diff --git "a/last_to_push_frequency_5039/checkpoint-40000/trainer_state.json" "b/last_to_push_frequency_5039/checkpoint-40000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last_to_push_frequency_5039/checkpoint-40000/trainer_state.json"
@@ -0,0 +1,6003 @@
+{
+  "best_global_step": 40000,
+  "best_metric": 3.5581958293914795,
+  "best_model_checkpoint": "/scratch/cl5625/exceptions/models/last_to_push_frequency_5039/checkpoint-40000",
+  "epoch": 11.648261401362921,
+  "eval_steps": 1000,
+  "global_step": 40000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.014561127613722406,
+      "grad_norm": 0.7142499685287476,
+      "learning_rate": 0.000294,
+      "loss": 8.4587,
+      "step": 50
+    },
+    {
+      "epoch": 0.029122255227444813,
+      "grad_norm": 0.5678296685218811,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 6.7387,
+      "step": 100
+    },
+    {
+      "epoch": 0.04368338284116722,
+      "grad_norm": 0.5606561303138733,
+      "learning_rate": 0.0005998286713286713,
+      "loss": 6.3655,
+      "step": 150
+    },
+    {
+      "epoch": 0.058244510454889625,
+      "grad_norm": 0.47408413887023926,
+      "learning_rate": 0.0005996538461538461,
+      "loss": 6.1262,
+      "step": 200
+    },
+    {
+      "epoch": 0.07280563806861204,
+      "grad_norm": 0.4732152223587036,
+      "learning_rate": 0.0005994790209790209,
+      "loss": 5.9972,
+      "step": 250
+    },
+    {
+      "epoch": 0.08736676568233444,
+      "grad_norm": 0.438856303691864,
+      "learning_rate": 0.0005993041958041958,
+      "loss": 5.8873,
+      "step": 300
+    },
+    {
+      "epoch": 0.10192789329605685,
+      "grad_norm": 0.4995385408401489,
+      "learning_rate": 0.0005991293706293705,
+      "loss": 5.7343,
+      "step": 350
+    },
+    {
+      "epoch": 0.11648902090977925,
+      "grad_norm": 0.4228087067604065,
+      "learning_rate": 0.0005989545454545454,
+      "loss": 5.63,
+      "step": 400
+    },
+    {
+      "epoch": 0.13105014852350166,
+      "grad_norm": 0.4598565399646759,
+      "learning_rate": 0.0005987797202797202,
+      "loss": 5.5141,
+      "step": 450
+    },
+    {
+      "epoch": 0.14561127613722408,
+      "grad_norm": 0.47586789727211,
+      "learning_rate": 0.000598604895104895,
+      "loss": 5.414,
+      "step": 500
+    },
+    {
+      "epoch": 0.16017240375094646,
+      "grad_norm": 0.477698415517807,
+      "learning_rate": 0.0005984300699300698,
+      "loss": 5.3322,
+      "step": 550
+    },
+    {
+      "epoch": 0.17473353136466888,
+      "grad_norm": 0.44980600476264954,
+      "learning_rate": 0.0005982552447552447,
+      "loss": 5.2471,
+      "step": 600
+    },
+    {
+      "epoch": 0.1892946589783913,
+      "grad_norm": 0.4625977873802185,
+      "learning_rate": 0.0005980804195804195,
+      "loss": 5.1807,
+      "step": 650
+    },
+    {
+      "epoch": 0.2038557865921137,
+      "grad_norm": 0.5372968912124634,
+      "learning_rate": 0.0005979055944055943,
+      "loss": 5.1209,
+      "step": 700
+    },
+    {
+      "epoch": 0.2184169142058361,
+      "grad_norm": 0.5401430726051331,
+      "learning_rate": 0.0005977307692307691,
+      "loss": 5.0708,
+      "step": 750
+    },
+    {
+      "epoch": 0.2329780418195585,
+      "grad_norm": 0.4346584379673004,
+      "learning_rate": 0.000597555944055944,
+      "loss": 5.0076,
+      "step": 800
+    },
+    {
+      "epoch": 0.24753916943328091,
+      "grad_norm": 0.460715115070343,
+      "learning_rate": 0.0005973811188811188,
+      "loss": 4.9693,
+      "step": 850
+    },
+    {
+      "epoch": 0.2621002970470033,
+      "grad_norm": 0.4268857538700104,
+      "learning_rate": 0.0005972062937062936,
+      "loss": 4.9309,
+      "step": 900
+    },
+    {
+      "epoch": 0.27666142466072574,
+      "grad_norm": 0.44858410954475403,
+      "learning_rate": 0.0005970314685314685,
+      "loss": 4.8723,
+      "step": 950
+    },
+    {
+      "epoch": 0.29122255227444815,
+      "grad_norm": 0.4247923493385315,
+      "learning_rate": 0.0005968566433566433,
+      "loss": 4.8399,
+      "step": 1000
+    },
+    {
+      "epoch": 0.29122255227444815,
+      "eval_accuracy": 0.2545978147932791,
+      "eval_loss": 4.757110118865967,
+      "eval_runtime": 177.2928,
+      "eval_samples_per_second": 93.89,
+      "eval_steps_per_second": 5.872,
+      "step": 1000
+    },
+    {
+      "epoch": 0.30578367988817057,
+      "grad_norm": 0.47108718752861023,
+      "learning_rate": 0.0005966818181818181,
+      "loss": 4.7695,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3203448075018929,
+      "grad_norm": 0.4697106182575226,
+      "learning_rate": 0.0005965069930069929,
+      "loss": 4.7315,
+      "step": 1100
+    },
+    {
+      "epoch": 0.33490593511561534,
+      "grad_norm": 0.4338913559913635,
+      "learning_rate": 0.0005963321678321677,
+      "loss": 4.7008,
+      "step": 1150
+    },
+    {
+      "epoch": 0.34946706272933775,
+      "grad_norm": 0.47174957394599915,
+      "learning_rate": 0.0005961573426573425,
+      "loss": 4.6608,
+      "step": 1200
+    },
+    {
+      "epoch": 0.36402819034306017,
+      "grad_norm": 0.4433532953262329,
+      "learning_rate": 0.0005959825174825174,
+      "loss": 4.632,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3785893179567826,
+      "grad_norm": 0.5037450194358826,
+      "learning_rate": 0.0005958076923076922,
+      "loss": 4.5994,
+      "step": 1300
+    },
+    {
+      "epoch": 0.393150445570505,
+      "grad_norm": 0.4283815622329712,
+      "learning_rate": 0.000595632867132867,
+      "loss": 4.567,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4077115731842274,
+      "grad_norm": 0.39518365263938904,
+      "learning_rate": 0.0005954580419580418,
+      "loss": 4.556,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4222727007979498,
+      "grad_norm": 0.435833215713501,
+      "learning_rate": 0.0005952832167832168,
+      "loss": 4.5333,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4368338284116722,
+      "grad_norm": 0.4726100265979767,
+      "learning_rate": 0.0005951083916083916,
+      "loss": 4.4928,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4513949560253946,
+      "grad_norm": 0.44213613867759705,
+      "learning_rate": 0.0005949335664335664,
+      "loss": 4.4892,
+      "step": 1550
+    },
+    {
+      "epoch": 0.465956083639117,
+      "grad_norm": 0.39726656675338745,
+      "learning_rate": 0.0005947587412587413,
+      "loss": 4.4738,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4805172112528394,
+      "grad_norm": 0.4227396547794342,
+      "learning_rate": 0.0005945839160839161,
+      "loss": 4.4391,
+      "step": 1650
+    },
+    {
+      "epoch": 0.49507833886656183,
+      "grad_norm": 0.4156210124492645,
+      "learning_rate": 0.0005944090909090909,
+      "loss": 4.4382,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5096394664802842,
+      "grad_norm": 0.4048631489276886,
+      "learning_rate": 0.0005942342657342657,
+      "loss": 4.4035,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5242005940940067,
+      "grad_norm": 0.3983406722545624,
+      "learning_rate": 0.0005940594405594406,
+      "loss": 4.3835,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5387617217077291,
+      "grad_norm": 0.4011361002922058,
+      "learning_rate": 0.0005938846153846153,
+      "loss": 4.3834,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5533228493214515,
+      "grad_norm": 0.4169325828552246,
+      "learning_rate": 0.0005937097902097902,
+      "loss": 4.3576,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5678839769351739,
+      "grad_norm": 0.4109379053115845,
+      "learning_rate": 0.000593534965034965,
+      "loss": 4.3456,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5824451045488963,
+      "grad_norm": 0.4288181960582733,
+      "learning_rate": 0.0005933601398601398,
+      "loss": 4.3272,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5824451045488963,
+      "eval_accuracy": 0.3003302568766484,
+      "eval_loss": 4.275340557098389,
+      "eval_runtime": 177.3788,
+      "eval_samples_per_second": 93.844,
+      "eval_steps_per_second": 5.869,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5970062321626187,
+      "grad_norm": 0.3974442183971405,
+      "learning_rate": 0.0005931853146853146,
+      "loss": 4.3343,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6115673597763411,
+      "grad_norm": 0.39755979180336,
+      "learning_rate": 0.0005930104895104895,
+      "loss": 4.3154,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6261284873900634,
+      "grad_norm": 0.3861689865589142,
+      "learning_rate": 0.0005928356643356643,
+      "loss": 4.2906,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6406896150037859,
+      "grad_norm": 0.3685368597507477,
+      "learning_rate": 0.0005926608391608391,
+      "loss": 4.2912,
+      "step": 2200
+    },
+    {
+      "epoch": 0.6552507426175083,
+      "grad_norm": 0.4314819574356079,
+      "learning_rate": 0.000592486013986014,
+      "loss": 4.2757,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6698118702312307,
+      "grad_norm": 0.38772013783454895,
+      "learning_rate": 0.0005923111888111888,
+      "loss": 4.2653,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6843729978449531,
+      "grad_norm": 0.36975449323654175,
+      "learning_rate": 0.0005921363636363636,
+      "loss": 4.2463,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6989341254586755,
+      "grad_norm": 0.374622642993927,
+      "learning_rate": 0.0005919615384615384,
+      "loss": 4.2391,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7134952530723979,
+      "grad_norm": 0.3807830214500427,
+      "learning_rate": 0.0005917867132867133,
+      "loss": 4.2362,
+      "step": 2450
+    },
+    {
+      "epoch": 0.7280563806861203,
+      "grad_norm": 0.35665392875671387,
+      "learning_rate": 0.0005916118881118881,
+      "loss": 4.2275,
+      "step": 2500
+    },
+    {
+      "epoch": 0.7426175082998427,
+      "grad_norm": 0.39687132835388184,
+      "learning_rate": 0.0005914370629370629,
+      "loss": 4.2086,
+      "step": 2550
+    },
+    {
+      "epoch": 0.7571786359135652,
+      "grad_norm": 0.3794260025024414,
+      "learning_rate": 0.0005912622377622377,
+      "loss": 4.2127,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7717397635272876,
+      "grad_norm": 0.40508922934532166,
+      "learning_rate": 0.0005910874125874125,
+      "loss": 4.2015,
+      "step": 2650
+    },
+    {
+      "epoch": 0.78630089114101,
+      "grad_norm": 0.3626640737056732,
+      "learning_rate": 0.0005909125874125873,
+      "loss": 4.187,
+      "step": 2700
+    },
+    {
+      "epoch": 0.8008620187547324,
+      "grad_norm": 0.36335116624832153,
+      "learning_rate": 0.0005907377622377622,
+      "loss": 4.173,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8154231463684548,
+      "grad_norm": 0.38054153323173523,
+      "learning_rate": 0.000590562937062937,
+      "loss": 4.1755,
+      "step": 2800
+    },
+    {
+      "epoch": 0.8299842739821772,
+      "grad_norm": 0.3813888132572174,
+      "learning_rate": 0.0005903881118881118,
+      "loss": 4.171,
+      "step": 2850
+    },
+    {
+      "epoch": 0.8445454015958996,
+      "grad_norm": 0.393034428358078,
+      "learning_rate": 0.0005902132867132867,
+      "loss": 4.154,
+      "step": 2900
+    },
+    {
+      "epoch": 0.8591065292096219,
+      "grad_norm": 0.36573460698127747,
+      "learning_rate": 0.0005900384615384615,
+      "loss": 4.1445,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8736676568233444,
+      "grad_norm": 0.36031314730644226,
+      "learning_rate": 0.0005898636363636363,
+      "loss": 4.1337,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8736676568233444,
+      "eval_accuracy": 0.3159021295996076,
+      "eval_loss": 4.092033386230469,
+      "eval_runtime": 177.2002,
+      "eval_samples_per_second": 93.939,
+      "eval_steps_per_second": 5.875,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8882287844370668,
+      "grad_norm": 0.33011677861213684,
+      "learning_rate": 0.0005896888111888111,
+      "loss": 4.1267,
+      "step": 3050
+    },
+    {
+      "epoch": 0.9027899120507892,
+      "grad_norm": 0.3441181480884552,
+      "learning_rate": 0.000589513986013986,
+      "loss": 4.1295,
+      "step": 3100
+    },
+    {
+      "epoch": 0.9173510396645116,
+      "grad_norm": 0.3503768742084503,
+      "learning_rate": 0.0005893391608391608,
+      "loss": 4.1174,
+      "step": 3150
+    },
+    {
+      "epoch": 0.931912167278234,
+      "grad_norm": 0.36162590980529785,
+      "learning_rate": 0.0005891643356643356,
+      "loss": 4.1218,
+      "step": 3200
+    },
+    {
+      "epoch": 0.9464732948919564,
+      "grad_norm": 0.3388279974460602,
+      "learning_rate": 0.0005889895104895104,
+      "loss": 4.1123,
+      "step": 3250
+    },
+    {
+      "epoch": 0.9610344225056788,
+      "grad_norm": 0.36532530188560486,
+      "learning_rate": 0.0005888146853146853,
+      "loss": 4.1076,
+      "step": 3300
+    },
+    {
+      "epoch": 0.9755955501194012,
+      "grad_norm": 0.42103567719459534,
+      "learning_rate": 0.00058863986013986,
+      "loss": 4.0885,
+      "step": 3350
+    },
+    {
+      "epoch": 0.9901566777331237,
+      "grad_norm": 0.354731947183609,
+      "learning_rate": 0.0005884650349650349,
+      "loss": 4.0847,
+      "step": 3400
+    },
+    {
+      "epoch": 1.004659560836391,
+      "grad_norm": 0.34602439403533936,
+      "learning_rate": 0.0005882902097902097,
+      "loss": 4.0677,
+      "step": 3450
+    },
+    {
+      "epoch": 1.0192206884501136,
+      "grad_norm": 0.3426474332809448,
+      "learning_rate": 0.0005881153846153845,
+      "loss": 4.0154,
+      "step": 3500
+    },
+    {
+      "epoch": 1.033781816063836,
+      "grad_norm": 0.34196737408638,
+      "learning_rate": 0.0005879405594405594,
+      "loss": 4.0117,
+      "step": 3550
+    },
+    {
+      "epoch": 1.0483429436775584,
+      "grad_norm": 0.35845261812210083,
+      "learning_rate": 0.0005877657342657342,
+      "loss": 4.0123,
+      "step": 3600
+    },
+    {
+      "epoch": 1.0629040712912807,
+      "grad_norm": 0.3470519185066223,
+      "learning_rate": 0.000587590909090909,
+      "loss": 4.0003,
+      "step": 3650
+    },
+    {
+      "epoch": 1.0774651989050033,
+      "grad_norm": 0.3520037531852722,
+      "learning_rate": 0.0005874160839160838,
+      "loss": 3.989,
+      "step": 3700
+    },
+    {
+      "epoch": 1.0920263265187256,
+      "grad_norm": 0.3449414074420929,
+      "learning_rate": 0.0005872412587412587,
+      "loss": 4.0093,
+      "step": 3750
+    },
+    {
+      "epoch": 1.106587454132448,
+      "grad_norm": 0.3322015106678009,
+      "learning_rate": 0.0005870664335664335,
+      "loss": 4.0085,
+      "step": 3800
+    },
+    {
+      "epoch": 1.1211485817461704,
+      "grad_norm": 0.35296934843063354,
+      "learning_rate": 0.0005868916083916083,
+      "loss": 3.9958,
+      "step": 3850
+    },
+    {
+      "epoch": 1.135709709359893,
+      "grad_norm": 0.3486291766166687,
+      "learning_rate": 0.0005867167832167831,
+      "loss": 4.0075,
+      "step": 3900
+    },
+    {
+      "epoch": 1.1502708369736152,
+      "grad_norm": 0.34043097496032715,
+      "learning_rate": 0.000586541958041958,
+      "loss": 3.9973,
+      "step": 3950
+    },
+    {
+      "epoch": 1.1648319645873377,
+      "grad_norm": 0.33894768357276917,
+      "learning_rate": 0.0005863671328671328,
+      "loss": 3.9915,
+      "step": 4000
+    },
+    {
+      "epoch": 1.1648319645873377,
+      "eval_accuracy": 0.32561280096909206,
+      "eval_loss": 3.9853665828704834,
+      "eval_runtime": 177.196,
+      "eval_samples_per_second": 93.941,
+      "eval_steps_per_second": 5.875,
+      "step": 4000
+    },
+    {
+      "epoch": 1.17939309220106,
+      "grad_norm": 0.3345903754234314,
+      "learning_rate": 0.0005861923076923076,
+      "loss": 3.9927,
+      "step": 4050
+    },
+    {
+      "epoch": 1.1939542198147826,
+      "grad_norm": 0.3380124270915985,
+      "learning_rate": 0.0005860174825174824,
+      "loss": 3.9764,
+      "step": 4100
+    },
+    {
+      "epoch": 1.2085153474285049,
+      "grad_norm": 0.3687117099761963,
+      "learning_rate": 0.0005858426573426573,
+      "loss": 3.985,
+      "step": 4150
+    },
+    {
+      "epoch": 1.2230764750422272,
+      "grad_norm": 0.3383759558200836,
+      "learning_rate": 0.000585667832167832,
+      "loss": 3.967,
+      "step": 4200
+    },
+    {
+      "epoch": 1.2376376026559497,
+      "grad_norm": 0.3614034354686737,
+      "learning_rate": 0.000585493006993007,
+      "loss": 3.9671,
+      "step": 4250
+    },
+    {
+      "epoch": 1.2521987302696722,
+      "grad_norm": 0.3286982476711273,
+      "learning_rate": 0.0005853181818181817,
+      "loss": 3.9622,
+      "step": 4300
+    },
+    {
+      "epoch": 1.2667598578833945,
+      "grad_norm": 0.34898343682289124,
+      "learning_rate": 0.0005851433566433565,
+      "loss": 3.9527,
+      "step": 4350
+    },
+    {
+      "epoch": 1.2813209854971168,
+      "grad_norm": 0.3299115300178528,
+      "learning_rate": 0.0005849685314685315,
+      "loss": 3.9661,
+      "step": 4400
+    },
+    {
+      "epoch": 1.2958821131108393,
+      "grad_norm": 0.3476330637931824,
+      "learning_rate": 0.0005847937062937063,
+      "loss": 3.9757,
+      "step": 4450
+    },
+    {
+      "epoch": 1.3104432407245616,
+      "grad_norm": 0.3423220217227936,
+      "learning_rate": 0.0005846188811188811,
+      "loss": 3.9664,
+      "step": 4500
+    },
+    {
+      "epoch": 1.3250043683382842,
+      "grad_norm": 0.36620113253593445,
+      "learning_rate": 0.0005844440559440559,
+      "loss": 3.9626,
+      "step": 4550
+    },
+    {
+      "epoch": 1.3395654959520065,
+      "grad_norm": 0.34613823890686035,
+      "learning_rate": 0.0005842692307692308,
+      "loss": 3.9469,
+      "step": 4600
+    },
+    {
+      "epoch": 1.354126623565729,
+      "grad_norm": 0.3371380865573883,
+      "learning_rate": 0.0005840944055944056,
+      "loss": 3.9431,
+      "step": 4650
+    },
+    {
+      "epoch": 1.3686877511794513,
+      "grad_norm": 0.3557433485984802,
+      "learning_rate": 0.0005839195804195804,
+      "loss": 3.9371,
+      "step": 4700
+    },
+    {
+      "epoch": 1.3832488787931738,
+      "grad_norm": 0.34661224484443665,
+      "learning_rate": 0.0005837447552447552,
+      "loss": 3.9475,
+      "step": 4750
+    },
+    {
+      "epoch": 1.3978100064068961,
+      "grad_norm": 0.3363933563232422,
+      "learning_rate": 0.0005835699300699301,
+      "loss": 3.9525,
+      "step": 4800
+    },
+    {
+      "epoch": 1.4123711340206184,
+      "grad_norm": 0.3473567068576813,
+      "learning_rate": 0.0005833951048951048,
+      "loss": 3.9478,
+      "step": 4850
+    },
+    {
+      "epoch": 1.426932261634341,
+      "grad_norm": 0.35772213339805603,
+      "learning_rate": 0.0005832202797202797,
+      "loss": 3.9494,
+      "step": 4900
+    },
+    {
+      "epoch": 1.4414933892480635,
+      "grad_norm": 0.3194500803947449,
+      "learning_rate": 0.0005830454545454546,
+      "loss": 3.9274,
+      "step": 4950
+    },
+    {
+      "epoch": 1.4560545168617858,
+      "grad_norm": 0.3184325397014618,
+      "learning_rate": 0.0005828706293706293,
+      "loss": 3.91,
+      "step": 5000
+    },
+    {
+      "epoch": 1.4560545168617858,
+      "eval_accuracy": 0.3322197019411703,
+      "eval_loss": 3.9121792316436768,
+      "eval_runtime": 177.1541,
+      "eval_samples_per_second": 93.963,
+      "eval_steps_per_second": 5.876,
+      "step": 5000
+    },
+    {
+      "epoch": 1.470615644475508,
+      "grad_norm": 0.334377646446228,
+      "learning_rate": 0.0005826958041958042,
+      "loss": 3.9292,
+      "step": 5050
+    },
+    {
+      "epoch": 1.4851767720892306,
+      "grad_norm": 0.3505079746246338,
+      "learning_rate": 0.000582520979020979,
+      "loss": 3.9197,
+      "step": 5100
+    },
+    {
+      "epoch": 1.4997378997029531,
+      "grad_norm": 0.3423149585723877,
+      "learning_rate": 0.0005823461538461538,
+      "loss": 3.9216,
+      "step": 5150
+    },
+    {
+      "epoch": 1.5142990273166754,
+      "grad_norm": 0.3242221772670746,
+      "learning_rate": 0.0005821713286713286,
+      "loss": 3.9264,
+      "step": 5200
+    },
+    {
+      "epoch": 1.5288601549303977,
+      "grad_norm": 0.3360047936439514,
+      "learning_rate": 0.0005819965034965035,
+      "loss": 3.9075,
+      "step": 5250
+    },
+    {
+      "epoch": 1.5434212825441203,
+      "grad_norm": 0.3281499445438385,
+      "learning_rate": 0.0005818216783216783,
+      "loss": 3.9156,
+      "step": 5300
+    },
+    {
+      "epoch": 1.5579824101578428,
+      "grad_norm": 0.3143256604671478,
+      "learning_rate": 0.0005816468531468531,
+      "loss": 3.9054,
+      "step": 5350
+    },
+    {
+      "epoch": 1.572543537771565,
+      "grad_norm": 0.34801602363586426,
+      "learning_rate": 0.0005814720279720279,
+      "loss": 3.9052,
+      "step": 5400
+    },
+    {
+      "epoch": 1.5871046653852874,
+      "grad_norm": 0.33646127581596375,
+      "learning_rate": 0.0005812972027972028,
+      "loss": 3.9049,
+      "step": 5450
+    },
+    {
+      "epoch": 1.6016657929990097,
+      "grad_norm": 0.34947100281715393,
+      "learning_rate": 0.0005811223776223776,
+      "loss": 3.9141,
+      "step": 5500
+    },
+    {
+      "epoch": 1.6162269206127322,
+      "grad_norm": 0.3342869281768799,
+      "learning_rate": 0.0005809475524475524,
+      "loss": 3.8802,
+      "step": 5550
+    },
+    {
+      "epoch": 1.6307880482264547,
+      "grad_norm": 0.3248964250087738,
+      "learning_rate": 0.0005807727272727272,
+      "loss": 3.9011,
+      "step": 5600
+    },
+    {
+      "epoch": 1.645349175840177,
+      "grad_norm": 0.3466763496398926,
+      "learning_rate": 0.0005805979020979021,
+      "loss": 3.8889,
+      "step": 5650
+    },
+    {
+      "epoch": 1.6599103034538993,
+      "grad_norm": 0.34471645951271057,
+      "learning_rate": 0.0005804230769230769,
+      "loss": 3.8915,
+      "step": 5700
+    },
+    {
+      "epoch": 1.6744714310676219,
+      "grad_norm": 0.3128034770488739,
+      "learning_rate": 0.0005802482517482517,
+      "loss": 3.8916,
+      "step": 5750
+    },
+    {
+      "epoch": 1.6890325586813444,
+      "grad_norm": 0.3492928445339203,
+      "learning_rate": 0.0005800734265734265,
+      "loss": 3.8693,
+      "step": 5800
+    },
+    {
+      "epoch": 1.7035936862950667,
+      "grad_norm": 0.3331936001777649,
+      "learning_rate": 0.0005798986013986013,
+      "loss": 3.8772,
+      "step": 5850
+    },
+    {
+      "epoch": 1.718154813908789,
+      "grad_norm": 0.3292458951473236,
+      "learning_rate": 0.0005797237762237762,
+      "loss": 3.874,
+      "step": 5900
+    },
+    {
+      "epoch": 1.7327159415225115,
+      "grad_norm": 0.3334185779094696,
+      "learning_rate": 0.000579548951048951,
+      "loss": 3.878,
+      "step": 5950
+    },
+    {
+      "epoch": 1.747277069136234,
+      "grad_norm": 0.3228028118610382,
+      "learning_rate": 0.0005793741258741258,
+      "loss": 3.8822,
+      "step": 6000
+    },
+    {
+      "epoch": 1.747277069136234,
+      "eval_accuracy": 0.33703835809241034,
+      "eval_loss": 3.8543479442596436,
+      "eval_runtime": 177.1903,
+      "eval_samples_per_second": 93.944,
+      "eval_steps_per_second": 5.875,
+      "step": 6000
+    },
+    {
+      "epoch": 1.7618381967499563,
+      "grad_norm": 0.35547763109207153,
+      "learning_rate": 0.0005791993006993006,
+      "loss": 3.8844,
+      "step": 6050
+    },
+    {
+      "epoch": 1.7763993243636786,
+      "grad_norm": 0.33292338252067566,
+      "learning_rate": 0.0005790244755244755,
+      "loss": 3.8799,
+      "step": 6100
+    },
+    {
+      "epoch": 1.7909604519774012,
+      "grad_norm": 0.3139176666736603,
+      "learning_rate": 0.0005788496503496503,
+      "loss": 3.869,
+      "step": 6150
+    },
+    {
+      "epoch": 1.8055215795911237,
+      "grad_norm": 0.3225059509277344,
+      "learning_rate": 0.0005786748251748251,
+      "loss": 3.8519,
+      "step": 6200
+    },
+    {
+      "epoch": 1.820082707204846,
+      "grad_norm": 0.3270617127418518,
+      "learning_rate": 0.0005784999999999999,
+      "loss": 3.8537,
+      "step": 6250
+    },
+    {
+      "epoch": 1.8346438348185683,
+      "grad_norm": 0.32356396317481995,
+      "learning_rate": 0.0005783251748251748,
+      "loss": 3.8637,
+      "step": 6300
+    },
+    {
+      "epoch": 1.8492049624322906,
+      "grad_norm": 0.34862807393074036,
+      "learning_rate": 0.0005781503496503496,
+      "loss": 3.8608,
+      "step": 6350
+    },
+    {
+      "epoch": 1.8637660900460131,
+      "grad_norm": 0.3327373266220093,
+      "learning_rate": 0.0005779755244755244,
+      "loss": 3.8592,
+      "step": 6400
+    },
+    {
+      "epoch": 1.8783272176597356,
+      "grad_norm": 0.323494553565979,
+      "learning_rate": 0.0005778006993006993,
+      "loss": 3.859,
+      "step": 6450
+    },
+    {
+      "epoch": 1.892888345273458,
+      "grad_norm": 0.3193258047103882,
+      "learning_rate": 0.000577625874125874,
+      "loss": 3.8555,
+      "step": 6500
+    },
+    {
+      "epoch": 1.9074494728871803,
+      "grad_norm": 0.34051451086997986,
+      "learning_rate": 0.0005774510489510489,
+      "loss": 3.8548,
+      "step": 6550
+    },
+    {
+      "epoch": 1.9220106005009028,
+      "grad_norm": 0.34305649995803833,
+      "learning_rate": 0.0005772762237762237,
+      "loss": 3.8472,
+      "step": 6600
+    },
+    {
+      "epoch": 1.9365717281146253,
+      "grad_norm": 0.3282603323459625,
+      "learning_rate": 0.0005771013986013985,
+      "loss": 3.8501,
+      "step": 6650
+    },
+    {
+      "epoch": 1.9511328557283476,
+      "grad_norm": 0.31083357334136963,
+      "learning_rate": 0.0005769265734265733,
+      "loss": 3.8399,
+      "step": 6700
+    },
+    {
+      "epoch": 1.96569398334207,
+      "grad_norm": 0.3474798798561096,
+      "learning_rate": 0.0005767517482517482,
+      "loss": 3.8562,
+      "step": 6750
+    },
+    {
+      "epoch": 1.9802551109557924,
+      "grad_norm": 0.3274882435798645,
+      "learning_rate": 0.000576576923076923,
+      "loss": 3.8483,
+      "step": 6800
+    },
+    {
+      "epoch": 1.994816238569515,
+      "grad_norm": 0.33568403124809265,
+      "learning_rate": 0.0005764020979020978,
+      "loss": 3.841,
+      "step": 6850
+    },
+    {
+      "epoch": 2.009319121672782,
+      "grad_norm": 0.3249611556529999,
+      "learning_rate": 0.0005762272727272726,
+      "loss": 3.7798,
+      "step": 6900
+    },
+    {
+      "epoch": 2.023880249286505,
+      "grad_norm": 0.3431794047355652,
+      "learning_rate": 0.0005760524475524475,
+      "loss": 3.7367,
+      "step": 6950
+    },
+    {
+      "epoch": 2.038441376900227,
+      "grad_norm": 0.322765976190567,
+      "learning_rate": 0.0005758776223776223,
+      "loss": 3.7549,
+      "step": 7000
+    },
+    {
+      "epoch": 2.038441376900227,
+      "eval_accuracy": 0.3418397325403657,
+      "eval_loss": 3.810866355895996,
+      "eval_runtime": 177.2789,
+      "eval_samples_per_second": 93.897,
+      "eval_steps_per_second": 5.872,
+      "step": 7000
+    },
+    {
+      "epoch": 2.0530025045139495,
+      "grad_norm": 0.3592188060283661,
+      "learning_rate": 0.0005757027972027971,
+      "loss": 3.7467,
+      "step": 7050
+    },
+    {
+      "epoch": 2.067563632127672,
+      "grad_norm": 0.3323240876197815,
+      "learning_rate": 0.000575527972027972,
+      "loss": 3.7268,
+      "step": 7100
+    },
+    {
+      "epoch": 2.0821247597413945,
+      "grad_norm": 0.3217570185661316,
+      "learning_rate": 0.0005753531468531468,
+      "loss": 3.7491,
+      "step": 7150
+    },
+    {
+      "epoch": 2.096685887355117,
+      "grad_norm": 0.33970892429351807,
+      "learning_rate": 0.0005751783216783216,
+      "loss": 3.7353,
+      "step": 7200
+    },
+    {
+      "epoch": 2.111247014968839,
+      "grad_norm": 0.33313965797424316,
+      "learning_rate": 0.0005750034965034964,
+      "loss": 3.7521,
+      "step": 7250
+    },
+    {
+      "epoch": 2.1258081425825615,
+      "grad_norm": 0.31688767671585083,
+      "learning_rate": 0.0005748286713286712,
+      "loss": 3.7518,
+      "step": 7300
+    },
+    {
+      "epoch": 2.140369270196284,
+      "grad_norm": 0.32191213965415955,
+      "learning_rate": 0.000574653846153846,
+      "loss": 3.7579,
+      "step": 7350
+    },
+    {
+      "epoch": 2.1549303978100065,
+      "grad_norm": 0.3361343741416931,
+      "learning_rate": 0.000574479020979021,
+      "loss": 3.7556,
+      "step": 7400
+    },
+    {
+      "epoch": 2.169491525423729,
+      "grad_norm": 0.31816932559013367,
+      "learning_rate": 0.0005743041958041958,
+      "loss": 3.7599,
+      "step": 7450
+    },
+    {
+      "epoch": 2.184052653037451,
+      "grad_norm": 0.32364562153816223,
+      "learning_rate": 0.0005741293706293706,
+      "loss": 3.7605,
+      "step": 7500
+    },
+    {
+      "epoch": 2.198613780651174,
+      "grad_norm": 0.32209041714668274,
+      "learning_rate": 0.0005739545454545454,
+      "loss": 3.7569,
+      "step": 7550
+    },
+    {
+      "epoch": 2.213174908264896,
+      "grad_norm": 0.33622536063194275,
+      "learning_rate": 0.0005737797202797203,
+      "loss": 3.7571,
+      "step": 7600
+    },
+    {
+      "epoch": 2.2277360358786185,
+      "grad_norm": 0.34482723474502563,
+      "learning_rate": 0.0005736048951048951,
+      "loss": 3.7517,
+      "step": 7650
+    },
+    {
+      "epoch": 2.2422971634923408,
+      "grad_norm": 0.322247177362442,
+      "learning_rate": 0.0005734300699300699,
+      "loss": 3.7608,
+      "step": 7700
+    },
+    {
+      "epoch": 2.256858291106063,
+      "grad_norm": 0.3295249044895172,
+      "learning_rate": 0.0005732552447552448,
+      "loss": 3.7445,
+      "step": 7750
+    },
+    {
+      "epoch": 2.271419418719786,
+      "grad_norm": 0.3132839798927307,
+      "learning_rate": 0.0005730804195804196,
+      "loss": 3.7522,
+      "step": 7800
+    },
+    {
+      "epoch": 2.285980546333508,
+      "grad_norm": 0.31943801045417786,
+      "learning_rate": 0.0005729055944055944,
+      "loss": 3.7567,
+      "step": 7850
+    },
+    {
+      "epoch": 2.3005416739472304,
+      "grad_norm": 0.3134506940841675,
+      "learning_rate": 0.0005727307692307692,
+      "loss": 3.7585,
+      "step": 7900
+    },
+    {
+      "epoch": 2.3151028015609527,
+      "grad_norm": 0.32046815752983093,
+      "learning_rate": 0.0005725559440559441,
+      "loss": 3.7384,
+      "step": 7950
+    },
+    {
+      "epoch": 2.3296639291746755,
+      "grad_norm": 0.34398654103279114,
+      "learning_rate": 0.0005723811188811188,
+      "loss": 3.751,
+      "step": 8000
+    },
+    {
+      "epoch": 2.3296639291746755,
+      "eval_accuracy": 0.34463184446561096,
+      "eval_loss": 3.7817397117614746,
+      "eval_runtime": 177.3233,
+      "eval_samples_per_second": 93.874,
+      "eval_steps_per_second": 5.871,
+      "step": 8000
+    },
+    {
+      "epoch": 2.3442250567883978,
+      "grad_norm": 0.3400885760784149,
+      "learning_rate": 0.0005722062937062937,
+      "loss": 3.7609,
+      "step": 8050
+    },
+    {
+      "epoch": 2.35878618440212,
+      "grad_norm": 0.30969181656837463,
+      "learning_rate": 0.0005720314685314685,
+      "loss": 3.754,
+      "step": 8100
+    },
+    {
+      "epoch": 2.3733473120158424,
+      "grad_norm": 0.35174381732940674,
+      "learning_rate": 0.0005718566433566433,
+      "loss": 3.7636,
+      "step": 8150
+    },
+    {
+      "epoch": 2.387908439629565,
+      "grad_norm": 0.30494076013565063,
+      "learning_rate": 0.0005716818181818181,
+      "loss": 3.7505,
+      "step": 8200
+    },
+    {
+      "epoch": 2.4024695672432874,
+      "grad_norm": 0.3317963480949402,
+      "learning_rate": 0.000571506993006993,
+      "loss": 3.7411,
+      "step": 8250
+    },
+    {
+      "epoch": 2.4170306948570097,
+      "grad_norm": 0.3330554962158203,
+      "learning_rate": 0.0005713321678321678,
+      "loss": 3.7478,
+      "step": 8300
+    },
+    {
+      "epoch": 2.431591822470732,
+      "grad_norm": 0.33387240767478943,
+      "learning_rate": 0.0005711573426573426,
+      "loss": 3.7595,
+      "step": 8350
+    },
+    {
+      "epoch": 2.4461529500844543,
+      "grad_norm": 0.3251585066318512,
+      "learning_rate": 0.0005709825174825175,
+      "loss": 3.744,
+      "step": 8400
+    },
+    {
+      "epoch": 2.460714077698177,
+      "grad_norm": 0.3279932141304016,
+      "learning_rate": 0.0005708076923076923,
+      "loss": 3.75,
+      "step": 8450
+    },
+    {
+      "epoch": 2.4752752053118994,
+      "grad_norm": 0.317627876996994,
+      "learning_rate": 0.0005706328671328671,
+      "loss": 3.7403,
+      "step": 8500
+    },
+    {
+      "epoch": 2.4898363329256217,
+      "grad_norm": 0.33804240822792053,
+      "learning_rate": 0.0005704580419580419,
+      "loss": 3.7408,
+      "step": 8550
+    },
+    {
+      "epoch": 2.5043974605393444,
+      "grad_norm": 0.333448588848114,
+      "learning_rate": 0.0005702832167832168,
+      "loss": 3.7505,
+      "step": 8600
+    },
+    {
+      "epoch": 2.5189585881530667,
+      "grad_norm": 0.325022429227829,
+      "learning_rate": 0.0005701083916083916,
+      "loss": 3.7527,
+      "step": 8650
+    },
+    {
+      "epoch": 2.533519715766789,
+      "grad_norm": 0.3052625358104706,
+      "learning_rate": 0.0005699335664335664,
+      "loss": 3.7333,
+      "step": 8700
+    },
+    {
+      "epoch": 2.5480808433805113,
+      "grad_norm": 0.3203795850276947,
+      "learning_rate": 0.0005697587412587412,
+      "loss": 3.7428,
+      "step": 8750
+    },
+    {
+      "epoch": 2.5626419709942336,
+      "grad_norm": 0.30669328570365906,
+      "learning_rate": 0.000569583916083916,
+      "loss": 3.7437,
+      "step": 8800
+    },
+    {
+      "epoch": 2.5772030986079564,
+      "grad_norm": 0.3302299976348877,
+      "learning_rate": 0.0005694090909090908,
+      "loss": 3.7527,
+      "step": 8850
+    },
+    {
+      "epoch": 2.5917642262216787,
+      "grad_norm": 0.3199024498462677,
+      "learning_rate": 0.0005692342657342657,
+      "loss": 3.7376,
+      "step": 8900
+    },
+    {
+      "epoch": 2.606325353835401,
+      "grad_norm": 0.3255689740180969,
+      "learning_rate": 0.0005690594405594405,
+      "loss": 3.7342,
+      "step": 8950
+    },
+    {
+      "epoch": 2.6208864814491233,
+      "grad_norm": 0.33225247263908386,
+      "learning_rate": 0.0005688846153846153,
+      "loss": 3.7396,
+      "step": 9000
+    },
+    {
+      "epoch": 2.6208864814491233,
+      "eval_accuracy": 0.347539990684339,
+      "eval_loss": 3.750495672225952,
+      "eval_runtime": 177.3467,
+      "eval_samples_per_second": 93.861,
+      "eval_steps_per_second": 5.87,
+      "step": 9000
+    },
+    {
+      "epoch": 2.6354476090628456,
+      "grad_norm": 0.31644511222839355,
+      "learning_rate": 0.0005687097902097901,
+      "loss": 3.737,
+      "step": 9050
+    },
+    {
+      "epoch": 2.6500087366765683,
+      "grad_norm": 0.32711660861968994,
+      "learning_rate": 0.000568534965034965,
+      "loss": 3.7258,
+      "step": 9100
+    },
+    {
+      "epoch": 2.6645698642902906,
+      "grad_norm": 0.32171499729156494,
+      "learning_rate": 0.0005683601398601398,
+      "loss": 3.7459,
+      "step": 9150
+    },
+    {
+      "epoch": 2.679130991904013,
+      "grad_norm": 0.30605417490005493,
+      "learning_rate": 0.0005681853146853146,
+      "loss": 3.7252,
+      "step": 9200
+    },
+    {
+      "epoch": 2.6936921195177357,
+      "grad_norm": 0.31815701723098755,
+      "learning_rate": 0.0005680104895104895,
+      "loss": 3.7312,
+      "step": 9250
+    },
+    {
+      "epoch": 2.708253247131458,
+      "grad_norm": 0.320014625787735,
+      "learning_rate": 0.0005678356643356643,
+      "loss": 3.7287,
+      "step": 9300
+    },
+    {
+      "epoch": 2.7228143747451803,
+      "grad_norm": 0.31022703647613525,
+      "learning_rate": 0.0005676608391608391,
+      "loss": 3.7285,
+      "step": 9350
+    },
+    {
+      "epoch": 2.7373755023589026,
+      "grad_norm": 0.33272796869277954,
+      "learning_rate": 0.0005674860139860139,
+      "loss": 3.7302,
+      "step": 9400
+    },
+    {
+      "epoch": 2.751936629972625,
+      "grad_norm": 0.336722731590271,
+      "learning_rate": 0.0005673111888111888,
+      "loss": 3.713,
+      "step": 9450
+    },
+    {
+      "epoch": 2.7664977575863476,
+      "grad_norm": 0.3396904468536377,
+      "learning_rate": 0.0005671363636363635,
+      "loss": 3.7374,
+      "step": 9500
+    },
+    {
+      "epoch": 2.78105888520007,
+      "grad_norm": 0.3151884973049164,
+      "learning_rate": 0.0005669615384615384,
+      "loss": 3.7306,
+      "step": 9550
+    },
+    {
+      "epoch": 2.7956200128137922,
+      "grad_norm": 0.31197404861450195,
+      "learning_rate": 0.0005667867132867132,
+      "loss": 3.7201,
+      "step": 9600
+    },
+    {
+      "epoch": 2.8101811404275145,
+      "grad_norm": 0.33173638582229614,
+      "learning_rate": 0.000566611888111888,
+      "loss": 3.7265,
+      "step": 9650
+    },
+    {
+      "epoch": 2.824742268041237,
+      "grad_norm": 0.31983959674835205,
+      "learning_rate": 0.0005664370629370628,
+      "loss": 3.7218,
+      "step": 9700
+    },
+    {
+      "epoch": 2.8393033956549596,
+      "grad_norm": 0.3080846071243286,
+      "learning_rate": 0.0005662622377622377,
+      "loss": 3.7198,
+      "step": 9750
+    },
+    {
+      "epoch": 2.853864523268682,
+      "grad_norm": 0.31331005692481995,
+      "learning_rate": 0.0005660874125874125,
+      "loss": 3.7078,
+      "step": 9800
+    },
+    {
+      "epoch": 2.868425650882404,
+      "grad_norm": 0.3520505428314209,
+      "learning_rate": 0.0005659125874125873,
+      "loss": 3.726,
+      "step": 9850
+    },
+    {
+      "epoch": 2.882986778496127,
+      "grad_norm": 0.3302844762802124,
+      "learning_rate": 0.0005657377622377622,
+      "loss": 3.7182,
+      "step": 9900
+    },
+    {
+      "epoch": 2.8975479061098492,
+      "grad_norm": 0.3250056803226471,
+      "learning_rate": 0.000565562937062937,
+      "loss": 3.7079,
+      "step": 9950
+    },
+    {
+      "epoch": 2.9121090337235715,
+      "grad_norm": 0.3227342963218689,
+      "learning_rate": 0.0005653881118881118,
+      "loss": 3.7105,
+      "step": 10000
+    },
+    {
+      "epoch": 2.9121090337235715,
+      "eval_accuracy": 0.3495654768468674,
+      "eval_loss": 3.724705457687378,
+      "eval_runtime": 177.4012,
+      "eval_samples_per_second": 93.833,
+      "eval_steps_per_second": 5.868,
+      "step": 10000
+    },
+    {
+      "epoch": 2.926670161337294,
+      "grad_norm": 0.32378628849983215,
+      "learning_rate": 0.0005652132867132866,
+      "loss": 3.7132,
+      "step": 10050
+    },
+    {
+      "epoch": 2.941231288951016,
+      "grad_norm": 0.3264535665512085,
+      "learning_rate": 0.0005650384615384615,
+      "loss": 3.72,
+      "step": 10100
+    },
+    {
+      "epoch": 2.955792416564739,
+      "grad_norm": 0.3335714042186737,
+      "learning_rate": 0.0005648636363636363,
+      "loss": 3.714,
+      "step": 10150
+    },
+    {
+      "epoch": 2.970353544178461,
+      "grad_norm": 0.3222605288028717,
+      "learning_rate": 0.0005646888111888111,
+      "loss": 3.7256,
+      "step": 10200
+    },
+    {
+      "epoch": 2.9849146717921835,
+      "grad_norm": 0.3324715495109558,
+      "learning_rate": 0.000564513986013986,
+      "loss": 3.7218,
+      "step": 10250
+    },
+    {
+      "epoch": 2.9994757994059063,
+      "grad_norm": 0.3224101960659027,
+      "learning_rate": 0.0005643391608391607,
+      "loss": 3.7147,
+      "step": 10300
+    },
+    {
+      "epoch": 3.0139786825091734,
+      "grad_norm": 0.3330100476741791,
+      "learning_rate": 0.0005641643356643355,
+      "loss": 3.6132,
+      "step": 10350
+    },
+    {
+      "epoch": 3.0285398101228957,
+      "grad_norm": 0.3183719515800476,
+      "learning_rate": 0.0005639895104895105,
+      "loss": 3.6057,
+      "step": 10400
+    },
+    {
+      "epoch": 3.0431009377366185,
+      "grad_norm": 0.31805160641670227,
+      "learning_rate": 0.0005638146853146853,
+      "loss": 3.616,
+      "step": 10450
+    },
+    {
+      "epoch": 3.057662065350341,
+      "grad_norm": 0.3436165153980255,
+      "learning_rate": 0.0005636398601398601,
+      "loss": 3.625,
+      "step": 10500
+    },
+    {
+      "epoch": 3.072223192964063,
+      "grad_norm": 0.32004937529563904,
+      "learning_rate": 0.000563465034965035,
+      "loss": 3.6283,
+      "step": 10550
+    },
+    {
+      "epoch": 3.0867843205777854,
+      "grad_norm": 0.3213531970977783,
+      "learning_rate": 0.0005632902097902098,
+      "loss": 3.6217,
+      "step": 10600
+    },
+    {
+      "epoch": 3.101345448191508,
+      "grad_norm": 0.31485000252723694,
+      "learning_rate": 0.0005631153846153846,
+      "loss": 3.6229,
+      "step": 10650
+    },
+    {
+      "epoch": 3.1159065758052304,
+      "grad_norm": 0.3263075649738312,
+      "learning_rate": 0.0005629405594405594,
+      "loss": 3.6317,
+      "step": 10700
+    },
+    {
+      "epoch": 3.1304677034189528,
+      "grad_norm": 0.34470370411872864,
+      "learning_rate": 0.0005627657342657343,
+      "loss": 3.6219,
+      "step": 10750
+    },
+    {
+      "epoch": 3.145028831032675,
+      "grad_norm": 0.33321434259414673,
+      "learning_rate": 0.0005625909090909091,
+      "loss": 3.6205,
+      "step": 10800
+    },
+    {
+      "epoch": 3.1595899586463974,
+      "grad_norm": 0.32217034697532654,
+      "learning_rate": 0.0005624160839160839,
+      "loss": 3.6309,
+      "step": 10850
+    },
+    {
+      "epoch": 3.17415108626012,
+      "grad_norm": 0.34395185112953186,
+      "learning_rate": 0.0005622412587412587,
+      "loss": 3.6366,
+      "step": 10900
+    },
+    {
+      "epoch": 3.1887122138738424,
+      "grad_norm": 0.3254980146884918,
+      "learning_rate": 0.0005620664335664336,
+      "loss": 3.6369,
+      "step": 10950
+    },
+    {
+      "epoch": 3.2032733414875647,
+      "grad_norm": 0.338420033454895,
+      "learning_rate": 0.0005618916083916083,
+      "loss": 3.6418,
+      "step": 11000
+    },
+    {
+      "epoch": 3.2032733414875647,
+      "eval_accuracy": 0.3518970960390101,
+      "eval_loss": 3.711723566055298,
+      "eval_runtime": 177.6335,
+      "eval_samples_per_second": 93.71,
+      "eval_steps_per_second": 5.86,
+      "step": 11000
+    },
+    {
+      "epoch": 3.217834469101287,
+      "grad_norm": 0.3291839361190796,
+      "learning_rate": 0.0005617167832167832,
+      "loss": 3.6324,
+      "step": 11050
+    },
+    {
+      "epoch": 3.2323955967150098,
+      "grad_norm": 0.3438150882720947,
+      "learning_rate": 0.000561541958041958,
+      "loss": 3.6253,
+      "step": 11100
+    },
+    {
+      "epoch": 3.246956724328732,
+      "grad_norm": 0.36098095774650574,
+      "learning_rate": 0.0005613671328671328,
+      "loss": 3.6401,
+      "step": 11150
+    },
+    {
+      "epoch": 3.2615178519424544,
+      "grad_norm": 0.351468026638031,
+      "learning_rate": 0.0005611923076923077,
+      "loss": 3.6395,
+      "step": 11200
+    },
+    {
+      "epoch": 3.2760789795561767,
+      "grad_norm": 0.3300299644470215,
+      "learning_rate": 0.0005610174825174825,
+      "loss": 3.644,
+      "step": 11250
+    },
+    {
+      "epoch": 3.2906401071698994,
+      "grad_norm": 0.3039533495903015,
+      "learning_rate": 0.0005608426573426573,
+      "loss": 3.6416,
+      "step": 11300
+    },
+    {
+      "epoch": 3.3052012347836217,
+      "grad_norm": 0.35001346468925476,
+      "learning_rate": 0.0005606678321678321,
+      "loss": 3.6403,
+      "step": 11350
+    },
+    {
+      "epoch": 3.319762362397344,
+      "grad_norm": 0.3169427514076233,
+      "learning_rate": 0.000560493006993007,
+      "loss": 3.6419,
+      "step": 11400
+    },
+    {
+      "epoch": 3.3343234900110663,
+      "grad_norm": 0.3217058777809143,
+      "learning_rate": 0.0005603181818181818,
+      "loss": 3.6448,
+      "step": 11450
+    },
+    {
+      "epoch": 3.3488846176247886,
+      "grad_norm": 0.3203667402267456,
+      "learning_rate": 0.0005601433566433566,
+      "loss": 3.6387,
+      "step": 11500
+    },
+    {
+      "epoch": 3.3634457452385114,
+      "grad_norm": 0.3251610994338989,
+      "learning_rate": 0.0005599685314685314,
+      "loss": 3.6272,
+      "step": 11550
+    },
+    {
+      "epoch": 3.3780068728522337,
+      "grad_norm": 0.329180508852005,
+      "learning_rate": 0.0005597937062937063,
+      "loss": 3.6358,
+      "step": 11600
+    },
+    {
+      "epoch": 3.392568000465956,
+      "grad_norm": 0.31434664130210876,
+      "learning_rate": 0.0005596188811188811,
+      "loss": 3.6503,
+      "step": 11650
+    },
+    {
+      "epoch": 3.4071291280796787,
+      "grad_norm": 0.3283628225326538,
+      "learning_rate": 0.0005594440559440559,
+      "loss": 3.6348,
+      "step": 11700
+    },
+    {
+      "epoch": 3.421690255693401,
+      "grad_norm": 0.3240419626235962,
+      "learning_rate": 0.0005592692307692307,
+      "loss": 3.6482,
+      "step": 11750
+    },
+    {
+      "epoch": 3.4362513833071233,
+      "grad_norm": 0.32653725147247314,
+      "learning_rate": 0.0005590944055944055,
+      "loss": 3.6465,
+      "step": 11800
+    },
+    {
+      "epoch": 3.4508125109208456,
+      "grad_norm": 0.3058265447616577,
+      "learning_rate": 0.0005589195804195803,
+      "loss": 3.6452,
+      "step": 11850
+    },
+    {
+      "epoch": 3.465373638534568,
+      "grad_norm": 0.3274167776107788,
+      "learning_rate": 0.0005587447552447552,
+      "loss": 3.6373,
+      "step": 11900
+    },
+    {
+      "epoch": 3.4799347661482907,
+      "grad_norm": 0.34979650378227234,
+      "learning_rate": 0.00055856993006993,
+      "loss": 3.6462,
+      "step": 11950
+    },
+    {
+      "epoch": 3.494495893762013,
+      "grad_norm": 0.3270816206932068,
+      "learning_rate": 0.0005583951048951048,
+      "loss": 3.6293,
+      "step": 12000
+    },
+    {
+      "epoch": 3.494495893762013,
+      "eval_accuracy": 0.3533076122023403,
+      "eval_loss": 3.6936569213867188,
+      "eval_runtime": 177.1638,
+      "eval_samples_per_second": 93.958,
+      "eval_steps_per_second": 5.876,
+      "step": 12000
+    },
+    {
+      "epoch": 3.5090570213757353,
+      "grad_norm": 0.3215574026107788,
+      "learning_rate": 0.0005582202797202797,
+      "loss": 3.632,
+      "step": 12050
+    },
+    {
+      "epoch": 3.523618148989458,
+      "grad_norm": 0.326255738735199,
+      "learning_rate": 0.0005580454545454545,
+      "loss": 3.6415,
+      "step": 12100
+    },
+    {
+      "epoch": 3.53817927660318,
+      "grad_norm": 0.33679527044296265,
+      "learning_rate": 0.0005578706293706293,
+      "loss": 3.6456,
+      "step": 12150
+    },
+    {
+      "epoch": 3.5527404042169026,
+      "grad_norm": 0.3434588313102722,
+      "learning_rate": 0.0005576958041958041,
+      "loss": 3.6466,
+      "step": 12200
+    },
+    {
+      "epoch": 3.567301531830625,
+      "grad_norm": 0.31146910786628723,
+      "learning_rate": 0.000557520979020979,
+      "loss": 3.6367,
+      "step": 12250
+    },
+    {
+      "epoch": 3.5818626594443472,
+      "grad_norm": 0.3134685158729553,
+      "learning_rate": 0.0005573461538461538,
+      "loss": 3.6253,
+      "step": 12300
+    },
+    {
+      "epoch": 3.59642378705807,
+      "grad_norm": 0.31741416454315186,
+      "learning_rate": 0.0005571713286713286,
+      "loss": 3.6344,
+      "step": 12350
+    },
+    {
+      "epoch": 3.6109849146717923,
+      "grad_norm": 0.3213517963886261,
+      "learning_rate": 0.0005569965034965034,
+      "loss": 3.6461,
+      "step": 12400
+    },
+    {
+      "epoch": 3.6255460422855146,
+      "grad_norm": 0.3236750662326813,
+      "learning_rate": 0.0005568216783216783,
+      "loss": 3.6363,
+      "step": 12450
+    },
+    {
+      "epoch": 3.640107169899237,
+      "grad_norm": 0.3350413143634796,
+      "learning_rate": 0.000556646853146853,
+      "loss": 3.6341,
+      "step": 12500
+    },
+    {
+      "epoch": 3.654668297512959,
+      "grad_norm": 0.3210690915584564,
+      "learning_rate": 0.0005564720279720279,
+      "loss": 3.6341,
+      "step": 12550
+    },
+    {
+      "epoch": 3.669229425126682,
+      "grad_norm": 0.31280940771102905,
+      "learning_rate": 0.0005562972027972027,
+      "loss": 3.6406,
+      "step": 12600
+    },
+    {
+      "epoch": 3.6837905527404042,
+      "grad_norm": 0.3478114902973175,
+      "learning_rate": 0.0005561223776223775,
+      "loss": 3.6437,
+      "step": 12650
+    },
+    {
+      "epoch": 3.6983516803541265,
+      "grad_norm": 0.3223908245563507,
+      "learning_rate": 0.0005559475524475524,
+      "loss": 3.6346,
+      "step": 12700
+    },
+    {
+      "epoch": 3.7129128079678493,
+      "grad_norm": 0.30570903420448303,
+      "learning_rate": 0.0005557727272727272,
+      "loss": 3.6475,
+      "step": 12750
+    },
+    {
+      "epoch": 3.7274739355815716,
+      "grad_norm": 0.3094755709171295,
+      "learning_rate": 0.000555597902097902,
+      "loss": 3.6307,
+      "step": 12800
+    },
+    {
+      "epoch": 3.742035063195294,
+      "grad_norm": 0.2975589334964752,
+      "learning_rate": 0.0005554230769230768,
+      "loss": 3.6396,
+      "step": 12850
+    },
+    {
+      "epoch": 3.756596190809016,
+      "grad_norm": 0.3289732038974762,
+      "learning_rate": 0.0005552482517482517,
+      "loss": 3.6434,
+      "step": 12900
+    },
+    {
+      "epoch": 3.7711573184227385,
+      "grad_norm": 0.32117950916290283,
+      "learning_rate": 0.0005550734265734265,
+      "loss": 3.6423,
+      "step": 12950
+    },
+    {
+      "epoch": 3.7857184460364612,
+      "grad_norm": 0.32686787843704224,
+      "learning_rate": 0.0005548986013986013,
+      "loss": 3.642,
+      "step": 13000
+    },
+    {
+      "epoch": 3.7857184460364612,
+      "eval_accuracy": 0.35558350671858546,
+      "eval_loss": 3.676309823989868,
+      "eval_runtime": 177.0944,
+      "eval_samples_per_second": 93.995,
+      "eval_steps_per_second": 5.878,
+      "step": 13000
+    },
+    {
+      "epoch": 3.8002795736501835,
+      "grad_norm": 0.3189132809638977,
+      "learning_rate": 0.0005547237762237761,
+      "loss": 3.6439,
+      "step": 13050
+    },
+    {
+      "epoch": 3.814840701263906,
+      "grad_norm": 0.33861854672431946,
+      "learning_rate": 0.000554548951048951,
+      "loss": 3.6343,
+      "step": 13100
+    },
+    {
+      "epoch": 3.829401828877628,
+      "grad_norm": 0.3242887854576111,
+      "learning_rate": 0.0005543741258741258,
+      "loss": 3.64,
+      "step": 13150
+    },
+    {
+      "epoch": 3.8439629564913504,
+      "grad_norm": 0.3283834457397461,
+      "learning_rate": 0.0005541993006993006,
+      "loss": 3.6236,
+      "step": 13200
+    },
+    {
+      "epoch": 3.858524084105073,
+      "grad_norm": 0.3189845085144043,
+      "learning_rate": 0.0005540244755244756,
+      "loss": 3.6343,
+      "step": 13250
+    },
+    {
+      "epoch": 3.8730852117187955,
+      "grad_norm": 0.3211795687675476,
+      "learning_rate": 0.0005538496503496502,
+      "loss": 3.6512,
+      "step": 13300
+    },
+    {
+      "epoch": 3.887646339332518,
+      "grad_norm": 0.3205946087837219,
+      "learning_rate": 0.0005536748251748252,
+      "loss": 3.63,
+      "step": 13350
+    },
+    {
+      "epoch": 3.9022074669462405,
+      "grad_norm": 0.3182278275489807,
+      "learning_rate": 0.0005535,
+      "loss": 3.6192,
+      "step": 13400
+    },
+    {
+      "epoch": 3.916768594559963,
+      "grad_norm": 0.3093646466732025,
+      "learning_rate": 0.0005533251748251748,
+      "loss": 3.629,
+      "step": 13450
+    },
+    {
+      "epoch": 3.931329722173685,
+      "grad_norm": 0.325284481048584,
+      "learning_rate": 0.0005531503496503496,
+      "loss": 3.6445,
+      "step": 13500
+    },
+    {
+      "epoch": 3.9458908497874075,
+      "grad_norm": 0.34685397148132324,
+      "learning_rate": 0.0005529755244755245,
+      "loss": 3.6269,
+      "step": 13550
+    },
+    {
+      "epoch": 3.9604519774011298,
+      "grad_norm": 0.3556680977344513,
+      "learning_rate": 0.0005528006993006993,
+      "loss": 3.6371,
+      "step": 13600
+    },
+    {
+      "epoch": 3.9750131050148525,
+      "grad_norm": 0.3256296217441559,
+      "learning_rate": 0.0005526258741258741,
+      "loss": 3.6367,
+      "step": 13650
+    },
+    {
+      "epoch": 3.989574232628575,
+      "grad_norm": 0.31314557790756226,
+      "learning_rate": 0.0005524510489510489,
+      "loss": 3.6367,
+      "step": 13700
+    },
+    {
+      "epoch": 4.004077115731842,
+      "grad_norm": 0.34284132719039917,
+      "learning_rate": 0.0005522762237762238,
+      "loss": 3.6039,
+      "step": 13750
+    },
+    {
+      "epoch": 4.018638243345564,
+      "grad_norm": 0.31531134247779846,
+      "learning_rate": 0.0005521013986013986,
+      "loss": 3.5281,
+      "step": 13800
+    },
+    {
+      "epoch": 4.033199370959287,
+      "grad_norm": 0.329609751701355,
+      "learning_rate": 0.0005519265734265734,
+      "loss": 3.5438,
+      "step": 13850
+    },
+    {
+      "epoch": 4.04776049857301,
+      "grad_norm": 0.3221539855003357,
+      "learning_rate": 0.0005517517482517482,
+      "loss": 3.5235,
+      "step": 13900
+    },
+    {
+      "epoch": 4.062321626186732,
+      "grad_norm": 0.3120812177658081,
+      "learning_rate": 0.0005515769230769231,
+      "loss": 3.5343,
+      "step": 13950
+    },
+    {
+      "epoch": 4.076882753800454,
+      "grad_norm": 0.32255053520202637,
+      "learning_rate": 0.0005514020979020979,
+      "loss": 3.543,
+      "step": 14000
+    },
+    {
+      "epoch": 4.076882753800454,
+      "eval_accuracy": 0.35604047257346666,
+      "eval_loss": 3.672227382659912,
+      "eval_runtime": 177.2502,
+      "eval_samples_per_second": 93.912,
+      "eval_steps_per_second": 5.873,
+      "step": 14000
+    },
+    {
+      "epoch": 4.091443881414177,
+      "grad_norm": 0.33779409527778625,
+      "learning_rate": 0.0005512272727272727,
+      "loss": 3.5375,
+      "step": 14050
+    },
+    {
+      "epoch": 4.106005009027899,
+      "grad_norm": 0.30933794379234314,
+      "learning_rate": 0.0005510524475524475,
+      "loss": 3.5311,
+      "step": 14100
+    },
+    {
+      "epoch": 4.120566136641622,
+      "grad_norm": 0.3431943356990814,
+      "learning_rate": 0.0005508776223776223,
+      "loss": 3.532,
+      "step": 14150
+    },
+    {
+      "epoch": 4.135127264255344,
+      "grad_norm": 0.31620094180107117,
+      "learning_rate": 0.0005507027972027972,
+      "loss": 3.5503,
+      "step": 14200
+    },
+    {
+      "epoch": 4.149688391869066,
+      "grad_norm": 0.3254013657569885,
+      "learning_rate": 0.000550527972027972,
+      "loss": 3.5538,
+      "step": 14250
+    },
+    {
+      "epoch": 4.164249519482789,
+      "grad_norm": 0.33322015404701233,
+      "learning_rate": 0.0005503531468531468,
+      "loss": 3.5521,
+      "step": 14300
+    },
+    {
+      "epoch": 4.178810647096511,
+      "grad_norm": 0.3261924386024475,
+      "learning_rate": 0.0005501783216783216,
+      "loss": 3.5439,
+      "step": 14350
+    },
+    {
+      "epoch": 4.193371774710234,
+      "grad_norm": 0.32681429386138916,
+      "learning_rate": 0.0005500034965034965,
+      "loss": 3.5543,
+      "step": 14400
+    },
+    {
+      "epoch": 4.207932902323956,
+      "grad_norm": 0.3240799307823181,
+      "learning_rate": 0.0005498286713286713,
+      "loss": 3.5622,
+      "step": 14450
+    },
+    {
+      "epoch": 4.222494029937678,
+      "grad_norm": 0.32154789566993713,
+      "learning_rate": 0.0005496538461538461,
+      "loss": 3.5425,
+      "step": 14500
+    },
+    {
+      "epoch": 4.237055157551401,
+      "grad_norm": 0.3319743275642395,
+      "learning_rate": 0.0005494790209790209,
+      "loss": 3.5755,
+      "step": 14550
+    },
+    {
+      "epoch": 4.251616285165123,
+      "grad_norm": 0.32192280888557434,
+      "learning_rate": 0.0005493041958041958,
+      "loss": 3.5474,
+      "step": 14600
+    },
+    {
+      "epoch": 4.266177412778846,
+      "grad_norm": 0.3264658451080322,
+      "learning_rate": 0.0005491293706293706,
+      "loss": 3.5525,
+      "step": 14650
+    },
+    {
+      "epoch": 4.280738540392568,
+      "grad_norm": 0.30953332781791687,
+      "learning_rate": 0.0005489545454545454,
+      "loss": 3.5686,
+      "step": 14700
+    },
+    {
+      "epoch": 4.29529966800629,
+      "grad_norm": 0.32248425483703613,
+      "learning_rate": 0.0005487797202797203,
+      "loss": 3.5576,
+      "step": 14750
+    },
+    {
+      "epoch": 4.309860795620013,
+      "grad_norm": 0.3248192071914673,
+      "learning_rate": 0.000548604895104895,
+      "loss": 3.5652,
+      "step": 14800
+    },
+    {
+      "epoch": 4.324421923233735,
+      "grad_norm": 0.3153315484523773,
+      "learning_rate": 0.0005484300699300699,
+      "loss": 3.5583,
+      "step": 14850
+    },
+    {
+      "epoch": 4.338983050847458,
+      "grad_norm": 0.33788663148880005,
+      "learning_rate": 0.0005482552447552447,
+      "loss": 3.5586,
+      "step": 14900
+    },
+    {
+      "epoch": 4.35354417846118,
+      "grad_norm": 0.3236865699291229,
+      "learning_rate": 0.0005480804195804195,
+      "loss": 3.5692,
+      "step": 14950
+    },
+    {
+      "epoch": 4.368105306074902,
+      "grad_norm": 0.3407519459724426,
+      "learning_rate": 0.0005479055944055943,
+      "loss": 3.5577,
+      "step": 15000
+    },
+    {
+      "epoch": 4.368105306074902,
+      "eval_accuracy": 0.3575495062017802,
+      "eval_loss": 3.658498764038086,
+      "eval_runtime": 176.9515,
+      "eval_samples_per_second": 94.071,
+      "eval_steps_per_second": 5.883,
+      "step": 15000
+    },
+    {
+      "epoch": 4.382666433688625,
+      "grad_norm": 0.3334459364414215,
+      "learning_rate": 0.0005477307692307692,
+      "loss": 3.5686,
+      "step": 15050
+    },
+    {
+      "epoch": 4.397227561302348,
+      "grad_norm": 0.32275843620300293,
+      "learning_rate": 0.000547555944055944,
+      "loss": 3.5695,
+      "step": 15100
+    },
+    {
+      "epoch": 4.41178868891607,
+      "grad_norm": 0.3123950660228729,
+      "learning_rate": 0.0005473811188811188,
+      "loss": 3.5732,
+      "step": 15150
+    },
+    {
+      "epoch": 4.426349816529792,
+      "grad_norm": 0.32868602871894836,
+      "learning_rate": 0.0005472062937062936,
+      "loss": 3.5836,
+      "step": 15200
+    },
+    {
+      "epoch": 4.440910944143514,
+      "grad_norm": 0.3395336866378784,
+      "learning_rate": 0.0005470314685314685,
+      "loss": 3.5748,
+      "step": 15250
+    },
+    {
+      "epoch": 4.455472071757237,
+      "grad_norm": 0.3110513687133789,
+      "learning_rate": 0.0005468566433566433,
+      "loss": 3.5758,
+      "step": 15300
+    },
+    {
+      "epoch": 4.47003319937096,
+      "grad_norm": 0.3269290328025818,
+      "learning_rate": 0.0005466818181818181,
+      "loss": 3.5562,
+      "step": 15350
+    },
+    {
+      "epoch": 4.4845943269846815,
+      "grad_norm": 0.32525432109832764,
+      "learning_rate": 0.000546506993006993,
+      "loss": 3.5786,
+      "step": 15400
+    },
+    {
+      "epoch": 4.499155454598404,
+      "grad_norm": 0.33424168825149536,
+      "learning_rate": 0.0005463321678321678,
+      "loss": 3.577,
+      "step": 15450
+    },
+    {
+      "epoch": 4.513716582212126,
+      "grad_norm": 0.31379204988479614,
+      "learning_rate": 0.0005461573426573426,
+      "loss": 3.5716,
+      "step": 15500
+    },
+    {
+      "epoch": 4.528277709825849,
+      "grad_norm": 0.33003076910972595,
+      "learning_rate": 0.0005459825174825174,
+      "loss": 3.5673,
+      "step": 15550
+    },
+    {
+      "epoch": 4.542838837439572,
+      "grad_norm": 0.31797513365745544,
+      "learning_rate": 0.0005458076923076922,
+      "loss": 3.5634,
+      "step": 15600
+    },
+    {
+      "epoch": 4.5573999650532935,
+      "grad_norm": 0.3288755416870117,
+      "learning_rate": 0.000545632867132867,
+      "loss": 3.5642,
+      "step": 15650
+    },
+    {
+      "epoch": 4.571961092667016,
+      "grad_norm": 0.3203357756137848,
+      "learning_rate": 0.0005454580419580419,
+      "loss": 3.5713,
+      "step": 15700
+    },
+    {
+      "epoch": 4.586522220280738,
+      "grad_norm": 0.31958380341529846,
+      "learning_rate": 0.0005452832167832167,
+      "loss": 3.5633,
+      "step": 15750
+    },
+    {
+      "epoch": 4.601083347894461,
+      "grad_norm": 0.33480215072631836,
+      "learning_rate": 0.0005451083916083915,
+      "loss": 3.5648,
+      "step": 15800
+    },
+    {
+      "epoch": 4.615644475508184,
+      "grad_norm": 0.31152933835983276,
+      "learning_rate": 0.0005449335664335663,
+      "loss": 3.5824,
+      "step": 15850
+    },
+    {
+      "epoch": 4.630205603121905,
+      "grad_norm": 0.31223437190055847,
+      "learning_rate": 0.0005447587412587412,
+      "loss": 3.5717,
+      "step": 15900
+    },
+    {
+      "epoch": 4.644766730735628,
+      "grad_norm": 0.32843029499053955,
+      "learning_rate": 0.000544583916083916,
+      "loss": 3.561,
+      "step": 15950
+    },
+    {
+      "epoch": 4.659327858349351,
+      "grad_norm": 0.3143811523914337,
+      "learning_rate": 0.0005444090909090908,
+      "loss": 3.5824,
+      "step": 16000
+    },
+    {
+      "epoch": 4.659327858349351,
+      "eval_accuracy": 0.35877027631680114,
+      "eval_loss": 3.646221399307251,
+      "eval_runtime": 176.8737,
+      "eval_samples_per_second": 94.112,
+      "eval_steps_per_second": 5.886,
+      "step": 16000
+    },
+    {
+      "epoch": 4.673888985963073,
+      "grad_norm": 0.32724830508232117,
+      "learning_rate": 0.0005442342657342657,
+      "loss": 3.5791,
+      "step": 16050
+    },
+    {
+      "epoch": 4.6884501135767955,
+      "grad_norm": 0.3239496648311615,
+      "learning_rate": 0.0005440594405594405,
+      "loss": 3.5755,
+      "step": 16100
+    },
+    {
+      "epoch": 4.703011241190518,
+      "grad_norm": 0.317526251077652,
+      "learning_rate": 0.0005438846153846153,
+      "loss": 3.5594,
+      "step": 16150
+    },
+    {
+      "epoch": 4.71757236880424,
+      "grad_norm": 0.32207709550857544,
+      "learning_rate": 0.0005437097902097901,
+      "loss": 3.566,
+      "step": 16200
+    },
+    {
+      "epoch": 4.732133496417963,
+      "grad_norm": 0.32053446769714355,
+      "learning_rate": 0.0005435349650349651,
+      "loss": 3.5819,
+      "step": 16250
+    },
+    {
+      "epoch": 4.746694624031685,
+      "grad_norm": 0.3298093378543854,
+      "learning_rate": 0.0005433601398601397,
+      "loss": 3.576,
+      "step": 16300
+    },
+    {
+      "epoch": 4.7612557516454075,
+      "grad_norm": 0.32744932174682617,
+      "learning_rate": 0.0005431853146853147,
+      "loss": 3.5798,
+      "step": 16350
+    },
+    {
+      "epoch": 4.77581687925913,
+      "grad_norm": 0.3389306664466858,
+      "learning_rate": 0.0005430104895104895,
+      "loss": 3.5866,
+      "step": 16400
+    },
+    {
+      "epoch": 4.790378006872852,
+      "grad_norm": 0.3112584352493286,
+      "learning_rate": 0.0005428356643356643,
+      "loss": 3.5773,
+      "step": 16450
+    },
+    {
+      "epoch": 4.804939134486575,
+      "grad_norm": 0.32686665654182434,
+      "learning_rate": 0.0005426608391608391,
+      "loss": 3.5815,
+      "step": 16500
+    },
+    {
+      "epoch": 4.819500262100297,
+      "grad_norm": 0.3334883749485016,
+      "learning_rate": 0.000542486013986014,
+      "loss": 3.5664,
+      "step": 16550
+    },
+    {
+      "epoch": 4.834061389714019,
+      "grad_norm": 0.3258337378501892,
+      "learning_rate": 0.0005423111888111888,
+      "loss": 3.5747,
+      "step": 16600
+    },
+    {
+      "epoch": 4.848622517327742,
+      "grad_norm": 0.29793718457221985,
+      "learning_rate": 0.0005421363636363636,
+      "loss": 3.5803,
+      "step": 16650
+    },
+    {
+      "epoch": 4.863183644941464,
+      "grad_norm": 0.3122703433036804,
+      "learning_rate": 0.0005419615384615385,
+      "loss": 3.5709,
+      "step": 16700
+    },
+    {
+      "epoch": 4.877744772555187,
+      "grad_norm": 0.31609809398651123,
+      "learning_rate": 0.0005417867132867133,
+      "loss": 3.5709,
+      "step": 16750
+    },
+    {
+      "epoch": 4.892305900168909,
+      "grad_norm": 0.3263167142868042,
+      "learning_rate": 0.0005416118881118881,
+      "loss": 3.5827,
+      "step": 16800
+    },
+    {
+      "epoch": 4.906867027782631,
+      "grad_norm": 0.3147113025188446,
+      "learning_rate": 0.0005414370629370629,
+      "loss": 3.5783,
+      "step": 16850
+    },
+    {
+      "epoch": 4.921428155396354,
+      "grad_norm": 0.3340342938899994,
+      "learning_rate": 0.0005412622377622378,
+      "loss": 3.5829,
+      "step": 16900
+    },
+    {
+      "epoch": 4.935989283010076,
+      "grad_norm": 0.3196568489074707,
+      "learning_rate": 0.0005410874125874126,
+      "loss": 3.5705,
+      "step": 16950
+    },
+    {
+      "epoch": 4.950550410623799,
+      "grad_norm": 0.3239845931529999,
+      "learning_rate": 0.0005409125874125874,
+      "loss": 3.5754,
+      "step": 17000
+    },
+    {
+      "epoch": 4.950550410623799,
+      "eval_accuracy": 0.3598982895346002,
+      "eval_loss": 3.6311697959899902,
+      "eval_runtime": 176.9446,
+      "eval_samples_per_second": 94.075,
+      "eval_steps_per_second": 5.883,
+      "step": 17000
+    },
+    {
+      "epoch": 4.9651115382375215,
+      "grad_norm": 0.3366827368736267,
+      "learning_rate": 0.0005407377622377622,
+      "loss": 3.5755,
+      "step": 17050
+    },
+    {
+      "epoch": 4.979672665851243,
+      "grad_norm": 0.3341779112815857,
+      "learning_rate": 0.000540562937062937,
+      "loss": 3.5782,
+      "step": 17100
+    },
+    {
+      "epoch": 4.994233793464966,
+      "grad_norm": 0.3099096119403839,
+      "learning_rate": 0.0005403881118881118,
+      "loss": 3.5812,
+      "step": 17150
+    },
+    {
+      "epoch": 5.008736676568233,
+      "grad_norm": 0.3530953824520111,
+      "learning_rate": 0.0005402132867132867,
+      "loss": 3.4902,
+      "step": 17200
+    },
+    {
+      "epoch": 5.023297804181956,
+      "grad_norm": 0.3362038731575012,
+      "learning_rate": 0.0005400384615384615,
+      "loss": 3.4618,
+      "step": 17250
+    },
+    {
+      "epoch": 5.037858931795678,
+      "grad_norm": 0.35798731446266174,
+      "learning_rate": 0.0005398636363636363,
+      "loss": 3.4807,
+      "step": 17300
+    },
+    {
+      "epoch": 5.052420059409401,
+      "grad_norm": 0.3367793560028076,
+      "learning_rate": 0.0005396888111888111,
+      "loss": 3.4656,
+      "step": 17350
+    },
+    {
+      "epoch": 5.066981187023123,
+      "grad_norm": 0.3349305987358093,
+      "learning_rate": 0.000539513986013986,
+      "loss": 3.4776,
+      "step": 17400
+    },
+    {
+      "epoch": 5.081542314636845,
+      "grad_norm": 0.31668785214424133,
+      "learning_rate": 0.0005393391608391608,
+      "loss": 3.4841,
+      "step": 17450
+    },
+    {
+      "epoch": 5.096103442250568,
+      "grad_norm": 0.32060810923576355,
+      "learning_rate": 0.0005391643356643356,
+      "loss": 3.4835,
+      "step": 17500
+    },
+    {
+      "epoch": 5.110664569864291,
+      "grad_norm": 0.31662890315055847,
+      "learning_rate": 0.0005389895104895105,
+      "loss": 3.4778,
+      "step": 17550
+    },
+    {
+      "epoch": 5.125225697478013,
+      "grad_norm": 0.3131362795829773,
+      "learning_rate": 0.0005388146853146853,
+      "loss": 3.4767,
+      "step": 17600
+    },
+    {
+      "epoch": 5.139786825091735,
+      "grad_norm": 0.32624542713165283,
+      "learning_rate": 0.0005386398601398601,
+      "loss": 3.4872,
+      "step": 17650
+    },
+    {
+      "epoch": 5.154347952705457,
+      "grad_norm": 0.3383287787437439,
+      "learning_rate": 0.0005384650349650349,
+      "loss": 3.4867,
+      "step": 17700
+    },
+    {
+      "epoch": 5.16890908031918,
+      "grad_norm": 0.3286052644252777,
+      "learning_rate": 0.0005382902097902098,
+      "loss": 3.4852,
+      "step": 17750
+    },
+    {
+      "epoch": 5.183470207932903,
+      "grad_norm": 0.31900569796562195,
+      "learning_rate": 0.0005381153846153845,
+      "loss": 3.4994,
+      "step": 17800
+    },
+    {
+      "epoch": 5.1980313355466246,
+      "grad_norm": 0.3237035572528839,
+      "learning_rate": 0.0005379405594405594,
+      "loss": 3.4957,
+      "step": 17850
+    },
+    {
+      "epoch": 5.212592463160347,
+      "grad_norm": 0.3051255941390991,
+      "learning_rate": 0.0005377657342657342,
+      "loss": 3.5044,
+      "step": 17900
+    },
+    {
+      "epoch": 5.227153590774069,
+      "grad_norm": 0.33518698811531067,
+      "learning_rate": 0.000537590909090909,
+      "loss": 3.4936,
+      "step": 17950
+    },
+    {
+      "epoch": 5.241714718387792,
+      "grad_norm": 0.3128952980041504,
+      "learning_rate": 0.0005374160839160838,
+      "loss": 3.5,
+      "step": 18000
+    },
+    {
+      "epoch": 5.241714718387792,
+      "eval_accuracy": 0.36015245989175304,
+      "eval_loss": 3.6348447799682617,
+      "eval_runtime": 176.8036,
+      "eval_samples_per_second": 94.15,
+      "eval_steps_per_second": 5.888,
+      "step": 18000
+    },
+    {
+      "epoch": 5.256275846001515,
+      "grad_norm": 0.33034732937812805,
+      "learning_rate": 0.0005372412587412587,
+      "loss": 3.509,
+      "step": 18050
+    },
+    {
+      "epoch": 5.2708369736152365,
+      "grad_norm": 0.3423735499382019,
+      "learning_rate": 0.0005370664335664335,
+      "loss": 3.5055,
+      "step": 18100
+    },
+    {
+      "epoch": 5.285398101228959,
+      "grad_norm": 0.3269577920436859,
+      "learning_rate": 0.0005368916083916083,
+      "loss": 3.5036,
+      "step": 18150
+    },
+    {
+      "epoch": 5.299959228842681,
+      "grad_norm": 0.3377140462398529,
+      "learning_rate": 0.0005367167832167832,
+      "loss": 3.5139,
+      "step": 18200
+    },
+    {
+      "epoch": 5.314520356456404,
+      "grad_norm": 0.3444065749645233,
+      "learning_rate": 0.000536541958041958,
+      "loss": 3.5031,
+      "step": 18250
+    },
+    {
+      "epoch": 5.329081484070127,
+      "grad_norm": 0.3035888671875,
+      "learning_rate": 0.0005363671328671328,
+      "loss": 3.4989,
+      "step": 18300
+    },
+    {
+      "epoch": 5.3436426116838485,
+      "grad_norm": 0.3229525089263916,
+      "learning_rate": 0.0005361923076923076,
+      "loss": 3.5053,
+      "step": 18350
+    },
+    {
+      "epoch": 5.358203739297571,
+      "grad_norm": 0.3133704960346222,
+      "learning_rate": 0.0005360174825174825,
+      "loss": 3.5075,
+      "step": 18400
+    },
+    {
+      "epoch": 5.372764866911294,
+      "grad_norm": 0.31471043825149536,
+      "learning_rate": 0.0005358426573426573,
+      "loss": 3.5161,
+      "step": 18450
+    },
+    {
+      "epoch": 5.387325994525016,
+      "grad_norm": 0.3169424533843994,
+      "learning_rate": 0.0005356678321678321,
+      "loss": 3.5031,
+      "step": 18500
+    },
+    {
+      "epoch": 5.401887122138739,
+      "grad_norm": 0.3456338047981262,
+      "learning_rate": 0.0005354930069930069,
+      "loss": 3.5103,
+      "step": 18550
+    },
+    {
+      "epoch": 5.41644824975246,
+      "grad_norm": 0.3145761489868164,
+      "learning_rate": 0.0005353181818181817,
+      "loss": 3.5202,
+      "step": 18600
+    },
+    {
+      "epoch": 5.431009377366183,
+      "grad_norm": 0.3344932198524475,
+      "learning_rate": 0.0005351433566433565,
+      "loss": 3.5115,
+      "step": 18650
+    },
+    {
+      "epoch": 5.445570504979906,
+      "grad_norm": 0.32802441716194153,
+      "learning_rate": 0.0005349685314685314,
+      "loss": 3.5108,
+      "step": 18700
+    },
+    {
+      "epoch": 5.460131632593628,
+      "grad_norm": 0.33433371782302856,
+      "learning_rate": 0.0005347937062937062,
+      "loss": 3.5175,
+      "step": 18750
+    },
+    {
+      "epoch": 5.4746927602073505,
+      "grad_norm": 0.3225354552268982,
+      "learning_rate": 0.000534618881118881,
+      "loss": 3.5294,
+      "step": 18800
+    },
+    {
+      "epoch": 5.489253887821073,
+      "grad_norm": 0.3271482586860657,
+      "learning_rate": 0.0005344440559440559,
+      "loss": 3.515,
+      "step": 18850
+    },
+    {
+      "epoch": 5.503815015434795,
+      "grad_norm": 0.32893088459968567,
+      "learning_rate": 0.0005342692307692307,
+      "loss": 3.517,
+      "step": 18900
+    },
+    {
+      "epoch": 5.518376143048518,
+      "grad_norm": 0.3370042145252228,
+      "learning_rate": 0.0005340944055944055,
+      "loss": 3.5241,
+      "step": 18950
+    },
+    {
+      "epoch": 5.53293727066224,
+      "grad_norm": 0.32803529500961304,
+      "learning_rate": 0.0005339195804195803,
+      "loss": 3.5244,
+      "step": 19000
+    },
+    {
+      "epoch": 5.53293727066224,
+      "eval_accuracy": 0.36118689327407866,
+      "eval_loss": 3.623746395111084,
+      "eval_runtime": 177.0774,
+      "eval_samples_per_second": 94.004,
+      "eval_steps_per_second": 5.879,
+      "step": 19000
+    },
+    {
+      "epoch": 5.5474983982759625,
+      "grad_norm": 0.325192391872406,
+      "learning_rate": 0.0005337447552447552,
+      "loss": 3.5244,
+      "step": 19050
+    },
+    {
+      "epoch": 5.562059525889685,
+      "grad_norm": 0.3339932858943939,
+      "learning_rate": 0.00053356993006993,
+      "loss": 3.5122,
+      "step": 19100
+    },
+    {
+      "epoch": 5.576620653503407,
+      "grad_norm": 0.3150177597999573,
+      "learning_rate": 0.0005333951048951048,
+      "loss": 3.5213,
+      "step": 19150
+    },
+    {
+      "epoch": 5.59118178111713,
+      "grad_norm": 0.3161846697330475,
+      "learning_rate": 0.0005332202797202796,
+      "loss": 3.5178,
+      "step": 19200
+    },
+    {
+      "epoch": 5.605742908730852,
+      "grad_norm": 0.3393925130367279,
+      "learning_rate": 0.0005330454545454546,
+      "loss": 3.5075,
+      "step": 19250
+    },
+    {
+      "epoch": 5.620304036344574,
+      "grad_norm": 0.3091968595981598,
+      "learning_rate": 0.0005328706293706292,
+      "loss": 3.5195,
+      "step": 19300
+    },
+    {
+      "epoch": 5.634865163958297,
+      "grad_norm": 0.3470529317855835,
+      "learning_rate": 0.0005326958041958042,
+      "loss": 3.5365,
+      "step": 19350
+    },
+    {
+      "epoch": 5.649426291572019,
+      "grad_norm": 0.3210427165031433,
+      "learning_rate": 0.000532520979020979,
+      "loss": 3.5194,
+      "step": 19400
+    },
+    {
+      "epoch": 5.663987419185742,
+      "grad_norm": 0.3233715295791626,
+      "learning_rate": 0.0005323461538461538,
+      "loss": 3.5358,
+      "step": 19450
+    },
+    {
+      "epoch": 5.6785485467994645,
+      "grad_norm": 0.3228999376296997,
+      "learning_rate": 0.0005321713286713287,
+      "loss": 3.5195,
+      "step": 19500
+    },
+    {
+      "epoch": 5.693109674413186,
+      "grad_norm": 0.32254138588905334,
+      "learning_rate": 0.0005319965034965035,
+      "loss": 3.5227,
+      "step": 19550
+    },
+    {
+      "epoch": 5.707670802026909,
+      "grad_norm": 0.3360702097415924,
+      "learning_rate": 0.0005318216783216783,
+      "loss": 3.5264,
+      "step": 19600
+    },
+    {
+      "epoch": 5.722231929640631,
+      "grad_norm": 0.32106834650039673,
+      "learning_rate": 0.0005316468531468531,
+      "loss": 3.5294,
+      "step": 19650
+    },
+    {
+      "epoch": 5.736793057254354,
+      "grad_norm": 0.3194294273853302,
+      "learning_rate": 0.000531472027972028,
+      "loss": 3.5303,
+      "step": 19700
+    },
+    {
+      "epoch": 5.7513541848680765,
+      "grad_norm": 0.31817537546157837,
+      "learning_rate": 0.0005312972027972028,
+      "loss": 3.5289,
+      "step": 19750
+    },
+    {
+      "epoch": 5.765915312481798,
+      "grad_norm": 0.32823848724365234,
+      "learning_rate": 0.0005311223776223776,
+      "loss": 3.542,
+      "step": 19800
+    },
+    {
+      "epoch": 5.780476440095521,
+      "grad_norm": 0.33044731616973877,
+      "learning_rate": 0.0005309475524475524,
+      "loss": 3.5308,
+      "step": 19850
+    },
+    {
+      "epoch": 5.795037567709244,
+      "grad_norm": 0.30992332100868225,
+      "learning_rate": 0.0005307727272727273,
+      "loss": 3.5285,
+      "step": 19900
+    },
+    {
+      "epoch": 5.809598695322966,
+      "grad_norm": 0.3125179409980774,
+      "learning_rate": 0.0005305979020979021,
+      "loss": 3.516,
+      "step": 19950
+    },
+    {
+      "epoch": 5.824159822936688,
+      "grad_norm": 0.32367056608200073,
+      "learning_rate": 0.0005304230769230769,
+      "loss": 3.5255,
+      "step": 20000
+    },
+    {
+      "epoch": 5.824159822936688,
+      "eval_accuracy": 0.36226388432027534,
+      "eval_loss": 3.613421678543091,
+      "eval_runtime": 176.9052,
+      "eval_samples_per_second": 94.096,
+      "eval_steps_per_second": 5.885,
+      "step": 20000
+    },
+    {
+      "epoch": 5.83872095055041,
+      "grad_norm": 0.3150468170642853,
+      "learning_rate": 0.0005302482517482517,
+      "loss": 3.5362,
+      "step": 20050
+    },
+    {
+      "epoch": 5.853282078164133,
+      "grad_norm": 0.2990545332431793,
+      "learning_rate": 0.0005300734265734265,
+      "loss": 3.5295,
+      "step": 20100
+    },
+    {
+      "epoch": 5.867843205777856,
+      "grad_norm": 0.29398471117019653,
+      "learning_rate": 0.0005298986013986013,
+      "loss": 3.5227,
+      "step": 20150
+    },
+    {
+      "epoch": 5.882404333391578,
+      "grad_norm": 0.30902978777885437,
+      "learning_rate": 0.0005297237762237762,
+      "loss": 3.5248,
+      "step": 20200
+    },
+    {
+      "epoch": 5.8969654610053,
+      "grad_norm": 0.3113495409488678,
+      "learning_rate": 0.000529548951048951,
+      "loss": 3.538,
+      "step": 20250
+    },
+    {
+      "epoch": 5.911526588619022,
+      "grad_norm": 0.31345126032829285,
+      "learning_rate": 0.0005293741258741258,
+      "loss": 3.5334,
+      "step": 20300
+    },
+    {
+      "epoch": 5.926087716232745,
+      "grad_norm": 0.29870733618736267,
+      "learning_rate": 0.0005291993006993007,
+      "loss": 3.5317,
+      "step": 20350
+    },
+    {
+      "epoch": 5.940648843846468,
+      "grad_norm": 0.32265445590019226,
+      "learning_rate": 0.0005290244755244755,
+      "loss": 3.5114,
+      "step": 20400
+    },
+    {
+      "epoch": 5.95520997146019,
+      "grad_norm": 0.34477177262306213,
+      "learning_rate": 0.0005288496503496503,
+      "loss": 3.5371,
+      "step": 20450
+    },
+    {
+      "epoch": 5.969771099073912,
+      "grad_norm": 0.3217330574989319,
+      "learning_rate": 0.0005286748251748251,
+      "loss": 3.541,
+      "step": 20500
+    },
+    {
+      "epoch": 5.984332226687634,
+      "grad_norm": 0.3138887882232666,
+      "learning_rate": 0.0005285,
+      "loss": 3.5157,
+      "step": 20550
+    },
+    {
+      "epoch": 5.998893354301357,
+      "grad_norm": 0.34451672434806824,
+      "learning_rate": 0.0005283251748251748,
+      "loss": 3.5283,
+      "step": 20600
+    },
+    {
+      "epoch": 6.013396237404625,
+      "grad_norm": 0.3342776894569397,
+      "learning_rate": 0.0005281503496503496,
+      "loss": 3.4197,
+      "step": 20650
+    },
+    {
+      "epoch": 6.027957365018347,
+      "grad_norm": 0.31999471783638,
+      "learning_rate": 0.0005279755244755244,
+      "loss": 3.4206,
+      "step": 20700
+    },
+    {
+      "epoch": 6.04251849263207,
+      "grad_norm": 0.3338138461112976,
+      "learning_rate": 0.0005278006993006993,
+      "loss": 3.4196,
+      "step": 20750
+    },
+    {
+      "epoch": 6.0570796202457915,
+      "grad_norm": 0.3524891436100006,
+      "learning_rate": 0.000527625874125874,
+      "loss": 3.4353,
+      "step": 20800
+    },
+    {
+      "epoch": 6.071640747859514,
+      "grad_norm": 0.3189772367477417,
+      "learning_rate": 0.0005274510489510489,
+      "loss": 3.4282,
+      "step": 20850
+    },
+    {
+      "epoch": 6.086201875473237,
+      "grad_norm": 0.3258211314678192,
+      "learning_rate": 0.0005272762237762238,
+      "loss": 3.4414,
+      "step": 20900
+    },
+    {
+      "epoch": 6.100763003086959,
+      "grad_norm": 0.33525657653808594,
+      "learning_rate": 0.0005271013986013985,
+      "loss": 3.4436,
+      "step": 20950
+    },
+    {
+      "epoch": 6.115324130700682,
+      "grad_norm": 0.3542822599411011,
+      "learning_rate": 0.0005269265734265734,
+      "loss": 3.4384,
+      "step": 21000
+    },
+    {
+      "epoch": 6.115324130700682,
+      "eval_accuracy": 0.3627670522798564,
+      "eval_loss": 3.6149725914001465,
+      "eval_runtime": 176.9568,
+      "eval_samples_per_second": 94.068,
+      "eval_steps_per_second": 5.883,
+      "step": 21000
+    },
+    {
+      "epoch": 6.1298852583144035,
+      "grad_norm": 0.3309166431427002,
+      "learning_rate": 0.0005267517482517482,
+      "loss": 3.4315,
+      "step": 21050
+    },
+    {
+      "epoch": 6.144446385928126,
+      "grad_norm": 0.3391399681568146,
+      "learning_rate": 0.000526576923076923,
+      "loss": 3.4495,
+      "step": 21100
+    },
+    {
+      "epoch": 6.159007513541849,
+      "grad_norm": 0.3149496018886566,
+      "learning_rate": 0.0005264020979020978,
+      "loss": 3.4587,
+      "step": 21150
+    },
+    {
+      "epoch": 6.173568641155571,
+      "grad_norm": 0.32553061842918396,
+      "learning_rate": 0.0005262272727272727,
+      "loss": 3.4494,
+      "step": 21200
+    },
+    {
+      "epoch": 6.1881297687692935,
+      "grad_norm": 0.33138251304626465,
+      "learning_rate": 0.0005260524475524475,
+      "loss": 3.4513,
+      "step": 21250
+    },
+    {
+      "epoch": 6.202690896383016,
+      "grad_norm": 0.32948899269104004,
+      "learning_rate": 0.0005258776223776223,
+      "loss": 3.4544,
+      "step": 21300
+    },
+    {
+      "epoch": 6.217252023996738,
+      "grad_norm": 0.341678261756897,
+      "learning_rate": 0.0005257027972027971,
+      "loss": 3.442,
+      "step": 21350
+    },
+    {
+      "epoch": 6.231813151610461,
+      "grad_norm": 0.33684396743774414,
+      "learning_rate": 0.000525527972027972,
+      "loss": 3.4379,
+      "step": 21400
+    },
+    {
+      "epoch": 6.246374279224183,
+      "grad_norm": 0.34825167059898376,
+      "learning_rate": 0.0005253531468531468,
+      "loss": 3.4584,
+      "step": 21450
+    },
+    {
+      "epoch": 6.2609354068379055,
+      "grad_norm": 0.335224449634552,
+      "learning_rate": 0.0005251783216783216,
+      "loss": 3.4535,
+      "step": 21500
+    },
+    {
+      "epoch": 6.275496534451628,
+      "grad_norm": 0.3297387659549713,
+      "learning_rate": 0.0005250034965034965,
+      "loss": 3.4562,
+      "step": 21550
+    },
+    {
+      "epoch": 6.29005766206535,
+      "grad_norm": 0.3344549238681793,
+      "learning_rate": 0.0005248286713286712,
+      "loss": 3.4579,
+      "step": 21600
+    },
+    {
+      "epoch": 6.304618789679073,
+      "grad_norm": 0.3421954810619354,
+      "learning_rate": 0.0005246538461538461,
+      "loss": 3.4619,
+      "step": 21650
+    },
+    {
+      "epoch": 6.319179917292795,
+      "grad_norm": 0.3300471007823944,
+      "learning_rate": 0.0005244790209790209,
+      "loss": 3.4587,
+      "step": 21700
+    },
+    {
+      "epoch": 6.3337410449065175,
+      "grad_norm": 0.3263568878173828,
+      "learning_rate": 0.0005243041958041957,
+      "loss": 3.4679,
+      "step": 21750
+    },
+    {
+      "epoch": 6.34830217252024,
+      "grad_norm": 0.33946892619132996,
+      "learning_rate": 0.0005241293706293705,
+      "loss": 3.4643,
+      "step": 21800
+    },
+    {
+      "epoch": 6.362863300133962,
+      "grad_norm": 0.3320576250553131,
+      "learning_rate": 0.0005239545454545454,
+      "loss": 3.4587,
+      "step": 21850
+    },
+    {
+      "epoch": 6.377424427747685,
+      "grad_norm": 0.327897846698761,
+      "learning_rate": 0.0005237797202797202,
+      "loss": 3.484,
+      "step": 21900
+    },
+    {
+      "epoch": 6.391985555361408,
+      "grad_norm": 0.32064688205718994,
+      "learning_rate": 0.000523604895104895,
+      "loss": 3.4837,
+      "step": 21950
+    },
+    {
+      "epoch": 6.406546682975129,
+      "grad_norm": 0.3372121751308441,
+      "learning_rate": 0.0005234300699300698,
+      "loss": 3.4611,
+      "step": 22000
+    },
+    {
+      "epoch": 6.406546682975129,
+      "eval_accuracy": 0.36337061870613885,
+      "eval_loss": 3.607774019241333,
+      "eval_runtime": 176.7678,
+      "eval_samples_per_second": 94.169,
+      "eval_steps_per_second": 5.889,
+      "step": 22000
+    },
+    {
+      "epoch": 6.421107810588852,
+      "grad_norm": 0.3300037384033203,
+      "learning_rate": 0.0005232552447552447,
+      "loss": 3.4701,
+      "step": 22050
+    },
+    {
+      "epoch": 6.435668938202574,
+      "grad_norm": 0.35658490657806396,
+      "learning_rate": 0.0005230804195804195,
+      "loss": 3.4581,
+      "step": 22100
+    },
+    {
+      "epoch": 6.450230065816297,
+      "grad_norm": 0.3384239077568054,
+      "learning_rate": 0.0005229055944055943,
+      "loss": 3.4855,
+      "step": 22150
+    },
+    {
+      "epoch": 6.4647911934300195,
+      "grad_norm": 0.3166094720363617,
+      "learning_rate": 0.0005227307692307691,
+      "loss": 3.4814,
+      "step": 22200
+    },
+    {
+      "epoch": 6.479352321043741,
+      "grad_norm": 0.31494995951652527,
+      "learning_rate": 0.0005225559440559441,
+      "loss": 3.4707,
+      "step": 22250
+    },
+    {
+      "epoch": 6.493913448657464,
+      "grad_norm": 0.33063846826553345,
+      "learning_rate": 0.0005223811188811189,
+      "loss": 3.49,
+      "step": 22300
+    },
+    {
+      "epoch": 6.508474576271187,
+      "grad_norm": 0.3317149579524994,
+      "learning_rate": 0.0005222062937062937,
+      "loss": 3.4876,
+      "step": 22350
+    },
+    {
+      "epoch": 6.523035703884909,
+      "grad_norm": 0.3236587941646576,
+      "learning_rate": 0.0005220314685314686,
+      "loss": 3.476,
+      "step": 22400
+    },
+    {
+      "epoch": 6.5375968314986315,
+      "grad_norm": 0.32496577501296997,
+      "learning_rate": 0.0005218566433566433,
+      "loss": 3.4801,
+      "step": 22450
+    },
+    {
+      "epoch": 6.552157959112353,
+      "grad_norm": 0.3040722906589508,
+      "learning_rate": 0.0005216818181818182,
+      "loss": 3.4714,
+      "step": 22500
+    },
+    {
+      "epoch": 6.566719086726076,
+      "grad_norm": 0.3201970160007477,
+      "learning_rate": 0.000521506993006993,
+      "loss": 3.4741,
+      "step": 22550
+    },
+    {
+      "epoch": 6.581280214339799,
+      "grad_norm": 0.3287389576435089,
+      "learning_rate": 0.0005213321678321678,
+      "loss": 3.4773,
+      "step": 22600
+    },
+    {
+      "epoch": 6.595841341953521,
+      "grad_norm": 0.3368239998817444,
+      "learning_rate": 0.0005211573426573426,
+      "loss": 3.4811,
+      "step": 22650
+    },
+    {
+      "epoch": 6.610402469567243,
+      "grad_norm": 0.341665655374527,
+      "learning_rate": 0.0005209825174825175,
+      "loss": 3.4828,
+      "step": 22700
+    },
+    {
+      "epoch": 6.624963597180965,
+      "grad_norm": 0.32684972882270813,
+      "learning_rate": 0.0005208076923076923,
+      "loss": 3.4814,
+      "step": 22750
+    },
+    {
+      "epoch": 6.639524724794688,
+      "grad_norm": 0.32613569498062134,
+      "learning_rate": 0.0005206328671328671,
+      "loss": 3.4808,
+      "step": 22800
+    },
+    {
+      "epoch": 6.654085852408411,
+      "grad_norm": 0.3358040452003479,
+      "learning_rate": 0.0005204580419580419,
+      "loss": 3.4861,
+      "step": 22850
+    },
+    {
+      "epoch": 6.668646980022133,
+      "grad_norm": 0.3069293797016144,
+      "learning_rate": 0.0005202832167832168,
+      "loss": 3.4817,
+      "step": 22900
+    },
+    {
+      "epoch": 6.683208107635855,
+      "grad_norm": 0.3199792504310608,
+      "learning_rate": 0.0005201083916083916,
+      "loss": 3.4666,
+      "step": 22950
+    },
+    {
+      "epoch": 6.697769235249577,
+      "grad_norm": 0.33008405566215515,
+      "learning_rate": 0.0005199335664335664,
+      "loss": 3.5017,
+      "step": 23000
+    },
+    {
+      "epoch": 6.697769235249577,
+      "eval_accuracy": 0.3638195903037183,
+      "eval_loss": 3.6003408432006836,
+      "eval_runtime": 176.8026,
+      "eval_samples_per_second": 94.15,
+      "eval_steps_per_second": 5.888,
+      "step": 23000
+    },
+    {
+      "epoch": 6.7123303628633,
+      "grad_norm": 0.3358388841152191,
+      "learning_rate": 0.0005197587412587413,
+      "loss": 3.4842,
+      "step": 23050
+    },
+    {
+      "epoch": 6.726891490477023,
+      "grad_norm": 0.32142195105552673,
+      "learning_rate": 0.0005195839160839161,
+      "loss": 3.4794,
+      "step": 23100
+    },
+    {
+      "epoch": 6.741452618090745,
+      "grad_norm": 0.3318839371204376,
+      "learning_rate": 0.0005194090909090909,
+      "loss": 3.4883,
+      "step": 23150
+    },
+    {
+      "epoch": 6.756013745704467,
+      "grad_norm": 0.33538463711738586,
+      "learning_rate": 0.0005192342657342657,
+      "loss": 3.4851,
+      "step": 23200
+    },
+    {
+      "epoch": 6.77057487331819,
+      "grad_norm": 0.31580328941345215,
+      "learning_rate": 0.0005190594405594405,
+      "loss": 3.482,
+      "step": 23250
+    },
+    {
+      "epoch": 6.785136000931912,
+      "grad_norm": 0.3154962658882141,
+      "learning_rate": 0.0005188846153846153,
+      "loss": 3.5067,
+      "step": 23300
+    },
+    {
+      "epoch": 6.799697128545635,
+      "grad_norm": 0.32202768325805664,
+      "learning_rate": 0.0005187097902097902,
+      "loss": 3.4836,
+      "step": 23350
+    },
+    {
+      "epoch": 6.814258256159357,
+      "grad_norm": 0.3195551335811615,
+      "learning_rate": 0.000518534965034965,
+      "loss": 3.4966,
+      "step": 23400
+    },
+    {
+      "epoch": 6.828819383773079,
+      "grad_norm": 0.33164918422698975,
+      "learning_rate": 0.0005183601398601398,
+      "loss": 3.4705,
+      "step": 23450
+    },
+    {
+      "epoch": 6.843380511386802,
+      "grad_norm": 0.32325154542922974,
+      "learning_rate": 0.0005181853146853146,
+      "loss": 3.487,
+      "step": 23500
+    },
+    {
+      "epoch": 6.857941639000524,
+      "grad_norm": 0.3257133662700653,
+      "learning_rate": 0.0005180104895104895,
+      "loss": 3.4918,
+      "step": 23550
+    },
+    {
+      "epoch": 6.872502766614247,
+      "grad_norm": 0.33086922764778137,
+      "learning_rate": 0.0005178356643356643,
+      "loss": 3.4844,
+      "step": 23600
+    },
+    {
+      "epoch": 6.887063894227969,
+      "grad_norm": 0.3378217816352844,
+      "learning_rate": 0.0005176608391608391,
+      "loss": 3.4906,
+      "step": 23650
+    },
+    {
+      "epoch": 6.901625021841691,
+      "grad_norm": 0.3248620629310608,
+      "learning_rate": 0.000517486013986014,
+      "loss": 3.4827,
+      "step": 23700
+    },
+    {
+      "epoch": 6.916186149455414,
+      "grad_norm": 0.3082542419433594,
+      "learning_rate": 0.0005173111888111888,
+      "loss": 3.4889,
+      "step": 23750
+    },
+    {
+      "epoch": 6.930747277069136,
+      "grad_norm": 0.326917439699173,
+      "learning_rate": 0.0005171363636363636,
+      "loss": 3.4922,
+      "step": 23800
+    },
+    {
+      "epoch": 6.945308404682859,
+      "grad_norm": 0.32005560398101807,
+      "learning_rate": 0.0005169615384615384,
+      "loss": 3.5049,
+      "step": 23850
+    },
+    {
+      "epoch": 6.959869532296581,
+      "grad_norm": 0.3334341049194336,
+      "learning_rate": 0.0005167867132867133,
+      "loss": 3.492,
+      "step": 23900
+    },
+    {
+      "epoch": 6.974430659910303,
+      "grad_norm": 0.32381048798561096,
+      "learning_rate": 0.000516611888111888,
+      "loss": 3.4998,
+      "step": 23950
+    },
+    {
+      "epoch": 6.988991787524026,
+      "grad_norm": 0.34249672293663025,
+      "learning_rate": 0.0005164370629370629,
+      "loss": 3.4861,
+      "step": 24000
+    },
+    {
+      "epoch": 6.988991787524026,
+      "eval_accuracy": 0.36504106579438345,
+      "eval_loss": 3.589481830596924,
+      "eval_runtime": 176.7844,
+      "eval_samples_per_second": 94.16,
+      "eval_steps_per_second": 5.889,
+      "step": 24000
+    },
+    {
+      "epoch": 7.003494670627293,
+      "grad_norm": 0.34342801570892334,
+      "learning_rate": 0.0005162622377622377,
+      "loss": 3.4622,
+      "step": 24050
+    },
+    {
+      "epoch": 7.018055798241016,
+      "grad_norm": 0.3223758935928345,
+      "learning_rate": 0.0005160874125874125,
+      "loss": 3.3769,
+      "step": 24100
+    },
+    {
+      "epoch": 7.032616925854738,
+      "grad_norm": 0.33334001898765564,
+      "learning_rate": 0.0005159125874125873,
+      "loss": 3.3793,
+      "step": 24150
+    },
+    {
+      "epoch": 7.0471780534684605,
+      "grad_norm": 0.3260599374771118,
+      "learning_rate": 0.0005157377622377622,
+      "loss": 3.3868,
+      "step": 24200
+    },
+    {
+      "epoch": 7.061739181082183,
+      "grad_norm": 0.32783156633377075,
+      "learning_rate": 0.000515562937062937,
+      "loss": 3.3913,
+      "step": 24250
+    },
+    {
+      "epoch": 7.076300308695905,
+      "grad_norm": 0.33441364765167236,
+      "learning_rate": 0.0005153881118881118,
+      "loss": 3.398,
+      "step": 24300
+    },
+    {
+      "epoch": 7.090861436309628,
+      "grad_norm": 0.31968820095062256,
+      "learning_rate": 0.0005152132867132867,
+      "loss": 3.4004,
+      "step": 24350
+    },
+    {
+      "epoch": 7.105422563923351,
+      "grad_norm": 0.3459835648536682,
+      "learning_rate": 0.0005150384615384615,
+      "loss": 3.3953,
+      "step": 24400
+    },
+    {
+      "epoch": 7.1199836915370724,
+      "grad_norm": 0.3527763783931732,
+      "learning_rate": 0.0005148636363636363,
+      "loss": 3.3997,
+      "step": 24450
+    },
+    {
+      "epoch": 7.134544819150795,
+      "grad_norm": 0.33494487404823303,
+      "learning_rate": 0.0005146888111888111,
+      "loss": 3.4178,
+      "step": 24500
+    },
+    {
+      "epoch": 7.149105946764517,
+      "grad_norm": 0.3221123218536377,
+      "learning_rate": 0.000514513986013986,
+      "loss": 3.4155,
+      "step": 24550
+    },
+    {
+      "epoch": 7.16366707437824,
+      "grad_norm": 0.34533846378326416,
+      "learning_rate": 0.0005143391608391608,
+      "loss": 3.4131,
+      "step": 24600
+    },
+    {
+      "epoch": 7.1782282019919625,
+      "grad_norm": 0.34984448552131653,
+      "learning_rate": 0.0005141643356643356,
+      "loss": 3.4036,
+      "step": 24650
+    },
+    {
+      "epoch": 7.192789329605684,
+      "grad_norm": 0.32715174555778503,
+      "learning_rate": 0.0005139895104895104,
+      "loss": 3.4013,
+      "step": 24700
+    },
+    {
+      "epoch": 7.207350457219407,
+      "grad_norm": 0.3404192626476288,
+      "learning_rate": 0.0005138146853146852,
+      "loss": 3.4106,
+      "step": 24750
+    },
+    {
+      "epoch": 7.22191158483313,
+      "grad_norm": 0.31030765175819397,
+      "learning_rate": 0.00051363986013986,
+      "loss": 3.4243,
+      "step": 24800
+    },
+    {
+      "epoch": 7.236472712446852,
+      "grad_norm": 0.3313232958316803,
+      "learning_rate": 0.0005134650349650349,
+      "loss": 3.422,
+      "step": 24850
+    },
+    {
+      "epoch": 7.2510338400605745,
+      "grad_norm": 0.31506332755088806,
+      "learning_rate": 0.0005132902097902097,
+      "loss": 3.4182,
+      "step": 24900
+    },
+    {
+      "epoch": 7.265594967674296,
+      "grad_norm": 0.32670408487319946,
+      "learning_rate": 0.0005131153846153845,
+      "loss": 3.4169,
+      "step": 24950
+    },
+    {
+      "epoch": 7.280156095288019,
+      "grad_norm": 0.3303210437297821,
+      "learning_rate": 0.0005129405594405594,
+      "loss": 3.4213,
+      "step": 25000
+    },
+    {
+      "epoch": 7.280156095288019,
+      "eval_accuracy": 0.3645839823768949,
+      "eval_loss": 3.5995564460754395,
+      "eval_runtime": 177.1172,
+      "eval_samples_per_second": 93.983,
+      "eval_steps_per_second": 5.877,
+      "step": 25000
+    },
+    {
+      "epoch": 7.294717222901742,
+      "grad_norm": 0.3087610900402069,
+      "learning_rate": 0.0005127657342657342,
+      "loss": 3.4324,
+      "step": 25050
+    },
+    {
+      "epoch": 7.309278350515464,
+      "grad_norm": 0.3354821801185608,
+      "learning_rate": 0.000512590909090909,
+      "loss": 3.4259,
+      "step": 25100
+    },
+    {
+      "epoch": 7.3238394781291865,
+      "grad_norm": 0.32773345708847046,
+      "learning_rate": 0.0005124160839160838,
+      "loss": 3.4342,
+      "step": 25150
+    },
+    {
+      "epoch": 7.338400605742908,
+      "grad_norm": 0.3321012854576111,
+      "learning_rate": 0.0005122412587412588,
+      "loss": 3.4347,
+      "step": 25200
+    },
+    {
+      "epoch": 7.352961733356631,
+      "grad_norm": 0.33389997482299805,
+      "learning_rate": 0.0005120664335664336,
+      "loss": 3.4243,
+      "step": 25250
+    },
+    {
+      "epoch": 7.367522860970354,
+      "grad_norm": 0.34130993485450745,
+      "learning_rate": 0.0005118916083916084,
+      "loss": 3.4286,
+      "step": 25300
+    },
+    {
+      "epoch": 7.382083988584076,
+      "grad_norm": 0.32038894295692444,
+      "learning_rate": 0.0005117167832167832,
+      "loss": 3.4303,
+      "step": 25350
+    },
+    {
+      "epoch": 7.396645116197798,
+      "grad_norm": 0.33677011728286743,
+      "learning_rate": 0.0005115419580419581,
+      "loss": 3.4209,
+      "step": 25400
+    },
+    {
+      "epoch": 7.411206243811521,
+      "grad_norm": 0.32208898663520813,
+      "learning_rate": 0.0005113671328671328,
+      "loss": 3.4312,
+      "step": 25450
+    },
+    {
+      "epoch": 7.425767371425243,
+      "grad_norm": 0.3141185939311981,
+      "learning_rate": 0.0005111923076923077,
+      "loss": 3.4408,
+      "step": 25500
+    },
+    {
+      "epoch": 7.440328499038966,
+      "grad_norm": 0.33646243810653687,
+      "learning_rate": 0.0005110174825174825,
+      "loss": 3.4376,
+      "step": 25550
+    },
+    {
+      "epoch": 7.454889626652688,
+      "grad_norm": 0.34247833490371704,
+      "learning_rate": 0.0005108426573426573,
+      "loss": 3.4332,
+      "step": 25600
+    },
+    {
+      "epoch": 7.46945075426641,
+      "grad_norm": 0.31767964363098145,
+      "learning_rate": 0.0005106678321678321,
+      "loss": 3.4316,
+      "step": 25650
+    },
+    {
+      "epoch": 7.484011881880133,
+      "grad_norm": 0.3620831072330475,
+      "learning_rate": 0.000510493006993007,
+      "loss": 3.4366,
+      "step": 25700
+    },
+    {
+      "epoch": 7.498573009493855,
+      "grad_norm": 0.34545835852622986,
+      "learning_rate": 0.0005103181818181818,
+      "loss": 3.4369,
+      "step": 25750
+    },
+    {
+      "epoch": 7.513134137107578,
+      "grad_norm": 0.35401225090026855,
+      "learning_rate": 0.0005101433566433566,
+      "loss": 3.4461,
+      "step": 25800
+    },
+    {
+      "epoch": 7.5276952647213005,
+      "grad_norm": 0.31585973501205444,
+      "learning_rate": 0.0005099685314685315,
+      "loss": 3.4509,
+      "step": 25850
+    },
+    {
+      "epoch": 7.542256392335022,
+      "grad_norm": 0.32004138827323914,
+      "learning_rate": 0.0005097937062937063,
+      "loss": 3.4571,
+      "step": 25900
+    },
+    {
+      "epoch": 7.556817519948745,
+      "grad_norm": 0.34505805373191833,
+      "learning_rate": 0.0005096188811188811,
+      "loss": 3.4596,
+      "step": 25950
+    },
+    {
+      "epoch": 7.571378647562467,
+      "grad_norm": 0.3367854356765747,
+      "learning_rate": 0.0005094440559440559,
+      "loss": 3.4574,
+      "step": 26000
+    },
+    {
+      "epoch": 7.571378647562467,
+      "eval_accuracy": 0.36489834478902566,
+      "eval_loss": 3.5899016857147217,
+      "eval_runtime": 177.0415,
+      "eval_samples_per_second": 94.023,
+      "eval_steps_per_second": 5.88,
+      "step": 26000
+    },
+    {
+      "epoch": 7.58593977517619,
+      "grad_norm": 0.31563594937324524,
+      "learning_rate": 0.0005092692307692308,
+      "loss": 3.4589,
+      "step": 26050
+    },
+    {
+      "epoch": 7.600500902789912,
+      "grad_norm": 0.3145526945590973,
+      "learning_rate": 0.0005090944055944056,
+      "loss": 3.4487,
+      "step": 26100
+    },
+    {
+      "epoch": 7.615062030403634,
+      "grad_norm": 0.3387201130390167,
+      "learning_rate": 0.0005089195804195804,
+      "loss": 3.4584,
+      "step": 26150
+    },
+    {
+      "epoch": 7.629623158017357,
+      "grad_norm": 0.3117615282535553,
+      "learning_rate": 0.0005087447552447552,
+      "loss": 3.4548,
+      "step": 26200
+    },
+    {
+      "epoch": 7.644184285631079,
+      "grad_norm": 0.3377816379070282,
+      "learning_rate": 0.00050856993006993,
+      "loss": 3.456,
+      "step": 26250
+    },
+    {
+      "epoch": 7.658745413244802,
+      "grad_norm": 0.36369189620018005,
+      "learning_rate": 0.0005083951048951048,
+      "loss": 3.4473,
+      "step": 26300
+    },
+    {
+      "epoch": 7.673306540858524,
+      "grad_norm": 0.3553776741027832,
+      "learning_rate": 0.0005082202797202797,
+      "loss": 3.4433,
+      "step": 26350
+    },
+    {
+      "epoch": 7.687867668472246,
+      "grad_norm": 0.33874639868736267,
+      "learning_rate": 0.0005080454545454545,
+      "loss": 3.4423,
+      "step": 26400
+    },
+    {
+      "epoch": 7.702428796085969,
+      "grad_norm": 0.33977967500686646,
+      "learning_rate": 0.0005078706293706293,
+      "loss": 3.4613,
+      "step": 26450
+    },
+    {
+      "epoch": 7.716989923699691,
+      "grad_norm": 0.3333614468574524,
+      "learning_rate": 0.0005076958041958042,
+      "loss": 3.4496,
+      "step": 26500
+    },
+    {
+      "epoch": 7.731551051313414,
+      "grad_norm": 0.3278277814388275,
+      "learning_rate": 0.000507520979020979,
+      "loss": 3.4582,
+      "step": 26550
+    },
+    {
+      "epoch": 7.746112178927136,
+      "grad_norm": 0.33077800273895264,
+      "learning_rate": 0.0005073461538461538,
+      "loss": 3.4525,
+      "step": 26600
+    },
+    {
+      "epoch": 7.760673306540858,
+      "grad_norm": 0.3300637900829315,
+      "learning_rate": 0.0005071713286713286,
+      "loss": 3.4449,
+      "step": 26650
+    },
+    {
+      "epoch": 7.775234434154581,
+      "grad_norm": 0.3329484760761261,
+      "learning_rate": 0.0005069965034965035,
+      "loss": 3.4613,
+      "step": 26700
+    },
+    {
+      "epoch": 7.789795561768304,
+      "grad_norm": 0.35463809967041016,
+      "learning_rate": 0.0005068216783216783,
+      "loss": 3.4637,
+      "step": 26750
+    },
+    {
+      "epoch": 7.8043566893820255,
+      "grad_norm": 0.3434412181377411,
+      "learning_rate": 0.0005066468531468531,
+      "loss": 3.4636,
+      "step": 26800
+    },
+    {
+      "epoch": 7.818917816995748,
+      "grad_norm": 0.33126378059387207,
+      "learning_rate": 0.0005064720279720279,
+      "loss": 3.4617,
+      "step": 26850
+    },
+    {
+      "epoch": 7.833478944609471,
+      "grad_norm": 0.3325732350349426,
+      "learning_rate": 0.0005062972027972028,
+      "loss": 3.4551,
+      "step": 26900
+    },
+    {
+      "epoch": 7.848040072223193,
+      "grad_norm": 0.3155645430088043,
+      "learning_rate": 0.0005061223776223775,
+      "loss": 3.4577,
+      "step": 26950
+    },
+    {
+      "epoch": 7.862601199836916,
+      "grad_norm": 0.3395119607448578,
+      "learning_rate": 0.0005059475524475524,
+      "loss": 3.4488,
+      "step": 27000
+    },
+    {
+      "epoch": 7.862601199836916,
+      "eval_accuracy": 0.36595053012506545,
+      "eval_loss": 3.581655263900757,
+      "eval_runtime": 176.4968,
+      "eval_samples_per_second": 94.313,
+      "eval_steps_per_second": 5.898,
+      "step": 27000
+    },
+    {
+      "epoch": 7.8771623274506375,
+      "grad_norm": 0.3462318480014801,
+      "learning_rate": 0.0005057727272727272,
+      "loss": 3.4536,
+      "step": 27050
+    },
+    {
+      "epoch": 7.89172345506436,
+      "grad_norm": 0.34466618299484253,
+      "learning_rate": 0.000505597902097902,
+      "loss": 3.45,
+      "step": 27100
+    },
+    {
+      "epoch": 7.906284582678083,
+      "grad_norm": 0.3201284110546112,
+      "learning_rate": 0.0005054230769230769,
+      "loss": 3.4584,
+      "step": 27150
+    },
+    {
+      "epoch": 7.920845710291805,
+      "grad_norm": 0.32837069034576416,
+      "learning_rate": 0.0005052482517482517,
+      "loss": 3.4479,
+      "step": 27200
+    },
+    {
+      "epoch": 7.935406837905528,
+      "grad_norm": 0.3236668109893799,
+      "learning_rate": 0.0005050734265734265,
+      "loss": 3.4531,
+      "step": 27250
+    },
+    {
+      "epoch": 7.9499679655192494,
+      "grad_norm": 0.34047573804855347,
+      "learning_rate": 0.0005048986013986013,
+      "loss": 3.4721,
+      "step": 27300
+    },
+    {
+      "epoch": 7.964529093132972,
+      "grad_norm": 0.3239133954048157,
+      "learning_rate": 0.0005047237762237762,
+      "loss": 3.4515,
+      "step": 27350
+    },
+    {
+      "epoch": 7.979090220746695,
+      "grad_norm": 0.32450243830680847,
+      "learning_rate": 0.000504548951048951,
+      "loss": 3.4584,
+      "step": 27400
+    },
+    {
+      "epoch": 7.993651348360417,
+      "grad_norm": 0.3248816132545471,
+      "learning_rate": 0.0005043741258741258,
+      "loss": 3.4567,
+      "step": 27450
+    },
+    {
+      "epoch": 8.008154231463685,
+      "grad_norm": 0.32652345299720764,
+      "learning_rate": 0.0005041993006993006,
+      "loss": 3.401,
+      "step": 27500
+    },
+    {
+      "epoch": 8.022715359077408,
+      "grad_norm": 0.3587798774242401,
+      "learning_rate": 0.0005040244755244755,
+      "loss": 3.3408,
+      "step": 27550
+    },
+    {
+      "epoch": 8.037276486691129,
+      "grad_norm": 0.3377417027950287,
+      "learning_rate": 0.0005038496503496503,
+      "loss": 3.3456,
+      "step": 27600
+    },
+    {
+      "epoch": 8.051837614304851,
+      "grad_norm": 0.3346528708934784,
+      "learning_rate": 0.0005036748251748251,
+      "loss": 3.3501,
+      "step": 27650
+    },
+    {
+      "epoch": 8.066398741918574,
+      "grad_norm": 0.35164663195610046,
+      "learning_rate": 0.0005034999999999999,
+      "loss": 3.3577,
+      "step": 27700
+    },
+    {
+      "epoch": 8.080959869532297,
+      "grad_norm": 0.3102585971355438,
+      "learning_rate": 0.0005033251748251747,
+      "loss": 3.375,
+      "step": 27750
+    },
+    {
+      "epoch": 8.09552099714602,
+      "grad_norm": 0.3394184708595276,
+      "learning_rate": 0.0005031503496503496,
+      "loss": 3.3755,
+      "step": 27800
+    },
+    {
+      "epoch": 8.11008212475974,
+      "grad_norm": 0.32756221294403076,
+      "learning_rate": 0.0005029755244755244,
+      "loss": 3.3746,
+      "step": 27850
+    },
+    {
+      "epoch": 8.124643252373463,
+      "grad_norm": 0.3159530758857727,
+      "learning_rate": 0.0005028006993006992,
+      "loss": 3.3724,
+      "step": 27900
+    },
+    {
+      "epoch": 8.139204379987186,
+      "grad_norm": 0.33860981464385986,
+      "learning_rate": 0.000502625874125874,
+      "loss": 3.3789,
+      "step": 27950
+    },
+    {
+      "epoch": 8.153765507600909,
+      "grad_norm": 0.34003502130508423,
+      "learning_rate": 0.000502451048951049,
+      "loss": 3.3653,
+      "step": 28000
+    },
+    {
+      "epoch": 8.153765507600909,
+      "eval_accuracy": 0.36557009752758785,
+      "eval_loss": 3.589177370071411,
+      "eval_runtime": 176.3386,
+      "eval_samples_per_second": 94.398,
+      "eval_steps_per_second": 5.903,
+      "step": 28000
+    },
+    {
+      "epoch": 8.168326635214632,
+      "grad_norm": 0.351193904876709,
+      "learning_rate": 0.0005022762237762237,
+      "loss": 3.3757,
+      "step": 28050
+    },
+    {
+      "epoch": 8.182887762828354,
+      "grad_norm": 0.3602054715156555,
+      "learning_rate": 0.0005021013986013985,
+      "loss": 3.3774,
+      "step": 28100
+    },
+    {
+      "epoch": 8.197448890442075,
+      "grad_norm": 0.31822383403778076,
+      "learning_rate": 0.0005019265734265733,
+      "loss": 3.3891,
+      "step": 28150
+    },
+    {
+      "epoch": 8.212010018055798,
+      "grad_norm": 0.3160878121852875,
+      "learning_rate": 0.0005017517482517483,
+      "loss": 3.3887,
+      "step": 28200
+    },
+    {
+      "epoch": 8.22657114566952,
+      "grad_norm": 0.3426283895969391,
+      "learning_rate": 0.0005015769230769231,
+      "loss": 3.3857,
+      "step": 28250
+    },
+    {
+      "epoch": 8.241132273283243,
+      "grad_norm": 0.3402179777622223,
+      "learning_rate": 0.0005014020979020979,
+      "loss": 3.3947,
+      "step": 28300
+    },
+    {
+      "epoch": 8.255693400896966,
+      "grad_norm": 0.3560255169868469,
+      "learning_rate": 0.0005012272727272727,
+      "loss": 3.3902,
+      "step": 28350
+    },
+    {
+      "epoch": 8.270254528510687,
+      "grad_norm": 0.3526296317577362,
+      "learning_rate": 0.0005010524475524476,
+      "loss": 3.392,
+      "step": 28400
+    },
+    {
+      "epoch": 8.28481565612441,
+      "grad_norm": 0.3429584503173828,
+      "learning_rate": 0.0005008776223776223,
+      "loss": 3.3907,
+      "step": 28450
+    },
+    {
+      "epoch": 8.299376783738133,
+      "grad_norm": 0.3741019070148468,
+      "learning_rate": 0.0005007027972027972,
+      "loss": 3.3867,
+      "step": 28500
+    },
+    {
+      "epoch": 8.313937911351855,
+      "grad_norm": 0.3399593234062195,
+      "learning_rate": 0.000500527972027972,
+      "loss": 3.4004,
+      "step": 28550
+    },
+    {
+      "epoch": 8.328499038965578,
+      "grad_norm": 0.3139590322971344,
+      "learning_rate": 0.0005003531468531468,
+      "loss": 3.4095,
+      "step": 28600
+    },
+    {
+      "epoch": 8.3430601665793,
+      "grad_norm": 0.3386857807636261,
+      "learning_rate": 0.0005001783216783217,
+      "loss": 3.4005,
+      "step": 28650
+    },
+    {
+      "epoch": 8.357621294193022,
+      "grad_norm": 0.34544748067855835,
+      "learning_rate": 0.0005000034965034965,
+      "loss": 3.4143,
+      "step": 28700
+    },
+    {
+      "epoch": 8.372182421806745,
+      "grad_norm": 0.3371305763721466,
+      "learning_rate": 0.0004998286713286713,
+      "loss": 3.4018,
+      "step": 28750
+    },
+    {
+      "epoch": 8.386743549420467,
+      "grad_norm": 0.3295310437679291,
+      "learning_rate": 0.0004996538461538461,
+      "loss": 3.3987,
+      "step": 28800
+    },
+    {
+      "epoch": 8.40130467703419,
+      "grad_norm": 0.3291172385215759,
+      "learning_rate": 0.000499479020979021,
+      "loss": 3.4058,
+      "step": 28850
+    },
+    {
+      "epoch": 8.415865804647911,
+      "grad_norm": 0.3469097912311554,
+      "learning_rate": 0.0004993041958041958,
+      "loss": 3.4172,
+      "step": 28900
+    },
+    {
+      "epoch": 8.430426932261634,
+      "grad_norm": 0.36806628108024597,
+      "learning_rate": 0.0004991293706293706,
+      "loss": 3.4096,
+      "step": 28950
+    },
+    {
+      "epoch": 8.444988059875357,
+      "grad_norm": 0.3716620206832886,
+      "learning_rate": 0.0004989545454545454,
+      "loss": 3.4157,
+      "step": 29000
+    },
+    {
+      "epoch": 8.444988059875357,
+      "eval_accuracy": 0.36656561768687107,
+      "eval_loss": 3.5810399055480957,
+      "eval_runtime": 176.3981,
+      "eval_samples_per_second": 94.366,
+      "eval_steps_per_second": 5.901,
+      "step": 29000
+    },
+    {
+      "epoch": 8.45954918748908,
+      "grad_norm": 0.35446473956108093,
+      "learning_rate": 0.0004987797202797203,
+      "loss": 3.4093,
+      "step": 29050
+    },
+    {
+      "epoch": 8.474110315102802,
+      "grad_norm": 0.3686777353286743,
+      "learning_rate": 0.0004986048951048951,
+      "loss": 3.4094,
+      "step": 29100
+    },
+    {
+      "epoch": 8.488671442716523,
+      "grad_norm": 0.35613036155700684,
+      "learning_rate": 0.0004984300699300699,
+      "loss": 3.3991,
+      "step": 29150
+    },
+    {
+      "epoch": 8.503232570330246,
+      "grad_norm": 0.32252880930900574,
+      "learning_rate": 0.0004982552447552448,
+      "loss": 3.4118,
+      "step": 29200
+    },
+    {
+      "epoch": 8.517793697943969,
+      "grad_norm": 0.3352912962436676,
+      "learning_rate": 0.0004980804195804195,
+      "loss": 3.4036,
+      "step": 29250
+    },
+    {
+      "epoch": 8.532354825557691,
+      "grad_norm": 0.31796717643737793,
+      "learning_rate": 0.0004979055944055944,
+      "loss": 3.4199,
+      "step": 29300
+    },
+    {
+      "epoch": 8.546915953171414,
+      "grad_norm": 0.32738158106803894,
+      "learning_rate": 0.0004977307692307692,
+      "loss": 3.4043,
+      "step": 29350
+    },
+    {
+      "epoch": 8.561477080785137,
+      "grad_norm": 0.33792996406555176,
+      "learning_rate": 0.000497555944055944,
+      "loss": 3.419,
+      "step": 29400
+    },
+    {
+      "epoch": 8.576038208398858,
+      "grad_norm": 0.3542904555797577,
+      "learning_rate": 0.0004973811188811188,
+      "loss": 3.4204,
+      "step": 29450
+    },
+    {
+      "epoch": 8.59059933601258,
+      "grad_norm": 0.3942735195159912,
+      "learning_rate": 0.0004972062937062937,
+      "loss": 3.4245,
+      "step": 29500
+    },
+    {
+      "epoch": 8.605160463626303,
+      "grad_norm": 0.31929633021354675,
+      "learning_rate": 0.0004970314685314685,
+      "loss": 3.4133,
+      "step": 29550
+    },
+    {
+      "epoch": 8.619721591240026,
+      "grad_norm": 0.3423312306404114,
+      "learning_rate": 0.0004968566433566433,
+      "loss": 3.4228,
+      "step": 29600
+    },
+    {
+      "epoch": 8.634282718853749,
+      "grad_norm": 0.3417450189590454,
+      "learning_rate": 0.0004966818181818181,
+      "loss": 3.4134,
+      "step": 29650
+    },
+    {
+      "epoch": 8.64884384646747,
+      "grad_norm": 0.3385900855064392,
+      "learning_rate": 0.000496506993006993,
+      "loss": 3.4247,
+      "step": 29700
+    },
+    {
+      "epoch": 8.663404974081192,
+      "grad_norm": 0.32018157839775085,
+      "learning_rate": 0.0004963321678321678,
+      "loss": 3.3997,
+      "step": 29750
+    },
+    {
+      "epoch": 8.677966101694915,
+      "grad_norm": 0.34629449248313904,
+      "learning_rate": 0.0004961573426573426,
+      "loss": 3.4207,
+      "step": 29800
+    },
+    {
+      "epoch": 8.692527229308638,
+      "grad_norm": 0.32567837834358215,
+      "learning_rate": 0.0004959825174825175,
+      "loss": 3.4374,
+      "step": 29850
+    },
+    {
+      "epoch": 8.70708835692236,
+      "grad_norm": 0.334450900554657,
+      "learning_rate": 0.0004958076923076923,
+      "loss": 3.4314,
+      "step": 29900
+    },
+    {
+      "epoch": 8.721649484536082,
+      "grad_norm": 0.3887094557285309,
+      "learning_rate": 0.0004956328671328671,
+      "loss": 3.4253,
+      "step": 29950
+    },
+    {
+      "epoch": 8.736210612149804,
+      "grad_norm": 0.3402259051799774,
+      "learning_rate": 0.0004954580419580419,
+      "loss": 3.4305,
+      "step": 30000
+    },
+    {
+      "epoch": 8.736210612149804,
+      "eval_accuracy": 0.36701117996883653,
+      "eval_loss": 3.574934244155884,
+      "eval_runtime": 176.9658,
+      "eval_samples_per_second": 94.063,
+      "eval_steps_per_second": 5.882,
+      "step": 30000
+    },
+    {
+      "epoch": 8.750771739763527,
+      "grad_norm": 0.3436213731765747,
+      "learning_rate": 0.0004952832167832167,
+      "loss": 3.4177,
+      "step": 30050
+    },
+    {
+      "epoch": 8.76533286737725,
+      "grad_norm": 0.328987717628479,
+      "learning_rate": 0.0004951083916083915,
+      "loss": 3.4336,
+      "step": 30100
+    },
+    {
+      "epoch": 8.779893994990973,
+      "grad_norm": 0.3385300934314728,
+      "learning_rate": 0.0004949335664335664,
+      "loss": 3.4331,
+      "step": 30150
+    },
+    {
+      "epoch": 8.794455122604695,
+      "grad_norm": 0.33109623193740845,
+      "learning_rate": 0.0004947587412587412,
+      "loss": 3.4107,
+      "step": 30200
+    },
+    {
+      "epoch": 8.809016250218416,
+      "grad_norm": 0.31443682312965393,
+      "learning_rate": 0.000494583916083916,
+      "loss": 3.43,
+      "step": 30250
+    },
+    {
+      "epoch": 8.82357737783214,
+      "grad_norm": 0.29897189140319824,
+      "learning_rate": 0.0004944090909090908,
+      "loss": 3.444,
+      "step": 30300
+    },
+    {
+      "epoch": 8.838138505445862,
+      "grad_norm": 0.34077444672584534,
+      "learning_rate": 0.0004942342657342657,
+      "loss": 3.4399,
+      "step": 30350
+    },
+    {
+      "epoch": 8.852699633059585,
+      "grad_norm": 0.3268566131591797,
+      "learning_rate": 0.0004940594405594405,
+      "loss": 3.4268,
+      "step": 30400
+    },
+    {
+      "epoch": 8.867260760673307,
+      "grad_norm": 0.32913246750831604,
+      "learning_rate": 0.0004938846153846153,
+      "loss": 3.433,
+      "step": 30450
+    },
+    {
+      "epoch": 8.881821888287028,
+      "grad_norm": 0.31822359561920166,
+      "learning_rate": 0.0004937097902097901,
+      "loss": 3.4374,
+      "step": 30500
+    },
+    {
+      "epoch": 8.896383015900751,
+      "grad_norm": 0.34571900963783264,
+      "learning_rate": 0.000493534965034965,
+      "loss": 3.4361,
+      "step": 30550
+    },
+    {
+      "epoch": 8.910944143514474,
+      "grad_norm": 0.3010501563549042,
+      "learning_rate": 0.0004933601398601398,
+      "loss": 3.4355,
+      "step": 30600
+    },
+    {
+      "epoch": 8.925505271128197,
+      "grad_norm": 0.3567883372306824,
+      "learning_rate": 0.0004931853146853146,
+      "loss": 3.4244,
+      "step": 30650
+    },
+    {
+      "epoch": 8.94006639874192,
+      "grad_norm": 0.32925522327423096,
+      "learning_rate": 0.0004930104895104895,
+      "loss": 3.4203,
+      "step": 30700
+    },
+    {
+      "epoch": 8.95462752635564,
+      "grad_norm": 0.33393311500549316,
+      "learning_rate": 0.0004928356643356642,
+      "loss": 3.4295,
+      "step": 30750
+    },
+    {
+      "epoch": 8.969188653969363,
+      "grad_norm": 0.34621018171310425,
+      "learning_rate": 0.0004926608391608391,
+      "loss": 3.437,
+      "step": 30800
+    },
+    {
+      "epoch": 8.983749781583086,
+      "grad_norm": 0.359050989151001,
+      "learning_rate": 0.0004924860139860139,
+      "loss": 3.4432,
+      "step": 30850
+    },
+    {
+      "epoch": 8.998310909196809,
+      "grad_norm": 0.329822838306427,
+      "learning_rate": 0.0004923111888111887,
+      "loss": 3.4229,
+      "step": 30900
+    },
+    {
+      "epoch": 9.012813792300076,
+      "grad_norm": 0.34674400091171265,
+      "learning_rate": 0.0004921363636363635,
+      "loss": 3.3361,
+      "step": 30950
+    },
+    {
+      "epoch": 9.027374919913798,
+      "grad_norm": 0.33987441658973694,
+      "learning_rate": 0.0004919615384615384,
+      "loss": 3.3188,
+      "step": 31000
+    },
+    {
+      "epoch": 9.027374919913798,
+      "eval_accuracy": 0.3674608569420602,
+      "eval_loss": 3.5798559188842773,
+      "eval_runtime": 176.6125,
+      "eval_samples_per_second": 94.252,
+      "eval_steps_per_second": 5.894,
+      "step": 31000
+    },
+    {
+      "epoch": 9.041936047527521,
+      "grad_norm": 0.33782774209976196,
+      "learning_rate": 0.0004917867132867132,
+      "loss": 3.327,
+      "step": 31050
+    },
+    {
+      "epoch": 9.056497175141242,
+      "grad_norm": 0.3608669340610504,
+      "learning_rate": 0.000491611888111888,
+      "loss": 3.3257,
+      "step": 31100
+    },
+    {
+      "epoch": 9.071058302754965,
+      "grad_norm": 0.33250337839126587,
+      "learning_rate": 0.0004914370629370628,
+      "loss": 3.3302,
+      "step": 31150
+    },
+    {
+      "epoch": 9.085619430368688,
+      "grad_norm": 0.380667120218277,
+      "learning_rate": 0.0004912622377622378,
+      "loss": 3.3367,
+      "step": 31200
+    },
+    {
+      "epoch": 9.10018055798241,
+      "grad_norm": 0.3177809715270996,
+      "learning_rate": 0.0004910874125874126,
+      "loss": 3.341,
+      "step": 31250
+    },
+    {
+      "epoch": 9.114741685596133,
+      "grad_norm": 0.3426327109336853,
+      "learning_rate": 0.0004909125874125874,
+      "loss": 3.3328,
+      "step": 31300
+    },
+    {
+      "epoch": 9.129302813209854,
+      "grad_norm": 0.3812158703804016,
+      "learning_rate": 0.0004907377622377623,
+      "loss": 3.3401,
+      "step": 31350
+    },
+    {
+      "epoch": 9.143863940823577,
+      "grad_norm": 0.3903525471687317,
+      "learning_rate": 0.0004905629370629371,
+      "loss": 3.358,
+      "step": 31400
+    },
+    {
+      "epoch": 9.1584250684373,
+      "grad_norm": 0.3273092210292816,
+      "learning_rate": 0.0004903881118881119,
+      "loss": 3.3457,
+      "step": 31450
+    },
+    {
+      "epoch": 9.172986196051022,
+      "grad_norm": 0.34901750087738037,
+      "learning_rate": 0.0004902132867132867,
+      "loss": 3.3489,
+      "step": 31500
+    },
+    {
+      "epoch": 9.187547323664745,
+      "grad_norm": 0.3518262803554535,
+      "learning_rate": 0.0004900384615384615,
+      "loss": 3.3524,
+      "step": 31550
+    },
+    {
+      "epoch": 9.202108451278466,
+      "grad_norm": 0.34695637226104736,
+      "learning_rate": 0.0004898636363636363,
+      "loss": 3.3472,
+      "step": 31600
+    },
+    {
+      "epoch": 9.216669578892189,
+      "grad_norm": 0.330387145280838,
+      "learning_rate": 0.0004896888111888112,
+      "loss": 3.3538,
+      "step": 31650
+    },
+    {
+      "epoch": 9.231230706505912,
+      "grad_norm": 0.36280760169029236,
+      "learning_rate": 0.000489513986013986,
+      "loss": 3.3641,
+      "step": 31700
+    },
+    {
+      "epoch": 9.245791834119634,
+      "grad_norm": 0.3654526472091675,
+      "learning_rate": 0.0004893391608391608,
+      "loss": 3.3713,
+      "step": 31750
+    },
+    {
+      "epoch": 9.260352961733357,
+      "grad_norm": 0.3399411737918854,
+      "learning_rate": 0.0004891643356643356,
+      "loss": 3.3678,
+      "step": 31800
+    },
+    {
+      "epoch": 9.27491408934708,
+      "grad_norm": 0.3387416899204254,
+      "learning_rate": 0.0004889895104895105,
+      "loss": 3.3738,
+      "step": 31850
+    },
+    {
+      "epoch": 9.2894752169608,
+      "grad_norm": 0.3442247807979584,
+      "learning_rate": 0.0004888146853146853,
+      "loss": 3.3624,
+      "step": 31900
+    },
+    {
+      "epoch": 9.304036344574524,
+      "grad_norm": 0.3599652647972107,
+      "learning_rate": 0.0004886398601398601,
+      "loss": 3.3695,
+      "step": 31950
+    },
+    {
+      "epoch": 9.318597472188246,
+      "grad_norm": 0.33344852924346924,
+      "learning_rate": 0.000488465034965035,
+      "loss": 3.367,
+      "step": 32000
+    },
+    {
+      "epoch": 9.318597472188246,
+      "eval_accuracy": 0.3671916385711629,
+      "eval_loss": 3.579378604888916,
+      "eval_runtime": 176.8517,
+      "eval_samples_per_second": 94.124,
+      "eval_steps_per_second": 5.886,
+      "step": 32000
+    },
+    {
+      "epoch": 9.333158599801969,
+      "grad_norm": 0.36125627160072327,
+      "learning_rate": 0.0004882902097902098,
+      "loss": 3.3745,
+      "step": 32050
+    },
+    {
+      "epoch": 9.347719727415692,
+      "grad_norm": 0.3284973204135895,
+      "learning_rate": 0.0004881153846153846,
+      "loss": 3.3784,
+      "step": 32100
+    },
+    {
+      "epoch": 9.362280855029413,
+      "grad_norm": 0.3623260259628296,
+      "learning_rate": 0.0004879405594405594,
+      "loss": 3.3884,
+      "step": 32150
+    },
+    {
+      "epoch": 9.376841982643136,
+      "grad_norm": 0.35829871892929077,
+      "learning_rate": 0.00048776573426573424,
+      "loss": 3.3828,
+      "step": 32200
+    },
+    {
+      "epoch": 9.391403110256858,
+      "grad_norm": 0.35492923855781555,
+      "learning_rate": 0.00048759090909090904,
+      "loss": 3.389,
+      "step": 32250
+    },
+    {
+      "epoch": 9.405964237870581,
+      "grad_norm": 0.38198089599609375,
+      "learning_rate": 0.0004874160839160839,
+      "loss": 3.3864,
+      "step": 32300
+    },
+    {
+      "epoch": 9.420525365484304,
+      "grad_norm": 0.3594827651977539,
+      "learning_rate": 0.0004872412587412587,
+      "loss": 3.3896,
+      "step": 32350
+    },
+    {
+      "epoch": 9.435086493098025,
+      "grad_norm": 0.3426271975040436,
+      "learning_rate": 0.00048706643356643354,
+      "loss": 3.3825,
+      "step": 32400
+    },
+    {
+      "epoch": 9.449647620711747,
+      "grad_norm": 0.3374452590942383,
+      "learning_rate": 0.00048689160839160834,
+      "loss": 3.3773,
+      "step": 32450
+    },
+    {
+      "epoch": 9.46420874832547,
+      "grad_norm": 0.3500727415084839,
+      "learning_rate": 0.0004867167832167832,
+      "loss": 3.3978,
+      "step": 32500
+    },
+    {
+      "epoch": 9.478769875939193,
+      "grad_norm": 0.3206511437892914,
+      "learning_rate": 0.00048654195804195794,
+      "loss": 3.3935,
+      "step": 32550
+    },
+    {
+      "epoch": 9.493331003552916,
+      "grad_norm": 0.3213861882686615,
+      "learning_rate": 0.00048636713286713285,
+      "loss": 3.3972,
+      "step": 32600
+    },
+    {
+      "epoch": 9.507892131166638,
+      "grad_norm": 0.3542587459087372,
+      "learning_rate": 0.0004861923076923077,
+      "loss": 3.3891,
+      "step": 32650
+    },
+    {
+      "epoch": 9.52245325878036,
+      "grad_norm": 0.35178741812705994,
+      "learning_rate": 0.00048601748251748245,
+      "loss": 3.3902,
+      "step": 32700
+    },
+    {
+      "epoch": 9.537014386394082,
+      "grad_norm": 0.3213842511177063,
+      "learning_rate": 0.0004858426573426573,
+      "loss": 3.3894,
+      "step": 32750
+    },
+    {
+      "epoch": 9.551575514007805,
+      "grad_norm": 0.3286217451095581,
+      "learning_rate": 0.0004856678321678321,
+      "loss": 3.3926,
+      "step": 32800
+    },
+    {
+      "epoch": 9.566136641621528,
+      "grad_norm": 0.3548128008842468,
+      "learning_rate": 0.00048549300699300696,
+      "loss": 3.3883,
+      "step": 32850
+    },
+    {
+      "epoch": 9.58069776923525,
+      "grad_norm": 0.3568916618824005,
+      "learning_rate": 0.00048531818181818176,
+      "loss": 3.4017,
+      "step": 32900
+    },
+    {
+      "epoch": 9.595258896848971,
+      "grad_norm": 0.3281558156013489,
+      "learning_rate": 0.0004851433566433566,
+      "loss": 3.4038,
+      "step": 32950
+    },
+    {
+      "epoch": 9.609820024462694,
+      "grad_norm": 0.35022494196891785,
+      "learning_rate": 0.0004849685314685314,
+      "loss": 3.4041,
+      "step": 33000
+    },
+    {
+      "epoch": 9.609820024462694,
+      "eval_accuracy": 0.3675565529044665,
+      "eval_loss": 3.573331594467163,
+      "eval_runtime": 177.4614,
+      "eval_samples_per_second": 93.801,
+      "eval_steps_per_second": 5.866,
+      "step": 33000
+    },
+    {
+      "epoch": 9.624381152076417,
+      "grad_norm": 0.34615230560302734,
+      "learning_rate": 0.00048479370629370627,
+      "loss": 3.3877,
+      "step": 33050
+    },
+    {
+      "epoch": 9.63894227969014,
+      "grad_norm": 0.3371587097644806,
+      "learning_rate": 0.00048461888111888106,
+      "loss": 3.3997,
+      "step": 33100
+    },
+    {
+      "epoch": 9.653503407303862,
+      "grad_norm": 0.32450369000434875,
+      "learning_rate": 0.0004844440559440559,
+      "loss": 3.3985,
+      "step": 33150
+    },
+    {
+      "epoch": 9.668064534917583,
+      "grad_norm": 0.3822547495365143,
+      "learning_rate": 0.0004842692307692307,
+      "loss": 3.396,
+      "step": 33200
+    },
+    {
+      "epoch": 9.682625662531306,
+      "grad_norm": 0.3404769003391266,
+      "learning_rate": 0.00048409440559440557,
+      "loss": 3.4074,
+      "step": 33250
+    },
+    {
+      "epoch": 9.697186790145029,
+      "grad_norm": 0.35298749804496765,
+      "learning_rate": 0.0004839195804195803,
+      "loss": 3.4048,
+      "step": 33300
+    },
+    {
+      "epoch": 9.711747917758752,
+      "grad_norm": 0.346872478723526,
+      "learning_rate": 0.0004837447552447552,
+      "loss": 3.3976,
+      "step": 33350
+    },
+    {
+      "epoch": 9.726309045372474,
+      "grad_norm": 0.32681095600128174,
+      "learning_rate": 0.0004835699300699301,
+      "loss": 3.4035,
+      "step": 33400
+    },
+    {
+      "epoch": 9.740870172986195,
+      "grad_norm": 0.3425065279006958,
+      "learning_rate": 0.0004833951048951048,
+      "loss": 3.3915,
+      "step": 33450
+    },
+    {
+      "epoch": 9.755431300599918,
+      "grad_norm": 0.33771848678588867,
+      "learning_rate": 0.0004832202797202797,
+      "loss": 3.3985,
+      "step": 33500
+    },
+    {
+      "epoch": 9.76999242821364,
+      "grad_norm": 0.3263910412788391,
+      "learning_rate": 0.0004830454545454545,
+      "loss": 3.4014,
+      "step": 33550
+    },
+    {
+      "epoch": 9.784553555827364,
+      "grad_norm": 0.332059770822525,
+      "learning_rate": 0.00048287062937062933,
+      "loss": 3.3994,
+      "step": 33600
+    },
+    {
+      "epoch": 9.799114683441086,
+      "grad_norm": 0.35489657521247864,
+      "learning_rate": 0.00048269580419580413,
+      "loss": 3.4078,
+      "step": 33650
+    },
+    {
+      "epoch": 9.813675811054807,
+      "grad_norm": 0.35381078720092773,
+      "learning_rate": 0.000482520979020979,
+      "loss": 3.4005,
+      "step": 33700
+    },
+    {
+      "epoch": 9.82823693866853,
+      "grad_norm": 0.324863463640213,
+      "learning_rate": 0.0004823461538461538,
+      "loss": 3.3952,
+      "step": 33750
+    },
+    {
+      "epoch": 9.842798066282253,
+      "grad_norm": 0.3365383446216583,
+      "learning_rate": 0.00048217132867132864,
+      "loss": 3.4049,
+      "step": 33800
+    },
+    {
+      "epoch": 9.857359193895975,
+      "grad_norm": 0.34103065729141235,
+      "learning_rate": 0.00048199650349650344,
+      "loss": 3.4089,
+      "step": 33850
+    },
+    {
+      "epoch": 9.871920321509698,
+      "grad_norm": 0.3343670070171356,
+      "learning_rate": 0.0004818216783216783,
+      "loss": 3.4227,
+      "step": 33900
+    },
+    {
+      "epoch": 9.88648144912342,
+      "grad_norm": 0.32671719789505005,
+      "learning_rate": 0.0004816468531468531,
+      "loss": 3.4102,
+      "step": 33950
+    },
+    {
+      "epoch": 9.901042576737142,
+      "grad_norm": 0.36884456872940063,
+      "learning_rate": 0.00048147202797202795,
+      "loss": 3.4055,
+      "step": 34000
+    },
+    {
+      "epoch": 9.901042576737142,
+      "eval_accuracy": 0.368497524013926,
+      "eval_loss": 3.564042329788208,
+      "eval_runtime": 176.835,
+      "eval_samples_per_second": 94.133,
+      "eval_steps_per_second": 5.887,
+      "step": 34000
+    },
+    {
+      "epoch": 9.915603704350865,
+      "grad_norm": 0.3321670591831207,
+      "learning_rate": 0.0004812972027972028,
+      "loss": 3.3949,
+      "step": 34050
+    },
+    {
+      "epoch": 9.930164831964587,
+      "grad_norm": 0.36186736822128296,
+      "learning_rate": 0.0004811223776223776,
+      "loss": 3.4091,
+      "step": 34100
+    },
+    {
+      "epoch": 9.94472595957831,
+      "grad_norm": 0.3428935408592224,
+      "learning_rate": 0.00048094755244755245,
+      "loss": 3.4074,
+      "step": 34150
+    },
+    {
+      "epoch": 9.959287087192033,
+      "grad_norm": 0.35069432854652405,
+      "learning_rate": 0.0004807727272727272,
+      "loss": 3.3979,
+      "step": 34200
+    },
+    {
+      "epoch": 9.973848214805754,
+      "grad_norm": 0.32886582612991333,
+      "learning_rate": 0.00048059790209790205,
+      "loss": 3.4107,
+      "step": 34250
+    },
+    {
+      "epoch": 9.988409342419477,
+      "grad_norm": 0.3682115972042084,
+      "learning_rate": 0.00048042307692307685,
+      "loss": 3.4094,
+      "step": 34300
+    },
+    {
+      "epoch": 10.002912225522744,
+      "grad_norm": 0.35744708776474,
+      "learning_rate": 0.0004802482517482517,
+      "loss": 3.3896,
+      "step": 34350
+    },
+    {
+      "epoch": 10.017473353136467,
+      "grad_norm": 0.35240817070007324,
+      "learning_rate": 0.0004800734265734265,
+      "loss": 3.3044,
+      "step": 34400
+    },
+    {
+      "epoch": 10.03203448075019,
+      "grad_norm": 0.34491005539894104,
+      "learning_rate": 0.00047989860139860136,
+      "loss": 3.2982,
+      "step": 34450
+    },
+    {
+      "epoch": 10.046595608363912,
+      "grad_norm": 0.33637934923171997,
+      "learning_rate": 0.00047972377622377616,
+      "loss": 3.3106,
+      "step": 34500
+    },
+    {
+      "epoch": 10.061156735977635,
+      "grad_norm": 0.3516424894332886,
+      "learning_rate": 0.000479548951048951,
+      "loss": 3.3193,
+      "step": 34550
+    },
+    {
+      "epoch": 10.075717863591356,
+      "grad_norm": 0.3707565665245056,
+      "learning_rate": 0.0004793741258741258,
+      "loss": 3.3098,
+      "step": 34600
+    },
+    {
+      "epoch": 10.090278991205079,
+      "grad_norm": 0.33687910437583923,
+      "learning_rate": 0.00047919930069930067,
+      "loss": 3.3132,
+      "step": 34650
+    },
+    {
+      "epoch": 10.104840118818801,
+      "grad_norm": 0.353137731552124,
+      "learning_rate": 0.0004790244755244755,
+      "loss": 3.3231,
+      "step": 34700
+    },
+    {
+      "epoch": 10.119401246432524,
+      "grad_norm": 0.36714377999305725,
+      "learning_rate": 0.0004788496503496503,
+      "loss": 3.324,
+      "step": 34750
+    },
+    {
+      "epoch": 10.133962374046247,
+      "grad_norm": 0.32502633333206177,
+      "learning_rate": 0.0004786748251748252,
+      "loss": 3.3239,
+      "step": 34800
+    },
+    {
+      "epoch": 10.148523501659968,
+      "grad_norm": 0.33693206310272217,
+      "learning_rate": 0.0004785,
+      "loss": 3.3299,
+      "step": 34850
+    },
+    {
+      "epoch": 10.16308462927369,
+      "grad_norm": 0.3602798581123352,
+      "learning_rate": 0.00047832517482517483,
+      "loss": 3.3361,
+      "step": 34900
+    },
+    {
+      "epoch": 10.177645756887413,
+      "grad_norm": 0.3324842154979706,
+      "learning_rate": 0.0004781503496503496,
+      "loss": 3.3397,
+      "step": 34950
+    },
+    {
+      "epoch": 10.192206884501136,
+      "grad_norm": 0.338860422372818,
+      "learning_rate": 0.00047797552447552443,
+      "loss": 3.3366,
+      "step": 35000
+    },
+    {
+      "epoch": 10.192206884501136,
+      "eval_accuracy": 0.3681085093461097,
+      "eval_loss": 3.572119951248169,
+      "eval_runtime": 176.7411,
+      "eval_samples_per_second": 94.183,
+      "eval_steps_per_second": 5.89,
+      "step": 35000
+    },
+    {
+      "epoch": 10.206768012114859,
+      "grad_norm": 0.37433600425720215,
+      "learning_rate": 0.00047780069930069923,
+      "loss": 3.3329,
+      "step": 35050
+    },
+    {
+      "epoch": 10.221329139728581,
+      "grad_norm": 0.3410218060016632,
+      "learning_rate": 0.0004776258741258741,
+      "loss": 3.3312,
+      "step": 35100
+    },
+    {
+      "epoch": 10.235890267342302,
+      "grad_norm": 0.33754661679267883,
+      "learning_rate": 0.0004774510489510489,
+      "loss": 3.3364,
+      "step": 35150
+    },
+    {
+      "epoch": 10.250451394956025,
+      "grad_norm": 0.3520354628562927,
+      "learning_rate": 0.00047727622377622374,
+      "loss": 3.3463,
+      "step": 35200
+    },
+    {
+      "epoch": 10.265012522569748,
+      "grad_norm": 0.3703749179840088,
+      "learning_rate": 0.00047710139860139854,
+      "loss": 3.352,
+      "step": 35250
+    },
+    {
+      "epoch": 10.27957365018347,
+      "grad_norm": 0.3496933877468109,
+      "learning_rate": 0.0004769265734265734,
+      "loss": 3.3469,
+      "step": 35300
+    },
+    {
+      "epoch": 10.294134777797193,
+      "grad_norm": 0.3573146164417267,
+      "learning_rate": 0.0004767517482517482,
+      "loss": 3.3455,
+      "step": 35350
+    },
+    {
+      "epoch": 10.308695905410914,
+      "grad_norm": 0.34615665674209595,
+      "learning_rate": 0.00047657692307692304,
+      "loss": 3.3578,
+      "step": 35400
+    },
+    {
+      "epoch": 10.323257033024637,
+      "grad_norm": 0.3539220690727234,
+      "learning_rate": 0.0004764020979020979,
+      "loss": 3.3548,
+      "step": 35450
+    },
+    {
+      "epoch": 10.33781816063836,
+      "grad_norm": 0.3664166033267975,
+      "learning_rate": 0.0004762272727272727,
+      "loss": 3.3533,
+      "step": 35500
+    },
+    {
+      "epoch": 10.352379288252083,
+      "grad_norm": 0.35653188824653625,
+      "learning_rate": 0.00047605244755244755,
+      "loss": 3.3557,
+      "step": 35550
+    },
+    {
+      "epoch": 10.366940415865805,
+      "grad_norm": 0.3385823965072632,
+      "learning_rate": 0.00047587762237762235,
+      "loss": 3.3517,
+      "step": 35600
+    },
+    {
+      "epoch": 10.381501543479526,
+      "grad_norm": 0.36101484298706055,
+      "learning_rate": 0.0004757027972027972,
+      "loss": 3.3607,
+      "step": 35650
+    },
+    {
+      "epoch": 10.396062671093249,
+      "grad_norm": 0.34660804271698,
+      "learning_rate": 0.00047552797202797195,
+      "loss": 3.3548,
+      "step": 35700
+    },
+    {
+      "epoch": 10.410623798706972,
+      "grad_norm": 0.3321472704410553,
+      "learning_rate": 0.0004753531468531468,
+      "loss": 3.3513,
+      "step": 35750
+    },
+    {
+      "epoch": 10.425184926320695,
+      "grad_norm": 0.3524532616138458,
+      "learning_rate": 0.0004751783216783216,
+      "loss": 3.3542,
+      "step": 35800
+    },
+    {
+      "epoch": 10.439746053934417,
+      "grad_norm": 0.33926790952682495,
+      "learning_rate": 0.00047500349650349646,
+      "loss": 3.3685,
+      "step": 35850
+    },
+    {
+      "epoch": 10.454307181548138,
+      "grad_norm": 0.3707771599292755,
+      "learning_rate": 0.00047482867132867126,
+      "loss": 3.3754,
+      "step": 35900
+    },
+    {
+      "epoch": 10.468868309161861,
+      "grad_norm": 0.36390265822410583,
+      "learning_rate": 0.0004746538461538461,
+      "loss": 3.3671,
+      "step": 35950
+    },
+    {
+      "epoch": 10.483429436775584,
+      "grad_norm": 0.32835954427719116,
+      "learning_rate": 0.0004744790209790209,
+      "loss": 3.3638,
+      "step": 36000
+    },
+    {
+      "epoch": 10.483429436775584,
+      "eval_accuracy": 0.36857488020958123,
+      "eval_loss": 3.5680227279663086,
+      "eval_runtime": 176.8143,
+      "eval_samples_per_second": 94.144,
+      "eval_steps_per_second": 5.888,
+      "step": 36000
+    },
+    {
+      "epoch": 10.497990564389307,
+      "grad_norm": 0.36081641912460327,
+      "learning_rate": 0.00047430419580419576,
+      "loss": 3.3636,
+      "step": 36050
+    },
+    {
+      "epoch": 10.51255169200303,
+      "grad_norm": 0.3701050579547882,
+      "learning_rate": 0.0004741293706293706,
+      "loss": 3.3462,
+      "step": 36100
+    },
+    {
+      "epoch": 10.52711281961675,
+      "grad_norm": 0.3523276150226593,
+      "learning_rate": 0.0004739545454545454,
+      "loss": 3.3713,
+      "step": 36150
+    },
+    {
+      "epoch": 10.541673947230473,
+      "grad_norm": 0.34822702407836914,
+      "learning_rate": 0.00047377972027972027,
+      "loss": 3.3808,
+      "step": 36200
+    },
+    {
+      "epoch": 10.556235074844196,
+      "grad_norm": 0.34755954146385193,
+      "learning_rate": 0.00047360489510489507,
+      "loss": 3.3718,
+      "step": 36250
+    },
+    {
+      "epoch": 10.570796202457919,
+      "grad_norm": 0.3505362868309021,
+      "learning_rate": 0.0004734300699300699,
+      "loss": 3.3684,
+      "step": 36300
+    },
+    {
+      "epoch": 10.585357330071641,
+      "grad_norm": 0.36556077003479004,
+      "learning_rate": 0.0004732552447552447,
+      "loss": 3.3585,
+      "step": 36350
+    },
+    {
+      "epoch": 10.599918457685362,
+      "grad_norm": 0.36185914278030396,
+      "learning_rate": 0.0004730804195804196,
+      "loss": 3.3745,
+      "step": 36400
+    },
+    {
+      "epoch": 10.614479585299085,
+      "grad_norm": 0.3472917973995209,
+      "learning_rate": 0.0004729055944055943,
+      "loss": 3.369,
+      "step": 36450
+    },
+    {
+      "epoch": 10.629040712912808,
+      "grad_norm": 0.3542565405368805,
+      "learning_rate": 0.0004727307692307692,
+      "loss": 3.377,
+      "step": 36500
+    },
+    {
+      "epoch": 10.64360184052653,
+      "grad_norm": 0.3553348779678345,
+      "learning_rate": 0.000472555944055944,
+      "loss": 3.3614,
+      "step": 36550
+    },
+    {
+      "epoch": 10.658162968140253,
+      "grad_norm": 0.3704865872859955,
+      "learning_rate": 0.00047238111888111883,
+      "loss": 3.3763,
+      "step": 36600
+    },
+    {
+      "epoch": 10.672724095753976,
+      "grad_norm": 0.3559943437576294,
+      "learning_rate": 0.00047220629370629363,
+      "loss": 3.3715,
+      "step": 36650
+    },
+    {
+      "epoch": 10.687285223367697,
+      "grad_norm": 0.34045952558517456,
+      "learning_rate": 0.0004720314685314685,
+      "loss": 3.3823,
+      "step": 36700
+    },
+    {
+      "epoch": 10.70184635098142,
+      "grad_norm": 0.3242470622062683,
+      "learning_rate": 0.0004718566433566433,
+      "loss": 3.3757,
+      "step": 36750
+    },
+    {
+      "epoch": 10.716407478595142,
+      "grad_norm": 0.35755789279937744,
+      "learning_rate": 0.00047168181818181814,
+      "loss": 3.3811,
+      "step": 36800
+    },
+    {
+      "epoch": 10.730968606208865,
+      "grad_norm": 0.37564322352409363,
+      "learning_rate": 0.000471506993006993,
+      "loss": 3.3723,
+      "step": 36850
+    },
+    {
+      "epoch": 10.745529733822588,
+      "grad_norm": 0.3541555404663086,
+      "learning_rate": 0.0004713321678321678,
+      "loss": 3.373,
+      "step": 36900
+    },
+    {
+      "epoch": 10.760090861436309,
+      "grad_norm": 0.3599989414215088,
+      "learning_rate": 0.00047115734265734265,
+      "loss": 3.3864,
+      "step": 36950
+    },
+    {
+      "epoch": 10.774651989050032,
+      "grad_norm": 0.35538411140441895,
+      "learning_rate": 0.00047098251748251745,
+      "loss": 3.3768,
+      "step": 37000
+    },
+    {
+      "epoch": 10.774651989050032,
+      "eval_accuracy": 0.3687768527690579,
+      "eval_loss": 3.5605952739715576,
+      "eval_runtime": 176.8123,
+      "eval_samples_per_second": 94.145,
+      "eval_steps_per_second": 5.888,
+      "step": 37000
+    },
+    {
+      "epoch": 10.789213116663754,
+      "grad_norm": 0.32753708958625793,
+      "learning_rate": 0.0004708076923076923,
+      "loss": 3.3872,
+      "step": 37050
+    },
+    {
+      "epoch": 10.803774244277477,
+      "grad_norm": 0.3197953402996063,
+      "learning_rate": 0.0004706328671328671,
+      "loss": 3.3832,
+      "step": 37100
+    },
+    {
+      "epoch": 10.8183353718912,
+      "grad_norm": 0.33160483837127686,
+      "learning_rate": 0.00047045804195804195,
+      "loss": 3.375,
+      "step": 37150
+    },
+    {
+      "epoch": 10.83289649950492,
+      "grad_norm": 0.36038827896118164,
+      "learning_rate": 0.0004702832167832167,
+      "loss": 3.3957,
+      "step": 37200
+    },
+    {
+      "epoch": 10.847457627118644,
+      "grad_norm": 0.33877766132354736,
+      "learning_rate": 0.00047010839160839155,
+      "loss": 3.3873,
+      "step": 37250
+    },
+    {
+      "epoch": 10.862018754732366,
+      "grad_norm": 0.34722471237182617,
+      "learning_rate": 0.00046993356643356635,
+      "loss": 3.3901,
+      "step": 37300
+    },
+    {
+      "epoch": 10.876579882346089,
+      "grad_norm": 0.3494495153427124,
+      "learning_rate": 0.0004697587412587412,
+      "loss": 3.3954,
+      "step": 37350
+    },
+    {
+      "epoch": 10.891141009959812,
+      "grad_norm": 0.34382206201553345,
+      "learning_rate": 0.000469583916083916,
+      "loss": 3.3816,
+      "step": 37400
+    },
+    {
+      "epoch": 10.905702137573535,
+      "grad_norm": 0.33641430735588074,
+      "learning_rate": 0.00046940909090909086,
+      "loss": 3.3772,
+      "step": 37450
+    },
+    {
+      "epoch": 10.920263265187256,
+      "grad_norm": 0.3528263568878174,
+      "learning_rate": 0.0004692342657342657,
+      "loss": 3.3876,
+      "step": 37500
+    },
+    {
+      "epoch": 10.934824392800978,
+      "grad_norm": 0.34849458932876587,
+      "learning_rate": 0.0004690594405594405,
+      "loss": 3.3914,
+      "step": 37550
+    },
+    {
+      "epoch": 10.949385520414701,
+      "grad_norm": 0.34705743193626404,
+      "learning_rate": 0.00046888461538461537,
+      "loss": 3.3793,
+      "step": 37600
+    },
+    {
+      "epoch": 10.963946648028424,
+      "grad_norm": 0.3272669017314911,
+      "learning_rate": 0.00046870979020979017,
+      "loss": 3.3788,
+      "step": 37650
+    },
+    {
+      "epoch": 10.978507775642147,
+      "grad_norm": 0.33112722635269165,
+      "learning_rate": 0.000468534965034965,
+      "loss": 3.3837,
+      "step": 37700
+    },
+    {
+      "epoch": 10.993068903255867,
+      "grad_norm": 0.3313654363155365,
+      "learning_rate": 0.0004683601398601398,
+      "loss": 3.3966,
+      "step": 37750
+    },
+    {
+      "epoch": 11.007571786359136,
+      "grad_norm": 0.3786119222640991,
+      "learning_rate": 0.0004681853146853147,
+      "loss": 3.3267,
+      "step": 37800
+    },
+    {
+      "epoch": 11.022132913972857,
+      "grad_norm": 0.35445159673690796,
+      "learning_rate": 0.0004680104895104895,
+      "loss": 3.2648,
+      "step": 37850
+    },
+    {
+      "epoch": 11.03669404158658,
+      "grad_norm": 0.3338647782802582,
+      "learning_rate": 0.00046783566433566433,
+      "loss": 3.2768,
+      "step": 37900
+    },
+    {
+      "epoch": 11.051255169200303,
+      "grad_norm": 0.33461859822273254,
+      "learning_rate": 0.0004676608391608391,
+      "loss": 3.2856,
+      "step": 37950
+    },
+    {
+      "epoch": 11.065816296814026,
+      "grad_norm": 0.3517346978187561,
+      "learning_rate": 0.00046748601398601393,
+      "loss": 3.2775,
+      "step": 38000
+    },
+    {
+      "epoch": 11.065816296814026,
+      "eval_accuracy": 0.36888348205395044,
+      "eval_loss": 3.570584297180176,
+      "eval_runtime": 176.7964,
+      "eval_samples_per_second": 94.153,
+      "eval_steps_per_second": 5.888,
+      "step": 38000
+    },
+    {
+      "epoch": 11.080377424427748,
+      "grad_norm": 0.3435273766517639,
+      "learning_rate": 0.00046731118881118873,
+      "loss": 3.2915,
+      "step": 38050
+    },
+    {
+      "epoch": 11.09493855204147,
+      "grad_norm": 0.3325134813785553,
+      "learning_rate": 0.0004671363636363636,
+      "loss": 3.2978,
+      "step": 38100
+    },
+    {
+      "epoch": 11.109499679655192,
+      "grad_norm": 0.36426037549972534,
+      "learning_rate": 0.00046696153846153844,
+      "loss": 3.3033,
+      "step": 38150
+    },
+    {
+      "epoch": 11.124060807268915,
+      "grad_norm": 0.327764093875885,
+      "learning_rate": 0.00046678671328671324,
+      "loss": 3.3043,
+      "step": 38200
+    },
+    {
+      "epoch": 11.138621934882638,
+      "grad_norm": 0.3818574845790863,
+      "learning_rate": 0.0004666118881118881,
+      "loss": 3.3042,
+      "step": 38250
+    },
+    {
+      "epoch": 11.15318306249636,
+      "grad_norm": 0.38499313592910767,
+      "learning_rate": 0.0004664370629370629,
+      "loss": 3.2968,
+      "step": 38300
+    },
+    {
+      "epoch": 11.167744190110081,
+      "grad_norm": 0.3457299768924713,
+      "learning_rate": 0.00046626223776223774,
+      "loss": 3.299,
+      "step": 38350
+    },
+    {
+      "epoch": 11.182305317723804,
+      "grad_norm": 0.37761256098747253,
+      "learning_rate": 0.00046608741258741254,
+      "loss": 3.3019,
+      "step": 38400
+    },
+    {
+      "epoch": 11.196866445337527,
+      "grad_norm": 0.35469600558280945,
+      "learning_rate": 0.0004659125874125874,
+      "loss": 3.3116,
+      "step": 38450
+    },
+    {
+      "epoch": 11.21142757295125,
+      "grad_norm": 0.3749145269393921,
+      "learning_rate": 0.0004657377622377622,
+      "loss": 3.3128,
+      "step": 38500
+    },
+    {
+      "epoch": 11.225988700564972,
+      "grad_norm": 0.38890284299850464,
+      "learning_rate": 0.00046556293706293705,
+      "loss": 3.332,
+      "step": 38550
+    },
+    {
+      "epoch": 11.240549828178693,
+      "grad_norm": 0.3799166679382324,
+      "learning_rate": 0.00046538811188811185,
+      "loss": 3.3111,
+      "step": 38600
+    },
+    {
+      "epoch": 11.255110955792416,
+      "grad_norm": 0.3511936366558075,
+      "learning_rate": 0.0004652132867132867,
+      "loss": 3.3251,
+      "step": 38650
+    },
+    {
+      "epoch": 11.269672083406139,
+      "grad_norm": 0.37755730748176575,
+      "learning_rate": 0.00046503846153846145,
+      "loss": 3.3257,
+      "step": 38700
+    },
+    {
+      "epoch": 11.284233211019862,
+      "grad_norm": 0.34114590287208557,
+      "learning_rate": 0.0004648636363636363,
+      "loss": 3.3212,
+      "step": 38750
+    },
+    {
+      "epoch": 11.298794338633584,
+      "grad_norm": 0.3562406897544861,
+      "learning_rate": 0.0004646888111888111,
+      "loss": 3.3259,
+      "step": 38800
+    },
+    {
+      "epoch": 11.313355466247307,
+      "grad_norm": 0.3757557272911072,
+      "learning_rate": 0.00046451398601398596,
+      "loss": 3.3395,
+      "step": 38850
+    },
+    {
+      "epoch": 11.327916593861028,
+      "grad_norm": 0.3693144619464874,
+      "learning_rate": 0.0004643391608391608,
+      "loss": 3.3458,
+      "step": 38900
+    },
+    {
+      "epoch": 11.34247772147475,
+      "grad_norm": 0.3550606071949005,
+      "learning_rate": 0.0004641643356643356,
+      "loss": 3.336,
+      "step": 38950
+    },
+    {
+      "epoch": 11.357038849088473,
+      "grad_norm": 0.36543989181518555,
+      "learning_rate": 0.00046398951048951046,
+      "loss": 3.331,
+      "step": 39000
+    },
+    {
+      "epoch": 11.357038849088473,
+      "eval_accuracy": 0.3695512376638617,
+      "eval_loss": 3.567185878753662,
+      "eval_runtime": 176.8379,
+      "eval_samples_per_second": 94.131,
+      "eval_steps_per_second": 5.887,
+      "step": 39000
+    },
+    {
+      "epoch": 11.371599976702196,
+      "grad_norm": 0.36690041422843933,
+      "learning_rate": 0.00046381468531468526,
+      "loss": 3.3383,
+      "step": 39050
+    },
+    {
+      "epoch": 11.386161104315919,
+      "grad_norm": 0.3439081013202667,
+      "learning_rate": 0.0004636398601398601,
+      "loss": 3.3272,
+      "step": 39100
+    },
+    {
+      "epoch": 11.40072223192964,
+      "grad_norm": 0.3694179654121399,
+      "learning_rate": 0.0004634650349650349,
+      "loss": 3.337,
+      "step": 39150
+    },
+    {
+      "epoch": 11.415283359543363,
+      "grad_norm": 0.36797547340393066,
+      "learning_rate": 0.00046329020979020977,
+      "loss": 3.3276,
+      "step": 39200
+    },
+    {
+      "epoch": 11.429844487157085,
+      "grad_norm": 0.40091055631637573,
+      "learning_rate": 0.00046311538461538457,
+      "loss": 3.3355,
+      "step": 39250
+    },
+    {
+      "epoch": 11.444405614770808,
+      "grad_norm": 0.34869131445884705,
+      "learning_rate": 0.0004629405594405594,
+      "loss": 3.3352,
+      "step": 39300
+    },
+    {
+      "epoch": 11.458966742384531,
+      "grad_norm": 0.35575979948043823,
+      "learning_rate": 0.0004627657342657342,
+      "loss": 3.3462,
+      "step": 39350
+    },
+    {
+      "epoch": 11.473527869998252,
+      "grad_norm": 0.3693273365497589,
+      "learning_rate": 0.0004625909090909091,
+      "loss": 3.3398,
+      "step": 39400
+    },
+    {
+      "epoch": 11.488088997611975,
+      "grad_norm": 0.3422088623046875,
+      "learning_rate": 0.0004624160839160838,
+      "loss": 3.3537,
+      "step": 39450
+    },
+    {
+      "epoch": 11.502650125225697,
+      "grad_norm": 0.35034725069999695,
+      "learning_rate": 0.0004622412587412587,
+      "loss": 3.3536,
+      "step": 39500
+    },
+    {
+      "epoch": 11.51721125283942,
+      "grad_norm": 0.3721543550491333,
+      "learning_rate": 0.00046206643356643353,
+      "loss": 3.3611,
+      "step": 39550
+    },
+    {
+      "epoch": 11.531772380453143,
+      "grad_norm": 0.3786129653453827,
+      "learning_rate": 0.00046189160839160833,
+      "loss": 3.3612,
+      "step": 39600
+    },
+    {
+      "epoch": 11.546333508066864,
+      "grad_norm": 0.37621766328811646,
+      "learning_rate": 0.0004617167832167832,
+      "loss": 3.3513,
+      "step": 39650
+    },
+    {
+      "epoch": 11.560894635680587,
+      "grad_norm": 0.33818793296813965,
+      "learning_rate": 0.000461541958041958,
+      "loss": 3.3585,
+      "step": 39700
+    },
+    {
+      "epoch": 11.57545576329431,
+      "grad_norm": 0.3617473840713501,
+      "learning_rate": 0.00046136713286713284,
+      "loss": 3.3416,
+      "step": 39750
+    },
+    {
+      "epoch": 11.590016890908032,
+      "grad_norm": 0.36892908811569214,
+      "learning_rate": 0.00046119230769230764,
+      "loss": 3.3564,
+      "step": 39800
+    },
+    {
+      "epoch": 11.604578018521755,
+      "grad_norm": 0.3816312849521637,
+      "learning_rate": 0.0004610174825174825,
+      "loss": 3.3536,
+      "step": 39850
+    },
+    {
+      "epoch": 11.619139146135478,
+      "grad_norm": 0.3597758114337921,
+      "learning_rate": 0.0004608426573426573,
+      "loss": 3.3576,
+      "step": 39900
+    },
+    {
+      "epoch": 11.633700273749199,
+      "grad_norm": 0.3639199733734131,
+      "learning_rate": 0.00046066783216783215,
+      "loss": 3.354,
+      "step": 39950
+    },
+    {
+      "epoch": 11.648261401362921,
+      "grad_norm": 0.3267276883125305,
+      "learning_rate": 0.00046049300699300695,
+      "loss": 3.3565,
+      "step": 40000
+    },
+    {
+      "epoch": 11.648261401362921,
+      "eval_accuracy": 0.36973110845315116,
+      "eval_loss": 3.5581958293914795,
+      "eval_runtime": 176.7361,
+      "eval_samples_per_second": 94.186,
+      "eval_steps_per_second": 5.89,
+      "step": 40000
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 171700,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 50,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 20,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.3607701815296e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}