad_norm": 0.02798384614288807,
      "learning_rate": 5.1623900085154694e-05,
      "loss": 0.0005644055083394051,
      "step": 170500
    },
    {
      "epoch": 48.39625319330116,
      "eval_accuracy": 0.9790805620906721,
      "eval_loss": 0.07664205133914948,
      "eval_runtime": 32.6579,
      "eval_samples_per_second": 481.567,
      "eval_steps_per_second": 7.533,
      "step": 170500
    },
    {
      "epoch": 48.399091683224526,
      "grad_norm": 0.08876006305217743,
      "learning_rate": 5.1621061595231336e-05,
      "loss": 0.000337301567196846,
      "step": 170510
    },
    {
      "epoch": 48.40193017314788,
      "grad_norm": 0.28436708450317383,
      "learning_rate": 5.1618223105307984e-05,
      "loss": 0.000609748438000679,
      "step": 170520
    },
    {
      "epoch": 48.404768663071245,
      "grad_norm": 0.18547315895557404,
      "learning_rate": 5.161538461538462e-05,
      "loss": 0.0004974557086825371,
      "step": 170530
    },
    {
      "epoch": 48.40760715299461,
      "grad_norm": 0.22587506473064423,
      "learning_rate": 5.161254612546126e-05,
      "loss": 0.0005632251501083374,
      "step": 170540
    },
    {
      "epoch": 48.41044564291797,
      "grad_norm": 0.07921502739191055,
      "learning_rate": 5.1609707635537894e-05,
      "loss": 0.00034118220210075376,
      "step": 170550
    },
    {
      "epoch": 48.413284132841326,
      "grad_norm": 0.2299228310585022,
      "learning_rate": 5.1606869145614536e-05,
      "loss": 0.003197053074836731,
      "step": 170560
    },
    {
      "epoch": 48.41612262276469,
      "grad_norm": 8.988582611083984,
      "learning_rate": 5.160403065569117e-05,
      "loss": 0.002694563381373882,
      "step": 170570
    },
    {
      "epoch": 48.41896111268805,
      "grad_norm": 15.185585975646973,
      "learning_rate": 5.160119216576781e-05,
      "loss": 0.006045353412628174,
      "step": 170580
    },
    {
      "epoch": 48.42179960261141,
      "grad_norm": 0.7228904962539673,
      "learning_rate": 5.159835367584446e-05,
      "loss": 0.004216354340314865,
      "step": 170590
    },
    {
      "epoch": 48.42463809253477,
      "grad_norm": 0.1329009234905243,
      "learning_rate": 5.1595515185921095e-05,
      "loss": 0.0031081357970833778,
      "step": 170600
    },
    {
      "epoch": 48.427476582458134,
      "grad_norm": 1.2691131830215454,
      "learning_rate": 5.1592676695997736e-05,
      "loss": 0.0004847286269068718,
      "step": 170610
    },
    {
      "epoch": 48.43031507238149,
      "grad_norm": 0.038418155163526535,
      "learning_rate": 5.158983820607437e-05,
      "loss": 0.0006129065528512001,
      "step": 170620
    },
    {
      "epoch": 48.43315356230485,
      "grad_norm": 6.66414737701416,
      "learning_rate": 5.1586999716151005e-05,
      "loss": 0.004517550393939018,
      "step": 170630
    },
    {
      "epoch": 48.435992052228215,
      "grad_norm": 0.13070984184741974,
      "learning_rate": 5.1584161226227646e-05,
      "loss": 0.006014755368232727,
      "step": 170640
    },
    {
      "epoch": 48.43883054215158,
      "grad_norm": 0.09078683704137802,
      "learning_rate": 5.1581322736304295e-05,
      "loss": 0.003291340544819832,
      "step": 170650
    },
    {
      "epoch": 48.441669032074934,
      "grad_norm": 1.7763590812683105,
      "learning_rate": 5.157848424638093e-05,
      "loss": 0.01226412206888199,
      "step": 170660
    },
    {
      "epoch": 48.4445075219983,
      "grad_norm": 0.47938188910484314,
      "learning_rate": 5.157564575645757e-05,
      "loss": 0.005168944969773293,
      "step": 170670
    },
    {
      "epoch": 48.44734601192166,
      "grad_norm": 0.19833579659461975,
      "learning_rate": 5.1572807266534205e-05,
      "loss": 0.001846395805478096,
      "step": 170680
    },
    {
      "epoch": 48.450184501845015,
      "grad_norm": 0.040801040828228,
      "learning_rate": 5.1569968776610847e-05,
      "loss": 0.002320140041410923,
      "step": 170690
    },
    {
      "epoch": 48.45302299176838,
      "grad_norm": 1.3466113805770874,
      "learning_rate": 5.156713028668748e-05,
      "loss": 0.0012026140466332436,
      "step": 170700
    },
    {
      "epoch": 48.45586148169174,
      "grad_norm": 0.04430330917239189,
      "learning_rate": 5.156429179676412e-05,
      "loss": 0.005154763162136078,
      "step": 170710
    },
    {
      "epoch": 48.458699971615104,
      "grad_norm": 0.022386042401194572,
      "learning_rate": 5.156145330684077e-05,
      "loss": 0.00400623008608818,
      "step": 170720
    },
    {
      "epoch": 48.46153846153846,
      "grad_norm": 0.5387179851531982,
      "learning_rate": 5.1558614816917405e-05,
      "loss": 0.0012103494256734848,
      "step": 170730
    },
    {
      "epoch": 48.46437695146182,
      "grad_norm": 0.08315584063529968,
      "learning_rate": 5.155577632699404e-05,
      "loss": 0.0011243775486946106,
      "step": 170740
    },
    {
      "epoch": 48.467215441385186,
      "grad_norm": 0.042945489287376404,
      "learning_rate": 5.155293783707068e-05,
      "loss": 0.0020023128017783165,
      "step": 170750
    },
    {
      "epoch": 48.47005393130854,
      "grad_norm": 2.7136924266815186,
      "learning_rate": 5.1550099347147316e-05,
      "loss": 0.0026473045349121095,
      "step": 170760
    },
    {
      "epoch": 48.472892421231904,
      "grad_norm": 7.623650074005127,
      "learning_rate": 5.154726085722396e-05,
      "loss": 0.0025663595646619798,
      "step": 170770
    },
    {
      "epoch": 48.47573091115527,
      "grad_norm": 0.38927021622657776,
      "learning_rate": 5.154442236730059e-05,
      "loss": 0.0039700761437416075,
      "step": 170780
    },
    {
      "epoch": 48.47856940107862,
      "grad_norm": 0.18257685005664825,
      "learning_rate": 5.154158387737724e-05,
      "loss": 0.0028970541432499887,
      "step": 170790
    },
    {
      "epoch": 48.481407891001986,
      "grad_norm": 0.0339609794318676,
      "learning_rate": 5.153874538745388e-05,
      "loss": 0.000954761914908886,
      "step": 170800
    },
    {
      "epoch": 48.48424638092535,
      "grad_norm": 0.017256328836083412,
      "learning_rate": 5.1535906897530516e-05,
      "loss": 0.0014236824586987495,
      "step": 170810
    },
    {
      "epoch": 48.48708487084871,
      "grad_norm": 2.4651925563812256,
      "learning_rate": 5.153306840760716e-05,
      "loss": 0.008707504719495773,
      "step": 170820
    },
    {
      "epoch": 48.48992336077207,
      "grad_norm": 0.0893767699599266,
      "learning_rate": 5.153022991768379e-05,
      "loss": 0.01208752691745758,
      "step": 170830
    },
    {
      "epoch": 48.49276185069543,
      "grad_norm": 0.3827891945838928,
      "learning_rate": 5.1527391427760427e-05,
      "loss": 0.009104026108980178,
      "step": 170840
    },
    {
      "epoch": 48.49560034061879,
      "grad_norm": 0.08280209451913834,
      "learning_rate": 5.152455293783708e-05,
      "loss": 0.0073622152209281925,
      "step": 170850
    },
    {
      "epoch": 48.49843883054215,
      "grad_norm": 0.225166454911232,
      "learning_rate": 5.1521714447913716e-05,
      "loss": 0.0005918486043810844,
      "step": 170860
    },
    {
      "epoch": 48.50127732046551,
      "grad_norm": 9.239640235900879,
      "learning_rate": 5.151887595799035e-05,
      "loss": 0.0026381196454167365,
      "step": 170870
    },
    {
      "epoch": 48.504115810388875,
      "grad_norm": 0.2799241244792938,
      "learning_rate": 5.151603746806699e-05,
      "loss": 0.001545260287821293,
      "step": 170880
    },
    {
      "epoch": 48.50695430031223,
      "grad_norm": 0.06842215359210968,
      "learning_rate": 5.151319897814363e-05,
      "loss": 0.012658520042896271,
      "step": 170890
    },
    {
      "epoch": 48.50979279023559,
      "grad_norm": 0.37595370411872864,
      "learning_rate": 5.151036048822027e-05,
      "loss": 0.001045878604054451,
      "step": 170900
    },
    {
      "epoch": 48.512631280158956,
      "grad_norm": 2.510371446609497,
      "learning_rate": 5.15075219982969e-05,
      "loss": 0.0049222812056541445,
      "step": 170910
    },
    {
      "epoch": 48.51546977008232,
      "grad_norm": 1.7015056610107422,
      "learning_rate": 5.150468350837355e-05,
      "loss": 0.002806585468351841,
      "step": 170920
    },
    {
      "epoch": 48.518308260005675,
      "grad_norm": 0.889167845249176,
      "learning_rate": 5.150184501845019e-05,
      "loss": 0.0012668151408433913,
      "step": 170930
    },
    {
      "epoch": 48.52114674992904,
      "grad_norm": 0.3165140151977539,
      "learning_rate": 5.149900652852683e-05,
      "loss": 0.0029624992981553077,
      "step": 170940
    },
    {
      "epoch": 48.5239852398524,
      "grad_norm": 11.839585304260254,
      "learning_rate": 5.149616803860347e-05,
      "loss": 0.00961458683013916,
      "step": 170950
    },
    {
      "epoch": 48.52682372977576,
      "grad_norm": 0.9960107803344727,
      "learning_rate": 5.14933295486801e-05,
      "loss": 0.004828505590558052,
      "step": 170960
    },
    {
      "epoch": 48.52966221969912,
      "grad_norm": 0.028587639331817627,
      "learning_rate": 5.149049105875674e-05,
      "loss": 0.0015985943377017976,
      "step": 170970
    },
    {
      "epoch": 48.53250070962248,
      "grad_norm": 0.1929071694612503,
      "learning_rate": 5.148765256883338e-05,
      "loss": 0.0021568873897194864,
      "step": 170980
    },
    {
      "epoch": 48.535339199545845,
      "grad_norm": 1.6953352689743042,
      "learning_rate": 5.148481407891003e-05,
      "loss": 0.0019011260941624642,
      "step": 170990
    },
    {
      "epoch": 48.5381776894692,
      "grad_norm": 0.13031187653541565,
      "learning_rate": 5.148197558898666e-05,
      "loss": 0.0009530404582619667,
      "step": 171000
    },
    {
      "epoch": 48.5381776894692,
      "eval_accuracy": 0.9766007503020283,
      "eval_loss": 0.08445987850427628,
      "eval_runtime": 32.589,
      "eval_samples_per_second": 482.586,
      "eval_steps_per_second": 7.549,
      "step": 171000
    },
    {
      "epoch": 48.541016179392564,
      "grad_norm": 0.17168587446212769,
      "learning_rate": 5.14791370990633e-05,
      "loss": 0.0004420937970280647,
      "step": 171010
    },
    {
      "epoch": 48.54385466931593,
      "grad_norm": 0.3086674213409424,
      "learning_rate": 5.147629860913994e-05,
      "loss": 0.0076403521001338955,
      "step": 171020
    },
    {
      "epoch": 48.54669315923928,
      "grad_norm": 3.1585206985473633,
      "learning_rate": 5.147346011921658e-05,
      "loss": 0.0017544565722346305,
      "step": 171030
    },
    {
      "epoch": 48.549531649162645,
      "grad_norm": 0.8581807613372803,
      "learning_rate": 5.147062162929321e-05,
      "loss": 0.0015395818278193473,
      "step": 171040
    },
    {
      "epoch": 48.55237013908601,
      "grad_norm": 0.3838559687137604,
      "learning_rate": 5.146778313936986e-05,
      "loss": 0.001527724601328373,
      "step": 171050
    },
    {
      "epoch": 48.555208629009364,
      "grad_norm": 1.5384438037872314,
      "learning_rate": 5.14649446494465e-05,
      "loss": 0.002065899595618248,
      "step": 171060
    },
    {
      "epoch": 48.55804711893273,
      "grad_norm": 0.1205187737941742,
      "learning_rate": 5.146210615952314e-05,
      "loss": 0.005208033323287964,
      "step": 171070
    },
    {
      "epoch": 48.56088560885609,
      "grad_norm": 0.2176227867603302,
      "learning_rate": 5.145926766959978e-05,
      "loss": 0.0022127697244286536,
      "step": 171080
    },
    {
      "epoch": 48.56372409877945,
      "grad_norm": 5.478058815002441,
      "learning_rate": 5.1456429179676413e-05,
      "loss": 0.0028652939945459366,
      "step": 171090
    },
    {
      "epoch": 48.56656258870281,
      "grad_norm": 0.1728636473417282,
      "learning_rate": 5.145359068975305e-05,
      "loss": 0.0022642796859145163,
      "step": 171100
    },
    {
      "epoch": 48.56940107862617,
      "grad_norm": 0.12026462703943253,
      "learning_rate": 5.145075219982969e-05,
      "loss": 0.001722855307161808,
      "step": 171110
    },
    {
      "epoch": 48.572239568549534,
      "grad_norm": 0.0746661052107811,
      "learning_rate": 5.144791370990634e-05,
      "loss": 0.00029281917959451676,
      "step": 171120
    },
    {
      "epoch": 48.57507805847289,
      "grad_norm": 3.2571768760681152,
      "learning_rate": 5.144507521998297e-05,
      "loss": 0.007282399386167526,
      "step": 171130
    },
    {
      "epoch": 48.57791654839625,
      "grad_norm": 0.14351454377174377,
      "learning_rate": 5.1442236730059614e-05,
      "loss": 0.0070213906466960905,
      "step": 171140
    },
    {
      "epoch": 48.580755038319616,
      "grad_norm": 14.091404914855957,
      "learning_rate": 5.143939824013625e-05,
      "loss": 0.0031679809093475343,
      "step": 171150
    },
    {
      "epoch": 48.58359352824297,
      "grad_norm": 6.517899036407471,
      "learning_rate": 5.143655975021289e-05,
      "loss": 0.004081464558839798,
      "step": 171160
    },
    {
      "epoch": 48.586432018166335,
      "grad_norm": 0.9132385849952698,
      "learning_rate": 5.1433721260289524e-05,
      "loss": 0.002209450490772724,
      "step": 171170
    },
    {
      "epoch": 48.5892705080897,
      "grad_norm": 0.5172933340072632,
      "learning_rate": 5.1430882770366166e-05,
      "loss": 0.005937594175338745,
      "step": 171180
    },
    {
      "epoch": 48.59210899801306,
      "grad_norm": 1.3507235050201416,
      "learning_rate": 5.1428044280442814e-05,
      "loss": 0.00198699701577425,
      "step": 171190
    },
    {
      "epoch": 48.594947487936416,
      "grad_norm": 0.1407298445701599,
      "learning_rate": 5.142520579051945e-05,
      "loss": 0.004060435295104981,
      "step": 171200
    },
    {
      "epoch": 48.59778597785978,
      "grad_norm": 10.12601375579834,
      "learning_rate": 5.142236730059608e-05,
      "loss": 0.004953330755233765,
      "step": 171210
    },
    {
      "epoch": 48.60062446778314,
      "grad_norm": 3.8021068572998047,
      "learning_rate": 5.1419528810672724e-05,
      "loss": 0.01492089033126831,
      "step": 171220
    },
    {
      "epoch": 48.6034629577065,
      "grad_norm": 0.21038328111171722,
      "learning_rate": 5.141669032074936e-05,
      "loss": 0.0019347172230482102,
      "step": 171230
    },
    {
      "epoch": 48.60630144762986,
      "grad_norm": 5.240849018096924,
      "learning_rate": 5.1413851830826e-05,
      "loss": 0.004002631455659866,
      "step": 171240
    },
    {
      "epoch": 48.60913993755322,
      "grad_norm": 0.09050235897302628,
      "learning_rate": 5.141101334090265e-05,
      "loss": 0.001175028830766678,
      "step": 171250
    },
    {
      "epoch": 48.61197842747658,
      "grad_norm": 0.30089071393013,
      "learning_rate": 5.140817485097928e-05,
      "loss": 0.00181279294192791,
      "step": 171260
    },
    {
      "epoch": 48.61481691739994,
      "grad_norm": 0.1463218331336975,
      "learning_rate": 5.1405336361055924e-05,
      "loss": 0.0015710178762674333,
      "step": 171270
    },
    {
      "epoch": 48.617655407323305,
      "grad_norm": 0.5719322562217712,
      "learning_rate": 5.140249787113256e-05,
      "loss": 0.0012796854600310326,
      "step": 171280
    },
    {
      "epoch": 48.62049389724667,
      "grad_norm": 0.16673468053340912,
      "learning_rate": 5.13996593812092e-05,
      "loss": 0.0007609622552990913,
      "step": 171290
    },
    {
      "epoch": 48.623332387170024,
      "grad_norm": 0.029619168490171432,
      "learning_rate": 5.1396820891285835e-05,
      "loss": 0.0006316835060715675,
      "step": 171300
    },
    {
      "epoch": 48.62617087709339,
      "grad_norm": 10.497674942016602,
      "learning_rate": 5.139398240136247e-05,
      "loss": 0.008860303461551667,
      "step": 171310
    },
    {
      "epoch": 48.62900936701675,
      "grad_norm": 0.23965345323085785,
      "learning_rate": 5.1391143911439124e-05,
      "loss": 0.00040855016559362414,
      "step": 171320
    },
    {
      "epoch": 48.631847856940105,
      "grad_norm": 0.04509411379694939,
      "learning_rate": 5.138830542151576e-05,
      "loss": 0.0014358246698975563,
      "step": 171330
    },
    {
      "epoch": 48.63468634686347,
      "grad_norm": 9.060681343078613,
      "learning_rate": 5.1385466931592394e-05,
      "loss": 0.00682150274515152,
      "step": 171340
    },
    {
      "epoch": 48.63752483678683,
      "grad_norm": 1.9924805164337158,
      "learning_rate": 5.1382628441669035e-05,
      "loss": 0.003715803474187851,
      "step": 171350
    },
    {
      "epoch": 48.64036332671019,
      "grad_norm": 1.9966981410980225,
      "learning_rate": 5.137978995174567e-05,
      "loss": 0.010807289183139801,
      "step": 171360
    },
    {
      "epoch": 48.64320181663355,
      "grad_norm": 0.692420482635498,
      "learning_rate": 5.137695146182231e-05,
      "loss": 0.0014898790046572684,
      "step": 171370
    },
    {
      "epoch": 48.64604030655691,
      "grad_norm": 0.02174954302608967,
      "learning_rate": 5.1374112971898946e-05,
      "loss": 0.002732396498322487,
      "step": 171380
    },
    {
      "epoch": 48.648878796480275,
      "grad_norm": 0.21062695980072021,
      "learning_rate": 5.1371274481975594e-05,
      "loss": 0.0020420346409082413,
      "step": 171390
    },
    {
      "epoch": 48.65171728640363,
      "grad_norm": 0.006979825906455517,
      "learning_rate": 5.1368435992052235e-05,
      "loss": 0.0032182361930608748,
      "step": 171400
    },
    {
      "epoch": 48.654555776326994,
      "grad_norm": 0.16062021255493164,
      "learning_rate": 5.136559750212887e-05,
      "loss": 0.0018060909584164619,
      "step": 171410
    },
    {
      "epoch": 48.65739426625036,
      "grad_norm": 0.06651894003152847,
      "learning_rate": 5.136275901220551e-05,
      "loss": 0.0008187636733055115,
      "step": 171420
    },
    {
      "epoch": 48.66023275617371,
      "grad_norm": 0.056935835629701614,
      "learning_rate": 5.1359920522282146e-05,
      "loss": 0.001069670170545578,
      "step": 171430
    },
    {
      "epoch": 48.663071246097076,
      "grad_norm": 0.05172022804617882,
      "learning_rate": 5.135708203235878e-05,
      "loss": 0.0004696100950241089,
      "step": 171440
    },
    {
      "epoch": 48.66590973602044,
      "grad_norm": 0.1213679164648056,
      "learning_rate": 5.135424354243543e-05,
      "loss": 0.001972397416830063,
      "step": 171450
    },
    {
      "epoch": 48.6687482259438,
      "grad_norm": 0.24569158256053925,
      "learning_rate": 5.135140505251207e-05,
      "loss": 0.0004940463230013847,
      "step": 171460
    },
    {
      "epoch": 48.67158671586716,
      "grad_norm": 0.036339204758405685,
      "learning_rate": 5.1348566562588704e-05,
      "loss": 0.0004220619797706604,
      "step": 171470
    },
    {
      "epoch": 48.67442520579052,
      "grad_norm": 0.05259789526462555,
      "learning_rate": 5.1345728072665346e-05,
      "loss": 0.0006614919751882553,
      "step": 171480
    },
    {
      "epoch": 48.67726369571388,
      "grad_norm": 0.07911424338817596,
      "learning_rate": 5.134288958274198e-05,
      "loss": 0.00023583080619573593,
      "step": 171490
    },
    {
      "epoch": 48.68010218563724,
      "grad_norm": 0.10449560731649399,
      "learning_rate": 5.134005109281862e-05,
      "loss": 0.0005123855546116829,
      "step": 171500
    },
    {
      "epoch": 48.68010218563724,
      "eval_accuracy": 0.977872448655179,
      "eval_loss": 0.0746883824467659,
      "eval_runtime": 32.4992,
      "eval_samples_per_second": 483.92,
      "eval_steps_per_second": 7.569,
      "step": 171500
    },
    {
      "epoch": 48.6829406755606,
      "grad_norm": 0.056315239518880844,
      "learning_rate": 5.1337212602895256e-05,
      "loss": 0.0010042695328593254,
      "step": 171510
    },
    {
      "epoch": 48.685779165483964,
      "grad_norm": 0.060811251401901245,
      "learning_rate": 5.1334374112971905e-05,
      "loss": 0.0005493801087141037,
      "step": 171520
    },
    {
      "epoch": 48.68861765540732,
      "grad_norm": 0.049916304647922516,
      "learning_rate": 5.1331535623048546e-05,
      "loss": 0.0009134382009506226,
      "step": 171530
    },
    {
      "epoch": 48.69145614533068,
      "grad_norm": 0.020978478714823723,
      "learning_rate": 5.132869713312518e-05,
      "loss": 0.00042226947844028475,
      "step": 171540
    },
    {
      "epoch": 48.694294635254046,
      "grad_norm": 1.654530644416809,
      "learning_rate": 5.132585864320182e-05,
      "loss": 0.0026064876466989515,
      "step": 171550
    },
    {
      "epoch": 48.69713312517741,
      "grad_norm": 0.2503092885017395,
      "learning_rate": 5.1323020153278456e-05,
      "loss": 0.0009323280304670334,
      "step": 171560
    },
    {
      "epoch": 48.699971615100765,
      "grad_norm": 0.09303180873394012,
      "learning_rate": 5.132018166335509e-05,
      "loss": 0.0003151062875986099,
      "step": 171570
    },
    {
      "epoch": 48.70281010502413,
      "grad_norm": 1.4312406778335571,
      "learning_rate": 5.131734317343174e-05,
      "loss": 0.0016324160620570182,
      "step": 171580
    },
    {
      "epoch": 48.70564859494749,
      "grad_norm": 0.3516513705253601,
      "learning_rate": 5.131450468350838e-05,
      "loss": 0.00466107577085495,
      "step": 171590
    },
    {
      "epoch": 48.708487084870846,
      "grad_norm": 0.32015863060951233,
      "learning_rate": 5.1311666193585015e-05,
      "loss": 0.00390034094452858,
      "step": 171600
    },
    {
      "epoch": 48.71132557479421,
      "grad_norm": 3.952726125717163,
      "learning_rate": 5.1308827703661657e-05,
      "loss": 0.004910657927393913,
      "step": 171610
    },
    {
      "epoch": 48.71416406471757,
      "grad_norm": 0.30980437994003296,
      "learning_rate": 5.130598921373829e-05,
      "loss": 0.001830965094268322,
      "step": 171620
    },
    {
      "epoch": 48.71700255464093,
      "grad_norm": 0.47968441247940063,
      "learning_rate": 5.130315072381493e-05,
      "loss": 0.0014920564368367194,
      "step": 171630
    },
    {
      "epoch": 48.71984104456429,
      "grad_norm": 0.056100063025951385,
      "learning_rate": 5.130031223389157e-05,
      "loss": 0.0008868316188454628,
      "step": 171640
    },
    {
      "epoch": 48.72267953448765,
      "grad_norm": 0.9908788800239563,
      "learning_rate": 5.1297473743968215e-05,
      "loss": 0.0017683958634734155,
      "step": 171650
    },
    {
      "epoch": 48.725518024411016,
      "grad_norm": 6.060591220855713,
      "learning_rate": 5.129463525404486e-05,
      "loss": 0.0015589671209454536,
      "step": 171660
    },
    {
      "epoch": 48.72835651433437,
      "grad_norm": 0.11978467553853989,
      "learning_rate": 5.129179676412149e-05,
      "loss": 0.0018501009792089463,
      "step": 171670
    },
    {
      "epoch": 48.731195004257735,
      "grad_norm": 0.09514296054840088,
      "learning_rate": 5.1288958274198126e-05,
      "loss": 0.0019240105524659158,
      "step": 171680
    },
    {
      "epoch": 48.7340334941811,
      "grad_norm": 0.02538694441318512,
      "learning_rate": 5.128611978427477e-05,
      "loss": 0.0005737496539950371,
      "step": 171690
    },
    {
      "epoch": 48.736871984104454,
      "grad_norm": 0.5992187857627869,
      "learning_rate": 5.12832812943514e-05,
      "loss": 0.0018726563081145286,
      "step": 171700
    },
    {
      "epoch": 48.73971047402782,
      "grad_norm": 0.06297511607408524,
      "learning_rate": 5.128044280442804e-05,
      "loss": 0.00034196358174085615,
      "step": 171710
    },
    {
      "epoch": 48.74254896395118,
      "grad_norm": 0.04354177042841911,
      "learning_rate": 5.127760431450469e-05,
      "loss": 0.0056247469037771225,
      "step": 171720
    },
    {
      "epoch": 48.745387453874535,
      "grad_norm": 0.15472543239593506,
      "learning_rate": 5.1274765824581326e-05,
      "loss": 0.0014159411191940307,
      "step": 171730
    },
    {
      "epoch": 48.7482259437979,
      "grad_norm": 0.6958574652671814,
      "learning_rate": 5.127192733465797e-05,
      "loss": 0.003483116254210472,
      "step": 171740
    },
    {
      "epoch": 48.75106443372126,
      "grad_norm": 0.13745221495628357,
      "learning_rate": 5.12690888447346e-05,
      "loss": 0.0018445182591676711,
      "step": 171750
    },
    {
      "epoch": 48.753902923644624,
      "grad_norm": 0.5727301836013794,
      "learning_rate": 5.126625035481124e-05,
      "loss": 0.0014442389830946923,
      "step": 171760
    },
    {
      "epoch": 48.75674141356798,
      "grad_norm": 0.03319557011127472,
      "learning_rate": 5.126341186488788e-05,
      "loss": 0.00045747347176074984,
      "step": 171770
    },
    {
      "epoch": 48.75957990349134,
      "grad_norm": 0.1531682163476944,
      "learning_rate": 5.1260573374964526e-05,
      "loss": 0.0010006040334701537,
      "step": 171780
    },
    {
      "epoch": 48.762418393414706,
      "grad_norm": 4.754243850708008,
      "learning_rate": 5.125773488504117e-05,
      "loss": 0.0009909659624099732,
      "step": 171790
    },
    {
      "epoch": 48.76525688333806,
      "grad_norm": 0.38498616218566895,
      "learning_rate": 5.12548963951178e-05,
      "loss": 0.0008324358612298965,
      "step": 171800
    },
    {
      "epoch": 48.768095373261424,
      "grad_norm": 0.1062794178724289,
      "learning_rate": 5.125205790519444e-05,
      "loss": 0.0004904672503471374,
      "step": 171810
    },
    {
      "epoch": 48.77093386318479,
      "grad_norm": 0.1093926876783371,
      "learning_rate": 5.124921941527108e-05,
      "loss": 0.0005360066890716553,
      "step": 171820
    },
    {
      "epoch": 48.77377235310815,
      "grad_norm": 0.018153851851820946,
      "learning_rate": 5.124638092534771e-05,
      "loss": 0.0010227154940366744,
      "step": 171830
    },
    {
      "epoch": 48.776610843031506,
      "grad_norm": 1.4325839281082153,
      "learning_rate": 5.1243542435424354e-05,
      "loss": 0.000879974290728569,
      "step": 171840
    },
    {
      "epoch": 48.77944933295487,
      "grad_norm": 0.0067153675481677055,
      "learning_rate": 5.1240703945501e-05,
      "loss": 0.0006906948983669281,
      "step": 171850
    },
    {
      "epoch": 48.78228782287823,
      "grad_norm": 0.11794984340667725,
      "learning_rate": 5.123786545557764e-05,
      "loss": 0.0006513083353638649,
      "step": 171860
    },
    {
      "epoch": 48.78512631280159,
      "grad_norm": 0.03766554221510887,
      "learning_rate": 5.123502696565428e-05,
      "loss": 0.0017198644578456878,
      "step": 171870
    },
    {
      "epoch": 48.78796480272495,
      "grad_norm": 0.36724409461021423,
      "learning_rate": 5.123218847573091e-05,
      "loss": 0.001821814477443695,
      "step": 171880
    },
    {
      "epoch": 48.79080329264831,
      "grad_norm": 1.3753690719604492,
      "learning_rate": 5.1229349985807554e-05,
      "loss": 0.0020224038511514665,
      "step": 171890
    },
    {
      "epoch": 48.79364178257167,
      "grad_norm": 9.933741569519043,
      "learning_rate": 5.122651149588419e-05,
      "loss": 0.011322507262229919,
      "step": 171900
    },
    {
      "epoch": 48.79648027249503,
      "grad_norm": 2.224764585494995,
      "learning_rate": 5.122367300596082e-05,
      "loss": 0.006666841357946396,
      "step": 171910
    },
    {
      "epoch": 48.799318762418395,
      "grad_norm": 0.19863854348659515,
      "learning_rate": 5.122083451603747e-05,
      "loss": 0.0015585789456963538,
      "step": 171920
    },
    {
      "epoch": 48.80215725234176,
      "grad_norm": 0.2558833062648773,
      "learning_rate": 5.121799602611411e-05,
      "loss": 0.004019550979137421,
      "step": 171930
    },
    {
      "epoch": 48.80499574226511,
      "grad_norm": 0.14411549270153046,
      "learning_rate": 5.121515753619075e-05,
      "loss": 0.0002683686092495918,
      "step": 171940
    },
    {
      "epoch": 48.807834232188476,
      "grad_norm": 0.17288127541542053,
      "learning_rate": 5.121231904626739e-05,
      "loss": 0.0005569819360971451,
      "step": 171950
    },
    {
      "epoch": 48.81067272211184,
      "grad_norm": 0.01469586230814457,
      "learning_rate": 5.120948055634402e-05,
      "loss": 0.0021493881940841675,
      "step": 171960
    },
    {
      "epoch": 48.813511212035195,
      "grad_norm": 0.04532536491751671,
      "learning_rate": 5.1206642066420665e-05,
      "loss": 0.0008125390857458114,
      "step": 171970
    },
    {
      "epoch": 48.81634970195856,
      "grad_norm": 0.05526276305317879,
      "learning_rate": 5.120380357649731e-05,
      "loss": 0.00020254477858543395,
      "step": 171980
    },
    {
      "epoch": 48.81918819188192,
      "grad_norm": 0.4725193679332733,
      "learning_rate": 5.120096508657395e-05,
      "loss": 0.0010133270174264909,
      "step": 171990
    },
    {
      "epoch": 48.822026681805276,
      "grad_norm": 5.523651123046875,
      "learning_rate": 5.119812659665059e-05,
      "loss": 0.005498120188713073,
      "step": 172000
    },
    {
      "epoch": 48.822026681805276,
      "eval_accuracy": 0.9779360335728365,
      "eval_loss": 0.07996834814548492,
      "eval_runtime": 32.5093,
      "eval_samples_per_second": 483.769,
      "eval_steps_per_second": 7.567,
      "step": 172000
    },
    {
      "epoch": 48.82486517172864,
      "grad_norm": 0.18724176287651062,
      "learning_rate": 5.1195288106727223e-05,
      "loss": 0.0008900420740246773,
      "step": 172010
    },
    {
      "epoch": 48.827703661652,
      "grad_norm": 2.2936644554138184,
      "learning_rate": 5.119244961680386e-05,
      "loss": 0.005339031666517257,
      "step": 172020
    },
    {
      "epoch": 48.830542151575365,
      "grad_norm": 0.3006945848464966,
      "learning_rate": 5.11896111268805e-05,
      "loss": 0.0007886774837970734,
      "step": 172030
    },
    {
      "epoch": 48.83338064149872,
      "grad_norm": 0.05989944934844971,
      "learning_rate": 5.1186772636957134e-05,
      "loss": 0.0011230131611227989,
      "step": 172040
    },
    {
      "epoch": 48.836219131422084,
      "grad_norm": 0.022534262388944626,
      "learning_rate": 5.118393414703378e-05,
      "loss": 0.0012328688055276871,
      "step": 172050
    },
    {
      "epoch": 48.83905762134545,
      "grad_norm": 10.427884101867676,
      "learning_rate": 5.1181095657110424e-05,
      "loss": 0.010115724802017213,
      "step": 172060
    },
    {
      "epoch": 48.8418961112688,
      "grad_norm": 1.2002466917037964,
      "learning_rate": 5.117825716718706e-05,
      "loss": 0.0031574338674545286,
      "step": 172070
    },
    {
      "epoch": 48.844734601192165,
      "grad_norm": 1.10024893283844,
      "learning_rate": 5.11754186772637e-05,
      "loss": 0.002984492480754852,
      "step": 172080
    },
    {
      "epoch": 48.84757309111553,
      "grad_norm": 0.20414020121097565,
      "learning_rate": 5.1172580187340334e-05,
      "loss": 0.0024426614865660666,
      "step": 172090
    },
    {
      "epoch": 48.850411581038884,
      "grad_norm": 0.3845486044883728,
      "learning_rate": 5.1169741697416975e-05,
      "loss": 0.005111327022314071,
      "step": 172100
    },
    {
      "epoch": 48.85325007096225,
      "grad_norm": 0.08331666886806488,
      "learning_rate": 5.116690320749361e-05,
      "loss": 0.0017271287739276885,
      "step": 172110
    },
    {
      "epoch": 48.85608856088561,
      "grad_norm": 6.246562957763672,
      "learning_rate": 5.116406471757026e-05,
      "loss": 0.003239167481660843,
      "step": 172120
    },
    {
      "epoch": 48.85892705080897,
      "grad_norm": 1.20320725440979,
      "learning_rate": 5.11612262276469e-05,
      "loss": 0.0011236019432544707,
      "step": 172130
    },
    {
      "epoch": 48.86176554073233,
      "grad_norm": 2.4668877124786377,
      "learning_rate": 5.1158387737723534e-05,
      "loss": 0.0016177073121070862,
      "step": 172140
    },
    {
      "epoch": 48.86460403065569,
      "grad_norm": 0.17081069946289062,
      "learning_rate": 5.115554924780017e-05,
      "loss": 0.0019013697281479835,
      "step": 172150
    },
    {
      "epoch": 48.867442520579054,
      "grad_norm": 0.0636497288942337,
      "learning_rate": 5.115271075787681e-05,
      "loss": 0.0009447552263736724,
      "step": 172160
    },
    {
      "epoch": 48.87028101050241,
      "grad_norm": 0.7080081701278687,
      "learning_rate": 5.1149872267953445e-05,
      "loss": 0.0006501093506813049,
      "step": 172170
    },
    {
      "epoch": 48.87311950042577,
      "grad_norm": 1.296241283416748,
      "learning_rate": 5.114731762702243e-05,
      "loss": 0.00796903446316719,
      "step": 172180
    },
    {
      "epoch": 48.875957990349136,
      "grad_norm": 0.4087015986442566,
      "learning_rate": 5.1144479137099066e-05,
      "loss": 0.002808142453432083,
      "step": 172190
    },
    {
      "epoch": 48.8787964802725,
      "grad_norm": 0.10935381799936295,
      "learning_rate": 5.11416406471757e-05,
      "loss": 0.0013132559135556221,
      "step": 172200
    },
    {
      "epoch": 48.881634970195854,
      "grad_norm": 0.10228131711483002,
      "learning_rate": 5.113880215725234e-05,
      "loss": 0.0010452998802065849,
      "step": 172210
    },
    {
      "epoch": 48.88447346011922,
      "grad_norm": 0.011633284389972687,
      "learning_rate": 5.113596366732898e-05,
      "loss": 0.0005109105259180069,
      "step": 172220
    },
    {
      "epoch": 48.88731195004258,
      "grad_norm": 7.366385459899902,
      "learning_rate": 5.1133125177405625e-05,
      "loss": 0.0042848832905292514,
      "step": 172230
    },
    {
      "epoch": 48.890150439965936,
      "grad_norm": 0.05976659432053566,
      "learning_rate": 5.1130286687482266e-05,
      "loss": 0.000361105240881443,
      "step": 172240
    },
    {
      "epoch": 48.8929889298893,
      "grad_norm": 0.1869400292634964,
      "learning_rate": 5.11274481975589e-05,
      "loss": 0.005433176457881927,
      "step": 172250
    },
    {
      "epoch": 48.89582741981266,
      "grad_norm": 0.3396529257297516,
      "learning_rate": 5.112460970763554e-05,
      "loss": 0.000697348639369011,
      "step": 172260
    },
    {
      "epoch": 48.89866590973602,
      "grad_norm": 0.04140743985772133,
      "learning_rate": 5.112177121771218e-05,
      "loss": 0.0008406154811382294,
      "step": 172270
    },
    {
      "epoch": 48.90150439965938,
      "grad_norm": 0.16311980783939362,
      "learning_rate": 5.111893272778882e-05,
      "loss": 0.0002374660223722458,
      "step": 172280
    },
    {
      "epoch": 48.90434288958274,
      "grad_norm": 0.06974630802869797,
      "learning_rate": 5.111609423786545e-05,
      "loss": 0.0002673132345080376,
      "step": 172290
    },
    {
      "epoch": 48.907181379506106,
      "grad_norm": 0.00860648974776268,
      "learning_rate": 5.11132557479421e-05,
      "loss": 0.0006512820720672608,
      "step": 172300
    },
    {
      "epoch": 48.91001986942946,
      "grad_norm": 0.3189452588558197,
      "learning_rate": 5.111041725801874e-05,
      "loss": 0.0008086696267127991,
      "step": 172310
    },
    {
      "epoch": 48.912858359352825,
      "grad_norm": 0.05550479143857956,
      "learning_rate": 5.110757876809538e-05,
      "loss": 0.0004661312326788902,
      "step": 172320
    },
    {
      "epoch": 48.91569684927619,
      "grad_norm": 0.6365115642547607,
      "learning_rate": 5.110474027817201e-05,
      "loss": 0.006756998598575592,
      "step": 172330
    },
    {
      "epoch": 48.91853533919954,
      "grad_norm": 0.013937837444245815,
      "learning_rate": 5.110190178824865e-05,
      "loss": 0.005842366814613342,
      "step": 172340
    },
    {
      "epoch": 48.921373829122906,
      "grad_norm": 0.023381661623716354,
      "learning_rate": 5.109906329832529e-05,
      "loss": 0.0057539395987987515,
      "step": 172350
    },
    {
      "epoch": 48.92421231904627,
      "grad_norm": 5.716965198516846,
      "learning_rate": 5.1096224808401936e-05,
      "loss": 0.004662581533193588,
      "step": 172360
    },
    {
      "epoch": 48.927050808969625,
      "grad_norm": 4.085431098937988,
      "learning_rate": 5.109367016747091e-05,
      "loss": 0.012604504823684692,
      "step": 172370
    },
    {
      "epoch": 48.92988929889299,
      "grad_norm": 0.062256645411252975,
      "learning_rate": 5.109083167754754e-05,
      "loss": 0.001525731012225151,
      "step": 172380
    },
    {
      "epoch": 48.93272778881635,
      "grad_norm": 0.30070123076438904,
      "learning_rate": 5.1087993187624185e-05,
      "loss": 0.009692084789276124,
      "step": 172390
    },
    {
      "epoch": 48.935566278739714,
      "grad_norm": 3.0615694522857666,
      "learning_rate": 5.108515469770082e-05,
      "loss": 0.0038044527173042297,
      "step": 172400
    },
    {
      "epoch": 48.93840476866307,
      "grad_norm": 8.432422637939453,
      "learning_rate": 5.108231620777747e-05,
      "loss": 0.007222647219896317,
      "step": 172410
    },
    {
      "epoch": 48.94124325858643,
      "grad_norm": 5.4584455490112305,
      "learning_rate": 5.107947771785411e-05,
      "loss": 0.00397547036409378,
      "step": 172420
    },
    {
      "epoch": 48.944081748509795,
      "grad_norm": 12.874959945678711,
      "learning_rate": 5.1076639227930743e-05,
      "loss": 0.0063656076788902284,
      "step": 172430
    },
    {
      "epoch": 48.94692023843315,
      "grad_norm": 0.10854694247245789,
      "learning_rate": 5.1073800738007385e-05,
      "loss": 0.001983114704489708,
      "step": 172440
    },
    {
      "epoch": 48.949758728356514,
      "grad_norm": 0.17944559454917908,
      "learning_rate": 5.107096224808402e-05,
      "loss": 0.000785338506102562,
      "step": 172450
    },
    {
      "epoch": 48.95259721827988,
      "grad_norm": 2.413919448852539,
      "learning_rate": 5.106812375816066e-05,
      "loss": 0.007317013293504715,
      "step": 172460
    },
    {
      "epoch": 48.95543570820323,
      "grad_norm": 2.8163747787475586,
      "learning_rate": 5.1065285268237295e-05,
      "loss": 0.004812180250883103,
      "step": 172470
    },
    {
      "epoch": 48.958274198126595,
      "grad_norm": 0.6288138031959534,
      "learning_rate": 5.1062446778313944e-05,
      "loss": 0.004755834490060807,
      "step": 172480
    },
    {
      "epoch": 48.96111268804996,
      "grad_norm": 13.701496124267578,
      "learning_rate": 5.105960828839058e-05,
      "loss": 0.009347520023584365,
      "step": 172490
    },
    {
      "epoch": 48.96395117797332,
      "grad_norm": 0.17137959599494934,
      "learning_rate": 5.105676979846722e-05,
      "loss": 0.001433514431118965,
      "step": 172500
    },
    {
      "epoch": 48.96395117797332,
      "eval_accuracy": 0.9775545240668914,
      "eval_loss": 0.08699586242437363,
      "eval_runtime": 32.6146,
      "eval_samples_per_second": 482.207,
      "eval_steps_per_second": 7.543,
      "step": 172500
    },
    {
      "epoch": 48.96678966789668,
      "grad_norm": 0.5708220601081848,
      "learning_rate": 5.1053931308543854e-05,
      "loss": 0.004009757936000824,
      "step": 172510
    },
    {
      "epoch": 48.96962815782004,
      "grad_norm": 0.34977808594703674,
      "learning_rate": 5.1051092818620496e-05,
      "loss": 0.0009630687534809112,
      "step": 172520
    },
    {
      "epoch": 48.9724666477434,
      "grad_norm": 13.491650581359863,
      "learning_rate": 5.104825432869713e-05,
      "loss": 0.004478693753480911,
      "step": 172530
    },
    {
      "epoch": 48.97530513766676,
      "grad_norm": 15.53306770324707,
      "learning_rate": 5.104541583877378e-05,
      "loss": 0.008974905312061309,
      "step": 172540
    },
    {
      "epoch": 48.97814362759012,
      "grad_norm": 0.057033028453588486,
      "learning_rate": 5.104257734885042e-05,
      "loss": 0.005124781653285027,
      "step": 172550
    },
    {
      "epoch": 48.980982117513484,
      "grad_norm": 0.7872393131256104,
      "learning_rate": 5.1039738858927054e-05,
      "loss": 0.002710718661546707,
      "step": 172560
    },
    {
      "epoch": 48.98382060743684,
      "grad_norm": 0.550419807434082,
      "learning_rate": 5.1036900369003696e-05,
      "loss": 0.0023342955857515333,
      "step": 172570
    },
    {
      "epoch": 48.9866590973602,
      "grad_norm": 1.0960137844085693,
      "learning_rate": 5.103406187908033e-05,
      "loss": 0.0026195770129561423,
      "step": 172580
    },
    {
      "epoch": 48.989497587283566,
      "grad_norm": 0.20212888717651367,
      "learning_rate": 5.1031223389156965e-05,
      "loss": 0.0021183524280786515,
      "step": 172590
    },
    {
      "epoch": 48.99233607720693,
      "grad_norm": 1.2823057174682617,
      "learning_rate": 5.1028384899233606e-05,
      "loss": 0.0014798562973737718,
      "step": 172600
    },
    {
      "epoch": 48.995174567130285,
      "grad_norm": 1.5165798664093018,
      "learning_rate": 5.1025546409310254e-05,
      "loss": 0.007388165593147278,
      "step": 172610
    },
    {
      "epoch": 48.99801305705365,
      "grad_norm": 0.2570347189903259,
      "learning_rate": 5.102270791938689e-05,
      "loss": 0.002708026394248009,
      "step": 172620
    },
    {
      "epoch": 49.00085154697701,
      "grad_norm": 0.011321907863020897,
      "learning_rate": 5.101986942946353e-05,
      "loss": 0.004464778676629066,
      "step": 172630
    },
    {
      "epoch": 49.003690036900366,
      "grad_norm": 0.13006596267223358,
      "learning_rate": 5.1017030939540165e-05,
      "loss": 0.0012980271130800247,
      "step": 172640
    },
    {
      "epoch": 49.00652852682373,
      "grad_norm": 0.31325390934944153,
      "learning_rate": 5.1014192449616806e-05,
      "loss": 0.00035820621997117996,
      "step": 172650
    },
    {
      "epoch": 49.00936701674709,
      "grad_norm": 0.05022968351840973,
      "learning_rate": 5.101135395969344e-05,
      "loss": 0.0007009321823716163,
      "step": 172660
    },
    {
      "epoch": 49.012205506670455,
      "grad_norm": 0.5981322526931763,
      "learning_rate": 5.100851546977008e-05,
      "loss": 0.0017971133813261986,
      "step": 172670
    },
    {
      "epoch": 49.01504399659381,
      "grad_norm": 0.010666106827557087,
      "learning_rate": 5.100567697984673e-05,
      "loss": 0.00042894445359706876,
      "step": 172680
    },
    {
      "epoch": 49.01788248651717,
      "grad_norm": 0.2878568470478058,
      "learning_rate": 5.1002838489923365e-05,
      "loss": 0.012633153796195984,
      "step": 172690
    },
    {
      "epoch": 49.020720976440536,
      "grad_norm": 0.02159702032804489,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 0.00570906400680542,
      "step": 172700
    },
    {
      "epoch": 49.02355946636389,
      "grad_norm": 0.2105906456708908,
      "learning_rate": 5.099716151007664e-05,
      "loss": 0.0004220360890030861,
      "step": 172710
    },
    {
      "epoch": 49.026397956287255,
      "grad_norm": 0.13678349554538727,
      "learning_rate": 5.0994323020153276e-05,
      "loss": 0.00037243887782096864,
      "step": 172720
    },
    {
      "epoch": 49.02923644621062,
      "grad_norm": 0.2747247517108917,
      "learning_rate": 5.099148453022992e-05,
      "loss": 0.004150266572833061,
      "step": 172730
    },
    {
      "epoch": 49.032074936133974,
      "grad_norm": 0.10745176672935486,
      "learning_rate": 5.0988646040306565e-05,
      "loss": 0.00033308565616607666,
      "step": 172740
    },
    {
      "epoch": 49.03491342605734,
      "grad_norm": 0.008018656633794308,
      "learning_rate": 5.09858075503832e-05,
      "loss": 0.0005752561613917351,
      "step": 172750
    },
    {
      "epoch": 49.0377519159807,
      "grad_norm": 0.013525616377592087,
      "learning_rate": 5.098296906045984e-05,
      "loss": 0.0004135463386774063,
      "step": 172760
    },
    {
      "epoch": 49.04059040590406,
      "grad_norm": 0.16271071135997772,
      "learning_rate": 5.0980130570536476e-05,
      "loss": 0.0003116823732852936,
      "step": 172770
    },
    {
      "epoch": 49.04342889582742,
      "grad_norm": 0.0540117546916008,
      "learning_rate": 5.097729208061312e-05,
      "loss": 0.0006945980712771416,
      "step": 172780
    },
    {
      "epoch": 49.04626738575078,
      "grad_norm": 0.1240612342953682,
      "learning_rate": 5.097445359068975e-05,
      "loss": 0.0005830543115735054,
      "step": 172790
    },
    {
      "epoch": 49.049105875674144,
      "grad_norm": 1.4307292699813843,
      "learning_rate": 5.097161510076639e-05,
      "loss": 0.0013180430978536606,
      "step": 172800
    },
    {
      "epoch": 49.0519443655975,
      "grad_norm": 0.1365261673927307,
      "learning_rate": 5.096877661084304e-05,
      "loss": 0.0004808057099580765,
      "step": 172810
    },
    {
      "epoch": 49.05478285552086,
      "grad_norm": 0.0333840548992157,
      "learning_rate": 5.0965938120919676e-05,
      "loss": 0.00033517032861709597,
      "step": 172820
    },
    {
      "epoch": 49.057621345444225,
      "grad_norm": 0.06113322824239731,
      "learning_rate": 5.096309963099632e-05,
      "loss": 0.00031305812299251554,
      "step": 172830
    },
    {
      "epoch": 49.06045983536758,
      "grad_norm": 0.09961499273777008,
      "learning_rate": 5.096026114107295e-05,
      "loss": 0.00017212294042110444,
      "step": 172840
    },
    {
      "epoch": 49.063298325290944,
      "grad_norm": 5.2695722579956055,
      "learning_rate": 5.0957422651149586e-05,
      "loss": 0.0011304600164294242,
      "step": 172850
    },
    {
      "epoch": 49.06613681521431,
      "grad_norm": 0.2888909578323364,
      "learning_rate": 5.095458416122623e-05,
      "loss": 0.0003456804901361465,
      "step": 172860
    },
    {
      "epoch": 49.06897530513767,
      "grad_norm": 0.05498090758919716,
      "learning_rate": 5.095174567130286e-05,
      "loss": 0.0005265653133392334,
      "step": 172870
    },
    {
      "epoch": 49.071813795061026,
      "grad_norm": 0.0338624007999897,
      "learning_rate": 5.094890718137951e-05,
      "loss": 0.0007517552003264427,
      "step": 172880
    },
    {
      "epoch": 49.07465228498439,
      "grad_norm": 1.2004364728927612,
      "learning_rate": 5.094606869145615e-05,
      "loss": 0.0006098782643675805,
      "step": 172890
    },
    {
      "epoch": 49.07749077490775,
      "grad_norm": 0.7085698843002319,
      "learning_rate": 5.0943230201532786e-05,
      "loss": 0.0010296367108821869,
      "step": 172900
    },
    {
      "epoch": 49.08032926483111,
      "grad_norm": 0.20880892872810364,
      "learning_rate": 5.094039171160943e-05,
      "loss": 0.0004785487428307533,
      "step": 172910
    },
    {
      "epoch": 49.08316775475447,
      "grad_norm": 0.6015651226043701,
      "learning_rate": 5.093755322168606e-05,
      "loss": 0.0006080139428377151,
      "step": 172920
    },
    {
      "epoch": 49.08600624467783,
      "grad_norm": 0.10781806707382202,
      "learning_rate": 5.0934714731762704e-05,
      "loss": 0.0022604286670684813,
      "step": 172930
    },
    {
      "epoch": 49.08884473460119,
      "grad_norm": 15.786813735961914,
      "learning_rate": 5.093187624183935e-05,
      "loss": 0.005497681349515915,
      "step": 172940
    },
    {
      "epoch": 49.09168322452455,
      "grad_norm": 0.014808000065386295,
      "learning_rate": 5.0929037751915987e-05,
      "loss": 0.0007314698770642281,
      "step": 172950
    },
    {
      "epoch": 49.094521714447914,
      "grad_norm": 15.514394760131836,
      "learning_rate": 5.092619926199262e-05,
      "loss": 0.013802260160446167,
      "step": 172960
    },
    {
      "epoch": 49.09736020437128,
      "grad_norm": 0.09080936014652252,
      "learning_rate": 5.092336077206926e-05,
      "loss": 0.005931773781776428,
      "step": 172970
    },
    {
      "epoch": 49.10019869429463,
      "grad_norm": 0.35394608974456787,
      "learning_rate": 5.09205222821459e-05,
      "loss": 0.0037084750831127165,
      "step": 172980
    },
    {
      "epoch": 49.103037184217996,
      "grad_norm": 0.18272143602371216,
      "learning_rate": 5.091768379222254e-05,
      "loss": 0.005183426290750503,
      "step": 172990
    },
    {
      "epoch": 49.10587567414136,
      "grad_norm": 0.030214393511414528,
      "learning_rate": 5.091484530229917e-05,
      "loss": 0.0004182647913694382,
      "step": 173000
    },
    {
      "epoch": 49.10587567414136,
      "eval_accuracy": 0.977045844725631,
      "eval_loss": 0.08575211465358734,
      "eval_runtime": 32.4098,
      "eval_samples_per_second": 485.255,
      "eval_steps_per_second": 7.59,
      "step": 173000
    },
    {
      "epoch": 49.108714164064715,
      "grad_norm": 0.12613116204738617,
      "learning_rate": 5.091200681237582e-05,
      "loss": 0.00155723188072443,
      "step": 173010
    },
    {
      "epoch": 49.11155265398808,
      "grad_norm": 0.09215208888053894,
      "learning_rate": 5.090916832245246e-05,
      "loss": 0.0025695061311125757,
      "step": 173020
    },
    {
      "epoch": 49.11439114391144,
      "grad_norm": 0.04675302281975746,
      "learning_rate": 5.09063298325291e-05,
      "loss": 0.0010211061686277389,
      "step": 173030
    },
    {
      "epoch": 49.1172296338348,
      "grad_norm": 0.15003721415996552,
      "learning_rate": 5.090349134260574e-05,
      "loss": 0.00041568893939256666,
      "step": 173040
    },
    {
      "epoch": 49.12006812375816,
      "grad_norm": 0.15591321885585785,
      "learning_rate": 5.090065285268237e-05,
      "loss": 0.001656535640358925,
      "step": 173050
    },
    {
      "epoch": 49.12290661368152,
      "grad_norm": 0.18946810066699982,
      "learning_rate": 5.089781436275901e-05,
      "loss": 0.0022673474624752997,
      "step": 173060
    },
    {
      "epoch": 49.125745103604885,
      "grad_norm": 0.43806007504463196,
      "learning_rate": 5.089497587283565e-05,
      "loss": 0.0005615003407001496,
      "step": 173070
    },
    {
      "epoch": 49.12858359352824,
      "grad_norm": 0.03690282627940178,
      "learning_rate": 5.08921373829123e-05,
      "loss": 0.0005021881312131882,
      "step": 173080
    },
    {
      "epoch": 49.131422083451604,
      "grad_norm": 0.389499306678772,
      "learning_rate": 5.088929889298893e-05,
      "loss": 0.001154518872499466,
      "step": 173090
    },
    {
      "epoch": 49.134260573374966,
      "grad_norm": 0.31621673703193665,
      "learning_rate": 5.088646040306557e-05,
      "loss": 0.0006406363099813461,
      "step": 173100
    },
    {
      "epoch": 49.13709906329832,
      "grad_norm": 0.7544806599617004,
      "learning_rate": 5.088362191314221e-05,
      "loss": 0.00033389125019311905,
      "step": 173110
    },
    {
      "epoch": 49.139937553221685,
      "grad_norm": 0.05639801174402237,
      "learning_rate": 5.088078342321885e-05,
      "loss": 0.000821429118514061,
      "step": 173120
    },
    {
      "epoch": 49.14277604314505,
      "grad_norm": 4.913681507110596,
      "learning_rate": 5.0877944933295484e-05,
      "loss": 0.0013337507843971252,
      "step": 173130
    },
    {
      "epoch": 49.14561453306841,
      "grad_norm": 0.8026243448257446,
      "learning_rate": 5.087510644337213e-05,
      "loss": 0.001129472069442272,
      "step": 173140
    },
    {
      "epoch": 49.14845302299177,
      "grad_norm": 9.207498550415039,
      "learning_rate": 5.087226795344877e-05,
      "loss": 0.004255690053105355,
      "step": 173150
    },
    {
      "epoch": 49.15129151291513,
      "grad_norm": 0.10501831769943237,
      "learning_rate": 5.086942946352541e-05,
      "loss": 0.0014639021828770638,
      "step": 173160
    },
    {
      "epoch": 49.15413000283849,
      "grad_norm": 0.1757003664970398,
      "learning_rate": 5.086659097360205e-05,
      "loss": 0.0011401575058698655,
      "step": 173170
    },
    {
      "epoch": 49.15696849276185,
      "grad_norm": 4.435089111328125,
      "learning_rate": 5.0863752483678684e-05,
      "loss": 0.002386649325489998,
      "step": 173180
    },
    {
      "epoch": 49.15980698268521,
      "grad_norm": 0.05456741899251938,
      "learning_rate": 5.086091399375532e-05,
      "loss": 0.00017779506742954254,
      "step": 173190
    },
    {
      "epoch": 49.162645472608574,
      "grad_norm": 0.2861558198928833,
      "learning_rate": 5.085807550383196e-05,
      "loss": 0.0006563998758792877,
      "step": 173200
    },
    {
      "epoch": 49.16548396253193,
      "grad_norm": 0.005178770050406456,
      "learning_rate": 5.085523701390861e-05,
      "loss": 0.00020892135798931122,
      "step": 173210
    },
    {
      "epoch": 49.16832245245529,
      "grad_norm": 1.080743432044983,
      "learning_rate": 5.085239852398524e-05,
      "loss": 0.000543312169611454,
      "step": 173220
    },
    {
      "epoch": 49.171160942378656,
      "grad_norm": 0.23394814133644104,
      "learning_rate": 5.0849560034061884e-05,
      "loss": 0.000432051345705986,
      "step": 173230
    },
    {
      "epoch": 49.17399943230202,
      "grad_norm": 0.1966971457004547,
      "learning_rate": 5.084672154413852e-05,
      "loss": 0.0004761883988976479,
      "step": 173240
    },
    {
      "epoch": 49.176837922225374,
      "grad_norm": 0.0465129129588604,
      "learning_rate": 5.084388305421516e-05,
      "loss": 0.0002175329253077507,
      "step": 173250
    },
    {
      "epoch": 49.17967641214874,
      "grad_norm": 0.0520174466073513,
      "learning_rate": 5.0841044564291795e-05,
      "loss": 0.0003683961927890778,
      "step": 173260
    },
    {
      "epoch": 49.1825149020721,
      "grad_norm": 0.02840975485742092,
      "learning_rate": 5.0838206074368436e-05,
      "loss": 0.0002465972676873207,
      "step": 173270
    },
    {
      "epoch": 49.185353391995456,
      "grad_norm": 0.1266687959432602,
      "learning_rate": 5.0835367584445084e-05,
      "loss": 0.0020509181544184686,
      "step": 173280
    },
    {
      "epoch": 49.18819188191882,
      "grad_norm": 0.015617385506629944,
      "learning_rate": 5.083252909452172e-05,
      "loss": 0.00039960816502571106,
      "step": 173290
    },
    {
      "epoch": 49.19103037184218,
      "grad_norm": 0.32313403487205505,
      "learning_rate": 5.082969060459836e-05,
      "loss": 0.0010525776073336602,
      "step": 173300
    },
    {
      "epoch": 49.19386886176554,
      "grad_norm": 0.01434552762657404,
      "learning_rate": 5.0826852114674995e-05,
      "loss": 0.0011583685874938965,
      "step": 173310
    },
    {
      "epoch": 49.1967073516889,
      "grad_norm": 0.011542484164237976,
      "learning_rate": 5.082401362475163e-05,
      "loss": 0.008152759075164795,
      "step": 173320
    },
    {
      "epoch": 49.19954584161226,
      "grad_norm": 0.02312752977013588,
      "learning_rate": 5.082117513482827e-05,
      "loss": 0.0035766873508691786,
      "step": 173330
    },
    {
      "epoch": 49.202384331535626,
      "grad_norm": 0.7306753396987915,
      "learning_rate": 5.081833664490492e-05,
      "loss": 0.0008628932759165764,
      "step": 173340
    },
    {
      "epoch": 49.20522282145898,
      "grad_norm": 0.08036074787378311,
      "learning_rate": 5.0815498154981553e-05,
      "loss": 0.001309681124985218,
      "step": 173350
    },
    {
      "epoch": 49.208061311382345,
      "grad_norm": 0.19072997570037842,
      "learning_rate": 5.0812659665058195e-05,
      "loss": 0.006792020797729492,
      "step": 173360
    },
    {
      "epoch": 49.21089980130571,
      "grad_norm": 2.7617101669311523,
      "learning_rate": 5.080982117513483e-05,
      "loss": 0.006394731998443604,
      "step": 173370
    },
    {
      "epoch": 49.21373829122906,
      "grad_norm": 2.834892511367798,
      "learning_rate": 5.080698268521147e-05,
      "loss": 0.002821844257414341,
      "step": 173380
    },
    {
      "epoch": 49.216576781152426,
      "grad_norm": 0.024129508063197136,
      "learning_rate": 5.0804144195288105e-05,
      "loss": 0.0021660128608345986,
      "step": 173390
    },
    {
      "epoch": 49.21941527107579,
      "grad_norm": 0.09458238631486893,
      "learning_rate": 5.080130570536475e-05,
      "loss": 0.0017084361985325812,
      "step": 173400
    },
    {
      "epoch": 49.22225376099915,
      "grad_norm": 0.024649200960993767,
      "learning_rate": 5.0798467215441395e-05,
      "loss": 0.005533041059970855,
      "step": 173410
    },
    {
      "epoch": 49.22509225092251,
      "grad_norm": 0.004047004040330648,
      "learning_rate": 5.079562872551803e-05,
      "loss": 0.003920510411262512,
      "step": 173420
    },
    {
      "epoch": 49.22793074084587,
      "grad_norm": 0.267958402633667,
      "learning_rate": 5.0792790235594664e-05,
      "loss": 0.01486324667930603,
      "step": 173430
    },
    {
      "epoch": 49.23076923076923,
      "grad_norm": 0.36376041173934937,
      "learning_rate": 5.0789951745671305e-05,
      "loss": 0.0010496888309717179,
      "step": 173440
    },
    {
      "epoch": 49.23360772069259,
      "grad_norm": 0.5658524632453918,
      "learning_rate": 5.078711325574794e-05,
      "loss": 0.0006689749658107758,
      "step": 173450
    },
    {
      "epoch": 49.23644621061595,
      "grad_norm": 0.1821237951517105,
      "learning_rate": 5.078427476582458e-05,
      "loss": 0.003654921054840088,
      "step": 173460
    },
    {
      "epoch": 49.239284700539315,
      "grad_norm": 0.05096187815070152,
      "learning_rate": 5.0781436275901216e-05,
      "loss": 0.0006558354943990707,
      "step": 173470
    },
    {
      "epoch": 49.24212319046267,
      "grad_norm": 5.636626720428467,
      "learning_rate": 5.0778597785977864e-05,
      "loss": 0.0021102624014019966,
      "step": 173480
    },
    {
      "epoch": 49.244961680386034,
      "grad_norm": 0.12602224946022034,
      "learning_rate": 5.0775759296054506e-05,
      "loss": 0.0022053269669413565,
      "step": 173490
    },
    {
      "epoch": 49.2478001703094,
      "grad_norm": 6.852323532104492,
      "learning_rate": 5.077292080613114e-05,
      "loss": 0.002283021621406078,
      "step": 173500
    },
    {
      "epoch": 49.2478001703094,
      "eval_accuracy": 0.9737394290074395,
      "eval_loss": 0.09817315638065338,
      "eval_runtime": 32.7124,
      "eval_samples_per_second": 480.765,
      "eval_steps_per_second": 7.52,
      "step": 173500
    },
    {
      "epoch": 49.25063866023276,
      "grad_norm": 6.805729389190674,
      "learning_rate": 5.077008231620778e-05,
      "loss": 0.0036963626742362978,
      "step": 173510
    },
    {
      "epoch": 49.253477150156115,
      "grad_norm": 2.2525441646575928,
      "learning_rate": 5.0767243826284416e-05,
      "loss": 0.0015326056629419327,
      "step": 173520
    },
    {
      "epoch": 49.25631564007948,
      "grad_norm": 0.33189085125923157,
      "learning_rate": 5.076440533636105e-05,
      "loss": 0.001049601472914219,
      "step": 173530
    },
    {
      "epoch": 49.25915413000284,
      "grad_norm": 0.1237260103225708,
      "learning_rate": 5.0761566846437706e-05,
      "loss": 0.0008194318041205406,
      "step": 173540
    },
    {
      "epoch": 49.2619926199262,
      "grad_norm": 0.08516093343496323,
      "learning_rate": 5.075872835651434e-05,
      "loss": 0.0005874171853065491,
      "step": 173550
    },
    {
      "epoch": 49.26483110984956,
      "grad_norm": 0.05770125240087509,
      "learning_rate": 5.0755889866590975e-05,
      "loss": 0.0006643613800406456,
      "step": 173560
    },
    {
      "epoch": 49.26766959977292,
      "grad_norm": 0.027042362838983536,
      "learning_rate": 5.0753051376667616e-05,
      "loss": 0.0003014327958226204,
      "step": 173570
    },
    {
      "epoch": 49.27050808969628,
      "grad_norm": 0.05333235114812851,
      "learning_rate": 5.075021288674425e-05,
      "loss": 0.000378747284412384,
      "step": 173580
    },
    {
      "epoch": 49.27334657961964,
      "grad_norm": 0.06277239322662354,
      "learning_rate": 5.074737439682089e-05,
      "loss": 0.000751454010605812,
      "step": 173590
    },
    {
      "epoch": 49.276185069543004,
      "grad_norm": 0.013695844449102879,
      "learning_rate": 5.074453590689753e-05,
      "loss": 0.0005365692079067231,
      "step": 173600
    },
    {
      "epoch": 49.27902355946637,
      "grad_norm": 0.8405146598815918,
      "learning_rate": 5.0741697416974175e-05,
      "loss": 0.00044798608869314196,
      "step": 173610
    },
    {
      "epoch": 49.28186204938972,
      "grad_norm": 0.028053848072886467,
      "learning_rate": 5.0738858927050816e-05,
      "loss": 0.0004679374396800995,
      "step": 173620
    },
    {
      "epoch": 49.284700539313086,
      "grad_norm": 0.06331754475831985,
      "learning_rate": 5.073602043712745e-05,
      "loss": 0.005630004033446312,
      "step": 173630
    },
    {
      "epoch": 49.28753902923645,
      "grad_norm": 1.0597145557403564,
      "learning_rate": 5.073318194720409e-05,
      "loss": 0.006686075031757355,
      "step": 173640
    },
    {
      "epoch": 49.290377519159804,
      "grad_norm": 0.01150739286094904,
      "learning_rate": 5.073034345728073e-05,
      "loss": 0.0003741389140486717,
      "step": 173650
    },
    {
      "epoch": 49.29321600908317,
      "grad_norm": 0.007601823192089796,
      "learning_rate": 5.072750496735736e-05,
      "loss": 0.001813308335840702,
      "step": 173660
    },
    {
      "epoch": 49.29605449900653,
      "grad_norm": 0.12178615480661392,
      "learning_rate": 5.072466647743401e-05,
      "loss": 0.0036561667919158937,
      "step": 173670
    },
    {
      "epoch": 49.298892988929886,
      "grad_norm": 0.09903335571289062,
      "learning_rate": 5.072182798751065e-05,
      "loss": 0.006555713713169098,
      "step": 173680
    },
    {
      "epoch": 49.30173147885325,
      "grad_norm": 0.07839127629995346,
      "learning_rate": 5.0718989497587286e-05,
      "loss": 0.0010848145931959153,
      "step": 173690
    },
    {
      "epoch": 49.30456996877661,
      "grad_norm": 0.07298515737056732,
      "learning_rate": 5.071615100766393e-05,
      "loss": 0.0077747412025928496,
      "step": 173700
    },
    {
      "epoch": 49.307408458699975,
      "grad_norm": 0.03916887566447258,
      "learning_rate": 5.071331251774056e-05,
      "loss": 0.0070011667907238,
      "step": 173710
    },
    {
      "epoch": 49.31024694862333,
      "grad_norm": 0.07214349508285522,
      "learning_rate": 5.07104740278172e-05,
      "loss": 0.0012357614934444428,
      "step": 173720
    },
    {
      "epoch": 49.31308543854669,
      "grad_norm": 0.7011746168136597,
      "learning_rate": 5.070763553789384e-05,
      "loss": 0.003491102159023285,
      "step": 173730
    },
    {
      "epoch": 49.315923928470056,
      "grad_norm": 0.09081356227397919,
      "learning_rate": 5.0704797047970486e-05,
      "loss": 0.007144607603549957,
      "step": 173740
    },
    {
      "epoch": 49.31876241839341,
      "grad_norm": 0.021544350311160088,
      "learning_rate": 5.070195855804713e-05,
      "loss": 0.005827154964208603,
      "step": 173750
    },
    {
      "epoch": 49.321600908316775,
      "grad_norm": 2.2530667781829834,
      "learning_rate": 5.069912006812376e-05,
      "loss": 0.007123856246471405,
      "step": 173760
    },
    {
      "epoch": 49.32443939824014,
      "grad_norm": 0.8668180704116821,
      "learning_rate": 5.06962815782004e-05,
      "loss": 0.005804193764925003,
      "step": 173770
    },
    {
      "epoch": 49.32727788816349,
      "grad_norm": 17.37893295288086,
      "learning_rate": 5.069344308827704e-05,
      "loss": 0.013988442718982697,
      "step": 173780
    },
    {
      "epoch": 49.330116378086856,
      "grad_norm": 0.2568034827709198,
      "learning_rate": 5.069060459835367e-05,
      "loss": 0.002850036509335041,
      "step": 173790
    },
    {
      "epoch": 49.33295486801022,
      "grad_norm": 0.3624034523963928,
      "learning_rate": 5.0687766108430314e-05,
      "loss": 0.0020668381825089456,
      "step": 173800
    },
    {
      "epoch": 49.33579335793358,
      "grad_norm": 0.11971236765384674,
      "learning_rate": 5.068492761850696e-05,
      "loss": 0.0018747434020042419,
      "step": 173810
    },
    {
      "epoch": 49.33863184785694,
      "grad_norm": 0.023391762748360634,
      "learning_rate": 5.0682089128583596e-05,
      "loss": 0.0005408026278018952,
      "step": 173820
    },
    {
      "epoch": 49.3414703377803,
      "grad_norm": 0.9461982250213623,
      "learning_rate": 5.067925063866024e-05,
      "loss": 0.001001550815999508,
      "step": 173830
    },
    {
      "epoch": 49.344308827703664,
      "grad_norm": 0.059548716992139816,
      "learning_rate": 5.067641214873687e-05,
      "loss": 0.004326153546571732,
      "step": 173840
    },
    {
      "epoch": 49.34714731762702,
      "grad_norm": 0.13651373982429504,
      "learning_rate": 5.0673573658813514e-05,
      "loss": 0.005768497288227081,
      "step": 173850
    },
    {
      "epoch": 49.34998580755038,
      "grad_norm": 0.08145644515752792,
      "learning_rate": 5.067073516889015e-05,
      "loss": 0.001885322853922844,
      "step": 173860
    },
    {
      "epoch": 49.352824297473745,
      "grad_norm": 0.14749662578105927,
      "learning_rate": 5.0667896678966796e-05,
      "loss": 0.0013126792386174203,
      "step": 173870
    },
    {
      "epoch": 49.35566278739711,
      "grad_norm": 0.47245869040489197,
      "learning_rate": 5.066505818904344e-05,
      "loss": 0.022169597446918488,
      "step": 173880
    },
    {
      "epoch": 49.358501277320464,
      "grad_norm": 0.13764461874961853,
      "learning_rate": 5.066221969912007e-05,
      "loss": 0.0028109893202781676,
      "step": 173890
    },
    {
      "epoch": 49.36133976724383,
      "grad_norm": 0.08403323590755463,
      "learning_rate": 5.065938120919671e-05,
      "loss": 0.007163897156715393,
      "step": 173900
    },
    {
      "epoch": 49.36417825716719,
      "grad_norm": 0.2353680282831192,
      "learning_rate": 5.065654271927335e-05,
      "loss": 0.002231322415173054,
      "step": 173910
    },
    {
      "epoch": 49.367016747090545,
      "grad_norm": 0.08180417865514755,
      "learning_rate": 5.065370422934998e-05,
      "loss": 0.008885100483894348,
      "step": 173920
    },
    {
      "epoch": 49.36985523701391,
      "grad_norm": 0.4971647560596466,
      "learning_rate": 5.0650865739426624e-05,
      "loss": 0.0025781694799661635,
      "step": 173930
    },
    {
      "epoch": 49.37269372693727,
      "grad_norm": 3.8541276454925537,
      "learning_rate": 5.064802724950327e-05,
      "loss": 0.000815461203455925,
      "step": 173940
    },
    {
      "epoch": 49.37553221686063,
      "grad_norm": 0.16474762558937073,
      "learning_rate": 5.064518875957991e-05,
      "loss": 0.0011359967291355133,
      "step": 173950
    },
    {
      "epoch": 49.37837070678399,
      "grad_norm": 0.15303142368793488,
      "learning_rate": 5.064235026965655e-05,
      "loss": 0.0018501266837120055,
      "step": 173960
    },
    {
      "epoch": 49.38120919670735,
      "grad_norm": 0.06253893673419952,
      "learning_rate": 5.063951177973318e-05,
      "loss": 0.006607683748006821,
      "step": 173970
    },
    {
      "epoch": 49.384047686630716,
      "grad_norm": 4.195091247558594,
      "learning_rate": 5.0636673289809825e-05,
      "loss": 0.004405527934432029,
      "step": 173980
    },
    {
      "epoch": 49.38688617655407,
      "grad_norm": 0.2868594527244568,
      "learning_rate": 5.063383479988646e-05,
      "loss": 0.0035739287734031676,
      "step": 173990
    },
    {
      "epoch": 49.389724666477434,
      "grad_norm": 0.09643464535474777,
      "learning_rate": 5.0630996309963094e-05,
      "loss": 0.009132219851016999,
      "step": 174000
    },
    {
      "epoch": 49.389724666477434,
      "eval_accuracy": 0.9771094296432886,
      "eval_loss": 0.08226440101861954,
      "eval_runtime": 32.2818,
      "eval_samples_per_second": 487.178,
      "eval_steps_per_second": 7.62,
      "step": 174000
    },
    {
      "epoch": 49.3925631564008,
      "grad_norm": 1.6048071384429932,
      "learning_rate": 5.062815782003975e-05,
      "loss": 0.003418990224599838,
      "step": 174010
    },
    {
      "epoch": 49.39540164632415,
      "grad_norm": 0.023365017026662827,
      "learning_rate": 5.062531933011638e-05,
      "loss": 0.00246965941041708,
      "step": 174020
    },
    {
      "epoch": 49.398240136247516,
      "grad_norm": 0.2861991226673126,
      "learning_rate": 5.062248084019302e-05,
      "loss": 0.0014650985598564149,
      "step": 174030
    },
    {
      "epoch": 49.40107862617088,
      "grad_norm": 0.48803043365478516,
      "learning_rate": 5.061964235026966e-05,
      "loss": 0.002677516266703606,
      "step": 174040
    },
    {
      "epoch": 49.403917116094235,
      "grad_norm": 0.050629086792469025,
      "learning_rate": 5.0616803860346294e-05,
      "loss": 0.0010527187958359718,
      "step": 174050
    },
    {
      "epoch": 49.4067556060176,
      "grad_norm": 0.12454574555158615,
      "learning_rate": 5.0613965370422935e-05,
      "loss": 0.011010089516639709,
      "step": 174060
    },
    {
      "epoch": 49.40959409594096,
      "grad_norm": 0.01813160814344883,
      "learning_rate": 5.061112688049958e-05,
      "loss": 0.0060628712177276615,
      "step": 174070
    },
    {
      "epoch": 49.41243258586432,
      "grad_norm": 0.13506701588630676,
      "learning_rate": 5.060828839057622e-05,
      "loss": 0.011777782440185547,
      "step": 174080
    },
    {
      "epoch": 49.41527107578768,
      "grad_norm": 0.047128163278102875,
      "learning_rate": 5.060544990065286e-05,
      "loss": 0.025590035319328307,
      "step": 174090
    },
    {
      "epoch": 49.41810956571104,
      "grad_norm": 0.43338021636009216,
      "learning_rate": 5.0602611410729494e-05,
      "loss": 0.008002544939517974,
      "step": 174100
    },
    {
      "epoch": 49.420948055634405,
      "grad_norm": 0.05247509106993675,
      "learning_rate": 5.0599772920806135e-05,
      "loss": 0.00029188450425863264,
      "step": 174110
    },
    {
      "epoch": 49.42378654555776,
      "grad_norm": 0.10351300984621048,
      "learning_rate": 5.059693443088277e-05,
      "loss": 0.0015583477914333343,
      "step": 174120
    },
    {
      "epoch": 49.42662503548112,
      "grad_norm": 0.04541488736867905,
      "learning_rate": 5.0594095940959404e-05,
      "loss": 0.0020386604592204094,
      "step": 174130
    },
    {
      "epoch": 49.429463525404486,
      "grad_norm": 0.04769646003842354,
      "learning_rate": 5.059125745103605e-05,
      "loss": 0.004193255305290222,
      "step": 174140
    },
    {
      "epoch": 49.43230201532784,
      "grad_norm": 0.0363120473921299,
      "learning_rate": 5.0588418961112694e-05,
      "loss": 0.000612378679215908,
      "step": 174150
    },
    {
      "epoch": 49.435140505251205,
      "grad_norm": 1.1993077993392944,
      "learning_rate": 5.058558047118933e-05,
      "loss": 0.0023366983979940413,
      "step": 174160
    },
    {
      "epoch": 49.43797899517457,
      "grad_norm": 0.21437665820121765,
      "learning_rate": 5.058274198126597e-05,
      "loss": 0.0014089716598391533,
      "step": 174170
    },
    {
      "epoch": 49.44081748509793,
      "grad_norm": 0.34311437606811523,
      "learning_rate": 5.0579903491342605e-05,
      "loss": 0.0007108062505722046,
      "step": 174180
    },
    {
      "epoch": 49.44365597502129,
      "grad_norm": 0.3064103424549103,
      "learning_rate": 5.0577065001419246e-05,
      "loss": 0.0007103627547621727,
      "step": 174190
    },
    {
      "epoch": 49.44649446494465,
      "grad_norm": 0.060519468039274216,
      "learning_rate": 5.057422651149588e-05,
      "loss": 0.0029462510719895363,
      "step": 174200
    },
    {
      "epoch": 49.44933295486801,
      "grad_norm": 0.08160970360040665,
      "learning_rate": 5.057138802157253e-05,
      "loss": 0.002896803803741932,
      "step": 174210
    },
    {
      "epoch": 49.45217144479137,
      "grad_norm": 0.2061879187822342,
      "learning_rate": 5.056854953164917e-05,
      "loss": 0.000956176221370697,
      "step": 174220
    },
    {
      "epoch": 49.45500993471473,
      "grad_norm": 0.027024801820516586,
      "learning_rate": 5.0565711041725805e-05,
      "loss": 0.0005043752491474152,
      "step": 174230
    },
    {
      "epoch": 49.457848424638094,
      "grad_norm": 0.19731086492538452,
      "learning_rate": 5.056287255180244e-05,
      "loss": 0.000964079424738884,
      "step": 174240
    },
    {
      "epoch": 49.46068691456146,
      "grad_norm": 0.4043397009372711,
      "learning_rate": 5.056003406187908e-05,
      "loss": 0.000355483777821064,
      "step": 174250
    },
    {
      "epoch": 49.46352540448481,
      "grad_norm": 0.02666647918522358,
      "learning_rate": 5.0557195571955715e-05,
      "loss": 0.004449402913451195,
      "step": 174260
    },
    {
      "epoch": 49.466363894408175,
      "grad_norm": 0.12688389420509338,
      "learning_rate": 5.0554357082032363e-05,
      "loss": 0.008807403594255447,
      "step": 174270
    },
    {
      "epoch": 49.46920238433154,
      "grad_norm": 0.15351034700870514,
      "learning_rate": 5.0551518592109005e-05,
      "loss": 0.00037506837397813795,
      "step": 174280
    },
    {
      "epoch": 49.472040874254894,
      "grad_norm": 0.22444675862789154,
      "learning_rate": 5.054868010218564e-05,
      "loss": 0.0022159697487950324,
      "step": 174290
    },
    {
      "epoch": 49.47487936417826,
      "grad_norm": 0.42663490772247314,
      "learning_rate": 5.054584161226228e-05,
      "loss": 0.0010086042806506158,
      "step": 174300
    },
    {
      "epoch": 49.47771785410162,
      "grad_norm": 0.05571358650922775,
      "learning_rate": 5.0543003122338915e-05,
      "loss": 0.001645335741341114,
      "step": 174310
    },
    {
      "epoch": 49.480556344024976,
      "grad_norm": 0.09672936052083969,
      "learning_rate": 5.054016463241556e-05,
      "loss": 0.0010848598554730414,
      "step": 174320
    },
    {
      "epoch": 49.48339483394834,
      "grad_norm": 0.03758428245782852,
      "learning_rate": 5.053732614249219e-05,
      "loss": 0.0004982728511095047,
      "step": 174330
    },
    {
      "epoch": 49.4862333238717,
      "grad_norm": 1.3514416217803955,
      "learning_rate": 5.053448765256884e-05,
      "loss": 0.0009937770664691925,
      "step": 174340
    },
    {
      "epoch": 49.489071813795064,
      "grad_norm": 0.10619716346263885,
      "learning_rate": 5.053164916264548e-05,
      "loss": 0.004194065183401108,
      "step": 174350
    },
    {
      "epoch": 49.49191030371842,
      "grad_norm": 1.3667157888412476,
      "learning_rate": 5.0528810672722115e-05,
      "loss": 0.00941515490412712,
      "step": 174360
    },
    {
      "epoch": 49.49474879364178,
      "grad_norm": 0.9064038395881653,
      "learning_rate": 5.052597218279875e-05,
      "loss": 0.01582586467266083,
      "step": 174370
    },
    {
      "epoch": 49.497587283565146,
      "grad_norm": 0.3536556661128998,
      "learning_rate": 5.052313369287539e-05,
      "loss": 0.0056289248168468475,
      "step": 174380
    },
    {
      "epoch": 49.5004257734885,
      "grad_norm": 0.0658712238073349,
      "learning_rate": 5.0520295202952026e-05,
      "loss": 0.0012128373607993125,
      "step": 174390
    },
    {
      "epoch": 49.503264263411864,
      "grad_norm": 0.14651751518249512,
      "learning_rate": 5.051745671302867e-05,
      "loss": 0.0003593388944864273,
      "step": 174400
    },
    {
      "epoch": 49.50610275333523,
      "grad_norm": 0.04782366380095482,
      "learning_rate": 5.0514618223105316e-05,
      "loss": 0.0006188154220581055,
      "step": 174410
    },
    {
      "epoch": 49.50894124325858,
      "grad_norm": 1.1683348417282104,
      "learning_rate": 5.051177973318195e-05,
      "loss": 0.0007347853854298592,
      "step": 174420
    },
    {
      "epoch": 49.511779733181946,
      "grad_norm": 0.8962810635566711,
      "learning_rate": 5.050894124325859e-05,
      "loss": 0.0015039181336760521,
      "step": 174430
    },
    {
      "epoch": 49.51461822310531,
      "grad_norm": 0.1680213212966919,
      "learning_rate": 5.0506102753335226e-05,
      "loss": 0.0026662373915314675,
      "step": 174440
    },
    {
      "epoch": 49.51745671302867,
      "grad_norm": 12.828865051269531,
      "learning_rate": 5.050326426341187e-05,
      "loss": 0.0023170053958892823,
      "step": 174450
    },
    {
      "epoch": 49.52029520295203,
      "grad_norm": 0.5616059899330139,
      "learning_rate": 5.05004257734885e-05,
      "loss": 0.004547327756881714,
      "step": 174460
    },
    {
      "epoch": 49.52313369287539,
      "grad_norm": 11.061531066894531,
      "learning_rate": 5.049758728356515e-05,
      "loss": 0.002197488769888878,
      "step": 174470
    },
    {
      "epoch": 49.52597218279875,
      "grad_norm": 0.262590616941452,
      "learning_rate": 5.049474879364179e-05,
      "loss": 0.0052189499139785765,
      "step": 174480
    },
    {
      "epoch": 49.52881067272211,
      "grad_norm": 0.6213337182998657,
      "learning_rate": 5.0491910303718426e-05,
      "loss": 0.00125887431204319,
      "step": 174490
    },
    {
      "epoch": 49.53164916264547,
      "grad_norm": 0.016047660261392593,
      "learning_rate": 5.048907181379506e-05,
      "loss": 0.0003376699984073639,
      "step": 174500
    },
    {
      "epoch": 49.53164916264547,
      "eval_accuracy": 0.9762828257137407,
      "eval_loss": 0.08225994557142258,
      "eval_runtime": 32.8805,
      "eval_samples_per_second": 478.308,
      "eval_steps_per_second": 7.482,
      "step": 174500
    },
    {
      "epoch": 49.534487652568835,
      "grad_norm": 0.09457311779260635,
      "learning_rate": 5.04862333238717e-05,
      "loss": 0.0007161464542150498,
      "step": 174510
    },
    {
      "epoch": 49.53732614249219,
      "grad_norm": 0.06291300058364868,
      "learning_rate": 5.048339483394834e-05,
      "loss": 0.0009557103738188743,
      "step": 174520
    },
    {
      "epoch": 49.540164632415554,
      "grad_norm": 0.04507777839899063,
      "learning_rate": 5.048055634402498e-05,
      "loss": 0.0007106008008122445,
      "step": 174530
    },
    {
      "epoch": 49.54300312233892,
      "grad_norm": 0.02703217975795269,
      "learning_rate": 5.0477717854101626e-05,
      "loss": 0.00061690304428339,
      "step": 174540
    },
    {
      "epoch": 49.54584161226228,
      "grad_norm": 0.020704973489046097,
      "learning_rate": 5.047487936417826e-05,
      "loss": 0.0011949406936764718,
      "step": 174550
    },
    {
      "epoch": 49.548680102185635,
      "grad_norm": 0.02430592104792595,
      "learning_rate": 5.04720408742549e-05,
      "loss": 0.00041502602398395537,
      "step": 174560
    },
    {
      "epoch": 49.551518592109,
      "grad_norm": 1.1303411722183228,
      "learning_rate": 5.046920238433154e-05,
      "loss": 0.0004589153453707695,
      "step": 174570
    },
    {
      "epoch": 49.55435708203236,
      "grad_norm": 0.08333845436573029,
      "learning_rate": 5.046636389440818e-05,
      "loss": 0.002490334399044514,
      "step": 174580
    },
    {
      "epoch": 49.55719557195572,
      "grad_norm": 2.0511646270751953,
      "learning_rate": 5.046352540448481e-05,
      "loss": 0.0012020649388432502,
      "step": 174590
    },
    {
      "epoch": 49.56003406187908,
      "grad_norm": 0.011292465031147003,
      "learning_rate": 5.046068691456145e-05,
      "loss": 0.0007081996649503708,
      "step": 174600
    },
    {
      "epoch": 49.56287255180244,
      "grad_norm": 0.0370759591460228,
      "learning_rate": 5.0457848424638096e-05,
      "loss": 0.0012294072657823563,
      "step": 174610
    },
    {
      "epoch": 49.565711041725805,
      "grad_norm": 0.1219487264752388,
      "learning_rate": 5.045500993471474e-05,
      "loss": 0.0030302319675683973,
      "step": 174620
    },
    {
      "epoch": 49.56854953164916,
      "grad_norm": 9.172586441040039,
      "learning_rate": 5.045217144479137e-05,
      "loss": 0.002801471762359142,
      "step": 174630
    },
    {
      "epoch": 49.571388021572524,
      "grad_norm": 0.10189906507730484,
      "learning_rate": 5.044933295486801e-05,
      "loss": 0.0002078576013445854,
      "step": 174640
    },
    {
      "epoch": 49.57422651149589,
      "grad_norm": 0.027512622997164726,
      "learning_rate": 5.044649446494465e-05,
      "loss": 0.0013089664280414581,
      "step": 174650
    },
    {
      "epoch": 49.57706500141924,
      "grad_norm": 7.966211318969727,
      "learning_rate": 5.044365597502129e-05,
      "loss": 0.0016883127391338348,
      "step": 174660
    },
    {
      "epoch": 49.579903491342606,
      "grad_norm": 2.1473007202148438,
      "learning_rate": 5.044081748509794e-05,
      "loss": 0.0038018207997083664,
      "step": 174670
    },
    {
      "epoch": 49.58274198126597,
      "grad_norm": 0.08792694658041,
      "learning_rate": 5.043797899517457e-05,
      "loss": 0.0035412661731243134,
      "step": 174680
    },
    {
      "epoch": 49.585580471189324,
      "grad_norm": 2.688619613647461,
      "learning_rate": 5.043514050525121e-05,
      "loss": 0.0013900652527809143,
      "step": 174690
    },
    {
      "epoch": 49.58841896111269,
      "grad_norm": 0.04414433240890503,
      "learning_rate": 5.043230201532785e-05,
      "loss": 0.00034538041800260546,
      "step": 174700
    },
    {
      "epoch": 49.59125745103605,
      "grad_norm": 0.5002400875091553,
      "learning_rate": 5.042946352540448e-05,
      "loss": 0.0006683496758341789,
      "step": 174710
    },
    {
      "epoch": 49.59409594095941,
      "grad_norm": 0.6112593412399292,
      "learning_rate": 5.0426625035481124e-05,
      "loss": 0.0008251957595348358,
      "step": 174720
    },
    {
      "epoch": 49.59693443088277,
      "grad_norm": 0.19376496970653534,
      "learning_rate": 5.042378654555776e-05,
      "loss": 0.00026604775339365003,
      "step": 174730
    },
    {
      "epoch": 49.59977292080613,
      "grad_norm": 0.020368777215480804,
      "learning_rate": 5.0420948055634406e-05,
      "loss": 0.0018425839021801948,
      "step": 174740
    },
    {
      "epoch": 49.602611410729494,
      "grad_norm": 0.010004108771681786,
      "learning_rate": 5.041810956571105e-05,
      "loss": 0.0015065003186464309,
      "step": 174750
    },
    {
      "epoch": 49.60544990065285,
      "grad_norm": 0.03885076567530632,
      "learning_rate": 5.041527107578768e-05,
      "loss": 0.0004271429032087326,
      "step": 174760
    },
    {
      "epoch": 49.60828839057621,
      "grad_norm": 1.101487159729004,
      "learning_rate": 5.0412432585864324e-05,
      "loss": 0.0006877217441797256,
      "step": 174770
    },
    {
      "epoch": 49.611126880499576,
      "grad_norm": 0.7713820934295654,
      "learning_rate": 5.040959409594096e-05,
      "loss": 0.0012678943574428558,
      "step": 174780
    },
    {
      "epoch": 49.61396537042293,
      "grad_norm": 0.027248458936810493,
      "learning_rate": 5.04067556060176e-05,
      "loss": 0.0003350406885147095,
      "step": 174790
    },
    {
      "epoch": 49.616803860346295,
      "grad_norm": 0.03299420326948166,
      "learning_rate": 5.040391711609425e-05,
      "loss": 0.0005190249532461167,
      "step": 174800
    },
    {
      "epoch": 49.61964235026966,
      "grad_norm": 0.22925002872943878,
      "learning_rate": 5.040107862617088e-05,
      "loss": 0.0018724218010902406,
      "step": 174810
    },
    {
      "epoch": 49.62248084019302,
      "grad_norm": 0.030862247571349144,
      "learning_rate": 5.0398240136247524e-05,
      "loss": 0.0006745636463165283,
      "step": 174820
    },
    {
      "epoch": 49.625319330116376,
      "grad_norm": 0.29360970854759216,
      "learning_rate": 5.039540164632416e-05,
      "loss": 0.0005116831511259079,
      "step": 174830
    },
    {
      "epoch": 49.62815782003974,
      "grad_norm": 6.483983993530273,
      "learning_rate": 5.039256315640079e-05,
      "loss": 0.0023964446038007734,
      "step": 174840
    },
    {
      "epoch": 49.6309963099631,
      "grad_norm": 1.4613324403762817,
      "learning_rate": 5.0389724666477434e-05,
      "loss": 0.001032392680644989,
      "step": 174850
    },
    {
      "epoch": 49.63383479988646,
      "grad_norm": 0.03379906341433525,
      "learning_rate": 5.038688617655407e-05,
      "loss": 0.0005904011428356171,
      "step": 174860
    },
    {
      "epoch": 49.63667328980982,
      "grad_norm": 0.05205250531435013,
      "learning_rate": 5.038404768663072e-05,
      "loss": 0.002123118005692959,
      "step": 174870
    },
    {
      "epoch": 49.63951177973318,
      "grad_norm": 0.017428213730454445,
      "learning_rate": 5.038120919670736e-05,
      "loss": 0.0006925828754901886,
      "step": 174880
    },
    {
      "epoch": 49.64235026965654,
      "grad_norm": 0.04248816519975662,
      "learning_rate": 5.037837070678399e-05,
      "loss": 0.004625849425792694,
      "step": 174890
    },
    {
      "epoch": 49.6451887595799,
      "grad_norm": 0.08191556483507156,
      "learning_rate": 5.0375532216860634e-05,
      "loss": 0.005106126889586449,
      "step": 174900
    },
    {
      "epoch": 49.648027249503265,
      "grad_norm": 0.04353724420070648,
      "learning_rate": 5.037269372693727e-05,
      "loss": 0.0011774029582738877,
      "step": 174910
    },
    {
      "epoch": 49.65086573942663,
      "grad_norm": 0.13645096123218536,
      "learning_rate": 5.036985523701391e-05,
      "loss": 0.0010683424770832062,
      "step": 174920
    },
    {
      "epoch": 49.653704229349984,
      "grad_norm": 0.010712743736803532,
      "learning_rate": 5.0367016747090545e-05,
      "loss": 0.0018770337104797364,
      "step": 174930
    },
    {
      "epoch": 49.65654271927335,
      "grad_norm": 2.0343637466430664,
      "learning_rate": 5.036417825716719e-05,
      "loss": 0.0024123938754200937,
      "step": 174940
    },
    {
      "epoch": 49.65938120919671,
      "grad_norm": 4.162909507751465,
      "learning_rate": 5.0361339767243835e-05,
      "loss": 0.005811486393213272,
      "step": 174950
    },
    {
      "epoch": 49.662219699120065,
      "grad_norm": 11.690189361572266,
      "learning_rate": 5.035850127732047e-05,
      "loss": 0.012012768536806107,
      "step": 174960
    },
    {
      "epoch": 49.66505818904343,
      "grad_norm": 6.432911396026611,
      "learning_rate": 5.0355662787397104e-05,
      "loss": 0.009525948762893676,
      "step": 174970
    },
    {
      "epoch": 49.66789667896679,
      "grad_norm": 0.5019909143447876,
      "learning_rate": 5.0352824297473745e-05,
      "loss": 0.00040815435349941256,
      "step": 174980
    },
    {
      "epoch": 49.67073516889015,
      "grad_norm": 0.08197805285453796,
      "learning_rate": 5.034998580755038e-05,
      "loss": 0.0010222945362329482,
      "step": 174990
    },
    {
      "epoch": 49.67357365881351,
      "grad_norm": 1.085543155670166,
      "learning_rate": 5.034714731762703e-05,
      "loss": 0.00593641847372055,
      "step": 175000
    },
    {
      "epoch": 49.67357365881351,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.08713710308074951,
      "eval_runtime": 32.8422,
      "eval_samples_per_second": 478.865,
      "eval_steps_per_second": 7.49,
      "step": 175000
    },
    {
      "epoch": 49.67641214873687,
      "grad_norm": 7.617450714111328,
      "learning_rate": 5.034430882770367e-05,
      "loss": 0.0028795527294278147,
      "step": 175010
    },
    {
      "epoch": 49.679250638660235,
      "grad_norm": 8.637292861938477,
      "learning_rate": 5.0341470337780304e-05,
      "loss": 0.005108456313610077,
      "step": 175020
    },
    {
      "epoch": 49.68208912858359,
      "grad_norm": 0.023896286264061928,
      "learning_rate": 5.0338631847856945e-05,
      "loss": 0.0012277422472834587,
      "step": 175030
    },
    {
      "epoch": 49.684927618506954,
      "grad_norm": 0.08153923600912094,
      "learning_rate": 5.033579335793358e-05,
      "loss": 0.008194569498300552,
      "step": 175040
    },
    {
      "epoch": 49.68776610843032,
      "grad_norm": 0.7620654702186584,
      "learning_rate": 5.033295486801022e-05,
      "loss": 0.002302536927163601,
      "step": 175050
    },
    {
      "epoch": 49.69060459835367,
      "grad_norm": 5.171926498413086,
      "learning_rate": 5.0330116378086856e-05,
      "loss": 0.001172569766640663,
      "step": 175060
    },
    {
      "epoch": 49.693443088277036,
      "grad_norm": 5.460896968841553,
      "learning_rate": 5.0327277888163504e-05,
      "loss": 0.0018888220191001892,
      "step": 175070
    },
    {
      "epoch": 49.6962815782004,
      "grad_norm": 1.1599875688552856,
      "learning_rate": 5.032443939824014e-05,
      "loss": 0.002580607309937477,
      "step": 175080
    },
    {
      "epoch": 49.69912006812376,
      "grad_norm": 0.4760423004627228,
      "learning_rate": 5.032160090831678e-05,
      "loss": 0.00266636461019516,
      "step": 175090
    },
    {
      "epoch": 49.70195855804712,
      "grad_norm": 0.3911263644695282,
      "learning_rate": 5.0318762418393415e-05,
      "loss": 0.0008474301546812057,
      "step": 175100
    },
    {
      "epoch": 49.70479704797048,
      "grad_norm": 1.6237149238586426,
      "learning_rate": 5.0315923928470056e-05,
      "loss": 0.0015234140679240226,
      "step": 175110
    },
    {
      "epoch": 49.70763553789384,
      "grad_norm": 0.2679058611392975,
      "learning_rate": 5.031308543854669e-05,
      "loss": 0.0026421252638101577,
      "step": 175120
    },
    {
      "epoch": 49.7104740278172,
      "grad_norm": 0.05848897248506546,
      "learning_rate": 5.031024694862333e-05,
      "loss": 0.006892288476228714,
      "step": 175130
    },
    {
      "epoch": 49.71331251774056,
      "grad_norm": 0.15998360514640808,
      "learning_rate": 5.030740845869998e-05,
      "loss": 0.0013499166816473007,
      "step": 175140
    },
    {
      "epoch": 49.716151007663925,
      "grad_norm": 0.25701963901519775,
      "learning_rate": 5.0304569968776615e-05,
      "loss": 0.004136099666357041,
      "step": 175150
    },
    {
      "epoch": 49.71898949758728,
      "grad_norm": 0.3531850278377533,
      "learning_rate": 5.0301731478853256e-05,
      "loss": 0.001028701476752758,
      "step": 175160
    },
    {
      "epoch": 49.72182798751064,
      "grad_norm": 0.177789568901062,
      "learning_rate": 5.029889298892989e-05,
      "loss": 0.0008823325857520104,
      "step": 175170
    },
    {
      "epoch": 49.724666477434006,
      "grad_norm": 0.044498004019260406,
      "learning_rate": 5.0296054499006525e-05,
      "loss": 0.0008461998775601387,
      "step": 175180
    },
    {
      "epoch": 49.72750496735737,
      "grad_norm": 0.09205108880996704,
      "learning_rate": 5.0293216009083167e-05,
      "loss": 0.0017942890524864196,
      "step": 175190
    },
    {
      "epoch": 49.730343457280725,
      "grad_norm": 1.0957984924316406,
      "learning_rate": 5.0290377519159815e-05,
      "loss": 0.0009455930441617966,
      "step": 175200
    },
    {
      "epoch": 49.73318194720409,
      "grad_norm": 0.14448638260364532,
      "learning_rate": 5.028753902923645e-05,
      "loss": 0.0026460645720362665,
      "step": 175210
    },
    {
      "epoch": 49.73602043712745,
      "grad_norm": 0.17612239718437195,
      "learning_rate": 5.028470053931309e-05,
      "loss": 0.0011737575754523278,
      "step": 175220
    },
    {
      "epoch": 49.738858927050806,
      "grad_norm": 0.5082954168319702,
      "learning_rate": 5.0281862049389725e-05,
      "loss": 0.0008183961734175682,
      "step": 175230
    },
    {
      "epoch": 49.74169741697417,
      "grad_norm": 0.27348488569259644,
      "learning_rate": 5.027902355946637e-05,
      "loss": 0.0007488925009965896,
      "step": 175240
    },
    {
      "epoch": 49.74453590689753,
      "grad_norm": 1.489981770515442,
      "learning_rate": 5.0276185069543e-05,
      "loss": 0.0008515970781445503,
      "step": 175250
    },
    {
      "epoch": 49.74737439682089,
      "grad_norm": 0.03553125634789467,
      "learning_rate": 5.027334657961964e-05,
      "loss": 0.0027698434889316557,
      "step": 175260
    },
    {
      "epoch": 49.75021288674425,
      "grad_norm": 0.5971530079841614,
      "learning_rate": 5.027050808969629e-05,
      "loss": 0.0009780852124094963,
      "step": 175270
    },
    {
      "epoch": 49.753051376667614,
      "grad_norm": 0.2780238687992096,
      "learning_rate": 5.0267669599772925e-05,
      "loss": 0.0014037508517503738,
      "step": 175280
    },
    {
      "epoch": 49.75588986659098,
      "grad_norm": 0.12754395604133606,
      "learning_rate": 5.026483110984957e-05,
      "loss": 0.000493558682501316,
      "step": 175290
    },
    {
      "epoch": 49.75872835651433,
      "grad_norm": 0.33132442831993103,
      "learning_rate": 5.02619926199262e-05,
      "loss": 0.0009608305990695953,
      "step": 175300
    },
    {
      "epoch": 49.761566846437695,
      "grad_norm": 0.03972368687391281,
      "learning_rate": 5.0259154130002836e-05,
      "loss": 0.0016151823103427887,
      "step": 175310
    },
    {
      "epoch": 49.76440533636106,
      "grad_norm": 0.7699304819107056,
      "learning_rate": 5.025631564007948e-05,
      "loss": 0.00039414633065462115,
      "step": 175320
    },
    {
      "epoch": 49.767243826284414,
      "grad_norm": 0.01738104037940502,
      "learning_rate": 5.025347715015611e-05,
      "loss": 0.0006595015525817871,
      "step": 175330
    },
    {
      "epoch": 49.77008231620778,
      "grad_norm": 2.201671838760376,
      "learning_rate": 5.025063866023276e-05,
      "loss": 0.0015249321237206458,
      "step": 175340
    },
    {
      "epoch": 49.77292080613114,
      "grad_norm": 0.012536357156932354,
      "learning_rate": 5.02478001703094e-05,
      "loss": 0.0011964490637183189,
      "step": 175350
    },
    {
      "epoch": 49.7757592960545,
      "grad_norm": 0.45947346091270447,
      "learning_rate": 5.0244961680386036e-05,
      "loss": 0.002071053348481655,
      "step": 175360
    },
    {
      "epoch": 49.77859778597786,
      "grad_norm": 1.0450935363769531,
      "learning_rate": 5.024212319046268e-05,
      "loss": 0.0012564558535814285,
      "step": 175370
    },
    {
      "epoch": 49.78143627590122,
      "grad_norm": 0.03372954949736595,
      "learning_rate": 5.023928470053931e-05,
      "loss": 0.0012946862727403641,
      "step": 175380
    },
    {
      "epoch": 49.784274765824584,
      "grad_norm": 0.15097546577453613,
      "learning_rate": 5.0236446210615953e-05,
      "loss": 0.0017166053876280785,
      "step": 175390
    },
    {
      "epoch": 49.78711325574794,
      "grad_norm": 0.3945470154285431,
      "learning_rate": 5.02336077206926e-05,
      "loss": 0.0020857272669672967,
      "step": 175400
    },
    {
      "epoch": 49.7899517456713,
      "grad_norm": 0.12289819121360779,
      "learning_rate": 5.0230769230769236e-05,
      "loss": 0.0007977455854415894,
      "step": 175410
    },
    {
      "epoch": 49.792790235594666,
      "grad_norm": 2.9969444274902344,
      "learning_rate": 5.022793074084588e-05,
      "loss": 0.003031211718916893,
      "step": 175420
    },
    {
      "epoch": 49.79562872551802,
      "grad_norm": 0.20918305218219757,
      "learning_rate": 5.022509225092251e-05,
      "loss": 0.0013728175312280656,
      "step": 175430
    },
    {
      "epoch": 49.798467215441384,
      "grad_norm": 0.5579810738563538,
      "learning_rate": 5.022225376099915e-05,
      "loss": 0.0017795367166399956,
      "step": 175440
    },
    {
      "epoch": 49.80130570536475,
      "grad_norm": 0.254611074924469,
      "learning_rate": 5.021941527107579e-05,
      "loss": 0.0014431871473789214,
      "step": 175450
    },
    {
      "epoch": 49.80414419528811,
      "grad_norm": 0.2649161219596863,
      "learning_rate": 5.021657678115242e-05,
      "loss": 0.0010729063302278518,
      "step": 175460
    },
    {
      "epoch": 49.806982685211466,
      "grad_norm": 0.037268947809934616,
      "learning_rate": 5.021373829122907e-05,
      "loss": 0.001562812365591526,
      "step": 175470
    },
    {
      "epoch": 49.80982117513483,
      "grad_norm": 0.8643369674682617,
      "learning_rate": 5.021089980130571e-05,
      "loss": 0.0012505130842328071,
      "step": 175480
    },
    {
      "epoch": 49.81265966505819,
      "grad_norm": 0.04570915177464485,
      "learning_rate": 5.020806131138235e-05,
      "loss": 0.0012321338057518006,
      "step": 175490
    },
    {
      "epoch": 49.81549815498155,
      "grad_norm": 1.856271505355835,
      "learning_rate": 5.020522282145899e-05,
      "loss": 0.0013563914224505424,
      "step": 175500
    },
    {
      "epoch": 49.81549815498155,
      "eval_accuracy": 0.972722070324919,
      "eval_loss": 0.10761478543281555,
      "eval_runtime": 32.4377,
      "eval_samples_per_second": 484.837,
      "eval_steps_per_second": 7.584,
      "step": 175500
    },
    {
      "epoch": 49.81833664490491,
      "grad_norm": 0.08523158729076385,
      "learning_rate": 5.020238433153562e-05,
      "loss": 0.003979005664587021,
      "step": 175510
    },
    {
      "epoch": 49.82117513482827,
      "grad_norm": 0.20873010158538818,
      "learning_rate": 5.0199545841612264e-05,
      "loss": 0.0012896081432700157,
      "step": 175520
    },
    {
      "epoch": 49.82401362475163,
      "grad_norm": 10.26535415649414,
      "learning_rate": 5.01967073516889e-05,
      "loss": 0.0028474293649196625,
      "step": 175530
    },
    {
      "epoch": 49.82685211467499,
      "grad_norm": 0.047118719667196274,
      "learning_rate": 5.019386886176555e-05,
      "loss": 0.000449572317302227,
      "step": 175540
    },
    {
      "epoch": 49.829690604598355,
      "grad_norm": 0.12147834151983261,
      "learning_rate": 5.019103037184218e-05,
      "loss": 0.0072993718087673186,
      "step": 175550
    },
    {
      "epoch": 49.83252909452172,
      "grad_norm": 0.22156867384910583,
      "learning_rate": 5.018819188191882e-05,
      "loss": 0.0014776095747947692,
      "step": 175560
    },
    {
      "epoch": 49.83536758444507,
      "grad_norm": 0.18199506402015686,
      "learning_rate": 5.018535339199546e-05,
      "loss": 0.0015635907649993897,
      "step": 175570
    },
    {
      "epoch": 49.838206074368436,
      "grad_norm": 0.0061673372983932495,
      "learning_rate": 5.01825149020721e-05,
      "loss": 0.004207996651530266,
      "step": 175580
    },
    {
      "epoch": 49.8410445642918,
      "grad_norm": 0.3063407242298126,
      "learning_rate": 5.0179676412148733e-05,
      "loss": 0.004273391515016556,
      "step": 175590
    },
    {
      "epoch": 49.843883054215155,
      "grad_norm": 0.18359903991222382,
      "learning_rate": 5.017683792222538e-05,
      "loss": 0.000860634632408619,
      "step": 175600
    },
    {
      "epoch": 49.84672154413852,
      "grad_norm": 0.6394650936126709,
      "learning_rate": 5.017399943230202e-05,
      "loss": 0.0022802097722887994,
      "step": 175610
    },
    {
      "epoch": 49.84956003406188,
      "grad_norm": 0.2010713666677475,
      "learning_rate": 5.017116094237866e-05,
      "loss": 0.000846301019191742,
      "step": 175620
    },
    {
      "epoch": 49.85239852398524,
      "grad_norm": 0.03978755697607994,
      "learning_rate": 5.01683224524553e-05,
      "loss": 0.0013067258521914482,
      "step": 175630
    },
    {
      "epoch": 49.8552370139086,
      "grad_norm": 0.0057344259694218636,
      "learning_rate": 5.0165483962531934e-05,
      "loss": 0.0007142610847949982,
      "step": 175640
    },
    {
      "epoch": 49.85807550383196,
      "grad_norm": 0.9011965394020081,
      "learning_rate": 5.016264547260857e-05,
      "loss": 0.0006152540445327759,
      "step": 175650
    },
    {
      "epoch": 49.860913993755325,
      "grad_norm": 0.10104791074991226,
      "learning_rate": 5.015980698268521e-05,
      "loss": 0.0005251636728644371,
      "step": 175660
    },
    {
      "epoch": 49.86375248367868,
      "grad_norm": 0.023529106751084328,
      "learning_rate": 5.015696849276186e-05,
      "loss": 0.002229335904121399,
      "step": 175670
    },
    {
      "epoch": 49.866590973602044,
      "grad_norm": 0.1872633844614029,
      "learning_rate": 5.015413000283849e-05,
      "loss": 0.004377792775630951,
      "step": 175680
    },
    {
      "epoch": 49.86942946352541,
      "grad_norm": 0.20909199118614197,
      "learning_rate": 5.0151291512915134e-05,
      "loss": 0.00166688933968544,
      "step": 175690
    },
    {
      "epoch": 49.87226795344876,
      "grad_norm": 0.10742336511611938,
      "learning_rate": 5.014845302299177e-05,
      "loss": 0.0012144222855567932,
      "step": 175700
    },
    {
      "epoch": 49.875106443372125,
      "grad_norm": 0.04432913288474083,
      "learning_rate": 5.014561453306841e-05,
      "loss": 0.0005228672176599503,
      "step": 175710
    },
    {
      "epoch": 49.87794493329549,
      "grad_norm": 0.06133781000971794,
      "learning_rate": 5.0142776043145044e-05,
      "loss": 0.0024652579799294473,
      "step": 175720
    },
    {
      "epoch": 49.880783423218844,
      "grad_norm": 0.02541930414736271,
      "learning_rate": 5.0139937553221686e-05,
      "loss": 0.0008047346025705337,
      "step": 175730
    },
    {
      "epoch": 49.88362191314221,
      "grad_norm": 0.10522088408470154,
      "learning_rate": 5.0137099063298334e-05,
      "loss": 0.013824224472045898,
      "step": 175740
    },
    {
      "epoch": 49.88646040306557,
      "grad_norm": 0.04452231526374817,
      "learning_rate": 5.013426057337497e-05,
      "loss": 0.002551809512078762,
      "step": 175750
    },
    {
      "epoch": 49.88929889298893,
      "grad_norm": 0.5581021308898926,
      "learning_rate": 5.013142208345161e-05,
      "loss": 0.002123919874429703,
      "step": 175760
    },
    {
      "epoch": 49.89213738291229,
      "grad_norm": 0.20246925950050354,
      "learning_rate": 5.0128583593528244e-05,
      "loss": 0.0008541466668248177,
      "step": 175770
    },
    {
      "epoch": 49.89497587283565,
      "grad_norm": 1.4875656366348267,
      "learning_rate": 5.012574510360488e-05,
      "loss": 0.0004628987982869148,
      "step": 175780
    },
    {
      "epoch": 49.897814362759014,
      "grad_norm": 0.18628545105457306,
      "learning_rate": 5.012290661368152e-05,
      "loss": 0.0006874006241559983,
      "step": 175790
    },
    {
      "epoch": 49.90065285268237,
      "grad_norm": 0.01880861073732376,
      "learning_rate": 5.012006812375817e-05,
      "loss": 0.000714532844722271,
      "step": 175800
    },
    {
      "epoch": 49.90349134260573,
      "grad_norm": 0.3142130374908447,
      "learning_rate": 5.01172296338348e-05,
      "loss": 0.0009714033454656601,
      "step": 175810
    },
    {
      "epoch": 49.906329832529096,
      "grad_norm": 0.03135831654071808,
      "learning_rate": 5.0114391143911444e-05,
      "loss": 0.0008255355060100555,
      "step": 175820
    },
    {
      "epoch": 49.90916832245246,
      "grad_norm": 0.11629349738359451,
      "learning_rate": 5.011155265398808e-05,
      "loss": 0.00041448790580034256,
      "step": 175830
    },
    {
      "epoch": 49.912006812375814,
      "grad_norm": 0.013747074641287327,
      "learning_rate": 5.010871416406472e-05,
      "loss": 0.0013789813965559006,
      "step": 175840
    },
    {
      "epoch": 49.91484530229918,
      "grad_norm": 0.01844453439116478,
      "learning_rate": 5.0105875674141355e-05,
      "loss": 0.0007831752300262451,
      "step": 175850
    },
    {
      "epoch": 49.91768379222254,
      "grad_norm": 0.05346502736210823,
      "learning_rate": 5.0103037184217996e-05,
      "loss": 0.0007181094959378242,
      "step": 175860
    },
    {
      "epoch": 49.920522282145896,
      "grad_norm": 7.567402362823486,
      "learning_rate": 5.0100198694294645e-05,
      "loss": 0.0020791573449969293,
      "step": 175870
    },
    {
      "epoch": 49.92336077206926,
      "grad_norm": 0.2983570396900177,
      "learning_rate": 5.009736020437128e-05,
      "loss": 0.0012483488768339156,
      "step": 175880
    },
    {
      "epoch": 49.92619926199262,
      "grad_norm": 0.021005239337682724,
      "learning_rate": 5.009452171444792e-05,
      "loss": 0.0011861098930239678,
      "step": 175890
    },
    {
      "epoch": 49.92903775191598,
      "grad_norm": 0.21002040803432465,
      "learning_rate": 5.0091683224524555e-05,
      "loss": 0.006847255676984787,
      "step": 175900
    },
    {
      "epoch": 49.93187624183934,
      "grad_norm": 0.06860771030187607,
      "learning_rate": 5.008884473460119e-05,
      "loss": 0.0009809765964746476,
      "step": 175910
    },
    {
      "epoch": 49.9347147317627,
      "grad_norm": 0.38959622383117676,
      "learning_rate": 5.008600624467783e-05,
      "loss": 0.0010251175612211228,
      "step": 175920
    },
    {
      "epoch": 49.937553221686066,
      "grad_norm": 0.2003687024116516,
      "learning_rate": 5.0083167754754466e-05,
      "loss": 0.0016873536631464958,
      "step": 175930
    },
    {
      "epoch": 49.94039171160942,
      "grad_norm": 0.7171655893325806,
      "learning_rate": 5.0080329264831114e-05,
      "loss": 0.005209941789507866,
      "step": 175940
    },
    {
      "epoch": 49.943230201532785,
      "grad_norm": 0.18133193254470825,
      "learning_rate": 5.0077490774907755e-05,
      "loss": 0.0007899625226855278,
      "step": 175950
    },
    {
      "epoch": 49.94606869145615,
      "grad_norm": 1.4987937211990356,
      "learning_rate": 5.007465228498439e-05,
      "loss": 0.0011442881077528,
      "step": 175960
    },
    {
      "epoch": 49.948907181379504,
      "grad_norm": 0.6878413558006287,
      "learning_rate": 5.007181379506103e-05,
      "loss": 0.0023087671026587485,
      "step": 175970
    },
    {
      "epoch": 49.95174567130287,
      "grad_norm": 0.13091632723808289,
      "learning_rate": 5.0068975305137666e-05,
      "loss": 0.006896325200796127,
      "step": 175980
    },
    {
      "epoch": 49.95458416122623,
      "grad_norm": 2.2272679805755615,
      "learning_rate": 5.006613681521431e-05,
      "loss": 0.012304463237524033,
      "step": 175990
    },
    {
      "epoch": 49.957422651149585,
      "grad_norm": 6.90089225769043,
      "learning_rate": 5.0063298325290955e-05,
      "loss": 0.0037517033517360686,
      "step": 176000
    },
    {
      "epoch": 49.957422651149585,
      "eval_accuracy": 0.9685890506771794,
      "eval_loss": 0.11283570528030396,
      "eval_runtime": 32.6927,
      "eval_samples_per_second": 481.055,
      "eval_steps_per_second": 7.525,
      "step": 176000
    },
    {
      "epoch": 49.96026114107295,
      "grad_norm": 0.6998717784881592,
      "learning_rate": 5.006045983536759e-05,
      "loss": 0.0036075346171855925,
      "step": 176010
    },
    {
      "epoch": 49.96309963099631,
      "grad_norm": 1.8785208463668823,
      "learning_rate": 5.0057621345444225e-05,
      "loss": 0.0015206236392259597,
      "step": 176020
    },
    {
      "epoch": 49.965938120919674,
      "grad_norm": 0.2636459767818451,
      "learning_rate": 5.0054782855520866e-05,
      "loss": 0.000851638801395893,
      "step": 176030
    },
    {
      "epoch": 49.96877661084303,
      "grad_norm": 0.7934018969535828,
      "learning_rate": 5.00519443655975e-05,
      "loss": 0.005938491225242615,
      "step": 176040
    },
    {
      "epoch": 49.97161510076639,
      "grad_norm": 0.10201861709356308,
      "learning_rate": 5.004910587567414e-05,
      "loss": 0.0025740481913089754,
      "step": 176050
    },
    {
      "epoch": 49.974453590689755,
      "grad_norm": 0.08057184517383575,
      "learning_rate": 5.0046267385750776e-05,
      "loss": 0.002357472479343414,
      "step": 176060
    },
    {
      "epoch": 49.97729208061311,
      "grad_norm": 3.975041151046753,
      "learning_rate": 5.0043428895827425e-05,
      "loss": 0.0011539261788129807,
      "step": 176070
    },
    {
      "epoch": 49.980130570536474,
      "grad_norm": 9.44887638092041,
      "learning_rate": 5.0040590405904066e-05,
      "loss": 0.002850980870425701,
      "step": 176080
    },
    {
      "epoch": 49.98296906045984,
      "grad_norm": 0.028105955570936203,
      "learning_rate": 5.00377519159807e-05,
      "loss": 0.003853920102119446,
      "step": 176090
    },
    {
      "epoch": 49.98580755038319,
      "grad_norm": 0.0209950041025877,
      "learning_rate": 5.003491342605734e-05,
      "loss": 0.0007055636495351791,
      "step": 176100
    },
    {
      "epoch": 49.988646040306556,
      "grad_norm": 2.3295576572418213,
      "learning_rate": 5.0032074936133977e-05,
      "loss": 0.002795134112238884,
      "step": 176110
    },
    {
      "epoch": 49.99148453022992,
      "grad_norm": 0.11044897139072418,
      "learning_rate": 5.002923644621061e-05,
      "loss": 0.00271771214902401,
      "step": 176120
    },
    {
      "epoch": 49.99432302015328,
      "grad_norm": 0.08135133981704712,
      "learning_rate": 5.0026397956287266e-05,
      "loss": 0.0027631301432847976,
      "step": 176130
    },
    {
      "epoch": 49.99716151007664,
      "grad_norm": 0.47417858242988586,
      "learning_rate": 5.00235594663639e-05,
      "loss": 0.0013587038964033126,
      "step": 176140
    },
    {
      "epoch": 50.0,
      "grad_norm": 0.189507856965065,
      "learning_rate": 5.0020720976440535e-05,
      "loss": 0.0007840776816010475,
      "step": 176150
    },
    {
      "epoch": 50.00283848992336,
      "grad_norm": 3.7724413871765137,
      "learning_rate": 5.001788248651718e-05,
      "loss": 0.0008784815669059754,
      "step": 176160
    },
    {
      "epoch": 50.00567697984672,
      "grad_norm": 0.2793523967266083,
      "learning_rate": 5.001504399659381e-05,
      "loss": 0.0010044803842902184,
      "step": 176170
    },
    {
      "epoch": 50.00851546977008,
      "grad_norm": 7.323808193206787,
      "learning_rate": 5.001220550667045e-05,
      "loss": 0.0017520207911729813,
      "step": 176180
    },
    {
      "epoch": 50.011353959693444,
      "grad_norm": 0.02302386984229088,
      "learning_rate": 5.000936701674709e-05,
      "loss": 0.0020531518384814262,
      "step": 176190
    },
    {
      "epoch": 50.01419244961681,
      "grad_norm": 0.2943992614746094,
      "learning_rate": 5.0006528526823735e-05,
      "loss": 0.0007233167067170143,
      "step": 176200
    },
    {
      "epoch": 50.01703093954016,
      "grad_norm": 0.007859306409955025,
      "learning_rate": 5.000369003690038e-05,
      "loss": 0.0010492430999875068,
      "step": 176210
    },
    {
      "epoch": 50.019869429463526,
      "grad_norm": 0.5912808775901794,
      "learning_rate": 5.000085154697701e-05,
      "loss": 0.0013657096773386002,
      "step": 176220
    },
    {
      "epoch": 50.02270791938689,
      "grad_norm": 1.8536204099655151,
      "learning_rate": 4.999801305705365e-05,
      "loss": 0.0008431902155280113,
      "step": 176230
    },
    {
      "epoch": 50.025546409310245,
      "grad_norm": 6.110299110412598,
      "learning_rate": 4.999517456713029e-05,
      "loss": 0.0024310681968927384,
      "step": 176240
    },
    {
      "epoch": 50.02838489923361,
      "grad_norm": 0.6818021535873413,
      "learning_rate": 4.999233607720693e-05,
      "loss": 0.005994686484336853,
      "step": 176250
    },
    {
      "epoch": 50.03122338915697,
      "grad_norm": 0.08530093729496002,
      "learning_rate": 4.998949758728357e-05,
      "loss": 0.0017384717240929604,
      "step": 176260
    },
    {
      "epoch": 50.034061879080326,
      "grad_norm": 1.48838210105896,
      "learning_rate": 4.9986659097360205e-05,
      "loss": 0.0006951028481125832,
      "step": 176270
    },
    {
      "epoch": 50.03690036900369,
      "grad_norm": 0.06297750771045685,
      "learning_rate": 4.9983820607436846e-05,
      "loss": 0.0005120839923620224,
      "step": 176280
    },
    {
      "epoch": 50.03973885892705,
      "grad_norm": 0.02089560218155384,
      "learning_rate": 4.998098211751349e-05,
      "loss": 0.0008932847529649734,
      "step": 176290
    },
    {
      "epoch": 50.042577348850415,
      "grad_norm": 0.06646229326725006,
      "learning_rate": 4.997814362759012e-05,
      "loss": 0.0009349917992949486,
      "step": 176300
    },
    {
      "epoch": 50.04541583877377,
      "grad_norm": 0.1431989222764969,
      "learning_rate": 4.997530513766676e-05,
      "loss": 0.00023363307118415833,
      "step": 176310
    },
    {
      "epoch": 50.04825432869713,
      "grad_norm": 0.020652173087000847,
      "learning_rate": 4.9972466647743405e-05,
      "loss": 0.0023580309003591537,
      "step": 176320
    },
    {
      "epoch": 50.051092818620496,
      "grad_norm": 0.05032404512166977,
      "learning_rate": 4.996962815782004e-05,
      "loss": 0.0007284469902515411,
      "step": 176330
    },
    {
      "epoch": 50.05393130854385,
      "grad_norm": 0.2373400777578354,
      "learning_rate": 4.996678966789668e-05,
      "loss": 0.0009558342397212982,
      "step": 176340
    },
    {
      "epoch": 50.056769798467215,
      "grad_norm": 5.793634414672852,
      "learning_rate": 4.996395117797332e-05,
      "loss": 0.0010962445288896562,
      "step": 176350
    },
    {
      "epoch": 50.05960828839058,
      "grad_norm": 0.13304241001605988,
      "learning_rate": 4.9961112688049963e-05,
      "loss": 0.0009592054411768913,
      "step": 176360
    },
    {
      "epoch": 50.062446778313934,
      "grad_norm": 2.5221426486968994,
      "learning_rate": 4.99582741981266e-05,
      "loss": 0.0008399656042456626,
      "step": 176370
    },
    {
      "epoch": 50.0652852682373,
      "grad_norm": 0.046720437705516815,
      "learning_rate": 4.995543570820324e-05,
      "loss": 0.0006598435342311859,
      "step": 176380
    },
    {
      "epoch": 50.06812375816066,
      "grad_norm": 0.4744413197040558,
      "learning_rate": 4.995259721827988e-05,
      "loss": 0.0022697716951370238,
      "step": 176390
    },
    {
      "epoch": 50.07096224808402,
      "grad_norm": 0.08460473269224167,
      "learning_rate": 4.9949758728356515e-05,
      "loss": 0.0026431258767843245,
      "step": 176400
    },
    {
      "epoch": 50.07380073800738,
      "grad_norm": 9.851479530334473,
      "learning_rate": 4.994692023843316e-05,
      "loss": 0.003089432045817375,
      "step": 176410
    },
    {
      "epoch": 50.07663922793074,
      "grad_norm": 0.8279853463172913,
      "learning_rate": 4.99440817485098e-05,
      "loss": 0.0012674272060394288,
      "step": 176420
    },
    {
      "epoch": 50.079477717854104,
      "grad_norm": 0.18671581149101257,
      "learning_rate": 4.994124325858643e-05,
      "loss": 0.00387532040476799,
      "step": 176430
    },
    {
      "epoch": 50.08231620777746,
      "grad_norm": 0.878939688205719,
      "learning_rate": 4.9938404768663074e-05,
      "loss": 0.0006136523559689521,
      "step": 176440
    },
    {
      "epoch": 50.08515469770082,
      "grad_norm": 0.11840343475341797,
      "learning_rate": 4.9935566278739716e-05,
      "loss": 0.009156681597232819,
      "step": 176450
    },
    {
      "epoch": 50.087993187624186,
      "grad_norm": 0.08839007467031479,
      "learning_rate": 4.993272778881635e-05,
      "loss": 0.0035822682082653047,
      "step": 176460
    },
    {
      "epoch": 50.09083167754754,
      "grad_norm": 0.08188849687576294,
      "learning_rate": 4.992988929889299e-05,
      "loss": 0.0022962478920817374,
      "step": 176470
    },
    {
      "epoch": 50.093670167470904,
      "grad_norm": 2.2806150913238525,
      "learning_rate": 4.992705080896963e-05,
      "loss": 0.0011621389538049697,
      "step": 176480
    },
    {
      "epoch": 50.09650865739427,
      "grad_norm": 0.2724352777004242,
      "learning_rate": 4.9924496168038606e-05,
      "loss": 0.0025118185207247735,
      "step": 176490
    },
    {
      "epoch": 50.09934714731763,
      "grad_norm": 1.164432406425476,
      "learning_rate": 4.992165767811525e-05,
      "loss": 0.0037373602390289307,
      "step": 176500
    },
    {
      "epoch": 50.09934714731763,
      "eval_accuracy": 0.9769822598079736,
      "eval_loss": 0.08625923097133636,
      "eval_runtime": 32.2141,
      "eval_samples_per_second": 488.202,
      "eval_steps_per_second": 7.636,
      "step": 176500
    },
    {
      "epoch": 50.102185637240986,
      "grad_norm": 1.7083172798156738,
      "learning_rate": 4.991910303718422e-05,
      "loss": 0.008183462172746658,
      "step": 176510
    },
    {
      "epoch": 50.10502412716435,
      "grad_norm": 1.135770320892334,
      "learning_rate": 4.991626454726086e-05,
      "loss": 0.0005703866481781006,
      "step": 176520
    },
    {
      "epoch": 50.10786261708771,
      "grad_norm": 0.09532252699136734,
      "learning_rate": 4.9913426057337497e-05,
      "loss": 0.0009808547794818879,
      "step": 176530
    },
    {
      "epoch": 50.11070110701107,
      "grad_norm": 0.4745347499847412,
      "learning_rate": 4.991058756741414e-05,
      "loss": 0.0028731677681207656,
      "step": 176540
    },
    {
      "epoch": 50.11353959693443,
      "grad_norm": 0.03605005145072937,
      "learning_rate": 4.990774907749078e-05,
      "loss": 0.0009749777615070343,
      "step": 176550
    },
    {
      "epoch": 50.11637808685779,
      "grad_norm": 0.07686666399240494,
      "learning_rate": 4.9904910587567414e-05,
      "loss": 0.0005639469251036644,
      "step": 176560
    },
    {
      "epoch": 50.119216576781156,
      "grad_norm": 0.006282479967921972,
      "learning_rate": 4.9902072097644055e-05,
      "loss": 0.0007820015773177147,
      "step": 176570
    },
    {
      "epoch": 50.12205506670451,
      "grad_norm": 0.02165645733475685,
      "learning_rate": 4.98992336077207e-05,
      "loss": 0.005430639907717705,
      "step": 176580
    },
    {
      "epoch": 50.124893556627875,
      "grad_norm": 15.569891929626465,
      "learning_rate": 4.989639511779733e-05,
      "loss": 0.0063955865800380705,
      "step": 176590
    },
    {
      "epoch": 50.12773204655124,
      "grad_norm": 0.2452220916748047,
      "learning_rate": 4.989355662787397e-05,
      "loss": 0.002078629843890667,
      "step": 176600
    },
    {
      "epoch": 50.13057053647459,
      "grad_norm": 0.0274581890553236,
      "learning_rate": 4.9890718137950614e-05,
      "loss": 0.0007554735988378525,
      "step": 176610
    },
    {
      "epoch": 50.133409026397956,
      "grad_norm": 1.1781631708145142,
      "learning_rate": 4.9887879648027255e-05,
      "loss": 0.000506405532360077,
      "step": 176620
    },
    {
      "epoch": 50.13624751632132,
      "grad_norm": 0.04099762812256813,
      "learning_rate": 4.988504115810389e-05,
      "loss": 0.0004459936171770096,
      "step": 176630
    },
    {
      "epoch": 50.139086006244675,
      "grad_norm": 0.028696412220597267,
      "learning_rate": 4.9882202668180525e-05,
      "loss": 0.002434944733977318,
      "step": 176640
    },
    {
      "epoch": 50.14192449616804,
      "grad_norm": 0.014150276780128479,
      "learning_rate": 4.987936417825717e-05,
      "loss": 0.0015130940824747085,
      "step": 176650
    },
    {
      "epoch": 50.1447629860914,
      "grad_norm": 0.3179625868797302,
      "learning_rate": 4.987652568833381e-05,
      "loss": 0.0006388736888766289,
      "step": 176660
    },
    {
      "epoch": 50.14760147601476,
      "grad_norm": 0.3257445991039276,
      "learning_rate": 4.987368719841045e-05,
      "loss": 0.0005356330424547195,
      "step": 176670
    },
    {
      "epoch": 50.15043996593812,
      "grad_norm": 0.028779519721865654,
      "learning_rate": 4.987084870848709e-05,
      "loss": 0.0004261115565896034,
      "step": 176680
    },
    {
      "epoch": 50.15327845586148,
      "grad_norm": 0.9095824360847473,
      "learning_rate": 4.9868010218563725e-05,
      "loss": 0.0019007945433259011,
      "step": 176690
    },
    {
      "epoch": 50.156116945784845,
      "grad_norm": 0.6246103048324585,
      "learning_rate": 4.9865171728640366e-05,
      "loss": 0.0005518237128853797,
      "step": 176700
    },
    {
      "epoch": 50.1589554357082,
      "grad_norm": 0.14597567915916443,
      "learning_rate": 4.986233323871701e-05,
      "loss": 0.0005241403356194496,
      "step": 176710
    },
    {
      "epoch": 50.161793925631564,
      "grad_norm": 0.062584787607193,
      "learning_rate": 4.985949474879364e-05,
      "loss": 0.0012103330343961715,
      "step": 176720
    },
    {
      "epoch": 50.16463241555493,
      "grad_norm": 0.15316635370254517,
      "learning_rate": 4.9856656258870283e-05,
      "loss": 0.0030560286715626715,
      "step": 176730
    },
    {
      "epoch": 50.16747090547828,
      "grad_norm": 0.02110178954899311,
      "learning_rate": 4.985381776894692e-05,
      "loss": 0.0046000611037015915,
      "step": 176740
    },
    {
      "epoch": 50.170309395401645,
      "grad_norm": 0.04300222545862198,
      "learning_rate": 4.9850979279023566e-05,
      "loss": 0.00060130525380373,
      "step": 176750
    },
    {
      "epoch": 50.17314788532501,
      "grad_norm": 0.6389912962913513,
      "learning_rate": 4.98481407891002e-05,
      "loss": 0.0014501934871077537,
      "step": 176760
    },
    {
      "epoch": 50.17598637524837,
      "grad_norm": 0.0434531532227993,
      "learning_rate": 4.9845302299176835e-05,
      "loss": 0.0004605159163475037,
      "step": 176770
    },
    {
      "epoch": 50.17882486517173,
      "grad_norm": 1.0250457525253296,
      "learning_rate": 4.9842463809253483e-05,
      "loss": 0.0008345304057002067,
      "step": 176780
    },
    {
      "epoch": 50.18166335509509,
      "grad_norm": 0.05345132201910019,
      "learning_rate": 4.983962531933012e-05,
      "loss": 0.0004011526703834534,
      "step": 176790
    },
    {
      "epoch": 50.18450184501845,
      "grad_norm": 1.5893410444259644,
      "learning_rate": 4.983678682940676e-05,
      "loss": 0.015104100108146667,
      "step": 176800
    },
    {
      "epoch": 50.18734033494181,
      "grad_norm": 2.648761510848999,
      "learning_rate": 4.98339483394834e-05,
      "loss": 0.0007993241772055626,
      "step": 176810
    },
    {
      "epoch": 50.19017882486517,
      "grad_norm": 0.4990914463996887,
      "learning_rate": 4.9831109849560035e-05,
      "loss": 0.002957923151552677,
      "step": 176820
    },
    {
      "epoch": 50.193017314788534,
      "grad_norm": 0.1258476823568344,
      "learning_rate": 4.982827135963668e-05,
      "loss": 0.0008682165294885636,
      "step": 176830
    },
    {
      "epoch": 50.19585580471189,
      "grad_norm": 0.21630902588367462,
      "learning_rate": 4.982543286971331e-05,
      "loss": 0.000866316445171833,
      "step": 176840
    },
    {
      "epoch": 50.19869429463525,
      "grad_norm": 0.09617657959461212,
      "learning_rate": 4.982259437978995e-05,
      "loss": 0.0005035484209656715,
      "step": 176850
    },
    {
      "epoch": 50.201532784558616,
      "grad_norm": 0.01744590885937214,
      "learning_rate": 4.9819755889866594e-05,
      "loss": 0.0004816923290491104,
      "step": 176860
    },
    {
      "epoch": 50.20437127448198,
      "grad_norm": 0.052794624119997025,
      "learning_rate": 4.981691739994323e-05,
      "loss": 0.0017745234072208405,
      "step": 176870
    },
    {
      "epoch": 50.207209764405334,
      "grad_norm": 0.07953483611345291,
      "learning_rate": 4.981407891001987e-05,
      "loss": 0.001958402805030346,
      "step": 176880
    },
    {
      "epoch": 50.2100482543287,
      "grad_norm": 0.11982344835996628,
      "learning_rate": 4.981124042009651e-05,
      "loss": 0.00464974045753479,
      "step": 176890
    },
    {
      "epoch": 50.21288674425206,
      "grad_norm": 0.5792988538742065,
      "learning_rate": 4.9808401930173146e-05,
      "loss": 0.001023976132273674,
      "step": 176900
    },
    {
      "epoch": 50.215725234175416,
      "grad_norm": 2.3434958457946777,
      "learning_rate": 4.9805563440249794e-05,
      "loss": 0.003122660145163536,
      "step": 176910
    },
    {
      "epoch": 50.21856372409878,
      "grad_norm": 0.6093628406524658,
      "learning_rate": 4.980272495032643e-05,
      "loss": 0.0005148198455572128,
      "step": 176920
    },
    {
      "epoch": 50.22140221402214,
      "grad_norm": 0.030891595408320427,
      "learning_rate": 4.9799886460403063e-05,
      "loss": 0.0015668433159589767,
      "step": 176930
    },
    {
      "epoch": 50.2242407039455,
      "grad_norm": 0.2774398922920227,
      "learning_rate": 4.979704797047971e-05,
      "loss": 0.0003255235031247139,
      "step": 176940
    },
    {
      "epoch": 50.22707919386886,
      "grad_norm": 8.698624610900879,
      "learning_rate": 4.9794209480556346e-05,
      "loss": 0.0018035238608717918,
      "step": 176950
    },
    {
      "epoch": 50.22991768379222,
      "grad_norm": 2.889691114425659,
      "learning_rate": 4.979137099063299e-05,
      "loss": 0.002668721787631512,
      "step": 176960
    },
    {
      "epoch": 50.232756173715586,
      "grad_norm": 7.766643524169922,
      "learning_rate": 4.978853250070962e-05,
      "loss": 0.00206392128020525,
      "step": 176970
    },
    {
      "epoch": 50.23559466363894,
      "grad_norm": 9.352635383605957,
      "learning_rate": 4.9785694010786264e-05,
      "loss": 0.0016606124117970467,
      "step": 176980
    },
    {
      "epoch": 50.238433153562305,
      "grad_norm": 1.1609859466552734,
      "learning_rate": 4.9782855520862905e-05,
      "loss": 0.0018675822764635086,
      "step": 176990
    },
    {
      "epoch": 50.24127164348567,
      "grad_norm": 0.11830850690603256,
      "learning_rate": 4.978001703093954e-05,
      "loss": 0.008323868364095688,
      "step": 177000
    },
    {
      "epoch": 50.24127164348567,
      "eval_accuracy": 0.9779360335728365,
      "eval_loss": 0.0856231302022934,
      "eval_runtime": 32.778,
      "eval_samples_per_second": 479.804,
      "eval_steps_per_second": 7.505,
      "step": 177000
    },
    {
      "epoch": 50.24411013340902,
      "grad_norm": 0.08256211876869202,
      "learning_rate": 4.977717854101618e-05,
      "loss": 0.002439967170357704,
      "step": 177010
    },
    {
      "epoch": 50.246948623332386,
      "grad_norm": 0.19302710890769958,
      "learning_rate": 4.977434005109282e-05,
      "loss": 0.0024215953424572946,
      "step": 177020
    },
    {
      "epoch": 50.24978711325575,
      "grad_norm": 0.1892315000295639,
      "learning_rate": 4.977150156116946e-05,
      "loss": 0.0012735802680253982,
      "step": 177030
    },
    {
      "epoch": 50.25262560317911,
      "grad_norm": 0.07982230186462402,
      "learning_rate": 4.9768663071246105e-05,
      "loss": 0.0009816793724894523,
      "step": 177040
    },
    {
      "epoch": 50.25546409310247,
      "grad_norm": 0.041709598153829575,
      "learning_rate": 4.976582458132274e-05,
      "loss": 0.0005294255912303925,
      "step": 177050
    },
    {
      "epoch": 50.25830258302583,
      "grad_norm": 2.273650884628296,
      "learning_rate": 4.9762986091399374e-05,
      "loss": 0.001293467730283737,
      "step": 177060
    },
    {
      "epoch": 50.261141072949194,
      "grad_norm": 0.1955438256263733,
      "learning_rate": 4.9760147601476016e-05,
      "loss": 0.0007217144593596458,
      "step": 177070
    },
    {
      "epoch": 50.26397956287255,
      "grad_norm": 0.39188092947006226,
      "learning_rate": 4.975730911155266e-05,
      "loss": 0.0003591010347008705,
      "step": 177080
    },
    {
      "epoch": 50.26681805279591,
      "grad_norm": 0.05045894905924797,
      "learning_rate": 4.97544706216293e-05,
      "loss": 0.0004187233746051788,
      "step": 177090
    },
    {
      "epoch": 50.269656542719275,
      "grad_norm": 0.04422523453831673,
      "learning_rate": 4.975163213170593e-05,
      "loss": 0.00031277742236852647,
      "step": 177100
    },
    {
      "epoch": 50.27249503264263,
      "grad_norm": 0.19990792870521545,
      "learning_rate": 4.9748793641782574e-05,
      "loss": 0.0007353005930781364,
      "step": 177110
    },
    {
      "epoch": 50.275333522565994,
      "grad_norm": 0.06600464135408401,
      "learning_rate": 4.9745955151859216e-05,
      "loss": 0.0002290608361363411,
      "step": 177120
    },
    {
      "epoch": 50.27817201248936,
      "grad_norm": 0.2030678689479828,
      "learning_rate": 4.974311666193585e-05,
      "loss": 0.0007766973227262497,
      "step": 177130
    },
    {
      "epoch": 50.28101050241272,
      "grad_norm": 0.7113814353942871,
      "learning_rate": 4.974027817201249e-05,
      "loss": 0.0005529282614588737,
      "step": 177140
    },
    {
      "epoch": 50.283848992336075,
      "grad_norm": 0.02137497067451477,
      "learning_rate": 4.973743968208913e-05,
      "loss": 0.0005377311259508133,
      "step": 177150
    },
    {
      "epoch": 50.28668748225944,
      "grad_norm": 0.2531367540359497,
      "learning_rate": 4.973460119216577e-05,
      "loss": 0.000562996044754982,
      "step": 177160
    },
    {
      "epoch": 50.2895259721828,
      "grad_norm": 0.45157748460769653,
      "learning_rate": 4.973176270224241e-05,
      "loss": 0.001690037176012993,
      "step": 177170
    },
    {
      "epoch": 50.29236446210616,
      "grad_norm": 0.05742061883211136,
      "learning_rate": 4.972892421231905e-05,
      "loss": 0.0003670386970043182,
      "step": 177180
    },
    {
      "epoch": 50.29520295202952,
      "grad_norm": 0.07757249474525452,
      "learning_rate": 4.9726085722395685e-05,
      "loss": 0.0037764310836791992,
      "step": 177190
    },
    {
      "epoch": 50.29804144195288,
      "grad_norm": 0.055256836116313934,
      "learning_rate": 4.9723247232472326e-05,
      "loss": 0.0029304096475243568,
      "step": 177200
    },
    {
      "epoch": 50.30087993187624,
      "grad_norm": 0.5304828882217407,
      "learning_rate": 4.972040874254897e-05,
      "loss": 0.002909406274557114,
      "step": 177210
    },
    {
      "epoch": 50.3037184217996,
      "grad_norm": 0.33761054277420044,
      "learning_rate": 4.971757025262561e-05,
      "loss": 0.00031282082200050353,
      "step": 177220
    },
    {
      "epoch": 50.306556911722964,
      "grad_norm": 0.15963734686374664,
      "learning_rate": 4.9714731762702244e-05,
      "loss": 0.0008289923891425133,
      "step": 177230
    },
    {
      "epoch": 50.30939540164633,
      "grad_norm": 1.5332616567611694,
      "learning_rate": 4.9711893272778885e-05,
      "loss": 0.0015551738440990448,
      "step": 177240
    },
    {
      "epoch": 50.31223389156968,
      "grad_norm": 0.3817329704761505,
      "learning_rate": 4.9709054782855526e-05,
      "loss": 0.0006673812866210937,
      "step": 177250
    },
    {
      "epoch": 50.315072381493046,
      "grad_norm": 1.096095323562622,
      "learning_rate": 4.970621629293216e-05,
      "loss": 0.003609811142086983,
      "step": 177260
    },
    {
      "epoch": 50.31791087141641,
      "grad_norm": 0.5516811013221741,
      "learning_rate": 4.97033778030088e-05,
      "loss": 0.003701005131006241,
      "step": 177270
    },
    {
      "epoch": 50.320749361339765,
      "grad_norm": 0.24736420810222626,
      "learning_rate": 4.9700539313085444e-05,
      "loss": 0.0010272614657878877,
      "step": 177280
    },
    {
      "epoch": 50.32358785126313,
      "grad_norm": 0.030993130058050156,
      "learning_rate": 4.969770082316208e-05,
      "loss": 0.00017837658524513244,
      "step": 177290
    },
    {
      "epoch": 50.32642634118649,
      "grad_norm": 0.24044036865234375,
      "learning_rate": 4.969486233323872e-05,
      "loss": 0.0009753476828336715,
      "step": 177300
    },
    {
      "epoch": 50.329264831109846,
      "grad_norm": 0.24940486252307892,
      "learning_rate": 4.969202384331536e-05,
      "loss": 0.0012340404093265534,
      "step": 177310
    },
    {
      "epoch": 50.33210332103321,
      "grad_norm": 0.1399960219860077,
      "learning_rate": 4.9689185353391996e-05,
      "loss": 0.0009991338476538659,
      "step": 177320
    },
    {
      "epoch": 50.33494181095657,
      "grad_norm": 0.1798580437898636,
      "learning_rate": 4.968634686346864e-05,
      "loss": 0.001671215333044529,
      "step": 177330
    },
    {
      "epoch": 50.337780300879935,
      "grad_norm": 1.7044034004211426,
      "learning_rate": 4.968350837354528e-05,
      "loss": 0.0015520596876740455,
      "step": 177340
    },
    {
      "epoch": 50.34061879080329,
      "grad_norm": 0.1086382195353508,
      "learning_rate": 4.968066988362191e-05,
      "loss": 0.0027397431433200834,
      "step": 177350
    },
    {
      "epoch": 50.34345728072665,
      "grad_norm": 0.1624901294708252,
      "learning_rate": 4.9677831393698554e-05,
      "loss": 0.0011369818821549416,
      "step": 177360
    },
    {
      "epoch": 50.346295770650016,
      "grad_norm": 0.6583725214004517,
      "learning_rate": 4.967499290377519e-05,
      "loss": 0.0006082981824874877,
      "step": 177370
    },
    {
      "epoch": 50.34913426057337,
      "grad_norm": 0.11071088910102844,
      "learning_rate": 4.967215441385184e-05,
      "loss": 0.0009104721248149872,
      "step": 177380
    },
    {
      "epoch": 50.351972750496735,
      "grad_norm": 0.2696910798549652,
      "learning_rate": 4.966931592392847e-05,
      "loss": 0.00036047566682100295,
      "step": 177390
    },
    {
      "epoch": 50.3548112404201,
      "grad_norm": 0.13019660115242004,
      "learning_rate": 4.9666477434005106e-05,
      "loss": 0.0005082773044705391,
      "step": 177400
    },
    {
      "epoch": 50.35764973034346,
      "grad_norm": 1.0782768726348877,
      "learning_rate": 4.9663638944081755e-05,
      "loss": 0.0008891647681593895,
      "step": 177410
    },
    {
      "epoch": 50.36048822026682,
      "grad_norm": 0.05414819344878197,
      "learning_rate": 4.966080045415839e-05,
      "loss": 0.0029627135023474692,
      "step": 177420
    },
    {
      "epoch": 50.36332671019018,
      "grad_norm": 0.4545585513114929,
      "learning_rate": 4.965796196423503e-05,
      "loss": 0.0007574697956442833,
      "step": 177430
    },
    {
      "epoch": 50.36616520011354,
      "grad_norm": 0.0877997875213623,
      "learning_rate": 4.965512347431167e-05,
      "loss": 0.006679043173789978,
      "step": 177440
    },
    {
      "epoch": 50.3690036900369,
      "grad_norm": 1.6697880029678345,
      "learning_rate": 4.9652284984388307e-05,
      "loss": 0.0015411537140607833,
      "step": 177450
    },
    {
      "epoch": 50.37184217996026,
      "grad_norm": 0.09440036863088608,
      "learning_rate": 4.964944649446495e-05,
      "loss": 0.000862899050116539,
      "step": 177460
    },
    {
      "epoch": 50.374680669883624,
      "grad_norm": 0.03411230072379112,
      "learning_rate": 4.964660800454158e-05,
      "loss": 0.00018202271312475204,
      "step": 177470
    },
    {
      "epoch": 50.37751915980698,
      "grad_norm": 0.01655156910419464,
      "learning_rate": 4.9643769514618224e-05,
      "loss": 0.004170828312635422,
      "step": 177480
    },
    {
      "epoch": 50.38035764973034,
      "grad_norm": 0.7139278650283813,
      "learning_rate": 4.9640931024694865e-05,
      "loss": 0.002236265130341053,
      "step": 177490
    },
    {
      "epoch": 50.383196139653705,
      "grad_norm": 0.014876579865813255,
      "learning_rate": 4.96380925347715e-05,
      "loss": 0.0014686668291687965,
      "step": 177500
    },
    {
      "epoch": 50.383196139653705,
      "eval_accuracy": 0.9734850893368093,
      "eval_loss": 0.09695994853973389,
      "eval_runtime": 31.8462,
      "eval_samples_per_second": 493.842,
      "eval_steps_per_second": 7.725,
      "step": 177500
    },
    {
      "epoch": 50.38603462957707,
      "grad_norm": 0.05836835876107216,
      "learning_rate": 4.963525404484815e-05,
      "loss": 0.005047713220119476,
      "step": 177510
    },
    {
      "epoch": 50.388873119500424,
      "grad_norm": 3.8501880168914795,
      "learning_rate": 4.963241555492478e-05,
      "loss": 0.001256764866411686,
      "step": 177520
    },
    {
      "epoch": 50.39171160942379,
      "grad_norm": 4.1185221672058105,
      "learning_rate": 4.962957706500142e-05,
      "loss": 0.0011665267869830132,
      "step": 177530
    },
    {
      "epoch": 50.39455009934715,
      "grad_norm": 0.35036447644233704,
      "learning_rate": 4.9626738575078065e-05,
      "loss": 0.0007903749123215675,
      "step": 177540
    },
    {
      "epoch": 50.397388589270506,
      "grad_norm": 0.024981167167425156,
      "learning_rate": 4.96239000851547e-05,
      "loss": 0.0005061108618974686,
      "step": 177550
    },
    {
      "epoch": 50.40022707919387,
      "grad_norm": 0.05890714004635811,
      "learning_rate": 4.962106159523134e-05,
      "loss": 0.0007998129352927208,
      "step": 177560
    },
    {
      "epoch": 50.40306556911723,
      "grad_norm": 21.335351943969727,
      "learning_rate": 4.9618223105307976e-05,
      "loss": 0.007528568059206009,
      "step": 177570
    },
    {
      "epoch": 50.40590405904059,
      "grad_norm": 0.0112476646900177,
      "learning_rate": 4.961538461538462e-05,
      "loss": 0.0004099428653717041,
      "step": 177580
    },
    {
      "epoch": 50.40874254896395,
      "grad_norm": 0.032216690480709076,
      "learning_rate": 4.961254612546126e-05,
      "loss": 0.0012944817543029785,
      "step": 177590
    },
    {
      "epoch": 50.41158103888731,
      "grad_norm": 0.3729014992713928,
      "learning_rate": 4.960970763553789e-05,
      "loss": 0.00034929364919662477,
      "step": 177600
    },
    {
      "epoch": 50.414419528810676,
      "grad_norm": 0.05313735455274582,
      "learning_rate": 4.9606869145614535e-05,
      "loss": 0.0041382797062397,
      "step": 177610
    },
    {
      "epoch": 50.41725801873403,
      "grad_norm": 0.07106073945760727,
      "learning_rate": 4.9604030655691176e-05,
      "loss": 0.0027822446078062057,
      "step": 177620
    },
    {
      "epoch": 50.420096508657394,
      "grad_norm": 0.6923015117645264,
      "learning_rate": 4.960119216576781e-05,
      "loss": 0.004545788839459419,
      "step": 177630
    },
    {
      "epoch": 50.42293499858076,
      "grad_norm": 0.026344675570726395,
      "learning_rate": 4.959835367584446e-05,
      "loss": 0.002169114165008068,
      "step": 177640
    },
    {
      "epoch": 50.42577348850411,
      "grad_norm": 0.8514630198478699,
      "learning_rate": 4.959551518592109e-05,
      "loss": 0.00214039608836174,
      "step": 177650
    },
    {
      "epoch": 50.428611978427476,
      "grad_norm": 0.5716798901557922,
      "learning_rate": 4.959267669599773e-05,
      "loss": 0.000827709399163723,
      "step": 177660
    },
    {
      "epoch": 50.43145046835084,
      "grad_norm": 0.10842598974704742,
      "learning_rate": 4.958983820607437e-05,
      "loss": 0.00046628396958112715,
      "step": 177670
    },
    {
      "epoch": 50.434288958274195,
      "grad_norm": 0.09555017203092575,
      "learning_rate": 4.958699971615101e-05,
      "loss": 0.003782496228814125,
      "step": 177680
    },
    {
      "epoch": 50.43712744819756,
      "grad_norm": 0.4245949983596802,
      "learning_rate": 4.958416122622765e-05,
      "loss": 0.0003491690382361412,
      "step": 177690
    },
    {
      "epoch": 50.43996593812092,
      "grad_norm": 0.32890450954437256,
      "learning_rate": 4.958132273630429e-05,
      "loss": 0.0003079770132899284,
      "step": 177700
    },
    {
      "epoch": 50.44280442804428,
      "grad_norm": 0.3969707489013672,
      "learning_rate": 4.957848424638093e-05,
      "loss": 0.0004984445869922638,
      "step": 177710
    },
    {
      "epoch": 50.44564291796764,
      "grad_norm": 0.04004652425646782,
      "learning_rate": 4.957564575645757e-05,
      "loss": 0.0008567538112401962,
      "step": 177720
    },
    {
      "epoch": 50.448481407891,
      "grad_norm": 0.058560408651828766,
      "learning_rate": 4.9572807266534204e-05,
      "loss": 0.003626786172389984,
      "step": 177730
    },
    {
      "epoch": 50.451319897814365,
      "grad_norm": 0.005180516745895147,
      "learning_rate": 4.9569968776610845e-05,
      "loss": 0.0005284024402499199,
      "step": 177740
    },
    {
      "epoch": 50.45415838773772,
      "grad_norm": 0.027004335075616837,
      "learning_rate": 4.956713028668749e-05,
      "loss": 0.004837240651249885,
      "step": 177750
    },
    {
      "epoch": 50.45699687766108,
      "grad_norm": 0.04382847994565964,
      "learning_rate": 4.956429179676412e-05,
      "loss": 0.0019002886489033698,
      "step": 177760
    },
    {
      "epoch": 50.459835367584446,
      "grad_norm": 4.791123390197754,
      "learning_rate": 4.956145330684076e-05,
      "loss": 0.0010612225160002708,
      "step": 177770
    },
    {
      "epoch": 50.46267385750781,
      "grad_norm": 0.4115423262119293,
      "learning_rate": 4.9558614816917404e-05,
      "loss": 0.001125345379114151,
      "step": 177780
    },
    {
      "epoch": 50.465512347431165,
      "grad_norm": 0.16639269888401031,
      "learning_rate": 4.955577632699404e-05,
      "loss": 0.0008646093308925628,
      "step": 177790
    },
    {
      "epoch": 50.46835083735453,
      "grad_norm": 0.32309722900390625,
      "learning_rate": 4.955293783707068e-05,
      "loss": 0.0010424047708511353,
      "step": 177800
    },
    {
      "epoch": 50.47118932727789,
      "grad_norm": 0.03370222821831703,
      "learning_rate": 4.955009934714732e-05,
      "loss": 0.001813761331140995,
      "step": 177810
    },
    {
      "epoch": 50.47402781720125,
      "grad_norm": 0.12336213886737823,
      "learning_rate": 4.9547260857223956e-05,
      "loss": 0.000892116129398346,
      "step": 177820
    },
    {
      "epoch": 50.47686630712461,
      "grad_norm": 2.7869176864624023,
      "learning_rate": 4.95444223673006e-05,
      "loss": 0.0014268592000007629,
      "step": 177830
    },
    {
      "epoch": 50.47970479704797,
      "grad_norm": 0.17018850147724152,
      "learning_rate": 4.954158387737724e-05,
      "loss": 0.0004818037152290344,
      "step": 177840
    },
    {
      "epoch": 50.48254328697133,
      "grad_norm": 0.029928898438811302,
      "learning_rate": 4.953874538745388e-05,
      "loss": 0.004192335158586502,
      "step": 177850
    },
    {
      "epoch": 50.48538177689469,
      "grad_norm": 0.12719881534576416,
      "learning_rate": 4.9535906897530515e-05,
      "loss": 0.0013388942927122117,
      "step": 177860
    },
    {
      "epoch": 50.488220266818054,
      "grad_norm": 0.2800428569316864,
      "learning_rate": 4.953306840760715e-05,
      "loss": 0.0008534926921129227,
      "step": 177870
    },
    {
      "epoch": 50.49105875674142,
      "grad_norm": 2.415271282196045,
      "learning_rate": 4.95302299176838e-05,
      "loss": 0.016553825139999388,
      "step": 177880
    },
    {
      "epoch": 50.49389724666477,
      "grad_norm": 0.05258976295590401,
      "learning_rate": 4.952739142776043e-05,
      "loss": 0.00021453816443681717,
      "step": 177890
    },
    {
      "epoch": 50.496735736588136,
      "grad_norm": 0.007284042425453663,
      "learning_rate": 4.9524552937837074e-05,
      "loss": 0.005002042278647423,
      "step": 177900
    },
    {
      "epoch": 50.4995742265115,
      "grad_norm": 0.09343832731246948,
      "learning_rate": 4.9521714447913715e-05,
      "loss": 0.0035232074558734896,
      "step": 177910
    },
    {
      "epoch": 50.502412716434854,
      "grad_norm": 0.05552128329873085,
      "learning_rate": 4.951887595799035e-05,
      "loss": 0.0007843418046832084,
      "step": 177920
    },
    {
      "epoch": 50.50525120635822,
      "grad_norm": 0.9645230770111084,
      "learning_rate": 4.951603746806699e-05,
      "loss": 0.0014153802767395973,
      "step": 177930
    },
    {
      "epoch": 50.50808969628158,
      "grad_norm": 0.0924747884273529,
      "learning_rate": 4.951319897814363e-05,
      "loss": 0.0008359776809811592,
      "step": 177940
    },
    {
      "epoch": 50.510928186204936,
      "grad_norm": 0.07320868968963623,
      "learning_rate": 4.951036048822027e-05,
      "loss": 0.0016082949936389924,
      "step": 177950
    },
    {
      "epoch": 50.5137666761283,
      "grad_norm": 1.2412312030792236,
      "learning_rate": 4.950752199829691e-05,
      "loss": 0.0036428429186344145,
      "step": 177960
    },
    {
      "epoch": 50.51660516605166,
      "grad_norm": 0.0991537794470787,
      "learning_rate": 4.950468350837354e-05,
      "loss": 0.0006276082247495651,
      "step": 177970
    },
    {
      "epoch": 50.519443655975024,
      "grad_norm": 8.58150863647461,
      "learning_rate": 4.950184501845019e-05,
      "loss": 0.00971858948469162,
      "step": 177980
    },
    {
      "epoch": 50.52228214589838,
      "grad_norm": 0.22930099070072174,
      "learning_rate": 4.9499006528526826e-05,
      "loss": 0.000516694039106369,
      "step": 177990
    },
    {
      "epoch": 50.52512063582174,
      "grad_norm": 0.22646944224834442,
      "learning_rate": 4.949616803860346e-05,
      "loss": 0.0010945620015263557,
      "step": 178000
    },
    {
      "epoch": 50.52512063582174,
      "eval_accuracy": 0.9749475424429326,
      "eval_loss": 0.08912072330713272,
      "eval_runtime": 33.1405,
      "eval_samples_per_second": 474.555,
      "eval_steps_per_second": 7.423,
      "step": 178000
    },
    {
      "epoch": 50.527959125745106,
      "grad_norm": 1.298165202140808,
      "learning_rate": 4.949332954868011e-05,
      "loss": 0.0007663747295737266,
      "step": 178010
    },
    {
      "epoch": 50.53079761566846,
      "grad_norm": 0.33097919821739197,
      "learning_rate": 4.949049105875674e-05,
      "loss": 0.0020462434738874437,
      "step": 178020
    },
    {
      "epoch": 50.533636105591825,
      "grad_norm": 0.4576626420021057,
      "learning_rate": 4.9487652568833384e-05,
      "loss": 0.003384851664304733,
      "step": 178030
    },
    {
      "epoch": 50.53647459551519,
      "grad_norm": 0.1948893964290619,
      "learning_rate": 4.9484814078910026e-05,
      "loss": 0.002604338712990284,
      "step": 178040
    },
    {
      "epoch": 50.53931308543854,
      "grad_norm": 3.9826879501342773,
      "learning_rate": 4.948197558898666e-05,
      "loss": 0.0019319184124469758,
      "step": 178050
    },
    {
      "epoch": 50.542151575361906,
      "grad_norm": 0.1248917356133461,
      "learning_rate": 4.94791370990633e-05,
      "loss": 0.007322826236486435,
      "step": 178060
    },
    {
      "epoch": 50.54499006528527,
      "grad_norm": 0.3652973771095276,
      "learning_rate": 4.9476298609139936e-05,
      "loss": 0.010866234451532364,
      "step": 178070
    },
    {
      "epoch": 50.54782855520863,
      "grad_norm": 0.18217891454696655,
      "learning_rate": 4.947346011921658e-05,
      "loss": 0.0010630324482917785,
      "step": 178080
    },
    {
      "epoch": 50.55066704513199,
      "grad_norm": 0.15132637321949005,
      "learning_rate": 4.947062162929322e-05,
      "loss": 0.001171666383743286,
      "step": 178090
    },
    {
      "epoch": 50.55350553505535,
      "grad_norm": 5.3725152015686035,
      "learning_rate": 4.9467783139369854e-05,
      "loss": 0.017546015977859496,
      "step": 178100
    },
    {
      "epoch": 50.55634402497871,
      "grad_norm": 0.033957455307245255,
      "learning_rate": 4.94649446494465e-05,
      "loss": 0.0017675373703241349,
      "step": 178110
    },
    {
      "epoch": 50.55918251490207,
      "grad_norm": 1.643140435218811,
      "learning_rate": 4.9462106159523136e-05,
      "loss": 0.0020062139257788656,
      "step": 178120
    },
    {
      "epoch": 50.56202100482543,
      "grad_norm": 0.14635750651359558,
      "learning_rate": 4.945926766959977e-05,
      "loss": 0.0037103869020938874,
      "step": 178130
    },
    {
      "epoch": 50.564859494748795,
      "grad_norm": 0.24740934371948242,
      "learning_rate": 4.945642917967642e-05,
      "loss": 0.004420775175094605,
      "step": 178140
    },
    {
      "epoch": 50.56769798467215,
      "grad_norm": 0.18603527545928955,
      "learning_rate": 4.9453590689753054e-05,
      "loss": 0.008466163277626037,
      "step": 178150
    },
    {
      "epoch": 50.570536474595514,
      "grad_norm": 14.504068374633789,
      "learning_rate": 4.9450752199829695e-05,
      "loss": 0.007627817243337632,
      "step": 178160
    },
    {
      "epoch": 50.57337496451888,
      "grad_norm": 0.22988714277744293,
      "learning_rate": 4.944791370990633e-05,
      "loss": 0.0030602091923356057,
      "step": 178170
    },
    {
      "epoch": 50.57621345444224,
      "grad_norm": 0.24587509036064148,
      "learning_rate": 4.944507521998297e-05,
      "loss": 0.0007264429703354835,
      "step": 178180
    },
    {
      "epoch": 50.579051944365595,
      "grad_norm": 0.4983689486980438,
      "learning_rate": 4.944223673005961e-05,
      "loss": 0.0042855173349380495,
      "step": 178190
    },
    {
      "epoch": 50.58189043428896,
      "grad_norm": 0.05465778708457947,
      "learning_rate": 4.943939824013625e-05,
      "loss": 0.0018220921978354455,
      "step": 178200
    },
    {
      "epoch": 50.58472892421232,
      "grad_norm": 0.20462851226329803,
      "learning_rate": 4.943655975021289e-05,
      "loss": 0.0008056052029132843,
      "step": 178210
    },
    {
      "epoch": 50.58756741413568,
      "grad_norm": 0.1778930425643921,
      "learning_rate": 4.943372126028953e-05,
      "loss": 0.0003673093393445015,
      "step": 178220
    },
    {
      "epoch": 50.59040590405904,
      "grad_norm": 0.22313767671585083,
      "learning_rate": 4.9430882770366164e-05,
      "loss": 0.0012095678597688676,
      "step": 178230
    },
    {
      "epoch": 50.5932443939824,
      "grad_norm": 2.6927945613861084,
      "learning_rate": 4.9428044280442806e-05,
      "loss": 0.006263354420661926,
      "step": 178240
    },
    {
      "epoch": 50.596082883905765,
      "grad_norm": 0.06281709671020508,
      "learning_rate": 4.942520579051945e-05,
      "loss": 0.0009450465440750122,
      "step": 178250
    },
    {
      "epoch": 50.59892137382912,
      "grad_norm": 9.54774284362793,
      "learning_rate": 4.942236730059608e-05,
      "loss": 0.0029483195394277573,
      "step": 178260
    },
    {
      "epoch": 50.601759863752484,
      "grad_norm": 17.819562911987305,
      "learning_rate": 4.941952881067273e-05,
      "loss": 0.012881028652191161,
      "step": 178270
    },
    {
      "epoch": 50.60459835367585,
      "grad_norm": 7.835615634918213,
      "learning_rate": 4.9416690320749364e-05,
      "loss": 0.016581083834171294,
      "step": 178280
    },
    {
      "epoch": 50.6074368435992,
      "grad_norm": 6.273597240447998,
      "learning_rate": 4.9413851830826e-05,
      "loss": 0.012644153833389283,
      "step": 178290
    },
    {
      "epoch": 50.610275333522566,
      "grad_norm": 4.978204250335693,
      "learning_rate": 4.941101334090264e-05,
      "loss": 0.005877928435802459,
      "step": 178300
    },
    {
      "epoch": 50.61311382344593,
      "grad_norm": 0.323255717754364,
      "learning_rate": 4.940817485097928e-05,
      "loss": 0.001150345429778099,
      "step": 178310
    },
    {
      "epoch": 50.615952313369284,
      "grad_norm": 6.733771800994873,
      "learning_rate": 4.940533636105592e-05,
      "loss": 0.0015473274514079093,
      "step": 178320
    },
    {
      "epoch": 50.61879080329265,
      "grad_norm": 0.3099766671657562,
      "learning_rate": 4.940249787113256e-05,
      "loss": 0.0008110819384455681,
      "step": 178330
    },
    {
      "epoch": 50.62162929321601,
      "grad_norm": 0.06060781702399254,
      "learning_rate": 4.93996593812092e-05,
      "loss": 0.003547689691185951,
      "step": 178340
    },
    {
      "epoch": 50.62446778313937,
      "grad_norm": 0.07607654482126236,
      "learning_rate": 4.939682089128584e-05,
      "loss": 0.0015209093689918518,
      "step": 178350
    },
    {
      "epoch": 50.62730627306273,
      "grad_norm": 0.299899160861969,
      "learning_rate": 4.9393982401362475e-05,
      "loss": 0.0015377139672636986,
      "step": 178360
    },
    {
      "epoch": 50.63014476298609,
      "grad_norm": 0.05727801471948624,
      "learning_rate": 4.9391143911439116e-05,
      "loss": 0.0020029693841934204,
      "step": 178370
    },
    {
      "epoch": 50.632983252909455,
      "grad_norm": 0.11769679188728333,
      "learning_rate": 4.938830542151576e-05,
      "loss": 0.003923571109771729,
      "step": 178380
    },
    {
      "epoch": 50.63582174283281,
      "grad_norm": 1.423798680305481,
      "learning_rate": 4.938546693159239e-05,
      "loss": 0.0036637675017118453,
      "step": 178390
    },
    {
      "epoch": 50.63866023275617,
      "grad_norm": 0.17247925698757172,
      "learning_rate": 4.9382628441669034e-05,
      "loss": 0.0014694530516862869,
      "step": 178400
    },
    {
      "epoch": 50.641498722679536,
      "grad_norm": 1.3631196022033691,
      "learning_rate": 4.9379789951745675e-05,
      "loss": 0.0015047889202833176,
      "step": 178410
    },
    {
      "epoch": 50.64433721260289,
      "grad_norm": 1.2381564378738403,
      "learning_rate": 4.937695146182231e-05,
      "loss": 0.008275463432073592,
      "step": 178420
    },
    {
      "epoch": 50.647175702526255,
      "grad_norm": 0.1554345041513443,
      "learning_rate": 4.937411297189895e-05,
      "loss": 0.011019628494977951,
      "step": 178430
    },
    {
      "epoch": 50.65001419244962,
      "grad_norm": 5.878764629364014,
      "learning_rate": 4.937127448197559e-05,
      "loss": 0.0020541947335004805,
      "step": 178440
    },
    {
      "epoch": 50.65285268237298,
      "grad_norm": 0.12758630514144897,
      "learning_rate": 4.9368435992052234e-05,
      "loss": 0.017905624210834505,
      "step": 178450
    },
    {
      "epoch": 50.655691172296336,
      "grad_norm": 0.11042506247758865,
      "learning_rate": 4.936559750212887e-05,
      "loss": 0.0018763771280646325,
      "step": 178460
    },
    {
      "epoch": 50.6585296622197,
      "grad_norm": 0.4924071133136749,
      "learning_rate": 4.936275901220551e-05,
      "loss": 0.001947961002588272,
      "step": 178470
    },
    {
      "epoch": 50.66136815214306,
      "grad_norm": 0.06848179548978806,
      "learning_rate": 4.935992052228215e-05,
      "loss": 0.0004623718559741974,
      "step": 178480
    },
    {
      "epoch": 50.66420664206642,
      "grad_norm": 0.05665064975619316,
      "learning_rate": 4.9357082032358786e-05,
      "loss": 0.000892830267548561,
      "step": 178490
    },
    {
      "epoch": 50.66704513198978,
      "grad_norm": 0.16801804304122925,
      "learning_rate": 4.935424354243543e-05,
      "loss": 0.0005209440365433693,
      "step": 178500
    },
    {
      "epoch": 50.66704513198978,
      "eval_accuracy": 0.97475678768996,
      "eval_loss": 0.08886351436376572,
      "eval_runtime": 32.6215,
      "eval_samples_per_second": 482.105,
      "eval_steps_per_second": 7.541,
      "step": 178500
    },
    {
      "epoch": 50.669883621913144,
      "grad_norm": 1.6069427728652954,
      "learning_rate": 4.935140505251207e-05,
      "loss": 0.0026915842667222024,
      "step": 178510
    },
    {
      "epoch": 50.67272211183651,
      "grad_norm": 0.15120673179626465,
      "learning_rate": 4.93485665625887e-05,
      "loss": 0.0005074212327599526,
      "step": 178520
    },
    {
      "epoch": 50.67556060175986,
      "grad_norm": 0.3359042704105377,
      "learning_rate": 4.9345728072665345e-05,
      "loss": 0.0019681137055158615,
      "step": 178530
    },
    {
      "epoch": 50.678399091683225,
      "grad_norm": 0.024479566141963005,
      "learning_rate": 4.9342889582741986e-05,
      "loss": 0.0022592568770051003,
      "step": 178540
    },
    {
      "epoch": 50.68123758160659,
      "grad_norm": 0.21077264845371246,
      "learning_rate": 4.934005109281862e-05,
      "loss": 0.002022714912891388,
      "step": 178550
    },
    {
      "epoch": 50.684076071529944,
      "grad_norm": 0.09945801645517349,
      "learning_rate": 4.933721260289526e-05,
      "loss": 0.0009286802262067795,
      "step": 178560
    },
    {
      "epoch": 50.68691456145331,
      "grad_norm": 0.8508023023605347,
      "learning_rate": 4.93343741129719e-05,
      "loss": 0.002763545513153076,
      "step": 178570
    },
    {
      "epoch": 50.68975305137667,
      "grad_norm": 0.2934623956680298,
      "learning_rate": 4.9331535623048545e-05,
      "loss": 0.00554654449224472,
      "step": 178580
    },
    {
      "epoch": 50.692591541300025,
      "grad_norm": 0.31937262415885925,
      "learning_rate": 4.932869713312518e-05,
      "loss": 0.003822198137640953,
      "step": 178590
    },
    {
      "epoch": 50.69543003122339,
      "grad_norm": 0.1112031638622284,
      "learning_rate": 4.9325858643201814e-05,
      "loss": 0.0033473484218120575,
      "step": 178600
    },
    {
      "epoch": 50.69826852114675,
      "grad_norm": 0.4033728837966919,
      "learning_rate": 4.932302015327846e-05,
      "loss": 0.000270291231572628,
      "step": 178610
    },
    {
      "epoch": 50.701107011070114,
      "grad_norm": 0.2086223065853119,
      "learning_rate": 4.93201816633551e-05,
      "loss": 0.0006224347278475761,
      "step": 178620
    },
    {
      "epoch": 50.70394550099347,
      "grad_norm": 2.71340012550354,
      "learning_rate": 4.931734317343174e-05,
      "loss": 0.001313019171357155,
      "step": 178630
    },
    {
      "epoch": 50.70678399091683,
      "grad_norm": 0.42862969636917114,
      "learning_rate": 4.931450468350838e-05,
      "loss": 0.0010548047721385956,
      "step": 178640
    },
    {
      "epoch": 50.709622480840196,
      "grad_norm": 0.14810322225093842,
      "learning_rate": 4.9311666193585014e-05,
      "loss": 0.001307407021522522,
      "step": 178650
    },
    {
      "epoch": 50.71246097076355,
      "grad_norm": 0.08268176019191742,
      "learning_rate": 4.9308827703661655e-05,
      "loss": 0.0032232481986284255,
      "step": 178660
    },
    {
      "epoch": 50.715299460686914,
      "grad_norm": 0.021411698311567307,
      "learning_rate": 4.93059892137383e-05,
      "loss": 0.0007195794954895973,
      "step": 178670
    },
    {
      "epoch": 50.71813795061028,
      "grad_norm": 6.673913955688477,
      "learning_rate": 4.930315072381493e-05,
      "loss": 0.0022778794169425963,
      "step": 178680
    },
    {
      "epoch": 50.72097644053363,
      "grad_norm": 0.48466309905052185,
      "learning_rate": 4.930031223389157e-05,
      "loss": 0.0021695947274565698,
      "step": 178690
    },
    {
      "epoch": 50.723814930456996,
      "grad_norm": 3.3368430137634277,
      "learning_rate": 4.929747374396821e-05,
      "loss": 0.0017875708639621735,
      "step": 178700
    },
    {
      "epoch": 50.72665342038036,
      "grad_norm": 0.0369260348379612,
      "learning_rate": 4.929463525404485e-05,
      "loss": 0.005566006153821945,
      "step": 178710
    },
    {
      "epoch": 50.72949191030372,
      "grad_norm": 4.205667972564697,
      "learning_rate": 4.929179676412149e-05,
      "loss": 0.0022798661142587663,
      "step": 178720
    },
    {
      "epoch": 50.73233040022708,
      "grad_norm": 0.022319070994853973,
      "learning_rate": 4.9288958274198125e-05,
      "loss": 0.0007448522374033928,
      "step": 178730
    },
    {
      "epoch": 50.73516889015044,
      "grad_norm": 0.010256805457174778,
      "learning_rate": 4.928611978427477e-05,
      "loss": 0.00022689811885356903,
      "step": 178740
    },
    {
      "epoch": 50.7380073800738,
      "grad_norm": 0.2731415927410126,
      "learning_rate": 4.928328129435141e-05,
      "loss": 0.005032230168581009,
      "step": 178750
    },
    {
      "epoch": 50.74084586999716,
      "grad_norm": 0.6842676997184753,
      "learning_rate": 4.928044280442804e-05,
      "loss": 0.004304073750972748,
      "step": 178760
    },
    {
      "epoch": 50.74368435992052,
      "grad_norm": 0.5024015307426453,
      "learning_rate": 4.927760431450469e-05,
      "loss": 0.0008814016357064248,
      "step": 178770
    },
    {
      "epoch": 50.746522849843885,
      "grad_norm": 2.7455503940582275,
      "learning_rate": 4.9274765824581325e-05,
      "loss": 0.0012464871630072593,
      "step": 178780
    },
    {
      "epoch": 50.74936133976724,
      "grad_norm": 5.226221084594727,
      "learning_rate": 4.9271927334657966e-05,
      "loss": 0.001848204992711544,
      "step": 178790
    },
    {
      "epoch": 50.7521998296906,
      "grad_norm": 0.10118193924427032,
      "learning_rate": 4.92690888447346e-05,
      "loss": 0.001522783562541008,
      "step": 178800
    },
    {
      "epoch": 50.755038319613966,
      "grad_norm": 2.0859222412109375,
      "learning_rate": 4.926625035481124e-05,
      "loss": 0.004864646494388581,
      "step": 178810
    },
    {
      "epoch": 50.75787680953733,
      "grad_norm": 0.3066481947898865,
      "learning_rate": 4.9263411864887883e-05,
      "loss": 0.0010640375316143037,
      "step": 178820
    },
    {
      "epoch": 50.760715299460685,
      "grad_norm": 0.09536521136760712,
      "learning_rate": 4.926057337496452e-05,
      "loss": 0.010853983461856842,
      "step": 178830
    },
    {
      "epoch": 50.76355378938405,
      "grad_norm": 0.32673078775405884,
      "learning_rate": 4.925773488504116e-05,
      "loss": 0.0016717568039894104,
      "step": 178840
    },
    {
      "epoch": 50.76639227930741,
      "grad_norm": 0.192287415266037,
      "learning_rate": 4.92548963951178e-05,
      "loss": 0.0014872996136546135,
      "step": 178850
    },
    {
      "epoch": 50.76923076923077,
      "grad_norm": 0.2720217704772949,
      "learning_rate": 4.9252057905194435e-05,
      "loss": 0.0011731542646884919,
      "step": 178860
    },
    {
      "epoch": 50.77206925915413,
      "grad_norm": 0.7712118625640869,
      "learning_rate": 4.9249219415271084e-05,
      "loss": 0.0014049982652068137,
      "step": 178870
    },
    {
      "epoch": 50.77490774907749,
      "grad_norm": 0.01615619845688343,
      "learning_rate": 4.924638092534772e-05,
      "loss": 0.004363653063774109,
      "step": 178880
    },
    {
      "epoch": 50.77774623900085,
      "grad_norm": 3.0605239868164062,
      "learning_rate": 4.924354243542435e-05,
      "loss": 0.003465009480714798,
      "step": 178890
    },
    {
      "epoch": 50.78058472892421,
      "grad_norm": 0.038306254893541336,
      "learning_rate": 4.9240703945500994e-05,
      "loss": 0.0007870623841881752,
      "step": 178900
    },
    {
      "epoch": 50.783423218847574,
      "grad_norm": 0.19215263426303864,
      "learning_rate": 4.9237865455577636e-05,
      "loss": 0.0005981309339404106,
      "step": 178910
    },
    {
      "epoch": 50.78626170877094,
      "grad_norm": 0.06646416336297989,
      "learning_rate": 4.923502696565428e-05,
      "loss": 0.000516049936413765,
      "step": 178920
    },
    {
      "epoch": 50.78910019869429,
      "grad_norm": 1.3104448318481445,
      "learning_rate": 4.923218847573091e-05,
      "loss": 0.0004753680899739265,
      "step": 178930
    },
    {
      "epoch": 50.791938688617655,
      "grad_norm": 1.341243028640747,
      "learning_rate": 4.922934998580755e-05,
      "loss": 0.003874048590660095,
      "step": 178940
    },
    {
      "epoch": 50.79477717854102,
      "grad_norm": 0.6288641095161438,
      "learning_rate": 4.9226511495884194e-05,
      "loss": 0.006040126830339432,
      "step": 178950
    },
    {
      "epoch": 50.797615668464374,
      "grad_norm": 2.2384088039398193,
      "learning_rate": 4.922367300596083e-05,
      "loss": 0.0010082099586725235,
      "step": 178960
    },
    {
      "epoch": 50.80045415838774,
      "grad_norm": 1.2499902248382568,
      "learning_rate": 4.922083451603747e-05,
      "loss": 0.005352294072508812,
      "step": 178970
    },
    {
      "epoch": 50.8032926483111,
      "grad_norm": 0.16821342706680298,
      "learning_rate": 4.921799602611411e-05,
      "loss": 0.0006681712344288826,
      "step": 178980
    },
    {
      "epoch": 50.80613113823446,
      "grad_norm": 0.017941802740097046,
      "learning_rate": 4.9215157536190746e-05,
      "loss": 0.003770666569471359,
      "step": 178990
    },
    {
      "epoch": 50.80896962815782,
      "grad_norm": 0.05893326550722122,
      "learning_rate": 4.921231904626739e-05,
      "loss": 0.0016591912135481834,
      "step": 179000
    },
    {
      "epoch": 50.80896962815782,
      "eval_accuracy": 0.9757741463724805,
      "eval_loss": 0.09130814671516418,
      "eval_runtime": 32.5125,
      "eval_samples_per_second": 483.721,
      "eval_steps_per_second": 7.566,
      "step": 179000
    },
    {
      "epoch": 50.81180811808118,
      "grad_norm": 0.05623128265142441,
      "learning_rate": 4.920948055634403e-05,
      "loss": 0.002649303339421749,
      "step": 179010
    },
    {
      "epoch": 50.814646608004544,
      "grad_norm": 22.51488494873047,
      "learning_rate": 4.9206642066420664e-05,
      "loss": 0.013958887755870819,
      "step": 179020
    },
    {
      "epoch": 50.8174850979279,
      "grad_norm": 0.018028760328888893,
      "learning_rate": 4.9203803576497305e-05,
      "loss": 0.0006909333169460297,
      "step": 179030
    },
    {
      "epoch": 50.82032358785126,
      "grad_norm": 0.049799077212810516,
      "learning_rate": 4.9200965086573946e-05,
      "loss": 0.0005367718636989594,
      "step": 179040
    },
    {
      "epoch": 50.823162077774626,
      "grad_norm": 0.024345967918634415,
      "learning_rate": 4.919812659665059e-05,
      "loss": 0.0006771998479962349,
      "step": 179050
    },
    {
      "epoch": 50.82600056769798,
      "grad_norm": 0.11605974286794662,
      "learning_rate": 4.919528810672722e-05,
      "loss": 0.0007713280618190765,
      "step": 179060
    },
    {
      "epoch": 50.828839057621344,
      "grad_norm": 0.29047030210494995,
      "learning_rate": 4.9192449616803864e-05,
      "loss": 0.000330309197306633,
      "step": 179070
    },
    {
      "epoch": 50.83167754754471,
      "grad_norm": 0.19378119707107544,
      "learning_rate": 4.9189611126880505e-05,
      "loss": 0.0008186059072613716,
      "step": 179080
    },
    {
      "epoch": 50.83451603746807,
      "grad_norm": 0.19748885929584503,
      "learning_rate": 4.918677263695714e-05,
      "loss": 0.0006861373782157898,
      "step": 179090
    },
    {
      "epoch": 50.837354527391426,
      "grad_norm": 0.3749293386936188,
      "learning_rate": 4.918393414703378e-05,
      "loss": 0.0015106894075870513,
      "step": 179100
    },
    {
      "epoch": 50.84019301731479,
      "grad_norm": 0.18366852402687073,
      "learning_rate": 4.918109565711042e-05,
      "loss": 0.0010784173384308815,
      "step": 179110
    },
    {
      "epoch": 50.84303150723815,
      "grad_norm": 0.18460649251937866,
      "learning_rate": 4.917825716718706e-05,
      "loss": 0.0017572684213519097,
      "step": 179120
    },
    {
      "epoch": 50.84586999716151,
      "grad_norm": 0.23588702082633972,
      "learning_rate": 4.91754186772637e-05,
      "loss": 0.0022995347157120706,
      "step": 179130
    },
    {
      "epoch": 50.84870848708487,
      "grad_norm": 0.24003693461418152,
      "learning_rate": 4.917258018734034e-05,
      "loss": 0.005397206544876099,
      "step": 179140
    },
    {
      "epoch": 50.85154697700823,
      "grad_norm": 0.7501184344291687,
      "learning_rate": 4.9169741697416974e-05,
      "loss": 0.006804966181516647,
      "step": 179150
    },
    {
      "epoch": 50.85438546693159,
      "grad_norm": 0.07630418986082077,
      "learning_rate": 4.9166903207493616e-05,
      "loss": 0.0011077145114541053,
      "step": 179160
    },
    {
      "epoch": 50.85722395685495,
      "grad_norm": 2.4324944019317627,
      "learning_rate": 4.916406471757026e-05,
      "loss": 0.003729851543903351,
      "step": 179170
    },
    {
      "epoch": 50.860062446778315,
      "grad_norm": 0.37616318464279175,
      "learning_rate": 4.916122622764689e-05,
      "loss": 0.0009873485192656516,
      "step": 179180
    },
    {
      "epoch": 50.86290093670168,
      "grad_norm": 0.08013641089200974,
      "learning_rate": 4.915838773772353e-05,
      "loss": 0.0012565094977617264,
      "step": 179190
    },
    {
      "epoch": 50.865739426625034,
      "grad_norm": 3.9856531620025635,
      "learning_rate": 4.915554924780017e-05,
      "loss": 0.0021447960287332537,
      "step": 179200
    },
    {
      "epoch": 50.868577916548396,
      "grad_norm": 0.040843307971954346,
      "learning_rate": 4.9152710757876816e-05,
      "loss": 0.0030737029388546943,
      "step": 179210
    },
    {
      "epoch": 50.87141640647176,
      "grad_norm": 0.12800690531730652,
      "learning_rate": 4.914987226795345e-05,
      "loss": 0.000683315098285675,
      "step": 179220
    },
    {
      "epoch": 50.874254896395115,
      "grad_norm": 0.21204321086406708,
      "learning_rate": 4.9147033778030085e-05,
      "loss": 0.0013605091720819474,
      "step": 179230
    },
    {
      "epoch": 50.87709338631848,
      "grad_norm": 0.02820630557835102,
      "learning_rate": 4.914419528810673e-05,
      "loss": 0.0009277824312448502,
      "step": 179240
    },
    {
      "epoch": 50.87993187624184,
      "grad_norm": 0.06900962442159653,
      "learning_rate": 4.914135679818337e-05,
      "loss": 0.002055242471396923,
      "step": 179250
    },
    {
      "epoch": 50.8827703661652,
      "grad_norm": 0.1176263764500618,
      "learning_rate": 4.913851830826001e-05,
      "loss": 0.009094487130641937,
      "step": 179260
    },
    {
      "epoch": 50.88560885608856,
      "grad_norm": 0.05066773295402527,
      "learning_rate": 4.913567981833665e-05,
      "loss": 0.0075528621673583984,
      "step": 179270
    },
    {
      "epoch": 50.88844734601192,
      "grad_norm": 0.05010965093970299,
      "learning_rate": 4.9132841328413285e-05,
      "loss": 0.0011628611013293267,
      "step": 179280
    },
    {
      "epoch": 50.891285835935285,
      "grad_norm": 0.5996926426887512,
      "learning_rate": 4.9130002838489926e-05,
      "loss": 0.004261218011379242,
      "step": 179290
    },
    {
      "epoch": 50.89412432585864,
      "grad_norm": 0.3129023015499115,
      "learning_rate": 4.912716434856656e-05,
      "loss": 0.0010878697037696838,
      "step": 179300
    },
    {
      "epoch": 50.896962815782004,
      "grad_norm": 0.015143539756536484,
      "learning_rate": 4.91243258586432e-05,
      "loss": 0.000557640939950943,
      "step": 179310
    },
    {
      "epoch": 50.89980130570537,
      "grad_norm": 0.015455203130841255,
      "learning_rate": 4.9121487368719844e-05,
      "loss": 0.0013242602348327638,
      "step": 179320
    },
    {
      "epoch": 50.90263979562872,
      "grad_norm": 0.09173936396837234,
      "learning_rate": 4.911864887879648e-05,
      "loss": 0.0007935767993330955,
      "step": 179330
    },
    {
      "epoch": 50.905478285552086,
      "grad_norm": 1.9988716840744019,
      "learning_rate": 4.9115810388873127e-05,
      "loss": 0.0011340508237481118,
      "step": 179340
    },
    {
      "epoch": 50.90831677547545,
      "grad_norm": 0.16947871446609497,
      "learning_rate": 4.911297189894976e-05,
      "loss": 0.002062997780740261,
      "step": 179350
    },
    {
      "epoch": 50.91115526539881,
      "grad_norm": 0.11406555771827698,
      "learning_rate": 4.9110133409026396e-05,
      "loss": 0.0010280873626470565,
      "step": 179360
    },
    {
      "epoch": 50.91399375532217,
      "grad_norm": 1.2644504308700562,
      "learning_rate": 4.9107294919103044e-05,
      "loss": 0.0014095757156610488,
      "step": 179370
    },
    {
      "epoch": 50.91683224524553,
      "grad_norm": 0.024860622361302376,
      "learning_rate": 4.910445642917968e-05,
      "loss": 0.0003766695037484169,
      "step": 179380
    },
    {
      "epoch": 50.91967073516889,
      "grad_norm": 8.560633659362793,
      "learning_rate": 4.910161793925632e-05,
      "loss": 0.0036016374826431273,
      "step": 179390
    },
    {
      "epoch": 50.92250922509225,
      "grad_norm": 0.04153590276837349,
      "learning_rate": 4.9098779449332954e-05,
      "loss": 0.0003776244819164276,
      "step": 179400
    },
    {
      "epoch": 50.92534771501561,
      "grad_norm": 1.5761879682540894,
      "learning_rate": 4.9095940959409596e-05,
      "loss": 0.0016282957047224044,
      "step": 179410
    },
    {
      "epoch": 50.928186204938974,
      "grad_norm": 0.5046162009239197,
      "learning_rate": 4.909310246948624e-05,
      "loss": 0.0026831995695829392,
      "step": 179420
    },
    {
      "epoch": 50.93102469486233,
      "grad_norm": 24.203166961669922,
      "learning_rate": 4.909026397956287e-05,
      "loss": 0.01174393817782402,
      "step": 179430
    },
    {
      "epoch": 50.93386318478569,
      "grad_norm": 0.7535839080810547,
      "learning_rate": 4.908742548963951e-05,
      "loss": 0.0023954806849360464,
      "step": 179440
    },
    {
      "epoch": 50.936701674709056,
      "grad_norm": 0.2901063859462738,
      "learning_rate": 4.9084586999716155e-05,
      "loss": 0.0008005311712622643,
      "step": 179450
    },
    {
      "epoch": 50.93954016463242,
      "grad_norm": 0.23094689846038818,
      "learning_rate": 4.908174850979279e-05,
      "loss": 0.001790887862443924,
      "step": 179460
    },
    {
      "epoch": 50.942378654555775,
      "grad_norm": 0.055532876402139664,
      "learning_rate": 4.907891001986944e-05,
      "loss": 0.0006162568926811219,
      "step": 179470
    },
    {
      "epoch": 50.94521714447914,
      "grad_norm": 4.2882981300354,
      "learning_rate": 4.907607152994607e-05,
      "loss": 0.0014024326577782632,
      "step": 179480
    },
    {
      "epoch": 50.9480556344025,
      "grad_norm": 0.021220499649643898,
      "learning_rate": 4.9073233040022707e-05,
      "loss": 0.004764564335346222,
      "step": 179490
    },
    {
      "epoch": 50.950894124325856,
      "grad_norm": 1.666385531425476,
      "learning_rate": 4.9070394550099355e-05,
      "loss": 0.0011418627575039864,
      "step": 179500
    },
    {
      "epoch": 50.950894124325856,
      "eval_accuracy": 0.9743752781840147,
      "eval_loss": 0.09286370128393173,
      "eval_runtime": 32.4708,
      "eval_samples_per_second": 484.344,
      "eval_steps_per_second": 7.576,
      "step": 179500
    },
    {
      "epoch": 50.95373261424922,
      "grad_norm": 0.16561318933963776,
      "learning_rate": 4.906755606017599e-05,
      "loss": 0.004696987569332123,
      "step": 179510
    },
    {
      "epoch": 50.95657110417258,
      "grad_norm": 0.07594060897827148,
      "learning_rate": 4.906471757025263e-05,
      "loss": 0.0009161263704299926,
      "step": 179520
    },
    {
      "epoch": 50.95940959409594,
      "grad_norm": 0.021128594875335693,
      "learning_rate": 4.9061879080329265e-05,
      "loss": 0.0011955585330724716,
      "step": 179530
    },
    {
      "epoch": 50.9622480840193,
      "grad_norm": 0.2908976376056671,
      "learning_rate": 4.905904059040591e-05,
      "loss": 0.0007049527019262313,
      "step": 179540
    },
    {
      "epoch": 50.96508657394266,
      "grad_norm": 9.747366905212402,
      "learning_rate": 4.905620210048255e-05,
      "loss": 0.0020374804735183716,
      "step": 179550
    },
    {
      "epoch": 50.967925063866026,
      "grad_norm": 0.05633821710944176,
      "learning_rate": 4.905336361055918e-05,
      "loss": 0.0004924533888697625,
      "step": 179560
    },
    {
      "epoch": 50.97076355378938,
      "grad_norm": 0.9094576239585876,
      "learning_rate": 4.9050525120635824e-05,
      "loss": 0.0017846930772066117,
      "step": 179570
    },
    {
      "epoch": 50.973602043712745,
      "grad_norm": 0.9898324012756348,
      "learning_rate": 4.9047686630712465e-05,
      "loss": 0.0005733927711844445,
      "step": 179580
    },
    {
      "epoch": 50.97644053363611,
      "grad_norm": 0.1377558708190918,
      "learning_rate": 4.90448481407891e-05,
      "loss": 0.0008543012663722038,
      "step": 179590
    },
    {
      "epoch": 50.979279023559464,
      "grad_norm": 0.30294978618621826,
      "learning_rate": 4.904200965086574e-05,
      "loss": 0.0017169589176774026,
      "step": 179600
    },
    {
      "epoch": 50.98211751348283,
      "grad_norm": 0.03887096047401428,
      "learning_rate": 4.903917116094238e-05,
      "loss": 0.0006159618496894836,
      "step": 179610
    },
    {
      "epoch": 50.98495600340619,
      "grad_norm": 0.07467129081487656,
      "learning_rate": 4.903633267101902e-05,
      "loss": 0.007407064735889435,
      "step": 179620
    },
    {
      "epoch": 50.987794493329545,
      "grad_norm": 0.026289066299796104,
      "learning_rate": 4.903349418109566e-05,
      "loss": 0.0006912363693118096,
      "step": 179630
    },
    {
      "epoch": 50.99063298325291,
      "grad_norm": 0.1056104227900505,
      "learning_rate": 4.90306556911723e-05,
      "loss": 0.00022765006870031356,
      "step": 179640
    },
    {
      "epoch": 50.99347147317627,
      "grad_norm": 0.04278640076518059,
      "learning_rate": 4.9027817201248935e-05,
      "loss": 0.0003650272265076637,
      "step": 179650
    },
    {
      "epoch": 50.996309963099634,
      "grad_norm": 0.4679018259048462,
      "learning_rate": 4.9024978711325576e-05,
      "loss": 0.0005566198378801346,
      "step": 179660
    },
    {
      "epoch": 50.99914845302299,
      "grad_norm": 0.020620033144950867,
      "learning_rate": 4.902214022140222e-05,
      "loss": 0.0005242584273219108,
      "step": 179670
    },
    {
      "epoch": 51.00198694294635,
      "grad_norm": 0.08303036540746689,
      "learning_rate": 4.901930173147886e-05,
      "loss": 0.0004657704383134842,
      "step": 179680
    },
    {
      "epoch": 51.004825432869715,
      "grad_norm": 0.15461541712284088,
      "learning_rate": 4.901646324155549e-05,
      "loss": 0.0004808560013771057,
      "step": 179690
    },
    {
      "epoch": 51.00766392279307,
      "grad_norm": 0.008365842513740063,
      "learning_rate": 4.9013624751632135e-05,
      "loss": 0.00023199021816253663,
      "step": 179700
    },
    {
      "epoch": 51.010502412716434,
      "grad_norm": 0.01753336377441883,
      "learning_rate": 4.9010786261708776e-05,
      "loss": 0.00024949200451374054,
      "step": 179710
    },
    {
      "epoch": 51.0133409026398,
      "grad_norm": 0.00863791722804308,
      "learning_rate": 4.900794777178541e-05,
      "loss": 0.00026589669287204745,
      "step": 179720
    },
    {
      "epoch": 51.01617939256316,
      "grad_norm": 0.2756747305393219,
      "learning_rate": 4.900510928186205e-05,
      "loss": 0.0007994692772626877,
      "step": 179730
    },
    {
      "epoch": 51.019017882486516,
      "grad_norm": 0.037193067371845245,
      "learning_rate": 4.9002270791938693e-05,
      "loss": 0.0008436944335699081,
      "step": 179740
    },
    {
      "epoch": 51.02185637240988,
      "grad_norm": 0.19737833738327026,
      "learning_rate": 4.899943230201533e-05,
      "loss": 0.0003716334700584412,
      "step": 179750
    },
    {
      "epoch": 51.02469486233324,
      "grad_norm": 2.1401031017303467,
      "learning_rate": 4.899659381209197e-05,
      "loss": 0.0008911542594432831,
      "step": 179760
    },
    {
      "epoch": 51.0275333522566,
      "grad_norm": 0.017984740436077118,
      "learning_rate": 4.899375532216861e-05,
      "loss": 0.0013545835390686989,
      "step": 179770
    },
    {
      "epoch": 51.03037184217996,
      "grad_norm": 0.011006012558937073,
      "learning_rate": 4.8990916832245245e-05,
      "loss": 0.0045109190046787265,
      "step": 179780
    },
    {
      "epoch": 51.03321033210332,
      "grad_norm": 0.1501239538192749,
      "learning_rate": 4.898807834232189e-05,
      "loss": 0.0006913190707564354,
      "step": 179790
    },
    {
      "epoch": 51.03604882202668,
      "grad_norm": 0.03708026558160782,
      "learning_rate": 4.898523985239853e-05,
      "loss": 0.0006171310320496559,
      "step": 179800
    },
    {
      "epoch": 51.03888731195004,
      "grad_norm": 0.717915415763855,
      "learning_rate": 4.898240136247517e-05,
      "loss": 0.00019870921969413757,
      "step": 179810
    },
    {
      "epoch": 51.041725801873405,
      "grad_norm": 0.0492561049759388,
      "learning_rate": 4.8979562872551804e-05,
      "loss": 0.0005968088284134865,
      "step": 179820
    },
    {
      "epoch": 51.04456429179677,
      "grad_norm": 2.8632595539093018,
      "learning_rate": 4.897672438262844e-05,
      "loss": 0.0010893121361732483,
      "step": 179830
    },
    {
      "epoch": 51.04740278172012,
      "grad_norm": 0.017252415418624878,
      "learning_rate": 4.897388589270509e-05,
      "loss": 0.00029211267828941343,
      "step": 179840
    },
    {
      "epoch": 51.050241271643486,
      "grad_norm": 0.15318366885185242,
      "learning_rate": 4.897104740278172e-05,
      "loss": 0.003840245306491852,
      "step": 179850
    },
    {
      "epoch": 51.05307976156685,
      "grad_norm": 0.00788168702274561,
      "learning_rate": 4.896820891285836e-05,
      "loss": 0.0019046250730752946,
      "step": 179860
    },
    {
      "epoch": 51.055918251490205,
      "grad_norm": 1.8242930173873901,
      "learning_rate": 4.8965370422935004e-05,
      "loss": 0.0009945519268512726,
      "step": 179870
    },
    {
      "epoch": 51.05875674141357,
      "grad_norm": 2.6354260444641113,
      "learning_rate": 4.896253193301164e-05,
      "loss": 0.0025006502866744993,
      "step": 179880
    },
    {
      "epoch": 51.06159523133693,
      "grad_norm": 0.11894006282091141,
      "learning_rate": 4.895969344308828e-05,
      "loss": 0.0011612219735980035,
      "step": 179890
    },
    {
      "epoch": 51.064433721260286,
      "grad_norm": 0.220514178276062,
      "learning_rate": 4.895685495316492e-05,
      "loss": 0.0004450822249054909,
      "step": 179900
    },
    {
      "epoch": 51.06727221118365,
      "grad_norm": 0.2767716944217682,
      "learning_rate": 4.8954016463241556e-05,
      "loss": 0.00031970683485269544,
      "step": 179910
    },
    {
      "epoch": 51.07011070110701,
      "grad_norm": 5.96195125579834,
      "learning_rate": 4.89511779733182e-05,
      "loss": 0.0029579455032944678,
      "step": 179920
    },
    {
      "epoch": 51.072949191030375,
      "grad_norm": 0.15815524756908417,
      "learning_rate": 4.894833948339483e-05,
      "loss": 0.0034724690020084383,
      "step": 179930
    },
    {
      "epoch": 51.07578768095373,
      "grad_norm": 0.5208958387374878,
      "learning_rate": 4.894550099347148e-05,
      "loss": 0.00039660818874835967,
      "step": 179940
    },
    {
      "epoch": 51.078626170877094,
      "grad_norm": 0.018688667565584183,
      "learning_rate": 4.8942662503548115e-05,
      "loss": 0.0009561549872159958,
      "step": 179950
    },
    {
      "epoch": 51.08146466080046,
      "grad_norm": 0.03799564018845558,
      "learning_rate": 4.893982401362475e-05,
      "loss": 0.0006402885541319847,
      "step": 179960
    },
    {
      "epoch": 51.08430315072381,
      "grad_norm": 6.493662357330322,
      "learning_rate": 4.89369855237014e-05,
      "loss": 0.0016361810266971588,
      "step": 179970
    },
    {
      "epoch": 51.087141640647175,
      "grad_norm": 18.471843719482422,
      "learning_rate": 4.893414703377803e-05,
      "loss": 0.0025501562282443046,
      "step": 179980
    },
    {
      "epoch": 51.08998013057054,
      "grad_norm": 0.53755784034729,
      "learning_rate": 4.8931308543854674e-05,
      "loss": 0.00204334557056427,
      "step": 179990
    },
    {
      "epoch": 51.092818620493894,
      "grad_norm": 0.2069019377231598,
      "learning_rate": 4.8928470053931315e-05,
      "loss": 0.007746119052171707,
      "step": 180000
    },
    {
      "epoch": 51.092818620493894,
      "eval_accuracy": 0.975837731290138,
      "eval_loss": 0.08971923589706421,
      "eval_runtime": 32.6144,
      "eval_samples_per_second": 482.211,
      "eval_steps_per_second": 7.543,
      "step": 180000
    },
    {
      "epoch": 51.09565711041726,
      "grad_norm": 0.38862478733062744,
      "learning_rate": 4.892563156400795e-05,
      "loss": 0.01819635331630707,
      "step": 180010
    },
    {
      "epoch": 51.09849560034062,
      "grad_norm": 0.39510875940322876,
      "learning_rate": 4.892279307408459e-05,
      "loss": 0.0018758226186037063,
      "step": 180020
    },
    {
      "epoch": 51.10133409026398,
      "grad_norm": 7.894891262054443,
      "learning_rate": 4.8919954584161226e-05,
      "loss": 0.0052969329059124,
      "step": 180030
    },
    {
      "epoch": 51.10417258018734,
      "grad_norm": 1.0198668241500854,
      "learning_rate": 4.891711609423787e-05,
      "loss": 0.012655633687973022,
      "step": 180040
    },
    {
      "epoch": 51.1070110701107,
      "grad_norm": 0.10073983669281006,
      "learning_rate": 4.891427760431451e-05,
      "loss": 0.006173787266016006,
      "step": 180050
    },
    {
      "epoch": 51.109849560034064,
      "grad_norm": 1.8462492227554321,
      "learning_rate": 4.891143911439114e-05,
      "loss": 0.0018522180616855621,
      "step": 180060
    },
    {
      "epoch": 51.11268804995742,
      "grad_norm": 0.10347336530685425,
      "learning_rate": 4.8908600624467784e-05,
      "loss": 0.0011287227272987366,
      "step": 180070
    },
    {
      "epoch": 51.11552653988078,
      "grad_norm": 0.07740754634141922,
      "learning_rate": 4.8905762134544426e-05,
      "loss": 0.00034088008105754853,
      "step": 180080
    },
    {
      "epoch": 51.118365029804146,
      "grad_norm": 0.06252571195363998,
      "learning_rate": 4.890292364462106e-05,
      "loss": 0.0022137897089123727,
      "step": 180090
    },
    {
      "epoch": 51.1212035197275,
      "grad_norm": 4.828277111053467,
      "learning_rate": 4.890008515469771e-05,
      "loss": 0.0016985338181257248,
      "step": 180100
    },
    {
      "epoch": 51.124042009650864,
      "grad_norm": 0.24035215377807617,
      "learning_rate": 4.889724666477434e-05,
      "loss": 0.0011762615293264389,
      "step": 180110
    },
    {
      "epoch": 51.12688049957423,
      "grad_norm": 0.011230031959712505,
      "learning_rate": 4.889440817485098e-05,
      "loss": 0.0009894957765936852,
      "step": 180120
    },
    {
      "epoch": 51.12971898949759,
      "grad_norm": 0.11687051504850388,
      "learning_rate": 4.889156968492762e-05,
      "loss": 0.0010034220293164253,
      "step": 180130
    },
    {
      "epoch": 51.132557479420946,
      "grad_norm": 3.432638645172119,
      "learning_rate": 4.888873119500426e-05,
      "loss": 0.005461366474628448,
      "step": 180140
    },
    {
      "epoch": 51.13539596934431,
      "grad_norm": 0.10272270441055298,
      "learning_rate": 4.88858927050809e-05,
      "loss": 0.0036858487874269485,
      "step": 180150
    },
    {
      "epoch": 51.13823445926767,
      "grad_norm": 0.6963006854057312,
      "learning_rate": 4.8883054215157536e-05,
      "loss": 0.0051358573138713835,
      "step": 180160
    },
    {
      "epoch": 51.14107294919103,
      "grad_norm": 5.506946086883545,
      "learning_rate": 4.888021572523418e-05,
      "loss": 0.008579801768064499,
      "step": 180170
    },
    {
      "epoch": 51.14391143911439,
      "grad_norm": 0.04704684391617775,
      "learning_rate": 4.887766108430316e-05,
      "loss": 0.003477158397436142,
      "step": 180180
    },
    {
      "epoch": 51.14674992903775,
      "grad_norm": 10.09178352355957,
      "learning_rate": 4.887482259437979e-05,
      "loss": 0.002485775202512741,
      "step": 180190
    },
    {
      "epoch": 51.149588418961116,
      "grad_norm": 0.4808120131492615,
      "learning_rate": 4.8871984104456434e-05,
      "loss": 0.0012505196034908294,
      "step": 180200
    },
    {
      "epoch": 51.15242690888447,
      "grad_norm": 0.07344196736812592,
      "learning_rate": 4.886914561453307e-05,
      "loss": 0.0018891185522079469,
      "step": 180210
    },
    {
      "epoch": 51.155265398807835,
      "grad_norm": 0.5004212856292725,
      "learning_rate": 4.886630712460971e-05,
      "loss": 0.0009931311011314391,
      "step": 180220
    },
    {
      "epoch": 51.1581038887312,
      "grad_norm": 2.5172946453094482,
      "learning_rate": 4.886346863468635e-05,
      "loss": 0.00524277463555336,
      "step": 180230
    },
    {
      "epoch": 51.16094237865455,
      "grad_norm": 0.08604004234075546,
      "learning_rate": 4.8860630144762986e-05,
      "loss": 0.0007053487002849579,
      "step": 180240
    },
    {
      "epoch": 51.163780868577916,
      "grad_norm": 1.566454291343689,
      "learning_rate": 4.885779165483963e-05,
      "loss": 0.00382319837808609,
      "step": 180250
    },
    {
      "epoch": 51.16661935850128,
      "grad_norm": 0.04597785323858261,
      "learning_rate": 4.885495316491627e-05,
      "loss": 0.0007988559082150459,
      "step": 180260
    },
    {
      "epoch": 51.169457848424635,
      "grad_norm": 0.06951954215765,
      "learning_rate": 4.88521146749929e-05,
      "loss": 0.001137450896203518,
      "step": 180270
    },
    {
      "epoch": 51.172296338348,
      "grad_norm": 0.1777539849281311,
      "learning_rate": 4.884927618506955e-05,
      "loss": 0.0053015690296888355,
      "step": 180280
    },
    {
      "epoch": 51.17513482827136,
      "grad_norm": 0.1762663573026657,
      "learning_rate": 4.8846437695146186e-05,
      "loss": 0.0038615263998508453,
      "step": 180290
    },
    {
      "epoch": 51.177973318194724,
      "grad_norm": 6.532079219818115,
      "learning_rate": 4.884359920522282e-05,
      "loss": 0.0020250273868441583,
      "step": 180300
    },
    {
      "epoch": 51.18081180811808,
      "grad_norm": 0.9653957486152649,
      "learning_rate": 4.884076071529946e-05,
      "loss": 0.0004380591213703156,
      "step": 180310
    },
    {
      "epoch": 51.18365029804144,
      "grad_norm": 0.050387512892484665,
      "learning_rate": 4.88379222253761e-05,
      "loss": 0.0029013574123382567,
      "step": 180320
    },
    {
      "epoch": 51.186488787964805,
      "grad_norm": 0.018023531883955002,
      "learning_rate": 4.8835083735452744e-05,
      "loss": 0.0013168906792998315,
      "step": 180330
    },
    {
      "epoch": 51.18932727788816,
      "grad_norm": 0.08583804965019226,
      "learning_rate": 4.883224524552938e-05,
      "loss": 0.0011037260293960572,
      "step": 180340
    },
    {
      "epoch": 51.192165767811524,
      "grad_norm": 0.1388053148984909,
      "learning_rate": 4.882940675560602e-05,
      "loss": 0.001609784923493862,
      "step": 180350
    },
    {
      "epoch": 51.19500425773489,
      "grad_norm": 0.024241134524345398,
      "learning_rate": 4.882656826568266e-05,
      "loss": 0.0015512334182858467,
      "step": 180360
    },
    {
      "epoch": 51.19784274765824,
      "grad_norm": 0.0329434908926487,
      "learning_rate": 4.8823729775759296e-05,
      "loss": 0.004232289642095566,
      "step": 180370
    },
    {
      "epoch": 51.200681237581605,
      "grad_norm": 0.0504172220826149,
      "learning_rate": 4.882089128583594e-05,
      "loss": 0.0009813189506530761,
      "step": 180380
    },
    {
      "epoch": 51.20351972750497,
      "grad_norm": 0.017061954364180565,
      "learning_rate": 4.881805279591258e-05,
      "loss": 0.0006106648594141007,
      "step": 180390
    },
    {
      "epoch": 51.20635821742833,
      "grad_norm": 0.3209627866744995,
      "learning_rate": 4.8815214305989214e-05,
      "loss": 0.0005499124526977539,
      "step": 180400
    },
    {
      "epoch": 51.20919670735169,
      "grad_norm": 3.182741641998291,
      "learning_rate": 4.8812375816065855e-05,
      "loss": 0.0008940545842051506,
      "step": 180410
    },
    {
      "epoch": 51.21203519727505,
      "grad_norm": 1.943410038948059,
      "learning_rate": 4.8809537326142496e-05,
      "loss": 0.0005300072953104973,
      "step": 180420
    },
    {
      "epoch": 51.21487368719841,
      "grad_norm": 0.04560404643416405,
      "learning_rate": 4.880669883621913e-05,
      "loss": 0.0009103551506996155,
      "step": 180430
    },
    {
      "epoch": 51.21771217712177,
      "grad_norm": 0.15289495885372162,
      "learning_rate": 4.880386034629577e-05,
      "loss": 0.0003016771748661995,
      "step": 180440
    },
    {
      "epoch": 51.22055066704513,
      "grad_norm": 0.1808972954750061,
      "learning_rate": 4.8801021856372414e-05,
      "loss": 0.00039109475910663604,
      "step": 180450
    },
    {
      "epoch": 51.223389156968494,
      "grad_norm": 0.2849595546722412,
      "learning_rate": 4.879818336644905e-05,
      "loss": 0.00045471377670764924,
      "step": 180460
    },
    {
      "epoch": 51.22622764689185,
      "grad_norm": 0.11283183097839355,
      "learning_rate": 4.879534487652569e-05,
      "loss": 0.002761823497712612,
      "step": 180470
    },
    {
      "epoch": 51.22906613681521,
      "grad_norm": 0.06183494254946709,
      "learning_rate": 4.879250638660233e-05,
      "loss": 0.0010355474427342416,
      "step": 180480
    },
    {
      "epoch": 51.231904626738576,
      "grad_norm": 0.07126171886920929,
      "learning_rate": 4.878966789667897e-05,
      "loss": 0.00028327684849500657,
      "step": 180490
    },
    {
      "epoch": 51.23474311666194,
      "grad_norm": 1.5550838708877563,
      "learning_rate": 4.878682940675561e-05,
      "loss": 0.0009197643026709557,
      "step": 180500
    },
    {
      "epoch": 51.23474311666194,
      "eval_accuracy": 0.977045844725631,
      "eval_loss": 0.08801767975091934,
      "eval_runtime": 32.6738,
      "eval_samples_per_second": 481.334,
      "eval_steps_per_second": 7.529,
      "step": 180500
    },
    {
      "epoch": 51.237581606585294,
      "grad_norm": 0.21365058422088623,
      "learning_rate": 4.878399091683225e-05,
      "loss": 0.000344541110098362,
      "step": 180510
    },
    {
      "epoch": 51.24042009650866,
      "grad_norm": 0.40562233328819275,
      "learning_rate": 4.878115242690889e-05,
      "loss": 0.0006029009819030762,
      "step": 180520
    },
    {
      "epoch": 51.24325858643202,
      "grad_norm": 0.04293714836239815,
      "learning_rate": 4.8778313936985524e-05,
      "loss": 0.001121486723423004,
      "step": 180530
    },
    {
      "epoch": 51.246097076355376,
      "grad_norm": 0.1427977979183197,
      "learning_rate": 4.8775475447062166e-05,
      "loss": 0.0028390195220708847,
      "step": 180540
    },
    {
      "epoch": 51.24893556627874,
      "grad_norm": 0.2921789288520813,
      "learning_rate": 4.877263695713881e-05,
      "loss": 0.003016146831214428,
      "step": 180550
    },
    {
      "epoch": 51.2517740562021,
      "grad_norm": 0.08235447853803635,
      "learning_rate": 4.876979846721544e-05,
      "loss": 0.0005880095064640045,
      "step": 180560
    },
    {
      "epoch": 51.254612546125465,
      "grad_norm": 0.0403439961373806,
      "learning_rate": 4.876695997729208e-05,
      "loss": 0.0011786442250013352,
      "step": 180570
    },
    {
      "epoch": 51.25745103604882,
      "grad_norm": 2.6774117946624756,
      "learning_rate": 4.8764121487368725e-05,
      "loss": 0.001136874221265316,
      "step": 180580
    },
    {
      "epoch": 51.26028952597218,
      "grad_norm": 0.11026768386363983,
      "learning_rate": 4.876128299744536e-05,
      "loss": 0.0003934638574719429,
      "step": 180590
    },
    {
      "epoch": 51.263128015895546,
      "grad_norm": 0.13399673998355865,
      "learning_rate": 4.8758444507522e-05,
      "loss": 0.0002336740493774414,
      "step": 180600
    },
    {
      "epoch": 51.2659665058189,
      "grad_norm": 0.11144089698791504,
      "learning_rate": 4.8755606017598635e-05,
      "loss": 0.0008956875652074813,
      "step": 180610
    },
    {
      "epoch": 51.268804995742265,
      "grad_norm": 0.28044575452804565,
      "learning_rate": 4.875276752767528e-05,
      "loss": 0.003932081162929535,
      "step": 180620
    },
    {
      "epoch": 51.27164348566563,
      "grad_norm": 0.1427331566810608,
      "learning_rate": 4.874992903775192e-05,
      "loss": 0.00586228296160698,
      "step": 180630
    },
    {
      "epoch": 51.274481975588984,
      "grad_norm": 0.3159143328666687,
      "learning_rate": 4.874709054782855e-05,
      "loss": 0.0013567646965384483,
      "step": 180640
    },
    {
      "epoch": 51.27732046551235,
      "grad_norm": 0.03928321227431297,
      "learning_rate": 4.87442520579052e-05,
      "loss": 0.000447937473654747,
      "step": 180650
    },
    {
      "epoch": 51.28015895543571,
      "grad_norm": 0.20527687668800354,
      "learning_rate": 4.8741413567981835e-05,
      "loss": 0.0024988552555441856,
      "step": 180660
    },
    {
      "epoch": 51.28299744535907,
      "grad_norm": 8.317633628845215,
      "learning_rate": 4.8738575078058477e-05,
      "loss": 0.004004713147878647,
      "step": 180670
    },
    {
      "epoch": 51.28583593528243,
      "grad_norm": 0.058568768203258514,
      "learning_rate": 4.873573658813512e-05,
      "loss": 0.00302742887288332,
      "step": 180680
    },
    {
      "epoch": 51.28867442520579,
      "grad_norm": 10.257464408874512,
      "learning_rate": 4.873289809821175e-05,
      "loss": 0.007389913499355316,
      "step": 180690
    },
    {
      "epoch": 51.291512915129154,
      "grad_norm": 0.24839149415493011,
      "learning_rate": 4.8730059608288394e-05,
      "loss": 0.00042100641876459123,
      "step": 180700
    },
    {
      "epoch": 51.29435140505251,
      "grad_norm": 16.516530990600586,
      "learning_rate": 4.872722111836503e-05,
      "loss": 0.00713203102350235,
      "step": 180710
    },
    {
      "epoch": 51.29718989497587,
      "grad_norm": 0.08153164386749268,
      "learning_rate": 4.872438262844167e-05,
      "loss": 0.0002146560698747635,
      "step": 180720
    },
    {
      "epoch": 51.300028384899235,
      "grad_norm": 0.004099681507796049,
      "learning_rate": 4.872154413851831e-05,
      "loss": 0.0004123905673623085,
      "step": 180730
    },
    {
      "epoch": 51.30286687482259,
      "grad_norm": 0.026017921045422554,
      "learning_rate": 4.8718705648594946e-05,
      "loss": 0.0005426462739706039,
      "step": 180740
    },
    {
      "epoch": 51.305705364745954,
      "grad_norm": 6.821203231811523,
      "learning_rate": 4.8715867158671594e-05,
      "loss": 0.00504787340760231,
      "step": 180750
    },
    {
      "epoch": 51.30854385466932,
      "grad_norm": 0.02022259496152401,
      "learning_rate": 4.871302866874823e-05,
      "loss": 0.0022015584632754325,
      "step": 180760
    },
    {
      "epoch": 51.31138234459268,
      "grad_norm": 3.727421283721924,
      "learning_rate": 4.871019017882486e-05,
      "loss": 0.0010901762172579766,
      "step": 180770
    },
    {
      "epoch": 51.314220834516036,
      "grad_norm": 0.06664789468050003,
      "learning_rate": 4.870735168890151e-05,
      "loss": 0.0006896872073411942,
      "step": 180780
    },
    {
      "epoch": 51.3170593244394,
      "grad_norm": 0.11181101202964783,
      "learning_rate": 4.8704513198978146e-05,
      "loss": 0.0011407820507884025,
      "step": 180790
    },
    {
      "epoch": 51.31989781436276,
      "grad_norm": 0.09372278302907944,
      "learning_rate": 4.870167470905479e-05,
      "loss": 0.0016520094126462937,
      "step": 180800
    },
    {
      "epoch": 51.32273630428612,
      "grad_norm": 0.055271636694669724,
      "learning_rate": 4.869883621913142e-05,
      "loss": 0.0005426280200481415,
      "step": 180810
    },
    {
      "epoch": 51.32557479420948,
      "grad_norm": 0.23932403326034546,
      "learning_rate": 4.869599772920806e-05,
      "loss": 0.0003681229427456856,
      "step": 180820
    },
    {
      "epoch": 51.32841328413284,
      "grad_norm": 0.03279741853475571,
      "learning_rate": 4.8693159239284705e-05,
      "loss": 0.0002476053312420845,
      "step": 180830
    },
    {
      "epoch": 51.3312517740562,
      "grad_norm": 0.3722105920314789,
      "learning_rate": 4.869032074936134e-05,
      "loss": 0.0006588449701666832,
      "step": 180840
    },
    {
      "epoch": 51.33409026397956,
      "grad_norm": 0.026470595970749855,
      "learning_rate": 4.868748225943798e-05,
      "loss": 0.0012213436886668206,
      "step": 180850
    },
    {
      "epoch": 51.336928753902924,
      "grad_norm": 0.0803748220205307,
      "learning_rate": 4.868464376951462e-05,
      "loss": 0.004335398226976395,
      "step": 180860
    },
    {
      "epoch": 51.33976724382629,
      "grad_norm": 0.0489346943795681,
      "learning_rate": 4.868180527959126e-05,
      "loss": 0.000897197239100933,
      "step": 180870
    },
    {
      "epoch": 51.34260573374964,
      "grad_norm": 0.15893396735191345,
      "learning_rate": 4.86789667896679e-05,
      "loss": 0.00025262255221605303,
      "step": 180880
    },
    {
      "epoch": 51.345444223673006,
      "grad_norm": 0.015027850866317749,
      "learning_rate": 4.867612829974454e-05,
      "loss": 0.0011533098295331001,
      "step": 180890
    },
    {
      "epoch": 51.34828271359637,
      "grad_norm": 0.6291434168815613,
      "learning_rate": 4.8673289809821174e-05,
      "loss": 0.0012927433475852014,
      "step": 180900
    },
    {
      "epoch": 51.351121203519725,
      "grad_norm": 0.8774008750915527,
      "learning_rate": 4.8670451319897815e-05,
      "loss": 0.0009439870715141297,
      "step": 180910
    },
    {
      "epoch": 51.35395969344309,
      "grad_norm": 0.04107557609677315,
      "learning_rate": 4.866761282997446e-05,
      "loss": 0.000671207718551159,
      "step": 180920
    },
    {
      "epoch": 51.35679818336645,
      "grad_norm": 0.35322102904319763,
      "learning_rate": 4.866477434005109e-05,
      "loss": 0.0003599001094698906,
      "step": 180930
    },
    {
      "epoch": 51.35963667328981,
      "grad_norm": 0.08499258011579514,
      "learning_rate": 4.866193585012773e-05,
      "loss": 0.0003437289968132973,
      "step": 180940
    },
    {
      "epoch": 51.36247516321317,
      "grad_norm": 0.07135917246341705,
      "learning_rate": 4.8659097360204374e-05,
      "loss": 0.005439245700836181,
      "step": 180950
    },
    {
      "epoch": 51.36531365313653,
      "grad_norm": 0.012154940515756607,
      "learning_rate": 4.8656258870281015e-05,
      "loss": 0.0032960034906864165,
      "step": 180960
    },
    {
      "epoch": 51.368152143059895,
      "grad_norm": 9.94111442565918,
      "learning_rate": 4.865342038035765e-05,
      "loss": 0.003684813156723976,
      "step": 180970
    },
    {
      "epoch": 51.37099063298325,
      "grad_norm": 0.032539766281843185,
      "learning_rate": 4.865058189043429e-05,
      "loss": 0.0010061433538794517,
      "step": 180980
    },
    {
      "epoch": 51.37382912290661,
      "grad_norm": 0.013985106721520424,
      "learning_rate": 4.864774340051093e-05,
      "loss": 0.00496593713760376,
      "step": 180990
    },
    {
      "epoch": 51.376667612829976,
      "grad_norm": 0.0863175019621849,
      "learning_rate": 4.864490491058757e-05,
      "loss": 0.0013937652111053467,
      "step": 181000
    },
    {
      "epoch": 51.376667612829976,
      "eval_accuracy": 0.9727856552425764,
      "eval_loss": 0.10613223910331726,
      "eval_runtime": 32.5739,
      "eval_samples_per_second": 482.81,
      "eval_steps_per_second": 7.552,
      "step": 181000
    },
    {
      "epoch": 51.37950610275333,
      "grad_norm": 1.217592716217041,
      "learning_rate": 4.864206642066421e-05,
      "loss": 0.0027171896770596506,
      "step": 181010
    },
    {
      "epoch": 51.382344592676695,
      "grad_norm": 0.02334403246641159,
      "learning_rate": 4.863922793074085e-05,
      "loss": 0.00397578626871109,
      "step": 181020
    },
    {
      "epoch": 51.38518308260006,
      "grad_norm": 0.614835798740387,
      "learning_rate": 4.8636389440817485e-05,
      "loss": 0.002307300828397274,
      "step": 181030
    },
    {
      "epoch": 51.38802157252342,
      "grad_norm": 0.11029272526502609,
      "learning_rate": 4.8633550950894126e-05,
      "loss": 0.007133351266384124,
      "step": 181040
    },
    {
      "epoch": 51.39086006244678,
      "grad_norm": 0.4057632088661194,
      "learning_rate": 4.863071246097077e-05,
      "loss": 0.003668259456753731,
      "step": 181050
    },
    {
      "epoch": 51.39369855237014,
      "grad_norm": 0.03350900113582611,
      "learning_rate": 4.86278739710474e-05,
      "loss": 0.002091832645237446,
      "step": 181060
    },
    {
      "epoch": 51.3965370422935,
      "grad_norm": 0.04151994735002518,
      "learning_rate": 4.8625035481124043e-05,
      "loss": 0.0015479713678359986,
      "step": 181070
    },
    {
      "epoch": 51.39937553221686,
      "grad_norm": 2.130741596221924,
      "learning_rate": 4.8622196991200685e-05,
      "loss": 0.0029365856200456618,
      "step": 181080
    },
    {
      "epoch": 51.40221402214022,
      "grad_norm": 0.05362674966454506,
      "learning_rate": 4.8619358501277326e-05,
      "loss": 0.0008785046637058258,
      "step": 181090
    },
    {
      "epoch": 51.405052512063584,
      "grad_norm": 14.57697868347168,
      "learning_rate": 4.861652001135396e-05,
      "loss": 0.005883552134037018,
      "step": 181100
    },
    {
      "epoch": 51.40789100198694,
      "grad_norm": 13.36205768585205,
      "learning_rate": 4.8613681521430595e-05,
      "loss": 0.0037763576954603197,
      "step": 181110
    },
    {
      "epoch": 51.4107294919103,
      "grad_norm": 0.015633828938007355,
      "learning_rate": 4.8610843031507244e-05,
      "loss": 0.0013071518391370774,
      "step": 181120
    },
    {
      "epoch": 51.413567981833665,
      "grad_norm": 0.03230821713805199,
      "learning_rate": 4.860800454158388e-05,
      "loss": 0.00046955756843090055,
      "step": 181130
    },
    {
      "epoch": 51.41640647175703,
      "grad_norm": 0.03939720615744591,
      "learning_rate": 4.860516605166052e-05,
      "loss": 0.0005629988387227059,
      "step": 181140
    },
    {
      "epoch": 51.419244961680384,
      "grad_norm": 0.03258715942502022,
      "learning_rate": 4.860232756173716e-05,
      "loss": 0.000768083706498146,
      "step": 181150
    },
    {
      "epoch": 51.42208345160375,
      "grad_norm": 0.3541862964630127,
      "learning_rate": 4.8599489071813796e-05,
      "loss": 0.001079992763698101,
      "step": 181160
    },
    {
      "epoch": 51.42492194152711,
      "grad_norm": 16.229467391967773,
      "learning_rate": 4.859665058189044e-05,
      "loss": 0.006977045536041259,
      "step": 181170
    },
    {
      "epoch": 51.427760431450466,
      "grad_norm": 1.6912175416946411,
      "learning_rate": 4.859381209196708e-05,
      "loss": 0.0006770160049200058,
      "step": 181180
    },
    {
      "epoch": 51.43059892137383,
      "grad_norm": 0.16344580054283142,
      "learning_rate": 4.859097360204371e-05,
      "loss": 0.0015537645667791367,
      "step": 181190
    },
    {
      "epoch": 51.43343741129719,
      "grad_norm": 0.021976374089717865,
      "learning_rate": 4.8588135112120354e-05,
      "loss": 0.002021470107138157,
      "step": 181200
    },
    {
      "epoch": 51.43627590122055,
      "grad_norm": 0.08889459818601608,
      "learning_rate": 4.8585296622196996e-05,
      "loss": 0.0006487442180514336,
      "step": 181210
    },
    {
      "epoch": 51.43911439114391,
      "grad_norm": 0.3332346975803375,
      "learning_rate": 4.858245813227364e-05,
      "loss": 0.0008749550208449364,
      "step": 181220
    },
    {
      "epoch": 51.44195288106727,
      "grad_norm": 0.6767814755439758,
      "learning_rate": 4.857961964235027e-05,
      "loss": 0.0006916450336575508,
      "step": 181230
    },
    {
      "epoch": 51.444791370990636,
      "grad_norm": 0.02048870176076889,
      "learning_rate": 4.8576781152426906e-05,
      "loss": 0.0010262200608849525,
      "step": 181240
    },
    {
      "epoch": 51.44762986091399,
      "grad_norm": 0.037782955914735794,
      "learning_rate": 4.8573942662503554e-05,
      "loss": 0.0003974897786974907,
      "step": 181250
    },
    {
      "epoch": 51.450468350837355,
      "grad_norm": 0.09114035964012146,
      "learning_rate": 4.857110417258019e-05,
      "loss": 0.0005215317010879517,
      "step": 181260
    },
    {
      "epoch": 51.45330684076072,
      "grad_norm": 0.054964207112789154,
      "learning_rate": 4.856826568265683e-05,
      "loss": 0.0018994625657796859,
      "step": 181270
    },
    {
      "epoch": 51.45614533068407,
      "grad_norm": 0.012142033316195011,
      "learning_rate": 4.856542719273347e-05,
      "loss": 0.0015295058488845824,
      "step": 181280
    },
    {
      "epoch": 51.458983820607436,
      "grad_norm": 0.07273957133293152,
      "learning_rate": 4.8562588702810106e-05,
      "loss": 0.0006283344700932503,
      "step": 181290
    },
    {
      "epoch": 51.4618223105308,
      "grad_norm": 0.11205682903528214,
      "learning_rate": 4.855975021288675e-05,
      "loss": 0.002934711053967476,
      "step": 181300
    },
    {
      "epoch": 51.464660800454155,
      "grad_norm": 0.45976153016090393,
      "learning_rate": 4.855691172296339e-05,
      "loss": 0.0006705144420266152,
      "step": 181310
    },
    {
      "epoch": 51.46749929037752,
      "grad_norm": 0.05666870251297951,
      "learning_rate": 4.8554073233040024e-05,
      "loss": 0.0005197223275899887,
      "step": 181320
    },
    {
      "epoch": 51.47033778030088,
      "grad_norm": 0.08819159865379333,
      "learning_rate": 4.8551234743116665e-05,
      "loss": 0.002376114763319492,
      "step": 181330
    },
    {
      "epoch": 51.47317627022424,
      "grad_norm": 0.059031471610069275,
      "learning_rate": 4.85483962531933e-05,
      "loss": 0.001571834273636341,
      "step": 181340
    },
    {
      "epoch": 51.4760147601476,
      "grad_norm": 0.03961729258298874,
      "learning_rate": 4.854555776326994e-05,
      "loss": 0.006097172200679779,
      "step": 181350
    },
    {
      "epoch": 51.47885325007096,
      "grad_norm": 0.37075722217559814,
      "learning_rate": 4.854271927334658e-05,
      "loss": 0.0006198080256581307,
      "step": 181360
    },
    {
      "epoch": 51.481691739994325,
      "grad_norm": 0.28974345326423645,
      "learning_rate": 4.853988078342322e-05,
      "loss": 0.0010362718254327773,
      "step": 181370
    },
    {
      "epoch": 51.48453022991768,
      "grad_norm": 1.238527536392212,
      "learning_rate": 4.8537042293499865e-05,
      "loss": 0.0005213011056184769,
      "step": 181380
    },
    {
      "epoch": 51.487368719841044,
      "grad_norm": 0.1479216367006302,
      "learning_rate": 4.85342038035765e-05,
      "loss": 0.0003423018380999565,
      "step": 181390
    },
    {
      "epoch": 51.49020720976441,
      "grad_norm": 0.17771846055984497,
      "learning_rate": 4.8531365313653134e-05,
      "loss": 0.001193930022418499,
      "step": 181400
    },
    {
      "epoch": 51.49304569968777,
      "grad_norm": 0.7106952667236328,
      "learning_rate": 4.852852682372978e-05,
      "loss": 0.00037361718714237215,
      "step": 181410
    },
    {
      "epoch": 51.495884189611125,
      "grad_norm": 0.02444724552333355,
      "learning_rate": 4.852568833380642e-05,
      "loss": 0.0016379190608859062,
      "step": 181420
    },
    {
      "epoch": 51.49872267953449,
      "grad_norm": 0.3195110857486725,
      "learning_rate": 4.852284984388306e-05,
      "loss": 0.004004678130149842,
      "step": 181430
    },
    {
      "epoch": 51.50156116945785,
      "grad_norm": 0.07824578136205673,
      "learning_rate": 4.852001135395969e-05,
      "loss": 0.0004601573571562767,
      "step": 181440
    },
    {
      "epoch": 51.50439965938121,
      "grad_norm": 0.06756619364023209,
      "learning_rate": 4.8517172864036334e-05,
      "loss": 0.0007849743589758873,
      "step": 181450
    },
    {
      "epoch": 51.50723814930457,
      "grad_norm": 8.194310188293457,
      "learning_rate": 4.8514334374112976e-05,
      "loss": 0.0022716274484992026,
      "step": 181460
    },
    {
      "epoch": 51.51007663922793,
      "grad_norm": 0.7217859029769897,
      "learning_rate": 4.851149588418961e-05,
      "loss": 0.0008877860382199287,
      "step": 181470
    },
    {
      "epoch": 51.51291512915129,
      "grad_norm": 0.08673321455717087,
      "learning_rate": 4.850865739426625e-05,
      "loss": 0.0003693487495183945,
      "step": 181480
    },
    {
      "epoch": 51.51575361907465,
      "grad_norm": 4.938827991485596,
      "learning_rate": 4.850581890434289e-05,
      "loss": 0.0008981414139270782,
      "step": 181490
    },
    {
      "epoch": 51.518592108998014,
      "grad_norm": 0.6222519278526306,
      "learning_rate": 4.850298041441953e-05,
      "loss": 0.0006748436018824578,
      "step": 181500
    },
    {
      "epoch": 51.518592108998014,
      "eval_accuracy": 0.9764735804667133,
      "eval_loss": 0.08601141721010208,
      "eval_runtime": 32.561,
      "eval_samples_per_second": 483.002,
      "eval_steps_per_second": 7.555,
      "step": 181500
    },
    {
      "epoch": 51.52143059892138,
      "grad_norm": 0.06601157784461975,
      "learning_rate": 4.8500141924496176e-05,
      "loss": 0.000710977241396904,
      "step": 181510
    },
    {
      "epoch": 51.52426908884473,
      "grad_norm": 0.49527034163475037,
      "learning_rate": 4.849730343457281e-05,
      "loss": 0.0006310498341917991,
      "step": 181520
    },
    {
      "epoch": 51.527107578768096,
      "grad_norm": 0.044814515858888626,
      "learning_rate": 4.8494464944649445e-05,
      "loss": 0.0023737261071801186,
      "step": 181530
    },
    {
      "epoch": 51.52994606869146,
      "grad_norm": 0.3510282635688782,
      "learning_rate": 4.8491626454726086e-05,
      "loss": 0.0007682416588068008,
      "step": 181540
    },
    {
      "epoch": 51.532784558614814,
      "grad_norm": 3.006649971008301,
      "learning_rate": 4.848878796480273e-05,
      "loss": 0.001893463358283043,
      "step": 181550
    },
    {
      "epoch": 51.53562304853818,
      "grad_norm": 1.2473188638687134,
      "learning_rate": 4.848594947487937e-05,
      "loss": 0.0004672415554523468,
      "step": 181560
    },
    {
      "epoch": 51.53846153846154,
      "grad_norm": 0.27910885214805603,
      "learning_rate": 4.8483110984956004e-05,
      "loss": 0.003571612760424614,
      "step": 181570
    },
    {
      "epoch": 51.541300028384896,
      "grad_norm": 0.20243224501609802,
      "learning_rate": 4.8480272495032645e-05,
      "loss": 0.0002509405836462975,
      "step": 181580
    },
    {
      "epoch": 51.54413851830826,
      "grad_norm": 0.04568805918097496,
      "learning_rate": 4.8477434005109287e-05,
      "loss": 0.0012143231928348542,
      "step": 181590
    },
    {
      "epoch": 51.54697700823162,
      "grad_norm": 0.23504769802093506,
      "learning_rate": 4.847459551518592e-05,
      "loss": 0.0004699230194091797,
      "step": 181600
    },
    {
      "epoch": 51.549815498154985,
      "grad_norm": 0.0416623055934906,
      "learning_rate": 4.847175702526256e-05,
      "loss": 0.0005585474893450737,
      "step": 181610
    },
    {
      "epoch": 51.55265398807834,
      "grad_norm": 0.11152494698762894,
      "learning_rate": 4.8468918535339204e-05,
      "loss": 0.0008632868528366088,
      "step": 181620
    },
    {
      "epoch": 51.5554924780017,
      "grad_norm": 13.38485336303711,
      "learning_rate": 4.846608004541584e-05,
      "loss": 0.004079372808337212,
      "step": 181630
    },
    {
      "epoch": 51.558330967925066,
      "grad_norm": 0.08424478769302368,
      "learning_rate": 4.846324155549248e-05,
      "loss": 0.0015043659135699273,
      "step": 181640
    },
    {
      "epoch": 51.56116945784842,
      "grad_norm": 3.046485662460327,
      "learning_rate": 4.846040306556912e-05,
      "loss": 0.0023901436477899553,
      "step": 181650
    },
    {
      "epoch": 51.564007947771785,
      "grad_norm": 0.2820039689540863,
      "learning_rate": 4.8457564575645756e-05,
      "loss": 0.0006506204605102539,
      "step": 181660
    },
    {
      "epoch": 51.56684643769515,
      "grad_norm": 0.11082679778337479,
      "learning_rate": 4.84547260857224e-05,
      "loss": 0.005555438995361328,
      "step": 181670
    },
    {
      "epoch": 51.56968492761851,
      "grad_norm": 0.1286991834640503,
      "learning_rate": 4.845188759579904e-05,
      "loss": 0.0005170630291104316,
      "step": 181680
    },
    {
      "epoch": 51.572523417541866,
      "grad_norm": 0.03374123573303223,
      "learning_rate": 4.844904910587568e-05,
      "loss": 0.0008635444566607476,
      "step": 181690
    },
    {
      "epoch": 51.57536190746523,
      "grad_norm": 0.13898295164108276,
      "learning_rate": 4.8446210615952315e-05,
      "loss": 0.003147616982460022,
      "step": 181700
    },
    {
      "epoch": 51.57820039738859,
      "grad_norm": 1.1623297929763794,
      "learning_rate": 4.8443372126028956e-05,
      "loss": 0.0006249109283089638,
      "step": 181710
    },
    {
      "epoch": 51.58103888731195,
      "grad_norm": 0.1495249718427658,
      "learning_rate": 4.84405336361056e-05,
      "loss": 0.0005292003974318504,
      "step": 181720
    },
    {
      "epoch": 51.58387737723531,
      "grad_norm": 0.07027989625930786,
      "learning_rate": 4.843769514618223e-05,
      "loss": 0.0012970726937055587,
      "step": 181730
    },
    {
      "epoch": 51.586715867158674,
      "grad_norm": 0.04861418157815933,
      "learning_rate": 4.843485665625887e-05,
      "loss": 0.0006477834656834602,
      "step": 181740
    },
    {
      "epoch": 51.58955435708203,
      "grad_norm": 0.3068985641002655,
      "learning_rate": 4.8432018166335515e-05,
      "loss": 0.0005508480593562126,
      "step": 181750
    },
    {
      "epoch": 51.59239284700539,
      "grad_norm": 0.29374852776527405,
      "learning_rate": 4.842917967641215e-05,
      "loss": 0.00033357590436935427,
      "step": 181760
    },
    {
      "epoch": 51.595231336928755,
      "grad_norm": 6.635913372039795,
      "learning_rate": 4.842634118648879e-05,
      "loss": 0.0017150992527604104,
      "step": 181770
    },
    {
      "epoch": 51.59806982685212,
      "grad_norm": 2.9442191123962402,
      "learning_rate": 4.842350269656543e-05,
      "loss": 0.001280095800757408,
      "step": 181780
    },
    {
      "epoch": 51.600908316775474,
      "grad_norm": 0.13151171803474426,
      "learning_rate": 4.8420664206642067e-05,
      "loss": 0.001746627502143383,
      "step": 181790
    },
    {
      "epoch": 51.60374680669884,
      "grad_norm": 0.11686323583126068,
      "learning_rate": 4.841782571671871e-05,
      "loss": 0.0040001943707466125,
      "step": 181800
    },
    {
      "epoch": 51.6065852966222,
      "grad_norm": 0.4985428750514984,
      "learning_rate": 4.841498722679535e-05,
      "loss": 0.0023714551702141763,
      "step": 181810
    },
    {
      "epoch": 51.609423786545555,
      "grad_norm": 0.020097579807043076,
      "learning_rate": 4.8412148736871984e-05,
      "loss": 0.008575651794672012,
      "step": 181820
    },
    {
      "epoch": 51.61226227646892,
      "grad_norm": 1.050213098526001,
      "learning_rate": 4.8409310246948625e-05,
      "loss": 0.0006308304145932198,
      "step": 181830
    },
    {
      "epoch": 51.61510076639228,
      "grad_norm": 0.2779833972454071,
      "learning_rate": 4.840647175702526e-05,
      "loss": 0.0024436334148049354,
      "step": 181840
    },
    {
      "epoch": 51.61793925631564,
      "grad_norm": 1.8336868286132812,
      "learning_rate": 4.840363326710191e-05,
      "loss": 0.0058536045253276825,
      "step": 181850
    },
    {
      "epoch": 51.620777746239,
      "grad_norm": 0.08608710020780563,
      "learning_rate": 4.840079477717854e-05,
      "loss": 0.01229449436068535,
      "step": 181860
    },
    {
      "epoch": 51.62361623616236,
      "grad_norm": 7.038613796234131,
      "learning_rate": 4.839795628725518e-05,
      "loss": 0.007699941098690033,
      "step": 181870
    },
    {
      "epoch": 51.626454726085726,
      "grad_norm": 15.456930160522461,
      "learning_rate": 4.8395117797331825e-05,
      "loss": 0.011249026656150818,
      "step": 181880
    },
    {
      "epoch": 51.62929321600908,
      "grad_norm": 1.8425384759902954,
      "learning_rate": 4.839227930740846e-05,
      "loss": 0.001823759078979492,
      "step": 181890
    },
    {
      "epoch": 51.632131705932444,
      "grad_norm": 0.4331265687942505,
      "learning_rate": 4.83894408174851e-05,
      "loss": 0.002013492025434971,
      "step": 181900
    },
    {
      "epoch": 51.63497019585581,
      "grad_norm": 0.8078008890151978,
      "learning_rate": 4.838660232756174e-05,
      "loss": 0.002008094824850559,
      "step": 181910
    },
    {
      "epoch": 51.63780868577916,
      "grad_norm": 0.22775200009346008,
      "learning_rate": 4.838376383763838e-05,
      "loss": 0.0004917610436677933,
      "step": 181920
    },
    {
      "epoch": 51.640647175702526,
      "grad_norm": 0.018183346837759018,
      "learning_rate": 4.838092534771502e-05,
      "loss": 0.0008903389796614646,
      "step": 181930
    },
    {
      "epoch": 51.64348566562589,
      "grad_norm": 0.11603771895170212,
      "learning_rate": 4.837808685779165e-05,
      "loss": 0.0004816558212041855,
      "step": 181940
    },
    {
      "epoch": 51.646324155549244,
      "grad_norm": 0.05986776575446129,
      "learning_rate": 4.8375248367868295e-05,
      "loss": 0.00038706324994564056,
      "step": 181950
    },
    {
      "epoch": 51.64916264547261,
      "grad_norm": 0.10174457728862762,
      "learning_rate": 4.8372409877944936e-05,
      "loss": 0.0005039626732468605,
      "step": 181960
    },
    {
      "epoch": 51.65200113539597,
      "grad_norm": 7.768913269042969,
      "learning_rate": 4.836957138802157e-05,
      "loss": 0.003003590553998947,
      "step": 181970
    },
    {
      "epoch": 51.65483962531933,
      "grad_norm": 0.0953037291765213,
      "learning_rate": 4.836673289809822e-05,
      "loss": 0.0007136184722185135,
      "step": 181980
    },
    {
      "epoch": 51.65767811524269,
      "grad_norm": 0.450955867767334,
      "learning_rate": 4.8363894408174853e-05,
      "loss": 0.002825357764959335,
      "step": 181990
    },
    {
      "epoch": 51.66051660516605,
      "grad_norm": 1.63673996925354,
      "learning_rate": 4.836105591825149e-05,
      "loss": 0.0014265939593315125,
      "step": 182000
    },
    {
      "epoch": 51.66051660516605,
      "eval_accuracy": 0.9742481083486997,
      "eval_loss": 0.09733858704566956,
      "eval_runtime": 32.4231,
      "eval_samples_per_second": 485.056,
      "eval_steps_per_second": 7.587,
      "step": 182000
    },
    {
      "epoch": 51.663355095089415,
      "grad_norm": 0.07253444939851761,
      "learning_rate": 4.8358217428328136e-05,
      "loss": 0.0018283488228917123,
      "step": 182010
    },
    {
      "epoch": 51.66619358501277,
      "grad_norm": 0.11729633808135986,
      "learning_rate": 4.835537893840477e-05,
      "loss": 0.001674944907426834,
      "step": 182020
    },
    {
      "epoch": 51.66903207493613,
      "grad_norm": 0.14158402383327484,
      "learning_rate": 4.835254044848141e-05,
      "loss": 0.0029733633622527123,
      "step": 182030
    },
    {
      "epoch": 51.671870564859496,
      "grad_norm": 2.668970823287964,
      "learning_rate": 4.834970195855805e-05,
      "loss": 0.005384045466780662,
      "step": 182040
    },
    {
      "epoch": 51.67470905478285,
      "grad_norm": 1.6362977027893066,
      "learning_rate": 4.834686346863469e-05,
      "loss": 0.0010470721870660781,
      "step": 182050
    },
    {
      "epoch": 51.677547544706215,
      "grad_norm": 0.2941102087497711,
      "learning_rate": 4.834402497871133e-05,
      "loss": 0.002261085994541645,
      "step": 182060
    },
    {
      "epoch": 51.68038603462958,
      "grad_norm": 0.8117838501930237,
      "learning_rate": 4.8341186488787964e-05,
      "loss": 0.000460306741297245,
      "step": 182070
    },
    {
      "epoch": 51.68322452455294,
      "grad_norm": 0.4067847728729248,
      "learning_rate": 4.8338347998864605e-05,
      "loss": 0.011284461617469788,
      "step": 182080
    },
    {
      "epoch": 51.6860630144763,
      "grad_norm": 10.003881454467773,
      "learning_rate": 4.833550950894125e-05,
      "loss": 0.005481652915477753,
      "step": 182090
    },
    {
      "epoch": 51.68890150439966,
      "grad_norm": 0.10897538810968399,
      "learning_rate": 4.833267101901788e-05,
      "loss": 0.0007394958287477493,
      "step": 182100
    },
    {
      "epoch": 51.69173999432302,
      "grad_norm": 6.69999361038208,
      "learning_rate": 4.832983252909453e-05,
      "loss": 0.017944443225860595,
      "step": 182110
    },
    {
      "epoch": 51.69457848424638,
      "grad_norm": 1.956730604171753,
      "learning_rate": 4.8326994039171164e-05,
      "loss": 0.0030435658991336823,
      "step": 182120
    },
    {
      "epoch": 51.69741697416974,
      "grad_norm": 0.934472382068634,
      "learning_rate": 4.83241555492478e-05,
      "loss": 0.010732844471931458,
      "step": 182130
    },
    {
      "epoch": 51.700255464093104,
      "grad_norm": 5.449108123779297,
      "learning_rate": 4.832131705932444e-05,
      "loss": 0.0023874159902334212,
      "step": 182140
    },
    {
      "epoch": 51.70309395401647,
      "grad_norm": 0.15881489217281342,
      "learning_rate": 4.831847856940108e-05,
      "loss": 0.0009800756350159645,
      "step": 182150
    },
    {
      "epoch": 51.70593244393982,
      "grad_norm": 13.191521644592285,
      "learning_rate": 4.831564007947772e-05,
      "loss": 0.0035543959587812425,
      "step": 182160
    },
    {
      "epoch": 51.708770933863185,
      "grad_norm": 4.631021022796631,
      "learning_rate": 4.831280158955436e-05,
      "loss": 0.0025419492274522783,
      "step": 182170
    },
    {
      "epoch": 51.71160942378655,
      "grad_norm": 0.06105424836277962,
      "learning_rate": 4.8309963099631e-05,
      "loss": 0.0008065775036811829,
      "step": 182180
    },
    {
      "epoch": 51.714447913709904,
      "grad_norm": 0.7231377363204956,
      "learning_rate": 4.830712460970764e-05,
      "loss": 0.0009502695873379707,
      "step": 182190
    },
    {
      "epoch": 51.71728640363327,
      "grad_norm": 0.36902305483818054,
      "learning_rate": 4.8304286119784275e-05,
      "loss": 0.0009264951571822166,
      "step": 182200
    },
    {
      "epoch": 51.72012489355663,
      "grad_norm": 0.06052348390221596,
      "learning_rate": 4.8301447629860916e-05,
      "loss": 0.0007219156250357628,
      "step": 182210
    },
    {
      "epoch": 51.722963383479986,
      "grad_norm": 0.1284666210412979,
      "learning_rate": 4.829860913993756e-05,
      "loss": 0.004949063062667847,
      "step": 182220
    },
    {
      "epoch": 51.72580187340335,
      "grad_norm": 0.509452223777771,
      "learning_rate": 4.829577065001419e-05,
      "loss": 0.00042866673320531847,
      "step": 182230
    },
    {
      "epoch": 51.72864036332671,
      "grad_norm": 0.2914709448814392,
      "learning_rate": 4.8292932160090834e-05,
      "loss": 0.00347202867269516,
      "step": 182240
    },
    {
      "epoch": 51.731478853250074,
      "grad_norm": 0.10682666301727295,
      "learning_rate": 4.8290093670167475e-05,
      "loss": 0.00019755307585000992,
      "step": 182250
    },
    {
      "epoch": 51.73431734317343,
      "grad_norm": 0.013890299014747143,
      "learning_rate": 4.828725518024411e-05,
      "loss": 0.00042697135359048843,
      "step": 182260
    },
    {
      "epoch": 51.73715583309679,
      "grad_norm": 1.1726351976394653,
      "learning_rate": 4.828441669032075e-05,
      "loss": 0.00043001510202884675,
      "step": 182270
    },
    {
      "epoch": 51.739994323020156,
      "grad_norm": 0.03203336149454117,
      "learning_rate": 4.828157820039739e-05,
      "loss": 0.00021441206336021424,
      "step": 182280
    },
    {
      "epoch": 51.74283281294351,
      "grad_norm": 0.059880051761865616,
      "learning_rate": 4.827873971047403e-05,
      "loss": 0.0006886312738060951,
      "step": 182290
    },
    {
      "epoch": 51.745671302866874,
      "grad_norm": 0.0780503898859024,
      "learning_rate": 4.827590122055067e-05,
      "loss": 0.00047068726271390913,
      "step": 182300
    },
    {
      "epoch": 51.74850979279024,
      "grad_norm": 0.06790386140346527,
      "learning_rate": 4.827306273062731e-05,
      "loss": 0.0002838464453816414,
      "step": 182310
    },
    {
      "epoch": 51.75134828271359,
      "grad_norm": 0.09590662270784378,
      "learning_rate": 4.827022424070395e-05,
      "loss": 0.0005536878481507302,
      "step": 182320
    },
    {
      "epoch": 51.754186772636956,
      "grad_norm": 0.015578238293528557,
      "learning_rate": 4.8267385750780586e-05,
      "loss": 0.0005279622972011566,
      "step": 182330
    },
    {
      "epoch": 51.75702526256032,
      "grad_norm": 0.958501398563385,
      "learning_rate": 4.826454726085722e-05,
      "loss": 0.0008314261212944984,
      "step": 182340
    },
    {
      "epoch": 51.75986375248368,
      "grad_norm": 0.14000454545021057,
      "learning_rate": 4.826170877093387e-05,
      "loss": 0.0009681940078735352,
      "step": 182350
    },
    {
      "epoch": 51.76270224240704,
      "grad_norm": 2.396366834640503,
      "learning_rate": 4.82588702810105e-05,
      "loss": 0.0008987881243228913,
      "step": 182360
    },
    {
      "epoch": 51.7655407323304,
      "grad_norm": 0.047076817601919174,
      "learning_rate": 4.8256031791087144e-05,
      "loss": 0.0007864663377404213,
      "step": 182370
    },
    {
      "epoch": 51.76837922225376,
      "grad_norm": 0.07016611844301224,
      "learning_rate": 4.8253193301163786e-05,
      "loss": 0.0008623959496617317,
      "step": 182380
    },
    {
      "epoch": 51.77121771217712,
      "grad_norm": 0.8966825604438782,
      "learning_rate": 4.825035481124042e-05,
      "loss": 0.00121581070125103,
      "step": 182390
    },
    {
      "epoch": 51.77405620210048,
      "grad_norm": 0.015400027856230736,
      "learning_rate": 4.824751632131706e-05,
      "loss": 0.0014248896390199662,
      "step": 182400
    },
    {
      "epoch": 51.776894692023845,
      "grad_norm": 0.052285660058259964,
      "learning_rate": 4.82446778313937e-05,
      "loss": 0.002049084007740021,
      "step": 182410
    },
    {
      "epoch": 51.7797331819472,
      "grad_norm": 0.015802329406142235,
      "learning_rate": 4.824183934147034e-05,
      "loss": 0.0025892384350299836,
      "step": 182420
    },
    {
      "epoch": 51.78257167187056,
      "grad_norm": 0.011272271163761616,
      "learning_rate": 4.823900085154698e-05,
      "loss": 0.0059128038585186,
      "step": 182430
    },
    {
      "epoch": 51.785410161793926,
      "grad_norm": 1.2104628086090088,
      "learning_rate": 4.8236162361623614e-05,
      "loss": 0.0006790978834033013,
      "step": 182440
    },
    {
      "epoch": 51.78824865171729,
      "grad_norm": 0.05820819362998009,
      "learning_rate": 4.823332387170026e-05,
      "loss": 0.015612594783306122,
      "step": 182450
    },
    {
      "epoch": 51.791087141640645,
      "grad_norm": 0.3581760823726654,
      "learning_rate": 4.8230485381776896e-05,
      "loss": 0.004312196373939514,
      "step": 182460
    },
    {
      "epoch": 51.79392563156401,
      "grad_norm": 0.08177873492240906,
      "learning_rate": 4.822764689185353e-05,
      "loss": 0.003569342941045761,
      "step": 182470
    },
    {
      "epoch": 51.79676412148737,
      "grad_norm": 0.05191491171717644,
      "learning_rate": 4.822480840193018e-05,
      "loss": 0.004858438298106194,
      "step": 182480
    },
    {
      "epoch": 51.79960261141073,
      "grad_norm": 0.03919440880417824,
      "learning_rate": 4.8221969912006814e-05,
      "loss": 0.0023818347603082658,
      "step": 182490
    },
    {
      "epoch": 51.80244110133409,
      "grad_norm": 0.12623044848442078,
      "learning_rate": 4.8219131422083455e-05,
      "loss": 0.009125805646181106,
      "step": 182500
    },
    {
      "epoch": 51.80244110133409,
      "eval_accuracy": 0.9734850893368093,
      "eval_loss": 0.09479783475399017,
      "eval_runtime": 32.4514,
      "eval_samples_per_second": 484.632,
      "eval_steps_per_second": 7.581,
      "step": 182500
    },
    {
      "epoch": 51.80527959125745,
      "grad_norm": 0.9531261324882507,
      "learning_rate": 4.8216292932160097e-05,
      "loss": 0.0010429421439766885,
      "step": 182510
    },
    {
      "epoch": 51.808118081180815,
      "grad_norm": 0.2409517765045166,
      "learning_rate": 4.821345444223673e-05,
      "loss": 0.0012009555473923683,
      "step": 182520
    },
    {
      "epoch": 51.81095657110417,
      "grad_norm": 0.2287381887435913,
      "learning_rate": 4.821061595231337e-05,
      "loss": 0.0013151943683624267,
      "step": 182530
    },
    {
      "epoch": 51.813795061027534,
      "grad_norm": 0.13147073984146118,
      "learning_rate": 4.8207777462390014e-05,
      "loss": 0.005388849601149559,
      "step": 182540
    },
    {
      "epoch": 51.8166335509509,
      "grad_norm": 0.07902240008115768,
      "learning_rate": 4.820493897246665e-05,
      "loss": 0.00048444122076034545,
      "step": 182550
    },
    {
      "epoch": 51.81947204087425,
      "grad_norm": 0.11406946182250977,
      "learning_rate": 4.820210048254329e-05,
      "loss": 0.0032390281558036804,
      "step": 182560
    },
    {
      "epoch": 51.822310530797616,
      "grad_norm": 0.10355222970247269,
      "learning_rate": 4.8199261992619924e-05,
      "loss": 0.0005752943456172943,
      "step": 182570
    },
    {
      "epoch": 51.82514902072098,
      "grad_norm": 0.3657318651676178,
      "learning_rate": 4.819642350269657e-05,
      "loss": 0.0014988573268055916,
      "step": 182580
    },
    {
      "epoch": 51.827987510644334,
      "grad_norm": 0.014269332401454449,
      "learning_rate": 4.819358501277321e-05,
      "loss": 0.010669919103384018,
      "step": 182590
    },
    {
      "epoch": 51.8308260005677,
      "grad_norm": 0.3653144836425781,
      "learning_rate": 4.819074652284984e-05,
      "loss": 0.0015364862978458405,
      "step": 182600
    },
    {
      "epoch": 51.83366449049106,
      "grad_norm": 0.162485733628273,
      "learning_rate": 4.818790803292649e-05,
      "loss": 0.0010139100253582002,
      "step": 182610
    },
    {
      "epoch": 51.83650298041442,
      "grad_norm": 0.16376501321792603,
      "learning_rate": 4.8185069543003125e-05,
      "loss": 0.0007864033803343772,
      "step": 182620
    },
    {
      "epoch": 51.83934147033778,
      "grad_norm": 0.10423777997493744,
      "learning_rate": 4.8182231053079766e-05,
      "loss": 0.0018844639882445336,
      "step": 182630
    },
    {
      "epoch": 51.84217996026114,
      "grad_norm": 0.41506412625312805,
      "learning_rate": 4.817939256315641e-05,
      "loss": 0.0006255175918340683,
      "step": 182640
    },
    {
      "epoch": 51.845018450184504,
      "grad_norm": 0.05386204272508621,
      "learning_rate": 4.817655407323304e-05,
      "loss": 0.00047949235886335373,
      "step": 182650
    },
    {
      "epoch": 51.84785694010786,
      "grad_norm": 0.06201737001538277,
      "learning_rate": 4.817371558330968e-05,
      "loss": 0.00077715665102005,
      "step": 182660
    },
    {
      "epoch": 51.85069543003122,
      "grad_norm": 0.22596240043640137,
      "learning_rate": 4.817087709338632e-05,
      "loss": 0.0015824969857931137,
      "step": 182670
    },
    {
      "epoch": 51.853533919954586,
      "grad_norm": 0.05849112942814827,
      "learning_rate": 4.816803860346296e-05,
      "loss": 0.0004113098606467247,
      "step": 182680
    },
    {
      "epoch": 51.85637240987794,
      "grad_norm": 0.10807622224092484,
      "learning_rate": 4.81652001135396e-05,
      "loss": 0.0003912497311830521,
      "step": 182690
    },
    {
      "epoch": 51.859210899801305,
      "grad_norm": 0.014222614467144012,
      "learning_rate": 4.8162361623616235e-05,
      "loss": 0.0005474014207720757,
      "step": 182700
    },
    {
      "epoch": 51.86204938972467,
      "grad_norm": 0.13809043169021606,
      "learning_rate": 4.8159523133692877e-05,
      "loss": 0.0005332110449671745,
      "step": 182710
    },
    {
      "epoch": 51.86488787964803,
      "grad_norm": 0.2838243246078491,
      "learning_rate": 4.815668464376952e-05,
      "loss": 0.0017236709594726563,
      "step": 182720
    },
    {
      "epoch": 51.867726369571386,
      "grad_norm": 0.1773257553577423,
      "learning_rate": 4.815384615384615e-05,
      "loss": 0.0016954099759459496,
      "step": 182730
    },
    {
      "epoch": 51.87056485949475,
      "grad_norm": 0.1669190675020218,
      "learning_rate": 4.81510076639228e-05,
      "loss": 0.0025507261976599692,
      "step": 182740
    },
    {
      "epoch": 51.87340334941811,
      "grad_norm": 0.09507136046886444,
      "learning_rate": 4.8148169173999435e-05,
      "loss": 0.0006633238866925239,
      "step": 182750
    },
    {
      "epoch": 51.87624183934147,
      "grad_norm": 0.16149522364139557,
      "learning_rate": 4.814533068407607e-05,
      "loss": 0.0041283808648586275,
      "step": 182760
    },
    {
      "epoch": 51.87908032926483,
      "grad_norm": 0.07657840847969055,
      "learning_rate": 4.814249219415271e-05,
      "loss": 0.0007986057549715042,
      "step": 182770
    },
    {
      "epoch": 51.88191881918819,
      "grad_norm": 0.21423909068107605,
      "learning_rate": 4.813965370422935e-05,
      "loss": 0.004993660002946853,
      "step": 182780
    },
    {
      "epoch": 51.88475730911155,
      "grad_norm": 0.5809764862060547,
      "learning_rate": 4.8136815214305994e-05,
      "loss": 0.0007304297760128975,
      "step": 182790
    },
    {
      "epoch": 51.88759579903491,
      "grad_norm": 1.5357874631881714,
      "learning_rate": 4.813397672438263e-05,
      "loss": 0.0007596768438816071,
      "step": 182800
    },
    {
      "epoch": 51.890434288958275,
      "grad_norm": 0.26369741559028625,
      "learning_rate": 4.813113823445927e-05,
      "loss": 0.00036773253232240676,
      "step": 182810
    },
    {
      "epoch": 51.89327277888164,
      "grad_norm": 0.2780047059059143,
      "learning_rate": 4.812829974453591e-05,
      "loss": 0.010024514794349671,
      "step": 182820
    },
    {
      "epoch": 51.896111268804994,
      "grad_norm": 0.12321890145540237,
      "learning_rate": 4.8125461254612546e-05,
      "loss": 0.0017388559877872466,
      "step": 182830
    },
    {
      "epoch": 51.89894975872836,
      "grad_norm": 1.678441047668457,
      "learning_rate": 4.812262276468919e-05,
      "loss": 0.0013358509168028831,
      "step": 182840
    },
    {
      "epoch": 51.90178824865172,
      "grad_norm": 0.35110482573509216,
      "learning_rate": 4.811978427476583e-05,
      "loss": 0.0017352698370814324,
      "step": 182850
    },
    {
      "epoch": 51.904626738575075,
      "grad_norm": 0.02774232253432274,
      "learning_rate": 4.811694578484246e-05,
      "loss": 0.0008305624127388,
      "step": 182860
    },
    {
      "epoch": 51.90746522849844,
      "grad_norm": 5.005304336547852,
      "learning_rate": 4.8114107294919105e-05,
      "loss": 0.0038928404450416566,
      "step": 182870
    },
    {
      "epoch": 51.9103037184218,
      "grad_norm": 4.239023208618164,
      "learning_rate": 4.8111268804995746e-05,
      "loss": 0.001853250525891781,
      "step": 182880
    },
    {
      "epoch": 51.913142208345164,
      "grad_norm": 14.860383033752441,
      "learning_rate": 4.810843031507238e-05,
      "loss": 0.007831629365682602,
      "step": 182890
    },
    {
      "epoch": 51.91598069826852,
      "grad_norm": 2.194275379180908,
      "learning_rate": 4.810559182514902e-05,
      "loss": 0.0012575622648000717,
      "step": 182900
    },
    {
      "epoch": 51.91881918819188,
      "grad_norm": 7.52362585067749,
      "learning_rate": 4.8102753335225663e-05,
      "loss": 0.0018234940245747567,
      "step": 182910
    },
    {
      "epoch": 51.921657678115245,
      "grad_norm": 0.20958444476127625,
      "learning_rate": 4.8099914845302305e-05,
      "loss": 0.00027543045580387117,
      "step": 182920
    },
    {
      "epoch": 51.9244961680386,
      "grad_norm": 0.03363151475787163,
      "learning_rate": 4.809707635537894e-05,
      "loss": 0.0011302854865789414,
      "step": 182930
    },
    {
      "epoch": 51.927334657961964,
      "grad_norm": 0.03240484744310379,
      "learning_rate": 4.809423786545558e-05,
      "loss": 0.0013516105711460114,
      "step": 182940
    },
    {
      "epoch": 51.93017314788533,
      "grad_norm": 0.18886597454547882,
      "learning_rate": 4.809139937553222e-05,
      "loss": 0.000336373969912529,
      "step": 182950
    },
    {
      "epoch": 51.93301163780868,
      "grad_norm": 0.1809409111738205,
      "learning_rate": 4.808856088560886e-05,
      "loss": 0.0010003114119172096,
      "step": 182960
    },
    {
      "epoch": 51.935850127732046,
      "grad_norm": 5.938627243041992,
      "learning_rate": 4.80857223956855e-05,
      "loss": 0.0027820289134979246,
      "step": 182970
    },
    {
      "epoch": 51.93868861765541,
      "grad_norm": 0.10181485861539841,
      "learning_rate": 4.808288390576214e-05,
      "loss": 0.0008155753836035729,
      "step": 182980
    },
    {
      "epoch": 51.94152710757877,
      "grad_norm": 0.31481724977493286,
      "learning_rate": 4.8080045415838774e-05,
      "loss": 0.0008360465988516807,
      "step": 182990
    },
    {
      "epoch": 51.94436559750213,
      "grad_norm": 0.14639221131801605,
      "learning_rate": 4.8077490774907754e-05,
      "loss": 0.010876993834972381,
      "step": 183000
    },
    {
      "epoch": 51.94436559750213,
      "eval_accuracy": 0.9767279201373434,
      "eval_loss": 0.08172839879989624,
      "eval_runtime": 32.4181,
      "eval_samples_per_second": 485.13,
      "eval_steps_per_second": 7.588,
      "step": 183000
    },
    {
      "epoch": 51.94720408742549,
      "grad_norm": 0.24690350890159607,
      "learning_rate": 4.807465228498439e-05,
      "loss": 0.00031044557690620425,
      "step": 183010
    },
    {
      "epoch": 51.95004257734885,
      "grad_norm": 0.048013534396886826,
      "learning_rate": 4.807181379506103e-05,
      "loss": 0.0006077568978071213,
      "step": 183020
    },
    {
      "epoch": 51.95288106727221,
      "grad_norm": 0.035825103521347046,
      "learning_rate": 4.806897530513767e-05,
      "loss": 0.0020835012197494506,
      "step": 183030
    },
    {
      "epoch": 51.95571955719557,
      "grad_norm": 0.06610327214002609,
      "learning_rate": 4.8066136815214306e-05,
      "loss": 0.0003492157906293869,
      "step": 183040
    },
    {
      "epoch": 51.958558047118935,
      "grad_norm": 0.6760591864585876,
      "learning_rate": 4.806329832529095e-05,
      "loss": 0.00025796499103307723,
      "step": 183050
    },
    {
      "epoch": 51.96139653704229,
      "grad_norm": 0.14896906912326813,
      "learning_rate": 4.806045983536759e-05,
      "loss": 0.000361521914601326,
      "step": 183060
    },
    {
      "epoch": 51.96423502696565,
      "grad_norm": 0.08462279289960861,
      "learning_rate": 4.805762134544422e-05,
      "loss": 0.0005348850041627884,
      "step": 183070
    },
    {
      "epoch": 51.967073516889016,
      "grad_norm": 0.9394897818565369,
      "learning_rate": 4.8054782855520865e-05,
      "loss": 0.0003594519570469856,
      "step": 183080
    },
    {
      "epoch": 51.96991200681238,
      "grad_norm": 3.250230550765991,
      "learning_rate": 4.8051944365597506e-05,
      "loss": 0.00048760809004306794,
      "step": 183090
    },
    {
      "epoch": 51.972750496735735,
      "grad_norm": 0.013411608524620533,
      "learning_rate": 4.804910587567415e-05,
      "loss": 0.0009473580867052078,
      "step": 183100
    },
    {
      "epoch": 51.9755889866591,
      "grad_norm": 0.07319562882184982,
      "learning_rate": 4.804626738575078e-05,
      "loss": 0.0006691792979836464,
      "step": 183110
    },
    {
      "epoch": 51.97842747658246,
      "grad_norm": 0.0706510841846466,
      "learning_rate": 4.8043428895827423e-05,
      "loss": 0.0005858691409230232,
      "step": 183120
    },
    {
      "epoch": 51.981265966505816,
      "grad_norm": 0.741838812828064,
      "learning_rate": 4.8040590405904065e-05,
      "loss": 0.0012753462418913842,
      "step": 183130
    },
    {
      "epoch": 51.98410445642918,
      "grad_norm": 0.07140675187110901,
      "learning_rate": 4.80377519159807e-05,
      "loss": 0.0004938645288348198,
      "step": 183140
    },
    {
      "epoch": 51.98694294635254,
      "grad_norm": 0.4427141845226288,
      "learning_rate": 4.803491342605734e-05,
      "loss": 0.005382928252220154,
      "step": 183150
    },
    {
      "epoch": 51.9897814362759,
      "grad_norm": 2.9594504833221436,
      "learning_rate": 4.803207493613398e-05,
      "loss": 0.0012240985408425332,
      "step": 183160
    },
    {
      "epoch": 51.99261992619926,
      "grad_norm": 0.4567752182483673,
      "learning_rate": 4.802923644621062e-05,
      "loss": 0.014815299212932587,
      "step": 183170
    },
    {
      "epoch": 51.995458416122624,
      "grad_norm": 0.09555605798959732,
      "learning_rate": 4.802639795628726e-05,
      "loss": 0.0009070837870240212,
      "step": 183180
    },
    {
      "epoch": 51.99829690604599,
      "grad_norm": 0.24449460208415985,
      "learning_rate": 4.80235594663639e-05,
      "loss": 0.0015728767961263657,
      "step": 183190
    },
    {
      "epoch": 52.00113539596934,
      "grad_norm": 0.1252463459968567,
      "learning_rate": 4.8020720976440534e-05,
      "loss": 0.0011314516887068748,
      "step": 183200
    },
    {
      "epoch": 52.003973885892705,
      "grad_norm": 0.17552568018436432,
      "learning_rate": 4.8017882486517175e-05,
      "loss": 0.0063804671168327335,
      "step": 183210
    },
    {
      "epoch": 52.00681237581607,
      "grad_norm": 0.136989563703537,
      "learning_rate": 4.801504399659382e-05,
      "loss": 0.0010395416989922523,
      "step": 183220
    },
    {
      "epoch": 52.009650865739424,
      "grad_norm": 4.250365257263184,
      "learning_rate": 4.801220550667045e-05,
      "loss": 0.001101909577846527,
      "step": 183230
    },
    {
      "epoch": 52.01248935566279,
      "grad_norm": 0.3450939953327179,
      "learning_rate": 4.800936701674709e-05,
      "loss": 0.0005226006731390953,
      "step": 183240
    },
    {
      "epoch": 52.01532784558615,
      "grad_norm": 0.9420564770698547,
      "learning_rate": 4.800652852682373e-05,
      "loss": 0.0012619102373719216,
      "step": 183250
    },
    {
      "epoch": 52.018166335509505,
      "grad_norm": 0.03450402244925499,
      "learning_rate": 4.8003690036900376e-05,
      "loss": 0.00045227594673633573,
      "step": 183260
    },
    {
      "epoch": 52.02100482543287,
      "grad_norm": 0.020162923261523247,
      "learning_rate": 4.800085154697701e-05,
      "loss": 0.000719030387699604,
      "step": 183270
    },
    {
      "epoch": 52.02384331535623,
      "grad_norm": 0.33634984493255615,
      "learning_rate": 4.7998013057053645e-05,
      "loss": 0.0009289706125855446,
      "step": 183280
    },
    {
      "epoch": 52.026681805279594,
      "grad_norm": 0.11174613982439041,
      "learning_rate": 4.799517456713029e-05,
      "loss": 0.001429719477891922,
      "step": 183290
    },
    {
      "epoch": 52.02952029520295,
      "grad_norm": 0.02389918640255928,
      "learning_rate": 4.799233607720693e-05,
      "loss": 0.0015846051275730133,
      "step": 183300
    },
    {
      "epoch": 52.03235878512631,
      "grad_norm": 0.21122485399246216,
      "learning_rate": 4.798949758728357e-05,
      "loss": 0.0014102132990956306,
      "step": 183310
    },
    {
      "epoch": 52.035197275049676,
      "grad_norm": 0.025034114718437195,
      "learning_rate": 4.798665909736021e-05,
      "loss": 0.0007824892178177834,
      "step": 183320
    },
    {
      "epoch": 52.03803576497303,
      "grad_norm": 1.564091682434082,
      "learning_rate": 4.7983820607436845e-05,
      "loss": 0.000429689884185791,
      "step": 183330
    },
    {
      "epoch": 52.040874254896394,
      "grad_norm": 0.26598823070526123,
      "learning_rate": 4.7980982117513486e-05,
      "loss": 0.0026240132749080656,
      "step": 183340
    },
    {
      "epoch": 52.04371274481976,
      "grad_norm": 4.473158359527588,
      "learning_rate": 4.797814362759012e-05,
      "loss": 0.002162446640431881,
      "step": 183350
    },
    {
      "epoch": 52.04655123474312,
      "grad_norm": 0.46561720967292786,
      "learning_rate": 4.797530513766676e-05,
      "loss": 0.005860726162791252,
      "step": 183360
    },
    {
      "epoch": 52.049389724666476,
      "grad_norm": 0.031022032722830772,
      "learning_rate": 4.7972466647743404e-05,
      "loss": 0.0005652682855725288,
      "step": 183370
    },
    {
      "epoch": 52.05222821458984,
      "grad_norm": 0.0655103474855423,
      "learning_rate": 4.796962815782004e-05,
      "loss": 0.0006406722590327263,
      "step": 183380
    },
    {
      "epoch": 52.0550667045132,
      "grad_norm": 8.915067672729492,
      "learning_rate": 4.7966789667896686e-05,
      "loss": 0.0018107224255800246,
      "step": 183390
    },
    {
      "epoch": 52.05790519443656,
      "grad_norm": 0.05562408268451691,
      "learning_rate": 4.796395117797332e-05,
      "loss": 0.004964353144168853,
      "step": 183400
    },
    {
      "epoch": 52.06074368435992,
      "grad_norm": 0.00911193247884512,
      "learning_rate": 4.7961112688049956e-05,
      "loss": 0.0023705745115876198,
      "step": 183410
    },
    {
      "epoch": 52.06358217428328,
      "grad_norm": 1.438262939453125,
      "learning_rate": 4.7958274198126604e-05,
      "loss": 0.001288844272494316,
      "step": 183420
    },
    {
      "epoch": 52.06642066420664,
      "grad_norm": 0.26436740159988403,
      "learning_rate": 4.795543570820324e-05,
      "loss": 0.00023554451763629913,
      "step": 183430
    },
    {
      "epoch": 52.06925915413,
      "grad_norm": 0.018051110208034515,
      "learning_rate": 4.795259721827988e-05,
      "loss": 0.0009310223162174225,
      "step": 183440
    },
    {
      "epoch": 52.072097644053365,
      "grad_norm": 2.080995559692383,
      "learning_rate": 4.7949758728356514e-05,
      "loss": 0.004716078191995621,
      "step": 183450
    },
    {
      "epoch": 52.07493613397673,
      "grad_norm": 0.39596524834632874,
      "learning_rate": 4.7946920238433156e-05,
      "loss": 0.003822232037782669,
      "step": 183460
    },
    {
      "epoch": 52.07777462390008,
      "grad_norm": 0.05577684938907623,
      "learning_rate": 4.79440817485098e-05,
      "loss": 0.0008683588355779647,
      "step": 183470
    },
    {
      "epoch": 52.080613113823446,
      "grad_norm": 0.06712818890810013,
      "learning_rate": 4.794124325858643e-05,
      "loss": 0.001541576161980629,
      "step": 183480
    },
    {
      "epoch": 52.08345160374681,
      "grad_norm": 0.22391265630722046,
      "learning_rate": 4.793840476866307e-05,
      "loss": 0.009365933388471604,
      "step": 183490
    },
    {
      "epoch": 52.086290093670165,
      "grad_norm": 0.1344362050294876,
      "learning_rate": 4.7935566278739714e-05,
      "loss": 0.0015834558755159378,
      "step": 183500
    },
    {
      "epoch": 52.086290093670165,
      "eval_accuracy": 0.9775545240668914,
      "eval_loss": 0.07857996970415115,
      "eval_runtime": 32.6865,
      "eval_samples_per_second": 481.146,
      "eval_steps_per_second": 7.526,
      "step": 183500
    },
    {
      "epoch": 52.08912858359353,
      "grad_norm": 0.15102817118167877,
      "learning_rate": 4.793272778881635e-05,
      "loss": 0.0009012224152684211,
      "step": 183510
    },
    {
      "epoch": 52.09196707351689,
      "grad_norm": 0.08173829317092896,
      "learning_rate": 4.7929889298893e-05,
      "loss": 0.0009478798136115074,
      "step": 183520
    },
    {
      "epoch": 52.09480556344025,
      "grad_norm": 0.024196408689022064,
      "learning_rate": 4.792705080896963e-05,
      "loss": 0.0007890662178397179,
      "step": 183530
    },
    {
      "epoch": 52.09764405336361,
      "grad_norm": 0.01097149308770895,
      "learning_rate": 4.7924212319046266e-05,
      "loss": 0.0004108775407075882,
      "step": 183540
    },
    {
      "epoch": 52.10048254328697,
      "grad_norm": 0.06965126842260361,
      "learning_rate": 4.792137382912291e-05,
      "loss": 0.0013867853209376334,
      "step": 183550
    },
    {
      "epoch": 52.103321033210335,
      "grad_norm": 0.1719314455986023,
      "learning_rate": 4.791853533919955e-05,
      "loss": 0.0038325943052768706,
      "step": 183560
    },
    {
      "epoch": 52.10615952313369,
      "grad_norm": 0.03397070989012718,
      "learning_rate": 4.791569684927619e-05,
      "loss": 0.00030316244810819627,
      "step": 183570
    },
    {
      "epoch": 52.108998013057054,
      "grad_norm": 0.0243116095662117,
      "learning_rate": 4.7912858359352825e-05,
      "loss": 0.0005455181002616882,
      "step": 183580
    },
    {
      "epoch": 52.11183650298042,
      "grad_norm": 1.5040119886398315,
      "learning_rate": 4.7910019869429466e-05,
      "loss": 0.0007761558517813683,
      "step": 183590
    },
    {
      "epoch": 52.11467499290377,
      "grad_norm": 0.30178847908973694,
      "learning_rate": 4.790718137950611e-05,
      "loss": 0.00043684784322977065,
      "step": 183600
    },
    {
      "epoch": 52.117513482827135,
      "grad_norm": 0.1653946489095688,
      "learning_rate": 4.790434288958274e-05,
      "loss": 0.005474375188350677,
      "step": 183610
    },
    {
      "epoch": 52.1203519727505,
      "grad_norm": 0.4449189305305481,
      "learning_rate": 4.7901504399659384e-05,
      "loss": 0.0003239834681153297,
      "step": 183620
    },
    {
      "epoch": 52.123190462673854,
      "grad_norm": 0.19438432157039642,
      "learning_rate": 4.7898665909736025e-05,
      "loss": 0.0009518662467598915,
      "step": 183630
    },
    {
      "epoch": 52.12602895259722,
      "grad_norm": 0.11911072582006454,
      "learning_rate": 4.789582741981266e-05,
      "loss": 0.0005753114819526672,
      "step": 183640
    },
    {
      "epoch": 52.12886744252058,
      "grad_norm": 0.030836397781968117,
      "learning_rate": 4.78929889298893e-05,
      "loss": 0.0013831334188580513,
      "step": 183650
    },
    {
      "epoch": 52.13170593244394,
      "grad_norm": 1.0039864778518677,
      "learning_rate": 4.789015043996594e-05,
      "loss": 0.0004141075536608696,
      "step": 183660
    },
    {
      "epoch": 52.1345444223673,
      "grad_norm": 0.08138184994459152,
      "learning_rate": 4.788731195004258e-05,
      "loss": 0.000584898516535759,
      "step": 183670
    },
    {
      "epoch": 52.13738291229066,
      "grad_norm": 0.04388280585408211,
      "learning_rate": 4.788447346011922e-05,
      "loss": 0.000628044456243515,
      "step": 183680
    },
    {
      "epoch": 52.140221402214024,
      "grad_norm": 21.203325271606445,
      "learning_rate": 4.788163497019586e-05,
      "loss": 0.009185659885406493,
      "step": 183690
    },
    {
      "epoch": 52.14305989213738,
      "grad_norm": 0.20027989149093628,
      "learning_rate": 4.7878796480272494e-05,
      "loss": 0.0003276161849498749,
      "step": 183700
    },
    {
      "epoch": 52.14589838206074,
      "grad_norm": 0.0775650218129158,
      "learning_rate": 4.7875957990349136e-05,
      "loss": 0.0003987180069088936,
      "step": 183710
    },
    {
      "epoch": 52.148736871984106,
      "grad_norm": 0.048422377556562424,
      "learning_rate": 4.787311950042578e-05,
      "loss": 0.0010486222803592683,
      "step": 183720
    },
    {
      "epoch": 52.15157536190747,
      "grad_norm": 5.641684055328369,
      "learning_rate": 4.787028101050242e-05,
      "loss": 0.0013720091432332993,
      "step": 183730
    },
    {
      "epoch": 52.154413851830824,
      "grad_norm": 0.057831570506095886,
      "learning_rate": 4.786744252057905e-05,
      "loss": 0.001147693768143654,
      "step": 183740
    },
    {
      "epoch": 52.15725234175419,
      "grad_norm": 0.09249673783779144,
      "learning_rate": 4.786460403065569e-05,
      "loss": 0.011926360428333282,
      "step": 183750
    },
    {
      "epoch": 52.16009083167755,
      "grad_norm": 0.012170925736427307,
      "learning_rate": 4.7861765540732336e-05,
      "loss": 0.0013270171359181405,
      "step": 183760
    },
    {
      "epoch": 52.162929321600906,
      "grad_norm": 0.08099169284105301,
      "learning_rate": 4.785892705080897e-05,
      "loss": 0.0007620403543114662,
      "step": 183770
    },
    {
      "epoch": 52.16576781152427,
      "grad_norm": 0.09675470739603043,
      "learning_rate": 4.785608856088561e-05,
      "loss": 0.004948173463344574,
      "step": 183780
    },
    {
      "epoch": 52.16860630144763,
      "grad_norm": 0.17229188978672028,
      "learning_rate": 4.785325007096225e-05,
      "loss": 0.0008942348882555962,
      "step": 183790
    },
    {
      "epoch": 52.17144479137099,
      "grad_norm": 2.768332004547119,
      "learning_rate": 4.785041158103889e-05,
      "loss": 0.0013775531202554702,
      "step": 183800
    },
    {
      "epoch": 52.17428328129435,
      "grad_norm": 0.02013053372502327,
      "learning_rate": 4.784757309111553e-05,
      "loss": 0.00039181839674711226,
      "step": 183810
    },
    {
      "epoch": 52.17712177121771,
      "grad_norm": 0.11338331550359726,
      "learning_rate": 4.784473460119217e-05,
      "loss": 0.00220294464379549,
      "step": 183820
    },
    {
      "epoch": 52.179960261141076,
      "grad_norm": 0.24792633950710297,
      "learning_rate": 4.7841896111268805e-05,
      "loss": 0.002382390759885311,
      "step": 183830
    },
    {
      "epoch": 52.18279875106443,
      "grad_norm": 0.03318275138735771,
      "learning_rate": 4.7839057621345447e-05,
      "loss": 0.0046615440398454664,
      "step": 183840
    },
    {
      "epoch": 52.185637240987795,
      "grad_norm": 0.7485451698303223,
      "learning_rate": 4.783621913142208e-05,
      "loss": 0.0020092591643333436,
      "step": 183850
    },
    {
      "epoch": 52.18847573091116,
      "grad_norm": 0.20020534098148346,
      "learning_rate": 4.783338064149873e-05,
      "loss": 0.006860274076461792,
      "step": 183860
    },
    {
      "epoch": 52.19131422083451,
      "grad_norm": 7.625820159912109,
      "learning_rate": 4.7830542151575364e-05,
      "loss": 0.008323963731527328,
      "step": 183870
    },
    {
      "epoch": 52.194152710757876,
      "grad_norm": 0.12388471513986588,
      "learning_rate": 4.7827703661652e-05,
      "loss": 0.0005649343132972717,
      "step": 183880
    },
    {
      "epoch": 52.19699120068124,
      "grad_norm": 21.527563095092773,
      "learning_rate": 4.782486517172865e-05,
      "loss": 0.005670201778411865,
      "step": 183890
    },
    {
      "epoch": 52.199829690604595,
      "grad_norm": 0.045184019953012466,
      "learning_rate": 4.782202668180528e-05,
      "loss": 0.002153523080050945,
      "step": 183900
    },
    {
      "epoch": 52.20266818052796,
      "grad_norm": 0.3825509250164032,
      "learning_rate": 4.781918819188192e-05,
      "loss": 0.0012369973585009575,
      "step": 183910
    },
    {
      "epoch": 52.20550667045132,
      "grad_norm": 0.05174838751554489,
      "learning_rate": 4.7816349701958564e-05,
      "loss": 0.002726239711046219,
      "step": 183920
    },
    {
      "epoch": 52.208345160374684,
      "grad_norm": 0.3884969651699066,
      "learning_rate": 4.78135112120352e-05,
      "loss": 0.005254042148590088,
      "step": 183930
    },
    {
      "epoch": 52.21118365029804,
      "grad_norm": 0.32423657178878784,
      "learning_rate": 4.781067272211184e-05,
      "loss": 0.004890571534633637,
      "step": 183940
    },
    {
      "epoch": 52.2140221402214,
      "grad_norm": 4.699454307556152,
      "learning_rate": 4.7807834232188475e-05,
      "loss": 0.0020103521645069124,
      "step": 183950
    },
    {
      "epoch": 52.216860630144765,
      "grad_norm": 2.506873846054077,
      "learning_rate": 4.7804995742265116e-05,
      "loss": 0.0008382083848118782,
      "step": 183960
    },
    {
      "epoch": 52.21969912006812,
      "grad_norm": 0.03574621304869652,
      "learning_rate": 4.780215725234176e-05,
      "loss": 0.0018414055928587913,
      "step": 183970
    },
    {
      "epoch": 52.222537609991484,
      "grad_norm": 0.00886073149740696,
      "learning_rate": 4.779931876241839e-05,
      "loss": 0.001334230788052082,
      "step": 183980
    },
    {
      "epoch": 52.22537609991485,
      "grad_norm": 0.3629567623138428,
      "learning_rate": 4.779648027249504e-05,
      "loss": 0.0005370743572711944,
      "step": 183990
    },
    {
      "epoch": 52.2282145898382,
      "grad_norm": 0.038913846015930176,
      "learning_rate": 4.7793641782571675e-05,
      "loss": 0.001140449196100235,
      "step": 184000
    },
    {
      "epoch": 52.2282145898382,
      "eval_accuracy": 0.9759649011254531,
      "eval_loss": 0.0853082463145256,
      "eval_runtime": 32.5995,
      "eval_samples_per_second": 482.431,
      "eval_steps_per_second": 7.546,
      "step": 184000
    },
    {
      "epoch": 52.231053079761566,
      "grad_norm": 0.12194695323705673,
      "learning_rate": 4.779080329264831e-05,
      "loss": 0.0014708971604704857,
      "step": 184010
    },
    {
      "epoch": 52.23389156968493,
      "grad_norm": 0.06730790436267853,
      "learning_rate": 4.778796480272496e-05,
      "loss": 0.00044020526111125945,
      "step": 184020
    },
    {
      "epoch": 52.23673005960829,
      "grad_norm": 21.550878524780273,
      "learning_rate": 4.778512631280159e-05,
      "loss": 0.005208000168204307,
      "step": 184030
    },
    {
      "epoch": 52.23956854953165,
      "grad_norm": 0.07917027175426483,
      "learning_rate": 4.778228782287823e-05,
      "loss": 0.00042155440896749494,
      "step": 184040
    },
    {
      "epoch": 52.24240703945501,
      "grad_norm": 0.20390169322490692,
      "learning_rate": 4.777944933295487e-05,
      "loss": 0.0012712359428405761,
      "step": 184050
    },
    {
      "epoch": 52.24524552937837,
      "grad_norm": 0.11394552141427994,
      "learning_rate": 4.777661084303151e-05,
      "loss": 0.0010580187663435936,
      "step": 184060
    },
    {
      "epoch": 52.24808401930173,
      "grad_norm": 0.07907111197710037,
      "learning_rate": 4.777377235310815e-05,
      "loss": 0.002778307907283306,
      "step": 184070
    },
    {
      "epoch": 52.25092250922509,
      "grad_norm": 0.4431132972240448,
      "learning_rate": 4.7770933863184785e-05,
      "loss": 0.002066568098962307,
      "step": 184080
    },
    {
      "epoch": 52.253760999148454,
      "grad_norm": 0.15419727563858032,
      "learning_rate": 4.776809537326143e-05,
      "loss": 0.00034906230866909026,
      "step": 184090
    },
    {
      "epoch": 52.25659948907182,
      "grad_norm": 2.434131622314453,
      "learning_rate": 4.776525688333807e-05,
      "loss": 0.0032071176916360854,
      "step": 184100
    },
    {
      "epoch": 52.25943797899517,
      "grad_norm": 0.40985676646232605,
      "learning_rate": 4.77624183934147e-05,
      "loss": 0.0009425994008779526,
      "step": 184110
    },
    {
      "epoch": 52.262276468918536,
      "grad_norm": 0.33366069197654724,
      "learning_rate": 4.7759579903491344e-05,
      "loss": 0.0006732998415827751,
      "step": 184120
    },
    {
      "epoch": 52.2651149588419,
      "grad_norm": 0.49148792028427124,
      "learning_rate": 4.7756741413567985e-05,
      "loss": 0.004441948980093003,
      "step": 184130
    },
    {
      "epoch": 52.267953448765255,
      "grad_norm": 0.07162880897521973,
      "learning_rate": 4.775390292364462e-05,
      "loss": 0.00047065373510122297,
      "step": 184140
    },
    {
      "epoch": 52.27079193868862,
      "grad_norm": 0.015660328790545464,
      "learning_rate": 4.775106443372126e-05,
      "loss": 0.0003439469262957573,
      "step": 184150
    },
    {
      "epoch": 52.27363042861198,
      "grad_norm": 0.07236897945404053,
      "learning_rate": 4.77482259437979e-05,
      "loss": 0.004213414341211319,
      "step": 184160
    },
    {
      "epoch": 52.276468918535336,
      "grad_norm": 0.04823959246277809,
      "learning_rate": 4.774538745387454e-05,
      "loss": 0.004126560688018799,
      "step": 184170
    },
    {
      "epoch": 52.2793074084587,
      "grad_norm": 2.9902758598327637,
      "learning_rate": 4.774254896395118e-05,
      "loss": 0.0023122189566493036,
      "step": 184180
    },
    {
      "epoch": 52.28214589838206,
      "grad_norm": 0.015466309152543545,
      "learning_rate": 4.773971047402782e-05,
      "loss": 0.0005363108590245247,
      "step": 184190
    },
    {
      "epoch": 52.284984388305425,
      "grad_norm": 1.1628447771072388,
      "learning_rate": 4.773687198410446e-05,
      "loss": 0.0013002697378396988,
      "step": 184200
    },
    {
      "epoch": 52.28782287822878,
      "grad_norm": 0.16024670004844666,
      "learning_rate": 4.7734033494181096e-05,
      "loss": 0.0004971085116267204,
      "step": 184210
    },
    {
      "epoch": 52.29066136815214,
      "grad_norm": 0.04302060231566429,
      "learning_rate": 4.773119500425774e-05,
      "loss": 0.0015284229069948197,
      "step": 184220
    },
    {
      "epoch": 52.293499858075506,
      "grad_norm": 0.09796073287725449,
      "learning_rate": 4.772835651433438e-05,
      "loss": 0.0007480617612600326,
      "step": 184230
    },
    {
      "epoch": 52.29633834799886,
      "grad_norm": 5.639986038208008,
      "learning_rate": 4.7725518024411013e-05,
      "loss": 0.003276053071022034,
      "step": 184240
    },
    {
      "epoch": 52.299176837922225,
      "grad_norm": 0.6984824538230896,
      "learning_rate": 4.7722679534487655e-05,
      "loss": 0.0024738194420933723,
      "step": 184250
    },
    {
      "epoch": 52.30201532784559,
      "grad_norm": 0.043838758021593094,
      "learning_rate": 4.7719841044564296e-05,
      "loss": 0.000746610201895237,
      "step": 184260
    },
    {
      "epoch": 52.304853817768944,
      "grad_norm": 0.29030969738960266,
      "learning_rate": 4.771700255464093e-05,
      "loss": 0.0034334458410739898,
      "step": 184270
    },
    {
      "epoch": 52.30769230769231,
      "grad_norm": 1.4350874423980713,
      "learning_rate": 4.771416406471757e-05,
      "loss": 0.0008537506684660912,
      "step": 184280
    },
    {
      "epoch": 52.31053079761567,
      "grad_norm": 0.02729605883359909,
      "learning_rate": 4.7711325574794214e-05,
      "loss": 0.00159029234200716,
      "step": 184290
    },
    {
      "epoch": 52.31336928753903,
      "grad_norm": 0.2532888352870941,
      "learning_rate": 4.770848708487085e-05,
      "loss": 0.0021553006023168563,
      "step": 184300
    },
    {
      "epoch": 52.31620777746239,
      "grad_norm": 0.13101926445960999,
      "learning_rate": 4.770564859494749e-05,
      "loss": 0.01797635108232498,
      "step": 184310
    },
    {
      "epoch": 52.31904626738575,
      "grad_norm": 0.5800873041152954,
      "learning_rate": 4.770281010502413e-05,
      "loss": 0.004862882941961288,
      "step": 184320
    },
    {
      "epoch": 52.321884757309114,
      "grad_norm": 0.03931891545653343,
      "learning_rate": 4.769997161510077e-05,
      "loss": 0.0023556262254714965,
      "step": 184330
    },
    {
      "epoch": 52.32472324723247,
      "grad_norm": 8.784247398376465,
      "learning_rate": 4.769713312517741e-05,
      "loss": 0.006658436357975006,
      "step": 184340
    },
    {
      "epoch": 52.32756173715583,
      "grad_norm": 3.140986919403076,
      "learning_rate": 4.769429463525405e-05,
      "loss": 0.0011419855058193208,
      "step": 184350
    },
    {
      "epoch": 52.330400227079195,
      "grad_norm": 12.199102401733398,
      "learning_rate": 4.769145614533069e-05,
      "loss": 0.0026628632098436356,
      "step": 184360
    },
    {
      "epoch": 52.33323871700255,
      "grad_norm": 0.20722025632858276,
      "learning_rate": 4.7688617655407324e-05,
      "loss": 0.002056492492556572,
      "step": 184370
    },
    {
      "epoch": 52.336077206925914,
      "grad_norm": 1.593297004699707,
      "learning_rate": 4.7685779165483966e-05,
      "loss": 0.0009723709896206856,
      "step": 184380
    },
    {
      "epoch": 52.33891569684928,
      "grad_norm": 1.6908036470413208,
      "learning_rate": 4.768294067556061e-05,
      "loss": 0.010466842353343964,
      "step": 184390
    },
    {
      "epoch": 52.34175418677264,
      "grad_norm": 10.084512710571289,
      "learning_rate": 4.768010218563724e-05,
      "loss": 0.002477515861392021,
      "step": 184400
    },
    {
      "epoch": 52.344592676695996,
      "grad_norm": 0.8554307222366333,
      "learning_rate": 4.767726369571388e-05,
      "loss": 0.0007222775369882583,
      "step": 184410
    },
    {
      "epoch": 52.34743116661936,
      "grad_norm": 0.06834488362073898,
      "learning_rate": 4.7674425205790524e-05,
      "loss": 0.0016830066218972207,
      "step": 184420
    },
    {
      "epoch": 52.35026965654272,
      "grad_norm": 0.041436124593019485,
      "learning_rate": 4.767158671586716e-05,
      "loss": 0.001099945232272148,
      "step": 184430
    },
    {
      "epoch": 52.35310814646608,
      "grad_norm": 1.5089346170425415,
      "learning_rate": 4.76687482259438e-05,
      "loss": 0.0026598405092954636,
      "step": 184440
    },
    {
      "epoch": 52.35594663638944,
      "grad_norm": 1.0984470844268799,
      "learning_rate": 4.766590973602044e-05,
      "loss": 0.0005424888804554939,
      "step": 184450
    },
    {
      "epoch": 52.3587851263128,
      "grad_norm": 0.2821011245250702,
      "learning_rate": 4.7663071246097076e-05,
      "loss": 0.00337773896753788,
      "step": 184460
    },
    {
      "epoch": 52.36162361623616,
      "grad_norm": 0.2699981927871704,
      "learning_rate": 4.766023275617372e-05,
      "loss": 0.0018408577889204025,
      "step": 184470
    },
    {
      "epoch": 52.36446210615952,
      "grad_norm": 0.05198495090007782,
      "learning_rate": 4.765739426625035e-05,
      "loss": 0.0010284805670380593,
      "step": 184480
    },
    {
      "epoch": 52.367300596082885,
      "grad_norm": 5.800675868988037,
      "learning_rate": 4.7654555776327e-05,
      "loss": 0.0014519184827804566,
      "step": 184490
    },
    {
      "epoch": 52.37013908600625,
      "grad_norm": 0.33598682284355164,
      "learning_rate": 4.7651717286403635e-05,
      "loss": 0.00037855394184589387,
      "step": 184500
    },
    {
      "epoch": 52.37013908600625,
      "eval_accuracy": 0.9760284860431105,
      "eval_loss": 0.09127943217754364,
      "eval_runtime": 32.5253,
      "eval_samples_per_second": 483.531,
      "eval_steps_per_second": 7.563,
      "step": 184500
    },
    {
      "epoch": 52.3729775759296,
      "grad_norm": 0.0159261804074049,
      "learning_rate": 4.7648878796480276e-05,
      "loss": 0.006323327124118805,
      "step": 184510
    },
    {
      "epoch": 52.375816065852966,
      "grad_norm": 0.14504724740982056,
      "learning_rate": 4.764604030655692e-05,
      "loss": 0.0005192440003156662,
      "step": 184520
    },
    {
      "epoch": 52.37865455577633,
      "grad_norm": 0.04463227465748787,
      "learning_rate": 4.764320181663355e-05,
      "loss": 0.0008708713576197624,
      "step": 184530
    },
    {
      "epoch": 52.381493045699685,
      "grad_norm": 0.010425101034343243,
      "learning_rate": 4.7640363326710194e-05,
      "loss": 0.001168590411543846,
      "step": 184540
    },
    {
      "epoch": 52.38433153562305,
      "grad_norm": 0.011947902850806713,
      "learning_rate": 4.7637524836786835e-05,
      "loss": 0.0034333862364292144,
      "step": 184550
    },
    {
      "epoch": 52.38717002554641,
      "grad_norm": 0.12935885787010193,
      "learning_rate": 4.763468634686347e-05,
      "loss": 0.0010515183210372924,
      "step": 184560
    },
    {
      "epoch": 52.39000851546977,
      "grad_norm": 4.406347274780273,
      "learning_rate": 4.763184785694011e-05,
      "loss": 0.002359734661877155,
      "step": 184570
    },
    {
      "epoch": 52.39284700539313,
      "grad_norm": 0.4258795380592346,
      "learning_rate": 4.7629009367016746e-05,
      "loss": 0.0021335961297154426,
      "step": 184580
    },
    {
      "epoch": 52.39568549531649,
      "grad_norm": 0.884838879108429,
      "learning_rate": 4.762617087709339e-05,
      "loss": 0.0044537119567394255,
      "step": 184590
    },
    {
      "epoch": 52.398523985239855,
      "grad_norm": 0.15563571453094482,
      "learning_rate": 4.762333238717003e-05,
      "loss": 0.0018550509586930275,
      "step": 184600
    },
    {
      "epoch": 52.40136247516321,
      "grad_norm": 0.036623600870370865,
      "learning_rate": 4.762049389724666e-05,
      "loss": 0.001363677904009819,
      "step": 184610
    },
    {
      "epoch": 52.404200965086574,
      "grad_norm": 0.14291466772556305,
      "learning_rate": 4.761765540732331e-05,
      "loss": 0.001742580346763134,
      "step": 184620
    },
    {
      "epoch": 52.40703945500994,
      "grad_norm": 0.01803308166563511,
      "learning_rate": 4.7614816917399946e-05,
      "loss": 0.00028180405497550963,
      "step": 184630
    },
    {
      "epoch": 52.40987794493329,
      "grad_norm": 0.07993413507938385,
      "learning_rate": 4.761197842747658e-05,
      "loss": 0.0006215602159500122,
      "step": 184640
    },
    {
      "epoch": 52.412716434856655,
      "grad_norm": 0.039400532841682434,
      "learning_rate": 4.760913993755323e-05,
      "loss": 0.000861317664384842,
      "step": 184650
    },
    {
      "epoch": 52.41555492478002,
      "grad_norm": 0.40908896923065186,
      "learning_rate": 4.760630144762986e-05,
      "loss": 0.0020953189581632615,
      "step": 184660
    },
    {
      "epoch": 52.41839341470338,
      "grad_norm": 0.007140699774026871,
      "learning_rate": 4.7603462957706504e-05,
      "loss": 0.0018901146948337554,
      "step": 184670
    },
    {
      "epoch": 52.42123190462674,
      "grad_norm": 5.458056449890137,
      "learning_rate": 4.760062446778314e-05,
      "loss": 0.004056227952241897,
      "step": 184680
    },
    {
      "epoch": 52.4240703945501,
      "grad_norm": 0.056706395000219345,
      "learning_rate": 4.759778597785978e-05,
      "loss": 0.0008314166218042374,
      "step": 184690
    },
    {
      "epoch": 52.42690888447346,
      "grad_norm": 0.055072952061891556,
      "learning_rate": 4.759494748793642e-05,
      "loss": 0.0037443917244672775,
      "step": 184700
    },
    {
      "epoch": 52.42974737439682,
      "grad_norm": 0.2812627851963043,
      "learning_rate": 4.7592108998013056e-05,
      "loss": 0.011396793276071548,
      "step": 184710
    },
    {
      "epoch": 52.43258586432018,
      "grad_norm": 0.3162594735622406,
      "learning_rate": 4.75892705080897e-05,
      "loss": 0.0013636523857712747,
      "step": 184720
    },
    {
      "epoch": 52.435424354243544,
      "grad_norm": 0.34498757123947144,
      "learning_rate": 4.758643201816634e-05,
      "loss": 0.0008828606456518173,
      "step": 184730
    },
    {
      "epoch": 52.4382628441669,
      "grad_norm": 0.11711649596691132,
      "learning_rate": 4.7583593528242974e-05,
      "loss": 0.003614000603556633,
      "step": 184740
    },
    {
      "epoch": 52.44110133409026,
      "grad_norm": 31.341411590576172,
      "learning_rate": 4.758075503831962e-05,
      "loss": 0.011059018969535827,
      "step": 184750
    },
    {
      "epoch": 52.443939824013626,
      "grad_norm": 0.10356653481721878,
      "learning_rate": 4.7577916548396257e-05,
      "loss": 0.000668397918343544,
      "step": 184760
    },
    {
      "epoch": 52.44677831393699,
      "grad_norm": 0.02240273542702198,
      "learning_rate": 4.757507805847289e-05,
      "loss": 0.0008125150576233864,
      "step": 184770
    },
    {
      "epoch": 52.449616803860344,
      "grad_norm": 0.4690690338611603,
      "learning_rate": 4.757223956854953e-05,
      "loss": 0.0026510611176490785,
      "step": 184780
    },
    {
      "epoch": 52.45245529378371,
      "grad_norm": 0.5949777364730835,
      "learning_rate": 4.7569401078626174e-05,
      "loss": 0.00033699870109558103,
      "step": 184790
    },
    {
      "epoch": 52.45529378370707,
      "grad_norm": 0.01026222389191389,
      "learning_rate": 4.7566562588702815e-05,
      "loss": 0.0003689641132950783,
      "step": 184800
    },
    {
      "epoch": 52.458132273630426,
      "grad_norm": 0.03486157953739166,
      "learning_rate": 4.756372409877945e-05,
      "loss": 0.00091804638504982,
      "step": 184810
    },
    {
      "epoch": 52.46097076355379,
      "grad_norm": 0.03475930169224739,
      "learning_rate": 4.756088560885609e-05,
      "loss": 0.0002218225970864296,
      "step": 184820
    },
    {
      "epoch": 52.46380925347715,
      "grad_norm": 0.15469048917293549,
      "learning_rate": 4.755804711893273e-05,
      "loss": 0.0004458896815776825,
      "step": 184830
    },
    {
      "epoch": 52.46664774340051,
      "grad_norm": 0.02663545124232769,
      "learning_rate": 4.755520862900937e-05,
      "loss": 0.001989557594060898,
      "step": 184840
    },
    {
      "epoch": 52.46948623332387,
      "grad_norm": 0.03500394895672798,
      "learning_rate": 4.755237013908601e-05,
      "loss": 0.0005112333223223686,
      "step": 184850
    },
    {
      "epoch": 52.47232472324723,
      "grad_norm": 0.14612731337547302,
      "learning_rate": 4.754953164916265e-05,
      "loss": 0.0006827004253864288,
      "step": 184860
    },
    {
      "epoch": 52.475163213170596,
      "grad_norm": 0.2962496876716614,
      "learning_rate": 4.7546693159239285e-05,
      "loss": 0.006271124631166458,
      "step": 184870
    },
    {
      "epoch": 52.47800170309395,
      "grad_norm": 0.05187174305319786,
      "learning_rate": 4.7543854669315926e-05,
      "loss": 0.0007941052317619324,
      "step": 184880
    },
    {
      "epoch": 52.480840193017315,
      "grad_norm": 0.15852448344230652,
      "learning_rate": 4.754101617939257e-05,
      "loss": 0.0010285772383213042,
      "step": 184890
    },
    {
      "epoch": 52.48367868294068,
      "grad_norm": 0.06756472587585449,
      "learning_rate": 4.75381776894692e-05,
      "loss": 0.0007333695888519287,
      "step": 184900
    },
    {
      "epoch": 52.48651717286403,
      "grad_norm": 0.91930091381073,
      "learning_rate": 4.753533919954584e-05,
      "loss": 0.0005421994253993034,
      "step": 184910
    },
    {
      "epoch": 52.489355662787396,
      "grad_norm": 0.026963118463754654,
      "learning_rate": 4.7532500709622485e-05,
      "loss": 0.0005008872598409652,
      "step": 184920
    },
    {
      "epoch": 52.49219415271076,
      "grad_norm": 0.02083013765513897,
      "learning_rate": 4.752966221969912e-05,
      "loss": 0.0010683590546250344,
      "step": 184930
    },
    {
      "epoch": 52.49503264263412,
      "grad_norm": 0.02074730210006237,
      "learning_rate": 4.752682372977576e-05,
      "loss": 0.001010383665561676,
      "step": 184940
    },
    {
      "epoch": 52.49787113255748,
      "grad_norm": 0.03191481530666351,
      "learning_rate": 4.75239852398524e-05,
      "loss": 0.0006425833329558372,
      "step": 184950
    },
    {
      "epoch": 52.50070962248084,
      "grad_norm": 1.4525662660598755,
      "learning_rate": 4.752114674992904e-05,
      "loss": 0.0005657251924276352,
      "step": 184960
    },
    {
      "epoch": 52.503548112404204,
      "grad_norm": 12.38812255859375,
      "learning_rate": 4.751830826000568e-05,
      "loss": 0.002616007253527641,
      "step": 184970
    },
    {
      "epoch": 52.50638660232756,
      "grad_norm": 0.9182807803153992,
      "learning_rate": 4.751546977008231e-05,
      "loss": 0.0019361067563295365,
      "step": 184980
    },
    {
      "epoch": 52.50922509225092,
      "grad_norm": 0.0336979404091835,
      "learning_rate": 4.751263128015896e-05,
      "loss": 0.0006442580372095108,
      "step": 184990
    },
    {
      "epoch": 52.512063582174285,
      "grad_norm": 0.038579318672418594,
      "learning_rate": 4.7509792790235595e-05,
      "loss": 0.0012181278318166733,
      "step": 185000
    },
    {
      "epoch": 52.512063582174285,
      "eval_accuracy": 0.9760920709607681,
      "eval_loss": 0.08229879289865494,
      "eval_runtime": 32.8146,
      "eval_samples_per_second": 479.269,
      "eval_steps_per_second": 7.497,
      "step": 185000
    },
    {
      "epoch": 52.51490207209764,
      "grad_norm": 0.2812656760215759,
      "learning_rate": 4.750695430031224e-05,
      "loss": 0.0019872616976499557,
      "step": 185010
    },
    {
      "epoch": 52.517740562021004,
      "grad_norm": 0.6278689503669739,
      "learning_rate": 4.750411581038888e-05,
      "loss": 0.0011812107637524604,
      "step": 185020
    },
    {
      "epoch": 52.52057905194437,
      "grad_norm": 0.10104627907276154,
      "learning_rate": 4.750127732046551e-05,
      "loss": 0.00023264791816473006,
      "step": 185030
    },
    {
      "epoch": 52.52341754186773,
      "grad_norm": 0.34198492765426636,
      "learning_rate": 4.7498438830542154e-05,
      "loss": 0.00035769641399383543,
      "step": 185040
    },
    {
      "epoch": 52.526256031791085,
      "grad_norm": 1.1983280181884766,
      "learning_rate": 4.7495600340618795e-05,
      "loss": 0.0004389923065900803,
      "step": 185050
    },
    {
      "epoch": 52.52909452171445,
      "grad_norm": 0.3388192057609558,
      "learning_rate": 4.749276185069543e-05,
      "loss": 0.00020617544651031495,
      "step": 185060
    },
    {
      "epoch": 52.53193301163781,
      "grad_norm": 0.02668917365372181,
      "learning_rate": 4.748992336077207e-05,
      "loss": 0.0008501425385475158,
      "step": 185070
    },
    {
      "epoch": 52.53477150156117,
      "grad_norm": 0.015657059848308563,
      "learning_rate": 4.7487084870848706e-05,
      "loss": 0.00034740213304758073,
      "step": 185080
    },
    {
      "epoch": 52.53760999148453,
      "grad_norm": 0.02161979116499424,
      "learning_rate": 4.7484246380925354e-05,
      "loss": 0.0009873220697045326,
      "step": 185090
    },
    {
      "epoch": 52.54044848140789,
      "grad_norm": 0.23935921490192413,
      "learning_rate": 4.748140789100199e-05,
      "loss": 0.0003578871488571167,
      "step": 185100
    },
    {
      "epoch": 52.54328697133125,
      "grad_norm": 0.15645882487297058,
      "learning_rate": 4.747856940107862e-05,
      "loss": 0.005178725719451905,
      "step": 185110
    },
    {
      "epoch": 52.54612546125461,
      "grad_norm": 0.07009433209896088,
      "learning_rate": 4.747573091115527e-05,
      "loss": 0.0007305731996893883,
      "step": 185120
    },
    {
      "epoch": 52.548963951177974,
      "grad_norm": 0.2810664176940918,
      "learning_rate": 4.7472892421231906e-05,
      "loss": 0.0008572075515985489,
      "step": 185130
    },
    {
      "epoch": 52.55180244110134,
      "grad_norm": 0.10936997830867767,
      "learning_rate": 4.747005393130855e-05,
      "loss": 0.0002504771575331688,
      "step": 185140
    },
    {
      "epoch": 52.55464093102469,
      "grad_norm": 0.1941082924604416,
      "learning_rate": 4.746721544138519e-05,
      "loss": 0.0029945846647024153,
      "step": 185150
    },
    {
      "epoch": 52.557479420948056,
      "grad_norm": 0.05512005090713501,
      "learning_rate": 4.7464376951461823e-05,
      "loss": 0.00023591574281454087,
      "step": 185160
    },
    {
      "epoch": 52.56031791087142,
      "grad_norm": 0.045048169791698456,
      "learning_rate": 4.7461538461538465e-05,
      "loss": 0.000295221246778965,
      "step": 185170
    },
    {
      "epoch": 52.563156400794774,
      "grad_norm": 0.03022882342338562,
      "learning_rate": 4.74586999716151e-05,
      "loss": 0.00033792518079280855,
      "step": 185180
    },
    {
      "epoch": 52.56599489071814,
      "grad_norm": 0.3429563045501709,
      "learning_rate": 4.745586148169174e-05,
      "loss": 0.000784645788371563,
      "step": 185190
    },
    {
      "epoch": 52.5688333806415,
      "grad_norm": 0.3609480559825897,
      "learning_rate": 4.745302299176838e-05,
      "loss": 0.003713177889585495,
      "step": 185200
    },
    {
      "epoch": 52.571671870564856,
      "grad_norm": 0.2989714741706848,
      "learning_rate": 4.745018450184502e-05,
      "loss": 0.0015159361064434052,
      "step": 185210
    },
    {
      "epoch": 52.57451036048822,
      "grad_norm": 0.20835088193416595,
      "learning_rate": 4.7447346011921665e-05,
      "loss": 0.0003001958131790161,
      "step": 185220
    },
    {
      "epoch": 52.57734885041158,
      "grad_norm": 0.2202584594488144,
      "learning_rate": 4.74445075219983e-05,
      "loss": 0.0003246534615755081,
      "step": 185230
    },
    {
      "epoch": 52.580187340334945,
      "grad_norm": 0.21264386177062988,
      "learning_rate": 4.7441669032074934e-05,
      "loss": 0.000213053822517395,
      "step": 185240
    },
    {
      "epoch": 52.5830258302583,
      "grad_norm": 0.27794283628463745,
      "learning_rate": 4.743883054215158e-05,
      "loss": 0.0003891423344612122,
      "step": 185250
    },
    {
      "epoch": 52.58586432018166,
      "grad_norm": 0.011930675245821476,
      "learning_rate": 4.743599205222822e-05,
      "loss": 0.0009061414748430252,
      "step": 185260
    },
    {
      "epoch": 52.588702810105026,
      "grad_norm": 0.6409693956375122,
      "learning_rate": 4.743315356230486e-05,
      "loss": 0.0006713824346661568,
      "step": 185270
    },
    {
      "epoch": 52.59154130002838,
      "grad_norm": 0.23920466005802155,
      "learning_rate": 4.743031507238149e-05,
      "loss": 0.0007942656055092811,
      "step": 185280
    },
    {
      "epoch": 52.594379789951745,
      "grad_norm": 0.13893775641918182,
      "learning_rate": 4.7427476582458134e-05,
      "loss": 0.0008586443960666657,
      "step": 185290
    },
    {
      "epoch": 52.59721827987511,
      "grad_norm": 2.483689546585083,
      "learning_rate": 4.7424638092534776e-05,
      "loss": 0.0005111707374453545,
      "step": 185300
    },
    {
      "epoch": 52.60005676979847,
      "grad_norm": 0.0080666933208704,
      "learning_rate": 4.742179960261141e-05,
      "loss": 0.00039760638028383253,
      "step": 185310
    },
    {
      "epoch": 52.60289525972183,
      "grad_norm": 0.021713584661483765,
      "learning_rate": 4.741896111268805e-05,
      "loss": 0.0005626743659377098,
      "step": 185320
    },
    {
      "epoch": 52.60573374964519,
      "grad_norm": 0.0566079318523407,
      "learning_rate": 4.741612262276469e-05,
      "loss": 0.0005209507420659065,
      "step": 185330
    },
    {
      "epoch": 52.60857223956855,
      "grad_norm": 0.019202305004000664,
      "learning_rate": 4.741328413284133e-05,
      "loss": 0.003702656924724579,
      "step": 185340
    },
    {
      "epoch": 52.61141072949191,
      "grad_norm": 1.7712969779968262,
      "learning_rate": 4.741044564291797e-05,
      "loss": 0.0023082386702299117,
      "step": 185350
    },
    {
      "epoch": 52.61424921941527,
      "grad_norm": 0.527144730091095,
      "learning_rate": 4.740760715299461e-05,
      "loss": 0.0005703317001461983,
      "step": 185360
    },
    {
      "epoch": 52.617087709338634,
      "grad_norm": 0.18114162981510162,
      "learning_rate": 4.7404768663071245e-05,
      "loss": 0.0005095869302749634,
      "step": 185370
    },
    {
      "epoch": 52.61992619926199,
      "grad_norm": 0.018949512392282486,
      "learning_rate": 4.7401930173147886e-05,
      "loss": 0.00036325007677078246,
      "step": 185380
    },
    {
      "epoch": 52.62276468918535,
      "grad_norm": 0.02048935927450657,
      "learning_rate": 4.739909168322453e-05,
      "loss": 0.0003137892112135887,
      "step": 185390
    },
    {
      "epoch": 52.625603179108715,
      "grad_norm": 0.08305579423904419,
      "learning_rate": 4.739625319330116e-05,
      "loss": 0.006444612145423889,
      "step": 185400
    },
    {
      "epoch": 52.62844166903208,
      "grad_norm": 0.28879034519195557,
      "learning_rate": 4.7393414703377804e-05,
      "loss": 0.0019057640805840493,
      "step": 185410
    },
    {
      "epoch": 52.631280158955434,
      "grad_norm": 0.027159877121448517,
      "learning_rate": 4.7390576213454445e-05,
      "loss": 0.0006318029016256332,
      "step": 185420
    },
    {
      "epoch": 52.6341186488788,
      "grad_norm": 0.11528678238391876,
      "learning_rate": 4.7387737723531086e-05,
      "loss": 0.0005651568993926048,
      "step": 185430
    },
    {
      "epoch": 52.63695713880216,
      "grad_norm": 0.20741763710975647,
      "learning_rate": 4.738489923360772e-05,
      "loss": 0.0013119548559188843,
      "step": 185440
    },
    {
      "epoch": 52.639795628725516,
      "grad_norm": 0.26731425523757935,
      "learning_rate": 4.738206074368436e-05,
      "loss": 0.0002934100106358528,
      "step": 185450
    },
    {
      "epoch": 52.64263411864888,
      "grad_norm": 0.03198122978210449,
      "learning_rate": 4.7379222253761004e-05,
      "loss": 0.001993107609450817,
      "step": 185460
    },
    {
      "epoch": 52.64547260857224,
      "grad_norm": 0.03491595387458801,
      "learning_rate": 4.737638376383764e-05,
      "loss": 0.0005677128210663795,
      "step": 185470
    },
    {
      "epoch": 52.6483110984956,
      "grad_norm": 0.09818445891141891,
      "learning_rate": 4.737354527391428e-05,
      "loss": 0.0006580352783203125,
      "step": 185480
    },
    {
      "epoch": 52.65114958841896,
      "grad_norm": 0.3002893626689911,
      "learning_rate": 4.737070678399092e-05,
      "loss": 0.007006453722715378,
      "step": 185490
    },
    {
      "epoch": 52.65398807834232,
      "grad_norm": 0.10338694602251053,
      "learning_rate": 4.7367868294067556e-05,
      "loss": 0.0080448217689991,
      "step": 185500
    },
    {
      "epoch": 52.65398807834232,
      "eval_accuracy": 0.9743752781840147,
      "eval_loss": 0.0890604555606842,
      "eval_runtime": 32.6632,
      "eval_samples_per_second": 481.489,
      "eval_steps_per_second": 7.531,
      "step": 185500
    },
    {
      "epoch": 52.656826568265686,
      "grad_norm": 0.12359783053398132,
      "learning_rate": 4.73650298041442e-05,
      "loss": 0.0018954411149024963,
      "step": 185510
    },
    {
      "epoch": 52.65966505818904,
      "grad_norm": 0.02318534255027771,
      "learning_rate": 4.736219131422084e-05,
      "loss": 0.0003147747367620468,
      "step": 185520
    },
    {
      "epoch": 52.662503548112404,
      "grad_norm": 3.2030482292175293,
      "learning_rate": 4.735935282429747e-05,
      "loss": 0.0017232675105333327,
      "step": 185530
    },
    {
      "epoch": 52.66534203803577,
      "grad_norm": 0.02708360180258751,
      "learning_rate": 4.7356514334374114e-05,
      "loss": 0.0010398253798484802,
      "step": 185540
    },
    {
      "epoch": 52.66818052795912,
      "grad_norm": 0.45479533076286316,
      "learning_rate": 4.7353675844450756e-05,
      "loss": 0.00043730009347200396,
      "step": 185550
    },
    {
      "epoch": 52.671019017882486,
      "grad_norm": 0.03316853940486908,
      "learning_rate": 4.73508373545274e-05,
      "loss": 0.0005436841398477555,
      "step": 185560
    },
    {
      "epoch": 52.67385750780585,
      "grad_norm": 0.036265190690755844,
      "learning_rate": 4.734799886460403e-05,
      "loss": 0.00048191547393798827,
      "step": 185570
    },
    {
      "epoch": 52.676695997729205,
      "grad_norm": 0.058726515620946884,
      "learning_rate": 4.734516037468067e-05,
      "loss": 0.004901925846934318,
      "step": 185580
    },
    {
      "epoch": 52.67953448765257,
      "grad_norm": 0.02433466538786888,
      "learning_rate": 4.7342321884757314e-05,
      "loss": 0.0009424816817045212,
      "step": 185590
    },
    {
      "epoch": 52.68237297757593,
      "grad_norm": 0.024017754942178726,
      "learning_rate": 4.733948339483395e-05,
      "loss": 0.00549938939511776,
      "step": 185600
    },
    {
      "epoch": 52.68521146749929,
      "grad_norm": 2.948828935623169,
      "learning_rate": 4.733664490491059e-05,
      "loss": 0.000913066603243351,
      "step": 185610
    },
    {
      "epoch": 52.68804995742265,
      "grad_norm": 0.25217127799987793,
      "learning_rate": 4.733380641498723e-05,
      "loss": 0.01155783236026764,
      "step": 185620
    },
    {
      "epoch": 52.69088844734601,
      "grad_norm": 3.9449269771575928,
      "learning_rate": 4.7330967925063866e-05,
      "loss": 0.001362660340964794,
      "step": 185630
    },
    {
      "epoch": 52.693726937269375,
      "grad_norm": 0.01123631838709116,
      "learning_rate": 4.732812943514051e-05,
      "loss": 0.0014895929023623466,
      "step": 185640
    },
    {
      "epoch": 52.69656542719273,
      "grad_norm": 0.9017177820205688,
      "learning_rate": 4.732529094521715e-05,
      "loss": 0.0006215654313564301,
      "step": 185650
    },
    {
      "epoch": 52.69940391711609,
      "grad_norm": 0.30196502804756165,
      "learning_rate": 4.7322452455293784e-05,
      "loss": 0.004408921301364899,
      "step": 185660
    },
    {
      "epoch": 52.702242407039456,
      "grad_norm": 9.108743667602539,
      "learning_rate": 4.7319613965370425e-05,
      "loss": 0.0025856684893369674,
      "step": 185670
    },
    {
      "epoch": 52.70508089696281,
      "grad_norm": 0.05668089911341667,
      "learning_rate": 4.7316775475447066e-05,
      "loss": 0.0004437718540430069,
      "step": 185680
    },
    {
      "epoch": 52.707919386886175,
      "grad_norm": 2.226569890975952,
      "learning_rate": 4.731393698552371e-05,
      "loss": 0.0009616909548640252,
      "step": 185690
    },
    {
      "epoch": 52.71075787680954,
      "grad_norm": 1.040649652481079,
      "learning_rate": 4.731109849560034e-05,
      "loss": 0.0008955013006925582,
      "step": 185700
    },
    {
      "epoch": 52.7135963667329,
      "grad_norm": 0.026919545605778694,
      "learning_rate": 4.730826000567698e-05,
      "loss": 0.0006707839667797089,
      "step": 185710
    },
    {
      "epoch": 52.71643485665626,
      "grad_norm": 0.1344549059867859,
      "learning_rate": 4.7305421515753625e-05,
      "loss": 0.0028807993978261946,
      "step": 185720
    },
    {
      "epoch": 52.71927334657962,
      "grad_norm": 0.2006743848323822,
      "learning_rate": 4.730258302583026e-05,
      "loss": 0.0011252373456954955,
      "step": 185730
    },
    {
      "epoch": 52.72211183650298,
      "grad_norm": 0.04570501297712326,
      "learning_rate": 4.72997445359069e-05,
      "loss": 0.0011382753029465676,
      "step": 185740
    },
    {
      "epoch": 52.72495032642634,
      "grad_norm": 0.018986402079463005,
      "learning_rate": 4.729690604598354e-05,
      "loss": 0.0004973847419023514,
      "step": 185750
    },
    {
      "epoch": 52.7277888163497,
      "grad_norm": 0.41940921545028687,
      "learning_rate": 4.729406755606018e-05,
      "loss": 0.0007176132872700691,
      "step": 185760
    },
    {
      "epoch": 52.730627306273064,
      "grad_norm": 0.08683253824710846,
      "learning_rate": 4.729122906613682e-05,
      "loss": 0.0006362447515130043,
      "step": 185770
    },
    {
      "epoch": 52.73346579619643,
      "grad_norm": 0.04611833766102791,
      "learning_rate": 4.728839057621346e-05,
      "loss": 0.0012508610263466835,
      "step": 185780
    },
    {
      "epoch": 52.73630428611978,
      "grad_norm": 1.2595043182373047,
      "learning_rate": 4.7285552086290094e-05,
      "loss": 0.0008790530264377594,
      "step": 185790
    },
    {
      "epoch": 52.739142776043145,
      "grad_norm": 0.06525368243455887,
      "learning_rate": 4.7282713596366736e-05,
      "loss": 0.0012218631803989411,
      "step": 185800
    },
    {
      "epoch": 52.74198126596651,
      "grad_norm": 2.1024703979492188,
      "learning_rate": 4.727987510644337e-05,
      "loss": 0.0010576851665973664,
      "step": 185810
    },
    {
      "epoch": 52.744819755889864,
      "grad_norm": 2.0177900791168213,
      "learning_rate": 4.727703661652001e-05,
      "loss": 0.006474721431732178,
      "step": 185820
    },
    {
      "epoch": 52.74765824581323,
      "grad_norm": 0.0555436797440052,
      "learning_rate": 4.727419812659665e-05,
      "loss": 0.0025393694639205934,
      "step": 185830
    },
    {
      "epoch": 52.75049673573659,
      "grad_norm": 0.5443676114082336,
      "learning_rate": 4.727135963667329e-05,
      "loss": 0.005507834255695343,
      "step": 185840
    },
    {
      "epoch": 52.753335225659946,
      "grad_norm": 0.05759714916348457,
      "learning_rate": 4.7268521146749936e-05,
      "loss": 0.0024169694632291796,
      "step": 185850
    },
    {
      "epoch": 52.75617371558331,
      "grad_norm": 0.45984742045402527,
      "learning_rate": 4.726568265682657e-05,
      "loss": 0.0012779790908098222,
      "step": 185860
    },
    {
      "epoch": 52.75901220550667,
      "grad_norm": 0.579883337020874,
      "learning_rate": 4.7262844166903205e-05,
      "loss": 0.0009206749498844146,
      "step": 185870
    },
    {
      "epoch": 52.761850695430034,
      "grad_norm": 1.9006707668304443,
      "learning_rate": 4.726000567697985e-05,
      "loss": 0.001442374475300312,
      "step": 185880
    },
    {
      "epoch": 52.76468918535339,
      "grad_norm": 0.6051242351531982,
      "learning_rate": 4.725716718705649e-05,
      "loss": 0.00028705839067697523,
      "step": 185890
    },
    {
      "epoch": 52.76752767527675,
      "grad_norm": 0.09879337251186371,
      "learning_rate": 4.725432869713313e-05,
      "loss": 0.0014525072649121285,
      "step": 185900
    },
    {
      "epoch": 52.770366165200116,
      "grad_norm": 0.2563883066177368,
      "learning_rate": 4.7251490207209764e-05,
      "loss": 0.0006884919479489326,
      "step": 185910
    },
    {
      "epoch": 52.77320465512347,
      "grad_norm": 0.02914213202893734,
      "learning_rate": 4.7248651717286405e-05,
      "loss": 0.0013490477576851845,
      "step": 185920
    },
    {
      "epoch": 52.776043145046835,
      "grad_norm": 0.0796378031373024,
      "learning_rate": 4.724581322736305e-05,
      "loss": 0.0013372285291552543,
      "step": 185930
    },
    {
      "epoch": 52.7788816349702,
      "grad_norm": 0.046775925904512405,
      "learning_rate": 4.724297473743968e-05,
      "loss": 0.0010764306411147118,
      "step": 185940
    },
    {
      "epoch": 52.78172012489355,
      "grad_norm": 1.439765214920044,
      "learning_rate": 4.724013624751632e-05,
      "loss": 0.005131819099187851,
      "step": 185950
    },
    {
      "epoch": 52.784558614816916,
      "grad_norm": 0.10606741160154343,
      "learning_rate": 4.7237297757592964e-05,
      "loss": 0.0022700073197484015,
      "step": 185960
    },
    {
      "epoch": 52.78739710474028,
      "grad_norm": 0.3596985340118408,
      "learning_rate": 4.72344592676696e-05,
      "loss": 0.0007230114191770553,
      "step": 185970
    },
    {
      "epoch": 52.79023559466364,
      "grad_norm": 0.07961712777614594,
      "learning_rate": 4.723162077774625e-05,
      "loss": 0.0010866424068808555,
      "step": 185980
    },
    {
      "epoch": 52.793074084587,
      "grad_norm": 0.04963191598653793,
      "learning_rate": 4.722878228782288e-05,
      "loss": 0.00038391612470149993,
      "step": 185990
    },
    {
      "epoch": 52.79591257451036,
      "grad_norm": 0.7350358366966248,
      "learning_rate": 4.7225943797899516e-05,
      "loss": 0.0004122616723179817,
      "step": 186000
    },
    {
      "epoch": 52.79591257451036,
      "eval_accuracy": 0.9782539581611241,
      "eval_loss": 0.07835876941680908,
      "eval_runtime": 32.8036,
      "eval_samples_per_second": 479.43,
      "eval_steps_per_second": 7.499,
      "step": 186000
    },
    {
      "epoch": 52.79875106443372,
      "grad_norm": 0.43226468563079834,
      "learning_rate": 4.722310530797616e-05,
      "loss": 0.000393989123404026,
      "step": 186010
    },
    {
      "epoch": 52.80158955435708,
      "grad_norm": 0.5629341006278992,
      "learning_rate": 4.72202668180528e-05,
      "loss": 0.006162229925394058,
      "step": 186020
    },
    {
      "epoch": 52.80442804428044,
      "grad_norm": 0.042067691683769226,
      "learning_rate": 4.721742832812944e-05,
      "loss": 0.0013523841276764869,
      "step": 186030
    },
    {
      "epoch": 52.807266534203805,
      "grad_norm": 0.07040529698133469,
      "learning_rate": 4.7214589838206075e-05,
      "loss": 0.0010930126532912253,
      "step": 186040
    },
    {
      "epoch": 52.81010502412717,
      "grad_norm": 0.39801710844039917,
      "learning_rate": 4.7211751348282716e-05,
      "loss": 0.00594882033765316,
      "step": 186050
    },
    {
      "epoch": 52.812943514050524,
      "grad_norm": 3.3025922775268555,
      "learning_rate": 4.720891285835936e-05,
      "loss": 0.0012652017176151276,
      "step": 186060
    },
    {
      "epoch": 52.81578200397389,
      "grad_norm": 5.5879011154174805,
      "learning_rate": 4.720607436843599e-05,
      "loss": 0.0012182939797639846,
      "step": 186070
    },
    {
      "epoch": 52.81862049389725,
      "grad_norm": 0.1153416708111763,
      "learning_rate": 4.720323587851263e-05,
      "loss": 0.001368202455341816,
      "step": 186080
    },
    {
      "epoch": 52.821458983820605,
      "grad_norm": 1.0654881000518799,
      "learning_rate": 4.7200397388589275e-05,
      "loss": 0.0005781034007668496,
      "step": 186090
    },
    {
      "epoch": 52.82429747374397,
      "grad_norm": 1.574164867401123,
      "learning_rate": 4.719755889866591e-05,
      "loss": 0.0009738141670823097,
      "step": 186100
    },
    {
      "epoch": 52.82713596366733,
      "grad_norm": 0.41416653990745544,
      "learning_rate": 4.719472040874255e-05,
      "loss": 0.0009036064147949219,
      "step": 186110
    },
    {
      "epoch": 52.82997445359069,
      "grad_norm": 0.10472802072763443,
      "learning_rate": 4.719188191881919e-05,
      "loss": 0.0007203890010714531,
      "step": 186120
    },
    {
      "epoch": 52.83281294351405,
      "grad_norm": 0.1843029409646988,
      "learning_rate": 4.718904342889583e-05,
      "loss": 0.0010707173496484756,
      "step": 186130
    },
    {
      "epoch": 52.83565143343741,
      "grad_norm": 0.47467759251594543,
      "learning_rate": 4.718620493897247e-05,
      "loss": 0.0006647901609539986,
      "step": 186140
    },
    {
      "epoch": 52.838489923360775,
      "grad_norm": 0.9722204208374023,
      "learning_rate": 4.718336644904911e-05,
      "loss": 0.0007989989593625068,
      "step": 186150
    },
    {
      "epoch": 52.84132841328413,
      "grad_norm": 0.036686379462480545,
      "learning_rate": 4.718052795912575e-05,
      "loss": 0.0006733773276209831,
      "step": 186160
    },
    {
      "epoch": 52.844166903207494,
      "grad_norm": 4.554087162017822,
      "learning_rate": 4.7177689469202385e-05,
      "loss": 0.0032998915761709214,
      "step": 186170
    },
    {
      "epoch": 52.84700539313086,
      "grad_norm": 0.0486115962266922,
      "learning_rate": 4.717485097927903e-05,
      "loss": 0.001992819271981716,
      "step": 186180
    },
    {
      "epoch": 52.84984388305421,
      "grad_norm": 1.5523966550827026,
      "learning_rate": 4.717201248935567e-05,
      "loss": 0.0010800985619425774,
      "step": 186190
    },
    {
      "epoch": 52.852682372977576,
      "grad_norm": 0.2830045819282532,
      "learning_rate": 4.71691739994323e-05,
      "loss": 0.0015748050063848495,
      "step": 186200
    },
    {
      "epoch": 52.85552086290094,
      "grad_norm": 0.33962830901145935,
      "learning_rate": 4.7166335509508944e-05,
      "loss": 0.0003678485751152039,
      "step": 186210
    },
    {
      "epoch": 52.858359352824294,
      "grad_norm": 0.0537160262465477,
      "learning_rate": 4.7163497019585586e-05,
      "loss": 0.0017336517572402953,
      "step": 186220
    },
    {
      "epoch": 52.86119784274766,
      "grad_norm": 0.04691077023744583,
      "learning_rate": 4.716065852966222e-05,
      "loss": 0.0009140096604824066,
      "step": 186230
    },
    {
      "epoch": 52.86403633267102,
      "grad_norm": 0.6862740516662598,
      "learning_rate": 4.715782003973886e-05,
      "loss": 0.0038388464599847794,
      "step": 186240
    },
    {
      "epoch": 52.86687482259438,
      "grad_norm": 0.05020619183778763,
      "learning_rate": 4.71549815498155e-05,
      "loss": 0.0009594792500138283,
      "step": 186250
    },
    {
      "epoch": 52.86971331251774,
      "grad_norm": 0.07432413846254349,
      "learning_rate": 4.715214305989214e-05,
      "loss": 0.0010832717642188072,
      "step": 186260
    },
    {
      "epoch": 52.8725518024411,
      "grad_norm": 0.5585078597068787,
      "learning_rate": 4.714930456996878e-05,
      "loss": 0.0011963484808802604,
      "step": 186270
    },
    {
      "epoch": 52.875390292364465,
      "grad_norm": 0.052770067006349564,
      "learning_rate": 4.714646608004542e-05,
      "loss": 0.0014254337176680565,
      "step": 186280
    },
    {
      "epoch": 52.87822878228782,
      "grad_norm": 0.11159157752990723,
      "learning_rate": 4.7143627590122055e-05,
      "loss": 0.0004890399053692817,
      "step": 186290
    },
    {
      "epoch": 52.88106727221118,
      "grad_norm": 1.0920945405960083,
      "learning_rate": 4.7140789100198696e-05,
      "loss": 0.001100941002368927,
      "step": 186300
    },
    {
      "epoch": 52.883905762134546,
      "grad_norm": 0.03142758458852768,
      "learning_rate": 4.713795061027533e-05,
      "loss": 0.0003737971186637878,
      "step": 186310
    },
    {
      "epoch": 52.8867442520579,
      "grad_norm": 0.7031912803649902,
      "learning_rate": 4.713511212035198e-05,
      "loss": 0.0006695404648780823,
      "step": 186320
    },
    {
      "epoch": 52.889582741981265,
      "grad_norm": 0.29146626591682434,
      "learning_rate": 4.7132273630428614e-05,
      "loss": 0.00019972715526819229,
      "step": 186330
    },
    {
      "epoch": 52.89242123190463,
      "grad_norm": 0.2578912079334259,
      "learning_rate": 4.712943514050525e-05,
      "loss": 0.0015983432531356812,
      "step": 186340
    },
    {
      "epoch": 52.89525972182799,
      "grad_norm": 0.26646336913108826,
      "learning_rate": 4.7126596650581896e-05,
      "loss": 0.0025143550708889963,
      "step": 186350
    },
    {
      "epoch": 52.898098211751346,
      "grad_norm": 0.03392914682626724,
      "learning_rate": 4.712375816065853e-05,
      "loss": 0.002839016355574131,
      "step": 186360
    },
    {
      "epoch": 52.90093670167471,
      "grad_norm": 0.5476954579353333,
      "learning_rate": 4.712091967073517e-05,
      "loss": 0.0005209110677242279,
      "step": 186370
    },
    {
      "epoch": 52.90377519159807,
      "grad_norm": 0.406361848115921,
      "learning_rate": 4.7118081180811814e-05,
      "loss": 0.0008580882102251052,
      "step": 186380
    },
    {
      "epoch": 52.90661368152143,
      "grad_norm": 0.06669175624847412,
      "learning_rate": 4.711524269088845e-05,
      "loss": 0.0032583050429821014,
      "step": 186390
    },
    {
      "epoch": 52.90945217144479,
      "grad_norm": 0.6824454069137573,
      "learning_rate": 4.711240420096509e-05,
      "loss": 0.0007803093641996384,
      "step": 186400
    },
    {
      "epoch": 52.912290661368154,
      "grad_norm": 0.14807383716106415,
      "learning_rate": 4.7109565711041724e-05,
      "loss": 0.00147513747215271,
      "step": 186410
    },
    {
      "epoch": 52.91512915129151,
      "grad_norm": 0.08323166519403458,
      "learning_rate": 4.7106727221118366e-05,
      "loss": 0.00987521857023239,
      "step": 186420
    },
    {
      "epoch": 52.91796764121487,
      "grad_norm": 0.03487098217010498,
      "learning_rate": 4.710388873119501e-05,
      "loss": 0.0008615966886281967,
      "step": 186430
    },
    {
      "epoch": 52.920806131138235,
      "grad_norm": 0.07758292555809021,
      "learning_rate": 4.710105024127164e-05,
      "loss": 0.0007901865988969803,
      "step": 186440
    },
    {
      "epoch": 52.9236446210616,
      "grad_norm": 0.40486881136894226,
      "learning_rate": 4.709821175134829e-05,
      "loss": 0.0014066126197576523,
      "step": 186450
    },
    {
      "epoch": 52.926483110984954,
      "grad_norm": 0.09714283794164658,
      "learning_rate": 4.7095373261424924e-05,
      "loss": 0.0012796740978956223,
      "step": 186460
    },
    {
      "epoch": 52.92932160090832,
      "grad_norm": 0.5464173555374146,
      "learning_rate": 4.709253477150156e-05,
      "loss": 0.0073540061712265015,
      "step": 186470
    },
    {
      "epoch": 52.93216009083168,
      "grad_norm": 1.4453067779541016,
      "learning_rate": 4.708969628157821e-05,
      "loss": 0.011840809881687165,
      "step": 186480
    },
    {
      "epoch": 52.934998580755035,
      "grad_norm": 0.15751346945762634,
      "learning_rate": 4.708685779165484e-05,
      "loss": 0.002746947854757309,
      "step": 186490
    },
    {
      "epoch": 52.9378370706784,
      "grad_norm": 3.499262809753418,
      "learning_rate": 4.708401930173148e-05,
      "loss": 0.002546422928571701,
      "step": 186500
    },
    {
      "epoch": 52.9378370706784,
      "eval_accuracy": 0.976918674890316,
      "eval_loss": 0.08579359203577042,
      "eval_runtime": 32.0221,
      "eval_samples_per_second": 491.129,
      "eval_steps_per_second": 7.682,
      "step": 186500
    },
    {
      "epoch": 52.94067556060176,
      "grad_norm": 0.12745369970798492,
      "learning_rate": 4.708118081180812e-05,
      "loss": 0.0011678164824843406,
      "step": 186510
    },
    {
      "epoch": 52.943514050525124,
      "grad_norm": 0.19829964637756348,
      "learning_rate": 4.707834232188476e-05,
      "loss": 0.003963455557823181,
      "step": 186520
    },
    {
      "epoch": 52.94635254044848,
      "grad_norm": 1.4347254037857056,
      "learning_rate": 4.70755038319614e-05,
      "loss": 0.001419261284172535,
      "step": 186530
    },
    {
      "epoch": 52.94919103037184,
      "grad_norm": 0.24300187826156616,
      "learning_rate": 4.7072665342038035e-05,
      "loss": 0.009653251618146896,
      "step": 186540
    },
    {
      "epoch": 52.952029520295206,
      "grad_norm": 0.28136777877807617,
      "learning_rate": 4.7069826852114676e-05,
      "loss": 0.0016790073364973067,
      "step": 186550
    },
    {
      "epoch": 52.95486801021856,
      "grad_norm": 0.1163061112165451,
      "learning_rate": 4.706698836219132e-05,
      "loss": 0.0007288211956620217,
      "step": 186560
    },
    {
      "epoch": 52.957706500141924,
      "grad_norm": 0.14876312017440796,
      "learning_rate": 4.706414987226795e-05,
      "loss": 0.002441161312162876,
      "step": 186570
    },
    {
      "epoch": 52.96054499006529,
      "grad_norm": 0.017498524859547615,
      "learning_rate": 4.706159523133693e-05,
      "loss": 0.01781250536441803,
      "step": 186580
    },
    {
      "epoch": 52.96338347998864,
      "grad_norm": 0.38986697793006897,
      "learning_rate": 4.705875674141357e-05,
      "loss": 0.0005735533311963081,
      "step": 186590
    },
    {
      "epoch": 52.966221969912006,
      "grad_norm": 0.26328518986701965,
      "learning_rate": 4.705591825149021e-05,
      "loss": 0.0020713524892926214,
      "step": 186600
    },
    {
      "epoch": 52.96906045983537,
      "grad_norm": 0.4067014455795288,
      "learning_rate": 4.705307976156685e-05,
      "loss": 0.0003866251558065414,
      "step": 186610
    },
    {
      "epoch": 52.97189894975873,
      "grad_norm": 0.03233945742249489,
      "learning_rate": 4.7050241271643484e-05,
      "loss": 0.0003457270562648773,
      "step": 186620
    },
    {
      "epoch": 52.97473743968209,
      "grad_norm": 0.48162275552749634,
      "learning_rate": 4.704740278172013e-05,
      "loss": 0.0005493048578500747,
      "step": 186630
    },
    {
      "epoch": 52.97757592960545,
      "grad_norm": 0.052287984639406204,
      "learning_rate": 4.704456429179677e-05,
      "loss": 0.0007736720144748688,
      "step": 186640
    },
    {
      "epoch": 52.98041441952881,
      "grad_norm": 0.1726934015750885,
      "learning_rate": 4.70417258018734e-05,
      "loss": 0.003924472630023957,
      "step": 186650
    },
    {
      "epoch": 52.98325290945217,
      "grad_norm": 0.05058598145842552,
      "learning_rate": 4.703888731195005e-05,
      "loss": 0.0014989161863923073,
      "step": 186660
    },
    {
      "epoch": 52.98609139937553,
      "grad_norm": 0.7435520887374878,
      "learning_rate": 4.7036048822026684e-05,
      "loss": 0.0009860554710030556,
      "step": 186670
    },
    {
      "epoch": 52.988929889298895,
      "grad_norm": 0.12682199478149414,
      "learning_rate": 4.7033210332103326e-05,
      "loss": 0.0006713606417179108,
      "step": 186680
    },
    {
      "epoch": 52.99176837922225,
      "grad_norm": 0.0371667705476284,
      "learning_rate": 4.703037184217996e-05,
      "loss": 0.007192297279834748,
      "step": 186690
    },
    {
      "epoch": 52.99460686914561,
      "grad_norm": 0.0615924634039402,
      "learning_rate": 4.70275333522566e-05,
      "loss": 0.0004767684265971184,
      "step": 186700
    },
    {
      "epoch": 52.997445359068976,
      "grad_norm": 3.853525161743164,
      "learning_rate": 4.702469486233324e-05,
      "loss": 0.0024151716381311415,
      "step": 186710
    },
    {
      "epoch": 53.00028384899234,
      "grad_norm": 0.3753420114517212,
      "learning_rate": 4.702185637240988e-05,
      "loss": 0.0005496107041835785,
      "step": 186720
    },
    {
      "epoch": 53.003122338915695,
      "grad_norm": 2.4691920280456543,
      "learning_rate": 4.701901788248652e-05,
      "loss": 0.0016416231170296668,
      "step": 186730
    },
    {
      "epoch": 53.00596082883906,
      "grad_norm": 0.4115088880062103,
      "learning_rate": 4.701617939256316e-05,
      "loss": 0.0009204635396599769,
      "step": 186740
    },
    {
      "epoch": 53.00879931876242,
      "grad_norm": 5.280255317687988,
      "learning_rate": 4.7013340902639795e-05,
      "loss": 0.0010163282975554466,
      "step": 186750
    },
    {
      "epoch": 53.01163780868578,
      "grad_norm": 0.16813445091247559,
      "learning_rate": 4.7010502412716436e-05,
      "loss": 0.0009657301008701325,
      "step": 186760
    },
    {
      "epoch": 53.01447629860914,
      "grad_norm": 0.02061653323471546,
      "learning_rate": 4.700766392279308e-05,
      "loss": 0.0038428518921136854,
      "step": 186770
    },
    {
      "epoch": 53.0173147885325,
      "grad_norm": 0.0222296342253685,
      "learning_rate": 4.700482543286971e-05,
      "loss": 0.0005093691870570183,
      "step": 186780
    },
    {
      "epoch": 53.02015327845586,
      "grad_norm": 0.031990401446819305,
      "learning_rate": 4.7001986942946354e-05,
      "loss": 0.0014039330184459687,
      "step": 186790
    },
    {
      "epoch": 53.02299176837922,
      "grad_norm": 0.11974051594734192,
      "learning_rate": 4.6999148453022995e-05,
      "loss": 0.0011685620993375778,
      "step": 186800
    },
    {
      "epoch": 53.025830258302584,
      "grad_norm": 0.02388240024447441,
      "learning_rate": 4.699630996309963e-05,
      "loss": 0.0014687377959489822,
      "step": 186810
    },
    {
      "epoch": 53.02866874822595,
      "grad_norm": 0.030611790716648102,
      "learning_rate": 4.699347147317627e-05,
      "loss": 0.00036798454821109774,
      "step": 186820
    },
    {
      "epoch": 53.0315072381493,
      "grad_norm": 0.2679491937160492,
      "learning_rate": 4.699063298325291e-05,
      "loss": 0.0007893437519669533,
      "step": 186830
    },
    {
      "epoch": 53.034345728072665,
      "grad_norm": 0.6006653308868408,
      "learning_rate": 4.6987794493329554e-05,
      "loss": 0.0006616206839680672,
      "step": 186840
    },
    {
      "epoch": 53.03718421799603,
      "grad_norm": 0.01810004934668541,
      "learning_rate": 4.698495600340619e-05,
      "loss": 0.0002793140709400177,
      "step": 186850
    },
    {
      "epoch": 53.040022707919384,
      "grad_norm": 0.007617782801389694,
      "learning_rate": 4.698211751348283e-05,
      "loss": 0.0007954850792884826,
      "step": 186860
    },
    {
      "epoch": 53.04286119784275,
      "grad_norm": 1.574866771697998,
      "learning_rate": 4.697927902355947e-05,
      "loss": 0.0006692333146929741,
      "step": 186870
    },
    {
      "epoch": 53.04569968776611,
      "grad_norm": 0.06003246083855629,
      "learning_rate": 4.6976440533636106e-05,
      "loss": 0.002177863568067551,
      "step": 186880
    },
    {
      "epoch": 53.04853817768947,
      "grad_norm": 0.48628848791122437,
      "learning_rate": 4.697360204371275e-05,
      "loss": 0.0011009020730853082,
      "step": 186890
    },
    {
      "epoch": 53.05137666761283,
      "grad_norm": 0.17109133303165436,
      "learning_rate": 4.697076355378939e-05,
      "loss": 0.0006325438618659974,
      "step": 186900
    },
    {
      "epoch": 53.05421515753619,
      "grad_norm": 0.12658186256885529,
      "learning_rate": 4.696792506386602e-05,
      "loss": 0.00023299697786569596,
      "step": 186910
    },
    {
      "epoch": 53.057053647459554,
      "grad_norm": 0.09961030632257462,
      "learning_rate": 4.6965086573942664e-05,
      "loss": 0.0007283028215169906,
      "step": 186920
    },
    {
      "epoch": 53.05989213738291,
      "grad_norm": 0.05417123809456825,
      "learning_rate": 4.6962248084019306e-05,
      "loss": 0.0003538798540830612,
      "step": 186930
    },
    {
      "epoch": 53.06273062730627,
      "grad_norm": 0.13278192281723022,
      "learning_rate": 4.695940959409594e-05,
      "loss": 0.0035180777311325075,
      "step": 186940
    },
    {
      "epoch": 53.065569117229636,
      "grad_norm": 0.12029147148132324,
      "learning_rate": 4.695657110417258e-05,
      "loss": 0.006178952753543854,
      "step": 186950
    },
    {
      "epoch": 53.06840760715299,
      "grad_norm": 0.031747233122587204,
      "learning_rate": 4.695373261424922e-05,
      "loss": 0.001958624832332134,
      "step": 186960
    },
    {
      "epoch": 53.071246097076354,
      "grad_norm": 0.08547956496477127,
      "learning_rate": 4.6950894124325865e-05,
      "loss": 0.00025147367268800734,
      "step": 186970
    },
    {
      "epoch": 53.07408458699972,
      "grad_norm": 1.0183758735656738,
      "learning_rate": 4.69480556344025e-05,
      "loss": 0.00044731739908456805,
      "step": 186980
    },
    {
      "epoch": 53.07692307692308,
      "grad_norm": 0.019348958507180214,
      "learning_rate": 4.6945217144479134e-05,
      "loss": 0.0015370257198810578,
      "step": 186990
    },
    {
      "epoch": 53.079761566846436,
      "grad_norm": 0.04783247411251068,
      "learning_rate": 4.694237865455578e-05,
      "loss": 0.0005344085395336152,
      "step": 187000
    },
    {
      "epoch": 53.079761566846436,
      "eval_accuracy": 0.9713867870541107,
      "eval_loss": 0.11089169234037399,
      "eval_runtime": 31.9512,
      "eval_samples_per_second": 492.22,
      "eval_steps_per_second": 7.699,
      "step": 187000
    },
    {
      "epoch": 53.0826000567698,
      "grad_norm": 20.596805572509766,
      "learning_rate": 4.6939540164632417e-05,
      "loss": 0.011659535020589829,
      "step": 187010
    },
    {
      "epoch": 53.08543854669316,
      "grad_norm": 2.3479671478271484,
      "learning_rate": 4.693670167470906e-05,
      "loss": 0.0012940609827637672,
      "step": 187020
    },
    {
      "epoch": 53.08827703661652,
      "grad_norm": 0.03114408068358898,
      "learning_rate": 4.69338631847857e-05,
      "loss": 0.0002849854528903961,
      "step": 187030
    },
    {
      "epoch": 53.09111552653988,
      "grad_norm": 1.9398494958877563,
      "learning_rate": 4.6931024694862334e-05,
      "loss": 0.0005921779200434684,
      "step": 187040
    },
    {
      "epoch": 53.09395401646324,
      "grad_norm": 0.709852397441864,
      "learning_rate": 4.6928186204938975e-05,
      "loss": 0.0006474701687693596,
      "step": 187050
    },
    {
      "epoch": 53.0967925063866,
      "grad_norm": 0.06263653188943863,
      "learning_rate": 4.6925347715015617e-05,
      "loss": 0.0006272278726100922,
      "step": 187060
    },
    {
      "epoch": 53.09963099630996,
      "grad_norm": 0.014547742903232574,
      "learning_rate": 4.692250922509225e-05,
      "loss": 0.0002681197598576546,
      "step": 187070
    },
    {
      "epoch": 53.102469486233325,
      "grad_norm": 0.03502983972430229,
      "learning_rate": 4.691967073516889e-05,
      "loss": 0.00023582149296998977,
      "step": 187080
    },
    {
      "epoch": 53.10530797615669,
      "grad_norm": 0.01701279543340206,
      "learning_rate": 4.691683224524553e-05,
      "loss": 0.00021918416023254396,
      "step": 187090
    },
    {
      "epoch": 53.10814646608004,
      "grad_norm": 0.4748169481754303,
      "learning_rate": 4.6913993755322175e-05,
      "loss": 0.0004496075212955475,
      "step": 187100
    },
    {
      "epoch": 53.110984956003406,
      "grad_norm": 0.06224515661597252,
      "learning_rate": 4.691115526539881e-05,
      "loss": 0.00013815425336360931,
      "step": 187110
    },
    {
      "epoch": 53.11382344592677,
      "grad_norm": 1.3196110725402832,
      "learning_rate": 4.6908316775475445e-05,
      "loss": 0.00038871299475431444,
      "step": 187120
    },
    {
      "epoch": 53.116661935850125,
      "grad_norm": 0.06916163861751556,
      "learning_rate": 4.690547828555209e-05,
      "loss": 0.0020409394055604935,
      "step": 187130
    },
    {
      "epoch": 53.11950042577349,
      "grad_norm": 0.19408981502056122,
      "learning_rate": 4.690263979562873e-05,
      "loss": 0.0043735247105360035,
      "step": 187140
    },
    {
      "epoch": 53.12233891569685,
      "grad_norm": 0.012711308896541595,
      "learning_rate": 4.689980130570537e-05,
      "loss": 0.0015541860833764077,
      "step": 187150
    },
    {
      "epoch": 53.12517740562021,
      "grad_norm": 0.0476750023663044,
      "learning_rate": 4.689696281578201e-05,
      "loss": 0.00039815939962863924,
      "step": 187160
    },
    {
      "epoch": 53.12801589554357,
      "grad_norm": 0.27740389108657837,
      "learning_rate": 4.6894124325858645e-05,
      "loss": 0.000287252850830555,
      "step": 187170
    },
    {
      "epoch": 53.13085438546693,
      "grad_norm": 0.01572190597653389,
      "learning_rate": 4.6891285835935286e-05,
      "loss": 0.00041622593998909,
      "step": 187180
    },
    {
      "epoch": 53.133692875390295,
      "grad_norm": 0.021731248125433922,
      "learning_rate": 4.688844734601193e-05,
      "loss": 0.0005726967006921768,
      "step": 187190
    },
    {
      "epoch": 53.13653136531365,
      "grad_norm": 0.8053247928619385,
      "learning_rate": 4.688560885608856e-05,
      "loss": 0.0004643416032195091,
      "step": 187200
    },
    {
      "epoch": 53.139369855237014,
      "grad_norm": 0.019197138026356697,
      "learning_rate": 4.68827703661652e-05,
      "loss": 0.000215916708111763,
      "step": 187210
    },
    {
      "epoch": 53.14220834516038,
      "grad_norm": 0.4081684350967407,
      "learning_rate": 4.687993187624184e-05,
      "loss": 0.0004559928551316261,
      "step": 187220
    },
    {
      "epoch": 53.14504683508373,
      "grad_norm": 0.04098568484187126,
      "learning_rate": 4.687709338631848e-05,
      "loss": 0.0008385729044675827,
      "step": 187230
    },
    {
      "epoch": 53.147885325007096,
      "grad_norm": 0.07545971125364304,
      "learning_rate": 4.687425489639512e-05,
      "loss": 0.00023680906742811202,
      "step": 187240
    },
    {
      "epoch": 53.15072381493046,
      "grad_norm": 0.17282487452030182,
      "learning_rate": 4.6871416406471755e-05,
      "loss": 0.0009677525609731674,
      "step": 187250
    },
    {
      "epoch": 53.15356230485382,
      "grad_norm": 0.9702873826026917,
      "learning_rate": 4.6868577916548403e-05,
      "loss": 0.0036900244653224946,
      "step": 187260
    },
    {
      "epoch": 53.15640079477718,
      "grad_norm": 0.52781742811203,
      "learning_rate": 4.686573942662504e-05,
      "loss": 0.0002894740551710129,
      "step": 187270
    },
    {
      "epoch": 53.15923928470054,
      "grad_norm": 0.09344497323036194,
      "learning_rate": 4.686290093670167e-05,
      "loss": 0.0003575272858142853,
      "step": 187280
    },
    {
      "epoch": 53.1620777746239,
      "grad_norm": 0.07455617934465408,
      "learning_rate": 4.686006244677832e-05,
      "loss": 0.00045306365936994555,
      "step": 187290
    },
    {
      "epoch": 53.16491626454726,
      "grad_norm": 10.327901840209961,
      "learning_rate": 4.6857223956854955e-05,
      "loss": 0.0033307231962680815,
      "step": 187300
    },
    {
      "epoch": 53.16775475447062,
      "grad_norm": 0.0852927714586258,
      "learning_rate": 4.68543854669316e-05,
      "loss": 0.0020900283008813856,
      "step": 187310
    },
    {
      "epoch": 53.170593244393984,
      "grad_norm": 0.331105500459671,
      "learning_rate": 4.685154697700823e-05,
      "loss": 0.000512843020260334,
      "step": 187320
    },
    {
      "epoch": 53.17343173431734,
      "grad_norm": 0.14804953336715698,
      "learning_rate": 4.684870848708487e-05,
      "loss": 0.0033414628356695174,
      "step": 187330
    },
    {
      "epoch": 53.1762702242407,
      "grad_norm": 0.027433590963482857,
      "learning_rate": 4.6845869997161514e-05,
      "loss": 0.004519091546535492,
      "step": 187340
    },
    {
      "epoch": 53.179108714164066,
      "grad_norm": 0.04188234731554985,
      "learning_rate": 4.684303150723815e-05,
      "loss": 0.00019401032477617264,
      "step": 187350
    },
    {
      "epoch": 53.18194720408743,
      "grad_norm": 0.43340104818344116,
      "learning_rate": 4.684019301731479e-05,
      "loss": 0.00018005650490522386,
      "step": 187360
    },
    {
      "epoch": 53.184785694010785,
      "grad_norm": 0.005684260278940201,
      "learning_rate": 4.683735452739143e-05,
      "loss": 0.00040181409567594526,
      "step": 187370
    },
    {
      "epoch": 53.18762418393415,
      "grad_norm": 0.043361764401197433,
      "learning_rate": 4.6834516037468066e-05,
      "loss": 0.00025251749902963636,
      "step": 187380
    },
    {
      "epoch": 53.19046267385751,
      "grad_norm": 0.007895417511463165,
      "learning_rate": 4.6831677547544714e-05,
      "loss": 0.0003590365871787071,
      "step": 187390
    },
    {
      "epoch": 53.193301163780866,
      "grad_norm": 0.03264365345239639,
      "learning_rate": 4.682883905762135e-05,
      "loss": 0.00030274651944637296,
      "step": 187400
    },
    {
      "epoch": 53.19613965370423,
      "grad_norm": 0.06636297702789307,
      "learning_rate": 4.6826000567697983e-05,
      "loss": 0.0016117408871650697,
      "step": 187410
    },
    {
      "epoch": 53.19897814362759,
      "grad_norm": 0.9597775936126709,
      "learning_rate": 4.6823162077774625e-05,
      "loss": 0.0017969051375985145,
      "step": 187420
    },
    {
      "epoch": 53.20181663355095,
      "grad_norm": 0.1412876397371292,
      "learning_rate": 4.6820323587851266e-05,
      "loss": 0.0005128502845764161,
      "step": 187430
    },
    {
      "epoch": 53.20465512347431,
      "grad_norm": 0.01800248958170414,
      "learning_rate": 4.681748509792791e-05,
      "loss": 0.005396492034196854,
      "step": 187440
    },
    {
      "epoch": 53.20749361339767,
      "grad_norm": 0.05267363414168358,
      "learning_rate": 4.681464660800454e-05,
      "loss": 0.00023108404129743577,
      "step": 187450
    },
    {
      "epoch": 53.210332103321036,
      "grad_norm": 2.6126646995544434,
      "learning_rate": 4.6811808118081184e-05,
      "loss": 0.000644209235906601,
      "step": 187460
    },
    {
      "epoch": 53.21317059324439,
      "grad_norm": 0.15921775996685028,
      "learning_rate": 4.6808969628157825e-05,
      "loss": 0.0002822166308760643,
      "step": 187470
    },
    {
      "epoch": 53.216009083167755,
      "grad_norm": 2.448462724685669,
      "learning_rate": 4.680613113823446e-05,
      "loss": 0.0008476134389638901,
      "step": 187480
    },
    {
      "epoch": 53.21884757309112,
      "grad_norm": 6.471462726593018,
      "learning_rate": 4.68032926483111e-05,
      "loss": 0.0024997176602482795,
      "step": 187490
    },
    {
      "epoch": 53.221686063014474,
      "grad_norm": 0.12524141371250153,
      "learning_rate": 4.680045415838774e-05,
      "loss": 0.0007732944563031196,
      "step": 187500
    },
    {
      "epoch": 53.221686063014474,
      "eval_accuracy": 0.9753926368665352,
      "eval_loss": 0.09537705034017563,
      "eval_runtime": 32.919,
      "eval_samples_per_second": 477.749,
      "eval_steps_per_second": 7.473,
      "step": 187500
    },
    {
      "epoch": 53.22452455293784,
      "grad_norm": 0.5902913808822632,
      "learning_rate": 4.679761566846438e-05,
      "loss": 0.0014896310865879058,
      "step": 187510
    },
    {
      "epoch": 53.2273630428612,
      "grad_norm": 0.20311912894248962,
      "learning_rate": 4.679477717854102e-05,
      "loss": 0.00042458828538656236,
      "step": 187520
    },
    {
      "epoch": 53.230201532784555,
      "grad_norm": 0.42265260219573975,
      "learning_rate": 4.679193868861766e-05,
      "loss": 0.00038704369217157364,
      "step": 187530
    },
    {
      "epoch": 53.23304002270792,
      "grad_norm": 0.029061755165457726,
      "learning_rate": 4.6789100198694294e-05,
      "loss": 0.0002748539671301842,
      "step": 187540
    },
    {
      "epoch": 53.23587851263128,
      "grad_norm": 0.0276603102684021,
      "learning_rate": 4.6786261708770936e-05,
      "loss": 0.000908457487821579,
      "step": 187550
    },
    {
      "epoch": 53.238717002554644,
      "grad_norm": 0.3567744493484497,
      "learning_rate": 4.678342321884758e-05,
      "loss": 0.0002970263361930847,
      "step": 187560
    },
    {
      "epoch": 53.241555492478,
      "grad_norm": 0.09496394544839859,
      "learning_rate": 4.678058472892422e-05,
      "loss": 0.0023291038349270822,
      "step": 187570
    },
    {
      "epoch": 53.24439398240136,
      "grad_norm": 0.22010588645935059,
      "learning_rate": 4.677774623900085e-05,
      "loss": 0.00023479852825403214,
      "step": 187580
    },
    {
      "epoch": 53.247232472324725,
      "grad_norm": 0.3240836560726166,
      "learning_rate": 4.6774907749077494e-05,
      "loss": 0.0007786054164171218,
      "step": 187590
    },
    {
      "epoch": 53.25007096224808,
      "grad_norm": 0.06833242624998093,
      "learning_rate": 4.6772069259154136e-05,
      "loss": 0.001326226256787777,
      "step": 187600
    },
    {
      "epoch": 53.252909452171444,
      "grad_norm": 0.09214320033788681,
      "learning_rate": 4.676923076923077e-05,
      "loss": 0.00024407096207141877,
      "step": 187610
    },
    {
      "epoch": 53.25574794209481,
      "grad_norm": 9.006688117980957,
      "learning_rate": 4.676639227930741e-05,
      "loss": 0.001843802258372307,
      "step": 187620
    },
    {
      "epoch": 53.25858643201816,
      "grad_norm": 0.016255563125014305,
      "learning_rate": 4.676355378938405e-05,
      "loss": 0.00045562312006950376,
      "step": 187630
    },
    {
      "epoch": 53.261424921941526,
      "grad_norm": 3.0736217498779297,
      "learning_rate": 4.676071529946069e-05,
      "loss": 0.0007492952048778534,
      "step": 187640
    },
    {
      "epoch": 53.26426341186489,
      "grad_norm": 0.0143976965919137,
      "learning_rate": 4.675787680953733e-05,
      "loss": 0.00026582367718219757,
      "step": 187650
    },
    {
      "epoch": 53.26710190178825,
      "grad_norm": 0.08633128553628922,
      "learning_rate": 4.675503831961397e-05,
      "loss": 0.0013951560482382773,
      "step": 187660
    },
    {
      "epoch": 53.26994039171161,
      "grad_norm": 0.07975874841213226,
      "learning_rate": 4.6752199829690605e-05,
      "loss": 0.0010851476341485976,
      "step": 187670
    },
    {
      "epoch": 53.27277888163497,
      "grad_norm": 3.7503960132598877,
      "learning_rate": 4.6749361339767246e-05,
      "loss": 0.0011142130941152572,
      "step": 187680
    },
    {
      "epoch": 53.27561737155833,
      "grad_norm": 0.08227263391017914,
      "learning_rate": 4.674652284984389e-05,
      "loss": 0.0005412077531218529,
      "step": 187690
    },
    {
      "epoch": 53.27845586148169,
      "grad_norm": 3.2993576526641846,
      "learning_rate": 4.674368435992052e-05,
      "loss": 0.003132442757487297,
      "step": 187700
    },
    {
      "epoch": 53.28129435140505,
      "grad_norm": 0.414966881275177,
      "learning_rate": 4.6740845869997164e-05,
      "loss": 0.0006981011480093003,
      "step": 187710
    },
    {
      "epoch": 53.284132841328415,
      "grad_norm": 0.13751912117004395,
      "learning_rate": 4.67380073800738e-05,
      "loss": 0.0011086151003837585,
      "step": 187720
    },
    {
      "epoch": 53.28697133125178,
      "grad_norm": 0.49899041652679443,
      "learning_rate": 4.6735168890150446e-05,
      "loss": 0.0039992030709981915,
      "step": 187730
    },
    {
      "epoch": 53.28980982117513,
      "grad_norm": 0.011489659547805786,
      "learning_rate": 4.673233040022708e-05,
      "loss": 0.0077802278101444244,
      "step": 187740
    },
    {
      "epoch": 53.292648311098496,
      "grad_norm": 0.052054475992918015,
      "learning_rate": 4.6729491910303716e-05,
      "loss": 0.004279228672385216,
      "step": 187750
    },
    {
      "epoch": 53.29548680102186,
      "grad_norm": 0.08907400816679001,
      "learning_rate": 4.6726653420380364e-05,
      "loss": 0.007800862193107605,
      "step": 187760
    },
    {
      "epoch": 53.298325290945215,
      "grad_norm": 4.105244159698486,
      "learning_rate": 4.6723814930457e-05,
      "loss": 0.0018029658123850823,
      "step": 187770
    },
    {
      "epoch": 53.30116378086858,
      "grad_norm": 0.9128137826919556,
      "learning_rate": 4.672097644053364e-05,
      "loss": 0.0006383150815963745,
      "step": 187780
    },
    {
      "epoch": 53.30400227079194,
      "grad_norm": 6.919089317321777,
      "learning_rate": 4.671813795061028e-05,
      "loss": 0.0031527493149042128,
      "step": 187790
    },
    {
      "epoch": 53.306840760715296,
      "grad_norm": 0.04528863728046417,
      "learning_rate": 4.6715299460686916e-05,
      "loss": 0.0020196642726659775,
      "step": 187800
    },
    {
      "epoch": 53.30967925063866,
      "grad_norm": 16.958234786987305,
      "learning_rate": 4.671246097076356e-05,
      "loss": 0.008579444885253907,
      "step": 187810
    },
    {
      "epoch": 53.31251774056202,
      "grad_norm": 0.09051081538200378,
      "learning_rate": 4.670962248084019e-05,
      "loss": 0.0010532019659876823,
      "step": 187820
    },
    {
      "epoch": 53.315356230485385,
      "grad_norm": 1.0442044734954834,
      "learning_rate": 4.670678399091683e-05,
      "loss": 0.002135928347706795,
      "step": 187830
    },
    {
      "epoch": 53.31819472040874,
      "grad_norm": 4.879495620727539,
      "learning_rate": 4.6703945500993474e-05,
      "loss": 0.0016322946175932884,
      "step": 187840
    },
    {
      "epoch": 53.321033210332104,
      "grad_norm": 0.013169843703508377,
      "learning_rate": 4.670110701107011e-05,
      "loss": 0.0007332935929298401,
      "step": 187850
    },
    {
      "epoch": 53.32387170025547,
      "grad_norm": 0.0702318474650383,
      "learning_rate": 4.669826852114676e-05,
      "loss": 0.00375099778175354,
      "step": 187860
    },
    {
      "epoch": 53.32671019017882,
      "grad_norm": 0.022802142426371574,
      "learning_rate": 4.669543003122339e-05,
      "loss": 0.0042123071849346164,
      "step": 187870
    },
    {
      "epoch": 53.329548680102185,
      "grad_norm": 1.2780405282974243,
      "learning_rate": 4.6692591541300026e-05,
      "loss": 0.008054125308990478,
      "step": 187880
    },
    {
      "epoch": 53.33238717002555,
      "grad_norm": 0.7242178916931152,
      "learning_rate": 4.6689753051376675e-05,
      "loss": 0.0015998652204871177,
      "step": 187890
    },
    {
      "epoch": 53.335225659948904,
      "grad_norm": 0.12846241891384125,
      "learning_rate": 4.668691456145331e-05,
      "loss": 0.0004772050306200981,
      "step": 187900
    },
    {
      "epoch": 53.33806414987227,
      "grad_norm": 0.22185134887695312,
      "learning_rate": 4.668407607152995e-05,
      "loss": 0.003217550367116928,
      "step": 187910
    },
    {
      "epoch": 53.34090263979563,
      "grad_norm": 0.11795251071453094,
      "learning_rate": 4.6681237581606585e-05,
      "loss": 0.000369369238615036,
      "step": 187920
    },
    {
      "epoch": 53.34374112971899,
      "grad_norm": 3.212143659591675,
      "learning_rate": 4.6678399091683226e-05,
      "loss": 0.008692929148674011,
      "step": 187930
    },
    {
      "epoch": 53.34657961964235,
      "grad_norm": 0.010836701840162277,
      "learning_rate": 4.667556060175987e-05,
      "loss": 0.00044226516038179395,
      "step": 187940
    },
    {
      "epoch": 53.34941810956571,
      "grad_norm": 0.01609104312956333,
      "learning_rate": 4.66727221118365e-05,
      "loss": 0.0005209090188145638,
      "step": 187950
    },
    {
      "epoch": 53.352256599489074,
      "grad_norm": 1.2381609678268433,
      "learning_rate": 4.6669883621913144e-05,
      "loss": 0.005836763232946396,
      "step": 187960
    },
    {
      "epoch": 53.35509508941243,
      "grad_norm": 0.10620569437742233,
      "learning_rate": 4.6667045131989785e-05,
      "loss": 0.0010946294292807578,
      "step": 187970
    },
    {
      "epoch": 53.35793357933579,
      "grad_norm": 0.07060200721025467,
      "learning_rate": 4.666420664206642e-05,
      "loss": 0.009801150113344193,
      "step": 187980
    },
    {
      "epoch": 53.360772069259156,
      "grad_norm": 0.0490715317428112,
      "learning_rate": 4.666136815214307e-05,
      "loss": 0.004727715998888016,
      "step": 187990
    },
    {
      "epoch": 53.36361055918251,
      "grad_norm": 0.05600271746516228,
      "learning_rate": 4.66585296622197e-05,
      "loss": 0.000669502466917038,
      "step": 188000
    },
    {
      "epoch": 53.36361055918251,
      "eval_accuracy": 0.9762828257137407,
      "eval_loss": 0.08823893964290619,
      "eval_runtime": 32.2328,
      "eval_samples_per_second": 487.92,
      "eval_steps_per_second": 7.632,
      "step": 188000
    },
    {
      "epoch": 53.366449049105874,
      "grad_norm": 5.185880661010742,
      "learning_rate": 4.665569117229634e-05,
      "loss": 0.004804622381925583,
      "step": 188010
    },
    {
      "epoch": 53.36928753902924,
      "grad_norm": 0.04575292766094208,
      "learning_rate": 4.665285268237298e-05,
      "loss": 0.000950300320982933,
      "step": 188020
    },
    {
      "epoch": 53.3721260289526,
      "grad_norm": 0.06251511722803116,
      "learning_rate": 4.665001419244962e-05,
      "loss": 0.0020166222006082536,
      "step": 188030
    },
    {
      "epoch": 53.374964518875956,
      "grad_norm": 0.011868019588291645,
      "learning_rate": 4.664717570252626e-05,
      "loss": 0.000693296454846859,
      "step": 188040
    },
    {
      "epoch": 53.37780300879932,
      "grad_norm": 6.547276973724365,
      "learning_rate": 4.6644337212602896e-05,
      "loss": 0.001462644524872303,
      "step": 188050
    },
    {
      "epoch": 53.38064149872268,
      "grad_norm": 1.185037612915039,
      "learning_rate": 4.664149872267954e-05,
      "loss": 0.0010535186156630516,
      "step": 188060
    },
    {
      "epoch": 53.38347998864604,
      "grad_norm": 0.838546097278595,
      "learning_rate": 4.663866023275618e-05,
      "loss": 0.0011111408472061156,
      "step": 188070
    },
    {
      "epoch": 53.3863184785694,
      "grad_norm": 0.5520305037498474,
      "learning_rate": 4.663582174283281e-05,
      "loss": 0.0008367469534277916,
      "step": 188080
    },
    {
      "epoch": 53.38915696849276,
      "grad_norm": 0.40938836336135864,
      "learning_rate": 4.6632983252909455e-05,
      "loss": 0.004319056868553162,
      "step": 188090
    },
    {
      "epoch": 53.391995458416126,
      "grad_norm": 0.08606300503015518,
      "learning_rate": 4.6630144762986096e-05,
      "loss": 0.0005734413862228394,
      "step": 188100
    },
    {
      "epoch": 53.39483394833948,
      "grad_norm": 0.7917957901954651,
      "learning_rate": 4.662730627306273e-05,
      "loss": 0.00047364681959152224,
      "step": 188110
    },
    {
      "epoch": 53.397672438262845,
      "grad_norm": 0.022059952840209007,
      "learning_rate": 4.662446778313937e-05,
      "loss": 0.0017526473850011826,
      "step": 188120
    },
    {
      "epoch": 53.40051092818621,
      "grad_norm": 2.4283320903778076,
      "learning_rate": 4.662162929321601e-05,
      "loss": 0.001446358673274517,
      "step": 188130
    },
    {
      "epoch": 53.40334941810956,
      "grad_norm": 0.05018947646021843,
      "learning_rate": 4.661879080329265e-05,
      "loss": 0.0008096132427453995,
      "step": 188140
    },
    {
      "epoch": 53.406187908032926,
      "grad_norm": 0.18982112407684326,
      "learning_rate": 4.661595231336929e-05,
      "loss": 0.0025374770164489748,
      "step": 188150
    },
    {
      "epoch": 53.40902639795629,
      "grad_norm": 0.03617572784423828,
      "learning_rate": 4.661311382344593e-05,
      "loss": 0.0006389845162630081,
      "step": 188160
    },
    {
      "epoch": 53.411864887879645,
      "grad_norm": 0.05680306628346443,
      "learning_rate": 4.6610275333522565e-05,
      "loss": 0.004915392398834229,
      "step": 188170
    },
    {
      "epoch": 53.41470337780301,
      "grad_norm": 0.35472002625465393,
      "learning_rate": 4.660743684359921e-05,
      "loss": 0.0021022846922278404,
      "step": 188180
    },
    {
      "epoch": 53.41754186772637,
      "grad_norm": 3.3389973640441895,
      "learning_rate": 4.660459835367585e-05,
      "loss": 0.004448512196540832,
      "step": 188190
    },
    {
      "epoch": 53.420380357649734,
      "grad_norm": 0.07949172705411911,
      "learning_rate": 4.660175986375249e-05,
      "loss": 0.0008588759228587151,
      "step": 188200
    },
    {
      "epoch": 53.42321884757309,
      "grad_norm": 0.0025089362170547247,
      "learning_rate": 4.6598921373829124e-05,
      "loss": 0.0019780004397034643,
      "step": 188210
    },
    {
      "epoch": 53.42605733749645,
      "grad_norm": 0.055848415940999985,
      "learning_rate": 4.659608288390576e-05,
      "loss": 0.0012480685487389564,
      "step": 188220
    },
    {
      "epoch": 53.428895827419815,
      "grad_norm": 0.644150972366333,
      "learning_rate": 4.659324439398241e-05,
      "loss": 0.0004986774176359177,
      "step": 188230
    },
    {
      "epoch": 53.43173431734317,
      "grad_norm": 0.29161128401756287,
      "learning_rate": 4.659040590405904e-05,
      "loss": 0.0006886543706059456,
      "step": 188240
    },
    {
      "epoch": 53.434572807266534,
      "grad_norm": 0.028612801805138588,
      "learning_rate": 4.658756741413568e-05,
      "loss": 0.0006515640765428543,
      "step": 188250
    },
    {
      "epoch": 53.4374112971899,
      "grad_norm": 0.031471848487854004,
      "learning_rate": 4.6584728924212324e-05,
      "loss": 0.0003195792436599731,
      "step": 188260
    },
    {
      "epoch": 53.44024978711325,
      "grad_norm": 0.062422677874565125,
      "learning_rate": 4.658189043428896e-05,
      "loss": 0.000794207863509655,
      "step": 188270
    },
    {
      "epoch": 53.443088277036615,
      "grad_norm": 0.07437340170145035,
      "learning_rate": 4.65790519443656e-05,
      "loss": 0.0002399832010269165,
      "step": 188280
    },
    {
      "epoch": 53.44592676695998,
      "grad_norm": 0.03113231062889099,
      "learning_rate": 4.657621345444224e-05,
      "loss": 0.0014466363936662674,
      "step": 188290
    },
    {
      "epoch": 53.44876525688334,
      "grad_norm": 0.062189262360334396,
      "learning_rate": 4.6573374964518876e-05,
      "loss": 0.0003046328201889992,
      "step": 188300
    },
    {
      "epoch": 53.4516037468067,
      "grad_norm": 0.24986520409584045,
      "learning_rate": 4.657053647459552e-05,
      "loss": 0.0009922612458467483,
      "step": 188310
    },
    {
      "epoch": 53.45444223673006,
      "grad_norm": 0.034784771502017975,
      "learning_rate": 4.656769798467215e-05,
      "loss": 0.0002948492765426636,
      "step": 188320
    },
    {
      "epoch": 53.45728072665342,
      "grad_norm": 2.984663724899292,
      "learning_rate": 4.65648594947488e-05,
      "loss": 0.0007460379973053932,
      "step": 188330
    },
    {
      "epoch": 53.46011921657678,
      "grad_norm": 0.015306895598769188,
      "learning_rate": 4.6562021004825435e-05,
      "loss": 0.000284622423350811,
      "step": 188340
    },
    {
      "epoch": 53.46295770650014,
      "grad_norm": 0.038686010986566544,
      "learning_rate": 4.655918251490207e-05,
      "loss": 0.0003249751403927803,
      "step": 188350
    },
    {
      "epoch": 53.465796196423504,
      "grad_norm": 4.617964267730713,
      "learning_rate": 4.655634402497872e-05,
      "loss": 0.003648388013243675,
      "step": 188360
    },
    {
      "epoch": 53.46863468634686,
      "grad_norm": 0.15376074612140656,
      "learning_rate": 4.655350553505535e-05,
      "loss": 0.00040807388722896576,
      "step": 188370
    },
    {
      "epoch": 53.47147317627022,
      "grad_norm": 0.7541678547859192,
      "learning_rate": 4.6550667045131993e-05,
      "loss": 0.0007031701505184173,
      "step": 188380
    },
    {
      "epoch": 53.474311666193586,
      "grad_norm": 0.12843947112560272,
      "learning_rate": 4.6547828555208635e-05,
      "loss": 0.004782354459166527,
      "step": 188390
    },
    {
      "epoch": 53.47715015611695,
      "grad_norm": 12.120165824890137,
      "learning_rate": 4.654499006528527e-05,
      "loss": 0.0024478064849972727,
      "step": 188400
    },
    {
      "epoch": 53.479988646040304,
      "grad_norm": 1.235603928565979,
      "learning_rate": 4.654215157536191e-05,
      "loss": 0.00048064384609460833,
      "step": 188410
    },
    {
      "epoch": 53.48282713596367,
      "grad_norm": 0.08865918219089508,
      "learning_rate": 4.6539313085438545e-05,
      "loss": 0.0010095691308379174,
      "step": 188420
    },
    {
      "epoch": 53.48566562588703,
      "grad_norm": 0.6434485912322998,
      "learning_rate": 4.653647459551519e-05,
      "loss": 0.0009179523214697838,
      "step": 188430
    },
    {
      "epoch": 53.488504115810386,
      "grad_norm": 0.04104013741016388,
      "learning_rate": 4.653363610559183e-05,
      "loss": 0.015262462198734283,
      "step": 188440
    },
    {
      "epoch": 53.49134260573375,
      "grad_norm": 0.16304485499858856,
      "learning_rate": 4.653079761566846e-05,
      "loss": 0.00320955328643322,
      "step": 188450
    },
    {
      "epoch": 53.49418109565711,
      "grad_norm": 0.041997287422418594,
      "learning_rate": 4.6527959125745104e-05,
      "loss": 0.001618834212422371,
      "step": 188460
    },
    {
      "epoch": 53.497019585580475,
      "grad_norm": 0.14904682338237762,
      "learning_rate": 4.6525120635821746e-05,
      "loss": 0.0018872521817684174,
      "step": 188470
    },
    {
      "epoch": 53.49985807550383,
      "grad_norm": 1.0034269094467163,
      "learning_rate": 4.652228214589838e-05,
      "loss": 0.0010557964444160462,
      "step": 188480
    },
    {
      "epoch": 53.50269656542719,
      "grad_norm": 0.039611585438251495,
      "learning_rate": 4.651944365597503e-05,
      "loss": 0.0005894061177968979,
      "step": 188490
    },
    {
      "epoch": 53.505535055350556,
      "grad_norm": 1.7711520195007324,
      "learning_rate": 4.651660516605166e-05,
      "loss": 0.0013704488053917884,
      "step": 188500
    },
    {
      "epoch": 53.505535055350556,
      "eval_accuracy": 0.9732943345838367,
      "eval_loss": 0.09527381509542465,
      "eval_runtime": 32.8245,
      "eval_samples_per_second": 479.124,
      "eval_steps_per_second": 7.494,
      "step": 188500
    },
    {
      "epoch": 53.50837354527391,
      "grad_norm": 1.8463540077209473,
      "learning_rate": 4.6513766676128304e-05,
      "loss": 0.005460367351770401,
      "step": 188510
    },
    {
      "epoch": 53.511212035197275,
      "grad_norm": 0.11110347509384155,
      "learning_rate": 4.6510928186204946e-05,
      "loss": 0.0009748652577400207,
      "step": 188520
    },
    {
      "epoch": 53.51405052512064,
      "grad_norm": 0.13024891912937164,
      "learning_rate": 4.650808969628158e-05,
      "loss": 0.00549941435456276,
      "step": 188530
    },
    {
      "epoch": 53.51688901504399,
      "grad_norm": 1.8181471824645996,
      "learning_rate": 4.650525120635822e-05,
      "loss": 0.005132311582565307,
      "step": 188540
    },
    {
      "epoch": 53.519727504967356,
      "grad_norm": 0.05729495361447334,
      "learning_rate": 4.6502412716434856e-05,
      "loss": 0.0017580313608050347,
      "step": 188550
    },
    {
      "epoch": 53.52256599489072,
      "grad_norm": 0.11723684519529343,
      "learning_rate": 4.64995742265115e-05,
      "loss": 0.00388554148375988,
      "step": 188560
    },
    {
      "epoch": 53.52540448481408,
      "grad_norm": 5.678378105163574,
      "learning_rate": 4.649673573658814e-05,
      "loss": 0.0029160957783460617,
      "step": 188570
    },
    {
      "epoch": 53.52824297473744,
      "grad_norm": 0.46963757276535034,
      "learning_rate": 4.6493897246664774e-05,
      "loss": 0.0022098880261182787,
      "step": 188580
    },
    {
      "epoch": 53.5310814646608,
      "grad_norm": 4.8113789558410645,
      "learning_rate": 4.6491058756741415e-05,
      "loss": 0.001940527930855751,
      "step": 188590
    },
    {
      "epoch": 53.533919954584164,
      "grad_norm": 9.30190372467041,
      "learning_rate": 4.6488220266818056e-05,
      "loss": 0.0018904400989413262,
      "step": 188600
    },
    {
      "epoch": 53.53675844450752,
      "grad_norm": 0.6030286550521851,
      "learning_rate": 4.648538177689469e-05,
      "loss": 0.002278678864240646,
      "step": 188610
    },
    {
      "epoch": 53.53959693443088,
      "grad_norm": 0.3396587371826172,
      "learning_rate": 4.648254328697134e-05,
      "loss": 0.005025509744882584,
      "step": 188620
    },
    {
      "epoch": 53.542435424354245,
      "grad_norm": 0.01260585431009531,
      "learning_rate": 4.6479704797047974e-05,
      "loss": 0.001262352056801319,
      "step": 188630
    },
    {
      "epoch": 53.5452739142776,
      "grad_norm": 0.09240014851093292,
      "learning_rate": 4.647686630712461e-05,
      "loss": 0.0048409514129161835,
      "step": 188640
    },
    {
      "epoch": 53.548112404200964,
      "grad_norm": 0.011994143016636372,
      "learning_rate": 4.647402781720125e-05,
      "loss": 0.0007047509774565696,
      "step": 188650
    },
    {
      "epoch": 53.55095089412433,
      "grad_norm": 0.06757330894470215,
      "learning_rate": 4.647118932727789e-05,
      "loss": 0.0009589048102498054,
      "step": 188660
    },
    {
      "epoch": 53.55378938404769,
      "grad_norm": 0.03461061790585518,
      "learning_rate": 4.646835083735453e-05,
      "loss": 0.0031213924288749696,
      "step": 188670
    },
    {
      "epoch": 53.556627873971046,
      "grad_norm": 0.9232422113418579,
      "learning_rate": 4.646551234743117e-05,
      "loss": 0.002472261153161526,
      "step": 188680
    },
    {
      "epoch": 53.55946636389441,
      "grad_norm": 3.757460832595825,
      "learning_rate": 4.646267385750781e-05,
      "loss": 0.0028620751574635504,
      "step": 188690
    },
    {
      "epoch": 53.56230485381777,
      "grad_norm": 0.05308603122830391,
      "learning_rate": 4.645983536758445e-05,
      "loss": 0.0006343841552734375,
      "step": 188700
    },
    {
      "epoch": 53.56514334374113,
      "grad_norm": 0.1945359855890274,
      "learning_rate": 4.6456996877661084e-05,
      "loss": 0.0047975372523069385,
      "step": 188710
    },
    {
      "epoch": 53.56798183366449,
      "grad_norm": 0.04334292188286781,
      "learning_rate": 4.6454158387737726e-05,
      "loss": 0.0011134112253785133,
      "step": 188720
    },
    {
      "epoch": 53.57082032358785,
      "grad_norm": 0.46612629294395447,
      "learning_rate": 4.645131989781437e-05,
      "loss": 0.0021153725683689117,
      "step": 188730
    },
    {
      "epoch": 53.57365881351121,
      "grad_norm": 0.017947977408766747,
      "learning_rate": 4.6448481407891e-05,
      "loss": 0.0009453095495700836,
      "step": 188740
    },
    {
      "epoch": 53.57649730343457,
      "grad_norm": 0.07249636948108673,
      "learning_rate": 4.644564291796764e-05,
      "loss": 0.0007467998191714286,
      "step": 188750
    },
    {
      "epoch": 53.579335793357934,
      "grad_norm": 0.028150703758001328,
      "learning_rate": 4.6442804428044284e-05,
      "loss": 0.002425053343176842,
      "step": 188760
    },
    {
      "epoch": 53.5821742832813,
      "grad_norm": 0.12547074258327484,
      "learning_rate": 4.643996593812092e-05,
      "loss": 0.0013765463605523109,
      "step": 188770
    },
    {
      "epoch": 53.58501277320465,
      "grad_norm": 3.709806203842163,
      "learning_rate": 4.643712744819756e-05,
      "loss": 0.0021960305050015448,
      "step": 188780
    },
    {
      "epoch": 53.587851263128016,
      "grad_norm": 0.17636407911777496,
      "learning_rate": 4.64342889582742e-05,
      "loss": 0.0013036832213401794,
      "step": 188790
    },
    {
      "epoch": 53.59068975305138,
      "grad_norm": 20.381999969482422,
      "learning_rate": 4.643145046835084e-05,
      "loss": 0.005438613146543503,
      "step": 188800
    },
    {
      "epoch": 53.593528242974735,
      "grad_norm": 0.3026883602142334,
      "learning_rate": 4.642861197842748e-05,
      "loss": 0.002001916617155075,
      "step": 188810
    },
    {
      "epoch": 53.5963667328981,
      "grad_norm": 0.1791962832212448,
      "learning_rate": 4.642577348850412e-05,
      "loss": 0.0005995603278279305,
      "step": 188820
    },
    {
      "epoch": 53.59920522282146,
      "grad_norm": 0.0670161023736,
      "learning_rate": 4.642293499858076e-05,
      "loss": 0.0018175350502133369,
      "step": 188830
    },
    {
      "epoch": 53.602043712744816,
      "grad_norm": 0.5099794268608093,
      "learning_rate": 4.6420096508657395e-05,
      "loss": 0.0035638704895973204,
      "step": 188840
    },
    {
      "epoch": 53.60488220266818,
      "grad_norm": 0.22814388573169708,
      "learning_rate": 4.6417258018734036e-05,
      "loss": 0.001583326980471611,
      "step": 188850
    },
    {
      "epoch": 53.60772069259154,
      "grad_norm": 0.037258636206388474,
      "learning_rate": 4.641441952881068e-05,
      "loss": 0.0024504045024514197,
      "step": 188860
    },
    {
      "epoch": 53.610559182514905,
      "grad_norm": 1.7237563133239746,
      "learning_rate": 4.641158103888731e-05,
      "loss": 0.002083294466137886,
      "step": 188870
    },
    {
      "epoch": 53.61339767243826,
      "grad_norm": 0.183261901140213,
      "learning_rate": 4.6408742548963954e-05,
      "loss": 0.0003583701327443123,
      "step": 188880
    },
    {
      "epoch": 53.61623616236162,
      "grad_norm": 0.011732768267393112,
      "learning_rate": 4.6405904059040595e-05,
      "loss": 0.0017732346430420875,
      "step": 188890
    },
    {
      "epoch": 53.619074652284986,
      "grad_norm": 0.028748994693160057,
      "learning_rate": 4.640306556911723e-05,
      "loss": 0.0004817206412553787,
      "step": 188900
    },
    {
      "epoch": 53.62191314220834,
      "grad_norm": 0.1272525191307068,
      "learning_rate": 4.640022707919387e-05,
      "loss": 0.0009630387648940086,
      "step": 188910
    },
    {
      "epoch": 53.624751632131705,
      "grad_norm": 4.592484951019287,
      "learning_rate": 4.639738858927051e-05,
      "loss": 0.0011784028261899947,
      "step": 188920
    },
    {
      "epoch": 53.62759012205507,
      "grad_norm": 0.0870702862739563,
      "learning_rate": 4.639455009934715e-05,
      "loss": 0.003085043281316757,
      "step": 188930
    },
    {
      "epoch": 53.63042861197843,
      "grad_norm": 0.0361563116312027,
      "learning_rate": 4.639171160942379e-05,
      "loss": 0.0009189007803797721,
      "step": 188940
    },
    {
      "epoch": 53.63326710190179,
      "grad_norm": 0.12388407438993454,
      "learning_rate": 4.638887311950042e-05,
      "loss": 0.0013304010033607482,
      "step": 188950
    },
    {
      "epoch": 53.63610559182515,
      "grad_norm": 0.6145620942115784,
      "learning_rate": 4.638603462957707e-05,
      "loss": 0.00045438967645168306,
      "step": 188960
    },
    {
      "epoch": 53.63894408174851,
      "grad_norm": 3.5897934436798096,
      "learning_rate": 4.6383196139653706e-05,
      "loss": 0.001373514160513878,
      "step": 188970
    },
    {
      "epoch": 53.64178257167187,
      "grad_norm": 1.138921856880188,
      "learning_rate": 4.638035764973034e-05,
      "loss": 0.007842230051755905,
      "step": 188980
    },
    {
      "epoch": 53.64462106159523,
      "grad_norm": 0.7738030552864075,
      "learning_rate": 4.637751915980699e-05,
      "loss": 0.0030894435942173002,
      "step": 188990
    },
    {
      "epoch": 53.647459551518594,
      "grad_norm": 0.30103740096092224,
      "learning_rate": 4.637468066988362e-05,
      "loss": 0.0022830281406641005,
      "step": 189000
    },
    {
      "epoch": 53.647459551518594,
      "eval_accuracy": 0.9762192407960831,
      "eval_loss": 0.08706940710544586,
      "eval_runtime": 32.4851,
      "eval_samples_per_second": 484.13,
      "eval_steps_per_second": 7.573,
      "step": 189000
    },
    {
      "epoch": 53.65029804144195,
      "grad_norm": 0.6918962001800537,
      "learning_rate": 4.6371842179960265e-05,
      "loss": 0.0010351579636335372,
      "step": 189010
    },
    {
      "epoch": 53.65313653136531,
      "grad_norm": 0.05111856758594513,
      "learning_rate": 4.6369003690036906e-05,
      "loss": 0.006209845095872879,
      "step": 189020
    },
    {
      "epoch": 53.655975021288675,
      "grad_norm": 0.49328985810279846,
      "learning_rate": 4.636616520011354e-05,
      "loss": 0.0013452405110001564,
      "step": 189030
    },
    {
      "epoch": 53.65881351121204,
      "grad_norm": 1.1391907930374146,
      "learning_rate": 4.636332671019018e-05,
      "loss": 0.002011183649301529,
      "step": 189040
    },
    {
      "epoch": 53.661652001135394,
      "grad_norm": 0.01998521387577057,
      "learning_rate": 4.6360488220266816e-05,
      "loss": 0.004966780170798301,
      "step": 189050
    },
    {
      "epoch": 53.66449049105876,
      "grad_norm": 0.3614199161529541,
      "learning_rate": 4.635764973034346e-05,
      "loss": 0.00042575821280479433,
      "step": 189060
    },
    {
      "epoch": 53.66732898098212,
      "grad_norm": 0.14182338118553162,
      "learning_rate": 4.63548112404201e-05,
      "loss": 0.0002698248252272606,
      "step": 189070
    },
    {
      "epoch": 53.670167470905476,
      "grad_norm": 0.15767528116703033,
      "learning_rate": 4.6351972750496734e-05,
      "loss": 0.0003722000867128372,
      "step": 189080
    },
    {
      "epoch": 53.67300596082884,
      "grad_norm": 0.02104571834206581,
      "learning_rate": 4.634913426057338e-05,
      "loss": 0.0005773192271590233,
      "step": 189090
    },
    {
      "epoch": 53.6758444507522,
      "grad_norm": 0.09053391218185425,
      "learning_rate": 4.6346295770650017e-05,
      "loss": 0.00041388403624296186,
      "step": 189100
    },
    {
      "epoch": 53.67868294067556,
      "grad_norm": 0.02403920888900757,
      "learning_rate": 4.634345728072665e-05,
      "loss": 0.00041633546352386474,
      "step": 189110
    },
    {
      "epoch": 53.68152143059892,
      "grad_norm": 0.06246112659573555,
      "learning_rate": 4.63406187908033e-05,
      "loss": 0.0012833327054977417,
      "step": 189120
    },
    {
      "epoch": 53.68435992052228,
      "grad_norm": 0.015193790197372437,
      "learning_rate": 4.6337780300879934e-05,
      "loss": 0.0009062131866812706,
      "step": 189130
    },
    {
      "epoch": 53.687198410445646,
      "grad_norm": 17.454011917114258,
      "learning_rate": 4.6334941810956575e-05,
      "loss": 0.004297524690628052,
      "step": 189140
    },
    {
      "epoch": 53.690036900369,
      "grad_norm": 0.08305272459983826,
      "learning_rate": 4.633210332103321e-05,
      "loss": 0.0017290262505412102,
      "step": 189150
    },
    {
      "epoch": 53.692875390292365,
      "grad_norm": 0.2885409891605377,
      "learning_rate": 4.632926483110985e-05,
      "loss": 0.0010030638426542283,
      "step": 189160
    },
    {
      "epoch": 53.69571388021573,
      "grad_norm": 0.014678586274385452,
      "learning_rate": 4.632642634118649e-05,
      "loss": 0.0031177390366792677,
      "step": 189170
    },
    {
      "epoch": 53.69855237013908,
      "grad_norm": 0.6939734220504761,
      "learning_rate": 4.632358785126313e-05,
      "loss": 0.0015747612342238426,
      "step": 189180
    },
    {
      "epoch": 53.701390860062446,
      "grad_norm": 8.897331237792969,
      "learning_rate": 4.632074936133977e-05,
      "loss": 0.0038195673376321794,
      "step": 189190
    },
    {
      "epoch": 53.70422934998581,
      "grad_norm": 0.08537156134843826,
      "learning_rate": 4.631791087141641e-05,
      "loss": 0.0034681320190429687,
      "step": 189200
    },
    {
      "epoch": 53.70706783990917,
      "grad_norm": 0.33852535486221313,
      "learning_rate": 4.6315072381493045e-05,
      "loss": 0.003675052523612976,
      "step": 189210
    },
    {
      "epoch": 53.70990632983253,
      "grad_norm": 0.03338174521923065,
      "learning_rate": 4.631223389156969e-05,
      "loss": 0.00034920331090688707,
      "step": 189220
    },
    {
      "epoch": 53.71274481975589,
      "grad_norm": 0.4796745777130127,
      "learning_rate": 4.630939540164633e-05,
      "loss": 0.0017383914440870286,
      "step": 189230
    },
    {
      "epoch": 53.71558330967925,
      "grad_norm": 0.23669879138469696,
      "learning_rate": 4.630655691172296e-05,
      "loss": 0.0014797637239098548,
      "step": 189240
    },
    {
      "epoch": 53.71842179960261,
      "grad_norm": 0.10383355617523193,
      "learning_rate": 4.63037184217996e-05,
      "loss": 0.0012917250394821167,
      "step": 189250
    },
    {
      "epoch": 53.72126028952597,
      "grad_norm": 0.17062482237815857,
      "learning_rate": 4.6300879931876245e-05,
      "loss": 0.006341440230607986,
      "step": 189260
    },
    {
      "epoch": 53.724098779449335,
      "grad_norm": 0.21293208003044128,
      "learning_rate": 4.6298041441952886e-05,
      "loss": 0.0025379808619618418,
      "step": 189270
    },
    {
      "epoch": 53.72693726937269,
      "grad_norm": 0.22762413322925568,
      "learning_rate": 4.629520295202952e-05,
      "loss": 0.01632435917854309,
      "step": 189280
    },
    {
      "epoch": 53.729775759296054,
      "grad_norm": 0.05195756256580353,
      "learning_rate": 4.629236446210616e-05,
      "loss": 0.00036417804658412934,
      "step": 189290
    },
    {
      "epoch": 53.73261424921942,
      "grad_norm": 0.14508557319641113,
      "learning_rate": 4.6289525972182803e-05,
      "loss": 0.0015911296010017394,
      "step": 189300
    },
    {
      "epoch": 53.73545273914278,
      "grad_norm": 0.025452902540564537,
      "learning_rate": 4.628668748225944e-05,
      "loss": 0.0008297970518469811,
      "step": 189310
    },
    {
      "epoch": 53.738291229066135,
      "grad_norm": 0.273532509803772,
      "learning_rate": 4.628384899233608e-05,
      "loss": 0.006817734241485596,
      "step": 189320
    },
    {
      "epoch": 53.7411297189895,
      "grad_norm": 0.06399763375520706,
      "learning_rate": 4.628101050241272e-05,
      "loss": 0.0016867170110344888,
      "step": 189330
    },
    {
      "epoch": 53.74396820891286,
      "grad_norm": 1.3764978647232056,
      "learning_rate": 4.6278172012489355e-05,
      "loss": 0.0013445986434817315,
      "step": 189340
    },
    {
      "epoch": 53.74680669883622,
      "grad_norm": 0.05525941401720047,
      "learning_rate": 4.6275333522566e-05,
      "loss": 0.001662706397473812,
      "step": 189350
    },
    {
      "epoch": 53.74964518875958,
      "grad_norm": 0.019307225942611694,
      "learning_rate": 4.627249503264264e-05,
      "loss": 0.004553106427192688,
      "step": 189360
    },
    {
      "epoch": 53.75248367868294,
      "grad_norm": 15.523204803466797,
      "learning_rate": 4.626965654271927e-05,
      "loss": 0.006425189226865769,
      "step": 189370
    },
    {
      "epoch": 53.7553221686063,
      "grad_norm": Infinity,
      "learning_rate": 4.6266818052795914e-05,
      "loss": 0.006315889954566956,
      "step": 189380
    },
    {
      "epoch": 53.75816065852966,
      "grad_norm": 2.023550271987915,
      "learning_rate": 4.626426341186489e-05,
      "loss": 0.010281020402908325,
      "step": 189390
    },
    {
      "epoch": 53.760999148453024,
      "grad_norm": 0.05651474371552467,
      "learning_rate": 4.626142492194153e-05,
      "loss": 0.0009592682123184204,
      "step": 189400
    },
    {
      "epoch": 53.76383763837639,
      "grad_norm": 10.953176498413086,
      "learning_rate": 4.625858643201817e-05,
      "loss": 0.012157021462917328,
      "step": 189410
    },
    {
      "epoch": 53.76667612829974,
      "grad_norm": 0.040942542254924774,
      "learning_rate": 4.6255747942094805e-05,
      "loss": 0.016399207711219787,
      "step": 189420
    },
    {
      "epoch": 53.769514618223106,
      "grad_norm": 0.15226218104362488,
      "learning_rate": 4.6252909452171446e-05,
      "loss": 0.004528618603944779,
      "step": 189430
    },
    {
      "epoch": 53.77235310814647,
      "grad_norm": 18.527626037597656,
      "learning_rate": 4.625007096224809e-05,
      "loss": 0.005434627085924149,
      "step": 189440
    },
    {
      "epoch": 53.775191598069824,
      "grad_norm": 0.02767735719680786,
      "learning_rate": 4.624723247232472e-05,
      "loss": 0.005245065316557884,
      "step": 189450
    },
    {
      "epoch": 53.77803008799319,
      "grad_norm": 0.39277541637420654,
      "learning_rate": 4.624439398240136e-05,
      "loss": 0.002950715832412243,
      "step": 189460
    },
    {
      "epoch": 53.78086857791655,
      "grad_norm": 0.06754680722951889,
      "learning_rate": 4.6241555492478005e-05,
      "loss": 0.0018896477296948433,
      "step": 189470
    },
    {
      "epoch": 53.783707067839906,
      "grad_norm": 0.024361910298466682,
      "learning_rate": 4.6238717002554646e-05,
      "loss": 0.0006793098524212837,
      "step": 189480
    },
    {
      "epoch": 53.78654555776327,
      "grad_norm": 18.31993293762207,
      "learning_rate": 4.623587851263128e-05,
      "loss": 0.005990404635667801,
      "step": 189490
    },
    {
      "epoch": 53.78938404768663,
      "grad_norm": 0.14143411815166473,
      "learning_rate": 4.623304002270792e-05,
      "loss": 0.00049457848072052,
      "step": 189500
    },
    {
      "epoch": 53.78938404768663,
      "eval_accuracy": 0.9772365994786036,
      "eval_loss": 0.08411398530006409,
      "eval_runtime": 32.9337,
      "eval_samples_per_second": 477.535,
      "eval_steps_per_second": 7.47,
      "step": 189500
    },
    {
      "epoch": 53.792222537609995,
      "grad_norm": 0.021601110696792603,
      "learning_rate": 4.6230201532784563e-05,
      "loss": 0.0021941764280200005,
      "step": 189510
    },
    {
      "epoch": 53.79506102753335,
      "grad_norm": 9.54564380645752,
      "learning_rate": 4.62273630428612e-05,
      "loss": 0.0025540245696902275,
      "step": 189520
    },
    {
      "epoch": 53.79789951745671,
      "grad_norm": 0.01287209801375866,
      "learning_rate": 4.622452455293784e-05,
      "loss": 0.0035553470253944395,
      "step": 189530
    },
    {
      "epoch": 53.800738007380076,
      "grad_norm": 0.025906408205628395,
      "learning_rate": 4.622168606301448e-05,
      "loss": 0.0006148003041744233,
      "step": 189540
    },
    {
      "epoch": 53.80357649730343,
      "grad_norm": 0.11060051620006561,
      "learning_rate": 4.6218847573091115e-05,
      "loss": 0.0012214817106723785,
      "step": 189550
    },
    {
      "epoch": 53.806414987226795,
      "grad_norm": 0.028840314596891403,
      "learning_rate": 4.621600908316776e-05,
      "loss": 0.00043843276798725126,
      "step": 189560
    },
    {
      "epoch": 53.80925347715016,
      "grad_norm": 0.052746452391147614,
      "learning_rate": 4.62131705932444e-05,
      "loss": 0.0007230622693896293,
      "step": 189570
    },
    {
      "epoch": 53.81209196707351,
      "grad_norm": 0.2922099530696869,
      "learning_rate": 4.621033210332103e-05,
      "loss": 0.001693853922188282,
      "step": 189580
    },
    {
      "epoch": 53.814930456996876,
      "grad_norm": 0.24571073055267334,
      "learning_rate": 4.6207493613397674e-05,
      "loss": 0.0003926454111933708,
      "step": 189590
    },
    {
      "epoch": 53.81776894692024,
      "grad_norm": 0.08323043584823608,
      "learning_rate": 4.6204655123474315e-05,
      "loss": 0.0006954755634069442,
      "step": 189600
    },
    {
      "epoch": 53.8206074368436,
      "grad_norm": 0.024788228794932365,
      "learning_rate": 4.620181663355096e-05,
      "loss": 0.00025607813149690627,
      "step": 189610
    },
    {
      "epoch": 53.82344592676696,
      "grad_norm": 0.2623211145401001,
      "learning_rate": 4.619897814362759e-05,
      "loss": 0.00037547256797552107,
      "step": 189620
    },
    {
      "epoch": 53.82628441669032,
      "grad_norm": 0.05043733865022659,
      "learning_rate": 4.6196139653704226e-05,
      "loss": 0.00036613382399082185,
      "step": 189630
    },
    {
      "epoch": 53.829122906613684,
      "grad_norm": 0.08899254351854324,
      "learning_rate": 4.6193301163780874e-05,
      "loss": 0.0002988923341035843,
      "step": 189640
    },
    {
      "epoch": 53.83196139653704,
      "grad_norm": 1.3906793594360352,
      "learning_rate": 4.619046267385751e-05,
      "loss": 0.0015650900080800057,
      "step": 189650
    },
    {
      "epoch": 53.8347998864604,
      "grad_norm": 0.08224312216043472,
      "learning_rate": 4.618762418393415e-05,
      "loss": 0.0003687510266900063,
      "step": 189660
    },
    {
      "epoch": 53.837638376383765,
      "grad_norm": 0.006065186113119125,
      "learning_rate": 4.618478569401079e-05,
      "loss": 0.0024584300816059114,
      "step": 189670
    },
    {
      "epoch": 53.84047686630713,
      "grad_norm": 0.01789562590420246,
      "learning_rate": 4.6181947204087426e-05,
      "loss": 0.0006448995321989059,
      "step": 189680
    },
    {
      "epoch": 53.843315356230484,
      "grad_norm": 2.570772647857666,
      "learning_rate": 4.617910871416407e-05,
      "loss": 0.0010420428588986397,
      "step": 189690
    },
    {
      "epoch": 53.84615384615385,
      "grad_norm": 0.22996391355991364,
      "learning_rate": 4.617627022424071e-05,
      "loss": 0.0009878776967525483,
      "step": 189700
    },
    {
      "epoch": 53.84899233607721,
      "grad_norm": 0.05956209450960159,
      "learning_rate": 4.6173431734317343e-05,
      "loss": 0.00040733665227890017,
      "step": 189710
    },
    {
      "epoch": 53.851830826000565,
      "grad_norm": 0.6274832487106323,
      "learning_rate": 4.6170593244393985e-05,
      "loss": 0.0010274490341544151,
      "step": 189720
    },
    {
      "epoch": 53.85466931592393,
      "grad_norm": 0.0415818989276886,
      "learning_rate": 4.616775475447062e-05,
      "loss": 0.0007280947640538216,
      "step": 189730
    },
    {
      "epoch": 53.85750780584729,
      "grad_norm": 0.0592634491622448,
      "learning_rate": 4.616491626454727e-05,
      "loss": 0.00041999686509370805,
      "step": 189740
    },
    {
      "epoch": 53.86034629577065,
      "grad_norm": 0.012188628315925598,
      "learning_rate": 4.61620777746239e-05,
      "loss": 0.0009052179753780365,
      "step": 189750
    },
    {
      "epoch": 53.86318478569401,
      "grad_norm": 0.18655340373516083,
      "learning_rate": 4.615923928470054e-05,
      "loss": 0.001091170869767666,
      "step": 189760
    },
    {
      "epoch": 53.86602327561737,
      "grad_norm": 0.08614689856767654,
      "learning_rate": 4.6156400794777185e-05,
      "loss": 0.0009487850591540336,
      "step": 189770
    },
    {
      "epoch": 53.868861765540736,
      "grad_norm": 0.04204598814249039,
      "learning_rate": 4.615356230485382e-05,
      "loss": 0.0014914494007825852,
      "step": 189780
    },
    {
      "epoch": 53.87170025546409,
      "grad_norm": 0.08111171424388885,
      "learning_rate": 4.615072381493046e-05,
      "loss": 0.002232424542307854,
      "step": 189790
    },
    {
      "epoch": 53.874538745387454,
      "grad_norm": 0.25810083746910095,
      "learning_rate": 4.61478853250071e-05,
      "loss": 0.0012791948392987251,
      "step": 189800
    },
    {
      "epoch": 53.87737723531082,
      "grad_norm": 0.19585120677947998,
      "learning_rate": 4.614504683508374e-05,
      "loss": 0.0003525331616401672,
      "step": 189810
    },
    {
      "epoch": 53.88021572523417,
      "grad_norm": 0.13614630699157715,
      "learning_rate": 4.614220834516038e-05,
      "loss": 0.0024169130250811578,
      "step": 189820
    },
    {
      "epoch": 53.883054215157536,
      "grad_norm": 0.014362146146595478,
      "learning_rate": 4.613936985523701e-05,
      "loss": 0.0010011771693825722,
      "step": 189830
    },
    {
      "epoch": 53.8858927050809,
      "grad_norm": 0.00514635443687439,
      "learning_rate": 4.6136531365313654e-05,
      "loss": 0.0032917216420173647,
      "step": 189840
    },
    {
      "epoch": 53.888731195004254,
      "grad_norm": 1.447511911392212,
      "learning_rate": 4.6133692875390296e-05,
      "loss": 0.0015402790158987045,
      "step": 189850
    },
    {
      "epoch": 53.89156968492762,
      "grad_norm": 0.5010764598846436,
      "learning_rate": 4.613085438546693e-05,
      "loss": 0.0006503740325570107,
      "step": 189860
    },
    {
      "epoch": 53.89440817485098,
      "grad_norm": 0.25249022245407104,
      "learning_rate": 4.612801589554357e-05,
      "loss": 0.0005416134372353554,
      "step": 189870
    },
    {
      "epoch": 53.89724666477434,
      "grad_norm": 0.07815248519182205,
      "learning_rate": 4.612517740562021e-05,
      "loss": 0.003799545019865036,
      "step": 189880
    },
    {
      "epoch": 53.9000851546977,
      "grad_norm": 0.05863397568464279,
      "learning_rate": 4.612233891569685e-05,
      "loss": 0.003936757147312164,
      "step": 189890
    },
    {
      "epoch": 53.90292364462106,
      "grad_norm": 0.03863385692238808,
      "learning_rate": 4.6119500425773496e-05,
      "loss": 0.0003835948184132576,
      "step": 189900
    },
    {
      "epoch": 53.905762134544425,
      "grad_norm": 4.592468738555908,
      "learning_rate": 4.611666193585013e-05,
      "loss": 0.0011073751375079155,
      "step": 189910
    },
    {
      "epoch": 53.90860062446778,
      "grad_norm": 8.734415054321289,
      "learning_rate": 4.6113823445926765e-05,
      "loss": 0.002851998247206211,
      "step": 189920
    },
    {
      "epoch": 53.91143911439114,
      "grad_norm": 0.20079463720321655,
      "learning_rate": 4.6110984956003406e-05,
      "loss": 0.0034388467669487,
      "step": 189930
    },
    {
      "epoch": 53.914277604314506,
      "grad_norm": 0.9759150743484497,
      "learning_rate": 4.610814646608005e-05,
      "loss": 0.0005862940102815628,
      "step": 189940
    },
    {
      "epoch": 53.91711609423786,
      "grad_norm": 0.07106303423643112,
      "learning_rate": 4.610530797615669e-05,
      "loss": 0.0006806841120123863,
      "step": 189950
    },
    {
      "epoch": 53.919954584161225,
      "grad_norm": 0.2194651961326599,
      "learning_rate": 4.6102469486233324e-05,
      "loss": 0.0002784525975584984,
      "step": 189960
    },
    {
      "epoch": 53.92279307408459,
      "grad_norm": 0.018796708434820175,
      "learning_rate": 4.6099630996309965e-05,
      "loss": 0.00027405768632888795,
      "step": 189970
    },
    {
      "epoch": 53.92563156400795,
      "grad_norm": 0.40771278738975525,
      "learning_rate": 4.6096792506386606e-05,
      "loss": 0.0004953578114509583,
      "step": 189980
    },
    {
      "epoch": 53.928470053931306,
      "grad_norm": 0.01927034556865692,
      "learning_rate": 4.609395401646324e-05,
      "loss": 0.0003737468272447586,
      "step": 189990
    },
    {
      "epoch": 53.93130854385467,
      "grad_norm": 0.27646583318710327,
      "learning_rate": 4.609111552653988e-05,
      "loss": 0.0006116572767496109,
      "step": 190000
    },
    {
      "epoch": 53.93130854385467,
      "eval_accuracy": 0.9788262224200419,
      "eval_loss": 0.07482069730758667,
      "eval_runtime": 32.6115,
      "eval_samples_per_second": 482.253,
      "eval_steps_per_second": 7.543,
      "step": 190000
    },
    {
      "epoch": 53.93414703377803,
      "grad_norm": 0.4058345556259155,
      "learning_rate": 4.6088277036616524e-05,
      "loss": 0.0002704864367842674,
      "step": 190010
    },
    {
      "epoch": 53.93698552370139,
      "grad_norm": 0.023325767368078232,
      "learning_rate": 4.608543854669316e-05,
      "loss": 0.0012160364538431167,
      "step": 190020
    },
    {
      "epoch": 53.93982401362475,
      "grad_norm": 0.2957730293273926,
      "learning_rate": 4.60826000567698e-05,
      "loss": 0.002289293520152569,
      "step": 190030
    },
    {
      "epoch": 53.942662503548114,
      "grad_norm": 0.20680375397205353,
      "learning_rate": 4.607976156684644e-05,
      "loss": 0.0005347423255443573,
      "step": 190040
    },
    {
      "epoch": 53.94550099347148,
      "grad_norm": 1.0706795454025269,
      "learning_rate": 4.6076923076923076e-05,
      "loss": 0.0006216930225491524,
      "step": 190050
    },
    {
      "epoch": 53.94833948339483,
      "grad_norm": 0.028042763471603394,
      "learning_rate": 4.607408458699972e-05,
      "loss": 0.0005715759471058846,
      "step": 190060
    },
    {
      "epoch": 53.951177973318195,
      "grad_norm": 0.022818604484200478,
      "learning_rate": 4.607124609707636e-05,
      "loss": 0.000824689120054245,
      "step": 190070
    },
    {
      "epoch": 53.95401646324156,
      "grad_norm": 1.9552831649780273,
      "learning_rate": 4.6068407607153e-05,
      "loss": 0.001447206549346447,
      "step": 190080
    },
    {
      "epoch": 53.956854953164914,
      "grad_norm": 0.13339178264141083,
      "learning_rate": 4.6065569117229634e-05,
      "loss": 0.0003341201692819595,
      "step": 190090
    },
    {
      "epoch": 53.95969344308828,
      "grad_norm": 0.37993094325065613,
      "learning_rate": 4.6062730627306276e-05,
      "loss": 0.0004221992567181587,
      "step": 190100
    },
    {
      "epoch": 53.96253193301164,
      "grad_norm": 0.16393068432807922,
      "learning_rate": 4.605989213738292e-05,
      "loss": 0.0015563966706395149,
      "step": 190110
    },
    {
      "epoch": 53.965370422934996,
      "grad_norm": 0.06460926681756973,
      "learning_rate": 4.605705364745955e-05,
      "loss": 0.004735343903303146,
      "step": 190120
    },
    {
      "epoch": 53.96820891285836,
      "grad_norm": 0.09176139533519745,
      "learning_rate": 4.605421515753619e-05,
      "loss": 0.0048593848943710325,
      "step": 190130
    },
    {
      "epoch": 53.97104740278172,
      "grad_norm": 3.7055065631866455,
      "learning_rate": 4.6051376667612835e-05,
      "loss": 0.0030064450576901434,
      "step": 190140
    },
    {
      "epoch": 53.973885892705084,
      "grad_norm": 0.1778849959373474,
      "learning_rate": 4.604853817768947e-05,
      "loss": 0.01610005795955658,
      "step": 190150
    },
    {
      "epoch": 53.97672438262844,
      "grad_norm": 0.14732691645622253,
      "learning_rate": 4.604569968776611e-05,
      "loss": 0.0024450957775115968,
      "step": 190160
    },
    {
      "epoch": 53.9795628725518,
      "grad_norm": 0.9671622514724731,
      "learning_rate": 4.604286119784275e-05,
      "loss": 0.0006380550563335419,
      "step": 190170
    },
    {
      "epoch": 53.982401362475166,
      "grad_norm": 0.10074272006750107,
      "learning_rate": 4.6040022707919386e-05,
      "loss": 0.0007599027827382088,
      "step": 190180
    },
    {
      "epoch": 53.98523985239852,
      "grad_norm": 0.008153036236763,
      "learning_rate": 4.603718421799603e-05,
      "loss": 0.0017909526824951172,
      "step": 190190
    },
    {
      "epoch": 53.988078342321884,
      "grad_norm": 12.13166332244873,
      "learning_rate": 4.603434572807267e-05,
      "loss": 0.0018133584409952164,
      "step": 190200
    },
    {
      "epoch": 53.99091683224525,
      "grad_norm": 0.18008406460285187,
      "learning_rate": 4.603150723814931e-05,
      "loss": 0.0021529370918869973,
      "step": 190210
    },
    {
      "epoch": 53.9937553221686,
      "grad_norm": 0.1943129152059555,
      "learning_rate": 4.6028668748225945e-05,
      "loss": 0.0006942916661500931,
      "step": 190220
    },
    {
      "epoch": 53.996593812091966,
      "grad_norm": 0.07712505757808685,
      "learning_rate": 4.6025830258302587e-05,
      "loss": 0.0045285351574420925,
      "step": 190230
    },
    {
      "epoch": 53.99943230201533,
      "grad_norm": 0.01945650205016136,
      "learning_rate": 4.602299176837923e-05,
      "loss": 0.0012741034850478172,
      "step": 190240
    },
    {
      "epoch": 54.00227079193869,
      "grad_norm": 0.04336756467819214,
      "learning_rate": 4.602015327845586e-05,
      "loss": 0.0004376681987196207,
      "step": 190250
    },
    {
      "epoch": 54.00510928186205,
      "grad_norm": 0.058626607060432434,
      "learning_rate": 4.6017314788532504e-05,
      "loss": 0.002380247227847576,
      "step": 190260
    },
    {
      "epoch": 54.00794777178541,
      "grad_norm": 0.025904756039381027,
      "learning_rate": 4.6014476298609145e-05,
      "loss": 0.00265309102833271,
      "step": 190270
    },
    {
      "epoch": 54.01078626170877,
      "grad_norm": 0.10968022048473358,
      "learning_rate": 4.601163780868578e-05,
      "loss": 0.0013353103771805764,
      "step": 190280
    },
    {
      "epoch": 54.01362475163213,
      "grad_norm": 0.08156608790159225,
      "learning_rate": 4.600879931876242e-05,
      "loss": 0.004735436290502548,
      "step": 190290
    },
    {
      "epoch": 54.01646324155549,
      "grad_norm": 0.02988324500620365,
      "learning_rate": 4.600596082883906e-05,
      "loss": 0.0008209524676203728,
      "step": 190300
    },
    {
      "epoch": 54.019301731478855,
      "grad_norm": 0.03307041525840759,
      "learning_rate": 4.60031223389157e-05,
      "loss": 0.0004901405423879624,
      "step": 190310
    },
    {
      "epoch": 54.02214022140221,
      "grad_norm": 0.2678154706954956,
      "learning_rate": 4.600028384899234e-05,
      "loss": 0.00025896038860082624,
      "step": 190320
    },
    {
      "epoch": 54.02497871132557,
      "grad_norm": 0.23837795853614807,
      "learning_rate": 4.599744535906898e-05,
      "loss": 0.0006772223860025406,
      "step": 190330
    },
    {
      "epoch": 54.027817201248936,
      "grad_norm": 0.1294577270746231,
      "learning_rate": 4.5994606869145615e-05,
      "loss": 0.0017785361036658288,
      "step": 190340
    },
    {
      "epoch": 54.0306556911723,
      "grad_norm": 0.33269912004470825,
      "learning_rate": 4.5991768379222256e-05,
      "loss": 0.00022858977317810058,
      "step": 190350
    },
    {
      "epoch": 54.033494181095655,
      "grad_norm": 0.008971957489848137,
      "learning_rate": 4.598892988929889e-05,
      "loss": 0.0008467480540275574,
      "step": 190360
    },
    {
      "epoch": 54.03633267101902,
      "grad_norm": 1.3784480094909668,
      "learning_rate": 4.598609139937554e-05,
      "loss": 0.0035009533166885376,
      "step": 190370
    },
    {
      "epoch": 54.03917116094238,
      "grad_norm": 0.03688269108533859,
      "learning_rate": 4.598325290945217e-05,
      "loss": 0.00036978591233491895,
      "step": 190380
    },
    {
      "epoch": 54.04200965086574,
      "grad_norm": 2.610508918762207,
      "learning_rate": 4.598041441952881e-05,
      "loss": 0.0009543802589178085,
      "step": 190390
    },
    {
      "epoch": 54.0448481407891,
      "grad_norm": 0.015336356125772,
      "learning_rate": 4.5977575929605456e-05,
      "loss": 0.0007450997829437256,
      "step": 190400
    },
    {
      "epoch": 54.04768663071246,
      "grad_norm": 0.07187741994857788,
      "learning_rate": 4.597473743968209e-05,
      "loss": 0.0008507147431373596,
      "step": 190410
    },
    {
      "epoch": 54.050525120635825,
      "grad_norm": 0.11248627305030823,
      "learning_rate": 4.597189894975873e-05,
      "loss": 0.00016391444951295852,
      "step": 190420
    },
    {
      "epoch": 54.05336361055918,
      "grad_norm": 0.18101564049720764,
      "learning_rate": 4.5969060459835373e-05,
      "loss": 0.0005840940400958061,
      "step": 190430
    },
    {
      "epoch": 54.056202100482544,
      "grad_norm": 0.06716714799404144,
      "learning_rate": 4.596622196991201e-05,
      "loss": 0.0004270797595381737,
      "step": 190440
    },
    {
      "epoch": 54.05904059040591,
      "grad_norm": 0.039141811430454254,
      "learning_rate": 4.596338347998865e-05,
      "loss": 0.000286259688436985,
      "step": 190450
    },
    {
      "epoch": 54.06187908032926,
      "grad_norm": 0.027115758508443832,
      "learning_rate": 4.5960544990065284e-05,
      "loss": 0.00038269311189651487,
      "step": 190460
    },
    {
      "epoch": 54.064717570252625,
      "grad_norm": 0.036761052906513214,
      "learning_rate": 4.5957706500141925e-05,
      "loss": 0.00014310982078313828,
      "step": 190470
    },
    {
      "epoch": 54.06755606017599,
      "grad_norm": 0.011145690456032753,
      "learning_rate": 4.595486801021857e-05,
      "loss": 0.0002455353736877441,
      "step": 190480
    },
    {
      "epoch": 54.070394550099344,
      "grad_norm": 0.007584948092699051,
      "learning_rate": 4.59520295202952e-05,
      "loss": 0.0004253394901752472,
      "step": 190490
    },
    {
      "epoch": 54.07323304002271,
      "grad_norm": 0.03196874260902405,
      "learning_rate": 4.594919103037185e-05,
      "loss": 0.0029203345999121665,
      "step": 190500
    },
    {
      "epoch": 54.07323304002271,
      "eval_accuracy": 0.9783811279964393,
      "eval_loss": 0.0743328109383583,
      "eval_runtime": 32.6456,
      "eval_samples_per_second": 481.749,
      "eval_steps_per_second": 7.535,
      "step": 190500
    },
    {
      "epoch": 54.07607152994607,
      "grad_norm": 7.898156642913818,
      "learning_rate": 4.5946352540448484e-05,
      "loss": 0.001129193603992462,
      "step": 190510
    },
    {
      "epoch": 54.07891001986943,
      "grad_norm": 0.10997507721185684,
      "learning_rate": 4.594351405052512e-05,
      "loss": 0.0012220559641718865,
      "step": 190520
    },
    {
      "epoch": 54.08174850979279,
      "grad_norm": 0.03728827089071274,
      "learning_rate": 4.594067556060177e-05,
      "loss": 0.0008155627176165581,
      "step": 190530
    },
    {
      "epoch": 54.08458699971615,
      "grad_norm": 1.9674068689346313,
      "learning_rate": 4.59378370706784e-05,
      "loss": 0.008432667702436447,
      "step": 190540
    },
    {
      "epoch": 54.087425489639514,
      "grad_norm": 0.020622720941901207,
      "learning_rate": 4.593499858075504e-05,
      "loss": 0.0005267409607768059,
      "step": 190550
    },
    {
      "epoch": 54.09026397956287,
      "grad_norm": 0.04700889810919762,
      "learning_rate": 4.593216009083168e-05,
      "loss": 0.0004438776522874832,
      "step": 190560
    },
    {
      "epoch": 54.09310246948623,
      "grad_norm": 0.052473366260528564,
      "learning_rate": 4.592932160090832e-05,
      "loss": 0.0005018325522542,
      "step": 190570
    },
    {
      "epoch": 54.095940959409596,
      "grad_norm": 0.1601518839597702,
      "learning_rate": 4.592648311098496e-05,
      "loss": 0.004880205541849136,
      "step": 190580
    },
    {
      "epoch": 54.09877944933295,
      "grad_norm": 0.056338779628276825,
      "learning_rate": 4.5923644621061595e-05,
      "loss": 0.004223726689815521,
      "step": 190590
    },
    {
      "epoch": 54.101617939256315,
      "grad_norm": 1.8295094966888428,
      "learning_rate": 4.5920806131138236e-05,
      "loss": 0.0006453650072216988,
      "step": 190600
    },
    {
      "epoch": 54.10445642917968,
      "grad_norm": 0.04320772737264633,
      "learning_rate": 4.591796764121488e-05,
      "loss": 0.000980471633374691,
      "step": 190610
    },
    {
      "epoch": 54.10729491910304,
      "grad_norm": 0.32836875319480896,
      "learning_rate": 4.591512915129151e-05,
      "loss": 0.0011575426906347275,
      "step": 190620
    },
    {
      "epoch": 54.110133409026396,
      "grad_norm": 0.10729886591434479,
      "learning_rate": 4.591229066136816e-05,
      "loss": 0.00022611748427152633,
      "step": 190630
    },
    {
      "epoch": 54.11297189894976,
      "grad_norm": 0.04957544431090355,
      "learning_rate": 4.5909452171444795e-05,
      "loss": 0.0008996525779366493,
      "step": 190640
    },
    {
      "epoch": 54.11581038887312,
      "grad_norm": 0.022691527381539345,
      "learning_rate": 4.590661368152143e-05,
      "loss": 0.0007292501628398895,
      "step": 190650
    },
    {
      "epoch": 54.11864887879648,
      "grad_norm": 0.4851768910884857,
      "learning_rate": 4.590377519159807e-05,
      "loss": 0.002762913890182972,
      "step": 190660
    },
    {
      "epoch": 54.12148736871984,
      "grad_norm": 0.043385524302721024,
      "learning_rate": 4.590093670167471e-05,
      "loss": 0.005570866167545319,
      "step": 190670
    },
    {
      "epoch": 54.1243258586432,
      "grad_norm": 0.5570418834686279,
      "learning_rate": 4.5898098211751354e-05,
      "loss": 0.0009375890716910362,
      "step": 190680
    },
    {
      "epoch": 54.12716434856656,
      "grad_norm": 0.050397127866744995,
      "learning_rate": 4.589525972182799e-05,
      "loss": 0.00035753827542066573,
      "step": 190690
    },
    {
      "epoch": 54.13000283848992,
      "grad_norm": 0.0205315463244915,
      "learning_rate": 4.589242123190463e-05,
      "loss": 0.0009131802245974541,
      "step": 190700
    },
    {
      "epoch": 54.132841328413285,
      "grad_norm": 0.05958734452724457,
      "learning_rate": 4.588958274198127e-05,
      "loss": 0.001697041280567646,
      "step": 190710
    },
    {
      "epoch": 54.13567981833665,
      "grad_norm": 6.207794189453125,
      "learning_rate": 4.5886744252057906e-05,
      "loss": 0.004091175645589829,
      "step": 190720
    },
    {
      "epoch": 54.138518308260004,
      "grad_norm": 0.05499005317687988,
      "learning_rate": 4.588390576213455e-05,
      "loss": 0.0006162609905004501,
      "step": 190730
    },
    {
      "epoch": 54.14135679818337,
      "grad_norm": 0.2337942272424698,
      "learning_rate": 4.588106727221119e-05,
      "loss": 0.0003521820530295372,
      "step": 190740
    },
    {
      "epoch": 54.14419528810673,
      "grad_norm": 0.11948271095752716,
      "learning_rate": 4.587822878228782e-05,
      "loss": 0.0003924405202269554,
      "step": 190750
    },
    {
      "epoch": 54.147033778030085,
      "grad_norm": 0.05974902585148811,
      "learning_rate": 4.5875390292364464e-05,
      "loss": 0.0004996908828616142,
      "step": 190760
    },
    {
      "epoch": 54.14987226795345,
      "grad_norm": 0.08094682544469833,
      "learning_rate": 4.5872551802441106e-05,
      "loss": 0.00017572548240423201,
      "step": 190770
    },
    {
      "epoch": 54.15271075787681,
      "grad_norm": 0.005863814614713192,
      "learning_rate": 4.586971331251774e-05,
      "loss": 0.0002920014783740044,
      "step": 190780
    },
    {
      "epoch": 54.15554924780017,
      "grad_norm": 0.15569227933883667,
      "learning_rate": 4.586687482259438e-05,
      "loss": 0.000244101881980896,
      "step": 190790
    },
    {
      "epoch": 54.15838773772353,
      "grad_norm": 0.014053049497306347,
      "learning_rate": 4.586403633267102e-05,
      "loss": 0.00016052387654781343,
      "step": 190800
    },
    {
      "epoch": 54.16122622764689,
      "grad_norm": 0.023193303495645523,
      "learning_rate": 4.586119784274766e-05,
      "loss": 0.0002686984837055206,
      "step": 190810
    },
    {
      "epoch": 54.164064717570255,
      "grad_norm": 6.790426254272461,
      "learning_rate": 4.58583593528243e-05,
      "loss": 0.0017693564295768738,
      "step": 190820
    },
    {
      "epoch": 54.16690320749361,
      "grad_norm": 0.17603667080402374,
      "learning_rate": 4.585552086290094e-05,
      "loss": 0.0005218934267759324,
      "step": 190830
    },
    {
      "epoch": 54.169741697416974,
      "grad_norm": 0.021246036514639854,
      "learning_rate": 4.585268237297758e-05,
      "loss": 0.0008829269558191299,
      "step": 190840
    },
    {
      "epoch": 54.17258018734034,
      "grad_norm": 0.054241664707660675,
      "learning_rate": 4.5849843883054216e-05,
      "loss": 0.0005923217162489891,
      "step": 190850
    },
    {
      "epoch": 54.17541867726369,
      "grad_norm": 0.010939565487205982,
      "learning_rate": 4.584700539313085e-05,
      "loss": 0.00018359683454036713,
      "step": 190860
    },
    {
      "epoch": 54.178257167187056,
      "grad_norm": 0.12357290089130402,
      "learning_rate": 4.58441669032075e-05,
      "loss": 0.00020604655146598815,
      "step": 190870
    },
    {
      "epoch": 54.18109565711042,
      "grad_norm": 0.0812787115573883,
      "learning_rate": 4.5841328413284134e-05,
      "loss": 0.00022991299629211427,
      "step": 190880
    },
    {
      "epoch": 54.18393414703378,
      "grad_norm": 0.02822262980043888,
      "learning_rate": 4.5838489923360775e-05,
      "loss": 0.00026959292590618135,
      "step": 190890
    },
    {
      "epoch": 54.18677263695714,
      "grad_norm": 0.05768641084432602,
      "learning_rate": 4.5835651433437416e-05,
      "loss": 0.00035768505185842516,
      "step": 190900
    },
    {
      "epoch": 54.1896111268805,
      "grad_norm": 0.30397072434425354,
      "learning_rate": 4.583281294351405e-05,
      "loss": 0.00040937867015600204,
      "step": 190910
    },
    {
      "epoch": 54.19244961680386,
      "grad_norm": 0.023846931755542755,
      "learning_rate": 4.582997445359069e-05,
      "loss": 0.0002432873472571373,
      "step": 190920
    },
    {
      "epoch": 54.19528810672722,
      "grad_norm": 2.2404913902282715,
      "learning_rate": 4.5827135963667334e-05,
      "loss": 0.0005140436813235282,
      "step": 190930
    },
    {
      "epoch": 54.19812659665058,
      "grad_norm": 0.12774401903152466,
      "learning_rate": 4.582429747374397e-05,
      "loss": 0.0016571661457419396,
      "step": 190940
    },
    {
      "epoch": 54.200965086573945,
      "grad_norm": 0.23690098524093628,
      "learning_rate": 4.582145898382061e-05,
      "loss": 0.0003392687067389488,
      "step": 190950
    },
    {
      "epoch": 54.2038035764973,
      "grad_norm": 0.03348930552601814,
      "learning_rate": 4.5818620493897244e-05,
      "loss": 0.0003225687891244888,
      "step": 190960
    },
    {
      "epoch": 54.20664206642066,
      "grad_norm": 0.04760325327515602,
      "learning_rate": 4.581578200397389e-05,
      "loss": 0.0013542346656322478,
      "step": 190970
    },
    {
      "epoch": 54.209480556344026,
      "grad_norm": 0.43742257356643677,
      "learning_rate": 4.581294351405053e-05,
      "loss": 0.0010713983327150344,
      "step": 190980
    },
    {
      "epoch": 54.21231904626739,
      "grad_norm": 0.006574250757694244,
      "learning_rate": 4.581010502412716e-05,
      "loss": 0.0007963255047798157,
      "step": 190990
    },
    {
      "epoch": 54.215157536190745,
      "grad_norm": 0.02654959261417389,
      "learning_rate": 4.580726653420381e-05,
      "loss": 0.0007808137685060501,
      "step": 191000
    },
    {
      "epoch": 54.215157536190745,
      "eval_accuracy": 0.9786354676670693,
      "eval_loss": 0.07990726828575134,
      "eval_runtime": 32.521,
      "eval_samples_per_second": 483.595,
      "eval_steps_per_second": 7.564,
      "step": 191000
    },
    {
      "epoch": 54.21799602611411,
      "grad_norm": 0.027145924046635628,
      "learning_rate": 4.5804428044280444e-05,
      "loss": 0.00040116701275110244,
      "step": 191010
    },
    {
      "epoch": 54.22083451603747,
      "grad_norm": 2.6159069538116455,
      "learning_rate": 4.5801589554357086e-05,
      "loss": 0.0005386099219322205,
      "step": 191020
    },
    {
      "epoch": 54.223673005960826,
      "grad_norm": 0.7345849871635437,
      "learning_rate": 4.579875106443373e-05,
      "loss": 0.0004670847207307816,
      "step": 191030
    },
    {
      "epoch": 54.22651149588419,
      "grad_norm": 0.3353499472141266,
      "learning_rate": 4.579591257451036e-05,
      "loss": 0.0005030401051044464,
      "step": 191040
    },
    {
      "epoch": 54.22934998580755,
      "grad_norm": 0.7025876641273499,
      "learning_rate": 4.5793074084587e-05,
      "loss": 0.005337405204772949,
      "step": 191050
    },
    {
      "epoch": 54.23218847573091,
      "grad_norm": 0.5406635403633118,
      "learning_rate": 4.579023559466364e-05,
      "loss": 0.003859071433544159,
      "step": 191060
    },
    {
      "epoch": 54.23502696565427,
      "grad_norm": 0.032845284789800644,
      "learning_rate": 4.578739710474028e-05,
      "loss": 0.00019309762865304947,
      "step": 191070
    },
    {
      "epoch": 54.237865455577634,
      "grad_norm": 0.07537852972745895,
      "learning_rate": 4.578455861481692e-05,
      "loss": 0.013812538981437684,
      "step": 191080
    },
    {
      "epoch": 54.240703945501,
      "grad_norm": 0.052238769829273224,
      "learning_rate": 4.5781720124893555e-05,
      "loss": 0.00025481339544057845,
      "step": 191090
    },
    {
      "epoch": 54.24354243542435,
      "grad_norm": 0.03162602335214615,
      "learning_rate": 4.57788816349702e-05,
      "loss": 0.0003665471449494362,
      "step": 191100
    },
    {
      "epoch": 54.246380925347715,
      "grad_norm": 0.03463372960686684,
      "learning_rate": 4.577604314504684e-05,
      "loss": 0.002721059322357178,
      "step": 191110
    },
    {
      "epoch": 54.24921941527108,
      "grad_norm": 0.00753134535625577,
      "learning_rate": 4.577320465512347e-05,
      "loss": 0.0022253615781664847,
      "step": 191120
    },
    {
      "epoch": 54.252057905194434,
      "grad_norm": 0.012002123519778252,
      "learning_rate": 4.577036616520012e-05,
      "loss": 0.0007774662226438522,
      "step": 191130
    },
    {
      "epoch": 54.2548963951178,
      "grad_norm": 0.09558827430009842,
      "learning_rate": 4.5767527675276755e-05,
      "loss": 0.0033770311623811723,
      "step": 191140
    },
    {
      "epoch": 54.25773488504116,
      "grad_norm": 0.05416702851653099,
      "learning_rate": 4.5764689185353397e-05,
      "loss": 0.0009279221296310425,
      "step": 191150
    },
    {
      "epoch": 54.260573374964515,
      "grad_norm": 0.13274389505386353,
      "learning_rate": 4.576185069543003e-05,
      "loss": 0.0014615826308727264,
      "step": 191160
    },
    {
      "epoch": 54.26341186488788,
      "grad_norm": 0.06981343030929565,
      "learning_rate": 4.575901220550667e-05,
      "loss": 0.004239465296268463,
      "step": 191170
    },
    {
      "epoch": 54.26625035481124,
      "grad_norm": 0.011452543549239635,
      "learning_rate": 4.5756173715583314e-05,
      "loss": 0.001320442743599415,
      "step": 191180
    },
    {
      "epoch": 54.269088844734604,
      "grad_norm": 0.6174089908599854,
      "learning_rate": 4.575333522565995e-05,
      "loss": 0.0014120515435934067,
      "step": 191190
    },
    {
      "epoch": 54.27192733465796,
      "grad_norm": 1.648452877998352,
      "learning_rate": 4.575049673573659e-05,
      "loss": 0.004841911047697068,
      "step": 191200
    },
    {
      "epoch": 54.27476582458132,
      "grad_norm": 4.7957844734191895,
      "learning_rate": 4.574765824581323e-05,
      "loss": 0.002529970556497574,
      "step": 191210
    },
    {
      "epoch": 54.277604314504686,
      "grad_norm": 0.18139295279979706,
      "learning_rate": 4.5744819755889866e-05,
      "loss": 0.000805499404668808,
      "step": 191220
    },
    {
      "epoch": 54.28044280442804,
      "grad_norm": 0.7347800135612488,
      "learning_rate": 4.574198126596651e-05,
      "loss": 0.0025661103427410126,
      "step": 191230
    },
    {
      "epoch": 54.283281294351404,
      "grad_norm": 0.7614655494689941,
      "learning_rate": 4.573914277604315e-05,
      "loss": 0.0005474206060171128,
      "step": 191240
    },
    {
      "epoch": 54.28611978427477,
      "grad_norm": 0.7470982670783997,
      "learning_rate": 4.573630428611978e-05,
      "loss": 0.00046431403607130053,
      "step": 191250
    },
    {
      "epoch": 54.28895827419813,
      "grad_norm": 0.1361372172832489,
      "learning_rate": 4.5733465796196425e-05,
      "loss": 0.000633927620947361,
      "step": 191260
    },
    {
      "epoch": 54.291796764121486,
      "grad_norm": 14.023747444152832,
      "learning_rate": 4.5730627306273066e-05,
      "loss": 0.004489714652299881,
      "step": 191270
    },
    {
      "epoch": 54.29463525404485,
      "grad_norm": 0.1560022234916687,
      "learning_rate": 4.57277888163497e-05,
      "loss": 0.00188782699406147,
      "step": 191280
    },
    {
      "epoch": 54.29747374396821,
      "grad_norm": 0.041179411113262177,
      "learning_rate": 4.572495032642634e-05,
      "loss": 0.0007719064131379127,
      "step": 191290
    },
    {
      "epoch": 54.30031223389157,
      "grad_norm": 0.20201550424098969,
      "learning_rate": 4.572211183650298e-05,
      "loss": 0.0011780181899666785,
      "step": 191300
    },
    {
      "epoch": 54.30315072381493,
      "grad_norm": 0.24065889418125153,
      "learning_rate": 4.5719273346579625e-05,
      "loss": 0.01725877970457077,
      "step": 191310
    },
    {
      "epoch": 54.30598921373829,
      "grad_norm": 0.06295903772115707,
      "learning_rate": 4.571643485665626e-05,
      "loss": 0.00045367050915956497,
      "step": 191320
    },
    {
      "epoch": 54.30882770366165,
      "grad_norm": 0.2680228352546692,
      "learning_rate": 4.57135963667329e-05,
      "loss": 0.0006093390285968781,
      "step": 191330
    },
    {
      "epoch": 54.31166619358501,
      "grad_norm": 0.08256310224533081,
      "learning_rate": 4.571075787680954e-05,
      "loss": 0.0002137446776032448,
      "step": 191340
    },
    {
      "epoch": 54.314504683508375,
      "grad_norm": 0.0482572503387928,
      "learning_rate": 4.5707919386886177e-05,
      "loss": 0.007931867986917496,
      "step": 191350
    },
    {
      "epoch": 54.31734317343174,
      "grad_norm": 0.019631007686257362,
      "learning_rate": 4.570508089696282e-05,
      "loss": 0.0003249365836381912,
      "step": 191360
    },
    {
      "epoch": 54.32018166335509,
      "grad_norm": 0.007392289116978645,
      "learning_rate": 4.570224240703946e-05,
      "loss": 0.000434829480946064,
      "step": 191370
    },
    {
      "epoch": 54.323020153278456,
      "grad_norm": 0.025458846241235733,
      "learning_rate": 4.5699403917116094e-05,
      "loss": 0.0002754811197519302,
      "step": 191380
    },
    {
      "epoch": 54.32585864320182,
      "grad_norm": 0.9495893716812134,
      "learning_rate": 4.5696565427192735e-05,
      "loss": 0.0008845260366797447,
      "step": 191390
    },
    {
      "epoch": 54.328697133125175,
      "grad_norm": 0.07244095206260681,
      "learning_rate": 4.569372693726938e-05,
      "loss": 0.0005084015429019928,
      "step": 191400
    },
    {
      "epoch": 54.33153562304854,
      "grad_norm": 0.04121003299951553,
      "learning_rate": 4.569088844734601e-05,
      "loss": 0.0019967040047049523,
      "step": 191410
    },
    {
      "epoch": 54.3343741129719,
      "grad_norm": 0.04737447202205658,
      "learning_rate": 4.568804995742265e-05,
      "loss": 0.0006064316257834435,
      "step": 191420
    },
    {
      "epoch": 54.33721260289526,
      "grad_norm": 0.11682509630918503,
      "learning_rate": 4.5685211467499294e-05,
      "loss": 0.003230658173561096,
      "step": 191430
    },
    {
      "epoch": 54.34005109281862,
      "grad_norm": 0.11286338418722153,
      "learning_rate": 4.5682372977575935e-05,
      "loss": 0.00032396893948316574,
      "step": 191440
    },
    {
      "epoch": 54.34288958274198,
      "grad_norm": 0.02628742903470993,
      "learning_rate": 4.567953448765257e-05,
      "loss": 0.0008755277842283249,
      "step": 191450
    },
    {
      "epoch": 54.345728072665345,
      "grad_norm": 0.10322128236293793,
      "learning_rate": 4.5676695997729205e-05,
      "loss": 0.0001767423003911972,
      "step": 191460
    },
    {
      "epoch": 54.3485665625887,
      "grad_norm": 0.429061621427536,
      "learning_rate": 4.567385750780585e-05,
      "loss": 0.00015796199440956115,
      "step": 191470
    },
    {
      "epoch": 54.351405052512064,
      "grad_norm": 0.03771941736340523,
      "learning_rate": 4.567101901788249e-05,
      "loss": 9.199436753988266e-05,
      "step": 191480
    },
    {
      "epoch": 54.35424354243543,
      "grad_norm": 0.03777378425002098,
      "learning_rate": 4.566818052795913e-05,
      "loss": 0.0006996814161539077,
      "step": 191490
    },
    {
      "epoch": 54.35708203235878,
      "grad_norm": 1.254267692565918,
      "learning_rate": 4.566534203803577e-05,
      "loss": 0.005466943979263306,
      "step": 191500
    },
    {
      "epoch": 54.35708203235878,
      "eval_accuracy": 0.9782539581611241,
      "eval_loss": 0.08096545934677124,
      "eval_runtime": 32.7311,
      "eval_samples_per_second": 480.49,
      "eval_steps_per_second": 7.516,
      "step": 191500
    },
    {
      "epoch": 54.359920522282145,
      "grad_norm": 0.8003575205802917,
      "learning_rate": 4.5662503548112405e-05,
      "loss": 0.0006389133632183075,
      "step": 191510
    },
    {
      "epoch": 54.36275901220551,
      "grad_norm": 0.2379906177520752,
      "learning_rate": 4.5659665058189046e-05,
      "loss": 0.0004051385447382927,
      "step": 191520
    },
    {
      "epoch": 54.365597502128864,
      "grad_norm": 0.20272108912467957,
      "learning_rate": 4.565682656826569e-05,
      "loss": 0.0007701417431235314,
      "step": 191530
    },
    {
      "epoch": 54.36843599205223,
      "grad_norm": 0.13115806877613068,
      "learning_rate": 4.565398807834232e-05,
      "loss": 0.000862884521484375,
      "step": 191540
    },
    {
      "epoch": 54.37127448197559,
      "grad_norm": 0.6040709614753723,
      "learning_rate": 4.5651149588418963e-05,
      "loss": 0.0005554372444748878,
      "step": 191550
    },
    {
      "epoch": 54.37411297189895,
      "grad_norm": 0.09156836569309235,
      "learning_rate": 4.5648311098495605e-05,
      "loss": 0.004020366817712784,
      "step": 191560
    },
    {
      "epoch": 54.37695146182231,
      "grad_norm": 0.43785208463668823,
      "learning_rate": 4.5645472608572246e-05,
      "loss": 0.011412200331687928,
      "step": 191570
    },
    {
      "epoch": 54.37978995174567,
      "grad_norm": 0.21632596850395203,
      "learning_rate": 4.564263411864888e-05,
      "loss": 0.009718668460845948,
      "step": 191580
    },
    {
      "epoch": 54.382628441669034,
      "grad_norm": 1.2963097095489502,
      "learning_rate": 4.5639795628725515e-05,
      "loss": 0.01190505251288414,
      "step": 191590
    },
    {
      "epoch": 54.38546693159239,
      "grad_norm": 0.03205480799078941,
      "learning_rate": 4.5636957138802164e-05,
      "loss": 0.009194850176572799,
      "step": 191600
    },
    {
      "epoch": 54.38830542151575,
      "grad_norm": 5.241995811462402,
      "learning_rate": 4.56341186488788e-05,
      "loss": 0.007095793634653092,
      "step": 191610
    },
    {
      "epoch": 54.391143911439116,
      "grad_norm": 4.867777347564697,
      "learning_rate": 4.563128015895544e-05,
      "loss": 0.001996377855539322,
      "step": 191620
    },
    {
      "epoch": 54.39398240136248,
      "grad_norm": 0.162331223487854,
      "learning_rate": 4.562844166903208e-05,
      "loss": 0.0006814096122980118,
      "step": 191630
    },
    {
      "epoch": 54.396820891285834,
      "grad_norm": 0.01757028140127659,
      "learning_rate": 4.5625603179108715e-05,
      "loss": 0.0007355289533734321,
      "step": 191640
    },
    {
      "epoch": 54.3996593812092,
      "grad_norm": 0.07736075669527054,
      "learning_rate": 4.562276468918536e-05,
      "loss": 0.0003492781892418861,
      "step": 191650
    },
    {
      "epoch": 54.40249787113256,
      "grad_norm": 0.40973594784736633,
      "learning_rate": 4.5619926199262e-05,
      "loss": 0.0018367763608694077,
      "step": 191660
    },
    {
      "epoch": 54.405336361055916,
      "grad_norm": 1.6062086820602417,
      "learning_rate": 4.561708770933863e-05,
      "loss": 0.0016108294948935509,
      "step": 191670
    },
    {
      "epoch": 54.40817485097928,
      "grad_norm": 0.07022950798273087,
      "learning_rate": 4.5614249219415274e-05,
      "loss": 0.005348193645477295,
      "step": 191680
    },
    {
      "epoch": 54.41101334090264,
      "grad_norm": 0.13323922455310822,
      "learning_rate": 4.561141072949191e-05,
      "loss": 0.00034225676208734514,
      "step": 191690
    },
    {
      "epoch": 54.413851830826,
      "grad_norm": 0.4637349545955658,
      "learning_rate": 4.560857223956855e-05,
      "loss": 0.0010190952569246292,
      "step": 191700
    },
    {
      "epoch": 54.41669032074936,
      "grad_norm": 5.711320400238037,
      "learning_rate": 4.560573374964519e-05,
      "loss": 0.008157269656658172,
      "step": 191710
    },
    {
      "epoch": 54.41952881067272,
      "grad_norm": 2.7402095794677734,
      "learning_rate": 4.5602895259721826e-05,
      "loss": 0.0021525029093027114,
      "step": 191720
    },
    {
      "epoch": 54.422367300596086,
      "grad_norm": 0.2618250846862793,
      "learning_rate": 4.5600056769798474e-05,
      "loss": 0.010515743494033813,
      "step": 191730
    },
    {
      "epoch": 54.42520579051944,
      "grad_norm": 0.0976276844739914,
      "learning_rate": 4.559721827987511e-05,
      "loss": 0.0018458623439073563,
      "step": 191740
    },
    {
      "epoch": 54.428044280442805,
      "grad_norm": 0.01917622983455658,
      "learning_rate": 4.5594379789951743e-05,
      "loss": 0.004306823760271072,
      "step": 191750
    },
    {
      "epoch": 54.43088277036617,
      "grad_norm": 0.1566542237997055,
      "learning_rate": 4.559154130002839e-05,
      "loss": 0.00032712016254663465,
      "step": 191760
    },
    {
      "epoch": 54.43372126028952,
      "grad_norm": 0.034231625497341156,
      "learning_rate": 4.5588702810105026e-05,
      "loss": 0.0007139647379517556,
      "step": 191770
    },
    {
      "epoch": 54.436559750212886,
      "grad_norm": 0.08058899641036987,
      "learning_rate": 4.558586432018167e-05,
      "loss": 0.0005681518465280533,
      "step": 191780
    },
    {
      "epoch": 54.43939824013625,
      "grad_norm": 0.23930566012859344,
      "learning_rate": 4.55830258302583e-05,
      "loss": 0.012514838576316833,
      "step": 191790
    },
    {
      "epoch": 54.442236730059605,
      "grad_norm": 0.02976282499730587,
      "learning_rate": 4.5580187340334944e-05,
      "loss": 0.001925123855471611,
      "step": 191800
    },
    {
      "epoch": 54.44507521998297,
      "grad_norm": 0.026894567534327507,
      "learning_rate": 4.5577348850411585e-05,
      "loss": 0.0004089120775461197,
      "step": 191810
    },
    {
      "epoch": 54.44791370990633,
      "grad_norm": 0.9611183404922485,
      "learning_rate": 4.557451036048822e-05,
      "loss": 0.001247238926589489,
      "step": 191820
    },
    {
      "epoch": 54.450752199829694,
      "grad_norm": 0.09460238367319107,
      "learning_rate": 4.557167187056486e-05,
      "loss": 0.0009379588067531585,
      "step": 191830
    },
    {
      "epoch": 54.45359068975305,
      "grad_norm": 0.175171360373497,
      "learning_rate": 4.55688333806415e-05,
      "loss": 0.0015018945559859276,
      "step": 191840
    },
    {
      "epoch": 54.45642917967641,
      "grad_norm": Infinity,
      "learning_rate": 4.556599489071814e-05,
      "loss": 0.020250675082206727,
      "step": 191850
    },
    {
      "epoch": 54.459267669599775,
      "grad_norm": 1.2165013551712036,
      "learning_rate": 4.556344024978712e-05,
      "loss": 0.00045145489275455475,
      "step": 191860
    },
    {
      "epoch": 54.46210615952313,
      "grad_norm": 3.818896770477295,
      "learning_rate": 4.556060175986375e-05,
      "loss": 0.0018362969160079956,
      "step": 191870
    },
    {
      "epoch": 54.464944649446494,
      "grad_norm": 0.08423987776041031,
      "learning_rate": 4.555776326994039e-05,
      "loss": 0.0026127370074391365,
      "step": 191880
    },
    {
      "epoch": 54.46778313936986,
      "grad_norm": 1.9855378866195679,
      "learning_rate": 4.5554924780017034e-05,
      "loss": 0.0009028496220707894,
      "step": 191890
    },
    {
      "epoch": 54.47062162929321,
      "grad_norm": 0.06720281392335892,
      "learning_rate": 4.555208629009367e-05,
      "loss": 0.0006246268749237061,
      "step": 191900
    },
    {
      "epoch": 54.473460119216575,
      "grad_norm": 0.00843675434589386,
      "learning_rate": 4.554924780017032e-05,
      "loss": 0.0008774057030677795,
      "step": 191910
    },
    {
      "epoch": 54.47629860913994,
      "grad_norm": 0.053035154938697815,
      "learning_rate": 4.554640931024695e-05,
      "loss": 0.0019910309463739397,
      "step": 191920
    },
    {
      "epoch": 54.4791370990633,
      "grad_norm": 0.06042381748557091,
      "learning_rate": 4.5543570820323586e-05,
      "loss": 0.00042624231427907945,
      "step": 191930
    },
    {
      "epoch": 54.48197558898666,
      "grad_norm": 0.05435601994395256,
      "learning_rate": 4.5540732330400234e-05,
      "loss": 0.0016485599800944329,
      "step": 191940
    },
    {
      "epoch": 54.48481407891002,
      "grad_norm": 16.76787757873535,
      "learning_rate": 4.553789384047687e-05,
      "loss": 0.006288465112447739,
      "step": 191950
    },
    {
      "epoch": 54.48765256883338,
      "grad_norm": 0.1326131671667099,
      "learning_rate": 4.553505535055351e-05,
      "loss": 0.0003082128241658211,
      "step": 191960
    },
    {
      "epoch": 54.49049105875674,
      "grad_norm": 1.0724918842315674,
      "learning_rate": 4.5532216860630145e-05,
      "loss": 0.0008005261421203613,
      "step": 191970
    },
    {
      "epoch": 54.4933295486801,
      "grad_norm": 0.06680697947740555,
      "learning_rate": 4.5529378370706786e-05,
      "loss": 0.0012105032801628113,
      "step": 191980
    },
    {
      "epoch": 54.496168038603464,
      "grad_norm": 0.06247660145163536,
      "learning_rate": 4.552653988078343e-05,
      "loss": 0.0002882188186049461,
      "step": 191990
    },
    {
      "epoch": 54.49900652852683,
      "grad_norm": 0.06340223550796509,
      "learning_rate": 4.552370139086006e-05,
      "loss": 0.0006563013419508934,
      "step": 192000
    },
    {
      "epoch": 54.49900652852683,
      "eval_accuracy": 0.9772365994786036,
      "eval_loss": 0.08066019415855408,
      "eval_runtime": 32.2524,
      "eval_samples_per_second": 487.622,
      "eval_steps_per_second": 7.627,
      "step": 192000
    },
    {
      "epoch": 54.50184501845018,
      "grad_norm": 0.07750023156404495,
      "learning_rate": 4.5520862900936704e-05,
      "loss": 0.00027098413556814194,
      "step": 192010
    },
    {
      "epoch": 54.504683508373546,
      "grad_norm": 0.03031003102660179,
      "learning_rate": 4.5518024411013345e-05,
      "loss": 0.00022284649312496186,
      "step": 192020
    },
    {
      "epoch": 54.50752199829691,
      "grad_norm": 0.09574250876903534,
      "learning_rate": 4.551518592108998e-05,
      "loss": 0.0011160939931869507,
      "step": 192030
    },
    {
      "epoch": 54.510360488220265,
      "grad_norm": 0.046857986599206924,
      "learning_rate": 4.551234743116662e-05,
      "loss": 0.0003237653523683548,
      "step": 192040
    },
    {
      "epoch": 54.51319897814363,
      "grad_norm": 0.3462381660938263,
      "learning_rate": 4.550950894124326e-05,
      "loss": 0.0025655729696154594,
      "step": 192050
    },
    {
      "epoch": 54.51603746806699,
      "grad_norm": 0.10085567831993103,
      "learning_rate": 4.55066704513199e-05,
      "loss": 0.0005363073199987411,
      "step": 192060
    },
    {
      "epoch": 54.518875957990346,
      "grad_norm": 0.07148906588554382,
      "learning_rate": 4.550383196139654e-05,
      "loss": 0.0006646843627095223,
      "step": 192070
    },
    {
      "epoch": 54.52171444791371,
      "grad_norm": 1.7412410974502563,
      "learning_rate": 4.550099347147318e-05,
      "loss": 0.0007324749603867531,
      "step": 192080
    },
    {
      "epoch": 54.52455293783707,
      "grad_norm": 0.09977587312459946,
      "learning_rate": 4.5498154981549814e-05,
      "loss": 0.00046803466975688936,
      "step": 192090
    },
    {
      "epoch": 54.527391427760435,
      "grad_norm": 0.042045172303915024,
      "learning_rate": 4.5495316491626456e-05,
      "loss": 0.0005581462755799294,
      "step": 192100
    },
    {
      "epoch": 54.53022991768379,
      "grad_norm": 0.407324880361557,
      "learning_rate": 4.54924780017031e-05,
      "loss": 0.002821612358093262,
      "step": 192110
    },
    {
      "epoch": 54.53306840760715,
      "grad_norm": 2.7339792251586914,
      "learning_rate": 4.548963951177974e-05,
      "loss": 0.0010560296475887298,
      "step": 192120
    },
    {
      "epoch": 54.535906897530516,
      "grad_norm": 0.02168787270784378,
      "learning_rate": 4.548680102185637e-05,
      "loss": 0.00045020710676908494,
      "step": 192130
    },
    {
      "epoch": 54.53874538745387,
      "grad_norm": 0.12661434710025787,
      "learning_rate": 4.5483962531933014e-05,
      "loss": 0.0032897204160690307,
      "step": 192140
    },
    {
      "epoch": 54.541583877377235,
      "grad_norm": 0.5909566879272461,
      "learning_rate": 4.5481124042009656e-05,
      "loss": 0.0002688605338335037,
      "step": 192150
    },
    {
      "epoch": 54.5444223673006,
      "grad_norm": 17.908164978027344,
      "learning_rate": 4.547828555208629e-05,
      "loss": 0.006518945097923279,
      "step": 192160
    },
    {
      "epoch": 54.547260857223954,
      "grad_norm": 0.06297691911458969,
      "learning_rate": 4.547544706216293e-05,
      "loss": 0.0009724490344524383,
      "step": 192170
    },
    {
      "epoch": 54.55009934714732,
      "grad_norm": 0.029532356187701225,
      "learning_rate": 4.547260857223957e-05,
      "loss": 0.00041862446814775465,
      "step": 192180
    },
    {
      "epoch": 54.55293783707068,
      "grad_norm": 0.19218584895133972,
      "learning_rate": 4.546977008231621e-05,
      "loss": 0.000959034264087677,
      "step": 192190
    },
    {
      "epoch": 54.55577632699404,
      "grad_norm": 0.24306516349315643,
      "learning_rate": 4.546693159239285e-05,
      "loss": 0.0015700815245509147,
      "step": 192200
    },
    {
      "epoch": 54.5586148169174,
      "grad_norm": 0.06988168507814407,
      "learning_rate": 4.546409310246949e-05,
      "loss": 0.00020565036684274673,
      "step": 192210
    },
    {
      "epoch": 54.56145330684076,
      "grad_norm": 0.060943227261304855,
      "learning_rate": 4.5461254612546125e-05,
      "loss": 0.0007553787901997567,
      "step": 192220
    },
    {
      "epoch": 54.564291796764124,
      "grad_norm": 0.02212204411625862,
      "learning_rate": 4.5458416122622766e-05,
      "loss": 0.0007668403908610344,
      "step": 192230
    },
    {
      "epoch": 54.56713028668748,
      "grad_norm": 0.2982671856880188,
      "learning_rate": 4.545557763269941e-05,
      "loss": 0.0010435983538627624,
      "step": 192240
    },
    {
      "epoch": 54.56996877661084,
      "grad_norm": 0.040205419063568115,
      "learning_rate": 4.545273914277605e-05,
      "loss": 0.00040983259677886965,
      "step": 192250
    },
    {
      "epoch": 54.572807266534205,
      "grad_norm": 0.011486396193504333,
      "learning_rate": 4.5449900652852684e-05,
      "loss": 0.001012999564409256,
      "step": 192260
    },
    {
      "epoch": 54.57564575645756,
      "grad_norm": 0.01004727091640234,
      "learning_rate": 4.544706216292932e-05,
      "loss": 0.0005482640117406845,
      "step": 192270
    },
    {
      "epoch": 54.578484246380924,
      "grad_norm": 0.02840992622077465,
      "learning_rate": 4.5444223673005967e-05,
      "loss": 0.00026895590126514436,
      "step": 192280
    },
    {
      "epoch": 54.58132273630429,
      "grad_norm": 0.05997065082192421,
      "learning_rate": 4.54413851830826e-05,
      "loss": 0.00432254858314991,
      "step": 192290
    },
    {
      "epoch": 54.58416122622765,
      "grad_norm": 0.012514862231910229,
      "learning_rate": 4.543854669315924e-05,
      "loss": 0.00021007414907217025,
      "step": 192300
    },
    {
      "epoch": 54.586999716151006,
      "grad_norm": 0.16145235300064087,
      "learning_rate": 4.5435708203235884e-05,
      "loss": 0.0004065090790390968,
      "step": 192310
    },
    {
      "epoch": 54.58983820607437,
      "grad_norm": 0.09747538715600967,
      "learning_rate": 4.543286971331252e-05,
      "loss": 0.0020884845405817034,
      "step": 192320
    },
    {
      "epoch": 54.59267669599773,
      "grad_norm": 0.03529169410467148,
      "learning_rate": 4.543003122338916e-05,
      "loss": 0.0011177945882081986,
      "step": 192330
    },
    {
      "epoch": 54.59551518592109,
      "grad_norm": 1.405574917793274,
      "learning_rate": 4.54271927334658e-05,
      "loss": 0.005006754398345947,
      "step": 192340
    },
    {
      "epoch": 54.59835367584445,
      "grad_norm": 0.3183739185333252,
      "learning_rate": 4.5424354243542436e-05,
      "loss": 0.0007190024480223656,
      "step": 192350
    },
    {
      "epoch": 54.60119216576781,
      "grad_norm": 0.09176784008741379,
      "learning_rate": 4.542151575361908e-05,
      "loss": 0.0009590830653905868,
      "step": 192360
    },
    {
      "epoch": 54.604030655691176,
      "grad_norm": 0.5007919669151306,
      "learning_rate": 4.541867726369571e-05,
      "loss": 0.006760498136281967,
      "step": 192370
    },
    {
      "epoch": 54.60686914561453,
      "grad_norm": 2.3587849140167236,
      "learning_rate": 4.541583877377236e-05,
      "loss": 0.0063875623047351835,
      "step": 192380
    },
    {
      "epoch": 54.609707635537895,
      "grad_norm": 5.596359729766846,
      "learning_rate": 4.5413000283848995e-05,
      "loss": 0.0062640272080898285,
      "step": 192390
    },
    {
      "epoch": 54.61254612546126,
      "grad_norm": 0.5585590600967407,
      "learning_rate": 4.541016179392563e-05,
      "loss": 0.004681891202926636,
      "step": 192400
    },
    {
      "epoch": 54.61538461538461,
      "grad_norm": 0.1267448216676712,
      "learning_rate": 4.540732330400228e-05,
      "loss": 0.005992340296506882,
      "step": 192410
    },
    {
      "epoch": 54.618223105307976,
      "grad_norm": 0.028188111260533333,
      "learning_rate": 4.540448481407891e-05,
      "loss": 0.0015598913654685021,
      "step": 192420
    },
    {
      "epoch": 54.62106159523134,
      "grad_norm": 0.17758294939994812,
      "learning_rate": 4.540164632415555e-05,
      "loss": 0.00221580695360899,
      "step": 192430
    },
    {
      "epoch": 54.623900085154695,
      "grad_norm": 0.03579934686422348,
      "learning_rate": 4.5398807834232195e-05,
      "loss": 0.0005748918280005455,
      "step": 192440
    },
    {
      "epoch": 54.62673857507806,
      "grad_norm": 1.8916223049163818,
      "learning_rate": 4.539596934430883e-05,
      "loss": 0.0012983901426196099,
      "step": 192450
    },
    {
      "epoch": 54.62957706500142,
      "grad_norm": 0.11436627060174942,
      "learning_rate": 4.539313085438547e-05,
      "loss": 0.0026648728176951408,
      "step": 192460
    },
    {
      "epoch": 54.63241555492478,
      "grad_norm": 0.04040074348449707,
      "learning_rate": 4.5390292364462105e-05,
      "loss": 0.0006347080692648887,
      "step": 192470
    },
    {
      "epoch": 54.63525404484814,
      "grad_norm": 0.13617153465747833,
      "learning_rate": 4.5387453874538747e-05,
      "loss": 0.0013882244005799294,
      "step": 192480
    },
    {
      "epoch": 54.6380925347715,
      "grad_norm": 0.27459990978240967,
      "learning_rate": 4.538461538461539e-05,
      "loss": 0.0011465493589639664,
      "step": 192490
    },
    {
      "epoch": 54.640931024694865,
      "grad_norm": 0.037793755531311035,
      "learning_rate": 4.538177689469202e-05,
      "loss": 0.0018102325499057769,
      "step": 192500
    },
    {
      "epoch": 54.640931024694865,
      "eval_accuracy": 0.9724677306542888,
      "eval_loss": 0.09608685970306396,
      "eval_runtime": 32.5171,
      "eval_samples_per_second": 483.653,
      "eval_steps_per_second": 7.565,
      "step": 192500
    },
    {
      "epoch": 54.64376951461822,
      "grad_norm": 0.35484978556632996,
      "learning_rate": 4.5378938404768664e-05,
      "loss": 0.0008193304762244225,
      "step": 192510
    },
    {
      "epoch": 54.646608004541584,
      "grad_norm": 0.3018094003200531,
      "learning_rate": 4.5376099914845305e-05,
      "loss": 0.0027299633249640463,
      "step": 192520
    },
    {
      "epoch": 54.64944649446495,
      "grad_norm": 0.4788801074028015,
      "learning_rate": 4.537326142492194e-05,
      "loss": 0.001204531267285347,
      "step": 192530
    },
    {
      "epoch": 54.6522849843883,
      "grad_norm": 0.03987901657819748,
      "learning_rate": 4.537042293499859e-05,
      "loss": 0.0049213066697120665,
      "step": 192540
    },
    {
      "epoch": 54.655123474311665,
      "grad_norm": 0.854216992855072,
      "learning_rate": 4.536758444507522e-05,
      "loss": 0.0009148487821221352,
      "step": 192550
    },
    {
      "epoch": 54.65796196423503,
      "grad_norm": 0.4054808020591736,
      "learning_rate": 4.536474595515186e-05,
      "loss": 0.0007602997124195099,
      "step": 192560
    },
    {
      "epoch": 54.66080045415839,
      "grad_norm": 0.3868764638900757,
      "learning_rate": 4.53619074652285e-05,
      "loss": 0.0044997379183769224,
      "step": 192570
    },
    {
      "epoch": 54.66363894408175,
      "grad_norm": 10.473978042602539,
      "learning_rate": 4.535906897530514e-05,
      "loss": 0.0036855705082416536,
      "step": 192580
    },
    {
      "epoch": 54.66647743400511,
      "grad_norm": 0.29570889472961426,
      "learning_rate": 4.535623048538178e-05,
      "loss": 0.000539948046207428,
      "step": 192590
    },
    {
      "epoch": 54.66931592392847,
      "grad_norm": 1.1597963571548462,
      "learning_rate": 4.5353391995458416e-05,
      "loss": 0.0014783261343836785,
      "step": 192600
    },
    {
      "epoch": 54.67215441385183,
      "grad_norm": 5.333010673522949,
      "learning_rate": 4.535055350553506e-05,
      "loss": 0.0015417825430631637,
      "step": 192610
    },
    {
      "epoch": 54.67499290377519,
      "grad_norm": 0.2151205837726593,
      "learning_rate": 4.53477150156117e-05,
      "loss": 0.0007778137922286987,
      "step": 192620
    },
    {
      "epoch": 54.677831393698554,
      "grad_norm": 0.05176310986280441,
      "learning_rate": 4.534487652568833e-05,
      "loss": 0.0006184954196214676,
      "step": 192630
    },
    {
      "epoch": 54.68066988362191,
      "grad_norm": 0.5517154932022095,
      "learning_rate": 4.5342038035764975e-05,
      "loss": 0.0014133095741271974,
      "step": 192640
    },
    {
      "epoch": 54.68350837354527,
      "grad_norm": 0.08922632038593292,
      "learning_rate": 4.5339199545841616e-05,
      "loss": 0.005697987973690033,
      "step": 192650
    },
    {
      "epoch": 54.686346863468636,
      "grad_norm": 0.24071922898292542,
      "learning_rate": 4.533636105591825e-05,
      "loss": 0.0003132052719593048,
      "step": 192660
    },
    {
      "epoch": 54.689185353392,
      "grad_norm": 0.07995827496051788,
      "learning_rate": 4.533352256599489e-05,
      "loss": 0.0011485984548926353,
      "step": 192670
    },
    {
      "epoch": 54.692023843315354,
      "grad_norm": 0.2503539025783539,
      "learning_rate": 4.5330684076071533e-05,
      "loss": 0.0024651190266013145,
      "step": 192680
    },
    {
      "epoch": 54.69486233323872,
      "grad_norm": 0.03767113387584686,
      "learning_rate": 4.532784558614817e-05,
      "loss": 0.00045919399708509444,
      "step": 192690
    },
    {
      "epoch": 54.69770082316208,
      "grad_norm": 9.668213844299316,
      "learning_rate": 4.532500709622481e-05,
      "loss": 0.004905151575803757,
      "step": 192700
    },
    {
      "epoch": 54.700539313085436,
      "grad_norm": 10.275238990783691,
      "learning_rate": 4.532216860630145e-05,
      "loss": 0.003293055295944214,
      "step": 192710
    },
    {
      "epoch": 54.7033778030088,
      "grad_norm": 5.07054328918457,
      "learning_rate": 4.531933011637809e-05,
      "loss": 0.002890988439321518,
      "step": 192720
    },
    {
      "epoch": 54.70621629293216,
      "grad_norm": 0.4986066520214081,
      "learning_rate": 4.531649162645473e-05,
      "loss": 0.0004779331386089325,
      "step": 192730
    },
    {
      "epoch": 54.70905478285552,
      "grad_norm": 0.6505165100097656,
      "learning_rate": 4.531365313653137e-05,
      "loss": 0.0008390314877033233,
      "step": 192740
    },
    {
      "epoch": 54.71189327277888,
      "grad_norm": 1.590277075767517,
      "learning_rate": 4.531081464660801e-05,
      "loss": 0.0017888875678181647,
      "step": 192750
    },
    {
      "epoch": 54.71473176270224,
      "grad_norm": 8.684060096740723,
      "learning_rate": 4.5307976156684644e-05,
      "loss": 0.0011608406901359558,
      "step": 192760
    },
    {
      "epoch": 54.717570252625606,
      "grad_norm": 4.239928722381592,
      "learning_rate": 4.5305137666761285e-05,
      "loss": 0.0036556873470544814,
      "step": 192770
    },
    {
      "epoch": 54.72040874254896,
      "grad_norm": 0.10527706146240234,
      "learning_rate": 4.530229917683793e-05,
      "loss": 0.0003862097859382629,
      "step": 192780
    },
    {
      "epoch": 54.723247232472325,
      "grad_norm": 9.572259902954102,
      "learning_rate": 4.529946068691456e-05,
      "loss": 0.0059012148529291155,
      "step": 192790
    },
    {
      "epoch": 54.72608572239569,
      "grad_norm": 0.07620464265346527,
      "learning_rate": 4.52966221969912e-05,
      "loss": 0.00047458857297897337,
      "step": 192800
    },
    {
      "epoch": 54.72892421231904,
      "grad_norm": 0.18030978739261627,
      "learning_rate": 4.5293783707067844e-05,
      "loss": 0.0010402575135231018,
      "step": 192810
    },
    {
      "epoch": 54.731762702242406,
      "grad_norm": 0.04602370783686638,
      "learning_rate": 4.529094521714448e-05,
      "loss": 0.00092079509049654,
      "step": 192820
    },
    {
      "epoch": 54.73460119216577,
      "grad_norm": 0.19807742536067963,
      "learning_rate": 4.528810672722112e-05,
      "loss": 0.0005450891330838203,
      "step": 192830
    },
    {
      "epoch": 54.73743968208913,
      "grad_norm": 0.06712987273931503,
      "learning_rate": 4.528526823729776e-05,
      "loss": 0.0008508488535881042,
      "step": 192840
    },
    {
      "epoch": 54.74027817201249,
      "grad_norm": 0.07866635918617249,
      "learning_rate": 4.52824297473744e-05,
      "loss": 0.0011670637875795364,
      "step": 192850
    },
    {
      "epoch": 54.74311666193585,
      "grad_norm": 13.692626953125,
      "learning_rate": 4.527959125745104e-05,
      "loss": 0.004864080995321274,
      "step": 192860
    },
    {
      "epoch": 54.745955151859214,
      "grad_norm": 0.2779467701911926,
      "learning_rate": 4.527675276752767e-05,
      "loss": 0.0006695106625556946,
      "step": 192870
    },
    {
      "epoch": 54.74879364178257,
      "grad_norm": 16.66049575805664,
      "learning_rate": 4.527391427760432e-05,
      "loss": 0.004803113266825676,
      "step": 192880
    },
    {
      "epoch": 54.75163213170593,
      "grad_norm": 0.13514883816242218,
      "learning_rate": 4.5271075787680955e-05,
      "loss": 0.0004820266738533974,
      "step": 192890
    },
    {
      "epoch": 54.754470621629295,
      "grad_norm": 0.05200890451669693,
      "learning_rate": 4.5268237297757596e-05,
      "loss": 0.001515892520546913,
      "step": 192900
    },
    {
      "epoch": 54.75730911155265,
      "grad_norm": 0.08665496110916138,
      "learning_rate": 4.526539880783424e-05,
      "loss": 0.004966672882437706,
      "step": 192910
    },
    {
      "epoch": 54.760147601476014,
      "grad_norm": 1.787520408630371,
      "learning_rate": 4.526256031791087e-05,
      "loss": 0.0015240205451846124,
      "step": 192920
    },
    {
      "epoch": 54.76298609139938,
      "grad_norm": 0.05158241093158722,
      "learning_rate": 4.5259721827987514e-05,
      "loss": 0.0010311653837561606,
      "step": 192930
    },
    {
      "epoch": 54.76582458132274,
      "grad_norm": 0.009568643756210804,
      "learning_rate": 4.5256883338064155e-05,
      "loss": 0.0012820202857255936,
      "step": 192940
    },
    {
      "epoch": 54.768663071246095,
      "grad_norm": 4.9963884353637695,
      "learning_rate": 4.525404484814079e-05,
      "loss": 0.002583321928977966,
      "step": 192950
    },
    {
      "epoch": 54.77150156116946,
      "grad_norm": 2.6551594734191895,
      "learning_rate": 4.525120635821743e-05,
      "loss": 0.002155821584165096,
      "step": 192960
    },
    {
      "epoch": 54.77434005109282,
      "grad_norm": 0.12301301956176758,
      "learning_rate": 4.5248367868294066e-05,
      "loss": 0.0009524689987301827,
      "step": 192970
    },
    {
      "epoch": 54.77717854101618,
      "grad_norm": 0.026012232527136803,
      "learning_rate": 4.524552937837071e-05,
      "loss": 0.0019410010427236557,
      "step": 192980
    },
    {
      "epoch": 54.78001703093954,
      "grad_norm": 0.44197386503219604,
      "learning_rate": 4.524269088844735e-05,
      "loss": 0.0011832265183329583,
      "step": 192990
    },
    {
      "epoch": 54.7828555208629,
      "grad_norm": 0.2571312189102173,
      "learning_rate": 4.523985239852398e-05,
      "loss": 0.004839348420500755,
      "step": 193000
    },
    {
      "epoch": 54.7828555208629,
      "eval_accuracy": 0.9767279201373434,
      "eval_loss": 0.08199021965265274,
      "eval_runtime": 33.2696,
      "eval_samples_per_second": 472.714,
      "eval_steps_per_second": 7.394,
      "step": 193000
    },
    {
      "epoch": 54.78569401078626,
      "grad_norm": 0.08331780135631561,
      "learning_rate": 4.523701390860063e-05,
      "loss": 0.0020491404458880423,
      "step": 193010
    },
    {
      "epoch": 54.78853250070962,
      "grad_norm": 0.19433677196502686,
      "learning_rate": 4.5234175418677266e-05,
      "loss": 0.0013044068589806556,
      "step": 193020
    },
    {
      "epoch": 54.791370990632984,
      "grad_norm": 0.21276892721652985,
      "learning_rate": 4.52313369287539e-05,
      "loss": 0.003513157367706299,
      "step": 193030
    },
    {
      "epoch": 54.79420948055635,
      "grad_norm": 0.3858931362628937,
      "learning_rate": 4.522849843883055e-05,
      "loss": 0.0008163414895534515,
      "step": 193040
    },
    {
      "epoch": 54.7970479704797,
      "grad_norm": 0.18437929451465607,
      "learning_rate": 4.522565994890718e-05,
      "loss": 0.0025547361001372337,
      "step": 193050
    },
    {
      "epoch": 54.799886460403066,
      "grad_norm": 4.0028605461120605,
      "learning_rate": 4.5222821458983824e-05,
      "loss": 0.001390901580452919,
      "step": 193060
    },
    {
      "epoch": 54.80272495032643,
      "grad_norm": 0.01603393256664276,
      "learning_rate": 4.521998296906046e-05,
      "loss": 0.0005649816244840622,
      "step": 193070
    },
    {
      "epoch": 54.805563440249784,
      "grad_norm": 0.08990688621997833,
      "learning_rate": 4.52171444791371e-05,
      "loss": 0.0020457480102777483,
      "step": 193080
    },
    {
      "epoch": 54.80840193017315,
      "grad_norm": 0.24544215202331543,
      "learning_rate": 4.521430598921374e-05,
      "loss": 0.007690902054309845,
      "step": 193090
    },
    {
      "epoch": 54.81124042009651,
      "grad_norm": 0.20418989658355713,
      "learning_rate": 4.5211467499290376e-05,
      "loss": 0.004023807495832444,
      "step": 193100
    },
    {
      "epoch": 54.814078910019866,
      "grad_norm": 7.985023498535156,
      "learning_rate": 4.520862900936702e-05,
      "loss": 0.001903153397142887,
      "step": 193110
    },
    {
      "epoch": 54.81691739994323,
      "grad_norm": 0.24844545125961304,
      "learning_rate": 4.520579051944366e-05,
      "loss": 0.006008155643939972,
      "step": 193120
    },
    {
      "epoch": 54.81975588986659,
      "grad_norm": 1.743600606918335,
      "learning_rate": 4.5202952029520294e-05,
      "loss": 0.0020038703456521033,
      "step": 193130
    },
    {
      "epoch": 54.822594379789955,
      "grad_norm": 0.07064063847064972,
      "learning_rate": 4.520011353959694e-05,
      "loss": 0.0005411414429545402,
      "step": 193140
    },
    {
      "epoch": 54.82543286971331,
      "grad_norm": 0.26500457525253296,
      "learning_rate": 4.5197275049673576e-05,
      "loss": 0.0008420482277870178,
      "step": 193150
    },
    {
      "epoch": 54.82827135963667,
      "grad_norm": 0.776692807674408,
      "learning_rate": 4.519443655975021e-05,
      "loss": 0.000926315225660801,
      "step": 193160
    },
    {
      "epoch": 54.831109849560036,
      "grad_norm": 0.21970584988594055,
      "learning_rate": 4.519159806982685e-05,
      "loss": 0.0006067922338843345,
      "step": 193170
    },
    {
      "epoch": 54.83394833948339,
      "grad_norm": 4.706866264343262,
      "learning_rate": 4.5188759579903494e-05,
      "loss": 0.004598168656229973,
      "step": 193180
    },
    {
      "epoch": 54.836786829406755,
      "grad_norm": 0.07605768740177155,
      "learning_rate": 4.5185921089980135e-05,
      "loss": 0.000839291326701641,
      "step": 193190
    },
    {
      "epoch": 54.83962531933012,
      "grad_norm": 0.1045413538813591,
      "learning_rate": 4.518308260005677e-05,
      "loss": 0.0004973091185092926,
      "step": 193200
    },
    {
      "epoch": 54.84246380925348,
      "grad_norm": 0.18285517394542694,
      "learning_rate": 4.518024411013341e-05,
      "loss": 0.00015832800418138503,
      "step": 193210
    },
    {
      "epoch": 54.845302299176836,
      "grad_norm": 1.668381929397583,
      "learning_rate": 4.517740562021005e-05,
      "loss": 0.0008169155567884445,
      "step": 193220
    },
    {
      "epoch": 54.8481407891002,
      "grad_norm": 0.08601323515176773,
      "learning_rate": 4.517456713028669e-05,
      "loss": 0.0003814948722720146,
      "step": 193230
    },
    {
      "epoch": 54.85097927902356,
      "grad_norm": 0.08280467242002487,
      "learning_rate": 4.517172864036333e-05,
      "loss": 0.0011677952483296395,
      "step": 193240
    },
    {
      "epoch": 54.85381776894692,
      "grad_norm": 0.02753811702132225,
      "learning_rate": 4.516889015043997e-05,
      "loss": 0.00021603237837553024,
      "step": 193250
    },
    {
      "epoch": 54.85665625887028,
      "grad_norm": 2.4973671436309814,
      "learning_rate": 4.5166051660516604e-05,
      "loss": 0.0008990088477730751,
      "step": 193260
    },
    {
      "epoch": 54.859494748793644,
      "grad_norm": 0.6331532001495361,
      "learning_rate": 4.516321317059325e-05,
      "loss": 0.0007856853306293487,
      "step": 193270
    },
    {
      "epoch": 54.862333238717,
      "grad_norm": 0.06639384478330612,
      "learning_rate": 4.516037468066989e-05,
      "loss": 0.0006695657968521119,
      "step": 193280
    },
    {
      "epoch": 54.86517172864036,
      "grad_norm": 0.01670117676258087,
      "learning_rate": 4.515753619074652e-05,
      "loss": 0.0005840536206960679,
      "step": 193290
    },
    {
      "epoch": 54.868010218563725,
      "grad_norm": 1.174407720565796,
      "learning_rate": 4.515469770082316e-05,
      "loss": 0.0006215361878275871,
      "step": 193300
    },
    {
      "epoch": 54.87084870848709,
      "grad_norm": 0.09297282248735428,
      "learning_rate": 4.5151859210899804e-05,
      "loss": 0.0003429649397730827,
      "step": 193310
    },
    {
      "epoch": 54.873687198410444,
      "grad_norm": 0.49915459752082825,
      "learning_rate": 4.5149020720976446e-05,
      "loss": 0.006991142034530639,
      "step": 193320
    },
    {
      "epoch": 54.87652568833381,
      "grad_norm": 1.1704343557357788,
      "learning_rate": 4.514618223105308e-05,
      "loss": 0.0017220772802829742,
      "step": 193330
    },
    {
      "epoch": 54.87936417825717,
      "grad_norm": 0.034017205238342285,
      "learning_rate": 4.514334374112972e-05,
      "loss": 0.0009599035605788231,
      "step": 193340
    },
    {
      "epoch": 54.882202668180526,
      "grad_norm": 0.03104356676340103,
      "learning_rate": 4.514050525120636e-05,
      "loss": 0.001101686805486679,
      "step": 193350
    },
    {
      "epoch": 54.88504115810389,
      "grad_norm": 0.10257542133331299,
      "learning_rate": 4.5137666761283e-05,
      "loss": 0.0011461110785603522,
      "step": 193360
    },
    {
      "epoch": 54.88787964802725,
      "grad_norm": 0.2692425847053528,
      "learning_rate": 4.513482827135964e-05,
      "loss": 0.000727594830095768,
      "step": 193370
    },
    {
      "epoch": 54.89071813795061,
      "grad_norm": 0.14733290672302246,
      "learning_rate": 4.513198978143628e-05,
      "loss": 0.0003593308851122856,
      "step": 193380
    },
    {
      "epoch": 54.89355662787397,
      "grad_norm": 0.078899085521698,
      "learning_rate": 4.5129151291512915e-05,
      "loss": 0.0006048738956451416,
      "step": 193390
    },
    {
      "epoch": 54.89639511779733,
      "grad_norm": 0.6950100064277649,
      "learning_rate": 4.5126312801589557e-05,
      "loss": 0.0010959558188915254,
      "step": 193400
    },
    {
      "epoch": 54.899233607720696,
      "grad_norm": 6.848596096038818,
      "learning_rate": 4.51234743116662e-05,
      "loss": 0.0018643075600266457,
      "step": 193410
    },
    {
      "epoch": 54.90207209764405,
      "grad_norm": 4.271066188812256,
      "learning_rate": 4.512063582174283e-05,
      "loss": 0.0010663041844964027,
      "step": 193420
    },
    {
      "epoch": 54.904910587567414,
      "grad_norm": 0.3417695462703705,
      "learning_rate": 4.5117797331819474e-05,
      "loss": 0.01135827973484993,
      "step": 193430
    },
    {
      "epoch": 54.90774907749078,
      "grad_norm": 0.35392799973487854,
      "learning_rate": 4.5114958841896115e-05,
      "loss": 0.006626708060503006,
      "step": 193440
    },
    {
      "epoch": 54.91058756741413,
      "grad_norm": 0.010730382986366749,
      "learning_rate": 4.511212035197275e-05,
      "loss": 0.0005681196227669716,
      "step": 193450
    },
    {
      "epoch": 54.913426057337496,
      "grad_norm": 0.28312432765960693,
      "learning_rate": 4.510928186204939e-05,
      "loss": 0.0034432146698236466,
      "step": 193460
    },
    {
      "epoch": 54.91626454726086,
      "grad_norm": 0.3397130072116852,
      "learning_rate": 4.510644337212603e-05,
      "loss": 0.0006520602852106095,
      "step": 193470
    },
    {
      "epoch": 54.919103037184215,
      "grad_norm": 0.05224641039967537,
      "learning_rate": 4.5103604882202674e-05,
      "loss": 0.0007474120706319809,
      "step": 193480
    },
    {
      "epoch": 54.92194152710758,
      "grad_norm": 0.0948987752199173,
      "learning_rate": 4.510076639227931e-05,
      "loss": 0.00033192243427038193,
      "step": 193490
    },
    {
      "epoch": 54.92478001703094,
      "grad_norm": 0.008073879405856133,
      "learning_rate": 4.509792790235594e-05,
      "loss": 0.00133442971855402,
      "step": 193500
    },
    {
      "epoch": 54.92478001703094,
      "eval_accuracy": 0.9786354676670693,
      "eval_loss": 0.07588300108909607,
      "eval_runtime": 32.7319,
      "eval_samples_per_second": 480.48,
      "eval_steps_per_second": 7.516,
      "step": 193500
    },
    {
      "epoch": 54.9276185069543,
      "grad_norm": 0.1921972930431366,
      "learning_rate": 4.509508941243259e-05,
      "loss": 0.0004267564043402672,
      "step": 193510
    },
    {
      "epoch": 54.93045699687766,
      "grad_norm": 0.6815023422241211,
      "learning_rate": 4.5092250922509226e-05,
      "loss": 0.0007739594206213951,
      "step": 193520
    },
    {
      "epoch": 54.93329548680102,
      "grad_norm": 0.09581562876701355,
      "learning_rate": 4.508941243258587e-05,
      "loss": 0.0017895305529236794,
      "step": 193530
    },
    {
      "epoch": 54.936133976724385,
      "grad_norm": 0.014870029874145985,
      "learning_rate": 4.508657394266251e-05,
      "loss": 0.0023726608604192735,
      "step": 193540
    },
    {
      "epoch": 54.93897246664774,
      "grad_norm": 0.04893457517027855,
      "learning_rate": 4.508373545273914e-05,
      "loss": 0.0003431588411331177,
      "step": 193550
    },
    {
      "epoch": 54.9418109565711,
      "grad_norm": 1.321195363998413,
      "learning_rate": 4.5080896962815785e-05,
      "loss": 0.001604944095015526,
      "step": 193560
    },
    {
      "epoch": 54.944649446494466,
      "grad_norm": 11.417872428894043,
      "learning_rate": 4.5078058472892426e-05,
      "loss": 0.0034023217856884004,
      "step": 193570
    },
    {
      "epoch": 54.94748793641783,
      "grad_norm": 0.7584550380706787,
      "learning_rate": 4.507521998296906e-05,
      "loss": 0.0006143055856227875,
      "step": 193580
    },
    {
      "epoch": 54.950326426341185,
      "grad_norm": 0.42939451336860657,
      "learning_rate": 4.50723814930457e-05,
      "loss": 0.0034742169082164764,
      "step": 193590
    },
    {
      "epoch": 54.95316491626455,
      "grad_norm": 0.020512780174613,
      "learning_rate": 4.5069543003122337e-05,
      "loss": 0.0006411071866750717,
      "step": 193600
    },
    {
      "epoch": 54.95600340618791,
      "grad_norm": 0.04974746331572533,
      "learning_rate": 4.5066704513198985e-05,
      "loss": 0.000949045829474926,
      "step": 193610
    },
    {
      "epoch": 54.95884189611127,
      "grad_norm": 0.8657918572425842,
      "learning_rate": 4.506386602327562e-05,
      "loss": 0.009335582703351974,
      "step": 193620
    },
    {
      "epoch": 54.96168038603463,
      "grad_norm": 0.019998984411358833,
      "learning_rate": 4.5061027533352254e-05,
      "loss": 0.0009070407599210739,
      "step": 193630
    },
    {
      "epoch": 54.96451887595799,
      "grad_norm": 0.10608475655317307,
      "learning_rate": 4.50581890434289e-05,
      "loss": 0.00045571401715278627,
      "step": 193640
    },
    {
      "epoch": 54.96735736588135,
      "grad_norm": 0.02978459931910038,
      "learning_rate": 4.505535055350554e-05,
      "loss": 0.001737365499138832,
      "step": 193650
    },
    {
      "epoch": 54.97019585580471,
      "grad_norm": 0.13439415395259857,
      "learning_rate": 4.505251206358218e-05,
      "loss": 0.0014687672257423401,
      "step": 193660
    },
    {
      "epoch": 54.973034345728074,
      "grad_norm": 0.202483668923378,
      "learning_rate": 4.504967357365882e-05,
      "loss": 0.005338087305426597,
      "step": 193670
    },
    {
      "epoch": 54.97587283565144,
      "grad_norm": 4.919798374176025,
      "learning_rate": 4.5046835083735454e-05,
      "loss": 0.0010338090360164642,
      "step": 193680
    },
    {
      "epoch": 54.97871132557479,
      "grad_norm": 6.3843183517456055,
      "learning_rate": 4.5043996593812095e-05,
      "loss": 0.0020647086203098297,
      "step": 193690
    },
    {
      "epoch": 54.981549815498155,
      "grad_norm": 0.18966571986675262,
      "learning_rate": 4.504115810388873e-05,
      "loss": 0.002423659712076187,
      "step": 193700
    },
    {
      "epoch": 54.98438830542152,
      "grad_norm": 0.2408057004213333,
      "learning_rate": 4.503831961396537e-05,
      "loss": 0.0005019035190343857,
      "step": 193710
    },
    {
      "epoch": 54.987226795344874,
      "grad_norm": 0.4083288609981537,
      "learning_rate": 4.503548112404201e-05,
      "loss": 0.0006309045478701591,
      "step": 193720
    },
    {
      "epoch": 54.99006528526824,
      "grad_norm": 0.09742613136768341,
      "learning_rate": 4.503264263411865e-05,
      "loss": 0.0005408357828855515,
      "step": 193730
    },
    {
      "epoch": 54.9929037751916,
      "grad_norm": 0.137253075838089,
      "learning_rate": 4.5029804144195296e-05,
      "loss": 0.000510379858314991,
      "step": 193740
    },
    {
      "epoch": 54.995742265114956,
      "grad_norm": 0.03280330449342728,
      "learning_rate": 4.502696565427193e-05,
      "loss": 0.0007651934400200843,
      "step": 193750
    },
    {
      "epoch": 54.99858075503832,
      "grad_norm": 0.17405132949352264,
      "learning_rate": 4.5024127164348565e-05,
      "loss": 0.00031260661780834197,
      "step": 193760
    },
    {
      "epoch": 55.00141924496168,
      "grad_norm": 0.02616366744041443,
      "learning_rate": 4.502128867442521e-05,
      "loss": 0.0008971180766820907,
      "step": 193770
    },
    {
      "epoch": 55.004257734885044,
      "grad_norm": 0.6239033937454224,
      "learning_rate": 4.501845018450185e-05,
      "loss": 0.00015637893229722977,
      "step": 193780
    },
    {
      "epoch": 55.0070962248084,
      "grad_norm": 0.12782365083694458,
      "learning_rate": 4.501561169457849e-05,
      "loss": 0.00028281304985284806,
      "step": 193790
    },
    {
      "epoch": 55.00993471473176,
      "grad_norm": 0.07772985845804214,
      "learning_rate": 4.5012773204655123e-05,
      "loss": 0.00031991302967071533,
      "step": 193800
    },
    {
      "epoch": 55.012773204655126,
      "grad_norm": 0.14819084107875824,
      "learning_rate": 4.5009934714731765e-05,
      "loss": 0.002174023352563381,
      "step": 193810
    },
    {
      "epoch": 55.01561169457848,
      "grad_norm": 0.10162334889173508,
      "learning_rate": 4.5007096224808406e-05,
      "loss": 0.0019311131909489632,
      "step": 193820
    },
    {
      "epoch": 55.018450184501845,
      "grad_norm": 0.009083440527319908,
      "learning_rate": 4.500425773488504e-05,
      "loss": 0.0012231888249516487,
      "step": 193830
    },
    {
      "epoch": 55.02128867442521,
      "grad_norm": 19.13519859313965,
      "learning_rate": 4.500141924496168e-05,
      "loss": 0.004359965398907662,
      "step": 193840
    },
    {
      "epoch": 55.02412716434856,
      "grad_norm": 20.1438045501709,
      "learning_rate": 4.4998580755038324e-05,
      "loss": 0.021589916944503785,
      "step": 193850
    },
    {
      "epoch": 55.026965654271926,
      "grad_norm": 0.11773016303777695,
      "learning_rate": 4.499574226511496e-05,
      "loss": 0.0017927112057805061,
      "step": 193860
    },
    {
      "epoch": 55.02980414419529,
      "grad_norm": 0.005287750158458948,
      "learning_rate": 4.49929037751916e-05,
      "loss": 0.0027223965153098106,
      "step": 193870
    },
    {
      "epoch": 55.03264263411865,
      "grad_norm": 24.35765838623047,
      "learning_rate": 4.499006528526824e-05,
      "loss": 0.006979382783174515,
      "step": 193880
    },
    {
      "epoch": 55.03548112404201,
      "grad_norm": 0.07231361418962479,
      "learning_rate": 4.4987226795344875e-05,
      "loss": 0.00035792943090200423,
      "step": 193890
    },
    {
      "epoch": 55.03831961396537,
      "grad_norm": 0.05828948691487312,
      "learning_rate": 4.498438830542152e-05,
      "loss": 0.0013010494410991668,
      "step": 193900
    },
    {
      "epoch": 55.04115810388873,
      "grad_norm": 0.03026546537876129,
      "learning_rate": 4.498154981549816e-05,
      "loss": 0.0007915075868368149,
      "step": 193910
    },
    {
      "epoch": 55.04399659381209,
      "grad_norm": 0.022937564179301262,
      "learning_rate": 4.497871132557479e-05,
      "loss": 0.0001715485006570816,
      "step": 193920
    },
    {
      "epoch": 55.04683508373545,
      "grad_norm": 0.013667608611285686,
      "learning_rate": 4.4975872835651434e-05,
      "loss": 0.00044150408357381823,
      "step": 193930
    },
    {
      "epoch": 55.049673573658815,
      "grad_norm": 0.15419965982437134,
      "learning_rate": 4.4973034345728076e-05,
      "loss": 0.003681786358356476,
      "step": 193940
    },
    {
      "epoch": 55.05251206358217,
      "grad_norm": 0.12160047143697739,
      "learning_rate": 4.497019585580472e-05,
      "loss": 0.0008872700855135918,
      "step": 193950
    },
    {
      "epoch": 55.055350553505534,
      "grad_norm": 9.364887237548828,
      "learning_rate": 4.496735736588135e-05,
      "loss": 0.006453837454319,
      "step": 193960
    },
    {
      "epoch": 55.0581890434289,
      "grad_norm": 2.1096715927124023,
      "learning_rate": 4.496451887595799e-05,
      "loss": 0.003289201110601425,
      "step": 193970
    },
    {
      "epoch": 55.06102753335226,
      "grad_norm": 0.014158018864691257,
      "learning_rate": 4.4961680386034634e-05,
      "loss": 0.003348197788000107,
      "step": 193980
    },
    {
      "epoch": 55.063866023275615,
      "grad_norm": 0.2770717442035675,
      "learning_rate": 4.495884189611127e-05,
      "loss": 0.0012485940009355546,
      "step": 193990
    },
    {
      "epoch": 55.06670451319898,
      "grad_norm": 0.22681371867656708,
      "learning_rate": 4.495600340618791e-05,
      "loss": 0.0007808644324541092,
      "step": 194000
    },
    {
      "epoch": 55.06670451319898,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.0930028110742569,
      "eval_runtime": 32.8128,
      "eval_samples_per_second": 479.294,
      "eval_steps_per_second": 7.497,
      "step": 194000
    },
    {
      "epoch": 55.06954300312234,
      "grad_norm": 0.029804272577166557,
      "learning_rate": 4.495316491626455e-05,
      "loss": 0.0011145146563649178,
      "step": 194010
    },
    {
      "epoch": 55.0723814930457,
      "grad_norm": 0.1192685067653656,
      "learning_rate": 4.4950610275333525e-05,
      "loss": 0.003115515597164631,
      "step": 194020
    },
    {
      "epoch": 55.07521998296906,
      "grad_norm": 0.020017346367239952,
      "learning_rate": 4.4947771785410166e-05,
      "loss": 0.0010362906381487847,
      "step": 194030
    },
    {
      "epoch": 55.07805847289242,
      "grad_norm": 0.4125311076641083,
      "learning_rate": 4.49449332954868e-05,
      "loss": 0.0002985116094350815,
      "step": 194040
    },
    {
      "epoch": 55.080896962815785,
      "grad_norm": 0.043651483952999115,
      "learning_rate": 4.494209480556344e-05,
      "loss": 0.012575843930244445,
      "step": 194050
    },
    {
      "epoch": 55.08373545273914,
      "grad_norm": 3.485382080078125,
      "learning_rate": 4.4939256315640084e-05,
      "loss": 0.000773235410451889,
      "step": 194060
    },
    {
      "epoch": 55.086573942662504,
      "grad_norm": 0.29404959082603455,
      "learning_rate": 4.493641782571672e-05,
      "loss": 0.00020912550389766692,
      "step": 194070
    },
    {
      "epoch": 55.08941243258587,
      "grad_norm": 0.11801493167877197,
      "learning_rate": 4.493357933579336e-05,
      "loss": 0.0007062975317239761,
      "step": 194080
    },
    {
      "epoch": 55.09225092250922,
      "grad_norm": 0.039400991052389145,
      "learning_rate": 4.493074084587e-05,
      "loss": 0.0002997206524014473,
      "step": 194090
    },
    {
      "epoch": 55.095089412432586,
      "grad_norm": 0.3497559428215027,
      "learning_rate": 4.4927902355946635e-05,
      "loss": 0.00040965750813484194,
      "step": 194100
    },
    {
      "epoch": 55.09792790235595,
      "grad_norm": 9.693792343139648,
      "learning_rate": 4.492506386602328e-05,
      "loss": 0.001997986435890198,
      "step": 194110
    },
    {
      "epoch": 55.100766392279304,
      "grad_norm": 0.0255607720464468,
      "learning_rate": 4.492222537609992e-05,
      "loss": 0.0002866353839635849,
      "step": 194120
    },
    {
      "epoch": 55.10360488220267,
      "grad_norm": 0.06255605816841125,
      "learning_rate": 4.491938688617656e-05,
      "loss": 0.0029277056455612183,
      "step": 194130
    },
    {
      "epoch": 55.10644337212603,
      "grad_norm": 0.013188205659389496,
      "learning_rate": 4.4916548396253194e-05,
      "loss": 0.00012948643416166306,
      "step": 194140
    },
    {
      "epoch": 55.10928186204939,
      "grad_norm": 0.06642240285873413,
      "learning_rate": 4.4913709906329836e-05,
      "loss": 0.0012414507567882539,
      "step": 194150
    },
    {
      "epoch": 55.11212035197275,
      "grad_norm": 0.6423050165176392,
      "learning_rate": 4.491087141640648e-05,
      "loss": 0.0002448853105306625,
      "step": 194160
    },
    {
      "epoch": 55.11495884189611,
      "grad_norm": 0.04526769742369652,
      "learning_rate": 4.490803292648311e-05,
      "loss": 0.0002661734819412231,
      "step": 194170
    },
    {
      "epoch": 55.117797331819474,
      "grad_norm": 0.6339390873908997,
      "learning_rate": 4.490519443655975e-05,
      "loss": 0.001001104898750782,
      "step": 194180
    },
    {
      "epoch": 55.12063582174283,
      "grad_norm": 0.02418268658220768,
      "learning_rate": 4.4902355946636394e-05,
      "loss": 0.00039238333702087405,
      "step": 194190
    },
    {
      "epoch": 55.12347431166619,
      "grad_norm": 0.7863162755966187,
      "learning_rate": 4.489951745671303e-05,
      "loss": 0.0004016367718577385,
      "step": 194200
    },
    {
      "epoch": 55.126312801589556,
      "grad_norm": 0.0613560825586319,
      "learning_rate": 4.489667896678967e-05,
      "loss": 0.0010295888409018517,
      "step": 194210
    },
    {
      "epoch": 55.12915129151291,
      "grad_norm": 0.010939335450530052,
      "learning_rate": 4.489384047686631e-05,
      "loss": 0.0012508980929851533,
      "step": 194220
    },
    {
      "epoch": 55.131989781436275,
      "grad_norm": 0.10237230360507965,
      "learning_rate": 4.4891001986942946e-05,
      "loss": 0.0010218488052487374,
      "step": 194230
    },
    {
      "epoch": 55.13482827135964,
      "grad_norm": 0.02024519257247448,
      "learning_rate": 4.488816349701959e-05,
      "loss": 0.0027677293866872786,
      "step": 194240
    },
    {
      "epoch": 55.137666761283,
      "grad_norm": 5.052537441253662,
      "learning_rate": 4.488532500709623e-05,
      "loss": 0.002382362261414528,
      "step": 194250
    },
    {
      "epoch": 55.140505251206356,
      "grad_norm": 0.8396875262260437,
      "learning_rate": 4.488248651717287e-05,
      "loss": 0.0004217498004436493,
      "step": 194260
    },
    {
      "epoch": 55.14334374112972,
      "grad_norm": 1.4175028800964355,
      "learning_rate": 4.4879648027249505e-05,
      "loss": 0.0008722422644495964,
      "step": 194270
    },
    {
      "epoch": 55.14618223105308,
      "grad_norm": 0.09628958255052567,
      "learning_rate": 4.487680953732614e-05,
      "loss": 0.0005133228376507759,
      "step": 194280
    },
    {
      "epoch": 55.14902072097644,
      "grad_norm": 0.11109407991170883,
      "learning_rate": 4.487397104740279e-05,
      "loss": 0.0007476769387722015,
      "step": 194290
    },
    {
      "epoch": 55.1518592108998,
      "grad_norm": 0.23278552293777466,
      "learning_rate": 4.487113255747942e-05,
      "loss": 0.0007284300401806832,
      "step": 194300
    },
    {
      "epoch": 55.154697700823164,
      "grad_norm": 0.22267752885818481,
      "learning_rate": 4.4868294067556064e-05,
      "loss": 0.0003184758126735687,
      "step": 194310
    },
    {
      "epoch": 55.15753619074652,
      "grad_norm": 0.17797483503818512,
      "learning_rate": 4.4865455577632705e-05,
      "loss": 0.0001699177548289299,
      "step": 194320
    },
    {
      "epoch": 55.16037468066988,
      "grad_norm": 0.06253586709499359,
      "learning_rate": 4.486261708770934e-05,
      "loss": 0.0012514950707554818,
      "step": 194330
    },
    {
      "epoch": 55.163213170593245,
      "grad_norm": 16.923171997070312,
      "learning_rate": 4.485977859778598e-05,
      "loss": 0.0036897551268339156,
      "step": 194340
    },
    {
      "epoch": 55.16605166051661,
      "grad_norm": 2.552764415740967,
      "learning_rate": 4.485694010786262e-05,
      "loss": 0.0005691768601536751,
      "step": 194350
    },
    {
      "epoch": 55.168890150439964,
      "grad_norm": 0.4761795699596405,
      "learning_rate": 4.485410161793926e-05,
      "loss": 0.0009078178554773331,
      "step": 194360
    },
    {
      "epoch": 55.17172864036333,
      "grad_norm": 0.0016080249333754182,
      "learning_rate": 4.48512631280159e-05,
      "loss": 0.000337614119052887,
      "step": 194370
    },
    {
      "epoch": 55.17456713028669,
      "grad_norm": 0.14274680614471436,
      "learning_rate": 4.484842463809253e-05,
      "loss": 0.0003740767017006874,
      "step": 194380
    },
    {
      "epoch": 55.177405620210045,
      "grad_norm": 0.044082410633563995,
      "learning_rate": 4.4845586148169174e-05,
      "loss": 0.0003754129633307457,
      "step": 194390
    },
    {
      "epoch": 55.18024411013341,
      "grad_norm": 0.00775542389601469,
      "learning_rate": 4.4842747658245816e-05,
      "loss": 0.0017359802499413491,
      "step": 194400
    },
    {
      "epoch": 55.18308260005677,
      "grad_norm": 0.038673534989356995,
      "learning_rate": 4.483990916832245e-05,
      "loss": 0.00019173212349414824,
      "step": 194410
    },
    {
      "epoch": 55.185921089980134,
      "grad_norm": 0.5505730509757996,
      "learning_rate": 4.48370706783991e-05,
      "loss": 0.00036329366266727445,
      "step": 194420
    },
    {
      "epoch": 55.18875957990349,
      "grad_norm": 0.2965264916419983,
      "learning_rate": 4.483423218847573e-05,
      "loss": 0.00038316380232572556,
      "step": 194430
    },
    {
      "epoch": 55.19159806982685,
      "grad_norm": 0.03618954122066498,
      "learning_rate": 4.483139369855237e-05,
      "loss": 0.0009590055793523789,
      "step": 194440
    },
    {
      "epoch": 55.194436559750216,
      "grad_norm": 0.06821884214878082,
      "learning_rate": 4.4828555208629016e-05,
      "loss": 0.000608360767364502,
      "step": 194450
    },
    {
      "epoch": 55.19727504967357,
      "grad_norm": 0.06027701869606972,
      "learning_rate": 4.482571671870565e-05,
      "loss": 0.0002906423062086105,
      "step": 194460
    },
    {
      "epoch": 55.200113539596934,
      "grad_norm": 0.0756693184375763,
      "learning_rate": 4.482287822878229e-05,
      "loss": 0.0008252138271927833,
      "step": 194470
    },
    {
      "epoch": 55.2029520295203,
      "grad_norm": 0.12636911869049072,
      "learning_rate": 4.4820039738858926e-05,
      "loss": 0.001247130148112774,
      "step": 194480
    },
    {
      "epoch": 55.20579051944365,
      "grad_norm": 0.1515313684940338,
      "learning_rate": 4.481720124893557e-05,
      "loss": 0.0009822078049182893,
      "step": 194490
    },
    {
      "epoch": 55.208629009367016,
      "grad_norm": 0.07619692385196686,
      "learning_rate": 4.481436275901221e-05,
      "loss": 0.00039006806910037997,
      "step": 194500
    },
    {
      "epoch": 55.208629009367016,
      "eval_accuracy": 0.9786990525847269,
      "eval_loss": 0.08498641848564148,
      "eval_runtime": 32.911,
      "eval_samples_per_second": 477.864,
      "eval_steps_per_second": 7.475,
      "step": 194500
    },
    {
      "epoch": 55.21146749929038,
      "grad_norm": 0.22936800122261047,
      "learning_rate": 4.4811524269088844e-05,
      "loss": 0.000566130317747593,
      "step": 194510
    },
    {
      "epoch": 55.21430598921374,
      "grad_norm": 14.665034294128418,
      "learning_rate": 4.4808685779165485e-05,
      "loss": 0.014037653803825378,
      "step": 194520
    },
    {
      "epoch": 55.2171444791371,
      "grad_norm": 0.038242414593696594,
      "learning_rate": 4.4805847289242126e-05,
      "loss": 0.0020670820027589797,
      "step": 194530
    },
    {
      "epoch": 55.21998296906046,
      "grad_norm": 0.07455743104219437,
      "learning_rate": 4.480300879931876e-05,
      "loss": 0.002356105111539364,
      "step": 194540
    },
    {
      "epoch": 55.22282145898382,
      "grad_norm": 0.24001234769821167,
      "learning_rate": 4.480017030939541e-05,
      "loss": 0.0019136454910039901,
      "step": 194550
    },
    {
      "epoch": 55.22565994890718,
      "grad_norm": 0.0326654314994812,
      "learning_rate": 4.4797331819472044e-05,
      "loss": 0.00044527091085910797,
      "step": 194560
    },
    {
      "epoch": 55.22849843883054,
      "grad_norm": 18.23886489868164,
      "learning_rate": 4.479449332954868e-05,
      "loss": 0.016130602359771727,
      "step": 194570
    },
    {
      "epoch": 55.231336928753905,
      "grad_norm": 0.0644712895154953,
      "learning_rate": 4.479165483962532e-05,
      "loss": 0.0022852616384625435,
      "step": 194580
    },
    {
      "epoch": 55.23417541867726,
      "grad_norm": 0.23194275796413422,
      "learning_rate": 4.478881634970196e-05,
      "loss": 0.00044996365904808043,
      "step": 194590
    },
    {
      "epoch": 55.23701390860062,
      "grad_norm": 0.044111207127571106,
      "learning_rate": 4.47859778597786e-05,
      "loss": 0.00026917587965726855,
      "step": 194600
    },
    {
      "epoch": 55.239852398523986,
      "grad_norm": 0.7198662757873535,
      "learning_rate": 4.478313936985524e-05,
      "loss": 0.0003333861008286476,
      "step": 194610
    },
    {
      "epoch": 55.24269088844735,
      "grad_norm": 0.3087533414363861,
      "learning_rate": 4.478030087993188e-05,
      "loss": 0.00114208422601223,
      "step": 194620
    },
    {
      "epoch": 55.245529378370705,
      "grad_norm": 3.340681791305542,
      "learning_rate": 4.477746239000852e-05,
      "loss": 0.001962008699774742,
      "step": 194630
    },
    {
      "epoch": 55.24836786829407,
      "grad_norm": 0.03065573237836361,
      "learning_rate": 4.4774623900085155e-05,
      "loss": 0.00047474950551986697,
      "step": 194640
    },
    {
      "epoch": 55.25120635821743,
      "grad_norm": 0.14722421765327454,
      "learning_rate": 4.4771785410161796e-05,
      "loss": 0.0006685396656394005,
      "step": 194650
    },
    {
      "epoch": 55.254044848140786,
      "grad_norm": 0.027249690145254135,
      "learning_rate": 4.476894692023844e-05,
      "loss": 0.00047531258314847946,
      "step": 194660
    },
    {
      "epoch": 55.25688333806415,
      "grad_norm": 0.05916985869407654,
      "learning_rate": 4.476610843031507e-05,
      "loss": 0.0006445864215493202,
      "step": 194670
    },
    {
      "epoch": 55.25972182798751,
      "grad_norm": 0.8173816800117493,
      "learning_rate": 4.476326994039171e-05,
      "loss": 0.0004665181040763855,
      "step": 194680
    },
    {
      "epoch": 55.26256031791087,
      "grad_norm": 0.2880288362503052,
      "learning_rate": 4.4760431450468355e-05,
      "loss": 0.00025918837636709215,
      "step": 194690
    },
    {
      "epoch": 55.26539880783423,
      "grad_norm": 3.3000266551971436,
      "learning_rate": 4.475759296054499e-05,
      "loss": 0.0010233314707875252,
      "step": 194700
    },
    {
      "epoch": 55.268237297757594,
      "grad_norm": 0.050387244671583176,
      "learning_rate": 4.475475447062163e-05,
      "loss": 0.00035357773303985597,
      "step": 194710
    },
    {
      "epoch": 55.27107578768096,
      "grad_norm": 0.07578729838132858,
      "learning_rate": 4.475191598069827e-05,
      "loss": 0.0015855209901928901,
      "step": 194720
    },
    {
      "epoch": 55.27391427760431,
      "grad_norm": 1.8770846128463745,
      "learning_rate": 4.474907749077491e-05,
      "loss": 0.0010290032252669335,
      "step": 194730
    },
    {
      "epoch": 55.276752767527675,
      "grad_norm": 0.023023571819067,
      "learning_rate": 4.474623900085155e-05,
      "loss": 0.00593048632144928,
      "step": 194740
    },
    {
      "epoch": 55.27959125745104,
      "grad_norm": 0.14154699444770813,
      "learning_rate": 4.474340051092819e-05,
      "loss": 0.0002583129331469536,
      "step": 194750
    },
    {
      "epoch": 55.282429747374394,
      "grad_norm": 0.06609191745519638,
      "learning_rate": 4.474056202100483e-05,
      "loss": 0.0024312878027558326,
      "step": 194760
    },
    {
      "epoch": 55.28526823729776,
      "grad_norm": 0.0890822559595108,
      "learning_rate": 4.4737723531081465e-05,
      "loss": 0.00032823048532009124,
      "step": 194770
    },
    {
      "epoch": 55.28810672722112,
      "grad_norm": 0.03643076494336128,
      "learning_rate": 4.473488504115811e-05,
      "loss": 0.0003332842141389847,
      "step": 194780
    },
    {
      "epoch": 55.29094521714448,
      "grad_norm": 0.05892584100365639,
      "learning_rate": 4.473204655123475e-05,
      "loss": 0.00022799372673034667,
      "step": 194790
    },
    {
      "epoch": 55.29378370706784,
      "grad_norm": 0.06062020733952522,
      "learning_rate": 4.472920806131138e-05,
      "loss": 0.00018019899725914003,
      "step": 194800
    },
    {
      "epoch": 55.2966221969912,
      "grad_norm": 0.15940389037132263,
      "learning_rate": 4.4726369571388024e-05,
      "loss": 0.0029679324477910997,
      "step": 194810
    },
    {
      "epoch": 55.299460686914564,
      "grad_norm": 0.009689678438007832,
      "learning_rate": 4.4723531081464665e-05,
      "loss": 0.0033068276941776277,
      "step": 194820
    },
    {
      "epoch": 55.30229917683792,
      "grad_norm": 0.24689918756484985,
      "learning_rate": 4.47206925915413e-05,
      "loss": 0.0003689061850309372,
      "step": 194830
    },
    {
      "epoch": 55.30513766676128,
      "grad_norm": 1.4904077053070068,
      "learning_rate": 4.471785410161794e-05,
      "loss": 0.0010180400684475898,
      "step": 194840
    },
    {
      "epoch": 55.307976156684646,
      "grad_norm": 0.35899242758750916,
      "learning_rate": 4.471501561169458e-05,
      "loss": 0.0007432864978909493,
      "step": 194850
    },
    {
      "epoch": 55.310814646608,
      "grad_norm": 0.003990854136645794,
      "learning_rate": 4.471217712177122e-05,
      "loss": 0.0012456996366381646,
      "step": 194860
    },
    {
      "epoch": 55.313653136531364,
      "grad_norm": 0.02826351672410965,
      "learning_rate": 4.470933863184786e-05,
      "loss": 0.0005749000236392021,
      "step": 194870
    },
    {
      "epoch": 55.31649162645473,
      "grad_norm": 0.9362640380859375,
      "learning_rate": 4.470650014192449e-05,
      "loss": 0.007070305198431015,
      "step": 194880
    },
    {
      "epoch": 55.31933011637809,
      "grad_norm": 7.644920349121094,
      "learning_rate": 4.470366165200114e-05,
      "loss": 0.004437017440795899,
      "step": 194890
    },
    {
      "epoch": 55.322168606301446,
      "grad_norm": 0.018548686057329178,
      "learning_rate": 4.4700823162077776e-05,
      "loss": 0.001053914614021778,
      "step": 194900
    },
    {
      "epoch": 55.32500709622481,
      "grad_norm": 7.442314147949219,
      "learning_rate": 4.469798467215441e-05,
      "loss": 0.003334449976682663,
      "step": 194910
    },
    {
      "epoch": 55.32784558614817,
      "grad_norm": 0.04250696673989296,
      "learning_rate": 4.469514618223106e-05,
      "loss": 0.001729109138250351,
      "step": 194920
    },
    {
      "epoch": 55.33068407607153,
      "grad_norm": 0.5079478621482849,
      "learning_rate": 4.4692307692307693e-05,
      "loss": 0.0005363667383790016,
      "step": 194930
    },
    {
      "epoch": 55.33352256599489,
      "grad_norm": 1.8001205921173096,
      "learning_rate": 4.4689469202384335e-05,
      "loss": 0.009835890680551528,
      "step": 194940
    },
    {
      "epoch": 55.33636105591825,
      "grad_norm": 0.09766422212123871,
      "learning_rate": 4.4686630712460976e-05,
      "loss": 0.00041944757103919985,
      "step": 194950
    },
    {
      "epoch": 55.33919954584161,
      "grad_norm": 0.15057848393917084,
      "learning_rate": 4.468379222253761e-05,
      "loss": 0.006303573399782181,
      "step": 194960
    },
    {
      "epoch": 55.34203803576497,
      "grad_norm": 0.7030815482139587,
      "learning_rate": 4.468095373261425e-05,
      "loss": 0.006318077445030212,
      "step": 194970
    },
    {
      "epoch": 55.344876525688335,
      "grad_norm": 4.12020206451416,
      "learning_rate": 4.4678115242690893e-05,
      "loss": 0.0008585747331380844,
      "step": 194980
    },
    {
      "epoch": 55.3477150156117,
      "grad_norm": 1.4110859632492065,
      "learning_rate": 4.467527675276753e-05,
      "loss": 0.0010818565264344216,
      "step": 194990
    },
    {
      "epoch": 55.35055350553505,
      "grad_norm": 0.005748797673732042,
      "learning_rate": 4.467243826284417e-05,
      "loss": 0.0004417736083269119,
      "step": 195000
    },
    {
      "epoch": 55.35055350553505,
      "eval_accuracy": 0.9761556558784257,
      "eval_loss": 0.0904388576745987,
      "eval_runtime": 32.0171,
      "eval_samples_per_second": 491.207,
      "eval_steps_per_second": 7.683,
      "step": 195000
    },
    {
      "epoch": 55.353391995458416,
      "grad_norm": 0.058975785970687866,
      "learning_rate": 4.4669599772920804e-05,
      "loss": 0.00155792236328125,
      "step": 195010
    },
    {
      "epoch": 55.35623048538178,
      "grad_norm": 0.7122920751571655,
      "learning_rate": 4.466676128299745e-05,
      "loss": 0.008237532526254653,
      "step": 195020
    },
    {
      "epoch": 55.359068975305135,
      "grad_norm": 0.18734413385391235,
      "learning_rate": 4.466392279307409e-05,
      "loss": 0.0009437600150704384,
      "step": 195030
    },
    {
      "epoch": 55.3619074652285,
      "grad_norm": 0.44869399070739746,
      "learning_rate": 4.466108430315072e-05,
      "loss": 0.005068421736359597,
      "step": 195040
    },
    {
      "epoch": 55.36474595515186,
      "grad_norm": 1.7386337518692017,
      "learning_rate": 4.465824581322737e-05,
      "loss": 0.004939820617437363,
      "step": 195050
    },
    {
      "epoch": 55.36758444507522,
      "grad_norm": 0.0189081821590662,
      "learning_rate": 4.4655407323304004e-05,
      "loss": 0.001970767229795456,
      "step": 195060
    },
    {
      "epoch": 55.37042293499858,
      "grad_norm": 0.992856502532959,
      "learning_rate": 4.4652568833380646e-05,
      "loss": 0.0006323590874671936,
      "step": 195070
    },
    {
      "epoch": 55.37326142492194,
      "grad_norm": 0.005068126134574413,
      "learning_rate": 4.464973034345729e-05,
      "loss": 0.001990712247788906,
      "step": 195080
    },
    {
      "epoch": 55.376099914845305,
      "grad_norm": 0.04581119492650032,
      "learning_rate": 4.464689185353392e-05,
      "loss": 0.002633177861571312,
      "step": 195090
    },
    {
      "epoch": 55.37893840476866,
      "grad_norm": 0.00856122374534607,
      "learning_rate": 4.464405336361056e-05,
      "loss": 0.00046063978224992753,
      "step": 195100
    },
    {
      "epoch": 55.381776894692024,
      "grad_norm": 0.3049808144569397,
      "learning_rate": 4.46412148736872e-05,
      "loss": 0.0004952913150191307,
      "step": 195110
    },
    {
      "epoch": 55.38461538461539,
      "grad_norm": 0.052736274898052216,
      "learning_rate": 4.463837638376384e-05,
      "loss": 0.0006919488310813904,
      "step": 195120
    },
    {
      "epoch": 55.38745387453874,
      "grad_norm": 0.2375338226556778,
      "learning_rate": 4.463553789384048e-05,
      "loss": 0.0026282791048288347,
      "step": 195130
    },
    {
      "epoch": 55.390292364462105,
      "grad_norm": 0.22425179183483124,
      "learning_rate": 4.4632699403917115e-05,
      "loss": 0.0011659577488899232,
      "step": 195140
    },
    {
      "epoch": 55.39313085438547,
      "grad_norm": 0.01393402274698019,
      "learning_rate": 4.4629860913993756e-05,
      "loss": 0.0003966502845287323,
      "step": 195150
    },
    {
      "epoch": 55.395969344308824,
      "grad_norm": 0.13200625777244568,
      "learning_rate": 4.46270224240704e-05,
      "loss": 0.0005212925374507904,
      "step": 195160
    },
    {
      "epoch": 55.39880783423219,
      "grad_norm": 0.018576575443148613,
      "learning_rate": 4.462418393414703e-05,
      "loss": 0.002523546852171421,
      "step": 195170
    },
    {
      "epoch": 55.40164632415555,
      "grad_norm": 9.146350860595703,
      "learning_rate": 4.462134544422368e-05,
      "loss": 0.0023242756724357605,
      "step": 195180
    },
    {
      "epoch": 55.40448481407891,
      "grad_norm": 0.1979316622018814,
      "learning_rate": 4.4618506954300315e-05,
      "loss": 0.0006057566031813621,
      "step": 195190
    },
    {
      "epoch": 55.40732330400227,
      "grad_norm": 1.0347381830215454,
      "learning_rate": 4.461566846437695e-05,
      "loss": 0.00044256746768951417,
      "step": 195200
    },
    {
      "epoch": 55.41016179392563,
      "grad_norm": 0.09901831299066544,
      "learning_rate": 4.461282997445359e-05,
      "loss": 0.0003788786008954048,
      "step": 195210
    },
    {
      "epoch": 55.413000283848994,
      "grad_norm": 0.03397703915834427,
      "learning_rate": 4.460999148453023e-05,
      "loss": 0.0013886306434869767,
      "step": 195220
    },
    {
      "epoch": 55.41583877377235,
      "grad_norm": 0.08982089906930923,
      "learning_rate": 4.4607152994606874e-05,
      "loss": 0.002927272766828537,
      "step": 195230
    },
    {
      "epoch": 55.41867726369571,
      "grad_norm": 0.1762295663356781,
      "learning_rate": 4.460431450468351e-05,
      "loss": 0.001263965666294098,
      "step": 195240
    },
    {
      "epoch": 55.421515753619076,
      "grad_norm": 0.04802100732922554,
      "learning_rate": 4.460147601476015e-05,
      "loss": 0.0007280142977833748,
      "step": 195250
    },
    {
      "epoch": 55.42435424354244,
      "grad_norm": 0.1878032684326172,
      "learning_rate": 4.459863752483679e-05,
      "loss": 0.0025816164910793305,
      "step": 195260
    },
    {
      "epoch": 55.427192733465795,
      "grad_norm": 0.13638068735599518,
      "learning_rate": 4.4595799034913426e-05,
      "loss": 0.0004608534276485443,
      "step": 195270
    },
    {
      "epoch": 55.43003122338916,
      "grad_norm": 0.1042189747095108,
      "learning_rate": 4.459296054499007e-05,
      "loss": 0.0004773736000061035,
      "step": 195280
    },
    {
      "epoch": 55.43286971331252,
      "grad_norm": 0.12997114658355713,
      "learning_rate": 4.459012205506671e-05,
      "loss": 0.0004634244367480278,
      "step": 195290
    },
    {
      "epoch": 55.435708203235876,
      "grad_norm": 0.06837407499551773,
      "learning_rate": 4.458728356514334e-05,
      "loss": 0.0003552177920937538,
      "step": 195300
    },
    {
      "epoch": 55.43854669315924,
      "grad_norm": 0.04790256917476654,
      "learning_rate": 4.4584445075219984e-05,
      "loss": 0.0008546538650989532,
      "step": 195310
    },
    {
      "epoch": 55.4413851830826,
      "grad_norm": 0.016681192442774773,
      "learning_rate": 4.4581606585296626e-05,
      "loss": 0.00011992193758487701,
      "step": 195320
    },
    {
      "epoch": 55.44422367300596,
      "grad_norm": 0.2294277399778366,
      "learning_rate": 4.457876809537326e-05,
      "loss": 0.0011796841397881507,
      "step": 195330
    },
    {
      "epoch": 55.44706216292932,
      "grad_norm": 0.286893755197525,
      "learning_rate": 4.45759296054499e-05,
      "loss": 0.0004792273044586182,
      "step": 195340
    },
    {
      "epoch": 55.44990065285268,
      "grad_norm": 0.0750778466463089,
      "learning_rate": 4.457309111552654e-05,
      "loss": 0.0007407324388623238,
      "step": 195350
    },
    {
      "epoch": 55.452739142776046,
      "grad_norm": 0.06433068215847015,
      "learning_rate": 4.4570252625603184e-05,
      "loss": 0.002376294508576393,
      "step": 195360
    },
    {
      "epoch": 55.4555776326994,
      "grad_norm": 0.3422260880470276,
      "learning_rate": 4.456741413567982e-05,
      "loss": 0.001585122011601925,
      "step": 195370
    },
    {
      "epoch": 55.458416122622765,
      "grad_norm": 0.06209409236907959,
      "learning_rate": 4.456457564575646e-05,
      "loss": 0.0007416512817144394,
      "step": 195380
    },
    {
      "epoch": 55.46125461254613,
      "grad_norm": 0.16264092922210693,
      "learning_rate": 4.45617371558331e-05,
      "loss": 0.0009405048564076424,
      "step": 195390
    },
    {
      "epoch": 55.464093102469484,
      "grad_norm": 0.36454451084136963,
      "learning_rate": 4.4558898665909736e-05,
      "loss": 0.0013256259262561798,
      "step": 195400
    },
    {
      "epoch": 55.46693159239285,
      "grad_norm": 0.42974647879600525,
      "learning_rate": 4.455606017598638e-05,
      "loss": 0.001060732640326023,
      "step": 195410
    },
    {
      "epoch": 55.46977008231621,
      "grad_norm": 0.0345328226685524,
      "learning_rate": 4.455322168606302e-05,
      "loss": 0.0026641532778739927,
      "step": 195420
    },
    {
      "epoch": 55.472608572239565,
      "grad_norm": 0.053902532905340195,
      "learning_rate": 4.4550383196139654e-05,
      "loss": 0.0015611572191119194,
      "step": 195430
    },
    {
      "epoch": 55.47544706216293,
      "grad_norm": 0.01404232531785965,
      "learning_rate": 4.4547544706216295e-05,
      "loss": 0.012929515540599823,
      "step": 195440
    },
    {
      "epoch": 55.47828555208629,
      "grad_norm": 0.021717660129070282,
      "learning_rate": 4.4544706216292936e-05,
      "loss": 0.0013281747698783875,
      "step": 195450
    },
    {
      "epoch": 55.481124042009654,
      "grad_norm": 0.16588552296161652,
      "learning_rate": 4.454186772636957e-05,
      "loss": 0.0003172192722558975,
      "step": 195460
    },
    {
      "epoch": 55.48396253193301,
      "grad_norm": 0.036945052444934845,
      "learning_rate": 4.453902923644621e-05,
      "loss": 0.0009128941223025322,
      "step": 195470
    },
    {
      "epoch": 55.48680102185637,
      "grad_norm": 0.11791518330574036,
      "learning_rate": 4.4536190746522854e-05,
      "loss": 0.005113641545176506,
      "step": 195480
    },
    {
      "epoch": 55.489639511779735,
      "grad_norm": 0.03821057826280594,
      "learning_rate": 4.4533352256599495e-05,
      "loss": 0.0004328705370426178,
      "step": 195490
    },
    {
      "epoch": 55.49247800170309,
      "grad_norm": 0.05291134864091873,
      "learning_rate": 4.453051376667613e-05,
      "loss": 0.0021774593740701677,
      "step": 195500
    },
    {
      "epoch": 55.49247800170309,
      "eval_accuracy": 0.9793984866789598,
      "eval_loss": 0.0788162350654602,
      "eval_runtime": 33.1417,
      "eval_samples_per_second": 474.538,
      "eval_steps_per_second": 7.423,
      "step": 195500
    },
    {
      "epoch": 55.495316491626454,
      "grad_norm": 0.140462264418602,
      "learning_rate": 4.4527675276752764e-05,
      "loss": 0.0003875315189361572,
      "step": 195510
    },
    {
      "epoch": 55.49815498154982,
      "grad_norm": 0.19697502255439758,
      "learning_rate": 4.452483678682941e-05,
      "loss": 0.0008436206728219986,
      "step": 195520
    },
    {
      "epoch": 55.50099347147318,
      "grad_norm": 0.8182134032249451,
      "learning_rate": 4.452199829690605e-05,
      "loss": 0.0004978477954864502,
      "step": 195530
    },
    {
      "epoch": 55.503831961396536,
      "grad_norm": 0.10129676759243011,
      "learning_rate": 4.451915980698269e-05,
      "loss": 0.0003723142668604851,
      "step": 195540
    },
    {
      "epoch": 55.5066704513199,
      "grad_norm": 0.13134445250034332,
      "learning_rate": 4.451632131705933e-05,
      "loss": 0.0011487668380141258,
      "step": 195550
    },
    {
      "epoch": 55.50950894124326,
      "grad_norm": 0.24510473012924194,
      "learning_rate": 4.4513482827135964e-05,
      "loss": 0.00030811242759227754,
      "step": 195560
    },
    {
      "epoch": 55.51234743116662,
      "grad_norm": 0.010434734635055065,
      "learning_rate": 4.4510644337212606e-05,
      "loss": 0.00041885655373334884,
      "step": 195570
    },
    {
      "epoch": 55.51518592108998,
      "grad_norm": 0.1905076801776886,
      "learning_rate": 4.450780584728925e-05,
      "loss": 0.0018813129514455794,
      "step": 195580
    },
    {
      "epoch": 55.51802441101334,
      "grad_norm": 0.07967260479927063,
      "learning_rate": 4.450496735736588e-05,
      "loss": 0.0002949539572000504,
      "step": 195590
    },
    {
      "epoch": 55.5208629009367,
      "grad_norm": 0.45000386238098145,
      "learning_rate": 4.450212886744252e-05,
      "loss": 0.004608586803078652,
      "step": 195600
    },
    {
      "epoch": 55.52370139086006,
      "grad_norm": 0.02075824700295925,
      "learning_rate": 4.449929037751916e-05,
      "loss": 0.0011772202327847482,
      "step": 195610
    },
    {
      "epoch": 55.526539880783425,
      "grad_norm": 0.02408762089908123,
      "learning_rate": 4.44964518875958e-05,
      "loss": 0.00045991577208042144,
      "step": 195620
    },
    {
      "epoch": 55.52937837070679,
      "grad_norm": 2.340108871459961,
      "learning_rate": 4.449361339767244e-05,
      "loss": 0.0025185955688357352,
      "step": 195630
    },
    {
      "epoch": 55.53221686063014,
      "grad_norm": 1.4925473928451538,
      "learning_rate": 4.4490774907749075e-05,
      "loss": 0.0008068062365055084,
      "step": 195640
    },
    {
      "epoch": 55.535055350553506,
      "grad_norm": 0.8110709190368652,
      "learning_rate": 4.448793641782572e-05,
      "loss": 0.0019405078142881393,
      "step": 195650
    },
    {
      "epoch": 55.53789384047687,
      "grad_norm": 0.12303122133016586,
      "learning_rate": 4.448509792790236e-05,
      "loss": 0.004039919748902321,
      "step": 195660
    },
    {
      "epoch": 55.540732330400225,
      "grad_norm": 0.08504562079906464,
      "learning_rate": 4.448225943797899e-05,
      "loss": 0.0009208496659994126,
      "step": 195670
    },
    {
      "epoch": 55.54357082032359,
      "grad_norm": 0.13055868446826935,
      "learning_rate": 4.447942094805564e-05,
      "loss": 0.00845632255077362,
      "step": 195680
    },
    {
      "epoch": 55.54640931024695,
      "grad_norm": 0.4772072732448578,
      "learning_rate": 4.4476582458132275e-05,
      "loss": 0.0010243121534585953,
      "step": 195690
    },
    {
      "epoch": 55.549247800170306,
      "grad_norm": 0.03264050930738449,
      "learning_rate": 4.447374396820892e-05,
      "loss": 0.0011041749268770218,
      "step": 195700
    },
    {
      "epoch": 55.55208629009367,
      "grad_norm": 0.07846280932426453,
      "learning_rate": 4.447090547828555e-05,
      "loss": 0.003601345047354698,
      "step": 195710
    },
    {
      "epoch": 55.55492478001703,
      "grad_norm": 0.06509603559970856,
      "learning_rate": 4.446806698836219e-05,
      "loss": 0.0007311167195439338,
      "step": 195720
    },
    {
      "epoch": 55.557763269940395,
      "grad_norm": 0.31171146035194397,
      "learning_rate": 4.4465228498438834e-05,
      "loss": 0.0019745094701647758,
      "step": 195730
    },
    {
      "epoch": 55.56060175986375,
      "grad_norm": 1.0746601819992065,
      "learning_rate": 4.446239000851547e-05,
      "loss": 0.0036881938576698303,
      "step": 195740
    },
    {
      "epoch": 55.563440249787114,
      "grad_norm": 0.11764934659004211,
      "learning_rate": 4.445955151859211e-05,
      "loss": 0.0004766654223203659,
      "step": 195750
    },
    {
      "epoch": 55.56627873971048,
      "grad_norm": 0.14857521653175354,
      "learning_rate": 4.445671302866875e-05,
      "loss": 0.0025995437055826185,
      "step": 195760
    },
    {
      "epoch": 55.56911722963383,
      "grad_norm": 0.06457062065601349,
      "learning_rate": 4.4453874538745386e-05,
      "loss": 0.00040244702249765397,
      "step": 195770
    },
    {
      "epoch": 55.571955719557195,
      "grad_norm": 0.024196013808250427,
      "learning_rate": 4.4451036048822034e-05,
      "loss": 0.0012528656050562858,
      "step": 195780
    },
    {
      "epoch": 55.57479420948056,
      "grad_norm": 0.01978454180061817,
      "learning_rate": 4.444819755889867e-05,
      "loss": 0.0004911383613944054,
      "step": 195790
    },
    {
      "epoch": 55.577632699403914,
      "grad_norm": 0.086908720433712,
      "learning_rate": 4.44453590689753e-05,
      "loss": 0.0003069400787353516,
      "step": 195800
    },
    {
      "epoch": 55.58047118932728,
      "grad_norm": 0.281536340713501,
      "learning_rate": 4.4442520579051945e-05,
      "loss": 0.0014177361503243447,
      "step": 195810
    },
    {
      "epoch": 55.58330967925064,
      "grad_norm": 0.33803147077560425,
      "learning_rate": 4.4439682089128586e-05,
      "loss": 0.0014886047691106796,
      "step": 195820
    },
    {
      "epoch": 55.586148169174,
      "grad_norm": 1.5873678922653198,
      "learning_rate": 4.443684359920523e-05,
      "loss": 0.0013248478993773461,
      "step": 195830
    },
    {
      "epoch": 55.58898665909736,
      "grad_norm": 0.1931331306695938,
      "learning_rate": 4.443400510928186e-05,
      "loss": 0.0005571950227022171,
      "step": 195840
    },
    {
      "epoch": 55.59182514902072,
      "grad_norm": 0.04527309164404869,
      "learning_rate": 4.44311666193585e-05,
      "loss": 0.0014125628396868706,
      "step": 195850
    },
    {
      "epoch": 55.594663638944084,
      "grad_norm": 0.29715847969055176,
      "learning_rate": 4.4428328129435145e-05,
      "loss": 0.0007061680778861045,
      "step": 195860
    },
    {
      "epoch": 55.59750212886744,
      "grad_norm": 0.032203949987888336,
      "learning_rate": 4.442548963951178e-05,
      "loss": 0.006834743916988373,
      "step": 195870
    },
    {
      "epoch": 55.6003406187908,
      "grad_norm": 0.03830215334892273,
      "learning_rate": 4.442265114958842e-05,
      "loss": 0.0006267936900258064,
      "step": 195880
    },
    {
      "epoch": 55.603179108714166,
      "grad_norm": 0.04029150679707527,
      "learning_rate": 4.441981265966506e-05,
      "loss": 0.0007859796285629272,
      "step": 195890
    },
    {
      "epoch": 55.60601759863752,
      "grad_norm": 0.5850456357002258,
      "learning_rate": 4.44169741697417e-05,
      "loss": 0.001617724820971489,
      "step": 195900
    },
    {
      "epoch": 55.608856088560884,
      "grad_norm": 0.040380753576755524,
      "learning_rate": 4.441413567981834e-05,
      "loss": 0.0007476085796952247,
      "step": 195910
    },
    {
      "epoch": 55.61169457848425,
      "grad_norm": 0.06166137754917145,
      "learning_rate": 4.441129718989498e-05,
      "loss": 0.0021175406873226166,
      "step": 195920
    },
    {
      "epoch": 55.61453306840761,
      "grad_norm": 2.293567419052124,
      "learning_rate": 4.4408458699971614e-05,
      "loss": 0.0009004078805446625,
      "step": 195930
    },
    {
      "epoch": 55.617371558330966,
      "grad_norm": 0.12644106149673462,
      "learning_rate": 4.4405620210048255e-05,
      "loss": 0.0004668345674872398,
      "step": 195940
    },
    {
      "epoch": 55.62021004825433,
      "grad_norm": 0.05381820723414421,
      "learning_rate": 4.44027817201249e-05,
      "loss": 0.00041090063750743867,
      "step": 195950
    },
    {
      "epoch": 55.62304853817769,
      "grad_norm": 0.6979149580001831,
      "learning_rate": 4.439994323020154e-05,
      "loss": 0.0011342067271471023,
      "step": 195960
    },
    {
      "epoch": 55.62588702810105,
      "grad_norm": 0.27880677580833435,
      "learning_rate": 4.439710474027817e-05,
      "loss": 0.0010289128869771958,
      "step": 195970
    },
    {
      "epoch": 55.62872551802441,
      "grad_norm": 0.13811397552490234,
      "learning_rate": 4.4394266250354814e-05,
      "loss": 0.0014708327129483224,
      "step": 195980
    },
    {
      "epoch": 55.63156400794777,
      "grad_norm": 0.1676754653453827,
      "learning_rate": 4.4391427760431456e-05,
      "loss": 0.000462646409869194,
      "step": 195990
    },
    {
      "epoch": 55.634402497871136,
      "grad_norm": 4.822806358337402,
      "learning_rate": 4.438858927050809e-05,
      "loss": 0.0012267749756574632,
      "step": 196000
    },
    {
      "epoch": 55.634402497871136,
      "eval_accuracy": 0.9785718827494119,
      "eval_loss": 0.07978913933038712,
      "eval_runtime": 32.4524,
      "eval_samples_per_second": 484.617,
      "eval_steps_per_second": 7.58,
      "step": 196000
    },
    {
      "epoch": 55.63724098779449,
      "grad_norm": 2.371077060699463,
      "learning_rate": 4.438575078058473e-05,
      "loss": 0.0020863503217697144,
      "step": 196010
    },
    {
      "epoch": 55.640079477717855,
      "grad_norm": 1.2471134662628174,
      "learning_rate": 4.438291229066137e-05,
      "loss": 0.0003457244485616684,
      "step": 196020
    },
    {
      "epoch": 55.64291796764122,
      "grad_norm": 0.007532291579991579,
      "learning_rate": 4.438007380073801e-05,
      "loss": 0.0007181014865636825,
      "step": 196030
    },
    {
      "epoch": 55.64575645756457,
      "grad_norm": 0.29052087664604187,
      "learning_rate": 4.437723531081465e-05,
      "loss": 0.01364491879940033,
      "step": 196040
    },
    {
      "epoch": 55.648594947487936,
      "grad_norm": 0.010884094052016735,
      "learning_rate": 4.437439682089129e-05,
      "loss": 0.00047542229294776915,
      "step": 196050
    },
    {
      "epoch": 55.6514334374113,
      "grad_norm": 0.3816167116165161,
      "learning_rate": 4.4371558330967925e-05,
      "loss": 0.0005589909851551056,
      "step": 196060
    },
    {
      "epoch": 55.654271927334655,
      "grad_norm": 0.14847809076309204,
      "learning_rate": 4.4368719841044566e-05,
      "loss": 0.0003341909497976303,
      "step": 196070
    },
    {
      "epoch": 55.65711041725802,
      "grad_norm": 3.6622815132141113,
      "learning_rate": 4.436588135112121e-05,
      "loss": 0.0018750831484794616,
      "step": 196080
    },
    {
      "epoch": 55.65994890718138,
      "grad_norm": 0.8110861778259277,
      "learning_rate": 4.436304286119784e-05,
      "loss": 0.00030072927474975584,
      "step": 196090
    },
    {
      "epoch": 55.66278739710474,
      "grad_norm": 0.14410029351711273,
      "learning_rate": 4.4360204371274484e-05,
      "loss": 0.0006238007918000221,
      "step": 196100
    },
    {
      "epoch": 55.6656258870281,
      "grad_norm": 11.876084327697754,
      "learning_rate": 4.435736588135112e-05,
      "loss": 0.003476811945438385,
      "step": 196110
    },
    {
      "epoch": 55.66846437695146,
      "grad_norm": 0.4048847258090973,
      "learning_rate": 4.4354527391427766e-05,
      "loss": 0.001298365369439125,
      "step": 196120
    },
    {
      "epoch": 55.671302866874825,
      "grad_norm": 0.8168792724609375,
      "learning_rate": 4.43516889015044e-05,
      "loss": 0.0018630454316735268,
      "step": 196130
    },
    {
      "epoch": 55.67414135679818,
      "grad_norm": 0.10523020476102829,
      "learning_rate": 4.4348850411581035e-05,
      "loss": 0.000249406136572361,
      "step": 196140
    },
    {
      "epoch": 55.676979846721544,
      "grad_norm": 0.2033424824476242,
      "learning_rate": 4.4346011921657684e-05,
      "loss": 0.0004924353212118149,
      "step": 196150
    },
    {
      "epoch": 55.67981833664491,
      "grad_norm": 0.08681827038526535,
      "learning_rate": 4.434317343173432e-05,
      "loss": 0.0001948842778801918,
      "step": 196160
    },
    {
      "epoch": 55.68265682656826,
      "grad_norm": 0.06687960773706436,
      "learning_rate": 4.434033494181096e-05,
      "loss": 0.0017615793272852897,
      "step": 196170
    },
    {
      "epoch": 55.685495316491625,
      "grad_norm": 0.032955352216959,
      "learning_rate": 4.43374964518876e-05,
      "loss": 0.0008970245718955994,
      "step": 196180
    },
    {
      "epoch": 55.68833380641499,
      "grad_norm": 0.0464039146900177,
      "learning_rate": 4.4334657961964236e-05,
      "loss": 0.0003467181697487831,
      "step": 196190
    },
    {
      "epoch": 55.69117229633835,
      "grad_norm": 0.05231625959277153,
      "learning_rate": 4.433181947204088e-05,
      "loss": 0.0006382651627063751,
      "step": 196200
    },
    {
      "epoch": 55.69401078626171,
      "grad_norm": 1.9588552713394165,
      "learning_rate": 4.432898098211752e-05,
      "loss": 0.00047942418605089187,
      "step": 196210
    },
    {
      "epoch": 55.69684927618507,
      "grad_norm": 4.796819686889648,
      "learning_rate": 4.432614249219415e-05,
      "loss": 0.0009731125086545944,
      "step": 196220
    },
    {
      "epoch": 55.69968776610843,
      "grad_norm": 0.08818718045949936,
      "learning_rate": 4.4323304002270794e-05,
      "loss": 0.00042299237102270124,
      "step": 196230
    },
    {
      "epoch": 55.70252625603179,
      "grad_norm": 0.016414295881986618,
      "learning_rate": 4.432046551234743e-05,
      "loss": 0.0005264440551400185,
      "step": 196240
    },
    {
      "epoch": 55.70536474595515,
      "grad_norm": 0.019471850246191025,
      "learning_rate": 4.431762702242408e-05,
      "loss": 0.00061210747808218,
      "step": 196250
    },
    {
      "epoch": 55.708203235878514,
      "grad_norm": 0.0701521560549736,
      "learning_rate": 4.431478853250071e-05,
      "loss": 0.00022021960467100143,
      "step": 196260
    },
    {
      "epoch": 55.71104172580187,
      "grad_norm": 0.20904599130153656,
      "learning_rate": 4.4311950042577346e-05,
      "loss": 0.00030242409557104113,
      "step": 196270
    },
    {
      "epoch": 55.71388021572523,
      "grad_norm": 0.01986934058368206,
      "learning_rate": 4.4309111552653994e-05,
      "loss": 0.00042535196989774705,
      "step": 196280
    },
    {
      "epoch": 55.716718705648596,
      "grad_norm": 0.0640273466706276,
      "learning_rate": 4.430627306273063e-05,
      "loss": 0.00015320442616939545,
      "step": 196290
    },
    {
      "epoch": 55.71955719557196,
      "grad_norm": 0.022599680349230766,
      "learning_rate": 4.430343457280727e-05,
      "loss": 0.00022713989019393921,
      "step": 196300
    },
    {
      "epoch": 55.722395685495314,
      "grad_norm": 1.0642094612121582,
      "learning_rate": 4.430059608288391e-05,
      "loss": 0.00043539702892303467,
      "step": 196310
    },
    {
      "epoch": 55.72523417541868,
      "grad_norm": 0.016858529299497604,
      "learning_rate": 4.4297757592960546e-05,
      "loss": 0.003183409571647644,
      "step": 196320
    },
    {
      "epoch": 55.72807266534204,
      "grad_norm": 0.11536920070648193,
      "learning_rate": 4.429491910303719e-05,
      "loss": 0.0002772163599729538,
      "step": 196330
    },
    {
      "epoch": 55.730911155265396,
      "grad_norm": 0.1606159508228302,
      "learning_rate": 4.429208061311382e-05,
      "loss": 0.00040016863495111464,
      "step": 196340
    },
    {
      "epoch": 55.73374964518876,
      "grad_norm": 0.022690167650580406,
      "learning_rate": 4.4289242123190464e-05,
      "loss": 0.0018143236637115478,
      "step": 196350
    },
    {
      "epoch": 55.73658813511212,
      "grad_norm": 0.055691879242658615,
      "learning_rate": 4.4286403633267105e-05,
      "loss": 0.00038042869418859483,
      "step": 196360
    },
    {
      "epoch": 55.739426625035485,
      "grad_norm": 0.442757248878479,
      "learning_rate": 4.428356514334374e-05,
      "loss": 0.00044394098222255707,
      "step": 196370
    },
    {
      "epoch": 55.74226511495884,
      "grad_norm": 0.02711881510913372,
      "learning_rate": 4.428072665342039e-05,
      "loss": 0.00020918566733598709,
      "step": 196380
    },
    {
      "epoch": 55.7451036048822,
      "grad_norm": 0.06608392298221588,
      "learning_rate": 4.427788816349702e-05,
      "loss": 0.0002472247928380966,
      "step": 196390
    },
    {
      "epoch": 55.747942094805566,
      "grad_norm": 0.13745005428791046,
      "learning_rate": 4.427504967357366e-05,
      "loss": 0.00017805453389883042,
      "step": 196400
    },
    {
      "epoch": 55.75078058472892,
      "grad_norm": 0.20907828211784363,
      "learning_rate": 4.4272211183650305e-05,
      "loss": 0.0005539601668715477,
      "step": 196410
    },
    {
      "epoch": 55.753619074652285,
      "grad_norm": 0.2176695317029953,
      "learning_rate": 4.426937269372694e-05,
      "loss": 0.0008491484448313713,
      "step": 196420
    },
    {
      "epoch": 55.75645756457565,
      "grad_norm": 0.23949824273586273,
      "learning_rate": 4.426653420380358e-05,
      "loss": 0.0002777740359306335,
      "step": 196430
    },
    {
      "epoch": 55.759296054499,
      "grad_norm": 0.015139023773372173,
      "learning_rate": 4.4263695713880216e-05,
      "loss": 0.0005654899403452873,
      "step": 196440
    },
    {
      "epoch": 55.762134544422366,
      "grad_norm": 0.24645769596099854,
      "learning_rate": 4.426085722395686e-05,
      "loss": 0.00032842159271240234,
      "step": 196450
    },
    {
      "epoch": 55.76497303434573,
      "grad_norm": 0.09163224697113037,
      "learning_rate": 4.42580187340335e-05,
      "loss": 0.002380048669874668,
      "step": 196460
    },
    {
      "epoch": 55.76781152426909,
      "grad_norm": 10.132923126220703,
      "learning_rate": 4.425518024411013e-05,
      "loss": 0.0020694958046078684,
      "step": 196470
    },
    {
      "epoch": 55.77065001419245,
      "grad_norm": 0.006881595589220524,
      "learning_rate": 4.4252341754186774e-05,
      "loss": 0.008043007552623748,
      "step": 196480
    },
    {
      "epoch": 55.77348850411581,
      "grad_norm": 0.03419148549437523,
      "learning_rate": 4.4249503264263416e-05,
      "loss": 0.001072332076728344,
      "step": 196490
    },
    {
      "epoch": 55.776326994039174,
      "grad_norm": 4.8492302894592285,
      "learning_rate": 4.424666477434005e-05,
      "loss": 0.001530216820538044,
      "step": 196500
    },
    {
      "epoch": 55.776326994039174,
      "eval_accuracy": 0.9759013162077955,
      "eval_loss": 0.08591056615114212,
      "eval_runtime": 32.8386,
      "eval_samples_per_second": 478.918,
      "eval_steps_per_second": 7.491,
      "step": 196500
    },
    {
      "epoch": 55.77916548396253,
      "grad_norm": 0.02104274183511734,
      "learning_rate": 4.424382628441669e-05,
      "loss": 0.00043503046035766604,
      "step": 196510
    },
    {
      "epoch": 55.78200397388589,
      "grad_norm": 0.1139107421040535,
      "learning_rate": 4.424098779449333e-05,
      "loss": 0.0008360268548130989,
      "step": 196520
    },
    {
      "epoch": 55.784842463809255,
      "grad_norm": 0.03754192590713501,
      "learning_rate": 4.423814930456997e-05,
      "loss": 0.0023066187277436256,
      "step": 196530
    },
    {
      "epoch": 55.78768095373261,
      "grad_norm": 0.013354557566344738,
      "learning_rate": 4.423531081464661e-05,
      "loss": 0.0019367378205060958,
      "step": 196540
    },
    {
      "epoch": 55.790519443655974,
      "grad_norm": 0.06813370436429977,
      "learning_rate": 4.423247232472325e-05,
      "loss": 0.0004153234884142876,
      "step": 196550
    },
    {
      "epoch": 55.79335793357934,
      "grad_norm": 0.11764467507600784,
      "learning_rate": 4.4229633834799885e-05,
      "loss": 0.0009926460683345796,
      "step": 196560
    },
    {
      "epoch": 55.7961964235027,
      "grad_norm": 0.02853783220052719,
      "learning_rate": 4.4226795344876526e-05,
      "loss": 0.0013534115627408029,
      "step": 196570
    },
    {
      "epoch": 55.799034913426055,
      "grad_norm": 0.03399305418133736,
      "learning_rate": 4.422395685495317e-05,
      "loss": 0.0002983277663588524,
      "step": 196580
    },
    {
      "epoch": 55.80187340334942,
      "grad_norm": 0.12297548353672028,
      "learning_rate": 4.422111836502981e-05,
      "loss": 0.0005769422277808189,
      "step": 196590
    },
    {
      "epoch": 55.80471189327278,
      "grad_norm": 0.01821609027683735,
      "learning_rate": 4.4218279875106444e-05,
      "loss": 0.00021558105945587158,
      "step": 196600
    },
    {
      "epoch": 55.80755038319614,
      "grad_norm": 0.030883856117725372,
      "learning_rate": 4.4215441385183085e-05,
      "loss": 0.00024609584361314776,
      "step": 196610
    },
    {
      "epoch": 55.8103888731195,
      "grad_norm": 0.024584682658314705,
      "learning_rate": 4.4212602895259727e-05,
      "loss": 0.0008576789870858193,
      "step": 196620
    },
    {
      "epoch": 55.81322736304286,
      "grad_norm": 0.09661012887954712,
      "learning_rate": 4.420976440533636e-05,
      "loss": 0.0009856108576059342,
      "step": 196630
    },
    {
      "epoch": 55.81606585296622,
      "grad_norm": 0.4234319031238556,
      "learning_rate": 4.4206925915413e-05,
      "loss": 0.00028709862381219863,
      "step": 196640
    },
    {
      "epoch": 55.81890434288958,
      "grad_norm": 0.14795376360416412,
      "learning_rate": 4.4204087425489644e-05,
      "loss": 0.0003234105184674263,
      "step": 196650
    },
    {
      "epoch": 55.821742832812944,
      "grad_norm": 0.2211301028728485,
      "learning_rate": 4.420124893556628e-05,
      "loss": 0.00029030777513980863,
      "step": 196660
    },
    {
      "epoch": 55.82458132273631,
      "grad_norm": 0.013068323023617268,
      "learning_rate": 4.419841044564292e-05,
      "loss": 0.00011935140937566758,
      "step": 196670
    },
    {
      "epoch": 55.82741981265966,
      "grad_norm": 0.15816088020801544,
      "learning_rate": 4.419557195571956e-05,
      "loss": 0.00017476733773946762,
      "step": 196680
    },
    {
      "epoch": 55.830258302583026,
      "grad_norm": 0.025062914937734604,
      "learning_rate": 4.4192733465796196e-05,
      "loss": 0.0009407781064510346,
      "step": 196690
    },
    {
      "epoch": 55.83309679250639,
      "grad_norm": 0.0777682363986969,
      "learning_rate": 4.418989497587284e-05,
      "loss": 0.0006402276456356048,
      "step": 196700
    },
    {
      "epoch": 55.835935282429745,
      "grad_norm": 0.04425423592329025,
      "learning_rate": 4.418705648594948e-05,
      "loss": 0.0003429219126701355,
      "step": 196710
    },
    {
      "epoch": 55.83877377235311,
      "grad_norm": 0.12589485943317413,
      "learning_rate": 4.418421799602612e-05,
      "loss": 0.000830291211605072,
      "step": 196720
    },
    {
      "epoch": 55.84161226227647,
      "grad_norm": 0.8923065066337585,
      "learning_rate": 4.4181379506102755e-05,
      "loss": 0.010044597089290619,
      "step": 196730
    },
    {
      "epoch": 55.84445075219983,
      "grad_norm": 0.05249821022152901,
      "learning_rate": 4.417854101617939e-05,
      "loss": 0.0013560689985752106,
      "step": 196740
    },
    {
      "epoch": 55.84728924212319,
      "grad_norm": 0.2111930549144745,
      "learning_rate": 4.417570252625604e-05,
      "loss": 0.0028815098106861115,
      "step": 196750
    },
    {
      "epoch": 55.85012773204655,
      "grad_norm": 0.699023962020874,
      "learning_rate": 4.417286403633267e-05,
      "loss": 0.0004907786846160888,
      "step": 196760
    },
    {
      "epoch": 55.852966221969915,
      "grad_norm": 0.30411437153816223,
      "learning_rate": 4.417002554640931e-05,
      "loss": 0.0019480813294649123,
      "step": 196770
    },
    {
      "epoch": 55.85580471189327,
      "grad_norm": 0.07168100029230118,
      "learning_rate": 4.4167187056485955e-05,
      "loss": 0.00045349579304456713,
      "step": 196780
    },
    {
      "epoch": 55.85864320181663,
      "grad_norm": 1.0561796426773071,
      "learning_rate": 4.416434856656259e-05,
      "loss": 0.0006963502615690232,
      "step": 196790
    },
    {
      "epoch": 55.861481691739996,
      "grad_norm": 0.014141416177153587,
      "learning_rate": 4.416151007663923e-05,
      "loss": 0.00199742317199707,
      "step": 196800
    },
    {
      "epoch": 55.86432018166335,
      "grad_norm": 0.04862400144338608,
      "learning_rate": 4.415867158671587e-05,
      "loss": 0.008266513794660568,
      "step": 196810
    },
    {
      "epoch": 55.867158671586715,
      "grad_norm": 0.6778897643089294,
      "learning_rate": 4.415583309679251e-05,
      "loss": 0.0005776094272732735,
      "step": 196820
    },
    {
      "epoch": 55.86999716151008,
      "grad_norm": 0.06471648812294006,
      "learning_rate": 4.415299460686915e-05,
      "loss": 0.0004810810089111328,
      "step": 196830
    },
    {
      "epoch": 55.87283565143344,
      "grad_norm": 0.0660003051161766,
      "learning_rate": 4.415015611694578e-05,
      "loss": 0.0006744973361492157,
      "step": 196840
    },
    {
      "epoch": 55.8756741413568,
      "grad_norm": 0.007100454065948725,
      "learning_rate": 4.414731762702243e-05,
      "loss": 0.00044870078563690184,
      "step": 196850
    },
    {
      "epoch": 55.87851263128016,
      "grad_norm": 8.078669548034668,
      "learning_rate": 4.4144479137099065e-05,
      "loss": 0.003866002708673477,
      "step": 196860
    },
    {
      "epoch": 55.88135112120352,
      "grad_norm": 0.08319126069545746,
      "learning_rate": 4.41416406471757e-05,
      "loss": 0.0008420303463935852,
      "step": 196870
    },
    {
      "epoch": 55.88418961112688,
      "grad_norm": 0.6044951677322388,
      "learning_rate": 4.413880215725235e-05,
      "loss": 0.0011138007044792174,
      "step": 196880
    },
    {
      "epoch": 55.88702810105024,
      "grad_norm": 0.2160164713859558,
      "learning_rate": 4.413596366732898e-05,
      "loss": 0.0006604541093111038,
      "step": 196890
    },
    {
      "epoch": 55.889866590973604,
      "grad_norm": 0.16837632656097412,
      "learning_rate": 4.4133125177405624e-05,
      "loss": 0.00035883933305740355,
      "step": 196900
    },
    {
      "epoch": 55.89270508089696,
      "grad_norm": 0.05623887851834297,
      "learning_rate": 4.4130286687482265e-05,
      "loss": 0.00036455150693655015,
      "step": 196910
    },
    {
      "epoch": 55.89554357082032,
      "grad_norm": 0.03406241536140442,
      "learning_rate": 4.41274481975589e-05,
      "loss": 0.0021186094731092454,
      "step": 196920
    },
    {
      "epoch": 55.898382060743685,
      "grad_norm": 0.0315786749124527,
      "learning_rate": 4.412460970763554e-05,
      "loss": 0.0005835970863699913,
      "step": 196930
    },
    {
      "epoch": 55.90122055066705,
      "grad_norm": 0.01840806193649769,
      "learning_rate": 4.4121771217712176e-05,
      "loss": 0.00031728055328130724,
      "step": 196940
    },
    {
      "epoch": 55.904059040590404,
      "grad_norm": 0.030644332990050316,
      "learning_rate": 4.411893272778882e-05,
      "loss": 0.0006682487204670906,
      "step": 196950
    },
    {
      "epoch": 55.90689753051377,
      "grad_norm": 0.053842999041080475,
      "learning_rate": 4.411609423786546e-05,
      "loss": 0.0009479392319917679,
      "step": 196960
    },
    {
      "epoch": 55.90973602043713,
      "grad_norm": 0.37311267852783203,
      "learning_rate": 4.411325574794209e-05,
      "loss": 0.0005013581365346908,
      "step": 196970
    },
    {
      "epoch": 55.912574510360486,
      "grad_norm": 0.2571651339530945,
      "learning_rate": 4.4110417258018735e-05,
      "loss": 0.0016391843557357789,
      "step": 196980
    },
    {
      "epoch": 55.91541300028385,
      "grad_norm": 3.8961353302001953,
      "learning_rate": 4.4107578768095376e-05,
      "loss": 0.003627689927816391,
      "step": 196990
    },
    {
      "epoch": 55.91825149020721,
      "grad_norm": 1.1127028465270996,
      "learning_rate": 4.410474027817201e-05,
      "loss": 0.0006626248359680176,
      "step": 197000
    },
    {
      "epoch": 55.91825149020721,
      "eval_accuracy": 0.9755198067018503,
      "eval_loss": 0.09028699994087219,
      "eval_runtime": 32.8866,
      "eval_samples_per_second": 478.22,
      "eval_steps_per_second": 7.48,
      "step": 197000
    },
    {
      "epoch": 55.92108998013057,
      "grad_norm": 0.07078777253627777,
      "learning_rate": 4.410190178824866e-05,
      "loss": 0.002096804790198803,
      "step": 197010
    },
    {
      "epoch": 55.92392847005393,
      "grad_norm": 0.029537471011281013,
      "learning_rate": 4.4099063298325293e-05,
      "loss": 0.000508381798863411,
      "step": 197020
    },
    {
      "epoch": 55.92676695997729,
      "grad_norm": 1.0426149368286133,
      "learning_rate": 4.409622480840193e-05,
      "loss": 0.001748926192522049,
      "step": 197030
    },
    {
      "epoch": 55.929605449900656,
      "grad_norm": 3.2299537658691406,
      "learning_rate": 4.409338631847857e-05,
      "loss": 0.004475123807787895,
      "step": 197040
    },
    {
      "epoch": 55.93244393982401,
      "grad_norm": 0.4356624186038971,
      "learning_rate": 4.409054782855521e-05,
      "loss": 0.002441550977528095,
      "step": 197050
    },
    {
      "epoch": 55.935282429747375,
      "grad_norm": 1.0316020250320435,
      "learning_rate": 4.408770933863185e-05,
      "loss": 0.002776379510760307,
      "step": 197060
    },
    {
      "epoch": 55.93812091967074,
      "grad_norm": 0.03970561549067497,
      "learning_rate": 4.408487084870849e-05,
      "loss": 0.0020995175465941427,
      "step": 197070
    },
    {
      "epoch": 55.94095940959409,
      "grad_norm": 5.274291515350342,
      "learning_rate": 4.408203235878513e-05,
      "loss": 0.0029045039787888526,
      "step": 197080
    },
    {
      "epoch": 55.943797899517456,
      "grad_norm": 1.0549286603927612,
      "learning_rate": 4.407919386886177e-05,
      "loss": 0.0016000311821699142,
      "step": 197090
    },
    {
      "epoch": 55.94663638944082,
      "grad_norm": 1.7141069173812866,
      "learning_rate": 4.4076355378938404e-05,
      "loss": 0.001901523768901825,
      "step": 197100
    },
    {
      "epoch": 55.949474879364175,
      "grad_norm": 0.2701321542263031,
      "learning_rate": 4.4073516889015046e-05,
      "loss": 0.0027799053117632867,
      "step": 197110
    },
    {
      "epoch": 55.95231336928754,
      "grad_norm": 0.32178449630737305,
      "learning_rate": 4.407067839909169e-05,
      "loss": 0.009899512678384782,
      "step": 197120
    },
    {
      "epoch": 55.9551518592109,
      "grad_norm": 0.05778244882822037,
      "learning_rate": 4.406783990916832e-05,
      "loss": 0.001419794373214245,
      "step": 197130
    },
    {
      "epoch": 55.95799034913426,
      "grad_norm": 0.06016255170106888,
      "learning_rate": 4.406500141924496e-05,
      "loss": 0.0010527152568101882,
      "step": 197140
    },
    {
      "epoch": 55.96082883905762,
      "grad_norm": 0.681505560874939,
      "learning_rate": 4.4062162929321604e-05,
      "loss": 0.0016550857573747635,
      "step": 197150
    },
    {
      "epoch": 55.96366732898098,
      "grad_norm": 0.17727383971214294,
      "learning_rate": 4.405932443939824e-05,
      "loss": 0.008316299319267273,
      "step": 197160
    },
    {
      "epoch": 55.966505818904345,
      "grad_norm": 0.15485012531280518,
      "learning_rate": 4.405648594947488e-05,
      "loss": 0.0011253084987401963,
      "step": 197170
    },
    {
      "epoch": 55.9693443088277,
      "grad_norm": 0.037993598729372025,
      "learning_rate": 4.405364745955152e-05,
      "loss": 0.0003286227583885193,
      "step": 197180
    },
    {
      "epoch": 55.972182798751064,
      "grad_norm": 0.0142594575881958,
      "learning_rate": 4.405080896962816e-05,
      "loss": 0.0006614217534661293,
      "step": 197190
    },
    {
      "epoch": 55.97502128867443,
      "grad_norm": 0.18776586651802063,
      "learning_rate": 4.40479704797048e-05,
      "loss": 0.00041862428188323977,
      "step": 197200
    },
    {
      "epoch": 55.97785977859779,
      "grad_norm": 0.11794696003198624,
      "learning_rate": 4.404513198978144e-05,
      "loss": 0.0006279835477471351,
      "step": 197210
    },
    {
      "epoch": 55.980698268521145,
      "grad_norm": 0.014981154352426529,
      "learning_rate": 4.404229349985808e-05,
      "loss": 0.0003104716539382935,
      "step": 197220
    },
    {
      "epoch": 55.98353675844451,
      "grad_norm": 0.04261564463376999,
      "learning_rate": 4.4039455009934715e-05,
      "loss": 0.00048018284142017367,
      "step": 197230
    },
    {
      "epoch": 55.98637524836787,
      "grad_norm": 0.13542304933071136,
      "learning_rate": 4.4036616520011356e-05,
      "loss": 0.00023575909435749053,
      "step": 197240
    },
    {
      "epoch": 55.98921373829123,
      "grad_norm": 0.017865128815174103,
      "learning_rate": 4.4033778030088e-05,
      "loss": 0.00027904212474823,
      "step": 197250
    },
    {
      "epoch": 55.99205222821459,
      "grad_norm": 0.04700678586959839,
      "learning_rate": 4.403093954016463e-05,
      "loss": 0.0004254782572388649,
      "step": 197260
    },
    {
      "epoch": 55.99489071813795,
      "grad_norm": 0.3910329043865204,
      "learning_rate": 4.4028101050241274e-05,
      "loss": 0.0003100203350186348,
      "step": 197270
    },
    {
      "epoch": 55.99772920806131,
      "grad_norm": 0.23227888345718384,
      "learning_rate": 4.4025262560317915e-05,
      "loss": 0.00044828802347183225,
      "step": 197280
    },
    {
      "epoch": 56.00056769798467,
      "grad_norm": 0.037090808153152466,
      "learning_rate": 4.402242407039455e-05,
      "loss": 0.00034112222492694856,
      "step": 197290
    },
    {
      "epoch": 56.003406187908034,
      "grad_norm": 0.14840741455554962,
      "learning_rate": 4.401958558047119e-05,
      "loss": 0.00019460096955299379,
      "step": 197300
    },
    {
      "epoch": 56.0062446778314,
      "grad_norm": 0.0092390775680542,
      "learning_rate": 4.401674709054783e-05,
      "loss": 0.0005055587738752365,
      "step": 197310
    },
    {
      "epoch": 56.00908316775475,
      "grad_norm": 0.012530405074357986,
      "learning_rate": 4.4013908600624474e-05,
      "loss": 0.00013594739139080047,
      "step": 197320
    },
    {
      "epoch": 56.011921657678116,
      "grad_norm": 0.047808460891246796,
      "learning_rate": 4.401107011070111e-05,
      "loss": 0.000666525773704052,
      "step": 197330
    },
    {
      "epoch": 56.01476014760148,
      "grad_norm": 0.041439563035964966,
      "learning_rate": 4.400823162077774e-05,
      "loss": 0.0018152600154280663,
      "step": 197340
    },
    {
      "epoch": 56.017598637524834,
      "grad_norm": 0.159916952252388,
      "learning_rate": 4.400539313085439e-05,
      "loss": 0.0005783768370747566,
      "step": 197350
    },
    {
      "epoch": 56.0204371274482,
      "grad_norm": 0.17146021127700806,
      "learning_rate": 4.4002554640931026e-05,
      "loss": 0.0037273682653903963,
      "step": 197360
    },
    {
      "epoch": 56.02327561737156,
      "grad_norm": 0.2154400646686554,
      "learning_rate": 4.399971615100767e-05,
      "loss": 0.008866132795810699,
      "step": 197370
    },
    {
      "epoch": 56.026114107294916,
      "grad_norm": 0.06461193412542343,
      "learning_rate": 4.399687766108431e-05,
      "loss": 0.0009569931775331497,
      "step": 197380
    },
    {
      "epoch": 56.02895259721828,
      "grad_norm": 0.09547378122806549,
      "learning_rate": 4.399403917116094e-05,
      "loss": 0.011964388936758042,
      "step": 197390
    },
    {
      "epoch": 56.03179108714164,
      "grad_norm": 0.0626763328909874,
      "learning_rate": 4.3991200681237584e-05,
      "loss": 0.0013785533607006072,
      "step": 197400
    },
    {
      "epoch": 56.034629577065004,
      "grad_norm": 0.07333990186452866,
      "learning_rate": 4.3988362191314226e-05,
      "loss": 0.00428132638335228,
      "step": 197410
    },
    {
      "epoch": 56.03746806698836,
      "grad_norm": 0.014438900165259838,
      "learning_rate": 4.398552370139086e-05,
      "loss": 0.0009747935459017753,
      "step": 197420
    },
    {
      "epoch": 56.04030655691172,
      "grad_norm": 0.04271167516708374,
      "learning_rate": 4.39826852114675e-05,
      "loss": 0.00028040036559104917,
      "step": 197430
    },
    {
      "epoch": 56.043145046835086,
      "grad_norm": 0.06776389479637146,
      "learning_rate": 4.3979846721544136e-05,
      "loss": 0.000681416317820549,
      "step": 197440
    },
    {
      "epoch": 56.04598353675844,
      "grad_norm": 0.45363032817840576,
      "learning_rate": 4.397700823162078e-05,
      "loss": 0.0002904154360294342,
      "step": 197450
    },
    {
      "epoch": 56.048822026681805,
      "grad_norm": 0.1014968752861023,
      "learning_rate": 4.397416974169742e-05,
      "loss": 0.00024654120206832884,
      "step": 197460
    },
    {
      "epoch": 56.05166051660517,
      "grad_norm": 0.08723069727420807,
      "learning_rate": 4.3971331251774054e-05,
      "loss": 0.0007890606299042702,
      "step": 197470
    },
    {
      "epoch": 56.05449900652852,
      "grad_norm": 0.01609191484749317,
      "learning_rate": 4.39684927618507e-05,
      "loss": 0.0005990006029605865,
      "step": 197480
    },
    {
      "epoch": 56.057337496451886,
      "grad_norm": 0.2650815546512604,
      "learning_rate": 4.3965654271927336e-05,
      "loss": 0.00029516369104385374,
      "step": 197490
    },
    {
      "epoch": 56.06017598637525,
      "grad_norm": 0.007899067364633083,
      "learning_rate": 4.396281578200397e-05,
      "loss": 0.0005149470642209053,
      "step": 197500
    },
    {
      "epoch": 56.06017598637525,
      "eval_accuracy": 0.9790805620906721,
      "eval_loss": 0.07877887040376663,
      "eval_runtime": 33.0609,
      "eval_samples_per_second": 475.698,
      "eval_steps_per_second": 7.441,
      "step": 197500
    },
    {
      "epoch": 56.06301447629861,
      "grad_norm": 0.07078234851360321,
      "learning_rate": 4.395997729208062e-05,
      "loss": 0.00019044410437345504,
      "step": 197510
    },
    {
      "epoch": 56.06585296622197,
      "grad_norm": 0.7675188183784485,
      "learning_rate": 4.3957138802157254e-05,
      "loss": 0.0002797111868858337,
      "step": 197520
    },
    {
      "epoch": 56.06869145614533,
      "grad_norm": 0.04986502602696419,
      "learning_rate": 4.3954300312233895e-05,
      "loss": 0.00024943500757217406,
      "step": 197530
    },
    {
      "epoch": 56.071529946068694,
      "grad_norm": 0.07227253913879395,
      "learning_rate": 4.3951461822310537e-05,
      "loss": 0.0005524942651391029,
      "step": 197540
    },
    {
      "epoch": 56.07436843599205,
      "grad_norm": 0.0565667487680912,
      "learning_rate": 4.394862333238717e-05,
      "loss": 0.0002444203943014145,
      "step": 197550
    },
    {
      "epoch": 56.07720692591541,
      "grad_norm": 2.75298810005188,
      "learning_rate": 4.394578484246381e-05,
      "loss": 0.0004672294482588768,
      "step": 197560
    },
    {
      "epoch": 56.080045415838775,
      "grad_norm": 0.009504813700914383,
      "learning_rate": 4.394294635254045e-05,
      "loss": 0.00012306235730648042,
      "step": 197570
    },
    {
      "epoch": 56.08288390576214,
      "grad_norm": 0.05454844981431961,
      "learning_rate": 4.394010786261709e-05,
      "loss": 0.0011475438252091408,
      "step": 197580
    },
    {
      "epoch": 56.085722395685494,
      "grad_norm": 0.00970033835619688,
      "learning_rate": 4.393726937269373e-05,
      "loss": 0.0003916572779417038,
      "step": 197590
    },
    {
      "epoch": 56.08856088560886,
      "grad_norm": 0.06024151295423508,
      "learning_rate": 4.3934430882770364e-05,
      "loss": 0.000473402813076973,
      "step": 197600
    },
    {
      "epoch": 56.09139937553222,
      "grad_norm": 0.09335778653621674,
      "learning_rate": 4.393159239284701e-05,
      "loss": 0.0005736978724598885,
      "step": 197610
    },
    {
      "epoch": 56.094237865455575,
      "grad_norm": 0.512845516204834,
      "learning_rate": 4.392875390292365e-05,
      "loss": 0.0012324657291173935,
      "step": 197620
    },
    {
      "epoch": 56.09707635537894,
      "grad_norm": 0.6239861249923706,
      "learning_rate": 4.392591541300028e-05,
      "loss": 0.009652135521173477,
      "step": 197630
    },
    {
      "epoch": 56.0999148453023,
      "grad_norm": 0.030514217913150787,
      "learning_rate": 4.392307692307693e-05,
      "loss": 0.0006255961954593659,
      "step": 197640
    },
    {
      "epoch": 56.10275333522566,
      "grad_norm": 3.043515682220459,
      "learning_rate": 4.3920238433153565e-05,
      "loss": 0.000696626678109169,
      "step": 197650
    },
    {
      "epoch": 56.10559182514902,
      "grad_norm": 0.016939260065555573,
      "learning_rate": 4.3917399943230206e-05,
      "loss": 0.0012533944100141525,
      "step": 197660
    },
    {
      "epoch": 56.10843031507238,
      "grad_norm": 0.34762042760849,
      "learning_rate": 4.391456145330684e-05,
      "loss": 0.0004999421536922455,
      "step": 197670
    },
    {
      "epoch": 56.111268804995746,
      "grad_norm": 0.03120344877243042,
      "learning_rate": 4.391172296338348e-05,
      "loss": 0.0007635621353983879,
      "step": 197680
    },
    {
      "epoch": 56.1141072949191,
      "grad_norm": 0.06392840296030045,
      "learning_rate": 4.390888447346012e-05,
      "loss": 0.005988557636737823,
      "step": 197690
    },
    {
      "epoch": 56.116945784842464,
      "grad_norm": 0.03505518659949303,
      "learning_rate": 4.390604598353676e-05,
      "loss": 0.0008239742368459701,
      "step": 197700
    },
    {
      "epoch": 56.11978427476583,
      "grad_norm": 0.035669561475515366,
      "learning_rate": 4.39032074936134e-05,
      "loss": 0.0005952177569270134,
      "step": 197710
    },
    {
      "epoch": 56.12262276468918,
      "grad_norm": 0.47885778546333313,
      "learning_rate": 4.390036900369004e-05,
      "loss": 0.0005559653043746948,
      "step": 197720
    },
    {
      "epoch": 56.125461254612546,
      "grad_norm": 0.1748822182416916,
      "learning_rate": 4.3897530513766675e-05,
      "loss": 0.0010783966630697251,
      "step": 197730
    },
    {
      "epoch": 56.12829974453591,
      "grad_norm": 0.20833677053451538,
      "learning_rate": 4.389469202384332e-05,
      "loss": 0.001283513940870762,
      "step": 197740
    },
    {
      "epoch": 56.131138234459264,
      "grad_norm": 4.170480251312256,
      "learning_rate": 4.389185353391996e-05,
      "loss": 0.0017899546772241592,
      "step": 197750
    },
    {
      "epoch": 56.13397672438263,
      "grad_norm": 4.9609832763671875,
      "learning_rate": 4.388901504399659e-05,
      "loss": 0.008066914975643158,
      "step": 197760
    },
    {
      "epoch": 56.13681521430599,
      "grad_norm": 0.1354832798242569,
      "learning_rate": 4.3886176554073234e-05,
      "loss": 0.0026313146576285363,
      "step": 197770
    },
    {
      "epoch": 56.13965370422935,
      "grad_norm": 2.690324306488037,
      "learning_rate": 4.3883338064149875e-05,
      "loss": 0.0008557170629501342,
      "step": 197780
    },
    {
      "epoch": 56.14249219415271,
      "grad_norm": 0.33090153336524963,
      "learning_rate": 4.388049957422652e-05,
      "loss": 0.0046662587672472,
      "step": 197790
    },
    {
      "epoch": 56.14533068407607,
      "grad_norm": 1.0361905097961426,
      "learning_rate": 4.387766108430315e-05,
      "loss": 0.000522649846971035,
      "step": 197800
    },
    {
      "epoch": 56.148169173999435,
      "grad_norm": 0.07533556967973709,
      "learning_rate": 4.387482259437979e-05,
      "loss": 0.004646455124020577,
      "step": 197810
    },
    {
      "epoch": 56.15100766392279,
      "grad_norm": 1.1133232116699219,
      "learning_rate": 4.3871984104456434e-05,
      "loss": 0.00034068115055561064,
      "step": 197820
    },
    {
      "epoch": 56.15384615384615,
      "grad_norm": 0.051488231867551804,
      "learning_rate": 4.386914561453307e-05,
      "loss": 0.00017730072140693666,
      "step": 197830
    },
    {
      "epoch": 56.156684643769516,
      "grad_norm": 0.01392693817615509,
      "learning_rate": 4.386630712460971e-05,
      "loss": 0.0013605743646621705,
      "step": 197840
    },
    {
      "epoch": 56.15952313369287,
      "grad_norm": 0.18626351654529572,
      "learning_rate": 4.386346863468635e-05,
      "loss": 0.00041651967912912367,
      "step": 197850
    },
    {
      "epoch": 56.162361623616235,
      "grad_norm": 0.3668014705181122,
      "learning_rate": 4.3860630144762986e-05,
      "loss": 0.004006946459412575,
      "step": 197860
    },
    {
      "epoch": 56.1652001135396,
      "grad_norm": 4.950279235839844,
      "learning_rate": 4.385779165483963e-05,
      "loss": 0.0015882330015301704,
      "step": 197870
    },
    {
      "epoch": 56.16803860346296,
      "grad_norm": 1.7250604629516602,
      "learning_rate": 4.385495316491627e-05,
      "loss": 0.005562891811132431,
      "step": 197880
    },
    {
      "epoch": 56.170877093386316,
      "grad_norm": 0.20169413089752197,
      "learning_rate": 4.38521146749929e-05,
      "loss": 0.0005310537293553352,
      "step": 197890
    },
    {
      "epoch": 56.17371558330968,
      "grad_norm": 2.4441425800323486,
      "learning_rate": 4.3849276185069545e-05,
      "loss": 0.008076748251914978,
      "step": 197900
    },
    {
      "epoch": 56.17655407323304,
      "grad_norm": 1.2797824144363403,
      "learning_rate": 4.3846437695146186e-05,
      "loss": 0.009489230811595917,
      "step": 197910
    },
    {
      "epoch": 56.1793925631564,
      "grad_norm": 1.0756442546844482,
      "learning_rate": 4.384359920522282e-05,
      "loss": 0.019038066267967224,
      "step": 197920
    },
    {
      "epoch": 56.18223105307976,
      "grad_norm": 2.0704245567321777,
      "learning_rate": 4.384076071529946e-05,
      "loss": 0.0006803514435887337,
      "step": 197930
    },
    {
      "epoch": 56.185069543003124,
      "grad_norm": 1.1507623195648193,
      "learning_rate": 4.3837922225376103e-05,
      "loss": 0.005391133576631546,
      "step": 197940
    },
    {
      "epoch": 56.18790803292649,
      "grad_norm": 0.4097041189670563,
      "learning_rate": 4.3835083735452745e-05,
      "loss": 0.011469721794128418,
      "step": 197950
    },
    {
      "epoch": 56.19074652284984,
      "grad_norm": 2.59089994430542,
      "learning_rate": 4.383224524552938e-05,
      "loss": 0.015905682742595673,
      "step": 197960
    },
    {
      "epoch": 56.193585012773205,
      "grad_norm": 1.6597341299057007,
      "learning_rate": 4.3829406755606014e-05,
      "loss": 0.00253769401460886,
      "step": 197970
    },
    {
      "epoch": 56.19642350269657,
      "grad_norm": 0.030487125739455223,
      "learning_rate": 4.382656826568266e-05,
      "loss": 0.0017853746190667152,
      "step": 197980
    },
    {
      "epoch": 56.199261992619924,
      "grad_norm": 0.08982520550489426,
      "learning_rate": 4.38237297757593e-05,
      "loss": 0.000512753427028656,
      "step": 197990
    },
    {
      "epoch": 56.20210048254329,
      "grad_norm": 0.14178965985774994,
      "learning_rate": 4.382089128583594e-05,
      "loss": 0.006114903837442398,
      "step": 198000
    },
    {
      "epoch": 56.20210048254329,
      "eval_accuracy": 0.9775545240668914,
      "eval_loss": 0.077818863093853,
      "eval_runtime": 32.4909,
      "eval_samples_per_second": 484.043,
      "eval_steps_per_second": 7.571,
      "step": 198000
    },
    {
      "epoch": 56.20493897246665,
      "grad_norm": 0.0845697894692421,
      "learning_rate": 4.381805279591258e-05,
      "loss": 0.0002902230247855186,
      "step": 198010
    },
    {
      "epoch": 56.207777462390005,
      "grad_norm": 0.22844667732715607,
      "learning_rate": 4.3815214305989214e-05,
      "loss": 0.00033297352492809293,
      "step": 198020
    },
    {
      "epoch": 56.21061595231337,
      "grad_norm": 0.012572943232953548,
      "learning_rate": 4.3812375816065855e-05,
      "loss": 0.0010914696380496026,
      "step": 198030
    },
    {
      "epoch": 56.21345444223673,
      "grad_norm": 0.5784156322479248,
      "learning_rate": 4.38095373261425e-05,
      "loss": 0.0011450149118900298,
      "step": 198040
    },
    {
      "epoch": 56.216292932160094,
      "grad_norm": 0.5785497426986694,
      "learning_rate": 4.380669883621913e-05,
      "loss": 0.000695529393851757,
      "step": 198050
    },
    {
      "epoch": 56.21913142208345,
      "grad_norm": 0.109353207051754,
      "learning_rate": 4.380386034629577e-05,
      "loss": 0.0005987774580717087,
      "step": 198060
    },
    {
      "epoch": 56.22196991200681,
      "grad_norm": 0.16899922490119934,
      "learning_rate": 4.380102185637241e-05,
      "loss": 0.0011895840987563133,
      "step": 198070
    },
    {
      "epoch": 56.224808401930176,
      "grad_norm": 0.644934356212616,
      "learning_rate": 4.3798183366449056e-05,
      "loss": 0.0005254877731204032,
      "step": 198080
    },
    {
      "epoch": 56.22764689185353,
      "grad_norm": 0.15639841556549072,
      "learning_rate": 4.379534487652569e-05,
      "loss": 0.0020797932520508766,
      "step": 198090
    },
    {
      "epoch": 56.230485381776894,
      "grad_norm": 0.015220246277749538,
      "learning_rate": 4.3792506386602325e-05,
      "loss": 0.0020483529195189478,
      "step": 198100
    },
    {
      "epoch": 56.23332387170026,
      "grad_norm": 0.031398188322782516,
      "learning_rate": 4.378966789667897e-05,
      "loss": 0.0006619745865464211,
      "step": 198110
    },
    {
      "epoch": 56.23616236162361,
      "grad_norm": 0.12131484597921371,
      "learning_rate": 4.378682940675561e-05,
      "loss": 0.00025120191276073456,
      "step": 198120
    },
    {
      "epoch": 56.239000851546976,
      "grad_norm": 0.4681147038936615,
      "learning_rate": 4.378399091683225e-05,
      "loss": 0.0012213705107569695,
      "step": 198130
    },
    {
      "epoch": 56.24183934147034,
      "grad_norm": 0.4163227379322052,
      "learning_rate": 4.378115242690889e-05,
      "loss": 0.0019706455990672112,
      "step": 198140
    },
    {
      "epoch": 56.2446778313937,
      "grad_norm": 0.037470411509275436,
      "learning_rate": 4.3778313936985525e-05,
      "loss": 0.000494806095957756,
      "step": 198150
    },
    {
      "epoch": 56.24751632131706,
      "grad_norm": 0.2693958878517151,
      "learning_rate": 4.3775475447062166e-05,
      "loss": 0.0010548725724220276,
      "step": 198160
    },
    {
      "epoch": 56.25035481124042,
      "grad_norm": 0.11869361251592636,
      "learning_rate": 4.37726369571388e-05,
      "loss": 0.0007648164406418801,
      "step": 198170
    },
    {
      "epoch": 56.25319330116378,
      "grad_norm": 0.21156053245067596,
      "learning_rate": 4.376979846721544e-05,
      "loss": 0.001613084226846695,
      "step": 198180
    },
    {
      "epoch": 56.25603179108714,
      "grad_norm": 12.185402870178223,
      "learning_rate": 4.3766959977292084e-05,
      "loss": 0.009113604575395584,
      "step": 198190
    },
    {
      "epoch": 56.2588702810105,
      "grad_norm": 0.05877240002155304,
      "learning_rate": 4.376412148736872e-05,
      "loss": 0.0006514696404337883,
      "step": 198200
    },
    {
      "epoch": 56.261708770933865,
      "grad_norm": 0.06642419099807739,
      "learning_rate": 4.3761282997445366e-05,
      "loss": 0.0006647750735282898,
      "step": 198210
    },
    {
      "epoch": 56.26454726085722,
      "grad_norm": 0.1845230609178543,
      "learning_rate": 4.3758444507522e-05,
      "loss": 0.0009336881339550018,
      "step": 198220
    },
    {
      "epoch": 56.26738575078058,
      "grad_norm": 1.1860766410827637,
      "learning_rate": 4.3755606017598636e-05,
      "loss": 0.0005762943997979164,
      "step": 198230
    },
    {
      "epoch": 56.270224240703946,
      "grad_norm": 0.033912964165210724,
      "learning_rate": 4.3752767527675284e-05,
      "loss": 0.0007745284587144851,
      "step": 198240
    },
    {
      "epoch": 56.27306273062731,
      "grad_norm": 0.18883812427520752,
      "learning_rate": 4.374992903775192e-05,
      "loss": 0.0006298128515481949,
      "step": 198250
    },
    {
      "epoch": 56.275901220550665,
      "grad_norm": 1.4490346908569336,
      "learning_rate": 4.374709054782856e-05,
      "loss": 0.004904014617204666,
      "step": 198260
    },
    {
      "epoch": 56.27873971047403,
      "grad_norm": 0.2459600269794464,
      "learning_rate": 4.3744252057905194e-05,
      "loss": 0.0005243731662631035,
      "step": 198270
    },
    {
      "epoch": 56.28157820039739,
      "grad_norm": 0.563421368598938,
      "learning_rate": 4.3741413567981836e-05,
      "loss": 0.00029081404209136964,
      "step": 198280
    },
    {
      "epoch": 56.28441669032075,
      "grad_norm": 0.3818355202674866,
      "learning_rate": 4.373857507805848e-05,
      "loss": 0.0006073202937841415,
      "step": 198290
    },
    {
      "epoch": 56.28725518024411,
      "grad_norm": 0.028666572645306587,
      "learning_rate": 4.373573658813511e-05,
      "loss": 0.0002596151083707809,
      "step": 198300
    },
    {
      "epoch": 56.29009367016747,
      "grad_norm": 0.1813509315252304,
      "learning_rate": 4.373289809821175e-05,
      "loss": 0.00020192544907331466,
      "step": 198310
    },
    {
      "epoch": 56.29293216009083,
      "grad_norm": 0.04363854601979256,
      "learning_rate": 4.373034345728073e-05,
      "loss": 0.00372762531042099,
      "step": 198320
    },
    {
      "epoch": 56.29577065001419,
      "grad_norm": 0.013831849209964275,
      "learning_rate": 4.372750496735737e-05,
      "loss": 0.0031982794404029845,
      "step": 198330
    },
    {
      "epoch": 56.298609139937554,
      "grad_norm": 0.015881000086665154,
      "learning_rate": 4.372466647743401e-05,
      "loss": 0.00013125278055667878,
      "step": 198340
    },
    {
      "epoch": 56.30144762986092,
      "grad_norm": 0.07827562093734741,
      "learning_rate": 4.3721827987510644e-05,
      "loss": 0.0012825321406126021,
      "step": 198350
    },
    {
      "epoch": 56.30428611978427,
      "grad_norm": 0.0406169667840004,
      "learning_rate": 4.3718989497587285e-05,
      "loss": 0.0017355794087052345,
      "step": 198360
    },
    {
      "epoch": 56.307124609707635,
      "grad_norm": 0.030913464725017548,
      "learning_rate": 4.3716151007663926e-05,
      "loss": 0.00036715492606163023,
      "step": 198370
    },
    {
      "epoch": 56.309963099631,
      "grad_norm": 0.007096961606293917,
      "learning_rate": 4.371331251774056e-05,
      "loss": 0.0002389693632721901,
      "step": 198380
    },
    {
      "epoch": 56.312801589554354,
      "grad_norm": 0.021570827811956406,
      "learning_rate": 4.37104740278172e-05,
      "loss": 0.00025997795164585115,
      "step": 198390
    },
    {
      "epoch": 56.31564007947772,
      "grad_norm": 0.01946287788450718,
      "learning_rate": 4.3707635537893844e-05,
      "loss": 0.000519716925919056,
      "step": 198400
    },
    {
      "epoch": 56.31847856940108,
      "grad_norm": 0.08321772515773773,
      "learning_rate": 4.370479704797048e-05,
      "loss": 0.0026102373376488686,
      "step": 198410
    },
    {
      "epoch": 56.32131705932444,
      "grad_norm": 0.0069947256706655025,
      "learning_rate": 4.3701958558047126e-05,
      "loss": 0.000380752794444561,
      "step": 198420
    },
    {
      "epoch": 56.3241555492478,
      "grad_norm": 14.97995376586914,
      "learning_rate": 4.369912006812376e-05,
      "loss": 0.007968942821025848,
      "step": 198430
    },
    {
      "epoch": 56.32699403917116,
      "grad_norm": 5.838681697845459,
      "learning_rate": 4.3696281578200396e-05,
      "loss": 0.001706031896173954,
      "step": 198440
    },
    {
      "epoch": 56.329832529094524,
      "grad_norm": 1.4455469846725464,
      "learning_rate": 4.369344308827704e-05,
      "loss": 0.006653010100126267,
      "step": 198450
    },
    {
      "epoch": 56.33267101901788,
      "grad_norm": 0.3108550012111664,
      "learning_rate": 4.369060459835368e-05,
      "loss": 0.00085301473736763,
      "step": 198460
    },
    {
      "epoch": 56.33550950894124,
      "grad_norm": 0.05176917836070061,
      "learning_rate": 4.368776610843032e-05,
      "loss": 0.0037717752158641813,
      "step": 198470
    },
    {
      "epoch": 56.338347998864606,
      "grad_norm": 0.10225865989923477,
      "learning_rate": 4.3684927618506954e-05,
      "loss": 0.0005568321794271469,
      "step": 198480
    },
    {
      "epoch": 56.34118648878796,
      "grad_norm": 0.0787891075015068,
      "learning_rate": 4.3682089128583596e-05,
      "loss": 0.001842343993484974,
      "step": 198490
    },
    {
      "epoch": 56.344024978711325,
      "grad_norm": 0.19564880430698395,
      "learning_rate": 4.367925063866024e-05,
      "loss": 0.003161030262708664,
      "step": 198500
    },
    {
      "epoch": 56.344024978711325,
      "eval_accuracy": 0.9762828257137407,
      "eval_loss": 0.08471360057592392,
      "eval_runtime": 32.197,
      "eval_samples_per_second": 488.462,
      "eval_steps_per_second": 7.64,
      "step": 198500
    },
    {
      "epoch": 56.34686346863469,
      "grad_norm": 0.07728838920593262,
      "learning_rate": 4.367641214873687e-05,
      "loss": 0.0004705877974629402,
      "step": 198510
    },
    {
      "epoch": 56.34970195855805,
      "grad_norm": 0.06463629752397537,
      "learning_rate": 4.367357365881351e-05,
      "loss": 0.0003051610663533211,
      "step": 198520
    },
    {
      "epoch": 56.352540448481406,
      "grad_norm": 0.017057890072464943,
      "learning_rate": 4.3670735168890154e-05,
      "loss": 0.0003836583346128464,
      "step": 198530
    },
    {
      "epoch": 56.35537893840477,
      "grad_norm": 0.0918227881193161,
      "learning_rate": 4.366789667896679e-05,
      "loss": 0.000265682116150856,
      "step": 198540
    },
    {
      "epoch": 56.35821742832813,
      "grad_norm": 0.028482750058174133,
      "learning_rate": 4.366505818904343e-05,
      "loss": 0.000638226605951786,
      "step": 198550
    },
    {
      "epoch": 56.36105591825149,
      "grad_norm": 0.06451809406280518,
      "learning_rate": 4.366221969912007e-05,
      "loss": 0.0007457753643393516,
      "step": 198560
    },
    {
      "epoch": 56.36389440817485,
      "grad_norm": 1.1632802486419678,
      "learning_rate": 4.3659381209196706e-05,
      "loss": 0.0024611489847302438,
      "step": 198570
    },
    {
      "epoch": 56.36673289809821,
      "grad_norm": 0.0964711382985115,
      "learning_rate": 4.365654271927335e-05,
      "loss": 0.0007909022271633149,
      "step": 198580
    },
    {
      "epoch": 56.36957138802157,
      "grad_norm": 0.019427409395575523,
      "learning_rate": 4.365370422934999e-05,
      "loss": 0.002803533896803856,
      "step": 198590
    },
    {
      "epoch": 56.37240987794493,
      "grad_norm": 0.14352138340473175,
      "learning_rate": 4.365086573942663e-05,
      "loss": 0.00416020043194294,
      "step": 198600
    },
    {
      "epoch": 56.375248367868295,
      "grad_norm": 0.67116379737854,
      "learning_rate": 4.3648027249503265e-05,
      "loss": 0.0006471769884228706,
      "step": 198610
    },
    {
      "epoch": 56.37808685779166,
      "grad_norm": 1.7457098960876465,
      "learning_rate": 4.3645188759579906e-05,
      "loss": 0.0010544246062636376,
      "step": 198620
    },
    {
      "epoch": 56.380925347715014,
      "grad_norm": 0.05199697986245155,
      "learning_rate": 4.364235026965655e-05,
      "loss": 0.0003349499776959419,
      "step": 198630
    },
    {
      "epoch": 56.38376383763838,
      "grad_norm": 3.4960880279541016,
      "learning_rate": 4.363951177973318e-05,
      "loss": 0.0023002691566944124,
      "step": 198640
    },
    {
      "epoch": 56.38660232756174,
      "grad_norm": 0.13748858869075775,
      "learning_rate": 4.3636673289809824e-05,
      "loss": 0.0025214117020368578,
      "step": 198650
    },
    {
      "epoch": 56.389440817485095,
      "grad_norm": 0.39394262433052063,
      "learning_rate": 4.3633834799886465e-05,
      "loss": 0.0012639863416552544,
      "step": 198660
    },
    {
      "epoch": 56.39227930740846,
      "grad_norm": 0.25414419174194336,
      "learning_rate": 4.36309963099631e-05,
      "loss": 0.00042229443788528443,
      "step": 198670
    },
    {
      "epoch": 56.39511779733182,
      "grad_norm": 0.7672843337059021,
      "learning_rate": 4.362815782003974e-05,
      "loss": 0.0007474036887288093,
      "step": 198680
    },
    {
      "epoch": 56.39795628725518,
      "grad_norm": 0.12842409312725067,
      "learning_rate": 4.362531933011638e-05,
      "loss": 0.004470654204487801,
      "step": 198690
    },
    {
      "epoch": 56.40079477717854,
      "grad_norm": 0.03979675471782684,
      "learning_rate": 4.362248084019302e-05,
      "loss": 0.0007620995864272117,
      "step": 198700
    },
    {
      "epoch": 56.4036332671019,
      "grad_norm": 0.0909903272986412,
      "learning_rate": 4.361964235026966e-05,
      "loss": 0.0024743080139160156,
      "step": 198710
    },
    {
      "epoch": 56.406471757025265,
      "grad_norm": 0.7290894985198975,
      "learning_rate": 4.36168038603463e-05,
      "loss": 0.0004583677276968956,
      "step": 198720
    },
    {
      "epoch": 56.40931024694862,
      "grad_norm": 0.4180063009262085,
      "learning_rate": 4.361396537042294e-05,
      "loss": 0.0011193234473466872,
      "step": 198730
    },
    {
      "epoch": 56.412148736871984,
      "grad_norm": 0.038974978029727936,
      "learning_rate": 4.3611126880499576e-05,
      "loss": 0.00048462022095918655,
      "step": 198740
    },
    {
      "epoch": 56.41498722679535,
      "grad_norm": 0.015640880912542343,
      "learning_rate": 4.360828839057621e-05,
      "loss": 0.0013642387464642524,
      "step": 198750
    },
    {
      "epoch": 56.4178257167187,
      "grad_norm": 0.033592142164707184,
      "learning_rate": 4.360544990065286e-05,
      "loss": 0.0003052284941077232,
      "step": 198760
    },
    {
      "epoch": 56.420664206642066,
      "grad_norm": 0.02572907879948616,
      "learning_rate": 4.360261141072949e-05,
      "loss": 0.0004195353016257286,
      "step": 198770
    },
    {
      "epoch": 56.42350269656543,
      "grad_norm": 2.2455687522888184,
      "learning_rate": 4.3599772920806135e-05,
      "loss": 0.005290890857577324,
      "step": 198780
    },
    {
      "epoch": 56.42634118648879,
      "grad_norm": 0.17791533470153809,
      "learning_rate": 4.3596934430882776e-05,
      "loss": 0.000258168950676918,
      "step": 198790
    },
    {
      "epoch": 56.42917967641215,
      "grad_norm": 0.022287752479314804,
      "learning_rate": 4.359409594095941e-05,
      "loss": 0.004058560356497765,
      "step": 198800
    },
    {
      "epoch": 56.43201816633551,
      "grad_norm": 0.16986031830310822,
      "learning_rate": 4.359125745103605e-05,
      "loss": 0.0016647623851895332,
      "step": 198810
    },
    {
      "epoch": 56.43485665625887,
      "grad_norm": 0.186801478266716,
      "learning_rate": 4.358841896111269e-05,
      "loss": 0.0004996253177523613,
      "step": 198820
    },
    {
      "epoch": 56.43769514618223,
      "grad_norm": 0.953207790851593,
      "learning_rate": 4.358558047118933e-05,
      "loss": 0.0003852887079119682,
      "step": 198830
    },
    {
      "epoch": 56.44053363610559,
      "grad_norm": 0.25107598304748535,
      "learning_rate": 4.358274198126597e-05,
      "loss": 0.0014645962044596672,
      "step": 198840
    },
    {
      "epoch": 56.443372126028954,
      "grad_norm": 0.06009405106306076,
      "learning_rate": 4.3579903491342604e-05,
      "loss": 0.0005159026011824607,
      "step": 198850
    },
    {
      "epoch": 56.44621061595231,
      "grad_norm": 0.42033523321151733,
      "learning_rate": 4.3577065001419245e-05,
      "loss": 0.00023478940129280091,
      "step": 198860
    },
    {
      "epoch": 56.44904910587567,
      "grad_norm": 0.03515841066837311,
      "learning_rate": 4.3574226511495887e-05,
      "loss": 0.00023357272148132323,
      "step": 198870
    },
    {
      "epoch": 56.451887595799036,
      "grad_norm": 0.1380787193775177,
      "learning_rate": 4.357138802157252e-05,
      "loss": 0.002481667883694172,
      "step": 198880
    },
    {
      "epoch": 56.4547260857224,
      "grad_norm": 0.06295204907655716,
      "learning_rate": 4.356854953164917e-05,
      "loss": 0.00037338715046644213,
      "step": 198890
    },
    {
      "epoch": 56.457564575645755,
      "grad_norm": 0.6336563229560852,
      "learning_rate": 4.3565711041725804e-05,
      "loss": 0.0007837843149900436,
      "step": 198900
    },
    {
      "epoch": 56.46040306556912,
      "grad_norm": 0.04138362780213356,
      "learning_rate": 4.356287255180244e-05,
      "loss": 0.0031766965985298157,
      "step": 198910
    },
    {
      "epoch": 56.46324155549248,
      "grad_norm": 0.1436455249786377,
      "learning_rate": 4.356003406187909e-05,
      "loss": 0.0003342380747199059,
      "step": 198920
    },
    {
      "epoch": 56.466080045415836,
      "grad_norm": 0.5750896334648132,
      "learning_rate": 4.355719557195572e-05,
      "loss": 0.0007151016965508461,
      "step": 198930
    },
    {
      "epoch": 56.4689185353392,
      "grad_norm": 0.11192185431718826,
      "learning_rate": 4.355435708203236e-05,
      "loss": 0.0023063011467456818,
      "step": 198940
    },
    {
      "epoch": 56.47175702526256,
      "grad_norm": 0.08794848620891571,
      "learning_rate": 4.3551518592109e-05,
      "loss": 0.0008930442854762077,
      "step": 198950
    },
    {
      "epoch": 56.47459551518592,
      "grad_norm": 0.32268255949020386,
      "learning_rate": 4.354868010218564e-05,
      "loss": 0.0010062240064144135,
      "step": 198960
    },
    {
      "epoch": 56.47743400510928,
      "grad_norm": 0.07041940838098526,
      "learning_rate": 4.354584161226228e-05,
      "loss": 0.00022476539015769958,
      "step": 198970
    },
    {
      "epoch": 56.480272495032644,
      "grad_norm": 0.402192085981369,
      "learning_rate": 4.3543003122338915e-05,
      "loss": 0.0008492724969983101,
      "step": 198980
    },
    {
      "epoch": 56.48311098495601,
      "grad_norm": 0.12153196334838867,
      "learning_rate": 4.3540164632415556e-05,
      "loss": 0.001067104935646057,
      "step": 198990
    },
    {
      "epoch": 56.48594947487936,
      "grad_norm": 0.7671191096305847,
      "learning_rate": 4.35373261424922e-05,
      "loss": 0.0003671528771519661,
      "step": 199000
    },
    {
      "epoch": 56.48594947487936,
      "eval_accuracy": 0.9766643352196859,
      "eval_loss": 0.07846841216087341,
      "eval_runtime": 33.4429,
      "eval_samples_per_second": 470.264,
      "eval_steps_per_second": 7.356,
      "step": 199000
    },
    {
      "epoch": 56.488787964802725,
      "grad_norm": 0.029213303700089455,
      "learning_rate": 4.353448765256883e-05,
      "loss": 0.0007071943953633308,
      "step": 199010
    },
    {
      "epoch": 56.49162645472609,
      "grad_norm": 0.057255640625953674,
      "learning_rate": 4.353164916264548e-05,
      "loss": 0.0015198281034827232,
      "step": 199020
    },
    {
      "epoch": 56.494464944649444,
      "grad_norm": 0.2156606912612915,
      "learning_rate": 4.3528810672722115e-05,
      "loss": 0.0020401576533913613,
      "step": 199030
    },
    {
      "epoch": 56.49730343457281,
      "grad_norm": 0.0633658692240715,
      "learning_rate": 4.352597218279875e-05,
      "loss": 0.0009381350129842759,
      "step": 199040
    },
    {
      "epoch": 56.50014192449617,
      "grad_norm": 0.12863902747631073,
      "learning_rate": 4.352313369287539e-05,
      "loss": 0.0039426635950803755,
      "step": 199050
    },
    {
      "epoch": 56.502980414419525,
      "grad_norm": 0.03906010463833809,
      "learning_rate": 4.352029520295203e-05,
      "loss": 0.0007244830951094628,
      "step": 199060
    },
    {
      "epoch": 56.50581890434289,
      "grad_norm": 0.08160391449928284,
      "learning_rate": 4.3517456713028673e-05,
      "loss": 0.0007255004718899726,
      "step": 199070
    },
    {
      "epoch": 56.50865739426625,
      "grad_norm": 0.0738561674952507,
      "learning_rate": 4.351461822310531e-05,
      "loss": 0.0009003078565001487,
      "step": 199080
    },
    {
      "epoch": 56.511495884189614,
      "grad_norm": 0.03525705635547638,
      "learning_rate": 4.351177973318195e-05,
      "loss": 0.0002653932198882103,
      "step": 199090
    },
    {
      "epoch": 56.51433437411297,
      "grad_norm": 0.35569605231285095,
      "learning_rate": 4.350894124325859e-05,
      "loss": 0.0006025351583957672,
      "step": 199100
    },
    {
      "epoch": 56.51717286403633,
      "grad_norm": 0.06256183236837387,
      "learning_rate": 4.3506102753335225e-05,
      "loss": 0.0005217483267188072,
      "step": 199110
    },
    {
      "epoch": 56.520011353959696,
      "grad_norm": 0.4175226092338562,
      "learning_rate": 4.350326426341187e-05,
      "loss": 0.00043813586235046386,
      "step": 199120
    },
    {
      "epoch": 56.52284984388305,
      "grad_norm": 1.0195633172988892,
      "learning_rate": 4.350042577348851e-05,
      "loss": 0.0005567774176597595,
      "step": 199130
    },
    {
      "epoch": 56.525688333806414,
      "grad_norm": 0.01752062700688839,
      "learning_rate": 4.349758728356514e-05,
      "loss": 0.0005378371104598045,
      "step": 199140
    },
    {
      "epoch": 56.52852682372978,
      "grad_norm": 9.5982027053833,
      "learning_rate": 4.3494748793641784e-05,
      "loss": 0.0048179224133491514,
      "step": 199150
    },
    {
      "epoch": 56.53136531365314,
      "grad_norm": 0.3466532826423645,
      "learning_rate": 4.3491910303718425e-05,
      "loss": 0.004235567152500152,
      "step": 199160
    },
    {
      "epoch": 56.534203803576496,
      "grad_norm": 0.027063734829425812,
      "learning_rate": 4.348907181379506e-05,
      "loss": 0.0003608273342251778,
      "step": 199170
    },
    {
      "epoch": 56.53704229349986,
      "grad_norm": 0.034100525081157684,
      "learning_rate": 4.34862333238717e-05,
      "loss": 0.0005609307438135148,
      "step": 199180
    },
    {
      "epoch": 56.53988078342322,
      "grad_norm": 0.17738543450832367,
      "learning_rate": 4.348339483394834e-05,
      "loss": 0.00014857780188322068,
      "step": 199190
    },
    {
      "epoch": 56.54271927334658,
      "grad_norm": 0.0913369208574295,
      "learning_rate": 4.348055634402498e-05,
      "loss": 0.00027710162103176115,
      "step": 199200
    },
    {
      "epoch": 56.54555776326994,
      "grad_norm": 0.05409266799688339,
      "learning_rate": 4.347771785410162e-05,
      "loss": 0.000291801430284977,
      "step": 199210
    },
    {
      "epoch": 56.5483962531933,
      "grad_norm": 0.47202062606811523,
      "learning_rate": 4.347487936417826e-05,
      "loss": 0.0002573749050498009,
      "step": 199220
    },
    {
      "epoch": 56.55123474311666,
      "grad_norm": 0.11567140370607376,
      "learning_rate": 4.34720408742549e-05,
      "loss": 0.000920984148979187,
      "step": 199230
    },
    {
      "epoch": 56.55407323304002,
      "grad_norm": 0.024555141106247902,
      "learning_rate": 4.3469202384331536e-05,
      "loss": 0.0001989012584090233,
      "step": 199240
    },
    {
      "epoch": 56.556911722963385,
      "grad_norm": 0.015964647755026817,
      "learning_rate": 4.346636389440818e-05,
      "loss": 0.00015730056911706926,
      "step": 199250
    },
    {
      "epoch": 56.55975021288675,
      "grad_norm": 0.0557699017226696,
      "learning_rate": 4.346352540448482e-05,
      "loss": 0.0006092982366681099,
      "step": 199260
    },
    {
      "epoch": 56.5625887028101,
      "grad_norm": 0.005719055887311697,
      "learning_rate": 4.3460686914561453e-05,
      "loss": 0.00035339519381523133,
      "step": 199270
    },
    {
      "epoch": 56.565427192733466,
      "grad_norm": 0.19153249263763428,
      "learning_rate": 4.3457848424638095e-05,
      "loss": 0.00018258262425661088,
      "step": 199280
    },
    {
      "epoch": 56.56826568265683,
      "grad_norm": 0.027012955397367477,
      "learning_rate": 4.3455009934714736e-05,
      "loss": 0.00018149204552173615,
      "step": 199290
    },
    {
      "epoch": 56.571104172580185,
      "grad_norm": 0.1535273641347885,
      "learning_rate": 4.345217144479137e-05,
      "loss": 0.00018101446330547332,
      "step": 199300
    },
    {
      "epoch": 56.57394266250355,
      "grad_norm": 0.07806742191314697,
      "learning_rate": 4.344933295486801e-05,
      "loss": 0.0002159474417567253,
      "step": 199310
    },
    {
      "epoch": 56.57678115242691,
      "grad_norm": 0.01677834242582321,
      "learning_rate": 4.3446494464944654e-05,
      "loss": 0.00019921697676181794,
      "step": 199320
    },
    {
      "epoch": 56.579619642350266,
      "grad_norm": 0.04106127843260765,
      "learning_rate": 4.344365597502129e-05,
      "loss": 0.0007447617128491402,
      "step": 199330
    },
    {
      "epoch": 56.58245813227363,
      "grad_norm": 0.0706578865647316,
      "learning_rate": 4.344081748509793e-05,
      "loss": 0.0010051688179373742,
      "step": 199340
    },
    {
      "epoch": 56.58529662219699,
      "grad_norm": 0.01734188012778759,
      "learning_rate": 4.343797899517457e-05,
      "loss": 0.00011336542665958404,
      "step": 199350
    },
    {
      "epoch": 56.588135112120355,
      "grad_norm": 0.015083505772054195,
      "learning_rate": 4.343514050525121e-05,
      "loss": 0.0003255097195506096,
      "step": 199360
    },
    {
      "epoch": 56.59097360204371,
      "grad_norm": 2.137889862060547,
      "learning_rate": 4.343230201532785e-05,
      "loss": 0.0004946982488036155,
      "step": 199370
    },
    {
      "epoch": 56.593812091967074,
      "grad_norm": 0.016169564798474312,
      "learning_rate": 4.342946352540448e-05,
      "loss": 0.000572783686220646,
      "step": 199380
    },
    {
      "epoch": 56.59665058189044,
      "grad_norm": 0.5501602292060852,
      "learning_rate": 4.342662503548113e-05,
      "loss": 0.00048679988831281664,
      "step": 199390
    },
    {
      "epoch": 56.59948907181379,
      "grad_norm": 0.14745880663394928,
      "learning_rate": 4.3423786545557764e-05,
      "loss": 0.0005479119718074799,
      "step": 199400
    },
    {
      "epoch": 56.602327561737155,
      "grad_norm": 0.7646116614341736,
      "learning_rate": 4.3420948055634406e-05,
      "loss": 0.0012604082003235817,
      "step": 199410
    },
    {
      "epoch": 56.60516605166052,
      "grad_norm": 0.004354759585112333,
      "learning_rate": 4.341810956571105e-05,
      "loss": 0.0003527868539094925,
      "step": 199420
    },
    {
      "epoch": 56.608004541583874,
      "grad_norm": 0.20170891284942627,
      "learning_rate": 4.341527107578768e-05,
      "loss": 0.0021626021713018416,
      "step": 199430
    },
    {
      "epoch": 56.61084303150724,
      "grad_norm": 0.026509013026952744,
      "learning_rate": 4.341243258586432e-05,
      "loss": 0.002615862712264061,
      "step": 199440
    },
    {
      "epoch": 56.6136815214306,
      "grad_norm": 0.02309305965900421,
      "learning_rate": 4.3409594095940964e-05,
      "loss": 0.0012685371562838554,
      "step": 199450
    },
    {
      "epoch": 56.61652001135396,
      "grad_norm": 0.023291172459721565,
      "learning_rate": 4.34067556060176e-05,
      "loss": 0.00025542806833982467,
      "step": 199460
    },
    {
      "epoch": 56.61935850127732,
      "grad_norm": 0.13455131649971008,
      "learning_rate": 4.340391711609424e-05,
      "loss": 0.0027252862229943274,
      "step": 199470
    },
    {
      "epoch": 56.62219699120068,
      "grad_norm": 0.09099557995796204,
      "learning_rate": 4.3401078626170875e-05,
      "loss": 0.0004426902160048485,
      "step": 199480
    },
    {
      "epoch": 56.625035481124044,
      "grad_norm": 0.04469424486160278,
      "learning_rate": 4.339824013624752e-05,
      "loss": 0.0003223340958356857,
      "step": 199490
    },
    {
      "epoch": 56.6278739710474,
      "grad_norm": 0.12165403366088867,
      "learning_rate": 4.339540164632416e-05,
      "loss": 0.0011931302025914193,
      "step": 199500
    },
    {
      "epoch": 56.6278739710474,
      "eval_accuracy": 0.9774909391492338,
      "eval_loss": 0.07720094174146652,
      "eval_runtime": 32.4595,
      "eval_samples_per_second": 484.512,
      "eval_steps_per_second": 7.579,
      "step": 199500
    },
    {
      "epoch": 56.63071246097076,
      "grad_norm": 0.16904130578041077,
      "learning_rate": 4.339256315640079e-05,
      "loss": 0.0004886258393526077,
      "step": 199510
    },
    {
      "epoch": 56.633550950894126,
      "grad_norm": 0.231682687997818,
      "learning_rate": 4.338972466647744e-05,
      "loss": 0.0004755785688757896,
      "step": 199520
    },
    {
      "epoch": 56.63638944081748,
      "grad_norm": 0.034110113978385925,
      "learning_rate": 4.3386886176554075e-05,
      "loss": 0.0012620704248547554,
      "step": 199530
    },
    {
      "epoch": 56.639227930740844,
      "grad_norm": 0.01764194667339325,
      "learning_rate": 4.3384047686630716e-05,
      "loss": 0.0003511270508170128,
      "step": 199540
    },
    {
      "epoch": 56.64206642066421,
      "grad_norm": 2.2723305225372314,
      "learning_rate": 4.338120919670736e-05,
      "loss": 0.0005339279770851136,
      "step": 199550
    },
    {
      "epoch": 56.64490491058757,
      "grad_norm": 0.06426084041595459,
      "learning_rate": 4.337837070678399e-05,
      "loss": 0.0001961858943104744,
      "step": 199560
    },
    {
      "epoch": 56.647743400510926,
      "grad_norm": 1.8229585886001587,
      "learning_rate": 4.3375532216860634e-05,
      "loss": 0.000505877286195755,
      "step": 199570
    },
    {
      "epoch": 56.65058189043429,
      "grad_norm": 0.05769103020429611,
      "learning_rate": 4.337269372693727e-05,
      "loss": 0.00017693303525447845,
      "step": 199580
    },
    {
      "epoch": 56.65342038035765,
      "grad_norm": 0.046078965067863464,
      "learning_rate": 4.336985523701391e-05,
      "loss": 0.00042354799807071687,
      "step": 199590
    },
    {
      "epoch": 56.65625887028101,
      "grad_norm": 0.11833233386278152,
      "learning_rate": 4.336701674709055e-05,
      "loss": 0.0015283428132534028,
      "step": 199600
    },
    {
      "epoch": 56.65909736020437,
      "grad_norm": 0.11072896420955658,
      "learning_rate": 4.3364178257167186e-05,
      "loss": 0.02097795307636261,
      "step": 199610
    },
    {
      "epoch": 56.66193585012773,
      "grad_norm": 0.570492684841156,
      "learning_rate": 4.336133976724383e-05,
      "loss": 0.00035146847367286684,
      "step": 199620
    },
    {
      "epoch": 56.664774340051096,
      "grad_norm": 0.5610129237174988,
      "learning_rate": 4.335850127732047e-05,
      "loss": 0.001219264790415764,
      "step": 199630
    },
    {
      "epoch": 56.66761282997445,
      "grad_norm": 0.1917383074760437,
      "learning_rate": 4.33556627873971e-05,
      "loss": 0.00019885897636413575,
      "step": 199640
    },
    {
      "epoch": 56.670451319897815,
      "grad_norm": 1.015963077545166,
      "learning_rate": 4.335282429747375e-05,
      "loss": 0.0007009373977780343,
      "step": 199650
    },
    {
      "epoch": 56.67328980982118,
      "grad_norm": 0.11489671468734741,
      "learning_rate": 4.3349985807550386e-05,
      "loss": 0.006462021172046662,
      "step": 199660
    },
    {
      "epoch": 56.67612829974453,
      "grad_norm": 0.9260806441307068,
      "learning_rate": 4.334714731762702e-05,
      "loss": 0.0013763448223471642,
      "step": 199670
    },
    {
      "epoch": 56.678966789667896,
      "grad_norm": 0.09579630941152573,
      "learning_rate": 4.334430882770366e-05,
      "loss": 0.000954856164753437,
      "step": 199680
    },
    {
      "epoch": 56.68180527959126,
      "grad_norm": 1.9144452810287476,
      "learning_rate": 4.33414703377803e-05,
      "loss": 0.0013401705771684646,
      "step": 199690
    },
    {
      "epoch": 56.684643769514615,
      "grad_norm": 0.011504733003675938,
      "learning_rate": 4.3338631847856945e-05,
      "loss": 0.000724194198846817,
      "step": 199700
    },
    {
      "epoch": 56.68748225943798,
      "grad_norm": 0.017461175099015236,
      "learning_rate": 4.333579335793358e-05,
      "loss": 0.0017755566164851188,
      "step": 199710
    },
    {
      "epoch": 56.69032074936134,
      "grad_norm": 0.056614961475133896,
      "learning_rate": 4.333295486801022e-05,
      "loss": 0.0012316390872001647,
      "step": 199720
    },
    {
      "epoch": 56.693159239284704,
      "grad_norm": 0.33073684573173523,
      "learning_rate": 4.333011637808686e-05,
      "loss": 0.001095731183886528,
      "step": 199730
    },
    {
      "epoch": 56.69599772920806,
      "grad_norm": 0.9772460460662842,
      "learning_rate": 4.3327277888163496e-05,
      "loss": 0.00040018819272518157,
      "step": 199740
    },
    {
      "epoch": 56.69883621913142,
      "grad_norm": 0.013589735142886639,
      "learning_rate": 4.332443939824014e-05,
      "loss": 0.0023385629057884215,
      "step": 199750
    },
    {
      "epoch": 56.701674709054785,
      "grad_norm": 0.10904175788164139,
      "learning_rate": 4.332160090831678e-05,
      "loss": 0.0005422079935669899,
      "step": 199760
    },
    {
      "epoch": 56.70451319897814,
      "grad_norm": 0.739193856716156,
      "learning_rate": 4.3318762418393414e-05,
      "loss": 0.0004178471863269806,
      "step": 199770
    },
    {
      "epoch": 56.707351688901504,
      "grad_norm": 0.10070403665304184,
      "learning_rate": 4.3315923928470055e-05,
      "loss": 0.007056562602519989,
      "step": 199780
    },
    {
      "epoch": 56.71019017882487,
      "grad_norm": 0.07318787276744843,
      "learning_rate": 4.3313085438546697e-05,
      "loss": 0.0062316469848155975,
      "step": 199790
    },
    {
      "epoch": 56.71302866874822,
      "grad_norm": 2.2195827960968018,
      "learning_rate": 4.331024694862333e-05,
      "loss": 0.00043890681117773054,
      "step": 199800
    },
    {
      "epoch": 56.715867158671585,
      "grad_norm": 0.6832600235939026,
      "learning_rate": 4.330740845869997e-05,
      "loss": 0.0004683392122387886,
      "step": 199810
    },
    {
      "epoch": 56.71870564859495,
      "grad_norm": 0.07580214738845825,
      "learning_rate": 4.3304569968776614e-05,
      "loss": 0.0014487143605947494,
      "step": 199820
    },
    {
      "epoch": 56.72154413851831,
      "grad_norm": 7.889248371124268,
      "learning_rate": 4.3301731478853255e-05,
      "loss": 0.0019018324092030525,
      "step": 199830
    },
    {
      "epoch": 56.72438262844167,
      "grad_norm": 0.2305252104997635,
      "learning_rate": 4.329889298892989e-05,
      "loss": 0.0014726314693689346,
      "step": 199840
    },
    {
      "epoch": 56.72722111836503,
      "grad_norm": 9.918656349182129,
      "learning_rate": 4.329605449900653e-05,
      "loss": 0.0029454467818140983,
      "step": 199850
    },
    {
      "epoch": 56.73005960828839,
      "grad_norm": 0.42369577288627625,
      "learning_rate": 4.329321600908317e-05,
      "loss": 0.005766977369785309,
      "step": 199860
    },
    {
      "epoch": 56.73289809821175,
      "grad_norm": 15.997648239135742,
      "learning_rate": 4.329037751915981e-05,
      "loss": 0.016415253281593323,
      "step": 199870
    },
    {
      "epoch": 56.73573658813511,
      "grad_norm": 0.0924142524600029,
      "learning_rate": 4.328753902923645e-05,
      "loss": 0.004418928176164627,
      "step": 199880
    },
    {
      "epoch": 56.738575078058474,
      "grad_norm": 5.531737327575684,
      "learning_rate": 4.328470053931309e-05,
      "loss": 0.0034918226301670074,
      "step": 199890
    },
    {
      "epoch": 56.74141356798184,
      "grad_norm": 0.2198348045349121,
      "learning_rate": 4.3281862049389725e-05,
      "loss": 0.014966781437397002,
      "step": 199900
    },
    {
      "epoch": 56.74425205790519,
      "grad_norm": 0.9527679085731506,
      "learning_rate": 4.3279023559466366e-05,
      "loss": 0.0010680094361305236,
      "step": 199910
    },
    {
      "epoch": 56.747090547828556,
      "grad_norm": 0.026231449097394943,
      "learning_rate": 4.327618506954301e-05,
      "loss": 0.005332134664058685,
      "step": 199920
    },
    {
      "epoch": 56.74992903775192,
      "grad_norm": 0.017704149708151817,
      "learning_rate": 4.327334657961964e-05,
      "loss": 0.004073721542954445,
      "step": 199930
    },
    {
      "epoch": 56.752767527675275,
      "grad_norm": 0.2289051115512848,
      "learning_rate": 4.327050808969628e-05,
      "loss": 0.0037243306636810304,
      "step": 199940
    },
    {
      "epoch": 56.75560601759864,
      "grad_norm": 0.07778113335371017,
      "learning_rate": 4.3267669599772925e-05,
      "loss": 0.007437411695718765,
      "step": 199950
    },
    {
      "epoch": 56.758444507522,
      "grad_norm": 0.018980856984853745,
      "learning_rate": 4.3264831109849566e-05,
      "loss": 0.00034308005124330523,
      "step": 199960
    },
    {
      "epoch": 56.761282997445356,
      "grad_norm": 3.053487777709961,
      "learning_rate": 4.32619926199262e-05,
      "loss": 0.0012752771377563476,
      "step": 199970
    },
    {
      "epoch": 56.76412148736872,
      "grad_norm": 0.05718189477920532,
      "learning_rate": 4.3259154130002835e-05,
      "loss": 0.001780446618795395,
      "step": 199980
    },
    {
      "epoch": 56.76695997729208,
      "grad_norm": 1.373928427696228,
      "learning_rate": 4.325631564007948e-05,
      "loss": 0.0015583224594593048,
      "step": 199990
    },
    {
      "epoch": 56.769798467215445,
      "grad_norm": 0.31937354803085327,
      "learning_rate": 4.325347715015612e-05,
      "loss": 0.00027578268200159075,
      "step": 200000
    },
    {
      "epoch": 56.769798467215445,
      "eval_accuracy": 0.9774273542315762,
      "eval_loss": 0.08635704219341278,
      "eval_runtime": 32.9425,
      "eval_samples_per_second": 477.407,
      "eval_steps_per_second": 7.468,
      "step": 200000
    },
    {
      "epoch": 56.7726369571388,
      "grad_norm": 0.052973486483097076,
      "learning_rate": 4.325063866023276e-05,
      "loss": 0.00818576142191887,
      "step": 200010
    },
    {
      "epoch": 56.77547544706216,
      "grad_norm": 0.23988540470600128,
      "learning_rate": 4.32478001703094e-05,
      "loss": 0.0007290612906217575,
      "step": 200020
    },
    {
      "epoch": 56.778313936985526,
      "grad_norm": 2.4585306644439697,
      "learning_rate": 4.3244961680386035e-05,
      "loss": 0.0039919808506965635,
      "step": 200030
    },
    {
      "epoch": 56.78115242690888,
      "grad_norm": 0.6307926774024963,
      "learning_rate": 4.324212319046268e-05,
      "loss": 0.0008143294602632522,
      "step": 200040
    },
    {
      "epoch": 56.783990916832245,
      "grad_norm": 0.22681139409542084,
      "learning_rate": 4.323928470053932e-05,
      "loss": 0.0012881381437182427,
      "step": 200050
    },
    {
      "epoch": 56.78682940675561,
      "grad_norm": 0.08073797076940536,
      "learning_rate": 4.323644621061595e-05,
      "loss": 0.004340262711048126,
      "step": 200060
    },
    {
      "epoch": 56.789667896678964,
      "grad_norm": 0.1734338104724884,
      "learning_rate": 4.3233607720692594e-05,
      "loss": 0.004114207997918129,
      "step": 200070
    },
    {
      "epoch": 56.79250638660233,
      "grad_norm": 0.016746170818805695,
      "learning_rate": 4.323076923076923e-05,
      "loss": 0.006683717668056488,
      "step": 200080
    },
    {
      "epoch": 56.79534487652569,
      "grad_norm": 0.603496789932251,
      "learning_rate": 4.322793074084587e-05,
      "loss": 0.003788972645998001,
      "step": 200090
    },
    {
      "epoch": 56.79818336644905,
      "grad_norm": 0.023494048044085503,
      "learning_rate": 4.322509225092251e-05,
      "loss": 0.0005455933511257172,
      "step": 200100
    },
    {
      "epoch": 56.80102185637241,
      "grad_norm": 0.38556423783302307,
      "learning_rate": 4.3222253760999146e-05,
      "loss": 0.000386604480445385,
      "step": 200110
    },
    {
      "epoch": 56.80386034629577,
      "grad_norm": 0.0393877774477005,
      "learning_rate": 4.3219415271075794e-05,
      "loss": 0.00021758172661066056,
      "step": 200120
    },
    {
      "epoch": 56.806698836219134,
      "grad_norm": 0.789720892906189,
      "learning_rate": 4.321657678115243e-05,
      "loss": 0.000470663420855999,
      "step": 200130
    },
    {
      "epoch": 56.80953732614249,
      "grad_norm": 0.389004647731781,
      "learning_rate": 4.321373829122906e-05,
      "loss": 0.005303002148866654,
      "step": 200140
    },
    {
      "epoch": 56.81237581606585,
      "grad_norm": 13.555556297302246,
      "learning_rate": 4.321089980130571e-05,
      "loss": 0.002979358844459057,
      "step": 200150
    },
    {
      "epoch": 56.815214305989215,
      "grad_norm": 0.19053852558135986,
      "learning_rate": 4.3208061311382346e-05,
      "loss": 0.0008209770545363426,
      "step": 200160
    },
    {
      "epoch": 56.81805279591257,
      "grad_norm": 0.27862349152565,
      "learning_rate": 4.320522282145899e-05,
      "loss": 0.0013585850596427918,
      "step": 200170
    },
    {
      "epoch": 56.820891285835934,
      "grad_norm": 0.11936115473508835,
      "learning_rate": 4.320238433153562e-05,
      "loss": 0.00029610954225063325,
      "step": 200180
    },
    {
      "epoch": 56.8237297757593,
      "grad_norm": 0.040944699198007584,
      "learning_rate": 4.3199545841612263e-05,
      "loss": 0.0006252303719520569,
      "step": 200190
    },
    {
      "epoch": 56.82656826568266,
      "grad_norm": 0.04554248973727226,
      "learning_rate": 4.3196707351688905e-05,
      "loss": 0.0005840733647346497,
      "step": 200200
    },
    {
      "epoch": 56.829406755606016,
      "grad_norm": 0.006429123226553202,
      "learning_rate": 4.319386886176554e-05,
      "loss": 0.0003351595252752304,
      "step": 200210
    },
    {
      "epoch": 56.83224524552938,
      "grad_norm": 0.3482067584991455,
      "learning_rate": 4.319103037184218e-05,
      "loss": 0.0005381667986512184,
      "step": 200220
    },
    {
      "epoch": 56.83508373545274,
      "grad_norm": 0.1282515525817871,
      "learning_rate": 4.318819188191882e-05,
      "loss": 0.0007099047303199768,
      "step": 200230
    },
    {
      "epoch": 56.8379222253761,
      "grad_norm": 0.06444639712572098,
      "learning_rate": 4.318535339199546e-05,
      "loss": 0.0006049908697605133,
      "step": 200240
    },
    {
      "epoch": 56.84076071529946,
      "grad_norm": 0.05016118288040161,
      "learning_rate": 4.3182514902072105e-05,
      "loss": 0.0002855706959962845,
      "step": 200250
    },
    {
      "epoch": 56.84359920522282,
      "grad_norm": 0.02094232477247715,
      "learning_rate": 4.317967641214874e-05,
      "loss": 0.0004952836781740189,
      "step": 200260
    },
    {
      "epoch": 56.84643769514618,
      "grad_norm": 0.14093153178691864,
      "learning_rate": 4.3176837922225374e-05,
      "loss": 0.0003348285332322121,
      "step": 200270
    },
    {
      "epoch": 56.84927618506954,
      "grad_norm": 0.05609801784157753,
      "learning_rate": 4.3173999432302015e-05,
      "loss": 0.0003192594274878502,
      "step": 200280
    },
    {
      "epoch": 56.852114674992904,
      "grad_norm": 0.18074922263622284,
      "learning_rate": 4.317116094237866e-05,
      "loss": 0.0002816483378410339,
      "step": 200290
    },
    {
      "epoch": 56.85495316491627,
      "grad_norm": 0.2897770404815674,
      "learning_rate": 4.31683224524553e-05,
      "loss": 0.00039952434599399567,
      "step": 200300
    },
    {
      "epoch": 56.85779165483962,
      "grad_norm": 0.7893989682197571,
      "learning_rate": 4.316548396253193e-05,
      "loss": 0.000849335826933384,
      "step": 200310
    },
    {
      "epoch": 56.860630144762986,
      "grad_norm": 0.15595951676368713,
      "learning_rate": 4.3162645472608574e-05,
      "loss": 0.001363135688006878,
      "step": 200320
    },
    {
      "epoch": 56.86346863468635,
      "grad_norm": 0.05356869474053383,
      "learning_rate": 4.3159806982685216e-05,
      "loss": 0.00030534714460372925,
      "step": 200330
    },
    {
      "epoch": 56.866307124609705,
      "grad_norm": 0.03867045044898987,
      "learning_rate": 4.315696849276185e-05,
      "loss": 0.00021122097969055175,
      "step": 200340
    },
    {
      "epoch": 56.86914561453307,
      "grad_norm": 1.7470285892486572,
      "learning_rate": 4.315413000283849e-05,
      "loss": 0.0006336702033877372,
      "step": 200350
    },
    {
      "epoch": 56.87198410445643,
      "grad_norm": 0.09334969520568848,
      "learning_rate": 4.315129151291513e-05,
      "loss": 0.002667706459760666,
      "step": 200360
    },
    {
      "epoch": 56.87482259437979,
      "grad_norm": 0.4083535373210907,
      "learning_rate": 4.314845302299177e-05,
      "loss": 0.0003425450995564461,
      "step": 200370
    },
    {
      "epoch": 56.87766108430315,
      "grad_norm": 0.01971421390771866,
      "learning_rate": 4.314561453306841e-05,
      "loss": 0.00040315445512533187,
      "step": 200380
    },
    {
      "epoch": 56.88049957422651,
      "grad_norm": 0.03132257238030434,
      "learning_rate": 4.314305989213738e-05,
      "loss": 0.002739547565579414,
      "step": 200390
    },
    {
      "epoch": 56.883338064149875,
      "grad_norm": 0.059551358222961426,
      "learning_rate": 4.3140221402214023e-05,
      "loss": 0.0033661499619483948,
      "step": 200400
    },
    {
      "epoch": 56.88617655407323,
      "grad_norm": 0.03028663620352745,
      "learning_rate": 4.3137382912290665e-05,
      "loss": 0.0003389192745089531,
      "step": 200410
    },
    {
      "epoch": 56.889015043996594,
      "grad_norm": 0.07753496617078781,
      "learning_rate": 4.31345444223673e-05,
      "loss": 0.004858072847127914,
      "step": 200420
    },
    {
      "epoch": 56.89185353391996,
      "grad_norm": 0.16182976961135864,
      "learning_rate": 4.313170593244395e-05,
      "loss": 0.0037418119609355927,
      "step": 200430
    },
    {
      "epoch": 56.89469202384331,
      "grad_norm": 3.6472718715667725,
      "learning_rate": 4.312886744252058e-05,
      "loss": 0.0014647191390395164,
      "step": 200440
    },
    {
      "epoch": 56.897530513766675,
      "grad_norm": 0.02206544578075409,
      "learning_rate": 4.312602895259722e-05,
      "loss": 0.0009286265820264816,
      "step": 200450
    },
    {
      "epoch": 56.90036900369004,
      "grad_norm": 0.11763380467891693,
      "learning_rate": 4.312319046267386e-05,
      "loss": 0.0002564597874879837,
      "step": 200460
    },
    {
      "epoch": 56.9032074936134,
      "grad_norm": 0.013742645271122456,
      "learning_rate": 4.31203519727505e-05,
      "loss": 0.0012595893815159798,
      "step": 200470
    },
    {
      "epoch": 56.90604598353676,
      "grad_norm": 0.06837963312864304,
      "learning_rate": 4.311751348282714e-05,
      "loss": 0.0005041493102908134,
      "step": 200480
    },
    {
      "epoch": 56.90888447346012,
      "grad_norm": 0.008963258937001228,
      "learning_rate": 4.3114674992903775e-05,
      "loss": 0.0005401270464062691,
      "step": 200490
    },
    {
      "epoch": 56.91172296338348,
      "grad_norm": 0.28664571046829224,
      "learning_rate": 4.311183650298042e-05,
      "loss": 0.000853918120265007,
      "step": 200500
    },
    {
      "epoch": 56.91172296338348,
      "eval_accuracy": 0.9788898073376995,
      "eval_loss": 0.07930737733840942,
      "eval_runtime": 32.9932,
      "eval_samples_per_second": 476.674,
      "eval_steps_per_second": 7.456,
      "step": 200500
    },
    {
      "epoch": 56.91456145330684,
      "grad_norm": 0.13170455396175385,
      "learning_rate": 4.310899801305706e-05,
      "loss": 0.001414143294095993,
      "step": 200510
    },
    {
      "epoch": 56.9173999432302,
      "grad_norm": 0.01512229349464178,
      "learning_rate": 4.310615952313369e-05,
      "loss": 0.00031428057700395585,
      "step": 200520
    },
    {
      "epoch": 56.920238433153564,
      "grad_norm": 0.028068216517567635,
      "learning_rate": 4.3103321033210334e-05,
      "loss": 0.0004225457087159157,
      "step": 200530
    },
    {
      "epoch": 56.92307692307692,
      "grad_norm": 0.2232406735420227,
      "learning_rate": 4.3100482543286976e-05,
      "loss": 0.004478540644049645,
      "step": 200540
    },
    {
      "epoch": 56.92591541300028,
      "grad_norm": 0.03122646175324917,
      "learning_rate": 4.309764405336361e-05,
      "loss": 0.005822714045643806,
      "step": 200550
    },
    {
      "epoch": 56.928753902923646,
      "grad_norm": 0.016302771866321564,
      "learning_rate": 4.309480556344025e-05,
      "loss": 0.002021363191306591,
      "step": 200560
    },
    {
      "epoch": 56.93159239284701,
      "grad_norm": 1.2137556076049805,
      "learning_rate": 4.309196707351689e-05,
      "loss": 0.0020448042079806326,
      "step": 200570
    },
    {
      "epoch": 56.934430882770364,
      "grad_norm": 1.5950778722763062,
      "learning_rate": 4.308912858359353e-05,
      "loss": 0.0023507248610258103,
      "step": 200580
    },
    {
      "epoch": 56.93726937269373,
      "grad_norm": 0.34476718306541443,
      "learning_rate": 4.308629009367017e-05,
      "loss": 0.001139584556221962,
      "step": 200590
    },
    {
      "epoch": 56.94010786261709,
      "grad_norm": 0.21791072189807892,
      "learning_rate": 4.308345160374681e-05,
      "loss": 0.003195559233427048,
      "step": 200600
    },
    {
      "epoch": 56.942946352540446,
      "grad_norm": 0.05086489021778107,
      "learning_rate": 4.3080613113823445e-05,
      "loss": 0.0014512546360492707,
      "step": 200610
    },
    {
      "epoch": 56.94578484246381,
      "grad_norm": 0.4523414373397827,
      "learning_rate": 4.3077774623900086e-05,
      "loss": 0.0006383338943123818,
      "step": 200620
    },
    {
      "epoch": 56.94862333238717,
      "grad_norm": 0.4168773293495178,
      "learning_rate": 4.307493613397673e-05,
      "loss": 0.0019120490178465843,
      "step": 200630
    },
    {
      "epoch": 56.95146182231053,
      "grad_norm": 0.05980618670582771,
      "learning_rate": 4.307209764405337e-05,
      "loss": 0.000257771834731102,
      "step": 200640
    },
    {
      "epoch": 56.95430031223389,
      "grad_norm": 0.038323335349559784,
      "learning_rate": 4.3069259154130004e-05,
      "loss": 0.0017015635967254639,
      "step": 200650
    },
    {
      "epoch": 56.95713880215725,
      "grad_norm": 4.837734699249268,
      "learning_rate": 4.306642066420664e-05,
      "loss": 0.001287921704351902,
      "step": 200660
    },
    {
      "epoch": 56.959977292080616,
      "grad_norm": 4.7121262550354,
      "learning_rate": 4.3063582174283286e-05,
      "loss": 0.0025889866054058077,
      "step": 200670
    },
    {
      "epoch": 56.96281578200397,
      "grad_norm": 0.3785896897315979,
      "learning_rate": 4.306074368435992e-05,
      "loss": 0.0009185686707496643,
      "step": 200680
    },
    {
      "epoch": 56.965654271927335,
      "grad_norm": 0.25577312707901,
      "learning_rate": 4.305790519443656e-05,
      "loss": 0.000325273722410202,
      "step": 200690
    },
    {
      "epoch": 56.9684927618507,
      "grad_norm": 0.7346470952033997,
      "learning_rate": 4.3055066704513204e-05,
      "loss": 0.0011619523167610169,
      "step": 200700
    },
    {
      "epoch": 56.97133125177405,
      "grad_norm": 0.16928669810295105,
      "learning_rate": 4.305222821458984e-05,
      "loss": 0.004123948514461517,
      "step": 200710
    },
    {
      "epoch": 56.974169741697416,
      "grad_norm": 0.2655104994773865,
      "learning_rate": 4.304938972466648e-05,
      "loss": 0.0011900918558239936,
      "step": 200720
    },
    {
      "epoch": 56.97700823162078,
      "grad_norm": 0.04529326409101486,
      "learning_rate": 4.304655123474312e-05,
      "loss": 0.00018385909497737886,
      "step": 200730
    },
    {
      "epoch": 56.97984672154414,
      "grad_norm": 8.621896743774414,
      "learning_rate": 4.3043712744819756e-05,
      "loss": 0.007953740656375885,
      "step": 200740
    },
    {
      "epoch": 56.9826852114675,
      "grad_norm": 0.07254883646965027,
      "learning_rate": 4.30408742548964e-05,
      "loss": 0.005233626812696457,
      "step": 200750
    },
    {
      "epoch": 56.98552370139086,
      "grad_norm": 0.03222733736038208,
      "learning_rate": 4.303803576497303e-05,
      "loss": 0.0007836554199457168,
      "step": 200760
    },
    {
      "epoch": 56.98836219131422,
      "grad_norm": 0.03387417644262314,
      "learning_rate": 4.303519727504968e-05,
      "loss": 0.003844743221998215,
      "step": 200770
    },
    {
      "epoch": 56.99120068123758,
      "grad_norm": 0.08525165170431137,
      "learning_rate": 4.3032358785126314e-05,
      "loss": 0.0006162222474813461,
      "step": 200780
    },
    {
      "epoch": 56.99403917116094,
      "grad_norm": 1.7434321641921997,
      "learning_rate": 4.302952029520295e-05,
      "loss": 0.00262595359236002,
      "step": 200790
    },
    {
      "epoch": 56.996877661084305,
      "grad_norm": 0.011591685935854912,
      "learning_rate": 4.30266818052796e-05,
      "loss": 0.007482858002185821,
      "step": 200800
    },
    {
      "epoch": 56.99971615100766,
      "grad_norm": 3.297243118286133,
      "learning_rate": 4.302384331535623e-05,
      "loss": 0.007903494685888291,
      "step": 200810
    },
    {
      "epoch": 57.002554640931024,
      "grad_norm": 0.40686121582984924,
      "learning_rate": 4.302100482543287e-05,
      "loss": 0.0005969660356640816,
      "step": 200820
    },
    {
      "epoch": 57.00539313085439,
      "grad_norm": 0.05055893957614899,
      "learning_rate": 4.3018166335509514e-05,
      "loss": 0.0006559722125530243,
      "step": 200830
    },
    {
      "epoch": 57.00823162077775,
      "grad_norm": 0.0697258785367012,
      "learning_rate": 4.301532784558615e-05,
      "loss": 0.0008485350757837295,
      "step": 200840
    },
    {
      "epoch": 57.011070110701105,
      "grad_norm": 0.021078651770949364,
      "learning_rate": 4.301248935566279e-05,
      "loss": 0.0029240841045975684,
      "step": 200850
    },
    {
      "epoch": 57.01390860062447,
      "grad_norm": 0.13679464161396027,
      "learning_rate": 4.3009650865739425e-05,
      "loss": 0.00034329015761613846,
      "step": 200860
    },
    {
      "epoch": 57.01674709054783,
      "grad_norm": 0.03644420579075813,
      "learning_rate": 4.3006812375816066e-05,
      "loss": 0.00019503328949213027,
      "step": 200870
    },
    {
      "epoch": 57.01958558047119,
      "grad_norm": 0.5464492440223694,
      "learning_rate": 4.300397388589271e-05,
      "loss": 0.0007966605946421624,
      "step": 200880
    },
    {
      "epoch": 57.02242407039455,
      "grad_norm": 0.11623863130807877,
      "learning_rate": 4.300113539596934e-05,
      "loss": 0.0004833795130252838,
      "step": 200890
    },
    {
      "epoch": 57.02526256031791,
      "grad_norm": 0.15781408548355103,
      "learning_rate": 4.299829690604599e-05,
      "loss": 0.00019017811864614488,
      "step": 200900
    },
    {
      "epoch": 57.02810105024127,
      "grad_norm": 0.028663834556937218,
      "learning_rate": 4.2995458416122625e-05,
      "loss": 0.0007589267566800118,
      "step": 200910
    },
    {
      "epoch": 57.03093954016463,
      "grad_norm": 0.04607165977358818,
      "learning_rate": 4.299261992619926e-05,
      "loss": 0.0003463162109255791,
      "step": 200920
    },
    {
      "epoch": 57.033778030087994,
      "grad_norm": 0.010459305718541145,
      "learning_rate": 4.298978143627591e-05,
      "loss": 0.00038949735462665556,
      "step": 200930
    },
    {
      "epoch": 57.03661652001136,
      "grad_norm": 0.018429765477776527,
      "learning_rate": 4.298694294635254e-05,
      "loss": 0.002263453043997288,
      "step": 200940
    },
    {
      "epoch": 57.03945500993471,
      "grad_norm": 0.02073507197201252,
      "learning_rate": 4.2984104456429184e-05,
      "loss": 0.0035685524344444275,
      "step": 200950
    },
    {
      "epoch": 57.042293499858076,
      "grad_norm": 0.11708080023527145,
      "learning_rate": 4.2981265966505825e-05,
      "loss": 0.00032347720116376877,
      "step": 200960
    },
    {
      "epoch": 57.04513198978144,
      "grad_norm": 0.4139687418937683,
      "learning_rate": 4.297842747658246e-05,
      "loss": 0.0005678731948137283,
      "step": 200970
    },
    {
      "epoch": 57.047970479704794,
      "grad_norm": 0.2454933226108551,
      "learning_rate": 4.29755889866591e-05,
      "loss": 0.00016815606504678727,
      "step": 200980
    },
    {
      "epoch": 57.05080896962816,
      "grad_norm": 0.03844744712114334,
      "learning_rate": 4.2972750496735736e-05,
      "loss": 0.0005626164376735688,
      "step": 200990
    },
    {
      "epoch": 57.05364745955152,
      "grad_norm": 0.05151917785406113,
      "learning_rate": 4.296991200681238e-05,
      "loss": 0.00024006012827157975,
      "step": 201000
    },
    {
      "epoch": 57.05364745955152,
      "eval_accuracy": 0.9773001843962612,
      "eval_loss": 0.08056609332561493,
      "eval_runtime": 32.6569,
      "eval_samples_per_second": 481.583,
      "eval_steps_per_second": 7.533,
      "step": 201000
    },
    {
      "epoch": 57.056485949474876,
      "grad_norm": 0.035329461097717285,
      "learning_rate": 4.296707351688902e-05,
      "loss": 0.0002843949943780899,
      "step": 201010
    },
    {
      "epoch": 57.05932443939824,
      "grad_norm": 0.07585558295249939,
      "learning_rate": 4.296423502696565e-05,
      "loss": 0.0009139340370893478,
      "step": 201020
    },
    {
      "epoch": 57.0621629293216,
      "grad_norm": 0.02679772675037384,
      "learning_rate": 4.2961396537042295e-05,
      "loss": 0.0007927954196929931,
      "step": 201030
    },
    {
      "epoch": 57.065001419244965,
      "grad_norm": 0.5146026015281677,
      "learning_rate": 4.2958558047118936e-05,
      "loss": 0.0006412042304873467,
      "step": 201040
    },
    {
      "epoch": 57.06783990916832,
      "grad_norm": 0.010220753960311413,
      "learning_rate": 4.295571955719557e-05,
      "loss": 0.00019828248769044877,
      "step": 201050
    },
    {
      "epoch": 57.07067839909168,
      "grad_norm": 0.10936890542507172,
      "learning_rate": 4.295288106727222e-05,
      "loss": 0.001105501502752304,
      "step": 201060
    },
    {
      "epoch": 57.073516889015046,
      "grad_norm": 0.4911111891269684,
      "learning_rate": 4.295004257734885e-05,
      "loss": 0.0003148248419165611,
      "step": 201070
    },
    {
      "epoch": 57.0763553789384,
      "grad_norm": 0.07574953138828278,
      "learning_rate": 4.294720408742549e-05,
      "loss": 0.0006164440885186196,
      "step": 201080
    },
    {
      "epoch": 57.079193868861765,
      "grad_norm": 0.03353596851229668,
      "learning_rate": 4.294436559750213e-05,
      "loss": 0.0006126653403043747,
      "step": 201090
    },
    {
      "epoch": 57.08203235878513,
      "grad_norm": 0.1112908199429512,
      "learning_rate": 4.294152710757877e-05,
      "loss": 0.0003598678857088089,
      "step": 201100
    },
    {
      "epoch": 57.08487084870849,
      "grad_norm": 0.2912595868110657,
      "learning_rate": 4.293868861765541e-05,
      "loss": 0.0004725232720375061,
      "step": 201110
    },
    {
      "epoch": 57.087709338631846,
      "grad_norm": 0.04724462702870369,
      "learning_rate": 4.2935850127732047e-05,
      "loss": 0.0014183890074491502,
      "step": 201120
    },
    {
      "epoch": 57.09054782855521,
      "grad_norm": 0.4213326573371887,
      "learning_rate": 4.293301163780869e-05,
      "loss": 0.000235789455473423,
      "step": 201130
    },
    {
      "epoch": 57.09338631847857,
      "grad_norm": 0.12265964597463608,
      "learning_rate": 4.293017314788533e-05,
      "loss": 0.00023546479642391204,
      "step": 201140
    },
    {
      "epoch": 57.09622480840193,
      "grad_norm": 0.23238897323608398,
      "learning_rate": 4.2927334657961964e-05,
      "loss": 0.00024346671998500823,
      "step": 201150
    },
    {
      "epoch": 57.09906329832529,
      "grad_norm": 0.0033123402390629053,
      "learning_rate": 4.2924496168038605e-05,
      "loss": 0.00014940686523914337,
      "step": 201160
    },
    {
      "epoch": 57.101901788248654,
      "grad_norm": 0.024003108963370323,
      "learning_rate": 4.292165767811525e-05,
      "loss": 0.0008556239306926727,
      "step": 201170
    },
    {
      "epoch": 57.10474027817201,
      "grad_norm": 0.08149160444736481,
      "learning_rate": 4.291881918819188e-05,
      "loss": 0.0004599355161190033,
      "step": 201180
    },
    {
      "epoch": 57.10757876809537,
      "grad_norm": 0.28362658619880676,
      "learning_rate": 4.291598069826852e-05,
      "loss": 0.0013384733349084855,
      "step": 201190
    },
    {
      "epoch": 57.110417258018735,
      "grad_norm": 0.754981517791748,
      "learning_rate": 4.2913142208345164e-05,
      "loss": 0.005869158729910851,
      "step": 201200
    },
    {
      "epoch": 57.1132557479421,
      "grad_norm": 0.16299238801002502,
      "learning_rate": 4.29103037184218e-05,
      "loss": 0.0004938431084156036,
      "step": 201210
    },
    {
      "epoch": 57.116094237865454,
      "grad_norm": 0.14035063982009888,
      "learning_rate": 4.290746522849844e-05,
      "loss": 0.000969882495701313,
      "step": 201220
    },
    {
      "epoch": 57.11893272778882,
      "grad_norm": 0.031112737953662872,
      "learning_rate": 4.290462673857508e-05,
      "loss": 0.0004536926746368408,
      "step": 201230
    },
    {
      "epoch": 57.12177121771218,
      "grad_norm": 0.19260765612125397,
      "learning_rate": 4.290178824865172e-05,
      "loss": 0.0005556778982281685,
      "step": 201240
    },
    {
      "epoch": 57.124609707635535,
      "grad_norm": 0.040867771953344345,
      "learning_rate": 4.289894975872836e-05,
      "loss": 0.00042935218662023546,
      "step": 201250
    },
    {
      "epoch": 57.1274481975589,
      "grad_norm": 0.0331200510263443,
      "learning_rate": 4.2896111268805e-05,
      "loss": 0.0009248403832316399,
      "step": 201260
    },
    {
      "epoch": 57.13028668748226,
      "grad_norm": 0.014002885669469833,
      "learning_rate": 4.289327277888164e-05,
      "loss": 0.0006891258060932159,
      "step": 201270
    },
    {
      "epoch": 57.13312517740562,
      "grad_norm": 0.021625259891152382,
      "learning_rate": 4.2890434288958275e-05,
      "loss": 0.0003234881907701492,
      "step": 201280
    },
    {
      "epoch": 57.13596366732898,
      "grad_norm": 0.19820933043956757,
      "learning_rate": 4.2887595799034916e-05,
      "loss": 0.0005907393991947174,
      "step": 201290
    },
    {
      "epoch": 57.13880215725234,
      "grad_norm": 0.3163262605667114,
      "learning_rate": 4.288475730911156e-05,
      "loss": 0.02052401453256607,
      "step": 201300
    },
    {
      "epoch": 57.141640647175706,
      "grad_norm": 0.9325040578842163,
      "learning_rate": 4.288191881918819e-05,
      "loss": 0.0026011385023593904,
      "step": 201310
    },
    {
      "epoch": 57.14447913709906,
      "grad_norm": 0.01992279477417469,
      "learning_rate": 4.2879080329264833e-05,
      "loss": 0.00027755871415138247,
      "step": 201320
    },
    {
      "epoch": 57.147317627022424,
      "grad_norm": 0.010596634820103645,
      "learning_rate": 4.2876241839341475e-05,
      "loss": 0.0016797471791505814,
      "step": 201330
    },
    {
      "epoch": 57.15015611694579,
      "grad_norm": 0.06688288599252701,
      "learning_rate": 4.287340334941811e-05,
      "loss": 0.0006846962496638298,
      "step": 201340
    },
    {
      "epoch": 57.15299460686914,
      "grad_norm": 0.7527656555175781,
      "learning_rate": 4.287056485949475e-05,
      "loss": 0.0031741395592689513,
      "step": 201350
    },
    {
      "epoch": 57.155833096792506,
      "grad_norm": 0.0603465735912323,
      "learning_rate": 4.286772636957139e-05,
      "loss": 0.00030042510479688644,
      "step": 201360
    },
    {
      "epoch": 57.15867158671587,
      "grad_norm": 0.019975975155830383,
      "learning_rate": 4.2864887879648034e-05,
      "loss": 0.005129379779100418,
      "step": 201370
    },
    {
      "epoch": 57.161510076639225,
      "grad_norm": 0.10040336847305298,
      "learning_rate": 4.286204938972467e-05,
      "loss": 0.0016369117423892022,
      "step": 201380
    },
    {
      "epoch": 57.16434856656259,
      "grad_norm": 7.092808246612549,
      "learning_rate": 4.28592108998013e-05,
      "loss": 0.0024403642863035204,
      "step": 201390
    },
    {
      "epoch": 57.16718705648595,
      "grad_norm": 0.021059805527329445,
      "learning_rate": 4.285637240987795e-05,
      "loss": 0.0003399364650249481,
      "step": 201400
    },
    {
      "epoch": 57.17002554640931,
      "grad_norm": 0.27285119891166687,
      "learning_rate": 4.2853533919954585e-05,
      "loss": 0.001135636866092682,
      "step": 201410
    },
    {
      "epoch": 57.17286403633267,
      "grad_norm": 0.21088802814483643,
      "learning_rate": 4.285069543003123e-05,
      "loss": 0.0054770026355981825,
      "step": 201420
    },
    {
      "epoch": 57.17570252625603,
      "grad_norm": 1.2385860681533813,
      "learning_rate": 4.284785694010787e-05,
      "loss": 0.005344578251242638,
      "step": 201430
    },
    {
      "epoch": 57.178541016179395,
      "grad_norm": 0.08587493002414703,
      "learning_rate": 4.28450184501845e-05,
      "loss": 0.002549692802131176,
      "step": 201440
    },
    {
      "epoch": 57.18137950610275,
      "grad_norm": 0.008264529518783092,
      "learning_rate": 4.2842179960261144e-05,
      "loss": 0.0005351455882191658,
      "step": 201450
    },
    {
      "epoch": 57.18421799602611,
      "grad_norm": 1.494195580482483,
      "learning_rate": 4.2839341470337786e-05,
      "loss": 0.0011081697419285774,
      "step": 201460
    },
    {
      "epoch": 57.187056485949476,
      "grad_norm": 0.017095206305384636,
      "learning_rate": 4.283650298041442e-05,
      "loss": 0.00044147055596113207,
      "step": 201470
    },
    {
      "epoch": 57.18989497587283,
      "grad_norm": 0.09560857713222504,
      "learning_rate": 4.283366449049106e-05,
      "loss": 0.003907162696123123,
      "step": 201480
    },
    {
      "epoch": 57.192733465796195,
      "grad_norm": 1.5653682947158813,
      "learning_rate": 4.2830826000567696e-05,
      "loss": 0.0010096924379467964,
      "step": 201490
    },
    {
      "epoch": 57.19557195571956,
      "grad_norm": 0.8284834027290344,
      "learning_rate": 4.282798751064434e-05,
      "loss": 0.00022936519235372543,
      "step": 201500
    },
    {
      "epoch": 57.19557195571956,
      "eval_accuracy": 0.9786990525847269,
      "eval_loss": 0.07789654284715652,
      "eval_runtime": 32.7108,
      "eval_samples_per_second": 480.789,
      "eval_steps_per_second": 7.52,
      "step": 201500
    },
    {
      "epoch": 57.19841044564292,
      "grad_norm": 1.8554421663284302,
      "learning_rate": 4.282514902072098e-05,
      "loss": 0.0018603634089231491,
      "step": 201510
    },
    {
      "epoch": 57.20124893556628,
      "grad_norm": 0.04776168614625931,
      "learning_rate": 4.2822310530797613e-05,
      "loss": 0.00028951317071914675,
      "step": 201520
    },
    {
      "epoch": 57.20408742548964,
      "grad_norm": 1.7256097793579102,
      "learning_rate": 4.281947204087426e-05,
      "loss": 0.001122703403234482,
      "step": 201530
    },
    {
      "epoch": 57.206925915413,
      "grad_norm": 0.0360376238822937,
      "learning_rate": 4.2816633550950896e-05,
      "loss": 0.00030745286494493484,
      "step": 201540
    },
    {
      "epoch": 57.20976440533636,
      "grad_norm": 0.0067825173027813435,
      "learning_rate": 4.281379506102753e-05,
      "loss": 0.0004044413566589355,
      "step": 201550
    },
    {
      "epoch": 57.21260289525972,
      "grad_norm": 0.20355620980262756,
      "learning_rate": 4.281095657110418e-05,
      "loss": 0.004194860160350799,
      "step": 201560
    },
    {
      "epoch": 57.215441385183084,
      "grad_norm": 0.48657962679862976,
      "learning_rate": 4.2808118081180814e-05,
      "loss": 0.0028144998475909235,
      "step": 201570
    },
    {
      "epoch": 57.21827987510645,
      "grad_norm": 2.9200165271759033,
      "learning_rate": 4.2805279591257455e-05,
      "loss": 0.004096401855349541,
      "step": 201580
    },
    {
      "epoch": 57.2211183650298,
      "grad_norm": 0.883172333240509,
      "learning_rate": 4.280244110133409e-05,
      "loss": 0.0019150709733366967,
      "step": 201590
    },
    {
      "epoch": 57.223956854953165,
      "grad_norm": 0.06598073989152908,
      "learning_rate": 4.279960261141073e-05,
      "loss": 0.0006409918889403344,
      "step": 201600
    },
    {
      "epoch": 57.22679534487653,
      "grad_norm": 0.11473830789327621,
      "learning_rate": 4.279676412148737e-05,
      "loss": 0.0017858995124697685,
      "step": 201610
    },
    {
      "epoch": 57.229633834799884,
      "grad_norm": 0.18643957376480103,
      "learning_rate": 4.279392563156401e-05,
      "loss": 0.00033036954700946806,
      "step": 201620
    },
    {
      "epoch": 57.23247232472325,
      "grad_norm": 0.09944051504135132,
      "learning_rate": 4.279108714164065e-05,
      "loss": 0.002861417457461357,
      "step": 201630
    },
    {
      "epoch": 57.23531081464661,
      "grad_norm": 0.5344258546829224,
      "learning_rate": 4.278824865171729e-05,
      "loss": 0.004497134312987328,
      "step": 201640
    },
    {
      "epoch": 57.238149304569966,
      "grad_norm": 0.03843424841761589,
      "learning_rate": 4.2785410161793924e-05,
      "loss": 0.002250286191701889,
      "step": 201650
    },
    {
      "epoch": 57.24098779449333,
      "grad_norm": 0.04939184710383415,
      "learning_rate": 4.278257167187057e-05,
      "loss": 0.0022885726764798166,
      "step": 201660
    },
    {
      "epoch": 57.24382628441669,
      "grad_norm": 1.6738266944885254,
      "learning_rate": 4.277973318194721e-05,
      "loss": 0.0015710202977061271,
      "step": 201670
    },
    {
      "epoch": 57.246664774340054,
      "grad_norm": 0.3871611952781677,
      "learning_rate": 4.277689469202384e-05,
      "loss": 0.005441091954708099,
      "step": 201680
    },
    {
      "epoch": 57.24950326426341,
      "grad_norm": 0.45166873931884766,
      "learning_rate": 4.277405620210048e-05,
      "loss": 0.0013358691707253457,
      "step": 201690
    },
    {
      "epoch": 57.25234175418677,
      "grad_norm": 0.36165645718574524,
      "learning_rate": 4.2771217712177124e-05,
      "loss": 0.0012647081166505814,
      "step": 201700
    },
    {
      "epoch": 57.255180244110136,
      "grad_norm": 12.646797180175781,
      "learning_rate": 4.2768379222253766e-05,
      "loss": 0.004315158724784851,
      "step": 201710
    },
    {
      "epoch": 57.25801873403349,
      "grad_norm": 0.5485801100730896,
      "learning_rate": 4.27655407323304e-05,
      "loss": 0.0031424075365066527,
      "step": 201720
    },
    {
      "epoch": 57.260857223956855,
      "grad_norm": 0.07665881514549255,
      "learning_rate": 4.276270224240704e-05,
      "loss": 0.0008447878062725067,
      "step": 201730
    },
    {
      "epoch": 57.26369571388022,
      "grad_norm": 0.026540782302618027,
      "learning_rate": 4.275986375248368e-05,
      "loss": 0.004280924052000046,
      "step": 201740
    },
    {
      "epoch": 57.26653420380357,
      "grad_norm": 0.06902874261140823,
      "learning_rate": 4.275702526256032e-05,
      "loss": 0.0004971720278263092,
      "step": 201750
    },
    {
      "epoch": 57.269372693726936,
      "grad_norm": 0.19614951312541962,
      "learning_rate": 4.275418677263696e-05,
      "loss": 0.0007437849417328835,
      "step": 201760
    },
    {
      "epoch": 57.2722111836503,
      "grad_norm": 0.017647884786128998,
      "learning_rate": 4.27513482827136e-05,
      "loss": 0.00023087374866008757,
      "step": 201770
    },
    {
      "epoch": 57.27504967357366,
      "grad_norm": 0.1867164522409439,
      "learning_rate": 4.2748509792790235e-05,
      "loss": 0.0006678322330117226,
      "step": 201780
    },
    {
      "epoch": 57.27788816349702,
      "grad_norm": 0.9830272793769836,
      "learning_rate": 4.2745671302866876e-05,
      "loss": 0.0005666259676218033,
      "step": 201790
    },
    {
      "epoch": 57.28072665342038,
      "grad_norm": 0.6811372637748718,
      "learning_rate": 4.274283281294352e-05,
      "loss": 0.0006288418546319008,
      "step": 201800
    },
    {
      "epoch": 57.28356514334374,
      "grad_norm": 0.03200864419341087,
      "learning_rate": 4.273999432302015e-05,
      "loss": 0.00039253849536180496,
      "step": 201810
    },
    {
      "epoch": 57.2864036332671,
      "grad_norm": 0.16454167664051056,
      "learning_rate": 4.2737155833096794e-05,
      "loss": 0.00019156821072101592,
      "step": 201820
    },
    {
      "epoch": 57.28924212319046,
      "grad_norm": 2.3114898204803467,
      "learning_rate": 4.2734317343173435e-05,
      "loss": 0.002325908653438091,
      "step": 201830
    },
    {
      "epoch": 57.292080613113825,
      "grad_norm": 3.676750659942627,
      "learning_rate": 4.2731478853250076e-05,
      "loss": 0.0007697217166423798,
      "step": 201840
    },
    {
      "epoch": 57.29491910303718,
      "grad_norm": 0.062299538403749466,
      "learning_rate": 4.272864036332671e-05,
      "loss": 0.0003022855147719383,
      "step": 201850
    },
    {
      "epoch": 57.297757592960544,
      "grad_norm": 5.26702356338501,
      "learning_rate": 4.272580187340335e-05,
      "loss": 0.0010319046676158905,
      "step": 201860
    },
    {
      "epoch": 57.30059608288391,
      "grad_norm": 0.010554893873631954,
      "learning_rate": 4.2722963383479994e-05,
      "loss": 0.0012429360300302505,
      "step": 201870
    },
    {
      "epoch": 57.30343457280727,
      "grad_norm": 0.15709945559501648,
      "learning_rate": 4.272012489355663e-05,
      "loss": 0.0036340445280075074,
      "step": 201880
    },
    {
      "epoch": 57.306273062730625,
      "grad_norm": 0.09297211468219757,
      "learning_rate": 4.271728640363327e-05,
      "loss": 0.0012716976925730705,
      "step": 201890
    },
    {
      "epoch": 57.30911155265399,
      "grad_norm": 0.2547995150089264,
      "learning_rate": 4.271444791370991e-05,
      "loss": 0.0037591468542814256,
      "step": 201900
    },
    {
      "epoch": 57.31195004257735,
      "grad_norm": 0.02291746996343136,
      "learning_rate": 4.2711609423786546e-05,
      "loss": 0.001446639932692051,
      "step": 201910
    },
    {
      "epoch": 57.31478853250071,
      "grad_norm": 0.1732814759016037,
      "learning_rate": 4.270877093386319e-05,
      "loss": 0.00046481769531965257,
      "step": 201920
    },
    {
      "epoch": 57.31762702242407,
      "grad_norm": 11.757360458374023,
      "learning_rate": 4.270593244393983e-05,
      "loss": 0.0018617138266563416,
      "step": 201930
    },
    {
      "epoch": 57.32046551234743,
      "grad_norm": 0.05320831760764122,
      "learning_rate": 4.270309395401646e-05,
      "loss": 0.00013310648500919342,
      "step": 201940
    },
    {
      "epoch": 57.323304002270795,
      "grad_norm": 5.967705726623535,
      "learning_rate": 4.2700255464093104e-05,
      "loss": 0.0008825855329632759,
      "step": 201950
    },
    {
      "epoch": 57.32614249219415,
      "grad_norm": 0.16669918596744537,
      "learning_rate": 4.2697416974169746e-05,
      "loss": 0.00038497764617204664,
      "step": 201960
    },
    {
      "epoch": 57.328980982117514,
      "grad_norm": 0.13413560390472412,
      "learning_rate": 4.269457848424638e-05,
      "loss": 0.0013362009078264237,
      "step": 201970
    },
    {
      "epoch": 57.33181947204088,
      "grad_norm": 0.010523278266191483,
      "learning_rate": 4.269173999432302e-05,
      "loss": 0.00022161733359098433,
      "step": 201980
    },
    {
      "epoch": 57.33465796196423,
      "grad_norm": 0.22267121076583862,
      "learning_rate": 4.2688901504399656e-05,
      "loss": 0.0005072617903351783,
      "step": 201990
    },
    {
      "epoch": 57.337496451887596,
      "grad_norm": 0.030805744230747223,
      "learning_rate": 4.2686063014476305e-05,
      "loss": 0.0009692342951893807,
      "step": 202000
    },
    {
      "epoch": 57.337496451887596,
      "eval_accuracy": 0.9809245247027405,
      "eval_loss": 0.07391611486673355,
      "eval_runtime": 32.1412,
      "eval_samples_per_second": 489.31,
      "eval_steps_per_second": 7.654,
      "step": 202000
    },
    {
      "epoch": 57.34033494181096,
      "grad_norm": 0.07943688333034515,
      "learning_rate": 4.268322452455294e-05,
      "loss": 0.00026950743049383166,
      "step": 202010
    },
    {
      "epoch": 57.343173431734314,
      "grad_norm": 0.048843540251255035,
      "learning_rate": 4.2680386034629574e-05,
      "loss": 0.0017082693055272102,
      "step": 202020
    },
    {
      "epoch": 57.34601192165768,
      "grad_norm": 0.03189486265182495,
      "learning_rate": 4.267754754470622e-05,
      "loss": 0.0003001859411597252,
      "step": 202030
    },
    {
      "epoch": 57.34885041158104,
      "grad_norm": 0.15821634232997894,
      "learning_rate": 4.2674709054782857e-05,
      "loss": 0.0006166325882077217,
      "step": 202040
    },
    {
      "epoch": 57.3516889015044,
      "grad_norm": 0.8173258900642395,
      "learning_rate": 4.26718705648595e-05,
      "loss": 0.002323352359235287,
      "step": 202050
    },
    {
      "epoch": 57.35452739142776,
      "grad_norm": 0.026219304651021957,
      "learning_rate": 4.266903207493614e-05,
      "loss": 0.0017041206359863282,
      "step": 202060
    },
    {
      "epoch": 57.35736588135112,
      "grad_norm": 0.04855605587363243,
      "learning_rate": 4.2666193585012774e-05,
      "loss": 0.0004995286464691163,
      "step": 202070
    },
    {
      "epoch": 57.360204371274484,
      "grad_norm": 0.7108160853385925,
      "learning_rate": 4.2663355095089415e-05,
      "loss": 0.00028795972466468813,
      "step": 202080
    },
    {
      "epoch": 57.36304286119784,
      "grad_norm": 0.3911873400211334,
      "learning_rate": 4.266051660516605e-05,
      "loss": 0.0010342035442590714,
      "step": 202090
    },
    {
      "epoch": 57.3658813511212,
      "grad_norm": 0.22070525586605072,
      "learning_rate": 4.265767811524269e-05,
      "loss": 0.001557663269340992,
      "step": 202100
    },
    {
      "epoch": 57.368719841044566,
      "grad_norm": 0.1147550493478775,
      "learning_rate": 4.265483962531933e-05,
      "loss": 0.0018103081732988358,
      "step": 202110
    },
    {
      "epoch": 57.37155833096792,
      "grad_norm": 0.029643462970852852,
      "learning_rate": 4.265200113539597e-05,
      "loss": 0.00048581007868051527,
      "step": 202120
    },
    {
      "epoch": 57.374396820891285,
      "grad_norm": 0.014745309948921204,
      "learning_rate": 4.2649162645472615e-05,
      "loss": 0.001324942521750927,
      "step": 202130
    },
    {
      "epoch": 57.37723531081465,
      "grad_norm": 0.23990780115127563,
      "learning_rate": 4.264632415554925e-05,
      "loss": 0.00026526432484388354,
      "step": 202140
    },
    {
      "epoch": 57.38007380073801,
      "grad_norm": 0.02432987093925476,
      "learning_rate": 4.2643485665625885e-05,
      "loss": 0.0018613256514072418,
      "step": 202150
    },
    {
      "epoch": 57.382912290661366,
      "grad_norm": 0.15802666544914246,
      "learning_rate": 4.264064717570253e-05,
      "loss": 0.0003154255449771881,
      "step": 202160
    },
    {
      "epoch": 57.38575078058473,
      "grad_norm": 0.1496344655752182,
      "learning_rate": 4.263780868577917e-05,
      "loss": 0.000443795882165432,
      "step": 202170
    },
    {
      "epoch": 57.38858927050809,
      "grad_norm": 0.21130171418190002,
      "learning_rate": 4.263497019585581e-05,
      "loss": 0.00035362783819437027,
      "step": 202180
    },
    {
      "epoch": 57.39142776043145,
      "grad_norm": 14.03533935546875,
      "learning_rate": 4.263213170593244e-05,
      "loss": 0.006137214973568917,
      "step": 202190
    },
    {
      "epoch": 57.39426625035481,
      "grad_norm": 0.10509560257196426,
      "learning_rate": 4.2629293216009085e-05,
      "loss": 0.00046892613172531126,
      "step": 202200
    },
    {
      "epoch": 57.39710474027817,
      "grad_norm": 0.1754673570394516,
      "learning_rate": 4.2626454726085726e-05,
      "loss": 0.0007714690640568733,
      "step": 202210
    },
    {
      "epoch": 57.39994323020153,
      "grad_norm": 2.183049440383911,
      "learning_rate": 4.262361623616236e-05,
      "loss": 0.0019114887341856956,
      "step": 202220
    },
    {
      "epoch": 57.40278172012489,
      "grad_norm": 0.33251047134399414,
      "learning_rate": 4.2620777746239e-05,
      "loss": 0.0008828723803162575,
      "step": 202230
    },
    {
      "epoch": 57.405620210048255,
      "grad_norm": 0.3166636526584625,
      "learning_rate": 4.261793925631564e-05,
      "loss": 0.0011380380019545555,
      "step": 202240
    },
    {
      "epoch": 57.40845869997162,
      "grad_norm": 0.1497737467288971,
      "learning_rate": 4.261510076639228e-05,
      "loss": 0.005512848496437073,
      "step": 202250
    },
    {
      "epoch": 57.411297189894974,
      "grad_norm": 0.022029288113117218,
      "learning_rate": 4.2612262276468926e-05,
      "loss": 0.0009461095556616783,
      "step": 202260
    },
    {
      "epoch": 57.41413567981834,
      "grad_norm": 1.2176225185394287,
      "learning_rate": 4.260942378654556e-05,
      "loss": 0.0016103994101285934,
      "step": 202270
    },
    {
      "epoch": 57.4169741697417,
      "grad_norm": 3.631279945373535,
      "learning_rate": 4.2606585296622195e-05,
      "loss": 0.002469648979604244,
      "step": 202280
    },
    {
      "epoch": 57.419812659665055,
      "grad_norm": 0.30400773882865906,
      "learning_rate": 4.2603746806698843e-05,
      "loss": 0.0025425424799323084,
      "step": 202290
    },
    {
      "epoch": 57.42265114958842,
      "grad_norm": 0.62889564037323,
      "learning_rate": 4.260090831677548e-05,
      "loss": 0.0006898233667016029,
      "step": 202300
    },
    {
      "epoch": 57.42548963951178,
      "grad_norm": 5.211442470550537,
      "learning_rate": 4.259806982685212e-05,
      "loss": 0.0012510219588875771,
      "step": 202310
    },
    {
      "epoch": 57.428328129435144,
      "grad_norm": 1.352169394493103,
      "learning_rate": 4.2595231336928754e-05,
      "loss": 0.0021158624440431595,
      "step": 202320
    },
    {
      "epoch": 57.4311666193585,
      "grad_norm": 0.5961002111434937,
      "learning_rate": 4.2592392847005395e-05,
      "loss": 0.0006390294060111046,
      "step": 202330
    },
    {
      "epoch": 57.43400510928186,
      "grad_norm": 0.18688765168190002,
      "learning_rate": 4.258955435708204e-05,
      "loss": 0.0005671286955475808,
      "step": 202340
    },
    {
      "epoch": 57.436843599205226,
      "grad_norm": 0.47930917143821716,
      "learning_rate": 4.258671586715867e-05,
      "loss": 0.000659513846039772,
      "step": 202350
    },
    {
      "epoch": 57.43968208912858,
      "grad_norm": 19.706541061401367,
      "learning_rate": 4.258387737723531e-05,
      "loss": 0.003128024563193321,
      "step": 202360
    },
    {
      "epoch": 57.442520579051944,
      "grad_norm": 1.855851411819458,
      "learning_rate": 4.2581038887311954e-05,
      "loss": 0.00028920676559209826,
      "step": 202370
    },
    {
      "epoch": 57.44535906897531,
      "grad_norm": 0.10312853753566742,
      "learning_rate": 4.257820039738859e-05,
      "loss": 0.0003955261781811714,
      "step": 202380
    },
    {
      "epoch": 57.44819755889866,
      "grad_norm": 0.013278954662382603,
      "learning_rate": 4.257536190746523e-05,
      "loss": 0.0009759977459907532,
      "step": 202390
    },
    {
      "epoch": 57.451036048822026,
      "grad_norm": 0.006509707774966955,
      "learning_rate": 4.257252341754187e-05,
      "loss": 0.00046918764710426333,
      "step": 202400
    },
    {
      "epoch": 57.45387453874539,
      "grad_norm": 0.22760948538780212,
      "learning_rate": 4.2569684927618506e-05,
      "loss": 0.002006787434220314,
      "step": 202410
    },
    {
      "epoch": 57.45671302866875,
      "grad_norm": 1.296704888343811,
      "learning_rate": 4.256684643769515e-05,
      "loss": 0.0034874189645051957,
      "step": 202420
    },
    {
      "epoch": 57.45955151859211,
      "grad_norm": 2.190035104751587,
      "learning_rate": 4.256400794777179e-05,
      "loss": 0.0036304362118244173,
      "step": 202430
    },
    {
      "epoch": 57.46239000851547,
      "grad_norm": 0.009228305891156197,
      "learning_rate": 4.2561169457848423e-05,
      "loss": 0.0006612570956349373,
      "step": 202440
    },
    {
      "epoch": 57.46522849843883,
      "grad_norm": 0.12334689497947693,
      "learning_rate": 4.2558614816917403e-05,
      "loss": 0.011062853038311005,
      "step": 202450
    },
    {
      "epoch": 57.46806698836219,
      "grad_norm": 0.08039363473653793,
      "learning_rate": 4.255577632699404e-05,
      "loss": 0.0007342813536524772,
      "step": 202460
    },
    {
      "epoch": 57.47090547828555,
      "grad_norm": 0.370477557182312,
      "learning_rate": 4.255293783707068e-05,
      "loss": 0.0006167536601424217,
      "step": 202470
    },
    {
      "epoch": 57.473743968208915,
      "grad_norm": 0.1946653574705124,
      "learning_rate": 4.255009934714732e-05,
      "loss": 0.0016571998596191407,
      "step": 202480
    },
    {
      "epoch": 57.47658245813227,
      "grad_norm": 0.06228036805987358,
      "learning_rate": 4.2547260857223955e-05,
      "loss": 0.0005370384082198143,
      "step": 202490
    },
    {
      "epoch": 57.47942094805563,
      "grad_norm": 0.023676464334130287,
      "learning_rate": 4.25444223673006e-05,
      "loss": 0.0004003720358014107,
      "step": 202500
    },
    {
      "epoch": 57.47942094805563,
      "eval_accuracy": 0.9772365994786036,
      "eval_loss": 0.0778462290763855,
      "eval_runtime": 32.0481,
      "eval_samples_per_second": 490.731,
      "eval_steps_per_second": 7.676,
      "step": 202500
    },
    {
      "epoch": 57.482259437978996,
      "grad_norm": 0.42736175656318665,
      "learning_rate": 4.254158387737724e-05,
      "loss": 0.0013916702941060067,
      "step": 202510
    },
    {
      "epoch": 57.48509792790236,
      "grad_norm": 0.1252487748861313,
      "learning_rate": 4.253874538745388e-05,
      "loss": 0.001075398176908493,
      "step": 202520
    },
    {
      "epoch": 57.487936417825715,
      "grad_norm": 0.20824357867240906,
      "learning_rate": 4.2535906897530514e-05,
      "loss": 0.0006401171907782554,
      "step": 202530
    },
    {
      "epoch": 57.49077490774908,
      "grad_norm": 0.1311708688735962,
      "learning_rate": 4.2533068407607155e-05,
      "loss": 0.0008078237995505333,
      "step": 202540
    },
    {
      "epoch": 57.49361339767244,
      "grad_norm": 0.01980063132941723,
      "learning_rate": 4.25302299176838e-05,
      "loss": 0.001834871806204319,
      "step": 202550
    },
    {
      "epoch": 57.496451887595796,
      "grad_norm": 0.15550297498703003,
      "learning_rate": 4.252739142776043e-05,
      "loss": 0.0002281099557876587,
      "step": 202560
    },
    {
      "epoch": 57.49929037751916,
      "grad_norm": 0.06680907309055328,
      "learning_rate": 4.252455293783707e-05,
      "loss": 0.004640765488147736,
      "step": 202570
    },
    {
      "epoch": 57.50212886744252,
      "grad_norm": 0.21559618413448334,
      "learning_rate": 4.2521714447913714e-05,
      "loss": 0.00037467554211616515,
      "step": 202580
    },
    {
      "epoch": 57.50496735736588,
      "grad_norm": 0.025715401396155357,
      "learning_rate": 4.251887595799035e-05,
      "loss": 0.00044564176350831987,
      "step": 202590
    },
    {
      "epoch": 57.50780584728924,
      "grad_norm": 0.3162256181240082,
      "learning_rate": 4.251603746806699e-05,
      "loss": 0.0005505038425326347,
      "step": 202600
    },
    {
      "epoch": 57.510644337212604,
      "grad_norm": 0.7635800838470459,
      "learning_rate": 4.251319897814363e-05,
      "loss": 0.0008844781666994094,
      "step": 202610
    },
    {
      "epoch": 57.51348282713597,
      "grad_norm": 0.005411970894783735,
      "learning_rate": 4.2510360488220266e-05,
      "loss": 0.000229620561003685,
      "step": 202620
    },
    {
      "epoch": 57.51632131705932,
      "grad_norm": 0.03309378772974014,
      "learning_rate": 4.250752199829691e-05,
      "loss": 0.00036442819982767104,
      "step": 202630
    },
    {
      "epoch": 57.519159806982685,
      "grad_norm": 0.023250920698046684,
      "learning_rate": 4.250468350837355e-05,
      "loss": 0.0031995609402656554,
      "step": 202640
    },
    {
      "epoch": 57.52199829690605,
      "grad_norm": 15.291648864746094,
      "learning_rate": 4.250184501845019e-05,
      "loss": 0.005511532723903656,
      "step": 202650
    },
    {
      "epoch": 57.524836786829404,
      "grad_norm": 0.10993843525648117,
      "learning_rate": 4.2499006528526825e-05,
      "loss": 0.0018990088254213333,
      "step": 202660
    },
    {
      "epoch": 57.52767527675277,
      "grad_norm": 2.188270330429077,
      "learning_rate": 4.2496168038603466e-05,
      "loss": 0.0013174111023545264,
      "step": 202670
    },
    {
      "epoch": 57.53051376667613,
      "grad_norm": 0.108439140021801,
      "learning_rate": 4.249332954868011e-05,
      "loss": 0.00425332672894001,
      "step": 202680
    },
    {
      "epoch": 57.533352256599485,
      "grad_norm": 0.04370484501123428,
      "learning_rate": 4.249049105875674e-05,
      "loss": 0.00023379530757665633,
      "step": 202690
    },
    {
      "epoch": 57.53619074652285,
      "grad_norm": 0.017790384590625763,
      "learning_rate": 4.2487652568833384e-05,
      "loss": 0.001262430101633072,
      "step": 202700
    },
    {
      "epoch": 57.53902923644621,
      "grad_norm": 0.05395805835723877,
      "learning_rate": 4.2484814078910025e-05,
      "loss": 0.0012628158554434776,
      "step": 202710
    },
    {
      "epoch": 57.541867726369574,
      "grad_norm": 1.0747252702713013,
      "learning_rate": 4.248197558898666e-05,
      "loss": 0.0020679650828242303,
      "step": 202720
    },
    {
      "epoch": 57.54470621629293,
      "grad_norm": 0.4443657696247101,
      "learning_rate": 4.24791370990633e-05,
      "loss": 0.0008516723290085792,
      "step": 202730
    },
    {
      "epoch": 57.54754470621629,
      "grad_norm": 7.216020107269287,
      "learning_rate": 4.247629860913994e-05,
      "loss": 0.003615783154964447,
      "step": 202740
    },
    {
      "epoch": 57.550383196139656,
      "grad_norm": 0.061488863080739975,
      "learning_rate": 4.247346011921658e-05,
      "loss": 0.0009046025574207306,
      "step": 202750
    },
    {
      "epoch": 57.55322168606301,
      "grad_norm": 1.0289504528045654,
      "learning_rate": 4.247062162929322e-05,
      "loss": 0.0007604986429214477,
      "step": 202760
    },
    {
      "epoch": 57.556060175986374,
      "grad_norm": 0.08776504546403885,
      "learning_rate": 4.246778313936986e-05,
      "loss": 0.007427095621824265,
      "step": 202770
    },
    {
      "epoch": 57.55889866590974,
      "grad_norm": 0.09619253873825073,
      "learning_rate": 4.2464944649446494e-05,
      "loss": 0.0010243801400065422,
      "step": 202780
    },
    {
      "epoch": 57.5617371558331,
      "grad_norm": 0.11272318661212921,
      "learning_rate": 4.2462106159523136e-05,
      "loss": 0.0013272235170006751,
      "step": 202790
    },
    {
      "epoch": 57.564575645756456,
      "grad_norm": 0.06961038708686829,
      "learning_rate": 4.245926766959977e-05,
      "loss": 0.001326056569814682,
      "step": 202800
    },
    {
      "epoch": 57.56741413567982,
      "grad_norm": 1.0135842561721802,
      "learning_rate": 4.245642917967642e-05,
      "loss": 0.0036885514855384827,
      "step": 202810
    },
    {
      "epoch": 57.57025262560318,
      "grad_norm": 0.0341981016099453,
      "learning_rate": 4.245359068975305e-05,
      "loss": 0.000358176976442337,
      "step": 202820
    },
    {
      "epoch": 57.57309111552654,
      "grad_norm": 0.03228076174855232,
      "learning_rate": 4.245075219982969e-05,
      "loss": 0.0005840960890054703,
      "step": 202830
    },
    {
      "epoch": 57.5759296054499,
      "grad_norm": 0.7097041010856628,
      "learning_rate": 4.2447913709906336e-05,
      "loss": 0.00025296267122030256,
      "step": 202840
    },
    {
      "epoch": 57.57876809537326,
      "grad_norm": 0.28014692664146423,
      "learning_rate": 4.244507521998297e-05,
      "loss": 0.000554264523088932,
      "step": 202850
    },
    {
      "epoch": 57.58160658529662,
      "grad_norm": 0.07063359767198563,
      "learning_rate": 4.244223673005961e-05,
      "loss": 0.0003155924379825592,
      "step": 202860
    },
    {
      "epoch": 57.58444507521998,
      "grad_norm": 0.10980930924415588,
      "learning_rate": 4.243939824013625e-05,
      "loss": 0.010909478366374969,
      "step": 202870
    },
    {
      "epoch": 57.587283565143345,
      "grad_norm": 0.030080821365118027,
      "learning_rate": 4.243655975021289e-05,
      "loss": 0.0005289612337946892,
      "step": 202880
    },
    {
      "epoch": 57.59012205506671,
      "grad_norm": 0.9363241195678711,
      "learning_rate": 4.243372126028953e-05,
      "loss": 0.0012692077085375785,
      "step": 202890
    },
    {
      "epoch": 57.59296054499006,
      "grad_norm": 0.9792172908782959,
      "learning_rate": 4.2430882770366164e-05,
      "loss": 0.005339708179235458,
      "step": 202900
    },
    {
      "epoch": 57.595799034913426,
      "grad_norm": 0.3905777633190155,
      "learning_rate": 4.2428044280442805e-05,
      "loss": 0.007431463152170181,
      "step": 202910
    },
    {
      "epoch": 57.59863752483679,
      "grad_norm": 1.0960655212402344,
      "learning_rate": 4.2425205790519446e-05,
      "loss": 0.0017599966377019882,
      "step": 202920
    },
    {
      "epoch": 57.601476014760145,
      "grad_norm": 13.955649375915527,
      "learning_rate": 4.242236730059608e-05,
      "loss": 0.006590297818183899,
      "step": 202930
    },
    {
      "epoch": 57.60431450468351,
      "grad_norm": 0.15392546355724335,
      "learning_rate": 4.241952881067273e-05,
      "loss": 0.0009449355304241181,
      "step": 202940
    },
    {
      "epoch": 57.60715299460687,
      "grad_norm": 0.05764875188469887,
      "learning_rate": 4.2416690320749364e-05,
      "loss": 0.0004433572292327881,
      "step": 202950
    },
    {
      "epoch": 57.60999148453023,
      "grad_norm": 1.4319692850112915,
      "learning_rate": 4.2413851830826e-05,
      "loss": 0.002892743796110153,
      "step": 202960
    },
    {
      "epoch": 57.61282997445359,
      "grad_norm": 0.14103437960147858,
      "learning_rate": 4.2411013340902646e-05,
      "loss": 0.00029717758297920227,
      "step": 202970
    },
    {
      "epoch": 57.61566846437695,
      "grad_norm": 0.15540529787540436,
      "learning_rate": 4.240817485097928e-05,
      "loss": 0.002683236077427864,
      "step": 202980
    },
    {
      "epoch": 57.618506954300315,
      "grad_norm": 2.6307213306427,
      "learning_rate": 4.240533636105592e-05,
      "loss": 0.0008884282782673836,
      "step": 202990
    },
    {
      "epoch": 57.62134544422367,
      "grad_norm": 1.4305577278137207,
      "learning_rate": 4.240249787113256e-05,
      "loss": 0.0009783884510397912,
      "step": 203000
    },
    {
      "epoch": 57.62134544422367,
      "eval_accuracy": 0.9762192407960831,
      "eval_loss": 0.0908336341381073,
      "eval_runtime": 32.5069,
      "eval_samples_per_second": 483.805,
      "eval_steps_per_second": 7.568,
      "step": 203000
    },
    {
      "epoch": 57.624183934147034,
      "grad_norm": 0.8108060956001282,
      "learning_rate": 4.23996593812092e-05,
      "loss": 0.003211665153503418,
      "step": 203010
    },
    {
      "epoch": 57.6270224240704,
      "grad_norm": 0.9907627105712891,
      "learning_rate": 4.239682089128584e-05,
      "loss": 0.0004545176401734352,
      "step": 203020
    },
    {
      "epoch": 57.62986091399375,
      "grad_norm": 0.7625811100006104,
      "learning_rate": 4.2393982401362474e-05,
      "loss": 0.00041805803775787356,
      "step": 203030
    },
    {
      "epoch": 57.632699403917115,
      "grad_norm": 0.3649080991744995,
      "learning_rate": 4.2391143911439116e-05,
      "loss": 0.0007667779922485351,
      "step": 203040
    },
    {
      "epoch": 57.63553789384048,
      "grad_norm": 1.927728533744812,
      "learning_rate": 4.238830542151576e-05,
      "loss": 0.003935283422470093,
      "step": 203050
    },
    {
      "epoch": 57.63837638376384,
      "grad_norm": 0.07767146080732346,
      "learning_rate": 4.238546693159239e-05,
      "loss": 0.0009949265047907829,
      "step": 203060
    },
    {
      "epoch": 57.6412148736872,
      "grad_norm": 2.8301377296447754,
      "learning_rate": 4.238262844166904e-05,
      "loss": 0.0012938981875777245,
      "step": 203070
    },
    {
      "epoch": 57.64405336361056,
      "grad_norm": 0.0549924261868,
      "learning_rate": 4.2379789951745674e-05,
      "loss": 0.0010164132341742515,
      "step": 203080
    },
    {
      "epoch": 57.64689185353392,
      "grad_norm": 0.11862672120332718,
      "learning_rate": 4.237695146182231e-05,
      "loss": 0.00030298028141260145,
      "step": 203090
    },
    {
      "epoch": 57.64973034345728,
      "grad_norm": 0.5455009341239929,
      "learning_rate": 4.237411297189895e-05,
      "loss": 0.0003002457320690155,
      "step": 203100
    },
    {
      "epoch": 57.65256883338064,
      "grad_norm": 4.040154933929443,
      "learning_rate": 4.237127448197559e-05,
      "loss": 0.0009656190872192383,
      "step": 203110
    },
    {
      "epoch": 57.655407323304004,
      "grad_norm": 0.021062998101115227,
      "learning_rate": 4.236843599205223e-05,
      "loss": 0.0003586918115615845,
      "step": 203120
    },
    {
      "epoch": 57.65824581322736,
      "grad_norm": 0.15249450504779816,
      "learning_rate": 4.236559750212887e-05,
      "loss": 0.00026847999542951585,
      "step": 203130
    },
    {
      "epoch": 57.66108430315072,
      "grad_norm": 0.021003063768148422,
      "learning_rate": 4.236275901220551e-05,
      "loss": 0.000362345390021801,
      "step": 203140
    },
    {
      "epoch": 57.663922793074086,
      "grad_norm": 0.012804241850972176,
      "learning_rate": 4.235992052228215e-05,
      "loss": 0.0003417976200580597,
      "step": 203150
    },
    {
      "epoch": 57.66676128299745,
      "grad_norm": 0.12278220057487488,
      "learning_rate": 4.2357082032358785e-05,
      "loss": 0.000321282260119915,
      "step": 203160
    },
    {
      "epoch": 57.669599772920805,
      "grad_norm": 0.3321686089038849,
      "learning_rate": 4.2354243542435427e-05,
      "loss": 0.00035560932010412214,
      "step": 203170
    },
    {
      "epoch": 57.67243826284417,
      "grad_norm": 0.012518245726823807,
      "learning_rate": 4.235140505251207e-05,
      "loss": 0.0002537654712796211,
      "step": 203180
    },
    {
      "epoch": 57.67527675276753,
      "grad_norm": 0.02378888800740242,
      "learning_rate": 4.23485665625887e-05,
      "loss": 0.0003287248313426971,
      "step": 203190
    },
    {
      "epoch": 57.678115242690886,
      "grad_norm": 0.013889407739043236,
      "learning_rate": 4.2345728072665344e-05,
      "loss": 0.0002557728439569473,
      "step": 203200
    },
    {
      "epoch": 57.68095373261425,
      "grad_norm": 0.025578226894140244,
      "learning_rate": 4.2342889582741985e-05,
      "loss": 0.0002968860790133476,
      "step": 203210
    },
    {
      "epoch": 57.68379222253761,
      "grad_norm": 0.024410929530858994,
      "learning_rate": 4.234005109281862e-05,
      "loss": 8.724108338356018e-05,
      "step": 203220
    },
    {
      "epoch": 57.68663071246097,
      "grad_norm": 0.06054743751883507,
      "learning_rate": 4.233721260289526e-05,
      "loss": 0.0007057260721921921,
      "step": 203230
    },
    {
      "epoch": 57.68946920238433,
      "grad_norm": 0.00766676664352417,
      "learning_rate": 4.23343741129719e-05,
      "loss": 0.0007407240569591522,
      "step": 203240
    },
    {
      "epoch": 57.69230769230769,
      "grad_norm": 0.007530054543167353,
      "learning_rate": 4.233153562304854e-05,
      "loss": 0.0003676220774650574,
      "step": 203250
    },
    {
      "epoch": 57.695146182231056,
      "grad_norm": 0.023047449067234993,
      "learning_rate": 4.232869713312518e-05,
      "loss": 0.00044903326779603957,
      "step": 203260
    },
    {
      "epoch": 57.69798467215441,
      "grad_norm": 0.07136601209640503,
      "learning_rate": 4.232585864320182e-05,
      "loss": 0.0002299761399626732,
      "step": 203270
    },
    {
      "epoch": 57.700823162077775,
      "grad_norm": 0.12068890780210495,
      "learning_rate": 4.232302015327846e-05,
      "loss": 0.0002027789130806923,
      "step": 203280
    },
    {
      "epoch": 57.70366165200114,
      "grad_norm": 0.031840648502111435,
      "learning_rate": 4.2320181663355096e-05,
      "loss": 0.0006800269708037377,
      "step": 203290
    },
    {
      "epoch": 57.706500141924494,
      "grad_norm": 0.020992863923311234,
      "learning_rate": 4.231734317343173e-05,
      "loss": 0.0005277404561638832,
      "step": 203300
    },
    {
      "epoch": 57.70933863184786,
      "grad_norm": 0.1057388037443161,
      "learning_rate": 4.231450468350838e-05,
      "loss": 0.0006167959421873092,
      "step": 203310
    },
    {
      "epoch": 57.71217712177122,
      "grad_norm": 0.03239342197775841,
      "learning_rate": 4.231166619358501e-05,
      "loss": 0.0004502894356846809,
      "step": 203320
    },
    {
      "epoch": 57.715015611694575,
      "grad_norm": 0.020041802898049355,
      "learning_rate": 4.2308827703661655e-05,
      "loss": 0.0004637433215975761,
      "step": 203330
    },
    {
      "epoch": 57.71785410161794,
      "grad_norm": 0.021435806527733803,
      "learning_rate": 4.2305989213738296e-05,
      "loss": 0.0014048116281628609,
      "step": 203340
    },
    {
      "epoch": 57.7206925915413,
      "grad_norm": 0.039801742881536484,
      "learning_rate": 4.230315072381493e-05,
      "loss": 0.0005146607756614685,
      "step": 203350
    },
    {
      "epoch": 57.723531081464664,
      "grad_norm": 0.6433714032173157,
      "learning_rate": 4.230031223389157e-05,
      "loss": 0.0005228357389569282,
      "step": 203360
    },
    {
      "epoch": 57.72636957138802,
      "grad_norm": 0.03833135962486267,
      "learning_rate": 4.229747374396821e-05,
      "loss": 0.0030122546479105948,
      "step": 203370
    },
    {
      "epoch": 57.72920806131138,
      "grad_norm": 0.06501774489879608,
      "learning_rate": 4.229463525404485e-05,
      "loss": 0.00021413061767816544,
      "step": 203380
    },
    {
      "epoch": 57.732046551234745,
      "grad_norm": 0.018152276054024696,
      "learning_rate": 4.229179676412149e-05,
      "loss": 0.0058472417294979095,
      "step": 203390
    },
    {
      "epoch": 57.7348850411581,
      "grad_norm": 0.151711568236351,
      "learning_rate": 4.2288958274198124e-05,
      "loss": 0.0020419908687472343,
      "step": 203400
    },
    {
      "epoch": 57.737723531081464,
      "grad_norm": 0.15659219026565552,
      "learning_rate": 4.228611978427477e-05,
      "loss": 0.001192198693752289,
      "step": 203410
    },
    {
      "epoch": 57.74056202100483,
      "grad_norm": 0.3061080873012543,
      "learning_rate": 4.228328129435141e-05,
      "loss": 0.0012196313589811325,
      "step": 203420
    },
    {
      "epoch": 57.74340051092818,
      "grad_norm": 1.2730869054794312,
      "learning_rate": 4.228044280442804e-05,
      "loss": 0.0008027389645576477,
      "step": 203430
    },
    {
      "epoch": 57.746239000851546,
      "grad_norm": 1.0387301445007324,
      "learning_rate": 4.227760431450469e-05,
      "loss": 0.0015052201226353646,
      "step": 203440
    },
    {
      "epoch": 57.74907749077491,
      "grad_norm": 0.4488503932952881,
      "learning_rate": 4.2274765824581324e-05,
      "loss": 0.0006218140944838524,
      "step": 203450
    },
    {
      "epoch": 57.75191598069827,
      "grad_norm": 8.58390998840332,
      "learning_rate": 4.2271927334657965e-05,
      "loss": 0.007500927150249481,
      "step": 203460
    },
    {
      "epoch": 57.75475447062163,
      "grad_norm": 0.5092456340789795,
      "learning_rate": 4.226908884473461e-05,
      "loss": 0.0009272841736674309,
      "step": 203470
    },
    {
      "epoch": 57.75759296054499,
      "grad_norm": 3.9844167232513428,
      "learning_rate": 4.226625035481124e-05,
      "loss": 0.0033251769840717317,
      "step": 203480
    },
    {
      "epoch": 57.76043145046835,
      "grad_norm": 20.932180404663086,
      "learning_rate": 4.226341186488788e-05,
      "loss": 0.015054890513420105,
      "step": 203490
    },
    {
      "epoch": 57.76326994039171,
      "grad_norm": 0.021061381325125694,
      "learning_rate": 4.226057337496452e-05,
      "loss": 0.0001088099554181099,
      "step": 203500
    },
    {
      "epoch": 57.76326994039171,
      "eval_accuracy": 0.9788262224200419,
      "eval_loss": 0.0806448757648468,
      "eval_runtime": 32.6241,
      "eval_samples_per_second": 482.068,
      "eval_steps_per_second": 7.54,
      "step": 203500
    },
    {
      "epoch": 57.76610843031507,
      "grad_norm": 0.1386331170797348,
      "learning_rate": 4.225773488504116e-05,
      "loss": 0.0008573323488235473,
      "step": 203510
    },
    {
      "epoch": 57.768946920238434,
      "grad_norm": 0.03176489472389221,
      "learning_rate": 4.22548963951178e-05,
      "loss": 0.0014109503477811812,
      "step": 203520
    },
    {
      "epoch": 57.7717854101618,
      "grad_norm": 0.0408705472946167,
      "learning_rate": 4.2252057905194435e-05,
      "loss": 0.001288481242954731,
      "step": 203530
    },
    {
      "epoch": 57.77462390008515,
      "grad_norm": 0.8526514172554016,
      "learning_rate": 4.224921941527108e-05,
      "loss": 0.0005118295550346375,
      "step": 203540
    },
    {
      "epoch": 57.777462390008516,
      "grad_norm": 1.2944879531860352,
      "learning_rate": 4.224638092534772e-05,
      "loss": 0.0012841032817959785,
      "step": 203550
    },
    {
      "epoch": 57.78030087993188,
      "grad_norm": 0.05061417445540428,
      "learning_rate": 4.224354243542435e-05,
      "loss": 0.0022646715864539147,
      "step": 203560
    },
    {
      "epoch": 57.783139369855235,
      "grad_norm": 0.28999507427215576,
      "learning_rate": 4.2240703945501e-05,
      "loss": 0.0009400840848684311,
      "step": 203570
    },
    {
      "epoch": 57.7859778597786,
      "grad_norm": 0.7716155648231506,
      "learning_rate": 4.2237865455577635e-05,
      "loss": 0.0007371071726083756,
      "step": 203580
    },
    {
      "epoch": 57.78881634970196,
      "grad_norm": 0.18266406655311584,
      "learning_rate": 4.2235026965654276e-05,
      "loss": 0.0003968212753534317,
      "step": 203590
    },
    {
      "epoch": 57.791654839625316,
      "grad_norm": 0.24841676652431488,
      "learning_rate": 4.223218847573091e-05,
      "loss": 0.0010218365117907525,
      "step": 203600
    },
    {
      "epoch": 57.79449332954868,
      "grad_norm": 0.23521652817726135,
      "learning_rate": 4.222934998580755e-05,
      "loss": 0.0007314315065741539,
      "step": 203610
    },
    {
      "epoch": 57.79733181947204,
      "grad_norm": 0.04392537474632263,
      "learning_rate": 4.2226511495884194e-05,
      "loss": 0.0026745861396193504,
      "step": 203620
    },
    {
      "epoch": 57.800170309395405,
      "grad_norm": 0.055736612528562546,
      "learning_rate": 4.222367300596083e-05,
      "loss": 0.002558356150984764,
      "step": 203630
    },
    {
      "epoch": 57.80300879931876,
      "grad_norm": 0.07415959239006042,
      "learning_rate": 4.222083451603747e-05,
      "loss": 0.0007418647408485412,
      "step": 203640
    },
    {
      "epoch": 57.805847289242124,
      "grad_norm": 0.02574933134019375,
      "learning_rate": 4.221799602611411e-05,
      "loss": 0.00044418778270483017,
      "step": 203650
    },
    {
      "epoch": 57.80868577916549,
      "grad_norm": 0.10044330358505249,
      "learning_rate": 4.2215157536190745e-05,
      "loss": 0.0009741054847836495,
      "step": 203660
    },
    {
      "epoch": 57.81152426908884,
      "grad_norm": 0.05653214082121849,
      "learning_rate": 4.221231904626739e-05,
      "loss": 0.0054633978754282,
      "step": 203670
    },
    {
      "epoch": 57.814362759012205,
      "grad_norm": 0.057849444448947906,
      "learning_rate": 4.220948055634403e-05,
      "loss": 0.0003483844920992851,
      "step": 203680
    },
    {
      "epoch": 57.81720124893557,
      "grad_norm": 0.14867746829986572,
      "learning_rate": 4.220664206642066e-05,
      "loss": 0.0003938857465982437,
      "step": 203690
    },
    {
      "epoch": 57.820039738858924,
      "grad_norm": 0.13294005393981934,
      "learning_rate": 4.2203803576497304e-05,
      "loss": 0.00025340262800455093,
      "step": 203700
    },
    {
      "epoch": 57.82287822878229,
      "grad_norm": 0.18580500781536102,
      "learning_rate": 4.2200965086573946e-05,
      "loss": 0.0013450739905238151,
      "step": 203710
    },
    {
      "epoch": 57.82571671870565,
      "grad_norm": 0.02587280422449112,
      "learning_rate": 4.219812659665058e-05,
      "loss": 0.0014976449310779572,
      "step": 203720
    },
    {
      "epoch": 57.82855520862901,
      "grad_norm": 0.05808544531464577,
      "learning_rate": 4.219528810672722e-05,
      "loss": 0.00041355248540639875,
      "step": 203730
    },
    {
      "epoch": 57.83139369855237,
      "grad_norm": 0.03316033259034157,
      "learning_rate": 4.219244961680386e-05,
      "loss": 0.0016160726547241211,
      "step": 203740
    },
    {
      "epoch": 57.83423218847573,
      "grad_norm": 0.32668185234069824,
      "learning_rate": 4.2189611126880504e-05,
      "loss": 0.0006963161751627922,
      "step": 203750
    },
    {
      "epoch": 57.837070678399094,
      "grad_norm": 0.32159045338630676,
      "learning_rate": 4.218677263695714e-05,
      "loss": 0.0003088017925620079,
      "step": 203760
    },
    {
      "epoch": 57.83990916832245,
      "grad_norm": 0.5166965126991272,
      "learning_rate": 4.218393414703378e-05,
      "loss": 0.0005923550575971604,
      "step": 203770
    },
    {
      "epoch": 57.84274765824581,
      "grad_norm": 0.04078520089387894,
      "learning_rate": 4.218109565711042e-05,
      "loss": 0.00023098234087228774,
      "step": 203780
    },
    {
      "epoch": 57.845586148169176,
      "grad_norm": 0.32828110456466675,
      "learning_rate": 4.2178257167187056e-05,
      "loss": 0.00035635977983474734,
      "step": 203790
    },
    {
      "epoch": 57.84842463809253,
      "grad_norm": 0.3432689905166626,
      "learning_rate": 4.21754186772637e-05,
      "loss": 0.0008433733135461807,
      "step": 203800
    },
    {
      "epoch": 57.851263128015894,
      "grad_norm": 0.04531968757510185,
      "learning_rate": 4.217258018734034e-05,
      "loss": 0.0003129448741674423,
      "step": 203810
    },
    {
      "epoch": 57.85410161793926,
      "grad_norm": 0.5039620995521545,
      "learning_rate": 4.2169741697416974e-05,
      "loss": 0.0013726573437452317,
      "step": 203820
    },
    {
      "epoch": 57.85694010786262,
      "grad_norm": 0.08485621213912964,
      "learning_rate": 4.2166903207493615e-05,
      "loss": 0.00023926272988319396,
      "step": 203830
    },
    {
      "epoch": 57.859778597785976,
      "grad_norm": 0.04285060614347458,
      "learning_rate": 4.2164064717570256e-05,
      "loss": 0.0004269793629646301,
      "step": 203840
    },
    {
      "epoch": 57.86261708770934,
      "grad_norm": 0.1039428561925888,
      "learning_rate": 4.216122622764689e-05,
      "loss": 0.0036132253706455232,
      "step": 203850
    },
    {
      "epoch": 57.8654555776327,
      "grad_norm": 0.5054091215133667,
      "learning_rate": 4.215838773772353e-05,
      "loss": 0.0037370655685663224,
      "step": 203860
    },
    {
      "epoch": 57.86829406755606,
      "grad_norm": 0.2088550329208374,
      "learning_rate": 4.2155549247800174e-05,
      "loss": 0.00022067539393901824,
      "step": 203870
    },
    {
      "epoch": 57.87113255747942,
      "grad_norm": 0.12185687571763992,
      "learning_rate": 4.2152710757876815e-05,
      "loss": 0.0004766346886754036,
      "step": 203880
    },
    {
      "epoch": 57.87397104740278,
      "grad_norm": 0.03746906295418739,
      "learning_rate": 4.214987226795345e-05,
      "loss": 0.0011947164312005043,
      "step": 203890
    },
    {
      "epoch": 57.876809537326146,
      "grad_norm": 0.22051361203193665,
      "learning_rate": 4.214703377803009e-05,
      "loss": 0.0002687947824597359,
      "step": 203900
    },
    {
      "epoch": 57.8796480272495,
      "grad_norm": 0.05351918935775757,
      "learning_rate": 4.214419528810673e-05,
      "loss": 0.00018464382737874984,
      "step": 203910
    },
    {
      "epoch": 57.882486517172865,
      "grad_norm": 0.028821086511015892,
      "learning_rate": 4.214135679818337e-05,
      "loss": 0.00046849120408296583,
      "step": 203920
    },
    {
      "epoch": 57.88532500709623,
      "grad_norm": 0.04851937294006348,
      "learning_rate": 4.213851830826001e-05,
      "loss": 0.000472315214574337,
      "step": 203930
    },
    {
      "epoch": 57.88816349701958,
      "grad_norm": 0.7905162572860718,
      "learning_rate": 4.213567981833665e-05,
      "loss": 0.0007357167080044747,
      "step": 203940
    },
    {
      "epoch": 57.891001986942946,
      "grad_norm": 0.6250839233398438,
      "learning_rate": 4.2132841328413284e-05,
      "loss": 0.0011721968650817872,
      "step": 203950
    },
    {
      "epoch": 57.89384047686631,
      "grad_norm": 4.825534343719482,
      "learning_rate": 4.2130002838489926e-05,
      "loss": 0.001409224234521389,
      "step": 203960
    },
    {
      "epoch": 57.896678966789665,
      "grad_norm": 0.22183270752429962,
      "learning_rate": 4.212716434856657e-05,
      "loss": 0.0003189507871866226,
      "step": 203970
    },
    {
      "epoch": 57.89951745671303,
      "grad_norm": 0.08644049614667892,
      "learning_rate": 4.21243258586432e-05,
      "loss": 0.0004677772521972656,
      "step": 203980
    },
    {
      "epoch": 57.90235594663639,
      "grad_norm": 0.02203330397605896,
      "learning_rate": 4.212148736871984e-05,
      "loss": 0.0003841409459710121,
      "step": 203990
    },
    {
      "epoch": 57.90519443655975,
      "grad_norm": 0.589742124080658,
      "learning_rate": 4.2118648878796484e-05,
      "loss": 0.00018324479460716248,
      "step": 204000
    },
    {
      "epoch": 57.90519443655975,
      "eval_accuracy": 0.9803522604438227,
      "eval_loss": 0.07444480806589127,
      "eval_runtime": 32.9807,
      "eval_samples_per_second": 476.854,
      "eval_steps_per_second": 7.459,
      "step": 204000
    },
    {
      "epoch": 57.90803292648311,
      "grad_norm": 0.010767682455480099,
      "learning_rate": 4.2115810388873126e-05,
      "loss": 0.0005605537444353104,
      "step": 204010
    },
    {
      "epoch": 57.91087141640647,
      "grad_norm": 0.09171885251998901,
      "learning_rate": 4.211297189894976e-05,
      "loss": 0.0002513719722628593,
      "step": 204020
    },
    {
      "epoch": 57.913709906329835,
      "grad_norm": 0.05096288397908211,
      "learning_rate": 4.2110133409026395e-05,
      "loss": 0.0004306025803089142,
      "step": 204030
    },
    {
      "epoch": 57.91654839625319,
      "grad_norm": 0.03095061145722866,
      "learning_rate": 4.210729491910304e-05,
      "loss": 0.00045613143593072893,
      "step": 204040
    },
    {
      "epoch": 57.919386886176554,
      "grad_norm": 0.29259613156318665,
      "learning_rate": 4.210445642917968e-05,
      "loss": 0.00023514945060014724,
      "step": 204050
    },
    {
      "epoch": 57.92222537609992,
      "grad_norm": 0.027359485626220703,
      "learning_rate": 4.210161793925632e-05,
      "loss": 0.00021498017013072969,
      "step": 204060
    },
    {
      "epoch": 57.92506386602327,
      "grad_norm": 0.021599385887384415,
      "learning_rate": 4.209877944933296e-05,
      "loss": 0.00017031989991664887,
      "step": 204070
    },
    {
      "epoch": 57.927902355946635,
      "grad_norm": 0.02814623713493347,
      "learning_rate": 4.2095940959409595e-05,
      "loss": 0.00016027409583330154,
      "step": 204080
    },
    {
      "epoch": 57.93074084587,
      "grad_norm": 0.006065940484404564,
      "learning_rate": 4.2093102469486236e-05,
      "loss": 0.00018275640904903413,
      "step": 204090
    },
    {
      "epoch": 57.93357933579336,
      "grad_norm": 0.05336625501513481,
      "learning_rate": 4.209026397956288e-05,
      "loss": 0.001657491736114025,
      "step": 204100
    },
    {
      "epoch": 57.93641782571672,
      "grad_norm": 0.14144258201122284,
      "learning_rate": 4.208742548963951e-05,
      "loss": 0.0003083301708102226,
      "step": 204110
    },
    {
      "epoch": 57.93925631564008,
      "grad_norm": 0.5792751908302307,
      "learning_rate": 4.2084586999716154e-05,
      "loss": 0.0003181589767336845,
      "step": 204120
    },
    {
      "epoch": 57.94209480556344,
      "grad_norm": 0.11705828458070755,
      "learning_rate": 4.208174850979279e-05,
      "loss": 0.00027374550700187683,
      "step": 204130
    },
    {
      "epoch": 57.9449332954868,
      "grad_norm": 0.12909115850925446,
      "learning_rate": 4.207891001986943e-05,
      "loss": 0.0009355930611491204,
      "step": 204140
    },
    {
      "epoch": 57.94777178541016,
      "grad_norm": 0.03992854803800583,
      "learning_rate": 4.207607152994607e-05,
      "loss": 0.004155614227056503,
      "step": 204150
    },
    {
      "epoch": 57.950610275333524,
      "grad_norm": 0.07783089578151703,
      "learning_rate": 4.2073233040022706e-05,
      "loss": 0.000640903040766716,
      "step": 204160
    },
    {
      "epoch": 57.95344876525688,
      "grad_norm": 0.11804375052452087,
      "learning_rate": 4.2070394550099354e-05,
      "loss": 0.00016187839210033417,
      "step": 204170
    },
    {
      "epoch": 57.95628725518024,
      "grad_norm": 0.3922516405582428,
      "learning_rate": 4.206755606017599e-05,
      "loss": 0.00034200269728899,
      "step": 204180
    },
    {
      "epoch": 57.959125745103606,
      "grad_norm": 0.05401405692100525,
      "learning_rate": 4.206471757025262e-05,
      "loss": 0.0001503054052591324,
      "step": 204190
    },
    {
      "epoch": 57.96196423502697,
      "grad_norm": 0.024301636964082718,
      "learning_rate": 4.206187908032927e-05,
      "loss": 0.00017855167388916015,
      "step": 204200
    },
    {
      "epoch": 57.964802724950324,
      "grad_norm": 0.010897928848862648,
      "learning_rate": 4.2059040590405906e-05,
      "loss": 0.00013245213776826858,
      "step": 204210
    },
    {
      "epoch": 57.96764121487369,
      "grad_norm": 0.35928940773010254,
      "learning_rate": 4.205620210048255e-05,
      "loss": 0.00038835816085338595,
      "step": 204220
    },
    {
      "epoch": 57.97047970479705,
      "grad_norm": 0.22269287705421448,
      "learning_rate": 4.205336361055918e-05,
      "loss": 0.000232057087123394,
      "step": 204230
    },
    {
      "epoch": 57.973318194720406,
      "grad_norm": 0.017344897612929344,
      "learning_rate": 4.205052512063582e-05,
      "loss": 0.00034070927649736403,
      "step": 204240
    },
    {
      "epoch": 57.97615668464377,
      "grad_norm": 0.03685707971453667,
      "learning_rate": 4.2047686630712465e-05,
      "loss": 0.0016710491850972175,
      "step": 204250
    },
    {
      "epoch": 57.97899517456713,
      "grad_norm": 0.022251730784773827,
      "learning_rate": 4.20448481407891e-05,
      "loss": 0.0005398474633693696,
      "step": 204260
    },
    {
      "epoch": 57.981833664490495,
      "grad_norm": 0.4273233115673065,
      "learning_rate": 4.204200965086574e-05,
      "loss": 0.000507097877562046,
      "step": 204270
    },
    {
      "epoch": 57.98467215441385,
      "grad_norm": 0.10712866485118866,
      "learning_rate": 4.203917116094238e-05,
      "loss": 0.0003894565626978874,
      "step": 204280
    },
    {
      "epoch": 57.98751064433721,
      "grad_norm": 0.14284949004650116,
      "learning_rate": 4.2036332671019017e-05,
      "loss": 0.00029521267861127854,
      "step": 204290
    },
    {
      "epoch": 57.990349134260576,
      "grad_norm": 0.06746077537536621,
      "learning_rate": 4.2033494181095665e-05,
      "loss": 0.0011087391525506974,
      "step": 204300
    },
    {
      "epoch": 57.99318762418393,
      "grad_norm": 0.030009908601641655,
      "learning_rate": 4.20306556911723e-05,
      "loss": 0.0005026973783969879,
      "step": 204310
    },
    {
      "epoch": 57.996026114107295,
      "grad_norm": 0.24859710037708282,
      "learning_rate": 4.2027817201248934e-05,
      "loss": 0.0002466266974806786,
      "step": 204320
    },
    {
      "epoch": 57.99886460403066,
      "grad_norm": 0.15011797845363617,
      "learning_rate": 4.2024978711325575e-05,
      "loss": 0.00021987874060869217,
      "step": 204330
    },
    {
      "epoch": 58.00170309395401,
      "grad_norm": 0.17467911541461945,
      "learning_rate": 4.202214022140222e-05,
      "loss": 0.00043325275182723997,
      "step": 204340
    },
    {
      "epoch": 58.004541583877376,
      "grad_norm": 0.021950123831629753,
      "learning_rate": 4.201930173147886e-05,
      "loss": 0.00016518235206604005,
      "step": 204350
    },
    {
      "epoch": 58.00738007380074,
      "grad_norm": 0.005392394959926605,
      "learning_rate": 4.201646324155549e-05,
      "loss": 0.0002975683659315109,
      "step": 204360
    },
    {
      "epoch": 58.0102185637241,
      "grad_norm": 0.7367399334907532,
      "learning_rate": 4.2013624751632134e-05,
      "loss": 0.00039396863430738447,
      "step": 204370
    },
    {
      "epoch": 58.01305705364746,
      "grad_norm": 2.967148542404175,
      "learning_rate": 4.2010786261708775e-05,
      "loss": 0.000678195059299469,
      "step": 204380
    },
    {
      "epoch": 58.01589554357082,
      "grad_norm": 0.06893350183963776,
      "learning_rate": 4.200794777178541e-05,
      "loss": 0.0004196334630250931,
      "step": 204390
    },
    {
      "epoch": 58.018734033494184,
      "grad_norm": 0.018050355836749077,
      "learning_rate": 4.200510928186205e-05,
      "loss": 0.0002667076885700226,
      "step": 204400
    },
    {
      "epoch": 58.02157252341754,
      "grad_norm": 0.020561840385198593,
      "learning_rate": 4.200227079193869e-05,
      "loss": 0.00018485207110643386,
      "step": 204410
    },
    {
      "epoch": 58.0244110133409,
      "grad_norm": 0.028353439643979073,
      "learning_rate": 4.199943230201533e-05,
      "loss": 0.0006221201270818711,
      "step": 204420
    },
    {
      "epoch": 58.027249503264265,
      "grad_norm": 0.05734207108616829,
      "learning_rate": 4.199659381209197e-05,
      "loss": 0.0006554437801241875,
      "step": 204430
    },
    {
      "epoch": 58.03008799318762,
      "grad_norm": 0.29836517572402954,
      "learning_rate": 4.199375532216861e-05,
      "loss": 0.0018628882244229317,
      "step": 204440
    },
    {
      "epoch": 58.032926483110984,
      "grad_norm": 0.016395173966884613,
      "learning_rate": 4.1990916832245245e-05,
      "loss": 0.0003655984997749329,
      "step": 204450
    },
    {
      "epoch": 58.03576497303435,
      "grad_norm": 0.13461102545261383,
      "learning_rate": 4.1988078342321886e-05,
      "loss": 0.0028987009078264236,
      "step": 204460
    },
    {
      "epoch": 58.03860346295771,
      "grad_norm": 0.06809957325458527,
      "learning_rate": 4.198523985239853e-05,
      "loss": 0.0012411508709192277,
      "step": 204470
    },
    {
      "epoch": 58.041441952881065,
      "grad_norm": 0.1384178102016449,
      "learning_rate": 4.198240136247517e-05,
      "loss": 0.0004287479445338249,
      "step": 204480
    },
    {
      "epoch": 58.04428044280443,
      "grad_norm": 0.10587082803249359,
      "learning_rate": 4.19795628725518e-05,
      "loss": 0.0002389878034591675,
      "step": 204490
    },
    {
      "epoch": 58.04711893272779,
      "grad_norm": 0.015492012724280357,
      "learning_rate": 4.1976724382628445e-05,
      "loss": 0.00026868730783462523,
      "step": 204500
    },
    {
      "epoch": 58.04711893272779,
      "eval_accuracy": 0.9804794302791378,
      "eval_loss": 0.07528763264417648,
      "eval_runtime": 32.8393,
      "eval_samples_per_second": 478.908,
      "eval_steps_per_second": 7.491,
      "step": 204500
    },
    {
      "epoch": 58.04995742265115,
      "grad_norm": 0.13865184783935547,
      "learning_rate": 4.1973885892705086e-05,
      "loss": 0.00037951022386550903,
      "step": 204510
    },
    {
      "epoch": 58.05279591257451,
      "grad_norm": 0.057282522320747375,
      "learning_rate": 4.197104740278172e-05,
      "loss": 0.0030770041048526764,
      "step": 204520
    },
    {
      "epoch": 58.05563440249787,
      "grad_norm": 0.16568686068058014,
      "learning_rate": 4.196820891285836e-05,
      "loss": 0.0013439642265439034,
      "step": 204530
    },
    {
      "epoch": 58.05847289242123,
      "grad_norm": 1.1781235933303833,
      "learning_rate": 4.1965370422935003e-05,
      "loss": 0.0007149338722229004,
      "step": 204540
    },
    {
      "epoch": 58.06131138234459,
      "grad_norm": 0.016168765723705292,
      "learning_rate": 4.196253193301164e-05,
      "loss": 0.000862187147140503,
      "step": 204550
    },
    {
      "epoch": 58.064149872267954,
      "grad_norm": 0.13774068653583527,
      "learning_rate": 4.195969344308828e-05,
      "loss": 0.0015854593366384506,
      "step": 204560
    },
    {
      "epoch": 58.06698836219132,
      "grad_norm": 0.108001209795475,
      "learning_rate": 4.195685495316492e-05,
      "loss": 0.00246198084205389,
      "step": 204570
    },
    {
      "epoch": 58.06982685211467,
      "grad_norm": 0.7792887091636658,
      "learning_rate": 4.1954016463241555e-05,
      "loss": 0.0006846090778708458,
      "step": 204580
    },
    {
      "epoch": 58.072665342038036,
      "grad_norm": 0.18153995275497437,
      "learning_rate": 4.1951461822310535e-05,
      "loss": 0.014819648861885071,
      "step": 204590
    },
    {
      "epoch": 58.0755038319614,
      "grad_norm": 0.06315401941537857,
      "learning_rate": 4.194862333238717e-05,
      "loss": 0.0008628163486719131,
      "step": 204600
    },
    {
      "epoch": 58.078342321884755,
      "grad_norm": 0.06517153233289719,
      "learning_rate": 4.194578484246381e-05,
      "loss": 0.000557304359972477,
      "step": 204610
    },
    {
      "epoch": 58.08118081180812,
      "grad_norm": 0.015062338672578335,
      "learning_rate": 4.194294635254045e-05,
      "loss": 0.0001638738438487053,
      "step": 204620
    },
    {
      "epoch": 58.08401930173148,
      "grad_norm": 0.061194922775030136,
      "learning_rate": 4.194010786261709e-05,
      "loss": 0.000349043495953083,
      "step": 204630
    },
    {
      "epoch": 58.086857791654836,
      "grad_norm": 0.01200471818447113,
      "learning_rate": 4.193726937269373e-05,
      "loss": 0.00016806256026029586,
      "step": 204640
    },
    {
      "epoch": 58.0896962815782,
      "grad_norm": 0.03657468035817146,
      "learning_rate": 4.193443088277037e-05,
      "loss": 0.00013396386057138444,
      "step": 204650
    },
    {
      "epoch": 58.09253477150156,
      "grad_norm": 0.47877076268196106,
      "learning_rate": 4.1931592392847005e-05,
      "loss": 0.0006796017289161683,
      "step": 204660
    },
    {
      "epoch": 58.095373261424925,
      "grad_norm": 0.008521954528987408,
      "learning_rate": 4.1928753902923646e-05,
      "loss": 0.00039926283061504365,
      "step": 204670
    },
    {
      "epoch": 58.09821175134828,
      "grad_norm": 0.013571600429713726,
      "learning_rate": 4.192591541300029e-05,
      "loss": 0.00044177062809467314,
      "step": 204680
    },
    {
      "epoch": 58.10105024127164,
      "grad_norm": 0.04802403971552849,
      "learning_rate": 4.192307692307693e-05,
      "loss": 0.0005490142852067947,
      "step": 204690
    },
    {
      "epoch": 58.103888731195006,
      "grad_norm": 0.2598487138748169,
      "learning_rate": 4.192023843315356e-05,
      "loss": 0.0001978311687707901,
      "step": 204700
    },
    {
      "epoch": 58.10672722111836,
      "grad_norm": 2.142902374267578,
      "learning_rate": 4.19173999432302e-05,
      "loss": 0.0009505357593297958,
      "step": 204710
    },
    {
      "epoch": 58.109565711041725,
      "grad_norm": 0.08950124680995941,
      "learning_rate": 4.1914561453306846e-05,
      "loss": 0.002841956354677677,
      "step": 204720
    },
    {
      "epoch": 58.11240420096509,
      "grad_norm": 0.6896215677261353,
      "learning_rate": 4.191172296338348e-05,
      "loss": 0.003446846827864647,
      "step": 204730
    },
    {
      "epoch": 58.11524269088845,
      "grad_norm": 0.09965222328901291,
      "learning_rate": 4.190888447346012e-05,
      "loss": 0.0014880875125527382,
      "step": 204740
    },
    {
      "epoch": 58.11808118081181,
      "grad_norm": 5.069331645965576,
      "learning_rate": 4.1906045983536763e-05,
      "loss": 0.008741626888513565,
      "step": 204750
    },
    {
      "epoch": 58.12091967073517,
      "grad_norm": 0.04064833000302315,
      "learning_rate": 4.19032074936134e-05,
      "loss": 0.0046473868191242215,
      "step": 204760
    },
    {
      "epoch": 58.12375816065853,
      "grad_norm": 2.1111674308776855,
      "learning_rate": 4.190036900369004e-05,
      "loss": 0.0010314088314771653,
      "step": 204770
    },
    {
      "epoch": 58.12659665058189,
      "grad_norm": 0.07537281513214111,
      "learning_rate": 4.189753051376668e-05,
      "loss": 0.002217673510313034,
      "step": 204780
    },
    {
      "epoch": 58.12943514050525,
      "grad_norm": 0.17979152500629425,
      "learning_rate": 4.1894692023843315e-05,
      "loss": 0.000832880474627018,
      "step": 204790
    },
    {
      "epoch": 58.132273630428614,
      "grad_norm": 0.05780456215143204,
      "learning_rate": 4.189185353391996e-05,
      "loss": 0.00038910023868083955,
      "step": 204800
    },
    {
      "epoch": 58.13511212035197,
      "grad_norm": 0.2759091258049011,
      "learning_rate": 4.188901504399659e-05,
      "loss": 0.0022787053138017656,
      "step": 204810
    },
    {
      "epoch": 58.13795061027533,
      "grad_norm": 1.2614411115646362,
      "learning_rate": 4.188617655407324e-05,
      "loss": 0.0036401398479938506,
      "step": 204820
    },
    {
      "epoch": 58.140789100198695,
      "grad_norm": 0.0448504202067852,
      "learning_rate": 4.1883338064149874e-05,
      "loss": 0.0014636456966400147,
      "step": 204830
    },
    {
      "epoch": 58.14362759012206,
      "grad_norm": 0.16074614226818085,
      "learning_rate": 4.188049957422651e-05,
      "loss": 0.0004017755389213562,
      "step": 204840
    },
    {
      "epoch": 58.146466080045414,
      "grad_norm": 0.5667877197265625,
      "learning_rate": 4.187766108430316e-05,
      "loss": 0.009945553541183472,
      "step": 204850
    },
    {
      "epoch": 58.14930456996878,
      "grad_norm": 0.09318536520004272,
      "learning_rate": 4.187482259437979e-05,
      "loss": 0.0011715713888406753,
      "step": 204860
    },
    {
      "epoch": 58.15214305989214,
      "grad_norm": 0.01850898191332817,
      "learning_rate": 4.187198410445643e-05,
      "loss": 0.0014135394245386123,
      "step": 204870
    },
    {
      "epoch": 58.154981549815496,
      "grad_norm": 0.018807750195264816,
      "learning_rate": 4.1869145614533074e-05,
      "loss": 0.00040181931108236314,
      "step": 204880
    },
    {
      "epoch": 58.15782003973886,
      "grad_norm": 11.1284761428833,
      "learning_rate": 4.186630712460971e-05,
      "loss": 0.004034848138689995,
      "step": 204890
    },
    {
      "epoch": 58.16065852966222,
      "grad_norm": 9.102705001831055,
      "learning_rate": 4.186346863468635e-05,
      "loss": 0.0016865378245711326,
      "step": 204900
    },
    {
      "epoch": 58.16349701958558,
      "grad_norm": 0.05338349565863609,
      "learning_rate": 4.1860630144762985e-05,
      "loss": 0.00025546513497829435,
      "step": 204910
    },
    {
      "epoch": 58.16633550950894,
      "grad_norm": 0.553736686706543,
      "learning_rate": 4.1857791654839626e-05,
      "loss": 0.0015936607494950294,
      "step": 204920
    },
    {
      "epoch": 58.1691739994323,
      "grad_norm": 6.255552291870117,
      "learning_rate": 4.185495316491627e-05,
      "loss": 0.0011901333928108215,
      "step": 204930
    },
    {
      "epoch": 58.172012489355666,
      "grad_norm": 1.2497267723083496,
      "learning_rate": 4.18521146749929e-05,
      "loss": 0.005077730864286423,
      "step": 204940
    },
    {
      "epoch": 58.17485097927902,
      "grad_norm": 2.7479703426361084,
      "learning_rate": 4.184927618506955e-05,
      "loss": 0.0009032720699906349,
      "step": 204950
    },
    {
      "epoch": 58.177689469202384,
      "grad_norm": 0.018021713942289352,
      "learning_rate": 4.1846437695146185e-05,
      "loss": 0.0004354001954197884,
      "step": 204960
    },
    {
      "epoch": 58.18052795912575,
      "grad_norm": 20.678144454956055,
      "learning_rate": 4.184359920522282e-05,
      "loss": 0.013803592324256897,
      "step": 204970
    },
    {
      "epoch": 58.1833664490491,
      "grad_norm": 0.168352410197258,
      "learning_rate": 4.184076071529947e-05,
      "loss": 0.008954645693302154,
      "step": 204980
    },
    {
      "epoch": 58.186204938972466,
      "grad_norm": 0.26497358083724976,
      "learning_rate": 4.18379222253761e-05,
      "loss": 0.0010895712301135063,
      "step": 204990
    },
    {
      "epoch": 58.18904342889583,
      "grad_norm": 0.031940143555402756,
      "learning_rate": 4.1835083735452744e-05,
      "loss": 0.0008628876879811287,
      "step": 205000
    },
    {
      "epoch": 58.18904342889583,
      "eval_accuracy": 0.9774273542315762,
      "eval_loss": 0.08469057828187943,
      "eval_runtime": 32.1376,
      "eval_samples_per_second": 489.364,
      "eval_steps_per_second": 7.655,
      "step": 205000
    },
    {
      "epoch": 58.191881918819185,
      "grad_norm": 0.5235498547554016,
      "learning_rate": 4.183224524552938e-05,
      "loss": 0.0013462038710713386,
      "step": 205010
    },
    {
      "epoch": 58.19472040874255,
      "grad_norm": 0.0942637175321579,
      "learning_rate": 4.182940675560602e-05,
      "loss": 0.0012478357180953025,
      "step": 205020
    },
    {
      "epoch": 58.19755889866591,
      "grad_norm": 0.21433603763580322,
      "learning_rate": 4.182656826568266e-05,
      "loss": 0.0010499902069568633,
      "step": 205030
    },
    {
      "epoch": 58.20039738858927,
      "grad_norm": 0.6372508406639099,
      "learning_rate": 4.1823729775759296e-05,
      "loss": 0.0014020510017871857,
      "step": 205040
    },
    {
      "epoch": 58.20323587851263,
      "grad_norm": 0.018086139112710953,
      "learning_rate": 4.182089128583594e-05,
      "loss": 0.00048404969274997713,
      "step": 205050
    },
    {
      "epoch": 58.20607436843599,
      "grad_norm": 0.1767113208770752,
      "learning_rate": 4.181805279591258e-05,
      "loss": 0.0003276534378528595,
      "step": 205060
    },
    {
      "epoch": 58.208912858359355,
      "grad_norm": 0.021839821711182594,
      "learning_rate": 4.181521430598921e-05,
      "loss": 0.0010433599352836608,
      "step": 205070
    },
    {
      "epoch": 58.21175134828271,
      "grad_norm": 1.037894368171692,
      "learning_rate": 4.1812375816065854e-05,
      "loss": 0.0004463814198970795,
      "step": 205080
    },
    {
      "epoch": 58.214589838206074,
      "grad_norm": 0.01604326069355011,
      "learning_rate": 4.1809537326142496e-05,
      "loss": 0.0005635779350996017,
      "step": 205090
    },
    {
      "epoch": 58.21742832812944,
      "grad_norm": 0.3372109830379486,
      "learning_rate": 4.180669883621913e-05,
      "loss": 0.0005124708637595177,
      "step": 205100
    },
    {
      "epoch": 58.2202668180528,
      "grad_norm": 0.35462090373039246,
      "learning_rate": 4.180386034629577e-05,
      "loss": 0.0010896431282162665,
      "step": 205110
    },
    {
      "epoch": 58.223105307976155,
      "grad_norm": 0.028064021840691566,
      "learning_rate": 4.180102185637241e-05,
      "loss": 0.003514605015516281,
      "step": 205120
    },
    {
      "epoch": 58.22594379789952,
      "grad_norm": 2.936357259750366,
      "learning_rate": 4.179818336644905e-05,
      "loss": 0.0016153039410710334,
      "step": 205130
    },
    {
      "epoch": 58.22878228782288,
      "grad_norm": 0.06186690554022789,
      "learning_rate": 4.179534487652569e-05,
      "loss": 0.0038416236639022827,
      "step": 205140
    },
    {
      "epoch": 58.23162077774624,
      "grad_norm": 0.152574822306633,
      "learning_rate": 4.179250638660233e-05,
      "loss": 0.00029898025095462797,
      "step": 205150
    },
    {
      "epoch": 58.2344592676696,
      "grad_norm": 0.05094078555703163,
      "learning_rate": 4.178966789667897e-05,
      "loss": 0.0002130495384335518,
      "step": 205160
    },
    {
      "epoch": 58.23729775759296,
      "grad_norm": 0.11865977197885513,
      "learning_rate": 4.1786829406755606e-05,
      "loss": 0.0003287043422460556,
      "step": 205170
    },
    {
      "epoch": 58.24013624751632,
      "grad_norm": 0.8597666621208191,
      "learning_rate": 4.178399091683225e-05,
      "loss": 0.0008318416774272918,
      "step": 205180
    },
    {
      "epoch": 58.24297473743968,
      "grad_norm": 0.016284247860312462,
      "learning_rate": 4.178115242690889e-05,
      "loss": 0.00032555852085351943,
      "step": 205190
    },
    {
      "epoch": 58.245813227363044,
      "grad_norm": 0.019429286941885948,
      "learning_rate": 4.1778313936985524e-05,
      "loss": 0.00021120253950357437,
      "step": 205200
    },
    {
      "epoch": 58.24865171728641,
      "grad_norm": 0.01344375405460596,
      "learning_rate": 4.1775475447062165e-05,
      "loss": 0.0006610799580812454,
      "step": 205210
    },
    {
      "epoch": 58.25149020720976,
      "grad_norm": 0.02961040660738945,
      "learning_rate": 4.1772636957138806e-05,
      "loss": 0.0002518216148018837,
      "step": 205220
    },
    {
      "epoch": 58.254328697133126,
      "grad_norm": 0.09081673622131348,
      "learning_rate": 4.176979846721544e-05,
      "loss": 0.00035782288759946825,
      "step": 205230
    },
    {
      "epoch": 58.25716718705649,
      "grad_norm": 0.09560229629278183,
      "learning_rate": 4.176695997729208e-05,
      "loss": 0.00015184488147497178,
      "step": 205240
    },
    {
      "epoch": 58.260005676979844,
      "grad_norm": 0.14729832112789154,
      "learning_rate": 4.1764121487368724e-05,
      "loss": 0.0004842573776841164,
      "step": 205250
    },
    {
      "epoch": 58.26284416690321,
      "grad_norm": 2.0042402744293213,
      "learning_rate": 4.176128299744536e-05,
      "loss": 0.0006337942555546761,
      "step": 205260
    },
    {
      "epoch": 58.26568265682657,
      "grad_norm": 0.011314044706523418,
      "learning_rate": 4.1758444507522e-05,
      "loss": 0.0039548899978399275,
      "step": 205270
    },
    {
      "epoch": 58.268521146749926,
      "grad_norm": 3.559664487838745,
      "learning_rate": 4.175560601759864e-05,
      "loss": 0.0007200537249445916,
      "step": 205280
    },
    {
      "epoch": 58.27135963667329,
      "grad_norm": 0.11831052601337433,
      "learning_rate": 4.175276752767528e-05,
      "loss": 0.00027854051440954206,
      "step": 205290
    },
    {
      "epoch": 58.27419812659665,
      "grad_norm": 0.0453171543776989,
      "learning_rate": 4.174992903775192e-05,
      "loss": 0.0001888323575258255,
      "step": 205300
    },
    {
      "epoch": 58.277036616520014,
      "grad_norm": 0.044095445424318314,
      "learning_rate": 4.174709054782855e-05,
      "loss": 0.00043348148465156556,
      "step": 205310
    },
    {
      "epoch": 58.27987510644337,
      "grad_norm": 0.19925419986248016,
      "learning_rate": 4.17442520579052e-05,
      "loss": 0.0024681663140654563,
      "step": 205320
    },
    {
      "epoch": 58.28271359636673,
      "grad_norm": 0.03999914228916168,
      "learning_rate": 4.1741413567981834e-05,
      "loss": 0.00012504421174526216,
      "step": 205330
    },
    {
      "epoch": 58.285552086290096,
      "grad_norm": 7.908804416656494,
      "learning_rate": 4.1738575078058476e-05,
      "loss": 0.0012836307287216187,
      "step": 205340
    },
    {
      "epoch": 58.28839057621345,
      "grad_norm": 0.24238896369934082,
      "learning_rate": 4.173573658813512e-05,
      "loss": 0.00035486966371536257,
      "step": 205350
    },
    {
      "epoch": 58.291229066136815,
      "grad_norm": 1.3363550901412964,
      "learning_rate": 4.173289809821175e-05,
      "loss": 0.00031757280230522157,
      "step": 205360
    },
    {
      "epoch": 58.29406755606018,
      "grad_norm": 0.18619368970394135,
      "learning_rate": 4.173005960828839e-05,
      "loss": 0.00023713260889053346,
      "step": 205370
    },
    {
      "epoch": 58.29690604598353,
      "grad_norm": 0.1024603545665741,
      "learning_rate": 4.1727221118365035e-05,
      "loss": 0.0004784869030117989,
      "step": 205380
    },
    {
      "epoch": 58.299744535906896,
      "grad_norm": 0.03788735717535019,
      "learning_rate": 4.172438262844167e-05,
      "loss": 0.00025156624615192414,
      "step": 205390
    },
    {
      "epoch": 58.30258302583026,
      "grad_norm": 0.14736196398735046,
      "learning_rate": 4.172154413851831e-05,
      "loss": 0.00027147121727466583,
      "step": 205400
    },
    {
      "epoch": 58.30542151575362,
      "grad_norm": 0.012785141356289387,
      "learning_rate": 4.1718705648594945e-05,
      "loss": 0.00024437569081783297,
      "step": 205410
    },
    {
      "epoch": 58.30826000567698,
      "grad_norm": 0.0817706435918808,
      "learning_rate": 4.1715867158671587e-05,
      "loss": 0.0004734601825475693,
      "step": 205420
    },
    {
      "epoch": 58.31109849560034,
      "grad_norm": 0.06922005861997604,
      "learning_rate": 4.171302866874823e-05,
      "loss": 0.004749102890491486,
      "step": 205430
    },
    {
      "epoch": 58.3139369855237,
      "grad_norm": 2.3305904865264893,
      "learning_rate": 4.171019017882486e-05,
      "loss": 0.0017081452533602715,
      "step": 205440
    },
    {
      "epoch": 58.31677547544706,
      "grad_norm": 0.13234132528305054,
      "learning_rate": 4.170735168890151e-05,
      "loss": 0.00035175904631614687,
      "step": 205450
    },
    {
      "epoch": 58.31961396537042,
      "grad_norm": 0.42978307604789734,
      "learning_rate": 4.1704513198978145e-05,
      "loss": 0.0003889555111527443,
      "step": 205460
    },
    {
      "epoch": 58.322452455293785,
      "grad_norm": 0.9656371474266052,
      "learning_rate": 4.1701674709054787e-05,
      "loss": 0.00044166650623083114,
      "step": 205470
    },
    {
      "epoch": 58.32529094521715,
      "grad_norm": 0.11956121027469635,
      "learning_rate": 4.169883621913143e-05,
      "loss": 0.0003294184803962708,
      "step": 205480
    },
    {
      "epoch": 58.328129435140504,
      "grad_norm": 0.5444433093070984,
      "learning_rate": 4.169599772920806e-05,
      "loss": 0.0005915256217122078,
      "step": 205490
    },
    {
      "epoch": 58.33096792506387,
      "grad_norm": 0.10612577944993973,
      "learning_rate": 4.1693159239284704e-05,
      "loss": 0.00018099993467330934,
      "step": 205500
    },
    {
      "epoch": 58.33096792506387,
      "eval_accuracy": 0.9802250906085077,
      "eval_loss": 0.0739494189620018,
      "eval_runtime": 32.3586,
      "eval_samples_per_second": 486.023,
      "eval_steps_per_second": 7.602,
      "step": 205500
    },
    {
      "epoch": 58.33380641498723,
      "grad_norm": 0.11806458979845047,
      "learning_rate": 4.169032074936134e-05,
      "loss": 0.0002736002206802368,
      "step": 205510
    },
    {
      "epoch": 58.336644904910585,
      "grad_norm": 0.055456411093473434,
      "learning_rate": 4.168748225943798e-05,
      "loss": 0.0002252276986837387,
      "step": 205520
    },
    {
      "epoch": 58.33948339483395,
      "grad_norm": 0.028017638251185417,
      "learning_rate": 4.168464376951462e-05,
      "loss": 0.0003796648234128952,
      "step": 205530
    },
    {
      "epoch": 58.34232188475731,
      "grad_norm": 0.022682668641209602,
      "learning_rate": 4.1681805279591256e-05,
      "loss": 0.00015981532633304595,
      "step": 205540
    },
    {
      "epoch": 58.34516037468067,
      "grad_norm": 0.13965179026126862,
      "learning_rate": 4.16789667896679e-05,
      "loss": 0.0004375981166958809,
      "step": 205550
    },
    {
      "epoch": 58.34799886460403,
      "grad_norm": 0.03178500384092331,
      "learning_rate": 4.167612829974454e-05,
      "loss": 0.0001471705734729767,
      "step": 205560
    },
    {
      "epoch": 58.35083735452739,
      "grad_norm": 0.10785835981369019,
      "learning_rate": 4.167328980982117e-05,
      "loss": 0.0002951962873339653,
      "step": 205570
    },
    {
      "epoch": 58.353675844450756,
      "grad_norm": 0.06760025024414062,
      "learning_rate": 4.167045131989782e-05,
      "loss": 0.0003511713817715645,
      "step": 205580
    },
    {
      "epoch": 58.35651433437411,
      "grad_norm": 0.029207304120063782,
      "learning_rate": 4.1667612829974456e-05,
      "loss": 0.00028275251388549806,
      "step": 205590
    },
    {
      "epoch": 58.359352824297474,
      "grad_norm": 0.022507071495056152,
      "learning_rate": 4.166477434005109e-05,
      "loss": 0.0007560320198535919,
      "step": 205600
    },
    {
      "epoch": 58.36219131422084,
      "grad_norm": 0.035710081458091736,
      "learning_rate": 4.166193585012774e-05,
      "loss": 0.0005072861909866333,
      "step": 205610
    },
    {
      "epoch": 58.36502980414419,
      "grad_norm": 0.13027210533618927,
      "learning_rate": 4.165909736020437e-05,
      "loss": 0.001174384169280529,
      "step": 205620
    },
    {
      "epoch": 58.367868294067556,
      "grad_norm": 0.012850498780608177,
      "learning_rate": 4.1656258870281015e-05,
      "loss": 0.0013997212052345275,
      "step": 205630
    },
    {
      "epoch": 58.37070678399092,
      "grad_norm": 0.0756896585226059,
      "learning_rate": 4.165342038035765e-05,
      "loss": 0.0005036361515522003,
      "step": 205640
    },
    {
      "epoch": 58.373545273914274,
      "grad_norm": 0.032071687281131744,
      "learning_rate": 4.165058189043429e-05,
      "loss": 0.0006042823195457458,
      "step": 205650
    },
    {
      "epoch": 58.37638376383764,
      "grad_norm": 0.15187151730060577,
      "learning_rate": 4.164774340051093e-05,
      "loss": 0.0006724942475557327,
      "step": 205660
    },
    {
      "epoch": 58.379222253761,
      "grad_norm": 3.1528849601745605,
      "learning_rate": 4.164490491058757e-05,
      "loss": 0.0008101312443614006,
      "step": 205670
    },
    {
      "epoch": 58.38206074368436,
      "grad_norm": 4.844971179962158,
      "learning_rate": 4.164206642066421e-05,
      "loss": 0.0017693787813186646,
      "step": 205680
    },
    {
      "epoch": 58.38489923360772,
      "grad_norm": 0.1397276371717453,
      "learning_rate": 4.163922793074085e-05,
      "loss": 0.00033976081758737566,
      "step": 205690
    },
    {
      "epoch": 58.38773772353108,
      "grad_norm": 0.10120905935764313,
      "learning_rate": 4.1636389440817484e-05,
      "loss": 0.0020681345835328103,
      "step": 205700
    },
    {
      "epoch": 58.390576213454445,
      "grad_norm": 0.04216722398996353,
      "learning_rate": 4.163355095089413e-05,
      "loss": 0.0001390049234032631,
      "step": 205710
    },
    {
      "epoch": 58.3934147033778,
      "grad_norm": 0.3242674171924591,
      "learning_rate": 4.163071246097077e-05,
      "loss": 0.00020927637815475463,
      "step": 205720
    },
    {
      "epoch": 58.39625319330116,
      "grad_norm": 0.007905598729848862,
      "learning_rate": 4.16278739710474e-05,
      "loss": 0.00011495891958475113,
      "step": 205730
    },
    {
      "epoch": 58.399091683224526,
      "grad_norm": 3.496248722076416,
      "learning_rate": 4.162503548112404e-05,
      "loss": 0.0007126154378056526,
      "step": 205740
    },
    {
      "epoch": 58.40193017314788,
      "grad_norm": 4.174905776977539,
      "learning_rate": 4.1622196991200684e-05,
      "loss": 0.0008264940232038497,
      "step": 205750
    },
    {
      "epoch": 58.404768663071245,
      "grad_norm": 2.5469467639923096,
      "learning_rate": 4.1619358501277325e-05,
      "loss": 0.0030883658677339555,
      "step": 205760
    },
    {
      "epoch": 58.40760715299461,
      "grad_norm": 0.8082329630851746,
      "learning_rate": 4.161652001135396e-05,
      "loss": 0.00023912861943244934,
      "step": 205770
    },
    {
      "epoch": 58.41044564291797,
      "grad_norm": 0.10769607126712799,
      "learning_rate": 4.16136815214306e-05,
      "loss": 0.0013112364336848258,
      "step": 205780
    },
    {
      "epoch": 58.413284132841326,
      "grad_norm": 0.06562838703393936,
      "learning_rate": 4.161084303150724e-05,
      "loss": 0.00045734159648418425,
      "step": 205790
    },
    {
      "epoch": 58.41612262276469,
      "grad_norm": 0.0805530771613121,
      "learning_rate": 4.160800454158388e-05,
      "loss": 0.00018837898969650268,
      "step": 205800
    },
    {
      "epoch": 58.41896111268805,
      "grad_norm": 0.6682500243186951,
      "learning_rate": 4.160516605166052e-05,
      "loss": 0.00035232454538345335,
      "step": 205810
    },
    {
      "epoch": 58.42179960261141,
      "grad_norm": 0.05829198285937309,
      "learning_rate": 4.160232756173716e-05,
      "loss": 0.00018559452146291733,
      "step": 205820
    },
    {
      "epoch": 58.42463809253477,
      "grad_norm": 0.01971971057355404,
      "learning_rate": 4.1599489071813795e-05,
      "loss": 0.003474518656730652,
      "step": 205830
    },
    {
      "epoch": 58.427476582458134,
      "grad_norm": 0.05474422872066498,
      "learning_rate": 4.1596650581890436e-05,
      "loss": 0.0005648896098136902,
      "step": 205840
    },
    {
      "epoch": 58.43031507238149,
      "grad_norm": 0.0166428592056036,
      "learning_rate": 4.159381209196708e-05,
      "loss": 0.00039085801690816877,
      "step": 205850
    },
    {
      "epoch": 58.43315356230485,
      "grad_norm": 0.21890917420387268,
      "learning_rate": 4.159097360204371e-05,
      "loss": 0.0002089180052280426,
      "step": 205860
    },
    {
      "epoch": 58.435992052228215,
      "grad_norm": 0.011787100695073605,
      "learning_rate": 4.1588135112120354e-05,
      "loss": 0.00021103490144014359,
      "step": 205870
    },
    {
      "epoch": 58.43883054215158,
      "grad_norm": 0.003612253349274397,
      "learning_rate": 4.1585296622196995e-05,
      "loss": 0.0001603182405233383,
      "step": 205880
    },
    {
      "epoch": 58.441669032074934,
      "grad_norm": 0.6743913888931274,
      "learning_rate": 4.158245813227363e-05,
      "loss": 0.0005043154582381248,
      "step": 205890
    },
    {
      "epoch": 58.4445075219983,
      "grad_norm": 0.3677307069301605,
      "learning_rate": 4.157961964235027e-05,
      "loss": 0.0008349336683750152,
      "step": 205900
    },
    {
      "epoch": 58.44734601192166,
      "grad_norm": 0.049873169511556625,
      "learning_rate": 4.157678115242691e-05,
      "loss": 0.0002532351762056351,
      "step": 205910
    },
    {
      "epoch": 58.450184501845015,
      "grad_norm": 0.04382982850074768,
      "learning_rate": 4.1573942662503554e-05,
      "loss": 0.0004124777391552925,
      "step": 205920
    },
    {
      "epoch": 58.45302299176838,
      "grad_norm": 0.07203242927789688,
      "learning_rate": 4.157110417258019e-05,
      "loss": 0.00010791867971420288,
      "step": 205930
    },
    {
      "epoch": 58.45586148169174,
      "grad_norm": 0.1392454355955124,
      "learning_rate": 4.156826568265683e-05,
      "loss": 0.005626077204942704,
      "step": 205940
    },
    {
      "epoch": 58.458699971615104,
      "grad_norm": 0.06474582850933075,
      "learning_rate": 4.156542719273347e-05,
      "loss": 0.0004236608743667603,
      "step": 205950
    },
    {
      "epoch": 58.46153846153846,
      "grad_norm": 0.0070573994889855385,
      "learning_rate": 4.1562588702810106e-05,
      "loss": 0.0006157249212265014,
      "step": 205960
    },
    {
      "epoch": 58.46437695146182,
      "grad_norm": 0.012061268091201782,
      "learning_rate": 4.155975021288675e-05,
      "loss": 0.00033986270427703856,
      "step": 205970
    },
    {
      "epoch": 58.467215441385186,
      "grad_norm": 0.023371854797005653,
      "learning_rate": 4.155691172296339e-05,
      "loss": 0.0008606761693954468,
      "step": 205980
    },
    {
      "epoch": 58.47005393130854,
      "grad_norm": 7.298279762268066,
      "learning_rate": 4.155407323304002e-05,
      "loss": 0.00149534922093153,
      "step": 205990
    },
    {
      "epoch": 58.472892421231904,
      "grad_norm": 4.245069980621338,
      "learning_rate": 4.1551234743116664e-05,
      "loss": 0.001015925221145153,
      "step": 206000
    },
    {
      "epoch": 58.472892421231904,
      "eval_accuracy": 0.9803522604438227,
      "eval_loss": 0.07817874848842621,
      "eval_runtime": 32.6958,
      "eval_samples_per_second": 481.01,
      "eval_steps_per_second": 7.524,
      "step": 206000
    },
    {
      "epoch": 58.47573091115527,
      "grad_norm": 0.031011363491415977,
      "learning_rate": 4.1548396253193306e-05,
      "loss": 0.0005143677815794944,
      "step": 206010
    },
    {
      "epoch": 58.47856940107862,
      "grad_norm": 0.17171825468540192,
      "learning_rate": 4.154555776326994e-05,
      "loss": 0.000588729977607727,
      "step": 206020
    },
    {
      "epoch": 58.481407891001986,
      "grad_norm": 0.2657344937324524,
      "learning_rate": 4.154271927334658e-05,
      "loss": 0.00047618672251701353,
      "step": 206030
    },
    {
      "epoch": 58.48424638092535,
      "grad_norm": 0.01347420085221529,
      "learning_rate": 4.1539880783423216e-05,
      "loss": 0.0002003578469157219,
      "step": 206040
    },
    {
      "epoch": 58.48708487084871,
      "grad_norm": 0.01635129004716873,
      "learning_rate": 4.1537042293499864e-05,
      "loss": 0.00023457687348127365,
      "step": 206050
    },
    {
      "epoch": 58.48992336077207,
      "grad_norm": 0.07479766756296158,
      "learning_rate": 4.15342038035765e-05,
      "loss": 0.00024643857032060625,
      "step": 206060
    },
    {
      "epoch": 58.49276185069543,
      "grad_norm": 1.0512486696243286,
      "learning_rate": 4.1531365313653134e-05,
      "loss": 0.013239280879497528,
      "step": 206070
    },
    {
      "epoch": 58.49560034061879,
      "grad_norm": 0.030059413984417915,
      "learning_rate": 4.152852682372978e-05,
      "loss": 0.0011805953457951547,
      "step": 206080
    },
    {
      "epoch": 58.49843883054215,
      "grad_norm": 0.0244869627058506,
      "learning_rate": 4.1525688333806416e-05,
      "loss": 0.002078399434685707,
      "step": 206090
    },
    {
      "epoch": 58.50127732046551,
      "grad_norm": 0.2963099181652069,
      "learning_rate": 4.152284984388306e-05,
      "loss": 0.005688820779323578,
      "step": 206100
    },
    {
      "epoch": 58.504115810388875,
      "grad_norm": 0.01306278258562088,
      "learning_rate": 4.15200113539597e-05,
      "loss": 0.00016800854355096816,
      "step": 206110
    },
    {
      "epoch": 58.50695430031223,
      "grad_norm": 0.024720774963498116,
      "learning_rate": 4.1517172864036334e-05,
      "loss": 0.00040089040994644164,
      "step": 206120
    },
    {
      "epoch": 58.50979279023559,
      "grad_norm": 0.06448198854923248,
      "learning_rate": 4.1514334374112975e-05,
      "loss": 0.00034239459782838824,
      "step": 206130
    },
    {
      "epoch": 58.512631280158956,
      "grad_norm": 0.26408785581588745,
      "learning_rate": 4.151149588418961e-05,
      "loss": 0.0008199870586395264,
      "step": 206140
    },
    {
      "epoch": 58.51546977008232,
      "grad_norm": 0.04131561890244484,
      "learning_rate": 4.150865739426625e-05,
      "loss": 0.00023290663957595826,
      "step": 206150
    },
    {
      "epoch": 58.518308260005675,
      "grad_norm": 0.02171259932219982,
      "learning_rate": 4.150581890434289e-05,
      "loss": 0.00013242699205875396,
      "step": 206160
    },
    {
      "epoch": 58.52114674992904,
      "grad_norm": 0.06444291770458221,
      "learning_rate": 4.150298041441953e-05,
      "loss": 0.0002375718206167221,
      "step": 206170
    },
    {
      "epoch": 58.5239852398524,
      "grad_norm": 0.0379272997379303,
      "learning_rate": 4.1500141924496175e-05,
      "loss": 0.0002611646428704262,
      "step": 206180
    },
    {
      "epoch": 58.52682372977576,
      "grad_norm": 0.03548593819141388,
      "learning_rate": 4.149730343457281e-05,
      "loss": 0.00016426313668489457,
      "step": 206190
    },
    {
      "epoch": 58.52966221969912,
      "grad_norm": 0.08589109778404236,
      "learning_rate": 4.1494464944649444e-05,
      "loss": 0.0004239514470100403,
      "step": 206200
    },
    {
      "epoch": 58.53250070962248,
      "grad_norm": 0.5127556324005127,
      "learning_rate": 4.149162645472609e-05,
      "loss": 0.0003514504060149193,
      "step": 206210
    },
    {
      "epoch": 58.535339199545845,
      "grad_norm": 0.42057403922080994,
      "learning_rate": 4.148878796480273e-05,
      "loss": 0.001057792454957962,
      "step": 206220
    },
    {
      "epoch": 58.5381776894692,
      "grad_norm": 0.15266260504722595,
      "learning_rate": 4.148594947487937e-05,
      "loss": 0.00032053161412477494,
      "step": 206230
    },
    {
      "epoch": 58.541016179392564,
      "grad_norm": 0.06631021201610565,
      "learning_rate": 4.1483110984956e-05,
      "loss": 0.0017901424318552018,
      "step": 206240
    },
    {
      "epoch": 58.54385466931593,
      "grad_norm": 0.16495361924171448,
      "learning_rate": 4.1480272495032644e-05,
      "loss": 0.00021556876599788667,
      "step": 206250
    },
    {
      "epoch": 58.54669315923928,
      "grad_norm": 0.27234265208244324,
      "learning_rate": 4.1477434005109286e-05,
      "loss": 0.00028309207409620284,
      "step": 206260
    },
    {
      "epoch": 58.549531649162645,
      "grad_norm": 0.02861521951854229,
      "learning_rate": 4.147459551518592e-05,
      "loss": 0.00024026613682508468,
      "step": 206270
    },
    {
      "epoch": 58.55237013908601,
      "grad_norm": 0.06954305619001389,
      "learning_rate": 4.147175702526256e-05,
      "loss": 0.0005324840545654297,
      "step": 206280
    },
    {
      "epoch": 58.555208629009364,
      "grad_norm": 0.07461442053318024,
      "learning_rate": 4.14689185353392e-05,
      "loss": 0.000377296656370163,
      "step": 206290
    },
    {
      "epoch": 58.55804711893273,
      "grad_norm": 0.19013510644435883,
      "learning_rate": 4.146608004541584e-05,
      "loss": 0.00018071476370096207,
      "step": 206300
    },
    {
      "epoch": 58.56088560885609,
      "grad_norm": 0.0616844967007637,
      "learning_rate": 4.146324155549248e-05,
      "loss": 0.00022306963801383973,
      "step": 206310
    },
    {
      "epoch": 58.56372409877945,
      "grad_norm": 0.00797318946570158,
      "learning_rate": 4.146040306556912e-05,
      "loss": 0.000749894417822361,
      "step": 206320
    },
    {
      "epoch": 58.56656258870281,
      "grad_norm": 4.22609281539917,
      "learning_rate": 4.1457564575645755e-05,
      "loss": 0.0006284125149250031,
      "step": 206330
    },
    {
      "epoch": 58.56940107862617,
      "grad_norm": 0.03797655552625656,
      "learning_rate": 4.1454726085722396e-05,
      "loss": 0.00021169427782297134,
      "step": 206340
    },
    {
      "epoch": 58.572239568549534,
      "grad_norm": 0.06143837049603462,
      "learning_rate": 4.145188759579904e-05,
      "loss": 0.001446019299328327,
      "step": 206350
    },
    {
      "epoch": 58.57507805847289,
      "grad_norm": 12.631304740905762,
      "learning_rate": 4.144904910587567e-05,
      "loss": 0.003256729245185852,
      "step": 206360
    },
    {
      "epoch": 58.57791654839625,
      "grad_norm": 0.3235250413417816,
      "learning_rate": 4.1446210615952314e-05,
      "loss": 0.0018879910930991173,
      "step": 206370
    },
    {
      "epoch": 58.580755038319616,
      "grad_norm": 0.03419598191976547,
      "learning_rate": 4.1443372126028955e-05,
      "loss": 0.0009353000670671463,
      "step": 206380
    },
    {
      "epoch": 58.58359352824297,
      "grad_norm": 0.04226769879460335,
      "learning_rate": 4.1440533636105597e-05,
      "loss": 0.0005440801382064819,
      "step": 206390
    },
    {
      "epoch": 58.586432018166335,
      "grad_norm": 0.029666220769286156,
      "learning_rate": 4.143769514618223e-05,
      "loss": 0.001789804734289646,
      "step": 206400
    },
    {
      "epoch": 58.5892705080897,
      "grad_norm": 0.5707429051399231,
      "learning_rate": 4.143485665625887e-05,
      "loss": 0.0008113857358694077,
      "step": 206410
    },
    {
      "epoch": 58.59210899801306,
      "grad_norm": 0.03844762220978737,
      "learning_rate": 4.1432018166335514e-05,
      "loss": 0.00048819109797477723,
      "step": 206420
    },
    {
      "epoch": 58.594947487936416,
      "grad_norm": 1.8750592470169067,
      "learning_rate": 4.142917967641215e-05,
      "loss": 0.004617907106876373,
      "step": 206430
    },
    {
      "epoch": 58.59778597785978,
      "grad_norm": 3.6845645904541016,
      "learning_rate": 4.142634118648879e-05,
      "loss": 0.0020926138386130333,
      "step": 206440
    },
    {
      "epoch": 58.60062446778314,
      "grad_norm": 0.3527293801307678,
      "learning_rate": 4.142350269656543e-05,
      "loss": 0.0016814867034554482,
      "step": 206450
    },
    {
      "epoch": 58.6034629577065,
      "grad_norm": 0.49289071559906006,
      "learning_rate": 4.1420664206642066e-05,
      "loss": 0.0007773702964186668,
      "step": 206460
    },
    {
      "epoch": 58.60630144762986,
      "grad_norm": 0.5054125189781189,
      "learning_rate": 4.141782571671871e-05,
      "loss": 0.0024444730952382088,
      "step": 206470
    },
    {
      "epoch": 58.60913993755322,
      "grad_norm": 0.2093491554260254,
      "learning_rate": 4.141527107578768e-05,
      "loss": 0.008703501522541046,
      "step": 206480
    },
    {
      "epoch": 58.61197842747658,
      "grad_norm": 0.044851645827293396,
      "learning_rate": 4.141243258586432e-05,
      "loss": 0.001154966838657856,
      "step": 206490
    },
    {
      "epoch": 58.61481691739994,
      "grad_norm": 0.12276975810527802,
      "learning_rate": 4.140959409594096e-05,
      "loss": 0.0002062743529677391,
      "step": 206500
    },
    {
      "epoch": 58.61481691739994,
      "eval_accuracy": 0.9754562217841928,
      "eval_loss": 0.09584749490022659,
      "eval_runtime": 33.2464,
      "eval_samples_per_second": 473.043,
      "eval_steps_per_second": 7.399,
      "step": 206500
    },
    {
      "epoch": 58.617655407323305,
      "grad_norm": 0.6599223017692566,
      "learning_rate": 4.14067556060176e-05,
      "loss": 0.0009128043428063392,
      "step": 206510
    },
    {
      "epoch": 58.62049389724667,
      "grad_norm": 0.08371642231941223,
      "learning_rate": 4.140391711609424e-05,
      "loss": 0.0018699953332543373,
      "step": 206520
    },
    {
      "epoch": 58.623332387170024,
      "grad_norm": 0.1008375883102417,
      "learning_rate": 4.140107862617088e-05,
      "loss": 0.002126074209809303,
      "step": 206530
    },
    {
      "epoch": 58.62617087709339,
      "grad_norm": 0.07552443444728851,
      "learning_rate": 4.1398240136247515e-05,
      "loss": 0.0011645831167697907,
      "step": 206540
    },
    {
      "epoch": 58.62900936701675,
      "grad_norm": 0.17458155751228333,
      "learning_rate": 4.1395401646324156e-05,
      "loss": 0.0004091557115316391,
      "step": 206550
    },
    {
      "epoch": 58.631847856940105,
      "grad_norm": 2.5185909271240234,
      "learning_rate": 4.13925631564008e-05,
      "loss": 0.0011567596346139907,
      "step": 206560
    },
    {
      "epoch": 58.63468634686347,
      "grad_norm": 0.3780214190483093,
      "learning_rate": 4.138972466647744e-05,
      "loss": 0.0012123353779315948,
      "step": 206570
    },
    {
      "epoch": 58.63752483678683,
      "grad_norm": 0.23872458934783936,
      "learning_rate": 4.1386886176554074e-05,
      "loss": 0.00043185334652662277,
      "step": 206580
    },
    {
      "epoch": 58.64036332671019,
      "grad_norm": 0.057355623692274094,
      "learning_rate": 4.1384047686630715e-05,
      "loss": 0.0009125780314207077,
      "step": 206590
    },
    {
      "epoch": 58.64320181663355,
      "grad_norm": 0.19321228563785553,
      "learning_rate": 4.1381209196707357e-05,
      "loss": 0.001856425404548645,
      "step": 206600
    },
    {
      "epoch": 58.64604030655691,
      "grad_norm": 4.119048118591309,
      "learning_rate": 4.137837070678399e-05,
      "loss": 0.003102562204003334,
      "step": 206610
    },
    {
      "epoch": 58.648878796480275,
      "grad_norm": 0.23236839473247528,
      "learning_rate": 4.137553221686063e-05,
      "loss": 0.0013714956119656562,
      "step": 206620
    },
    {
      "epoch": 58.65171728640363,
      "grad_norm": 0.09808182716369629,
      "learning_rate": 4.1372693726937274e-05,
      "loss": 0.0005684921517968178,
      "step": 206630
    },
    {
      "epoch": 58.654555776326994,
      "grad_norm": 0.29362431168556213,
      "learning_rate": 4.136985523701391e-05,
      "loss": 0.0020220573991537095,
      "step": 206640
    },
    {
      "epoch": 58.65739426625036,
      "grad_norm": 0.3885773718357086,
      "learning_rate": 4.136701674709055e-05,
      "loss": 0.001153378374874592,
      "step": 206650
    },
    {
      "epoch": 58.66023275617371,
      "grad_norm": 0.18055403232574463,
      "learning_rate": 4.136417825716719e-05,
      "loss": 0.0007753573358058929,
      "step": 206660
    },
    {
      "epoch": 58.663071246097076,
      "grad_norm": 0.009069940075278282,
      "learning_rate": 4.1361339767243826e-05,
      "loss": 0.0004912793636322021,
      "step": 206670
    },
    {
      "epoch": 58.66590973602044,
      "grad_norm": 0.103178471326828,
      "learning_rate": 4.135850127732047e-05,
      "loss": 0.001505832001566887,
      "step": 206680
    },
    {
      "epoch": 58.6687482259438,
      "grad_norm": 0.06005262956023216,
      "learning_rate": 4.135566278739711e-05,
      "loss": 0.00049851443618536,
      "step": 206690
    },
    {
      "epoch": 58.67158671586716,
      "grad_norm": 0.020493220537900925,
      "learning_rate": 4.135282429747375e-05,
      "loss": 0.00025725141167640684,
      "step": 206700
    },
    {
      "epoch": 58.67442520579052,
      "grad_norm": 0.03973439708352089,
      "learning_rate": 4.1349985807550385e-05,
      "loss": 0.0012383138760924338,
      "step": 206710
    },
    {
      "epoch": 58.67726369571388,
      "grad_norm": 0.2591654062271118,
      "learning_rate": 4.134714731762702e-05,
      "loss": 0.00027418825775384905,
      "step": 206720
    },
    {
      "epoch": 58.68010218563724,
      "grad_norm": 9.220125198364258,
      "learning_rate": 4.134430882770367e-05,
      "loss": 0.002757217362523079,
      "step": 206730
    },
    {
      "epoch": 58.6829406755606,
      "grad_norm": 2.395211696624756,
      "learning_rate": 4.13414703377803e-05,
      "loss": 0.0012048156931996346,
      "step": 206740
    },
    {
      "epoch": 58.685779165483964,
      "grad_norm": 0.02052822895348072,
      "learning_rate": 4.133863184785694e-05,
      "loss": 0.0004437889903783798,
      "step": 206750
    },
    {
      "epoch": 58.68861765540732,
      "grad_norm": 0.3433036804199219,
      "learning_rate": 4.1335793357933585e-05,
      "loss": 0.0005371766164898872,
      "step": 206760
    },
    {
      "epoch": 58.69145614533068,
      "grad_norm": 0.16759438812732697,
      "learning_rate": 4.133295486801022e-05,
      "loss": 0.0006929744035005569,
      "step": 206770
    },
    {
      "epoch": 58.694294635254046,
      "grad_norm": 0.10425325483083725,
      "learning_rate": 4.133011637808686e-05,
      "loss": 0.0022274473682045936,
      "step": 206780
    },
    {
      "epoch": 58.69713312517741,
      "grad_norm": 0.00831339880824089,
      "learning_rate": 4.13272778881635e-05,
      "loss": 0.005134113878011703,
      "step": 206790
    },
    {
      "epoch": 58.699971615100765,
      "grad_norm": 0.04423481225967407,
      "learning_rate": 4.132443939824014e-05,
      "loss": 0.00044206548482179643,
      "step": 206800
    },
    {
      "epoch": 58.70281010502413,
      "grad_norm": 0.406277060508728,
      "learning_rate": 4.132160090831678e-05,
      "loss": 0.000527748093008995,
      "step": 206810
    },
    {
      "epoch": 58.70564859494749,
      "grad_norm": 0.041793614625930786,
      "learning_rate": 4.131876241839341e-05,
      "loss": 0.00043745264410972596,
      "step": 206820
    },
    {
      "epoch": 58.708487084870846,
      "grad_norm": 0.015254014171659946,
      "learning_rate": 4.1315923928470054e-05,
      "loss": 0.0012932082638144494,
      "step": 206830
    },
    {
      "epoch": 58.71132557479421,
      "grad_norm": 0.6041697859764099,
      "learning_rate": 4.1313085438546695e-05,
      "loss": 0.0015688037499785423,
      "step": 206840
    },
    {
      "epoch": 58.71416406471757,
      "grad_norm": 0.26576244831085205,
      "learning_rate": 4.131024694862333e-05,
      "loss": 0.0005255449563264847,
      "step": 206850
    },
    {
      "epoch": 58.71700255464093,
      "grad_norm": 0.015513017773628235,
      "learning_rate": 4.130740845869998e-05,
      "loss": 0.0026957720518112184,
      "step": 206860
    },
    {
      "epoch": 58.71984104456429,
      "grad_norm": 0.3130224645137787,
      "learning_rate": 4.130456996877661e-05,
      "loss": 0.000860651209950447,
      "step": 206870
    },
    {
      "epoch": 58.72267953448765,
      "grad_norm": 20.887805938720703,
      "learning_rate": 4.130173147885325e-05,
      "loss": 0.007472198456525803,
      "step": 206880
    },
    {
      "epoch": 58.725518024411016,
      "grad_norm": 0.009199848398566246,
      "learning_rate": 4.1298892988929895e-05,
      "loss": 0.00017117243260145187,
      "step": 206890
    },
    {
      "epoch": 58.72835651433437,
      "grad_norm": 0.8444786071777344,
      "learning_rate": 4.129605449900653e-05,
      "loss": 0.0017740720883011817,
      "step": 206900
    },
    {
      "epoch": 58.731195004257735,
      "grad_norm": 0.15847627818584442,
      "learning_rate": 4.129321600908317e-05,
      "loss": 0.0009329678490757942,
      "step": 206910
    },
    {
      "epoch": 58.7340334941811,
      "grad_norm": 1.2502888441085815,
      "learning_rate": 4.1290377519159806e-05,
      "loss": 0.0013886304572224618,
      "step": 206920
    },
    {
      "epoch": 58.736871984104454,
      "grad_norm": 0.5318707823753357,
      "learning_rate": 4.128753902923645e-05,
      "loss": 0.0006603360176086426,
      "step": 206930
    },
    {
      "epoch": 58.73971047402782,
      "grad_norm": 0.04900744557380676,
      "learning_rate": 4.128470053931309e-05,
      "loss": 0.0002483395859599113,
      "step": 206940
    },
    {
      "epoch": 58.74254896395118,
      "grad_norm": 0.031023385003209114,
      "learning_rate": 4.128186204938972e-05,
      "loss": 0.0005301982164382935,
      "step": 206950
    },
    {
      "epoch": 58.745387453874535,
      "grad_norm": 0.02497422695159912,
      "learning_rate": 4.1279023559466365e-05,
      "loss": 0.0007561206817626953,
      "step": 206960
    },
    {
      "epoch": 58.7482259437979,
      "grad_norm": 0.026071587577462196,
      "learning_rate": 4.1276185069543006e-05,
      "loss": 0.0022153478115797044,
      "step": 206970
    },
    {
      "epoch": 58.75106443372126,
      "grad_norm": 0.07425837963819504,
      "learning_rate": 4.127334657961964e-05,
      "loss": 0.0007734159007668495,
      "step": 206980
    },
    {
      "epoch": 58.753902923644624,
      "grad_norm": 0.03809712454676628,
      "learning_rate": 4.127050808969629e-05,
      "loss": 0.004260948672890663,
      "step": 206990
    },
    {
      "epoch": 58.75674141356798,
      "grad_norm": 0.059499625116586685,
      "learning_rate": 4.1267669599772923e-05,
      "loss": 0.0008134795352816582,
      "step": 207000
    },
    {
      "epoch": 58.75674141356798,
      "eval_accuracy": 0.9771730145609462,
      "eval_loss": 0.0834764838218689,
      "eval_runtime": 32.8295,
      "eval_samples_per_second": 479.051,
      "eval_steps_per_second": 7.493,
      "step": 207000
    },
    {
      "epoch": 58.75957990349134,
      "grad_norm": 0.20077790319919586,
      "learning_rate": 4.126483110984956e-05,
      "loss": 0.0015691665932536126,
      "step": 207010
    },
    {
      "epoch": 58.762418393414706,
      "grad_norm": 0.02881387621164322,
      "learning_rate": 4.12619926199262e-05,
      "loss": 0.0011094870045781136,
      "step": 207020
    },
    {
      "epoch": 58.76525688333806,
      "grad_norm": 1.3382675647735596,
      "learning_rate": 4.125915413000284e-05,
      "loss": 0.001234646886587143,
      "step": 207030
    },
    {
      "epoch": 58.768095373261424,
      "grad_norm": 0.3627988398075104,
      "learning_rate": 4.125631564007948e-05,
      "loss": 0.0021222662180662155,
      "step": 207040
    },
    {
      "epoch": 58.77093386318479,
      "grad_norm": 0.3301372230052948,
      "learning_rate": 4.125347715015612e-05,
      "loss": 0.004397808387875557,
      "step": 207050
    },
    {
      "epoch": 58.77377235310815,
      "grad_norm": 0.03294415399432182,
      "learning_rate": 4.125063866023276e-05,
      "loss": 0.007307543605566025,
      "step": 207060
    },
    {
      "epoch": 58.776610843031506,
      "grad_norm": 0.003059299662709236,
      "learning_rate": 4.12478001703094e-05,
      "loss": 0.0007802657783031464,
      "step": 207070
    },
    {
      "epoch": 58.77944933295487,
      "grad_norm": 0.13260625302791595,
      "learning_rate": 4.1244961680386034e-05,
      "loss": 0.0011958673596382142,
      "step": 207080
    },
    {
      "epoch": 58.78228782287823,
      "grad_norm": 2.005964756011963,
      "learning_rate": 4.1242123190462676e-05,
      "loss": 0.0012364739552140236,
      "step": 207090
    },
    {
      "epoch": 58.78512631280159,
      "grad_norm": 1.0927907228469849,
      "learning_rate": 4.123928470053932e-05,
      "loss": 0.0010178759694099426,
      "step": 207100
    },
    {
      "epoch": 58.78796480272495,
      "grad_norm": 0.13922160863876343,
      "learning_rate": 4.123644621061595e-05,
      "loss": 0.0034227706491947174,
      "step": 207110
    },
    {
      "epoch": 58.79080329264831,
      "grad_norm": 0.060910664498806,
      "learning_rate": 4.123360772069259e-05,
      "loss": 0.006601770967245102,
      "step": 207120
    },
    {
      "epoch": 58.79364178257167,
      "grad_norm": 0.7579424381256104,
      "learning_rate": 4.1230769230769234e-05,
      "loss": 0.011407683789730071,
      "step": 207130
    },
    {
      "epoch": 58.79648027249503,
      "grad_norm": 0.058132000267505646,
      "learning_rate": 4.122793074084587e-05,
      "loss": 0.0010198086500167846,
      "step": 207140
    },
    {
      "epoch": 58.799318762418395,
      "grad_norm": 1.208670735359192,
      "learning_rate": 4.122509225092251e-05,
      "loss": 0.00250982828438282,
      "step": 207150
    },
    {
      "epoch": 58.80215725234176,
      "grad_norm": 1.1116269826889038,
      "learning_rate": 4.122225376099915e-05,
      "loss": 0.0016339613124728204,
      "step": 207160
    },
    {
      "epoch": 58.80499574226511,
      "grad_norm": 0.011944854632019997,
      "learning_rate": 4.121941527107579e-05,
      "loss": 0.0005118245258927345,
      "step": 207170
    },
    {
      "epoch": 58.807834232188476,
      "grad_norm": 1.5985960960388184,
      "learning_rate": 4.121657678115243e-05,
      "loss": 0.0010832170024514197,
      "step": 207180
    },
    {
      "epoch": 58.81067272211184,
      "grad_norm": 0.36284491419792175,
      "learning_rate": 4.121373829122907e-05,
      "loss": 0.0016591813415288924,
      "step": 207190
    },
    {
      "epoch": 58.813511212035195,
      "grad_norm": 0.016494128853082657,
      "learning_rate": 4.121089980130571e-05,
      "loss": 0.0008052796125411988,
      "step": 207200
    },
    {
      "epoch": 58.81634970195856,
      "grad_norm": 0.12156101316213608,
      "learning_rate": 4.1208061311382345e-05,
      "loss": 0.0009025674313306809,
      "step": 207210
    },
    {
      "epoch": 58.81918819188192,
      "grad_norm": 0.10179951786994934,
      "learning_rate": 4.1205222821458986e-05,
      "loss": 0.00030388925224542616,
      "step": 207220
    },
    {
      "epoch": 58.822026681805276,
      "grad_norm": 0.017110340297222137,
      "learning_rate": 4.120238433153563e-05,
      "loss": 0.0003111656755208969,
      "step": 207230
    },
    {
      "epoch": 58.82486517172864,
      "grad_norm": 0.06375014781951904,
      "learning_rate": 4.119954584161226e-05,
      "loss": 0.0002698222175240517,
      "step": 207240
    },
    {
      "epoch": 58.827703661652,
      "grad_norm": 0.017609983682632446,
      "learning_rate": 4.1196707351688904e-05,
      "loss": 0.00033736247569322587,
      "step": 207250
    },
    {
      "epoch": 58.830542151575365,
      "grad_norm": 0.7699373960494995,
      "learning_rate": 4.1193868861765545e-05,
      "loss": 0.0032134436070919035,
      "step": 207260
    },
    {
      "epoch": 58.83338064149872,
      "grad_norm": 0.15301477909088135,
      "learning_rate": 4.119103037184218e-05,
      "loss": 0.0004494383931159973,
      "step": 207270
    },
    {
      "epoch": 58.836219131422084,
      "grad_norm": 0.30342134833335876,
      "learning_rate": 4.118819188191882e-05,
      "loss": 0.00041185654699802397,
      "step": 207280
    },
    {
      "epoch": 58.83905762134545,
      "grad_norm": 0.05601118877530098,
      "learning_rate": 4.118535339199546e-05,
      "loss": 0.0006584875285625458,
      "step": 207290
    },
    {
      "epoch": 58.8418961112688,
      "grad_norm": 0.03268035501241684,
      "learning_rate": 4.11825149020721e-05,
      "loss": 0.0011105112731456756,
      "step": 207300
    },
    {
      "epoch": 58.844734601192165,
      "grad_norm": 0.4537128508090973,
      "learning_rate": 4.117967641214874e-05,
      "loss": 0.006664318591356277,
      "step": 207310
    },
    {
      "epoch": 58.84757309111553,
      "grad_norm": 0.08298700302839279,
      "learning_rate": 4.117683792222538e-05,
      "loss": 0.0033212848007678985,
      "step": 207320
    },
    {
      "epoch": 58.850411581038884,
      "grad_norm": 0.01407330110669136,
      "learning_rate": 4.117399943230202e-05,
      "loss": 0.0068441122770309445,
      "step": 207330
    },
    {
      "epoch": 58.85325007096225,
      "grad_norm": 7.157983779907227,
      "learning_rate": 4.1171160942378656e-05,
      "loss": 0.003793060779571533,
      "step": 207340
    },
    {
      "epoch": 58.85608856088561,
      "grad_norm": 0.24867653846740723,
      "learning_rate": 4.116832245245529e-05,
      "loss": 0.0012763502076268197,
      "step": 207350
    },
    {
      "epoch": 58.85892705080897,
      "grad_norm": 0.0388803631067276,
      "learning_rate": 4.116548396253194e-05,
      "loss": 0.0005054570734500885,
      "step": 207360
    },
    {
      "epoch": 58.86176554073233,
      "grad_norm": 0.15159553289413452,
      "learning_rate": 4.116264547260857e-05,
      "loss": 0.0015129465609788895,
      "step": 207370
    },
    {
      "epoch": 58.86460403065569,
      "grad_norm": 0.06402606517076492,
      "learning_rate": 4.1159806982685214e-05,
      "loss": 0.0016532553359866141,
      "step": 207380
    },
    {
      "epoch": 58.867442520579054,
      "grad_norm": 0.054266732186079025,
      "learning_rate": 4.1156968492761856e-05,
      "loss": 0.0004450865089893341,
      "step": 207390
    },
    {
      "epoch": 58.87028101050241,
      "grad_norm": 0.16818787157535553,
      "learning_rate": 4.115413000283849e-05,
      "loss": 0.0009078368544578552,
      "step": 207400
    },
    {
      "epoch": 58.87311950042577,
      "grad_norm": 0.09061696380376816,
      "learning_rate": 4.115129151291513e-05,
      "loss": 0.002191256172955036,
      "step": 207410
    },
    {
      "epoch": 58.875957990349136,
      "grad_norm": 0.19953051209449768,
      "learning_rate": 4.114845302299177e-05,
      "loss": 0.000559241883456707,
      "step": 207420
    },
    {
      "epoch": 58.8787964802725,
      "grad_norm": 0.12617988884449005,
      "learning_rate": 4.114561453306841e-05,
      "loss": 0.0008060453459620476,
      "step": 207430
    },
    {
      "epoch": 58.881634970195854,
      "grad_norm": 0.010966803878545761,
      "learning_rate": 4.114277604314505e-05,
      "loss": 0.0015103500336408615,
      "step": 207440
    },
    {
      "epoch": 58.88447346011922,
      "grad_norm": 2.135359525680542,
      "learning_rate": 4.1139937553221684e-05,
      "loss": 0.0007600663229823112,
      "step": 207450
    },
    {
      "epoch": 58.88731195004258,
      "grad_norm": 0.04405699670314789,
      "learning_rate": 4.113709906329833e-05,
      "loss": 0.0029303736984729765,
      "step": 207460
    },
    {
      "epoch": 58.890150439965936,
      "grad_norm": 16.221555709838867,
      "learning_rate": 4.1134260573374966e-05,
      "loss": 0.003999515995383262,
      "step": 207470
    },
    {
      "epoch": 58.8929889298893,
      "grad_norm": 0.43182840943336487,
      "learning_rate": 4.11314220834516e-05,
      "loss": 0.00041981544345617296,
      "step": 207480
    },
    {
      "epoch": 58.89582741981266,
      "grad_norm": 0.04110600799322128,
      "learning_rate": 4.112858359352825e-05,
      "loss": 0.0004885423928499222,
      "step": 207490
    },
    {
      "epoch": 58.89866590973602,
      "grad_norm": 0.6182302832603455,
      "learning_rate": 4.1125745103604884e-05,
      "loss": 0.0003967374563217163,
      "step": 207500
    },
    {
      "epoch": 58.89866590973602,
      "eval_accuracy": 0.9788262224200419,
      "eval_loss": 0.07703840732574463,
      "eval_runtime": 33.065,
      "eval_samples_per_second": 475.639,
      "eval_steps_per_second": 7.44,
      "step": 207500
    },
    {
      "epoch": 58.90150439965938,
      "grad_norm": 0.040107447654008865,
      "learning_rate": 4.1122906613681525e-05,
      "loss": 0.00020810551941394806,
      "step": 207510
    },
    {
      "epoch": 58.90434288958274,
      "grad_norm": 0.030618177726864815,
      "learning_rate": 4.1120068123758167e-05,
      "loss": 0.0006165483966469765,
      "step": 207520
    },
    {
      "epoch": 58.907181379506106,
      "grad_norm": 0.04630030691623688,
      "learning_rate": 4.11172296338348e-05,
      "loss": 0.00020969510078430175,
      "step": 207530
    },
    {
      "epoch": 58.91001986942946,
      "grad_norm": 0.11447747051715851,
      "learning_rate": 4.111439114391144e-05,
      "loss": 0.00016018860042095185,
      "step": 207540
    },
    {
      "epoch": 58.912858359352825,
      "grad_norm": 0.008581391535699368,
      "learning_rate": 4.111155265398808e-05,
      "loss": 0.00017898213118314744,
      "step": 207550
    },
    {
      "epoch": 58.91569684927619,
      "grad_norm": 0.501482367515564,
      "learning_rate": 4.110871416406472e-05,
      "loss": 0.00027753394097089766,
      "step": 207560
    },
    {
      "epoch": 58.91853533919954,
      "grad_norm": 0.46124380826950073,
      "learning_rate": 4.110587567414136e-05,
      "loss": 0.0009124785661697388,
      "step": 207570
    },
    {
      "epoch": 58.921373829122906,
      "grad_norm": 0.7475137114524841,
      "learning_rate": 4.1103037184217994e-05,
      "loss": 0.001150272972881794,
      "step": 207580
    },
    {
      "epoch": 58.92421231904627,
      "grad_norm": 1.1510826349258423,
      "learning_rate": 4.110019869429464e-05,
      "loss": 0.0015948090702295303,
      "step": 207590
    },
    {
      "epoch": 58.927050808969625,
      "grad_norm": 3.4067542552948,
      "learning_rate": 4.109736020437128e-05,
      "loss": 0.010500933229923248,
      "step": 207600
    },
    {
      "epoch": 58.92988929889299,
      "grad_norm": 0.27619796991348267,
      "learning_rate": 4.109452171444791e-05,
      "loss": 0.0005762262269854545,
      "step": 207610
    },
    {
      "epoch": 58.93272778881635,
      "grad_norm": 10.329214096069336,
      "learning_rate": 4.109168322452456e-05,
      "loss": 0.0017792237922549248,
      "step": 207620
    },
    {
      "epoch": 58.935566278739714,
      "grad_norm": 0.510172963142395,
      "learning_rate": 4.1088844734601195e-05,
      "loss": 0.001700356788933277,
      "step": 207630
    },
    {
      "epoch": 58.93840476866307,
      "grad_norm": 0.007970583625137806,
      "learning_rate": 4.1086006244677836e-05,
      "loss": 0.0006431018933653832,
      "step": 207640
    },
    {
      "epoch": 58.94124325858643,
      "grad_norm": 0.055422693490982056,
      "learning_rate": 4.108316775475447e-05,
      "loss": 0.00013470035046339034,
      "step": 207650
    },
    {
      "epoch": 58.944081748509795,
      "grad_norm": 0.2304726243019104,
      "learning_rate": 4.108032926483111e-05,
      "loss": 0.00023025237023830414,
      "step": 207660
    },
    {
      "epoch": 58.94692023843315,
      "grad_norm": 0.19219844043254852,
      "learning_rate": 4.107749077490775e-05,
      "loss": 0.00028163082897663115,
      "step": 207670
    },
    {
      "epoch": 58.949758728356514,
      "grad_norm": 0.022780882194638252,
      "learning_rate": 4.107465228498439e-05,
      "loss": 0.0003609348088502884,
      "step": 207680
    },
    {
      "epoch": 58.95259721827988,
      "grad_norm": 0.07804735004901886,
      "learning_rate": 4.107181379506103e-05,
      "loss": 0.002425245754420757,
      "step": 207690
    },
    {
      "epoch": 58.95543570820323,
      "grad_norm": 0.21282333135604858,
      "learning_rate": 4.106897530513767e-05,
      "loss": 0.0003826828673481941,
      "step": 207700
    },
    {
      "epoch": 58.958274198126595,
      "grad_norm": 0.12351103872060776,
      "learning_rate": 4.1066136815214305e-05,
      "loss": 0.0003309847787022591,
      "step": 207710
    },
    {
      "epoch": 58.96111268804996,
      "grad_norm": 0.02407628484070301,
      "learning_rate": 4.1063298325290947e-05,
      "loss": 0.0009531006217002869,
      "step": 207720
    },
    {
      "epoch": 58.96395117797332,
      "grad_norm": 0.026386631652712822,
      "learning_rate": 4.106045983536759e-05,
      "loss": 0.00043917614966630935,
      "step": 207730
    },
    {
      "epoch": 58.96678966789668,
      "grad_norm": 0.3386162519454956,
      "learning_rate": 4.105762134544422e-05,
      "loss": 0.0007970765233039856,
      "step": 207740
    },
    {
      "epoch": 58.96962815782004,
      "grad_norm": 2.255995988845825,
      "learning_rate": 4.1054782855520864e-05,
      "loss": 0.0011912908405065537,
      "step": 207750
    },
    {
      "epoch": 58.9724666477434,
      "grad_norm": 0.018130864948034286,
      "learning_rate": 4.1051944365597505e-05,
      "loss": 0.0003508782014250755,
      "step": 207760
    },
    {
      "epoch": 58.97530513766676,
      "grad_norm": 0.11074202507734299,
      "learning_rate": 4.104910587567414e-05,
      "loss": 0.0004644874483346939,
      "step": 207770
    },
    {
      "epoch": 58.97814362759012,
      "grad_norm": 0.11638941615819931,
      "learning_rate": 4.104626738575078e-05,
      "loss": 0.0005673544481396675,
      "step": 207780
    },
    {
      "epoch": 58.980982117513484,
      "grad_norm": 0.05714520439505577,
      "learning_rate": 4.104342889582742e-05,
      "loss": 0.0037725768983364106,
      "step": 207790
    },
    {
      "epoch": 58.98382060743684,
      "grad_norm": 0.28198277950286865,
      "learning_rate": 4.1040590405904064e-05,
      "loss": 0.0009155942127108574,
      "step": 207800
    },
    {
      "epoch": 58.9866590973602,
      "grad_norm": 0.04862555116415024,
      "learning_rate": 4.10377519159807e-05,
      "loss": 0.00025427136570215224,
      "step": 207810
    },
    {
      "epoch": 58.989497587283566,
      "grad_norm": 0.623757541179657,
      "learning_rate": 4.103491342605734e-05,
      "loss": 0.0008239630609750747,
      "step": 207820
    },
    {
      "epoch": 58.99233607720693,
      "grad_norm": 0.049276720732450485,
      "learning_rate": 4.103207493613398e-05,
      "loss": 0.0006139809265732765,
      "step": 207830
    },
    {
      "epoch": 58.995174567130285,
      "grad_norm": 0.34119123220443726,
      "learning_rate": 4.1029236446210616e-05,
      "loss": 0.0004357393831014633,
      "step": 207840
    },
    {
      "epoch": 58.99801305705365,
      "grad_norm": 0.02481074258685112,
      "learning_rate": 4.102639795628726e-05,
      "loss": 0.00015672259032726288,
      "step": 207850
    },
    {
      "epoch": 59.00085154697701,
      "grad_norm": 0.02579890750348568,
      "learning_rate": 4.10235594663639e-05,
      "loss": 0.0004602399654686451,
      "step": 207860
    },
    {
      "epoch": 59.003690036900366,
      "grad_norm": 0.44813501834869385,
      "learning_rate": 4.102072097644053e-05,
      "loss": 0.000399518758058548,
      "step": 207870
    },
    {
      "epoch": 59.00652852682373,
      "grad_norm": 0.021876564249396324,
      "learning_rate": 4.1017882486517175e-05,
      "loss": 0.00020244549959897996,
      "step": 207880
    },
    {
      "epoch": 59.00936701674709,
      "grad_norm": 0.01540295034646988,
      "learning_rate": 4.1015043996593816e-05,
      "loss": 0.0017710911110043527,
      "step": 207890
    },
    {
      "epoch": 59.012205506670455,
      "grad_norm": 0.4906509518623352,
      "learning_rate": 4.101220550667045e-05,
      "loss": 0.00041702501475811006,
      "step": 207900
    },
    {
      "epoch": 59.01504399659381,
      "grad_norm": 0.012258261442184448,
      "learning_rate": 4.100936701674709e-05,
      "loss": 0.0006076676771044731,
      "step": 207910
    },
    {
      "epoch": 59.01788248651717,
      "grad_norm": 0.1737445890903473,
      "learning_rate": 4.1006528526823733e-05,
      "loss": 0.0004614373669028282,
      "step": 207920
    },
    {
      "epoch": 59.020720976440536,
      "grad_norm": 0.009206118062138557,
      "learning_rate": 4.1003690036900375e-05,
      "loss": 0.0003240685909986496,
      "step": 207930
    },
    {
      "epoch": 59.02355946636389,
      "grad_norm": 0.06124449521303177,
      "learning_rate": 4.100085154697701e-05,
      "loss": 0.006179551780223847,
      "step": 207940
    },
    {
      "epoch": 59.026397956287255,
      "grad_norm": 0.04354212433099747,
      "learning_rate": 4.0998013057053644e-05,
      "loss": 0.0024033065885305403,
      "step": 207950
    },
    {
      "epoch": 59.02923644621062,
      "grad_norm": 0.04144112765789032,
      "learning_rate": 4.099517456713029e-05,
      "loss": 0.0004903672263026238,
      "step": 207960
    },
    {
      "epoch": 59.032074936133974,
      "grad_norm": 0.08474123477935791,
      "learning_rate": 4.099233607720693e-05,
      "loss": 0.0013550233095884323,
      "step": 207970
    },
    {
      "epoch": 59.03491342605734,
      "grad_norm": 0.1835555136203766,
      "learning_rate": 4.098949758728357e-05,
      "loss": 0.0008627686649560929,
      "step": 207980
    },
    {
      "epoch": 59.0377519159807,
      "grad_norm": 7.0196533203125,
      "learning_rate": 4.098665909736021e-05,
      "loss": 0.003278651461005211,
      "step": 207990
    },
    {
      "epoch": 59.04059040590406,
      "grad_norm": 7.421067714691162,
      "learning_rate": 4.0983820607436844e-05,
      "loss": 0.004430201277136803,
      "step": 208000
    },
    {
      "epoch": 59.04059040590406,
      "eval_accuracy": 0.9776816939022064,
      "eval_loss": 0.0822572410106659,
      "eval_runtime": 32.2419,
      "eval_samples_per_second": 487.781,
      "eval_steps_per_second": 7.63,
      "step": 208000
    },
    {
      "epoch": 59.04342889582742,
      "grad_norm": 8.766963958740234,
      "learning_rate": 4.0980982117513485e-05,
      "loss": 0.0015285126864910126,
      "step": 208010
    },
    {
      "epoch": 59.04626738575078,
      "grad_norm": 1.4790773391723633,
      "learning_rate": 4.097814362759013e-05,
      "loss": 0.0004593798890709877,
      "step": 208020
    },
    {
      "epoch": 59.049105875674144,
      "grad_norm": 0.006965484004467726,
      "learning_rate": 4.097530513766676e-05,
      "loss": 0.0006050080060958863,
      "step": 208030
    },
    {
      "epoch": 59.0519443655975,
      "grad_norm": 10.768735885620117,
      "learning_rate": 4.09724666477434e-05,
      "loss": 0.00208019707351923,
      "step": 208040
    },
    {
      "epoch": 59.05478285552086,
      "grad_norm": 0.7936661243438721,
      "learning_rate": 4.096962815782004e-05,
      "loss": 0.0005237067118287087,
      "step": 208050
    },
    {
      "epoch": 59.057621345444225,
      "grad_norm": 0.2742912173271179,
      "learning_rate": 4.0966789667896686e-05,
      "loss": 0.0008844098076224328,
      "step": 208060
    },
    {
      "epoch": 59.06045983536758,
      "grad_norm": 0.09872126579284668,
      "learning_rate": 4.096395117797332e-05,
      "loss": 0.0004022387787699699,
      "step": 208070
    },
    {
      "epoch": 59.063298325290944,
      "grad_norm": 0.17566470801830292,
      "learning_rate": 4.0961112688049955e-05,
      "loss": 0.0033900637179613115,
      "step": 208080
    },
    {
      "epoch": 59.06613681521431,
      "grad_norm": 0.29065167903900146,
      "learning_rate": 4.09582741981266e-05,
      "loss": 0.0002166997641324997,
      "step": 208090
    },
    {
      "epoch": 59.06897530513767,
      "grad_norm": 0.059227947145700455,
      "learning_rate": 4.095543570820324e-05,
      "loss": 0.0015466881915926934,
      "step": 208100
    },
    {
      "epoch": 59.071813795061026,
      "grad_norm": 0.11734378337860107,
      "learning_rate": 4.095259721827988e-05,
      "loss": 0.0003778437152504921,
      "step": 208110
    },
    {
      "epoch": 59.07465228498439,
      "grad_norm": 1.2174646854400635,
      "learning_rate": 4.094975872835652e-05,
      "loss": 0.00027200598269701005,
      "step": 208120
    },
    {
      "epoch": 59.07749077490775,
      "grad_norm": 7.156342029571533,
      "learning_rate": 4.0946920238433155e-05,
      "loss": 0.0010393979027867317,
      "step": 208130
    },
    {
      "epoch": 59.08032926483111,
      "grad_norm": 0.04469839110970497,
      "learning_rate": 4.0944081748509796e-05,
      "loss": 0.000218113511800766,
      "step": 208140
    },
    {
      "epoch": 59.08316775475447,
      "grad_norm": 0.015037835575640202,
      "learning_rate": 4.094124325858643e-05,
      "loss": 0.0016883708536624908,
      "step": 208150
    },
    {
      "epoch": 59.08600624467783,
      "grad_norm": 0.024703199043869972,
      "learning_rate": 4.093840476866307e-05,
      "loss": 0.00028570666909217836,
      "step": 208160
    },
    {
      "epoch": 59.08884473460119,
      "grad_norm": 0.07027429342269897,
      "learning_rate": 4.0935566278739714e-05,
      "loss": 0.0036558333784341814,
      "step": 208170
    },
    {
      "epoch": 59.09168322452455,
      "grad_norm": 0.0906556248664856,
      "learning_rate": 4.093272778881635e-05,
      "loss": 0.00019251257181167601,
      "step": 208180
    },
    {
      "epoch": 59.094521714447914,
      "grad_norm": 0.08480474352836609,
      "learning_rate": 4.092988929889299e-05,
      "loss": 0.0002675136551260948,
      "step": 208190
    },
    {
      "epoch": 59.09736020437128,
      "grad_norm": 0.2063034176826477,
      "learning_rate": 4.092705080896963e-05,
      "loss": 0.0002925315871834755,
      "step": 208200
    },
    {
      "epoch": 59.10019869429463,
      "grad_norm": 0.06712369620800018,
      "learning_rate": 4.0924212319046266e-05,
      "loss": 0.004868596792221069,
      "step": 208210
    },
    {
      "epoch": 59.103037184217996,
      "grad_norm": 0.08728727698326111,
      "learning_rate": 4.0921373829122914e-05,
      "loss": 0.0023691341280937194,
      "step": 208220
    },
    {
      "epoch": 59.10587567414136,
      "grad_norm": 0.3144252300262451,
      "learning_rate": 4.091853533919955e-05,
      "loss": 0.00021848194301128389,
      "step": 208230
    },
    {
      "epoch": 59.108714164064715,
      "grad_norm": 0.0677976906299591,
      "learning_rate": 4.091569684927618e-05,
      "loss": 0.00472303219139576,
      "step": 208240
    },
    {
      "epoch": 59.11155265398808,
      "grad_norm": 0.5081467032432556,
      "learning_rate": 4.0912858359352824e-05,
      "loss": 0.01782566010951996,
      "step": 208250
    },
    {
      "epoch": 59.11439114391144,
      "grad_norm": 0.10976193100214005,
      "learning_rate": 4.0910019869429466e-05,
      "loss": 0.0005334103479981422,
      "step": 208260
    },
    {
      "epoch": 59.1172296338348,
      "grad_norm": 0.02519877627491951,
      "learning_rate": 4.090718137950611e-05,
      "loss": 0.0018647167831659316,
      "step": 208270
    },
    {
      "epoch": 59.12006812375816,
      "grad_norm": 13.728065490722656,
      "learning_rate": 4.090434288958274e-05,
      "loss": 0.0034889984875917436,
      "step": 208280
    },
    {
      "epoch": 59.12290661368152,
      "grad_norm": 0.4617295563220978,
      "learning_rate": 4.090150439965938e-05,
      "loss": 0.0001879347488284111,
      "step": 208290
    },
    {
      "epoch": 59.125745103604885,
      "grad_norm": 0.04866599664092064,
      "learning_rate": 4.0898665909736024e-05,
      "loss": 0.0016242315992712974,
      "step": 208300
    },
    {
      "epoch": 59.12858359352824,
      "grad_norm": 0.0794895738363266,
      "learning_rate": 4.089582741981266e-05,
      "loss": 0.005256699025630951,
      "step": 208310
    },
    {
      "epoch": 59.131422083451604,
      "grad_norm": 8.032265663146973,
      "learning_rate": 4.08929889298893e-05,
      "loss": 0.0014704780653119088,
      "step": 208320
    },
    {
      "epoch": 59.134260573374966,
      "grad_norm": 15.124927520751953,
      "learning_rate": 4.089015043996594e-05,
      "loss": 0.005256590247154236,
      "step": 208330
    },
    {
      "epoch": 59.13709906329832,
      "grad_norm": 0.014676358550786972,
      "learning_rate": 4.0887311950042576e-05,
      "loss": 0.001370241679251194,
      "step": 208340
    },
    {
      "epoch": 59.139937553221685,
      "grad_norm": 0.016551930457353592,
      "learning_rate": 4.088447346011922e-05,
      "loss": 0.0004966679960489273,
      "step": 208350
    },
    {
      "epoch": 59.14277604314505,
      "grad_norm": 7.34307861328125,
      "learning_rate": 4.088163497019586e-05,
      "loss": 0.02061762511730194,
      "step": 208360
    },
    {
      "epoch": 59.14561453306841,
      "grad_norm": 0.07635661214590073,
      "learning_rate": 4.0878796480272494e-05,
      "loss": 0.00036470629274845125,
      "step": 208370
    },
    {
      "epoch": 59.14845302299177,
      "grad_norm": 0.3257458209991455,
      "learning_rate": 4.0875957990349135e-05,
      "loss": 0.0023926762863993646,
      "step": 208380
    },
    {
      "epoch": 59.15129151291513,
      "grad_norm": 0.11409034579992294,
      "learning_rate": 4.0873119500425776e-05,
      "loss": 0.00026971213519573213,
      "step": 208390
    },
    {
      "epoch": 59.15413000283849,
      "grad_norm": 0.2311750054359436,
      "learning_rate": 4.087028101050242e-05,
      "loss": 0.002975240163505077,
      "step": 208400
    },
    {
      "epoch": 59.15696849276185,
      "grad_norm": 0.018550384789705276,
      "learning_rate": 4.086744252057905e-05,
      "loss": 0.0003716014325618744,
      "step": 208410
    },
    {
      "epoch": 59.15980698268521,
      "grad_norm": 0.6655106544494629,
      "learning_rate": 4.0864604030655694e-05,
      "loss": 0.0007350720465183258,
      "step": 208420
    },
    {
      "epoch": 59.162645472608574,
      "grad_norm": 0.19258610904216766,
      "learning_rate": 4.0861765540732335e-05,
      "loss": 0.0004562605172395706,
      "step": 208430
    },
    {
      "epoch": 59.16548396253193,
      "grad_norm": 1.0891563892364502,
      "learning_rate": 4.085892705080897e-05,
      "loss": 0.0005592022091150284,
      "step": 208440
    },
    {
      "epoch": 59.16832245245529,
      "grad_norm": 0.27627140283584595,
      "learning_rate": 4.085608856088561e-05,
      "loss": 0.0005465507507324219,
      "step": 208450
    },
    {
      "epoch": 59.171160942378656,
      "grad_norm": 0.02419949881732464,
      "learning_rate": 4.085325007096225e-05,
      "loss": 0.002882329374551773,
      "step": 208460
    },
    {
      "epoch": 59.17399943230202,
      "grad_norm": 0.07663603127002716,
      "learning_rate": 4.085041158103889e-05,
      "loss": 0.0009726541116833687,
      "step": 208470
    },
    {
      "epoch": 59.176837922225374,
      "grad_norm": 0.11414188891649246,
      "learning_rate": 4.084757309111553e-05,
      "loss": 0.00023626722395420074,
      "step": 208480
    },
    {
      "epoch": 59.17967641214874,
      "grad_norm": 0.11437925696372986,
      "learning_rate": 4.084473460119217e-05,
      "loss": 0.002907714806497097,
      "step": 208490
    },
    {
      "epoch": 59.1825149020721,
      "grad_norm": 4.730220317840576,
      "learning_rate": 4.0841896111268804e-05,
      "loss": 0.0012253113090991974,
      "step": 208500
    },
    {
      "epoch": 59.1825149020721,
      "eval_accuracy": 0.9784447129140967,
      "eval_loss": 0.07730703055858612,
      "eval_runtime": 32.709,
      "eval_samples_per_second": 480.815,
      "eval_steps_per_second": 7.521,
      "step": 208500
    },
    {
      "epoch": 59.185353391995456,
      "grad_norm": 0.08481720834970474,
      "learning_rate": 4.0839057621345446e-05,
      "loss": 0.0003444690257310867,
      "step": 208510
    },
    {
      "epoch": 59.18819188191882,
      "grad_norm": 0.47381699085235596,
      "learning_rate": 4.083621913142209e-05,
      "loss": 0.00035023875534534454,
      "step": 208520
    },
    {
      "epoch": 59.19103037184218,
      "grad_norm": 0.05361727252602577,
      "learning_rate": 4.083338064149873e-05,
      "loss": 0.0005422268062829971,
      "step": 208530
    },
    {
      "epoch": 59.19386886176554,
      "grad_norm": 0.401660293340683,
      "learning_rate": 4.083054215157536e-05,
      "loss": 0.0010594256222248078,
      "step": 208540
    },
    {
      "epoch": 59.1967073516889,
      "grad_norm": 0.4858897626399994,
      "learning_rate": 4.0827703661652e-05,
      "loss": 0.00082086231559515,
      "step": 208550
    },
    {
      "epoch": 59.19954584161226,
      "grad_norm": 0.38886117935180664,
      "learning_rate": 4.0824865171728646e-05,
      "loss": 0.0008142786100506783,
      "step": 208560
    },
    {
      "epoch": 59.202384331535626,
      "grad_norm": 0.08062991499900818,
      "learning_rate": 4.082202668180528e-05,
      "loss": 0.00039433352649211886,
      "step": 208570
    },
    {
      "epoch": 59.20522282145898,
      "grad_norm": 0.020408710464835167,
      "learning_rate": 4.081918819188192e-05,
      "loss": 0.00301948357373476,
      "step": 208580
    },
    {
      "epoch": 59.208061311382345,
      "grad_norm": 0.07235443592071533,
      "learning_rate": 4.081634970195856e-05,
      "loss": 0.00036010369658470156,
      "step": 208590
    },
    {
      "epoch": 59.21089980130571,
      "grad_norm": 0.07540994137525558,
      "learning_rate": 4.08135112120352e-05,
      "loss": 0.0006806753575801849,
      "step": 208600
    },
    {
      "epoch": 59.21373829122906,
      "grad_norm": 1.5043736696243286,
      "learning_rate": 4.081067272211184e-05,
      "loss": 0.0006596093997359275,
      "step": 208610
    },
    {
      "epoch": 59.216576781152426,
      "grad_norm": 0.3405483365058899,
      "learning_rate": 4.080783423218848e-05,
      "loss": 0.0007652267813682556,
      "step": 208620
    },
    {
      "epoch": 59.21941527107579,
      "grad_norm": 0.18182450532913208,
      "learning_rate": 4.0804995742265115e-05,
      "loss": 0.0022856274619698523,
      "step": 208630
    },
    {
      "epoch": 59.22225376099915,
      "grad_norm": 0.14827436208724976,
      "learning_rate": 4.0802157252341757e-05,
      "loss": 0.0002925468608736992,
      "step": 208640
    },
    {
      "epoch": 59.22509225092251,
      "grad_norm": 0.0358758345246315,
      "learning_rate": 4.07993187624184e-05,
      "loss": 0.0009020799770951271,
      "step": 208650
    },
    {
      "epoch": 59.22793074084587,
      "grad_norm": 0.032166801393032074,
      "learning_rate": 4.079648027249503e-05,
      "loss": 0.0005771357566118241,
      "step": 208660
    },
    {
      "epoch": 59.23076923076923,
      "grad_norm": 0.18071867525577545,
      "learning_rate": 4.0793641782571674e-05,
      "loss": 0.000900956615805626,
      "step": 208670
    },
    {
      "epoch": 59.23360772069259,
      "grad_norm": 5.668272018432617,
      "learning_rate": 4.079080329264831e-05,
      "loss": 0.0012646080926060677,
      "step": 208680
    },
    {
      "epoch": 59.23644621061595,
      "grad_norm": 0.0711614117026329,
      "learning_rate": 4.078796480272496e-05,
      "loss": 0.00026070233434438705,
      "step": 208690
    },
    {
      "epoch": 59.239284700539315,
      "grad_norm": 0.08946266770362854,
      "learning_rate": 4.078512631280159e-05,
      "loss": 0.0004927709698677063,
      "step": 208700
    },
    {
      "epoch": 59.24212319046267,
      "grad_norm": 0.03241831809282303,
      "learning_rate": 4.0782287822878226e-05,
      "loss": 0.00018174443393945695,
      "step": 208710
    },
    {
      "epoch": 59.244961680386034,
      "grad_norm": 0.049535248428583145,
      "learning_rate": 4.0779449332954874e-05,
      "loss": 0.0003244033083319664,
      "step": 208720
    },
    {
      "epoch": 59.2478001703094,
      "grad_norm": 0.07869278639554977,
      "learning_rate": 4.077661084303151e-05,
      "loss": 0.00038533341139554975,
      "step": 208730
    },
    {
      "epoch": 59.25063866023276,
      "grad_norm": 0.017420105636119843,
      "learning_rate": 4.077377235310815e-05,
      "loss": 0.00015230514109134674,
      "step": 208740
    },
    {
      "epoch": 59.253477150156115,
      "grad_norm": 0.11073766648769379,
      "learning_rate": 4.077093386318479e-05,
      "loss": 0.0003903353586792946,
      "step": 208750
    },
    {
      "epoch": 59.25631564007948,
      "grad_norm": 0.8112896680831909,
      "learning_rate": 4.0768095373261426e-05,
      "loss": 0.0008689681068062783,
      "step": 208760
    },
    {
      "epoch": 59.25915413000284,
      "grad_norm": 0.06953651458024979,
      "learning_rate": 4.076525688333807e-05,
      "loss": 0.0003301454707980156,
      "step": 208770
    },
    {
      "epoch": 59.2619926199262,
      "grad_norm": 0.00554344616830349,
      "learning_rate": 4.07624183934147e-05,
      "loss": 0.000978374108672142,
      "step": 208780
    },
    {
      "epoch": 59.26483110984956,
      "grad_norm": 0.04279300570487976,
      "learning_rate": 4.075957990349134e-05,
      "loss": 0.00013325214385986328,
      "step": 208790
    },
    {
      "epoch": 59.26766959977292,
      "grad_norm": 0.11866388469934464,
      "learning_rate": 4.0756741413567985e-05,
      "loss": 0.0020523719489574433,
      "step": 208800
    },
    {
      "epoch": 59.27050808969628,
      "grad_norm": 0.08551599085330963,
      "learning_rate": 4.075390292364462e-05,
      "loss": 0.0012163668870925903,
      "step": 208810
    },
    {
      "epoch": 59.27334657961964,
      "grad_norm": 0.5311707258224487,
      "learning_rate": 4.075106443372127e-05,
      "loss": 0.01011124551296234,
      "step": 208820
    },
    {
      "epoch": 59.276185069543004,
      "grad_norm": 0.018165912479162216,
      "learning_rate": 4.07482259437979e-05,
      "loss": 0.00016736853867769242,
      "step": 208830
    },
    {
      "epoch": 59.27902355946637,
      "grad_norm": 0.1225375235080719,
      "learning_rate": 4.074538745387454e-05,
      "loss": 0.00044635478407144545,
      "step": 208840
    },
    {
      "epoch": 59.28186204938972,
      "grad_norm": 0.012719024904072285,
      "learning_rate": 4.0742548963951185e-05,
      "loss": 0.00028905365616083145,
      "step": 208850
    },
    {
      "epoch": 59.284700539313086,
      "grad_norm": 0.09274417906999588,
      "learning_rate": 4.073971047402782e-05,
      "loss": 0.0001270987093448639,
      "step": 208860
    },
    {
      "epoch": 59.28753902923645,
      "grad_norm": 0.046678781509399414,
      "learning_rate": 4.073687198410446e-05,
      "loss": 0.00027902983129024506,
      "step": 208870
    },
    {
      "epoch": 59.290377519159804,
      "grad_norm": 0.13408954441547394,
      "learning_rate": 4.0734033494181095e-05,
      "loss": 0.0002171434462070465,
      "step": 208880
    },
    {
      "epoch": 59.29321600908317,
      "grad_norm": 0.015426598489284515,
      "learning_rate": 4.073119500425774e-05,
      "loss": 0.00027035679668188097,
      "step": 208890
    },
    {
      "epoch": 59.29605449900653,
      "grad_norm": 0.2719776928424835,
      "learning_rate": 4.072835651433438e-05,
      "loss": 0.0003261376172304153,
      "step": 208900
    },
    {
      "epoch": 59.298892988929886,
      "grad_norm": 0.06029966101050377,
      "learning_rate": 4.072551802441101e-05,
      "loss": 0.001478903368115425,
      "step": 208910
    },
    {
      "epoch": 59.30173147885325,
      "grad_norm": 0.01664654165506363,
      "learning_rate": 4.0722679534487654e-05,
      "loss": 0.001374727301299572,
      "step": 208920
    },
    {
      "epoch": 59.30456996877661,
      "grad_norm": 0.04944661632180214,
      "learning_rate": 4.0719841044564295e-05,
      "loss": 0.005882376432418823,
      "step": 208930
    },
    {
      "epoch": 59.307408458699975,
      "grad_norm": 0.4178543984889984,
      "learning_rate": 4.071700255464093e-05,
      "loss": 0.0002276284620165825,
      "step": 208940
    },
    {
      "epoch": 59.31024694862333,
      "grad_norm": 0.34803563356399536,
      "learning_rate": 4.071416406471758e-05,
      "loss": 0.0006628014147281647,
      "step": 208950
    },
    {
      "epoch": 59.31308543854669,
      "grad_norm": 0.06936804950237274,
      "learning_rate": 4.071132557479421e-05,
      "loss": 0.00011117346584796906,
      "step": 208960
    },
    {
      "epoch": 59.315923928470056,
      "grad_norm": 0.10818737000226974,
      "learning_rate": 4.070848708487085e-05,
      "loss": 0.0011971769854426384,
      "step": 208970
    },
    {
      "epoch": 59.31876241839341,
      "grad_norm": 0.0684070959687233,
      "learning_rate": 4.070564859494749e-05,
      "loss": 0.0035743337124586106,
      "step": 208980
    },
    {
      "epoch": 59.321600908316775,
      "grad_norm": 0.1310119777917862,
      "learning_rate": 4.070281010502413e-05,
      "loss": 0.0003977134823799133,
      "step": 208990
    },
    {
      "epoch": 59.32443939824014,
      "grad_norm": 16.986696243286133,
      "learning_rate": 4.069997161510077e-05,
      "loss": 0.004254086315631867,
      "step": 209000
    },
    {
      "epoch": 59.32443939824014,
      "eval_accuracy": 0.9785718827494119,
      "eval_loss": 0.07572280615568161,
      "eval_runtime": 32.2164,
      "eval_samples_per_second": 488.167,
      "eval_steps_per_second": 7.636,
      "step": 209000
    },
    {
      "epoch": 59.32727788816349,
      "grad_norm": 4.5801496505737305,
      "learning_rate": 4.0697133125177406e-05,
      "loss": 0.0011519081890583038,
      "step": 209010
    },
    {
      "epoch": 59.330116378086856,
      "grad_norm": 0.07664173096418381,
      "learning_rate": 4.069429463525405e-05,
      "loss": 0.0027889978140592576,
      "step": 209020
    },
    {
      "epoch": 59.33295486801022,
      "grad_norm": 0.4735420346260071,
      "learning_rate": 4.069145614533069e-05,
      "loss": 0.000784255750477314,
      "step": 209030
    },
    {
      "epoch": 59.33579335793358,
      "grad_norm": 0.030794506892561913,
      "learning_rate": 4.0688617655407323e-05,
      "loss": 0.0007298711687326431,
      "step": 209040
    },
    {
      "epoch": 59.33863184785694,
      "grad_norm": 0.05999572575092316,
      "learning_rate": 4.0685779165483965e-05,
      "loss": 0.0011223534122109414,
      "step": 209050
    },
    {
      "epoch": 59.3414703377803,
      "grad_norm": 0.3529675602912903,
      "learning_rate": 4.0682940675560606e-05,
      "loss": 0.0014237245544791222,
      "step": 209060
    },
    {
      "epoch": 59.344308827703664,
      "grad_norm": 0.12415836751461029,
      "learning_rate": 4.068010218563724e-05,
      "loss": 0.0009725471958518028,
      "step": 209070
    },
    {
      "epoch": 59.34714731762702,
      "grad_norm": 0.02911434695124626,
      "learning_rate": 4.067726369571388e-05,
      "loss": 0.00031318534165620805,
      "step": 209080
    },
    {
      "epoch": 59.34998580755038,
      "grad_norm": 0.29892754554748535,
      "learning_rate": 4.0674425205790524e-05,
      "loss": 0.000219133123755455,
      "step": 209090
    },
    {
      "epoch": 59.352824297473745,
      "grad_norm": 0.05873022601008415,
      "learning_rate": 4.067158671586716e-05,
      "loss": 0.00021644551306962967,
      "step": 209100
    },
    {
      "epoch": 59.35566278739711,
      "grad_norm": 0.0941539853811264,
      "learning_rate": 4.06687482259438e-05,
      "loss": 0.00045733116567134855,
      "step": 209110
    },
    {
      "epoch": 59.358501277320464,
      "grad_norm": 0.032827965915203094,
      "learning_rate": 4.066590973602044e-05,
      "loss": 0.00032626911997795104,
      "step": 209120
    },
    {
      "epoch": 59.36133976724383,
      "grad_norm": 0.0531652569770813,
      "learning_rate": 4.0663071246097076e-05,
      "loss": 0.00023807883262634278,
      "step": 209130
    },
    {
      "epoch": 59.36417825716719,
      "grad_norm": 0.23717254400253296,
      "learning_rate": 4.066023275617372e-05,
      "loss": 0.0005705766379833221,
      "step": 209140
    },
    {
      "epoch": 59.367016747090545,
      "grad_norm": 0.04272928088903427,
      "learning_rate": 4.065739426625036e-05,
      "loss": 0.00021221786737442018,
      "step": 209150
    },
    {
      "epoch": 59.36985523701391,
      "grad_norm": 0.03711260110139847,
      "learning_rate": 4.0654555776327e-05,
      "loss": 0.0003359878435730934,
      "step": 209160
    },
    {
      "epoch": 59.37269372693727,
      "grad_norm": 0.06563179939985275,
      "learning_rate": 4.0651717286403634e-05,
      "loss": 0.0002058323472738266,
      "step": 209170
    },
    {
      "epoch": 59.37553221686063,
      "grad_norm": 0.08368227630853653,
      "learning_rate": 4.064887879648027e-05,
      "loss": 0.0001309594139456749,
      "step": 209180
    },
    {
      "epoch": 59.37837070678399,
      "grad_norm": 0.1590418964624405,
      "learning_rate": 4.064604030655692e-05,
      "loss": 0.0012128213420510292,
      "step": 209190
    },
    {
      "epoch": 59.38120919670735,
      "grad_norm": 0.010934002697467804,
      "learning_rate": 4.064320181663355e-05,
      "loss": 0.0003749147057533264,
      "step": 209200
    },
    {
      "epoch": 59.384047686630716,
      "grad_norm": 0.16526152193546295,
      "learning_rate": 4.064036332671019e-05,
      "loss": 0.00020508281886577606,
      "step": 209210
    },
    {
      "epoch": 59.38688617655407,
      "grad_norm": 0.20335642993450165,
      "learning_rate": 4.0637524836786834e-05,
      "loss": 0.0005382472649216652,
      "step": 209220
    },
    {
      "epoch": 59.389724666477434,
      "grad_norm": 0.03179871663451195,
      "learning_rate": 4.063468634686347e-05,
      "loss": 0.0005037881433963776,
      "step": 209230
    },
    {
      "epoch": 59.3925631564008,
      "grad_norm": 0.02179662324488163,
      "learning_rate": 4.063184785694011e-05,
      "loss": 0.001602536253631115,
      "step": 209240
    },
    {
      "epoch": 59.39540164632415,
      "grad_norm": 0.011978219263255596,
      "learning_rate": 4.062900936701675e-05,
      "loss": 0.001953318156301975,
      "step": 209250
    },
    {
      "epoch": 59.398240136247516,
      "grad_norm": 0.9310401678085327,
      "learning_rate": 4.0626170877093386e-05,
      "loss": 0.0006333401426672936,
      "step": 209260
    },
    {
      "epoch": 59.40107862617088,
      "grad_norm": 0.09759822487831116,
      "learning_rate": 4.062333238717003e-05,
      "loss": 0.00022952873259782792,
      "step": 209270
    },
    {
      "epoch": 59.403917116094235,
      "grad_norm": 0.02207164838910103,
      "learning_rate": 4.062049389724666e-05,
      "loss": 0.000393638014793396,
      "step": 209280
    },
    {
      "epoch": 59.4067556060176,
      "grad_norm": 0.01912384107708931,
      "learning_rate": 4.061765540732331e-05,
      "loss": 0.00027714092284440993,
      "step": 209290
    },
    {
      "epoch": 59.40959409594096,
      "grad_norm": 0.10842636227607727,
      "learning_rate": 4.0614816917399945e-05,
      "loss": 0.0004163352772593498,
      "step": 209300
    },
    {
      "epoch": 59.41243258586432,
      "grad_norm": 0.029367420822381973,
      "learning_rate": 4.061197842747658e-05,
      "loss": 0.0003536500036716461,
      "step": 209310
    },
    {
      "epoch": 59.41527107578768,
      "grad_norm": 0.016597745940089226,
      "learning_rate": 4.060913993755323e-05,
      "loss": 0.000644204206764698,
      "step": 209320
    },
    {
      "epoch": 59.41810956571104,
      "grad_norm": 10.22087574005127,
      "learning_rate": 4.060630144762986e-05,
      "loss": 0.00272232536226511,
      "step": 209330
    },
    {
      "epoch": 59.420948055634405,
      "grad_norm": 0.04013590142130852,
      "learning_rate": 4.0603462957706504e-05,
      "loss": 0.0004520803689956665,
      "step": 209340
    },
    {
      "epoch": 59.42378654555776,
      "grad_norm": 0.11514192819595337,
      "learning_rate": 4.0600624467783145e-05,
      "loss": 0.0005468053743243218,
      "step": 209350
    },
    {
      "epoch": 59.42662503548112,
      "grad_norm": 0.020102227106690407,
      "learning_rate": 4.059778597785978e-05,
      "loss": 0.00038982946425676347,
      "step": 209360
    },
    {
      "epoch": 59.429463525404486,
      "grad_norm": 0.025421474128961563,
      "learning_rate": 4.059494748793642e-05,
      "loss": 0.0003222128376364708,
      "step": 209370
    },
    {
      "epoch": 59.43230201532784,
      "grad_norm": 0.04584435746073723,
      "learning_rate": 4.0592108998013056e-05,
      "loss": 0.00035425964742898943,
      "step": 209380
    },
    {
      "epoch": 59.435140505251205,
      "grad_norm": 0.10768184065818787,
      "learning_rate": 4.05892705080897e-05,
      "loss": 0.000529603473842144,
      "step": 209390
    },
    {
      "epoch": 59.43797899517457,
      "grad_norm": 0.12898315489292145,
      "learning_rate": 4.058643201816634e-05,
      "loss": 0.00023846402764320372,
      "step": 209400
    },
    {
      "epoch": 59.44081748509793,
      "grad_norm": 0.18818816542625427,
      "learning_rate": 4.058359352824297e-05,
      "loss": 0.0008022658526897431,
      "step": 209410
    },
    {
      "epoch": 59.44365597502129,
      "grad_norm": 0.018970981240272522,
      "learning_rate": 4.0580755038319614e-05,
      "loss": 0.0002594944089651108,
      "step": 209420
    },
    {
      "epoch": 59.44649446494465,
      "grad_norm": 0.92731112241745,
      "learning_rate": 4.0577916548396256e-05,
      "loss": 0.00035805404186248777,
      "step": 209430
    },
    {
      "epoch": 59.44933295486801,
      "grad_norm": 0.0064618997275829315,
      "learning_rate": 4.057507805847289e-05,
      "loss": 0.0002789141610264778,
      "step": 209440
    },
    {
      "epoch": 59.45217144479137,
      "grad_norm": 0.12312639504671097,
      "learning_rate": 4.057223956854954e-05,
      "loss": 0.000753352977335453,
      "step": 209450
    },
    {
      "epoch": 59.45500993471473,
      "grad_norm": 0.024005092680454254,
      "learning_rate": 4.056940107862617e-05,
      "loss": 0.00020345598459243775,
      "step": 209460
    },
    {
      "epoch": 59.457848424638094,
      "grad_norm": 0.04824695736169815,
      "learning_rate": 4.0566562588702814e-05,
      "loss": 0.0020585261285305024,
      "step": 209470
    },
    {
      "epoch": 59.46068691456146,
      "grad_norm": 0.02838551625609398,
      "learning_rate": 4.056372409877945e-05,
      "loss": 0.000652468018233776,
      "step": 209480
    },
    {
      "epoch": 59.46352540448481,
      "grad_norm": 0.36792969703674316,
      "learning_rate": 4.056088560885609e-05,
      "loss": 0.00019667893648147582,
      "step": 209490
    },
    {
      "epoch": 59.466363894408175,
      "grad_norm": 0.034894924610853195,
      "learning_rate": 4.055804711893273e-05,
      "loss": 0.0005311084911227226,
      "step": 209500
    },
    {
      "epoch": 59.466363894408175,
      "eval_accuracy": 0.9791441470083296,
      "eval_loss": 0.0780511349439621,
      "eval_runtime": 32.3685,
      "eval_samples_per_second": 485.874,
      "eval_steps_per_second": 7.6,
      "step": 209500
    },
    {
      "epoch": 59.46920238433154,
      "grad_norm": 0.006686811335384846,
      "learning_rate": 4.0555208629009366e-05,
      "loss": 0.0003284519538283348,
      "step": 209510
    },
    {
      "epoch": 59.472040874254894,
      "grad_norm": 0.037137459963560104,
      "learning_rate": 4.055237013908601e-05,
      "loss": 0.002166958898305893,
      "step": 209520
    },
    {
      "epoch": 59.47487936417826,
      "grad_norm": 1.5821013450622559,
      "learning_rate": 4.054953164916265e-05,
      "loss": 0.0006052663549780846,
      "step": 209530
    },
    {
      "epoch": 59.47771785410162,
      "grad_norm": 0.11672259867191315,
      "learning_rate": 4.0546693159239284e-05,
      "loss": 0.000413600355386734,
      "step": 209540
    },
    {
      "epoch": 59.480556344024976,
      "grad_norm": 0.037047822028398514,
      "learning_rate": 4.0543854669315925e-05,
      "loss": 0.0001835683360695839,
      "step": 209550
    },
    {
      "epoch": 59.48339483394834,
      "grad_norm": 0.17093174159526825,
      "learning_rate": 4.0541016179392567e-05,
      "loss": 0.00030646361410617827,
      "step": 209560
    },
    {
      "epoch": 59.4862333238717,
      "grad_norm": 0.05653679370880127,
      "learning_rate": 4.05381776894692e-05,
      "loss": 0.0003068855032324791,
      "step": 209570
    },
    {
      "epoch": 59.489071813795064,
      "grad_norm": 8.30449390411377,
      "learning_rate": 4.053533919954584e-05,
      "loss": 0.0016902435570955276,
      "step": 209580
    },
    {
      "epoch": 59.49191030371842,
      "grad_norm": 0.3221970200538635,
      "learning_rate": 4.0532500709622484e-05,
      "loss": 0.0003624701872467995,
      "step": 209590
    },
    {
      "epoch": 59.49474879364178,
      "grad_norm": 0.03315882384777069,
      "learning_rate": 4.052966221969912e-05,
      "loss": 0.0019966138526797293,
      "step": 209600
    },
    {
      "epoch": 59.497587283565146,
      "grad_norm": 0.05506691709160805,
      "learning_rate": 4.052682372977576e-05,
      "loss": 0.00032933969050645827,
      "step": 209610
    },
    {
      "epoch": 59.5004257734885,
      "grad_norm": 0.10939708352088928,
      "learning_rate": 4.05239852398524e-05,
      "loss": 0.0012741835787892341,
      "step": 209620
    },
    {
      "epoch": 59.503264263411864,
      "grad_norm": 0.532612144947052,
      "learning_rate": 4.052114674992904e-05,
      "loss": 0.0015688519924879074,
      "step": 209630
    },
    {
      "epoch": 59.50610275333523,
      "grad_norm": 0.19323134422302246,
      "learning_rate": 4.051830826000568e-05,
      "loss": 0.0005793161690235138,
      "step": 209640
    },
    {
      "epoch": 59.50894124325858,
      "grad_norm": 0.06315860897302628,
      "learning_rate": 4.051546977008232e-05,
      "loss": 0.00038368292152881624,
      "step": 209650
    },
    {
      "epoch": 59.511779733181946,
      "grad_norm": 0.03336186707019806,
      "learning_rate": 4.051263128015896e-05,
      "loss": 0.0004970654845237732,
      "step": 209660
    },
    {
      "epoch": 59.51461822310531,
      "grad_norm": 0.06280676275491714,
      "learning_rate": 4.0509792790235595e-05,
      "loss": 0.0005774416029453277,
      "step": 209670
    },
    {
      "epoch": 59.51745671302867,
      "grad_norm": 0.10459858924150467,
      "learning_rate": 4.0506954300312236e-05,
      "loss": 0.004001777619123459,
      "step": 209680
    },
    {
      "epoch": 59.52029520295203,
      "grad_norm": 0.17402468621730804,
      "learning_rate": 4.050411581038888e-05,
      "loss": 0.0036673620343208313,
      "step": 209690
    },
    {
      "epoch": 59.52313369287539,
      "grad_norm": 0.21374420821666718,
      "learning_rate": 4.050127732046551e-05,
      "loss": 0.0003277307376265526,
      "step": 209700
    },
    {
      "epoch": 59.52597218279875,
      "grad_norm": 0.01814146898686886,
      "learning_rate": 4.049843883054215e-05,
      "loss": 0.0007129870355129242,
      "step": 209710
    },
    {
      "epoch": 59.52881067272211,
      "grad_norm": 0.10264556109905243,
      "learning_rate": 4.0495600340618795e-05,
      "loss": 0.0006165146827697754,
      "step": 209720
    },
    {
      "epoch": 59.53164916264547,
      "grad_norm": 0.034034572541713715,
      "learning_rate": 4.049276185069543e-05,
      "loss": 0.0002610733732581139,
      "step": 209730
    },
    {
      "epoch": 59.534487652568835,
      "grad_norm": 0.005010182037949562,
      "learning_rate": 4.048992336077207e-05,
      "loss": 0.0014965349808335304,
      "step": 209740
    },
    {
      "epoch": 59.53732614249219,
      "grad_norm": 0.09571391344070435,
      "learning_rate": 4.048708487084871e-05,
      "loss": 0.0007876871153712272,
      "step": 209750
    },
    {
      "epoch": 59.540164632415554,
      "grad_norm": 0.0985109955072403,
      "learning_rate": 4.048424638092535e-05,
      "loss": 0.0006372373551130295,
      "step": 209760
    },
    {
      "epoch": 59.54300312233892,
      "grad_norm": 0.13740719854831696,
      "learning_rate": 4.048140789100199e-05,
      "loss": 0.0007777223363518715,
      "step": 209770
    },
    {
      "epoch": 59.54584161226228,
      "grad_norm": 0.2751544713973999,
      "learning_rate": 4.047856940107862e-05,
      "loss": 0.0005603168159723282,
      "step": 209780
    },
    {
      "epoch": 59.548680102185635,
      "grad_norm": 0.05920284241437912,
      "learning_rate": 4.047573091115527e-05,
      "loss": 0.0023684268817305565,
      "step": 209790
    },
    {
      "epoch": 59.551518592109,
      "grad_norm": 0.006170699838548899,
      "learning_rate": 4.0472892421231905e-05,
      "loss": 0.0005256429314613343,
      "step": 209800
    },
    {
      "epoch": 59.55435708203236,
      "grad_norm": 0.33179283142089844,
      "learning_rate": 4.047005393130855e-05,
      "loss": 0.0015065357089042663,
      "step": 209810
    },
    {
      "epoch": 59.55719557195572,
      "grad_norm": 0.03181219846010208,
      "learning_rate": 4.046721544138519e-05,
      "loss": 0.0002704611048102379,
      "step": 209820
    },
    {
      "epoch": 59.56003406187908,
      "grad_norm": 0.060679949820041656,
      "learning_rate": 4.046437695146182e-05,
      "loss": 0.0006097139790654183,
      "step": 209830
    },
    {
      "epoch": 59.56287255180244,
      "grad_norm": 0.16131196916103363,
      "learning_rate": 4.0461538461538464e-05,
      "loss": 0.0023860838264226913,
      "step": 209840
    },
    {
      "epoch": 59.565711041725805,
      "grad_norm": 0.02322583645582199,
      "learning_rate": 4.0458699971615105e-05,
      "loss": 0.0022280106320977213,
      "step": 209850
    },
    {
      "epoch": 59.56854953164916,
      "grad_norm": 0.014256380498409271,
      "learning_rate": 4.045586148169174e-05,
      "loss": 0.0007122134789824486,
      "step": 209860
    },
    {
      "epoch": 59.571388021572524,
      "grad_norm": 0.11740683019161224,
      "learning_rate": 4.045302299176838e-05,
      "loss": 0.0016997901722788811,
      "step": 209870
    },
    {
      "epoch": 59.57422651149589,
      "grad_norm": 0.017211854457855225,
      "learning_rate": 4.0450184501845016e-05,
      "loss": 0.0005740350112318993,
      "step": 209880
    },
    {
      "epoch": 59.57706500141924,
      "grad_norm": 0.017044732347130775,
      "learning_rate": 4.044734601192166e-05,
      "loss": 0.0004944240674376488,
      "step": 209890
    },
    {
      "epoch": 59.579903491342606,
      "grad_norm": 0.04875865578651428,
      "learning_rate": 4.04445075219983e-05,
      "loss": 0.00034034717828035355,
      "step": 209900
    },
    {
      "epoch": 59.58274198126597,
      "grad_norm": 0.12790019810199738,
      "learning_rate": 4.044166903207493e-05,
      "loss": 0.0005253316834568977,
      "step": 209910
    },
    {
      "epoch": 59.585580471189324,
      "grad_norm": 0.6637232303619385,
      "learning_rate": 4.043883054215158e-05,
      "loss": 0.0008790219202637672,
      "step": 209920
    },
    {
      "epoch": 59.58841896111269,
      "grad_norm": 0.032912835478782654,
      "learning_rate": 4.0435992052228216e-05,
      "loss": 0.0020432541146874427,
      "step": 209930
    },
    {
      "epoch": 59.59125745103605,
      "grad_norm": 0.09318677335977554,
      "learning_rate": 4.043315356230486e-05,
      "loss": 0.0016679927706718446,
      "step": 209940
    },
    {
      "epoch": 59.59409594095941,
      "grad_norm": 0.016086770221590996,
      "learning_rate": 4.04303150723815e-05,
      "loss": 0.0019808027893304823,
      "step": 209950
    },
    {
      "epoch": 59.59693443088277,
      "grad_norm": 0.2881268560886383,
      "learning_rate": 4.0427476582458133e-05,
      "loss": 0.00036832373589277265,
      "step": 209960
    },
    {
      "epoch": 59.59977292080613,
      "grad_norm": 0.45400920510292053,
      "learning_rate": 4.042492194152711e-05,
      "loss": 0.008103539049625397,
      "step": 209970
    },
    {
      "epoch": 59.602611410729494,
      "grad_norm": 0.04531880468130112,
      "learning_rate": 4.042208345160375e-05,
      "loss": 0.00017920136451721191,
      "step": 209980
    },
    {
      "epoch": 59.60544990065285,
      "grad_norm": 1.6649457216262817,
      "learning_rate": 4.041924496168038e-05,
      "loss": 0.0029488641768693922,
      "step": 209990
    },
    {
      "epoch": 59.60828839057621,
      "grad_norm": 0.12997424602508545,
      "learning_rate": 4.041640647175703e-05,
      "loss": 0.0008970633149147034,
      "step": 210000
    },
    {
      "epoch": 59.60828839057621,
      "eval_accuracy": 0.9756469765371654,
      "eval_loss": 0.08912095427513123,
      "eval_runtime": 32.2349,
      "eval_samples_per_second": 487.887,
      "eval_steps_per_second": 7.631,
      "step": 210000
    },
    {
      "epoch": 59.611126880499576,
      "grad_norm": 0.0691925436258316,
      "learning_rate": 4.0413567981833665e-05,
      "loss": 0.003855973482131958,
      "step": 210010
    },
    {
      "epoch": 59.61396537042293,
      "grad_norm": 0.017711127176880836,
      "learning_rate": 4.041072949191031e-05,
      "loss": 0.0008747655898332596,
      "step": 210020
    },
    {
      "epoch": 59.616803860346295,
      "grad_norm": 1.0098774433135986,
      "learning_rate": 4.040789100198695e-05,
      "loss": 0.0004560330882668495,
      "step": 210030
    },
    {
      "epoch": 59.61964235026966,
      "grad_norm": 0.010808572173118591,
      "learning_rate": 4.040505251206358e-05,
      "loss": 0.0019031204283237457,
      "step": 210040
    },
    {
      "epoch": 59.62248084019302,
      "grad_norm": 0.05235874280333519,
      "learning_rate": 4.0402214022140224e-05,
      "loss": 0.0012681057676672935,
      "step": 210050
    },
    {
      "epoch": 59.625319330116376,
      "grad_norm": 0.6612359285354614,
      "learning_rate": 4.039937553221686e-05,
      "loss": 0.006590881943702697,
      "step": 210060
    },
    {
      "epoch": 59.62815782003974,
      "grad_norm": 0.020638419315218925,
      "learning_rate": 4.03965370422935e-05,
      "loss": 0.0025346560403704645,
      "step": 210070
    },
    {
      "epoch": 59.6309963099631,
      "grad_norm": 0.036871735006570816,
      "learning_rate": 4.039369855237014e-05,
      "loss": 0.0002831108868122101,
      "step": 210080
    },
    {
      "epoch": 59.63383479988646,
      "grad_norm": 0.5298957824707031,
      "learning_rate": 4.0390860062446776e-05,
      "loss": 0.0028772229328751565,
      "step": 210090
    },
    {
      "epoch": 59.63667328980982,
      "grad_norm": 0.01628963276743889,
      "learning_rate": 4.0388021572523424e-05,
      "loss": 0.00022240150719881057,
      "step": 210100
    },
    {
      "epoch": 59.63951177973318,
      "grad_norm": 0.004757504444569349,
      "learning_rate": 4.038518308260006e-05,
      "loss": 0.001880733110010624,
      "step": 210110
    },
    {
      "epoch": 59.64235026965654,
      "grad_norm": 0.09656154364347458,
      "learning_rate": 4.038234459267669e-05,
      "loss": 0.0008432568982243537,
      "step": 210120
    },
    {
      "epoch": 59.6451887595799,
      "grad_norm": 0.05068759247660637,
      "learning_rate": 4.037950610275334e-05,
      "loss": 0.007057788968086243,
      "step": 210130
    },
    {
      "epoch": 59.648027249503265,
      "grad_norm": 0.013869618065655231,
      "learning_rate": 4.0376667612829976e-05,
      "loss": 0.0023025652393698693,
      "step": 210140
    },
    {
      "epoch": 59.65086573942663,
      "grad_norm": 0.009449547156691551,
      "learning_rate": 4.037382912290662e-05,
      "loss": 0.0009074300527572632,
      "step": 210150
    },
    {
      "epoch": 59.653704229349984,
      "grad_norm": 0.3935542702674866,
      "learning_rate": 4.037099063298325e-05,
      "loss": 0.0011463796719908715,
      "step": 210160
    },
    {
      "epoch": 59.65654271927335,
      "grad_norm": 9.317618370056152,
      "learning_rate": 4.0368152143059893e-05,
      "loss": 0.002867347747087479,
      "step": 210170
    },
    {
      "epoch": 59.65938120919671,
      "grad_norm": 0.0292881578207016,
      "learning_rate": 4.0365313653136535e-05,
      "loss": 0.009246638417243958,
      "step": 210180
    },
    {
      "epoch": 59.662219699120065,
      "grad_norm": 3.4665985107421875,
      "learning_rate": 4.036247516321317e-05,
      "loss": 0.011939964443445205,
      "step": 210190
    },
    {
      "epoch": 59.66505818904343,
      "grad_norm": 1.0699975490570068,
      "learning_rate": 4.035963667328981e-05,
      "loss": 0.0005017891526222229,
      "step": 210200
    },
    {
      "epoch": 59.66789667896679,
      "grad_norm": 5.382299423217773,
      "learning_rate": 4.035679818336645e-05,
      "loss": 0.0055031668394804,
      "step": 210210
    },
    {
      "epoch": 59.67073516889015,
      "grad_norm": 0.012563210912048817,
      "learning_rate": 4.035395969344309e-05,
      "loss": 0.0008338011801242828,
      "step": 210220
    },
    {
      "epoch": 59.67357365881351,
      "grad_norm": 0.09791925549507141,
      "learning_rate": 4.0351121203519735e-05,
      "loss": 0.0018773911520838737,
      "step": 210230
    },
    {
      "epoch": 59.67641214873687,
      "grad_norm": 0.14343303442001343,
      "learning_rate": 4.034828271359637e-05,
      "loss": 0.0007456514984369278,
      "step": 210240
    },
    {
      "epoch": 59.679250638660235,
      "grad_norm": 0.024527471512556076,
      "learning_rate": 4.0345444223673004e-05,
      "loss": 0.0003636350855231285,
      "step": 210250
    },
    {
      "epoch": 59.68208912858359,
      "grad_norm": 0.013003482483327389,
      "learning_rate": 4.0342605733749645e-05,
      "loss": 0.00018302742391824723,
      "step": 210260
    },
    {
      "epoch": 59.684927618506954,
      "grad_norm": 2.3842275142669678,
      "learning_rate": 4.033976724382629e-05,
      "loss": 0.00068411435931921,
      "step": 210270
    },
    {
      "epoch": 59.68776610843032,
      "grad_norm": 0.21631591022014618,
      "learning_rate": 4.033692875390293e-05,
      "loss": 0.0005968298763036727,
      "step": 210280
    },
    {
      "epoch": 59.69060459835367,
      "grad_norm": 0.06462179124355316,
      "learning_rate": 4.033409026397956e-05,
      "loss": 0.0010957717895507812,
      "step": 210290
    },
    {
      "epoch": 59.693443088277036,
      "grad_norm": 0.09878084808588028,
      "learning_rate": 4.0331251774056204e-05,
      "loss": 0.002641037665307522,
      "step": 210300
    },
    {
      "epoch": 59.6962815782004,
      "grad_norm": 0.01770390197634697,
      "learning_rate": 4.0328413284132846e-05,
      "loss": 0.0013707032427191733,
      "step": 210310
    },
    {
      "epoch": 59.69912006812376,
      "grad_norm": 0.013016712851822376,
      "learning_rate": 4.032557479420948e-05,
      "loss": 0.0003132253885269165,
      "step": 210320
    },
    {
      "epoch": 59.70195855804712,
      "grad_norm": 0.05980139598250389,
      "learning_rate": 4.032273630428612e-05,
      "loss": 0.0006172595545649529,
      "step": 210330
    },
    {
      "epoch": 59.70479704797048,
      "grad_norm": 0.21739649772644043,
      "learning_rate": 4.031989781436276e-05,
      "loss": 0.0006977930665016174,
      "step": 210340
    },
    {
      "epoch": 59.70763553789384,
      "grad_norm": 0.08454285562038422,
      "learning_rate": 4.03170593244394e-05,
      "loss": 0.0003020014613866806,
      "step": 210350
    },
    {
      "epoch": 59.7104740278172,
      "grad_norm": 0.6004785299301147,
      "learning_rate": 4.031422083451604e-05,
      "loss": 0.000492185726761818,
      "step": 210360
    },
    {
      "epoch": 59.71331251774056,
      "grad_norm": 0.09345073252916336,
      "learning_rate": 4.031138234459268e-05,
      "loss": 0.00031666383147239686,
      "step": 210370
    },
    {
      "epoch": 59.716151007663925,
      "grad_norm": 0.017778825014829636,
      "learning_rate": 4.0308543854669315e-05,
      "loss": 0.00031419284641742706,
      "step": 210380
    },
    {
      "epoch": 59.71898949758728,
      "grad_norm": 0.08248747140169144,
      "learning_rate": 4.0305705364745956e-05,
      "loss": 0.0005164919421076775,
      "step": 210390
    },
    {
      "epoch": 59.72182798751064,
      "grad_norm": 2.7587473392486572,
      "learning_rate": 4.03028668748226e-05,
      "loss": 0.0025609491392970087,
      "step": 210400
    },
    {
      "epoch": 59.724666477434006,
      "grad_norm": 0.03173038363456726,
      "learning_rate": 4.030002838489923e-05,
      "loss": 0.00046058837324380875,
      "step": 210410
    },
    {
      "epoch": 59.72750496735737,
      "grad_norm": 0.16959813237190247,
      "learning_rate": 4.0297189894975874e-05,
      "loss": 0.003878995031118393,
      "step": 210420
    },
    {
      "epoch": 59.730343457280725,
      "grad_norm": 0.2533380687236786,
      "learning_rate": 4.0294351405052515e-05,
      "loss": 0.00029757600277662276,
      "step": 210430
    },
    {
      "epoch": 59.73318194720409,
      "grad_norm": 1.6967889070510864,
      "learning_rate": 4.0291512915129156e-05,
      "loss": 0.003287537395954132,
      "step": 210440
    },
    {
      "epoch": 59.73602043712745,
      "grad_norm": 0.01348136831074953,
      "learning_rate": 4.028867442520579e-05,
      "loss": 0.003794805705547333,
      "step": 210450
    },
    {
      "epoch": 59.738858927050806,
      "grad_norm": 1.0062379837036133,
      "learning_rate": 4.028583593528243e-05,
      "loss": 0.0009785156697034835,
      "step": 210460
    },
    {
      "epoch": 59.74169741697417,
      "grad_norm": 1.1179393529891968,
      "learning_rate": 4.0282997445359074e-05,
      "loss": 0.0007082734256982804,
      "step": 210470
    },
    {
      "epoch": 59.74453590689753,
      "grad_norm": 0.07823655754327774,
      "learning_rate": 4.028015895543571e-05,
      "loss": 0.0005363060161471366,
      "step": 210480
    },
    {
      "epoch": 59.74737439682089,
      "grad_norm": 0.1506243795156479,
      "learning_rate": 4.027732046551235e-05,
      "loss": 0.0011738160625100135,
      "step": 210490
    },
    {
      "epoch": 59.75021288674425,
      "grad_norm": 0.2562336325645447,
      "learning_rate": 4.027448197558899e-05,
      "loss": 0.0036632973700761796,
      "step": 210500
    },
    {
      "epoch": 59.75021288674425,
      "eval_accuracy": 0.9784447129140967,
      "eval_loss": 0.0848691314458847,
      "eval_runtime": 32.42,
      "eval_samples_per_second": 485.102,
      "eval_steps_per_second": 7.588,
      "step": 210500
    },
    {
      "epoch": 59.753051376667614,
      "grad_norm": 0.6554681658744812,
      "learning_rate": 4.0271643485665626e-05,
      "loss": 0.0009761733934283256,
      "step": 210510
    },
    {
      "epoch": 59.75588986659098,
      "grad_norm": 0.3382176458835602,
      "learning_rate": 4.026880499574227e-05,
      "loss": 0.0010560821741819383,
      "step": 210520
    },
    {
      "epoch": 59.75872835651433,
      "grad_norm": 0.6081188321113586,
      "learning_rate": 4.026596650581891e-05,
      "loss": 0.0015516530722379685,
      "step": 210530
    },
    {
      "epoch": 59.761566846437695,
      "grad_norm": 0.11982458084821701,
      "learning_rate": 4.026312801589554e-05,
      "loss": 0.0031986549496650697,
      "step": 210540
    },
    {
      "epoch": 59.76440533636106,
      "grad_norm": 0.08546780794858932,
      "learning_rate": 4.0260289525972184e-05,
      "loss": 0.0005486521869897843,
      "step": 210550
    },
    {
      "epoch": 59.767243826284414,
      "grad_norm": 0.04678124189376831,
      "learning_rate": 4.0257451036048826e-05,
      "loss": 0.001836995780467987,
      "step": 210560
    },
    {
      "epoch": 59.77008231620778,
      "grad_norm": 0.13658803701400757,
      "learning_rate": 4.025461254612547e-05,
      "loss": 0.0005544567480683327,
      "step": 210570
    },
    {
      "epoch": 59.77292080613114,
      "grad_norm": 0.1765584796667099,
      "learning_rate": 4.02517740562021e-05,
      "loss": 0.003435543179512024,
      "step": 210580
    },
    {
      "epoch": 59.7757592960545,
      "grad_norm": 0.08655203878879547,
      "learning_rate": 4.0248935566278736e-05,
      "loss": 0.011410191655158997,
      "step": 210590
    },
    {
      "epoch": 59.77859778597786,
      "grad_norm": 0.1530972123146057,
      "learning_rate": 4.0246097076355384e-05,
      "loss": 0.003193509578704834,
      "step": 210600
    },
    {
      "epoch": 59.78143627590122,
      "grad_norm": 0.10806639492511749,
      "learning_rate": 4.024325858643202e-05,
      "loss": 0.0005395375192165375,
      "step": 210610
    },
    {
      "epoch": 59.784274765824584,
      "grad_norm": 0.33152273297309875,
      "learning_rate": 4.024042009650866e-05,
      "loss": 0.0003270072862505913,
      "step": 210620
    },
    {
      "epoch": 59.78711325574794,
      "grad_norm": 0.007902678102254868,
      "learning_rate": 4.02375816065853e-05,
      "loss": 0.00020964201539754867,
      "step": 210630
    },
    {
      "epoch": 59.7899517456713,
      "grad_norm": 0.16649530827999115,
      "learning_rate": 4.0234743116661936e-05,
      "loss": 0.0004693400114774704,
      "step": 210640
    },
    {
      "epoch": 59.792790235594666,
      "grad_norm": 0.19708706438541412,
      "learning_rate": 4.023190462673858e-05,
      "loss": 0.0005341922864317894,
      "step": 210650
    },
    {
      "epoch": 59.79562872551802,
      "grad_norm": 0.0457807257771492,
      "learning_rate": 4.022906613681522e-05,
      "loss": 0.0014165516942739487,
      "step": 210660
    },
    {
      "epoch": 59.798467215441384,
      "grad_norm": 0.036932822316884995,
      "learning_rate": 4.0226227646891854e-05,
      "loss": 0.00040470268577337266,
      "step": 210670
    },
    {
      "epoch": 59.80130570536475,
      "grad_norm": 0.13668781518936157,
      "learning_rate": 4.0223389156968495e-05,
      "loss": 0.00048705916851758956,
      "step": 210680
    },
    {
      "epoch": 59.80414419528811,
      "grad_norm": 0.16861586272716522,
      "learning_rate": 4.022055066704513e-05,
      "loss": 0.003967031836509705,
      "step": 210690
    },
    {
      "epoch": 59.806982685211466,
      "grad_norm": 0.23143552243709564,
      "learning_rate": 4.021771217712178e-05,
      "loss": 0.00048358254134655,
      "step": 210700
    },
    {
      "epoch": 59.80982117513483,
      "grad_norm": 0.2588290572166443,
      "learning_rate": 4.021487368719841e-05,
      "loss": 0.0012872416526079177,
      "step": 210710
    },
    {
      "epoch": 59.81265966505819,
      "grad_norm": 0.010121159255504608,
      "learning_rate": 4.021203519727505e-05,
      "loss": 0.004198487102985382,
      "step": 210720
    },
    {
      "epoch": 59.81549815498155,
      "grad_norm": 0.1647483855485916,
      "learning_rate": 4.0209196707351695e-05,
      "loss": 0.0010204460471868516,
      "step": 210730
    },
    {
      "epoch": 59.81833664490491,
      "grad_norm": 0.09879885613918304,
      "learning_rate": 4.020635821742833e-05,
      "loss": 0.0009374627843499184,
      "step": 210740
    },
    {
      "epoch": 59.82117513482827,
      "grad_norm": 0.11048033833503723,
      "learning_rate": 4.020351972750497e-05,
      "loss": 0.0010093601420521736,
      "step": 210750
    },
    {
      "epoch": 59.82401362475163,
      "grad_norm": 0.035544995218515396,
      "learning_rate": 4.020068123758161e-05,
      "loss": 0.0022141106426715853,
      "step": 210760
    },
    {
      "epoch": 59.82685211467499,
      "grad_norm": 0.558935821056366,
      "learning_rate": 4.019784274765825e-05,
      "loss": 0.0016254760324954987,
      "step": 210770
    },
    {
      "epoch": 59.829690604598355,
      "grad_norm": 0.10218647122383118,
      "learning_rate": 4.019500425773489e-05,
      "loss": 0.00023770444095134736,
      "step": 210780
    },
    {
      "epoch": 59.83252909452172,
      "grad_norm": 0.024288896471261978,
      "learning_rate": 4.019216576781152e-05,
      "loss": 0.0006508735939860343,
      "step": 210790
    },
    {
      "epoch": 59.83536758444507,
      "grad_norm": 0.05534564331173897,
      "learning_rate": 4.0189327277888165e-05,
      "loss": 0.0007842393592000007,
      "step": 210800
    },
    {
      "epoch": 59.838206074368436,
      "grad_norm": 0.05127452686429024,
      "learning_rate": 4.0186488787964806e-05,
      "loss": 0.0006188103929162025,
      "step": 210810
    },
    {
      "epoch": 59.8410445642918,
      "grad_norm": 0.21261249482631683,
      "learning_rate": 4.018365029804144e-05,
      "loss": 0.0010247547179460526,
      "step": 210820
    },
    {
      "epoch": 59.843883054215155,
      "grad_norm": 0.016320256516337395,
      "learning_rate": 4.018081180811808e-05,
      "loss": 0.00053226538002491,
      "step": 210830
    },
    {
      "epoch": 59.84672154413852,
      "grad_norm": 0.074388287961483,
      "learning_rate": 4.017797331819472e-05,
      "loss": 0.005699943006038666,
      "step": 210840
    },
    {
      "epoch": 59.84956003406188,
      "grad_norm": 0.04246428608894348,
      "learning_rate": 4.017513482827136e-05,
      "loss": 0.0010278725996613503,
      "step": 210850
    },
    {
      "epoch": 59.85239852398524,
      "grad_norm": 0.015400438569486141,
      "learning_rate": 4.0172296338348006e-05,
      "loss": 0.00028403401374816896,
      "step": 210860
    },
    {
      "epoch": 59.8552370139086,
      "grad_norm": 0.2638717591762543,
      "learning_rate": 4.016945784842464e-05,
      "loss": 0.0018944313749670982,
      "step": 210870
    },
    {
      "epoch": 59.85807550383196,
      "grad_norm": 0.004116884898394346,
      "learning_rate": 4.0166619358501275e-05,
      "loss": 0.00039854831993579865,
      "step": 210880
    },
    {
      "epoch": 59.860913993755325,
      "grad_norm": 0.0444093756377697,
      "learning_rate": 4.0163780868577917e-05,
      "loss": 0.00032896660268306733,
      "step": 210890
    },
    {
      "epoch": 59.86375248367868,
      "grad_norm": 0.040541838854551315,
      "learning_rate": 4.016094237865456e-05,
      "loss": 0.0003893911838531494,
      "step": 210900
    },
    {
      "epoch": 59.866590973602044,
      "grad_norm": 0.026285527274012566,
      "learning_rate": 4.01581038887312e-05,
      "loss": 0.00041197557002305987,
      "step": 210910
    },
    {
      "epoch": 59.86942946352541,
      "grad_norm": 0.17691224813461304,
      "learning_rate": 4.0155265398807834e-05,
      "loss": 0.00020665060728788375,
      "step": 210920
    },
    {
      "epoch": 59.87226795344876,
      "grad_norm": 0.0331229530274868,
      "learning_rate": 4.0152426908884475e-05,
      "loss": 0.00011796709150075912,
      "step": 210930
    },
    {
      "epoch": 59.875106443372125,
      "grad_norm": 0.030056696385145187,
      "learning_rate": 4.014958841896112e-05,
      "loss": 0.0006768664345145226,
      "step": 210940
    },
    {
      "epoch": 59.87794493329549,
      "grad_norm": 0.15968599915504456,
      "learning_rate": 4.014674992903775e-05,
      "loss": 0.0002116737887263298,
      "step": 210950
    },
    {
      "epoch": 59.880783423218844,
      "grad_norm": 0.021295493468642235,
      "learning_rate": 4.014391143911439e-05,
      "loss": 0.0007686126977205276,
      "step": 210960
    },
    {
      "epoch": 59.88362191314221,
      "grad_norm": 0.3816732168197632,
      "learning_rate": 4.0141072949191034e-05,
      "loss": 0.0002393105998635292,
      "step": 210970
    },
    {
      "epoch": 59.88646040306557,
      "grad_norm": 0.012521645054221153,
      "learning_rate": 4.013823445926767e-05,
      "loss": 0.00029872506856918333,
      "step": 210980
    },
    {
      "epoch": 59.88929889298893,
      "grad_norm": 0.0838887020945549,
      "learning_rate": 4.013539596934431e-05,
      "loss": 0.0003717290237545967,
      "step": 210990
    },
    {
      "epoch": 59.89213738291229,
      "grad_norm": 0.0840463861823082,
      "learning_rate": 4.013255747942095e-05,
      "loss": 0.00020768344402313232,
      "step": 211000
    },
    {
      "epoch": 59.89213738291229,
      "eval_accuracy": 0.979779996184905,
      "eval_loss": 0.07591086626052856,
      "eval_runtime": 32.9449,
      "eval_samples_per_second": 477.372,
      "eval_steps_per_second": 7.467,
      "step": 211000
    },
    {
      "epoch": 59.89497587283565,
      "grad_norm": 0.07254587858915329,
      "learning_rate": 4.0129718989497586e-05,
      "loss": 0.00016259998083114623,
      "step": 211010
    },
    {
      "epoch": 59.897814362759014,
      "grad_norm": 0.007282676640897989,
      "learning_rate": 4.012688049957423e-05,
      "loss": 0.00018723011016845702,
      "step": 211020
    },
    {
      "epoch": 59.90065285268237,
      "grad_norm": 0.03641302138566971,
      "learning_rate": 4.012404200965087e-05,
      "loss": 0.0009120993316173553,
      "step": 211030
    },
    {
      "epoch": 59.90349134260573,
      "grad_norm": 0.029335318133234978,
      "learning_rate": 4.012120351972751e-05,
      "loss": 0.0012496639043092727,
      "step": 211040
    },
    {
      "epoch": 59.906329832529096,
      "grad_norm": 3.793973922729492,
      "learning_rate": 4.0118365029804145e-05,
      "loss": 0.00088198222219944,
      "step": 211050
    },
    {
      "epoch": 59.90916832245246,
      "grad_norm": 0.031840045005083084,
      "learning_rate": 4.0115526539880786e-05,
      "loss": 0.001798710972070694,
      "step": 211060
    },
    {
      "epoch": 59.912006812375814,
      "grad_norm": 0.03873513266444206,
      "learning_rate": 4.011268804995743e-05,
      "loss": 0.0021045539528131486,
      "step": 211070
    },
    {
      "epoch": 59.91484530229918,
      "grad_norm": 0.019491327926516533,
      "learning_rate": 4.010984956003406e-05,
      "loss": 0.0003326389938592911,
      "step": 211080
    },
    {
      "epoch": 59.91768379222254,
      "grad_norm": 0.033514540642499924,
      "learning_rate": 4.0107011070110703e-05,
      "loss": 0.0003185467794537544,
      "step": 211090
    },
    {
      "epoch": 59.920522282145896,
      "grad_norm": 0.06798397749662399,
      "learning_rate": 4.0104172580187345e-05,
      "loss": 0.00032975375652313235,
      "step": 211100
    },
    {
      "epoch": 59.92336077206926,
      "grad_norm": 0.013308022171258926,
      "learning_rate": 4.010133409026398e-05,
      "loss": 0.00027086157351732256,
      "step": 211110
    },
    {
      "epoch": 59.92619926199262,
      "grad_norm": 15.317023277282715,
      "learning_rate": 4.009849560034062e-05,
      "loss": 0.004647216200828553,
      "step": 211120
    },
    {
      "epoch": 59.92903775191598,
      "grad_norm": 0.0389261357486248,
      "learning_rate": 4.009565711041726e-05,
      "loss": 0.0007181994616985321,
      "step": 211130
    },
    {
      "epoch": 59.93187624183934,
      "grad_norm": 0.03205104172229767,
      "learning_rate": 4.00928186204939e-05,
      "loss": 0.007881271839141845,
      "step": 211140
    },
    {
      "epoch": 59.9347147317627,
      "grad_norm": 0.013270176015794277,
      "learning_rate": 4.008998013057054e-05,
      "loss": 0.0005778089165687561,
      "step": 211150
    },
    {
      "epoch": 59.937553221686066,
      "grad_norm": 0.1708674281835556,
      "learning_rate": 4.008714164064718e-05,
      "loss": 0.000528215616941452,
      "step": 211160
    },
    {
      "epoch": 59.94039171160942,
      "grad_norm": 0.010576403699815273,
      "learning_rate": 4.008430315072382e-05,
      "loss": 0.002294227108359337,
      "step": 211170
    },
    {
      "epoch": 59.943230201532785,
      "grad_norm": 0.21540026366710663,
      "learning_rate": 4.0081464660800455e-05,
      "loss": 0.001164194382727146,
      "step": 211180
    },
    {
      "epoch": 59.94606869145615,
      "grad_norm": 0.04402341693639755,
      "learning_rate": 4.007862617087709e-05,
      "loss": 0.0010401466861367226,
      "step": 211190
    },
    {
      "epoch": 59.948907181379504,
      "grad_norm": 0.03376477211713791,
      "learning_rate": 4.007578768095374e-05,
      "loss": 0.000520293414592743,
      "step": 211200
    },
    {
      "epoch": 59.95174567130287,
      "grad_norm": 0.2946440875530243,
      "learning_rate": 4.007294919103037e-05,
      "loss": 0.0001273093745112419,
      "step": 211210
    },
    {
      "epoch": 59.95458416122623,
      "grad_norm": 0.06882165372371674,
      "learning_rate": 4.0070110701107014e-05,
      "loss": 0.0007314592599868775,
      "step": 211220
    },
    {
      "epoch": 59.957422651149585,
      "grad_norm": 1.1934900283813477,
      "learning_rate": 4.0067272211183656e-05,
      "loss": 0.0006105324253439903,
      "step": 211230
    },
    {
      "epoch": 59.96026114107295,
      "grad_norm": 0.0524231381714344,
      "learning_rate": 4.006443372126029e-05,
      "loss": 0.0005541209131479264,
      "step": 211240
    },
    {
      "epoch": 59.96309963099631,
      "grad_norm": 0.023173538967967033,
      "learning_rate": 4.006159523133693e-05,
      "loss": 0.002034512721002102,
      "step": 211250
    },
    {
      "epoch": 59.965938120919674,
      "grad_norm": 0.13710714876651764,
      "learning_rate": 4.005875674141357e-05,
      "loss": 0.0002684036269783974,
      "step": 211260
    },
    {
      "epoch": 59.96877661084303,
      "grad_norm": 0.03873835876584053,
      "learning_rate": 4.005591825149021e-05,
      "loss": 0.00039734169840812685,
      "step": 211270
    },
    {
      "epoch": 59.97161510076639,
      "grad_norm": 0.04001837223768234,
      "learning_rate": 4.005307976156685e-05,
      "loss": 0.00018114298582077025,
      "step": 211280
    },
    {
      "epoch": 59.974453590689755,
      "grad_norm": 0.013866518624126911,
      "learning_rate": 4.0050241271643483e-05,
      "loss": 0.000873916782438755,
      "step": 211290
    },
    {
      "epoch": 59.97729208061311,
      "grad_norm": 0.041815295815467834,
      "learning_rate": 4.0047402781720125e-05,
      "loss": 0.00014318842440843582,
      "step": 211300
    },
    {
      "epoch": 59.980130570536474,
      "grad_norm": 0.898971438407898,
      "learning_rate": 4.0044564291796766e-05,
      "loss": 0.0006424741819500923,
      "step": 211310
    },
    {
      "epoch": 59.98296906045984,
      "grad_norm": 0.01878456585109234,
      "learning_rate": 4.00417258018734e-05,
      "loss": 0.00011926684528589249,
      "step": 211320
    },
    {
      "epoch": 59.98580755038319,
      "grad_norm": 0.0746101513504982,
      "learning_rate": 4.003888731195005e-05,
      "loss": 0.004294080659747124,
      "step": 211330
    },
    {
      "epoch": 59.988646040306556,
      "grad_norm": 0.13133519887924194,
      "learning_rate": 4.0036048822026684e-05,
      "loss": 0.0015269083902239799,
      "step": 211340
    },
    {
      "epoch": 59.99148453022992,
      "grad_norm": 0.03762432560324669,
      "learning_rate": 4.003321033210332e-05,
      "loss": 0.0003268269822001457,
      "step": 211350
    },
    {
      "epoch": 59.99432302015328,
      "grad_norm": 0.06108442321419716,
      "learning_rate": 4.0030371842179966e-05,
      "loss": 0.0007362829521298409,
      "step": 211360
    },
    {
      "epoch": 59.99716151007664,
      "grad_norm": 0.0351344496011734,
      "learning_rate": 4.00275333522566e-05,
      "loss": 0.0020063603296875954,
      "step": 211370
    },
    {
      "epoch": 60.0,
      "grad_norm": 0.002290131291374564,
      "learning_rate": 4.002469486233324e-05,
      "loss": 0.00046923854388296603,
      "step": 211380
    },
    {
      "epoch": 60.00283848992336,
      "grad_norm": 0.1879780888557434,
      "learning_rate": 4.002185637240988e-05,
      "loss": 0.0007167583331465721,
      "step": 211390
    },
    {
      "epoch": 60.00567697984672,
      "grad_norm": 0.24673084914684296,
      "learning_rate": 4.001901788248652e-05,
      "loss": 0.0003234041854739189,
      "step": 211400
    },
    {
      "epoch": 60.00851546977008,
      "grad_norm": 0.5992982983589172,
      "learning_rate": 4.001617939256316e-05,
      "loss": 0.00522833839058876,
      "step": 211410
    },
    {
      "epoch": 60.011353959693444,
      "grad_norm": 3.305943489074707,
      "learning_rate": 4.0013340902639794e-05,
      "loss": 0.0008543869480490685,
      "step": 211420
    },
    {
      "epoch": 60.01419244961681,
      "grad_norm": 6.160771369934082,
      "learning_rate": 4.0010502412716436e-05,
      "loss": 0.0008202288299798965,
      "step": 211430
    },
    {
      "epoch": 60.01703093954016,
      "grad_norm": 0.9627863168716431,
      "learning_rate": 4.000766392279308e-05,
      "loss": 0.0002718629315495491,
      "step": 211440
    },
    {
      "epoch": 60.019869429463526,
      "grad_norm": 0.021657360717654228,
      "learning_rate": 4.000482543286971e-05,
      "loss": 0.0003370434045791626,
      "step": 211450
    },
    {
      "epoch": 60.02270791938689,
      "grad_norm": 0.025875987485051155,
      "learning_rate": 4.000198694294636e-05,
      "loss": 0.0037871208041906357,
      "step": 211460
    },
    {
      "epoch": 60.025546409310245,
      "grad_norm": 0.021952837705612183,
      "learning_rate": 3.9999148453022994e-05,
      "loss": 0.0014201078563928604,
      "step": 211470
    },
    {
      "epoch": 60.02838489923361,
      "grad_norm": 0.6657183766365051,
      "learning_rate": 3.999630996309963e-05,
      "loss": 0.00039950571954250336,
      "step": 211480
    },
    {
      "epoch": 60.03122338915697,
      "grad_norm": 0.03816501051187515,
      "learning_rate": 3.999347147317627e-05,
      "loss": 0.0010640479624271393,
      "step": 211490
    },
    {
      "epoch": 60.034061879080326,
      "grad_norm": 0.018671464174985886,
      "learning_rate": 3.999063298325291e-05,
      "loss": 0.0004991693422198296,
      "step": 211500
    },
    {
      "epoch": 60.034061879080326,
      "eval_accuracy": 0.9791441470083296,
      "eval_loss": 0.0776624083518982,
      "eval_runtime": 32.9764,
      "eval_samples_per_second": 476.917,
      "eval_steps_per_second": 7.46,
      "step": 211500
    },
    {
      "epoch": 60.03690036900369,
      "grad_norm": 0.020353300496935844,
      "learning_rate": 3.998779449332955e-05,
      "loss": 0.002411297336220741,
      "step": 211510
    },
    {
      "epoch": 60.03973885892705,
      "grad_norm": 0.11460176855325699,
      "learning_rate": 3.998495600340619e-05,
      "loss": 0.0033170025795698165,
      "step": 211520
    },
    {
      "epoch": 60.042577348850415,
      "grad_norm": 0.028718817979097366,
      "learning_rate": 3.998211751348283e-05,
      "loss": 0.00036693159490823745,
      "step": 211530
    },
    {
      "epoch": 60.04541583877377,
      "grad_norm": 0.03270211070775986,
      "learning_rate": 3.997927902355947e-05,
      "loss": 0.0003207683563232422,
      "step": 211540
    },
    {
      "epoch": 60.04825432869713,
      "grad_norm": 0.18885424733161926,
      "learning_rate": 3.9976440533636105e-05,
      "loss": 0.00022667460143566132,
      "step": 211550
    },
    {
      "epoch": 60.051092818620496,
      "grad_norm": 0.04822346195578575,
      "learning_rate": 3.9973602043712746e-05,
      "loss": 0.00021669473499059678,
      "step": 211560
    },
    {
      "epoch": 60.05393130854385,
      "grad_norm": 0.0725894421339035,
      "learning_rate": 3.997076355378939e-05,
      "loss": 0.00016036089509725572,
      "step": 211570
    },
    {
      "epoch": 60.056769798467215,
      "grad_norm": 0.03381437063217163,
      "learning_rate": 3.996792506386602e-05,
      "loss": 0.0001748962327837944,
      "step": 211580
    },
    {
      "epoch": 60.05960828839058,
      "grad_norm": 0.002026122761890292,
      "learning_rate": 3.9965086573942664e-05,
      "loss": 0.0001487504690885544,
      "step": 211590
    },
    {
      "epoch": 60.062446778313934,
      "grad_norm": 0.04309766739606857,
      "learning_rate": 3.9962248084019305e-05,
      "loss": 0.0001339823007583618,
      "step": 211600
    },
    {
      "epoch": 60.0652852682373,
      "grad_norm": 0.015863489359617233,
      "learning_rate": 3.995940959409594e-05,
      "loss": 0.00028341971337795255,
      "step": 211610
    },
    {
      "epoch": 60.06812375816066,
      "grad_norm": 0.0060238041914999485,
      "learning_rate": 3.995657110417258e-05,
      "loss": 0.00032542645931243896,
      "step": 211620
    },
    {
      "epoch": 60.07096224808402,
      "grad_norm": 0.03819098323583603,
      "learning_rate": 3.995373261424922e-05,
      "loss": 0.00019183997064828873,
      "step": 211630
    },
    {
      "epoch": 60.07380073800738,
      "grad_norm": 2.3974475860595703,
      "learning_rate": 3.9950894124325864e-05,
      "loss": 0.0005217483267188072,
      "step": 211640
    },
    {
      "epoch": 60.07663922793074,
      "grad_norm": 0.04059465602040291,
      "learning_rate": 3.99480556344025e-05,
      "loss": 0.0011557983234524727,
      "step": 211650
    },
    {
      "epoch": 60.079477717854104,
      "grad_norm": 0.1702825278043747,
      "learning_rate": 3.994521714447914e-05,
      "loss": 0.00036404523998498914,
      "step": 211660
    },
    {
      "epoch": 60.08231620777746,
      "grad_norm": 0.06325136870145798,
      "learning_rate": 3.994237865455578e-05,
      "loss": 0.0010485131293535232,
      "step": 211670
    },
    {
      "epoch": 60.08515469770082,
      "grad_norm": 0.5279908776283264,
      "learning_rate": 3.9939540164632416e-05,
      "loss": 0.0001712080091238022,
      "step": 211680
    },
    {
      "epoch": 60.087993187624186,
      "grad_norm": 0.012116272002458572,
      "learning_rate": 3.993670167470906e-05,
      "loss": 0.0006543787196278573,
      "step": 211690
    },
    {
      "epoch": 60.09083167754754,
      "grad_norm": 0.4666938781738281,
      "learning_rate": 3.99338631847857e-05,
      "loss": 0.0004225371405482292,
      "step": 211700
    },
    {
      "epoch": 60.093670167470904,
      "grad_norm": 0.3102080225944519,
      "learning_rate": 3.993102469486233e-05,
      "loss": 0.0001693522557616234,
      "step": 211710
    },
    {
      "epoch": 60.09650865739427,
      "grad_norm": 3.2004315853118896,
      "learning_rate": 3.9928186204938974e-05,
      "loss": 0.002993980050086975,
      "step": 211720
    },
    {
      "epoch": 60.09934714731763,
      "grad_norm": 0.05361263453960419,
      "learning_rate": 3.9925347715015616e-05,
      "loss": 0.000644853338599205,
      "step": 211730
    },
    {
      "epoch": 60.102185637240986,
      "grad_norm": 2.5461912155151367,
      "learning_rate": 3.992250922509225e-05,
      "loss": 0.0009059201925992966,
      "step": 211740
    },
    {
      "epoch": 60.10502412716435,
      "grad_norm": 0.009339678101241589,
      "learning_rate": 3.991967073516889e-05,
      "loss": 0.0006791934370994568,
      "step": 211750
    },
    {
      "epoch": 60.10786261708771,
      "grad_norm": 0.5242336988449097,
      "learning_rate": 3.991683224524553e-05,
      "loss": 0.00024700555950403215,
      "step": 211760
    },
    {
      "epoch": 60.11070110701107,
      "grad_norm": 0.26593783497810364,
      "learning_rate": 3.991399375532217e-05,
      "loss": 0.00022332407534122466,
      "step": 211770
    },
    {
      "epoch": 60.11353959693443,
      "grad_norm": 0.012928388081490993,
      "learning_rate": 3.991115526539881e-05,
      "loss": 0.008300552517175675,
      "step": 211780
    },
    {
      "epoch": 60.11637808685779,
      "grad_norm": 0.023058686405420303,
      "learning_rate": 3.990831677547545e-05,
      "loss": 0.0002999361604452133,
      "step": 211790
    },
    {
      "epoch": 60.119216576781156,
      "grad_norm": 0.3228331208229065,
      "learning_rate": 3.990547828555209e-05,
      "loss": 0.0012249099090695381,
      "step": 211800
    },
    {
      "epoch": 60.12205506670451,
      "grad_norm": 0.053107134997844696,
      "learning_rate": 3.9902639795628727e-05,
      "loss": 0.00047668609768152236,
      "step": 211810
    },
    {
      "epoch": 60.124893556627875,
      "grad_norm": 0.14417718350887299,
      "learning_rate": 3.989980130570536e-05,
      "loss": 0.000939561054110527,
      "step": 211820
    },
    {
      "epoch": 60.12773204655124,
      "grad_norm": 0.06709910184144974,
      "learning_rate": 3.989696281578201e-05,
      "loss": 0.0006642637774348259,
      "step": 211830
    },
    {
      "epoch": 60.13057053647459,
      "grad_norm": 0.42655712366104126,
      "learning_rate": 3.9894124325858644e-05,
      "loss": 0.003091414086520672,
      "step": 211840
    },
    {
      "epoch": 60.133409026397956,
      "grad_norm": 0.032784584909677505,
      "learning_rate": 3.9891285835935285e-05,
      "loss": 0.003954719752073288,
      "step": 211850
    },
    {
      "epoch": 60.13624751632132,
      "grad_norm": 0.08716005086898804,
      "learning_rate": 3.988844734601193e-05,
      "loss": 0.0018747624009847641,
      "step": 211860
    },
    {
      "epoch": 60.139086006244675,
      "grad_norm": 0.15843982994556427,
      "learning_rate": 3.988560885608856e-05,
      "loss": 0.00017254594713449477,
      "step": 211870
    },
    {
      "epoch": 60.14192449616804,
      "grad_norm": 0.4905562698841095,
      "learning_rate": 3.98827703661652e-05,
      "loss": 0.0009080246090888977,
      "step": 211880
    },
    {
      "epoch": 60.1447629860914,
      "grad_norm": 0.02251243032515049,
      "learning_rate": 3.9879931876241844e-05,
      "loss": 0.0002464715391397476,
      "step": 211890
    },
    {
      "epoch": 60.14760147601476,
      "grad_norm": 0.009806018322706223,
      "learning_rate": 3.987709338631848e-05,
      "loss": 0.0007453549653291703,
      "step": 211900
    },
    {
      "epoch": 60.15043996593812,
      "grad_norm": 0.10268932580947876,
      "learning_rate": 3.987425489639512e-05,
      "loss": 0.0008415870368480682,
      "step": 211910
    },
    {
      "epoch": 60.15327845586148,
      "grad_norm": 0.15582704544067383,
      "learning_rate": 3.9871416406471755e-05,
      "loss": 0.000776963122189045,
      "step": 211920
    },
    {
      "epoch": 60.156116945784845,
      "grad_norm": 0.1932487040758133,
      "learning_rate": 3.98685779165484e-05,
      "loss": 0.008484239131212235,
      "step": 211930
    },
    {
      "epoch": 60.1589554357082,
      "grad_norm": 0.03270683065056801,
      "learning_rate": 3.986573942662504e-05,
      "loss": 0.00016201771795749665,
      "step": 211940
    },
    {
      "epoch": 60.161793925631564,
      "grad_norm": 0.45067501068115234,
      "learning_rate": 3.986290093670167e-05,
      "loss": 0.0008119748905301094,
      "step": 211950
    },
    {
      "epoch": 60.16463241555493,
      "grad_norm": 0.01790882833302021,
      "learning_rate": 3.986006244677832e-05,
      "loss": 0.00020392239093780518,
      "step": 211960
    },
    {
      "epoch": 60.16747090547828,
      "grad_norm": 0.052002277225255966,
      "learning_rate": 3.9857223956854955e-05,
      "loss": 0.0023178767412900926,
      "step": 211970
    },
    {
      "epoch": 60.170309395401645,
      "grad_norm": 0.19074980914592743,
      "learning_rate": 3.9854385466931596e-05,
      "loss": 0.0009038664400577545,
      "step": 211980
    },
    {
      "epoch": 60.17314788532501,
      "grad_norm": 20.184677124023438,
      "learning_rate": 3.985154697700824e-05,
      "loss": 0.011955204606056213,
      "step": 211990
    },
    {
      "epoch": 60.17598637524837,
      "grad_norm": 0.3196350038051605,
      "learning_rate": 3.984870848708487e-05,
      "loss": 0.0003661520779132843,
      "step": 212000
    },
    {
      "epoch": 60.17598637524837,
      "eval_accuracy": 0.9708781077128505,
      "eval_loss": 0.11928322166204453,
      "eval_runtime": 32.6167,
      "eval_samples_per_second": 482.176,
      "eval_steps_per_second": 7.542,
      "step": 212000
    },
    {
      "epoch": 60.17882486517173,
      "grad_norm": 0.02429787814617157,
      "learning_rate": 3.984586999716151e-05,
      "loss": 0.0035189088433980943,
      "step": 212010
    },
    {
      "epoch": 60.18166335509509,
      "grad_norm": 0.13881230354309082,
      "learning_rate": 3.984303150723815e-05,
      "loss": 0.0012731853872537612,
      "step": 212020
    },
    {
      "epoch": 60.18450184501845,
      "grad_norm": 0.31600329279899597,
      "learning_rate": 3.984019301731479e-05,
      "loss": 0.00039008036255836485,
      "step": 212030
    },
    {
      "epoch": 60.18734033494181,
      "grad_norm": 1.4159480333328247,
      "learning_rate": 3.983735452739143e-05,
      "loss": 0.0005129110068082809,
      "step": 212040
    },
    {
      "epoch": 60.19017882486517,
      "grad_norm": 3.4235918521881104,
      "learning_rate": 3.9834516037468065e-05,
      "loss": 0.0004810469225049019,
      "step": 212050
    },
    {
      "epoch": 60.193017314788534,
      "grad_norm": 0.04775285720825195,
      "learning_rate": 3.9831677547544713e-05,
      "loss": 0.00014351978898048402,
      "step": 212060
    },
    {
      "epoch": 60.19585580471189,
      "grad_norm": 0.20962877571582794,
      "learning_rate": 3.982883905762135e-05,
      "loss": 0.0025354975834488867,
      "step": 212070
    },
    {
      "epoch": 60.19869429463525,
      "grad_norm": 0.054245539009571075,
      "learning_rate": 3.982600056769798e-05,
      "loss": 0.0006039263680577278,
      "step": 212080
    },
    {
      "epoch": 60.201532784558616,
      "grad_norm": 0.1491963118314743,
      "learning_rate": 3.982316207777463e-05,
      "loss": 0.0005111075937747955,
      "step": 212090
    },
    {
      "epoch": 60.20437127448198,
      "grad_norm": 0.7442827820777893,
      "learning_rate": 3.9820323587851265e-05,
      "loss": 0.0005620602518320083,
      "step": 212100
    },
    {
      "epoch": 60.207209764405334,
      "grad_norm": 0.8530558347702026,
      "learning_rate": 3.981748509792791e-05,
      "loss": 0.0006948946043848992,
      "step": 212110
    },
    {
      "epoch": 60.2100482543287,
      "grad_norm": 0.0891876220703125,
      "learning_rate": 3.981464660800454e-05,
      "loss": 0.0005162809044122696,
      "step": 212120
    },
    {
      "epoch": 60.21288674425206,
      "grad_norm": 0.03871693089604378,
      "learning_rate": 3.981180811808118e-05,
      "loss": 0.00027050022035837176,
      "step": 212130
    },
    {
      "epoch": 60.215725234175416,
      "grad_norm": 0.132669135928154,
      "learning_rate": 3.9808969628157824e-05,
      "loss": 0.0005691112950444221,
      "step": 212140
    },
    {
      "epoch": 60.21856372409878,
      "grad_norm": 0.5355362892150879,
      "learning_rate": 3.980613113823446e-05,
      "loss": 0.001176760345697403,
      "step": 212150
    },
    {
      "epoch": 60.22140221402214,
      "grad_norm": 1.4414494037628174,
      "learning_rate": 3.98032926483111e-05,
      "loss": 0.000702892430126667,
      "step": 212160
    },
    {
      "epoch": 60.2242407039455,
      "grad_norm": 0.34948575496673584,
      "learning_rate": 3.980045415838774e-05,
      "loss": 0.00034031178802251816,
      "step": 212170
    },
    {
      "epoch": 60.22707919386886,
      "grad_norm": 0.18872098624706268,
      "learning_rate": 3.9797615668464376e-05,
      "loss": 0.0003749992698431015,
      "step": 212180
    },
    {
      "epoch": 60.22991768379222,
      "grad_norm": 0.1686212420463562,
      "learning_rate": 3.979477717854102e-05,
      "loss": 0.0004081733524799347,
      "step": 212190
    },
    {
      "epoch": 60.232756173715586,
      "grad_norm": 0.11145982146263123,
      "learning_rate": 3.979193868861766e-05,
      "loss": 0.0008456252515316009,
      "step": 212200
    },
    {
      "epoch": 60.23559466363894,
      "grad_norm": 0.07316011935472488,
      "learning_rate": 3.9789100198694293e-05,
      "loss": 0.0005921712145209312,
      "step": 212210
    },
    {
      "epoch": 60.238433153562305,
      "grad_norm": 0.09485489875078201,
      "learning_rate": 3.9786261708770935e-05,
      "loss": 0.002485986053943634,
      "step": 212220
    },
    {
      "epoch": 60.24127164348567,
      "grad_norm": 0.045263953506946564,
      "learning_rate": 3.9783423218847576e-05,
      "loss": 0.0005272664129734039,
      "step": 212230
    },
    {
      "epoch": 60.24411013340902,
      "grad_norm": 0.10300051420927048,
      "learning_rate": 3.978058472892421e-05,
      "loss": 0.0001920163631439209,
      "step": 212240
    },
    {
      "epoch": 60.246948623332386,
      "grad_norm": 0.13175544142723083,
      "learning_rate": 3.977774623900085e-05,
      "loss": 0.0002532050013542175,
      "step": 212250
    },
    {
      "epoch": 60.24978711325575,
      "grad_norm": 0.1971817910671234,
      "learning_rate": 3.9774907749077494e-05,
      "loss": 0.0012876838445663452,
      "step": 212260
    },
    {
      "epoch": 60.25262560317911,
      "grad_norm": 0.21744953095912933,
      "learning_rate": 3.9772069259154135e-05,
      "loss": 0.0004742663353681564,
      "step": 212270
    },
    {
      "epoch": 60.25546409310247,
      "grad_norm": 0.05233932286500931,
      "learning_rate": 3.976923076923077e-05,
      "loss": 0.013525614142417907,
      "step": 212280
    },
    {
      "epoch": 60.25830258302583,
      "grad_norm": 0.03086390532553196,
      "learning_rate": 3.976639227930741e-05,
      "loss": 0.0009335322305560112,
      "step": 212290
    },
    {
      "epoch": 60.261141072949194,
      "grad_norm": 0.015302402898669243,
      "learning_rate": 3.976355378938405e-05,
      "loss": 0.0003717370331287384,
      "step": 212300
    },
    {
      "epoch": 60.26397956287255,
      "grad_norm": 0.05667561665177345,
      "learning_rate": 3.976071529946069e-05,
      "loss": 0.005561895295977592,
      "step": 212310
    },
    {
      "epoch": 60.26681805279591,
      "grad_norm": 0.6445702314376831,
      "learning_rate": 3.975787680953733e-05,
      "loss": 0.0024014754220843316,
      "step": 212320
    },
    {
      "epoch": 60.269656542719275,
      "grad_norm": 0.5079824328422546,
      "learning_rate": 3.975503831961397e-05,
      "loss": 0.0016381965950131416,
      "step": 212330
    },
    {
      "epoch": 60.27249503264263,
      "grad_norm": 0.07116887718439102,
      "learning_rate": 3.9752199829690604e-05,
      "loss": 0.00028000157326459887,
      "step": 212340
    },
    {
      "epoch": 60.275333522565994,
      "grad_norm": 0.2740797996520996,
      "learning_rate": 3.9749361339767246e-05,
      "loss": 0.0006645273417234421,
      "step": 212350
    },
    {
      "epoch": 60.27817201248936,
      "grad_norm": 0.8726823925971985,
      "learning_rate": 3.974652284984389e-05,
      "loss": 0.0008790990337729454,
      "step": 212360
    },
    {
      "epoch": 60.28101050241272,
      "grad_norm": 0.024547360837459564,
      "learning_rate": 3.974368435992052e-05,
      "loss": 0.0012587549164891243,
      "step": 212370
    },
    {
      "epoch": 60.283848992336075,
      "grad_norm": 0.023781413212418556,
      "learning_rate": 3.974084586999716e-05,
      "loss": 0.0010516205802559853,
      "step": 212380
    },
    {
      "epoch": 60.28668748225944,
      "grad_norm": 5.040438652038574,
      "learning_rate": 3.9738007380073804e-05,
      "loss": 0.0026662316173315047,
      "step": 212390
    },
    {
      "epoch": 60.2895259721828,
      "grad_norm": 0.9050663113594055,
      "learning_rate": 3.9735168890150446e-05,
      "loss": 0.00043822582811117174,
      "step": 212400
    },
    {
      "epoch": 60.29236446210616,
      "grad_norm": 0.2255355715751648,
      "learning_rate": 3.973233040022708e-05,
      "loss": 0.0004996765404939651,
      "step": 212410
    },
    {
      "epoch": 60.29520295202952,
      "grad_norm": 0.669003963470459,
      "learning_rate": 3.9729491910303715e-05,
      "loss": 0.0008336296305060387,
      "step": 212420
    },
    {
      "epoch": 60.29804144195288,
      "grad_norm": 0.011262867599725723,
      "learning_rate": 3.972665342038036e-05,
      "loss": 0.0005226518958806991,
      "step": 212430
    },
    {
      "epoch": 60.30087993187624,
      "grad_norm": 1.670279622077942,
      "learning_rate": 3.9723814930457e-05,
      "loss": 0.0005961356684565545,
      "step": 212440
    },
    {
      "epoch": 60.3037184217996,
      "grad_norm": 0.1535409688949585,
      "learning_rate": 3.972097644053364e-05,
      "loss": 0.0008101677522063255,
      "step": 212450
    },
    {
      "epoch": 60.306556911722964,
      "grad_norm": 0.8125361800193787,
      "learning_rate": 3.971813795061028e-05,
      "loss": 0.010066947340965271,
      "step": 212460
    },
    {
      "epoch": 60.30939540164633,
      "grad_norm": 6.673725128173828,
      "learning_rate": 3.9715299460686915e-05,
      "loss": 0.0014988793060183524,
      "step": 212470
    },
    {
      "epoch": 60.31223389156968,
      "grad_norm": 0.1491834968328476,
      "learning_rate": 3.9712460970763556e-05,
      "loss": 0.0008095607161521911,
      "step": 212480
    },
    {
      "epoch": 60.315072381493046,
      "grad_norm": 0.002669596578925848,
      "learning_rate": 3.97096224808402e-05,
      "loss": 0.0015592427924275398,
      "step": 212490
    },
    {
      "epoch": 60.31791087141641,
      "grad_norm": 0.03614400699734688,
      "learning_rate": 3.970678399091683e-05,
      "loss": 0.0006926767528057098,
      "step": 212500
    },
    {
      "epoch": 60.31791087141641,
      "eval_accuracy": 0.9792713168436447,
      "eval_loss": 0.07853108644485474,
      "eval_runtime": 32.6081,
      "eval_samples_per_second": 482.303,
      "eval_steps_per_second": 7.544,
      "step": 212500
    },
    {
      "epoch": 60.320749361339765,
      "grad_norm": 0.013016685843467712,
      "learning_rate": 3.9703945500993474e-05,
      "loss": 0.00036813076585531237,
      "step": 212510
    },
    {
      "epoch": 60.32358785126313,
      "grad_norm": 0.07877673208713531,
      "learning_rate": 3.970110701107011e-05,
      "loss": 0.0037849336862564088,
      "step": 212520
    },
    {
      "epoch": 60.32642634118649,
      "grad_norm": 0.03298484534025192,
      "learning_rate": 3.9698268521146756e-05,
      "loss": 0.0012386031448841095,
      "step": 212530
    },
    {
      "epoch": 60.329264831109846,
      "grad_norm": 0.02003333531320095,
      "learning_rate": 3.969543003122339e-05,
      "loss": 0.0005031388252973556,
      "step": 212540
    },
    {
      "epoch": 60.33210332103321,
      "grad_norm": 0.060534264892339706,
      "learning_rate": 3.9692591541300026e-05,
      "loss": 0.0005312932655215264,
      "step": 212550
    },
    {
      "epoch": 60.33494181095657,
      "grad_norm": 1.5442663431167603,
      "learning_rate": 3.9689753051376674e-05,
      "loss": 0.0002979043871164322,
      "step": 212560
    },
    {
      "epoch": 60.337780300879935,
      "grad_norm": 0.818830132484436,
      "learning_rate": 3.968691456145331e-05,
      "loss": 0.0005384659394621849,
      "step": 212570
    },
    {
      "epoch": 60.34061879080329,
      "grad_norm": 0.07287649810314178,
      "learning_rate": 3.968407607152995e-05,
      "loss": 0.002614613249897957,
      "step": 212580
    },
    {
      "epoch": 60.34345728072665,
      "grad_norm": 0.25836342573165894,
      "learning_rate": 3.968123758160659e-05,
      "loss": 0.00305231511592865,
      "step": 212590
    },
    {
      "epoch": 60.346295770650016,
      "grad_norm": 1.3756757974624634,
      "learning_rate": 3.9678399091683226e-05,
      "loss": 0.00048483498394489286,
      "step": 212600
    },
    {
      "epoch": 60.34913426057337,
      "grad_norm": 9.199005126953125,
      "learning_rate": 3.967556060175987e-05,
      "loss": 0.007254482805728912,
      "step": 212610
    },
    {
      "epoch": 60.351972750496735,
      "grad_norm": 0.01796429045498371,
      "learning_rate": 3.96727221118365e-05,
      "loss": 0.008895892649888992,
      "step": 212620
    },
    {
      "epoch": 60.3548112404201,
      "grad_norm": 0.1044369786977768,
      "learning_rate": 3.966988362191314e-05,
      "loss": 0.0006351402029395103,
      "step": 212630
    },
    {
      "epoch": 60.35764973034346,
      "grad_norm": 0.49802166223526,
      "learning_rate": 3.9667045131989784e-05,
      "loss": 0.0006324104964733124,
      "step": 212640
    },
    {
      "epoch": 60.36048822026682,
      "grad_norm": 0.009968497790396214,
      "learning_rate": 3.966420664206642e-05,
      "loss": 0.00044434331357479095,
      "step": 212650
    },
    {
      "epoch": 60.36332671019018,
      "grad_norm": 0.07250282913446426,
      "learning_rate": 3.966136815214306e-05,
      "loss": 0.00029443632811307905,
      "step": 212660
    },
    {
      "epoch": 60.36616520011354,
      "grad_norm": 0.021006111055612564,
      "learning_rate": 3.96585296622197e-05,
      "loss": 0.0017744384706020356,
      "step": 212670
    },
    {
      "epoch": 60.3690036900369,
      "grad_norm": 0.11253514140844345,
      "learning_rate": 3.9655691172296336e-05,
      "loss": 0.004662338271737098,
      "step": 212680
    },
    {
      "epoch": 60.37184217996026,
      "grad_norm": 0.42191165685653687,
      "learning_rate": 3.9652852682372985e-05,
      "loss": 0.0006789162755012513,
      "step": 212690
    },
    {
      "epoch": 60.374680669883624,
      "grad_norm": 0.18644465506076813,
      "learning_rate": 3.965001419244962e-05,
      "loss": 0.00025683846324682237,
      "step": 212700
    },
    {
      "epoch": 60.37751915980698,
      "grad_norm": 0.03106045350432396,
      "learning_rate": 3.9647175702526254e-05,
      "loss": 0.0029737846925854683,
      "step": 212710
    },
    {
      "epoch": 60.38035764973034,
      "grad_norm": 0.18261921405792236,
      "learning_rate": 3.9644337212602895e-05,
      "loss": 0.00019237026572227478,
      "step": 212720
    },
    {
      "epoch": 60.383196139653705,
      "grad_norm": 0.03888605162501335,
      "learning_rate": 3.9641498722679536e-05,
      "loss": 0.000781780295073986,
      "step": 212730
    },
    {
      "epoch": 60.38603462957707,
      "grad_norm": 0.4467582702636719,
      "learning_rate": 3.963866023275618e-05,
      "loss": 0.0003185665234923363,
      "step": 212740
    },
    {
      "epoch": 60.388873119500424,
      "grad_norm": 1.0408800840377808,
      "learning_rate": 3.963582174283281e-05,
      "loss": 0.0004212265834212303,
      "step": 212750
    },
    {
      "epoch": 60.39171160942379,
      "grad_norm": 0.20046047866344452,
      "learning_rate": 3.9632983252909454e-05,
      "loss": 0.0002834515646100044,
      "step": 212760
    },
    {
      "epoch": 60.39455009934715,
      "grad_norm": 0.3705109655857086,
      "learning_rate": 3.9630144762986095e-05,
      "loss": 0.00023512355983257294,
      "step": 212770
    },
    {
      "epoch": 60.397388589270506,
      "grad_norm": 0.014628705568611622,
      "learning_rate": 3.962730627306273e-05,
      "loss": 0.00016522016376256942,
      "step": 212780
    },
    {
      "epoch": 60.40022707919387,
      "grad_norm": 0.019992034882307053,
      "learning_rate": 3.962446778313937e-05,
      "loss": 0.00013082921504974364,
      "step": 212790
    },
    {
      "epoch": 60.40306556911723,
      "grad_norm": 0.04282299801707268,
      "learning_rate": 3.962162929321601e-05,
      "loss": 0.0004411064088344574,
      "step": 212800
    },
    {
      "epoch": 60.40590405904059,
      "grad_norm": 0.03131137415766716,
      "learning_rate": 3.961879080329265e-05,
      "loss": 0.0006004448980093003,
      "step": 212810
    },
    {
      "epoch": 60.40874254896395,
      "grad_norm": 0.062434159219264984,
      "learning_rate": 3.961595231336929e-05,
      "loss": 0.00012311618775129318,
      "step": 212820
    },
    {
      "epoch": 60.41158103888731,
      "grad_norm": 0.15706795454025269,
      "learning_rate": 3.961311382344593e-05,
      "loss": 0.001236712746322155,
      "step": 212830
    },
    {
      "epoch": 60.414419528810676,
      "grad_norm": 0.1135672926902771,
      "learning_rate": 3.9610275333522565e-05,
      "loss": 0.0005148442462086678,
      "step": 212840
    },
    {
      "epoch": 60.41725801873403,
      "grad_norm": 0.00707873422652483,
      "learning_rate": 3.9607436843599206e-05,
      "loss": 0.00018717125058174134,
      "step": 212850
    },
    {
      "epoch": 60.420096508657394,
      "grad_norm": 0.0035596087109297514,
      "learning_rate": 3.960459835367585e-05,
      "loss": 0.0005172420293092728,
      "step": 212860
    },
    {
      "epoch": 60.42293499858076,
      "grad_norm": 0.025607600808143616,
      "learning_rate": 3.960175986375249e-05,
      "loss": 0.00017489343881607055,
      "step": 212870
    },
    {
      "epoch": 60.42577348850411,
      "grad_norm": 0.017175517976284027,
      "learning_rate": 3.959892137382912e-05,
      "loss": 0.00019865315407514573,
      "step": 212880
    },
    {
      "epoch": 60.428611978427476,
      "grad_norm": 1.673065423965454,
      "learning_rate": 3.9596082883905765e-05,
      "loss": 0.0025501398369669913,
      "step": 212890
    },
    {
      "epoch": 60.43145046835084,
      "grad_norm": 0.5908607244491577,
      "learning_rate": 3.9593244393982406e-05,
      "loss": 0.0002895643934607506,
      "step": 212900
    },
    {
      "epoch": 60.434288958274195,
      "grad_norm": 1.4015148878097534,
      "learning_rate": 3.959040590405904e-05,
      "loss": 0.00031952802091836927,
      "step": 212910
    },
    {
      "epoch": 60.43712744819756,
      "grad_norm": 0.00445850845426321,
      "learning_rate": 3.958756741413568e-05,
      "loss": 0.0002652568742632866,
      "step": 212920
    },
    {
      "epoch": 60.43996593812092,
      "grad_norm": 0.01709609664976597,
      "learning_rate": 3.958472892421232e-05,
      "loss": 0.0010497985407710076,
      "step": 212930
    },
    {
      "epoch": 60.44280442804428,
      "grad_norm": 0.014654036611318588,
      "learning_rate": 3.958189043428896e-05,
      "loss": 0.00033137742429971694,
      "step": 212940
    },
    {
      "epoch": 60.44564291796764,
      "grad_norm": 0.05428967997431755,
      "learning_rate": 3.95790519443656e-05,
      "loss": 0.0006460757926106453,
      "step": 212950
    },
    {
      "epoch": 60.448481407891,
      "grad_norm": 0.03374316915869713,
      "learning_rate": 3.957621345444224e-05,
      "loss": 0.00018990486860275267,
      "step": 212960
    },
    {
      "epoch": 60.451319897814365,
      "grad_norm": 0.4393336772918701,
      "learning_rate": 3.9573374964518875e-05,
      "loss": 0.00019049141556024552,
      "step": 212970
    },
    {
      "epoch": 60.45415838773772,
      "grad_norm": 0.028057726100087166,
      "learning_rate": 3.957053647459552e-05,
      "loss": 0.00011762399226427079,
      "step": 212980
    },
    {
      "epoch": 60.45699687766108,
      "grad_norm": 0.04605928808450699,
      "learning_rate": 3.956769798467216e-05,
      "loss": 0.0002554373815655708,
      "step": 212990
    },
    {
      "epoch": 60.459835367584446,
      "grad_norm": 0.008764546364545822,
      "learning_rate": 3.95648594947488e-05,
      "loss": 0.0001500586047768593,
      "step": 213000
    },
    {
      "epoch": 60.459835367584446,
      "eval_accuracy": 0.9802886755261652,
      "eval_loss": 0.07303733378648758,
      "eval_runtime": 32.0496,
      "eval_samples_per_second": 490.708,
      "eval_steps_per_second": 7.676,
      "step": 213000
    },
    {
      "epoch": 60.46267385750781,
      "grad_norm": 0.05356809124350548,
      "learning_rate": 3.9562021004825434e-05,
      "loss": 0.00010720957070589065,
      "step": 213010
    },
    {
      "epoch": 60.465512347431165,
      "grad_norm": 0.008991366252303123,
      "learning_rate": 3.9559182514902075e-05,
      "loss": 9.149257093667983e-05,
      "step": 213020
    },
    {
      "epoch": 60.46835083735453,
      "grad_norm": 0.022711465135216713,
      "learning_rate": 3.955634402497872e-05,
      "loss": 7.473956793546677e-05,
      "step": 213030
    },
    {
      "epoch": 60.47118932727789,
      "grad_norm": 0.011164833791553974,
      "learning_rate": 3.955350553505535e-05,
      "loss": 7.481146603822708e-05,
      "step": 213040
    },
    {
      "epoch": 60.47402781720125,
      "grad_norm": 0.01893482357263565,
      "learning_rate": 3.955066704513199e-05,
      "loss": 0.00011330880224704742,
      "step": 213050
    },
    {
      "epoch": 60.47686630712461,
      "grad_norm": 0.08045006543397903,
      "learning_rate": 3.9547828555208634e-05,
      "loss": 0.000163152813911438,
      "step": 213060
    },
    {
      "epoch": 60.47970479704797,
      "grad_norm": 0.08550207316875458,
      "learning_rate": 3.954499006528527e-05,
      "loss": 0.0001562533900141716,
      "step": 213070
    },
    {
      "epoch": 60.48254328697133,
      "grad_norm": 0.021030280739068985,
      "learning_rate": 3.954215157536191e-05,
      "loss": 0.00041407085955142976,
      "step": 213080
    },
    {
      "epoch": 60.48538177689469,
      "grad_norm": 0.0052907331846654415,
      "learning_rate": 3.953931308543855e-05,
      "loss": 7.603410631418228e-05,
      "step": 213090
    },
    {
      "epoch": 60.488220266818054,
      "grad_norm": 0.06101894751191139,
      "learning_rate": 3.9536474595515186e-05,
      "loss": 0.000521058402955532,
      "step": 213100
    },
    {
      "epoch": 60.49105875674142,
      "grad_norm": 0.1702054888010025,
      "learning_rate": 3.953363610559183e-05,
      "loss": 0.00023218858987092972,
      "step": 213110
    },
    {
      "epoch": 60.49389724666477,
      "grad_norm": 0.019629674032330513,
      "learning_rate": 3.953079761566847e-05,
      "loss": 0.0008514957502484321,
      "step": 213120
    },
    {
      "epoch": 60.496735736588136,
      "grad_norm": 1.5716602802276611,
      "learning_rate": 3.9527959125745103e-05,
      "loss": 0.0005959402769804001,
      "step": 213130
    },
    {
      "epoch": 60.4995742265115,
      "grad_norm": 0.01242644153535366,
      "learning_rate": 3.9525120635821745e-05,
      "loss": 0.0005464812740683556,
      "step": 213140
    },
    {
      "epoch": 60.502412716434854,
      "grad_norm": 0.2799952030181885,
      "learning_rate": 3.952228214589838e-05,
      "loss": 0.00022779665887355803,
      "step": 213150
    },
    {
      "epoch": 60.50525120635822,
      "grad_norm": 0.359395295381546,
      "learning_rate": 3.951944365597503e-05,
      "loss": 0.0024619530886411667,
      "step": 213160
    },
    {
      "epoch": 60.50808969628158,
      "grad_norm": 0.03196197748184204,
      "learning_rate": 3.951660516605166e-05,
      "loss": 0.0030092371627688406,
      "step": 213170
    },
    {
      "epoch": 60.510928186204936,
      "grad_norm": 0.0396360382437706,
      "learning_rate": 3.95137666761283e-05,
      "loss": 0.0002860262989997864,
      "step": 213180
    },
    {
      "epoch": 60.5137666761283,
      "grad_norm": 0.04495998099446297,
      "learning_rate": 3.9510928186204945e-05,
      "loss": 0.00377996489405632,
      "step": 213190
    },
    {
      "epoch": 60.51660516605166,
      "grad_norm": 0.46605342626571655,
      "learning_rate": 3.950808969628158e-05,
      "loss": 0.00023503229022026062,
      "step": 213200
    },
    {
      "epoch": 60.519443655975024,
      "grad_norm": 4.214391231536865,
      "learning_rate": 3.950525120635822e-05,
      "loss": 0.0010533083230257034,
      "step": 213210
    },
    {
      "epoch": 60.52228214589838,
      "grad_norm": 0.016126129776239395,
      "learning_rate": 3.950241271643486e-05,
      "loss": 0.0003232551738619804,
      "step": 213220
    },
    {
      "epoch": 60.52512063582174,
      "grad_norm": 0.10853899270296097,
      "learning_rate": 3.94995742265115e-05,
      "loss": 0.0015983885154128076,
      "step": 213230
    },
    {
      "epoch": 60.527959125745106,
      "grad_norm": 0.014764778316020966,
      "learning_rate": 3.949673573658814e-05,
      "loss": 0.00025718845427036285,
      "step": 213240
    },
    {
      "epoch": 60.53079761566846,
      "grad_norm": 0.3328261971473694,
      "learning_rate": 3.949389724666477e-05,
      "loss": 0.005609837919473648,
      "step": 213250
    },
    {
      "epoch": 60.533636105591825,
      "grad_norm": 0.661686360836029,
      "learning_rate": 3.9491058756741414e-05,
      "loss": 0.00042172782123088836,
      "step": 213260
    },
    {
      "epoch": 60.53647459551519,
      "grad_norm": 0.02894427813589573,
      "learning_rate": 3.9488220266818056e-05,
      "loss": 0.00015194080770015718,
      "step": 213270
    },
    {
      "epoch": 60.53931308543854,
      "grad_norm": 0.005786092486232519,
      "learning_rate": 3.948538177689469e-05,
      "loss": 0.00017364732921123504,
      "step": 213280
    },
    {
      "epoch": 60.542151575361906,
      "grad_norm": 0.5930266976356506,
      "learning_rate": 3.948254328697134e-05,
      "loss": 0.0007741134613752365,
      "step": 213290
    },
    {
      "epoch": 60.54499006528527,
      "grad_norm": 0.47336289286613464,
      "learning_rate": 3.947970479704797e-05,
      "loss": 0.00017344187945127486,
      "step": 213300
    },
    {
      "epoch": 60.54782855520863,
      "grad_norm": 0.008377163670957088,
      "learning_rate": 3.947686630712461e-05,
      "loss": 0.000228220596909523,
      "step": 213310
    },
    {
      "epoch": 60.55066704513199,
      "grad_norm": 0.08914519101381302,
      "learning_rate": 3.9474027817201256e-05,
      "loss": 0.0001415695995092392,
      "step": 213320
    },
    {
      "epoch": 60.55350553505535,
      "grad_norm": 0.22376513481140137,
      "learning_rate": 3.947118932727789e-05,
      "loss": 0.00012744441628456115,
      "step": 213330
    },
    {
      "epoch": 60.55634402497871,
      "grad_norm": 1.1880178451538086,
      "learning_rate": 3.946835083735453e-05,
      "loss": 0.0003081619739532471,
      "step": 213340
    },
    {
      "epoch": 60.55918251490207,
      "grad_norm": 0.014955968596041203,
      "learning_rate": 3.9465512347431166e-05,
      "loss": 0.00019318703562021255,
      "step": 213350
    },
    {
      "epoch": 60.56202100482543,
      "grad_norm": 0.4637303948402405,
      "learning_rate": 3.946267385750781e-05,
      "loss": 0.00038141980767250063,
      "step": 213360
    },
    {
      "epoch": 60.564859494748795,
      "grad_norm": 0.12317363172769547,
      "learning_rate": 3.945983536758445e-05,
      "loss": 0.0006155364215373992,
      "step": 213370
    },
    {
      "epoch": 60.56769798467215,
      "grad_norm": 0.06244570016860962,
      "learning_rate": 3.9456996877661084e-05,
      "loss": 0.004873473569750786,
      "step": 213380
    },
    {
      "epoch": 60.570536474595514,
      "grad_norm": 0.04374377429485321,
      "learning_rate": 3.9454158387737725e-05,
      "loss": 0.0011515742167830466,
      "step": 213390
    },
    {
      "epoch": 60.57337496451888,
      "grad_norm": 0.022432034835219383,
      "learning_rate": 3.9451319897814366e-05,
      "loss": 0.00026555471122264863,
      "step": 213400
    },
    {
      "epoch": 60.57621345444224,
      "grad_norm": 0.036803606897592545,
      "learning_rate": 3.9448481407891e-05,
      "loss": 0.0009046507999300957,
      "step": 213410
    },
    {
      "epoch": 60.579051944365595,
      "grad_norm": 0.1278969645500183,
      "learning_rate": 3.944564291796764e-05,
      "loss": 0.00033058617264032365,
      "step": 213420
    },
    {
      "epoch": 60.58189043428896,
      "grad_norm": 0.5922592282295227,
      "learning_rate": 3.9442804428044284e-05,
      "loss": 0.0006589632481336594,
      "step": 213430
    },
    {
      "epoch": 60.58472892421232,
      "grad_norm": 0.13460838794708252,
      "learning_rate": 3.943996593812092e-05,
      "loss": 0.0002545630559325218,
      "step": 213440
    },
    {
      "epoch": 60.58756741413568,
      "grad_norm": 1.6269841194152832,
      "learning_rate": 3.943712744819756e-05,
      "loss": 0.0009586678817868233,
      "step": 213450
    },
    {
      "epoch": 60.59040590405904,
      "grad_norm": 0.13093166053295135,
      "learning_rate": 3.94342889582742e-05,
      "loss": 0.0023082077503204347,
      "step": 213460
    },
    {
      "epoch": 60.5932443939824,
      "grad_norm": 0.06929690390825272,
      "learning_rate": 3.943145046835084e-05,
      "loss": 0.00014449022710323335,
      "step": 213470
    },
    {
      "epoch": 60.596082883905765,
      "grad_norm": 0.023822026327252388,
      "learning_rate": 3.942861197842748e-05,
      "loss": 0.0009205533191561699,
      "step": 213480
    },
    {
      "epoch": 60.59892137382912,
      "grad_norm": 0.3524034023284912,
      "learning_rate": 3.942577348850412e-05,
      "loss": 0.0012259647250175477,
      "step": 213490
    },
    {
      "epoch": 60.601759863752484,
      "grad_norm": 0.394624263048172,
      "learning_rate": 3.942293499858076e-05,
      "loss": 0.001963217742741108,
      "step": 213500
    },
    {
      "epoch": 60.601759863752484,
      "eval_accuracy": 0.9763464106313983,
      "eval_loss": 0.09106861799955368,
      "eval_runtime": 32.5414,
      "eval_samples_per_second": 483.293,
      "eval_steps_per_second": 7.56,
      "step": 213500
    },
    {
      "epoch": 60.60459835367585,
      "grad_norm": 0.107668936252594,
      "learning_rate": 3.9420096508657394e-05,
      "loss": 0.0006617605686187744,
      "step": 213510
    },
    {
      "epoch": 60.6074368435992,
      "grad_norm": 0.2477286458015442,
      "learning_rate": 3.9417258018734036e-05,
      "loss": 0.0013930074870586394,
      "step": 213520
    },
    {
      "epoch": 60.610275333522566,
      "grad_norm": 0.03896278515458107,
      "learning_rate": 3.941441952881068e-05,
      "loss": 0.002101821079850197,
      "step": 213530
    },
    {
      "epoch": 60.61311382344593,
      "grad_norm": 0.07699217647314072,
      "learning_rate": 3.941158103888731e-05,
      "loss": 0.0008196931332349778,
      "step": 213540
    },
    {
      "epoch": 60.615952313369284,
      "grad_norm": 0.14922674000263214,
      "learning_rate": 3.940874254896395e-05,
      "loss": 0.0009234042838215828,
      "step": 213550
    },
    {
      "epoch": 60.61879080329265,
      "grad_norm": 3.742690086364746,
      "learning_rate": 3.9405904059040594e-05,
      "loss": 0.0004824059084057808,
      "step": 213560
    },
    {
      "epoch": 60.62162929321601,
      "grad_norm": 0.022591421380639076,
      "learning_rate": 3.940306556911723e-05,
      "loss": 0.00030651278793811796,
      "step": 213570
    },
    {
      "epoch": 60.62446778313937,
      "grad_norm": 0.03171566128730774,
      "learning_rate": 3.940022707919387e-05,
      "loss": 0.0005651500076055526,
      "step": 213580
    },
    {
      "epoch": 60.62730627306273,
      "grad_norm": 0.2640872895717621,
      "learning_rate": 3.939738858927051e-05,
      "loss": 0.001060624234378338,
      "step": 213590
    },
    {
      "epoch": 60.63014476298609,
      "grad_norm": 0.21782411634922028,
      "learning_rate": 3.9394550099347146e-05,
      "loss": 0.0014341872185468674,
      "step": 213600
    },
    {
      "epoch": 60.632983252909455,
      "grad_norm": 0.1470104604959488,
      "learning_rate": 3.939171160942379e-05,
      "loss": 0.0005581220611929893,
      "step": 213610
    },
    {
      "epoch": 60.63582174283281,
      "grad_norm": 0.2718030512332916,
      "learning_rate": 3.938887311950043e-05,
      "loss": 0.0005162440240383149,
      "step": 213620
    },
    {
      "epoch": 60.63866023275617,
      "grad_norm": 0.03086242452263832,
      "learning_rate": 3.938603462957707e-05,
      "loss": 0.0006310911849141121,
      "step": 213630
    },
    {
      "epoch": 60.641498722679536,
      "grad_norm": 2.541426658630371,
      "learning_rate": 3.9383196139653705e-05,
      "loss": 0.0004966450855135917,
      "step": 213640
    },
    {
      "epoch": 60.64433721260289,
      "grad_norm": 1.334401249885559,
      "learning_rate": 3.938035764973034e-05,
      "loss": 0.0010387519374489783,
      "step": 213650
    },
    {
      "epoch": 60.647175702526255,
      "grad_norm": 0.20200476050376892,
      "learning_rate": 3.937751915980699e-05,
      "loss": 0.000436924584209919,
      "step": 213660
    },
    {
      "epoch": 60.65001419244962,
      "grad_norm": 0.03386653959751129,
      "learning_rate": 3.937468066988362e-05,
      "loss": 0.00026781000196933745,
      "step": 213670
    },
    {
      "epoch": 60.65285268237298,
      "grad_norm": 0.1357858031988144,
      "learning_rate": 3.9371842179960264e-05,
      "loss": 0.00089748315513134,
      "step": 213680
    },
    {
      "epoch": 60.655691172296336,
      "grad_norm": 0.17049528658390045,
      "learning_rate": 3.9369003690036905e-05,
      "loss": 0.0004184199497103691,
      "step": 213690
    },
    {
      "epoch": 60.6585296622197,
      "grad_norm": 0.020961789414286613,
      "learning_rate": 3.936616520011354e-05,
      "loss": 0.00025675203651189806,
      "step": 213700
    },
    {
      "epoch": 60.66136815214306,
      "grad_norm": 0.1327710896730423,
      "learning_rate": 3.936332671019018e-05,
      "loss": 0.0014912977814674378,
      "step": 213710
    },
    {
      "epoch": 60.66420664206642,
      "grad_norm": 0.02027597650885582,
      "learning_rate": 3.936048822026682e-05,
      "loss": 0.0003780985251069069,
      "step": 213720
    },
    {
      "epoch": 60.66704513198978,
      "grad_norm": 0.11193887889385223,
      "learning_rate": 3.935764973034346e-05,
      "loss": 0.0010601077228784561,
      "step": 213730
    },
    {
      "epoch": 60.669883621913144,
      "grad_norm": 0.17878660559654236,
      "learning_rate": 3.93548112404201e-05,
      "loss": 0.005622755736112595,
      "step": 213740
    },
    {
      "epoch": 60.67272211183651,
      "grad_norm": 0.18254338204860687,
      "learning_rate": 3.935197275049673e-05,
      "loss": 0.0005600804463028908,
      "step": 213750
    },
    {
      "epoch": 60.67556060175986,
      "grad_norm": 0.09618737548589706,
      "learning_rate": 3.934913426057338e-05,
      "loss": 0.0004000140354037285,
      "step": 213760
    },
    {
      "epoch": 60.678399091683225,
      "grad_norm": 0.01292925514280796,
      "learning_rate": 3.9346295770650016e-05,
      "loss": 0.0003779040649533272,
      "step": 213770
    },
    {
      "epoch": 60.68123758160659,
      "grad_norm": 0.44061172008514404,
      "learning_rate": 3.934345728072665e-05,
      "loss": 0.00024771355092525484,
      "step": 213780
    },
    {
      "epoch": 60.684076071529944,
      "grad_norm": 0.43400344252586365,
      "learning_rate": 3.93406187908033e-05,
      "loss": 0.0004198167473077774,
      "step": 213790
    },
    {
      "epoch": 60.68691456145331,
      "grad_norm": 0.7531759142875671,
      "learning_rate": 3.933778030087993e-05,
      "loss": 0.00033037513494491575,
      "step": 213800
    },
    {
      "epoch": 60.68975305137667,
      "grad_norm": 0.843131959438324,
      "learning_rate": 3.9334941810956575e-05,
      "loss": 0.0018951039761304854,
      "step": 213810
    },
    {
      "epoch": 60.692591541300025,
      "grad_norm": 0.7763177156448364,
      "learning_rate": 3.9332103321033216e-05,
      "loss": 0.015769125521183015,
      "step": 213820
    },
    {
      "epoch": 60.69543003122339,
      "grad_norm": 0.03397354483604431,
      "learning_rate": 3.932926483110985e-05,
      "loss": 0.00046726595610380173,
      "step": 213830
    },
    {
      "epoch": 60.69826852114675,
      "grad_norm": 0.02210998348891735,
      "learning_rate": 3.932642634118649e-05,
      "loss": 0.0005125179886817932,
      "step": 213840
    },
    {
      "epoch": 60.701107011070114,
      "grad_norm": 0.23833899199962616,
      "learning_rate": 3.9323587851263127e-05,
      "loss": 0.0004067091271281242,
      "step": 213850
    },
    {
      "epoch": 60.70394550099347,
      "grad_norm": 0.326462984085083,
      "learning_rate": 3.932074936133977e-05,
      "loss": 0.0007082685828208923,
      "step": 213860
    },
    {
      "epoch": 60.70678399091683,
      "grad_norm": 0.35567107796669006,
      "learning_rate": 3.931791087141641e-05,
      "loss": 0.0003374157473444939,
      "step": 213870
    },
    {
      "epoch": 60.709622480840196,
      "grad_norm": 0.09233374148607254,
      "learning_rate": 3.9315072381493044e-05,
      "loss": 0.0006197206676006317,
      "step": 213880
    },
    {
      "epoch": 60.71246097076355,
      "grad_norm": 0.0896356925368309,
      "learning_rate": 3.9312233891569685e-05,
      "loss": 0.00026288367807865144,
      "step": 213890
    },
    {
      "epoch": 60.715299460686914,
      "grad_norm": 0.1026448905467987,
      "learning_rate": 3.930939540164633e-05,
      "loss": 0.0009186679497361183,
      "step": 213900
    },
    {
      "epoch": 60.71813795061028,
      "grad_norm": 0.006846647709608078,
      "learning_rate": 3.930655691172296e-05,
      "loss": 0.00015797819942235946,
      "step": 213910
    },
    {
      "epoch": 60.72097644053363,
      "grad_norm": 0.0875532478094101,
      "learning_rate": 3.930371842179961e-05,
      "loss": 0.00025676488876342775,
      "step": 213920
    },
    {
      "epoch": 60.723814930456996,
      "grad_norm": 0.06060396507382393,
      "learning_rate": 3.9300879931876244e-05,
      "loss": 0.0005107516422867775,
      "step": 213930
    },
    {
      "epoch": 60.72665342038036,
      "grad_norm": 0.07572657614946365,
      "learning_rate": 3.9298041441952885e-05,
      "loss": 0.00024239439517259598,
      "step": 213940
    },
    {
      "epoch": 60.72949191030372,
      "grad_norm": 0.037467021495103836,
      "learning_rate": 3.929520295202952e-05,
      "loss": 0.0001464400440454483,
      "step": 213950
    },
    {
      "epoch": 60.73233040022708,
      "grad_norm": 0.02191409096121788,
      "learning_rate": 3.929236446210616e-05,
      "loss": 0.0009377025067806244,
      "step": 213960
    },
    {
      "epoch": 60.73516889015044,
      "grad_norm": 1.418107271194458,
      "learning_rate": 3.92895259721828e-05,
      "loss": 0.0009064683690667152,
      "step": 213970
    },
    {
      "epoch": 60.7380073800738,
      "grad_norm": 3.0805814266204834,
      "learning_rate": 3.928668748225944e-05,
      "loss": 0.0007204145193099975,
      "step": 213980
    },
    {
      "epoch": 60.74084586999716,
      "grad_norm": 0.22702492773532867,
      "learning_rate": 3.928384899233608e-05,
      "loss": 0.0013763925060629844,
      "step": 213990
    },
    {
      "epoch": 60.74368435992052,
      "grad_norm": 0.08493654429912567,
      "learning_rate": 3.928101050241272e-05,
      "loss": 0.00012333616614341737,
      "step": 214000
    },
    {
      "epoch": 60.74368435992052,
      "eval_accuracy": 0.9752654670312202,
      "eval_loss": 0.09392672032117844,
      "eval_runtime": 33.0374,
      "eval_samples_per_second": 476.037,
      "eval_steps_per_second": 7.446,
      "step": 214000
    },
    {
      "epoch": 60.746522849843885,
      "grad_norm": 0.056936852633953094,
      "learning_rate": 3.9278172012489355e-05,
      "loss": 0.00044082049280405047,
      "step": 214010
    },
    {
      "epoch": 60.74936133976724,
      "grad_norm": 12.9418363571167,
      "learning_rate": 3.9275617371558335e-05,
      "loss": 0.010830791294574737,
      "step": 214020
    },
    {
      "epoch": 60.7521998296906,
      "grad_norm": 0.162174791097641,
      "learning_rate": 3.927277888163497e-05,
      "loss": 0.0013185026124119758,
      "step": 214030
    },
    {
      "epoch": 60.755038319613966,
      "grad_norm": 0.09921158850193024,
      "learning_rate": 3.926994039171161e-05,
      "loss": 0.001711965724825859,
      "step": 214040
    },
    {
      "epoch": 60.75787680953733,
      "grad_norm": 0.19495490193367004,
      "learning_rate": 3.926710190178825e-05,
      "loss": 0.00107131190598011,
      "step": 214050
    },
    {
      "epoch": 60.760715299460685,
      "grad_norm": 0.11336076259613037,
      "learning_rate": 3.9264263411864887e-05,
      "loss": 0.008304493129253387,
      "step": 214060
    },
    {
      "epoch": 60.76355378938405,
      "grad_norm": 0.5726718306541443,
      "learning_rate": 3.926142492194153e-05,
      "loss": 0.0007468827068805694,
      "step": 214070
    },
    {
      "epoch": 60.76639227930741,
      "grad_norm": 0.11420349776744843,
      "learning_rate": 3.925858643201817e-05,
      "loss": 0.0028796155005693437,
      "step": 214080
    },
    {
      "epoch": 60.76923076923077,
      "grad_norm": 3.2791385650634766,
      "learning_rate": 3.9255747942094804e-05,
      "loss": 0.0009003689512610436,
      "step": 214090
    },
    {
      "epoch": 60.77206925915413,
      "grad_norm": 0.07733231037855148,
      "learning_rate": 3.925290945217145e-05,
      "loss": 0.0014156935736536979,
      "step": 214100
    },
    {
      "epoch": 60.77490774907749,
      "grad_norm": 0.016510190442204475,
      "learning_rate": 3.925007096224809e-05,
      "loss": 0.003924942389130592,
      "step": 214110
    },
    {
      "epoch": 60.77774623900085,
      "grad_norm": 0.046731382608413696,
      "learning_rate": 3.924723247232472e-05,
      "loss": 0.00037130918353796006,
      "step": 214120
    },
    {
      "epoch": 60.78058472892421,
      "grad_norm": 0.020450467243790627,
      "learning_rate": 3.924439398240136e-05,
      "loss": 0.0005197262391448021,
      "step": 214130
    },
    {
      "epoch": 60.783423218847574,
      "grad_norm": 9.076140403747559,
      "learning_rate": 3.9241555492478004e-05,
      "loss": 0.003102618455886841,
      "step": 214140
    },
    {
      "epoch": 60.78626170877094,
      "grad_norm": 0.12189380079507828,
      "learning_rate": 3.9238717002554645e-05,
      "loss": 0.000965387374162674,
      "step": 214150
    },
    {
      "epoch": 60.78910019869429,
      "grad_norm": 0.08119045197963715,
      "learning_rate": 3.923587851263128e-05,
      "loss": 0.0003419533371925354,
      "step": 214160
    },
    {
      "epoch": 60.791938688617655,
      "grad_norm": 0.13633762300014496,
      "learning_rate": 3.923304002270792e-05,
      "loss": 0.0003166003152728081,
      "step": 214170
    },
    {
      "epoch": 60.79477717854102,
      "grad_norm": 0.07971606403589249,
      "learning_rate": 3.923020153278456e-05,
      "loss": 0.0014098899438977242,
      "step": 214180
    },
    {
      "epoch": 60.797615668464374,
      "grad_norm": 0.2866664528846741,
      "learning_rate": 3.92273630428612e-05,
      "loss": 0.0008065858855843544,
      "step": 214190
    },
    {
      "epoch": 60.80045415838774,
      "grad_norm": 0.07133712619543076,
      "learning_rate": 3.922452455293784e-05,
      "loss": 0.00034293755888938906,
      "step": 214200
    },
    {
      "epoch": 60.8032926483111,
      "grad_norm": 0.15783825516700745,
      "learning_rate": 3.922168606301448e-05,
      "loss": 0.00032199863344430926,
      "step": 214210
    },
    {
      "epoch": 60.80613113823446,
      "grad_norm": 0.0934293270111084,
      "learning_rate": 3.9218847573091115e-05,
      "loss": 0.0006432369351387024,
      "step": 214220
    },
    {
      "epoch": 60.80896962815782,
      "grad_norm": 0.0640489012002945,
      "learning_rate": 3.9216009083167756e-05,
      "loss": 0.001894128881394863,
      "step": 214230
    },
    {
      "epoch": 60.81180811808118,
      "grad_norm": 6.179540634155273,
      "learning_rate": 3.92131705932444e-05,
      "loss": 0.0021935222670435906,
      "step": 214240
    },
    {
      "epoch": 60.814646608004544,
      "grad_norm": 0.187752828001976,
      "learning_rate": 3.921033210332103e-05,
      "loss": 0.0005717670544981956,
      "step": 214250
    },
    {
      "epoch": 60.8174850979279,
      "grad_norm": 0.13186945021152496,
      "learning_rate": 3.920749361339767e-05,
      "loss": 0.0011180918663740158,
      "step": 214260
    },
    {
      "epoch": 60.82032358785126,
      "grad_norm": 0.07031745463609695,
      "learning_rate": 3.9204655123474315e-05,
      "loss": 0.00028594136238098146,
      "step": 214270
    },
    {
      "epoch": 60.823162077774626,
      "grad_norm": 0.12241843342781067,
      "learning_rate": 3.9201816633550956e-05,
      "loss": 0.002047180198132992,
      "step": 214280
    },
    {
      "epoch": 60.82600056769798,
      "grad_norm": 0.4714779257774353,
      "learning_rate": 3.919897814362759e-05,
      "loss": 0.0009456511586904525,
      "step": 214290
    },
    {
      "epoch": 60.828839057621344,
      "grad_norm": 0.09269167482852936,
      "learning_rate": 3.919613965370423e-05,
      "loss": 0.003608642518520355,
      "step": 214300
    },
    {
      "epoch": 60.83167754754471,
      "grad_norm": 0.9019885659217834,
      "learning_rate": 3.9193301163780873e-05,
      "loss": 0.001398182474076748,
      "step": 214310
    },
    {
      "epoch": 60.83451603746807,
      "grad_norm": 0.09004561603069305,
      "learning_rate": 3.919046267385751e-05,
      "loss": 0.0006853125989437104,
      "step": 214320
    },
    {
      "epoch": 60.837354527391426,
      "grad_norm": 0.06747071444988251,
      "learning_rate": 3.918762418393415e-05,
      "loss": 0.0008158458396792412,
      "step": 214330
    },
    {
      "epoch": 60.84019301731479,
      "grad_norm": 0.010834372602403164,
      "learning_rate": 3.918478569401079e-05,
      "loss": 0.00540047362446785,
      "step": 214340
    },
    {
      "epoch": 60.84303150723815,
      "grad_norm": 0.060086917132139206,
      "learning_rate": 3.9181947204087425e-05,
      "loss": 0.00035485439002513885,
      "step": 214350
    },
    {
      "epoch": 60.84586999716151,
      "grad_norm": 1.0271942615509033,
      "learning_rate": 3.917910871416407e-05,
      "loss": 0.0004329778254032135,
      "step": 214360
    },
    {
      "epoch": 60.84870848708487,
      "grad_norm": 2.325047016143799,
      "learning_rate": 3.917627022424071e-05,
      "loss": 0.0005090530961751938,
      "step": 214370
    },
    {
      "epoch": 60.85154697700823,
      "grad_norm": 0.015145102515816689,
      "learning_rate": 3.917343173431734e-05,
      "loss": 0.0003334898501634598,
      "step": 214380
    },
    {
      "epoch": 60.85438546693159,
      "grad_norm": 0.013235276564955711,
      "learning_rate": 3.9170593244393984e-05,
      "loss": 0.0027549231424927713,
      "step": 214390
    },
    {
      "epoch": 60.85722395685495,
      "grad_norm": 0.05014194920659065,
      "learning_rate": 3.9167754754470626e-05,
      "loss": 0.00021939016878604888,
      "step": 214400
    },
    {
      "epoch": 60.860062446778315,
      "grad_norm": 0.06601856648921967,
      "learning_rate": 3.916491626454726e-05,
      "loss": 0.002479120343923569,
      "step": 214410
    },
    {
      "epoch": 60.86290093670168,
      "grad_norm": 3.7311959266662598,
      "learning_rate": 3.91620777746239e-05,
      "loss": 0.0009954076260328293,
      "step": 214420
    },
    {
      "epoch": 60.865739426625034,
      "grad_norm": 0.03158235177397728,
      "learning_rate": 3.9159239284700536e-05,
      "loss": 0.0002541018649935722,
      "step": 214430
    },
    {
      "epoch": 60.868577916548396,
      "grad_norm": 0.014159257523715496,
      "learning_rate": 3.9156400794777184e-05,
      "loss": 0.0007186410948634148,
      "step": 214440
    },
    {
      "epoch": 60.87141640647176,
      "grad_norm": 0.029498783871531487,
      "learning_rate": 3.915356230485382e-05,
      "loss": 0.00030751824378967286,
      "step": 214450
    },
    {
      "epoch": 60.874254896395115,
      "grad_norm": 0.07901179790496826,
      "learning_rate": 3.9150723814930453e-05,
      "loss": 0.00029143039137125015,
      "step": 214460
    },
    {
      "epoch": 60.87709338631848,
      "grad_norm": 0.027470100671052933,
      "learning_rate": 3.91478853250071e-05,
      "loss": 0.0002888062968850136,
      "step": 214470
    },
    {
      "epoch": 60.87993187624184,
      "grad_norm": 0.2429257333278656,
      "learning_rate": 3.9145046835083736e-05,
      "loss": 0.00023186914622783661,
      "step": 214480
    },
    {
      "epoch": 60.8827703661652,
      "grad_norm": 0.01785132847726345,
      "learning_rate": 3.914220834516038e-05,
      "loss": 0.0005063196644186973,
      "step": 214490
    },
    {
      "epoch": 60.88560885608856,
      "grad_norm": 0.20747493207454681,
      "learning_rate": 3.913936985523702e-05,
      "loss": 0.0077833980321884155,
      "step": 214500
    },
    {
      "epoch": 60.88560885608856,
      "eval_accuracy": 0.9767279201373434,
      "eval_loss": 0.09059061110019684,
      "eval_runtime": 32.3287,
      "eval_samples_per_second": 486.471,
      "eval_steps_per_second": 7.609,
      "step": 214500
    },
    {
      "epoch": 60.88844734601192,
      "grad_norm": 0.0623568631708622,
      "learning_rate": 3.9136815214305985e-05,
      "loss": 0.003043646551668644,
      "step": 214510
    },
    {
      "epoch": 60.891285835935285,
      "grad_norm": 0.024578966200351715,
      "learning_rate": 3.9133976724382633e-05,
      "loss": 0.002449057437479496,
      "step": 214520
    },
    {
      "epoch": 60.89412432585864,
      "grad_norm": 0.07419086247682571,
      "learning_rate": 3.913113823445927e-05,
      "loss": 0.0013263147324323655,
      "step": 214530
    },
    {
      "epoch": 60.896962815782004,
      "grad_norm": 9.104339599609375,
      "learning_rate": 3.912829974453591e-05,
      "loss": 0.002042484097182751,
      "step": 214540
    },
    {
      "epoch": 60.89980130570537,
      "grad_norm": 0.39052385091781616,
      "learning_rate": 3.912546125461255e-05,
      "loss": 0.0003722306340932846,
      "step": 214550
    },
    {
      "epoch": 60.90263979562872,
      "grad_norm": 0.08576544374227524,
      "learning_rate": 3.9122622764689185e-05,
      "loss": 0.0004353789612650871,
      "step": 214560
    },
    {
      "epoch": 60.905478285552086,
      "grad_norm": 0.20577400922775269,
      "learning_rate": 3.911978427476583e-05,
      "loss": 0.00039111729711294174,
      "step": 214570
    },
    {
      "epoch": 60.90831677547545,
      "grad_norm": 0.07964835315942764,
      "learning_rate": 3.911694578484247e-05,
      "loss": 0.006367252767086029,
      "step": 214580
    },
    {
      "epoch": 60.91115526539881,
      "grad_norm": 0.025012154132127762,
      "learning_rate": 3.91141072949191e-05,
      "loss": 0.0006810370832681656,
      "step": 214590
    },
    {
      "epoch": 60.91399375532217,
      "grad_norm": 0.025986049324274063,
      "learning_rate": 3.9111268804995744e-05,
      "loss": 0.0009364180266857147,
      "step": 214600
    },
    {
      "epoch": 60.91683224524553,
      "grad_norm": 0.0074682217091321945,
      "learning_rate": 3.910843031507238e-05,
      "loss": 0.0003288388252258301,
      "step": 214610
    },
    {
      "epoch": 60.91967073516889,
      "grad_norm": 0.042985301464796066,
      "learning_rate": 3.910559182514903e-05,
      "loss": 0.00040201954543590547,
      "step": 214620
    },
    {
      "epoch": 60.92250922509225,
      "grad_norm": 1.8868718147277832,
      "learning_rate": 3.910275333522566e-05,
      "loss": 0.0006316378712654114,
      "step": 214630
    },
    {
      "epoch": 60.92534771501561,
      "grad_norm": 0.5271281003952026,
      "learning_rate": 3.9099914845302296e-05,
      "loss": 0.002761477418243885,
      "step": 214640
    },
    {
      "epoch": 60.928186204938974,
      "grad_norm": 1.2164965867996216,
      "learning_rate": 3.9097076355378944e-05,
      "loss": 0.0023493416607379912,
      "step": 214650
    },
    {
      "epoch": 60.93102469486233,
      "grad_norm": 0.014424446038901806,
      "learning_rate": 3.909423786545558e-05,
      "loss": 0.0005723649635910987,
      "step": 214660
    },
    {
      "epoch": 60.93386318478569,
      "grad_norm": 0.06557957082986832,
      "learning_rate": 3.909139937553222e-05,
      "loss": 0.0008646249771118164,
      "step": 214670
    },
    {
      "epoch": 60.936701674709056,
      "grad_norm": 1.0704470872879028,
      "learning_rate": 3.908856088560886e-05,
      "loss": 0.0006767518818378449,
      "step": 214680
    },
    {
      "epoch": 60.93954016463242,
      "grad_norm": 0.021511416882276535,
      "learning_rate": 3.9085722395685496e-05,
      "loss": 0.0003639969974756241,
      "step": 214690
    },
    {
      "epoch": 60.942378654555775,
      "grad_norm": 0.05119406431913376,
      "learning_rate": 3.908288390576214e-05,
      "loss": 0.00014451835304498672,
      "step": 214700
    },
    {
      "epoch": 60.94521714447914,
      "grad_norm": 0.010633128695189953,
      "learning_rate": 3.908004541583877e-05,
      "loss": 0.00034739989787340163,
      "step": 214710
    },
    {
      "epoch": 60.9480556344025,
      "grad_norm": 0.009206034243106842,
      "learning_rate": 3.9077206925915414e-05,
      "loss": 0.0005860179662704468,
      "step": 214720
    },
    {
      "epoch": 60.950894124325856,
      "grad_norm": 0.024251770228147507,
      "learning_rate": 3.9074368435992055e-05,
      "loss": 0.00026971139013767245,
      "step": 214730
    },
    {
      "epoch": 60.95373261424922,
      "grad_norm": 17.094083786010742,
      "learning_rate": 3.907152994606869e-05,
      "loss": 0.005224929749965667,
      "step": 214740
    },
    {
      "epoch": 60.95657110417258,
      "grad_norm": 0.02884203940629959,
      "learning_rate": 3.906869145614534e-05,
      "loss": 0.00043208450078964234,
      "step": 214750
    },
    {
      "epoch": 60.95940959409594,
      "grad_norm": 0.07443677634000778,
      "learning_rate": 3.906585296622197e-05,
      "loss": 0.0013178279623389244,
      "step": 214760
    },
    {
      "epoch": 60.9622480840193,
      "grad_norm": 0.26190799474716187,
      "learning_rate": 3.906301447629861e-05,
      "loss": 0.0005351247265934945,
      "step": 214770
    },
    {
      "epoch": 60.96508657394266,
      "grad_norm": 0.13104768097400665,
      "learning_rate": 3.9060175986375255e-05,
      "loss": 0.0004167964681982994,
      "step": 214780
    },
    {
      "epoch": 60.967925063866026,
      "grad_norm": 0.693864107131958,
      "learning_rate": 3.905733749645189e-05,
      "loss": 0.0002873716875910759,
      "step": 214790
    },
    {
      "epoch": 60.97076355378938,
      "grad_norm": 0.06054842472076416,
      "learning_rate": 3.905449900652853e-05,
      "loss": 0.0009095653891563416,
      "step": 214800
    },
    {
      "epoch": 60.973602043712745,
      "grad_norm": 0.020037194713950157,
      "learning_rate": 3.9051660516605166e-05,
      "loss": 0.00032813418656587603,
      "step": 214810
    },
    {
      "epoch": 60.97644053363611,
      "grad_norm": 0.13205863535404205,
      "learning_rate": 3.904882202668181e-05,
      "loss": 0.0006993519142270088,
      "step": 214820
    },
    {
      "epoch": 60.979279023559464,
      "grad_norm": 0.44881370663642883,
      "learning_rate": 3.904598353675845e-05,
      "loss": 0.0011865904554724694,
      "step": 214830
    },
    {
      "epoch": 60.98211751348283,
      "grad_norm": 0.5846525430679321,
      "learning_rate": 3.904314504683508e-05,
      "loss": 0.000391811691224575,
      "step": 214840
    },
    {
      "epoch": 60.98495600340619,
      "grad_norm": 0.01980476640164852,
      "learning_rate": 3.9040306556911724e-05,
      "loss": 0.00026873238384723664,
      "step": 214850
    },
    {
      "epoch": 60.987794493329545,
      "grad_norm": 0.03612753003835678,
      "learning_rate": 3.9037468066988366e-05,
      "loss": 0.0008358336985111236,
      "step": 214860
    },
    {
      "epoch": 60.99063298325291,
      "grad_norm": 0.5984060168266296,
      "learning_rate": 3.9034629577065e-05,
      "loss": 0.00033920686691999433,
      "step": 214870
    },
    {
      "epoch": 60.99347147317627,
      "grad_norm": 0.05385841429233551,
      "learning_rate": 3.903179108714164e-05,
      "loss": 0.0002008087933063507,
      "step": 214880
    },
    {
      "epoch": 60.996309963099634,
      "grad_norm": 0.09474235773086548,
      "learning_rate": 3.902895259721828e-05,
      "loss": 0.0002620801329612732,
      "step": 214890
    },
    {
      "epoch": 60.99914845302299,
      "grad_norm": 3.9117348194122314,
      "learning_rate": 3.902611410729492e-05,
      "loss": 0.000969577208161354,
      "step": 214900
    },
    {
      "epoch": 61.00198694294635,
      "grad_norm": 0.06659076362848282,
      "learning_rate": 3.902327561737156e-05,
      "loss": 0.0004684463143348694,
      "step": 214910
    },
    {
      "epoch": 61.004825432869715,
      "grad_norm": 0.03915082663297653,
      "learning_rate": 3.90204371274482e-05,
      "loss": 0.0001376800239086151,
      "step": 214920
    },
    {
      "epoch": 61.00766392279307,
      "grad_norm": 0.038332995027303696,
      "learning_rate": 3.9017598637524835e-05,
      "loss": 0.0003470022231340408,
      "step": 214930
    },
    {
      "epoch": 61.010502412716434,
      "grad_norm": 0.014502450823783875,
      "learning_rate": 3.9014760147601476e-05,
      "loss": 0.00037841424345970155,
      "step": 214940
    },
    {
      "epoch": 61.0133409026398,
      "grad_norm": 12.313241958618164,
      "learning_rate": 3.901192165767812e-05,
      "loss": 0.002149241603910923,
      "step": 214950
    },
    {
      "epoch": 61.01617939256316,
      "grad_norm": 0.01935967244207859,
      "learning_rate": 3.900908316775476e-05,
      "loss": 0.0009054835885763168,
      "step": 214960
    },
    {
      "epoch": 61.019017882486516,
      "grad_norm": 0.035346534103155136,
      "learning_rate": 3.9006244677831394e-05,
      "loss": 0.00027482658624649047,
      "step": 214970
    },
    {
      "epoch": 61.02185637240988,
      "grad_norm": 0.009410804137587547,
      "learning_rate": 3.9003406187908035e-05,
      "loss": 0.004024219885468483,
      "step": 214980
    },
    {
      "epoch": 61.02469486233324,
      "grad_norm": 0.9445405602455139,
      "learning_rate": 3.9000567697984676e-05,
      "loss": 0.00029649678617715836,
      "step": 214990
    },
    {
      "epoch": 61.0275333522566,
      "grad_norm": 1.7651691436767578,
      "learning_rate": 3.899772920806131e-05,
      "loss": 0.001547003537416458,
      "step": 215000
    },
    {
      "epoch": 61.0275333522566,
      "eval_accuracy": 0.9764099955490557,
      "eval_loss": 0.08333548903465271,
      "eval_runtime": 33.3129,
      "eval_samples_per_second": 472.1,
      "eval_steps_per_second": 7.385,
      "step": 215000
    },
    {
      "epoch": 61.03037184217996,
      "grad_norm": 0.21056205034255981,
      "learning_rate": 3.899489071813795e-05,
      "loss": 0.0014349974691867829,
      "step": 215010
    },
    {
      "epoch": 61.03321033210332,
      "grad_norm": 0.37836697697639465,
      "learning_rate": 3.8992052228214594e-05,
      "loss": 0.0018550721928477288,
      "step": 215020
    },
    {
      "epoch": 61.03604882202668,
      "grad_norm": 0.08101577311754227,
      "learning_rate": 3.898921373829123e-05,
      "loss": 0.0011468274518847466,
      "step": 215030
    },
    {
      "epoch": 61.03888731195004,
      "grad_norm": 0.02415461279451847,
      "learning_rate": 3.898637524836787e-05,
      "loss": 0.00023990217596292496,
      "step": 215040
    },
    {
      "epoch": 61.041725801873405,
      "grad_norm": 1.2065339088439941,
      "learning_rate": 3.898353675844451e-05,
      "loss": 0.0002829842269420624,
      "step": 215050
    },
    {
      "epoch": 61.04456429179677,
      "grad_norm": 0.0822652280330658,
      "learning_rate": 3.8980698268521146e-05,
      "loss": 0.001963541842997074,
      "step": 215060
    },
    {
      "epoch": 61.04740278172012,
      "grad_norm": 0.04683750495314598,
      "learning_rate": 3.897785977859779e-05,
      "loss": 0.000877559371292591,
      "step": 215070
    },
    {
      "epoch": 61.050241271643486,
      "grad_norm": 0.009820149280130863,
      "learning_rate": 3.897502128867443e-05,
      "loss": 0.00011243615299463272,
      "step": 215080
    },
    {
      "epoch": 61.05307976156685,
      "grad_norm": 0.009107191115617752,
      "learning_rate": 3.897218279875107e-05,
      "loss": 0.0004445452243089676,
      "step": 215090
    },
    {
      "epoch": 61.055918251490205,
      "grad_norm": 0.03630382940173149,
      "learning_rate": 3.8969344308827704e-05,
      "loss": 0.00037032756954431535,
      "step": 215100
    },
    {
      "epoch": 61.05875674141357,
      "grad_norm": 0.0142793795093894,
      "learning_rate": 3.8966505818904346e-05,
      "loss": 0.0006057983264327049,
      "step": 215110
    },
    {
      "epoch": 61.06159523133693,
      "grad_norm": 0.05844442918896675,
      "learning_rate": 3.896366732898099e-05,
      "loss": 0.001671992801129818,
      "step": 215120
    },
    {
      "epoch": 61.064433721260286,
      "grad_norm": 0.03213059902191162,
      "learning_rate": 3.896082883905762e-05,
      "loss": 0.00021991375833749772,
      "step": 215130
    },
    {
      "epoch": 61.06727221118365,
      "grad_norm": 0.015917714685201645,
      "learning_rate": 3.895799034913426e-05,
      "loss": 0.00019307900220155716,
      "step": 215140
    },
    {
      "epoch": 61.07011070110701,
      "grad_norm": 0.005449051968753338,
      "learning_rate": 3.8955151859210905e-05,
      "loss": 0.00017181970179080963,
      "step": 215150
    },
    {
      "epoch": 61.072949191030375,
      "grad_norm": 1.1079127788543701,
      "learning_rate": 3.895231336928754e-05,
      "loss": 0.00029346011579036715,
      "step": 215160
    },
    {
      "epoch": 61.07578768095373,
      "grad_norm": 0.8667793869972229,
      "learning_rate": 3.894947487936418e-05,
      "loss": 0.0069108143448829654,
      "step": 215170
    },
    {
      "epoch": 61.078626170877094,
      "grad_norm": 3.7374625205993652,
      "learning_rate": 3.894663638944082e-05,
      "loss": 0.00342666357755661,
      "step": 215180
    },
    {
      "epoch": 61.08146466080046,
      "grad_norm": 0.023040490224957466,
      "learning_rate": 3.8943797899517457e-05,
      "loss": 0.0025723539292812347,
      "step": 215190
    },
    {
      "epoch": 61.08430315072381,
      "grad_norm": 0.10249919444322586,
      "learning_rate": 3.89409594095941e-05,
      "loss": 0.0026480862870812416,
      "step": 215200
    },
    {
      "epoch": 61.087141640647175,
      "grad_norm": 0.04386807233095169,
      "learning_rate": 3.893812091967074e-05,
      "loss": 0.0007886188104748726,
      "step": 215210
    },
    {
      "epoch": 61.08998013057054,
      "grad_norm": 0.06683900207281113,
      "learning_rate": 3.893528242974738e-05,
      "loss": 0.0005765730515122413,
      "step": 215220
    },
    {
      "epoch": 61.092818620493894,
      "grad_norm": 0.13995268940925598,
      "learning_rate": 3.8932443939824015e-05,
      "loss": 0.002858196571469307,
      "step": 215230
    },
    {
      "epoch": 61.09565711041726,
      "grad_norm": 0.10146021842956543,
      "learning_rate": 3.892960544990065e-05,
      "loss": 0.0001858748495578766,
      "step": 215240
    },
    {
      "epoch": 61.09849560034062,
      "grad_norm": 0.13982917368412018,
      "learning_rate": 3.89267669599773e-05,
      "loss": 0.0004026645794510841,
      "step": 215250
    },
    {
      "epoch": 61.10133409026398,
      "grad_norm": 0.02427654340863228,
      "learning_rate": 3.892392847005393e-05,
      "loss": 0.00022544655948877334,
      "step": 215260
    },
    {
      "epoch": 61.10417258018734,
      "grad_norm": 0.14845460653305054,
      "learning_rate": 3.8921089980130574e-05,
      "loss": 0.0003768356516957283,
      "step": 215270
    },
    {
      "epoch": 61.1070110701107,
      "grad_norm": 0.021916588768363,
      "learning_rate": 3.8918251490207215e-05,
      "loss": 0.00034391209483146666,
      "step": 215280
    },
    {
      "epoch": 61.109849560034064,
      "grad_norm": 0.09462939202785492,
      "learning_rate": 3.891541300028385e-05,
      "loss": 0.0024301664903759955,
      "step": 215290
    },
    {
      "epoch": 61.11268804995742,
      "grad_norm": 0.00822365004569292,
      "learning_rate": 3.891257451036049e-05,
      "loss": 0.01724238395690918,
      "step": 215300
    },
    {
      "epoch": 61.11552653988078,
      "grad_norm": 0.013373183086514473,
      "learning_rate": 3.890973602043713e-05,
      "loss": 0.002230127342045307,
      "step": 215310
    },
    {
      "epoch": 61.118365029804146,
      "grad_norm": 2.709699869155884,
      "learning_rate": 3.890689753051377e-05,
      "loss": 0.0018983447924256324,
      "step": 215320
    },
    {
      "epoch": 61.1212035197275,
      "grad_norm": 0.09977324306964874,
      "learning_rate": 3.890405904059041e-05,
      "loss": 0.0002844927832484245,
      "step": 215330
    },
    {
      "epoch": 61.124042009650864,
      "grad_norm": 0.03055824525654316,
      "learning_rate": 3.890122055066704e-05,
      "loss": 0.0004997871816158295,
      "step": 215340
    },
    {
      "epoch": 61.12688049957423,
      "grad_norm": 11.654694557189941,
      "learning_rate": 3.8898382060743685e-05,
      "loss": 0.0028288539499044417,
      "step": 215350
    },
    {
      "epoch": 61.12971898949759,
      "grad_norm": 1.0620603561401367,
      "learning_rate": 3.8895543570820326e-05,
      "loss": 0.001242346316576004,
      "step": 215360
    },
    {
      "epoch": 61.132557479420946,
      "grad_norm": 0.04451851546764374,
      "learning_rate": 3.889270508089696e-05,
      "loss": 0.00031450409442186357,
      "step": 215370
    },
    {
      "epoch": 61.13539596934431,
      "grad_norm": 4.111286640167236,
      "learning_rate": 3.888986659097361e-05,
      "loss": 0.000819564238190651,
      "step": 215380
    },
    {
      "epoch": 61.13823445926767,
      "grad_norm": 0.7345172166824341,
      "learning_rate": 3.888702810105024e-05,
      "loss": 0.00025739409029483794,
      "step": 215390
    },
    {
      "epoch": 61.14107294919103,
      "grad_norm": 0.11299198865890503,
      "learning_rate": 3.888418961112688e-05,
      "loss": 0.0002423936501145363,
      "step": 215400
    },
    {
      "epoch": 61.14391143911439,
      "grad_norm": 0.025248145684599876,
      "learning_rate": 3.8881351121203526e-05,
      "loss": 0.0002275649458169937,
      "step": 215410
    },
    {
      "epoch": 61.14674992903775,
      "grad_norm": 0.11199043691158295,
      "learning_rate": 3.887851263128016e-05,
      "loss": 0.00026700273156166077,
      "step": 215420
    },
    {
      "epoch": 61.149588418961116,
      "grad_norm": 0.23489168286323547,
      "learning_rate": 3.88756741413568e-05,
      "loss": 0.0002180822193622589,
      "step": 215430
    },
    {
      "epoch": 61.15242690888447,
      "grad_norm": 0.030860157683491707,
      "learning_rate": 3.887283565143344e-05,
      "loss": 0.00043433494865894315,
      "step": 215440
    },
    {
      "epoch": 61.155265398807835,
      "grad_norm": 0.11463258415460587,
      "learning_rate": 3.886999716151008e-05,
      "loss": 0.0001647079363465309,
      "step": 215450
    },
    {
      "epoch": 61.1581038887312,
      "grad_norm": 0.03646363690495491,
      "learning_rate": 3.886715867158672e-05,
      "loss": 0.00045794956386089323,
      "step": 215460
    },
    {
      "epoch": 61.16094237865455,
      "grad_norm": 0.0644049420952797,
      "learning_rate": 3.8864320181663354e-05,
      "loss": 0.0005653230473399162,
      "step": 215470
    },
    {
      "epoch": 61.163780868577916,
      "grad_norm": 0.0036862005945295095,
      "learning_rate": 3.8861481691739995e-05,
      "loss": 0.0005882274359464646,
      "step": 215480
    },
    {
      "epoch": 61.16661935850128,
      "grad_norm": 0.2289619743824005,
      "learning_rate": 3.885864320181664e-05,
      "loss": 0.00014341063797473907,
      "step": 215490
    },
    {
      "epoch": 61.169457848424635,
      "grad_norm": 0.03673364594578743,
      "learning_rate": 3.885580471189327e-05,
      "loss": 0.00042535662651062013,
      "step": 215500
    },
    {
      "epoch": 61.169457848424635,
      "eval_accuracy": 0.9795256565142748,
      "eval_loss": 0.07414978742599487,
      "eval_runtime": 32.228,
      "eval_samples_per_second": 487.992,
      "eval_steps_per_second": 7.633,
      "step": 215500
    },
    {
      "epoch": 61.172296338348,
      "grad_norm": 0.03363264352083206,
      "learning_rate": 3.885296622196992e-05,
      "loss": 0.0003963438794016838,
      "step": 215510
    },
    {
      "epoch": 61.17513482827136,
      "grad_norm": 0.23795604705810547,
      "learning_rate": 3.8850127732046554e-05,
      "loss": 0.00019614100456237792,
      "step": 215520
    },
    {
      "epoch": 61.177973318194724,
      "grad_norm": 0.05833710730075836,
      "learning_rate": 3.884728924212319e-05,
      "loss": 0.00027827657759189607,
      "step": 215530
    },
    {
      "epoch": 61.18081180811808,
      "grad_norm": 0.07046153396368027,
      "learning_rate": 3.884445075219983e-05,
      "loss": 0.0001988256350159645,
      "step": 215540
    },
    {
      "epoch": 61.18365029804144,
      "grad_norm": 0.03421470522880554,
      "learning_rate": 3.884161226227647e-05,
      "loss": 0.0002629075199365616,
      "step": 215550
    },
    {
      "epoch": 61.186488787964805,
      "grad_norm": 0.02755540795624256,
      "learning_rate": 3.883877377235311e-05,
      "loss": 0.00019657928496599197,
      "step": 215560
    },
    {
      "epoch": 61.18932727788816,
      "grad_norm": 0.379353791475296,
      "learning_rate": 3.883593528242975e-05,
      "loss": 0.0002679511904716492,
      "step": 215570
    },
    {
      "epoch": 61.192165767811524,
      "grad_norm": 0.020547306165099144,
      "learning_rate": 3.883309679250639e-05,
      "loss": 0.0023767832666635514,
      "step": 215580
    },
    {
      "epoch": 61.19500425773489,
      "grad_norm": 0.02539360150694847,
      "learning_rate": 3.883025830258303e-05,
      "loss": 8.322466164827347e-05,
      "step": 215590
    },
    {
      "epoch": 61.19784274765824,
      "grad_norm": 0.01864849030971527,
      "learning_rate": 3.8827419812659665e-05,
      "loss": 0.00016844216734170913,
      "step": 215600
    },
    {
      "epoch": 61.200681237581605,
      "grad_norm": 0.045883096754550934,
      "learning_rate": 3.8824581322736306e-05,
      "loss": 0.0001425381749868393,
      "step": 215610
    },
    {
      "epoch": 61.20351972750497,
      "grad_norm": 0.004148701671510935,
      "learning_rate": 3.882174283281295e-05,
      "loss": 0.0004144530743360519,
      "step": 215620
    },
    {
      "epoch": 61.20635821742833,
      "grad_norm": 0.025514276698231697,
      "learning_rate": 3.881890434288958e-05,
      "loss": 0.0004511922597885132,
      "step": 215630
    },
    {
      "epoch": 61.20919670735169,
      "grad_norm": 0.05463980510830879,
      "learning_rate": 3.8816065852966223e-05,
      "loss": 0.0008060790598392487,
      "step": 215640
    },
    {
      "epoch": 61.21203519727505,
      "grad_norm": 0.03285587206482887,
      "learning_rate": 3.8813227363042865e-05,
      "loss": 0.002129584550857544,
      "step": 215650
    },
    {
      "epoch": 61.21487368719841,
      "grad_norm": 0.05309877544641495,
      "learning_rate": 3.88103888731195e-05,
      "loss": 0.0002880467101931572,
      "step": 215660
    },
    {
      "epoch": 61.21771217712177,
      "grad_norm": 0.051354002207517624,
      "learning_rate": 3.880755038319614e-05,
      "loss": 0.0003772500902414322,
      "step": 215670
    },
    {
      "epoch": 61.22055066704513,
      "grad_norm": 0.12273658812046051,
      "learning_rate": 3.880471189327278e-05,
      "loss": 0.000333344005048275,
      "step": 215680
    },
    {
      "epoch": 61.223389156968494,
      "grad_norm": 0.35780057311058044,
      "learning_rate": 3.8801873403349424e-05,
      "loss": 0.0004122009500861168,
      "step": 215690
    },
    {
      "epoch": 61.22622764689185,
      "grad_norm": 0.010805453173816204,
      "learning_rate": 3.879903491342606e-05,
      "loss": 0.0022517744451761245,
      "step": 215700
    },
    {
      "epoch": 61.22906613681521,
      "grad_norm": 0.06760265678167343,
      "learning_rate": 3.87961964235027e-05,
      "loss": 0.00019772257655858994,
      "step": 215710
    },
    {
      "epoch": 61.231904626738576,
      "grad_norm": 0.004587364383041859,
      "learning_rate": 3.879335793357934e-05,
      "loss": 0.00012964140623807907,
      "step": 215720
    },
    {
      "epoch": 61.23474311666194,
      "grad_norm": 0.07136555016040802,
      "learning_rate": 3.8790519443655976e-05,
      "loss": 0.00017781164497137069,
      "step": 215730
    },
    {
      "epoch": 61.237581606585294,
      "grad_norm": 0.01658431813120842,
      "learning_rate": 3.878768095373262e-05,
      "loss": 0.0004226746037602425,
      "step": 215740
    },
    {
      "epoch": 61.24042009650866,
      "grad_norm": 0.009985741227865219,
      "learning_rate": 3.878484246380926e-05,
      "loss": 0.00011813361197710037,
      "step": 215750
    },
    {
      "epoch": 61.24325858643202,
      "grad_norm": 0.05464964359998703,
      "learning_rate": 3.878200397388589e-05,
      "loss": 0.00017763879150152208,
      "step": 215760
    },
    {
      "epoch": 61.246097076355376,
      "grad_norm": 0.01882336288690567,
      "learning_rate": 3.8779165483962534e-05,
      "loss": 0.00017423294484615327,
      "step": 215770
    },
    {
      "epoch": 61.24893556627874,
      "grad_norm": 0.032023943960666656,
      "learning_rate": 3.8776326994039176e-05,
      "loss": 0.00017803702503442763,
      "step": 215780
    },
    {
      "epoch": 61.2517740562021,
      "grad_norm": 0.008610665798187256,
      "learning_rate": 3.877348850411581e-05,
      "loss": 0.00021099820733070374,
      "step": 215790
    },
    {
      "epoch": 61.254612546125465,
      "grad_norm": 0.004117421340197325,
      "learning_rate": 3.877065001419245e-05,
      "loss": 0.00021890066564083098,
      "step": 215800
    },
    {
      "epoch": 61.25745103604882,
      "grad_norm": 0.030683910474181175,
      "learning_rate": 3.876781152426909e-05,
      "loss": 0.00015872325748205184,
      "step": 215810
    },
    {
      "epoch": 61.26028952597218,
      "grad_norm": 0.08489096164703369,
      "learning_rate": 3.876497303434573e-05,
      "loss": 0.00024553630501031873,
      "step": 215820
    },
    {
      "epoch": 61.263128015895546,
      "grad_norm": 0.03258201479911804,
      "learning_rate": 3.876213454442237e-05,
      "loss": 0.00021523721516132356,
      "step": 215830
    },
    {
      "epoch": 61.2659665058189,
      "grad_norm": 0.029003724455833435,
      "learning_rate": 3.8759296054499004e-05,
      "loss": 0.000756535679101944,
      "step": 215840
    },
    {
      "epoch": 61.268804995742265,
      "grad_norm": 0.008857635781168938,
      "learning_rate": 3.875645756457565e-05,
      "loss": 0.0004236387088894844,
      "step": 215850
    },
    {
      "epoch": 61.27164348566563,
      "grad_norm": 0.0161388348788023,
      "learning_rate": 3.8753619074652286e-05,
      "loss": 0.0008206432685256004,
      "step": 215860
    },
    {
      "epoch": 61.274481975588984,
      "grad_norm": 1.2961947917938232,
      "learning_rate": 3.875078058472892e-05,
      "loss": 0.003052062168717384,
      "step": 215870
    },
    {
      "epoch": 61.27732046551235,
      "grad_norm": 0.07350016385316849,
      "learning_rate": 3.874794209480557e-05,
      "loss": 0.00021300595253705978,
      "step": 215880
    },
    {
      "epoch": 61.28015895543571,
      "grad_norm": 0.0889420211315155,
      "learning_rate": 3.8745103604882204e-05,
      "loss": 0.000515422597527504,
      "step": 215890
    },
    {
      "epoch": 61.28299744535907,
      "grad_norm": 0.725948691368103,
      "learning_rate": 3.8742265114958845e-05,
      "loss": 0.0008290417492389679,
      "step": 215900
    },
    {
      "epoch": 61.28583593528243,
      "grad_norm": 0.13537763059139252,
      "learning_rate": 3.8739426625035486e-05,
      "loss": 0.00027867071330547335,
      "step": 215910
    },
    {
      "epoch": 61.28867442520579,
      "grad_norm": 0.01632503606379032,
      "learning_rate": 3.873658813511212e-05,
      "loss": 0.0034831054508686066,
      "step": 215920
    },
    {
      "epoch": 61.291512915129154,
      "grad_norm": 0.08921882510185242,
      "learning_rate": 3.873374964518876e-05,
      "loss": 0.01066528782248497,
      "step": 215930
    },
    {
      "epoch": 61.29435140505251,
      "grad_norm": 1.153261661529541,
      "learning_rate": 3.87309111552654e-05,
      "loss": 0.019118335843086243,
      "step": 215940
    },
    {
      "epoch": 61.29718989497587,
      "grad_norm": 0.031276777386665344,
      "learning_rate": 3.872807266534204e-05,
      "loss": 0.00027886852622032164,
      "step": 215950
    },
    {
      "epoch": 61.300028384899235,
      "grad_norm": 0.33014097809791565,
      "learning_rate": 3.872523417541868e-05,
      "loss": 0.0037278912961483,
      "step": 215960
    },
    {
      "epoch": 61.30286687482259,
      "grad_norm": 0.02982609160244465,
      "learning_rate": 3.8722395685495314e-05,
      "loss": 0.0031723715364933014,
      "step": 215970
    },
    {
      "epoch": 61.305705364745954,
      "grad_norm": 0.04600028693675995,
      "learning_rate": 3.871955719557196e-05,
      "loss": 0.0001853032037615776,
      "step": 215980
    },
    {
      "epoch": 61.30854385466932,
      "grad_norm": 0.0322023406624794,
      "learning_rate": 3.87167187056486e-05,
      "loss": 0.0008390305563807487,
      "step": 215990
    },
    {
      "epoch": 61.31138234459268,
      "grad_norm": 0.02932199090719223,
      "learning_rate": 3.871388021572523e-05,
      "loss": 0.0013613628223538398,
      "step": 216000
    },
    {
      "epoch": 61.31138234459268,
      "eval_accuracy": 0.9746296178546449,
      "eval_loss": 0.10100961476564407,
      "eval_runtime": 32.8879,
      "eval_samples_per_second": 478.201,
      "eval_steps_per_second": 7.48,
      "step": 216000
    },
    {
      "epoch": 61.314220834516036,
      "grad_norm": 0.10337033867835999,
      "learning_rate": 3.871104172580188e-05,
      "loss": 0.0007008964195847512,
      "step": 216010
    },
    {
      "epoch": 61.3170593244394,
      "grad_norm": 0.5335471034049988,
      "learning_rate": 3.8708203235878514e-05,
      "loss": 0.0003115927800536156,
      "step": 216020
    },
    {
      "epoch": 61.31989781436276,
      "grad_norm": 14.409452438354492,
      "learning_rate": 3.8705364745955156e-05,
      "loss": 0.0036960549652576445,
      "step": 216030
    },
    {
      "epoch": 61.32273630428612,
      "grad_norm": 3.6708271503448486,
      "learning_rate": 3.870252625603179e-05,
      "loss": 0.00253945030272007,
      "step": 216040
    },
    {
      "epoch": 61.32557479420948,
      "grad_norm": 0.005125561263412237,
      "learning_rate": 3.869968776610843e-05,
      "loss": 0.00023162476718425752,
      "step": 216050
    },
    {
      "epoch": 61.32841328413284,
      "grad_norm": 0.025265710428357124,
      "learning_rate": 3.869684927618507e-05,
      "loss": 0.003088388778269291,
      "step": 216060
    },
    {
      "epoch": 61.3312517740562,
      "grad_norm": 0.19009411334991455,
      "learning_rate": 3.869401078626171e-05,
      "loss": 0.00037227123975753785,
      "step": 216070
    },
    {
      "epoch": 61.33409026397956,
      "grad_norm": 4.930599212646484,
      "learning_rate": 3.869117229633835e-05,
      "loss": 0.012727208435535431,
      "step": 216080
    },
    {
      "epoch": 61.336928753902924,
      "grad_norm": 0.7179930210113525,
      "learning_rate": 3.868833380641499e-05,
      "loss": 0.01046122908592224,
      "step": 216090
    },
    {
      "epoch": 61.33976724382629,
      "grad_norm": 0.1588418036699295,
      "learning_rate": 3.8685495316491625e-05,
      "loss": 0.009204002469778061,
      "step": 216100
    },
    {
      "epoch": 61.34260573374964,
      "grad_norm": 0.45971256494522095,
      "learning_rate": 3.8682656826568266e-05,
      "loss": 0.06700435876846314,
      "step": 216110
    },
    {
      "epoch": 61.345444223673006,
      "grad_norm": 1.2575825452804565,
      "learning_rate": 3.867981833664491e-05,
      "loss": 0.00048235543072223664,
      "step": 216120
    },
    {
      "epoch": 61.34828271359637,
      "grad_norm": 0.7565178871154785,
      "learning_rate": 3.867697984672154e-05,
      "loss": 0.0031509183347225187,
      "step": 216130
    },
    {
      "epoch": 61.351121203519725,
      "grad_norm": 0.7258362174034119,
      "learning_rate": 3.8674141356798184e-05,
      "loss": 0.0016086721792817117,
      "step": 216140
    },
    {
      "epoch": 61.35395969344309,
      "grad_norm": 0.565324604511261,
      "learning_rate": 3.8671302866874825e-05,
      "loss": 0.000313415564596653,
      "step": 216150
    },
    {
      "epoch": 61.35679818336645,
      "grad_norm": 3.1536357402801514,
      "learning_rate": 3.8668464376951467e-05,
      "loss": 0.0015474297106266022,
      "step": 216160
    },
    {
      "epoch": 61.35963667328981,
      "grad_norm": 0.2877085506916046,
      "learning_rate": 3.86656258870281e-05,
      "loss": 0.0007391052320599556,
      "step": 216170
    },
    {
      "epoch": 61.36247516321317,
      "grad_norm": 3.4423561096191406,
      "learning_rate": 3.866278739710474e-05,
      "loss": 0.0009197505190968513,
      "step": 216180
    },
    {
      "epoch": 61.36531365313653,
      "grad_norm": 0.3582790195941925,
      "learning_rate": 3.8659948907181384e-05,
      "loss": 0.004711570590734482,
      "step": 216190
    },
    {
      "epoch": 61.368152143059895,
      "grad_norm": 0.12755458056926727,
      "learning_rate": 3.865711041725802e-05,
      "loss": 0.00040833484381437303,
      "step": 216200
    },
    {
      "epoch": 61.37099063298325,
      "grad_norm": 0.02035434916615486,
      "learning_rate": 3.865427192733466e-05,
      "loss": 0.00023771263659000397,
      "step": 216210
    },
    {
      "epoch": 61.37382912290661,
      "grad_norm": 0.07076713442802429,
      "learning_rate": 3.86514334374113e-05,
      "loss": 0.0015757745131850243,
      "step": 216220
    },
    {
      "epoch": 61.376667612829976,
      "grad_norm": 0.06178713217377663,
      "learning_rate": 3.8648594947487936e-05,
      "loss": 0.0005832139402627945,
      "step": 216230
    },
    {
      "epoch": 61.37950610275333,
      "grad_norm": 9.393887519836426,
      "learning_rate": 3.864575645756458e-05,
      "loss": 0.0054584208875894545,
      "step": 216240
    },
    {
      "epoch": 61.382344592676695,
      "grad_norm": 0.3173576295375824,
      "learning_rate": 3.864291796764122e-05,
      "loss": 0.001106717437505722,
      "step": 216250
    },
    {
      "epoch": 61.38518308260006,
      "grad_norm": 0.036778200417757034,
      "learning_rate": 3.864007947771785e-05,
      "loss": 0.0002189323306083679,
      "step": 216260
    },
    {
      "epoch": 61.38802157252342,
      "grad_norm": 0.023009805008769035,
      "learning_rate": 3.8637240987794495e-05,
      "loss": 0.0003602391108870506,
      "step": 216270
    },
    {
      "epoch": 61.39086006244678,
      "grad_norm": 0.0994749590754509,
      "learning_rate": 3.8634402497871136e-05,
      "loss": 0.00029295478016138076,
      "step": 216280
    },
    {
      "epoch": 61.39369855237014,
      "grad_norm": 0.5378846526145935,
      "learning_rate": 3.863156400794777e-05,
      "loss": 0.000308367982506752,
      "step": 216290
    },
    {
      "epoch": 61.3965370422935,
      "grad_norm": 0.007245919201523066,
      "learning_rate": 3.862872551802441e-05,
      "loss": 0.0004074612632393837,
      "step": 216300
    },
    {
      "epoch": 61.39937553221686,
      "grad_norm": 0.3359479010105133,
      "learning_rate": 3.862588702810105e-05,
      "loss": 0.0004731917753815651,
      "step": 216310
    },
    {
      "epoch": 61.40221402214022,
      "grad_norm": 0.04678356647491455,
      "learning_rate": 3.8623048538177695e-05,
      "loss": 0.005140122026205063,
      "step": 216320
    },
    {
      "epoch": 61.405052512063584,
      "grad_norm": 12.768224716186523,
      "learning_rate": 3.862021004825433e-05,
      "loss": 0.0012680761516094207,
      "step": 216330
    },
    {
      "epoch": 61.40789100198694,
      "grad_norm": 0.032740410417318344,
      "learning_rate": 3.861737155833097e-05,
      "loss": 0.0010481109842658044,
      "step": 216340
    },
    {
      "epoch": 61.4107294919103,
      "grad_norm": 0.8256431818008423,
      "learning_rate": 3.861453306840761e-05,
      "loss": 0.0005402708426117897,
      "step": 216350
    },
    {
      "epoch": 61.413567981833665,
      "grad_norm": 0.030901433899998665,
      "learning_rate": 3.861169457848425e-05,
      "loss": 0.0001170651987195015,
      "step": 216360
    },
    {
      "epoch": 61.41640647175703,
      "grad_norm": 0.0486987829208374,
      "learning_rate": 3.860885608856089e-05,
      "loss": 0.0001894034445285797,
      "step": 216370
    },
    {
      "epoch": 61.419244961680384,
      "grad_norm": 0.03201828524470329,
      "learning_rate": 3.860601759863753e-05,
      "loss": 0.0003419985994696617,
      "step": 216380
    },
    {
      "epoch": 61.42208345160375,
      "grad_norm": 0.0667727068066597,
      "learning_rate": 3.8603179108714164e-05,
      "loss": 0.00010155383497476578,
      "step": 216390
    },
    {
      "epoch": 61.42492194152711,
      "grad_norm": 0.4444575905799866,
      "learning_rate": 3.8600340618790805e-05,
      "loss": 0.0012884242460131645,
      "step": 216400
    },
    {
      "epoch": 61.427760431450466,
      "grad_norm": 0.14145974814891815,
      "learning_rate": 3.859750212886745e-05,
      "loss": 0.0002982020378112793,
      "step": 216410
    },
    {
      "epoch": 61.43059892137383,
      "grad_norm": 0.5499688386917114,
      "learning_rate": 3.859466363894408e-05,
      "loss": 0.0002449605613946915,
      "step": 216420
    },
    {
      "epoch": 61.43343741129719,
      "grad_norm": 0.4462849199771881,
      "learning_rate": 3.859182514902072e-05,
      "loss": 0.00024079959839582444,
      "step": 216430
    },
    {
      "epoch": 61.43627590122055,
      "grad_norm": 0.04889913648366928,
      "learning_rate": 3.8588986659097364e-05,
      "loss": 0.0002414163202047348,
      "step": 216440
    },
    {
      "epoch": 61.43911439114391,
      "grad_norm": 0.08770814538002014,
      "learning_rate": 3.8586148169174005e-05,
      "loss": 0.001789306290447712,
      "step": 216450
    },
    {
      "epoch": 61.44195288106727,
      "grad_norm": 0.06726711243391037,
      "learning_rate": 3.858330967925064e-05,
      "loss": 0.0001368945464491844,
      "step": 216460
    },
    {
      "epoch": 61.444791370990636,
      "grad_norm": 0.0825401023030281,
      "learning_rate": 3.8580471189327275e-05,
      "loss": 0.00024643111974000933,
      "step": 216470
    },
    {
      "epoch": 61.44762986091399,
      "grad_norm": 0.38674819469451904,
      "learning_rate": 3.857763269940392e-05,
      "loss": 0.0002783354371786118,
      "step": 216480
    },
    {
      "epoch": 61.450468350837355,
      "grad_norm": 0.003995418548583984,
      "learning_rate": 3.857479420948056e-05,
      "loss": 0.0004913210868835449,
      "step": 216490
    },
    {
      "epoch": 61.45330684076072,
      "grad_norm": 0.08872446417808533,
      "learning_rate": 3.85719557195572e-05,
      "loss": 0.0002280188724398613,
      "step": 216500
    },
    {
      "epoch": 61.45330684076072,
      "eval_accuracy": 0.9781903732434667,
      "eval_loss": 0.07973441481590271,
      "eval_runtime": 32.5395,
      "eval_samples_per_second": 483.32,
      "eval_steps_per_second": 7.56,
      "step": 216500
    },
    {
      "epoch": 61.45614533068407,
      "grad_norm": 0.018108388409018517,
      "learning_rate": 3.856911722963384e-05,
      "loss": 0.0005022900179028511,
      "step": 216510
    },
    {
      "epoch": 61.458983820607436,
      "grad_norm": 0.015644313767552376,
      "learning_rate": 3.8566278739710475e-05,
      "loss": 0.00031647440046072005,
      "step": 216520
    },
    {
      "epoch": 61.4618223105308,
      "grad_norm": 0.038114290684461594,
      "learning_rate": 3.8563440249787116e-05,
      "loss": 0.0005209870636463165,
      "step": 216530
    },
    {
      "epoch": 61.464660800454155,
      "grad_norm": 0.0938488319516182,
      "learning_rate": 3.856060175986376e-05,
      "loss": 0.0017789263278245927,
      "step": 216540
    },
    {
      "epoch": 61.46749929037752,
      "grad_norm": 0.05963509902358055,
      "learning_rate": 3.855776326994039e-05,
      "loss": 0.000708799809217453,
      "step": 216550
    },
    {
      "epoch": 61.47033778030088,
      "grad_norm": 0.16830037534236908,
      "learning_rate": 3.8554924780017033e-05,
      "loss": 0.0007388941943645477,
      "step": 216560
    },
    {
      "epoch": 61.47317627022424,
      "grad_norm": 1.579108715057373,
      "learning_rate": 3.855208629009367e-05,
      "loss": 0.0012810301035642625,
      "step": 216570
    },
    {
      "epoch": 61.4760147601476,
      "grad_norm": 0.032023195177316666,
      "learning_rate": 3.854924780017031e-05,
      "loss": 0.0009687226265668869,
      "step": 216580
    },
    {
      "epoch": 61.47885325007096,
      "grad_norm": 0.08872319757938385,
      "learning_rate": 3.854640931024695e-05,
      "loss": 0.0015518367290496826,
      "step": 216590
    },
    {
      "epoch": 61.481691739994325,
      "grad_norm": 1.124248743057251,
      "learning_rate": 3.8543570820323585e-05,
      "loss": 0.0030122024938464163,
      "step": 216600
    },
    {
      "epoch": 61.48453022991768,
      "grad_norm": 0.06267736107110977,
      "learning_rate": 3.8540732330400234e-05,
      "loss": 0.002822236716747284,
      "step": 216610
    },
    {
      "epoch": 61.487368719841044,
      "grad_norm": 0.029576007276773453,
      "learning_rate": 3.853789384047687e-05,
      "loss": 0.00046252775937318804,
      "step": 216620
    },
    {
      "epoch": 61.49020720976441,
      "grad_norm": 0.07619720697402954,
      "learning_rate": 3.85350553505535e-05,
      "loss": 0.0010661303997039795,
      "step": 216630
    },
    {
      "epoch": 61.49304569968777,
      "grad_norm": 0.13890773057937622,
      "learning_rate": 3.853221686063015e-05,
      "loss": 0.0005632080137729645,
      "step": 216640
    },
    {
      "epoch": 61.495884189611125,
      "grad_norm": 0.03242727369070053,
      "learning_rate": 3.8529378370706786e-05,
      "loss": 0.0021468615159392357,
      "step": 216650
    },
    {
      "epoch": 61.49872267953449,
      "grad_norm": 0.036812588572502136,
      "learning_rate": 3.852653988078343e-05,
      "loss": 0.0024068178609013557,
      "step": 216660
    },
    {
      "epoch": 61.50156116945785,
      "grad_norm": 2.6250011920928955,
      "learning_rate": 3.852370139086006e-05,
      "loss": 0.0007956758141517639,
      "step": 216670
    },
    {
      "epoch": 61.50439965938121,
      "grad_norm": 0.10710686445236206,
      "learning_rate": 3.85208629009367e-05,
      "loss": 0.0008166905492544175,
      "step": 216680
    },
    {
      "epoch": 61.50723814930457,
      "grad_norm": 0.028865627944469452,
      "learning_rate": 3.8518024411013344e-05,
      "loss": 0.00037157442420721054,
      "step": 216690
    },
    {
      "epoch": 61.51007663922793,
      "grad_norm": 0.06609141826629639,
      "learning_rate": 3.851518592108998e-05,
      "loss": 0.00530930832028389,
      "step": 216700
    },
    {
      "epoch": 61.51291512915129,
      "grad_norm": 0.011858562007546425,
      "learning_rate": 3.851234743116662e-05,
      "loss": 0.00030265618115663526,
      "step": 216710
    },
    {
      "epoch": 61.51575361907465,
      "grad_norm": 0.0744750127196312,
      "learning_rate": 3.850950894124326e-05,
      "loss": 0.0016878196969628335,
      "step": 216720
    },
    {
      "epoch": 61.518592108998014,
      "grad_norm": 2.3313846588134766,
      "learning_rate": 3.8506670451319896e-05,
      "loss": 0.0006332077085971832,
      "step": 216730
    },
    {
      "epoch": 61.52143059892138,
      "grad_norm": 0.09978590905666351,
      "learning_rate": 3.8503831961396544e-05,
      "loss": 0.0015172585844993591,
      "step": 216740
    },
    {
      "epoch": 61.52426908884473,
      "grad_norm": 0.3876282572746277,
      "learning_rate": 3.850099347147318e-05,
      "loss": 0.0030840219929814338,
      "step": 216750
    },
    {
      "epoch": 61.527107578768096,
      "grad_norm": 0.8153766989707947,
      "learning_rate": 3.8498154981549814e-05,
      "loss": 0.0008133532479405403,
      "step": 216760
    },
    {
      "epoch": 61.52994606869146,
      "grad_norm": 4.6972880363464355,
      "learning_rate": 3.8495316491626455e-05,
      "loss": 0.003050798922777176,
      "step": 216770
    },
    {
      "epoch": 61.532784558614814,
      "grad_norm": 0.04355913773179054,
      "learning_rate": 3.8492478001703096e-05,
      "loss": 0.003405255824327469,
      "step": 216780
    },
    {
      "epoch": 61.53562304853818,
      "grad_norm": 0.017452916130423546,
      "learning_rate": 3.848963951177974e-05,
      "loss": 0.0009114744141697883,
      "step": 216790
    },
    {
      "epoch": 61.53846153846154,
      "grad_norm": 0.2788509130477905,
      "learning_rate": 3.848680102185637e-05,
      "loss": 0.001282534934580326,
      "step": 216800
    },
    {
      "epoch": 61.541300028384896,
      "grad_norm": 0.04370063915848732,
      "learning_rate": 3.8484246380925345e-05,
      "loss": 0.014765578508377075,
      "step": 216810
    },
    {
      "epoch": 61.54413851830826,
      "grad_norm": 0.47571852803230286,
      "learning_rate": 3.8481407891001994e-05,
      "loss": 0.0013713359832763672,
      "step": 216820
    },
    {
      "epoch": 61.54697700823162,
      "grad_norm": 0.20616039633750916,
      "learning_rate": 3.847856940107863e-05,
      "loss": 0.0008811812847852706,
      "step": 216830
    },
    {
      "epoch": 61.549815498154985,
      "grad_norm": 0.1657957136631012,
      "learning_rate": 3.847573091115527e-05,
      "loss": 0.0005546921864151954,
      "step": 216840
    },
    {
      "epoch": 61.55265398807834,
      "grad_norm": 5.899930477142334,
      "learning_rate": 3.8472892421231904e-05,
      "loss": 0.0016560763120651244,
      "step": 216850
    },
    {
      "epoch": 61.5554924780017,
      "grad_norm": 0.1331617683172226,
      "learning_rate": 3.8470053931308546e-05,
      "loss": 0.0008631845936179161,
      "step": 216860
    },
    {
      "epoch": 61.558330967925066,
      "grad_norm": 0.025597110390663147,
      "learning_rate": 3.846721544138519e-05,
      "loss": 0.0007656343281269073,
      "step": 216870
    },
    {
      "epoch": 61.56116945784842,
      "grad_norm": 0.03603353351354599,
      "learning_rate": 3.846437695146182e-05,
      "loss": 0.0002874046564102173,
      "step": 216880
    },
    {
      "epoch": 61.564007947771785,
      "grad_norm": 0.8880316615104675,
      "learning_rate": 3.846153846153846e-05,
      "loss": 0.0023166486993432046,
      "step": 216890
    },
    {
      "epoch": 61.56684643769515,
      "grad_norm": 0.039738018065690994,
      "learning_rate": 3.8458699971615104e-05,
      "loss": 0.0010177794843912124,
      "step": 216900
    },
    {
      "epoch": 61.56968492761851,
      "grad_norm": 0.2883826494216919,
      "learning_rate": 3.845586148169174e-05,
      "loss": 0.0002113472670316696,
      "step": 216910
    },
    {
      "epoch": 61.572523417541866,
      "grad_norm": 0.06658594310283661,
      "learning_rate": 3.845302299176839e-05,
      "loss": 0.0016385991126298904,
      "step": 216920
    },
    {
      "epoch": 61.57536190746523,
      "grad_norm": 0.16823053359985352,
      "learning_rate": 3.845018450184502e-05,
      "loss": 0.0006068894639611244,
      "step": 216930
    },
    {
      "epoch": 61.57820039738859,
      "grad_norm": 2.0332601070404053,
      "learning_rate": 3.8447346011921656e-05,
      "loss": 0.0008418796584010124,
      "step": 216940
    },
    {
      "epoch": 61.58103888731195,
      "grad_norm": 0.09676778316497803,
      "learning_rate": 3.84445075219983e-05,
      "loss": 0.00016237571835517884,
      "step": 216950
    },
    {
      "epoch": 61.58387737723531,
      "grad_norm": 0.1492823213338852,
      "learning_rate": 3.844166903207494e-05,
      "loss": 0.0004834461957216263,
      "step": 216960
    },
    {
      "epoch": 61.586715867158674,
      "grad_norm": 0.044536590576171875,
      "learning_rate": 3.843883054215158e-05,
      "loss": 0.000254264660179615,
      "step": 216970
    },
    {
      "epoch": 61.58955435708203,
      "grad_norm": 0.03837936371564865,
      "learning_rate": 3.8435992052228215e-05,
      "loss": 0.0003126336261630058,
      "step": 216980
    },
    {
      "epoch": 61.59239284700539,
      "grad_norm": 0.013523315079510212,
      "learning_rate": 3.8433153562304856e-05,
      "loss": 0.00010235235095024109,
      "step": 216990
    },
    {
      "epoch": 61.595231336928755,
      "grad_norm": 0.031085077673196793,
      "learning_rate": 3.84303150723815e-05,
      "loss": 0.00018512457609176636,
      "step": 217000
    },
    {
      "epoch": 61.595231336928755,
      "eval_accuracy": 0.9781267883258091,
      "eval_loss": 0.07921256870031357,
      "eval_runtime": 32.7485,
      "eval_samples_per_second": 480.236,
      "eval_steps_per_second": 7.512,
      "step": 217000
    },
    {
      "epoch": 61.59806982685212,
      "grad_norm": 0.032644327729940414,
      "learning_rate": 3.842747658245813e-05,
      "loss": 0.0008241433650255203,
      "step": 217010
    },
    {
      "epoch": 61.600908316775474,
      "grad_norm": 0.0199650377035141,
      "learning_rate": 3.8424638092534774e-05,
      "loss": 0.00022299699485301973,
      "step": 217020
    },
    {
      "epoch": 61.60374680669884,
      "grad_norm": 1.0879168510437012,
      "learning_rate": 3.8421799602611415e-05,
      "loss": 0.00024086497724056243,
      "step": 217030
    },
    {
      "epoch": 61.6065852966222,
      "grad_norm": 0.016900500282645226,
      "learning_rate": 3.841896111268805e-05,
      "loss": 0.00015025362372398377,
      "step": 217040
    },
    {
      "epoch": 61.609423786545555,
      "grad_norm": 0.10442055761814117,
      "learning_rate": 3.841612262276469e-05,
      "loss": 0.001385180465877056,
      "step": 217050
    },
    {
      "epoch": 61.61226227646892,
      "grad_norm": 0.3163195848464966,
      "learning_rate": 3.841328413284133e-05,
      "loss": 0.0011114569380879403,
      "step": 217060
    },
    {
      "epoch": 61.61510076639228,
      "grad_norm": 0.07231635600328445,
      "learning_rate": 3.841044564291797e-05,
      "loss": 0.0003801813349127769,
      "step": 217070
    },
    {
      "epoch": 61.61793925631564,
      "grad_norm": 0.0867600291967392,
      "learning_rate": 3.840760715299461e-05,
      "loss": 0.00032310187816619873,
      "step": 217080
    },
    {
      "epoch": 61.620777746239,
      "grad_norm": 0.14584192633628845,
      "learning_rate": 3.840476866307125e-05,
      "loss": 0.0011425664648413657,
      "step": 217090
    },
    {
      "epoch": 61.62361623616236,
      "grad_norm": 0.4013562798500061,
      "learning_rate": 3.8401930173147884e-05,
      "loss": 0.00196463018655777,
      "step": 217100
    },
    {
      "epoch": 61.626454726085726,
      "grad_norm": 0.03987845033407211,
      "learning_rate": 3.8399091683224526e-05,
      "loss": 0.0003237469121813774,
      "step": 217110
    },
    {
      "epoch": 61.62929321600908,
      "grad_norm": 0.4696276783943176,
      "learning_rate": 3.839625319330117e-05,
      "loss": 0.00028607696294784545,
      "step": 217120
    },
    {
      "epoch": 61.632131705932444,
      "grad_norm": 0.007483228109776974,
      "learning_rate": 3.839341470337781e-05,
      "loss": 0.0009665843099355698,
      "step": 217130
    },
    {
      "epoch": 61.63497019585581,
      "grad_norm": 0.06793699413537979,
      "learning_rate": 3.839057621345444e-05,
      "loss": 0.0004978250712156296,
      "step": 217140
    },
    {
      "epoch": 61.63780868577916,
      "grad_norm": 0.03911016136407852,
      "learning_rate": 3.838773772353108e-05,
      "loss": 0.0018804160878062247,
      "step": 217150
    },
    {
      "epoch": 61.640647175702526,
      "grad_norm": 0.01066668052226305,
      "learning_rate": 3.8384899233607726e-05,
      "loss": 0.0006683653220534325,
      "step": 217160
    },
    {
      "epoch": 61.64348566562589,
      "grad_norm": 0.06693126261234283,
      "learning_rate": 3.838206074368436e-05,
      "loss": 0.0006136143580079079,
      "step": 217170
    },
    {
      "epoch": 61.646324155549244,
      "grad_norm": 0.015554950572550297,
      "learning_rate": 3.8379222253761e-05,
      "loss": 0.00018095728009939193,
      "step": 217180
    },
    {
      "epoch": 61.64916264547261,
      "grad_norm": 0.013605231419205666,
      "learning_rate": 3.837638376383764e-05,
      "loss": 0.0033058110624551774,
      "step": 217190
    },
    {
      "epoch": 61.65200113539597,
      "grad_norm": 0.03259960189461708,
      "learning_rate": 3.837354527391428e-05,
      "loss": 0.000248744897544384,
      "step": 217200
    },
    {
      "epoch": 61.65483962531933,
      "grad_norm": 0.4730927348136902,
      "learning_rate": 3.837070678399092e-05,
      "loss": 0.0006757820025086403,
      "step": 217210
    },
    {
      "epoch": 61.65767811524269,
      "grad_norm": 0.009730007499456406,
      "learning_rate": 3.836786829406756e-05,
      "loss": 0.0009155202656984329,
      "step": 217220
    },
    {
      "epoch": 61.66051660516605,
      "grad_norm": 2.490445375442505,
      "learning_rate": 3.8365029804144195e-05,
      "loss": 0.005714148283004761,
      "step": 217230
    },
    {
      "epoch": 61.663355095089415,
      "grad_norm": 0.02824172005057335,
      "learning_rate": 3.8362191314220836e-05,
      "loss": 0.0009842133149504662,
      "step": 217240
    },
    {
      "epoch": 61.66619358501277,
      "grad_norm": 0.13608810305595398,
      "learning_rate": 3.835935282429747e-05,
      "loss": 0.0010041950270533561,
      "step": 217250
    },
    {
      "epoch": 61.66903207493613,
      "grad_norm": 0.12815061211585999,
      "learning_rate": 3.835651433437412e-05,
      "loss": 0.0009767692536115646,
      "step": 217260
    },
    {
      "epoch": 61.671870564859496,
      "grad_norm": 0.17982962727546692,
      "learning_rate": 3.8353675844450754e-05,
      "loss": 0.002981898933649063,
      "step": 217270
    },
    {
      "epoch": 61.67470905478285,
      "grad_norm": 0.04425077885389328,
      "learning_rate": 3.835083735452739e-05,
      "loss": 0.0013688374310731888,
      "step": 217280
    },
    {
      "epoch": 61.677547544706215,
      "grad_norm": 0.2152532935142517,
      "learning_rate": 3.8347998864604037e-05,
      "loss": 0.0006952086463570595,
      "step": 217290
    },
    {
      "epoch": 61.68038603462958,
      "grad_norm": 0.2223113775253296,
      "learning_rate": 3.834516037468067e-05,
      "loss": 0.0007308265194296837,
      "step": 217300
    },
    {
      "epoch": 61.68322452455294,
      "grad_norm": 0.0898200273513794,
      "learning_rate": 3.834232188475731e-05,
      "loss": 0.007946072518825531,
      "step": 217310
    },
    {
      "epoch": 61.6860630144763,
      "grad_norm": 13.807066917419434,
      "learning_rate": 3.8339483394833954e-05,
      "loss": 0.0027885353192687035,
      "step": 217320
    },
    {
      "epoch": 61.68890150439966,
      "grad_norm": 0.048954565078020096,
      "learning_rate": 3.833664490491059e-05,
      "loss": 0.0003602685406804085,
      "step": 217330
    },
    {
      "epoch": 61.69173999432302,
      "grad_norm": 0.07136213779449463,
      "learning_rate": 3.833380641498723e-05,
      "loss": 0.00544116348028183,
      "step": 217340
    },
    {
      "epoch": 61.69457848424638,
      "grad_norm": 0.008069753646850586,
      "learning_rate": 3.8330967925063864e-05,
      "loss": 0.00019188523292541505,
      "step": 217350
    },
    {
      "epoch": 61.69741697416974,
      "grad_norm": 0.18848802149295807,
      "learning_rate": 3.8328129435140506e-05,
      "loss": 0.00023871399462223054,
      "step": 217360
    },
    {
      "epoch": 61.700255464093104,
      "grad_norm": 0.2632492184638977,
      "learning_rate": 3.832529094521715e-05,
      "loss": 0.0003208689391613007,
      "step": 217370
    },
    {
      "epoch": 61.70309395401647,
      "grad_norm": 0.017448343336582184,
      "learning_rate": 3.832245245529378e-05,
      "loss": 0.00023610778152942656,
      "step": 217380
    },
    {
      "epoch": 61.70593244393982,
      "grad_norm": 0.24063870310783386,
      "learning_rate": 3.831961396537043e-05,
      "loss": 0.00018811449408531188,
      "step": 217390
    },
    {
      "epoch": 61.708770933863185,
      "grad_norm": 0.09818367660045624,
      "learning_rate": 3.8316775475447065e-05,
      "loss": 0.00014753080904483795,
      "step": 217400
    },
    {
      "epoch": 61.71160942378655,
      "grad_norm": 0.08642785996198654,
      "learning_rate": 3.83139369855237e-05,
      "loss": 0.00015975721180438995,
      "step": 217410
    },
    {
      "epoch": 61.714447913709904,
      "grad_norm": 0.013072864152491093,
      "learning_rate": 3.831109849560035e-05,
      "loss": 0.001008422113955021,
      "step": 217420
    },
    {
      "epoch": 61.71728640363327,
      "grad_norm": 0.017754510045051575,
      "learning_rate": 3.830826000567698e-05,
      "loss": 0.00015929248183965683,
      "step": 217430
    },
    {
      "epoch": 61.72012489355663,
      "grad_norm": 0.15794241428375244,
      "learning_rate": 3.830542151575362e-05,
      "loss": 0.00020736884325742722,
      "step": 217440
    },
    {
      "epoch": 61.722963383479986,
      "grad_norm": 0.010171509347856045,
      "learning_rate": 3.830258302583026e-05,
      "loss": 0.00011540111154317856,
      "step": 217450
    },
    {
      "epoch": 61.72580187340335,
      "grad_norm": 0.05871286243200302,
      "learning_rate": 3.82997445359069e-05,
      "loss": 0.00014006700366735457,
      "step": 217460
    },
    {
      "epoch": 61.72864036332671,
      "grad_norm": 0.04364423081278801,
      "learning_rate": 3.829690604598354e-05,
      "loss": 0.00011705216020345688,
      "step": 217470
    },
    {
      "epoch": 61.731478853250074,
      "grad_norm": 0.08226942270994186,
      "learning_rate": 3.8294067556060175e-05,
      "loss": 0.00022198930382728578,
      "step": 217480
    },
    {
      "epoch": 61.73431734317343,
      "grad_norm": 0.030558815225958824,
      "learning_rate": 3.8291229066136817e-05,
      "loss": 0.0004575246945023537,
      "step": 217490
    },
    {
      "epoch": 61.73715583309679,
      "grad_norm": 0.022238383069634438,
      "learning_rate": 3.828839057621346e-05,
      "loss": 0.00035485122352838514,
      "step": 217500
    },
    {
      "epoch": 61.73715583309679,
      "eval_accuracy": 0.9781267883258091,
      "eval_loss": 0.07809249311685562,
      "eval_runtime": 32.3747,
      "eval_samples_per_second": 485.781,
      "eval_steps_per_second": 7.599,
      "step": 217500
    },
    {
      "epoch": 61.739994323020156,
      "grad_norm": 0.04953913763165474,
      "learning_rate": 3.828555208629009e-05,
      "loss": 0.00041917525231838226,
      "step": 217510
    },
    {
      "epoch": 61.74283281294351,
      "grad_norm": 0.010976020246744156,
      "learning_rate": 3.8282713596366734e-05,
      "loss": 0.0002754954621195793,
      "step": 217520
    },
    {
      "epoch": 61.745671302866874,
      "grad_norm": 0.1964559257030487,
      "learning_rate": 3.8279875106443375e-05,
      "loss": 0.0004724917933344841,
      "step": 217530
    },
    {
      "epoch": 61.74850979279024,
      "grad_norm": 0.9222380518913269,
      "learning_rate": 3.827703661652001e-05,
      "loss": 0.0007005743682384491,
      "step": 217540
    },
    {
      "epoch": 61.75134828271359,
      "grad_norm": 0.07457002252340317,
      "learning_rate": 3.827419812659665e-05,
      "loss": 0.0008480541408061981,
      "step": 217550
    },
    {
      "epoch": 61.754186772636956,
      "grad_norm": 0.23058953881263733,
      "learning_rate": 3.827135963667329e-05,
      "loss": 0.0002143649384379387,
      "step": 217560
    },
    {
      "epoch": 61.75702526256032,
      "grad_norm": 0.018108025193214417,
      "learning_rate": 3.826852114674993e-05,
      "loss": 0.0014776678755879403,
      "step": 217570
    },
    {
      "epoch": 61.75986375248368,
      "grad_norm": 0.41060641407966614,
      "learning_rate": 3.826568265682657e-05,
      "loss": 0.0020970242097973824,
      "step": 217580
    },
    {
      "epoch": 61.76270224240704,
      "grad_norm": 0.06185018643736839,
      "learning_rate": 3.826284416690321e-05,
      "loss": 0.0004948245361447335,
      "step": 217590
    },
    {
      "epoch": 61.7655407323304,
      "grad_norm": 0.019518911838531494,
      "learning_rate": 3.826000567697985e-05,
      "loss": 0.0006813149899244309,
      "step": 217600
    },
    {
      "epoch": 61.76837922225376,
      "grad_norm": 0.030522342771291733,
      "learning_rate": 3.8257167187056486e-05,
      "loss": 0.00018682610243558884,
      "step": 217610
    },
    {
      "epoch": 61.77121771217712,
      "grad_norm": 0.07573335617780685,
      "learning_rate": 3.825432869713313e-05,
      "loss": 0.0001832040026783943,
      "step": 217620
    },
    {
      "epoch": 61.77405620210048,
      "grad_norm": 0.33344289660453796,
      "learning_rate": 3.825149020720977e-05,
      "loss": 0.00022789575159549714,
      "step": 217630
    },
    {
      "epoch": 61.776894692023845,
      "grad_norm": 0.9697441458702087,
      "learning_rate": 3.82486517172864e-05,
      "loss": 0.0006881020963191986,
      "step": 217640
    },
    {
      "epoch": 61.7797331819472,
      "grad_norm": 0.03366716578602791,
      "learning_rate": 3.8245813227363045e-05,
      "loss": 0.00013221874833106994,
      "step": 217650
    },
    {
      "epoch": 61.78257167187056,
      "grad_norm": 0.1800769418478012,
      "learning_rate": 3.8242974737439686e-05,
      "loss": 0.0001564299687743187,
      "step": 217660
    },
    {
      "epoch": 61.785410161793926,
      "grad_norm": 0.04002588614821434,
      "learning_rate": 3.824013624751632e-05,
      "loss": 0.0001950930804014206,
      "step": 217670
    },
    {
      "epoch": 61.78824865171729,
      "grad_norm": 0.07872151583433151,
      "learning_rate": 3.823729775759296e-05,
      "loss": 0.00012639742344617845,
      "step": 217680
    },
    {
      "epoch": 61.791087141640645,
      "grad_norm": 0.01501628290861845,
      "learning_rate": 3.8234459267669603e-05,
      "loss": 0.00010744854807853699,
      "step": 217690
    },
    {
      "epoch": 61.79392563156401,
      "grad_norm": 0.5224476456642151,
      "learning_rate": 3.823162077774624e-05,
      "loss": 0.00015937108546495437,
      "step": 217700
    },
    {
      "epoch": 61.79676412148737,
      "grad_norm": 0.14197589457035065,
      "learning_rate": 3.822878228782288e-05,
      "loss": 0.0002589670941233635,
      "step": 217710
    },
    {
      "epoch": 61.79960261141073,
      "grad_norm": 0.07508742809295654,
      "learning_rate": 3.822594379789952e-05,
      "loss": 0.00014753807336091994,
      "step": 217720
    },
    {
      "epoch": 61.80244110133409,
      "grad_norm": 0.028729557991027832,
      "learning_rate": 3.822310530797616e-05,
      "loss": 0.00019843950867652892,
      "step": 217730
    },
    {
      "epoch": 61.80527959125745,
      "grad_norm": 0.1470598429441452,
      "learning_rate": 3.82202668180528e-05,
      "loss": 0.0002770736813545227,
      "step": 217740
    },
    {
      "epoch": 61.808118081180815,
      "grad_norm": 0.03908504173159599,
      "learning_rate": 3.821742832812943e-05,
      "loss": 0.0002984924241900444,
      "step": 217750
    },
    {
      "epoch": 61.81095657110417,
      "grad_norm": 0.17754800617694855,
      "learning_rate": 3.821458983820608e-05,
      "loss": 0.00022840015590190886,
      "step": 217760
    },
    {
      "epoch": 61.813795061027534,
      "grad_norm": 1.180654525756836,
      "learning_rate": 3.8211751348282714e-05,
      "loss": 0.0010582447052001954,
      "step": 217770
    },
    {
      "epoch": 61.8166335509509,
      "grad_norm": 0.2780829966068268,
      "learning_rate": 3.8208912858359355e-05,
      "loss": 0.001037704385817051,
      "step": 217780
    },
    {
      "epoch": 61.81947204087425,
      "grad_norm": 0.04650702700018883,
      "learning_rate": 3.8206074368436e-05,
      "loss": 0.0002873804420232773,
      "step": 217790
    },
    {
      "epoch": 61.822310530797616,
      "grad_norm": 0.1960422396659851,
      "learning_rate": 3.820323587851263e-05,
      "loss": 0.0005781611427664757,
      "step": 217800
    },
    {
      "epoch": 61.82514902072098,
      "grad_norm": 0.1644396334886551,
      "learning_rate": 3.820039738858927e-05,
      "loss": 0.0025014590471982958,
      "step": 217810
    },
    {
      "epoch": 61.827987510644334,
      "grad_norm": 0.02390863373875618,
      "learning_rate": 3.8197558898665914e-05,
      "loss": 0.0002692295238375664,
      "step": 217820
    },
    {
      "epoch": 61.8308260005677,
      "grad_norm": 0.02874595671892166,
      "learning_rate": 3.819472040874255e-05,
      "loss": 0.00030554290860891343,
      "step": 217830
    },
    {
      "epoch": 61.83366449049106,
      "grad_norm": 0.23946499824523926,
      "learning_rate": 3.819188191881919e-05,
      "loss": 0.0008040284737944603,
      "step": 217840
    },
    {
      "epoch": 61.83650298041442,
      "grad_norm": 0.1272350549697876,
      "learning_rate": 3.8189043428895825e-05,
      "loss": 0.00016713421791791916,
      "step": 217850
    },
    {
      "epoch": 61.83934147033778,
      "grad_norm": 0.04596645012497902,
      "learning_rate": 3.818620493897247e-05,
      "loss": 0.00022974777966737746,
      "step": 217860
    },
    {
      "epoch": 61.84217996026114,
      "grad_norm": 0.09199264645576477,
      "learning_rate": 3.818336644904911e-05,
      "loss": 0.00024253036826848984,
      "step": 217870
    },
    {
      "epoch": 61.845018450184504,
      "grad_norm": 0.8560591340065002,
      "learning_rate": 3.818052795912574e-05,
      "loss": 0.001969059370458126,
      "step": 217880
    },
    {
      "epoch": 61.84785694010786,
      "grad_norm": 5.865027904510498,
      "learning_rate": 3.817768946920239e-05,
      "loss": 0.002126147411763668,
      "step": 217890
    },
    {
      "epoch": 61.85069543003122,
      "grad_norm": 0.029618939384818077,
      "learning_rate": 3.8174850979279025e-05,
      "loss": 0.00042810067534446714,
      "step": 217900
    },
    {
      "epoch": 61.853533919954586,
      "grad_norm": 1.212518334388733,
      "learning_rate": 3.8172012489355666e-05,
      "loss": 0.0010757675394415856,
      "step": 217910
    },
    {
      "epoch": 61.85637240987794,
      "grad_norm": 0.11654415726661682,
      "learning_rate": 3.816917399943231e-05,
      "loss": 0.0011976499110460282,
      "step": 217920
    },
    {
      "epoch": 61.859210899801305,
      "grad_norm": 0.8900585770606995,
      "learning_rate": 3.816633550950894e-05,
      "loss": 0.0013878131285309792,
      "step": 217930
    },
    {
      "epoch": 61.86204938972467,
      "grad_norm": 0.06424091756343842,
      "learning_rate": 3.8163497019585584e-05,
      "loss": 0.0008834740146994591,
      "step": 217940
    },
    {
      "epoch": 61.86488787964803,
      "grad_norm": 1.0066320896148682,
      "learning_rate": 3.816065852966222e-05,
      "loss": 0.0010118091478943825,
      "step": 217950
    },
    {
      "epoch": 61.867726369571386,
      "grad_norm": 0.0073267812840640545,
      "learning_rate": 3.815782003973886e-05,
      "loss": 0.006073351949453354,
      "step": 217960
    },
    {
      "epoch": 61.87056485949475,
      "grad_norm": 0.1754620224237442,
      "learning_rate": 3.81549815498155e-05,
      "loss": 0.0007122492417693139,
      "step": 217970
    },
    {
      "epoch": 61.87340334941811,
      "grad_norm": 0.2582088112831116,
      "learning_rate": 3.8152143059892136e-05,
      "loss": 0.0003916662186384201,
      "step": 217980
    },
    {
      "epoch": 61.87624183934147,
      "grad_norm": 0.08018472790718079,
      "learning_rate": 3.814930456996878e-05,
      "loss": 0.00019261129200458526,
      "step": 217990
    },
    {
      "epoch": 61.87908032926483,
      "grad_norm": 0.0822259932756424,
      "learning_rate": 3.814646608004542e-05,
      "loss": 0.0001356717199087143,
      "step": 218000
    },
    {
      "epoch": 61.87908032926483,
      "eval_accuracy": 0.9776181089845488,
      "eval_loss": 0.0806901752948761,
      "eval_runtime": 32.9001,
      "eval_samples_per_second": 478.023,
      "eval_steps_per_second": 7.477,
      "step": 218000
    },
    {
      "epoch": 61.88191881918819,
      "grad_norm": 0.10943414270877838,
      "learning_rate": 3.814362759012205e-05,
      "loss": 0.00022743958979845046,
      "step": 218010
    },
    {
      "epoch": 61.88475730911155,
      "grad_norm": 0.34840843081474304,
      "learning_rate": 3.81407891001987e-05,
      "loss": 0.00040565002709627154,
      "step": 218020
    },
    {
      "epoch": 61.88759579903491,
      "grad_norm": 0.08743405342102051,
      "learning_rate": 3.8137950610275336e-05,
      "loss": 0.00041940119117498396,
      "step": 218030
    },
    {
      "epoch": 61.890434288958275,
      "grad_norm": 0.06301245838403702,
      "learning_rate": 3.813511212035197e-05,
      "loss": 0.0002447642385959625,
      "step": 218040
    },
    {
      "epoch": 61.89327277888164,
      "grad_norm": 0.06410734355449677,
      "learning_rate": 3.813227363042862e-05,
      "loss": 0.0016137035563588142,
      "step": 218050
    },
    {
      "epoch": 61.896111268804994,
      "grad_norm": 0.5701583027839661,
      "learning_rate": 3.812943514050525e-05,
      "loss": 0.0005459537729620934,
      "step": 218060
    },
    {
      "epoch": 61.89894975872836,
      "grad_norm": 0.589994490146637,
      "learning_rate": 3.8126596650581894e-05,
      "loss": 0.0018453694880008697,
      "step": 218070
    },
    {
      "epoch": 61.90178824865172,
      "grad_norm": 0.07172094285488129,
      "learning_rate": 3.812375816065853e-05,
      "loss": 0.0007000437006354332,
      "step": 218080
    },
    {
      "epoch": 61.904626738575075,
      "grad_norm": 0.010028081014752388,
      "learning_rate": 3.812091967073517e-05,
      "loss": 0.0053277663886547085,
      "step": 218090
    },
    {
      "epoch": 61.90746522849844,
      "grad_norm": 0.16390958428382874,
      "learning_rate": 3.811808118081181e-05,
      "loss": 0.0007159151136875153,
      "step": 218100
    },
    {
      "epoch": 61.9103037184218,
      "grad_norm": 0.6031491160392761,
      "learning_rate": 3.8115242690888446e-05,
      "loss": 0.0006224658340215683,
      "step": 218110
    },
    {
      "epoch": 61.913142208345164,
      "grad_norm": 0.3407686948776245,
      "learning_rate": 3.811240420096509e-05,
      "loss": 0.0004311058670282364,
      "step": 218120
    },
    {
      "epoch": 61.91598069826852,
      "grad_norm": 0.019475551322102547,
      "learning_rate": 3.810956571104173e-05,
      "loss": 0.000987803004682064,
      "step": 218130
    },
    {
      "epoch": 61.91881918819188,
      "grad_norm": 0.19623319804668427,
      "learning_rate": 3.8106727221118364e-05,
      "loss": 0.0005460405722260475,
      "step": 218140
    },
    {
      "epoch": 61.921657678115245,
      "grad_norm": 4.246057033538818,
      "learning_rate": 3.810388873119501e-05,
      "loss": 0.0005904318764805794,
      "step": 218150
    },
    {
      "epoch": 61.9244961680386,
      "grad_norm": 0.019376257434487343,
      "learning_rate": 3.8101050241271646e-05,
      "loss": 0.001971563883125782,
      "step": 218160
    },
    {
      "epoch": 61.927334657961964,
      "grad_norm": 10.392424583435059,
      "learning_rate": 3.809821175134828e-05,
      "loss": 0.005836671218276024,
      "step": 218170
    },
    {
      "epoch": 61.93017314788533,
      "grad_norm": 1.6899350881576538,
      "learning_rate": 3.809537326142492e-05,
      "loss": 0.02230727970600128,
      "step": 218180
    },
    {
      "epoch": 61.93301163780868,
      "grad_norm": 0.08037459850311279,
      "learning_rate": 3.8092534771501564e-05,
      "loss": 0.00044489521533250807,
      "step": 218190
    },
    {
      "epoch": 61.935850127732046,
      "grad_norm": 0.01890498772263527,
      "learning_rate": 3.8089696281578205e-05,
      "loss": 0.0017836006358265878,
      "step": 218200
    },
    {
      "epoch": 61.93868861765541,
      "grad_norm": 0.035886578261852264,
      "learning_rate": 3.808685779165484e-05,
      "loss": 0.002017683908343315,
      "step": 218210
    },
    {
      "epoch": 61.94152710757877,
      "grad_norm": 0.1778830587863922,
      "learning_rate": 3.808401930173148e-05,
      "loss": 0.0005489835515618324,
      "step": 218220
    },
    {
      "epoch": 61.94436559750213,
      "grad_norm": 0.14478468894958496,
      "learning_rate": 3.808118081180812e-05,
      "loss": 0.0002561919391155243,
      "step": 218230
    },
    {
      "epoch": 61.94720408742549,
      "grad_norm": 0.7613511085510254,
      "learning_rate": 3.807834232188476e-05,
      "loss": 0.0017649397253990174,
      "step": 218240
    },
    {
      "epoch": 61.95004257734885,
      "grad_norm": 0.1005609929561615,
      "learning_rate": 3.80755038319614e-05,
      "loss": 0.001559816487133503,
      "step": 218250
    },
    {
      "epoch": 61.95288106727221,
      "grad_norm": 0.13863050937652588,
      "learning_rate": 3.807266534203804e-05,
      "loss": 0.0014277094975113868,
      "step": 218260
    },
    {
      "epoch": 61.95571955719557,
      "grad_norm": 0.02590377815067768,
      "learning_rate": 3.8069826852114674e-05,
      "loss": 0.0016290482133626937,
      "step": 218270
    },
    {
      "epoch": 61.958558047118935,
      "grad_norm": 0.08019240200519562,
      "learning_rate": 3.8066988362191316e-05,
      "loss": 0.0006424658000469208,
      "step": 218280
    },
    {
      "epoch": 61.96139653704229,
      "grad_norm": 0.05989689379930496,
      "learning_rate": 3.806414987226796e-05,
      "loss": 0.0006014382466673851,
      "step": 218290
    },
    {
      "epoch": 61.96423502696565,
      "grad_norm": 0.0120118148624897,
      "learning_rate": 3.806131138234459e-05,
      "loss": 0.001042616367340088,
      "step": 218300
    },
    {
      "epoch": 61.967073516889016,
      "grad_norm": 0.010656965896487236,
      "learning_rate": 3.805847289242123e-05,
      "loss": 0.0007512902840971947,
      "step": 218310
    },
    {
      "epoch": 61.96991200681238,
      "grad_norm": 0.1776086390018463,
      "learning_rate": 3.8055634402497875e-05,
      "loss": 0.00037555508315563203,
      "step": 218320
    },
    {
      "epoch": 61.972750496735735,
      "grad_norm": 0.005589599721133709,
      "learning_rate": 3.8052795912574516e-05,
      "loss": 0.0005252476781606674,
      "step": 218330
    },
    {
      "epoch": 61.9755889866591,
      "grad_norm": 0.047544583678245544,
      "learning_rate": 3.804995742265115e-05,
      "loss": 0.0003793686628341675,
      "step": 218340
    },
    {
      "epoch": 61.97842747658246,
      "grad_norm": 0.40727704763412476,
      "learning_rate": 3.804711893272779e-05,
      "loss": 0.00035220328718423846,
      "step": 218350
    },
    {
      "epoch": 61.981265966505816,
      "grad_norm": 0.0571572482585907,
      "learning_rate": 3.804428044280443e-05,
      "loss": 0.0030634259805083276,
      "step": 218360
    },
    {
      "epoch": 61.98410445642918,
      "grad_norm": 1.0696675777435303,
      "learning_rate": 3.804144195288107e-05,
      "loss": 0.000828913226723671,
      "step": 218370
    },
    {
      "epoch": 61.98694294635254,
      "grad_norm": 0.5296390652656555,
      "learning_rate": 3.803860346295771e-05,
      "loss": 0.0011889945715665817,
      "step": 218380
    },
    {
      "epoch": 61.9897814362759,
      "grad_norm": 0.03299457207322121,
      "learning_rate": 3.803576497303435e-05,
      "loss": 0.005045325309038162,
      "step": 218390
    },
    {
      "epoch": 61.99261992619926,
      "grad_norm": 0.041387103497982025,
      "learning_rate": 3.8032926483110985e-05,
      "loss": 0.00047532860189676286,
      "step": 218400
    },
    {
      "epoch": 61.995458416122624,
      "grad_norm": 0.1026676744222641,
      "learning_rate": 3.8030087993187627e-05,
      "loss": 0.0024150582030415535,
      "step": 218410
    },
    {
      "epoch": 61.99829690604599,
      "grad_norm": 0.014434552751481533,
      "learning_rate": 3.802724950326427e-05,
      "loss": 0.0003462545573711395,
      "step": 218420
    },
    {
      "epoch": 62.00113539596934,
      "grad_norm": 0.10152295231819153,
      "learning_rate": 3.80244110133409e-05,
      "loss": 0.00018874008674174547,
      "step": 218430
    },
    {
      "epoch": 62.003973885892705,
      "grad_norm": 0.1566774547100067,
      "learning_rate": 3.8021572523417544e-05,
      "loss": 0.0007962495088577271,
      "step": 218440
    },
    {
      "epoch": 62.00681237581607,
      "grad_norm": 0.008062702603638172,
      "learning_rate": 3.8018734033494185e-05,
      "loss": 0.00019996333867311478,
      "step": 218450
    },
    {
      "epoch": 62.009650865739424,
      "grad_norm": 3.7242159843444824,
      "learning_rate": 3.801589554357082e-05,
      "loss": 0.0012314040213823318,
      "step": 218460
    },
    {
      "epoch": 62.01248935566279,
      "grad_norm": 0.040409885346889496,
      "learning_rate": 3.801305705364746e-05,
      "loss": 0.0002881327643990517,
      "step": 218470
    },
    {
      "epoch": 62.01532784558615,
      "grad_norm": 0.03602750599384308,
      "learning_rate": 3.8010218563724096e-05,
      "loss": 0.002608831599354744,
      "step": 218480
    },
    {
      "epoch": 62.018166335509505,
      "grad_norm": 0.008669503964483738,
      "learning_rate": 3.8007380073800744e-05,
      "loss": 0.0019926443696022035,
      "step": 218490
    },
    {
      "epoch": 62.02100482543287,
      "grad_norm": 0.006281090900301933,
      "learning_rate": 3.800454158387738e-05,
      "loss": 0.0001863257959485054,
      "step": 218500
    },
    {
      "epoch": 62.02100482543287,
      "eval_accuracy": 0.9781903732434667,
      "eval_loss": 0.07797908782958984,
      "eval_runtime": 33.1756,
      "eval_samples_per_second": 474.054,
      "eval_steps_per_second": 7.415,
      "step": 218500
    },
    {
      "epoch": 62.02384331535623,
      "grad_norm": 0.025058424100279808,
      "learning_rate": 3.800170309395401e-05,
      "loss": 0.0004523938521742821,
      "step": 218510
    },
    {
      "epoch": 62.026681805279594,
      "grad_norm": 0.031698934733867645,
      "learning_rate": 3.799886460403066e-05,
      "loss": 0.000352846086025238,
      "step": 218520
    },
    {
      "epoch": 62.02952029520295,
      "grad_norm": 3.3013434410095215,
      "learning_rate": 3.7996026114107296e-05,
      "loss": 0.007594221085309982,
      "step": 218530
    },
    {
      "epoch": 62.03235878512631,
      "grad_norm": 0.053129155188798904,
      "learning_rate": 3.799318762418394e-05,
      "loss": 0.0003856644034385681,
      "step": 218540
    },
    {
      "epoch": 62.035197275049676,
      "grad_norm": 0.11171478033065796,
      "learning_rate": 3.799034913426058e-05,
      "loss": 0.008635406196117402,
      "step": 218550
    },
    {
      "epoch": 62.03803576497303,
      "grad_norm": 0.4059250056743622,
      "learning_rate": 3.798751064433721e-05,
      "loss": 0.0002635778859257698,
      "step": 218560
    },
    {
      "epoch": 62.040874254896394,
      "grad_norm": 0.003278650576248765,
      "learning_rate": 3.7984672154413855e-05,
      "loss": 0.000611814297735691,
      "step": 218570
    },
    {
      "epoch": 62.04371274481976,
      "grad_norm": 0.06394561380147934,
      "learning_rate": 3.798183366449049e-05,
      "loss": 0.0011465108022093774,
      "step": 218580
    },
    {
      "epoch": 62.04655123474312,
      "grad_norm": 2.870055913925171,
      "learning_rate": 3.797899517456713e-05,
      "loss": 0.0030003517866134645,
      "step": 218590
    },
    {
      "epoch": 62.049389724666476,
      "grad_norm": 0.06478840857744217,
      "learning_rate": 3.797615668464377e-05,
      "loss": 0.0003382187336683273,
      "step": 218600
    },
    {
      "epoch": 62.05222821458984,
      "grad_norm": 5.38042688369751,
      "learning_rate": 3.797331819472041e-05,
      "loss": 0.00080911535769701,
      "step": 218610
    },
    {
      "epoch": 62.0550667045132,
      "grad_norm": 0.009469506330788136,
      "learning_rate": 3.7970479704797055e-05,
      "loss": 0.00022126678377389907,
      "step": 218620
    },
    {
      "epoch": 62.05790519443656,
      "grad_norm": 0.0882493108510971,
      "learning_rate": 3.796764121487369e-05,
      "loss": 0.00010358002036809921,
      "step": 218630
    },
    {
      "epoch": 62.06074368435992,
      "grad_norm": 0.21512244641780853,
      "learning_rate": 3.7964802724950324e-05,
      "loss": 0.00036390181630849836,
      "step": 218640
    },
    {
      "epoch": 62.06358217428328,
      "grad_norm": 0.09913040697574615,
      "learning_rate": 3.796196423502697e-05,
      "loss": 0.00013126786798238753,
      "step": 218650
    },
    {
      "epoch": 62.06642066420664,
      "grad_norm": 3.011153221130371,
      "learning_rate": 3.795912574510361e-05,
      "loss": 0.0008127864450216294,
      "step": 218660
    },
    {
      "epoch": 62.06925915413,
      "grad_norm": 0.07965370267629623,
      "learning_rate": 3.795628725518025e-05,
      "loss": 0.00017133615911006926,
      "step": 218670
    },
    {
      "epoch": 62.072097644053365,
      "grad_norm": 0.04524220526218414,
      "learning_rate": 3.795344876525688e-05,
      "loss": 0.00029147397726774215,
      "step": 218680
    },
    {
      "epoch": 62.07493613397673,
      "grad_norm": 0.01728985644876957,
      "learning_rate": 3.7950610275333524e-05,
      "loss": 0.0002709727734327316,
      "step": 218690
    },
    {
      "epoch": 62.07777462390008,
      "grad_norm": 0.7607879638671875,
      "learning_rate": 3.7947771785410165e-05,
      "loss": 0.0008877310901880264,
      "step": 218700
    },
    {
      "epoch": 62.080613113823446,
      "grad_norm": 0.022659361362457275,
      "learning_rate": 3.79449332954868e-05,
      "loss": 0.00017004497349262237,
      "step": 218710
    },
    {
      "epoch": 62.08345160374681,
      "grad_norm": 0.16287976503372192,
      "learning_rate": 3.794209480556344e-05,
      "loss": 0.0003151360899209976,
      "step": 218720
    },
    {
      "epoch": 62.086290093670165,
      "grad_norm": 0.1375291496515274,
      "learning_rate": 3.793925631564008e-05,
      "loss": 0.0023483801633119583,
      "step": 218730
    },
    {
      "epoch": 62.08912858359353,
      "grad_norm": 0.030452726408839226,
      "learning_rate": 3.793641782571672e-05,
      "loss": 0.0003312036395072937,
      "step": 218740
    },
    {
      "epoch": 62.09196707351689,
      "grad_norm": 0.017766226083040237,
      "learning_rate": 3.7933579335793366e-05,
      "loss": 0.0002023354172706604,
      "step": 218750
    },
    {
      "epoch": 62.09480556344025,
      "grad_norm": 2.9425907135009766,
      "learning_rate": 3.793074084587e-05,
      "loss": 0.0016248567029833795,
      "step": 218760
    },
    {
      "epoch": 62.09764405336361,
      "grad_norm": 0.007457878440618515,
      "learning_rate": 3.7927902355946635e-05,
      "loss": 0.0004856141284108162,
      "step": 218770
    },
    {
      "epoch": 62.10048254328697,
      "grad_norm": 0.9341108202934265,
      "learning_rate": 3.7925063866023276e-05,
      "loss": 0.0009186632931232452,
      "step": 218780
    },
    {
      "epoch": 62.103321033210335,
      "grad_norm": 0.020968137308955193,
      "learning_rate": 3.792222537609992e-05,
      "loss": 0.0001819279044866562,
      "step": 218790
    },
    {
      "epoch": 62.10615952313369,
      "grad_norm": 0.8765019178390503,
      "learning_rate": 3.791938688617656e-05,
      "loss": 0.0002856217324733734,
      "step": 218800
    },
    {
      "epoch": 62.108998013057054,
      "grad_norm": 0.1387549340724945,
      "learning_rate": 3.7916548396253193e-05,
      "loss": 0.0037482578307390215,
      "step": 218810
    },
    {
      "epoch": 62.11183650298042,
      "grad_norm": 0.058692507445812225,
      "learning_rate": 3.7913709906329835e-05,
      "loss": 0.0005689345300197602,
      "step": 218820
    },
    {
      "epoch": 62.11467499290377,
      "grad_norm": 0.054790280759334564,
      "learning_rate": 3.7910871416406476e-05,
      "loss": 0.0002366473898291588,
      "step": 218830
    },
    {
      "epoch": 62.117513482827135,
      "grad_norm": 0.004854176193475723,
      "learning_rate": 3.790803292648311e-05,
      "loss": 0.00016262289136648178,
      "step": 218840
    },
    {
      "epoch": 62.1203519727505,
      "grad_norm": 0.01813781075179577,
      "learning_rate": 3.790519443655975e-05,
      "loss": 0.00011467225849628448,
      "step": 218850
    },
    {
      "epoch": 62.123190462673854,
      "grad_norm": 0.13857761025428772,
      "learning_rate": 3.7902355946636394e-05,
      "loss": 0.0004955129697918892,
      "step": 218860
    },
    {
      "epoch": 62.12602895259722,
      "grad_norm": 0.031174857169389725,
      "learning_rate": 3.789951745671303e-05,
      "loss": 0.00036786533892154696,
      "step": 218870
    },
    {
      "epoch": 62.12886744252058,
      "grad_norm": 0.03192077949643135,
      "learning_rate": 3.789667896678967e-05,
      "loss": 0.0005792295560240746,
      "step": 218880
    },
    {
      "epoch": 62.13170593244394,
      "grad_norm": 0.05309580639004707,
      "learning_rate": 3.789384047686631e-05,
      "loss": 0.0003389021381735802,
      "step": 218890
    },
    {
      "epoch": 62.1345444223673,
      "grad_norm": 0.012357905507087708,
      "learning_rate": 3.7891001986942946e-05,
      "loss": 0.0020979007706046103,
      "step": 218900
    },
    {
      "epoch": 62.13738291229066,
      "grad_norm": 0.06666938215494156,
      "learning_rate": 3.788816349701959e-05,
      "loss": 0.00020651239901781082,
      "step": 218910
    },
    {
      "epoch": 62.140221402214024,
      "grad_norm": 0.01389849092811346,
      "learning_rate": 3.788532500709623e-05,
      "loss": 0.00039290226995944975,
      "step": 218920
    },
    {
      "epoch": 62.14305989213738,
      "grad_norm": 0.01174547802656889,
      "learning_rate": 3.788248651717286e-05,
      "loss": 0.0001665867865085602,
      "step": 218930
    },
    {
      "epoch": 62.14589838206074,
      "grad_norm": 0.022854190319776535,
      "learning_rate": 3.7879648027249504e-05,
      "loss": 0.00037089120596647265,
      "step": 218940
    },
    {
      "epoch": 62.148736871984106,
      "grad_norm": 0.21299807727336884,
      "learning_rate": 3.7876809537326146e-05,
      "loss": 0.00032324548810720445,
      "step": 218950
    },
    {
      "epoch": 62.15157536190747,
      "grad_norm": 0.04487968608736992,
      "learning_rate": 3.787397104740279e-05,
      "loss": 0.000251220166683197,
      "step": 218960
    },
    {
      "epoch": 62.154413851830824,
      "grad_norm": 0.5494627356529236,
      "learning_rate": 3.787113255747942e-05,
      "loss": 0.00022986680269241334,
      "step": 218970
    },
    {
      "epoch": 62.15725234175419,
      "grad_norm": 0.019761867821216583,
      "learning_rate": 3.7868294067556056e-05,
      "loss": 0.0005991479381918907,
      "step": 218980
    },
    {
      "epoch": 62.16009083167755,
      "grad_norm": 0.06674439460039139,
      "learning_rate": 3.7865455577632704e-05,
      "loss": 0.0010180534794926644,
      "step": 218990
    },
    {
      "epoch": 62.162929321600906,
      "grad_norm": 0.018801558762788773,
      "learning_rate": 3.786261708770934e-05,
      "loss": 0.00019344966858625412,
      "step": 219000
    },
    {
      "epoch": 62.162929321600906,
      "eval_accuracy": 0.9799707509378776,
      "eval_loss": 0.07611135393381119,
      "eval_runtime": 32.1576,
      "eval_samples_per_second": 489.061,
      "eval_steps_per_second": 7.65,
      "step": 219000
    },
    {
      "epoch": 62.16576781152427,
      "grad_norm": 0.4616508185863495,
      "learning_rate": 3.785977859778598e-05,
      "loss": 0.0004517126828432083,
      "step": 219010
    },
    {
      "epoch": 62.16860630144763,
      "grad_norm": 0.4357989728450775,
      "learning_rate": 3.785694010786262e-05,
      "loss": 0.0007189519703388215,
      "step": 219020
    },
    {
      "epoch": 62.17144479137099,
      "grad_norm": 0.13999664783477783,
      "learning_rate": 3.7854101617939256e-05,
      "loss": 0.0009274106472730636,
      "step": 219030
    },
    {
      "epoch": 62.17428328129435,
      "grad_norm": 0.015074108727276325,
      "learning_rate": 3.78512631280159e-05,
      "loss": 0.000232531875371933,
      "step": 219040
    },
    {
      "epoch": 62.17712177121771,
      "grad_norm": 0.06282586604356766,
      "learning_rate": 3.784842463809254e-05,
      "loss": 0.0023205233737826347,
      "step": 219050
    },
    {
      "epoch": 62.179960261141076,
      "grad_norm": 0.07035356014966965,
      "learning_rate": 3.7845586148169174e-05,
      "loss": 0.0026645392179489137,
      "step": 219060
    },
    {
      "epoch": 62.18279875106443,
      "grad_norm": 0.07992282509803772,
      "learning_rate": 3.7842747658245815e-05,
      "loss": 0.0002038177102804184,
      "step": 219070
    },
    {
      "epoch": 62.185637240987795,
      "grad_norm": 0.01094980537891388,
      "learning_rate": 3.783990916832245e-05,
      "loss": 0.0007394351065158844,
      "step": 219080
    },
    {
      "epoch": 62.18847573091116,
      "grad_norm": 0.01801287941634655,
      "learning_rate": 3.78370706783991e-05,
      "loss": 0.006691852211952209,
      "step": 219090
    },
    {
      "epoch": 62.19131422083451,
      "grad_norm": 0.9573699831962585,
      "learning_rate": 3.783423218847573e-05,
      "loss": 0.0031136304140090944,
      "step": 219100
    },
    {
      "epoch": 62.194152710757876,
      "grad_norm": 0.0036239661276340485,
      "learning_rate": 3.783139369855237e-05,
      "loss": 0.0018950289115309716,
      "step": 219110
    },
    {
      "epoch": 62.19699120068124,
      "grad_norm": 0.03517189249396324,
      "learning_rate": 3.7828555208629015e-05,
      "loss": 0.0001710960641503334,
      "step": 219120
    },
    {
      "epoch": 62.199829690604595,
      "grad_norm": 0.01670032925903797,
      "learning_rate": 3.782571671870565e-05,
      "loss": 0.0008104411885142326,
      "step": 219130
    },
    {
      "epoch": 62.20266818052796,
      "grad_norm": 0.07388734072446823,
      "learning_rate": 3.782287822878229e-05,
      "loss": 0.0008274240419268609,
      "step": 219140
    },
    {
      "epoch": 62.20550667045132,
      "grad_norm": 0.04889446496963501,
      "learning_rate": 3.782003973885893e-05,
      "loss": 0.0004162942990660667,
      "step": 219150
    },
    {
      "epoch": 62.208345160374684,
      "grad_norm": 12.564270973205566,
      "learning_rate": 3.781720124893557e-05,
      "loss": 0.00367276668548584,
      "step": 219160
    },
    {
      "epoch": 62.21118365029804,
      "grad_norm": 0.2573581337928772,
      "learning_rate": 3.781436275901221e-05,
      "loss": 0.015656030178070067,
      "step": 219170
    },
    {
      "epoch": 62.2140221402214,
      "grad_norm": 3.346475124359131,
      "learning_rate": 3.781152426908884e-05,
      "loss": 0.001143277995288372,
      "step": 219180
    },
    {
      "epoch": 62.216860630144765,
      "grad_norm": 1.0355018377304077,
      "learning_rate": 3.7808685779165484e-05,
      "loss": 0.009125328809022903,
      "step": 219190
    },
    {
      "epoch": 62.21969912006812,
      "grad_norm": 8.296284675598145,
      "learning_rate": 3.7805847289242126e-05,
      "loss": 0.010005916655063628,
      "step": 219200
    },
    {
      "epoch": 62.222537609991484,
      "grad_norm": 1.6540566682815552,
      "learning_rate": 3.780300879931876e-05,
      "loss": 0.011241798847913742,
      "step": 219210
    },
    {
      "epoch": 62.22537609991485,
      "grad_norm": 0.7496623396873474,
      "learning_rate": 3.780017030939541e-05,
      "loss": 0.0006888814270496369,
      "step": 219220
    },
    {
      "epoch": 62.2282145898382,
      "grad_norm": 0.1709733009338379,
      "learning_rate": 3.779733181947204e-05,
      "loss": 0.005229627341032028,
      "step": 219230
    },
    {
      "epoch": 62.231053079761566,
      "grad_norm": 0.04142606630921364,
      "learning_rate": 3.779449332954868e-05,
      "loss": 0.0001645239070057869,
      "step": 219240
    },
    {
      "epoch": 62.23389156968493,
      "grad_norm": 0.005263874307274818,
      "learning_rate": 3.7791654839625326e-05,
      "loss": 0.00043667983263731005,
      "step": 219250
    },
    {
      "epoch": 62.23673005960829,
      "grad_norm": 0.3858121335506439,
      "learning_rate": 3.778881634970196e-05,
      "loss": 0.002494099922478199,
      "step": 219260
    },
    {
      "epoch": 62.23956854953165,
      "grad_norm": 6.464031219482422,
      "learning_rate": 3.77859778597786e-05,
      "loss": 0.00207147803157568,
      "step": 219270
    },
    {
      "epoch": 62.24240703945501,
      "grad_norm": 0.13995666801929474,
      "learning_rate": 3.7783139369855236e-05,
      "loss": 0.0005194349214434624,
      "step": 219280
    },
    {
      "epoch": 62.24524552937837,
      "grad_norm": 0.017478762194514275,
      "learning_rate": 3.778030087993188e-05,
      "loss": 0.0008244156837463379,
      "step": 219290
    },
    {
      "epoch": 62.24808401930173,
      "grad_norm": 0.13240978121757507,
      "learning_rate": 3.777746239000852e-05,
      "loss": 0.0010995114222168922,
      "step": 219300
    },
    {
      "epoch": 62.25092250922509,
      "grad_norm": 0.0076134842820465565,
      "learning_rate": 3.7774623900085154e-05,
      "loss": 0.000849650613963604,
      "step": 219310
    },
    {
      "epoch": 62.253760999148454,
      "grad_norm": 0.5563639998435974,
      "learning_rate": 3.7771785410161795e-05,
      "loss": 0.0005636958405375481,
      "step": 219320
    },
    {
      "epoch": 62.25659948907182,
      "grad_norm": 0.0216164942830801,
      "learning_rate": 3.7768946920238437e-05,
      "loss": 0.00012986231595277787,
      "step": 219330
    },
    {
      "epoch": 62.25943797899517,
      "grad_norm": 0.006106148008257151,
      "learning_rate": 3.776610843031507e-05,
      "loss": 0.0004336489364504814,
      "step": 219340
    },
    {
      "epoch": 62.262276468918536,
      "grad_norm": 0.028716305270791054,
      "learning_rate": 3.776326994039171e-05,
      "loss": 0.0004390360787510872,
      "step": 219350
    },
    {
      "epoch": 62.2651149588419,
      "grad_norm": 0.07563501596450806,
      "learning_rate": 3.7760431450468354e-05,
      "loss": 0.006640709191560745,
      "step": 219360
    },
    {
      "epoch": 62.267953448765255,
      "grad_norm": 0.05550495162606239,
      "learning_rate": 3.775759296054499e-05,
      "loss": 0.0004650451242923737,
      "step": 219370
    },
    {
      "epoch": 62.27079193868862,
      "grad_norm": 0.09874530881643295,
      "learning_rate": 3.775475447062164e-05,
      "loss": 0.0005641063675284385,
      "step": 219380
    },
    {
      "epoch": 62.27363042861198,
      "grad_norm": 0.017176497727632523,
      "learning_rate": 3.775191598069827e-05,
      "loss": 0.00022655315697193146,
      "step": 219390
    },
    {
      "epoch": 62.276468918535336,
      "grad_norm": 0.4507908523082733,
      "learning_rate": 3.7749077490774906e-05,
      "loss": 0.000202757865190506,
      "step": 219400
    },
    {
      "epoch": 62.2793074084587,
      "grad_norm": 0.11348843574523926,
      "learning_rate": 3.774623900085155e-05,
      "loss": 0.0002904435619711876,
      "step": 219410
    },
    {
      "epoch": 62.28214589838206,
      "grad_norm": 0.016872573643922806,
      "learning_rate": 3.774340051092819e-05,
      "loss": 0.0004588117823004723,
      "step": 219420
    },
    {
      "epoch": 62.284984388305425,
      "grad_norm": 0.07719072699546814,
      "learning_rate": 3.774056202100483e-05,
      "loss": 0.00019363835453987122,
      "step": 219430
    },
    {
      "epoch": 62.28782287822878,
      "grad_norm": 0.5255810022354126,
      "learning_rate": 3.7737723531081465e-05,
      "loss": 0.0011194759979844093,
      "step": 219440
    },
    {
      "epoch": 62.29066136815214,
      "grad_norm": 0.04389384388923645,
      "learning_rate": 3.7734885041158106e-05,
      "loss": 0.0002085333690047264,
      "step": 219450
    },
    {
      "epoch": 62.293499858075506,
      "grad_norm": 0.01597927138209343,
      "learning_rate": 3.773204655123475e-05,
      "loss": 0.000486685149371624,
      "step": 219460
    },
    {
      "epoch": 62.29633834799886,
      "grad_norm": 0.02351010963320732,
      "learning_rate": 3.772920806131138e-05,
      "loss": 0.0002326617017388344,
      "step": 219470
    },
    {
      "epoch": 62.299176837922225,
      "grad_norm": 0.14318598806858063,
      "learning_rate": 3.772636957138802e-05,
      "loss": 0.0002879161387681961,
      "step": 219480
    },
    {
      "epoch": 62.30201532784559,
      "grad_norm": 0.14352332055568695,
      "learning_rate": 3.7723531081464665e-05,
      "loss": 0.00012005586177110672,
      "step": 219490
    },
    {
      "epoch": 62.304853817768944,
      "grad_norm": 0.2028595209121704,
      "learning_rate": 3.77206925915413e-05,
      "loss": 0.00018038004636764527,
      "step": 219500
    },
    {
      "epoch": 62.304853817768944,
      "eval_accuracy": 0.978953392255357,
      "eval_loss": 0.07779113203287125,
      "eval_runtime": 32.6883,
      "eval_samples_per_second": 481.12,
      "eval_steps_per_second": 7.526,
      "step": 219500
    },
    {
      "epoch": 62.30769230769231,
      "grad_norm": 0.02011045068502426,
      "learning_rate": 3.771785410161794e-05,
      "loss": 0.00012589525431394577,
      "step": 219510
    },
    {
      "epoch": 62.31053079761567,
      "grad_norm": 0.010861233808100224,
      "learning_rate": 3.771501561169458e-05,
      "loss": 0.0005329407751560212,
      "step": 219520
    },
    {
      "epoch": 62.31336928753903,
      "grad_norm": 0.052278075367212296,
      "learning_rate": 3.7712177121771217e-05,
      "loss": 0.005591506510972977,
      "step": 219530
    },
    {
      "epoch": 62.31620777746239,
      "grad_norm": 0.18878480792045593,
      "learning_rate": 3.770933863184786e-05,
      "loss": 0.0003347795456647873,
      "step": 219540
    },
    {
      "epoch": 62.31904626738575,
      "grad_norm": 0.026020199060440063,
      "learning_rate": 3.77065001419245e-05,
      "loss": 0.000538238137960434,
      "step": 219550
    },
    {
      "epoch": 62.321884757309114,
      "grad_norm": 0.0202209260314703,
      "learning_rate": 3.770366165200114e-05,
      "loss": 0.00016019903123378753,
      "step": 219560
    },
    {
      "epoch": 62.32472324723247,
      "grad_norm": 0.06138351187109947,
      "learning_rate": 3.7700823162077775e-05,
      "loss": 0.00035441704094409945,
      "step": 219570
    },
    {
      "epoch": 62.32756173715583,
      "grad_norm": 0.06058640033006668,
      "learning_rate": 3.769798467215442e-05,
      "loss": 0.00031191930174827575,
      "step": 219580
    },
    {
      "epoch": 62.330400227079195,
      "grad_norm": 0.20166485011577606,
      "learning_rate": 3.769514618223106e-05,
      "loss": 0.0019159005954861642,
      "step": 219590
    },
    {
      "epoch": 62.33323871700255,
      "grad_norm": 0.04015286639332771,
      "learning_rate": 3.769230769230769e-05,
      "loss": 0.0003723150119185448,
      "step": 219600
    },
    {
      "epoch": 62.336077206925914,
      "grad_norm": 23.481773376464844,
      "learning_rate": 3.7689469202384334e-05,
      "loss": 0.00763494074344635,
      "step": 219610
    },
    {
      "epoch": 62.33891569684928,
      "grad_norm": 0.01890701800584793,
      "learning_rate": 3.7686630712460975e-05,
      "loss": 0.00042074769735336306,
      "step": 219620
    },
    {
      "epoch": 62.34175418677264,
      "grad_norm": 0.08658425509929657,
      "learning_rate": 3.768379222253761e-05,
      "loss": 0.00028143078088760376,
      "step": 219630
    },
    {
      "epoch": 62.344592676695996,
      "grad_norm": 0.2765031158924103,
      "learning_rate": 3.768095373261425e-05,
      "loss": 0.00246813353151083,
      "step": 219640
    },
    {
      "epoch": 62.34743116661936,
      "grad_norm": 3.382819890975952,
      "learning_rate": 3.7678399091683225e-05,
      "loss": 0.016927482187747957,
      "step": 219650
    },
    {
      "epoch": 62.35026965654272,
      "grad_norm": 0.012410280294716358,
      "learning_rate": 3.7675560601759866e-05,
      "loss": 0.0012172715738415717,
      "step": 219660
    },
    {
      "epoch": 62.35310814646608,
      "grad_norm": 1.3713065385818481,
      "learning_rate": 3.767272211183651e-05,
      "loss": 0.0007400527596473694,
      "step": 219670
    },
    {
      "epoch": 62.35594663638944,
      "grad_norm": 22.14598274230957,
      "learning_rate": 3.766988362191314e-05,
      "loss": 0.013165083527565003,
      "step": 219680
    },
    {
      "epoch": 62.3587851263128,
      "grad_norm": 0.32524076104164124,
      "learning_rate": 3.766704513198978e-05,
      "loss": 0.00033500846475362776,
      "step": 219690
    },
    {
      "epoch": 62.36162361623616,
      "grad_norm": 0.006239022593945265,
      "learning_rate": 3.7664206642066425e-05,
      "loss": 0.007063467800617218,
      "step": 219700
    },
    {
      "epoch": 62.36446210615952,
      "grad_norm": 0.0684872567653656,
      "learning_rate": 3.766136815214306e-05,
      "loss": 0.0007958961650729179,
      "step": 219710
    },
    {
      "epoch": 62.367300596082885,
      "grad_norm": 0.03391709923744202,
      "learning_rate": 3.76585296622197e-05,
      "loss": 0.00037691351026296615,
      "step": 219720
    },
    {
      "epoch": 62.37013908600625,
      "grad_norm": 0.02030166983604431,
      "learning_rate": 3.765569117229634e-05,
      "loss": 0.00037884432822465897,
      "step": 219730
    },
    {
      "epoch": 62.3729775759296,
      "grad_norm": 0.09407872706651688,
      "learning_rate": 3.7652852682372977e-05,
      "loss": 0.0011985165998339652,
      "step": 219740
    },
    {
      "epoch": 62.375816065852966,
      "grad_norm": 0.3466847836971283,
      "learning_rate": 3.765001419244962e-05,
      "loss": 0.00045046042650938035,
      "step": 219750
    },
    {
      "epoch": 62.37865455577633,
      "grad_norm": 0.007372156251221895,
      "learning_rate": 3.764717570252626e-05,
      "loss": 0.00034062806516885755,
      "step": 219760
    },
    {
      "epoch": 62.381493045699685,
      "grad_norm": 0.015063219703733921,
      "learning_rate": 3.76443372126029e-05,
      "loss": 0.001486562006175518,
      "step": 219770
    },
    {
      "epoch": 62.38433153562305,
      "grad_norm": 0.015805484727025032,
      "learning_rate": 3.7641498722679535e-05,
      "loss": 0.00020647961646318435,
      "step": 219780
    },
    {
      "epoch": 62.38717002554641,
      "grad_norm": 0.11486828327178955,
      "learning_rate": 3.763866023275617e-05,
      "loss": 0.00048134420067071917,
      "step": 219790
    },
    {
      "epoch": 62.39000851546977,
      "grad_norm": 0.783648669719696,
      "learning_rate": 3.763582174283282e-05,
      "loss": 0.00043035224080085755,
      "step": 219800
    },
    {
      "epoch": 62.39284700539313,
      "grad_norm": 0.020256655290722847,
      "learning_rate": 3.763298325290945e-05,
      "loss": 0.0007004821673035622,
      "step": 219810
    },
    {
      "epoch": 62.39568549531649,
      "grad_norm": 0.16775614023208618,
      "learning_rate": 3.7630144762986094e-05,
      "loss": 0.0005523417145013809,
      "step": 219820
    },
    {
      "epoch": 62.398523985239855,
      "grad_norm": 0.04210258647799492,
      "learning_rate": 3.7627306273062735e-05,
      "loss": 0.000672232173383236,
      "step": 219830
    },
    {
      "epoch": 62.40136247516321,
      "grad_norm": 0.011634457856416702,
      "learning_rate": 3.762446778313937e-05,
      "loss": 0.00016869492828845977,
      "step": 219840
    },
    {
      "epoch": 62.404200965086574,
      "grad_norm": 0.0229943934828043,
      "learning_rate": 3.762162929321601e-05,
      "loss": 0.00037203337997198104,
      "step": 219850
    },
    {
      "epoch": 62.40703945500994,
      "grad_norm": 0.005593643523752689,
      "learning_rate": 3.761879080329265e-05,
      "loss": 0.0006641879677772522,
      "step": 219860
    },
    {
      "epoch": 62.40987794493329,
      "grad_norm": 0.03380071744322777,
      "learning_rate": 3.761595231336929e-05,
      "loss": 0.0013285016641020775,
      "step": 219870
    },
    {
      "epoch": 62.412716434856655,
      "grad_norm": 0.05653723329305649,
      "learning_rate": 3.761311382344593e-05,
      "loss": 0.00026184637099504473,
      "step": 219880
    },
    {
      "epoch": 62.41555492478002,
      "grad_norm": 0.12228763103485107,
      "learning_rate": 3.761027533352256e-05,
      "loss": 0.00034721381962299347,
      "step": 219890
    },
    {
      "epoch": 62.41839341470338,
      "grad_norm": 0.05563554912805557,
      "learning_rate": 3.760743684359921e-05,
      "loss": 0.0002539144828915596,
      "step": 219900
    },
    {
      "epoch": 62.42123190462674,
      "grad_norm": 0.007663853466510773,
      "learning_rate": 3.7604598353675846e-05,
      "loss": 0.0003175731748342514,
      "step": 219910
    },
    {
      "epoch": 62.4240703945501,
      "grad_norm": 0.06445333361625671,
      "learning_rate": 3.760175986375248e-05,
      "loss": 0.0005133796483278274,
      "step": 219920
    },
    {
      "epoch": 62.42690888447346,
      "grad_norm": 0.19025073945522308,
      "learning_rate": 3.759892137382913e-05,
      "loss": 0.00032522790133953093,
      "step": 219930
    },
    {
      "epoch": 62.42974737439682,
      "grad_norm": 0.011832769960165024,
      "learning_rate": 3.7596082883905763e-05,
      "loss": 0.0002409936860203743,
      "step": 219940
    },
    {
      "epoch": 62.43258586432018,
      "grad_norm": 0.39448243379592896,
      "learning_rate": 3.7593244393982405e-05,
      "loss": 0.0004256727173924446,
      "step": 219950
    },
    {
      "epoch": 62.435424354243544,
      "grad_norm": 0.04302350431680679,
      "learning_rate": 3.7590405904059046e-05,
      "loss": 0.00019156653434038162,
      "step": 219960
    },
    {
      "epoch": 62.4382628441669,
      "grad_norm": 0.02866055630147457,
      "learning_rate": 3.758756741413568e-05,
      "loss": 0.00047951415181159975,
      "step": 219970
    },
    {
      "epoch": 62.44110133409026,
      "grad_norm": 0.3037232458591461,
      "learning_rate": 3.758472892421232e-05,
      "loss": 0.0008851617574691772,
      "step": 219980
    },
    {
      "epoch": 62.443939824013626,
      "grad_norm": 0.2409208118915558,
      "learning_rate": 3.758189043428896e-05,
      "loss": 0.0007486550137400627,
      "step": 219990
    },
    {
      "epoch": 62.44677831393699,
      "grad_norm": 0.04556864872574806,
      "learning_rate": 3.75790519443656e-05,
      "loss": 0.00035252124071121215,
      "step": 220000
    },
    {
      "epoch": 62.44677831393699,
      "eval_accuracy": 0.9761556558784257,
      "eval_loss": 0.09264658391475677,
      "eval_runtime": 32.2571,
      "eval_samples_per_second": 487.551,
      "eval_steps_per_second": 7.626,
      "step": 220000
    },
    {
      "epoch": 62.449616803860344,
      "grad_norm": 0.5532296299934387,
      "learning_rate": 3.757621345444224e-05,
      "loss": 0.0020470358431339264,
      "step": 220010
    },
    {
      "epoch": 62.45245529378371,
      "grad_norm": 0.2950701117515564,
      "learning_rate": 3.7573374964518874e-05,
      "loss": 0.0008316058665513992,
      "step": 220020
    },
    {
      "epoch": 62.45529378370707,
      "grad_norm": 0.045538172125816345,
      "learning_rate": 3.757053647459552e-05,
      "loss": 0.0003891659900546074,
      "step": 220030
    },
    {
      "epoch": 62.458132273630426,
      "grad_norm": 0.03646159917116165,
      "learning_rate": 3.756769798467216e-05,
      "loss": 0.0021652096882462503,
      "step": 220040
    },
    {
      "epoch": 62.46097076355379,
      "grad_norm": 0.13905228674411774,
      "learning_rate": 3.756485949474879e-05,
      "loss": 0.00118691585958004,
      "step": 220050
    },
    {
      "epoch": 62.46380925347715,
      "grad_norm": 0.018202180042862892,
      "learning_rate": 3.756202100482544e-05,
      "loss": 0.0005172975361347198,
      "step": 220060
    },
    {
      "epoch": 62.46664774340051,
      "grad_norm": 0.12948396801948547,
      "learning_rate": 3.7559182514902074e-05,
      "loss": 0.00012460928410291673,
      "step": 220070
    },
    {
      "epoch": 62.46948623332387,
      "grad_norm": 0.02978682890534401,
      "learning_rate": 3.7556344024978716e-05,
      "loss": 0.0007194764912128448,
      "step": 220080
    },
    {
      "epoch": 62.47232472324723,
      "grad_norm": 0.1586718112230301,
      "learning_rate": 3.755350553505535e-05,
      "loss": 0.008529847115278244,
      "step": 220090
    },
    {
      "epoch": 62.475163213170596,
      "grad_norm": 0.028984202072024345,
      "learning_rate": 3.755066704513199e-05,
      "loss": 0.0005697222426533699,
      "step": 220100
    },
    {
      "epoch": 62.47800170309395,
      "grad_norm": 0.08513357490301132,
      "learning_rate": 3.754782855520863e-05,
      "loss": 0.0005765927955508233,
      "step": 220110
    },
    {
      "epoch": 62.480840193017315,
      "grad_norm": 0.057428762316703796,
      "learning_rate": 3.754499006528527e-05,
      "loss": 0.0028147926554083825,
      "step": 220120
    },
    {
      "epoch": 62.48367868294068,
      "grad_norm": 0.04622599110007286,
      "learning_rate": 3.754215157536191e-05,
      "loss": 0.0009506290778517723,
      "step": 220130
    },
    {
      "epoch": 62.48651717286403,
      "grad_norm": 8.83021068572998,
      "learning_rate": 3.753931308543855e-05,
      "loss": 0.002114705741405487,
      "step": 220140
    },
    {
      "epoch": 62.489355662787396,
      "grad_norm": 0.022999458014965057,
      "learning_rate": 3.7536474595515185e-05,
      "loss": 0.0015185102820396423,
      "step": 220150
    },
    {
      "epoch": 62.49219415271076,
      "grad_norm": 0.6429575085639954,
      "learning_rate": 3.7533636105591826e-05,
      "loss": 0.001180780865252018,
      "step": 220160
    },
    {
      "epoch": 62.49503264263412,
      "grad_norm": 0.05055927485227585,
      "learning_rate": 3.753079761566847e-05,
      "loss": 0.0002830306068062782,
      "step": 220170
    },
    {
      "epoch": 62.49787113255748,
      "grad_norm": 0.01604345254600048,
      "learning_rate": 3.75279591257451e-05,
      "loss": 0.00020285006612539292,
      "step": 220180
    },
    {
      "epoch": 62.50070962248084,
      "grad_norm": 1.6896758079528809,
      "learning_rate": 3.7525120635821744e-05,
      "loss": 0.0003705553710460663,
      "step": 220190
    },
    {
      "epoch": 62.503548112404204,
      "grad_norm": 0.00716337189078331,
      "learning_rate": 3.7522282145898385e-05,
      "loss": 0.00024059005081653596,
      "step": 220200
    },
    {
      "epoch": 62.50638660232756,
      "grad_norm": 0.06678484380245209,
      "learning_rate": 3.751944365597502e-05,
      "loss": 0.000671716034412384,
      "step": 220210
    },
    {
      "epoch": 62.50922509225092,
      "grad_norm": 0.09156182408332825,
      "learning_rate": 3.751660516605166e-05,
      "loss": 0.00020352285355329514,
      "step": 220220
    },
    {
      "epoch": 62.512063582174285,
      "grad_norm": 0.19403307139873505,
      "learning_rate": 3.75137666761283e-05,
      "loss": 0.01471591293811798,
      "step": 220230
    },
    {
      "epoch": 62.51490207209764,
      "grad_norm": 0.0022113036829978228,
      "learning_rate": 3.7510928186204944e-05,
      "loss": 0.00021448787301778794,
      "step": 220240
    },
    {
      "epoch": 62.517740562021004,
      "grad_norm": 0.04682810232043266,
      "learning_rate": 3.750808969628158e-05,
      "loss": 0.00014065522700548171,
      "step": 220250
    },
    {
      "epoch": 62.52057905194437,
      "grad_norm": 0.06576584279537201,
      "learning_rate": 3.750525120635822e-05,
      "loss": 0.0002667957916855812,
      "step": 220260
    },
    {
      "epoch": 62.52341754186773,
      "grad_norm": 0.02518932893872261,
      "learning_rate": 3.750241271643486e-05,
      "loss": 0.00015745870769023895,
      "step": 220270
    },
    {
      "epoch": 62.526256031791085,
      "grad_norm": 0.07633817195892334,
      "learning_rate": 3.7499574226511496e-05,
      "loss": 0.00020928401499986649,
      "step": 220280
    },
    {
      "epoch": 62.52909452171445,
      "grad_norm": 0.013097911141812801,
      "learning_rate": 3.749673573658814e-05,
      "loss": 0.00011100433766841888,
      "step": 220290
    },
    {
      "epoch": 62.53193301163781,
      "grad_norm": 0.02696506679058075,
      "learning_rate": 3.749389724666478e-05,
      "loss": 6.610751152038574e-05,
      "step": 220300
    },
    {
      "epoch": 62.53477150156117,
      "grad_norm": 0.04498637467622757,
      "learning_rate": 3.749105875674141e-05,
      "loss": 0.0006032867357134819,
      "step": 220310
    },
    {
      "epoch": 62.53760999148453,
      "grad_norm": 0.06372048705816269,
      "learning_rate": 3.7488220266818054e-05,
      "loss": 0.0005412934347987175,
      "step": 220320
    },
    {
      "epoch": 62.54044848140789,
      "grad_norm": 0.00394667312502861,
      "learning_rate": 3.7485381776894696e-05,
      "loss": 0.0013209940865635873,
      "step": 220330
    },
    {
      "epoch": 62.54328697133125,
      "grad_norm": 0.036725301295518875,
      "learning_rate": 3.748254328697133e-05,
      "loss": 0.0012402640655636787,
      "step": 220340
    },
    {
      "epoch": 62.54612546125461,
      "grad_norm": 0.15138983726501465,
      "learning_rate": 3.747970479704797e-05,
      "loss": 0.0002975022420287132,
      "step": 220350
    },
    {
      "epoch": 62.548963951177974,
      "grad_norm": 0.2635360360145569,
      "learning_rate": 3.747686630712461e-05,
      "loss": 0.0005023496225476265,
      "step": 220360
    },
    {
      "epoch": 62.55180244110134,
      "grad_norm": 0.047502949833869934,
      "learning_rate": 3.7474027817201254e-05,
      "loss": 0.0018935756757855414,
      "step": 220370
    },
    {
      "epoch": 62.55464093102469,
      "grad_norm": 0.7674239873886108,
      "learning_rate": 3.747118932727789e-05,
      "loss": 0.0016071878373622895,
      "step": 220380
    },
    {
      "epoch": 62.557479420948056,
      "grad_norm": 0.017517661675810814,
      "learning_rate": 3.7468350837354524e-05,
      "loss": 0.00019743293523788452,
      "step": 220390
    },
    {
      "epoch": 62.56031791087142,
      "grad_norm": 0.045989491045475006,
      "learning_rate": 3.746551234743117e-05,
      "loss": 0.0007046166807413101,
      "step": 220400
    },
    {
      "epoch": 62.563156400794774,
      "grad_norm": 0.018961738795042038,
      "learning_rate": 3.7462673857507806e-05,
      "loss": 0.0011204695329070091,
      "step": 220410
    },
    {
      "epoch": 62.56599489071814,
      "grad_norm": 0.23179659247398376,
      "learning_rate": 3.745983536758445e-05,
      "loss": 0.0003025852143764496,
      "step": 220420
    },
    {
      "epoch": 62.5688333806415,
      "grad_norm": 0.14665548503398895,
      "learning_rate": 3.745699687766109e-05,
      "loss": 0.00046889930963516234,
      "step": 220430
    },
    {
      "epoch": 62.571671870564856,
      "grad_norm": 0.7377319931983948,
      "learning_rate": 3.7454158387737724e-05,
      "loss": 0.0007603036239743233,
      "step": 220440
    },
    {
      "epoch": 62.57451036048822,
      "grad_norm": 0.19971708953380585,
      "learning_rate": 3.7451319897814365e-05,
      "loss": 0.0013747775927186012,
      "step": 220450
    },
    {
      "epoch": 62.57734885041158,
      "grad_norm": 0.16037562489509583,
      "learning_rate": 3.7448481407891006e-05,
      "loss": 0.0042459025979042055,
      "step": 220460
    },
    {
      "epoch": 62.580187340334945,
      "grad_norm": 5.44827938079834,
      "learning_rate": 3.744564291796764e-05,
      "loss": 0.001369336247444153,
      "step": 220470
    },
    {
      "epoch": 62.5830258302583,
      "grad_norm": 0.03660914674401283,
      "learning_rate": 3.744280442804428e-05,
      "loss": 0.0006473138928413391,
      "step": 220480
    },
    {
      "epoch": 62.58586432018166,
      "grad_norm": 0.16802188754081726,
      "learning_rate": 3.743996593812092e-05,
      "loss": 0.00017759446054697037,
      "step": 220490
    },
    {
      "epoch": 62.588702810105026,
      "grad_norm": 0.03436180576682091,
      "learning_rate": 3.7437127448197565e-05,
      "loss": 0.0004955517128109932,
      "step": 220500
    },
    {
      "epoch": 62.588702810105026,
      "eval_accuracy": 0.9804158453614803,
      "eval_loss": 0.07600770890712738,
      "eval_runtime": 32.8252,
      "eval_samples_per_second": 479.114,
      "eval_steps_per_second": 7.494,
      "step": 220500
    },
    {
      "epoch": 62.59154130002838,
      "grad_norm": 0.1225135400891304,
      "learning_rate": 3.74342889582742e-05,
      "loss": 0.0004224097356200218,
      "step": 220510
    },
    {
      "epoch": 62.594379789951745,
      "grad_norm": 0.04546502232551575,
      "learning_rate": 3.7431450468350834e-05,
      "loss": 0.001396300457417965,
      "step": 220520
    },
    {
      "epoch": 62.59721827987511,
      "grad_norm": 0.05635577440261841,
      "learning_rate": 3.742861197842748e-05,
      "loss": 0.001854581944644451,
      "step": 220530
    },
    {
      "epoch": 62.60005676979847,
      "grad_norm": 0.03737569600343704,
      "learning_rate": 3.742577348850412e-05,
      "loss": 0.00524551123380661,
      "step": 220540
    },
    {
      "epoch": 62.60289525972183,
      "grad_norm": 1.5329885482788086,
      "learning_rate": 3.742293499858076e-05,
      "loss": 0.000603405013680458,
      "step": 220550
    },
    {
      "epoch": 62.60573374964519,
      "grad_norm": 0.03986736759543419,
      "learning_rate": 3.74200965086574e-05,
      "loss": 0.0007313285022974014,
      "step": 220560
    },
    {
      "epoch": 62.60857223956855,
      "grad_norm": 0.3240727484226227,
      "learning_rate": 3.7417258018734035e-05,
      "loss": 0.0005974115803837777,
      "step": 220570
    },
    {
      "epoch": 62.61141072949191,
      "grad_norm": 0.03289702162146568,
      "learning_rate": 3.7414419528810676e-05,
      "loss": 0.0006734821945428848,
      "step": 220580
    },
    {
      "epoch": 62.61424921941527,
      "grad_norm": 0.09203065931797028,
      "learning_rate": 3.741158103888731e-05,
      "loss": 0.007814368605613709,
      "step": 220590
    },
    {
      "epoch": 62.617087709338634,
      "grad_norm": 0.07434184849262238,
      "learning_rate": 3.740874254896395e-05,
      "loss": 0.0015671016648411752,
      "step": 220600
    },
    {
      "epoch": 62.61992619926199,
      "grad_norm": 0.03134223818778992,
      "learning_rate": 3.740590405904059e-05,
      "loss": 0.0007429582998156548,
      "step": 220610
    },
    {
      "epoch": 62.62276468918535,
      "grad_norm": 0.1887994110584259,
      "learning_rate": 3.740306556911723e-05,
      "loss": 0.00022660382091999054,
      "step": 220620
    },
    {
      "epoch": 62.625603179108715,
      "grad_norm": 0.05799306556582451,
      "learning_rate": 3.740022707919387e-05,
      "loss": 0.0006130540743470192,
      "step": 220630
    },
    {
      "epoch": 62.62844166903208,
      "grad_norm": 0.01516047865152359,
      "learning_rate": 3.739738858927051e-05,
      "loss": 0.00103780347853899,
      "step": 220640
    },
    {
      "epoch": 62.631280158955434,
      "grad_norm": 0.9778589010238647,
      "learning_rate": 3.7394550099347145e-05,
      "loss": 0.0007133908569812774,
      "step": 220650
    },
    {
      "epoch": 62.6341186488788,
      "grad_norm": 0.9045343995094299,
      "learning_rate": 3.739171160942379e-05,
      "loss": 0.002260998263955116,
      "step": 220660
    },
    {
      "epoch": 62.63695713880216,
      "grad_norm": 0.48831111192703247,
      "learning_rate": 3.738887311950043e-05,
      "loss": 0.0003881271928548813,
      "step": 220670
    },
    {
      "epoch": 62.639795628725516,
      "grad_norm": 0.7465749382972717,
      "learning_rate": 3.738603462957706e-05,
      "loss": 0.00029463451355695727,
      "step": 220680
    },
    {
      "epoch": 62.64263411864888,
      "grad_norm": 0.06525035947561264,
      "learning_rate": 3.7383196139653704e-05,
      "loss": 0.00025153327733278273,
      "step": 220690
    },
    {
      "epoch": 62.64547260857224,
      "grad_norm": 0.2418093979358673,
      "learning_rate": 3.7380357649730345e-05,
      "loss": 0.000286782905459404,
      "step": 220700
    },
    {
      "epoch": 62.6483110984956,
      "grad_norm": 0.596157431602478,
      "learning_rate": 3.737751915980699e-05,
      "loss": 0.0022610699757933616,
      "step": 220710
    },
    {
      "epoch": 62.65114958841896,
      "grad_norm": 0.1989627331495285,
      "learning_rate": 3.737468066988362e-05,
      "loss": 0.0005165455862879753,
      "step": 220720
    },
    {
      "epoch": 62.65398807834232,
      "grad_norm": 0.21977877616882324,
      "learning_rate": 3.737184217996026e-05,
      "loss": 0.000334666483104229,
      "step": 220730
    },
    {
      "epoch": 62.656826568265686,
      "grad_norm": 0.016593199223279953,
      "learning_rate": 3.7369003690036904e-05,
      "loss": 0.0005463510751724243,
      "step": 220740
    },
    {
      "epoch": 62.65966505818904,
      "grad_norm": 2.721832513809204,
      "learning_rate": 3.736616520011354e-05,
      "loss": 0.0008658876642584801,
      "step": 220750
    },
    {
      "epoch": 62.662503548112404,
      "grad_norm": 0.2064489722251892,
      "learning_rate": 3.736332671019018e-05,
      "loss": 0.0004147423431277275,
      "step": 220760
    },
    {
      "epoch": 62.66534203803577,
      "grad_norm": 0.07577303797006607,
      "learning_rate": 3.736048822026682e-05,
      "loss": 0.0003905940800905228,
      "step": 220770
    },
    {
      "epoch": 62.66818052795912,
      "grad_norm": 3.1362242698669434,
      "learning_rate": 3.7357649730343456e-05,
      "loss": 0.001255347765982151,
      "step": 220780
    },
    {
      "epoch": 62.671019017882486,
      "grad_norm": 0.08183205127716064,
      "learning_rate": 3.73548112404201e-05,
      "loss": 0.0004605952650308609,
      "step": 220790
    },
    {
      "epoch": 62.67385750780585,
      "grad_norm": 0.9915308952331543,
      "learning_rate": 3.735197275049674e-05,
      "loss": 0.00033500026911497115,
      "step": 220800
    },
    {
      "epoch": 62.676695997729205,
      "grad_norm": 0.09687336534261703,
      "learning_rate": 3.734913426057337e-05,
      "loss": 0.00018446575850248337,
      "step": 220810
    },
    {
      "epoch": 62.67953448765257,
      "grad_norm": 0.010253356769680977,
      "learning_rate": 3.7346295770650015e-05,
      "loss": 0.00047416072338819505,
      "step": 220820
    },
    {
      "epoch": 62.68237297757593,
      "grad_norm": 0.047340117394924164,
      "learning_rate": 3.7343457280726656e-05,
      "loss": 0.0017058374360203743,
      "step": 220830
    },
    {
      "epoch": 62.68521146749929,
      "grad_norm": 0.5831284523010254,
      "learning_rate": 3.73406187908033e-05,
      "loss": 0.00045192502439022063,
      "step": 220840
    },
    {
      "epoch": 62.68804995742265,
      "grad_norm": 0.21998797357082367,
      "learning_rate": 3.733778030087993e-05,
      "loss": 0.0002714775502681732,
      "step": 220850
    },
    {
      "epoch": 62.69088844734601,
      "grad_norm": 0.048612214624881744,
      "learning_rate": 3.7334941810956573e-05,
      "loss": 0.0006042754277586937,
      "step": 220860
    },
    {
      "epoch": 62.693726937269375,
      "grad_norm": 0.03342023864388466,
      "learning_rate": 3.7332103321033215e-05,
      "loss": 0.0012373730540275573,
      "step": 220870
    },
    {
      "epoch": 62.69656542719273,
      "grad_norm": 0.02772674150764942,
      "learning_rate": 3.732926483110985e-05,
      "loss": 0.0004272120073437691,
      "step": 220880
    },
    {
      "epoch": 62.69940391711609,
      "grad_norm": 0.01515103131532669,
      "learning_rate": 3.732642634118649e-05,
      "loss": 0.00019667018204927444,
      "step": 220890
    },
    {
      "epoch": 62.702242407039456,
      "grad_norm": 0.24886518716812134,
      "learning_rate": 3.732358785126313e-05,
      "loss": 0.00013839714229106902,
      "step": 220900
    },
    {
      "epoch": 62.70508089696281,
      "grad_norm": 1.2524290084838867,
      "learning_rate": 3.732074936133977e-05,
      "loss": 0.0004319304600358009,
      "step": 220910
    },
    {
      "epoch": 62.707919386886175,
      "grad_norm": 0.008192543871700764,
      "learning_rate": 3.731791087141641e-05,
      "loss": 0.0009183449670672417,
      "step": 220920
    },
    {
      "epoch": 62.71075787680954,
      "grad_norm": 0.10828685760498047,
      "learning_rate": 3.731507238149305e-05,
      "loss": 0.004860294610261917,
      "step": 220930
    },
    {
      "epoch": 62.7135963667329,
      "grad_norm": 17.202125549316406,
      "learning_rate": 3.7312233891569684e-05,
      "loss": 0.005052722990512848,
      "step": 220940
    },
    {
      "epoch": 62.71643485665626,
      "grad_norm": 0.30226701498031616,
      "learning_rate": 3.7309395401646325e-05,
      "loss": 0.00030899811536073686,
      "step": 220950
    },
    {
      "epoch": 62.71927334657962,
      "grad_norm": 0.013504995964467525,
      "learning_rate": 3.730655691172297e-05,
      "loss": 0.00032200515270233156,
      "step": 220960
    },
    {
      "epoch": 62.72211183650298,
      "grad_norm": 0.01178754959255457,
      "learning_rate": 3.730371842179961e-05,
      "loss": 0.0010052049532532693,
      "step": 220970
    },
    {
      "epoch": 62.72495032642634,
      "grad_norm": 0.05268789455294609,
      "learning_rate": 3.730087993187624e-05,
      "loss": 0.0002615513280034065,
      "step": 220980
    },
    {
      "epoch": 62.7277888163497,
      "grad_norm": 0.014984403736889362,
      "learning_rate": 3.729804144195288e-05,
      "loss": 0.0003170343115925789,
      "step": 220990
    },
    {
      "epoch": 62.730627306273064,
      "grad_norm": 1.610073447227478,
      "learning_rate": 3.7295202952029526e-05,
      "loss": 0.0005050806328654289,
      "step": 221000
    },
    {
      "epoch": 62.730627306273064,
      "eval_accuracy": 0.9790805620906721,
      "eval_loss": 0.07736451923847198,
      "eval_runtime": 32.6568,
      "eval_samples_per_second": 481.584,
      "eval_steps_per_second": 7.533,
      "step": 221000
    },
    {
      "epoch": 62.73346579619643,
      "grad_norm": 0.029072754085063934,
      "learning_rate": 3.729236446210616e-05,
      "loss": 0.00037256386131048205,
      "step": 221010
    },
    {
      "epoch": 62.73630428611978,
      "grad_norm": 0.15091291069984436,
      "learning_rate": 3.72895259721828e-05,
      "loss": 0.0010726409032940864,
      "step": 221020
    },
    {
      "epoch": 62.739142776043145,
      "grad_norm": 0.01626225747168064,
      "learning_rate": 3.728668748225944e-05,
      "loss": 0.0005709551274776459,
      "step": 221030
    },
    {
      "epoch": 62.74198126596651,
      "grad_norm": 0.735421359539032,
      "learning_rate": 3.728384899233608e-05,
      "loss": 0.0005545029416680336,
      "step": 221040
    },
    {
      "epoch": 62.744819755889864,
      "grad_norm": 0.03571925684809685,
      "learning_rate": 3.728101050241272e-05,
      "loss": 0.0019419748336076737,
      "step": 221050
    },
    {
      "epoch": 62.74765824581323,
      "grad_norm": 0.029926322400569916,
      "learning_rate": 3.727817201248936e-05,
      "loss": 0.0021665140986442564,
      "step": 221060
    },
    {
      "epoch": 62.75049673573659,
      "grad_norm": 0.04182121157646179,
      "learning_rate": 3.7275333522565995e-05,
      "loss": 0.00050969198346138,
      "step": 221070
    },
    {
      "epoch": 62.753335225659946,
      "grad_norm": 0.07838650047779083,
      "learning_rate": 3.7272495032642636e-05,
      "loss": 0.005736139416694641,
      "step": 221080
    },
    {
      "epoch": 62.75617371558331,
      "grad_norm": 0.03903937339782715,
      "learning_rate": 3.726965654271928e-05,
      "loss": 0.0003226315602660179,
      "step": 221090
    },
    {
      "epoch": 62.75901220550667,
      "grad_norm": 0.12722247838974,
      "learning_rate": 3.726681805279591e-05,
      "loss": 0.00256072711199522,
      "step": 221100
    },
    {
      "epoch": 62.761850695430034,
      "grad_norm": 0.45306870341300964,
      "learning_rate": 3.7263979562872554e-05,
      "loss": 0.00023096241056919098,
      "step": 221110
    },
    {
      "epoch": 62.76468918535339,
      "grad_norm": 0.7984899878501892,
      "learning_rate": 3.726114107294919e-05,
      "loss": 0.00027173981070518496,
      "step": 221120
    },
    {
      "epoch": 62.76752767527675,
      "grad_norm": 0.05048202723264694,
      "learning_rate": 3.7258302583025836e-05,
      "loss": 0.00018072295933961867,
      "step": 221130
    },
    {
      "epoch": 62.770366165200116,
      "grad_norm": 0.02276439033448696,
      "learning_rate": 3.725546409310247e-05,
      "loss": 0.0002523284405469894,
      "step": 221140
    },
    {
      "epoch": 62.77320465512347,
      "grad_norm": 0.12624478340148926,
      "learning_rate": 3.7252625603179105e-05,
      "loss": 0.00025866702198982237,
      "step": 221150
    },
    {
      "epoch": 62.776043145046835,
      "grad_norm": 0.020636245608329773,
      "learning_rate": 3.7249787113255754e-05,
      "loss": 0.00036194305866956713,
      "step": 221160
    },
    {
      "epoch": 62.7788816349702,
      "grad_norm": 2.036579132080078,
      "learning_rate": 3.724694862333239e-05,
      "loss": 0.0008509108796715736,
      "step": 221170
    },
    {
      "epoch": 62.78172012489355,
      "grad_norm": 0.14460164308547974,
      "learning_rate": 3.724411013340903e-05,
      "loss": 0.00016202405095100403,
      "step": 221180
    },
    {
      "epoch": 62.784558614816916,
      "grad_norm": 0.10844504833221436,
      "learning_rate": 3.724127164348567e-05,
      "loss": 0.0032292120158672333,
      "step": 221190
    },
    {
      "epoch": 62.78739710474028,
      "grad_norm": 0.12857072055339813,
      "learning_rate": 3.7238433153562306e-05,
      "loss": 0.0003208884969353676,
      "step": 221200
    },
    {
      "epoch": 62.79023559466364,
      "grad_norm": 0.022208433598279953,
      "learning_rate": 3.723559466363895e-05,
      "loss": 9.976215660572052e-05,
      "step": 221210
    },
    {
      "epoch": 62.793074084587,
      "grad_norm": 0.03178482502698898,
      "learning_rate": 3.723275617371558e-05,
      "loss": 0.00018774624913930893,
      "step": 221220
    },
    {
      "epoch": 62.79591257451036,
      "grad_norm": 0.3308212459087372,
      "learning_rate": 3.722991768379222e-05,
      "loss": 0.0009816626086831092,
      "step": 221230
    },
    {
      "epoch": 62.79875106443372,
      "grad_norm": 0.012342538684606552,
      "learning_rate": 3.7227079193868864e-05,
      "loss": 0.0002688990905880928,
      "step": 221240
    },
    {
      "epoch": 62.80158955435708,
      "grad_norm": 0.048857398331165314,
      "learning_rate": 3.72242407039455e-05,
      "loss": 0.00040606539696455004,
      "step": 221250
    },
    {
      "epoch": 62.80442804428044,
      "grad_norm": 0.03173855319619179,
      "learning_rate": 3.722140221402215e-05,
      "loss": 0.0006343487650156022,
      "step": 221260
    },
    {
      "epoch": 62.807266534203805,
      "grad_norm": 0.14571121335029602,
      "learning_rate": 3.721856372409878e-05,
      "loss": 0.00032192151993513105,
      "step": 221270
    },
    {
      "epoch": 62.81010502412717,
      "grad_norm": 1.6813653707504272,
      "learning_rate": 3.7215725234175416e-05,
      "loss": 0.00040964093059301374,
      "step": 221280
    },
    {
      "epoch": 62.812943514050524,
      "grad_norm": 0.09138424694538116,
      "learning_rate": 3.7212886744252064e-05,
      "loss": 0.0002347376197576523,
      "step": 221290
    },
    {
      "epoch": 62.81578200397389,
      "grad_norm": 0.03747747838497162,
      "learning_rate": 3.72100482543287e-05,
      "loss": 0.00014528799802064895,
      "step": 221300
    },
    {
      "epoch": 62.81862049389725,
      "grad_norm": 0.017470844089984894,
      "learning_rate": 3.720720976440534e-05,
      "loss": 0.0003998352214694023,
      "step": 221310
    },
    {
      "epoch": 62.821458983820605,
      "grad_norm": 0.007343909703195095,
      "learning_rate": 3.7204371274481975e-05,
      "loss": 0.000683414377272129,
      "step": 221320
    },
    {
      "epoch": 62.82429747374397,
      "grad_norm": 0.04427691921591759,
      "learning_rate": 3.7201532784558616e-05,
      "loss": 0.00034426674246788023,
      "step": 221330
    },
    {
      "epoch": 62.82713596366733,
      "grad_norm": 0.0723504051566124,
      "learning_rate": 3.719869429463526e-05,
      "loss": 0.0002257300540804863,
      "step": 221340
    },
    {
      "epoch": 62.82997445359069,
      "grad_norm": 0.01600656844675541,
      "learning_rate": 3.719585580471189e-05,
      "loss": 0.0006023889407515525,
      "step": 221350
    },
    {
      "epoch": 62.83281294351405,
      "grad_norm": 0.023682599887251854,
      "learning_rate": 3.7193017314788534e-05,
      "loss": 0.0006419939920306206,
      "step": 221360
    },
    {
      "epoch": 62.83565143343741,
      "grad_norm": 0.012780014425516129,
      "learning_rate": 3.7190178824865175e-05,
      "loss": 0.0006440529599785805,
      "step": 221370
    },
    {
      "epoch": 62.838489923360775,
      "grad_norm": 0.002825644565746188,
      "learning_rate": 3.718734033494181e-05,
      "loss": 0.00020237155258655547,
      "step": 221380
    },
    {
      "epoch": 62.84132841328413,
      "grad_norm": 0.06286982446908951,
      "learning_rate": 3.718450184501846e-05,
      "loss": 0.00034492947161197663,
      "step": 221390
    },
    {
      "epoch": 62.844166903207494,
      "grad_norm": 0.0709298700094223,
      "learning_rate": 3.718166335509509e-05,
      "loss": 0.0003029840067028999,
      "step": 221400
    },
    {
      "epoch": 62.84700539313086,
      "grad_norm": 0.004451101645827293,
      "learning_rate": 3.717882486517173e-05,
      "loss": 7.170941680669784e-05,
      "step": 221410
    },
    {
      "epoch": 62.84984388305421,
      "grad_norm": 0.019794009625911713,
      "learning_rate": 3.717598637524837e-05,
      "loss": 0.00023233871906995772,
      "step": 221420
    },
    {
      "epoch": 62.852682372977576,
      "grad_norm": 0.2831297814846039,
      "learning_rate": 3.717314788532501e-05,
      "loss": 0.0018797386437654495,
      "step": 221430
    },
    {
      "epoch": 62.85552086290094,
      "grad_norm": 0.2952229976654053,
      "learning_rate": 3.717030939540165e-05,
      "loss": 0.0002957427874207497,
      "step": 221440
    },
    {
      "epoch": 62.858359352824294,
      "grad_norm": 0.07380744069814682,
      "learning_rate": 3.7167470905478286e-05,
      "loss": 0.0011314468458294869,
      "step": 221450
    },
    {
      "epoch": 62.86119784274766,
      "grad_norm": 0.20523954927921295,
      "learning_rate": 3.716463241555493e-05,
      "loss": 0.00029276981949806214,
      "step": 221460
    },
    {
      "epoch": 62.86403633267102,
      "grad_norm": 0.18051712214946747,
      "learning_rate": 3.716179392563157e-05,
      "loss": 0.0004997003823518753,
      "step": 221470
    },
    {
      "epoch": 62.86687482259438,
      "grad_norm": 0.20822861790657043,
      "learning_rate": 3.71589554357082e-05,
      "loss": 0.0005903499200940132,
      "step": 221480
    },
    {
      "epoch": 62.86971331251774,
      "grad_norm": 0.061056673526763916,
      "learning_rate": 3.7156116945784844e-05,
      "loss": 0.0006460011005401612,
      "step": 221490
    },
    {
      "epoch": 62.8725518024411,
      "grad_norm": 0.06393151730298996,
      "learning_rate": 3.7153278455861486e-05,
      "loss": 0.0003125712275505066,
      "step": 221500
    },
    {
      "epoch": 62.8725518024411,
      "eval_accuracy": 0.9773637693139188,
      "eval_loss": 0.0836043655872345,
      "eval_runtime": 33.1109,
      "eval_samples_per_second": 474.98,
      "eval_steps_per_second": 7.43,
      "step": 221500
    },
    {
      "epoch": 62.875390292364465,
      "grad_norm": 0.0933319702744484,
      "learning_rate": 3.715043996593812e-05,
      "loss": 0.00033258162438869475,
      "step": 221510
    },
    {
      "epoch": 62.87822878228782,
      "grad_norm": 0.5218175053596497,
      "learning_rate": 3.714760147601476e-05,
      "loss": 0.0005800547078251839,
      "step": 221520
    },
    {
      "epoch": 62.88106727221118,
      "grad_norm": 0.027496440336108208,
      "learning_rate": 3.71447629860914e-05,
      "loss": 0.0005764544010162353,
      "step": 221530
    },
    {
      "epoch": 62.883905762134546,
      "grad_norm": 0.16000771522521973,
      "learning_rate": 3.714192449616804e-05,
      "loss": 0.005874023586511612,
      "step": 221540
    },
    {
      "epoch": 62.8867442520579,
      "grad_norm": 0.1213483139872551,
      "learning_rate": 3.713908600624468e-05,
      "loss": 0.0005395777523517608,
      "step": 221550
    },
    {
      "epoch": 62.889582741981265,
      "grad_norm": 0.0938689187169075,
      "learning_rate": 3.713624751632132e-05,
      "loss": 0.003927844017744065,
      "step": 221560
    },
    {
      "epoch": 62.89242123190463,
      "grad_norm": 0.03203064948320389,
      "learning_rate": 3.7133409026397955e-05,
      "loss": 0.004167410731315613,
      "step": 221570
    },
    {
      "epoch": 62.89525972182799,
      "grad_norm": 0.10928358137607574,
      "learning_rate": 3.7130570536474597e-05,
      "loss": 0.0003794051706790924,
      "step": 221580
    },
    {
      "epoch": 62.898098211751346,
      "grad_norm": 0.34167349338531494,
      "learning_rate": 3.712773204655124e-05,
      "loss": 0.0011486953124403953,
      "step": 221590
    },
    {
      "epoch": 62.90093670167471,
      "grad_norm": 0.2215585559606552,
      "learning_rate": 3.712489355662788e-05,
      "loss": 0.0002844205126166344,
      "step": 221600
    },
    {
      "epoch": 62.90377519159807,
      "grad_norm": 0.037220295518636703,
      "learning_rate": 3.7122055066704514e-05,
      "loss": 0.00016481857746839524,
      "step": 221610
    },
    {
      "epoch": 62.90661368152143,
      "grad_norm": 0.18866892158985138,
      "learning_rate": 3.711921657678115e-05,
      "loss": 0.00014819875359535218,
      "step": 221620
    },
    {
      "epoch": 62.90945217144479,
      "grad_norm": 0.017261669039726257,
      "learning_rate": 3.71163780868578e-05,
      "loss": 0.00011001136153936386,
      "step": 221630
    },
    {
      "epoch": 62.912290661368154,
      "grad_norm": 0.0674494057893753,
      "learning_rate": 3.711353959693443e-05,
      "loss": 0.00021523162722587587,
      "step": 221640
    },
    {
      "epoch": 62.91512915129151,
      "grad_norm": 0.13221725821495056,
      "learning_rate": 3.711070110701107e-05,
      "loss": 0.0022959308698773384,
      "step": 221650
    },
    {
      "epoch": 62.91796764121487,
      "grad_norm": 10.170283317565918,
      "learning_rate": 3.7107862617087714e-05,
      "loss": 0.002906334586441517,
      "step": 221660
    },
    {
      "epoch": 62.920806131138235,
      "grad_norm": 0.9249834418296814,
      "learning_rate": 3.710502412716435e-05,
      "loss": 0.00035566221922636034,
      "step": 221670
    },
    {
      "epoch": 62.9236446210616,
      "grad_norm": 0.06344310939311981,
      "learning_rate": 3.710218563724099e-05,
      "loss": 0.000892285630106926,
      "step": 221680
    },
    {
      "epoch": 62.926483110984954,
      "grad_norm": 0.014313209801912308,
      "learning_rate": 3.709934714731763e-05,
      "loss": 0.012858664989471436,
      "step": 221690
    },
    {
      "epoch": 62.92932160090832,
      "grad_norm": 0.036715563386678696,
      "learning_rate": 3.7096508657394266e-05,
      "loss": 0.0012296322733163834,
      "step": 221700
    },
    {
      "epoch": 62.93216009083168,
      "grad_norm": 1.1415212154388428,
      "learning_rate": 3.709367016747091e-05,
      "loss": 0.0004183949902653694,
      "step": 221710
    },
    {
      "epoch": 62.934998580755035,
      "grad_norm": 0.7998591661453247,
      "learning_rate": 3.709083167754754e-05,
      "loss": 0.0004766814410686493,
      "step": 221720
    },
    {
      "epoch": 62.9378370706784,
      "grad_norm": 0.13821984827518463,
      "learning_rate": 3.708799318762419e-05,
      "loss": 0.0003136558458209038,
      "step": 221730
    },
    {
      "epoch": 62.94067556060176,
      "grad_norm": 0.008431218564510345,
      "learning_rate": 3.7085154697700825e-05,
      "loss": 0.003783559799194336,
      "step": 221740
    },
    {
      "epoch": 62.943514050525124,
      "grad_norm": 0.03993619233369827,
      "learning_rate": 3.708231620777746e-05,
      "loss": 0.00023132599890232086,
      "step": 221750
    },
    {
      "epoch": 62.94635254044848,
      "grad_norm": 0.08023405820131302,
      "learning_rate": 3.707947771785411e-05,
      "loss": 0.0002563139423727989,
      "step": 221760
    },
    {
      "epoch": 62.94919103037184,
      "grad_norm": 0.002759313676506281,
      "learning_rate": 3.707663922793074e-05,
      "loss": 0.00020506344735622407,
      "step": 221770
    },
    {
      "epoch": 62.952029520295206,
      "grad_norm": 0.659589946269989,
      "learning_rate": 3.707380073800738e-05,
      "loss": 0.00026085879653692245,
      "step": 221780
    },
    {
      "epoch": 62.95486801021856,
      "grad_norm": 0.016593962907791138,
      "learning_rate": 3.7070962248084025e-05,
      "loss": 0.0005251606926321983,
      "step": 221790
    },
    {
      "epoch": 62.957706500141924,
      "grad_norm": 0.029335135594010353,
      "learning_rate": 3.706812375816066e-05,
      "loss": 0.00024507790803909304,
      "step": 221800
    },
    {
      "epoch": 62.96054499006529,
      "grad_norm": 0.016252443194389343,
      "learning_rate": 3.70652852682373e-05,
      "loss": 0.00018752869218587875,
      "step": 221810
    },
    {
      "epoch": 62.96338347998864,
      "grad_norm": 0.051973696798086166,
      "learning_rate": 3.7062446778313935e-05,
      "loss": 0.00037641599774360655,
      "step": 221820
    },
    {
      "epoch": 62.966221969912006,
      "grad_norm": 0.013584896922111511,
      "learning_rate": 3.705960828839058e-05,
      "loss": 0.00022341348230838775,
      "step": 221830
    },
    {
      "epoch": 62.96906045983537,
      "grad_norm": 0.021034209057688713,
      "learning_rate": 3.705676979846722e-05,
      "loss": 0.002510124258697033,
      "step": 221840
    },
    {
      "epoch": 62.97189894975873,
      "grad_norm": 0.24433168768882751,
      "learning_rate": 3.705393130854385e-05,
      "loss": 0.0003959193825721741,
      "step": 221850
    },
    {
      "epoch": 62.97473743968209,
      "grad_norm": 0.05004934221506119,
      "learning_rate": 3.70510928186205e-05,
      "loss": 0.00017364602535963058,
      "step": 221860
    },
    {
      "epoch": 62.97757592960545,
      "grad_norm": 0.3371748626232147,
      "learning_rate": 3.7048254328697135e-05,
      "loss": 0.00035803411155939103,
      "step": 221870
    },
    {
      "epoch": 62.98041441952881,
      "grad_norm": 0.09849247336387634,
      "learning_rate": 3.704541583877377e-05,
      "loss": 0.00011715590953826904,
      "step": 221880
    },
    {
      "epoch": 62.98325290945217,
      "grad_norm": 0.09758304059505463,
      "learning_rate": 3.704257734885042e-05,
      "loss": 0.00020075794309377671,
      "step": 221890
    },
    {
      "epoch": 62.98609139937553,
      "grad_norm": 5.555059432983398,
      "learning_rate": 3.703973885892705e-05,
      "loss": 0.0013607826083898544,
      "step": 221900
    },
    {
      "epoch": 62.988929889298895,
      "grad_norm": 0.02382068522274494,
      "learning_rate": 3.7036900369003694e-05,
      "loss": 0.0015381064265966415,
      "step": 221910
    },
    {
      "epoch": 62.99176837922225,
      "grad_norm": 0.03526509925723076,
      "learning_rate": 3.703406187908033e-05,
      "loss": 0.0019354859367012977,
      "step": 221920
    },
    {
      "epoch": 62.99460686914561,
      "grad_norm": 0.017935989424586296,
      "learning_rate": 3.703122338915697e-05,
      "loss": 0.0007024912163615226,
      "step": 221930
    },
    {
      "epoch": 62.997445359068976,
      "grad_norm": 0.508703351020813,
      "learning_rate": 3.702838489923361e-05,
      "loss": 0.0002306222915649414,
      "step": 221940
    },
    {
      "epoch": 63.00028384899234,
      "grad_norm": 0.09749090671539307,
      "learning_rate": 3.7025546409310246e-05,
      "loss": 0.0003968656063079834,
      "step": 221950
    },
    {
      "epoch": 63.003122338915695,
      "grad_norm": 0.11861522495746613,
      "learning_rate": 3.702270791938689e-05,
      "loss": 0.0009935002774000168,
      "step": 221960
    },
    {
      "epoch": 63.00596082883906,
      "grad_norm": 0.13088586926460266,
      "learning_rate": 3.701986942946353e-05,
      "loss": 0.0005473753437399865,
      "step": 221970
    },
    {
      "epoch": 63.00879931876242,
      "grad_norm": 14.756282806396484,
      "learning_rate": 3.7017030939540163e-05,
      "loss": 0.0015325598418712617,
      "step": 221980
    },
    {
      "epoch": 63.01163780868578,
      "grad_norm": 1.800063967704773,
      "learning_rate": 3.7014192449616805e-05,
      "loss": 0.0018998807296156882,
      "step": 221990
    },
    {
      "epoch": 63.01447629860914,
      "grad_norm": 0.3618490695953369,
      "learning_rate": 3.7011353959693446e-05,
      "loss": 0.0001423874869942665,
      "step": 222000
    },
    {
      "epoch": 63.01447629860914,
      "eval_accuracy": 0.9743752781840147,
      "eval_loss": 0.09624533355236053,
      "eval_runtime": 33.1445,
      "eval_samples_per_second": 474.498,
      "eval_steps_per_second": 7.422,
      "step": 222000
    },
    {
      "epoch": 63.0173147885325,
      "grad_norm": 0.04066452383995056,
      "learning_rate": 3.700851546977008e-05,
      "loss": 0.0006798496469855309,
      "step": 222010
    },
    {
      "epoch": 63.02015327845586,
      "grad_norm": 0.05239201337099075,
      "learning_rate": 3.700567697984672e-05,
      "loss": 0.001550297997891903,
      "step": 222020
    },
    {
      "epoch": 63.02299176837922,
      "grad_norm": 0.03552217781543732,
      "learning_rate": 3.7002838489923364e-05,
      "loss": 0.002151862159371376,
      "step": 222030
    },
    {
      "epoch": 63.025830258302584,
      "grad_norm": 0.04686582833528519,
      "learning_rate": 3.7e-05,
      "loss": 0.00017016418278217315,
      "step": 222040
    },
    {
      "epoch": 63.02866874822595,
      "grad_norm": 0.20809105038642883,
      "learning_rate": 3.699716151007664e-05,
      "loss": 0.0003229530528187752,
      "step": 222050
    },
    {
      "epoch": 63.0315072381493,
      "grad_norm": 0.1375904530286789,
      "learning_rate": 3.699432302015328e-05,
      "loss": 0.003982384502887726,
      "step": 222060
    },
    {
      "epoch": 63.034345728072665,
      "grad_norm": 0.11510719358921051,
      "learning_rate": 3.699148453022992e-05,
      "loss": 0.00020192936062812805,
      "step": 222070
    },
    {
      "epoch": 63.03718421799603,
      "grad_norm": 0.022936230525374413,
      "learning_rate": 3.698864604030656e-05,
      "loss": 0.00023498386144638062,
      "step": 222080
    },
    {
      "epoch": 63.040022707919384,
      "grad_norm": 0.025212842971086502,
      "learning_rate": 3.69858075503832e-05,
      "loss": 0.00019252020865678787,
      "step": 222090
    },
    {
      "epoch": 63.04286119784275,
      "grad_norm": 0.0306193009018898,
      "learning_rate": 3.698296906045984e-05,
      "loss": 0.00015121083706617355,
      "step": 222100
    },
    {
      "epoch": 63.04569968776611,
      "grad_norm": 0.029223550111055374,
      "learning_rate": 3.6980130570536474e-05,
      "loss": 8.840765804052353e-05,
      "step": 222110
    },
    {
      "epoch": 63.04853817768947,
      "grad_norm": 0.0823156088590622,
      "learning_rate": 3.6977292080613116e-05,
      "loss": 0.0001844773069024086,
      "step": 222120
    },
    {
      "epoch": 63.05137666761283,
      "grad_norm": 0.010103429667651653,
      "learning_rate": 3.697445359068976e-05,
      "loss": 0.00011989008635282516,
      "step": 222130
    },
    {
      "epoch": 63.05421515753619,
      "grad_norm": 0.0016102748923003674,
      "learning_rate": 3.697161510076639e-05,
      "loss": 7.132887840270996e-05,
      "step": 222140
    },
    {
      "epoch": 63.057053647459554,
      "grad_norm": 0.006803066935390234,
      "learning_rate": 3.696877661084303e-05,
      "loss": 0.00019915085285902024,
      "step": 222150
    },
    {
      "epoch": 63.05989213738291,
      "grad_norm": 0.03678028658032417,
      "learning_rate": 3.6965938120919674e-05,
      "loss": 0.004399031400680542,
      "step": 222160
    },
    {
      "epoch": 63.06273062730627,
      "grad_norm": 0.27749061584472656,
      "learning_rate": 3.696309963099631e-05,
      "loss": 0.0027730124071240424,
      "step": 222170
    },
    {
      "epoch": 63.065569117229636,
      "grad_norm": 0.12102781981229782,
      "learning_rate": 3.696026114107295e-05,
      "loss": 0.00028396695852279664,
      "step": 222180
    },
    {
      "epoch": 63.06840760715299,
      "grad_norm": 0.22646041214466095,
      "learning_rate": 3.695742265114959e-05,
      "loss": 0.00043182596564292906,
      "step": 222190
    },
    {
      "epoch": 63.071246097076354,
      "grad_norm": 0.02128811739385128,
      "learning_rate": 3.695458416122623e-05,
      "loss": 0.0011990437284111976,
      "step": 222200
    },
    {
      "epoch": 63.07408458699972,
      "grad_norm": 2.4359209537506104,
      "learning_rate": 3.695174567130287e-05,
      "loss": 0.0004339456558227539,
      "step": 222210
    },
    {
      "epoch": 63.07692307692308,
      "grad_norm": 1.7531226873397827,
      "learning_rate": 3.69489071813795e-05,
      "loss": 0.0005901388823986054,
      "step": 222220
    },
    {
      "epoch": 63.079761566846436,
      "grad_norm": 0.0870654433965683,
      "learning_rate": 3.694606869145615e-05,
      "loss": 0.0003680797293782234,
      "step": 222230
    },
    {
      "epoch": 63.0826000567698,
      "grad_norm": 0.09533345699310303,
      "learning_rate": 3.6943230201532785e-05,
      "loss": 0.00015436746180057525,
      "step": 222240
    },
    {
      "epoch": 63.08543854669316,
      "grad_norm": 0.021106502041220665,
      "learning_rate": 3.6940391711609426e-05,
      "loss": 8.522514253854752e-05,
      "step": 222250
    },
    {
      "epoch": 63.08827703661652,
      "grad_norm": 0.03947500139474869,
      "learning_rate": 3.693755322168607e-05,
      "loss": 8.559450507164001e-05,
      "step": 222260
    },
    {
      "epoch": 63.09111552653988,
      "grad_norm": 0.01650817133486271,
      "learning_rate": 3.69347147317627e-05,
      "loss": 0.0001304119825363159,
      "step": 222270
    },
    {
      "epoch": 63.09395401646324,
      "grad_norm": 0.484923779964447,
      "learning_rate": 3.6931876241839344e-05,
      "loss": 0.0003352878615260124,
      "step": 222280
    },
    {
      "epoch": 63.0967925063866,
      "grad_norm": 0.4122466444969177,
      "learning_rate": 3.6929037751915985e-05,
      "loss": 0.00021409206092357636,
      "step": 222290
    },
    {
      "epoch": 63.09963099630996,
      "grad_norm": 0.11534779518842697,
      "learning_rate": 3.692619926199262e-05,
      "loss": 0.00029919333755970003,
      "step": 222300
    },
    {
      "epoch": 63.102469486233325,
      "grad_norm": 4.089741230010986,
      "learning_rate": 3.692336077206926e-05,
      "loss": 0.0005991507321596145,
      "step": 222310
    },
    {
      "epoch": 63.10530797615669,
      "grad_norm": 0.07785464078187943,
      "learning_rate": 3.69205222821459e-05,
      "loss": 0.0001516314223408699,
      "step": 222320
    },
    {
      "epoch": 63.10814646608004,
      "grad_norm": 0.014666670002043247,
      "learning_rate": 3.6917683792222544e-05,
      "loss": 0.0024389436468482016,
      "step": 222330
    },
    {
      "epoch": 63.110984956003406,
      "grad_norm": 0.007482943125069141,
      "learning_rate": 3.691484530229918e-05,
      "loss": 0.0005008239299058914,
      "step": 222340
    },
    {
      "epoch": 63.11382344592677,
      "grad_norm": 0.010378503240644932,
      "learning_rate": 3.691200681237581e-05,
      "loss": 0.00019440222531557082,
      "step": 222350
    },
    {
      "epoch": 63.116661935850125,
      "grad_norm": 0.15446656942367554,
      "learning_rate": 3.690916832245246e-05,
      "loss": 0.00037731193006038664,
      "step": 222360
    },
    {
      "epoch": 63.11950042577349,
      "grad_norm": 0.16951552033424377,
      "learning_rate": 3.6906329832529096e-05,
      "loss": 0.0002531193196773529,
      "step": 222370
    },
    {
      "epoch": 63.12233891569685,
      "grad_norm": 0.1492270678281784,
      "learning_rate": 3.690349134260574e-05,
      "loss": 0.00012338235974311828,
      "step": 222380
    },
    {
      "epoch": 63.12517740562021,
      "grad_norm": 0.014598102308809757,
      "learning_rate": 3.690065285268238e-05,
      "loss": 0.0002052679657936096,
      "step": 222390
    },
    {
      "epoch": 63.12801589554357,
      "grad_norm": 0.015792399644851685,
      "learning_rate": 3.689781436275901e-05,
      "loss": 0.0001588882878422737,
      "step": 222400
    },
    {
      "epoch": 63.13085438546693,
      "grad_norm": 0.12372888624668121,
      "learning_rate": 3.6894975872835654e-05,
      "loss": 0.00013833194971084594,
      "step": 222410
    },
    {
      "epoch": 63.133692875390295,
      "grad_norm": 0.017437366768717766,
      "learning_rate": 3.6892137382912296e-05,
      "loss": 0.00010833889245986938,
      "step": 222420
    },
    {
      "epoch": 63.13653136531365,
      "grad_norm": 0.03344358131289482,
      "learning_rate": 3.688929889298893e-05,
      "loss": 0.00022490303963422774,
      "step": 222430
    },
    {
      "epoch": 63.139369855237014,
      "grad_norm": 0.05987146869301796,
      "learning_rate": 3.688646040306557e-05,
      "loss": 0.00013126730918884278,
      "step": 222440
    },
    {
      "epoch": 63.14220834516038,
      "grad_norm": 0.01600753329694271,
      "learning_rate": 3.6883621913142206e-05,
      "loss": 0.00027775615453720095,
      "step": 222450
    },
    {
      "epoch": 63.14504683508373,
      "grad_norm": 0.044473182410001755,
      "learning_rate": 3.688078342321885e-05,
      "loss": 0.0005924517288804054,
      "step": 222460
    },
    {
      "epoch": 63.147885325007096,
      "grad_norm": 3.4597675800323486,
      "learning_rate": 3.687794493329549e-05,
      "loss": 0.0017128989100456237,
      "step": 222470
    },
    {
      "epoch": 63.15072381493046,
      "grad_norm": 0.040201932191848755,
      "learning_rate": 3.6875106443372124e-05,
      "loss": 0.001567579433321953,
      "step": 222480
    },
    {
      "epoch": 63.15356230485382,
      "grad_norm": 0.015649186447262764,
      "learning_rate": 3.687226795344877e-05,
      "loss": 0.0014405371621251106,
      "step": 222490
    },
    {
      "epoch": 63.15640079477718,
      "grad_norm": 0.046983592212200165,
      "learning_rate": 3.6869429463525406e-05,
      "loss": 0.0018902258947491647,
      "step": 222500
    },
    {
      "epoch": 63.15640079477718,
      "eval_accuracy": 0.9739301837604121,
      "eval_loss": 0.10085046291351318,
      "eval_runtime": 32.588,
      "eval_samples_per_second": 482.6,
      "eval_steps_per_second": 7.549,
      "step": 222500
    },
    {
      "epoch": 63.15923928470054,
      "grad_norm": 0.6243544220924377,
      "learning_rate": 3.686659097360204e-05,
      "loss": 0.0014897385612130166,
      "step": 222510
    },
    {
      "epoch": 63.1620777746239,
      "grad_norm": 0.5263832807540894,
      "learning_rate": 3.686375248367869e-05,
      "loss": 0.002801300399005413,
      "step": 222520
    },
    {
      "epoch": 63.16491626454726,
      "grad_norm": 0.07941048592329025,
      "learning_rate": 3.6860913993755324e-05,
      "loss": 0.0010701123625040053,
      "step": 222530
    },
    {
      "epoch": 63.16775475447062,
      "grad_norm": 0.014640825800597668,
      "learning_rate": 3.6858075503831965e-05,
      "loss": 0.0009040471166372299,
      "step": 222540
    },
    {
      "epoch": 63.170593244393984,
      "grad_norm": 0.34598857164382935,
      "learning_rate": 3.68552370139086e-05,
      "loss": 0.00030457302927970885,
      "step": 222550
    },
    {
      "epoch": 63.17343173431734,
      "grad_norm": 0.01571085676550865,
      "learning_rate": 3.685239852398524e-05,
      "loss": 0.0004549348726868629,
      "step": 222560
    },
    {
      "epoch": 63.1762702242407,
      "grad_norm": 0.1959463655948639,
      "learning_rate": 3.684956003406188e-05,
      "loss": 0.0003264063969254494,
      "step": 222570
    },
    {
      "epoch": 63.179108714164066,
      "grad_norm": 0.0394119992852211,
      "learning_rate": 3.684672154413852e-05,
      "loss": 0.00040846075862646105,
      "step": 222580
    },
    {
      "epoch": 63.18194720408743,
      "grad_norm": 0.21493591368198395,
      "learning_rate": 3.684388305421516e-05,
      "loss": 0.0007286196574568749,
      "step": 222590
    },
    {
      "epoch": 63.184785694010785,
      "grad_norm": 0.014978724531829357,
      "learning_rate": 3.68410445642918e-05,
      "loss": 0.0014208262786269189,
      "step": 222600
    },
    {
      "epoch": 63.18762418393415,
      "grad_norm": 0.07457029074430466,
      "learning_rate": 3.6838206074368435e-05,
      "loss": 0.00015747901052236556,
      "step": 222610
    },
    {
      "epoch": 63.19046267385751,
      "grad_norm": 0.08383733034133911,
      "learning_rate": 3.683536758444508e-05,
      "loss": 0.0006684379652142525,
      "step": 222620
    },
    {
      "epoch": 63.193301163780866,
      "grad_norm": 0.10786515474319458,
      "learning_rate": 3.683252909452172e-05,
      "loss": 0.000507449172437191,
      "step": 222630
    },
    {
      "epoch": 63.19613965370423,
      "grad_norm": 0.04150820150971413,
      "learning_rate": 3.682969060459835e-05,
      "loss": 0.005959928035736084,
      "step": 222640
    },
    {
      "epoch": 63.19897814362759,
      "grad_norm": 0.16716830432415009,
      "learning_rate": 3.682685211467499e-05,
      "loss": 0.0005399556830525398,
      "step": 222650
    },
    {
      "epoch": 63.20181663355095,
      "grad_norm": 0.06488361954689026,
      "learning_rate": 3.6824013624751635e-05,
      "loss": 0.00014287084341049194,
      "step": 222660
    },
    {
      "epoch": 63.20465512347431,
      "grad_norm": 0.03859921172261238,
      "learning_rate": 3.6821175134828276e-05,
      "loss": 0.0013493884354829789,
      "step": 222670
    },
    {
      "epoch": 63.20749361339767,
      "grad_norm": 0.08405287563800812,
      "learning_rate": 3.681833664490491e-05,
      "loss": 0.0005160909146070481,
      "step": 222680
    },
    {
      "epoch": 63.210332103321036,
      "grad_norm": 0.062039799988269806,
      "learning_rate": 3.681549815498155e-05,
      "loss": 0.00036007892340421676,
      "step": 222690
    },
    {
      "epoch": 63.21317059324439,
      "grad_norm": 0.16438180208206177,
      "learning_rate": 3.681265966505819e-05,
      "loss": 0.000617966242134571,
      "step": 222700
    },
    {
      "epoch": 63.216009083167755,
      "grad_norm": 0.5380982756614685,
      "learning_rate": 3.680982117513483e-05,
      "loss": 0.00027030911296606065,
      "step": 222710
    },
    {
      "epoch": 63.21884757309112,
      "grad_norm": 0.2200469970703125,
      "learning_rate": 3.680698268521147e-05,
      "loss": 0.0008650330826640129,
      "step": 222720
    },
    {
      "epoch": 63.221686063014474,
      "grad_norm": 0.019728846848011017,
      "learning_rate": 3.680414419528811e-05,
      "loss": 0.005314484611153603,
      "step": 222730
    },
    {
      "epoch": 63.22452455293784,
      "grad_norm": 2.4692025184631348,
      "learning_rate": 3.6801305705364745e-05,
      "loss": 0.0018581535667181015,
      "step": 222740
    },
    {
      "epoch": 63.2273630428612,
      "grad_norm": 0.039046015590429306,
      "learning_rate": 3.679846721544139e-05,
      "loss": 0.001369146816432476,
      "step": 222750
    },
    {
      "epoch": 63.230201532784555,
      "grad_norm": 0.043520327657461166,
      "learning_rate": 3.679562872551803e-05,
      "loss": 0.0004117781296372414,
      "step": 222760
    },
    {
      "epoch": 63.23304002270792,
      "grad_norm": 0.03483003005385399,
      "learning_rate": 3.679279023559466e-05,
      "loss": 0.002009091153740883,
      "step": 222770
    },
    {
      "epoch": 63.23587851263128,
      "grad_norm": 0.05678689107298851,
      "learning_rate": 3.6789951745671304e-05,
      "loss": 0.00043452270328998567,
      "step": 222780
    },
    {
      "epoch": 63.238717002554644,
      "grad_norm": 0.16596706211566925,
      "learning_rate": 3.6787113255747945e-05,
      "loss": 0.0009283162653446198,
      "step": 222790
    },
    {
      "epoch": 63.241555492478,
      "grad_norm": 0.039930060505867004,
      "learning_rate": 3.678427476582459e-05,
      "loss": 0.0003033565357327461,
      "step": 222800
    },
    {
      "epoch": 63.24439398240136,
      "grad_norm": 1.528653860092163,
      "learning_rate": 3.678143627590122e-05,
      "loss": 0.0011295134201645852,
      "step": 222810
    },
    {
      "epoch": 63.247232472324725,
      "grad_norm": 0.4009380340576172,
      "learning_rate": 3.677859778597786e-05,
      "loss": 0.0012015856802463532,
      "step": 222820
    },
    {
      "epoch": 63.25007096224808,
      "grad_norm": 0.29122796654701233,
      "learning_rate": 3.6775759296054504e-05,
      "loss": 0.0008800780400633812,
      "step": 222830
    },
    {
      "epoch": 63.252909452171444,
      "grad_norm": 0.29701393842697144,
      "learning_rate": 3.677292080613114e-05,
      "loss": 0.0006461620330810546,
      "step": 222840
    },
    {
      "epoch": 63.25574794209481,
      "grad_norm": 0.028278445824980736,
      "learning_rate": 3.677008231620778e-05,
      "loss": 0.0007592126727104187,
      "step": 222850
    },
    {
      "epoch": 63.25858643201816,
      "grad_norm": 0.044579699635505676,
      "learning_rate": 3.676724382628442e-05,
      "loss": 0.0014387547969818115,
      "step": 222860
    },
    {
      "epoch": 63.261424921941526,
      "grad_norm": 4.254112243652344,
      "learning_rate": 3.6764405336361056e-05,
      "loss": 0.0008183998987078666,
      "step": 222870
    },
    {
      "epoch": 63.26426341186489,
      "grad_norm": 0.35128235816955566,
      "learning_rate": 3.67615668464377e-05,
      "loss": 0.00154423788189888,
      "step": 222880
    },
    {
      "epoch": 63.26710190178825,
      "grad_norm": 0.01831040345132351,
      "learning_rate": 3.675872835651434e-05,
      "loss": 0.0011856276541948318,
      "step": 222890
    },
    {
      "epoch": 63.26994039171161,
      "grad_norm": 1.7920013666152954,
      "learning_rate": 3.675588986659097e-05,
      "loss": 0.00034234002232551574,
      "step": 222900
    },
    {
      "epoch": 63.27277888163497,
      "grad_norm": 0.03529277816414833,
      "learning_rate": 3.6753051376667615e-05,
      "loss": 0.000364505872130394,
      "step": 222910
    },
    {
      "epoch": 63.27561737155833,
      "grad_norm": 0.30966782569885254,
      "learning_rate": 3.6750212886744256e-05,
      "loss": 0.0014404036104679109,
      "step": 222920
    },
    {
      "epoch": 63.27845586148169,
      "grad_norm": 0.06424389779567719,
      "learning_rate": 3.674737439682089e-05,
      "loss": 0.00038585904985666275,
      "step": 222930
    },
    {
      "epoch": 63.28129435140505,
      "grad_norm": 0.03836692497134209,
      "learning_rate": 3.674453590689753e-05,
      "loss": 0.00038603637367486956,
      "step": 222940
    },
    {
      "epoch": 63.284132841328415,
      "grad_norm": 0.3858807682991028,
      "learning_rate": 3.674169741697417e-05,
      "loss": 0.000572367012500763,
      "step": 222950
    },
    {
      "epoch": 63.28697133125178,
      "grad_norm": 0.5881721377372742,
      "learning_rate": 3.6738858927050815e-05,
      "loss": 0.0007988136261701584,
      "step": 222960
    },
    {
      "epoch": 63.28980982117513,
      "grad_norm": 0.04903225228190422,
      "learning_rate": 3.673602043712745e-05,
      "loss": 0.0023206396028399466,
      "step": 222970
    },
    {
      "epoch": 63.292648311098496,
      "grad_norm": 0.46652910113334656,
      "learning_rate": 3.6733181947204084e-05,
      "loss": 0.0017977748066186905,
      "step": 222980
    },
    {
      "epoch": 63.29548680102186,
      "grad_norm": 0.021855583414435387,
      "learning_rate": 3.673034345728073e-05,
      "loss": 0.0006477676331996917,
      "step": 222990
    },
    {
      "epoch": 63.298325290945215,
      "grad_norm": 0.024461625143885612,
      "learning_rate": 3.672750496735737e-05,
      "loss": 0.00020331274718046188,
      "step": 223000
    },
    {
      "epoch": 63.298325290945215,
      "eval_accuracy": 0.9792077319259872,
      "eval_loss": 0.07978944480419159,
      "eval_runtime": 33.1666,
      "eval_samples_per_second": 474.181,
      "eval_steps_per_second": 7.417,
      "step": 223000
    },
    {
      "epoch": 63.30116378086858,
      "grad_norm": 0.05832643434405327,
      "learning_rate": 3.672466647743401e-05,
      "loss": 0.0014985736459493637,
      "step": 223010
    },
    {
      "epoch": 63.30400227079194,
      "grad_norm": 0.06378906965255737,
      "learning_rate": 3.672182798751065e-05,
      "loss": 0.0010667722672224045,
      "step": 223020
    },
    {
      "epoch": 63.306840760715296,
      "grad_norm": 1.0933951139450073,
      "learning_rate": 3.6718989497587284e-05,
      "loss": 0.0014186076819896697,
      "step": 223030
    },
    {
      "epoch": 63.30967925063866,
      "grad_norm": 0.9511200785636902,
      "learning_rate": 3.6716151007663926e-05,
      "loss": 0.001658061146736145,
      "step": 223040
    },
    {
      "epoch": 63.31251774056202,
      "grad_norm": 0.11717627197504044,
      "learning_rate": 3.671331251774056e-05,
      "loss": 0.0003696547821164131,
      "step": 223050
    },
    {
      "epoch": 63.315356230485385,
      "grad_norm": 0.3648156225681305,
      "learning_rate": 3.67104740278172e-05,
      "loss": 0.001051601395010948,
      "step": 223060
    },
    {
      "epoch": 63.31819472040874,
      "grad_norm": 0.05299392715096474,
      "learning_rate": 3.670763553789384e-05,
      "loss": 0.0008675988763570785,
      "step": 223070
    },
    {
      "epoch": 63.321033210332104,
      "grad_norm": 0.01848066784441471,
      "learning_rate": 3.670479704797048e-05,
      "loss": 0.0018120346590876578,
      "step": 223080
    },
    {
      "epoch": 63.32387170025547,
      "grad_norm": 0.26518502831459045,
      "learning_rate": 3.6701958558047126e-05,
      "loss": 0.0007869521155953408,
      "step": 223090
    },
    {
      "epoch": 63.32671019017882,
      "grad_norm": 0.011847668327391148,
      "learning_rate": 3.669912006812376e-05,
      "loss": 0.00030502546578645704,
      "step": 223100
    },
    {
      "epoch": 63.329548680102185,
      "grad_norm": 1.7225652933120728,
      "learning_rate": 3.6696281578200395e-05,
      "loss": 0.0005635194480419159,
      "step": 223110
    },
    {
      "epoch": 63.33238717002555,
      "grad_norm": 0.07520072907209396,
      "learning_rate": 3.669344308827704e-05,
      "loss": 0.0007388025522232055,
      "step": 223120
    },
    {
      "epoch": 63.335225659948904,
      "grad_norm": 0.018903842195868492,
      "learning_rate": 3.669060459835368e-05,
      "loss": 0.00024162307381629943,
      "step": 223130
    },
    {
      "epoch": 63.33806414987227,
      "grad_norm": 0.020696748048067093,
      "learning_rate": 3.668776610843032e-05,
      "loss": 0.0003941576927900314,
      "step": 223140
    },
    {
      "epoch": 63.34090263979563,
      "grad_norm": 0.11120280623435974,
      "learning_rate": 3.6684927618506954e-05,
      "loss": 0.000406407006084919,
      "step": 223150
    },
    {
      "epoch": 63.34374112971899,
      "grad_norm": 0.04610070586204529,
      "learning_rate": 3.6682089128583595e-05,
      "loss": 0.00027455557137727736,
      "step": 223160
    },
    {
      "epoch": 63.34657961964235,
      "grad_norm": 0.0141451982781291,
      "learning_rate": 3.6679250638660236e-05,
      "loss": 0.00012269150465726852,
      "step": 223170
    },
    {
      "epoch": 63.34941810956571,
      "grad_norm": 0.6683109402656555,
      "learning_rate": 3.667641214873687e-05,
      "loss": 0.0022162292152643204,
      "step": 223180
    },
    {
      "epoch": 63.352256599489074,
      "grad_norm": 7.353968620300293,
      "learning_rate": 3.667357365881351e-05,
      "loss": 0.003374944627285004,
      "step": 223190
    },
    {
      "epoch": 63.35509508941243,
      "grad_norm": 0.18984372913837433,
      "learning_rate": 3.6670735168890154e-05,
      "loss": 0.00028263144195079805,
      "step": 223200
    },
    {
      "epoch": 63.35793357933579,
      "grad_norm": 0.054205141961574554,
      "learning_rate": 3.666789667896679e-05,
      "loss": 0.0001990489661693573,
      "step": 223210
    },
    {
      "epoch": 63.360772069259156,
      "grad_norm": 0.13059395551681519,
      "learning_rate": 3.6665058189043436e-05,
      "loss": 0.0005018491297960282,
      "step": 223220
    },
    {
      "epoch": 63.36361055918251,
      "grad_norm": 0.015749845653772354,
      "learning_rate": 3.666221969912007e-05,
      "loss": 0.001197049394249916,
      "step": 223230
    },
    {
      "epoch": 63.366449049105874,
      "grad_norm": 0.12327847629785538,
      "learning_rate": 3.6659381209196706e-05,
      "loss": 0.0003326380625367165,
      "step": 223240
    },
    {
      "epoch": 63.36928753902924,
      "grad_norm": 1.1081219911575317,
      "learning_rate": 3.665654271927335e-05,
      "loss": 0.0010806074365973472,
      "step": 223250
    },
    {
      "epoch": 63.3721260289526,
      "grad_norm": 2.3641059398651123,
      "learning_rate": 3.665370422934999e-05,
      "loss": 0.0004905544221401215,
      "step": 223260
    },
    {
      "epoch": 63.374964518875956,
      "grad_norm": 0.15047495067119598,
      "learning_rate": 3.665086573942663e-05,
      "loss": 0.00026196129620075225,
      "step": 223270
    },
    {
      "epoch": 63.37780300879932,
      "grad_norm": 0.1565430611371994,
      "learning_rate": 3.6648027249503264e-05,
      "loss": 0.00047994628548622134,
      "step": 223280
    },
    {
      "epoch": 63.38064149872268,
      "grad_norm": 0.25629186630249023,
      "learning_rate": 3.6645188759579906e-05,
      "loss": 0.00038407668471336366,
      "step": 223290
    },
    {
      "epoch": 63.38347998864604,
      "grad_norm": 0.11215836554765701,
      "learning_rate": 3.664235026965655e-05,
      "loss": 0.0002856113016605377,
      "step": 223300
    },
    {
      "epoch": 63.3863184785694,
      "grad_norm": 0.07265391945838928,
      "learning_rate": 3.663951177973318e-05,
      "loss": 0.0005048206076025963,
      "step": 223310
    },
    {
      "epoch": 63.38915696849276,
      "grad_norm": 0.30489373207092285,
      "learning_rate": 3.663667328980982e-05,
      "loss": 0.0028773272410035134,
      "step": 223320
    },
    {
      "epoch": 63.391995458416126,
      "grad_norm": 0.0663236528635025,
      "learning_rate": 3.6633834799886464e-05,
      "loss": 0.0003373030573129654,
      "step": 223330
    },
    {
      "epoch": 63.39483394833948,
      "grad_norm": 0.5260716080665588,
      "learning_rate": 3.66309963099631e-05,
      "loss": 0.0003296591341495514,
      "step": 223340
    },
    {
      "epoch": 63.397672438262845,
      "grad_norm": 0.0467161126434803,
      "learning_rate": 3.662815782003974e-05,
      "loss": 0.00022893212735652924,
      "step": 223350
    },
    {
      "epoch": 63.40051092818621,
      "grad_norm": 0.017364367842674255,
      "learning_rate": 3.662531933011638e-05,
      "loss": 0.00010982230305671692,
      "step": 223360
    },
    {
      "epoch": 63.40334941810956,
      "grad_norm": 0.028566643595695496,
      "learning_rate": 3.6622480840193016e-05,
      "loss": 0.00024002380669116974,
      "step": 223370
    },
    {
      "epoch": 63.406187908032926,
      "grad_norm": 0.011106804944574833,
      "learning_rate": 3.661964235026966e-05,
      "loss": 0.00011408347636461258,
      "step": 223380
    },
    {
      "epoch": 63.40902639795629,
      "grad_norm": 0.03660682216286659,
      "learning_rate": 3.66168038603463e-05,
      "loss": 0.0005589857697486878,
      "step": 223390
    },
    {
      "epoch": 63.411864887879645,
      "grad_norm": 0.047657184302806854,
      "learning_rate": 3.6613965370422934e-05,
      "loss": 0.0005269348621368408,
      "step": 223400
    },
    {
      "epoch": 63.41470337780301,
      "grad_norm": 0.015334545634686947,
      "learning_rate": 3.6611126880499575e-05,
      "loss": 0.0003244917839765549,
      "step": 223410
    },
    {
      "epoch": 63.41754186772637,
      "grad_norm": 1.2991056442260742,
      "learning_rate": 3.6608288390576216e-05,
      "loss": 0.0007083222270011901,
      "step": 223420
    },
    {
      "epoch": 63.420380357649734,
      "grad_norm": 0.03295411542057991,
      "learning_rate": 3.660544990065286e-05,
      "loss": 0.000199834443628788,
      "step": 223430
    },
    {
      "epoch": 63.42321884757309,
      "grad_norm": 0.09537182748317719,
      "learning_rate": 3.660261141072949e-05,
      "loss": 0.00046051088720560073,
      "step": 223440
    },
    {
      "epoch": 63.42605733749645,
      "grad_norm": 0.14992915093898773,
      "learning_rate": 3.659977292080613e-05,
      "loss": 0.00046184565871953964,
      "step": 223450
    },
    {
      "epoch": 63.428895827419815,
      "grad_norm": 1.1201171875,
      "learning_rate": 3.6596934430882775e-05,
      "loss": 0.0005515260621905327,
      "step": 223460
    },
    {
      "epoch": 63.43173431734317,
      "grad_norm": 0.03545794636011124,
      "learning_rate": 3.659409594095941e-05,
      "loss": 0.00019689593464136123,
      "step": 223470
    },
    {
      "epoch": 63.434572807266534,
      "grad_norm": 0.07283949851989746,
      "learning_rate": 3.659125745103605e-05,
      "loss": 0.00022553279995918274,
      "step": 223480
    },
    {
      "epoch": 63.4374112971899,
      "grad_norm": 0.017604945227503777,
      "learning_rate": 3.658841896111269e-05,
      "loss": 0.00045907646417617796,
      "step": 223490
    },
    {
      "epoch": 63.44024978711325,
      "grad_norm": 0.42313745617866516,
      "learning_rate": 3.658558047118933e-05,
      "loss": 0.0010253774002194405,
      "step": 223500
    },
    {
      "epoch": 63.44024978711325,
      "eval_accuracy": 0.9783811279964393,
      "eval_loss": 0.08207947760820389,
      "eval_runtime": 32.6471,
      "eval_samples_per_second": 481.728,
      "eval_steps_per_second": 7.535,
      "step": 223500
    },
    {
      "epoch": 63.443088277036615,
      "grad_norm": 8.376821517944336,
      "learning_rate": 3.658274198126597e-05,
      "loss": 0.001646113581955433,
      "step": 223510
    },
    {
      "epoch": 63.44592676695998,
      "grad_norm": 0.04202789440751076,
      "learning_rate": 3.657990349134261e-05,
      "loss": 0.00015992391854524612,
      "step": 223520
    },
    {
      "epoch": 63.44876525688334,
      "grad_norm": 0.05442344769835472,
      "learning_rate": 3.6577065001419244e-05,
      "loss": 0.00017020665109157562,
      "step": 223530
    },
    {
      "epoch": 63.4516037468067,
      "grad_norm": 0.029028786346316338,
      "learning_rate": 3.6574226511495886e-05,
      "loss": 0.00011553820222616195,
      "step": 223540
    },
    {
      "epoch": 63.45444223673006,
      "grad_norm": 0.0681825801730156,
      "learning_rate": 3.657138802157252e-05,
      "loss": 0.0001388832926750183,
      "step": 223550
    },
    {
      "epoch": 63.45728072665342,
      "grad_norm": 0.028805602341890335,
      "learning_rate": 3.656854953164917e-05,
      "loss": 0.00012537594884634017,
      "step": 223560
    },
    {
      "epoch": 63.46011921657678,
      "grad_norm": 0.34578147530555725,
      "learning_rate": 3.65657110417258e-05,
      "loss": 0.00016492772847414016,
      "step": 223570
    },
    {
      "epoch": 63.46295770650014,
      "grad_norm": 0.037857603281736374,
      "learning_rate": 3.656287255180244e-05,
      "loss": 0.00011042803525924682,
      "step": 223580
    },
    {
      "epoch": 63.465796196423504,
      "grad_norm": 0.20112371444702148,
      "learning_rate": 3.6560034061879086e-05,
      "loss": 0.00026157274842262266,
      "step": 223590
    },
    {
      "epoch": 63.46863468634686,
      "grad_norm": 0.009926735423505306,
      "learning_rate": 3.655719557195572e-05,
      "loss": 0.00022343844175338746,
      "step": 223600
    },
    {
      "epoch": 63.47147317627022,
      "grad_norm": 1.3026700019836426,
      "learning_rate": 3.655435708203236e-05,
      "loss": 0.0002471527084708214,
      "step": 223610
    },
    {
      "epoch": 63.474311666193586,
      "grad_norm": 0.047060754150152206,
      "learning_rate": 3.6551518592109e-05,
      "loss": 0.0002538273110985756,
      "step": 223620
    },
    {
      "epoch": 63.47715015611695,
      "grad_norm": 0.052999403327703476,
      "learning_rate": 3.654868010218564e-05,
      "loss": 0.00015318337827920913,
      "step": 223630
    },
    {
      "epoch": 63.479988646040304,
      "grad_norm": 0.009487537667155266,
      "learning_rate": 3.654584161226228e-05,
      "loss": 0.00029899366199970245,
      "step": 223640
    },
    {
      "epoch": 63.48282713596367,
      "grad_norm": 0.19196276366710663,
      "learning_rate": 3.654300312233892e-05,
      "loss": 0.00016205664724111557,
      "step": 223650
    },
    {
      "epoch": 63.48566562588703,
      "grad_norm": 0.15439818799495697,
      "learning_rate": 3.6540164632415555e-05,
      "loss": 0.0002551918849349022,
      "step": 223660
    },
    {
      "epoch": 63.488504115810386,
      "grad_norm": 0.008861695416271687,
      "learning_rate": 3.6537326142492197e-05,
      "loss": 0.00017722360789775848,
      "step": 223670
    },
    {
      "epoch": 63.49134260573375,
      "grad_norm": 0.021694336086511612,
      "learning_rate": 3.653448765256883e-05,
      "loss": 0.00015702378004789352,
      "step": 223680
    },
    {
      "epoch": 63.49418109565711,
      "grad_norm": 0.027342243120074272,
      "learning_rate": 3.653164916264548e-05,
      "loss": 0.0002125358209013939,
      "step": 223690
    },
    {
      "epoch": 63.497019585580475,
      "grad_norm": 0.045691367238759995,
      "learning_rate": 3.6528810672722114e-05,
      "loss": 0.00037649646401405334,
      "step": 223700
    },
    {
      "epoch": 63.49985807550383,
      "grad_norm": 0.011465326882898808,
      "learning_rate": 3.652597218279875e-05,
      "loss": 0.0002548329532146454,
      "step": 223710
    },
    {
      "epoch": 63.50269656542719,
      "grad_norm": 0.2503974437713623,
      "learning_rate": 3.65231336928754e-05,
      "loss": 0.00043977871537208556,
      "step": 223720
    },
    {
      "epoch": 63.505535055350556,
      "grad_norm": 0.17101015150547028,
      "learning_rate": 3.652029520295203e-05,
      "loss": 0.0004071470350027084,
      "step": 223730
    },
    {
      "epoch": 63.50837354527391,
      "grad_norm": 0.22685126960277557,
      "learning_rate": 3.651745671302867e-05,
      "loss": 0.0002570757642388344,
      "step": 223740
    },
    {
      "epoch": 63.511212035197275,
      "grad_norm": 0.027577245607972145,
      "learning_rate": 3.6514618223105314e-05,
      "loss": 0.0005594300106167793,
      "step": 223750
    },
    {
      "epoch": 63.51405052512064,
      "grad_norm": 0.132013201713562,
      "learning_rate": 3.651206358217428e-05,
      "loss": 0.0026603545993566513,
      "step": 223760
    },
    {
      "epoch": 63.51688901504399,
      "grad_norm": 0.004958732053637505,
      "learning_rate": 3.650922509225093e-05,
      "loss": 0.0014467721804976464,
      "step": 223770
    },
    {
      "epoch": 63.519727504967356,
      "grad_norm": 3.806523084640503,
      "learning_rate": 3.650638660232756e-05,
      "loss": 0.0006628228351473808,
      "step": 223780
    },
    {
      "epoch": 63.52256599489072,
      "grad_norm": 0.02572963386774063,
      "learning_rate": 3.65035481124042e-05,
      "loss": 0.00012870635837316513,
      "step": 223790
    },
    {
      "epoch": 63.52540448481408,
      "grad_norm": 0.035188227891922,
      "learning_rate": 3.6500709622480846e-05,
      "loss": 9.522289037704468e-05,
      "step": 223800
    },
    {
      "epoch": 63.52824297473744,
      "grad_norm": 0.029137175530195236,
      "learning_rate": 3.649787113255748e-05,
      "loss": 0.000906759686768055,
      "step": 223810
    },
    {
      "epoch": 63.5310814646608,
      "grad_norm": 0.058693256229162216,
      "learning_rate": 3.649503264263412e-05,
      "loss": 0.0019776618108153345,
      "step": 223820
    },
    {
      "epoch": 63.533919954584164,
      "grad_norm": 4.727992057800293,
      "learning_rate": 3.6492194152710757e-05,
      "loss": 0.0017699170857667923,
      "step": 223830
    },
    {
      "epoch": 63.53675844450752,
      "grad_norm": 0.03395581617951393,
      "learning_rate": 3.64893556627874e-05,
      "loss": 0.0015294859185814857,
      "step": 223840
    },
    {
      "epoch": 63.53959693443088,
      "grad_norm": 0.004711443092674017,
      "learning_rate": 3.648651717286404e-05,
      "loss": 0.0014114465564489364,
      "step": 223850
    },
    {
      "epoch": 63.542435424354245,
      "grad_norm": 0.00806459505110979,
      "learning_rate": 3.6483678682940674e-05,
      "loss": 0.0003004983067512512,
      "step": 223860
    },
    {
      "epoch": 63.5452739142776,
      "grad_norm": 0.016837652772665024,
      "learning_rate": 3.6480840193017315e-05,
      "loss": 0.000626312755048275,
      "step": 223870
    },
    {
      "epoch": 63.548112404200964,
      "grad_norm": 0.07849552482366562,
      "learning_rate": 3.647800170309396e-05,
      "loss": 0.00037410594522953035,
      "step": 223880
    },
    {
      "epoch": 63.55095089412433,
      "grad_norm": 0.704382061958313,
      "learning_rate": 3.647516321317059e-05,
      "loss": 0.0023267984390258787,
      "step": 223890
    },
    {
      "epoch": 63.55378938404769,
      "grad_norm": 0.042670633643865585,
      "learning_rate": 3.647232472324724e-05,
      "loss": 0.0008226621896028519,
      "step": 223900
    },
    {
      "epoch": 63.556627873971046,
      "grad_norm": 0.01936558447778225,
      "learning_rate": 3.6469486233323874e-05,
      "loss": 0.000643080472946167,
      "step": 223910
    },
    {
      "epoch": 63.55946636389441,
      "grad_norm": 0.19851155579090118,
      "learning_rate": 3.646664774340051e-05,
      "loss": 0.00026398710906505585,
      "step": 223920
    },
    {
      "epoch": 63.56230485381777,
      "grad_norm": 0.06842292845249176,
      "learning_rate": 3.646380925347715e-05,
      "loss": 0.0014824427664279938,
      "step": 223930
    },
    {
      "epoch": 63.56514334374113,
      "grad_norm": 0.1126151904463768,
      "learning_rate": 3.646097076355379e-05,
      "loss": 0.00032440591603517533,
      "step": 223940
    },
    {
      "epoch": 63.56798183366449,
      "grad_norm": 0.015271972864866257,
      "learning_rate": 3.645813227363043e-05,
      "loss": 0.00023623406887054443,
      "step": 223950
    },
    {
      "epoch": 63.57082032358785,
      "grad_norm": 0.11235082149505615,
      "learning_rate": 3.645529378370707e-05,
      "loss": 0.0001907750964164734,
      "step": 223960
    },
    {
      "epoch": 63.57365881351121,
      "grad_norm": 0.8468502163887024,
      "learning_rate": 3.645245529378371e-05,
      "loss": 0.00027998797595500945,
      "step": 223970
    },
    {
      "epoch": 63.57649730343457,
      "grad_norm": 0.10770338773727417,
      "learning_rate": 3.644961680386035e-05,
      "loss": 0.00010765604674816132,
      "step": 223980
    },
    {
      "epoch": 63.579335793357934,
      "grad_norm": 0.20835553109645844,
      "learning_rate": 3.6446778313936985e-05,
      "loss": 0.0008030563592910766,
      "step": 223990
    },
    {
      "epoch": 63.5821742832813,
      "grad_norm": 1.3233823776245117,
      "learning_rate": 3.6443939824013626e-05,
      "loss": 0.0008538063615560531,
      "step": 224000
    },
    {
      "epoch": 63.5821742832813,
      "eval_accuracy": 0.9780632034081516,
      "eval_loss": 0.07864438742399216,
      "eval_runtime": 32.8511,
      "eval_samples_per_second": 478.736,
      "eval_steps_per_second": 7.488,
      "step": 224000
    },
    {
      "epoch": 63.58501277320465,
      "grad_norm": 1.145976185798645,
      "learning_rate": 3.644110133409027e-05,
      "loss": 0.0005381714552640914,
      "step": 224010
    },
    {
      "epoch": 63.587851263128016,
      "grad_norm": 0.08711237460374832,
      "learning_rate": 3.64382628441669e-05,
      "loss": 0.004125714302062988,
      "step": 224020
    },
    {
      "epoch": 63.59068975305138,
      "grad_norm": 2.6091971397399902,
      "learning_rate": 3.643542435424355e-05,
      "loss": 0.0013687342405319213,
      "step": 224030
    },
    {
      "epoch": 63.593528242974735,
      "grad_norm": 4.404272556304932,
      "learning_rate": 3.6432585864320185e-05,
      "loss": 0.0008155850693583489,
      "step": 224040
    },
    {
      "epoch": 63.5963667328981,
      "grad_norm": 0.04506436735391617,
      "learning_rate": 3.642974737439682e-05,
      "loss": 0.0017896765843033791,
      "step": 224050
    },
    {
      "epoch": 63.59920522282146,
      "grad_norm": 0.6834333539009094,
      "learning_rate": 3.642690888447346e-05,
      "loss": 0.0020935479551553726,
      "step": 224060
    },
    {
      "epoch": 63.602043712744816,
      "grad_norm": 0.037196241319179535,
      "learning_rate": 3.64240703945501e-05,
      "loss": 0.005231668800115585,
      "step": 224070
    },
    {
      "epoch": 63.60488220266818,
      "grad_norm": 0.5990810394287109,
      "learning_rate": 3.6421231904626743e-05,
      "loss": 0.0007414866238832473,
      "step": 224080
    },
    {
      "epoch": 63.60772069259154,
      "grad_norm": 0.01359111163765192,
      "learning_rate": 3.641839341470338e-05,
      "loss": 0.0005565857514739037,
      "step": 224090
    },
    {
      "epoch": 63.610559182514905,
      "grad_norm": 0.05971262976527214,
      "learning_rate": 3.641555492478002e-05,
      "loss": 0.0012302357703447343,
      "step": 224100
    },
    {
      "epoch": 63.61339767243826,
      "grad_norm": 0.8995712995529175,
      "learning_rate": 3.641271643485666e-05,
      "loss": 0.0019820192828774452,
      "step": 224110
    },
    {
      "epoch": 63.61623616236162,
      "grad_norm": 0.11152498424053192,
      "learning_rate": 3.6409877944933295e-05,
      "loss": 0.0012632576748728753,
      "step": 224120
    },
    {
      "epoch": 63.619074652284986,
      "grad_norm": 0.06649849563837051,
      "learning_rate": 3.640703945500994e-05,
      "loss": 0.007293985784053802,
      "step": 224130
    },
    {
      "epoch": 63.62191314220834,
      "grad_norm": 0.0432928241789341,
      "learning_rate": 3.640420096508658e-05,
      "loss": 0.0018134333193302154,
      "step": 224140
    },
    {
      "epoch": 63.624751632131705,
      "grad_norm": 0.5464154481887817,
      "learning_rate": 3.640136247516321e-05,
      "loss": 0.0006778333336114883,
      "step": 224150
    },
    {
      "epoch": 63.62759012205507,
      "grad_norm": 0.0472242571413517,
      "learning_rate": 3.6398523985239854e-05,
      "loss": 0.0007098762318491936,
      "step": 224160
    },
    {
      "epoch": 63.63042861197843,
      "grad_norm": 0.6130415201187134,
      "learning_rate": 3.6395685495316495e-05,
      "loss": 0.0002561876550316811,
      "step": 224170
    },
    {
      "epoch": 63.63326710190179,
      "grad_norm": 0.10148940235376358,
      "learning_rate": 3.639284700539313e-05,
      "loss": 0.0006665991619229317,
      "step": 224180
    },
    {
      "epoch": 63.63610559182515,
      "grad_norm": 0.4504221975803375,
      "learning_rate": 3.639000851546977e-05,
      "loss": 0.00023106206208467484,
      "step": 224190
    },
    {
      "epoch": 63.63894408174851,
      "grad_norm": 0.15223711729049683,
      "learning_rate": 3.638717002554641e-05,
      "loss": 0.0002560446038842201,
      "step": 224200
    },
    {
      "epoch": 63.64178257167187,
      "grad_norm": 0.09493300318717957,
      "learning_rate": 3.638433153562305e-05,
      "loss": 0.0017477802932262421,
      "step": 224210
    },
    {
      "epoch": 63.64462106159523,
      "grad_norm": 0.02578337863087654,
      "learning_rate": 3.638149304569969e-05,
      "loss": 0.0004847826436161995,
      "step": 224220
    },
    {
      "epoch": 63.647459551518594,
      "grad_norm": 0.017281465232372284,
      "learning_rate": 3.637865455577633e-05,
      "loss": 0.00016190093010663985,
      "step": 224230
    },
    {
      "epoch": 63.65029804144195,
      "grad_norm": 0.020872266963124275,
      "learning_rate": 3.637581606585297e-05,
      "loss": 0.0004323353990912437,
      "step": 224240
    },
    {
      "epoch": 63.65313653136531,
      "grad_norm": 0.013306882232427597,
      "learning_rate": 3.6372977575929606e-05,
      "loss": 0.0009837733581662179,
      "step": 224250
    },
    {
      "epoch": 63.655975021288675,
      "grad_norm": 0.021725419908761978,
      "learning_rate": 3.637013908600624e-05,
      "loss": 0.00024094954133033753,
      "step": 224260
    },
    {
      "epoch": 63.65881351121204,
      "grad_norm": 0.14972157776355743,
      "learning_rate": 3.636730059608289e-05,
      "loss": 0.004216253012418747,
      "step": 224270
    },
    {
      "epoch": 63.661652001135394,
      "grad_norm": 0.017985695973038673,
      "learning_rate": 3.6364462106159524e-05,
      "loss": 0.00021517854183912278,
      "step": 224280
    },
    {
      "epoch": 63.66449049105876,
      "grad_norm": 0.04004627838730812,
      "learning_rate": 3.6361623616236165e-05,
      "loss": 0.00022351332008838654,
      "step": 224290
    },
    {
      "epoch": 63.66732898098212,
      "grad_norm": 0.004322052001953125,
      "learning_rate": 3.6358785126312806e-05,
      "loss": 0.0006223093718290329,
      "step": 224300
    },
    {
      "epoch": 63.670167470905476,
      "grad_norm": 0.07481026649475098,
      "learning_rate": 3.635594663638944e-05,
      "loss": 0.000769777037203312,
      "step": 224310
    },
    {
      "epoch": 63.67300596082884,
      "grad_norm": 0.20656420290470123,
      "learning_rate": 3.635310814646608e-05,
      "loss": 0.0011296188458800316,
      "step": 224320
    },
    {
      "epoch": 63.6758444507522,
      "grad_norm": 0.08076050132513046,
      "learning_rate": 3.6350269656542724e-05,
      "loss": 0.0003452707082033157,
      "step": 224330
    },
    {
      "epoch": 63.67868294067556,
      "grad_norm": 0.019852444529533386,
      "learning_rate": 3.634743116661936e-05,
      "loss": 0.0001354474574327469,
      "step": 224340
    },
    {
      "epoch": 63.68152143059892,
      "grad_norm": 0.009588783606886864,
      "learning_rate": 3.6344592676696e-05,
      "loss": 0.0001936035230755806,
      "step": 224350
    },
    {
      "epoch": 63.68435992052228,
      "grad_norm": 0.11474621295928955,
      "learning_rate": 3.6341754186772634e-05,
      "loss": 0.0015029072761535644,
      "step": 224360
    },
    {
      "epoch": 63.687198410445646,
      "grad_norm": 0.3391484320163727,
      "learning_rate": 3.633891569684928e-05,
      "loss": 0.00025910399854183196,
      "step": 224370
    },
    {
      "epoch": 63.690036900369,
      "grad_norm": 5.7421345710754395,
      "learning_rate": 3.633607720692592e-05,
      "loss": 0.0015565535053610803,
      "step": 224380
    },
    {
      "epoch": 63.692875390292365,
      "grad_norm": 0.12042143195867538,
      "learning_rate": 3.633323871700255e-05,
      "loss": 0.00026180408895015717,
      "step": 224390
    },
    {
      "epoch": 63.69571388021573,
      "grad_norm": 0.004368194844573736,
      "learning_rate": 3.63304002270792e-05,
      "loss": 0.00029761362820863725,
      "step": 224400
    },
    {
      "epoch": 63.69855237013908,
      "grad_norm": 0.04389287531375885,
      "learning_rate": 3.6327561737155834e-05,
      "loss": 0.00028510894626379015,
      "step": 224410
    },
    {
      "epoch": 63.701390860062446,
      "grad_norm": 0.005231706891208887,
      "learning_rate": 3.6324723247232476e-05,
      "loss": 0.00035474412143230436,
      "step": 224420
    },
    {
      "epoch": 63.70422934998581,
      "grad_norm": 0.08212676644325256,
      "learning_rate": 3.632188475730912e-05,
      "loss": 0.00018621403723955153,
      "step": 224430
    },
    {
      "epoch": 63.70706783990917,
      "grad_norm": 0.8004197478294373,
      "learning_rate": 3.631904626738575e-05,
      "loss": 0.0004247302189469337,
      "step": 224440
    },
    {
      "epoch": 63.70990632983253,
      "grad_norm": 0.034846000373363495,
      "learning_rate": 3.631620777746239e-05,
      "loss": 0.0002795059233903885,
      "step": 224450
    },
    {
      "epoch": 63.71274481975589,
      "grad_norm": 0.015181692317128181,
      "learning_rate": 3.631336928753903e-05,
      "loss": 0.004319984465837479,
      "step": 224460
    },
    {
      "epoch": 63.71558330967925,
      "grad_norm": 0.014265087433159351,
      "learning_rate": 3.631053079761567e-05,
      "loss": 0.00015086904168128967,
      "step": 224470
    },
    {
      "epoch": 63.71842179960261,
      "grad_norm": 0.05621592327952385,
      "learning_rate": 3.630769230769231e-05,
      "loss": 0.00047922786325216293,
      "step": 224480
    },
    {
      "epoch": 63.72126028952597,
      "grad_norm": 0.0355403907597065,
      "learning_rate": 3.6304853817768945e-05,
      "loss": 0.0001550225540995598,
      "step": 224490
    },
    {
      "epoch": 63.724098779449335,
      "grad_norm": 0.055215127766132355,
      "learning_rate": 3.630201532784559e-05,
      "loss": 0.00046750903129577637,
      "step": 224500
    },
    {
      "epoch": 63.724098779449335,
      "eval_accuracy": 0.9804158453614803,
      "eval_loss": 0.07156595587730408,
      "eval_runtime": 32.7439,
      "eval_samples_per_second": 480.303,
      "eval_steps_per_second": 7.513,
      "step": 224500
    },
    {
      "epoch": 63.72693726937269,
      "grad_norm": 0.014392265118658543,
      "learning_rate": 3.629917683792223e-05,
      "loss": 0.0008051542565226555,
      "step": 224510
    },
    {
      "epoch": 63.729775759296054,
      "grad_norm": 1.1169310808181763,
      "learning_rate": 3.629633834799886e-05,
      "loss": 0.0028339531272649765,
      "step": 224520
    },
    {
      "epoch": 63.73261424921942,
      "grad_norm": 0.13225238025188446,
      "learning_rate": 3.629349985807551e-05,
      "loss": 0.0003013983368873596,
      "step": 224530
    },
    {
      "epoch": 63.73545273914278,
      "grad_norm": 0.8949704766273499,
      "learning_rate": 3.6290661368152145e-05,
      "loss": 0.0010794878005981445,
      "step": 224540
    },
    {
      "epoch": 63.738291229066135,
      "grad_norm": 1.656553030014038,
      "learning_rate": 3.6287822878228786e-05,
      "loss": 0.006018198654055595,
      "step": 224550
    },
    {
      "epoch": 63.7411297189895,
      "grad_norm": 0.035306207835674286,
      "learning_rate": 3.628498438830542e-05,
      "loss": 0.0009844541549682618,
      "step": 224560
    },
    {
      "epoch": 63.74396820891286,
      "grad_norm": 0.4865168631076813,
      "learning_rate": 3.628214589838206e-05,
      "loss": 0.0009433574974536896,
      "step": 224570
    },
    {
      "epoch": 63.74680669883622,
      "grad_norm": 0.32857009768486023,
      "learning_rate": 3.6279307408458704e-05,
      "loss": 0.005086997896432877,
      "step": 224580
    },
    {
      "epoch": 63.74964518875958,
      "grad_norm": 0.2155779004096985,
      "learning_rate": 3.627646891853534e-05,
      "loss": 0.0009758591651916504,
      "step": 224590
    },
    {
      "epoch": 63.75248367868294,
      "grad_norm": 0.010447981767356396,
      "learning_rate": 3.627363042861198e-05,
      "loss": 0.0008790446445345879,
      "step": 224600
    },
    {
      "epoch": 63.7553221686063,
      "grad_norm": 1.984439492225647,
      "learning_rate": 3.627079193868862e-05,
      "loss": 0.0005144774913787842,
      "step": 224610
    },
    {
      "epoch": 63.75816065852966,
      "grad_norm": 0.31451982259750366,
      "learning_rate": 3.6267953448765256e-05,
      "loss": 0.0003846365958452225,
      "step": 224620
    },
    {
      "epoch": 63.760999148453024,
      "grad_norm": 0.027535755187273026,
      "learning_rate": 3.62651149588419e-05,
      "loss": 0.0004064567387104034,
      "step": 224630
    },
    {
      "epoch": 63.76383763837639,
      "grad_norm": 1.6327887773513794,
      "learning_rate": 3.626227646891854e-05,
      "loss": 0.0005753325298428535,
      "step": 224640
    },
    {
      "epoch": 63.76667612829974,
      "grad_norm": 0.22552283108234406,
      "learning_rate": 3.625943797899517e-05,
      "loss": 0.00018854141235351561,
      "step": 224650
    },
    {
      "epoch": 63.769514618223106,
      "grad_norm": 0.2663899064064026,
      "learning_rate": 3.6256599489071814e-05,
      "loss": 0.0005783155560493469,
      "step": 224660
    },
    {
      "epoch": 63.77235310814647,
      "grad_norm": 0.8291921019554138,
      "learning_rate": 3.6253760999148456e-05,
      "loss": 0.002137721888720989,
      "step": 224670
    },
    {
      "epoch": 63.775191598069824,
      "grad_norm": 1.6700938940048218,
      "learning_rate": 3.625092250922509e-05,
      "loss": 0.0011700239032506944,
      "step": 224680
    },
    {
      "epoch": 63.77803008799319,
      "grad_norm": 0.041978947818279266,
      "learning_rate": 3.624808401930173e-05,
      "loss": 0.0007059192284941673,
      "step": 224690
    },
    {
      "epoch": 63.78086857791655,
      "grad_norm": 0.033448975533246994,
      "learning_rate": 3.624524552937837e-05,
      "loss": 0.002449842169880867,
      "step": 224700
    },
    {
      "epoch": 63.783707067839906,
      "grad_norm": 1.2871676683425903,
      "learning_rate": 3.6242407039455015e-05,
      "loss": 0.002666942775249481,
      "step": 224710
    },
    {
      "epoch": 63.78654555776327,
      "grad_norm": 3.1064724922180176,
      "learning_rate": 3.623956854953165e-05,
      "loss": 0.0027611643075942994,
      "step": 224720
    },
    {
      "epoch": 63.78938404768663,
      "grad_norm": 3.698308229446411,
      "learning_rate": 3.623673005960829e-05,
      "loss": 0.0021921943873167036,
      "step": 224730
    },
    {
      "epoch": 63.792222537609995,
      "grad_norm": 0.19769923388957977,
      "learning_rate": 3.623389156968493e-05,
      "loss": 0.000431833416223526,
      "step": 224740
    },
    {
      "epoch": 63.79506102753335,
      "grad_norm": 1.7552073001861572,
      "learning_rate": 3.6231053079761566e-05,
      "loss": 0.0004585022106766701,
      "step": 224750
    },
    {
      "epoch": 63.79789951745671,
      "grad_norm": 0.00428078044205904,
      "learning_rate": 3.622821458983821e-05,
      "loss": 0.0007421964779496193,
      "step": 224760
    },
    {
      "epoch": 63.800738007380076,
      "grad_norm": 0.6080329418182373,
      "learning_rate": 3.622537609991485e-05,
      "loss": 0.0004448911175131798,
      "step": 224770
    },
    {
      "epoch": 63.80357649730343,
      "grad_norm": 0.027296708896756172,
      "learning_rate": 3.6222537609991484e-05,
      "loss": 0.0003587124869227409,
      "step": 224780
    },
    {
      "epoch": 63.806414987226795,
      "grad_norm": 0.006950725801289082,
      "learning_rate": 3.6219699120068125e-05,
      "loss": 0.001052931509912014,
      "step": 224790
    },
    {
      "epoch": 63.80925347715016,
      "grad_norm": 0.15644274652004242,
      "learning_rate": 3.6216860630144767e-05,
      "loss": 0.0010136395692825317,
      "step": 224800
    },
    {
      "epoch": 63.81209196707351,
      "grad_norm": 0.008039982058107853,
      "learning_rate": 3.62140221402214e-05,
      "loss": 0.0001883283257484436,
      "step": 224810
    },
    {
      "epoch": 63.814930456996876,
      "grad_norm": 2.18320894241333,
      "learning_rate": 3.621118365029804e-05,
      "loss": 0.0010849986225366592,
      "step": 224820
    },
    {
      "epoch": 63.81776894692024,
      "grad_norm": 0.026601672172546387,
      "learning_rate": 3.6208345160374684e-05,
      "loss": 0.000851152092218399,
      "step": 224830
    },
    {
      "epoch": 63.8206074368436,
      "grad_norm": 0.11838595569133759,
      "learning_rate": 3.6205506670451325e-05,
      "loss": 0.004031016677618027,
      "step": 224840
    },
    {
      "epoch": 63.82344592676696,
      "grad_norm": 0.2057223916053772,
      "learning_rate": 3.620266818052796e-05,
      "loss": 0.003005993738770485,
      "step": 224850
    },
    {
      "epoch": 63.82628441669032,
      "grad_norm": 0.04075217247009277,
      "learning_rate": 3.6199829690604594e-05,
      "loss": 0.00020407196134328843,
      "step": 224860
    },
    {
      "epoch": 63.829122906613684,
      "grad_norm": 0.07504704594612122,
      "learning_rate": 3.619699120068124e-05,
      "loss": 0.00040705259889364245,
      "step": 224870
    },
    {
      "epoch": 63.83196139653704,
      "grad_norm": 0.07844492793083191,
      "learning_rate": 3.619415271075788e-05,
      "loss": 7.885117083787918e-05,
      "step": 224880
    },
    {
      "epoch": 63.8347998864604,
      "grad_norm": 0.023320579901337624,
      "learning_rate": 3.619131422083452e-05,
      "loss": 0.00020247306674718856,
      "step": 224890
    },
    {
      "epoch": 63.837638376383765,
      "grad_norm": 0.1836031824350357,
      "learning_rate": 3.618847573091116e-05,
      "loss": 0.0005699736997485161,
      "step": 224900
    },
    {
      "epoch": 63.84047686630713,
      "grad_norm": 0.01980968564748764,
      "learning_rate": 3.6185637240987795e-05,
      "loss": 0.000213627889752388,
      "step": 224910
    },
    {
      "epoch": 63.843315356230484,
      "grad_norm": 0.013688120990991592,
      "learning_rate": 3.6182798751064436e-05,
      "loss": 0.0028121789917349817,
      "step": 224920
    },
    {
      "epoch": 63.84615384615385,
      "grad_norm": 0.34825628995895386,
      "learning_rate": 3.617996026114108e-05,
      "loss": 0.00046895071864128113,
      "step": 224930
    },
    {
      "epoch": 63.84899233607721,
      "grad_norm": 0.02100256271660328,
      "learning_rate": 3.617712177121771e-05,
      "loss": 0.00024995915591716764,
      "step": 224940
    },
    {
      "epoch": 63.851830826000565,
      "grad_norm": 0.16673873364925385,
      "learning_rate": 3.617428328129435e-05,
      "loss": 0.0004692522808909416,
      "step": 224950
    },
    {
      "epoch": 63.85466931592393,
      "grad_norm": 0.03671998530626297,
      "learning_rate": 3.617144479137099e-05,
      "loss": 0.0002580223605036736,
      "step": 224960
    },
    {
      "epoch": 63.85750780584729,
      "grad_norm": 0.04799570515751839,
      "learning_rate": 3.6168606301447636e-05,
      "loss": 0.00015658382326364518,
      "step": 224970
    },
    {
      "epoch": 63.86034629577065,
      "grad_norm": 0.09831889718770981,
      "learning_rate": 3.616576781152427e-05,
      "loss": 0.0004115656018257141,
      "step": 224980
    },
    {
      "epoch": 63.86318478569401,
      "grad_norm": 0.0906229242682457,
      "learning_rate": 3.6162929321600905e-05,
      "loss": 0.00021391455084085464,
      "step": 224990
    },
    {
      "epoch": 63.86602327561737,
      "grad_norm": 0.02601264789700508,
      "learning_rate": 3.6160090831677553e-05,
      "loss": 0.00031835511326789857,
      "step": 225000
    },
    {
      "epoch": 63.86602327561737,
      "eval_accuracy": 0.9790805620906721,
      "eval_loss": 0.07967314124107361,
      "eval_runtime": 32.7919,
      "eval_samples_per_second": 479.601,
      "eval_steps_per_second": 7.502,
      "step": 225000
    },
    {
      "epoch": 63.868861765540736,
      "grad_norm": 0.03787478804588318,
      "learning_rate": 3.615725234175419e-05,
      "loss": 0.00886744111776352,
      "step": 225010
    },
    {
      "epoch": 63.87170025546409,
      "grad_norm": 0.08542962372303009,
      "learning_rate": 3.615441385183083e-05,
      "loss": 0.00013766512274742125,
      "step": 225020
    },
    {
      "epoch": 63.874538745387454,
      "grad_norm": 0.05806276574730873,
      "learning_rate": 3.615157536190747e-05,
      "loss": 0.00015752632170915604,
      "step": 225030
    },
    {
      "epoch": 63.87737723531082,
      "grad_norm": 0.3203323781490326,
      "learning_rate": 3.6148736871984105e-05,
      "loss": 0.0033796004951000213,
      "step": 225040
    },
    {
      "epoch": 63.88021572523417,
      "grad_norm": 0.06530045717954636,
      "learning_rate": 3.614589838206075e-05,
      "loss": 9.647663682699203e-05,
      "step": 225050
    },
    {
      "epoch": 63.883054215157536,
      "grad_norm": 0.06559520959854126,
      "learning_rate": 3.614305989213738e-05,
      "loss": 0.0011992694810032845,
      "step": 225060
    },
    {
      "epoch": 63.8858927050809,
      "grad_norm": 0.17584893107414246,
      "learning_rate": 3.614022140221402e-05,
      "loss": 0.001924452744424343,
      "step": 225070
    },
    {
      "epoch": 63.888731195004254,
      "grad_norm": 1.372194528579712,
      "learning_rate": 3.6137382912290664e-05,
      "loss": 0.0056559242308139804,
      "step": 225080
    },
    {
      "epoch": 63.89156968492762,
      "grad_norm": 0.0700661763548851,
      "learning_rate": 3.61345444223673e-05,
      "loss": 0.0005012307316064835,
      "step": 225090
    },
    {
      "epoch": 63.89440817485098,
      "grad_norm": 2.214586019515991,
      "learning_rate": 3.613170593244394e-05,
      "loss": 0.0030909327790141104,
      "step": 225100
    },
    {
      "epoch": 63.89724666477434,
      "grad_norm": 1.3360726833343506,
      "learning_rate": 3.612886744252058e-05,
      "loss": 0.0004929795861244202,
      "step": 225110
    },
    {
      "epoch": 63.9000851546977,
      "grad_norm": 0.07528794556856155,
      "learning_rate": 3.6126028952597216e-05,
      "loss": 0.000261850468814373,
      "step": 225120
    },
    {
      "epoch": 63.90292364462106,
      "grad_norm": 0.046959083527326584,
      "learning_rate": 3.6123190462673864e-05,
      "loss": 0.0008098874241113663,
      "step": 225130
    },
    {
      "epoch": 63.905762134544425,
      "grad_norm": 0.04502107575535774,
      "learning_rate": 3.61203519727505e-05,
      "loss": 0.00029917098581790926,
      "step": 225140
    },
    {
      "epoch": 63.90860062446778,
      "grad_norm": 0.4759259819984436,
      "learning_rate": 3.611751348282713e-05,
      "loss": 0.0004285890609025955,
      "step": 225150
    },
    {
      "epoch": 63.91143911439114,
      "grad_norm": 0.027916869148612022,
      "learning_rate": 3.6114674992903775e-05,
      "loss": 0.0004437034949660301,
      "step": 225160
    },
    {
      "epoch": 63.914277604314506,
      "grad_norm": 0.08362825959920883,
      "learning_rate": 3.6111836502980416e-05,
      "loss": 0.003944588080048561,
      "step": 225170
    },
    {
      "epoch": 63.91711609423786,
      "grad_norm": 3.528416872024536,
      "learning_rate": 3.610899801305706e-05,
      "loss": 0.0008362419903278351,
      "step": 225180
    },
    {
      "epoch": 63.919954584161225,
      "grad_norm": 0.4539260268211365,
      "learning_rate": 3.610615952313369e-05,
      "loss": 0.0002358408644795418,
      "step": 225190
    },
    {
      "epoch": 63.92279307408459,
      "grad_norm": 0.06131996214389801,
      "learning_rate": 3.6103321033210333e-05,
      "loss": 0.0004313550889492035,
      "step": 225200
    },
    {
      "epoch": 63.92563156400795,
      "grad_norm": 0.04696110263466835,
      "learning_rate": 3.6100482543286975e-05,
      "loss": 0.00038882624357938766,
      "step": 225210
    },
    {
      "epoch": 63.928470053931306,
      "grad_norm": 0.035636257380247116,
      "learning_rate": 3.609764405336361e-05,
      "loss": 0.0003195410594344139,
      "step": 225220
    },
    {
      "epoch": 63.93130854385467,
      "grad_norm": 0.023019082844257355,
      "learning_rate": 3.609480556344025e-05,
      "loss": 0.0005151335150003433,
      "step": 225230
    },
    {
      "epoch": 63.93414703377803,
      "grad_norm": 0.14070335030555725,
      "learning_rate": 3.609196707351689e-05,
      "loss": 0.00017023365944623947,
      "step": 225240
    },
    {
      "epoch": 63.93698552370139,
      "grad_norm": 0.02505980059504509,
      "learning_rate": 3.608912858359353e-05,
      "loss": 0.0029459377750754357,
      "step": 225250
    },
    {
      "epoch": 63.93982401362475,
      "grad_norm": 0.030038023367524147,
      "learning_rate": 3.608629009367017e-05,
      "loss": 0.0002865279093384743,
      "step": 225260
    },
    {
      "epoch": 63.942662503548114,
      "grad_norm": 0.5393161773681641,
      "learning_rate": 3.608345160374681e-05,
      "loss": 0.00026196874678134917,
      "step": 225270
    },
    {
      "epoch": 63.94550099347148,
      "grad_norm": 0.050458863377571106,
      "learning_rate": 3.6080613113823444e-05,
      "loss": 9.706318378448487e-05,
      "step": 225280
    },
    {
      "epoch": 63.94833948339483,
      "grad_norm": 0.05576249584555626,
      "learning_rate": 3.6077774623900086e-05,
      "loss": 0.0019566601142287253,
      "step": 225290
    },
    {
      "epoch": 63.951177973318195,
      "grad_norm": 1.0488667488098145,
      "learning_rate": 3.607493613397673e-05,
      "loss": 0.0012837210670113564,
      "step": 225300
    },
    {
      "epoch": 63.95401646324156,
      "grad_norm": 0.23463396728038788,
      "learning_rate": 3.607209764405337e-05,
      "loss": 0.0006284246221184731,
      "step": 225310
    },
    {
      "epoch": 63.956854953164914,
      "grad_norm": 0.3566974103450775,
      "learning_rate": 3.606925915413e-05,
      "loss": 0.008516659587621688,
      "step": 225320
    },
    {
      "epoch": 63.95969344308828,
      "grad_norm": 0.24091586470603943,
      "learning_rate": 3.6066420664206644e-05,
      "loss": 0.0012085732072591782,
      "step": 225330
    },
    {
      "epoch": 63.96253193301164,
      "grad_norm": 0.030770020559430122,
      "learning_rate": 3.6063582174283286e-05,
      "loss": 0.0002991087734699249,
      "step": 225340
    },
    {
      "epoch": 63.965370422934996,
      "grad_norm": 0.011261243373155594,
      "learning_rate": 3.606074368435992e-05,
      "loss": 0.0008709967136383056,
      "step": 225350
    },
    {
      "epoch": 63.96820891285836,
      "grad_norm": 0.060628827661275864,
      "learning_rate": 3.605790519443656e-05,
      "loss": 0.005290616303682327,
      "step": 225360
    },
    {
      "epoch": 63.97104740278172,
      "grad_norm": 0.08735974878072739,
      "learning_rate": 3.60550667045132e-05,
      "loss": 0.0012508576735854148,
      "step": 225370
    },
    {
      "epoch": 63.973885892705084,
      "grad_norm": 0.036764901131391525,
      "learning_rate": 3.605222821458984e-05,
      "loss": 0.0002717345952987671,
      "step": 225380
    },
    {
      "epoch": 63.97672438262844,
      "grad_norm": 0.816257655620575,
      "learning_rate": 3.604938972466648e-05,
      "loss": 0.00155891552567482,
      "step": 225390
    },
    {
      "epoch": 63.9795628725518,
      "grad_norm": 0.024156149476766586,
      "learning_rate": 3.604655123474312e-05,
      "loss": 0.0014712136238813401,
      "step": 225400
    },
    {
      "epoch": 63.982401362475166,
      "grad_norm": 0.02049519307911396,
      "learning_rate": 3.6043712744819755e-05,
      "loss": 0.00024126041680574417,
      "step": 225410
    },
    {
      "epoch": 63.98523985239852,
      "grad_norm": 0.06827552616596222,
      "learning_rate": 3.6040874254896396e-05,
      "loss": 0.0003369135782122612,
      "step": 225420
    },
    {
      "epoch": 63.988078342321884,
      "grad_norm": 0.10673180222511292,
      "learning_rate": 3.603803576497304e-05,
      "loss": 0.0005344051867723465,
      "step": 225430
    },
    {
      "epoch": 63.99091683224525,
      "grad_norm": 0.029519351199269295,
      "learning_rate": 3.603519727504968e-05,
      "loss": 0.0019487669691443444,
      "step": 225440
    },
    {
      "epoch": 63.9937553221686,
      "grad_norm": 2.0460259914398193,
      "learning_rate": 3.6032358785126314e-05,
      "loss": 0.0005363475531339646,
      "step": 225450
    },
    {
      "epoch": 63.996593812091966,
      "grad_norm": 0.045117057859897614,
      "learning_rate": 3.6029520295202955e-05,
      "loss": 0.00029621664434671403,
      "step": 225460
    },
    {
      "epoch": 63.99943230201533,
      "grad_norm": 0.13858415186405182,
      "learning_rate": 3.6026681805279596e-05,
      "loss": 0.00020586680620908736,
      "step": 225470
    },
    {
      "epoch": 64.00227079193868,
      "grad_norm": 0.008380487561225891,
      "learning_rate": 3.602384331535623e-05,
      "loss": 0.0022238465026021004,
      "step": 225480
    },
    {
      "epoch": 64.00510928186205,
      "grad_norm": 0.020235488191246986,
      "learning_rate": 3.602100482543287e-05,
      "loss": 9.889211505651475e-05,
      "step": 225490
    },
    {
      "epoch": 64.00794777178541,
      "grad_norm": 0.019735604524612427,
      "learning_rate": 3.6018166335509514e-05,
      "loss": 0.0008364584296941757,
      "step": 225500
    },
    {
      "epoch": 64.00794777178541,
      "eval_accuracy": 0.9795892414319324,
      "eval_loss": 0.07281606644392014,
      "eval_runtime": 33.0857,
      "eval_samples_per_second": 475.342,
      "eval_steps_per_second": 7.435,
      "step": 225500
    },
    {
      "epoch": 64.01078626170877,
      "grad_norm": 0.05334628000855446,
      "learning_rate": 3.601532784558615e-05,
      "loss": 0.0005608545616269112,
      "step": 225510
    },
    {
      "epoch": 64.01362475163214,
      "grad_norm": 0.0789429247379303,
      "learning_rate": 3.601248935566279e-05,
      "loss": 0.0009571185335516929,
      "step": 225520
    },
    {
      "epoch": 64.0164632415555,
      "grad_norm": 0.023302171379327774,
      "learning_rate": 3.600965086573943e-05,
      "loss": 0.00013711117208003998,
      "step": 225530
    },
    {
      "epoch": 64.01930173147885,
      "grad_norm": 0.0636732429265976,
      "learning_rate": 3.6006812375816066e-05,
      "loss": 0.0017106911167502404,
      "step": 225540
    },
    {
      "epoch": 64.02214022140221,
      "grad_norm": 1.7744396924972534,
      "learning_rate": 3.600397388589271e-05,
      "loss": 0.0002216283231973648,
      "step": 225550
    },
    {
      "epoch": 64.02497871132557,
      "grad_norm": 2.597858190536499,
      "learning_rate": 3.600113539596935e-05,
      "loss": 0.002311082184314728,
      "step": 225560
    },
    {
      "epoch": 64.02781720124894,
      "grad_norm": 0.047048892825841904,
      "learning_rate": 3.599829690604598e-05,
      "loss": 0.0001324361190199852,
      "step": 225570
    },
    {
      "epoch": 64.0306556911723,
      "grad_norm": 0.15715128183364868,
      "learning_rate": 3.5995458416122624e-05,
      "loss": 0.0006887489929795265,
      "step": 225580
    },
    {
      "epoch": 64.03349418109566,
      "grad_norm": 0.09049355983734131,
      "learning_rate": 3.599261992619926e-05,
      "loss": 0.0004651915282011032,
      "step": 225590
    },
    {
      "epoch": 64.03633267101901,
      "grad_norm": 0.04880692437291145,
      "learning_rate": 3.598978143627591e-05,
      "loss": 0.0005904717370867729,
      "step": 225600
    },
    {
      "epoch": 64.03917116094237,
      "grad_norm": 0.6023997068405151,
      "learning_rate": 3.598694294635254e-05,
      "loss": 0.00021831318736076355,
      "step": 225610
    },
    {
      "epoch": 64.04200965086574,
      "grad_norm": 0.006442782934755087,
      "learning_rate": 3.5984104456429176e-05,
      "loss": 0.0004392102360725403,
      "step": 225620
    },
    {
      "epoch": 64.0448481407891,
      "grad_norm": 9.170734405517578,
      "learning_rate": 3.5981265966505824e-05,
      "loss": 0.004915034025907516,
      "step": 225630
    },
    {
      "epoch": 64.04768663071246,
      "grad_norm": 0.032801248133182526,
      "learning_rate": 3.597842747658246e-05,
      "loss": 0.0008158409968018532,
      "step": 225640
    },
    {
      "epoch": 64.05052512063583,
      "grad_norm": 0.009990845806896687,
      "learning_rate": 3.59755889866591e-05,
      "loss": 0.0013832228258252143,
      "step": 225650
    },
    {
      "epoch": 64.05336361055919,
      "grad_norm": 0.2922166585922241,
      "learning_rate": 3.597275049673574e-05,
      "loss": 0.0003526870161294937,
      "step": 225660
    },
    {
      "epoch": 64.05620210048254,
      "grad_norm": 0.02581685408949852,
      "learning_rate": 3.5969912006812376e-05,
      "loss": 0.00012815054506063462,
      "step": 225670
    },
    {
      "epoch": 64.0590405904059,
      "grad_norm": 0.04092688113451004,
      "learning_rate": 3.596707351688902e-05,
      "loss": 0.00014609433710575104,
      "step": 225680
    },
    {
      "epoch": 64.06187908032926,
      "grad_norm": 0.12763315439224243,
      "learning_rate": 3.596423502696565e-05,
      "loss": 0.00024637095630168914,
      "step": 225690
    },
    {
      "epoch": 64.06471757025263,
      "grad_norm": 0.05214500054717064,
      "learning_rate": 3.5961396537042294e-05,
      "loss": 0.00020516272634267806,
      "step": 225700
    },
    {
      "epoch": 64.06755606017599,
      "grad_norm": 0.06695237010717392,
      "learning_rate": 3.5958558047118935e-05,
      "loss": 0.0003362055867910385,
      "step": 225710
    },
    {
      "epoch": 64.07039455009935,
      "grad_norm": 0.053965870290994644,
      "learning_rate": 3.595571955719557e-05,
      "loss": 0.00010519232600927352,
      "step": 225720
    },
    {
      "epoch": 64.07323304002271,
      "grad_norm": 1.744786262512207,
      "learning_rate": 3.595288106727222e-05,
      "loss": 0.0005765903741121292,
      "step": 225730
    },
    {
      "epoch": 64.07607152994606,
      "grad_norm": 7.338283061981201,
      "learning_rate": 3.595004257734885e-05,
      "loss": 0.001788925565779209,
      "step": 225740
    },
    {
      "epoch": 64.07891001986943,
      "grad_norm": 1.5170080661773682,
      "learning_rate": 3.594720408742549e-05,
      "loss": 0.003729585185647011,
      "step": 225750
    },
    {
      "epoch": 64.08174850979279,
      "grad_norm": 0.2071271687746048,
      "learning_rate": 3.5944365597502135e-05,
      "loss": 0.0015554523095488548,
      "step": 225760
    },
    {
      "epoch": 64.08458699971615,
      "grad_norm": 0.034490037709474564,
      "learning_rate": 3.594152710757877e-05,
      "loss": 0.0002894224599003792,
      "step": 225770
    },
    {
      "epoch": 64.08742548963951,
      "grad_norm": 0.019468436017632484,
      "learning_rate": 3.593868861765541e-05,
      "loss": 0.00031349435448646545,
      "step": 225780
    },
    {
      "epoch": 64.09026397956288,
      "grad_norm": 0.010690311901271343,
      "learning_rate": 3.5935850127732046e-05,
      "loss": 0.0012890243902802466,
      "step": 225790
    },
    {
      "epoch": 64.09310246948624,
      "grad_norm": 0.29279813170433044,
      "learning_rate": 3.593301163780869e-05,
      "loss": 0.0016189467161893845,
      "step": 225800
    },
    {
      "epoch": 64.09594095940959,
      "grad_norm": 0.2605352997779846,
      "learning_rate": 3.593017314788533e-05,
      "loss": 0.0010280540212988854,
      "step": 225810
    },
    {
      "epoch": 64.09877944933295,
      "grad_norm": 0.008754221722483635,
      "learning_rate": 3.592733465796196e-05,
      "loss": 0.00015467945486307143,
      "step": 225820
    },
    {
      "epoch": 64.10161793925631,
      "grad_norm": 0.02535228244960308,
      "learning_rate": 3.5924496168038605e-05,
      "loss": 0.0003149053081870079,
      "step": 225830
    },
    {
      "epoch": 64.10445642917968,
      "grad_norm": 0.007408665493130684,
      "learning_rate": 3.5921657678115246e-05,
      "loss": 0.0002118196338415146,
      "step": 225840
    },
    {
      "epoch": 64.10729491910304,
      "grad_norm": 0.5629767179489136,
      "learning_rate": 3.591881918819188e-05,
      "loss": 0.0003276322036981583,
      "step": 225850
    },
    {
      "epoch": 64.1101334090264,
      "grad_norm": 0.009962372481822968,
      "learning_rate": 3.591598069826853e-05,
      "loss": 0.00028475429862737653,
      "step": 225860
    },
    {
      "epoch": 64.11297189894975,
      "grad_norm": 0.01532883569598198,
      "learning_rate": 3.591314220834516e-05,
      "loss": 9.420327842235565e-05,
      "step": 225870
    },
    {
      "epoch": 64.11581038887311,
      "grad_norm": 0.009321934543550014,
      "learning_rate": 3.59103037184218e-05,
      "loss": 0.00047872476279735564,
      "step": 225880
    },
    {
      "epoch": 64.11864887879648,
      "grad_norm": 0.030270352959632874,
      "learning_rate": 3.590746522849844e-05,
      "loss": 0.00024196412414312363,
      "step": 225890
    },
    {
      "epoch": 64.12148736871984,
      "grad_norm": 0.4954870343208313,
      "learning_rate": 3.590462673857508e-05,
      "loss": 0.0002753160893917084,
      "step": 225900
    },
    {
      "epoch": 64.1243258586432,
      "grad_norm": 0.10643914341926575,
      "learning_rate": 3.590178824865172e-05,
      "loss": 0.0008167998865246772,
      "step": 225910
    },
    {
      "epoch": 64.12716434856657,
      "grad_norm": 0.006969409063458443,
      "learning_rate": 3.5898949758728357e-05,
      "loss": 0.0008104793727397918,
      "step": 225920
    },
    {
      "epoch": 64.13000283848993,
      "grad_norm": 0.3603917062282562,
      "learning_rate": 3.5896111268805e-05,
      "loss": 0.001976025477051735,
      "step": 225930
    },
    {
      "epoch": 64.13284132841328,
      "grad_norm": 0.010034373961389065,
      "learning_rate": 3.589327277888164e-05,
      "loss": 0.000402035191655159,
      "step": 225940
    },
    {
      "epoch": 64.13567981833664,
      "grad_norm": 4.584012985229492,
      "learning_rate": 3.5890434288958274e-05,
      "loss": 0.0011951865628361702,
      "step": 225950
    },
    {
      "epoch": 64.13851830826,
      "grad_norm": 0.24615073204040527,
      "learning_rate": 3.5887879648027254e-05,
      "loss": 0.0021301694214344025,
      "step": 225960
    },
    {
      "epoch": 64.14135679818337,
      "grad_norm": 0.2506524920463562,
      "learning_rate": 3.588504115810389e-05,
      "loss": 0.005956707894802094,
      "step": 225970
    },
    {
      "epoch": 64.14419528810673,
      "grad_norm": 0.29895687103271484,
      "learning_rate": 3.588220266818053e-05,
      "loss": 0.0008015289902687072,
      "step": 225980
    },
    {
      "epoch": 64.14703377803009,
      "grad_norm": 0.006833396852016449,
      "learning_rate": 3.587936417825717e-05,
      "loss": 0.0016420749947428704,
      "step": 225990
    },
    {
      "epoch": 64.14987226795346,
      "grad_norm": 0.06541809439659119,
      "learning_rate": 3.5876525688333806e-05,
      "loss": 0.00042325574904680254,
      "step": 226000
    },
    {
      "epoch": 64.14987226795346,
      "eval_accuracy": 0.9776816939022064,
      "eval_loss": 0.08531413227319717,
      "eval_runtime": 32.9806,
      "eval_samples_per_second": 476.856,
      "eval_steps_per_second": 7.459,
      "step": 226000
    },
    {
      "epoch": 64.1527107578768,
      "grad_norm": 6.881387233734131,
      "learning_rate": 3.587368719841045e-05,
      "loss": 0.0012235604226589203,
      "step": 226010
    },
    {
      "epoch": 64.15554924780017,
      "grad_norm": 0.04838798940181732,
      "learning_rate": 3.587084870848709e-05,
      "loss": 0.00017306357622146607,
      "step": 226020
    },
    {
      "epoch": 64.15838773772353,
      "grad_norm": 0.09536255896091461,
      "learning_rate": 3.586801021856372e-05,
      "loss": 0.0002011701464653015,
      "step": 226030
    },
    {
      "epoch": 64.16122622764689,
      "grad_norm": 0.10439273715019226,
      "learning_rate": 3.5865171728640365e-05,
      "loss": 0.00012302305549383163,
      "step": 226040
    },
    {
      "epoch": 64.16406471757026,
      "grad_norm": 0.1740400493144989,
      "learning_rate": 3.5862333238717006e-05,
      "loss": 0.00012539587914943694,
      "step": 226050
    },
    {
      "epoch": 64.16690320749362,
      "grad_norm": 0.12171534448862076,
      "learning_rate": 3.585949474879364e-05,
      "loss": 0.0003893587738275528,
      "step": 226060
    },
    {
      "epoch": 64.16974169741698,
      "grad_norm": 0.016652582213282585,
      "learning_rate": 3.585665625887028e-05,
      "loss": 0.00011251289397478104,
      "step": 226070
    },
    {
      "epoch": 64.17258018734033,
      "grad_norm": 0.020085537806153297,
      "learning_rate": 3.585381776894692e-05,
      "loss": 0.00038974378257989885,
      "step": 226080
    },
    {
      "epoch": 64.17541867726369,
      "grad_norm": 0.01949652098119259,
      "learning_rate": 3.585097927902356e-05,
      "loss": 0.00022530127316713333,
      "step": 226090
    },
    {
      "epoch": 64.17825716718706,
      "grad_norm": 0.0192548930644989,
      "learning_rate": 3.58481407891002e-05,
      "loss": 0.00018048398196697236,
      "step": 226100
    },
    {
      "epoch": 64.18109565711042,
      "grad_norm": 0.00596956629306078,
      "learning_rate": 3.584530229917684e-05,
      "loss": 6.864462047815323e-05,
      "step": 226110
    },
    {
      "epoch": 64.18393414703378,
      "grad_norm": 0.009871427901089191,
      "learning_rate": 3.584246380925348e-05,
      "loss": 0.00010660532861948014,
      "step": 226120
    },
    {
      "epoch": 64.18677263695714,
      "grad_norm": 1.5881246328353882,
      "learning_rate": 3.5839625319330117e-05,
      "loss": 0.0009110776707530021,
      "step": 226130
    },
    {
      "epoch": 64.1896111268805,
      "grad_norm": 0.058314915746450424,
      "learning_rate": 3.583678682940676e-05,
      "loss": 0.0002647867426276207,
      "step": 226140
    },
    {
      "epoch": 64.19244961680386,
      "grad_norm": 0.026866160333156586,
      "learning_rate": 3.58339483394834e-05,
      "loss": 0.000495917908847332,
      "step": 226150
    },
    {
      "epoch": 64.19528810672722,
      "grad_norm": 0.04071198031306267,
      "learning_rate": 3.5831109849560034e-05,
      "loss": 7.717311382293701e-05,
      "step": 226160
    },
    {
      "epoch": 64.19812659665058,
      "grad_norm": 0.032703083008527756,
      "learning_rate": 3.5828271359636675e-05,
      "loss": 7.667187601327896e-05,
      "step": 226170
    },
    {
      "epoch": 64.20096508657394,
      "grad_norm": 0.24686218798160553,
      "learning_rate": 3.582543286971332e-05,
      "loss": 0.00014807265251874923,
      "step": 226180
    },
    {
      "epoch": 64.20380357649731,
      "grad_norm": 0.00932091660797596,
      "learning_rate": 3.582259437978995e-05,
      "loss": 0.0001563597470521927,
      "step": 226190
    },
    {
      "epoch": 64.20664206642067,
      "grad_norm": 0.04663356393575668,
      "learning_rate": 3.581975588986659e-05,
      "loss": 9.128805249929428e-05,
      "step": 226200
    },
    {
      "epoch": 64.20948055634402,
      "grad_norm": 0.009335332550108433,
      "learning_rate": 3.5816917399943234e-05,
      "loss": 0.00013524889945983886,
      "step": 226210
    },
    {
      "epoch": 64.21231904626738,
      "grad_norm": 0.002837138483300805,
      "learning_rate": 3.581407891001987e-05,
      "loss": 6.556175649166107e-05,
      "step": 226220
    },
    {
      "epoch": 64.21515753619074,
      "grad_norm": 0.01361202634871006,
      "learning_rate": 3.581124042009651e-05,
      "loss": 0.00012867078185081483,
      "step": 226230
    },
    {
      "epoch": 64.21799602611411,
      "grad_norm": 0.07660821825265884,
      "learning_rate": 3.580840193017315e-05,
      "loss": 0.00012727770954370498,
      "step": 226240
    },
    {
      "epoch": 64.22083451603747,
      "grad_norm": 0.011597616598010063,
      "learning_rate": 3.580556344024979e-05,
      "loss": 0.00045406799763441085,
      "step": 226250
    },
    {
      "epoch": 64.22367300596083,
      "grad_norm": 0.013767405413091183,
      "learning_rate": 3.580272495032643e-05,
      "loss": 0.0012319032102823257,
      "step": 226260
    },
    {
      "epoch": 64.2265114958842,
      "grad_norm": 0.05508313328027725,
      "learning_rate": 3.579988646040306e-05,
      "loss": 0.00045019239187240603,
      "step": 226270
    },
    {
      "epoch": 64.22934998580754,
      "grad_norm": 0.10069742053747177,
      "learning_rate": 3.579704797047971e-05,
      "loss": 0.0015166960656642914,
      "step": 226280
    },
    {
      "epoch": 64.23218847573091,
      "grad_norm": 0.07478328049182892,
      "learning_rate": 3.5794209480556345e-05,
      "loss": 0.00028510801494121554,
      "step": 226290
    },
    {
      "epoch": 64.23502696565427,
      "grad_norm": 0.1531326025724411,
      "learning_rate": 3.5791370990632986e-05,
      "loss": 0.0009127387776970864,
      "step": 226300
    },
    {
      "epoch": 64.23786545557763,
      "grad_norm": 0.5119642615318298,
      "learning_rate": 3.578853250070963e-05,
      "loss": 0.0002471761777997017,
      "step": 226310
    },
    {
      "epoch": 64.240703945501,
      "grad_norm": 0.010212230496108532,
      "learning_rate": 3.578569401078626e-05,
      "loss": 0.00016364865005016326,
      "step": 226320
    },
    {
      "epoch": 64.24354243542436,
      "grad_norm": 0.08899784833192825,
      "learning_rate": 3.5782855520862903e-05,
      "loss": 0.00043965857475996016,
      "step": 226330
    },
    {
      "epoch": 64.24638092534771,
      "grad_norm": 0.5204827189445496,
      "learning_rate": 3.5780017030939545e-05,
      "loss": 0.0003086550161242485,
      "step": 226340
    },
    {
      "epoch": 64.24921941527107,
      "grad_norm": 0.13571280241012573,
      "learning_rate": 3.577717854101618e-05,
      "loss": 0.0001589776948094368,
      "step": 226350
    },
    {
      "epoch": 64.25205790519443,
      "grad_norm": 0.01725863665342331,
      "learning_rate": 3.577434005109282e-05,
      "loss": 0.0015774711966514588,
      "step": 226360
    },
    {
      "epoch": 64.2548963951178,
      "grad_norm": 0.04782959073781967,
      "learning_rate": 3.5771501561169455e-05,
      "loss": 0.0001449989154934883,
      "step": 226370
    },
    {
      "epoch": 64.25773488504116,
      "grad_norm": 1.7088624238967896,
      "learning_rate": 3.5768663071246104e-05,
      "loss": 0.0008180089294910431,
      "step": 226380
    },
    {
      "epoch": 64.26057337496452,
      "grad_norm": 0.18049293756484985,
      "learning_rate": 3.576582458132274e-05,
      "loss": 0.00014406889677047728,
      "step": 226390
    },
    {
      "epoch": 64.26341186488789,
      "grad_norm": 0.04832873493432999,
      "learning_rate": 3.576298609139937e-05,
      "loss": 0.0017211440950632094,
      "step": 226400
    },
    {
      "epoch": 64.26625035481123,
      "grad_norm": 0.5989809036254883,
      "learning_rate": 3.576014760147602e-05,
      "loss": 0.0010254276916384698,
      "step": 226410
    },
    {
      "epoch": 64.2690888447346,
      "grad_norm": 0.13365912437438965,
      "learning_rate": 3.5757309111552655e-05,
      "loss": 0.0003421025350689888,
      "step": 226420
    },
    {
      "epoch": 64.27192733465796,
      "grad_norm": 0.7348617315292358,
      "learning_rate": 3.57544706216293e-05,
      "loss": 0.0008985867723822593,
      "step": 226430
    },
    {
      "epoch": 64.27476582458132,
      "grad_norm": 0.21525119245052338,
      "learning_rate": 3.575163213170594e-05,
      "loss": 0.00020962301641702652,
      "step": 226440
    },
    {
      "epoch": 64.27760431450469,
      "grad_norm": 0.08904733508825302,
      "learning_rate": 3.574879364178257e-05,
      "loss": 0.00015540234744548798,
      "step": 226450
    },
    {
      "epoch": 64.28044280442805,
      "grad_norm": 0.014033038169145584,
      "learning_rate": 3.5745955151859214e-05,
      "loss": 0.0003388112410902977,
      "step": 226460
    },
    {
      "epoch": 64.28328129435141,
      "grad_norm": 0.4618314206600189,
      "learning_rate": 3.574311666193585e-05,
      "loss": 0.0004490155726671219,
      "step": 226470
    },
    {
      "epoch": 64.28611978427476,
      "grad_norm": 0.03341633081436157,
      "learning_rate": 3.574027817201249e-05,
      "loss": 0.00018044523894786835,
      "step": 226480
    },
    {
      "epoch": 64.28895827419812,
      "grad_norm": 0.027076125144958496,
      "learning_rate": 3.573743968208913e-05,
      "loss": 0.0011175639927387237,
      "step": 226490
    },
    {
      "epoch": 64.29179676412149,
      "grad_norm": 0.020943593233823776,
      "learning_rate": 3.5734601192165766e-05,
      "loss": 0.0007911078631877899,
      "step": 226500
    },
    {
      "epoch": 64.29179676412149,
      "eval_accuracy": 0.9722133909836587,
      "eval_loss": 0.11135087162256241,
      "eval_runtime": 33.167,
      "eval_samples_per_second": 474.176,
      "eval_steps_per_second": 7.417,
      "step": 226500
    },
    {
      "epoch": 64.29463525404485,
      "grad_norm": 0.0023058445658534765,
      "learning_rate": 3.573176270224241e-05,
      "loss": 0.0013081364333629609,
      "step": 226510
    },
    {
      "epoch": 64.29747374396821,
      "grad_norm": 0.028749050572514534,
      "learning_rate": 3.572892421231905e-05,
      "loss": 0.0003888057544827461,
      "step": 226520
    },
    {
      "epoch": 64.30031223389157,
      "grad_norm": 0.03666175156831741,
      "learning_rate": 3.5726085722395684e-05,
      "loss": 0.0007046950981020927,
      "step": 226530
    },
    {
      "epoch": 64.30315072381494,
      "grad_norm": 0.028619147837162018,
      "learning_rate": 3.572324723247233e-05,
      "loss": 0.0003707041963934898,
      "step": 226540
    },
    {
      "epoch": 64.30598921373829,
      "grad_norm": 3.833080291748047,
      "learning_rate": 3.5720408742548966e-05,
      "loss": 0.0005193129181861877,
      "step": 226550
    },
    {
      "epoch": 64.30882770366165,
      "grad_norm": 0.021174633875489235,
      "learning_rate": 3.57175702526256e-05,
      "loss": 0.0009632449597120285,
      "step": 226560
    },
    {
      "epoch": 64.31166619358501,
      "grad_norm": 0.04642188549041748,
      "learning_rate": 3.571473176270224e-05,
      "loss": 0.00011688489466905594,
      "step": 226570
    },
    {
      "epoch": 64.31450468350837,
      "grad_norm": 0.11116960644721985,
      "learning_rate": 3.5711893272778884e-05,
      "loss": 0.0009990982711315156,
      "step": 226580
    },
    {
      "epoch": 64.31734317343174,
      "grad_norm": 0.07844068855047226,
      "learning_rate": 3.5709054782855525e-05,
      "loss": 0.0007017653435468674,
      "step": 226590
    },
    {
      "epoch": 64.3201816633551,
      "grad_norm": 0.07922183722257614,
      "learning_rate": 3.570621629293216e-05,
      "loss": 0.0009196262806653976,
      "step": 226600
    },
    {
      "epoch": 64.32302015327845,
      "grad_norm": 0.0707586407661438,
      "learning_rate": 3.57033778030088e-05,
      "loss": 0.0005971027538180351,
      "step": 226610
    },
    {
      "epoch": 64.32585864320181,
      "grad_norm": 0.056553035974502563,
      "learning_rate": 3.570053931308544e-05,
      "loss": 0.0007618982344865799,
      "step": 226620
    },
    {
      "epoch": 64.32869713312517,
      "grad_norm": 0.02442164346575737,
      "learning_rate": 3.569770082316208e-05,
      "loss": 0.0022794131189584734,
      "step": 226630
    },
    {
      "epoch": 64.33153562304854,
      "grad_norm": 0.23808389902114868,
      "learning_rate": 3.569486233323872e-05,
      "loss": 0.007311782985925675,
      "step": 226640
    },
    {
      "epoch": 64.3343741129719,
      "grad_norm": 0.45268985629081726,
      "learning_rate": 3.569202384331536e-05,
      "loss": 0.0060161460191011425,
      "step": 226650
    },
    {
      "epoch": 64.33721260289526,
      "grad_norm": 0.041465748101472855,
      "learning_rate": 3.5689185353391994e-05,
      "loss": 0.00014994069933891297,
      "step": 226660
    },
    {
      "epoch": 64.34005109281863,
      "grad_norm": 12.896570205688477,
      "learning_rate": 3.5686346863468636e-05,
      "loss": 0.0032412678003311155,
      "step": 226670
    },
    {
      "epoch": 64.34288958274198,
      "grad_norm": 0.9873967170715332,
      "learning_rate": 3.568350837354528e-05,
      "loss": 0.00032231807708740237,
      "step": 226680
    },
    {
      "epoch": 64.34572807266534,
      "grad_norm": 0.05810638144612312,
      "learning_rate": 3.568066988362191e-05,
      "loss": 0.00010569561272859574,
      "step": 226690
    },
    {
      "epoch": 64.3485665625887,
      "grad_norm": 0.1932067573070526,
      "learning_rate": 3.567783139369855e-05,
      "loss": 0.00012325253337621688,
      "step": 226700
    },
    {
      "epoch": 64.35140505251206,
      "grad_norm": 0.27922871708869934,
      "learning_rate": 3.5674992903775194e-05,
      "loss": 0.0003151927143335342,
      "step": 226710
    },
    {
      "epoch": 64.35424354243543,
      "grad_norm": 0.03479532152414322,
      "learning_rate": 3.5672154413851836e-05,
      "loss": 0.0005983352661132812,
      "step": 226720
    },
    {
      "epoch": 64.35708203235879,
      "grad_norm": 0.36491966247558594,
      "learning_rate": 3.566931592392847e-05,
      "loss": 0.0003227923065423965,
      "step": 226730
    },
    {
      "epoch": 64.35992052228215,
      "grad_norm": 0.19436660408973694,
      "learning_rate": 3.566647743400511e-05,
      "loss": 0.0003984479233622551,
      "step": 226740
    },
    {
      "epoch": 64.3627590122055,
      "grad_norm": 1.1209758520126343,
      "learning_rate": 3.566363894408175e-05,
      "loss": 0.0008572598919272423,
      "step": 226750
    },
    {
      "epoch": 64.36559750212886,
      "grad_norm": 0.05188921093940735,
      "learning_rate": 3.566080045415839e-05,
      "loss": 0.00045839790254831314,
      "step": 226760
    },
    {
      "epoch": 64.36843599205223,
      "grad_norm": 0.3226931095123291,
      "learning_rate": 3.565796196423503e-05,
      "loss": 0.00035271886736154556,
      "step": 226770
    },
    {
      "epoch": 64.37127448197559,
      "grad_norm": 0.05054834857583046,
      "learning_rate": 3.565512347431167e-05,
      "loss": 0.0006363483145833016,
      "step": 226780
    },
    {
      "epoch": 64.37411297189895,
      "grad_norm": 0.037618428468704224,
      "learning_rate": 3.5652284984388305e-05,
      "loss": 0.0002492522820830345,
      "step": 226790
    },
    {
      "epoch": 64.37695146182232,
      "grad_norm": 0.12460628896951675,
      "learning_rate": 3.5649446494464946e-05,
      "loss": 0.0001988973468542099,
      "step": 226800
    },
    {
      "epoch": 64.37978995174566,
      "grad_norm": 0.0705265998840332,
      "learning_rate": 3.564660800454159e-05,
      "loss": 0.0033376879990100862,
      "step": 226810
    },
    {
      "epoch": 64.38262844166903,
      "grad_norm": 0.012924819253385067,
      "learning_rate": 3.564376951461822e-05,
      "loss": 0.0004001487046480179,
      "step": 226820
    },
    {
      "epoch": 64.38546693159239,
      "grad_norm": 0.03740304335951805,
      "learning_rate": 3.5640931024694864e-05,
      "loss": 0.00034913234412670135,
      "step": 226830
    },
    {
      "epoch": 64.38830542151575,
      "grad_norm": 0.12138964235782623,
      "learning_rate": 3.5638092534771505e-05,
      "loss": 0.00036101825535297396,
      "step": 226840
    },
    {
      "epoch": 64.39114391143912,
      "grad_norm": 0.016875706613063812,
      "learning_rate": 3.563525404484814e-05,
      "loss": 0.00023956168442964553,
      "step": 226850
    },
    {
      "epoch": 64.39398240136248,
      "grad_norm": 0.017406277358531952,
      "learning_rate": 3.563241555492478e-05,
      "loss": 0.0029905539005994795,
      "step": 226860
    },
    {
      "epoch": 64.39682089128584,
      "grad_norm": 0.05565466359257698,
      "learning_rate": 3.5629577065001416e-05,
      "loss": 0.0001872614026069641,
      "step": 226870
    },
    {
      "epoch": 64.39965938120919,
      "grad_norm": 0.07277591526508331,
      "learning_rate": 3.5626738575078064e-05,
      "loss": 0.00014880988746881484,
      "step": 226880
    },
    {
      "epoch": 64.40249787113255,
      "grad_norm": 0.04102430120110512,
      "learning_rate": 3.56239000851547e-05,
      "loss": 0.0012762289494276046,
      "step": 226890
    },
    {
      "epoch": 64.40533636105592,
      "grad_norm": 0.04281811788678169,
      "learning_rate": 3.562106159523134e-05,
      "loss": 0.0009597392752766609,
      "step": 226900
    },
    {
      "epoch": 64.40817485097928,
      "grad_norm": 0.010992499068379402,
      "learning_rate": 3.561822310530798e-05,
      "loss": 0.00015740711241960526,
      "step": 226910
    },
    {
      "epoch": 64.41101334090264,
      "grad_norm": 0.11131087690591812,
      "learning_rate": 3.5615384615384616e-05,
      "loss": 0.0005279656499624253,
      "step": 226920
    },
    {
      "epoch": 64.413851830826,
      "grad_norm": 9.382285118103027,
      "learning_rate": 3.561254612546126e-05,
      "loss": 0.002839912474155426,
      "step": 226930
    },
    {
      "epoch": 64.41669032074937,
      "grad_norm": 0.9375768899917603,
      "learning_rate": 3.56097076355379e-05,
      "loss": 0.0002333007752895355,
      "step": 226940
    },
    {
      "epoch": 64.41952881067272,
      "grad_norm": 0.14448152482509613,
      "learning_rate": 3.560686914561453e-05,
      "loss": 0.00012990422546863556,
      "step": 226950
    },
    {
      "epoch": 64.42236730059608,
      "grad_norm": 0.021059956401586533,
      "learning_rate": 3.5604030655691175e-05,
      "loss": 0.0009376503527164459,
      "step": 226960
    },
    {
      "epoch": 64.42520579051944,
      "grad_norm": 0.041144099086523056,
      "learning_rate": 3.560119216576781e-05,
      "loss": 0.0002996537834405899,
      "step": 226970
    },
    {
      "epoch": 64.4280442804428,
      "grad_norm": 2.10440993309021,
      "learning_rate": 3.559835367584445e-05,
      "loss": 0.0013378538191318513,
      "step": 226980
    },
    {
      "epoch": 64.43088277036617,
      "grad_norm": 0.03850625827908516,
      "learning_rate": 3.559551518592109e-05,
      "loss": 0.0006754685193300248,
      "step": 226990
    },
    {
      "epoch": 64.43372126028953,
      "grad_norm": 0.010102166794240475,
      "learning_rate": 3.5592676695997726e-05,
      "loss": 0.0004506360739469528,
      "step": 227000
    },
    {
      "epoch": 64.43372126028953,
      "eval_accuracy": 0.9779360335728365,
      "eval_loss": 0.0847136378288269,
      "eval_runtime": 32.4832,
      "eval_samples_per_second": 484.159,
      "eval_steps_per_second": 7.573,
      "step": 227000
    },
    {
      "epoch": 64.4365597502129,
      "grad_norm": 0.12759935855865479,
      "learning_rate": 3.5589838206074375e-05,
      "loss": 0.0005242889747023582,
      "step": 227010
    },
    {
      "epoch": 64.43939824013624,
      "grad_norm": 0.0180843286216259,
      "learning_rate": 3.558699971615101e-05,
      "loss": 0.0009631512686610222,
      "step": 227020
    },
    {
      "epoch": 64.4422367300596,
      "grad_norm": 0.08088906109333038,
      "learning_rate": 3.5584161226227644e-05,
      "loss": 0.001322108879685402,
      "step": 227030
    },
    {
      "epoch": 64.44507521998297,
      "grad_norm": 0.04792669788002968,
      "learning_rate": 3.558132273630429e-05,
      "loss": 0.0066739052534103395,
      "step": 227040
    },
    {
      "epoch": 64.44791370990633,
      "grad_norm": 0.11347947269678116,
      "learning_rate": 3.5578484246380927e-05,
      "loss": 0.002053220756351948,
      "step": 227050
    },
    {
      "epoch": 64.4507521998297,
      "grad_norm": 0.04892696067690849,
      "learning_rate": 3.557564575645757e-05,
      "loss": 0.0009788483381271361,
      "step": 227060
    },
    {
      "epoch": 64.45359068975306,
      "grad_norm": 0.01286416593939066,
      "learning_rate": 3.557280726653421e-05,
      "loss": 0.0008971275761723519,
      "step": 227070
    },
    {
      "epoch": 64.4564291796764,
      "grad_norm": 0.03952132165431976,
      "learning_rate": 3.5569968776610844e-05,
      "loss": 0.0003102939575910568,
      "step": 227080
    },
    {
      "epoch": 64.45926766959977,
      "grad_norm": 0.013046781532466412,
      "learning_rate": 3.5567130286687485e-05,
      "loss": 0.0016698693856596948,
      "step": 227090
    },
    {
      "epoch": 64.46210615952313,
      "grad_norm": 0.007003657054156065,
      "learning_rate": 3.556429179676412e-05,
      "loss": 0.0011827267706394195,
      "step": 227100
    },
    {
      "epoch": 64.4649446494465,
      "grad_norm": 0.04076068475842476,
      "learning_rate": 3.556145330684076e-05,
      "loss": 0.004574423283338546,
      "step": 227110
    },
    {
      "epoch": 64.46778313936986,
      "grad_norm": 0.03721284121274948,
      "learning_rate": 3.55586148169174e-05,
      "loss": 0.00030366312712430953,
      "step": 227120
    },
    {
      "epoch": 64.47062162929322,
      "grad_norm": 0.344247967004776,
      "learning_rate": 3.555577632699404e-05,
      "loss": 0.00024676043540239334,
      "step": 227130
    },
    {
      "epoch": 64.47346011921658,
      "grad_norm": 0.11584793031215668,
      "learning_rate": 3.5552937837070685e-05,
      "loss": 0.00033060908317565917,
      "step": 227140
    },
    {
      "epoch": 64.47629860913993,
      "grad_norm": 0.2506316006183624,
      "learning_rate": 3.555009934714732e-05,
      "loss": 0.000381021574139595,
      "step": 227150
    },
    {
      "epoch": 64.4791370990633,
      "grad_norm": 0.009774028323590755,
      "learning_rate": 3.5547260857223955e-05,
      "loss": 0.00024173445999622344,
      "step": 227160
    },
    {
      "epoch": 64.48197558898666,
      "grad_norm": 0.027842149138450623,
      "learning_rate": 3.55444223673006e-05,
      "loss": 0.000182328000664711,
      "step": 227170
    },
    {
      "epoch": 64.48481407891002,
      "grad_norm": 0.14081759750843048,
      "learning_rate": 3.554158387737724e-05,
      "loss": 0.00017489120364189148,
      "step": 227180
    },
    {
      "epoch": 64.48765256883338,
      "grad_norm": 0.03934396430850029,
      "learning_rate": 3.553874538745388e-05,
      "loss": 0.0002364378422498703,
      "step": 227190
    },
    {
      "epoch": 64.49049105875675,
      "grad_norm": 0.011172533966600895,
      "learning_rate": 3.553590689753051e-05,
      "loss": 0.0003931697458028793,
      "step": 227200
    },
    {
      "epoch": 64.49332954868011,
      "grad_norm": 0.004419004078954458,
      "learning_rate": 3.5533068407607155e-05,
      "loss": 0.0009636769071221351,
      "step": 227210
    },
    {
      "epoch": 64.49616803860346,
      "grad_norm": 2.2440853118896484,
      "learning_rate": 3.5530229917683796e-05,
      "loss": 0.0003806058317422867,
      "step": 227220
    },
    {
      "epoch": 64.49900652852682,
      "grad_norm": 0.023585433140397072,
      "learning_rate": 3.552739142776043e-05,
      "loss": 0.0008241184055805207,
      "step": 227230
    },
    {
      "epoch": 64.50184501845018,
      "grad_norm": 0.058894336223602295,
      "learning_rate": 3.552455293783707e-05,
      "loss": 0.0014057930558919907,
      "step": 227240
    },
    {
      "epoch": 64.50468350837355,
      "grad_norm": 0.3639346659183502,
      "learning_rate": 3.5521714447913713e-05,
      "loss": 0.0007924098521471023,
      "step": 227250
    },
    {
      "epoch": 64.50752199829691,
      "grad_norm": 0.3888349235057831,
      "learning_rate": 3.551887595799035e-05,
      "loss": 0.0003626767545938492,
      "step": 227260
    },
    {
      "epoch": 64.51036048822027,
      "grad_norm": 0.13349321484565735,
      "learning_rate": 3.551603746806699e-05,
      "loss": 0.0005032308399677277,
      "step": 227270
    },
    {
      "epoch": 64.51319897814363,
      "grad_norm": 0.009500262327492237,
      "learning_rate": 3.551319897814363e-05,
      "loss": 0.00012325141578912736,
      "step": 227280
    },
    {
      "epoch": 64.51603746806698,
      "grad_norm": 0.015890706330537796,
      "learning_rate": 3.5510360488220265e-05,
      "loss": 0.00012912489473819732,
      "step": 227290
    },
    {
      "epoch": 64.51887595799035,
      "grad_norm": 0.023742029443383217,
      "learning_rate": 3.550752199829691e-05,
      "loss": 0.00023947730660438537,
      "step": 227300
    },
    {
      "epoch": 64.52171444791371,
      "grad_norm": 9.8718900680542,
      "learning_rate": 3.550468350837355e-05,
      "loss": 0.0035428866744041444,
      "step": 227310
    },
    {
      "epoch": 64.52455293783707,
      "grad_norm": 0.41373974084854126,
      "learning_rate": 3.550184501845018e-05,
      "loss": 0.0009513290598988533,
      "step": 227320
    },
    {
      "epoch": 64.52739142776043,
      "grad_norm": 0.01422911137342453,
      "learning_rate": 3.5499006528526824e-05,
      "loss": 0.00010710060596466064,
      "step": 227330
    },
    {
      "epoch": 64.5302299176838,
      "grad_norm": 0.010934261605143547,
      "learning_rate": 3.5496168038603465e-05,
      "loss": 0.0017043884843587875,
      "step": 227340
    },
    {
      "epoch": 64.53306840760715,
      "grad_norm": 0.1288539618253708,
      "learning_rate": 3.549332954868011e-05,
      "loss": 0.0002602929249405861,
      "step": 227350
    },
    {
      "epoch": 64.53590689753051,
      "grad_norm": 0.06147071719169617,
      "learning_rate": 3.549049105875674e-05,
      "loss": 0.00019090678542852402,
      "step": 227360
    },
    {
      "epoch": 64.53874538745387,
      "grad_norm": 0.023164156824350357,
      "learning_rate": 3.548765256883338e-05,
      "loss": 7.439423352479934e-05,
      "step": 227370
    },
    {
      "epoch": 64.54158387737724,
      "grad_norm": 0.01951715350151062,
      "learning_rate": 3.5484814078910024e-05,
      "loss": 0.00026624277234077454,
      "step": 227380
    },
    {
      "epoch": 64.5444223673006,
      "grad_norm": 0.043041449040174484,
      "learning_rate": 3.548197558898666e-05,
      "loss": 0.00014764182269573213,
      "step": 227390
    },
    {
      "epoch": 64.54726085722396,
      "grad_norm": 0.6787921190261841,
      "learning_rate": 3.54791370990633e-05,
      "loss": 0.0003999501466751099,
      "step": 227400
    },
    {
      "epoch": 64.55009934714732,
      "grad_norm": 0.14617577195167542,
      "learning_rate": 3.547629860913994e-05,
      "loss": 0.00032883621752262117,
      "step": 227410
    },
    {
      "epoch": 64.55293783707067,
      "grad_norm": 0.017888391390442848,
      "learning_rate": 3.5473460119216576e-05,
      "loss": 0.0004807775840163231,
      "step": 227420
    },
    {
      "epoch": 64.55577632699404,
      "grad_norm": 4.158189296722412,
      "learning_rate": 3.547062162929322e-05,
      "loss": 0.0006672345101833343,
      "step": 227430
    },
    {
      "epoch": 64.5586148169174,
      "grad_norm": 0.010387646965682507,
      "learning_rate": 3.546778313936986e-05,
      "loss": 0.00028142724186182023,
      "step": 227440
    },
    {
      "epoch": 64.56145330684076,
      "grad_norm": 0.024116594344377518,
      "learning_rate": 3.5464944649446493e-05,
      "loss": 0.0030637938529253007,
      "step": 227450
    },
    {
      "epoch": 64.56429179676412,
      "grad_norm": 0.03489859402179718,
      "learning_rate": 3.5462106159523135e-05,
      "loss": 0.0025845248252153396,
      "step": 227460
    },
    {
      "epoch": 64.56713028668749,
      "grad_norm": 0.013325735926628113,
      "learning_rate": 3.5459267669599776e-05,
      "loss": 0.000940852239727974,
      "step": 227470
    },
    {
      "epoch": 64.56996877661085,
      "grad_norm": 0.18206597864627838,
      "learning_rate": 3.545642917967642e-05,
      "loss": 0.001006605662405491,
      "step": 227480
    },
    {
      "epoch": 64.5728072665342,
      "grad_norm": 1.6445246934890747,
      "learning_rate": 3.545359068975305e-05,
      "loss": 0.0027904415503144262,
      "step": 227490
    },
    {
      "epoch": 64.57564575645756,
      "grad_norm": 0.056534599512815475,
      "learning_rate": 3.545075219982969e-05,
      "loss": 0.00031729955226182937,
      "step": 227500
    },
    {
      "epoch": 64.57564575645756,
      "eval_accuracy": 0.9781903732434667,
      "eval_loss": 0.0824812799692154,
      "eval_runtime": 33.2725,
      "eval_samples_per_second": 472.673,
      "eval_steps_per_second": 7.394,
      "step": 227500
    },
    {
      "epoch": 64.57848424638092,
      "grad_norm": 6.02877140045166,
      "learning_rate": 3.5447913709906335e-05,
      "loss": 0.001592763513326645,
      "step": 227510
    },
    {
      "epoch": 64.58132273630429,
      "grad_norm": 0.036787260323762894,
      "learning_rate": 3.544507521998297e-05,
      "loss": 0.0017896093428134917,
      "step": 227520
    },
    {
      "epoch": 64.58416122622765,
      "grad_norm": 0.10064104199409485,
      "learning_rate": 3.544223673005961e-05,
      "loss": 0.0010635115206241608,
      "step": 227530
    },
    {
      "epoch": 64.58699971615101,
      "grad_norm": 0.014023300260305405,
      "learning_rate": 3.543939824013625e-05,
      "loss": 0.004633621126413345,
      "step": 227540
    },
    {
      "epoch": 64.58983820607436,
      "grad_norm": 0.06034943833947182,
      "learning_rate": 3.543655975021289e-05,
      "loss": 0.005195567011833191,
      "step": 227550
    },
    {
      "epoch": 64.59267669599772,
      "grad_norm": 0.23157483339309692,
      "learning_rate": 3.543372126028953e-05,
      "loss": 0.00028497520834207534,
      "step": 227560
    },
    {
      "epoch": 64.59551518592109,
      "grad_norm": 0.0328245609998703,
      "learning_rate": 3.543088277036617e-05,
      "loss": 0.000813632644712925,
      "step": 227570
    },
    {
      "epoch": 64.59835367584445,
      "grad_norm": 0.018730374053120613,
      "learning_rate": 3.5428044280442804e-05,
      "loss": 0.0003656305372714996,
      "step": 227580
    },
    {
      "epoch": 64.60119216576781,
      "grad_norm": 2.44191312789917,
      "learning_rate": 3.5425205790519446e-05,
      "loss": 0.000599067471921444,
      "step": 227590
    },
    {
      "epoch": 64.60403065569118,
      "grad_norm": 0.024014819413423538,
      "learning_rate": 3.542236730059608e-05,
      "loss": 0.006177405267953873,
      "step": 227600
    },
    {
      "epoch": 64.60686914561454,
      "grad_norm": 0.02569492533802986,
      "learning_rate": 3.541952881067273e-05,
      "loss": 0.0004070069640874863,
      "step": 227610
    },
    {
      "epoch": 64.60970763553789,
      "grad_norm": 0.11033676564693451,
      "learning_rate": 3.541669032074936e-05,
      "loss": 0.006826505064964294,
      "step": 227620
    },
    {
      "epoch": 64.61254612546125,
      "grad_norm": 0.23225270211696625,
      "learning_rate": 3.5413851830826e-05,
      "loss": 0.0003704478964209557,
      "step": 227630
    },
    {
      "epoch": 64.61538461538461,
      "grad_norm": 0.042143069207668304,
      "learning_rate": 3.5411013340902646e-05,
      "loss": 0.0008974594995379448,
      "step": 227640
    },
    {
      "epoch": 64.61822310530798,
      "grad_norm": 0.24286215007305145,
      "learning_rate": 3.540817485097928e-05,
      "loss": 0.0005177156999707222,
      "step": 227650
    },
    {
      "epoch": 64.62106159523134,
      "grad_norm": 0.14003713428974152,
      "learning_rate": 3.540533636105592e-05,
      "loss": 0.0005303861573338509,
      "step": 227660
    },
    {
      "epoch": 64.6239000851547,
      "grad_norm": 0.23829707503318787,
      "learning_rate": 3.540249787113256e-05,
      "loss": 0.000249442458152771,
      "step": 227670
    },
    {
      "epoch": 64.62673857507806,
      "grad_norm": 0.016767079010605812,
      "learning_rate": 3.53996593812092e-05,
      "loss": 0.000661054439842701,
      "step": 227680
    },
    {
      "epoch": 64.62957706500141,
      "grad_norm": 0.07220937311649323,
      "learning_rate": 3.539682089128584e-05,
      "loss": 0.01052992269396782,
      "step": 227690
    },
    {
      "epoch": 64.63241555492478,
      "grad_norm": 18.694955825805664,
      "learning_rate": 3.5393982401362474e-05,
      "loss": 0.00579703152179718,
      "step": 227700
    },
    {
      "epoch": 64.63525404484814,
      "grad_norm": 0.02216956950724125,
      "learning_rate": 3.5391143911439115e-05,
      "loss": 0.0010020527988672256,
      "step": 227710
    },
    {
      "epoch": 64.6380925347715,
      "grad_norm": 0.29641762375831604,
      "learning_rate": 3.5388305421515756e-05,
      "loss": 0.0022799234837293627,
      "step": 227720
    },
    {
      "epoch": 64.64093102469486,
      "grad_norm": 0.04273602366447449,
      "learning_rate": 3.538546693159239e-05,
      "loss": 0.00022718701511621476,
      "step": 227730
    },
    {
      "epoch": 64.64376951461823,
      "grad_norm": 0.09753295034170151,
      "learning_rate": 3.538262844166903e-05,
      "loss": 0.00011789780110120774,
      "step": 227740
    },
    {
      "epoch": 64.64660800454159,
      "grad_norm": 0.3434138298034668,
      "learning_rate": 3.5379789951745674e-05,
      "loss": 0.00020439419895410538,
      "step": 227750
    },
    {
      "epoch": 64.64944649446494,
      "grad_norm": 0.3323395252227783,
      "learning_rate": 3.537695146182231e-05,
      "loss": 0.00035300571471452713,
      "step": 227760
    },
    {
      "epoch": 64.6522849843883,
      "grad_norm": 0.04733119532465935,
      "learning_rate": 3.5374112971898956e-05,
      "loss": 0.0005828278139233589,
      "step": 227770
    },
    {
      "epoch": 64.65512347431167,
      "grad_norm": 0.21523348987102509,
      "learning_rate": 3.537127448197559e-05,
      "loss": 0.0005199907347559929,
      "step": 227780
    },
    {
      "epoch": 64.65796196423503,
      "grad_norm": 0.021186986938118935,
      "learning_rate": 3.5368435992052226e-05,
      "loss": 0.0001672951504588127,
      "step": 227790
    },
    {
      "epoch": 64.66080045415839,
      "grad_norm": 0.008240382187068462,
      "learning_rate": 3.536559750212887e-05,
      "loss": 0.0029375005513429643,
      "step": 227800
    },
    {
      "epoch": 64.66363894408175,
      "grad_norm": 0.01379274483770132,
      "learning_rate": 3.536275901220551e-05,
      "loss": 7.382091134786606e-05,
      "step": 227810
    },
    {
      "epoch": 64.6664774340051,
      "grad_norm": 0.12929123640060425,
      "learning_rate": 3.535992052228215e-05,
      "loss": 0.00016041845083236694,
      "step": 227820
    },
    {
      "epoch": 64.66931592392847,
      "grad_norm": 0.03246399760246277,
      "learning_rate": 3.5357082032358784e-05,
      "loss": 0.00022974200546741486,
      "step": 227830
    },
    {
      "epoch": 64.67215441385183,
      "grad_norm": 0.009126763790845871,
      "learning_rate": 3.5354243542435426e-05,
      "loss": 0.00011737309396266937,
      "step": 227840
    },
    {
      "epoch": 64.67499290377519,
      "grad_norm": 0.018399422988295555,
      "learning_rate": 3.535140505251207e-05,
      "loss": 0.00013364888727664947,
      "step": 227850
    },
    {
      "epoch": 64.67783139369855,
      "grad_norm": 0.10503066331148148,
      "learning_rate": 3.53485665625887e-05,
      "loss": 0.0005435882136225701,
      "step": 227860
    },
    {
      "epoch": 64.68066988362192,
      "grad_norm": 0.08922678232192993,
      "learning_rate": 3.534572807266534e-05,
      "loss": 0.009031365811824798,
      "step": 227870
    },
    {
      "epoch": 64.68350837354528,
      "grad_norm": 0.005905136466026306,
      "learning_rate": 3.5342889582741984e-05,
      "loss": 0.0002526151016354561,
      "step": 227880
    },
    {
      "epoch": 64.68634686346863,
      "grad_norm": 0.022331571206450462,
      "learning_rate": 3.534005109281862e-05,
      "loss": 0.0002189716324210167,
      "step": 227890
    },
    {
      "epoch": 64.68918535339199,
      "grad_norm": 0.09813796728849411,
      "learning_rate": 3.533721260289526e-05,
      "loss": 0.0001441851258277893,
      "step": 227900
    },
    {
      "epoch": 64.69202384331535,
      "grad_norm": 0.008964224718511105,
      "learning_rate": 3.53343741129719e-05,
      "loss": 0.0002774640917778015,
      "step": 227910
    },
    {
      "epoch": 64.69486233323872,
      "grad_norm": 0.05893057957291603,
      "learning_rate": 3.5331535623048536e-05,
      "loss": 0.00014841370284557343,
      "step": 227920
    },
    {
      "epoch": 64.69770082316208,
      "grad_norm": 0.10634230822324753,
      "learning_rate": 3.532869713312518e-05,
      "loss": 0.00013045892119407655,
      "step": 227930
    },
    {
      "epoch": 64.70053931308544,
      "grad_norm": 0.11731266975402832,
      "learning_rate": 3.532585864320182e-05,
      "loss": 0.00023052543401718139,
      "step": 227940
    },
    {
      "epoch": 64.7033778030088,
      "grad_norm": 0.1258549839258194,
      "learning_rate": 3.532302015327846e-05,
      "loss": 0.00013660378754138948,
      "step": 227950
    },
    {
      "epoch": 64.70621629293215,
      "grad_norm": 0.027348896488547325,
      "learning_rate": 3.5320181663355095e-05,
      "loss": 0.00021154098212718963,
      "step": 227960
    },
    {
      "epoch": 64.70905478285552,
      "grad_norm": 0.017850887030363083,
      "learning_rate": 3.5317343173431737e-05,
      "loss": 0.0002308053895831108,
      "step": 227970
    },
    {
      "epoch": 64.71189327277888,
      "grad_norm": 0.06431466341018677,
      "learning_rate": 3.531450468350838e-05,
      "loss": 0.00019279178231954575,
      "step": 227980
    },
    {
      "epoch": 64.71473176270224,
      "grad_norm": 0.03229772299528122,
      "learning_rate": 3.531166619358501e-05,
      "loss": 0.00014930013567209243,
      "step": 227990
    },
    {
      "epoch": 64.7175702526256,
      "grad_norm": 0.10112423449754715,
      "learning_rate": 3.5308827703661654e-05,
      "loss": 0.0001798415556550026,
      "step": 228000
    },
    {
      "epoch": 64.7175702526256,
      "eval_accuracy": 0.9802250906085077,
      "eval_loss": 0.07291392236948013,
      "eval_runtime": 32.9945,
      "eval_samples_per_second": 476.655,
      "eval_steps_per_second": 7.456,
      "step": 228000
    },
    {
      "epoch": 64.72040874254897,
      "grad_norm": 0.08900847285985947,
      "learning_rate": 3.5305989213738295e-05,
      "loss": 8.714385330677032e-05,
      "step": 228010
    },
    {
      "epoch": 64.72324723247232,
      "grad_norm": 0.04261748865246773,
      "learning_rate": 3.530315072381493e-05,
      "loss": 0.00013610608875751494,
      "step": 228020
    },
    {
      "epoch": 64.72608572239568,
      "grad_norm": 1.2438918352127075,
      "learning_rate": 3.530031223389157e-05,
      "loss": 0.00025221072137355806,
      "step": 228030
    },
    {
      "epoch": 64.72892421231904,
      "grad_norm": 0.020528972148895264,
      "learning_rate": 3.529747374396821e-05,
      "loss": 0.00025026053190231324,
      "step": 228040
    },
    {
      "epoch": 64.7317627022424,
      "grad_norm": 0.024411732330918312,
      "learning_rate": 3.529463525404485e-05,
      "loss": 0.00010906551033258438,
      "step": 228050
    },
    {
      "epoch": 64.73460119216577,
      "grad_norm": 0.00716741569340229,
      "learning_rate": 3.529179676412149e-05,
      "loss": 0.0001208556815981865,
      "step": 228060
    },
    {
      "epoch": 64.73743968208913,
      "grad_norm": 1.1603209972381592,
      "learning_rate": 3.528895827419813e-05,
      "loss": 0.0005732202902436256,
      "step": 228070
    },
    {
      "epoch": 64.7402781720125,
      "grad_norm": 0.014557943679392338,
      "learning_rate": 3.528611978427477e-05,
      "loss": 0.00025891028344631194,
      "step": 228080
    },
    {
      "epoch": 64.74311666193584,
      "grad_norm": 1.268306016921997,
      "learning_rate": 3.5283281294351406e-05,
      "loss": 0.000368807278573513,
      "step": 228090
    },
    {
      "epoch": 64.7459551518592,
      "grad_norm": 0.29416602849960327,
      "learning_rate": 3.528044280442804e-05,
      "loss": 0.0003156732767820358,
      "step": 228100
    },
    {
      "epoch": 64.74879364178257,
      "grad_norm": 0.9271993041038513,
      "learning_rate": 3.527760431450469e-05,
      "loss": 0.004461620002985001,
      "step": 228110
    },
    {
      "epoch": 64.75163213170593,
      "grad_norm": 0.03562605753540993,
      "learning_rate": 3.527476582458132e-05,
      "loss": 0.00035455580800771715,
      "step": 228120
    },
    {
      "epoch": 64.7544706216293,
      "grad_norm": 0.03544740378856659,
      "learning_rate": 3.5271927334657965e-05,
      "loss": 0.0014394694939255715,
      "step": 228130
    },
    {
      "epoch": 64.75730911155266,
      "grad_norm": 0.7851232886314392,
      "learning_rate": 3.5269088844734606e-05,
      "loss": 0.00033451486378908155,
      "step": 228140
    },
    {
      "epoch": 64.76014760147602,
      "grad_norm": 0.17049427330493927,
      "learning_rate": 3.526625035481124e-05,
      "loss": 0.00018730293959379196,
      "step": 228150
    },
    {
      "epoch": 64.76298609139937,
      "grad_norm": 0.011601299978792667,
      "learning_rate": 3.526341186488788e-05,
      "loss": 0.000165439210832119,
      "step": 228160
    },
    {
      "epoch": 64.76582458132273,
      "grad_norm": 0.02351550944149494,
      "learning_rate": 3.526057337496452e-05,
      "loss": 0.00011493787169456482,
      "step": 228170
    },
    {
      "epoch": 64.7686630712461,
      "grad_norm": 0.1096300259232521,
      "learning_rate": 3.525773488504116e-05,
      "loss": 9.328555315732956e-05,
      "step": 228180
    },
    {
      "epoch": 64.77150156116946,
      "grad_norm": 0.02514713630080223,
      "learning_rate": 3.52548963951178e-05,
      "loss": 0.0001400751993060112,
      "step": 228190
    },
    {
      "epoch": 64.77434005109282,
      "grad_norm": 0.011365020647644997,
      "learning_rate": 3.5252057905194434e-05,
      "loss": 0.00016196947544813156,
      "step": 228200
    },
    {
      "epoch": 64.77717854101618,
      "grad_norm": 0.09121741354465485,
      "learning_rate": 3.5249219415271075e-05,
      "loss": 0.00020008999854326248,
      "step": 228210
    },
    {
      "epoch": 64.78001703093955,
      "grad_norm": 0.024293268099427223,
      "learning_rate": 3.524638092534772e-05,
      "loss": 9.87498089671135e-05,
      "step": 228220
    },
    {
      "epoch": 64.7828555208629,
      "grad_norm": 0.32961133122444153,
      "learning_rate": 3.524354243542435e-05,
      "loss": 0.00041998010128736497,
      "step": 228230
    },
    {
      "epoch": 64.78569401078626,
      "grad_norm": 0.03152919188141823,
      "learning_rate": 3.5240703945501e-05,
      "loss": 0.00010145101696252823,
      "step": 228240
    },
    {
      "epoch": 64.78853250070962,
      "grad_norm": 0.006745504215359688,
      "learning_rate": 3.5237865455577634e-05,
      "loss": 0.00013104677200317382,
      "step": 228250
    },
    {
      "epoch": 64.79137099063298,
      "grad_norm": 0.01531310472637415,
      "learning_rate": 3.523502696565427e-05,
      "loss": 0.0005022125318646431,
      "step": 228260
    },
    {
      "epoch": 64.79420948055635,
      "grad_norm": 0.246595099568367,
      "learning_rate": 3.523218847573092e-05,
      "loss": 0.00013446249067783356,
      "step": 228270
    },
    {
      "epoch": 64.79704797047971,
      "grad_norm": 0.9065418839454651,
      "learning_rate": 3.522934998580755e-05,
      "loss": 0.004447096213698387,
      "step": 228280
    },
    {
      "epoch": 64.79988646040306,
      "grad_norm": 0.04589758813381195,
      "learning_rate": 3.522651149588419e-05,
      "loss": 0.0002563228830695152,
      "step": 228290
    },
    {
      "epoch": 64.80272495032642,
      "grad_norm": 0.019310442730784416,
      "learning_rate": 3.522367300596083e-05,
      "loss": 0.00034010689705610275,
      "step": 228300
    },
    {
      "epoch": 64.80556344024978,
      "grad_norm": 0.14963369071483612,
      "learning_rate": 3.522083451603747e-05,
      "loss": 0.00018470417708158494,
      "step": 228310
    },
    {
      "epoch": 64.80840193017315,
      "grad_norm": 2.7597739696502686,
      "learning_rate": 3.521799602611411e-05,
      "loss": 0.0008137373253703118,
      "step": 228320
    },
    {
      "epoch": 64.81124042009651,
      "grad_norm": 0.20758791267871857,
      "learning_rate": 3.5215157536190745e-05,
      "loss": 0.0013518983498215676,
      "step": 228330
    },
    {
      "epoch": 64.81407891001987,
      "grad_norm": 0.05411308631300926,
      "learning_rate": 3.5212319046267386e-05,
      "loss": 0.0012829853221774101,
      "step": 228340
    },
    {
      "epoch": 64.81691739994324,
      "grad_norm": 0.029415912926197052,
      "learning_rate": 3.520948055634403e-05,
      "loss": 0.00015726890414953233,
      "step": 228350
    },
    {
      "epoch": 64.81975588986658,
      "grad_norm": 0.061615124344825745,
      "learning_rate": 3.5206925915413e-05,
      "loss": 0.005569319427013397,
      "step": 228360
    },
    {
      "epoch": 64.82259437978995,
      "grad_norm": 0.07015086710453033,
      "learning_rate": 3.520408742548964e-05,
      "loss": 0.00023839082568883897,
      "step": 228370
    },
    {
      "epoch": 64.82543286971331,
      "grad_norm": 0.3058590590953827,
      "learning_rate": 3.5201248935566277e-05,
      "loss": 0.002316715009510517,
      "step": 228380
    },
    {
      "epoch": 64.82827135963667,
      "grad_norm": 0.02330838143825531,
      "learning_rate": 3.519841044564292e-05,
      "loss": 0.0004554379731416702,
      "step": 228390
    },
    {
      "epoch": 64.83110984956004,
      "grad_norm": 0.02738683484494686,
      "learning_rate": 3.519557195571956e-05,
      "loss": 0.00012701880186796188,
      "step": 228400
    },
    {
      "epoch": 64.8339483394834,
      "grad_norm": 0.12166569381952286,
      "learning_rate": 3.5192733465796194e-05,
      "loss": 0.00044478066265583036,
      "step": 228410
    },
    {
      "epoch": 64.83678682940676,
      "grad_norm": 0.04078010097146034,
      "learning_rate": 3.518989497587284e-05,
      "loss": 0.00025309715420007706,
      "step": 228420
    },
    {
      "epoch": 64.83962531933011,
      "grad_norm": 0.0959625095129013,
      "learning_rate": 3.518705648594948e-05,
      "loss": 0.00013638250529766083,
      "step": 228430
    },
    {
      "epoch": 64.84246380925347,
      "grad_norm": 0.0832049548625946,
      "learning_rate": 3.518421799602611e-05,
      "loss": 0.00046504531055688856,
      "step": 228440
    },
    {
      "epoch": 64.84530229917684,
      "grad_norm": 0.008345777168869972,
      "learning_rate": 3.518137950610276e-05,
      "loss": 0.0006667375564575195,
      "step": 228450
    },
    {
      "epoch": 64.8481407891002,
      "grad_norm": 0.20776358246803284,
      "learning_rate": 3.5178541016179394e-05,
      "loss": 0.00013624019920825957,
      "step": 228460
    },
    {
      "epoch": 64.85097927902356,
      "grad_norm": 0.16698192059993744,
      "learning_rate": 3.5175702526256035e-05,
      "loss": 0.0001493966206908226,
      "step": 228470
    },
    {
      "epoch": 64.85381776894693,
      "grad_norm": 0.05590372532606125,
      "learning_rate": 3.517286403633267e-05,
      "loss": 0.0002258399501442909,
      "step": 228480
    },
    {
      "epoch": 64.85665625887029,
      "grad_norm": 0.08549810200929642,
      "learning_rate": 3.517002554640931e-05,
      "loss": 0.00024744980037212373,
      "step": 228490
    },
    {
      "epoch": 64.85949474879364,
      "grad_norm": 0.08570590615272522,
      "learning_rate": 3.516718705648595e-05,
      "loss": 0.00013787485659122467,
      "step": 228500
    },
    {
      "epoch": 64.85949474879364,
      "eval_accuracy": 0.9793349017613022,
      "eval_loss": 0.08059171587228775,
      "eval_runtime": 32.6254,
      "eval_samples_per_second": 482.048,
      "eval_steps_per_second": 7.54,
      "step": 228500
    },
    {
      "epoch": 64.862333238717,
      "grad_norm": 0.006238734349608421,
      "learning_rate": 3.516434856656259e-05,
      "loss": 0.00011661574244499206,
      "step": 228510
    },
    {
      "epoch": 64.86517172864036,
      "grad_norm": 0.04444466903805733,
      "learning_rate": 3.516151007663923e-05,
      "loss": 9.542964398860932e-05,
      "step": 228520
    },
    {
      "epoch": 64.86801021856373,
      "grad_norm": 0.004281532485038042,
      "learning_rate": 3.515867158671587e-05,
      "loss": 8.04506242275238e-05,
      "step": 228530
    },
    {
      "epoch": 64.87084870848709,
      "grad_norm": 0.0077082389034330845,
      "learning_rate": 3.5155833096792505e-05,
      "loss": 0.00014811176806688308,
      "step": 228540
    },
    {
      "epoch": 64.87368719841045,
      "grad_norm": 0.015262167900800705,
      "learning_rate": 3.515299460686915e-05,
      "loss": 0.0005406780168414116,
      "step": 228550
    },
    {
      "epoch": 64.8765256883338,
      "grad_norm": 0.011923297308385372,
      "learning_rate": 3.515015611694579e-05,
      "loss": 0.0001754138618707657,
      "step": 228560
    },
    {
      "epoch": 64.87936417825716,
      "grad_norm": 0.15329918265342712,
      "learning_rate": 3.514731762702242e-05,
      "loss": 0.0013900475576519966,
      "step": 228570
    },
    {
      "epoch": 64.88220266818053,
      "grad_norm": 7.012753486633301,
      "learning_rate": 3.5144479137099063e-05,
      "loss": 0.0025579500943422318,
      "step": 228580
    },
    {
      "epoch": 64.88504115810389,
      "grad_norm": 1.0779067277908325,
      "learning_rate": 3.5141640647175705e-05,
      "loss": 0.0003856310620903969,
      "step": 228590
    },
    {
      "epoch": 64.88787964802725,
      "grad_norm": 0.014730863273143768,
      "learning_rate": 3.5138802157252346e-05,
      "loss": 0.003830450028181076,
      "step": 228600
    },
    {
      "epoch": 64.89071813795061,
      "grad_norm": 0.7305921912193298,
      "learning_rate": 3.513596366732898e-05,
      "loss": 0.0022347010672092438,
      "step": 228610
    },
    {
      "epoch": 64.89355662787398,
      "grad_norm": 0.2584199905395508,
      "learning_rate": 3.513312517740562e-05,
      "loss": 0.004457659646868705,
      "step": 228620
    },
    {
      "epoch": 64.89639511779733,
      "grad_norm": 0.33071789145469666,
      "learning_rate": 3.5130286687482264e-05,
      "loss": 0.011424675583839417,
      "step": 228630
    },
    {
      "epoch": 64.89923360772069,
      "grad_norm": 0.03937506675720215,
      "learning_rate": 3.51274481975589e-05,
      "loss": 0.0004631970077753067,
      "step": 228640
    },
    {
      "epoch": 64.90207209764405,
      "grad_norm": 0.05989633500576019,
      "learning_rate": 3.512460970763554e-05,
      "loss": 0.0002565082162618637,
      "step": 228650
    },
    {
      "epoch": 64.90491058756741,
      "grad_norm": 0.35470426082611084,
      "learning_rate": 3.512177121771218e-05,
      "loss": 0.0005886577069759369,
      "step": 228660
    },
    {
      "epoch": 64.90774907749078,
      "grad_norm": 0.23171883821487427,
      "learning_rate": 3.5118932727788815e-05,
      "loss": 0.0005947344005107879,
      "step": 228670
    },
    {
      "epoch": 64.91058756741414,
      "grad_norm": 0.05296129360795021,
      "learning_rate": 3.511609423786546e-05,
      "loss": 0.00033188238739967346,
      "step": 228680
    },
    {
      "epoch": 64.9134260573375,
      "grad_norm": 0.07429060339927673,
      "learning_rate": 3.51132557479421e-05,
      "loss": 0.00011734943836927414,
      "step": 228690
    },
    {
      "epoch": 64.91626454726085,
      "grad_norm": 0.016929257661104202,
      "learning_rate": 3.511041725801873e-05,
      "loss": 0.0003141514956951141,
      "step": 228700
    },
    {
      "epoch": 64.91910303718421,
      "grad_norm": 0.0515906885266304,
      "learning_rate": 3.5107578768095374e-05,
      "loss": 0.00045821554958820344,
      "step": 228710
    },
    {
      "epoch": 64.92194152710758,
      "grad_norm": 0.018287762999534607,
      "learning_rate": 3.5104740278172016e-05,
      "loss": 0.0002114308997988701,
      "step": 228720
    },
    {
      "epoch": 64.92478001703094,
      "grad_norm": 0.10078737884759903,
      "learning_rate": 3.510190178824865e-05,
      "loss": 0.00017918720841407775,
      "step": 228730
    },
    {
      "epoch": 64.9276185069543,
      "grad_norm": 0.005425699055194855,
      "learning_rate": 3.509906329832529e-05,
      "loss": 0.000498165562748909,
      "step": 228740
    },
    {
      "epoch": 64.93045699687767,
      "grad_norm": 0.1361352503299713,
      "learning_rate": 3.509622480840193e-05,
      "loss": 0.000264924019575119,
      "step": 228750
    },
    {
      "epoch": 64.93329548680101,
      "grad_norm": 0.01587316393852234,
      "learning_rate": 3.5093386318478574e-05,
      "loss": 0.0007388696074485778,
      "step": 228760
    },
    {
      "epoch": 64.93613397672438,
      "grad_norm": 0.042246703058481216,
      "learning_rate": 3.509054782855521e-05,
      "loss": 0.00021332018077373504,
      "step": 228770
    },
    {
      "epoch": 64.93897246664774,
      "grad_norm": 0.02428055927157402,
      "learning_rate": 3.508770933863185e-05,
      "loss": 0.00021994765847921371,
      "step": 228780
    },
    {
      "epoch": 64.9418109565711,
      "grad_norm": 0.042926132678985596,
      "learning_rate": 3.508487084870849e-05,
      "loss": 0.00024486444890499116,
      "step": 228790
    },
    {
      "epoch": 64.94464944649447,
      "grad_norm": 0.036877937614917755,
      "learning_rate": 3.5082032358785126e-05,
      "loss": 0.001571900025010109,
      "step": 228800
    },
    {
      "epoch": 64.94748793641783,
      "grad_norm": 1.2337578535079956,
      "learning_rate": 3.507919386886177e-05,
      "loss": 0.0006897265091538429,
      "step": 228810
    },
    {
      "epoch": 64.95032642634119,
      "grad_norm": 0.039720479398965836,
      "learning_rate": 3.507635537893841e-05,
      "loss": 0.00016620457172393798,
      "step": 228820
    },
    {
      "epoch": 64.95316491626454,
      "grad_norm": 0.012899552471935749,
      "learning_rate": 3.5073516889015044e-05,
      "loss": 0.00014179348945617676,
      "step": 228830
    },
    {
      "epoch": 64.9560034061879,
      "grad_norm": 0.9321927428245544,
      "learning_rate": 3.5070678399091685e-05,
      "loss": 0.00030822567641735076,
      "step": 228840
    },
    {
      "epoch": 64.95884189611127,
      "grad_norm": 0.021039534360170364,
      "learning_rate": 3.5067839909168326e-05,
      "loss": 0.0003265434876084328,
      "step": 228850
    },
    {
      "epoch": 64.96168038603463,
      "grad_norm": 0.0178206954151392,
      "learning_rate": 3.506500141924496e-05,
      "loss": 0.00022183302789926528,
      "step": 228860
    },
    {
      "epoch": 64.96451887595799,
      "grad_norm": 0.10837355256080627,
      "learning_rate": 3.50621629293216e-05,
      "loss": 0.0014633018523454665,
      "step": 228870
    },
    {
      "epoch": 64.96735736588136,
      "grad_norm": 0.008110346272587776,
      "learning_rate": 3.5059324439398244e-05,
      "loss": 0.00011632349342107772,
      "step": 228880
    },
    {
      "epoch": 64.97019585580472,
      "grad_norm": 0.058470018208026886,
      "learning_rate": 3.5056485949474885e-05,
      "loss": 0.0013840999454259872,
      "step": 228890
    },
    {
      "epoch": 64.97303434572807,
      "grad_norm": 0.010778256691992283,
      "learning_rate": 3.505364745955152e-05,
      "loss": 0.0017486196011304855,
      "step": 228900
    },
    {
      "epoch": 64.97587283565143,
      "grad_norm": 0.047925375401973724,
      "learning_rate": 3.5050808969628154e-05,
      "loss": 0.00029247943311929705,
      "step": 228910
    },
    {
      "epoch": 64.97871132557479,
      "grad_norm": 0.16158564388751984,
      "learning_rate": 3.50479704797048e-05,
      "loss": 0.0024313317611813546,
      "step": 228920
    },
    {
      "epoch": 64.98154981549816,
      "grad_norm": 20.1882266998291,
      "learning_rate": 3.504513198978144e-05,
      "loss": 0.007141861319541931,
      "step": 228930
    },
    {
      "epoch": 64.98438830542152,
      "grad_norm": 0.002830962184816599,
      "learning_rate": 3.504229349985808e-05,
      "loss": 0.0009547272697091102,
      "step": 228940
    },
    {
      "epoch": 64.98722679534488,
      "grad_norm": 0.01265433244407177,
      "learning_rate": 3.503945500993472e-05,
      "loss": 0.002023368515074253,
      "step": 228950
    },
    {
      "epoch": 64.99006528526824,
      "grad_norm": 0.0354875773191452,
      "learning_rate": 3.5036616520011354e-05,
      "loss": 0.0022039802744984625,
      "step": 228960
    },
    {
      "epoch": 64.99290377519159,
      "grad_norm": 0.29998937249183655,
      "learning_rate": 3.5033778030087996e-05,
      "loss": 0.0007000364363193512,
      "step": 228970
    },
    {
      "epoch": 64.99574226511496,
      "grad_norm": 0.41380774974823,
      "learning_rate": 3.503093954016464e-05,
      "loss": 0.0009984109550714494,
      "step": 228980
    },
    {
      "epoch": 64.99858075503832,
      "grad_norm": 0.0715368390083313,
      "learning_rate": 3.502810105024127e-05,
      "loss": 0.000525078922510147,
      "step": 228990
    },
    {
      "epoch": 65.00141924496168,
      "grad_norm": 0.5037841796875,
      "learning_rate": 3.502526256031791e-05,
      "loss": 0.0006352453492581844,
      "step": 229000
    },
    {
      "epoch": 65.00141924496168,
      "eval_accuracy": 0.9757741463724805,
      "eval_loss": 0.09193598479032516,
      "eval_runtime": 33.2982,
      "eval_samples_per_second": 472.308,
      "eval_steps_per_second": 7.388,
      "step": 229000
    },
    {
      "epoch": 65.00425773488504,
      "grad_norm": 0.05479614809155464,
      "learning_rate": 3.502242407039455e-05,
      "loss": 0.003186298906803131,
      "step": 229010
    },
    {
      "epoch": 65.00709622480841,
      "grad_norm": 0.08068352192640305,
      "learning_rate": 3.5019585580471196e-05,
      "loss": 0.011520585417747498,
      "step": 229020
    },
    {
      "epoch": 65.00993471473176,
      "grad_norm": 0.031193336471915245,
      "learning_rate": 3.501674709054783e-05,
      "loss": 0.000502321869134903,
      "step": 229030
    },
    {
      "epoch": 65.01277320465512,
      "grad_norm": 0.060957327485084534,
      "learning_rate": 3.5013908600624465e-05,
      "loss": 0.0013095280155539513,
      "step": 229040
    },
    {
      "epoch": 65.01561169457848,
      "grad_norm": 0.036372117698192596,
      "learning_rate": 3.501107011070111e-05,
      "loss": 0.0007886707782745362,
      "step": 229050
    },
    {
      "epoch": 65.01845018450184,
      "grad_norm": 0.08591330796480179,
      "learning_rate": 3.500823162077775e-05,
      "loss": 0.00048009101301431655,
      "step": 229060
    },
    {
      "epoch": 65.02128867442521,
      "grad_norm": 0.06429411470890045,
      "learning_rate": 3.500539313085439e-05,
      "loss": 0.0003225667402148247,
      "step": 229070
    },
    {
      "epoch": 65.02412716434857,
      "grad_norm": 2.2827236652374268,
      "learning_rate": 3.500255464093103e-05,
      "loss": 0.0005907127633690834,
      "step": 229080
    },
    {
      "epoch": 65.02696565427193,
      "grad_norm": 0.8169274926185608,
      "learning_rate": 3.4999716151007665e-05,
      "loss": 0.0007720300927758217,
      "step": 229090
    },
    {
      "epoch": 65.02980414419528,
      "grad_norm": 0.0725683942437172,
      "learning_rate": 3.4996877661084307e-05,
      "loss": 0.0002673149108886719,
      "step": 229100
    },
    {
      "epoch": 65.03264263411864,
      "grad_norm": 0.004973485134541988,
      "learning_rate": 3.499403917116094e-05,
      "loss": 0.00018357466906309127,
      "step": 229110
    },
    {
      "epoch": 65.03548112404201,
      "grad_norm": 0.03577915579080582,
      "learning_rate": 3.499120068123758e-05,
      "loss": 0.0020963814109563827,
      "step": 229120
    },
    {
      "epoch": 65.03831961396537,
      "grad_norm": 0.022209584712982178,
      "learning_rate": 3.4988362191314224e-05,
      "loss": 0.0024253327399492266,
      "step": 229130
    },
    {
      "epoch": 65.04115810388873,
      "grad_norm": 0.008739454671740532,
      "learning_rate": 3.498552370139086e-05,
      "loss": 0.0008276848122477532,
      "step": 229140
    },
    {
      "epoch": 65.0439965938121,
      "grad_norm": 0.6820820569992065,
      "learning_rate": 3.49826852114675e-05,
      "loss": 0.00024139154702425004,
      "step": 229150
    },
    {
      "epoch": 65.04683508373546,
      "grad_norm": 0.14321482181549072,
      "learning_rate": 3.497984672154414e-05,
      "loss": 0.0008460866287350654,
      "step": 229160
    },
    {
      "epoch": 65.04967357365881,
      "grad_norm": 0.06330700218677521,
      "learning_rate": 3.4977008231620776e-05,
      "loss": 0.0006181022152304649,
      "step": 229170
    },
    {
      "epoch": 65.05251206358217,
      "grad_norm": 0.030846847221255302,
      "learning_rate": 3.4974169741697424e-05,
      "loss": 0.00036870408803224564,
      "step": 229180
    },
    {
      "epoch": 65.05535055350553,
      "grad_norm": 0.33206796646118164,
      "learning_rate": 3.497133125177406e-05,
      "loss": 0.0007896116003394127,
      "step": 229190
    },
    {
      "epoch": 65.0581890434289,
      "grad_norm": 0.6131073236465454,
      "learning_rate": 3.496849276185069e-05,
      "loss": 0.0016240157186985017,
      "step": 229200
    },
    {
      "epoch": 65.06102753335226,
      "grad_norm": 0.13126453757286072,
      "learning_rate": 3.4965654271927335e-05,
      "loss": 0.0004170423373579979,
      "step": 229210
    },
    {
      "epoch": 65.06386602327562,
      "grad_norm": 0.02850036695599556,
      "learning_rate": 3.4962815782003976e-05,
      "loss": 0.0008174004033207893,
      "step": 229220
    },
    {
      "epoch": 65.06670451319899,
      "grad_norm": 0.015924982726573944,
      "learning_rate": 3.495997729208062e-05,
      "loss": 0.00026204828172922135,
      "step": 229230
    },
    {
      "epoch": 65.06954300312233,
      "grad_norm": 2.3965506553649902,
      "learning_rate": 3.495713880215725e-05,
      "loss": 0.001095571368932724,
      "step": 229240
    },
    {
      "epoch": 65.0723814930457,
      "grad_norm": 2.221008777618408,
      "learning_rate": 3.495430031223389e-05,
      "loss": 0.004897543415427208,
      "step": 229250
    },
    {
      "epoch": 65.07521998296906,
      "grad_norm": 0.4559348523616791,
      "learning_rate": 3.4951461822310535e-05,
      "loss": 0.0004163816571235657,
      "step": 229260
    },
    {
      "epoch": 65.07805847289242,
      "grad_norm": 3.349370002746582,
      "learning_rate": 3.494862333238717e-05,
      "loss": 0.0029130171984434127,
      "step": 229270
    },
    {
      "epoch": 65.08089696281579,
      "grad_norm": 0.0927416980266571,
      "learning_rate": 3.494578484246381e-05,
      "loss": 0.0003390710800886154,
      "step": 229280
    },
    {
      "epoch": 65.08373545273915,
      "grad_norm": 0.5207904577255249,
      "learning_rate": 3.494294635254045e-05,
      "loss": 0.00040192436426877975,
      "step": 229290
    },
    {
      "epoch": 65.0865739426625,
      "grad_norm": 0.061975330114364624,
      "learning_rate": 3.4940107862617087e-05,
      "loss": 0.0001819005236029625,
      "step": 229300
    },
    {
      "epoch": 65.08941243258586,
      "grad_norm": 0.31487739086151123,
      "learning_rate": 3.493726937269373e-05,
      "loss": 0.00038095135241746905,
      "step": 229310
    },
    {
      "epoch": 65.09225092250922,
      "grad_norm": 0.4930785596370697,
      "learning_rate": 3.493443088277037e-05,
      "loss": 0.0014207342639565468,
      "step": 229320
    },
    {
      "epoch": 65.09508941243259,
      "grad_norm": 0.07682589441537857,
      "learning_rate": 3.4931592392847004e-05,
      "loss": 0.0026118090376257897,
      "step": 229330
    },
    {
      "epoch": 65.09792790235595,
      "grad_norm": 0.013613464310765266,
      "learning_rate": 3.4928753902923645e-05,
      "loss": 0.0002553405240178108,
      "step": 229340
    },
    {
      "epoch": 65.10076639227931,
      "grad_norm": 13.439159393310547,
      "learning_rate": 3.492591541300029e-05,
      "loss": 0.002045172452926636,
      "step": 229350
    },
    {
      "epoch": 65.10360488220267,
      "grad_norm": 0.15311965346336365,
      "learning_rate": 3.492307692307693e-05,
      "loss": 0.0002773746848106384,
      "step": 229360
    },
    {
      "epoch": 65.10644337212602,
      "grad_norm": 0.10589984059333801,
      "learning_rate": 3.492023843315356e-05,
      "loss": 0.00018305163830518724,
      "step": 229370
    },
    {
      "epoch": 65.10928186204939,
      "grad_norm": 0.0267547108232975,
      "learning_rate": 3.4917399943230204e-05,
      "loss": 0.00010274406522512436,
      "step": 229380
    },
    {
      "epoch": 65.11212035197275,
      "grad_norm": 0.3524516224861145,
      "learning_rate": 3.4914561453306845e-05,
      "loss": 0.0001599179580807686,
      "step": 229390
    },
    {
      "epoch": 65.11495884189611,
      "grad_norm": 0.19062456488609314,
      "learning_rate": 3.491172296338348e-05,
      "loss": 0.00011992137879133225,
      "step": 229400
    },
    {
      "epoch": 65.11779733181947,
      "grad_norm": 0.00820415560156107,
      "learning_rate": 3.490888447346012e-05,
      "loss": 0.0004836037755012512,
      "step": 229410
    },
    {
      "epoch": 65.12063582174284,
      "grad_norm": 0.17747282981872559,
      "learning_rate": 3.490604598353676e-05,
      "loss": 0.004467607289552688,
      "step": 229420
    },
    {
      "epoch": 65.1234743116662,
      "grad_norm": 0.009906544350087643,
      "learning_rate": 3.49032074936134e-05,
      "loss": 0.0007542431354522705,
      "step": 229430
    },
    {
      "epoch": 65.12631280158955,
      "grad_norm": 0.3473424017429352,
      "learning_rate": 3.490036900369004e-05,
      "loss": 0.00026749838143587114,
      "step": 229440
    },
    {
      "epoch": 65.12915129151291,
      "grad_norm": 0.05080820992588997,
      "learning_rate": 3.489753051376668e-05,
      "loss": 0.00022383183240890503,
      "step": 229450
    },
    {
      "epoch": 65.13198978143627,
      "grad_norm": 0.01427767239511013,
      "learning_rate": 3.4894692023843315e-05,
      "loss": 0.00016125701367855071,
      "step": 229460
    },
    {
      "epoch": 65.13482827135964,
      "grad_norm": 0.0320495180785656,
      "learning_rate": 3.4891853533919956e-05,
      "loss": 0.00010007526725530624,
      "step": 229470
    },
    {
      "epoch": 65.137666761283,
      "grad_norm": 0.123896025121212,
      "learning_rate": 3.48890150439966e-05,
      "loss": 0.000272434763610363,
      "step": 229480
    },
    {
      "epoch": 65.14050525120636,
      "grad_norm": 0.14948038756847382,
      "learning_rate": 3.488617655407324e-05,
      "loss": 9.644664824008942e-05,
      "step": 229490
    },
    {
      "epoch": 65.14334374112971,
      "grad_norm": 0.03119046241044998,
      "learning_rate": 3.4883338064149873e-05,
      "loss": 0.0020104911178350448,
      "step": 229500
    },
    {
      "epoch": 65.14334374112971,
      "eval_accuracy": 0.9791441470083296,
      "eval_loss": 0.07697154581546783,
      "eval_runtime": 33.0737,
      "eval_samples_per_second": 475.513,
      "eval_steps_per_second": 7.438,
      "step": 229500
    },
    {
      "epoch": 65.14618223105307,
      "grad_norm": 0.050489313900470734,
      "learning_rate": 3.488049957422651e-05,
      "loss": 0.0013831224292516708,
      "step": 229510
    },
    {
      "epoch": 65.14902072097644,
      "grad_norm": 0.02803208865225315,
      "learning_rate": 3.4877661084303156e-05,
      "loss": 0.0006270445883274078,
      "step": 229520
    },
    {
      "epoch": 65.1518592108998,
      "grad_norm": 0.24929587543010712,
      "learning_rate": 3.487482259437979e-05,
      "loss": 0.0005738938227295876,
      "step": 229530
    },
    {
      "epoch": 65.15469770082316,
      "grad_norm": 0.013013454154133797,
      "learning_rate": 3.487198410445643e-05,
      "loss": 0.0004579050466418266,
      "step": 229540
    },
    {
      "epoch": 65.15753619074653,
      "grad_norm": 0.020105259492993355,
      "learning_rate": 3.4869145614533074e-05,
      "loss": 0.0005306549370288849,
      "step": 229550
    },
    {
      "epoch": 65.16037468066989,
      "grad_norm": 0.030942514538764954,
      "learning_rate": 3.486630712460971e-05,
      "loss": 0.0012645466253161431,
      "step": 229560
    },
    {
      "epoch": 65.16321317059324,
      "grad_norm": 0.0073573836125433445,
      "learning_rate": 3.486346863468635e-05,
      "loss": 0.00012910030782222747,
      "step": 229570
    },
    {
      "epoch": 65.1660516605166,
      "grad_norm": 0.008817413821816444,
      "learning_rate": 3.486063014476299e-05,
      "loss": 0.00011296533048152924,
      "step": 229580
    },
    {
      "epoch": 65.16889015043996,
      "grad_norm": 0.039188235998153687,
      "learning_rate": 3.4857791654839625e-05,
      "loss": 0.00022115688771009445,
      "step": 229590
    },
    {
      "epoch": 65.17172864036333,
      "grad_norm": 0.05255655571818352,
      "learning_rate": 3.485495316491627e-05,
      "loss": 0.0002570157870650291,
      "step": 229600
    },
    {
      "epoch": 65.17456713028669,
      "grad_norm": 0.022985966876149178,
      "learning_rate": 3.48521146749929e-05,
      "loss": 0.00017924681305885314,
      "step": 229610
    },
    {
      "epoch": 65.17740562021005,
      "grad_norm": 0.01840195246040821,
      "learning_rate": 3.484927618506954e-05,
      "loss": 5.7394057512283327e-05,
      "step": 229620
    },
    {
      "epoch": 65.18024411013342,
      "grad_norm": 0.3032020032405853,
      "learning_rate": 3.4846437695146184e-05,
      "loss": 0.0001550896093249321,
      "step": 229630
    },
    {
      "epoch": 65.18308260005676,
      "grad_norm": 0.009323233738541603,
      "learning_rate": 3.484359920522282e-05,
      "loss": 0.00015639737248420715,
      "step": 229640
    },
    {
      "epoch": 65.18592108998013,
      "grad_norm": 0.008873976767063141,
      "learning_rate": 3.484076071529947e-05,
      "loss": 0.000606117770075798,
      "step": 229650
    },
    {
      "epoch": 65.18875957990349,
      "grad_norm": 0.9524775743484497,
      "learning_rate": 3.48379222253761e-05,
      "loss": 0.0002569831907749176,
      "step": 229660
    },
    {
      "epoch": 65.19159806982685,
      "grad_norm": 0.010795419104397297,
      "learning_rate": 3.4835083735452736e-05,
      "loss": 0.00017020776867866517,
      "step": 229670
    },
    {
      "epoch": 65.19443655975022,
      "grad_norm": 0.058275166898965836,
      "learning_rate": 3.4832245245529384e-05,
      "loss": 0.0008393898606300354,
      "step": 229680
    },
    {
      "epoch": 65.19727504967358,
      "grad_norm": 0.037776391953229904,
      "learning_rate": 3.482940675560602e-05,
      "loss": 0.00029291976243257524,
      "step": 229690
    },
    {
      "epoch": 65.20011353959694,
      "grad_norm": 0.03828248754143715,
      "learning_rate": 3.482656826568266e-05,
      "loss": 0.00015361644327640533,
      "step": 229700
    },
    {
      "epoch": 65.20295202952029,
      "grad_norm": 0.03192436695098877,
      "learning_rate": 3.4823729775759295e-05,
      "loss": 0.0006170524284243584,
      "step": 229710
    },
    {
      "epoch": 65.20579051944365,
      "grad_norm": 0.0719391331076622,
      "learning_rate": 3.4820891285835936e-05,
      "loss": 0.0002520453184843063,
      "step": 229720
    },
    {
      "epoch": 65.20862900936702,
      "grad_norm": 0.03251367807388306,
      "learning_rate": 3.481805279591258e-05,
      "loss": 0.0004302188754081726,
      "step": 229730
    },
    {
      "epoch": 65.21146749929038,
      "grad_norm": 0.012099698185920715,
      "learning_rate": 3.481521430598921e-05,
      "loss": 0.0032926045358181,
      "step": 229740
    },
    {
      "epoch": 65.21430598921374,
      "grad_norm": 0.10387854278087616,
      "learning_rate": 3.4812375816065854e-05,
      "loss": 0.0005930440500378608,
      "step": 229750
    },
    {
      "epoch": 65.2171444791371,
      "grad_norm": 0.02268698439002037,
      "learning_rate": 3.4809537326142495e-05,
      "loss": 0.000557217001914978,
      "step": 229760
    },
    {
      "epoch": 65.21998296906045,
      "grad_norm": 0.02164512500166893,
      "learning_rate": 3.480669883621913e-05,
      "loss": 0.004361123964190483,
      "step": 229770
    },
    {
      "epoch": 65.22282145898382,
      "grad_norm": 0.08999887108802795,
      "learning_rate": 3.480386034629578e-05,
      "loss": 0.0004393072798848152,
      "step": 229780
    },
    {
      "epoch": 65.22565994890718,
      "grad_norm": 0.006770040839910507,
      "learning_rate": 3.480102185637241e-05,
      "loss": 0.006859703361988068,
      "step": 229790
    },
    {
      "epoch": 65.22849843883054,
      "grad_norm": 0.055072225630283356,
      "learning_rate": 3.479818336644905e-05,
      "loss": 0.0005920510739088058,
      "step": 229800
    },
    {
      "epoch": 65.2313369287539,
      "grad_norm": 0.14166371524333954,
      "learning_rate": 3.479534487652569e-05,
      "loss": 0.00046294480562210084,
      "step": 229810
    },
    {
      "epoch": 65.23417541867727,
      "grad_norm": 0.025354202836751938,
      "learning_rate": 3.479250638660233e-05,
      "loss": 0.0002845579758286476,
      "step": 229820
    },
    {
      "epoch": 65.23701390860063,
      "grad_norm": 0.08689596503973007,
      "learning_rate": 3.478966789667897e-05,
      "loss": 9.42939892411232e-05,
      "step": 229830
    },
    {
      "epoch": 65.23985239852398,
      "grad_norm": 0.011884348466992378,
      "learning_rate": 3.4786829406755606e-05,
      "loss": 0.0002040725201368332,
      "step": 229840
    },
    {
      "epoch": 65.24269088844734,
      "grad_norm": 0.23543092608451843,
      "learning_rate": 3.478399091683225e-05,
      "loss": 0.00026011522859334944,
      "step": 229850
    },
    {
      "epoch": 65.2455293783707,
      "grad_norm": 0.14591899514198303,
      "learning_rate": 3.478115242690889e-05,
      "loss": 0.0006381263956427574,
      "step": 229860
    },
    {
      "epoch": 65.24836786829407,
      "grad_norm": 0.056799184530973434,
      "learning_rate": 3.477831393698552e-05,
      "loss": 0.0015196120366454124,
      "step": 229870
    },
    {
      "epoch": 65.25120635821743,
      "grad_norm": 0.06244302913546562,
      "learning_rate": 3.4775475447062164e-05,
      "loss": 0.00046346522867679596,
      "step": 229880
    },
    {
      "epoch": 65.2540448481408,
      "grad_norm": 0.19157803058624268,
      "learning_rate": 3.4772636957138806e-05,
      "loss": 0.0052953936159610745,
      "step": 229890
    },
    {
      "epoch": 65.25688333806416,
      "grad_norm": 0.01785968244075775,
      "learning_rate": 3.476979846721544e-05,
      "loss": 0.0005230056121945381,
      "step": 229900
    },
    {
      "epoch": 65.2597218279875,
      "grad_norm": 3.848658323287964,
      "learning_rate": 3.476695997729208e-05,
      "loss": 0.001254352554678917,
      "step": 229910
    },
    {
      "epoch": 65.26256031791087,
      "grad_norm": 0.010446425527334213,
      "learning_rate": 3.476412148736872e-05,
      "loss": 0.000417804904282093,
      "step": 229920
    },
    {
      "epoch": 65.26539880783423,
      "grad_norm": 0.0497196689248085,
      "learning_rate": 3.476128299744536e-05,
      "loss": 0.00048804152756929395,
      "step": 229930
    },
    {
      "epoch": 65.2682372977576,
      "grad_norm": 0.04484446719288826,
      "learning_rate": 3.4758444507522e-05,
      "loss": 0.0005066849291324616,
      "step": 229940
    },
    {
      "epoch": 65.27107578768096,
      "grad_norm": 0.1865955889225006,
      "learning_rate": 3.475560601759864e-05,
      "loss": 0.0009935196489095689,
      "step": 229950
    },
    {
      "epoch": 65.27391427760432,
      "grad_norm": 0.680886447429657,
      "learning_rate": 3.475276752767528e-05,
      "loss": 0.0010285081341862678,
      "step": 229960
    },
    {
      "epoch": 65.27675276752768,
      "grad_norm": 0.2630820572376251,
      "learning_rate": 3.4749929037751916e-05,
      "loss": 0.00044944640249013903,
      "step": 229970
    },
    {
      "epoch": 65.27959125745103,
      "grad_norm": 0.06253282725811005,
      "learning_rate": 3.474709054782856e-05,
      "loss": 0.0018233034759759903,
      "step": 229980
    },
    {
      "epoch": 65.2824297473744,
      "grad_norm": 0.052729811519384384,
      "learning_rate": 3.47442520579052e-05,
      "loss": 0.0015331929549574851,
      "step": 229990
    },
    {
      "epoch": 65.28526823729776,
      "grad_norm": 0.10477558523416519,
      "learning_rate": 3.4741413567981834e-05,
      "loss": 0.00046506300568580625,
      "step": 230000
    },
    {
      "epoch": 65.28526823729776,
      "eval_accuracy": 0.9705601831245628,
      "eval_loss": 0.1166166365146637,
      "eval_runtime": 32.9518,
      "eval_samples_per_second": 477.272,
      "eval_steps_per_second": 7.465,
      "step": 230000
    },
    {
      "epoch": 65.28810672722112,
      "grad_norm": 0.08292065560817719,
      "learning_rate": 3.4738575078058475e-05,
      "loss": 0.0033504992723464968,
      "step": 230010
    },
    {
      "epoch": 65.29094521714448,
      "grad_norm": 0.9038577079772949,
      "learning_rate": 3.4735736588135116e-05,
      "loss": 0.0005789464339613914,
      "step": 230020
    },
    {
      "epoch": 65.29378370706785,
      "grad_norm": 0.020823480561375618,
      "learning_rate": 3.473289809821175e-05,
      "loss": 0.0006854185834527015,
      "step": 230030
    },
    {
      "epoch": 65.2966221969912,
      "grad_norm": 1.0181567668914795,
      "learning_rate": 3.473005960828839e-05,
      "loss": 0.0008813267573714257,
      "step": 230040
    },
    {
      "epoch": 65.29946068691456,
      "grad_norm": 0.03446287289261818,
      "learning_rate": 3.4727221118365034e-05,
      "loss": 0.007146850228309631,
      "step": 230050
    },
    {
      "epoch": 65.30229917683792,
      "grad_norm": 0.44271478056907654,
      "learning_rate": 3.472438262844167e-05,
      "loss": 0.0002869222313165665,
      "step": 230060
    },
    {
      "epoch": 65.30513766676128,
      "grad_norm": 0.3417847156524658,
      "learning_rate": 3.472154413851831e-05,
      "loss": 0.00022647716104984283,
      "step": 230070
    },
    {
      "epoch": 65.30797615668465,
      "grad_norm": 3.1347646713256836,
      "learning_rate": 3.471870564859495e-05,
      "loss": 0.0006802832707762718,
      "step": 230080
    },
    {
      "epoch": 65.31081464660801,
      "grad_norm": 0.07173875719308853,
      "learning_rate": 3.4715867158671586e-05,
      "loss": 0.00018881857395172119,
      "step": 230090
    },
    {
      "epoch": 65.31365313653137,
      "grad_norm": 0.011403699405491352,
      "learning_rate": 3.471302866874823e-05,
      "loss": 0.00010371096432209015,
      "step": 230100
    },
    {
      "epoch": 65.31649162645472,
      "grad_norm": 0.015037884935736656,
      "learning_rate": 3.471019017882487e-05,
      "loss": 0.00010374002158641815,
      "step": 230110
    },
    {
      "epoch": 65.31933011637808,
      "grad_norm": 0.017012938857078552,
      "learning_rate": 3.470735168890151e-05,
      "loss": 0.0005794903263449669,
      "step": 230120
    },
    {
      "epoch": 65.32216860630145,
      "grad_norm": 0.03473366051912308,
      "learning_rate": 3.4704513198978144e-05,
      "loss": 0.00012717004865407943,
      "step": 230130
    },
    {
      "epoch": 65.32500709622481,
      "grad_norm": 0.008127683773636818,
      "learning_rate": 3.470167470905478e-05,
      "loss": 0.0003181645646691322,
      "step": 230140
    },
    {
      "epoch": 65.32784558614817,
      "grad_norm": 0.10164839774370193,
      "learning_rate": 3.469883621913143e-05,
      "loss": 0.00010783802717924118,
      "step": 230150
    },
    {
      "epoch": 65.33068407607153,
      "grad_norm": 0.5886454582214355,
      "learning_rate": 3.469599772920806e-05,
      "loss": 0.00025347527116537094,
      "step": 230160
    },
    {
      "epoch": 65.3335225659949,
      "grad_norm": 0.008659115992486477,
      "learning_rate": 3.46931592392847e-05,
      "loss": 0.0013678845018148423,
      "step": 230170
    },
    {
      "epoch": 65.33636105591825,
      "grad_norm": 2.6044373512268066,
      "learning_rate": 3.4690320749361345e-05,
      "loss": 0.0006966769695281982,
      "step": 230180
    },
    {
      "epoch": 65.33919954584161,
      "grad_norm": 0.4325608015060425,
      "learning_rate": 3.468748225943798e-05,
      "loss": 0.0008468374609947205,
      "step": 230190
    },
    {
      "epoch": 65.34203803576497,
      "grad_norm": 0.08342748135328293,
      "learning_rate": 3.468464376951462e-05,
      "loss": 0.0001406598836183548,
      "step": 230200
    },
    {
      "epoch": 65.34487652568833,
      "grad_norm": 0.007242447696626186,
      "learning_rate": 3.468180527959126e-05,
      "loss": 0.00013828482478857041,
      "step": 230210
    },
    {
      "epoch": 65.3477150156117,
      "grad_norm": 0.011243588291108608,
      "learning_rate": 3.4678966789667897e-05,
      "loss": 0.0003251615911722183,
      "step": 230220
    },
    {
      "epoch": 65.35055350553506,
      "grad_norm": 0.6143301129341125,
      "learning_rate": 3.467612829974454e-05,
      "loss": 0.0009912194684147835,
      "step": 230230
    },
    {
      "epoch": 65.35339199545841,
      "grad_norm": 1.3268275260925293,
      "learning_rate": 3.467328980982117e-05,
      "loss": 0.00035418737679719924,
      "step": 230240
    },
    {
      "epoch": 65.35623048538177,
      "grad_norm": 0.1074412539601326,
      "learning_rate": 3.467045131989782e-05,
      "loss": 0.0001795763149857521,
      "step": 230250
    },
    {
      "epoch": 65.35906897530514,
      "grad_norm": 0.029931599274277687,
      "learning_rate": 3.4667612829974455e-05,
      "loss": 0.00033677797764539716,
      "step": 230260
    },
    {
      "epoch": 65.3619074652285,
      "grad_norm": 0.01415445189923048,
      "learning_rate": 3.466477434005109e-05,
      "loss": 0.00013789571821689605,
      "step": 230270
    },
    {
      "epoch": 65.36474595515186,
      "grad_norm": 0.10055750608444214,
      "learning_rate": 3.466193585012774e-05,
      "loss": 0.00010680314153432846,
      "step": 230280
    },
    {
      "epoch": 65.36758444507522,
      "grad_norm": 0.040038395673036575,
      "learning_rate": 3.465909736020437e-05,
      "loss": 0.00020609032362699508,
      "step": 230290
    },
    {
      "epoch": 65.37042293499859,
      "grad_norm": 0.27710017561912537,
      "learning_rate": 3.4656258870281014e-05,
      "loss": 0.004402939975261688,
      "step": 230300
    },
    {
      "epoch": 65.37326142492194,
      "grad_norm": 0.026688730344176292,
      "learning_rate": 3.4653420380357655e-05,
      "loss": 0.00030074585229158404,
      "step": 230310
    },
    {
      "epoch": 65.3760999148453,
      "grad_norm": 0.725716769695282,
      "learning_rate": 3.465058189043429e-05,
      "loss": 0.0006171843037009239,
      "step": 230320
    },
    {
      "epoch": 65.37893840476866,
      "grad_norm": 0.029741698876023293,
      "learning_rate": 3.464774340051093e-05,
      "loss": 0.0003771902993321419,
      "step": 230330
    },
    {
      "epoch": 65.38177689469202,
      "grad_norm": 0.02311135083436966,
      "learning_rate": 3.4644904910587566e-05,
      "loss": 0.00014627501368522645,
      "step": 230340
    },
    {
      "epoch": 65.38461538461539,
      "grad_norm": 0.08507902175188065,
      "learning_rate": 3.464206642066421e-05,
      "loss": 0.0002539176493883133,
      "step": 230350
    },
    {
      "epoch": 65.38745387453875,
      "grad_norm": 0.08854518085718155,
      "learning_rate": 3.463922793074085e-05,
      "loss": 0.00014759786427021027,
      "step": 230360
    },
    {
      "epoch": 65.39029236446211,
      "grad_norm": 0.01596117950975895,
      "learning_rate": 3.463638944081748e-05,
      "loss": 0.00015472415834665297,
      "step": 230370
    },
    {
      "epoch": 65.39313085438546,
      "grad_norm": 0.016562962904572487,
      "learning_rate": 3.463355095089413e-05,
      "loss": 0.00011501666158437729,
      "step": 230380
    },
    {
      "epoch": 65.39596934430882,
      "grad_norm": 0.019831513985991478,
      "learning_rate": 3.4630712460970766e-05,
      "loss": 0.00014644060283899307,
      "step": 230390
    },
    {
      "epoch": 65.39880783423219,
      "grad_norm": 0.13157080113887787,
      "learning_rate": 3.46278739710474e-05,
      "loss": 0.00022806357592344284,
      "step": 230400
    },
    {
      "epoch": 65.40164632415555,
      "grad_norm": 0.005039254203438759,
      "learning_rate": 3.462503548112405e-05,
      "loss": 9.468793869018554e-05,
      "step": 230410
    },
    {
      "epoch": 65.40448481407891,
      "grad_norm": 0.017745940014719963,
      "learning_rate": 3.462219699120068e-05,
      "loss": 0.00018891878426074982,
      "step": 230420
    },
    {
      "epoch": 65.40732330400228,
      "grad_norm": 0.013486118987202644,
      "learning_rate": 3.4619358501277325e-05,
      "loss": 0.00016265679150819778,
      "step": 230430
    },
    {
      "epoch": 65.41016179392564,
      "grad_norm": 0.018304064869880676,
      "learning_rate": 3.461652001135396e-05,
      "loss": 0.0004008414223790169,
      "step": 230440
    },
    {
      "epoch": 65.41300028384899,
      "grad_norm": 0.00928535033017397,
      "learning_rate": 3.46136815214306e-05,
      "loss": 0.000920371524989605,
      "step": 230450
    },
    {
      "epoch": 65.41583877377235,
      "grad_norm": 0.07542373985052109,
      "learning_rate": 3.461084303150724e-05,
      "loss": 0.0001936875283718109,
      "step": 230460
    },
    {
      "epoch": 65.41867726369571,
      "grad_norm": 0.03510186821222305,
      "learning_rate": 3.460800454158388e-05,
      "loss": 0.0013897791504859924,
      "step": 230470
    },
    {
      "epoch": 65.42151575361908,
      "grad_norm": 0.016283566132187843,
      "learning_rate": 3.460516605166052e-05,
      "loss": 0.00034122951328754427,
      "step": 230480
    },
    {
      "epoch": 65.42435424354244,
      "grad_norm": 0.03949473425745964,
      "learning_rate": 3.460232756173716e-05,
      "loss": 0.0009489970281720162,
      "step": 230490
    },
    {
      "epoch": 65.4271927334658,
      "grad_norm": 2.7300937175750732,
      "learning_rate": 3.4599489071813794e-05,
      "loss": 0.0010180028155446052,
      "step": 230500
    },
    {
      "epoch": 65.4271927334658,
      "eval_accuracy": 0.9783175430787817,
      "eval_loss": 0.07831234484910965,
      "eval_runtime": 32.8777,
      "eval_samples_per_second": 478.348,
      "eval_steps_per_second": 7.482,
      "step": 230500
    },
    {
      "epoch": 65.43003122338915,
      "grad_norm": 0.2614760100841522,
      "learning_rate": 3.4596650581890435e-05,
      "loss": 0.0005953298881649971,
      "step": 230510
    },
    {
      "epoch": 65.43286971331251,
      "grad_norm": 0.3019252419471741,
      "learning_rate": 3.459381209196708e-05,
      "loss": 0.0004292204976081848,
      "step": 230520
    },
    {
      "epoch": 65.43570820323588,
      "grad_norm": 0.06626644730567932,
      "learning_rate": 3.459097360204371e-05,
      "loss": 0.0002352869138121605,
      "step": 230530
    },
    {
      "epoch": 65.43854669315924,
      "grad_norm": 0.039274945855140686,
      "learning_rate": 3.458813511212035e-05,
      "loss": 0.0018612436950206756,
      "step": 230540
    },
    {
      "epoch": 65.4413851830826,
      "grad_norm": 0.01713317632675171,
      "learning_rate": 3.4585296622196994e-05,
      "loss": 0.0005646152421832085,
      "step": 230550
    },
    {
      "epoch": 65.44422367300596,
      "grad_norm": 0.11152097582817078,
      "learning_rate": 3.458245813227363e-05,
      "loss": 0.0008381742984056473,
      "step": 230560
    },
    {
      "epoch": 65.44706216292933,
      "grad_norm": 0.002642684616148472,
      "learning_rate": 3.457961964235027e-05,
      "loss": 0.00011054836213588714,
      "step": 230570
    },
    {
      "epoch": 65.44990065285268,
      "grad_norm": 3.1664795875549316,
      "learning_rate": 3.457678115242691e-05,
      "loss": 0.0009320586919784546,
      "step": 230580
    },
    {
      "epoch": 65.45273914277604,
      "grad_norm": 0.008346828632056713,
      "learning_rate": 3.457394266250355e-05,
      "loss": 0.0003162892535328865,
      "step": 230590
    },
    {
      "epoch": 65.4555776326994,
      "grad_norm": 0.08433346450328827,
      "learning_rate": 3.4571388021572526e-05,
      "loss": 0.0037971440702676772,
      "step": 230600
    },
    {
      "epoch": 65.45841612262276,
      "grad_norm": 0.24099792540073395,
      "learning_rate": 3.456854953164916e-05,
      "loss": 0.010692233592271805,
      "step": 230610
    },
    {
      "epoch": 65.46125461254613,
      "grad_norm": 0.06739385426044464,
      "learning_rate": 3.45657110417258e-05,
      "loss": 0.0004597621038556099,
      "step": 230620
    },
    {
      "epoch": 65.46409310246949,
      "grad_norm": 0.01220053993165493,
      "learning_rate": 3.456287255180244e-05,
      "loss": 0.00041505414992570877,
      "step": 230630
    },
    {
      "epoch": 65.46693159239285,
      "grad_norm": 0.019190708175301552,
      "learning_rate": 3.4560034061879085e-05,
      "loss": 0.00036629755049943924,
      "step": 230640
    },
    {
      "epoch": 65.4697700823162,
      "grad_norm": 0.03015018254518509,
      "learning_rate": 3.455719557195572e-05,
      "loss": 8.632075041532516e-05,
      "step": 230650
    },
    {
      "epoch": 65.47260857223957,
      "grad_norm": 4.916889190673828,
      "learning_rate": 3.455435708203236e-05,
      "loss": 0.0012139411643147468,
      "step": 230660
    },
    {
      "epoch": 65.47544706216293,
      "grad_norm": 0.028063824400305748,
      "learning_rate": 3.4551518592109e-05,
      "loss": 0.0003934372216463089,
      "step": 230670
    },
    {
      "epoch": 65.47828555208629,
      "grad_norm": 0.050168950110673904,
      "learning_rate": 3.454868010218564e-05,
      "loss": 0.0004121895879507065,
      "step": 230680
    },
    {
      "epoch": 65.48112404200965,
      "grad_norm": 0.0836583599448204,
      "learning_rate": 3.454584161226228e-05,
      "loss": 0.0004968998953700065,
      "step": 230690
    },
    {
      "epoch": 65.48396253193302,
      "grad_norm": 0.02380485087633133,
      "learning_rate": 3.454300312233892e-05,
      "loss": 0.0004846619442105293,
      "step": 230700
    },
    {
      "epoch": 65.48680102185637,
      "grad_norm": 0.10654915124177933,
      "learning_rate": 3.4540164632415554e-05,
      "loss": 0.00037540644407272337,
      "step": 230710
    },
    {
      "epoch": 65.48963951177973,
      "grad_norm": 0.06196329742670059,
      "learning_rate": 3.4537326142492195e-05,
      "loss": 0.00023442618548870088,
      "step": 230720
    },
    {
      "epoch": 65.49247800170309,
      "grad_norm": 0.013339802622795105,
      "learning_rate": 3.453448765256884e-05,
      "loss": 0.00013826563954353333,
      "step": 230730
    },
    {
      "epoch": 65.49531649162645,
      "grad_norm": 0.3302919566631317,
      "learning_rate": 3.453164916264547e-05,
      "loss": 0.00036805160343647005,
      "step": 230740
    },
    {
      "epoch": 65.49815498154982,
      "grad_norm": 0.01947767846286297,
      "learning_rate": 3.452881067272211e-05,
      "loss": 0.0003106825053691864,
      "step": 230750
    },
    {
      "epoch": 65.50099347147318,
      "grad_norm": 0.06973744183778763,
      "learning_rate": 3.4525972182798754e-05,
      "loss": 0.00019247308373451232,
      "step": 230760
    },
    {
      "epoch": 65.50383196139654,
      "grad_norm": 0.04456651210784912,
      "learning_rate": 3.4523133692875396e-05,
      "loss": 0.0019077777862548829,
      "step": 230770
    },
    {
      "epoch": 65.50667045131989,
      "grad_norm": 0.1005941778421402,
      "learning_rate": 3.452029520295203e-05,
      "loss": 0.0002575790509581566,
      "step": 230780
    },
    {
      "epoch": 65.50950894124325,
      "grad_norm": 0.0442071259021759,
      "learning_rate": 3.451745671302867e-05,
      "loss": 0.0004042450338602066,
      "step": 230790
    },
    {
      "epoch": 65.51234743116662,
      "grad_norm": 0.2052815556526184,
      "learning_rate": 3.451461822310531e-05,
      "loss": 0.0002114584669470787,
      "step": 230800
    },
    {
      "epoch": 65.51518592108998,
      "grad_norm": 0.09678258746862411,
      "learning_rate": 3.451177973318195e-05,
      "loss": 0.00029650796204805374,
      "step": 230810
    },
    {
      "epoch": 65.51802441101334,
      "grad_norm": 0.026263011619448662,
      "learning_rate": 3.450894124325859e-05,
      "loss": 0.0002931110560894012,
      "step": 230820
    },
    {
      "epoch": 65.5208629009367,
      "grad_norm": 0.4610153138637543,
      "learning_rate": 3.450610275333523e-05,
      "loss": 0.0003575865179300308,
      "step": 230830
    },
    {
      "epoch": 65.52370139086007,
      "grad_norm": 0.007482717279344797,
      "learning_rate": 3.4503264263411865e-05,
      "loss": 0.0002703210338950157,
      "step": 230840
    },
    {
      "epoch": 65.52653988078342,
      "grad_norm": 0.021622737869620323,
      "learning_rate": 3.4500425773488506e-05,
      "loss": 0.00036898069083690643,
      "step": 230850
    },
    {
      "epoch": 65.52937837070678,
      "grad_norm": 0.068491131067276,
      "learning_rate": 3.449758728356515e-05,
      "loss": 0.00039840303361415863,
      "step": 230860
    },
    {
      "epoch": 65.53221686063014,
      "grad_norm": 0.02338431030511856,
      "learning_rate": 3.449474879364178e-05,
      "loss": 0.0004004988819360733,
      "step": 230870
    },
    {
      "epoch": 65.5350553505535,
      "grad_norm": 0.1574171930551529,
      "learning_rate": 3.4491910303718424e-05,
      "loss": 0.0004371890798211098,
      "step": 230880
    },
    {
      "epoch": 65.53789384047687,
      "grad_norm": 0.01621890813112259,
      "learning_rate": 3.4489071813795065e-05,
      "loss": 0.00015305392444133758,
      "step": 230890
    },
    {
      "epoch": 65.54073233040023,
      "grad_norm": 0.049762826412916183,
      "learning_rate": 3.44862333238717e-05,
      "loss": 0.00012197401374578476,
      "step": 230900
    },
    {
      "epoch": 65.5435708203236,
      "grad_norm": 0.05430343747138977,
      "learning_rate": 3.448339483394834e-05,
      "loss": 0.00033241156488657,
      "step": 230910
    },
    {
      "epoch": 65.54640931024694,
      "grad_norm": 0.20650848746299744,
      "learning_rate": 3.4480556344024975e-05,
      "loss": 0.0002188732847571373,
      "step": 230920
    },
    {
      "epoch": 65.5492478001703,
      "grad_norm": 0.021488307043910027,
      "learning_rate": 3.4477717854101624e-05,
      "loss": 0.00014530587941408157,
      "step": 230930
    },
    {
      "epoch": 65.55208629009367,
      "grad_norm": 0.15565164387226105,
      "learning_rate": 3.447487936417826e-05,
      "loss": 0.0006895087659358978,
      "step": 230940
    },
    {
      "epoch": 65.55492478001703,
      "grad_norm": 0.08786024153232574,
      "learning_rate": 3.447204087425489e-05,
      "loss": 0.009687669575214386,
      "step": 230950
    },
    {
      "epoch": 65.5577632699404,
      "grad_norm": 0.07408939301967621,
      "learning_rate": 3.446920238433154e-05,
      "loss": 0.0005891254171729088,
      "step": 230960
    },
    {
      "epoch": 65.56060175986376,
      "grad_norm": 0.0551992692053318,
      "learning_rate": 3.4466363894408176e-05,
      "loss": 0.00021363552659749984,
      "step": 230970
    },
    {
      "epoch": 65.5634402497871,
      "grad_norm": 0.015658782795071602,
      "learning_rate": 3.446352540448482e-05,
      "loss": 0.00010338053107261658,
      "step": 230980
    },
    {
      "epoch": 65.56627873971047,
      "grad_norm": 0.02230912074446678,
      "learning_rate": 3.446068691456146e-05,
      "loss": 0.00015950407832860948,
      "step": 230990
    },
    {
      "epoch": 65.56911722963383,
      "grad_norm": 0.3710186183452606,
      "learning_rate": 3.445784842463809e-05,
      "loss": 0.00012608412653207778,
      "step": 231000
    },
    {
      "epoch": 65.56911722963383,
      "eval_accuracy": 0.9790805620906721,
      "eval_loss": 0.07118082791566849,
      "eval_runtime": 33.0474,
      "eval_samples_per_second": 475.893,
      "eval_steps_per_second": 7.444,
      "step": 231000
    },
    {
      "epoch": 65.5719557195572,
      "grad_norm": 0.025240499526262283,
      "learning_rate": 3.4455009934714734e-05,
      "loss": 0.00010130684822797775,
      "step": 231010
    },
    {
      "epoch": 65.57479420948056,
      "grad_norm": 0.0051840199157595634,
      "learning_rate": 3.445217144479137e-05,
      "loss": 9.222421795129776e-05,
      "step": 231020
    },
    {
      "epoch": 65.57763269940392,
      "grad_norm": 0.0024436688981950283,
      "learning_rate": 3.444933295486801e-05,
      "loss": 8.761975914239883e-05,
      "step": 231030
    },
    {
      "epoch": 65.58047118932728,
      "grad_norm": 0.006105129607021809,
      "learning_rate": 3.444649446494465e-05,
      "loss": 6.300210952758789e-05,
      "step": 231040
    },
    {
      "epoch": 65.58330967925063,
      "grad_norm": 0.10514181852340698,
      "learning_rate": 3.4443655975021286e-05,
      "loss": 0.0001981062814593315,
      "step": 231050
    },
    {
      "epoch": 65.586148169174,
      "grad_norm": 0.06704547256231308,
      "learning_rate": 3.4440817485097934e-05,
      "loss": 0.0014233734458684922,
      "step": 231060
    },
    {
      "epoch": 65.58898665909736,
      "grad_norm": 0.017359917983412743,
      "learning_rate": 3.443797899517457e-05,
      "loss": 0.0008830849081277847,
      "step": 231070
    },
    {
      "epoch": 65.59182514902072,
      "grad_norm": 0.7944636940956116,
      "learning_rate": 3.4435140505251204e-05,
      "loss": 0.0001443542540073395,
      "step": 231080
    },
    {
      "epoch": 65.59466363894408,
      "grad_norm": 0.911065936088562,
      "learning_rate": 3.443230201532785e-05,
      "loss": 0.0002723760902881622,
      "step": 231090
    },
    {
      "epoch": 65.59750212886745,
      "grad_norm": 0.0750395655632019,
      "learning_rate": 3.4429463525404486e-05,
      "loss": 0.00011400282382965088,
      "step": 231100
    },
    {
      "epoch": 65.60034061879081,
      "grad_norm": 0.09097669273614883,
      "learning_rate": 3.442662503548113e-05,
      "loss": 0.00015837717801332475,
      "step": 231110
    },
    {
      "epoch": 65.60317910871416,
      "grad_norm": 0.018021922558546066,
      "learning_rate": 3.442378654555776e-05,
      "loss": 7.276460528373718e-05,
      "step": 231120
    },
    {
      "epoch": 65.60601759863752,
      "grad_norm": 0.009925609454512596,
      "learning_rate": 3.4420948055634404e-05,
      "loss": 0.00012372098863124847,
      "step": 231130
    },
    {
      "epoch": 65.60885608856088,
      "grad_norm": 0.016456788405776024,
      "learning_rate": 3.4418109565711045e-05,
      "loss": 0.0001631377264857292,
      "step": 231140
    },
    {
      "epoch": 65.61169457848425,
      "grad_norm": 0.3011413514614105,
      "learning_rate": 3.441527107578768e-05,
      "loss": 0.0002016771584749222,
      "step": 231150
    },
    {
      "epoch": 65.61453306840761,
      "grad_norm": 0.018949178978800774,
      "learning_rate": 3.441243258586432e-05,
      "loss": 0.0007985735312104226,
      "step": 231160
    },
    {
      "epoch": 65.61737155833097,
      "grad_norm": 0.04617413133382797,
      "learning_rate": 3.440959409594096e-05,
      "loss": 0.0008039426058530808,
      "step": 231170
    },
    {
      "epoch": 65.62021004825434,
      "grad_norm": 0.3014024794101715,
      "learning_rate": 3.44067556060176e-05,
      "loss": 0.0009149014949798584,
      "step": 231180
    },
    {
      "epoch": 65.62304853817768,
      "grad_norm": 0.018783049657940865,
      "learning_rate": 3.4403917116094245e-05,
      "loss": 0.0010236378759145737,
      "step": 231190
    },
    {
      "epoch": 65.62588702810105,
      "grad_norm": 0.0816238522529602,
      "learning_rate": 3.440107862617088e-05,
      "loss": 0.0007648428902029992,
      "step": 231200
    },
    {
      "epoch": 65.62872551802441,
      "grad_norm": 0.023723937571048737,
      "learning_rate": 3.4398240136247514e-05,
      "loss": 0.0009662536904215813,
      "step": 231210
    },
    {
      "epoch": 65.63156400794777,
      "grad_norm": 0.02421645075082779,
      "learning_rate": 3.4395401646324156e-05,
      "loss": 0.0009558778256177902,
      "step": 231220
    },
    {
      "epoch": 65.63440249787114,
      "grad_norm": 0.030539389699697495,
      "learning_rate": 3.43925631564008e-05,
      "loss": 0.0012198476120829581,
      "step": 231230
    },
    {
      "epoch": 65.6372409877945,
      "grad_norm": 0.3311913311481476,
      "learning_rate": 3.438972466647744e-05,
      "loss": 0.008245759457349778,
      "step": 231240
    },
    {
      "epoch": 65.64007947771785,
      "grad_norm": 0.09007830172777176,
      "learning_rate": 3.438688617655407e-05,
      "loss": 0.00038088858127593994,
      "step": 231250
    },
    {
      "epoch": 65.64291796764121,
      "grad_norm": 0.07241680473089218,
      "learning_rate": 3.4384047686630714e-05,
      "loss": 0.00047122817486524584,
      "step": 231260
    },
    {
      "epoch": 65.64575645756457,
      "grad_norm": 0.43515029549598694,
      "learning_rate": 3.4381209196707356e-05,
      "loss": 0.0005272591486573219,
      "step": 231270
    },
    {
      "epoch": 65.64859494748794,
      "grad_norm": 0.042012769728899,
      "learning_rate": 3.437837070678399e-05,
      "loss": 0.000399027019739151,
      "step": 231280
    },
    {
      "epoch": 65.6514334374113,
      "grad_norm": 0.02603253535926342,
      "learning_rate": 3.437553221686063e-05,
      "loss": 0.000435580313205719,
      "step": 231290
    },
    {
      "epoch": 65.65427192733466,
      "grad_norm": 0.01575154811143875,
      "learning_rate": 3.437269372693727e-05,
      "loss": 0.000903334654867649,
      "step": 231300
    },
    {
      "epoch": 65.65711041725802,
      "grad_norm": 0.050132691860198975,
      "learning_rate": 3.436985523701391e-05,
      "loss": 0.0007720757275819778,
      "step": 231310
    },
    {
      "epoch": 65.65994890718137,
      "grad_norm": 0.03551299124956131,
      "learning_rate": 3.436701674709055e-05,
      "loss": 0.00015041660517454148,
      "step": 231320
    },
    {
      "epoch": 65.66278739710474,
      "grad_norm": 0.02867250144481659,
      "learning_rate": 3.436417825716719e-05,
      "loss": 0.0003714941442012787,
      "step": 231330
    },
    {
      "epoch": 65.6656258870281,
      "grad_norm": 0.01396878995001316,
      "learning_rate": 3.4361339767243825e-05,
      "loss": 8.286423981189728e-05,
      "step": 231340
    },
    {
      "epoch": 65.66846437695146,
      "grad_norm": 0.003421800909563899,
      "learning_rate": 3.4358501277320467e-05,
      "loss": 0.0004221295937895775,
      "step": 231350
    },
    {
      "epoch": 65.67130286687483,
      "grad_norm": 0.006106202024966478,
      "learning_rate": 3.435566278739711e-05,
      "loss": 0.0002126520499587059,
      "step": 231360
    },
    {
      "epoch": 65.67414135679819,
      "grad_norm": 0.09426310658454895,
      "learning_rate": 3.435282429747374e-05,
      "loss": 0.00020603574812412263,
      "step": 231370
    },
    {
      "epoch": 65.67697984672155,
      "grad_norm": 0.011370202526450157,
      "learning_rate": 3.4349985807550384e-05,
      "loss": 0.0002227712422609329,
      "step": 231380
    },
    {
      "epoch": 65.6798183366449,
      "grad_norm": 0.013650527223944664,
      "learning_rate": 3.4347147317627025e-05,
      "loss": 0.0003604702651500702,
      "step": 231390
    },
    {
      "epoch": 65.68265682656826,
      "grad_norm": 0.017411530017852783,
      "learning_rate": 3.4344308827703667e-05,
      "loss": 0.00010625701397657395,
      "step": 231400
    },
    {
      "epoch": 65.68549531649163,
      "grad_norm": 0.09672416746616364,
      "learning_rate": 3.43414703377803e-05,
      "loss": 0.00021499376744031905,
      "step": 231410
    },
    {
      "epoch": 65.68833380641499,
      "grad_norm": 0.06511159986257553,
      "learning_rate": 3.4338631847856936e-05,
      "loss": 0.00035912711173295974,
      "step": 231420
    },
    {
      "epoch": 65.69117229633835,
      "grad_norm": 0.017238302156329155,
      "learning_rate": 3.4335793357933584e-05,
      "loss": 0.0014703292399644852,
      "step": 231430
    },
    {
      "epoch": 65.69401078626171,
      "grad_norm": 0.016446134075522423,
      "learning_rate": 3.433295486801022e-05,
      "loss": 0.0022742325440049172,
      "step": 231440
    },
    {
      "epoch": 65.69684927618506,
      "grad_norm": 0.3073159456253052,
      "learning_rate": 3.433011637808686e-05,
      "loss": 0.0021853266283869744,
      "step": 231450
    },
    {
      "epoch": 65.69968776610843,
      "grad_norm": 1.3966937065124512,
      "learning_rate": 3.43272778881635e-05,
      "loss": 0.004742862284183502,
      "step": 231460
    },
    {
      "epoch": 65.70252625603179,
      "grad_norm": 0.07108964025974274,
      "learning_rate": 3.4324439398240136e-05,
      "loss": 0.00022201966494321822,
      "step": 231470
    },
    {
      "epoch": 65.70536474595515,
      "grad_norm": 0.07788769900798798,
      "learning_rate": 3.432160090831678e-05,
      "loss": 0.00041264742612838744,
      "step": 231480
    },
    {
      "epoch": 65.70820323587851,
      "grad_norm": 0.8292551636695862,
      "learning_rate": 3.431876241839342e-05,
      "loss": 0.0015250638127326966,
      "step": 231490
    },
    {
      "epoch": 65.71104172580188,
      "grad_norm": 0.7885483503341675,
      "learning_rate": 3.431592392847005e-05,
      "loss": 0.0006885522976517678,
      "step": 231500
    },
    {
      "epoch": 65.71104172580188,
      "eval_accuracy": 0.9771094296432886,
      "eval_loss": 0.08630332350730896,
      "eval_runtime": 32.822,
      "eval_samples_per_second": 479.16,
      "eval_steps_per_second": 7.495,
      "step": 231500
    },
    {
      "epoch": 65.71388021572524,
      "grad_norm": 0.0921783596277237,
      "learning_rate": 3.4313085438546695e-05,
      "loss": 0.001633262261748314,
      "step": 231510
    },
    {
      "epoch": 65.71671870564859,
      "grad_norm": 0.05505790933966637,
      "learning_rate": 3.431024694862333e-05,
      "loss": 0.0015654198825359345,
      "step": 231520
    },
    {
      "epoch": 65.71955719557195,
      "grad_norm": 0.1669234186410904,
      "learning_rate": 3.430740845869998e-05,
      "loss": 0.0003402795642614365,
      "step": 231530
    },
    {
      "epoch": 65.72239568549531,
      "grad_norm": 1.4129343032836914,
      "learning_rate": 3.430456996877661e-05,
      "loss": 0.004977821186184883,
      "step": 231540
    },
    {
      "epoch": 65.72523417541868,
      "grad_norm": 0.030669525265693665,
      "learning_rate": 3.4301731478853247e-05,
      "loss": 0.00026328042149543763,
      "step": 231550
    },
    {
      "epoch": 65.72807266534204,
      "grad_norm": 0.2314763367176056,
      "learning_rate": 3.4298892988929895e-05,
      "loss": 0.0009986249729990958,
      "step": 231560
    },
    {
      "epoch": 65.7309111552654,
      "grad_norm": 0.0065121129155159,
      "learning_rate": 3.429605449900653e-05,
      "loss": 0.002551092579960823,
      "step": 231570
    },
    {
      "epoch": 65.73374964518877,
      "grad_norm": 0.06986025720834732,
      "learning_rate": 3.429321600908317e-05,
      "loss": 0.0008673910051584244,
      "step": 231580
    },
    {
      "epoch": 65.73658813511211,
      "grad_norm": 0.1898462474346161,
      "learning_rate": 3.429037751915981e-05,
      "loss": 0.0010048115625977516,
      "step": 231590
    },
    {
      "epoch": 65.73942662503548,
      "grad_norm": 0.11327793449163437,
      "learning_rate": 3.428753902923645e-05,
      "loss": 0.0013318700715899468,
      "step": 231600
    },
    {
      "epoch": 65.74226511495884,
      "grad_norm": 0.020107487216591835,
      "learning_rate": 3.428470053931309e-05,
      "loss": 0.004937111586332321,
      "step": 231610
    },
    {
      "epoch": 65.7451036048822,
      "grad_norm": 0.03388233110308647,
      "learning_rate": 3.428186204938972e-05,
      "loss": 0.00048382841050624846,
      "step": 231620
    },
    {
      "epoch": 65.74794209480557,
      "grad_norm": 0.06529341638088226,
      "learning_rate": 3.4279023559466364e-05,
      "loss": 0.001849549449980259,
      "step": 231630
    },
    {
      "epoch": 65.75078058472893,
      "grad_norm": 1.0802533626556396,
      "learning_rate": 3.4276185069543005e-05,
      "loss": 0.00024856925010681153,
      "step": 231640
    },
    {
      "epoch": 65.75361907465229,
      "grad_norm": 0.013305902481079102,
      "learning_rate": 3.427334657961964e-05,
      "loss": 0.0003539029508829117,
      "step": 231650
    },
    {
      "epoch": 65.75645756457564,
      "grad_norm": 0.006168501451611519,
      "learning_rate": 3.427050808969629e-05,
      "loss": 0.0008251743391156196,
      "step": 231660
    },
    {
      "epoch": 65.759296054499,
      "grad_norm": 0.03794547915458679,
      "learning_rate": 3.426766959977292e-05,
      "loss": 0.0021901126950979234,
      "step": 231670
    },
    {
      "epoch": 65.76213454442237,
      "grad_norm": 0.0072954557836055756,
      "learning_rate": 3.426483110984956e-05,
      "loss": 0.00040808729827404024,
      "step": 231680
    },
    {
      "epoch": 65.76497303434573,
      "grad_norm": 0.025351418182253838,
      "learning_rate": 3.4261992619926205e-05,
      "loss": 0.00020852982997894288,
      "step": 231690
    },
    {
      "epoch": 65.76781152426909,
      "grad_norm": 0.011151624843478203,
      "learning_rate": 3.425915413000284e-05,
      "loss": 0.0012509185820817948,
      "step": 231700
    },
    {
      "epoch": 65.77065001419246,
      "grad_norm": 0.9641257524490356,
      "learning_rate": 3.425631564007948e-05,
      "loss": 0.00034696385264396665,
      "step": 231710
    },
    {
      "epoch": 65.7734885041158,
      "grad_norm": 0.23323750495910645,
      "learning_rate": 3.425347715015612e-05,
      "loss": 0.0003921885043382645,
      "step": 231720
    },
    {
      "epoch": 65.77632699403917,
      "grad_norm": 2.4443697929382324,
      "learning_rate": 3.425063866023276e-05,
      "loss": 0.001122518628835678,
      "step": 231730
    },
    {
      "epoch": 65.77916548396253,
      "grad_norm": 0.18842996656894684,
      "learning_rate": 3.42478001703094e-05,
      "loss": 0.0005273265764117241,
      "step": 231740
    },
    {
      "epoch": 65.78200397388589,
      "grad_norm": 0.19303396344184875,
      "learning_rate": 3.4244961680386033e-05,
      "loss": 0.00039974935352802274,
      "step": 231750
    },
    {
      "epoch": 65.78484246380926,
      "grad_norm": 0.02381720580160618,
      "learning_rate": 3.4242123190462675e-05,
      "loss": 0.0001455085352063179,
      "step": 231760
    },
    {
      "epoch": 65.78768095373262,
      "grad_norm": 0.08430250734090805,
      "learning_rate": 3.4239284700539316e-05,
      "loss": 0.0004067510366439819,
      "step": 231770
    },
    {
      "epoch": 65.79051944365598,
      "grad_norm": 0.09176656603813171,
      "learning_rate": 3.423644621061595e-05,
      "loss": 0.0002033540979027748,
      "step": 231780
    },
    {
      "epoch": 65.79335793357933,
      "grad_norm": 0.23469425737857819,
      "learning_rate": 3.423360772069259e-05,
      "loss": 0.0003663640469312668,
      "step": 231790
    },
    {
      "epoch": 65.79619642350269,
      "grad_norm": 0.027860073372721672,
      "learning_rate": 3.4230769230769234e-05,
      "loss": 9.603500366210938e-05,
      "step": 231800
    },
    {
      "epoch": 65.79903491342606,
      "grad_norm": 0.1567106693983078,
      "learning_rate": 3.422793074084587e-05,
      "loss": 0.0001372080296278,
      "step": 231810
    },
    {
      "epoch": 65.80187340334942,
      "grad_norm": 0.022114720195531845,
      "learning_rate": 3.4225092250922516e-05,
      "loss": 9.746979922056198e-05,
      "step": 231820
    },
    {
      "epoch": 65.80471189327278,
      "grad_norm": 0.013876757584512234,
      "learning_rate": 3.422225376099915e-05,
      "loss": 0.0002264091745018959,
      "step": 231830
    },
    {
      "epoch": 65.80755038319614,
      "grad_norm": 0.006722757127135992,
      "learning_rate": 3.4219415271075785e-05,
      "loss": 0.00011606048792600632,
      "step": 231840
    },
    {
      "epoch": 65.8103888731195,
      "grad_norm": 0.029834387823939323,
      "learning_rate": 3.421657678115243e-05,
      "loss": 0.000258730910718441,
      "step": 231850
    },
    {
      "epoch": 65.81322736304286,
      "grad_norm": 0.1048969104886055,
      "learning_rate": 3.421373829122907e-05,
      "loss": 0.0014695528894662856,
      "step": 231860
    },
    {
      "epoch": 65.81606585296622,
      "grad_norm": 0.5295614004135132,
      "learning_rate": 3.421089980130571e-05,
      "loss": 0.0002588074654340744,
      "step": 231870
    },
    {
      "epoch": 65.81890434288958,
      "grad_norm": 0.08855278044939041,
      "learning_rate": 3.4208061311382344e-05,
      "loss": 0.0016183022409677506,
      "step": 231880
    },
    {
      "epoch": 65.82174283281294,
      "grad_norm": 0.17444369196891785,
      "learning_rate": 3.4205222821458986e-05,
      "loss": 0.0003184320405125618,
      "step": 231890
    },
    {
      "epoch": 65.82458132273631,
      "grad_norm": 0.06564556807279587,
      "learning_rate": 3.420238433153563e-05,
      "loss": 0.00018995590507984162,
      "step": 231900
    },
    {
      "epoch": 65.82741981265967,
      "grad_norm": 0.1842729151248932,
      "learning_rate": 3.419954584161226e-05,
      "loss": 0.001046636886894703,
      "step": 231910
    },
    {
      "epoch": 65.83025830258302,
      "grad_norm": 0.08029181510210037,
      "learning_rate": 3.41967073516889e-05,
      "loss": 0.00022217053920030593,
      "step": 231920
    },
    {
      "epoch": 65.83309679250638,
      "grad_norm": 0.12270428985357285,
      "learning_rate": 3.4193868861765544e-05,
      "loss": 0.00029267873615026476,
      "step": 231930
    },
    {
      "epoch": 65.83593528242974,
      "grad_norm": 0.02139168418943882,
      "learning_rate": 3.419103037184218e-05,
      "loss": 0.0002420227974653244,
      "step": 231940
    },
    {
      "epoch": 65.83877377235311,
      "grad_norm": 0.03319339081645012,
      "learning_rate": 3.418819188191882e-05,
      "loss": 0.00037816055119037627,
      "step": 231950
    },
    {
      "epoch": 65.84161226227647,
      "grad_norm": 0.1663677990436554,
      "learning_rate": 3.418535339199546e-05,
      "loss": 0.0010635420680046082,
      "step": 231960
    },
    {
      "epoch": 65.84445075219983,
      "grad_norm": 0.019096067175269127,
      "learning_rate": 3.4182514902072096e-05,
      "loss": 0.0001076769083738327,
      "step": 231970
    },
    {
      "epoch": 65.8472892421232,
      "grad_norm": 0.043338995426893234,
      "learning_rate": 3.417967641214874e-05,
      "loss": 0.0003419457003474236,
      "step": 231980
    },
    {
      "epoch": 65.85012773204654,
      "grad_norm": 0.061249688267707825,
      "learning_rate": 3.417683792222538e-05,
      "loss": 0.00017466023564338684,
      "step": 231990
    },
    {
      "epoch": 65.85296622196991,
      "grad_norm": 0.17403727769851685,
      "learning_rate": 3.417399943230202e-05,
      "loss": 0.0010902166366577148,
      "step": 232000
    },
    {
      "epoch": 65.85296622196991,
      "eval_accuracy": 0.9783811279964393,
      "eval_loss": 0.07650379836559296,
      "eval_runtime": 32.6485,
      "eval_samples_per_second": 481.706,
      "eval_steps_per_second": 7.535,
      "step": 232000
    },
    {
      "epoch": 65.85580471189327,
      "grad_norm": 0.033439431339502335,
      "learning_rate": 3.4171160942378655e-05,
      "loss": 0.0003294188529253006,
      "step": 232010
    },
    {
      "epoch": 65.85864320181663,
      "grad_norm": 5.664687633514404,
      "learning_rate": 3.4168322452455296e-05,
      "loss": 0.0005848128348588943,
      "step": 232020
    },
    {
      "epoch": 65.86148169174,
      "grad_norm": 0.0294631440192461,
      "learning_rate": 3.416548396253194e-05,
      "loss": 8.283164352178574e-05,
      "step": 232030
    },
    {
      "epoch": 65.86432018166336,
      "grad_norm": 0.022696491330862045,
      "learning_rate": 3.416264547260857e-05,
      "loss": 0.00017242655158042908,
      "step": 232040
    },
    {
      "epoch": 65.86715867158672,
      "grad_norm": 0.05554160475730896,
      "learning_rate": 3.4159806982685214e-05,
      "loss": 0.0005330335348844528,
      "step": 232050
    },
    {
      "epoch": 65.86999716151007,
      "grad_norm": 0.1565648913383484,
      "learning_rate": 3.4156968492761855e-05,
      "loss": 0.00017522424459457398,
      "step": 232060
    },
    {
      "epoch": 65.87283565143343,
      "grad_norm": 0.009212814271450043,
      "learning_rate": 3.415413000283849e-05,
      "loss": 9.92894172668457e-05,
      "step": 232070
    },
    {
      "epoch": 65.8756741413568,
      "grad_norm": 0.04812614992260933,
      "learning_rate": 3.415129151291513e-05,
      "loss": 0.00022208280861377717,
      "step": 232080
    },
    {
      "epoch": 65.87851263128016,
      "grad_norm": 0.02265743725001812,
      "learning_rate": 3.414845302299177e-05,
      "loss": 5.548335611820221e-05,
      "step": 232090
    },
    {
      "epoch": 65.88135112120352,
      "grad_norm": 0.03143341466784477,
      "learning_rate": 3.414561453306841e-05,
      "loss": 0.00014950018376111985,
      "step": 232100
    },
    {
      "epoch": 65.88418961112689,
      "grad_norm": 0.09819622337818146,
      "learning_rate": 3.414277604314505e-05,
      "loss": 0.0004644446074962616,
      "step": 232110
    },
    {
      "epoch": 65.88702810105025,
      "grad_norm": 0.01867305114865303,
      "learning_rate": 3.413993755322169e-05,
      "loss": 0.0003481244668364525,
      "step": 232120
    },
    {
      "epoch": 65.8898665909736,
      "grad_norm": 0.001553376903757453,
      "learning_rate": 3.413709906329833e-05,
      "loss": 0.0031631909310817718,
      "step": 232130
    },
    {
      "epoch": 65.89270508089696,
      "grad_norm": 1.1753427982330322,
      "learning_rate": 3.4134260573374966e-05,
      "loss": 0.0007697656750679016,
      "step": 232140
    },
    {
      "epoch": 65.89554357082032,
      "grad_norm": 0.40565139055252075,
      "learning_rate": 3.41314220834516e-05,
      "loss": 0.0068141654133796695,
      "step": 232150
    },
    {
      "epoch": 65.89838206074369,
      "grad_norm": 0.07705086469650269,
      "learning_rate": 3.412858359352825e-05,
      "loss": 0.00012869518250226974,
      "step": 232160
    },
    {
      "epoch": 65.90122055066705,
      "grad_norm": 0.03706358000636101,
      "learning_rate": 3.412574510360488e-05,
      "loss": 0.00018549133092164992,
      "step": 232170
    },
    {
      "epoch": 65.90405904059041,
      "grad_norm": 0.016309289261698723,
      "learning_rate": 3.4122906613681524e-05,
      "loss": 0.0007084537297487259,
      "step": 232180
    },
    {
      "epoch": 65.90689753051376,
      "grad_norm": 0.015377521514892578,
      "learning_rate": 3.4120068123758166e-05,
      "loss": 0.00025738198310136794,
      "step": 232190
    },
    {
      "epoch": 65.90973602043712,
      "grad_norm": 0.016931679099798203,
      "learning_rate": 3.41172296338348e-05,
      "loss": 0.00020135659724473953,
      "step": 232200
    },
    {
      "epoch": 65.91257451036049,
      "grad_norm": 0.18335749208927155,
      "learning_rate": 3.411439114391144e-05,
      "loss": 0.00016441401094198228,
      "step": 232210
    },
    {
      "epoch": 65.91541300028385,
      "grad_norm": 0.03733351081609726,
      "learning_rate": 3.411155265398808e-05,
      "loss": 0.00143485888838768,
      "step": 232220
    },
    {
      "epoch": 65.91825149020721,
      "grad_norm": 0.0708928182721138,
      "learning_rate": 3.410871416406472e-05,
      "loss": 0.0003153594210743904,
      "step": 232230
    },
    {
      "epoch": 65.92108998013057,
      "grad_norm": 0.05072992667555809,
      "learning_rate": 3.410587567414136e-05,
      "loss": 0.00040464773774147035,
      "step": 232240
    },
    {
      "epoch": 65.92392847005394,
      "grad_norm": 0.015472141094505787,
      "learning_rate": 3.4103037184217994e-05,
      "loss": 0.0003967206925153732,
      "step": 232250
    },
    {
      "epoch": 65.92676695997729,
      "grad_norm": 0.10730788856744766,
      "learning_rate": 3.4100198694294635e-05,
      "loss": 0.0004399750381708145,
      "step": 232260
    },
    {
      "epoch": 65.92960544990065,
      "grad_norm": 8.229979515075684,
      "learning_rate": 3.4097360204371276e-05,
      "loss": 0.002939872071146965,
      "step": 232270
    },
    {
      "epoch": 65.93244393982401,
      "grad_norm": 0.016142047941684723,
      "learning_rate": 3.409452171444791e-05,
      "loss": 0.0002686440944671631,
      "step": 232280
    },
    {
      "epoch": 65.93528242974737,
      "grad_norm": 16.668062210083008,
      "learning_rate": 3.409168322452456e-05,
      "loss": 0.003414344787597656,
      "step": 232290
    },
    {
      "epoch": 65.93812091967074,
      "grad_norm": 0.28718504309654236,
      "learning_rate": 3.4088844734601194e-05,
      "loss": 0.0007456950843334198,
      "step": 232300
    },
    {
      "epoch": 65.9409594095941,
      "grad_norm": 0.027415936812758446,
      "learning_rate": 3.408600624467783e-05,
      "loss": 0.0003516515716910362,
      "step": 232310
    },
    {
      "epoch": 65.94379789951746,
      "grad_norm": 0.009290101006627083,
      "learning_rate": 3.4083167754754477e-05,
      "loss": 0.004120992869138718,
      "step": 232320
    },
    {
      "epoch": 65.94663638944081,
      "grad_norm": 0.03956007584929466,
      "learning_rate": 3.408032926483111e-05,
      "loss": 0.0002362443134188652,
      "step": 232330
    },
    {
      "epoch": 65.94947487936417,
      "grad_norm": 0.024351069703698158,
      "learning_rate": 3.407749077490775e-05,
      "loss": 0.00025674719363451005,
      "step": 232340
    },
    {
      "epoch": 65.95231336928754,
      "grad_norm": 0.6920332908630371,
      "learning_rate": 3.407465228498439e-05,
      "loss": 0.0003547629341483116,
      "step": 232350
    },
    {
      "epoch": 65.9551518592109,
      "grad_norm": 0.10338646918535233,
      "learning_rate": 3.407181379506103e-05,
      "loss": 0.0008632360026240349,
      "step": 232360
    },
    {
      "epoch": 65.95799034913426,
      "grad_norm": 4.255619049072266,
      "learning_rate": 3.406897530513767e-05,
      "loss": 0.0006251191720366478,
      "step": 232370
    },
    {
      "epoch": 65.96082883905763,
      "grad_norm": 0.6003546118736267,
      "learning_rate": 3.4066136815214304e-05,
      "loss": 0.00033625159412622454,
      "step": 232380
    },
    {
      "epoch": 65.96366732898099,
      "grad_norm": 1.0008975267410278,
      "learning_rate": 3.4063298325290946e-05,
      "loss": 0.00020723268389701843,
      "step": 232390
    },
    {
      "epoch": 65.96650581890434,
      "grad_norm": 0.22079218924045563,
      "learning_rate": 3.406045983536759e-05,
      "loss": 0.00025177020579576493,
      "step": 232400
    },
    {
      "epoch": 65.9693443088277,
      "grad_norm": 0.07953465729951859,
      "learning_rate": 3.405762134544422e-05,
      "loss": 0.00020938832312822342,
      "step": 232410
    },
    {
      "epoch": 65.97218279875106,
      "grad_norm": 0.008531776256859303,
      "learning_rate": 3.405478285552087e-05,
      "loss": 0.00025482177734375,
      "step": 232420
    },
    {
      "epoch": 65.97502128867443,
      "grad_norm": 0.02798614650964737,
      "learning_rate": 3.4051944365597505e-05,
      "loss": 0.00021680444478988647,
      "step": 232430
    },
    {
      "epoch": 65.97785977859779,
      "grad_norm": 0.03798951953649521,
      "learning_rate": 3.404910587567414e-05,
      "loss": 0.0005261555314064025,
      "step": 232440
    },
    {
      "epoch": 65.98069826852115,
      "grad_norm": 0.0229087732732296,
      "learning_rate": 3.404626738575078e-05,
      "loss": 0.0001209799200296402,
      "step": 232450
    },
    {
      "epoch": 65.9835367584445,
      "grad_norm": 0.0735493153333664,
      "learning_rate": 3.404342889582742e-05,
      "loss": 0.0005402615293860436,
      "step": 232460
    },
    {
      "epoch": 65.98637524836786,
      "grad_norm": 0.039110127836465836,
      "learning_rate": 3.404059040590406e-05,
      "loss": 0.00024355165660381318,
      "step": 232470
    },
    {
      "epoch": 65.98921373829123,
      "grad_norm": 0.014102994464337826,
      "learning_rate": 3.40377519159807e-05,
      "loss": 0.0003142576664686203,
      "step": 232480
    },
    {
      "epoch": 65.99205222821459,
      "grad_norm": 0.19636334478855133,
      "learning_rate": 3.403491342605734e-05,
      "loss": 0.000492800772190094,
      "step": 232490
    },
    {
      "epoch": 65.99489071813795,
      "grad_norm": 0.028438128530979156,
      "learning_rate": 3.403207493613398e-05,
      "loss": 0.00031008850783109665,
      "step": 232500
    },
    {
      "epoch": 65.99489071813795,
      "eval_accuracy": 0.9788898073376995,
      "eval_loss": 0.07891220599412918,
      "eval_runtime": 32.7657,
      "eval_samples_per_second": 479.983,
      "eval_steps_per_second": 7.508,
      "step": 232500
    },
    {
      "epoch": 65.99772920806132,
      "grad_norm": 0.027484824880957603,
      "learning_rate": 3.4029236446210615e-05,
      "loss": 0.00034668296575546265,
      "step": 232510
    },
    {
      "epoch": 66.00056769798468,
      "grad_norm": 0.0113536911085248,
      "learning_rate": 3.402639795628726e-05,
      "loss": 0.0010433056391775607,
      "step": 232520
    },
    {
      "epoch": 66.00340618790803,
      "grad_norm": 0.07524393498897552,
      "learning_rate": 3.40235594663639e-05,
      "loss": 0.00024408455938100814,
      "step": 232530
    },
    {
      "epoch": 66.00624467783139,
      "grad_norm": 0.0058694458566606045,
      "learning_rate": 3.402072097644053e-05,
      "loss": 9.697992354631423e-05,
      "step": 232540
    },
    {
      "epoch": 66.00908316775475,
      "grad_norm": 0.034445468336343765,
      "learning_rate": 3.4017882486517174e-05,
      "loss": 0.00012011956423521041,
      "step": 232550
    },
    {
      "epoch": 66.01192165767812,
      "grad_norm": 0.022343549877405167,
      "learning_rate": 3.4015043996593815e-05,
      "loss": 0.00018851533532142638,
      "step": 232560
    },
    {
      "epoch": 66.01476014760148,
      "grad_norm": 0.1105780079960823,
      "learning_rate": 3.401220550667045e-05,
      "loss": 0.00010188203305006027,
      "step": 232570
    },
    {
      "epoch": 66.01759863752484,
      "grad_norm": 0.009078293107450008,
      "learning_rate": 3.400936701674709e-05,
      "loss": 7.758978754281997e-05,
      "step": 232580
    },
    {
      "epoch": 66.0204371274482,
      "grad_norm": 0.03215876594185829,
      "learning_rate": 3.400652852682373e-05,
      "loss": 0.0001323288306593895,
      "step": 232590
    },
    {
      "epoch": 66.02327561737155,
      "grad_norm": 0.015220828354358673,
      "learning_rate": 3.4003690036900374e-05,
      "loss": 9.748991578817368e-05,
      "step": 232600
    },
    {
      "epoch": 66.02611410729492,
      "grad_norm": 0.003970368765294552,
      "learning_rate": 3.400085154697701e-05,
      "loss": 9.426083415746689e-05,
      "step": 232610
    },
    {
      "epoch": 66.02895259721828,
      "grad_norm": 0.005593647249042988,
      "learning_rate": 3.399801305705365e-05,
      "loss": 7.393192499876023e-05,
      "step": 232620
    },
    {
      "epoch": 66.03179108714164,
      "grad_norm": 0.47783565521240234,
      "learning_rate": 3.399517456713029e-05,
      "loss": 0.00019857939332723618,
      "step": 232630
    },
    {
      "epoch": 66.034629577065,
      "grad_norm": 0.05755291134119034,
      "learning_rate": 3.3992336077206926e-05,
      "loss": 8.904524147510529e-05,
      "step": 232640
    },
    {
      "epoch": 66.03746806698837,
      "grad_norm": 0.05923078954219818,
      "learning_rate": 3.398949758728357e-05,
      "loss": 6.714519113302231e-05,
      "step": 232650
    },
    {
      "epoch": 66.04030655691172,
      "grad_norm": 0.011720946989953518,
      "learning_rate": 3.398665909736021e-05,
      "loss": 0.00010413862764835358,
      "step": 232660
    },
    {
      "epoch": 66.04314504683508,
      "grad_norm": 0.07505165040493011,
      "learning_rate": 3.398382060743684e-05,
      "loss": 0.0001877306029200554,
      "step": 232670
    },
    {
      "epoch": 66.04598353675844,
      "grad_norm": 0.024829482659697533,
      "learning_rate": 3.3980982117513485e-05,
      "loss": 5.5584311485290525e-05,
      "step": 232680
    },
    {
      "epoch": 66.0488220266818,
      "grad_norm": 0.0053470553830266,
      "learning_rate": 3.3978143627590126e-05,
      "loss": 8.112583309412002e-05,
      "step": 232690
    },
    {
      "epoch": 66.05166051660517,
      "grad_norm": 0.12765564024448395,
      "learning_rate": 3.397530513766676e-05,
      "loss": 0.0001225791871547699,
      "step": 232700
    },
    {
      "epoch": 66.05449900652853,
      "grad_norm": 0.019378110766410828,
      "learning_rate": 3.39724666477434e-05,
      "loss": 0.00013296455144882203,
      "step": 232710
    },
    {
      "epoch": 66.0573374964519,
      "grad_norm": 0.007320824544876814,
      "learning_rate": 3.3969628157820043e-05,
      "loss": 0.00010502766817808152,
      "step": 232720
    },
    {
      "epoch": 66.06017598637524,
      "grad_norm": 0.05537530779838562,
      "learning_rate": 3.396678966789668e-05,
      "loss": 9.499993175268173e-05,
      "step": 232730
    },
    {
      "epoch": 66.0630144762986,
      "grad_norm": 0.3489086627960205,
      "learning_rate": 3.396395117797332e-05,
      "loss": 0.00015031024813652038,
      "step": 232740
    },
    {
      "epoch": 66.06585296622197,
      "grad_norm": 0.3592424690723419,
      "learning_rate": 3.3961112688049954e-05,
      "loss": 0.00036058723926544187,
      "step": 232750
    },
    {
      "epoch": 66.06869145614533,
      "grad_norm": 0.4094430208206177,
      "learning_rate": 3.39582741981266e-05,
      "loss": 0.00020624678581953048,
      "step": 232760
    },
    {
      "epoch": 66.0715299460687,
      "grad_norm": 0.11119238287210464,
      "learning_rate": 3.395543570820324e-05,
      "loss": 0.0001426978036761284,
      "step": 232770
    },
    {
      "epoch": 66.07436843599206,
      "grad_norm": 0.006968047935515642,
      "learning_rate": 3.395259721827987e-05,
      "loss": 9.83385369181633e-05,
      "step": 232780
    },
    {
      "epoch": 66.07720692591542,
      "grad_norm": 0.016121739521622658,
      "learning_rate": 3.394975872835652e-05,
      "loss": 0.0010469211265444756,
      "step": 232790
    },
    {
      "epoch": 66.08004541583877,
      "grad_norm": 0.018648609519004822,
      "learning_rate": 3.3946920238433154e-05,
      "loss": 0.00022006873041391373,
      "step": 232800
    },
    {
      "epoch": 66.08288390576213,
      "grad_norm": 0.00786043331027031,
      "learning_rate": 3.3944081748509796e-05,
      "loss": 7.970035076141358e-05,
      "step": 232810
    },
    {
      "epoch": 66.0857223956855,
      "grad_norm": 0.10460779815912247,
      "learning_rate": 3.394124325858644e-05,
      "loss": 0.0002209451049566269,
      "step": 232820
    },
    {
      "epoch": 66.08856088560886,
      "grad_norm": 0.002288075629621744,
      "learning_rate": 3.393840476866307e-05,
      "loss": 0.0008338961750268936,
      "step": 232830
    },
    {
      "epoch": 66.09139937553222,
      "grad_norm": 0.004492133855819702,
      "learning_rate": 3.393556627873971e-05,
      "loss": 0.0007001109421253205,
      "step": 232840
    },
    {
      "epoch": 66.09423786545558,
      "grad_norm": 0.12490373104810715,
      "learning_rate": 3.393272778881635e-05,
      "loss": 0.0007220461964607238,
      "step": 232850
    },
    {
      "epoch": 66.09707635537895,
      "grad_norm": 0.007942186668515205,
      "learning_rate": 3.392988929889299e-05,
      "loss": 0.0007060157135128975,
      "step": 232860
    },
    {
      "epoch": 66.0999148453023,
      "grad_norm": 0.010541570372879505,
      "learning_rate": 3.392705080896963e-05,
      "loss": 0.0002139287069439888,
      "step": 232870
    },
    {
      "epoch": 66.10275333522566,
      "grad_norm": 0.011709324084222317,
      "learning_rate": 3.3924212319046265e-05,
      "loss": 0.0002936314791440964,
      "step": 232880
    },
    {
      "epoch": 66.10559182514902,
      "grad_norm": 0.10702188313007355,
      "learning_rate": 3.392137382912291e-05,
      "loss": 0.00042288657277822495,
      "step": 232890
    },
    {
      "epoch": 66.10843031507238,
      "grad_norm": 0.007727478630840778,
      "learning_rate": 3.391853533919955e-05,
      "loss": 9.479895234107972e-05,
      "step": 232900
    },
    {
      "epoch": 66.11126880499575,
      "grad_norm": 0.007817096076905727,
      "learning_rate": 3.391569684927618e-05,
      "loss": 0.0002653332427144051,
      "step": 232910
    },
    {
      "epoch": 66.11410729491911,
      "grad_norm": 0.14381739497184753,
      "learning_rate": 3.391285835935283e-05,
      "loss": 0.00012149270623922348,
      "step": 232920
    },
    {
      "epoch": 66.11694578484246,
      "grad_norm": 0.01283132191747427,
      "learning_rate": 3.3910019869429465e-05,
      "loss": 8.952505886554718e-05,
      "step": 232930
    },
    {
      "epoch": 66.11978427476582,
      "grad_norm": 0.02057928778231144,
      "learning_rate": 3.3907181379506106e-05,
      "loss": 0.00014406237751245498,
      "step": 232940
    },
    {
      "epoch": 66.12262276468918,
      "grad_norm": 0.0033099802676588297,
      "learning_rate": 3.390434288958274e-05,
      "loss": 0.0001329910010099411,
      "step": 232950
    },
    {
      "epoch": 66.12546125461255,
      "grad_norm": 0.012909701094031334,
      "learning_rate": 3.390150439965938e-05,
      "loss": 0.00015131011605262756,
      "step": 232960
    },
    {
      "epoch": 66.12829974453591,
      "grad_norm": 0.031819023191928864,
      "learning_rate": 3.3898665909736024e-05,
      "loss": 6.848890334367752e-05,
      "step": 232970
    },
    {
      "epoch": 66.13113823445927,
      "grad_norm": 0.026120631024241447,
      "learning_rate": 3.389582741981266e-05,
      "loss": 0.00014063343405723572,
      "step": 232980
    },
    {
      "epoch": 66.13397672438263,
      "grad_norm": 0.07679503411054611,
      "learning_rate": 3.38929889298893e-05,
      "loss": 0.0005698882043361664,
      "step": 232990
    },
    {
      "epoch": 66.13681521430598,
      "grad_norm": 0.033672019839286804,
      "learning_rate": 3.389015043996594e-05,
      "loss": 0.00016231313347816467,
      "step": 233000
    },
    {
      "epoch": 66.13681521430598,
      "eval_accuracy": 0.9806701850321103,
      "eval_loss": 0.06895015388727188,
      "eval_runtime": 33.4699,
      "eval_samples_per_second": 469.885,
      "eval_steps_per_second": 7.35,
      "step": 233000
    },
    {
      "epoch": 66.13965370422935,
      "grad_norm": 0.006140344776213169,
      "learning_rate": 3.3887311950042576e-05,
      "loss": 0.00010113269090652465,
      "step": 233010
    },
    {
      "epoch": 66.14249219415271,
      "grad_norm": 0.01756472885608673,
      "learning_rate": 3.3884473460119224e-05,
      "loss": 0.00012804381549358368,
      "step": 233020
    },
    {
      "epoch": 66.14533068407607,
      "grad_norm": 0.04008111730217934,
      "learning_rate": 3.388163497019586e-05,
      "loss": 0.00011715982109308242,
      "step": 233030
    },
    {
      "epoch": 66.14816917399943,
      "grad_norm": 0.015966100618243217,
      "learning_rate": 3.387879648027249e-05,
      "loss": 0.0001583324745297432,
      "step": 233040
    },
    {
      "epoch": 66.1510076639228,
      "grad_norm": 0.017154641449451447,
      "learning_rate": 3.387595799034914e-05,
      "loss": 8.74202698469162e-05,
      "step": 233050
    },
    {
      "epoch": 66.15384615384616,
      "grad_norm": 0.004196998663246632,
      "learning_rate": 3.3873119500425776e-05,
      "loss": 5.8916211128234866e-05,
      "step": 233060
    },
    {
      "epoch": 66.15668464376951,
      "grad_norm": 0.0245505440980196,
      "learning_rate": 3.387028101050242e-05,
      "loss": 9.220130741596221e-05,
      "step": 233070
    },
    {
      "epoch": 66.15952313369287,
      "grad_norm": 0.02695852518081665,
      "learning_rate": 3.386744252057905e-05,
      "loss": 9.327903389930726e-05,
      "step": 233080
    },
    {
      "epoch": 66.16236162361623,
      "grad_norm": 0.025310788303613663,
      "learning_rate": 3.386460403065569e-05,
      "loss": 7.252376526594162e-05,
      "step": 233090
    },
    {
      "epoch": 66.1652001135396,
      "grad_norm": 0.004187425132840872,
      "learning_rate": 3.3861765540732334e-05,
      "loss": 6.744340062141419e-05,
      "step": 233100
    },
    {
      "epoch": 66.16803860346296,
      "grad_norm": 0.023863060399889946,
      "learning_rate": 3.385892705080897e-05,
      "loss": 6.474070250988007e-05,
      "step": 233110
    },
    {
      "epoch": 66.17087709338632,
      "grad_norm": 0.024018751457333565,
      "learning_rate": 3.385608856088561e-05,
      "loss": 5.5402703583240506e-05,
      "step": 233120
    },
    {
      "epoch": 66.17371558330967,
      "grad_norm": 0.003072138875722885,
      "learning_rate": 3.385325007096225e-05,
      "loss": 0.0002827305346727371,
      "step": 233130
    },
    {
      "epoch": 66.17655407323304,
      "grad_norm": 0.05802115797996521,
      "learning_rate": 3.3850411581038886e-05,
      "loss": 6.876196712255478e-05,
      "step": 233140
    },
    {
      "epoch": 66.1793925631564,
      "grad_norm": 0.05674227699637413,
      "learning_rate": 3.384757309111553e-05,
      "loss": 0.00018185805529356003,
      "step": 233150
    },
    {
      "epoch": 66.18223105307976,
      "grad_norm": 0.05866632238030434,
      "learning_rate": 3.384473460119217e-05,
      "loss": 0.00018876884132623672,
      "step": 233160
    },
    {
      "epoch": 66.18506954300312,
      "grad_norm": 0.01557983923703432,
      "learning_rate": 3.3841896111268804e-05,
      "loss": 0.00028312448412179945,
      "step": 233170
    },
    {
      "epoch": 66.18790803292649,
      "grad_norm": 0.04470216482877731,
      "learning_rate": 3.3839057621345445e-05,
      "loss": 0.00042217858135700225,
      "step": 233180
    },
    {
      "epoch": 66.19074652284985,
      "grad_norm": 0.013816723600029945,
      "learning_rate": 3.3836219131422086e-05,
      "loss": 0.0003851616755127907,
      "step": 233190
    },
    {
      "epoch": 66.1935850127732,
      "grad_norm": 0.047734037041664124,
      "learning_rate": 3.383338064149872e-05,
      "loss": 0.0038664262741804124,
      "step": 233200
    },
    {
      "epoch": 66.19642350269656,
      "grad_norm": 0.010320871137082577,
      "learning_rate": 3.38308260005677e-05,
      "loss": 0.010418751835823059,
      "step": 233210
    },
    {
      "epoch": 66.19926199261992,
      "grad_norm": 0.17280521988868713,
      "learning_rate": 3.3827987510644336e-05,
      "loss": 0.001406494528055191,
      "step": 233220
    },
    {
      "epoch": 66.20210048254329,
      "grad_norm": 0.04633255675435066,
      "learning_rate": 3.382514902072098e-05,
      "loss": 0.0002667145803570747,
      "step": 233230
    },
    {
      "epoch": 66.20493897246665,
      "grad_norm": 0.04391682893037796,
      "learning_rate": 3.382231053079762e-05,
      "loss": 0.0001771513372659683,
      "step": 233240
    },
    {
      "epoch": 66.20777746239001,
      "grad_norm": 0.014845605008304119,
      "learning_rate": 3.381947204087425e-05,
      "loss": 0.00022705458104610443,
      "step": 233250
    },
    {
      "epoch": 66.21061595231338,
      "grad_norm": 4.894590377807617,
      "learning_rate": 3.3816633550950894e-05,
      "loss": 0.0009203938767313957,
      "step": 233260
    },
    {
      "epoch": 66.21345444223672,
      "grad_norm": 0.4172787070274353,
      "learning_rate": 3.3813795061027536e-05,
      "loss": 0.0009730629622936249,
      "step": 233270
    },
    {
      "epoch": 66.21629293216009,
      "grad_norm": 0.8809516429901123,
      "learning_rate": 3.381095657110418e-05,
      "loss": 0.00041850320994853976,
      "step": 233280
    },
    {
      "epoch": 66.21913142208345,
      "grad_norm": 0.467478483915329,
      "learning_rate": 3.380811808118081e-05,
      "loss": 0.00015327110886573792,
      "step": 233290
    },
    {
      "epoch": 66.22196991200681,
      "grad_norm": 0.016467327252030373,
      "learning_rate": 3.380527959125745e-05,
      "loss": 0.00015284866094589233,
      "step": 233300
    },
    {
      "epoch": 66.22480840193018,
      "grad_norm": 15.403519630432129,
      "learning_rate": 3.3802441101334094e-05,
      "loss": 0.001954525150358677,
      "step": 233310
    },
    {
      "epoch": 66.22764689185354,
      "grad_norm": 0.03341085836291313,
      "learning_rate": 3.379960261141073e-05,
      "loss": 0.0007009433582425117,
      "step": 233320
    },
    {
      "epoch": 66.2304853817769,
      "grad_norm": 0.1050800085067749,
      "learning_rate": 3.379676412148737e-05,
      "loss": 0.004274044930934906,
      "step": 233330
    },
    {
      "epoch": 66.23332387170025,
      "grad_norm": 0.09115778654813766,
      "learning_rate": 3.379392563156401e-05,
      "loss": 0.001788605935871601,
      "step": 233340
    },
    {
      "epoch": 66.23616236162361,
      "grad_norm": 0.02427731454372406,
      "learning_rate": 3.3791087141640646e-05,
      "loss": 0.0003413060680031776,
      "step": 233350
    },
    {
      "epoch": 66.23900085154698,
      "grad_norm": 2.3370800018310547,
      "learning_rate": 3.378824865171729e-05,
      "loss": 0.0005744678899645805,
      "step": 233360
    },
    {
      "epoch": 66.24183934147034,
      "grad_norm": 0.23938773572444916,
      "learning_rate": 3.378541016179393e-05,
      "loss": 0.00043056365102529527,
      "step": 233370
    },
    {
      "epoch": 66.2446778313937,
      "grad_norm": 0.014057081192731857,
      "learning_rate": 3.3782571671870564e-05,
      "loss": 0.0003341382369399071,
      "step": 233380
    },
    {
      "epoch": 66.24751632131706,
      "grad_norm": 0.04909496009349823,
      "learning_rate": 3.3779733181947205e-05,
      "loss": 0.00040461476892232894,
      "step": 233390
    },
    {
      "epoch": 66.25035481124041,
      "grad_norm": 0.12517689168453217,
      "learning_rate": 3.3776894692023846e-05,
      "loss": 0.00032141637057065965,
      "step": 233400
    },
    {
      "epoch": 66.25319330116378,
      "grad_norm": 0.407720148563385,
      "learning_rate": 3.377405620210049e-05,
      "loss": 0.0004510160535573959,
      "step": 233410
    },
    {
      "epoch": 66.25603179108714,
      "grad_norm": 0.022809023037552834,
      "learning_rate": 3.377121771217712e-05,
      "loss": 0.00026702694594860077,
      "step": 233420
    },
    {
      "epoch": 66.2588702810105,
      "grad_norm": 0.08170762658119202,
      "learning_rate": 3.3768379222253764e-05,
      "loss": 0.001897326484322548,
      "step": 233430
    },
    {
      "epoch": 66.26170877093386,
      "grad_norm": 0.09008754789829254,
      "learning_rate": 3.3765540732330405e-05,
      "loss": 0.0006391020491719246,
      "step": 233440
    },
    {
      "epoch": 66.26454726085723,
      "grad_norm": 0.02216252125799656,
      "learning_rate": 3.376270224240704e-05,
      "loss": 0.0011498067528009416,
      "step": 233450
    },
    {
      "epoch": 66.26738575078059,
      "grad_norm": 0.05876651406288147,
      "learning_rate": 3.375986375248368e-05,
      "loss": 0.0004479650408029556,
      "step": 233460
    },
    {
      "epoch": 66.27022424070394,
      "grad_norm": 0.02546064555644989,
      "learning_rate": 3.375702526256032e-05,
      "loss": 0.0015215225517749786,
      "step": 233470
    },
    {
      "epoch": 66.2730627306273,
      "grad_norm": 0.19079254567623138,
      "learning_rate": 3.375418677263696e-05,
      "loss": 0.0022386979311704636,
      "step": 233480
    },
    {
      "epoch": 66.27590122055067,
      "grad_norm": 0.045802898705005646,
      "learning_rate": 3.37513482827136e-05,
      "loss": 0.00035517290234565735,
      "step": 233490
    },
    {
      "epoch": 66.27873971047403,
      "grad_norm": 0.7737349271774292,
      "learning_rate": 3.374850979279024e-05,
      "loss": 0.0003695761784911156,
      "step": 233500
    },
    {
      "epoch": 66.27873971047403,
      "eval_accuracy": 0.9774273542315762,
      "eval_loss": 0.0814509391784668,
      "eval_runtime": 32.3127,
      "eval_samples_per_second": 486.713,
      "eval_steps_per_second": 7.613,
      "step": 233500
    },
    {
      "epoch": 66.28157820039739,
      "grad_norm": 0.1334647238254547,
      "learning_rate": 3.3745671302866874e-05,
      "loss": 0.0022493500262498857,
      "step": 233510
    },
    {
      "epoch": 66.28441669032075,
      "grad_norm": 0.044024188071489334,
      "learning_rate": 3.3742832812943516e-05,
      "loss": 0.0004253266379237175,
      "step": 233520
    },
    {
      "epoch": 66.28725518024412,
      "grad_norm": 0.022740356624126434,
      "learning_rate": 3.373999432302016e-05,
      "loss": 0.002336791902780533,
      "step": 233530
    },
    {
      "epoch": 66.29009367016747,
      "grad_norm": 0.8399578332901001,
      "learning_rate": 3.373715583309679e-05,
      "loss": 0.01088150218129158,
      "step": 233540
    },
    {
      "epoch": 66.29293216009083,
      "grad_norm": 0.016112834215164185,
      "learning_rate": 3.373431734317343e-05,
      "loss": 0.005236045643687248,
      "step": 233550
    },
    {
      "epoch": 66.29577065001419,
      "grad_norm": 0.005988808814436197,
      "learning_rate": 3.373147885325007e-05,
      "loss": 0.0002348063513636589,
      "step": 233560
    },
    {
      "epoch": 66.29860913993755,
      "grad_norm": 0.02491454780101776,
      "learning_rate": 3.3728640363326716e-05,
      "loss": 0.00030242446810007094,
      "step": 233570
    },
    {
      "epoch": 66.30144762986092,
      "grad_norm": 0.47121721506118774,
      "learning_rate": 3.372580187340335e-05,
      "loss": 0.0002539573237299919,
      "step": 233580
    },
    {
      "epoch": 66.30428611978428,
      "grad_norm": 0.019390897825360298,
      "learning_rate": 3.3722963383479985e-05,
      "loss": 0.00037574581801891327,
      "step": 233590
    },
    {
      "epoch": 66.30712460970764,
      "grad_norm": 0.07653500884771347,
      "learning_rate": 3.372012489355663e-05,
      "loss": 0.0004108380526304245,
      "step": 233600
    },
    {
      "epoch": 66.30996309963099,
      "grad_norm": 0.022595003247261047,
      "learning_rate": 3.371728640363327e-05,
      "loss": 0.0017533550038933755,
      "step": 233610
    },
    {
      "epoch": 66.31280158955435,
      "grad_norm": 0.12804417312145233,
      "learning_rate": 3.371444791370991e-05,
      "loss": 0.00099282618612051,
      "step": 233620
    },
    {
      "epoch": 66.31564007947772,
      "grad_norm": 0.011718346737325191,
      "learning_rate": 3.371160942378655e-05,
      "loss": 0.00020436365157365798,
      "step": 233630
    },
    {
      "epoch": 66.31847856940108,
      "grad_norm": 0.053090233355760574,
      "learning_rate": 3.3708770933863185e-05,
      "loss": 0.0037699155509471893,
      "step": 233640
    },
    {
      "epoch": 66.32131705932444,
      "grad_norm": 0.025927796959877014,
      "learning_rate": 3.3705932443939827e-05,
      "loss": 0.0005738068372011185,
      "step": 233650
    },
    {
      "epoch": 66.3241555492478,
      "grad_norm": 0.09879015386104584,
      "learning_rate": 3.370309395401646e-05,
      "loss": 0.0015709768980741502,
      "step": 233660
    },
    {
      "epoch": 66.32699403917115,
      "grad_norm": 0.03706477954983711,
      "learning_rate": 3.37002554640931e-05,
      "loss": 0.00750417560338974,
      "step": 233670
    },
    {
      "epoch": 66.32983252909452,
      "grad_norm": 0.05168786272406578,
      "learning_rate": 3.3697416974169744e-05,
      "loss": 0.0002309763804078102,
      "step": 233680
    },
    {
      "epoch": 66.33267101901788,
      "grad_norm": 0.34112548828125,
      "learning_rate": 3.369457848424638e-05,
      "loss": 0.0004037173464894295,
      "step": 233690
    },
    {
      "epoch": 66.33550950894124,
      "grad_norm": 0.075618214905262,
      "learning_rate": 3.369173999432303e-05,
      "loss": 0.0028164977207779883,
      "step": 233700
    },
    {
      "epoch": 66.3383479988646,
      "grad_norm": 0.06753815710544586,
      "learning_rate": 3.368890150439966e-05,
      "loss": 0.00039541702717542647,
      "step": 233710
    },
    {
      "epoch": 66.34118648878797,
      "grad_norm": 0.01673709787428379,
      "learning_rate": 3.3686063014476296e-05,
      "loss": 0.0016311874613165855,
      "step": 233720
    },
    {
      "epoch": 66.34402497871133,
      "grad_norm": 0.06891032308340073,
      "learning_rate": 3.3683224524552944e-05,
      "loss": 0.002042749896645546,
      "step": 233730
    },
    {
      "epoch": 66.34686346863468,
      "grad_norm": 0.17845384776592255,
      "learning_rate": 3.368038603462958e-05,
      "loss": 0.00040593873709440233,
      "step": 233740
    },
    {
      "epoch": 66.34970195855804,
      "grad_norm": 0.0527908019721508,
      "learning_rate": 3.367754754470622e-05,
      "loss": 0.0005944699048995971,
      "step": 233750
    },
    {
      "epoch": 66.3525404484814,
      "grad_norm": 3.797447443008423,
      "learning_rate": 3.3674709054782855e-05,
      "loss": 0.0007732525467872619,
      "step": 233760
    },
    {
      "epoch": 66.35537893840477,
      "grad_norm": 0.7438935041427612,
      "learning_rate": 3.3671870564859496e-05,
      "loss": 0.0006641054525971413,
      "step": 233770
    },
    {
      "epoch": 66.35821742832813,
      "grad_norm": 0.036317892372608185,
      "learning_rate": 3.366903207493614e-05,
      "loss": 0.00031139124184846877,
      "step": 233780
    },
    {
      "epoch": 66.3610559182515,
      "grad_norm": 0.3196953535079956,
      "learning_rate": 3.366619358501277e-05,
      "loss": 0.00027534011751413346,
      "step": 233790
    },
    {
      "epoch": 66.36389440817486,
      "grad_norm": 0.16758906841278076,
      "learning_rate": 3.366335509508941e-05,
      "loss": 0.0004460209980607033,
      "step": 233800
    },
    {
      "epoch": 66.3667328980982,
      "grad_norm": 0.04394743591547012,
      "learning_rate": 3.3660516605166055e-05,
      "loss": 0.00039349719882011416,
      "step": 233810
    },
    {
      "epoch": 66.36957138802157,
      "grad_norm": 0.026773609220981598,
      "learning_rate": 3.365767811524269e-05,
      "loss": 0.0003765624016523361,
      "step": 233820
    },
    {
      "epoch": 66.37240987794493,
      "grad_norm": 0.011990039609372616,
      "learning_rate": 3.365483962531934e-05,
      "loss": 0.0022205490618944167,
      "step": 233830
    },
    {
      "epoch": 66.3752483678683,
      "grad_norm": 0.2686128616333008,
      "learning_rate": 3.365200113539597e-05,
      "loss": 0.0003711296245455742,
      "step": 233840
    },
    {
      "epoch": 66.37808685779166,
      "grad_norm": 0.8872512578964233,
      "learning_rate": 3.364916264547261e-05,
      "loss": 0.00031977221369743346,
      "step": 233850
    },
    {
      "epoch": 66.38092534771502,
      "grad_norm": 0.10426130145788193,
      "learning_rate": 3.364632415554925e-05,
      "loss": 0.0003214588388800621,
      "step": 233860
    },
    {
      "epoch": 66.38376383763837,
      "grad_norm": 0.2205161303281784,
      "learning_rate": 3.364348566562589e-05,
      "loss": 0.0002000289037823677,
      "step": 233870
    },
    {
      "epoch": 66.38660232756173,
      "grad_norm": 2.6638786792755127,
      "learning_rate": 3.364064717570253e-05,
      "loss": 0.000481300987303257,
      "step": 233880
    },
    {
      "epoch": 66.3894408174851,
      "grad_norm": 0.016233140602707863,
      "learning_rate": 3.3637808685779165e-05,
      "loss": 0.0003298191353678703,
      "step": 233890
    },
    {
      "epoch": 66.39227930740846,
      "grad_norm": 0.036975957453250885,
      "learning_rate": 3.363497019585581e-05,
      "loss": 0.0002529740333557129,
      "step": 233900
    },
    {
      "epoch": 66.39511779733182,
      "grad_norm": 3.6877834796905518,
      "learning_rate": 3.363213170593245e-05,
      "loss": 0.0009254509583115578,
      "step": 233910
    },
    {
      "epoch": 66.39795628725518,
      "grad_norm": 0.050573185086250305,
      "learning_rate": 3.362929321600908e-05,
      "loss": 0.00644277036190033,
      "step": 233920
    },
    {
      "epoch": 66.40079477717855,
      "grad_norm": 0.04055232182145119,
      "learning_rate": 3.3626454726085724e-05,
      "loss": 0.0003745067864656448,
      "step": 233930
    },
    {
      "epoch": 66.4036332671019,
      "grad_norm": 0.03012969344854355,
      "learning_rate": 3.3623616236162365e-05,
      "loss": 0.012364494800567626,
      "step": 233940
    },
    {
      "epoch": 66.40647175702526,
      "grad_norm": 0.007608199492096901,
      "learning_rate": 3.3620777746239e-05,
      "loss": 0.0003255065530538559,
      "step": 233950
    },
    {
      "epoch": 66.40931024694862,
      "grad_norm": 0.14091965556144714,
      "learning_rate": 3.361793925631564e-05,
      "loss": 0.00043797027319669724,
      "step": 233960
    },
    {
      "epoch": 66.41214873687198,
      "grad_norm": 0.013468250632286072,
      "learning_rate": 3.361510076639228e-05,
      "loss": 0.0003254568204283714,
      "step": 233970
    },
    {
      "epoch": 66.41498722679535,
      "grad_norm": 0.019465981051325798,
      "learning_rate": 3.361226227646892e-05,
      "loss": 0.00018247999250888824,
      "step": 233980
    },
    {
      "epoch": 66.41782571671871,
      "grad_norm": 0.0021905768662691116,
      "learning_rate": 3.360942378654556e-05,
      "loss": 0.0005058735609054565,
      "step": 233990
    },
    {
      "epoch": 66.42066420664207,
      "grad_norm": 0.07378256320953369,
      "learning_rate": 3.36065852966222e-05,
      "loss": 0.00017308015376329421,
      "step": 234000
    },
    {
      "epoch": 66.42066420664207,
      "eval_accuracy": 0.977745278819864,
      "eval_loss": 0.08340617269277573,
      "eval_runtime": 32.698,
      "eval_samples_per_second": 480.977,
      "eval_steps_per_second": 7.523,
      "step": 234000
    },
    {
      "epoch": 66.42350269656542,
      "grad_norm": 0.06589341908693314,
      "learning_rate": 3.3603746806698835e-05,
      "loss": 0.0004718272015452385,
      "step": 234010
    },
    {
      "epoch": 66.42634118648878,
      "grad_norm": 0.2931106388568878,
      "learning_rate": 3.3600908316775476e-05,
      "loss": 0.000377688929438591,
      "step": 234020
    },
    {
      "epoch": 66.42917967641215,
      "grad_norm": 3.7266225814819336,
      "learning_rate": 3.359806982685212e-05,
      "loss": 0.0007166512310504914,
      "step": 234030
    },
    {
      "epoch": 66.43201816633551,
      "grad_norm": 0.04121459648013115,
      "learning_rate": 3.359523133692876e-05,
      "loss": 0.00024150721728801727,
      "step": 234040
    },
    {
      "epoch": 66.43485665625887,
      "grad_norm": 0.048184558749198914,
      "learning_rate": 3.3592392847005394e-05,
      "loss": 0.0021021107211709024,
      "step": 234050
    },
    {
      "epoch": 66.43769514618224,
      "grad_norm": 0.007347151171416044,
      "learning_rate": 3.358955435708203e-05,
      "loss": 0.0003110002726316452,
      "step": 234060
    },
    {
      "epoch": 66.4405336361056,
      "grad_norm": 0.058627285063266754,
      "learning_rate": 3.3586715867158676e-05,
      "loss": 0.00026902761310338975,
      "step": 234070
    },
    {
      "epoch": 66.44337212602895,
      "grad_norm": 0.01747162826359272,
      "learning_rate": 3.358387737723531e-05,
      "loss": 0.0001813279464840889,
      "step": 234080
    },
    {
      "epoch": 66.44621061595231,
      "grad_norm": 0.10107839107513428,
      "learning_rate": 3.358103888731195e-05,
      "loss": 0.0001451432704925537,
      "step": 234090
    },
    {
      "epoch": 66.44904910587567,
      "grad_norm": 0.0721186175942421,
      "learning_rate": 3.3578200397388594e-05,
      "loss": 0.00015164129436016083,
      "step": 234100
    },
    {
      "epoch": 66.45188759579904,
      "grad_norm": 0.029579203575849533,
      "learning_rate": 3.357536190746523e-05,
      "loss": 0.00018994417041540146,
      "step": 234110
    },
    {
      "epoch": 66.4547260857224,
      "grad_norm": 0.04557596147060394,
      "learning_rate": 3.357252341754187e-05,
      "loss": 0.00017257016152143477,
      "step": 234120
    },
    {
      "epoch": 66.45756457564576,
      "grad_norm": 0.11672721803188324,
      "learning_rate": 3.356968492761851e-05,
      "loss": 0.0001310892403125763,
      "step": 234130
    },
    {
      "epoch": 66.46040306556911,
      "grad_norm": 10.214113235473633,
      "learning_rate": 3.3566846437695146e-05,
      "loss": 0.0016144296154379846,
      "step": 234140
    },
    {
      "epoch": 66.46324155549247,
      "grad_norm": 0.012520212680101395,
      "learning_rate": 3.356400794777179e-05,
      "loss": 0.0002703780308365822,
      "step": 234150
    },
    {
      "epoch": 66.46608004541584,
      "grad_norm": 0.015992283821105957,
      "learning_rate": 3.356116945784842e-05,
      "loss": 0.0002806922420859337,
      "step": 234160
    },
    {
      "epoch": 66.4689185353392,
      "grad_norm": 0.10218379646539688,
      "learning_rate": 3.355833096792507e-05,
      "loss": 0.0001648576930165291,
      "step": 234170
    },
    {
      "epoch": 66.47175702526256,
      "grad_norm": 0.026773007586598396,
      "learning_rate": 3.3555492478001704e-05,
      "loss": 0.00011934079229831695,
      "step": 234180
    },
    {
      "epoch": 66.47459551518592,
      "grad_norm": 0.0887552872300148,
      "learning_rate": 3.355265398807834e-05,
      "loss": 0.000319310836493969,
      "step": 234190
    },
    {
      "epoch": 66.47743400510929,
      "grad_norm": 0.06088581681251526,
      "learning_rate": 3.354981549815499e-05,
      "loss": 0.00015680640935897828,
      "step": 234200
    },
    {
      "epoch": 66.48027249503264,
      "grad_norm": 0.027781978249549866,
      "learning_rate": 3.354697700823162e-05,
      "loss": 0.0002354322001338005,
      "step": 234210
    },
    {
      "epoch": 66.483110984956,
      "grad_norm": 0.26573261618614197,
      "learning_rate": 3.354413851830826e-05,
      "loss": 0.0003344796597957611,
      "step": 234220
    },
    {
      "epoch": 66.48594947487936,
      "grad_norm": 0.12593974173069,
      "learning_rate": 3.3541300028384904e-05,
      "loss": 0.0006211671978235244,
      "step": 234230
    },
    {
      "epoch": 66.48878796480273,
      "grad_norm": 0.044329237192869186,
      "learning_rate": 3.353846153846154e-05,
      "loss": 0.0008790517225861549,
      "step": 234240
    },
    {
      "epoch": 66.49162645472609,
      "grad_norm": 0.053087569773197174,
      "learning_rate": 3.353562304853818e-05,
      "loss": 0.0009939033538103105,
      "step": 234250
    },
    {
      "epoch": 66.49446494464945,
      "grad_norm": 0.014603345654904842,
      "learning_rate": 3.3532784558614815e-05,
      "loss": 0.00015420559793710709,
      "step": 234260
    },
    {
      "epoch": 66.49730343457281,
      "grad_norm": 0.027813229709863663,
      "learning_rate": 3.3529946068691456e-05,
      "loss": 0.00025293733924627306,
      "step": 234270
    },
    {
      "epoch": 66.50014192449616,
      "grad_norm": 0.5406585335731506,
      "learning_rate": 3.35271075787681e-05,
      "loss": 0.0003246856853365898,
      "step": 234280
    },
    {
      "epoch": 66.50298041441953,
      "grad_norm": 0.021148554980754852,
      "learning_rate": 3.352426908884473e-05,
      "loss": 0.00020490512251853944,
      "step": 234290
    },
    {
      "epoch": 66.50581890434289,
      "grad_norm": 0.6980146765708923,
      "learning_rate": 3.352143059892138e-05,
      "loss": 0.0002507088705897331,
      "step": 234300
    },
    {
      "epoch": 66.50865739426625,
      "grad_norm": 0.1494617909193039,
      "learning_rate": 3.3518592108998015e-05,
      "loss": 0.0009024079889059066,
      "step": 234310
    },
    {
      "epoch": 66.51149588418961,
      "grad_norm": 0.018520154058933258,
      "learning_rate": 3.351575361907465e-05,
      "loss": 9.401645511388778e-05,
      "step": 234320
    },
    {
      "epoch": 66.51433437411298,
      "grad_norm": 0.023435363546013832,
      "learning_rate": 3.35129151291513e-05,
      "loss": 0.00019619259983301164,
      "step": 234330
    },
    {
      "epoch": 66.51717286403633,
      "grad_norm": 0.007228726055473089,
      "learning_rate": 3.351007663922793e-05,
      "loss": 0.0002901187166571617,
      "step": 234340
    },
    {
      "epoch": 66.52001135395969,
      "grad_norm": 0.061569694429636,
      "learning_rate": 3.3507238149304574e-05,
      "loss": 0.0001936899498105049,
      "step": 234350
    },
    {
      "epoch": 66.52284984388305,
      "grad_norm": 0.01215165015310049,
      "learning_rate": 3.350439965938121e-05,
      "loss": 0.00015190783888101577,
      "step": 234360
    },
    {
      "epoch": 66.52568833380641,
      "grad_norm": 0.02288607880473137,
      "learning_rate": 3.350156116945785e-05,
      "loss": 0.000410778820514679,
      "step": 234370
    },
    {
      "epoch": 66.52852682372978,
      "grad_norm": 0.224228173494339,
      "learning_rate": 3.349872267953449e-05,
      "loss": 0.0008495932444930077,
      "step": 234380
    },
    {
      "epoch": 66.53136531365314,
      "grad_norm": 0.03365559130907059,
      "learning_rate": 3.3495884189611126e-05,
      "loss": 0.00028967615216970446,
      "step": 234390
    },
    {
      "epoch": 66.5342038035765,
      "grad_norm": 0.021349044516682625,
      "learning_rate": 3.349304569968777e-05,
      "loss": 8.82122665643692e-05,
      "step": 234400
    },
    {
      "epoch": 66.53704229349985,
      "grad_norm": 0.010580348782241344,
      "learning_rate": 3.349020720976441e-05,
      "loss": 0.00011894870549440384,
      "step": 234410
    },
    {
      "epoch": 66.53988078342321,
      "grad_norm": 0.007859664037823677,
      "learning_rate": 3.348736871984104e-05,
      "loss": 0.00027098767459392547,
      "step": 234420
    },
    {
      "epoch": 66.54271927334658,
      "grad_norm": 0.17356199026107788,
      "learning_rate": 3.3484530229917684e-05,
      "loss": 0.00022247806191444397,
      "step": 234430
    },
    {
      "epoch": 66.54555776326994,
      "grad_norm": 0.48101910948753357,
      "learning_rate": 3.3481691739994326e-05,
      "loss": 0.0006773065775632858,
      "step": 234440
    },
    {
      "epoch": 66.5483962531933,
      "grad_norm": 0.01915062591433525,
      "learning_rate": 3.347885325007096e-05,
      "loss": 0.0001407390460371971,
      "step": 234450
    },
    {
      "epoch": 66.55123474311667,
      "grad_norm": 0.10281495004892349,
      "learning_rate": 3.34760147601476e-05,
      "loss": 0.0003403542563319206,
      "step": 234460
    },
    {
      "epoch": 66.55407323304003,
      "grad_norm": 0.09245515614748001,
      "learning_rate": 3.347317627022424e-05,
      "loss": 0.0005809513852000237,
      "step": 234470
    },
    {
      "epoch": 66.55691172296338,
      "grad_norm": 0.016669712960720062,
      "learning_rate": 3.347033778030088e-05,
      "loss": 0.00020486824214458466,
      "step": 234480
    },
    {
      "epoch": 66.55975021288674,
      "grad_norm": 0.016073841601610184,
      "learning_rate": 3.346749929037752e-05,
      "loss": 0.0020621690899133684,
      "step": 234490
    },
    {
      "epoch": 66.5625887028101,
      "grad_norm": 0.07154697924852371,
      "learning_rate": 3.346466080045416e-05,
      "loss": 0.00030030272901058195,
      "step": 234500
    },
    {
      "epoch": 66.5625887028101,
      "eval_accuracy": 0.975837731290138,
      "eval_loss": 0.09168843924999237,
      "eval_runtime": 32.922,
      "eval_samples_per_second": 477.705,
      "eval_steps_per_second": 7.472,
      "step": 234500
    },
    {
      "epoch": 66.56542719273347,
      "grad_norm": 0.07031968235969543,
      "learning_rate": 3.34618223105308e-05,
      "loss": 0.002065764181315899,
      "step": 234510
    },
    {
      "epoch": 66.56826568265683,
      "grad_norm": 0.22080214321613312,
      "learning_rate": 3.3458983820607436e-05,
      "loss": 0.0002716893330216408,
      "step": 234520
    },
    {
      "epoch": 66.57110417258019,
      "grad_norm": 0.6440873146057129,
      "learning_rate": 3.345614533068408e-05,
      "loss": 0.00027624107897281645,
      "step": 234530
    },
    {
      "epoch": 66.57394266250355,
      "grad_norm": 0.08325404673814774,
      "learning_rate": 3.345330684076072e-05,
      "loss": 0.0019241798669099807,
      "step": 234540
    },
    {
      "epoch": 66.5767811524269,
      "grad_norm": 0.010023496113717556,
      "learning_rate": 3.3450468350837354e-05,
      "loss": 0.0005561424419283867,
      "step": 234550
    },
    {
      "epoch": 66.57961964235027,
      "grad_norm": 0.8107993006706238,
      "learning_rate": 3.3447629860913995e-05,
      "loss": 0.0032136693596839907,
      "step": 234560
    },
    {
      "epoch": 66.58245813227363,
      "grad_norm": 0.03810854256153107,
      "learning_rate": 3.3444791370990637e-05,
      "loss": 0.0003692707046866417,
      "step": 234570
    },
    {
      "epoch": 66.58529662219699,
      "grad_norm": 0.13931377232074738,
      "learning_rate": 3.344195288106727e-05,
      "loss": 0.00019149426370859146,
      "step": 234580
    },
    {
      "epoch": 66.58813511212036,
      "grad_norm": 0.0054755667224526405,
      "learning_rate": 3.343911439114391e-05,
      "loss": 0.00030373167246580126,
      "step": 234590
    },
    {
      "epoch": 66.59097360204372,
      "grad_norm": 0.010160554200410843,
      "learning_rate": 3.3436275901220554e-05,
      "loss": 0.00018461793661117554,
      "step": 234600
    },
    {
      "epoch": 66.59381209196707,
      "grad_norm": 0.02645380049943924,
      "learning_rate": 3.343343741129719e-05,
      "loss": 0.0006392018869519234,
      "step": 234610
    },
    {
      "epoch": 66.59665058189043,
      "grad_norm": 0.21067896485328674,
      "learning_rate": 3.343059892137383e-05,
      "loss": 0.00034037008881568907,
      "step": 234620
    },
    {
      "epoch": 66.59948907181379,
      "grad_norm": 0.009810393676161766,
      "learning_rate": 3.342776043145047e-05,
      "loss": 0.00025226250290870665,
      "step": 234630
    },
    {
      "epoch": 66.60232756173716,
      "grad_norm": 0.017139559611678123,
      "learning_rate": 3.342492194152711e-05,
      "loss": 0.0002998366951942444,
      "step": 234640
    },
    {
      "epoch": 66.60516605166052,
      "grad_norm": 0.04511978477239609,
      "learning_rate": 3.342208345160375e-05,
      "loss": 0.00017644967883825302,
      "step": 234650
    },
    {
      "epoch": 66.60800454158388,
      "grad_norm": 0.04664217680692673,
      "learning_rate": 3.341924496168038e-05,
      "loss": 0.000860685296356678,
      "step": 234660
    },
    {
      "epoch": 66.61084303150724,
      "grad_norm": 1.1817643642425537,
      "learning_rate": 3.341640647175703e-05,
      "loss": 0.0003343012183904648,
      "step": 234670
    },
    {
      "epoch": 66.61368152143059,
      "grad_norm": 0.017945392057299614,
      "learning_rate": 3.3413567981833665e-05,
      "loss": 0.003979169577360153,
      "step": 234680
    },
    {
      "epoch": 66.61652001135396,
      "grad_norm": 0.014816789887845516,
      "learning_rate": 3.3410729491910306e-05,
      "loss": 0.00016241706907749176,
      "step": 234690
    },
    {
      "epoch": 66.61935850127732,
      "grad_norm": 0.007268806919455528,
      "learning_rate": 3.340789100198695e-05,
      "loss": 0.0011585166677832604,
      "step": 234700
    },
    {
      "epoch": 66.62219699120068,
      "grad_norm": 0.46014514565467834,
      "learning_rate": 3.340505251206358e-05,
      "loss": 0.00252775214612484,
      "step": 234710
    },
    {
      "epoch": 66.62503548112404,
      "grad_norm": 0.11951228231191635,
      "learning_rate": 3.340221402214022e-05,
      "loss": 0.0001375945284962654,
      "step": 234720
    },
    {
      "epoch": 66.6278739710474,
      "grad_norm": 0.07979801297187805,
      "learning_rate": 3.3399375532216865e-05,
      "loss": 0.0038548178970813753,
      "step": 234730
    },
    {
      "epoch": 66.63071246097077,
      "grad_norm": 0.5874942541122437,
      "learning_rate": 3.33965370422935e-05,
      "loss": 0.0004411023110151291,
      "step": 234740
    },
    {
      "epoch": 66.63355095089412,
      "grad_norm": 0.02940540947020054,
      "learning_rate": 3.339369855237014e-05,
      "loss": 0.0011964548379182816,
      "step": 234750
    },
    {
      "epoch": 66.63638944081748,
      "grad_norm": 0.2681537866592407,
      "learning_rate": 3.339086006244678e-05,
      "loss": 0.002540280483663082,
      "step": 234760
    },
    {
      "epoch": 66.63922793074084,
      "grad_norm": 6.239147186279297,
      "learning_rate": 3.3388021572523423e-05,
      "loss": 0.001138824038207531,
      "step": 234770
    },
    {
      "epoch": 66.64206642066421,
      "grad_norm": 0.10596702247858047,
      "learning_rate": 3.338518308260006e-05,
      "loss": 0.00023179370909929276,
      "step": 234780
    },
    {
      "epoch": 66.64490491058757,
      "grad_norm": 0.7297640442848206,
      "learning_rate": 3.338234459267669e-05,
      "loss": 0.00035857707262039186,
      "step": 234790
    },
    {
      "epoch": 66.64774340051093,
      "grad_norm": 0.03160128369927406,
      "learning_rate": 3.337950610275334e-05,
      "loss": 0.0005787720903754235,
      "step": 234800
    },
    {
      "epoch": 66.6505818904343,
      "grad_norm": 0.1653982400894165,
      "learning_rate": 3.3376667612829975e-05,
      "loss": 0.00025734826922416685,
      "step": 234810
    },
    {
      "epoch": 66.65342038035764,
      "grad_norm": 0.0072740837931632996,
      "learning_rate": 3.337382912290662e-05,
      "loss": 0.0003133151680231094,
      "step": 234820
    },
    {
      "epoch": 66.65625887028101,
      "grad_norm": 0.127198725938797,
      "learning_rate": 3.337099063298326e-05,
      "loss": 0.001746891252696514,
      "step": 234830
    },
    {
      "epoch": 66.65909736020437,
      "grad_norm": 0.01219212356954813,
      "learning_rate": 3.336815214305989e-05,
      "loss": 0.0003551337867975235,
      "step": 234840
    },
    {
      "epoch": 66.66193585012773,
      "grad_norm": 0.022250259295105934,
      "learning_rate": 3.3365313653136534e-05,
      "loss": 0.0051961980760097505,
      "step": 234850
    },
    {
      "epoch": 66.6647743400511,
      "grad_norm": 0.0717395693063736,
      "learning_rate": 3.3362475163213175e-05,
      "loss": 0.0003151543438434601,
      "step": 234860
    },
    {
      "epoch": 66.66761282997446,
      "grad_norm": 0.3620184659957886,
      "learning_rate": 3.335963667328981e-05,
      "loss": 0.00012973099946975709,
      "step": 234870
    },
    {
      "epoch": 66.67045131989781,
      "grad_norm": 0.8965954184532166,
      "learning_rate": 3.335679818336645e-05,
      "loss": 0.0005037592723965645,
      "step": 234880
    },
    {
      "epoch": 66.67328980982117,
      "grad_norm": 0.0550510510802269,
      "learning_rate": 3.3353959693443086e-05,
      "loss": 0.00017678085714578629,
      "step": 234890
    },
    {
      "epoch": 66.67612829974453,
      "grad_norm": 0.49060869216918945,
      "learning_rate": 3.335112120351973e-05,
      "loss": 0.0018667595461010933,
      "step": 234900
    },
    {
      "epoch": 66.6789667896679,
      "grad_norm": 0.09097166359424591,
      "learning_rate": 3.334828271359637e-05,
      "loss": 0.0006500331684947014,
      "step": 234910
    },
    {
      "epoch": 66.68180527959126,
      "grad_norm": 0.22833950817584991,
      "learning_rate": 3.3345444223673e-05,
      "loss": 0.00037789177149534227,
      "step": 234920
    },
    {
      "epoch": 66.68464376951462,
      "grad_norm": 0.0640188530087471,
      "learning_rate": 3.334260573374965e-05,
      "loss": 0.00014430582523345948,
      "step": 234930
    },
    {
      "epoch": 66.68748225943799,
      "grad_norm": 0.09780203551054001,
      "learning_rate": 3.3339767243826286e-05,
      "loss": 0.0003250345587730408,
      "step": 234940
    },
    {
      "epoch": 66.69032074936133,
      "grad_norm": 0.09841728210449219,
      "learning_rate": 3.333692875390292e-05,
      "loss": 0.0013363007456064224,
      "step": 234950
    },
    {
      "epoch": 66.6931592392847,
      "grad_norm": 0.05241420120000839,
      "learning_rate": 3.333409026397957e-05,
      "loss": 0.0004927139729261398,
      "step": 234960
    },
    {
      "epoch": 66.69599772920806,
      "grad_norm": 0.21192508935928345,
      "learning_rate": 3.3331251774056203e-05,
      "loss": 0.0004928139969706535,
      "step": 234970
    },
    {
      "epoch": 66.69883621913142,
      "grad_norm": 0.016908567398786545,
      "learning_rate": 3.3328413284132845e-05,
      "loss": 0.0012470027431845665,
      "step": 234980
    },
    {
      "epoch": 66.70167470905479,
      "grad_norm": 0.014833590015769005,
      "learning_rate": 3.332557479420948e-05,
      "loss": 0.0005156420171260834,
      "step": 234990
    },
    {
      "epoch": 66.70451319897815,
      "grad_norm": 0.05100671947002411,
      "learning_rate": 3.332273630428612e-05,
      "loss": 0.00035948809236288073,
      "step": 235000
    },
    {
      "epoch": 66.70451319897815,
      "eval_accuracy": 0.9776816939022064,
      "eval_loss": 0.08278652280569077,
      "eval_runtime": 32.9879,
      "eval_samples_per_second": 476.751,
      "eval_steps_per_second": 7.457,
      "step": 235000
    },
    {
      "epoch": 66.70735168890151,
      "grad_norm": 0.046206194907426834,
      "learning_rate": 3.331989781436276e-05,
      "loss": 0.00023380685597658158,
      "step": 235010
    },
    {
      "epoch": 66.71019017882486,
      "grad_norm": 0.7454606890678406,
      "learning_rate": 3.33170593244394e-05,
      "loss": 0.00045396313071250917,
      "step": 235020
    },
    {
      "epoch": 66.71302866874822,
      "grad_norm": 1.1556059122085571,
      "learning_rate": 3.331422083451604e-05,
      "loss": 0.0007166549563407898,
      "step": 235030
    },
    {
      "epoch": 66.71586715867159,
      "grad_norm": 0.3082740306854248,
      "learning_rate": 3.331138234459268e-05,
      "loss": 0.0019123798236250877,
      "step": 235040
    },
    {
      "epoch": 66.71870564859495,
      "grad_norm": 0.26297125220298767,
      "learning_rate": 3.3308543854669314e-05,
      "loss": 0.0004970375448465347,
      "step": 235050
    },
    {
      "epoch": 66.72154413851831,
      "grad_norm": 0.06687350571155548,
      "learning_rate": 3.330570536474596e-05,
      "loss": 0.0052892804145812985,
      "step": 235060
    },
    {
      "epoch": 66.72438262844167,
      "grad_norm": 4.687132835388184,
      "learning_rate": 3.33028668748226e-05,
      "loss": 0.002394319698214531,
      "step": 235070
    },
    {
      "epoch": 66.72722111836502,
      "grad_norm": 0.0725567638874054,
      "learning_rate": 3.330002838489923e-05,
      "loss": 0.00046697091311216355,
      "step": 235080
    },
    {
      "epoch": 66.73005960828839,
      "grad_norm": 0.0570666529238224,
      "learning_rate": 3.329718989497587e-05,
      "loss": 0.009010647982358932,
      "step": 235090
    },
    {
      "epoch": 66.73289809821175,
      "grad_norm": 0.051440946757793427,
      "learning_rate": 3.3294351405052514e-05,
      "loss": 0.00022988952696323395,
      "step": 235100
    },
    {
      "epoch": 66.73573658813511,
      "grad_norm": 0.08631904423236847,
      "learning_rate": 3.3291512915129156e-05,
      "loss": 0.0002056039869785309,
      "step": 235110
    },
    {
      "epoch": 66.73857507805847,
      "grad_norm": 0.00829800870269537,
      "learning_rate": 3.328867442520579e-05,
      "loss": 0.0003940058872103691,
      "step": 235120
    },
    {
      "epoch": 66.74141356798184,
      "grad_norm": 0.037061478942632675,
      "learning_rate": 3.328583593528243e-05,
      "loss": 0.0004971334710717201,
      "step": 235130
    },
    {
      "epoch": 66.7442520579052,
      "grad_norm": 0.022929465398192406,
      "learning_rate": 3.328299744535907e-05,
      "loss": 0.0016278211027383805,
      "step": 235140
    },
    {
      "epoch": 66.74709054782855,
      "grad_norm": 0.1881955862045288,
      "learning_rate": 3.328015895543571e-05,
      "loss": 0.00035176407545804976,
      "step": 235150
    },
    {
      "epoch": 66.74992903775191,
      "grad_norm": 0.035488251596689224,
      "learning_rate": 3.327732046551235e-05,
      "loss": 0.00022542625665664673,
      "step": 235160
    },
    {
      "epoch": 66.75276752767527,
      "grad_norm": 0.03501230850815773,
      "learning_rate": 3.327448197558899e-05,
      "loss": 0.0013707667589187623,
      "step": 235170
    },
    {
      "epoch": 66.75560601759864,
      "grad_norm": 0.014143884181976318,
      "learning_rate": 3.3271643485665625e-05,
      "loss": 0.0011176902800798417,
      "step": 235180
    },
    {
      "epoch": 66.758444507522,
      "grad_norm": 0.028458207845687866,
      "learning_rate": 3.3268804995742266e-05,
      "loss": 0.00018136892467737197,
      "step": 235190
    },
    {
      "epoch": 66.76128299744536,
      "grad_norm": 0.04401281103491783,
      "learning_rate": 3.326596650581891e-05,
      "loss": 0.0010701535269618035,
      "step": 235200
    },
    {
      "epoch": 66.76412148736873,
      "grad_norm": 0.1862560659646988,
      "learning_rate": 3.326312801589554e-05,
      "loss": 0.00041711609810590744,
      "step": 235210
    },
    {
      "epoch": 66.76695997729207,
      "grad_norm": 0.028274785727262497,
      "learning_rate": 3.3260289525972184e-05,
      "loss": 0.0008732037618756294,
      "step": 235220
    },
    {
      "epoch": 66.76979846721544,
      "grad_norm": 0.08757782727479935,
      "learning_rate": 3.3257451036048825e-05,
      "loss": 0.0005283936858177185,
      "step": 235230
    },
    {
      "epoch": 66.7726369571388,
      "grad_norm": 0.13802844285964966,
      "learning_rate": 3.3254612546125466e-05,
      "loss": 0.0023586975410580637,
      "step": 235240
    },
    {
      "epoch": 66.77547544706216,
      "grad_norm": 0.08174683898687363,
      "learning_rate": 3.32517740562021e-05,
      "loss": 0.0002231638878583908,
      "step": 235250
    },
    {
      "epoch": 66.77831393698553,
      "grad_norm": 0.00873940996825695,
      "learning_rate": 3.324893556627874e-05,
      "loss": 0.0005353214219212532,
      "step": 235260
    },
    {
      "epoch": 66.78115242690889,
      "grad_norm": 0.027028804644942284,
      "learning_rate": 3.3246097076355384e-05,
      "loss": 0.0006082732230424881,
      "step": 235270
    },
    {
      "epoch": 66.78399091683225,
      "grad_norm": 0.044380154460668564,
      "learning_rate": 3.324325858643202e-05,
      "loss": 0.00013571232557296753,
      "step": 235280
    },
    {
      "epoch": 66.7868294067556,
      "grad_norm": 0.682490348815918,
      "learning_rate": 3.324042009650866e-05,
      "loss": 0.0002600835636258125,
      "step": 235290
    },
    {
      "epoch": 66.78966789667896,
      "grad_norm": 0.0991189032793045,
      "learning_rate": 3.32375816065853e-05,
      "loss": 0.00036035943776369095,
      "step": 235300
    },
    {
      "epoch": 66.79250638660233,
      "grad_norm": 0.14955998957157135,
      "learning_rate": 3.3234743116661936e-05,
      "loss": 0.00011868607252836227,
      "step": 235310
    },
    {
      "epoch": 66.79534487652569,
      "grad_norm": 0.0032521849498152733,
      "learning_rate": 3.323190462673858e-05,
      "loss": 7.670726627111435e-05,
      "step": 235320
    },
    {
      "epoch": 66.79818336644905,
      "grad_norm": 0.07133053243160248,
      "learning_rate": 3.322906613681522e-05,
      "loss": 0.00019389744848012923,
      "step": 235330
    },
    {
      "epoch": 66.80102185637242,
      "grad_norm": 0.022408898919820786,
      "learning_rate": 3.322622764689185e-05,
      "loss": 0.00011443421244621277,
      "step": 235340
    },
    {
      "epoch": 66.80386034629576,
      "grad_norm": 0.014638759195804596,
      "learning_rate": 3.3223389156968494e-05,
      "loss": 0.00016908291727304458,
      "step": 235350
    },
    {
      "epoch": 66.80669883621913,
      "grad_norm": 0.014730146154761314,
      "learning_rate": 3.3220550667045136e-05,
      "loss": 0.00011240243911743164,
      "step": 235360
    },
    {
      "epoch": 66.80953732614249,
      "grad_norm": 1.3466508388519287,
      "learning_rate": 3.321771217712177e-05,
      "loss": 0.0005625693127512931,
      "step": 235370
    },
    {
      "epoch": 66.81237581606585,
      "grad_norm": 0.08747067302465439,
      "learning_rate": 3.321487368719841e-05,
      "loss": 0.00019540563225746156,
      "step": 235380
    },
    {
      "epoch": 66.81521430598922,
      "grad_norm": 0.08474695682525635,
      "learning_rate": 3.3212035197275046e-05,
      "loss": 0.00019948557019233704,
      "step": 235390
    },
    {
      "epoch": 66.81805279591258,
      "grad_norm": 0.08218106627464294,
      "learning_rate": 3.3209196707351694e-05,
      "loss": 0.00010387785732746124,
      "step": 235400
    },
    {
      "epoch": 66.82089128583594,
      "grad_norm": 0.06416016072034836,
      "learning_rate": 3.320635821742833e-05,
      "loss": 0.0005580108612775802,
      "step": 235410
    },
    {
      "epoch": 66.82372977575929,
      "grad_norm": 0.16667059063911438,
      "learning_rate": 3.32038035764973e-05,
      "loss": 0.004357136785984039,
      "step": 235420
    },
    {
      "epoch": 66.82656826568265,
      "grad_norm": 0.018854497000575066,
      "learning_rate": 3.3200965086573944e-05,
      "loss": 0.0002620039507746696,
      "step": 235430
    },
    {
      "epoch": 66.82940675560602,
      "grad_norm": 0.01709855906665325,
      "learning_rate": 3.3198126596650585e-05,
      "loss": 9.83433797955513e-05,
      "step": 235440
    },
    {
      "epoch": 66.83224524552938,
      "grad_norm": 0.2866460084915161,
      "learning_rate": 3.3195288106727226e-05,
      "loss": 0.00020274538546800614,
      "step": 235450
    },
    {
      "epoch": 66.83508373545274,
      "grad_norm": 0.020833667367696762,
      "learning_rate": 3.319244961680386e-05,
      "loss": 0.0002082526683807373,
      "step": 235460
    },
    {
      "epoch": 66.8379222253761,
      "grad_norm": 0.013686666265130043,
      "learning_rate": 3.3189611126880496e-05,
      "loss": 8.748061954975129e-05,
      "step": 235470
    },
    {
      "epoch": 66.84076071529947,
      "grad_norm": 0.007855470292270184,
      "learning_rate": 3.3186772636957144e-05,
      "loss": 0.0003111237660050392,
      "step": 235480
    },
    {
      "epoch": 66.84359920522282,
      "grad_norm": 0.0247490294277668,
      "learning_rate": 3.318393414703378e-05,
      "loss": 9.73207876086235e-05,
      "step": 235490
    },
    {
      "epoch": 66.84643769514618,
      "grad_norm": 0.056430425494909286,
      "learning_rate": 3.318109565711042e-05,
      "loss": 0.0012490108609199523,
      "step": 235500
    },
    {
      "epoch": 66.84643769514618,
      "eval_accuracy": 0.9812424492910282,
      "eval_loss": 0.07150499522686005,
      "eval_runtime": 32.4294,
      "eval_samples_per_second": 484.961,
      "eval_steps_per_second": 7.586,
      "step": 235500
    },
    {
      "epoch": 66.84927618506954,
      "grad_norm": 0.13038544356822968,
      "learning_rate": 3.317825716718706e-05,
      "loss": 0.00013990867882966994,
      "step": 235510
    },
    {
      "epoch": 66.8521146749929,
      "grad_norm": 0.012005079537630081,
      "learning_rate": 3.3175418677263696e-05,
      "loss": 0.0002109117805957794,
      "step": 235520
    },
    {
      "epoch": 66.85495316491627,
      "grad_norm": 1.156873106956482,
      "learning_rate": 3.317258018734034e-05,
      "loss": 0.0002911534160375595,
      "step": 235530
    },
    {
      "epoch": 66.85779165483963,
      "grad_norm": 0.01825658790767193,
      "learning_rate": 3.316974169741698e-05,
      "loss": 0.00010190289467573165,
      "step": 235540
    },
    {
      "epoch": 66.86063014476298,
      "grad_norm": 0.11198166757822037,
      "learning_rate": 3.316690320749361e-05,
      "loss": 0.0008565917611122132,
      "step": 235550
    },
    {
      "epoch": 66.86346863468634,
      "grad_norm": 0.011861014179885387,
      "learning_rate": 3.3164064717570254e-05,
      "loss": 0.00015586987137794495,
      "step": 235560
    },
    {
      "epoch": 66.8663071246097,
      "grad_norm": 0.008553036488592625,
      "learning_rate": 3.316122622764689e-05,
      "loss": 0.0001423928886651993,
      "step": 235570
    },
    {
      "epoch": 66.86914561453307,
      "grad_norm": 0.21452787518501282,
      "learning_rate": 3.315838773772354e-05,
      "loss": 0.0011643495410680771,
      "step": 235580
    },
    {
      "epoch": 66.87198410445643,
      "grad_norm": 0.028578972443938255,
      "learning_rate": 3.315554924780017e-05,
      "loss": 0.0001456761732697487,
      "step": 235590
    },
    {
      "epoch": 66.8748225943798,
      "grad_norm": 0.46401247382164,
      "learning_rate": 3.3152710757876806e-05,
      "loss": 0.0009338928386569024,
      "step": 235600
    },
    {
      "epoch": 66.87766108430316,
      "grad_norm": 0.020954176783561707,
      "learning_rate": 3.3149872267953454e-05,
      "loss": 0.00014038663357496262,
      "step": 235610
    },
    {
      "epoch": 66.8804995742265,
      "grad_norm": 0.4040793478488922,
      "learning_rate": 3.314703377803009e-05,
      "loss": 0.0003499394282698631,
      "step": 235620
    },
    {
      "epoch": 66.88333806414987,
      "grad_norm": 0.058987364172935486,
      "learning_rate": 3.314419528810673e-05,
      "loss": 0.00220202449709177,
      "step": 235630
    },
    {
      "epoch": 66.88617655407323,
      "grad_norm": 0.32006245851516724,
      "learning_rate": 3.314135679818337e-05,
      "loss": 0.00024179145693778992,
      "step": 235640
    },
    {
      "epoch": 66.8890150439966,
      "grad_norm": 0.19428128004074097,
      "learning_rate": 3.3138518308260006e-05,
      "loss": 0.0007002253085374832,
      "step": 235650
    },
    {
      "epoch": 66.89185353391996,
      "grad_norm": 0.13538797199726105,
      "learning_rate": 3.313567981833665e-05,
      "loss": 0.00021080709993839263,
      "step": 235660
    },
    {
      "epoch": 66.89469202384332,
      "grad_norm": 6.9469733238220215,
      "learning_rate": 3.313284132841328e-05,
      "loss": 0.005340106040239334,
      "step": 235670
    },
    {
      "epoch": 66.89753051376668,
      "grad_norm": 0.06877300143241882,
      "learning_rate": 3.3130002838489924e-05,
      "loss": 0.0005626648664474488,
      "step": 235680
    },
    {
      "epoch": 66.90036900369003,
      "grad_norm": 4.9698052406311035,
      "learning_rate": 3.3127164348566565e-05,
      "loss": 0.0007434243336319924,
      "step": 235690
    },
    {
      "epoch": 66.9032074936134,
      "grad_norm": 3.7375853061676025,
      "learning_rate": 3.31243258586432e-05,
      "loss": 0.0018354076892137527,
      "step": 235700
    },
    {
      "epoch": 66.90604598353676,
      "grad_norm": 0.010282821953296661,
      "learning_rate": 3.312148736871985e-05,
      "loss": 0.0051972895860672,
      "step": 235710
    },
    {
      "epoch": 66.90888447346012,
      "grad_norm": 0.3919256627559662,
      "learning_rate": 3.311864887879648e-05,
      "loss": 0.00022554751485586165,
      "step": 235720
    },
    {
      "epoch": 66.91172296338348,
      "grad_norm": 0.387600839138031,
      "learning_rate": 3.311581038887312e-05,
      "loss": 0.0003899676725268364,
      "step": 235730
    },
    {
      "epoch": 66.91456145330685,
      "grad_norm": 0.2507328391075134,
      "learning_rate": 3.3112971898949765e-05,
      "loss": 0.00034430380910634997,
      "step": 235740
    },
    {
      "epoch": 66.91739994323021,
      "grad_norm": 0.014165028929710388,
      "learning_rate": 3.31101334090264e-05,
      "loss": 0.0001960117369890213,
      "step": 235750
    },
    {
      "epoch": 66.92023843315356,
      "grad_norm": 0.029842572286725044,
      "learning_rate": 3.310729491910304e-05,
      "loss": 0.0002470238134264946,
      "step": 235760
    },
    {
      "epoch": 66.92307692307692,
      "grad_norm": 0.58656907081604,
      "learning_rate": 3.3104456429179676e-05,
      "loss": 0.0002323698252439499,
      "step": 235770
    },
    {
      "epoch": 66.92591541300028,
      "grad_norm": 0.1217036321759224,
      "learning_rate": 3.310161793925632e-05,
      "loss": 0.00032173618674278257,
      "step": 235780
    },
    {
      "epoch": 66.92875390292365,
      "grad_norm": 0.014387066476047039,
      "learning_rate": 3.309877944933296e-05,
      "loss": 0.006360673904418945,
      "step": 235790
    },
    {
      "epoch": 66.93159239284701,
      "grad_norm": 2.6611039638519287,
      "learning_rate": 3.309594095940959e-05,
      "loss": 0.000576053187251091,
      "step": 235800
    },
    {
      "epoch": 66.93443088277037,
      "grad_norm": 0.014019868336617947,
      "learning_rate": 3.3093102469486235e-05,
      "loss": 0.0005475856363773346,
      "step": 235810
    },
    {
      "epoch": 66.93726937269372,
      "grad_norm": 0.014015834778547287,
      "learning_rate": 3.3090263979562876e-05,
      "loss": 0.0019443454220890999,
      "step": 235820
    },
    {
      "epoch": 66.94010786261708,
      "grad_norm": 0.25374945998191833,
      "learning_rate": 3.308742548963951e-05,
      "loss": 0.000277322344481945,
      "step": 235830
    },
    {
      "epoch": 66.94294635254045,
      "grad_norm": 0.0697697252035141,
      "learning_rate": 3.308458699971615e-05,
      "loss": 0.0034619767218828203,
      "step": 235840
    },
    {
      "epoch": 66.94578484246381,
      "grad_norm": 0.02397332713007927,
      "learning_rate": 3.308174850979279e-05,
      "loss": 0.0021901439875364305,
      "step": 235850
    },
    {
      "epoch": 66.94862333238717,
      "grad_norm": 0.10184179991483688,
      "learning_rate": 3.307891001986943e-05,
      "loss": 0.0005130277946591377,
      "step": 235860
    },
    {
      "epoch": 66.95146182231053,
      "grad_norm": 0.00733168376609683,
      "learning_rate": 3.307607152994607e-05,
      "loss": 0.006280055642127991,
      "step": 235870
    },
    {
      "epoch": 66.9543003122339,
      "grad_norm": 16.944761276245117,
      "learning_rate": 3.307323304002271e-05,
      "loss": 0.0033344339579343798,
      "step": 235880
    },
    {
      "epoch": 66.95713880215725,
      "grad_norm": 1.4785228967666626,
      "learning_rate": 3.3070394550099345e-05,
      "loss": 0.001107693649828434,
      "step": 235890
    },
    {
      "epoch": 66.95997729208061,
      "grad_norm": 0.10607023537158966,
      "learning_rate": 3.3067556060175987e-05,
      "loss": 0.0003042459487915039,
      "step": 235900
    },
    {
      "epoch": 66.96281578200397,
      "grad_norm": 0.45717352628707886,
      "learning_rate": 3.306471757025263e-05,
      "loss": 0.00026686694473028184,
      "step": 235910
    },
    {
      "epoch": 66.96565427192733,
      "grad_norm": 0.029951220378279686,
      "learning_rate": 3.306187908032927e-05,
      "loss": 0.0001962263137102127,
      "step": 235920
    },
    {
      "epoch": 66.9684927618507,
      "grad_norm": 0.10945891588926315,
      "learning_rate": 3.3059040590405904e-05,
      "loss": 0.00030364785343408587,
      "step": 235930
    },
    {
      "epoch": 66.97133125177406,
      "grad_norm": 0.014496280811727047,
      "learning_rate": 3.3056202100482545e-05,
      "loss": 0.0002502577379345894,
      "step": 235940
    },
    {
      "epoch": 66.97416974169742,
      "grad_norm": 0.0050577437505126,
      "learning_rate": 3.305336361055919e-05,
      "loss": 0.005977569520473481,
      "step": 235950
    },
    {
      "epoch": 66.97700823162077,
      "grad_norm": 0.005154362879693508,
      "learning_rate": 3.305052512063582e-05,
      "loss": 0.0002639744430780411,
      "step": 235960
    },
    {
      "epoch": 66.97984672154413,
      "grad_norm": 0.027526360005140305,
      "learning_rate": 3.304768663071246e-05,
      "loss": 0.0016635648906230926,
      "step": 235970
    },
    {
      "epoch": 66.9826852114675,
      "grad_norm": 0.03489028289914131,
      "learning_rate": 3.3044848140789104e-05,
      "loss": 0.00024952106177806855,
      "step": 235980
    },
    {
      "epoch": 66.98552370139086,
      "grad_norm": 0.0067343455739319324,
      "learning_rate": 3.304200965086574e-05,
      "loss": 0.00011092647910118102,
      "step": 235990
    },
    {
      "epoch": 66.98836219131422,
      "grad_norm": 0.25556960701942444,
      "learning_rate": 3.303917116094238e-05,
      "loss": 0.00021039173007011413,
      "step": 236000
    },
    {
      "epoch": 66.98836219131422,
      "eval_accuracy": 0.9786990525847269,
      "eval_loss": 0.07723768800497055,
      "eval_runtime": 32.4536,
      "eval_samples_per_second": 484.6,
      "eval_steps_per_second": 7.58,
      "step": 236000
    },
    {
      "epoch": 66.99120068123759,
      "grad_norm": 0.006481558084487915,
      "learning_rate": 3.303633267101902e-05,
      "loss": 0.00045680683106184004,
      "step": 236010
    },
    {
      "epoch": 66.99403917116095,
      "grad_norm": 0.011692636646330357,
      "learning_rate": 3.3033494181095656e-05,
      "loss": 0.00029914453625679016,
      "step": 236020
    },
    {
      "epoch": 66.9968776610843,
      "grad_norm": 0.4017930030822754,
      "learning_rate": 3.30306556911723e-05,
      "loss": 0.0002335941419005394,
      "step": 236030
    },
    {
      "epoch": 66.99971615100766,
      "grad_norm": 0.0355517640709877,
      "learning_rate": 3.302781720124894e-05,
      "loss": 0.0002097446471452713,
      "step": 236040
    },
    {
      "epoch": 67.00255464093102,
      "grad_norm": 0.0060869003646075726,
      "learning_rate": 3.302497871132558e-05,
      "loss": 0.00012331365142017602,
      "step": 236050
    },
    {
      "epoch": 67.00539313085439,
      "grad_norm": 0.00845679547637701,
      "learning_rate": 3.3022140221402215e-05,
      "loss": 0.00012698248028755187,
      "step": 236060
    },
    {
      "epoch": 67.00823162077775,
      "grad_norm": 0.0244741328060627,
      "learning_rate": 3.301930173147885e-05,
      "loss": 0.0001370362937450409,
      "step": 236070
    },
    {
      "epoch": 67.01107011070111,
      "grad_norm": 0.003217400750145316,
      "learning_rate": 3.30164632415555e-05,
      "loss": 0.00016958098858594895,
      "step": 236080
    },
    {
      "epoch": 67.01390860062446,
      "grad_norm": 0.05049051344394684,
      "learning_rate": 3.301362475163213e-05,
      "loss": 9.31398943066597e-05,
      "step": 236090
    },
    {
      "epoch": 67.01674709054782,
      "grad_norm": 0.004134889226406813,
      "learning_rate": 3.3010786261708773e-05,
      "loss": 9.798165410757065e-05,
      "step": 236100
    },
    {
      "epoch": 67.01958558047119,
      "grad_norm": 0.09039821475744247,
      "learning_rate": 3.3007947771785415e-05,
      "loss": 9.245406836271286e-05,
      "step": 236110
    },
    {
      "epoch": 67.02242407039455,
      "grad_norm": 0.06599873304367065,
      "learning_rate": 3.300510928186205e-05,
      "loss": 8.301381021738052e-05,
      "step": 236120
    },
    {
      "epoch": 67.02526256031791,
      "grad_norm": 0.026155656203627586,
      "learning_rate": 3.300227079193869e-05,
      "loss": 0.00023772474378347397,
      "step": 236130
    },
    {
      "epoch": 67.02810105024128,
      "grad_norm": 0.06915438920259476,
      "learning_rate": 3.299943230201533e-05,
      "loss": 0.0001914042979478836,
      "step": 236140
    },
    {
      "epoch": 67.03093954016464,
      "grad_norm": 0.027031930163502693,
      "learning_rate": 3.299659381209197e-05,
      "loss": 0.00032555293291807174,
      "step": 236150
    },
    {
      "epoch": 67.03377803008799,
      "grad_norm": 0.014741223305463791,
      "learning_rate": 3.299375532216861e-05,
      "loss": 0.00104784294962883,
      "step": 236160
    },
    {
      "epoch": 67.03661652001135,
      "grad_norm": 0.0518384650349617,
      "learning_rate": 3.299091683224524e-05,
      "loss": 8.876565843820571e-05,
      "step": 236170
    },
    {
      "epoch": 67.03945500993471,
      "grad_norm": 0.012188301421701908,
      "learning_rate": 3.298807834232189e-05,
      "loss": 0.0017452221363782883,
      "step": 236180
    },
    {
      "epoch": 67.04229349985808,
      "grad_norm": 0.7081728577613831,
      "learning_rate": 3.2985239852398525e-05,
      "loss": 0.00017871428281068802,
      "step": 236190
    },
    {
      "epoch": 67.04513198978144,
      "grad_norm": 0.0620727464556694,
      "learning_rate": 3.298240136247516e-05,
      "loss": 0.00018891040235757828,
      "step": 236200
    },
    {
      "epoch": 67.0479704797048,
      "grad_norm": 0.0177723728120327,
      "learning_rate": 3.297956287255181e-05,
      "loss": 0.0001420043408870697,
      "step": 236210
    },
    {
      "epoch": 67.05080896962816,
      "grad_norm": 0.042667847126722336,
      "learning_rate": 3.297672438262844e-05,
      "loss": 0.00010299999266862869,
      "step": 236220
    },
    {
      "epoch": 67.05364745955151,
      "grad_norm": 0.0521695651113987,
      "learning_rate": 3.2973885892705084e-05,
      "loss": 0.00011220201849937439,
      "step": 236230
    },
    {
      "epoch": 67.05648594947488,
      "grad_norm": 0.016194134950637817,
      "learning_rate": 3.2971047402781726e-05,
      "loss": 7.477141916751861e-05,
      "step": 236240
    },
    {
      "epoch": 67.05932443939824,
      "grad_norm": 0.1353268027305603,
      "learning_rate": 3.296820891285836e-05,
      "loss": 8.58057290315628e-05,
      "step": 236250
    },
    {
      "epoch": 67.0621629293216,
      "grad_norm": 0.029034795239567757,
      "learning_rate": 3.2965370422935e-05,
      "loss": 0.00016581341624259948,
      "step": 236260
    },
    {
      "epoch": 67.06500141924496,
      "grad_norm": 0.09992945194244385,
      "learning_rate": 3.2962531933011636e-05,
      "loss": 0.0004752945154905319,
      "step": 236270
    },
    {
      "epoch": 67.06783990916833,
      "grad_norm": 1.022908329963684,
      "learning_rate": 3.295969344308828e-05,
      "loss": 0.00034801457077264785,
      "step": 236280
    },
    {
      "epoch": 67.07067839909168,
      "grad_norm": 0.0667208656668663,
      "learning_rate": 3.295685495316492e-05,
      "loss": 0.00013106036931276321,
      "step": 236290
    },
    {
      "epoch": 67.07351688901504,
      "grad_norm": 0.050853487104177475,
      "learning_rate": 3.2954016463241553e-05,
      "loss": 0.0003141326829791069,
      "step": 236300
    },
    {
      "epoch": 67.0763553789384,
      "grad_norm": 0.011382436379790306,
      "learning_rate": 3.2951177973318195e-05,
      "loss": 0.005402231961488724,
      "step": 236310
    },
    {
      "epoch": 67.07919386886176,
      "grad_norm": 0.0899774581193924,
      "learning_rate": 3.2948339483394836e-05,
      "loss": 0.0026313124224543573,
      "step": 236320
    },
    {
      "epoch": 67.08203235878513,
      "grad_norm": 0.010656670667231083,
      "learning_rate": 3.294550099347147e-05,
      "loss": 0.0004744578152894974,
      "step": 236330
    },
    {
      "epoch": 67.08487084870849,
      "grad_norm": 0.04786789417266846,
      "learning_rate": 3.294266250354812e-05,
      "loss": 0.00016459617763757705,
      "step": 236340
    },
    {
      "epoch": 67.08770933863185,
      "grad_norm": 0.023630335927009583,
      "learning_rate": 3.2939824013624754e-05,
      "loss": 0.002091119438409805,
      "step": 236350
    },
    {
      "epoch": 67.0905478285552,
      "grad_norm": 0.06522274017333984,
      "learning_rate": 3.293698552370139e-05,
      "loss": 0.0006758859381079674,
      "step": 236360
    },
    {
      "epoch": 67.09338631847857,
      "grad_norm": 0.04336344823241234,
      "learning_rate": 3.293414703377803e-05,
      "loss": 0.008488357812166215,
      "step": 236370
    },
    {
      "epoch": 67.09622480840193,
      "grad_norm": 0.06604571640491486,
      "learning_rate": 3.293130854385467e-05,
      "loss": 0.0002777904272079468,
      "step": 236380
    },
    {
      "epoch": 67.09906329832529,
      "grad_norm": 9.959754943847656,
      "learning_rate": 3.292847005393131e-05,
      "loss": 0.001907011680305004,
      "step": 236390
    },
    {
      "epoch": 67.10190178824865,
      "grad_norm": 0.0034141696523875,
      "learning_rate": 3.292563156400795e-05,
      "loss": 0.00022412482649087907,
      "step": 236400
    },
    {
      "epoch": 67.10474027817202,
      "grad_norm": 0.005250183399766684,
      "learning_rate": 3.292279307408459e-05,
      "loss": 0.0001677820459008217,
      "step": 236410
    },
    {
      "epoch": 67.10757876809538,
      "grad_norm": 0.004074434284120798,
      "learning_rate": 3.291995458416123e-05,
      "loss": 0.00010859258472919464,
      "step": 236420
    },
    {
      "epoch": 67.11041725801873,
      "grad_norm": 0.18865327537059784,
      "learning_rate": 3.2917116094237864e-05,
      "loss": 0.00025615859776735307,
      "step": 236430
    },
    {
      "epoch": 67.11325574794209,
      "grad_norm": 0.05772285908460617,
      "learning_rate": 3.2914277604314506e-05,
      "loss": 0.00013003647327423096,
      "step": 236440
    },
    {
      "epoch": 67.11609423786545,
      "grad_norm": 0.05750928446650505,
      "learning_rate": 3.291143911439115e-05,
      "loss": 0.0003446422517299652,
      "step": 236450
    },
    {
      "epoch": 67.11893272778882,
      "grad_norm": 0.001692999736405909,
      "learning_rate": 3.290860062446778e-05,
      "loss": 0.00010253842920064926,
      "step": 236460
    },
    {
      "epoch": 67.12177121771218,
      "grad_norm": 0.052037693560123444,
      "learning_rate": 3.290576213454443e-05,
      "loss": 0.00011726785451173782,
      "step": 236470
    },
    {
      "epoch": 67.12460970763554,
      "grad_norm": 0.11049552261829376,
      "learning_rate": 3.2902923644621064e-05,
      "loss": 7.203705608844757e-05,
      "step": 236480
    },
    {
      "epoch": 67.1274481975589,
      "grad_norm": 0.0294700525701046,
      "learning_rate": 3.29000851546977e-05,
      "loss": 0.00013865381479263305,
      "step": 236490
    },
    {
      "epoch": 67.13028668748225,
      "grad_norm": 0.006669621914625168,
      "learning_rate": 3.289724666477434e-05,
      "loss": 6.206240504980087e-05,
      "step": 236500
    },
    {
      "epoch": 67.13028668748225,
      "eval_accuracy": 0.9792713168436447,
      "eval_loss": 0.0756475180387497,
      "eval_runtime": 32.7558,
      "eval_samples_per_second": 480.129,
      "eval_steps_per_second": 7.51,
      "step": 236500
    },
    {
      "epoch": 67.13312517740562,
      "grad_norm": 0.01817850023508072,
      "learning_rate": 3.289440817485098e-05,
      "loss": 0.00012988485395908355,
      "step": 236510
    },
    {
      "epoch": 67.13596366732898,
      "grad_norm": 0.018373096361756325,
      "learning_rate": 3.289156968492762e-05,
      "loss": 9.049717336893082e-05,
      "step": 236520
    },
    {
      "epoch": 67.13880215725234,
      "grad_norm": 0.007237971760332584,
      "learning_rate": 3.288873119500426e-05,
      "loss": 9.212139993906022e-05,
      "step": 236530
    },
    {
      "epoch": 67.1416406471757,
      "grad_norm": 0.06076979637145996,
      "learning_rate": 3.28858927050809e-05,
      "loss": 0.00045496057718992234,
      "step": 236540
    },
    {
      "epoch": 67.14447913709907,
      "grad_norm": 0.01005479320883751,
      "learning_rate": 3.288305421515754e-05,
      "loss": 0.00016459915786981583,
      "step": 236550
    },
    {
      "epoch": 67.14731762702242,
      "grad_norm": 0.030428579077124596,
      "learning_rate": 3.2880215725234175e-05,
      "loss": 0.00015869475901126863,
      "step": 236560
    },
    {
      "epoch": 67.15015611694578,
      "grad_norm": 0.03179007023572922,
      "learning_rate": 3.2877377235310816e-05,
      "loss": 0.00010192859917879105,
      "step": 236570
    },
    {
      "epoch": 67.15299460686914,
      "grad_norm": 0.04424202814698219,
      "learning_rate": 3.287453874538746e-05,
      "loss": 7.434152066707612e-05,
      "step": 236580
    },
    {
      "epoch": 67.1558330967925,
      "grad_norm": 0.018560616299510002,
      "learning_rate": 3.287170025546409e-05,
      "loss": 9.98741015791893e-05,
      "step": 236590
    },
    {
      "epoch": 67.15867158671587,
      "grad_norm": 0.0035243912134319544,
      "learning_rate": 3.2868861765540734e-05,
      "loss": 4.1796080768108365e-05,
      "step": 236600
    },
    {
      "epoch": 67.16151007663923,
      "grad_norm": 0.004361484199762344,
      "learning_rate": 3.2866023275617375e-05,
      "loss": 7.771067321300506e-05,
      "step": 236610
    },
    {
      "epoch": 67.1643485665626,
      "grad_norm": 0.03471178188920021,
      "learning_rate": 3.286318478569401e-05,
      "loss": 5.748681724071503e-05,
      "step": 236620
    },
    {
      "epoch": 67.16718705648594,
      "grad_norm": 0.012607697397470474,
      "learning_rate": 3.286034629577065e-05,
      "loss": 5.701519548892975e-05,
      "step": 236630
    },
    {
      "epoch": 67.1700255464093,
      "grad_norm": 0.04695073142647743,
      "learning_rate": 3.285750780584729e-05,
      "loss": 0.00010181255638599395,
      "step": 236640
    },
    {
      "epoch": 67.17286403633267,
      "grad_norm": 0.022482389584183693,
      "learning_rate": 3.2854669315923934e-05,
      "loss": 0.00010940190404653549,
      "step": 236650
    },
    {
      "epoch": 67.17570252625603,
      "grad_norm": 0.011296030133962631,
      "learning_rate": 3.285183082600057e-05,
      "loss": 0.00041452422738075256,
      "step": 236660
    },
    {
      "epoch": 67.1785410161794,
      "grad_norm": 0.10751219093799591,
      "learning_rate": 3.284899233607721e-05,
      "loss": 0.00025536324828863145,
      "step": 236670
    },
    {
      "epoch": 67.18137950610276,
      "grad_norm": 0.026243533939123154,
      "learning_rate": 3.284615384615385e-05,
      "loss": 0.00016429182142019272,
      "step": 236680
    },
    {
      "epoch": 67.18421799602612,
      "grad_norm": 0.03632938116788864,
      "learning_rate": 3.2843315356230486e-05,
      "loss": 0.0002470681443810463,
      "step": 236690
    },
    {
      "epoch": 67.18705648594947,
      "grad_norm": 0.11133371293544769,
      "learning_rate": 3.284047686630713e-05,
      "loss": 0.00030075274407863616,
      "step": 236700
    },
    {
      "epoch": 67.18989497587283,
      "grad_norm": 0.005575345829129219,
      "learning_rate": 3.283763837638377e-05,
      "loss": 0.0004521509632468224,
      "step": 236710
    },
    {
      "epoch": 67.1927334657962,
      "grad_norm": 0.018629826605319977,
      "learning_rate": 3.283508373545274e-05,
      "loss": 0.0037490658462047575,
      "step": 236720
    },
    {
      "epoch": 67.19557195571956,
      "grad_norm": 0.7727771401405334,
      "learning_rate": 3.283224524552938e-05,
      "loss": 0.0008363824337720871,
      "step": 236730
    },
    {
      "epoch": 67.19841044564292,
      "grad_norm": 0.3189336061477661,
      "learning_rate": 3.282940675560602e-05,
      "loss": 0.004271965846419335,
      "step": 236740
    },
    {
      "epoch": 67.20124893556628,
      "grad_norm": 0.0867658331990242,
      "learning_rate": 3.282656826568265e-05,
      "loss": 0.005433711037039757,
      "step": 236750
    },
    {
      "epoch": 67.20408742548965,
      "grad_norm": 0.14835533499717712,
      "learning_rate": 3.28237297757593e-05,
      "loss": 0.0007791556417942047,
      "step": 236760
    },
    {
      "epoch": 67.206925915413,
      "grad_norm": 0.014016584493219852,
      "learning_rate": 3.2820891285835935e-05,
      "loss": 0.0014142265543341636,
      "step": 236770
    },
    {
      "epoch": 67.20976440533636,
      "grad_norm": 0.6574096083641052,
      "learning_rate": 3.2818052795912576e-05,
      "loss": 0.0007308794185519219,
      "step": 236780
    },
    {
      "epoch": 67.21260289525972,
      "grad_norm": 0.17764997482299805,
      "learning_rate": 3.281521430598922e-05,
      "loss": 0.00012448765337467194,
      "step": 236790
    },
    {
      "epoch": 67.21544138518308,
      "grad_norm": 1.0512490272521973,
      "learning_rate": 3.281237581606585e-05,
      "loss": 0.0016630355268716813,
      "step": 236800
    },
    {
      "epoch": 67.21827987510645,
      "grad_norm": 0.02039116434752941,
      "learning_rate": 3.2809537326142494e-05,
      "loss": 0.0005183152854442597,
      "step": 236810
    },
    {
      "epoch": 67.22111836502981,
      "grad_norm": 0.32698631286621094,
      "learning_rate": 3.2806698836219135e-05,
      "loss": 0.00025154203176498414,
      "step": 236820
    },
    {
      "epoch": 67.22395685495316,
      "grad_norm": 0.48315221071243286,
      "learning_rate": 3.280386034629577e-05,
      "loss": 0.0011653421446681024,
      "step": 236830
    },
    {
      "epoch": 67.22679534487652,
      "grad_norm": 0.033660952001810074,
      "learning_rate": 3.280102185637241e-05,
      "loss": 0.0005238119512796402,
      "step": 236840
    },
    {
      "epoch": 67.22963383479988,
      "grad_norm": 0.015927234664559364,
      "learning_rate": 3.279818336644905e-05,
      "loss": 0.0007678842172026634,
      "step": 236850
    },
    {
      "epoch": 67.23247232472325,
      "grad_norm": 17.38915252685547,
      "learning_rate": 3.2795344876525694e-05,
      "loss": 0.005421456322073937,
      "step": 236860
    },
    {
      "epoch": 67.23531081464661,
      "grad_norm": 0.11791999638080597,
      "learning_rate": 3.279250638660233e-05,
      "loss": 0.0041209693998098375,
      "step": 236870
    },
    {
      "epoch": 67.23814930456997,
      "grad_norm": 0.08793721348047256,
      "learning_rate": 3.278966789667896e-05,
      "loss": 0.00012165363878011704,
      "step": 236880
    },
    {
      "epoch": 67.24098779449334,
      "grad_norm": 0.020970268175005913,
      "learning_rate": 3.278682940675561e-05,
      "loss": 8.084066212177277e-05,
      "step": 236890
    },
    {
      "epoch": 67.24382628441668,
      "grad_norm": 0.05765959993004799,
      "learning_rate": 3.2783990916832246e-05,
      "loss": 0.00013734959065914154,
      "step": 236900
    },
    {
      "epoch": 67.24666477434005,
      "grad_norm": 0.00980284158140421,
      "learning_rate": 3.278115242690889e-05,
      "loss": 9.114164859056473e-05,
      "step": 236910
    },
    {
      "epoch": 67.24950326426341,
      "grad_norm": 0.308773398399353,
      "learning_rate": 3.277831393698553e-05,
      "loss": 0.00011258665472269058,
      "step": 236920
    },
    {
      "epoch": 67.25234175418677,
      "grad_norm": 0.07570528984069824,
      "learning_rate": 3.277547544706216e-05,
      "loss": 0.0001067792996764183,
      "step": 236930
    },
    {
      "epoch": 67.25518024411014,
      "grad_norm": 0.018651176244020462,
      "learning_rate": 3.2772636957138805e-05,
      "loss": 7.504243403673172e-05,
      "step": 236940
    },
    {
      "epoch": 67.2580187340335,
      "grad_norm": 0.0510127991437912,
      "learning_rate": 3.2769798467215446e-05,
      "loss": 0.00013084281235933303,
      "step": 236950
    },
    {
      "epoch": 67.26085722395686,
      "grad_norm": 0.011131802573800087,
      "learning_rate": 3.276695997729208e-05,
      "loss": 0.00013439562171697618,
      "step": 236960
    },
    {
      "epoch": 67.26369571388021,
      "grad_norm": 0.030319981276988983,
      "learning_rate": 3.276412148736872e-05,
      "loss": 8.776187896728516e-05,
      "step": 236970
    },
    {
      "epoch": 67.26653420380357,
      "grad_norm": 0.026257796213030815,
      "learning_rate": 3.2761282997445356e-05,
      "loss": 8.591786026954651e-05,
      "step": 236980
    },
    {
      "epoch": 67.26937269372694,
      "grad_norm": 0.050062086433172226,
      "learning_rate": 3.2758444507522005e-05,
      "loss": 0.00012008324265480041,
      "step": 236990
    },
    {
      "epoch": 67.2722111836503,
      "grad_norm": 0.04342683032155037,
      "learning_rate": 3.275560601759864e-05,
      "loss": 0.001168622449040413,
      "step": 237000
    },
    {
      "epoch": 67.2722111836503,
      "eval_accuracy": 0.9797164112672474,
      "eval_loss": 0.07643184810876846,
      "eval_runtime": 32.5734,
      "eval_samples_per_second": 482.817,
      "eval_steps_per_second": 7.552,
      "step": 237000
    },
    {
      "epoch": 67.27504967357366,
      "grad_norm": 0.092112235724926,
      "learning_rate": 3.2752767527675274e-05,
      "loss": 0.00023942440748214722,
      "step": 237010
    },
    {
      "epoch": 67.27788816349702,
      "grad_norm": 0.012762495316565037,
      "learning_rate": 3.274992903775192e-05,
      "loss": 0.00011216085404157639,
      "step": 237020
    },
    {
      "epoch": 67.28072665342037,
      "grad_norm": 0.04242315888404846,
      "learning_rate": 3.2747090547828557e-05,
      "loss": 0.0006746476516127587,
      "step": 237030
    },
    {
      "epoch": 67.28356514334374,
      "grad_norm": 0.01878080517053604,
      "learning_rate": 3.27442520579052e-05,
      "loss": 0.00019951723515987396,
      "step": 237040
    },
    {
      "epoch": 67.2864036332671,
      "grad_norm": 2.0583741664886475,
      "learning_rate": 3.274141356798184e-05,
      "loss": 0.0003763450309634209,
      "step": 237050
    },
    {
      "epoch": 67.28924212319046,
      "grad_norm": 0.07831819355487823,
      "learning_rate": 3.2738575078058474e-05,
      "loss": 0.0023138301447033884,
      "step": 237060
    },
    {
      "epoch": 67.29208061311382,
      "grad_norm": 0.2165016084909439,
      "learning_rate": 3.2735736588135115e-05,
      "loss": 0.00015714094042778014,
      "step": 237070
    },
    {
      "epoch": 67.29491910303719,
      "grad_norm": 0.04398176446557045,
      "learning_rate": 3.273289809821175e-05,
      "loss": 0.0009617276489734649,
      "step": 237080
    },
    {
      "epoch": 67.29775759296055,
      "grad_norm": 0.0849243700504303,
      "learning_rate": 3.273005960828839e-05,
      "loss": 0.00013422854244709014,
      "step": 237090
    },
    {
      "epoch": 67.3005960828839,
      "grad_norm": 0.03056160733103752,
      "learning_rate": 3.272722111836503e-05,
      "loss": 0.00042616799473762514,
      "step": 237100
    },
    {
      "epoch": 67.30343457280726,
      "grad_norm": 0.008664845488965511,
      "learning_rate": 3.272438262844167e-05,
      "loss": 0.00017385035753250122,
      "step": 237110
    },
    {
      "epoch": 67.30627306273063,
      "grad_norm": 0.29482415318489075,
      "learning_rate": 3.272154413851831e-05,
      "loss": 0.00025800149887800217,
      "step": 237120
    },
    {
      "epoch": 67.30911155265399,
      "grad_norm": 0.10153143107891083,
      "learning_rate": 3.271870564859495e-05,
      "loss": 0.00010078474879264832,
      "step": 237130
    },
    {
      "epoch": 67.31195004257735,
      "grad_norm": 0.021328706294298172,
      "learning_rate": 3.2715867158671585e-05,
      "loss": 9.536109864711761e-05,
      "step": 237140
    },
    {
      "epoch": 67.31478853250071,
      "grad_norm": 0.03882073611021042,
      "learning_rate": 3.271302866874823e-05,
      "loss": 0.00020025428384542465,
      "step": 237150
    },
    {
      "epoch": 67.31762702242408,
      "grad_norm": 0.049883902072906494,
      "learning_rate": 3.271019017882487e-05,
      "loss": 8.650124073028564e-05,
      "step": 237160
    },
    {
      "epoch": 67.32046551234743,
      "grad_norm": 0.022513367235660553,
      "learning_rate": 3.27073516889015e-05,
      "loss": 7.63261690735817e-05,
      "step": 237170
    },
    {
      "epoch": 67.32330400227079,
      "grad_norm": 0.031981728971004486,
      "learning_rate": 3.270451319897814e-05,
      "loss": 9.247064590454102e-05,
      "step": 237180
    },
    {
      "epoch": 67.32614249219415,
      "grad_norm": 0.1449345350265503,
      "learning_rate": 3.2701674709054785e-05,
      "loss": 0.0003418838605284691,
      "step": 237190
    },
    {
      "epoch": 67.32898098211751,
      "grad_norm": 0.02989945374429226,
      "learning_rate": 3.2698836219131426e-05,
      "loss": 0.00032199248671531676,
      "step": 237200
    },
    {
      "epoch": 67.33181947204088,
      "grad_norm": 0.02557341195642948,
      "learning_rate": 3.269599772920806e-05,
      "loss": 0.0047337919473648075,
      "step": 237210
    },
    {
      "epoch": 67.33465796196424,
      "grad_norm": 0.9418328404426575,
      "learning_rate": 3.26931592392847e-05,
      "loss": 0.0008564088493585587,
      "step": 237220
    },
    {
      "epoch": 67.3374964518876,
      "grad_norm": 0.0908663421869278,
      "learning_rate": 3.2690320749361343e-05,
      "loss": 0.003247436136007309,
      "step": 237230
    },
    {
      "epoch": 67.34033494181095,
      "grad_norm": 4.545805931091309,
      "learning_rate": 3.268748225943798e-05,
      "loss": 0.0019252188503742218,
      "step": 237240
    },
    {
      "epoch": 67.34317343173431,
      "grad_norm": 0.6400017142295837,
      "learning_rate": 3.268464376951462e-05,
      "loss": 0.0005923531949520111,
      "step": 237250
    },
    {
      "epoch": 67.34601192165768,
      "grad_norm": 0.15722693502902985,
      "learning_rate": 3.268180527959126e-05,
      "loss": 0.0003538699820637703,
      "step": 237260
    },
    {
      "epoch": 67.34885041158104,
      "grad_norm": 0.0171974990516901,
      "learning_rate": 3.2678966789667895e-05,
      "loss": 0.0005620758980512619,
      "step": 237270
    },
    {
      "epoch": 67.3516889015044,
      "grad_norm": 0.10647895187139511,
      "learning_rate": 3.267612829974454e-05,
      "loss": 0.0011112306267023087,
      "step": 237280
    },
    {
      "epoch": 67.35452739142777,
      "grad_norm": 0.028096171095967293,
      "learning_rate": 3.267328980982118e-05,
      "loss": 0.00046977773308753967,
      "step": 237290
    },
    {
      "epoch": 67.35736588135111,
      "grad_norm": 0.025773795321583748,
      "learning_rate": 3.267045131989781e-05,
      "loss": 0.004715617001056671,
      "step": 237300
    },
    {
      "epoch": 67.36020437127448,
      "grad_norm": 1.836323857307434,
      "learning_rate": 3.2667612829974454e-05,
      "loss": 0.0007401857525110245,
      "step": 237310
    },
    {
      "epoch": 67.36304286119784,
      "grad_norm": 0.15430571138858795,
      "learning_rate": 3.2664774340051095e-05,
      "loss": 0.0005004128441214561,
      "step": 237320
    },
    {
      "epoch": 67.3658813511212,
      "grad_norm": 0.03876098245382309,
      "learning_rate": 3.266193585012774e-05,
      "loss": 0.0006790626794099808,
      "step": 237330
    },
    {
      "epoch": 67.36871984104457,
      "grad_norm": 0.050470758229494095,
      "learning_rate": 3.265909736020437e-05,
      "loss": 0.00024403799325227736,
      "step": 237340
    },
    {
      "epoch": 67.37155833096793,
      "grad_norm": 0.012902499176561832,
      "learning_rate": 3.265625887028101e-05,
      "loss": 0.0016576904803514481,
      "step": 237350
    },
    {
      "epoch": 67.37439682089129,
      "grad_norm": 0.008782655000686646,
      "learning_rate": 3.2653420380357654e-05,
      "loss": 0.0021958498284220694,
      "step": 237360
    },
    {
      "epoch": 67.37723531081464,
      "grad_norm": 0.03203536942601204,
      "learning_rate": 3.265058189043429e-05,
      "loss": 0.0003221284598112106,
      "step": 237370
    },
    {
      "epoch": 67.380073800738,
      "grad_norm": 8.82541561126709,
      "learning_rate": 3.264774340051093e-05,
      "loss": 0.0018380090594291687,
      "step": 237380
    },
    {
      "epoch": 67.38291229066137,
      "grad_norm": 0.11666781455278397,
      "learning_rate": 3.264490491058757e-05,
      "loss": 0.00030213315039873123,
      "step": 237390
    },
    {
      "epoch": 67.38575078058473,
      "grad_norm": 0.07662409543991089,
      "learning_rate": 3.2642066420664206e-05,
      "loss": 0.0011859867721796037,
      "step": 237400
    },
    {
      "epoch": 67.38858927050809,
      "grad_norm": 0.23841720819473267,
      "learning_rate": 3.263922793074085e-05,
      "loss": 0.0003433629870414734,
      "step": 237410
    },
    {
      "epoch": 67.39142776043145,
      "grad_norm": 0.026715701445937157,
      "learning_rate": 3.263638944081749e-05,
      "loss": 0.0002550181001424789,
      "step": 237420
    },
    {
      "epoch": 67.39426625035482,
      "grad_norm": 0.18612638115882874,
      "learning_rate": 3.2633550950894123e-05,
      "loss": 0.00021160747855901717,
      "step": 237430
    },
    {
      "epoch": 67.39710474027817,
      "grad_norm": 0.08201069384813309,
      "learning_rate": 3.2630712460970765e-05,
      "loss": 0.002547897957265377,
      "step": 237440
    },
    {
      "epoch": 67.39994323020153,
      "grad_norm": 0.14940835535526276,
      "learning_rate": 3.2627873971047406e-05,
      "loss": 0.00025792531669139863,
      "step": 237450
    },
    {
      "epoch": 67.40278172012489,
      "grad_norm": 0.021731631830334663,
      "learning_rate": 3.262503548112405e-05,
      "loss": 0.00010238997638225555,
      "step": 237460
    },
    {
      "epoch": 67.40562021004826,
      "grad_norm": 0.00626297015696764,
      "learning_rate": 3.262219699120068e-05,
      "loss": 0.00019033048301935195,
      "step": 237470
    },
    {
      "epoch": 67.40845869997162,
      "grad_norm": 0.036828748881816864,
      "learning_rate": 3.261935850127732e-05,
      "loss": 0.0001714169979095459,
      "step": 237480
    },
    {
      "epoch": 67.41129718989498,
      "grad_norm": 0.19025029242038727,
      "learning_rate": 3.2616520011353965e-05,
      "loss": 0.0002760464325547218,
      "step": 237490
    },
    {
      "epoch": 67.41413567981834,
      "grad_norm": 0.013818745501339436,
      "learning_rate": 3.26136815214306e-05,
      "loss": 0.0005547253414988518,
      "step": 237500
    },
    {
      "epoch": 67.41413567981834,
      "eval_accuracy": 0.9779360335728365,
      "eval_loss": 0.08018671721220016,
      "eval_runtime": 33.0481,
      "eval_samples_per_second": 475.882,
      "eval_steps_per_second": 7.444,
      "step": 237500
    },
    {
      "epoch": 67.41697416974169,
      "grad_norm": 0.05217554420232773,
      "learning_rate": 3.261084303150724e-05,
      "loss": 0.00018909256905317307,
      "step": 237510
    },
    {
      "epoch": 67.41981265966506,
      "grad_norm": 0.36982622742652893,
      "learning_rate": 3.260800454158388e-05,
      "loss": 0.00015080366283655168,
      "step": 237520
    },
    {
      "epoch": 67.42265114958842,
      "grad_norm": 0.012583848088979721,
      "learning_rate": 3.260516605166052e-05,
      "loss": 0.00018460489809513092,
      "step": 237530
    },
    {
      "epoch": 67.42548963951178,
      "grad_norm": 0.003976361360400915,
      "learning_rate": 3.260232756173716e-05,
      "loss": 0.0025164172053337096,
      "step": 237540
    },
    {
      "epoch": 67.42832812943514,
      "grad_norm": 0.021744264289736748,
      "learning_rate": 3.25994890718138e-05,
      "loss": 0.00041102934628725054,
      "step": 237550
    },
    {
      "epoch": 67.4311666193585,
      "grad_norm": 0.06705716252326965,
      "learning_rate": 3.2596650581890434e-05,
      "loss": 0.00017867665737867356,
      "step": 237560
    },
    {
      "epoch": 67.43400510928186,
      "grad_norm": 0.010521214455366135,
      "learning_rate": 3.2593812091967076e-05,
      "loss": 0.00027242992073297503,
      "step": 237570
    },
    {
      "epoch": 67.43684359920522,
      "grad_norm": 0.019489815458655357,
      "learning_rate": 3.259097360204371e-05,
      "loss": 0.00027269795536994935,
      "step": 237580
    },
    {
      "epoch": 67.43968208912858,
      "grad_norm": 0.01643611676990986,
      "learning_rate": 3.258813511212035e-05,
      "loss": 0.00013379808515310288,
      "step": 237590
    },
    {
      "epoch": 67.44252057905194,
      "grad_norm": 0.08060578256845474,
      "learning_rate": 3.258529662219699e-05,
      "loss": 0.00011404510587453842,
      "step": 237600
    },
    {
      "epoch": 67.4453590689753,
      "grad_norm": 0.011957686394453049,
      "learning_rate": 3.258245813227363e-05,
      "loss": 0.00023435931652784347,
      "step": 237610
    },
    {
      "epoch": 67.44819755889867,
      "grad_norm": 3.2465033531188965,
      "learning_rate": 3.2579619642350276e-05,
      "loss": 0.0003473928198218346,
      "step": 237620
    },
    {
      "epoch": 67.45103604882203,
      "grad_norm": 0.042978618294000626,
      "learning_rate": 3.257678115242691e-05,
      "loss": 0.0008670385926961899,
      "step": 237630
    },
    {
      "epoch": 67.45387453874538,
      "grad_norm": 0.7776144742965698,
      "learning_rate": 3.2573942662503545e-05,
      "loss": 0.000506916269659996,
      "step": 237640
    },
    {
      "epoch": 67.45671302866874,
      "grad_norm": 3.677915334701538,
      "learning_rate": 3.257110417258019e-05,
      "loss": 0.0005712827667593956,
      "step": 237650
    },
    {
      "epoch": 67.45955151859211,
      "grad_norm": 0.02805529162287712,
      "learning_rate": 3.256826568265683e-05,
      "loss": 8.428264409303665e-05,
      "step": 237660
    },
    {
      "epoch": 67.46239000851547,
      "grad_norm": 0.021795719861984253,
      "learning_rate": 3.256542719273347e-05,
      "loss": 0.00023926720023155213,
      "step": 237670
    },
    {
      "epoch": 67.46522849843883,
      "grad_norm": 1.1590192317962646,
      "learning_rate": 3.2562588702810104e-05,
      "loss": 0.0003183698281645775,
      "step": 237680
    },
    {
      "epoch": 67.4680669883622,
      "grad_norm": 0.06659658253192902,
      "learning_rate": 3.2559750212886745e-05,
      "loss": 0.0001026008278131485,
      "step": 237690
    },
    {
      "epoch": 67.47090547828556,
      "grad_norm": 0.23256492614746094,
      "learning_rate": 3.2556911722963386e-05,
      "loss": 0.00030873361974954606,
      "step": 237700
    },
    {
      "epoch": 67.47374396820891,
      "grad_norm": 13.114998817443848,
      "learning_rate": 3.255407323304002e-05,
      "loss": 0.003718564659357071,
      "step": 237710
    },
    {
      "epoch": 67.47658245813227,
      "grad_norm": 0.1175011396408081,
      "learning_rate": 3.255123474311666e-05,
      "loss": 0.0004353182390332222,
      "step": 237720
    },
    {
      "epoch": 67.47942094805563,
      "grad_norm": 0.02598547376692295,
      "learning_rate": 3.2548396253193304e-05,
      "loss": 0.00028145574033260344,
      "step": 237730
    },
    {
      "epoch": 67.482259437979,
      "grad_norm": 0.08206906169652939,
      "learning_rate": 3.254555776326994e-05,
      "loss": 0.00013472083956003189,
      "step": 237740
    },
    {
      "epoch": 67.48509792790236,
      "grad_norm": 5.496104717254639,
      "learning_rate": 3.2542719273346586e-05,
      "loss": 0.0006842566654086113,
      "step": 237750
    },
    {
      "epoch": 67.48793641782572,
      "grad_norm": 0.03762378543615341,
      "learning_rate": 3.253988078342322e-05,
      "loss": 0.00013122893869876862,
      "step": 237760
    },
    {
      "epoch": 67.49077490774907,
      "grad_norm": 0.07827834784984589,
      "learning_rate": 3.2537042293499856e-05,
      "loss": 0.00046607572585344315,
      "step": 237770
    },
    {
      "epoch": 67.49361339767243,
      "grad_norm": 0.12565934658050537,
      "learning_rate": 3.25342038035765e-05,
      "loss": 0.00012675393372774124,
      "step": 237780
    },
    {
      "epoch": 67.4964518875958,
      "grad_norm": 0.42539021372795105,
      "learning_rate": 3.253136531365314e-05,
      "loss": 0.0003472644835710526,
      "step": 237790
    },
    {
      "epoch": 67.49929037751916,
      "grad_norm": 5.199791431427002,
      "learning_rate": 3.252852682372978e-05,
      "loss": 0.002033526450395584,
      "step": 237800
    },
    {
      "epoch": 67.50212886744252,
      "grad_norm": 9.261383056640625,
      "learning_rate": 3.2525688333806414e-05,
      "loss": 0.0019907845184206963,
      "step": 237810
    },
    {
      "epoch": 67.50496735736589,
      "grad_norm": 0.01654800772666931,
      "learning_rate": 3.2522849843883056e-05,
      "loss": 0.0001631900668144226,
      "step": 237820
    },
    {
      "epoch": 67.50780584728925,
      "grad_norm": 0.021316828206181526,
      "learning_rate": 3.25200113539597e-05,
      "loss": 0.00020019393414258957,
      "step": 237830
    },
    {
      "epoch": 67.5106443372126,
      "grad_norm": 0.03504329174757004,
      "learning_rate": 3.251717286403633e-05,
      "loss": 0.00022271033376455306,
      "step": 237840
    },
    {
      "epoch": 67.51348282713596,
      "grad_norm": 4.582546710968018,
      "learning_rate": 3.251433437411297e-05,
      "loss": 0.0007292604073882103,
      "step": 237850
    },
    {
      "epoch": 67.51632131705932,
      "grad_norm": 0.04747350886464119,
      "learning_rate": 3.2511495884189614e-05,
      "loss": 0.0002691950649023056,
      "step": 237860
    },
    {
      "epoch": 67.51915980698269,
      "grad_norm": 0.013321001082658768,
      "learning_rate": 3.250865739426625e-05,
      "loss": 0.0008635884150862694,
      "step": 237870
    },
    {
      "epoch": 67.52199829690605,
      "grad_norm": 0.016327951103448868,
      "learning_rate": 3.250581890434289e-05,
      "loss": 0.0001387348398566246,
      "step": 237880
    },
    {
      "epoch": 67.52483678682941,
      "grad_norm": 0.007938256487250328,
      "learning_rate": 3.250298041441953e-05,
      "loss": 0.0007561881095170975,
      "step": 237890
    },
    {
      "epoch": 67.52767527675277,
      "grad_norm": 0.34171441197395325,
      "learning_rate": 3.2500141924496166e-05,
      "loss": 0.0009125951677560806,
      "step": 237900
    },
    {
      "epoch": 67.53051376667612,
      "grad_norm": 0.06743182986974716,
      "learning_rate": 3.249730343457281e-05,
      "loss": 0.007697883248329163,
      "step": 237910
    },
    {
      "epoch": 67.53335225659949,
      "grad_norm": 0.018767064437270164,
      "learning_rate": 3.249446494464945e-05,
      "loss": 0.0007141321897506714,
      "step": 237920
    },
    {
      "epoch": 67.53619074652285,
      "grad_norm": 0.11455602198839188,
      "learning_rate": 3.249162645472609e-05,
      "loss": 0.00010518655180931091,
      "step": 237930
    },
    {
      "epoch": 67.53902923644621,
      "grad_norm": 0.04021434113383293,
      "learning_rate": 3.2488787964802725e-05,
      "loss": 0.0001988012343645096,
      "step": 237940
    },
    {
      "epoch": 67.54186772636957,
      "grad_norm": 0.02181648276746273,
      "learning_rate": 3.2485949474879367e-05,
      "loss": 0.00028188060969114304,
      "step": 237950
    },
    {
      "epoch": 67.54470621629294,
      "grad_norm": 0.038336578756570816,
      "learning_rate": 3.248311098495601e-05,
      "loss": 0.00033471304923295977,
      "step": 237960
    },
    {
      "epoch": 67.5475447062163,
      "grad_norm": 0.06564854085445404,
      "learning_rate": 3.248027249503264e-05,
      "loss": 0.00010924879461526871,
      "step": 237970
    },
    {
      "epoch": 67.55038319613965,
      "grad_norm": 6.99076509475708,
      "learning_rate": 3.2477434005109284e-05,
      "loss": 0.0008259061723947525,
      "step": 237980
    },
    {
      "epoch": 67.55322168606301,
      "grad_norm": 0.17916826903820038,
      "learning_rate": 3.2474595515185925e-05,
      "loss": 7.845032960176467e-05,
      "step": 237990
    },
    {
      "epoch": 67.55606017598637,
      "grad_norm": 0.2866898775100708,
      "learning_rate": 3.247175702526256e-05,
      "loss": 0.00046958867460489273,
      "step": 238000
    },
    {
      "epoch": 67.55606017598637,
      "eval_accuracy": 0.9800343358555351,
      "eval_loss": 0.07462452352046967,
      "eval_runtime": 32.847,
      "eval_samples_per_second": 478.796,
      "eval_steps_per_second": 7.489,
      "step": 238000
    },
    {
      "epoch": 67.55889866590974,
      "grad_norm": 0.028617478907108307,
      "learning_rate": 3.24689185353392e-05,
      "loss": 0.00017633382230997086,
      "step": 238010
    },
    {
      "epoch": 67.5617371558331,
      "grad_norm": 0.006552290637046099,
      "learning_rate": 3.246608004541584e-05,
      "loss": 0.0003404369577765465,
      "step": 238020
    },
    {
      "epoch": 67.56457564575646,
      "grad_norm": 3.0830724239349365,
      "learning_rate": 3.246324155549248e-05,
      "loss": 0.00047385673969984053,
      "step": 238030
    },
    {
      "epoch": 67.56741413567981,
      "grad_norm": 0.04404829069972038,
      "learning_rate": 3.246040306556912e-05,
      "loss": 0.0007031964138150215,
      "step": 238040
    },
    {
      "epoch": 67.57025262560317,
      "grad_norm": 0.5890796780586243,
      "learning_rate": 3.245756457564576e-05,
      "loss": 0.00034064315259456633,
      "step": 238050
    },
    {
      "epoch": 67.57309111552654,
      "grad_norm": 1.2230132818222046,
      "learning_rate": 3.2454726085722395e-05,
      "loss": 0.00346125066280365,
      "step": 238060
    },
    {
      "epoch": 67.5759296054499,
      "grad_norm": 0.010996498167514801,
      "learning_rate": 3.2451887595799036e-05,
      "loss": 0.00019328780472278596,
      "step": 238070
    },
    {
      "epoch": 67.57876809537326,
      "grad_norm": 0.2707473933696747,
      "learning_rate": 3.244904910587567e-05,
      "loss": 0.0001236187294125557,
      "step": 238080
    },
    {
      "epoch": 67.58160658529663,
      "grad_norm": 0.03466835618019104,
      "learning_rate": 3.244621061595232e-05,
      "loss": 0.0014740046113729476,
      "step": 238090
    },
    {
      "epoch": 67.58444507521999,
      "grad_norm": 8.493850708007812,
      "learning_rate": 3.244337212602895e-05,
      "loss": 0.0033713944256305696,
      "step": 238100
    },
    {
      "epoch": 67.58728356514334,
      "grad_norm": 0.0152988750487566,
      "learning_rate": 3.244053363610559e-05,
      "loss": 0.00224619023501873,
      "step": 238110
    },
    {
      "epoch": 67.5901220550667,
      "grad_norm": 0.15839128196239471,
      "learning_rate": 3.2437695146182236e-05,
      "loss": 0.0003886362537741661,
      "step": 238120
    },
    {
      "epoch": 67.59296054499006,
      "grad_norm": 5.970061302185059,
      "learning_rate": 3.243485665625887e-05,
      "loss": 0.0015026973560452461,
      "step": 238130
    },
    {
      "epoch": 67.59579903491343,
      "grad_norm": 0.044396527111530304,
      "learning_rate": 3.243201816633551e-05,
      "loss": 0.010679760575294494,
      "step": 238140
    },
    {
      "epoch": 67.59863752483679,
      "grad_norm": 0.16105493903160095,
      "learning_rate": 3.242917967641215e-05,
      "loss": 0.00022237971425056457,
      "step": 238150
    },
    {
      "epoch": 67.60147601476015,
      "grad_norm": 0.35511645674705505,
      "learning_rate": 3.242634118648879e-05,
      "loss": 0.00040302090346813204,
      "step": 238160
    },
    {
      "epoch": 67.60431450468351,
      "grad_norm": 0.021419376134872437,
      "learning_rate": 3.242350269656543e-05,
      "loss": 0.00015153884887695313,
      "step": 238170
    },
    {
      "epoch": 67.60715299460686,
      "grad_norm": 0.010115640237927437,
      "learning_rate": 3.242066420664207e-05,
      "loss": 0.0003530070185661316,
      "step": 238180
    },
    {
      "epoch": 67.60999148453023,
      "grad_norm": 0.7202189564704895,
      "learning_rate": 3.2417825716718705e-05,
      "loss": 0.0002615800127387047,
      "step": 238190
    },
    {
      "epoch": 67.61282997445359,
      "grad_norm": 0.181496262550354,
      "learning_rate": 3.241498722679535e-05,
      "loss": 0.0006333751603960991,
      "step": 238200
    },
    {
      "epoch": 67.61566846437695,
      "grad_norm": 0.028508830815553665,
      "learning_rate": 3.241214873687198e-05,
      "loss": 0.0011337369680404664,
      "step": 238210
    },
    {
      "epoch": 67.61850695430032,
      "grad_norm": 0.2563294470310211,
      "learning_rate": 3.240931024694863e-05,
      "loss": 0.0006260344758629799,
      "step": 238220
    },
    {
      "epoch": 67.62134544422368,
      "grad_norm": 0.005411229562014341,
      "learning_rate": 3.2406471757025264e-05,
      "loss": 0.00011811982840299606,
      "step": 238230
    },
    {
      "epoch": 67.62418393414703,
      "grad_norm": 0.026394879445433617,
      "learning_rate": 3.24036332671019e-05,
      "loss": 0.0001025468111038208,
      "step": 238240
    },
    {
      "epoch": 67.62702242407039,
      "grad_norm": 0.04029536247253418,
      "learning_rate": 3.240079477717855e-05,
      "loss": 0.0005399642512202262,
      "step": 238250
    },
    {
      "epoch": 67.62986091399375,
      "grad_norm": 0.07684516161680222,
      "learning_rate": 3.239795628725518e-05,
      "loss": 0.0011085314676165582,
      "step": 238260
    },
    {
      "epoch": 67.63269940391712,
      "grad_norm": 1.394189715385437,
      "learning_rate": 3.239511779733182e-05,
      "loss": 0.00030375011265277864,
      "step": 238270
    },
    {
      "epoch": 67.63553789384048,
      "grad_norm": 1.8145999908447266,
      "learning_rate": 3.2392279307408464e-05,
      "loss": 0.0004547718912363052,
      "step": 238280
    },
    {
      "epoch": 67.63837638376384,
      "grad_norm": 1.5817158222198486,
      "learning_rate": 3.23894408174851e-05,
      "loss": 0.00028676725924015045,
      "step": 238290
    },
    {
      "epoch": 67.6412148736872,
      "grad_norm": 0.2051060050725937,
      "learning_rate": 3.238660232756174e-05,
      "loss": 0.00020140688866376877,
      "step": 238300
    },
    {
      "epoch": 67.64405336361055,
      "grad_norm": 0.018723484128713608,
      "learning_rate": 3.2383763837638375e-05,
      "loss": 0.006840449571609497,
      "step": 238310
    },
    {
      "epoch": 67.64689185353392,
      "grad_norm": 0.039222218096256256,
      "learning_rate": 3.2380925347715016e-05,
      "loss": 0.0017077362164855004,
      "step": 238320
    },
    {
      "epoch": 67.64973034345728,
      "grad_norm": 0.08973008394241333,
      "learning_rate": 3.237808685779166e-05,
      "loss": 0.00044512040913105013,
      "step": 238330
    },
    {
      "epoch": 67.65256883338064,
      "grad_norm": 0.034558989107608795,
      "learning_rate": 3.237524836786829e-05,
      "loss": 0.00023877546191215516,
      "step": 238340
    },
    {
      "epoch": 67.655407323304,
      "grad_norm": 0.05780749395489693,
      "learning_rate": 3.237240987794494e-05,
      "loss": 0.014533048868179322,
      "step": 238350
    },
    {
      "epoch": 67.65824581322737,
      "grad_norm": 0.21600428223609924,
      "learning_rate": 3.2369571388021575e-05,
      "loss": 0.002072193473577499,
      "step": 238360
    },
    {
      "epoch": 67.66108430315073,
      "grad_norm": 0.6035264134407043,
      "learning_rate": 3.236673289809821e-05,
      "loss": 0.005184681713581085,
      "step": 238370
    },
    {
      "epoch": 67.66392279307408,
      "grad_norm": 6.263185501098633,
      "learning_rate": 3.236389440817486e-05,
      "loss": 0.0019714921712875367,
      "step": 238380
    },
    {
      "epoch": 67.66676128299744,
      "grad_norm": 0.07276687026023865,
      "learning_rate": 3.236105591825149e-05,
      "loss": 0.006644345819950104,
      "step": 238390
    },
    {
      "epoch": 67.6695997729208,
      "grad_norm": 0.9308815598487854,
      "learning_rate": 3.2358217428328134e-05,
      "loss": 0.002763892337679863,
      "step": 238400
    },
    {
      "epoch": 67.67243826284417,
      "grad_norm": 0.10474570095539093,
      "learning_rate": 3.235537893840477e-05,
      "loss": 0.006199084594845772,
      "step": 238410
    },
    {
      "epoch": 67.67527675276753,
      "grad_norm": 0.006782242562621832,
      "learning_rate": 3.235254044848141e-05,
      "loss": 0.0019528087228536605,
      "step": 238420
    },
    {
      "epoch": 67.6781152426909,
      "grad_norm": 8.153241157531738,
      "learning_rate": 3.234970195855805e-05,
      "loss": 0.0029659675434231757,
      "step": 238430
    },
    {
      "epoch": 67.68095373261426,
      "grad_norm": 0.15283608436584473,
      "learning_rate": 3.2346863468634685e-05,
      "loss": 0.006448875367641449,
      "step": 238440
    },
    {
      "epoch": 67.6837922225376,
      "grad_norm": 0.07770597189664841,
      "learning_rate": 3.234402497871133e-05,
      "loss": 0.0019217194989323617,
      "step": 238450
    },
    {
      "epoch": 67.68663071246097,
      "grad_norm": 0.01755458675324917,
      "learning_rate": 3.234118648878797e-05,
      "loss": 0.0003551799803972244,
      "step": 238460
    },
    {
      "epoch": 67.68946920238433,
      "grad_norm": 2.079482078552246,
      "learning_rate": 3.23383479988646e-05,
      "loss": 0.0006287086755037308,
      "step": 238470
    },
    {
      "epoch": 67.6923076923077,
      "grad_norm": 0.03817075118422508,
      "learning_rate": 3.2335509508941244e-05,
      "loss": 0.0002635115757584572,
      "step": 238480
    },
    {
      "epoch": 67.69514618223106,
      "grad_norm": 2.0910568237304688,
      "learning_rate": 3.2332671019017886e-05,
      "loss": 0.0007419925183057785,
      "step": 238490
    },
    {
      "epoch": 67.69798467215442,
      "grad_norm": 0.012498090043663979,
      "learning_rate": 3.232983252909452e-05,
      "loss": 0.0005417261272668839,
      "step": 238500
    },
    {
      "epoch": 67.69798467215442,
      "eval_accuracy": 0.9769822598079736,
      "eval_loss": 0.09116441011428833,
      "eval_runtime": 32.7187,
      "eval_samples_per_second": 480.673,
      "eval_steps_per_second": 7.519,
      "step": 238500
    },
    {
      "epoch": 67.70082316207777,
      "grad_norm": 0.04386280104517937,
      "learning_rate": 3.232699403917116e-05,
      "loss": 0.0015194511041045188,
      "step": 238510
    },
    {
      "epoch": 67.70366165200113,
      "grad_norm": 0.419534295797348,
      "learning_rate": 3.23241555492478e-05,
      "loss": 0.00018956176936626434,
      "step": 238520
    },
    {
      "epoch": 67.7065001419245,
      "grad_norm": 0.004012344870716333,
      "learning_rate": 3.232131705932444e-05,
      "loss": 0.00010520517826080322,
      "step": 238530
    },
    {
      "epoch": 67.70933863184786,
      "grad_norm": 0.01730985939502716,
      "learning_rate": 3.231847856940108e-05,
      "loss": 0.0001427123323082924,
      "step": 238540
    },
    {
      "epoch": 67.71217712177122,
      "grad_norm": 0.11343719065189362,
      "learning_rate": 3.231564007947772e-05,
      "loss": 0.00010906457901000976,
      "step": 238550
    },
    {
      "epoch": 67.71501561169458,
      "grad_norm": 0.032431695610284805,
      "learning_rate": 3.231280158955436e-05,
      "loss": 9.471699595451356e-05,
      "step": 238560
    },
    {
      "epoch": 67.71785410161795,
      "grad_norm": 0.028268298134207726,
      "learning_rate": 3.2309963099630996e-05,
      "loss": 0.00031718965619802474,
      "step": 238570
    },
    {
      "epoch": 67.7206925915413,
      "grad_norm": 0.04011308401823044,
      "learning_rate": 3.230712460970764e-05,
      "loss": 0.0008948517963290215,
      "step": 238580
    },
    {
      "epoch": 67.72353108146466,
      "grad_norm": 0.02729710005223751,
      "learning_rate": 3.230428611978428e-05,
      "loss": 0.0005768748000264168,
      "step": 238590
    },
    {
      "epoch": 67.72636957138802,
      "grad_norm": 0.009874220006167889,
      "learning_rate": 3.2301447629860914e-05,
      "loss": 0.00023008212447166442,
      "step": 238600
    },
    {
      "epoch": 67.72920806131138,
      "grad_norm": 0.1460953801870346,
      "learning_rate": 3.2298609139937555e-05,
      "loss": 0.00011670701205730438,
      "step": 238610
    },
    {
      "epoch": 67.73204655123475,
      "grad_norm": 0.0077936421148478985,
      "learning_rate": 3.2295770650014196e-05,
      "loss": 0.0011231679469347,
      "step": 238620
    },
    {
      "epoch": 67.73488504115811,
      "grad_norm": 1.9368336200714111,
      "learning_rate": 3.229293216009083e-05,
      "loss": 0.0004939043894410134,
      "step": 238630
    },
    {
      "epoch": 67.73772353108147,
      "grad_norm": 0.024576999247074127,
      "learning_rate": 3.229009367016747e-05,
      "loss": 0.0004694797098636627,
      "step": 238640
    },
    {
      "epoch": 67.74056202100482,
      "grad_norm": 11.660805702209473,
      "learning_rate": 3.2287255180244114e-05,
      "loss": 0.0031213566660881044,
      "step": 238650
    },
    {
      "epoch": 67.74340051092818,
      "grad_norm": 0.08496185392141342,
      "learning_rate": 3.228441669032075e-05,
      "loss": 0.005450227856636047,
      "step": 238660
    },
    {
      "epoch": 67.74623900085155,
      "grad_norm": 0.08500491827726364,
      "learning_rate": 3.228157820039739e-05,
      "loss": 0.00013103038072586058,
      "step": 238670
    },
    {
      "epoch": 67.74907749077491,
      "grad_norm": 0.009951264597475529,
      "learning_rate": 3.227873971047403e-05,
      "loss": 0.00030443388968706133,
      "step": 238680
    },
    {
      "epoch": 67.75191598069827,
      "grad_norm": 0.26333028078079224,
      "learning_rate": 3.227590122055067e-05,
      "loss": 0.0004402417689561844,
      "step": 238690
    },
    {
      "epoch": 67.75475447062163,
      "grad_norm": 0.03946924954652786,
      "learning_rate": 3.227306273062731e-05,
      "loss": 0.00012253690510988235,
      "step": 238700
    },
    {
      "epoch": 67.757592960545,
      "grad_norm": 0.007652658503502607,
      "learning_rate": 3.227022424070394e-05,
      "loss": 0.00025032274425029755,
      "step": 238710
    },
    {
      "epoch": 67.76043145046835,
      "grad_norm": 0.02098272740840912,
      "learning_rate": 3.226738575078059e-05,
      "loss": 0.0020718835294246674,
      "step": 238720
    },
    {
      "epoch": 67.76326994039171,
      "grad_norm": 0.0514928437769413,
      "learning_rate": 3.2264547260857224e-05,
      "loss": 0.00013334508985280992,
      "step": 238730
    },
    {
      "epoch": 67.76610843031507,
      "grad_norm": 0.012351004406809807,
      "learning_rate": 3.2261708770933866e-05,
      "loss": 0.001666281372308731,
      "step": 238740
    },
    {
      "epoch": 67.76894692023843,
      "grad_norm": 0.0037720007821917534,
      "learning_rate": 3.225887028101051e-05,
      "loss": 0.0009134652093052865,
      "step": 238750
    },
    {
      "epoch": 67.7717854101618,
      "grad_norm": 2.0431694984436035,
      "learning_rate": 3.225603179108714e-05,
      "loss": 0.000365748442709446,
      "step": 238760
    },
    {
      "epoch": 67.77462390008516,
      "grad_norm": 0.06703159213066101,
      "learning_rate": 3.225319330116378e-05,
      "loss": 0.0005766186863183975,
      "step": 238770
    },
    {
      "epoch": 67.77746239000851,
      "grad_norm": 17.542850494384766,
      "learning_rate": 3.2250354811240424e-05,
      "loss": 0.0034675572067499163,
      "step": 238780
    },
    {
      "epoch": 67.78030087993187,
      "grad_norm": 0.05222545936703682,
      "learning_rate": 3.224751632131706e-05,
      "loss": 0.0011237161234021187,
      "step": 238790
    },
    {
      "epoch": 67.78313936985523,
      "grad_norm": 0.6651024222373962,
      "learning_rate": 3.22446778313937e-05,
      "loss": 0.0012725075706839562,
      "step": 238800
    },
    {
      "epoch": 67.7859778597786,
      "grad_norm": 1.940011739730835,
      "learning_rate": 3.2241839341470335e-05,
      "loss": 0.0006208794191479682,
      "step": 238810
    },
    {
      "epoch": 67.78881634970196,
      "grad_norm": 0.12565436959266663,
      "learning_rate": 3.223900085154698e-05,
      "loss": 0.0002481468021869659,
      "step": 238820
    },
    {
      "epoch": 67.79165483962532,
      "grad_norm": 0.1693343073129654,
      "learning_rate": 3.223616236162362e-05,
      "loss": 0.00035248007625341413,
      "step": 238830
    },
    {
      "epoch": 67.79449332954869,
      "grad_norm": 0.23362968862056732,
      "learning_rate": 3.223332387170025e-05,
      "loss": 0.0004392711445689201,
      "step": 238840
    },
    {
      "epoch": 67.79733181947203,
      "grad_norm": 0.13409173488616943,
      "learning_rate": 3.22304853817769e-05,
      "loss": 0.001935061626136303,
      "step": 238850
    },
    {
      "epoch": 67.8001703093954,
      "grad_norm": 0.02549491636455059,
      "learning_rate": 3.2227646891853535e-05,
      "loss": 0.00033534429967403413,
      "step": 238860
    },
    {
      "epoch": 67.80300879931876,
      "grad_norm": 0.03381893038749695,
      "learning_rate": 3.2224808401930177e-05,
      "loss": 0.00023301038891077042,
      "step": 238870
    },
    {
      "epoch": 67.80584728924212,
      "grad_norm": 1.7774205207824707,
      "learning_rate": 3.222196991200682e-05,
      "loss": 0.00046814922243356707,
      "step": 238880
    },
    {
      "epoch": 67.80868577916549,
      "grad_norm": 0.019339250400662422,
      "learning_rate": 3.221913142208345e-05,
      "loss": 0.00023904722183942795,
      "step": 238890
    },
    {
      "epoch": 67.81152426908885,
      "grad_norm": 0.1288164258003235,
      "learning_rate": 3.2216292932160094e-05,
      "loss": 0.0018879311159253121,
      "step": 238900
    },
    {
      "epoch": 67.81436275901221,
      "grad_norm": 0.057192686945199966,
      "learning_rate": 3.221345444223673e-05,
      "loss": 0.0005377192050218582,
      "step": 238910
    },
    {
      "epoch": 67.81720124893556,
      "grad_norm": 0.6482744812965393,
      "learning_rate": 3.221061595231337e-05,
      "loss": 0.0004144478589296341,
      "step": 238920
    },
    {
      "epoch": 67.82003973885892,
      "grad_norm": 0.00415580440312624,
      "learning_rate": 3.220777746239001e-05,
      "loss": 0.00017345640808343887,
      "step": 238930
    },
    {
      "epoch": 67.82287822878229,
      "grad_norm": 0.0669969841837883,
      "learning_rate": 3.2204938972466646e-05,
      "loss": 0.0003446388989686966,
      "step": 238940
    },
    {
      "epoch": 67.82571671870565,
      "grad_norm": 0.016007406637072563,
      "learning_rate": 3.220210048254329e-05,
      "loss": 0.00019704941660165788,
      "step": 238950
    },
    {
      "epoch": 67.82855520862901,
      "grad_norm": 0.057914961129426956,
      "learning_rate": 3.219926199261993e-05,
      "loss": 0.00020583122968673707,
      "step": 238960
    },
    {
      "epoch": 67.83139369855238,
      "grad_norm": 0.06766346096992493,
      "learning_rate": 3.219642350269656e-05,
      "loss": 0.006274516880512238,
      "step": 238970
    },
    {
      "epoch": 67.83423218847572,
      "grad_norm": 0.046752460300922394,
      "learning_rate": 3.219358501277321e-05,
      "loss": 0.00026107355952262876,
      "step": 238980
    },
    {
      "epoch": 67.83707067839909,
      "grad_norm": 0.03323877230286598,
      "learning_rate": 3.2190746522849846e-05,
      "loss": 0.00024107899516820908,
      "step": 238990
    },
    {
      "epoch": 67.83990916832245,
      "grad_norm": 0.5278711915016174,
      "learning_rate": 3.218790803292648e-05,
      "loss": 0.0002707468345761299,
      "step": 239000
    },
    {
      "epoch": 67.83990916832245,
      "eval_accuracy": 0.9790169771730146,
      "eval_loss": 0.0726800188422203,
      "eval_runtime": 32.7351,
      "eval_samples_per_second": 480.432,
      "eval_steps_per_second": 7.515,
      "step": 239000
    },
    {
      "epoch": 67.84274765824581,
      "grad_norm": 0.08891092240810394,
      "learning_rate": 3.218506954300312e-05,
      "loss": 0.00024964120239019395,
      "step": 239010
    },
    {
      "epoch": 67.84558614816918,
      "grad_norm": 0.010999798774719238,
      "learning_rate": 3.218223105307976e-05,
      "loss": 0.0009563446044921875,
      "step": 239020
    },
    {
      "epoch": 67.84842463809254,
      "grad_norm": 0.09018875658512115,
      "learning_rate": 3.2179392563156405e-05,
      "loss": 0.0006124503910541534,
      "step": 239030
    },
    {
      "epoch": 67.8512631280159,
      "grad_norm": 0.11560098826885223,
      "learning_rate": 3.217655407323304e-05,
      "loss": 0.0005053102970123291,
      "step": 239040
    },
    {
      "epoch": 67.85410161793925,
      "grad_norm": 0.0025701841805130243,
      "learning_rate": 3.217371558330968e-05,
      "loss": 0.0006140125915408134,
      "step": 239050
    },
    {
      "epoch": 67.85694010786261,
      "grad_norm": 0.01077443640679121,
      "learning_rate": 3.217087709338632e-05,
      "loss": 8.740071207284928e-05,
      "step": 239060
    },
    {
      "epoch": 67.85977859778598,
      "grad_norm": 0.027358269318938255,
      "learning_rate": 3.2168038603462957e-05,
      "loss": 9.042751044034958e-05,
      "step": 239070
    },
    {
      "epoch": 67.86261708770934,
      "grad_norm": 0.049554865807294846,
      "learning_rate": 3.21652001135396e-05,
      "loss": 0.00022278353571891785,
      "step": 239080
    },
    {
      "epoch": 67.8654555776327,
      "grad_norm": 0.027202622964978218,
      "learning_rate": 3.216236162361624e-05,
      "loss": 6.805919110774994e-05,
      "step": 239090
    },
    {
      "epoch": 67.86829406755606,
      "grad_norm": 0.18732291460037231,
      "learning_rate": 3.2159523133692874e-05,
      "loss": 9.880848228931427e-05,
      "step": 239100
    },
    {
      "epoch": 67.87113255747943,
      "grad_norm": 0.08058620244264603,
      "learning_rate": 3.2156684643769515e-05,
      "loss": 9.300950914621353e-05,
      "step": 239110
    },
    {
      "epoch": 67.87397104740278,
      "grad_norm": 0.09692882746458054,
      "learning_rate": 3.215384615384616e-05,
      "loss": 0.00022635981440544128,
      "step": 239120
    },
    {
      "epoch": 67.87680953732614,
      "grad_norm": 0.03624165803194046,
      "learning_rate": 3.215100766392279e-05,
      "loss": 0.0003705183044075966,
      "step": 239130
    },
    {
      "epoch": 67.8796480272495,
      "grad_norm": 0.008550352416932583,
      "learning_rate": 3.214816917399943e-05,
      "loss": 0.0002673080191016197,
      "step": 239140
    },
    {
      "epoch": 67.88248651717286,
      "grad_norm": 0.022851472720503807,
      "learning_rate": 3.2145330684076074e-05,
      "loss": 0.00014623478055000305,
      "step": 239150
    },
    {
      "epoch": 67.88532500709623,
      "grad_norm": 0.032431501895189285,
      "learning_rate": 3.2142492194152715e-05,
      "loss": 0.0001826748251914978,
      "step": 239160
    },
    {
      "epoch": 67.88816349701959,
      "grad_norm": 0.0038551161997020245,
      "learning_rate": 3.213965370422935e-05,
      "loss": 0.0011141641065478326,
      "step": 239170
    },
    {
      "epoch": 67.89100198694295,
      "grad_norm": 0.055525630712509155,
      "learning_rate": 3.213681521430599e-05,
      "loss": 0.0006488323211669922,
      "step": 239180
    },
    {
      "epoch": 67.8938404768663,
      "grad_norm": 0.1834893822669983,
      "learning_rate": 3.213397672438263e-05,
      "loss": 0.0006844073534011841,
      "step": 239190
    },
    {
      "epoch": 67.89667896678966,
      "grad_norm": 0.025406358763575554,
      "learning_rate": 3.213113823445927e-05,
      "loss": 0.007703074812889099,
      "step": 239200
    },
    {
      "epoch": 67.89951745671303,
      "grad_norm": 0.056245721876621246,
      "learning_rate": 3.212829974453591e-05,
      "loss": 0.00015945956110954284,
      "step": 239210
    },
    {
      "epoch": 67.90235594663639,
      "grad_norm": 0.017768876627087593,
      "learning_rate": 3.212546125461255e-05,
      "loss": 0.00016051698476076125,
      "step": 239220
    },
    {
      "epoch": 67.90519443655975,
      "grad_norm": 0.010318760760128498,
      "learning_rate": 3.2122622764689185e-05,
      "loss": 0.00022235512733459473,
      "step": 239230
    },
    {
      "epoch": 67.90803292648312,
      "grad_norm": 0.12045524269342422,
      "learning_rate": 3.2119784274765826e-05,
      "loss": 0.0001407502219080925,
      "step": 239240
    },
    {
      "epoch": 67.91087141640647,
      "grad_norm": 0.020207002758979797,
      "learning_rate": 3.211694578484247e-05,
      "loss": 0.00024699252098798753,
      "step": 239250
    },
    {
      "epoch": 67.91370990632983,
      "grad_norm": 0.012565678916871548,
      "learning_rate": 3.21141072949191e-05,
      "loss": 0.00024842359125614165,
      "step": 239260
    },
    {
      "epoch": 67.91654839625319,
      "grad_norm": 0.015937281772494316,
      "learning_rate": 3.2111268804995743e-05,
      "loss": 0.0003131747245788574,
      "step": 239270
    },
    {
      "epoch": 67.91938688617655,
      "grad_norm": 0.015091205015778542,
      "learning_rate": 3.2108430315072385e-05,
      "loss": 0.00020168200135231018,
      "step": 239280
    },
    {
      "epoch": 67.92222537609992,
      "grad_norm": 1.3054227828979492,
      "learning_rate": 3.2105591825149026e-05,
      "loss": 0.00245161484926939,
      "step": 239290
    },
    {
      "epoch": 67.92506386602328,
      "grad_norm": 0.5125373601913452,
      "learning_rate": 3.210275333522566e-05,
      "loss": 0.00034980140626430514,
      "step": 239300
    },
    {
      "epoch": 67.92790235594664,
      "grad_norm": 0.2078283131122589,
      "learning_rate": 3.2099914845302295e-05,
      "loss": 0.0002550220116972923,
      "step": 239310
    },
    {
      "epoch": 67.93074084586999,
      "grad_norm": 0.011252552270889282,
      "learning_rate": 3.2097076355378943e-05,
      "loss": 0.0008575843647122383,
      "step": 239320
    },
    {
      "epoch": 67.93357933579335,
      "grad_norm": 0.05154822766780853,
      "learning_rate": 3.209423786545558e-05,
      "loss": 0.00016746986657381057,
      "step": 239330
    },
    {
      "epoch": 67.93641782571672,
      "grad_norm": 0.26395636796951294,
      "learning_rate": 3.209139937553222e-05,
      "loss": 0.0006748367100954056,
      "step": 239340
    },
    {
      "epoch": 67.93925631564008,
      "grad_norm": 0.018109094351530075,
      "learning_rate": 3.208856088560886e-05,
      "loss": 0.00014802105724811553,
      "step": 239350
    },
    {
      "epoch": 67.94209480556344,
      "grad_norm": 0.8936353325843811,
      "learning_rate": 3.2085722395685495e-05,
      "loss": 0.000566783919930458,
      "step": 239360
    },
    {
      "epoch": 67.9449332954868,
      "grad_norm": 5.948686599731445,
      "learning_rate": 3.208288390576214e-05,
      "loss": 0.0006142910569906235,
      "step": 239370
    },
    {
      "epoch": 67.94777178541017,
      "grad_norm": 2.0921225547790527,
      "learning_rate": 3.208004541583878e-05,
      "loss": 0.006413586437702179,
      "step": 239380
    },
    {
      "epoch": 67.95061027533352,
      "grad_norm": 0.41564705967903137,
      "learning_rate": 3.207720692591541e-05,
      "loss": 0.0048904623836278915,
      "step": 239390
    },
    {
      "epoch": 67.95344876525688,
      "grad_norm": 0.025140922516584396,
      "learning_rate": 3.2074368435992054e-05,
      "loss": 0.00023095104843378066,
      "step": 239400
    },
    {
      "epoch": 67.95628725518024,
      "grad_norm": 0.01304276566952467,
      "learning_rate": 3.2071529946068696e-05,
      "loss": 0.0035524383187294007,
      "step": 239410
    },
    {
      "epoch": 67.9591257451036,
      "grad_norm": 0.01920982077717781,
      "learning_rate": 3.206869145614533e-05,
      "loss": 0.0005229920148849488,
      "step": 239420
    },
    {
      "epoch": 67.96196423502697,
      "grad_norm": 0.024764643982052803,
      "learning_rate": 3.206585296622197e-05,
      "loss": 0.00040982067584991453,
      "step": 239430
    },
    {
      "epoch": 67.96480272495033,
      "grad_norm": 0.1697756052017212,
      "learning_rate": 3.2063014476298606e-05,
      "loss": 0.0008992278948426246,
      "step": 239440
    },
    {
      "epoch": 67.96764121487368,
      "grad_norm": 0.0502573698759079,
      "learning_rate": 3.2060175986375254e-05,
      "loss": 0.00046400073915719986,
      "step": 239450
    },
    {
      "epoch": 67.97047970479704,
      "grad_norm": 0.4333125352859497,
      "learning_rate": 3.205733749645189e-05,
      "loss": 0.00027176514267921446,
      "step": 239460
    },
    {
      "epoch": 67.9733181947204,
      "grad_norm": 0.27061450481414795,
      "learning_rate": 3.2054499006528523e-05,
      "loss": 0.00045074857771396635,
      "step": 239470
    },
    {
      "epoch": 67.97615668464377,
      "grad_norm": 0.020894957706332207,
      "learning_rate": 3.205166051660517e-05,
      "loss": 0.00021269712597131728,
      "step": 239480
    },
    {
      "epoch": 67.97899517456713,
      "grad_norm": 0.018411338329315186,
      "learning_rate": 3.2048822026681806e-05,
      "loss": 0.00020320788025856017,
      "step": 239490
    },
    {
      "epoch": 67.9818336644905,
      "grad_norm": 0.011295558884739876,
      "learning_rate": 3.204598353675845e-05,
      "loss": 0.00048033874481916427,
      "step": 239500
    },
    {
      "epoch": 67.9818336644905,
      "eval_accuracy": 0.9797164112672474,
      "eval_loss": 0.07567338645458221,
      "eval_runtime": 32.7689,
      "eval_samples_per_second": 479.937,
      "eval_steps_per_second": 7.507,
      "step": 239500
    },
    {
      "epoch": 67.98467215441386,
      "grad_norm": 0.03979369252920151,
      "learning_rate": 3.204314504683509e-05,
      "loss": 0.00019617974758148193,
      "step": 239510
    },
    {
      "epoch": 67.9875106443372,
      "grad_norm": 0.020957887172698975,
      "learning_rate": 3.2040306556911724e-05,
      "loss": 0.0007206477224826813,
      "step": 239520
    },
    {
      "epoch": 67.99034913426057,
      "grad_norm": 1.2131863832473755,
      "learning_rate": 3.2037468066988365e-05,
      "loss": 0.0008562853559851647,
      "step": 239530
    },
    {
      "epoch": 67.99318762418393,
      "grad_norm": 0.07259994745254517,
      "learning_rate": 3.2034629577065e-05,
      "loss": 0.0002510426566004753,
      "step": 239540
    },
    {
      "epoch": 67.9960261141073,
      "grad_norm": 0.10652611404657364,
      "learning_rate": 3.203179108714164e-05,
      "loss": 0.00014256089925765992,
      "step": 239550
    },
    {
      "epoch": 67.99886460403066,
      "grad_norm": 0.4638199806213379,
      "learning_rate": 3.202895259721828e-05,
      "loss": 0.00019073598086833955,
      "step": 239560
    },
    {
      "epoch": 68.00170309395402,
      "grad_norm": 0.01527088787406683,
      "learning_rate": 3.202611410729492e-05,
      "loss": 0.0014104208908975123,
      "step": 239570
    },
    {
      "epoch": 68.00454158387738,
      "grad_norm": 0.672959566116333,
      "learning_rate": 3.2023275617371565e-05,
      "loss": 0.0001778785139322281,
      "step": 239580
    },
    {
      "epoch": 68.00738007380073,
      "grad_norm": 0.14551571011543274,
      "learning_rate": 3.20204371274482e-05,
      "loss": 0.0003165684640407562,
      "step": 239590
    },
    {
      "epoch": 68.0102185637241,
      "grad_norm": 0.026756251230835915,
      "learning_rate": 3.2017598637524834e-05,
      "loss": 0.00011974945664405823,
      "step": 239600
    },
    {
      "epoch": 68.01305705364746,
      "grad_norm": 0.05024335905909538,
      "learning_rate": 3.201476014760148e-05,
      "loss": 0.0001674514263868332,
      "step": 239610
    },
    {
      "epoch": 68.01589554357082,
      "grad_norm": 0.03060646913945675,
      "learning_rate": 3.201192165767812e-05,
      "loss": 0.0003626551479101181,
      "step": 239620
    },
    {
      "epoch": 68.01873403349418,
      "grad_norm": 0.023660900071263313,
      "learning_rate": 3.200908316775476e-05,
      "loss": 0.0005819117650389671,
      "step": 239630
    },
    {
      "epoch": 68.02157252341755,
      "grad_norm": 0.10477801412343979,
      "learning_rate": 3.200624467783139e-05,
      "loss": 0.0010633833706378936,
      "step": 239640
    },
    {
      "epoch": 68.02441101334091,
      "grad_norm": 0.06702311336994171,
      "learning_rate": 3.2003406187908034e-05,
      "loss": 0.0002253001555800438,
      "step": 239650
    },
    {
      "epoch": 68.02724950326426,
      "grad_norm": 0.03948019817471504,
      "learning_rate": 3.2000567697984676e-05,
      "loss": 8.474420756101608e-05,
      "step": 239660
    },
    {
      "epoch": 68.03008799318762,
      "grad_norm": 0.0032265442423522472,
      "learning_rate": 3.199772920806131e-05,
      "loss": 0.00021944548934698104,
      "step": 239670
    },
    {
      "epoch": 68.03292648311098,
      "grad_norm": 0.11221322417259216,
      "learning_rate": 3.199489071813795e-05,
      "loss": 0.00021465197205543517,
      "step": 239680
    },
    {
      "epoch": 68.03576497303435,
      "grad_norm": 0.3614286184310913,
      "learning_rate": 3.199205222821459e-05,
      "loss": 0.0001493636518716812,
      "step": 239690
    },
    {
      "epoch": 68.03860346295771,
      "grad_norm": 0.016395343467593193,
      "learning_rate": 3.198921373829123e-05,
      "loss": 0.00014828331768512726,
      "step": 239700
    },
    {
      "epoch": 68.04144195288107,
      "grad_norm": 0.05956179276108742,
      "learning_rate": 3.1986375248367876e-05,
      "loss": 0.00017050504684448242,
      "step": 239710
    },
    {
      "epoch": 68.04428044280442,
      "grad_norm": 0.015231164172291756,
      "learning_rate": 3.198353675844451e-05,
      "loss": 9.020362049341201e-05,
      "step": 239720
    },
    {
      "epoch": 68.04711893272778,
      "grad_norm": 0.048278044909238815,
      "learning_rate": 3.1980698268521145e-05,
      "loss": 0.00017325934022665023,
      "step": 239730
    },
    {
      "epoch": 68.04995742265115,
      "grad_norm": 0.14237728714942932,
      "learning_rate": 3.1977859778597786e-05,
      "loss": 8.321087807416916e-05,
      "step": 239740
    },
    {
      "epoch": 68.05279591257451,
      "grad_norm": 0.02084796316921711,
      "learning_rate": 3.197502128867443e-05,
      "loss": 9.985268115997314e-05,
      "step": 239750
    },
    {
      "epoch": 68.05563440249787,
      "grad_norm": 0.0071514141745865345,
      "learning_rate": 3.197218279875107e-05,
      "loss": 9.639766067266464e-05,
      "step": 239760
    },
    {
      "epoch": 68.05847289242124,
      "grad_norm": 0.007235251367092133,
      "learning_rate": 3.1969344308827704e-05,
      "loss": 4.810839891433716e-05,
      "step": 239770
    },
    {
      "epoch": 68.0613113823446,
      "grad_norm": 0.36641374230384827,
      "learning_rate": 3.1966505818904345e-05,
      "loss": 0.00017156768590211868,
      "step": 239780
    },
    {
      "epoch": 68.06414987226795,
      "grad_norm": 0.3767191171646118,
      "learning_rate": 3.1963667328980986e-05,
      "loss": 0.00010190773755311966,
      "step": 239790
    },
    {
      "epoch": 68.06698836219131,
      "grad_norm": 0.01663331314921379,
      "learning_rate": 3.196082883905762e-05,
      "loss": 0.00015477370470762253,
      "step": 239800
    },
    {
      "epoch": 68.06982685211467,
      "grad_norm": 0.018213288858532906,
      "learning_rate": 3.195799034913426e-05,
      "loss": 6.615892052650452e-05,
      "step": 239810
    },
    {
      "epoch": 68.07266534203804,
      "grad_norm": 0.02156265266239643,
      "learning_rate": 3.1955151859210904e-05,
      "loss": 9.41479578614235e-05,
      "step": 239820
    },
    {
      "epoch": 68.0755038319614,
      "grad_norm": 0.009384107775986195,
      "learning_rate": 3.195231336928754e-05,
      "loss": 5.077756941318512e-05,
      "step": 239830
    },
    {
      "epoch": 68.07834232188476,
      "grad_norm": 0.009477132931351662,
      "learning_rate": 3.194947487936418e-05,
      "loss": 7.892083376646042e-05,
      "step": 239840
    },
    {
      "epoch": 68.08118081180812,
      "grad_norm": 0.019423970952630043,
      "learning_rate": 3.194663638944082e-05,
      "loss": 7.848497480154038e-05,
      "step": 239850
    },
    {
      "epoch": 68.08401930173147,
      "grad_norm": 0.029515594244003296,
      "learning_rate": 3.1943797899517456e-05,
      "loss": 8.597448468208312e-05,
      "step": 239860
    },
    {
      "epoch": 68.08685779165484,
      "grad_norm": 0.08134137094020844,
      "learning_rate": 3.19409594095941e-05,
      "loss": 6.95016235113144e-05,
      "step": 239870
    },
    {
      "epoch": 68.0896962815782,
      "grad_norm": 0.01796061173081398,
      "learning_rate": 3.193812091967074e-05,
      "loss": 9.283702820539474e-05,
      "step": 239880
    },
    {
      "epoch": 68.09253477150156,
      "grad_norm": 0.0036668998654931784,
      "learning_rate": 3.193528242974737e-05,
      "loss": 0.00021968837827444077,
      "step": 239890
    },
    {
      "epoch": 68.09537326142492,
      "grad_norm": 0.014884999953210354,
      "learning_rate": 3.1932443939824014e-05,
      "loss": 0.00020617879927158355,
      "step": 239900
    },
    {
      "epoch": 68.09821175134829,
      "grad_norm": 0.02411527745425701,
      "learning_rate": 3.1929605449900656e-05,
      "loss": 0.0003760108724236488,
      "step": 239910
    },
    {
      "epoch": 68.10105024127165,
      "grad_norm": 0.20173344016075134,
      "learning_rate": 3.19267669599773e-05,
      "loss": 0.00012201908975839615,
      "step": 239920
    },
    {
      "epoch": 68.103888731195,
      "grad_norm": 0.009099855087697506,
      "learning_rate": 3.192392847005393e-05,
      "loss": 0.00021886173635721207,
      "step": 239930
    },
    {
      "epoch": 68.10672722111836,
      "grad_norm": 0.025608666241168976,
      "learning_rate": 3.1921089980130566e-05,
      "loss": 0.00010616090148687362,
      "step": 239940
    },
    {
      "epoch": 68.10956571104172,
      "grad_norm": 0.08217760175466537,
      "learning_rate": 3.1918251490207215e-05,
      "loss": 0.0012155858799815177,
      "step": 239950
    },
    {
      "epoch": 68.11240420096509,
      "grad_norm": 0.11911110579967499,
      "learning_rate": 3.191541300028385e-05,
      "loss": 0.00010257884860038757,
      "step": 239960
    },
    {
      "epoch": 68.11524269088845,
      "grad_norm": 0.061949148774147034,
      "learning_rate": 3.191257451036049e-05,
      "loss": 0.0002506054937839508,
      "step": 239970
    },
    {
      "epoch": 68.11808118081181,
      "grad_norm": 0.009435508400201797,
      "learning_rate": 3.190973602043713e-05,
      "loss": 0.0008081628009676933,
      "step": 239980
    },
    {
      "epoch": 68.12091967073516,
      "grad_norm": 0.38836342096328735,
      "learning_rate": 3.1906897530513767e-05,
      "loss": 0.001847013644874096,
      "step": 239990
    },
    {
      "epoch": 68.12375816065853,
      "grad_norm": 0.40096041560173035,
      "learning_rate": 3.190405904059041e-05,
      "loss": 0.009953907877206802,
      "step": 240000
    },
    {
      "epoch": 68.12375816065853,
      "eval_accuracy": 0.9759649011254531,
      "eval_loss": 0.09003924578428268,
      "eval_runtime": 32.2303,
      "eval_samples_per_second": 487.957,
      "eval_steps_per_second": 7.633,
      "step": 240000
    },
    {
      "epoch": 68.12659665058189,
      "grad_norm": 0.03186274319887161,
      "learning_rate": 3.190122055066705e-05,
      "loss": 0.0025125518441200256,
      "step": 240010
    },
    {
      "epoch": 68.12943514050525,
      "grad_norm": 0.04875575378537178,
      "learning_rate": 3.1898382060743684e-05,
      "loss": 0.00014507118612527847,
      "step": 240020
    },
    {
      "epoch": 68.13227363042861,
      "grad_norm": 0.15903933346271515,
      "learning_rate": 3.1895543570820325e-05,
      "loss": 0.0008055312559008598,
      "step": 240030
    },
    {
      "epoch": 68.13511212035198,
      "grad_norm": 0.028924977406859398,
      "learning_rate": 3.189270508089696e-05,
      "loss": 0.000626072846353054,
      "step": 240040
    },
    {
      "epoch": 68.13795061027534,
      "grad_norm": 0.023683154955506325,
      "learning_rate": 3.188986659097361e-05,
      "loss": 0.00023929905146360397,
      "step": 240050
    },
    {
      "epoch": 68.14078910019869,
      "grad_norm": 0.09223537147045135,
      "learning_rate": 3.188702810105024e-05,
      "loss": 0.00028151758015155793,
      "step": 240060
    },
    {
      "epoch": 68.14362759012205,
      "grad_norm": 0.15968625247478485,
      "learning_rate": 3.188418961112688e-05,
      "loss": 0.00018699206411838533,
      "step": 240070
    },
    {
      "epoch": 68.14646608004541,
      "grad_norm": 0.03196307271718979,
      "learning_rate": 3.1881351121203525e-05,
      "loss": 0.0007780559360980987,
      "step": 240080
    },
    {
      "epoch": 68.14930456996878,
      "grad_norm": 0.3090411424636841,
      "learning_rate": 3.187851263128016e-05,
      "loss": 0.002107728458940983,
      "step": 240090
    },
    {
      "epoch": 68.15214305989214,
      "grad_norm": 0.04469125717878342,
      "learning_rate": 3.18756741413568e-05,
      "loss": 0.018842391669750214,
      "step": 240100
    },
    {
      "epoch": 68.1549815498155,
      "grad_norm": 0.04250493273139,
      "learning_rate": 3.187283565143344e-05,
      "loss": 0.0026277715340256693,
      "step": 240110
    },
    {
      "epoch": 68.15782003973887,
      "grad_norm": 0.22871221601963043,
      "learning_rate": 3.186999716151008e-05,
      "loss": 0.0003323571756482124,
      "step": 240120
    },
    {
      "epoch": 68.16065852966221,
      "grad_norm": 0.00623375503346324,
      "learning_rate": 3.186715867158672e-05,
      "loss": 0.00044661667197942734,
      "step": 240130
    },
    {
      "epoch": 68.16349701958558,
      "grad_norm": 0.06654512137174606,
      "learning_rate": 3.186432018166335e-05,
      "loss": 0.00046465154737234115,
      "step": 240140
    },
    {
      "epoch": 68.16633550950894,
      "grad_norm": 0.16486942768096924,
      "learning_rate": 3.1861481691739995e-05,
      "loss": 0.0005045639351010323,
      "step": 240150
    },
    {
      "epoch": 68.1691739994323,
      "grad_norm": 0.0629473403096199,
      "learning_rate": 3.1858643201816636e-05,
      "loss": 0.006023868173360825,
      "step": 240160
    },
    {
      "epoch": 68.17201248935567,
      "grad_norm": 0.04611983522772789,
      "learning_rate": 3.185580471189327e-05,
      "loss": 0.0003699101507663727,
      "step": 240170
    },
    {
      "epoch": 68.17485097927903,
      "grad_norm": 0.1408795416355133,
      "learning_rate": 3.185296622196992e-05,
      "loss": 0.0002438744530081749,
      "step": 240180
    },
    {
      "epoch": 68.17768946920238,
      "grad_norm": 0.02219211496412754,
      "learning_rate": 3.185012773204655e-05,
      "loss": 0.0004327865317463875,
      "step": 240190
    },
    {
      "epoch": 68.18052795912574,
      "grad_norm": 0.06677475571632385,
      "learning_rate": 3.184728924212319e-05,
      "loss": 0.0001890163868665695,
      "step": 240200
    },
    {
      "epoch": 68.1833664490491,
      "grad_norm": 0.035636577755212784,
      "learning_rate": 3.1844450752199836e-05,
      "loss": 0.00013316087424755096,
      "step": 240210
    },
    {
      "epoch": 68.18620493897247,
      "grad_norm": 0.03240849822759628,
      "learning_rate": 3.184161226227647e-05,
      "loss": 0.00017379093915224076,
      "step": 240220
    },
    {
      "epoch": 68.18904342889583,
      "grad_norm": 0.004642915446311235,
      "learning_rate": 3.183877377235311e-05,
      "loss": 0.0001903582364320755,
      "step": 240230
    },
    {
      "epoch": 68.19188191881919,
      "grad_norm": 0.05091605335474014,
      "learning_rate": 3.183593528242975e-05,
      "loss": 0.00011349134147167205,
      "step": 240240
    },
    {
      "epoch": 68.19472040874255,
      "grad_norm": 0.07790250331163406,
      "learning_rate": 3.183309679250639e-05,
      "loss": 0.00011327918618917466,
      "step": 240250
    },
    {
      "epoch": 68.1975588986659,
      "grad_norm": 0.015658201649785042,
      "learning_rate": 3.183025830258303e-05,
      "loss": 9.598899632692336e-05,
      "step": 240260
    },
    {
      "epoch": 68.20039738858927,
      "grad_norm": 0.020649883896112442,
      "learning_rate": 3.1827419812659664e-05,
      "loss": 0.0001398073509335518,
      "step": 240270
    },
    {
      "epoch": 68.20323587851263,
      "grad_norm": 0.019328974187374115,
      "learning_rate": 3.1824581322736305e-05,
      "loss": 0.00028530675917863845,
      "step": 240280
    },
    {
      "epoch": 68.20607436843599,
      "grad_norm": 0.045291125774383545,
      "learning_rate": 3.182174283281295e-05,
      "loss": 8.118730038404464e-05,
      "step": 240290
    },
    {
      "epoch": 68.20891285835935,
      "grad_norm": 0.03199334442615509,
      "learning_rate": 3.181890434288958e-05,
      "loss": 0.00014265794306993483,
      "step": 240300
    },
    {
      "epoch": 68.21175134828272,
      "grad_norm": 0.14129292964935303,
      "learning_rate": 3.181606585296622e-05,
      "loss": 0.00032215677201747895,
      "step": 240310
    },
    {
      "epoch": 68.21458983820608,
      "grad_norm": 0.007534525822848082,
      "learning_rate": 3.1813227363042864e-05,
      "loss": 9.383615106344223e-05,
      "step": 240320
    },
    {
      "epoch": 68.21742832812943,
      "grad_norm": 0.03280192241072655,
      "learning_rate": 3.18103888731195e-05,
      "loss": 0.00014167334884405137,
      "step": 240330
    },
    {
      "epoch": 68.22026681805279,
      "grad_norm": 0.016721315681934357,
      "learning_rate": 3.180755038319614e-05,
      "loss": 9.130574762821198e-05,
      "step": 240340
    },
    {
      "epoch": 68.22310530797616,
      "grad_norm": 0.0037161665968596935,
      "learning_rate": 3.180471189327278e-05,
      "loss": 0.00033390335738658905,
      "step": 240350
    },
    {
      "epoch": 68.22594379789952,
      "grad_norm": 0.011655728332698345,
      "learning_rate": 3.1801873403349416e-05,
      "loss": 0.00029209908097982407,
      "step": 240360
    },
    {
      "epoch": 68.22878228782288,
      "grad_norm": 0.2952776253223419,
      "learning_rate": 3.179903491342606e-05,
      "loss": 0.00014993790537118912,
      "step": 240370
    },
    {
      "epoch": 68.23162077774624,
      "grad_norm": 0.037166547030210495,
      "learning_rate": 3.17961964235027e-05,
      "loss": 0.0001272350549697876,
      "step": 240380
    },
    {
      "epoch": 68.2344592676696,
      "grad_norm": 0.21284136176109314,
      "learning_rate": 3.179335793357934e-05,
      "loss": 0.00010514315217733384,
      "step": 240390
    },
    {
      "epoch": 68.23729775759296,
      "grad_norm": 0.03443075343966484,
      "learning_rate": 3.1790519443655975e-05,
      "loss": 0.00011640992015600204,
      "step": 240400
    },
    {
      "epoch": 68.24013624751632,
      "grad_norm": 0.004973619244992733,
      "learning_rate": 3.1787680953732616e-05,
      "loss": 0.00010218154639005662,
      "step": 240410
    },
    {
      "epoch": 68.24297473743968,
      "grad_norm": 2.0718460083007812,
      "learning_rate": 3.178484246380926e-05,
      "loss": 0.00034552887082099913,
      "step": 240420
    },
    {
      "epoch": 68.24581322736304,
      "grad_norm": 0.029423242434859276,
      "learning_rate": 3.178200397388589e-05,
      "loss": 0.00013344772160053253,
      "step": 240430
    },
    {
      "epoch": 68.2486517172864,
      "grad_norm": 0.018500659614801407,
      "learning_rate": 3.1779165483962534e-05,
      "loss": 0.00037356726825237273,
      "step": 240440
    },
    {
      "epoch": 68.25149020720977,
      "grad_norm": 0.03547638654708862,
      "learning_rate": 3.1776326994039175e-05,
      "loss": 0.00010527763515710831,
      "step": 240450
    },
    {
      "epoch": 68.25432869713312,
      "grad_norm": 0.06997086852788925,
      "learning_rate": 3.177348850411581e-05,
      "loss": 0.00012017395347356797,
      "step": 240460
    },
    {
      "epoch": 68.25716718705648,
      "grad_norm": 0.05963843688368797,
      "learning_rate": 3.177065001419245e-05,
      "loss": 0.00015008337795734404,
      "step": 240470
    },
    {
      "epoch": 68.26000567697984,
      "grad_norm": 0.007205261383205652,
      "learning_rate": 3.176781152426909e-05,
      "loss": 0.00014366414397954942,
      "step": 240480
    },
    {
      "epoch": 68.26284416690321,
      "grad_norm": 0.08358176052570343,
      "learning_rate": 3.176497303434573e-05,
      "loss": 0.00010570883750915527,
      "step": 240490
    },
    {
      "epoch": 68.26568265682657,
      "grad_norm": 0.16327393054962158,
      "learning_rate": 3.176213454442237e-05,
      "loss": 0.00011135190725326538,
      "step": 240500
    },
    {
      "epoch": 68.26568265682657,
      "eval_accuracy": 0.9813696191263432,
      "eval_loss": 0.07210800796747208,
      "eval_runtime": 33.0941,
      "eval_samples_per_second": 475.22,
      "eval_steps_per_second": 7.433,
      "step": 240500
    },
    {
      "epoch": 68.26852114674993,
      "grad_norm": 0.2555639147758484,
      "learning_rate": 3.175929605449901e-05,
      "loss": 0.00024756919592618944,
      "step": 240510
    },
    {
      "epoch": 68.2713596366733,
      "grad_norm": 0.020155230537056923,
      "learning_rate": 3.175645756457565e-05,
      "loss": 0.006480222940444947,
      "step": 240520
    },
    {
      "epoch": 68.27419812659664,
      "grad_norm": 0.20619036257266998,
      "learning_rate": 3.1753619074652286e-05,
      "loss": 0.0003016818314790726,
      "step": 240530
    },
    {
      "epoch": 68.27703661652001,
      "grad_norm": 0.05673827975988388,
      "learning_rate": 3.175078058472892e-05,
      "loss": 7.147938013076782e-05,
      "step": 240540
    },
    {
      "epoch": 68.27987510644337,
      "grad_norm": 0.08694854378700256,
      "learning_rate": 3.174794209480557e-05,
      "loss": 0.00044311489909887314,
      "step": 240550
    },
    {
      "epoch": 68.28271359636673,
      "grad_norm": 0.08901243656873703,
      "learning_rate": 3.17451036048822e-05,
      "loss": 0.0002748033031821251,
      "step": 240560
    },
    {
      "epoch": 68.2855520862901,
      "grad_norm": 0.04600939527153969,
      "learning_rate": 3.1742265114958844e-05,
      "loss": 0.00024493541568517687,
      "step": 240570
    },
    {
      "epoch": 68.28839057621346,
      "grad_norm": 0.030282894149422646,
      "learning_rate": 3.1739426625035486e-05,
      "loss": 0.00032851714640855787,
      "step": 240580
    },
    {
      "epoch": 68.29122906613682,
      "grad_norm": 0.02183794043958187,
      "learning_rate": 3.173658813511212e-05,
      "loss": 0.00020386949181556703,
      "step": 240590
    },
    {
      "epoch": 68.29406755606017,
      "grad_norm": 0.016404390335083008,
      "learning_rate": 3.173374964518876e-05,
      "loss": 0.0004429707303643227,
      "step": 240600
    },
    {
      "epoch": 68.29690604598353,
      "grad_norm": 0.06766848266124725,
      "learning_rate": 3.17309111552654e-05,
      "loss": 0.003656989336013794,
      "step": 240610
    },
    {
      "epoch": 68.2997445359069,
      "grad_norm": 0.12801086902618408,
      "learning_rate": 3.172807266534204e-05,
      "loss": 0.0003150664269924164,
      "step": 240620
    },
    {
      "epoch": 68.30258302583026,
      "grad_norm": 0.014854386448860168,
      "learning_rate": 3.172523417541868e-05,
      "loss": 0.0001234179362654686,
      "step": 240630
    },
    {
      "epoch": 68.30542151575362,
      "grad_norm": 0.023155972361564636,
      "learning_rate": 3.1722395685495314e-05,
      "loss": 0.00023823492228984832,
      "step": 240640
    },
    {
      "epoch": 68.30826000567698,
      "grad_norm": 0.03179442882537842,
      "learning_rate": 3.171955719557196e-05,
      "loss": 0.004988441988825798,
      "step": 240650
    },
    {
      "epoch": 68.31109849560033,
      "grad_norm": 0.074126698076725,
      "learning_rate": 3.1716718705648596e-05,
      "loss": 0.002777617797255516,
      "step": 240660
    },
    {
      "epoch": 68.3139369855237,
      "grad_norm": 0.02923712134361267,
      "learning_rate": 3.171388021572523e-05,
      "loss": 0.00030777230858802795,
      "step": 240670
    },
    {
      "epoch": 68.31677547544706,
      "grad_norm": 0.0108087919652462,
      "learning_rate": 3.171104172580188e-05,
      "loss": 0.0021909184753894804,
      "step": 240680
    },
    {
      "epoch": 68.31961396537042,
      "grad_norm": 0.028870506212115288,
      "learning_rate": 3.1708203235878514e-05,
      "loss": 0.0001361273229122162,
      "step": 240690
    },
    {
      "epoch": 68.32245245529379,
      "grad_norm": 0.019778065383434296,
      "learning_rate": 3.1705364745955155e-05,
      "loss": 0.0007711024954915047,
      "step": 240700
    },
    {
      "epoch": 68.32529094521715,
      "grad_norm": 0.4539341628551483,
      "learning_rate": 3.1702526256031796e-05,
      "loss": 0.0001540055498480797,
      "step": 240710
    },
    {
      "epoch": 68.32812943514051,
      "grad_norm": 0.007098381407558918,
      "learning_rate": 3.169968776610843e-05,
      "loss": 0.00012554153800010682,
      "step": 240720
    },
    {
      "epoch": 68.33096792506386,
      "grad_norm": 0.05288795009255409,
      "learning_rate": 3.169684927618507e-05,
      "loss": 0.00014040619134902954,
      "step": 240730
    },
    {
      "epoch": 68.33380641498722,
      "grad_norm": 0.11315499246120453,
      "learning_rate": 3.1694010786261714e-05,
      "loss": 0.00018118899315595628,
      "step": 240740
    },
    {
      "epoch": 68.33664490491059,
      "grad_norm": 0.019296443089842796,
      "learning_rate": 3.169117229633835e-05,
      "loss": 0.00017240773886442184,
      "step": 240750
    },
    {
      "epoch": 68.33948339483395,
      "grad_norm": 0.04340608790516853,
      "learning_rate": 3.168833380641499e-05,
      "loss": 0.00013088025152683259,
      "step": 240760
    },
    {
      "epoch": 68.34232188475731,
      "grad_norm": 0.19458745419979095,
      "learning_rate": 3.1685495316491624e-05,
      "loss": 0.0001453593373298645,
      "step": 240770
    },
    {
      "epoch": 68.34516037468067,
      "grad_norm": 0.10309793055057526,
      "learning_rate": 3.1682656826568266e-05,
      "loss": 0.00020161587744951248,
      "step": 240780
    },
    {
      "epoch": 68.34799886460404,
      "grad_norm": 0.018246382474899292,
      "learning_rate": 3.167981833664491e-05,
      "loss": 7.169153541326523e-05,
      "step": 240790
    },
    {
      "epoch": 68.35083735452739,
      "grad_norm": 0.03828425705432892,
      "learning_rate": 3.167697984672154e-05,
      "loss": 0.0001765858381986618,
      "step": 240800
    },
    {
      "epoch": 68.35367584445075,
      "grad_norm": 0.015065059997141361,
      "learning_rate": 3.167414135679819e-05,
      "loss": 6.811358034610748e-05,
      "step": 240810
    },
    {
      "epoch": 68.35651433437411,
      "grad_norm": 0.10984756052494049,
      "learning_rate": 3.1671302866874824e-05,
      "loss": 9.066332131624222e-05,
      "step": 240820
    },
    {
      "epoch": 68.35935282429747,
      "grad_norm": 0.005068101920187473,
      "learning_rate": 3.166846437695146e-05,
      "loss": 0.00018775071948766707,
      "step": 240830
    },
    {
      "epoch": 68.36219131422084,
      "grad_norm": 0.019757352769374847,
      "learning_rate": 3.166562588702811e-05,
      "loss": 0.00018245410174131394,
      "step": 240840
    },
    {
      "epoch": 68.3650298041442,
      "grad_norm": 0.017005158588290215,
      "learning_rate": 3.166278739710474e-05,
      "loss": 9.020064026117325e-05,
      "step": 240850
    },
    {
      "epoch": 68.36786829406756,
      "grad_norm": 0.027766020968556404,
      "learning_rate": 3.165994890718138e-05,
      "loss": 0.0007189234718680381,
      "step": 240860
    },
    {
      "epoch": 68.37070678399091,
      "grad_norm": 0.04713226854801178,
      "learning_rate": 3.165711041725802e-05,
      "loss": 0.00016366541385650634,
      "step": 240870
    },
    {
      "epoch": 68.37354527391427,
      "grad_norm": 0.5719568729400635,
      "learning_rate": 3.165427192733466e-05,
      "loss": 0.0006589418277144432,
      "step": 240880
    },
    {
      "epoch": 68.37638376383764,
      "grad_norm": 0.011872733011841774,
      "learning_rate": 3.16514334374113e-05,
      "loss": 0.00018921252340078353,
      "step": 240890
    },
    {
      "epoch": 68.379222253761,
      "grad_norm": 0.03641471266746521,
      "learning_rate": 3.1648594947487935e-05,
      "loss": 0.0002554502338171005,
      "step": 240900
    },
    {
      "epoch": 68.38206074368436,
      "grad_norm": 0.013758386485278606,
      "learning_rate": 3.1645756457564576e-05,
      "loss": 0.0002008989453315735,
      "step": 240910
    },
    {
      "epoch": 68.38489923360773,
      "grad_norm": 0.007431195117533207,
      "learning_rate": 3.164291796764122e-05,
      "loss": 0.00019383709877729416,
      "step": 240920
    },
    {
      "epoch": 68.38773772353107,
      "grad_norm": 0.025080854073166847,
      "learning_rate": 3.164007947771785e-05,
      "loss": 8.591152727603913e-05,
      "step": 240930
    },
    {
      "epoch": 68.39057621345444,
      "grad_norm": 0.057088349014520645,
      "learning_rate": 3.16372409877945e-05,
      "loss": 0.00016024746000766754,
      "step": 240940
    },
    {
      "epoch": 68.3934147033778,
      "grad_norm": 0.008913598954677582,
      "learning_rate": 3.1634402497871135e-05,
      "loss": 0.0001309666782617569,
      "step": 240950
    },
    {
      "epoch": 68.39625319330116,
      "grad_norm": 0.38123971223831177,
      "learning_rate": 3.163156400794777e-05,
      "loss": 0.00014156270772218703,
      "step": 240960
    },
    {
      "epoch": 68.39909168322453,
      "grad_norm": 0.012061921879649162,
      "learning_rate": 3.162872551802441e-05,
      "loss": 8.814632892608642e-05,
      "step": 240970
    },
    {
      "epoch": 68.40193017314789,
      "grad_norm": 0.016239678487181664,
      "learning_rate": 3.162588702810105e-05,
      "loss": 0.0001998158171772957,
      "step": 240980
    },
    {
      "epoch": 68.40476866307125,
      "grad_norm": 2.3886466026306152,
      "learning_rate": 3.1623048538177694e-05,
      "loss": 0.000376242958009243,
      "step": 240990
    },
    {
      "epoch": 68.4076071529946,
      "grad_norm": 0.03412915766239166,
      "learning_rate": 3.162021004825433e-05,
      "loss": 0.0010751022025942802,
      "step": 241000
    },
    {
      "epoch": 68.4076071529946,
      "eval_accuracy": 0.9796528263495898,
      "eval_loss": 0.07736816257238388,
      "eval_runtime": 33.2466,
      "eval_samples_per_second": 473.041,
      "eval_steps_per_second": 7.399,
      "step": 241000
    },
    {
      "epoch": 68.41044564291796,
      "grad_norm": 0.0825410708785057,
      "learning_rate": 3.161737155833097e-05,
      "loss": 0.00025074072182178496,
      "step": 241010
    },
    {
      "epoch": 68.41328413284133,
      "grad_norm": 0.30977728962898254,
      "learning_rate": 3.161453306840761e-05,
      "loss": 0.0009983306750655173,
      "step": 241020
    },
    {
      "epoch": 68.41612262276469,
      "grad_norm": 0.13181768357753754,
      "learning_rate": 3.1611694578484246e-05,
      "loss": 0.0002491135150194168,
      "step": 241030
    },
    {
      "epoch": 68.41896111268805,
      "grad_norm": 0.011325988918542862,
      "learning_rate": 3.160885608856089e-05,
      "loss": 0.00018528103828430176,
      "step": 241040
    },
    {
      "epoch": 68.42179960261142,
      "grad_norm": 0.1627286672592163,
      "learning_rate": 3.160601759863753e-05,
      "loss": 0.0002881409600377083,
      "step": 241050
    },
    {
      "epoch": 68.42463809253478,
      "grad_norm": 0.05323462933301926,
      "learning_rate": 3.160317910871416e-05,
      "loss": 8.702389895915985e-05,
      "step": 241060
    },
    {
      "epoch": 68.42747658245813,
      "grad_norm": 0.06607991456985474,
      "learning_rate": 3.1600340618790805e-05,
      "loss": 0.00015744995325803757,
      "step": 241070
    },
    {
      "epoch": 68.43031507238149,
      "grad_norm": 0.1369789093732834,
      "learning_rate": 3.1597502128867446e-05,
      "loss": 0.00010715480893850327,
      "step": 241080
    },
    {
      "epoch": 68.43315356230485,
      "grad_norm": 0.012049149721860886,
      "learning_rate": 3.159466363894408e-05,
      "loss": 7.893908768892288e-05,
      "step": 241090
    },
    {
      "epoch": 68.43599205222822,
      "grad_norm": 0.1613868772983551,
      "learning_rate": 3.159182514902072e-05,
      "loss": 0.0005562428385019302,
      "step": 241100
    },
    {
      "epoch": 68.43883054215158,
      "grad_norm": 0.030495062470436096,
      "learning_rate": 3.158898665909736e-05,
      "loss": 0.0003192981705069542,
      "step": 241110
    },
    {
      "epoch": 68.44166903207494,
      "grad_norm": 0.00868602842092514,
      "learning_rate": 3.1586432018166336e-05,
      "loss": 0.01341680735349655,
      "step": 241120
    },
    {
      "epoch": 68.4445075219983,
      "grad_norm": 0.023221289739012718,
      "learning_rate": 3.158359352824298e-05,
      "loss": 0.0013208821415901183,
      "step": 241130
    },
    {
      "epoch": 68.44734601192165,
      "grad_norm": 0.03161539509892464,
      "learning_rate": 3.158075503831961e-05,
      "loss": 0.00045418161898851394,
      "step": 241140
    },
    {
      "epoch": 68.45018450184502,
      "grad_norm": 0.011472048237919807,
      "learning_rate": 3.1577916548396254e-05,
      "loss": 0.0008943324908614158,
      "step": 241150
    },
    {
      "epoch": 68.45302299176838,
      "grad_norm": 3.1632308959960938,
      "learning_rate": 3.1575078058472895e-05,
      "loss": 0.0006538273766636849,
      "step": 241160
    },
    {
      "epoch": 68.45586148169174,
      "grad_norm": 4.361955642700195,
      "learning_rate": 3.157223956854953e-05,
      "loss": 0.00293334499001503,
      "step": 241170
    },
    {
      "epoch": 68.4586999716151,
      "grad_norm": 0.5901647806167603,
      "learning_rate": 3.156940107862617e-05,
      "loss": 0.00047734398394823075,
      "step": 241180
    },
    {
      "epoch": 68.46153846153847,
      "grad_norm": 0.008668467402458191,
      "learning_rate": 3.156656258870281e-05,
      "loss": 0.002595830149948597,
      "step": 241190
    },
    {
      "epoch": 68.46437695146182,
      "grad_norm": 0.017537206411361694,
      "learning_rate": 3.1563724098779454e-05,
      "loss": 8.814036846160889e-05,
      "step": 241200
    },
    {
      "epoch": 68.46721544138518,
      "grad_norm": 0.04018620774149895,
      "learning_rate": 3.156088560885609e-05,
      "loss": 0.0002083677798509598,
      "step": 241210
    },
    {
      "epoch": 68.47005393130854,
      "grad_norm": 0.16401775181293488,
      "learning_rate": 3.155804711893273e-05,
      "loss": 0.0001417912542819977,
      "step": 241220
    },
    {
      "epoch": 68.4728924212319,
      "grad_norm": 0.002853651763871312,
      "learning_rate": 3.155520862900937e-05,
      "loss": 0.00016058720648288728,
      "step": 241230
    },
    {
      "epoch": 68.47573091115527,
      "grad_norm": 0.025632109493017197,
      "learning_rate": 3.1552370139086006e-05,
      "loss": 0.005535374209284782,
      "step": 241240
    },
    {
      "epoch": 68.47856940107863,
      "grad_norm": 0.0072481646202504635,
      "learning_rate": 3.154953164916265e-05,
      "loss": 0.00026509743183851243,
      "step": 241250
    },
    {
      "epoch": 68.481407891002,
      "grad_norm": 0.020258864387869835,
      "learning_rate": 3.154669315923929e-05,
      "loss": 8.84251669049263e-05,
      "step": 241260
    },
    {
      "epoch": 68.48424638092534,
      "grad_norm": 16.19149398803711,
      "learning_rate": 3.154385466931592e-05,
      "loss": 0.002504400722682476,
      "step": 241270
    },
    {
      "epoch": 68.4870848708487,
      "grad_norm": 0.016419917345046997,
      "learning_rate": 3.1541016179392565e-05,
      "loss": 7.138215005397797e-05,
      "step": 241280
    },
    {
      "epoch": 68.48992336077207,
      "grad_norm": 0.42231255769729614,
      "learning_rate": 3.1538177689469206e-05,
      "loss": 0.0002232946455478668,
      "step": 241290
    },
    {
      "epoch": 68.49276185069543,
      "grad_norm": 0.006525796838104725,
      "learning_rate": 3.153533919954584e-05,
      "loss": 7.52357766032219e-05,
      "step": 241300
    },
    {
      "epoch": 68.4956003406188,
      "grad_norm": 0.005107279401272535,
      "learning_rate": 3.153250070962248e-05,
      "loss": 0.0049756869673728945,
      "step": 241310
    },
    {
      "epoch": 68.49843883054216,
      "grad_norm": 0.03064819984138012,
      "learning_rate": 3.152966221969912e-05,
      "loss": 0.00026240907609462736,
      "step": 241320
    },
    {
      "epoch": 68.50127732046552,
      "grad_norm": 0.022127514705061913,
      "learning_rate": 3.1526823729775765e-05,
      "loss": 0.0006256481632590294,
      "step": 241330
    },
    {
      "epoch": 68.50411581038887,
      "grad_norm": 10.596150398254395,
      "learning_rate": 3.15239852398524e-05,
      "loss": 0.0017246780917048454,
      "step": 241340
    },
    {
      "epoch": 68.50695430031223,
      "grad_norm": 0.011785885319113731,
      "learning_rate": 3.1521146749929034e-05,
      "loss": 0.0106500044465065,
      "step": 241350
    },
    {
      "epoch": 68.5097927902356,
      "grad_norm": 0.05056037753820419,
      "learning_rate": 3.151830826000568e-05,
      "loss": 0.0004159990698099136,
      "step": 241360
    },
    {
      "epoch": 68.51263128015896,
      "grad_norm": 0.017659587785601616,
      "learning_rate": 3.151546977008232e-05,
      "loss": 0.0006187362596392632,
      "step": 241370
    },
    {
      "epoch": 68.51546977008232,
      "grad_norm": 0.04636266082525253,
      "learning_rate": 3.151263128015896e-05,
      "loss": 0.012124112248420716,
      "step": 241380
    },
    {
      "epoch": 68.51830826000568,
      "grad_norm": 0.020287949591875076,
      "learning_rate": 3.15097927902356e-05,
      "loss": 0.0002757444977760315,
      "step": 241390
    },
    {
      "epoch": 68.52114674992903,
      "grad_norm": 0.009766354225575924,
      "learning_rate": 3.1506954300312234e-05,
      "loss": 0.0006340218707919121,
      "step": 241400
    },
    {
      "epoch": 68.5239852398524,
      "grad_norm": 0.0530916228890419,
      "learning_rate": 3.1504115810388875e-05,
      "loss": 0.0010753296315670013,
      "step": 241410
    },
    {
      "epoch": 68.52682372977576,
      "grad_norm": 0.022510992363095284,
      "learning_rate": 3.150127732046552e-05,
      "loss": 0.0005237562581896782,
      "step": 241420
    },
    {
      "epoch": 68.52966221969912,
      "grad_norm": 0.017935626208782196,
      "learning_rate": 3.149843883054215e-05,
      "loss": 0.0001840416342020035,
      "step": 241430
    },
    {
      "epoch": 68.53250070962248,
      "grad_norm": 0.11937280744314194,
      "learning_rate": 3.149560034061879e-05,
      "loss": 0.00029893908649683,
      "step": 241440
    },
    {
      "epoch": 68.53533919954585,
      "grad_norm": 0.04870352894067764,
      "learning_rate": 3.149276185069543e-05,
      "loss": 0.0001819547265768051,
      "step": 241450
    },
    {
      "epoch": 68.53817768946921,
      "grad_norm": 0.026257047429680824,
      "learning_rate": 3.1489923360772075e-05,
      "loss": 0.0003120535984635353,
      "step": 241460
    },
    {
      "epoch": 68.54101617939256,
      "grad_norm": 0.019711166620254517,
      "learning_rate": 3.148708487084871e-05,
      "loss": 0.0002435309812426567,
      "step": 241470
    },
    {
      "epoch": 68.54385466931592,
      "grad_norm": 0.1013936772942543,
      "learning_rate": 3.1484246380925345e-05,
      "loss": 0.0008049558848142624,
      "step": 241480
    },
    {
      "epoch": 68.54669315923928,
      "grad_norm": 0.11699660867452621,
      "learning_rate": 3.148140789100199e-05,
      "loss": 0.0013045551255345344,
      "step": 241490
    },
    {
      "epoch": 68.54953164916265,
      "grad_norm": 0.5633865594863892,
      "learning_rate": 3.147856940107863e-05,
      "loss": 0.0005780719220638275,
      "step": 241500
    },
    {
      "epoch": 68.54953164916265,
      "eval_accuracy": 0.9786990525847269,
      "eval_loss": 0.08171876519918442,
      "eval_runtime": 33.3069,
      "eval_samples_per_second": 472.184,
      "eval_steps_per_second": 7.386,
      "step": 241500
    },
    {
      "epoch": 68.55237013908601,
      "grad_norm": 0.07555211335420609,
      "learning_rate": 3.147573091115527e-05,
      "loss": 0.0015536515042185784,
      "step": 241510
    },
    {
      "epoch": 68.55520862900937,
      "grad_norm": 0.06259137392044067,
      "learning_rate": 3.147289242123191e-05,
      "loss": 0.003832193836569786,
      "step": 241520
    },
    {
      "epoch": 68.55804711893273,
      "grad_norm": 0.007841531187295914,
      "learning_rate": 3.1470053931308545e-05,
      "loss": 0.0008786646649241447,
      "step": 241530
    },
    {
      "epoch": 68.56088560885608,
      "grad_norm": 0.04828483611345291,
      "learning_rate": 3.1467215441385186e-05,
      "loss": 0.00023389942944049836,
      "step": 241540
    },
    {
      "epoch": 68.56372409877945,
      "grad_norm": 0.02786937728524208,
      "learning_rate": 3.146437695146182e-05,
      "loss": 0.0003346838057041168,
      "step": 241550
    },
    {
      "epoch": 68.56656258870281,
      "grad_norm": 1.5541677474975586,
      "learning_rate": 3.146153846153846e-05,
      "loss": 0.0006526540964841843,
      "step": 241560
    },
    {
      "epoch": 68.56940107862617,
      "grad_norm": 0.018594902008771896,
      "learning_rate": 3.1458699971615103e-05,
      "loss": 0.000270833820104599,
      "step": 241570
    },
    {
      "epoch": 68.57223956854953,
      "grad_norm": 0.020332293584942818,
      "learning_rate": 3.145586148169174e-05,
      "loss": 0.00022978167980909349,
      "step": 241580
    },
    {
      "epoch": 68.5750780584729,
      "grad_norm": 0.008709891699254513,
      "learning_rate": 3.145302299176838e-05,
      "loss": 0.000669936090707779,
      "step": 241590
    },
    {
      "epoch": 68.57791654839626,
      "grad_norm": 0.20084045827388763,
      "learning_rate": 3.145018450184502e-05,
      "loss": 0.01737380027770996,
      "step": 241600
    },
    {
      "epoch": 68.58075503831961,
      "grad_norm": 0.003968459088355303,
      "learning_rate": 3.1447346011921655e-05,
      "loss": 0.00020779892802238464,
      "step": 241610
    },
    {
      "epoch": 68.58359352824297,
      "grad_norm": 0.010130992159247398,
      "learning_rate": 3.1444507521998304e-05,
      "loss": 0.0008029919117689133,
      "step": 241620
    },
    {
      "epoch": 68.58643201816633,
      "grad_norm": 0.008957494050264359,
      "learning_rate": 3.144166903207494e-05,
      "loss": 0.00015440434217453002,
      "step": 241630
    },
    {
      "epoch": 68.5892705080897,
      "grad_norm": 0.032504402101039886,
      "learning_rate": 3.143883054215157e-05,
      "loss": 6.893835961818695e-05,
      "step": 241640
    },
    {
      "epoch": 68.59210899801306,
      "grad_norm": 0.04427509382367134,
      "learning_rate": 3.1435992052228214e-05,
      "loss": 0.00017499253153800963,
      "step": 241650
    },
    {
      "epoch": 68.59494748793642,
      "grad_norm": 0.16161316633224487,
      "learning_rate": 3.1433153562304856e-05,
      "loss": 0.00047724880278110504,
      "step": 241660
    },
    {
      "epoch": 68.59778597785977,
      "grad_norm": 0.03222566843032837,
      "learning_rate": 3.14303150723815e-05,
      "loss": 0.0006421253085136414,
      "step": 241670
    },
    {
      "epoch": 68.60062446778313,
      "grad_norm": 0.007077581249177456,
      "learning_rate": 3.142747658245813e-05,
      "loss": 0.001435224898159504,
      "step": 241680
    },
    {
      "epoch": 68.6034629577065,
      "grad_norm": 0.010933981277048588,
      "learning_rate": 3.142463809253477e-05,
      "loss": 0.00019157901406288146,
      "step": 241690
    },
    {
      "epoch": 68.60630144762986,
      "grad_norm": 0.44706401228904724,
      "learning_rate": 3.1421799602611414e-05,
      "loss": 0.00028158985078334807,
      "step": 241700
    },
    {
      "epoch": 68.60913993755322,
      "grad_norm": 0.03889862820506096,
      "learning_rate": 3.141896111268805e-05,
      "loss": 0.001148807629942894,
      "step": 241710
    },
    {
      "epoch": 68.61197842747659,
      "grad_norm": 0.041643064469099045,
      "learning_rate": 3.141612262276469e-05,
      "loss": 0.0002703838050365448,
      "step": 241720
    },
    {
      "epoch": 68.61481691739995,
      "grad_norm": 0.018283188343048096,
      "learning_rate": 3.141328413284133e-05,
      "loss": 0.0004537126049399376,
      "step": 241730
    },
    {
      "epoch": 68.6176554073233,
      "grad_norm": 12.459197998046875,
      "learning_rate": 3.1410445642917966e-05,
      "loss": 0.004809238761663437,
      "step": 241740
    },
    {
      "epoch": 68.62049389724666,
      "grad_norm": 0.35728105902671814,
      "learning_rate": 3.140760715299461e-05,
      "loss": 0.00019841250032186507,
      "step": 241750
    },
    {
      "epoch": 68.62333238717002,
      "grad_norm": 0.018190165981650352,
      "learning_rate": 3.140476866307125e-05,
      "loss": 0.00012420006096363067,
      "step": 241760
    },
    {
      "epoch": 68.62617087709339,
      "grad_norm": 0.044130682945251465,
      "learning_rate": 3.1401930173147884e-05,
      "loss": 0.00015591122210025787,
      "step": 241770
    },
    {
      "epoch": 68.62900936701675,
      "grad_norm": 0.03572811931371689,
      "learning_rate": 3.1399091683224525e-05,
      "loss": 0.00026751793920993806,
      "step": 241780
    },
    {
      "epoch": 68.63184785694011,
      "grad_norm": 0.016564469784498215,
      "learning_rate": 3.1396253193301166e-05,
      "loss": 0.0001751726493239403,
      "step": 241790
    },
    {
      "epoch": 68.63468634686348,
      "grad_norm": 0.005307013634592295,
      "learning_rate": 3.139341470337781e-05,
      "loss": 0.00013709627091884613,
      "step": 241800
    },
    {
      "epoch": 68.63752483678682,
      "grad_norm": 0.03194558247923851,
      "learning_rate": 3.139057621345444e-05,
      "loss": 0.00025252960622310636,
      "step": 241810
    },
    {
      "epoch": 68.64036332671019,
      "grad_norm": 0.05026521161198616,
      "learning_rate": 3.1387737723531084e-05,
      "loss": 0.0004874167963862419,
      "step": 241820
    },
    {
      "epoch": 68.64320181663355,
      "grad_norm": 0.07109824568033218,
      "learning_rate": 3.1384899233607725e-05,
      "loss": 0.0002678379416465759,
      "step": 241830
    },
    {
      "epoch": 68.64604030655691,
      "grad_norm": 0.18799768388271332,
      "learning_rate": 3.138206074368436e-05,
      "loss": 0.0005527874454855919,
      "step": 241840
    },
    {
      "epoch": 68.64887879648028,
      "grad_norm": 0.013537528924643993,
      "learning_rate": 3.1379222253761e-05,
      "loss": 9.795837104320526e-05,
      "step": 241850
    },
    {
      "epoch": 68.65171728640364,
      "grad_norm": 0.025443900376558304,
      "learning_rate": 3.137638376383764e-05,
      "loss": 0.00013652704656124114,
      "step": 241860
    },
    {
      "epoch": 68.65455577632699,
      "grad_norm": 0.03835247829556465,
      "learning_rate": 3.137354527391428e-05,
      "loss": 0.0003083115443587303,
      "step": 241870
    },
    {
      "epoch": 68.65739426625035,
      "grad_norm": 0.007184120826423168,
      "learning_rate": 3.137070678399092e-05,
      "loss": 0.0001445794478058815,
      "step": 241880
    },
    {
      "epoch": 68.66023275617371,
      "grad_norm": 1.0054808855056763,
      "learning_rate": 3.136786829406756e-05,
      "loss": 0.00035395976155996325,
      "step": 241890
    },
    {
      "epoch": 68.66307124609708,
      "grad_norm": 0.18899841606616974,
      "learning_rate": 3.1365029804144194e-05,
      "loss": 0.0001590600237250328,
      "step": 241900
    },
    {
      "epoch": 68.66590973602044,
      "grad_norm": 0.011053543537855148,
      "learning_rate": 3.1362191314220836e-05,
      "loss": 0.002397079765796661,
      "step": 241910
    },
    {
      "epoch": 68.6687482259438,
      "grad_norm": 0.4623057544231415,
      "learning_rate": 3.135963667328981e-05,
      "loss": 0.004659312963485718,
      "step": 241920
    },
    {
      "epoch": 68.67158671586716,
      "grad_norm": 0.031379587948322296,
      "learning_rate": 3.135679818336645e-05,
      "loss": 0.0005584681406617164,
      "step": 241930
    },
    {
      "epoch": 68.67442520579051,
      "grad_norm": 0.3290727138519287,
      "learning_rate": 3.135395969344309e-05,
      "loss": 0.0001809835433959961,
      "step": 241940
    },
    {
      "epoch": 68.67726369571388,
      "grad_norm": 0.8122276663780212,
      "learning_rate": 3.1351121203519726e-05,
      "loss": 0.00046157632023096084,
      "step": 241950
    },
    {
      "epoch": 68.68010218563724,
      "grad_norm": 0.017355209216475487,
      "learning_rate": 3.134828271359637e-05,
      "loss": 0.0004881162196397781,
      "step": 241960
    },
    {
      "epoch": 68.6829406755606,
      "grad_norm": 1.5341253280639648,
      "learning_rate": 3.134544422367301e-05,
      "loss": 0.0007528502494096756,
      "step": 241970
    },
    {
      "epoch": 68.68577916548396,
      "grad_norm": 0.13411690294742584,
      "learning_rate": 3.134260573374965e-05,
      "loss": 0.0002561355009675026,
      "step": 241980
    },
    {
      "epoch": 68.68861765540733,
      "grad_norm": 0.15526454150676727,
      "learning_rate": 3.1339767243826285e-05,
      "loss": 0.0005258312448859215,
      "step": 241990
    },
    {
      "epoch": 68.69145614533069,
      "grad_norm": 0.042030785232782364,
      "learning_rate": 3.1336928753902926e-05,
      "loss": 0.0010858336463570595,
      "step": 242000
    },
    {
      "epoch": 68.69145614533069,
      "eval_accuracy": 0.9779360335728365,
      "eval_loss": 0.08356881886720657,
      "eval_runtime": 33.3327,
      "eval_samples_per_second": 471.819,
      "eval_steps_per_second": 7.38,
      "step": 242000
    }
  ],
  "logging_steps": 10,
  "max_steps": 352300,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 100,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6285092873819138e+21,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}