{
  "best_global_step": 2750,
  "best_metric": 0.8479926448053938,
  "best_model_checkpoint": "dinov2-fonts-with-subfonts-2/checkpoint-2750",
  "epoch": 1.0,
  "eval_steps": 50,
  "global_step": 2752,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003633720930232558,
      "grad_norm": 15.892566680908203,
      "learning_rate": 9.967296511627907e-05,
      "loss": 2.5094,
      "step": 10
    },
    {
      "epoch": 0.007267441860465116,
      "grad_norm": 13.693017959594727,
      "learning_rate": 9.930959302325582e-05,
      "loss": 2.4643,
      "step": 20
    },
    {
      "epoch": 0.010901162790697675,
      "grad_norm": 16.248794555664062,
      "learning_rate": 9.894622093023256e-05,
      "loss": 2.4288,
      "step": 30
    },
    {
      "epoch": 0.014534883720930232,
      "grad_norm": 14.924266815185547,
      "learning_rate": 9.85828488372093e-05,
      "loss": 2.1981,
      "step": 40
    },
    {
      "epoch": 0.018168604651162792,
      "grad_norm": 17.095857620239258,
      "learning_rate": 9.821947674418606e-05,
      "loss": 2.3055,
      "step": 50
    },
    {
      "epoch": 0.018168604651162792,
      "eval_accuracy": 0.28256205945448976,
      "eval_loss": 2.221597671508789,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 45.2864,
      "eval_samples_per_second": 72.053,
      "eval_steps_per_second": 2.252,
      "step": 50
    },
    {
      "epoch": 0.02180232558139535,
      "grad_norm": 15.624826431274414,
      "learning_rate": 9.78561046511628e-05,
      "loss": 2.2321,
      "step": 60
    },
    {
      "epoch": 0.025436046511627907,
      "grad_norm": 15.980948448181152,
      "learning_rate": 9.749273255813955e-05,
      "loss": 2.1783,
      "step": 70
    },
    {
      "epoch": 0.029069767441860465,
      "grad_norm": 16.875125885009766,
      "learning_rate": 9.712936046511628e-05,
      "loss": 2.1456,
      "step": 80
    },
    {
      "epoch": 0.032703488372093026,
      "grad_norm": 14.524847030639648,
      "learning_rate": 9.676598837209303e-05,
      "loss": 2.0414,
      "step": 90
    },
    {
      "epoch": 0.036337209302325583,
      "grad_norm": 19.023685455322266,
      "learning_rate": 9.640261627906977e-05,
      "loss": 2.0327,
      "step": 100
    },
    {
      "epoch": 0.036337209302325583,
      "eval_accuracy": 0.3444682807232608,
      "eval_loss": 2.0124428272247314,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 55.6946,
      "eval_samples_per_second": 58.587,
      "eval_steps_per_second": 1.831,
      "step": 100
    },
    {
      "epoch": 0.03997093023255814,
      "grad_norm": 18.07843589782715,
      "learning_rate": 9.603924418604651e-05,
      "loss": 1.9096,
      "step": 110
    },
    {
      "epoch": 0.0436046511627907,
      "grad_norm": 15.953253746032715,
      "learning_rate": 9.567587209302325e-05,
      "loss": 1.9289,
      "step": 120
    },
    {
      "epoch": 0.047238372093023256,
      "grad_norm": 16.393022537231445,
      "learning_rate": 9.53125e-05,
      "loss": 1.898,
      "step": 130
    },
    {
      "epoch": 0.050872093023255814,
      "grad_norm": 15.503422737121582,
      "learning_rate": 9.494912790697676e-05,
      "loss": 1.9764,
      "step": 140
    },
    {
      "epoch": 0.05450581395348837,
      "grad_norm": 17.98733139038086,
      "learning_rate": 9.458575581395349e-05,
      "loss": 1.8097,
      "step": 150
    },
    {
      "epoch": 0.05450581395348837,
      "eval_accuracy": 0.3781795893349678,
      "eval_loss": 1.8244749307632446,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.7499,
      "eval_samples_per_second": 64.296,
      "eval_steps_per_second": 2.01,
      "step": 150
    },
    {
      "epoch": 0.05813953488372093,
      "grad_norm": 22.832233428955078,
      "learning_rate": 9.422238372093024e-05,
      "loss": 1.8411,
      "step": 160
    },
    {
      "epoch": 0.06177325581395349,
      "grad_norm": 17.715545654296875,
      "learning_rate": 9.385901162790698e-05,
      "loss": 1.7486,
      "step": 170
    },
    {
      "epoch": 0.06540697674418605,
      "grad_norm": 18.01825714111328,
      "learning_rate": 9.349563953488372e-05,
      "loss": 1.7325,
      "step": 180
    },
    {
      "epoch": 0.0690406976744186,
      "grad_norm": 17.555883407592773,
      "learning_rate": 9.313226744186046e-05,
      "loss": 1.772,
      "step": 190
    },
    {
      "epoch": 0.07267441860465117,
      "grad_norm": 14.410489082336426,
      "learning_rate": 9.276889534883722e-05,
      "loss": 1.7379,
      "step": 200
    },
    {
      "epoch": 0.07267441860465117,
      "eval_accuracy": 0.4112779650628256,
      "eval_loss": 1.7361937761306763,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.8089,
      "eval_samples_per_second": 68.251,
      "eval_steps_per_second": 2.133,
      "step": 200
    },
    {
      "epoch": 0.07630813953488372,
      "grad_norm": 17.80829429626465,
      "learning_rate": 9.240552325581396e-05,
      "loss": 1.7533,
      "step": 210
    },
    {
      "epoch": 0.07994186046511628,
      "grad_norm": 19.89047622680664,
      "learning_rate": 9.20421511627907e-05,
      "loss": 1.7556,
      "step": 220
    },
    {
      "epoch": 0.08357558139534883,
      "grad_norm": 17.062580108642578,
      "learning_rate": 9.167877906976745e-05,
      "loss": 1.6221,
      "step": 230
    },
    {
      "epoch": 0.0872093023255814,
      "grad_norm": 20.477750778198242,
      "learning_rate": 9.131540697674419e-05,
      "loss": 1.6176,
      "step": 240
    },
    {
      "epoch": 0.09084302325581395,
      "grad_norm": 19.113245010375977,
      "learning_rate": 9.095203488372093e-05,
      "loss": 1.7862,
      "step": 250
    },
    {
      "epoch": 0.09084302325581395,
      "eval_accuracy": 0.4511186025130248,
      "eval_loss": 1.5864135026931763,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.5524,
      "eval_samples_per_second": 68.619,
      "eval_steps_per_second": 2.145,
      "step": 250
    },
    {
      "epoch": 0.09447674418604651,
      "grad_norm": 17.91750144958496,
      "learning_rate": 9.058866279069767e-05,
      "loss": 1.5788,
      "step": 260
    },
    {
      "epoch": 0.09811046511627906,
      "grad_norm": 19.93201446533203,
      "learning_rate": 9.022529069767443e-05,
      "loss": 1.6572,
      "step": 270
    },
    {
      "epoch": 0.10174418604651163,
      "grad_norm": 14.712456703186035,
      "learning_rate": 8.986191860465117e-05,
      "loss": 1.6514,
      "step": 280
    },
    {
      "epoch": 0.10537790697674419,
      "grad_norm": 17.98506736755371,
      "learning_rate": 8.94985465116279e-05,
      "loss": 1.555,
      "step": 290
    },
    {
      "epoch": 0.10901162790697674,
      "grad_norm": 18.016740798950195,
      "learning_rate": 8.913517441860465e-05,
      "loss": 1.4337,
      "step": 300
    },
    {
      "epoch": 0.10901162790697674,
      "eval_accuracy": 0.4566349984676678,
      "eval_loss": 1.5329961776733398,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 49.8886,
      "eval_samples_per_second": 65.406,
      "eval_steps_per_second": 2.045,
      "step": 300
    },
    {
      "epoch": 0.11264534883720931,
      "grad_norm": 18.44303321838379,
      "learning_rate": 8.87718023255814e-05,
      "loss": 1.6438,
      "step": 310
    },
    {
      "epoch": 0.11627906976744186,
      "grad_norm": 19.64215660095215,
      "learning_rate": 8.840843023255815e-05,
      "loss": 1.4625,
      "step": 320
    },
    {
      "epoch": 0.11991279069767442,
      "grad_norm": 23.59469985961914,
      "learning_rate": 8.804505813953488e-05,
      "loss": 1.5685,
      "step": 330
    },
    {
      "epoch": 0.12354651162790697,
      "grad_norm": 19.073352813720703,
      "learning_rate": 8.768168604651164e-05,
      "loss": 1.4693,
      "step": 340
    },
    {
      "epoch": 0.12718023255813954,
      "grad_norm": 16.183441162109375,
      "learning_rate": 8.731831395348838e-05,
      "loss": 1.4731,
      "step": 350
    },
    {
      "epoch": 0.12718023255813954,
      "eval_accuracy": 0.5099601593625498,
      "eval_loss": 1.4183763265609741,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.7302,
      "eval_samples_per_second": 68.363,
      "eval_steps_per_second": 2.137,
      "step": 350
    },
    {
      "epoch": 0.1308139534883721,
      "grad_norm": 22.20075035095215,
      "learning_rate": 8.695494186046512e-05,
      "loss": 1.4671,
      "step": 360
    },
    {
      "epoch": 0.13444767441860464,
      "grad_norm": 19.052248001098633,
      "learning_rate": 8.659156976744186e-05,
      "loss": 1.4519,
      "step": 370
    },
    {
      "epoch": 0.1380813953488372,
      "grad_norm": 17.498851776123047,
      "learning_rate": 8.622819767441861e-05,
      "loss": 1.4895,
      "step": 380
    },
    {
      "epoch": 0.14171511627906977,
      "grad_norm": 17.81024932861328,
      "learning_rate": 8.586482558139536e-05,
      "loss": 1.5058,
      "step": 390
    },
    {
      "epoch": 0.14534883720930233,
      "grad_norm": 22.155494689941406,
      "learning_rate": 8.550145348837209e-05,
      "loss": 1.4192,
      "step": 400
    },
    {
      "epoch": 0.14534883720930233,
      "eval_accuracy": 0.5163959546429666,
      "eval_loss": 1.3699551820755005,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.8531,
      "eval_samples_per_second": 68.188,
      "eval_steps_per_second": 2.132,
      "step": 400
    },
    {
      "epoch": 0.14898255813953487,
      "grad_norm": 17.518573760986328,
      "learning_rate": 8.513808139534885e-05,
      "loss": 1.3937,
      "step": 410
    },
    {
      "epoch": 0.15261627906976744,
      "grad_norm": 18.192092895507812,
      "learning_rate": 8.477470930232559e-05,
      "loss": 1.3298,
      "step": 420
    },
    {
      "epoch": 0.15625,
      "grad_norm": 19.992328643798828,
      "learning_rate": 8.441133720930233e-05,
      "loss": 1.3802,
      "step": 430
    },
    {
      "epoch": 0.15988372093023256,
      "grad_norm": 27.304393768310547,
      "learning_rate": 8.404796511627907e-05,
      "loss": 1.3234,
      "step": 440
    },
    {
      "epoch": 0.16351744186046513,
      "grad_norm": 24.619159698486328,
      "learning_rate": 8.368459302325582e-05,
      "loss": 1.415,
      "step": 450
    },
    {
      "epoch": 0.16351744186046513,
      "eval_accuracy": 0.5262028807845541,
      "eval_loss": 1.3008671998977661,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.1216,
      "eval_samples_per_second": 67.807,
      "eval_steps_per_second": 2.12,
      "step": 450
    },
    {
      "epoch": 0.16715116279069767,
      "grad_norm": 19.95876121520996,
      "learning_rate": 8.332122093023256e-05,
      "loss": 1.3458,
      "step": 460
    },
    {
      "epoch": 0.17078488372093023,
      "grad_norm": 18.42659568786621,
      "learning_rate": 8.29578488372093e-05,
      "loss": 1.4143,
      "step": 470
    },
    {
      "epoch": 0.1744186046511628,
      "grad_norm": 19.090171813964844,
      "learning_rate": 8.259447674418606e-05,
      "loss": 1.3685,
      "step": 480
    },
    {
      "epoch": 0.17805232558139536,
      "grad_norm": 23.138212203979492,
      "learning_rate": 8.22311046511628e-05,
      "loss": 1.3492,
      "step": 490
    },
    {
      "epoch": 0.1816860465116279,
      "grad_norm": 24.334434509277344,
      "learning_rate": 8.186773255813954e-05,
      "loss": 1.4102,
      "step": 500
    },
    {
      "epoch": 0.1816860465116279,
      "eval_accuracy": 0.5455102666258045,
      "eval_loss": 1.2436403036117554,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.5657,
      "eval_samples_per_second": 67.187,
      "eval_steps_per_second": 2.1,
      "step": 500
    },
    {
      "epoch": 0.18531976744186046,
      "grad_norm": 20.45575714111328,
      "learning_rate": 8.150436046511628e-05,
      "loss": 1.2621,
      "step": 510
    },
    {
      "epoch": 0.18895348837209303,
      "grad_norm": 15.790328979492188,
      "learning_rate": 8.114098837209303e-05,
      "loss": 1.2724,
      "step": 520
    },
    {
      "epoch": 0.1925872093023256,
      "grad_norm": 21.53169059753418,
      "learning_rate": 8.077761627906977e-05,
      "loss": 1.3039,
      "step": 530
    },
    {
      "epoch": 0.19622093023255813,
      "grad_norm": 17.059200286865234,
      "learning_rate": 8.041424418604651e-05,
      "loss": 1.2784,
      "step": 540
    },
    {
      "epoch": 0.1998546511627907,
      "grad_norm": 19.423311233520508,
      "learning_rate": 8.005087209302325e-05,
      "loss": 1.2175,
      "step": 550
    },
    {
      "epoch": 0.1998546511627907,
      "eval_accuracy": 0.5675758504443763,
      "eval_loss": 1.1887367963790894,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.5535,
      "eval_samples_per_second": 67.204,
      "eval_steps_per_second": 2.101,
      "step": 550
    },
    {
      "epoch": 0.20348837209302326,
      "grad_norm": 21.524168014526367,
      "learning_rate": 7.96875e-05,
      "loss": 1.1391,
      "step": 560
    },
    {
      "epoch": 0.20712209302325582,
      "grad_norm": 15.95803165435791,
      "learning_rate": 7.932412790697675e-05,
      "loss": 1.1714,
      "step": 570
    },
    {
      "epoch": 0.21075581395348839,
      "grad_norm": 17.44819450378418,
      "learning_rate": 7.896075581395349e-05,
      "loss": 1.2085,
      "step": 580
    },
    {
      "epoch": 0.21438953488372092,
      "grad_norm": 20.14234161376953,
      "learning_rate": 7.859738372093024e-05,
      "loss": 1.1547,
      "step": 590
    },
    {
      "epoch": 0.2180232558139535,
      "grad_norm": 22.286245346069336,
      "learning_rate": 7.823401162790698e-05,
      "loss": 1.1583,
      "step": 600
    },
    {
      "epoch": 0.2180232558139535,
      "eval_accuracy": 0.5770763101440393,
      "eval_loss": 1.1267926692962646,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.6542,
      "eval_samples_per_second": 68.472,
      "eval_steps_per_second": 2.14,
      "step": 600
    },
    {
      "epoch": 0.22165697674418605,
      "grad_norm": 24.456897735595703,
      "learning_rate": 7.787063953488372e-05,
      "loss": 1.1663,
      "step": 610
    },
    {
      "epoch": 0.22529069767441862,
      "grad_norm": 19.741336822509766,
      "learning_rate": 7.750726744186046e-05,
      "loss": 1.1761,
      "step": 620
    },
    {
      "epoch": 0.22892441860465115,
      "grad_norm": 19.847604751586914,
      "learning_rate": 7.714389534883722e-05,
      "loss": 1.1423,
      "step": 630
    },
    {
      "epoch": 0.23255813953488372,
      "grad_norm": 16.832901000976562,
      "learning_rate": 7.678052325581396e-05,
      "loss": 1.0611,
      "step": 640
    },
    {
      "epoch": 0.23619186046511628,
      "grad_norm": 19.98552703857422,
      "learning_rate": 7.64171511627907e-05,
      "loss": 1.1858,
      "step": 650
    },
    {
      "epoch": 0.23619186046511628,
      "eval_accuracy": 0.6196751455715599,
      "eval_loss": 1.0860530138015747,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.2713,
      "eval_samples_per_second": 64.908,
      "eval_steps_per_second": 2.029,
      "step": 650
    },
    {
      "epoch": 0.23982558139534885,
      "grad_norm": 22.90323257446289,
      "learning_rate": 7.605377906976745e-05,
      "loss": 1.2555,
      "step": 660
    },
    {
      "epoch": 0.24345930232558138,
      "grad_norm": 18.252866744995117,
      "learning_rate": 7.569040697674419e-05,
      "loss": 1.1472,
      "step": 670
    },
    {
      "epoch": 0.24709302325581395,
      "grad_norm": 29.458932876586914,
      "learning_rate": 7.532703488372093e-05,
      "loss": 1.2043,
      "step": 680
    },
    {
      "epoch": 0.2507267441860465,
      "grad_norm": 26.31637954711914,
      "learning_rate": 7.496366279069767e-05,
      "loss": 1.1212,
      "step": 690
    },
    {
      "epoch": 0.2543604651162791,
      "grad_norm": 19.821483612060547,
      "learning_rate": 7.460029069767443e-05,
      "loss": 1.0821,
      "step": 700
    },
    {
      "epoch": 0.2543604651162791,
      "eval_accuracy": 0.622739809990806,
      "eval_loss": 1.0653626918792725,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.867,
      "eval_samples_per_second": 68.168,
      "eval_steps_per_second": 2.131,
      "step": 700
    },
    {
      "epoch": 0.25799418604651164,
      "grad_norm": 20.264490127563477,
      "learning_rate": 7.423691860465117e-05,
      "loss": 1.0817,
      "step": 710
    },
    {
      "epoch": 0.2616279069767442,
      "grad_norm": 16.901201248168945,
      "learning_rate": 7.38735465116279e-05,
      "loss": 1.0836,
      "step": 720
    },
    {
      "epoch": 0.26526162790697677,
      "grad_norm": 21.42996597290039,
      "learning_rate": 7.351017441860465e-05,
      "loss": 1.0995,
      "step": 730
    },
    {
      "epoch": 0.2688953488372093,
      "grad_norm": 27.091205596923828,
      "learning_rate": 7.31468023255814e-05,
      "loss": 1.2028,
      "step": 740
    },
    {
      "epoch": 0.27252906976744184,
      "grad_norm": 26.723642349243164,
      "learning_rate": 7.278343023255814e-05,
      "loss": 1.1347,
      "step": 750
    },
    {
      "epoch": 0.27252906976744184,
      "eval_accuracy": 0.6356114005516396,
      "eval_loss": 1.023024082183838,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.1088,
      "eval_samples_per_second": 67.825,
      "eval_steps_per_second": 2.12,
      "step": 750
    },
    {
      "epoch": 0.2761627906976744,
      "grad_norm": 15.63192367553711,
      "learning_rate": 7.242005813953488e-05,
      "loss": 1.1222,
      "step": 760
    },
    {
      "epoch": 0.279796511627907,
      "grad_norm": 16.611923217773438,
      "learning_rate": 7.205668604651164e-05,
      "loss": 1.1061,
      "step": 770
    },
    {
      "epoch": 0.28343023255813954,
      "grad_norm": 21.829530715942383,
      "learning_rate": 7.169331395348838e-05,
      "loss": 1.0047,
      "step": 780
    },
    {
      "epoch": 0.2870639534883721,
      "grad_norm": 22.85479736328125,
      "learning_rate": 7.132994186046512e-05,
      "loss": 1.1656,
      "step": 790
    },
    {
      "epoch": 0.29069767441860467,
      "grad_norm": 24.8774356842041,
      "learning_rate": 7.096656976744186e-05,
      "loss": 1.0365,
      "step": 800
    },
    {
      "epoch": 0.29069767441860467,
      "eval_accuracy": 0.6472571253447748,
      "eval_loss": 0.9910984635353088,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 49.0197,
      "eval_samples_per_second": 66.565,
      "eval_steps_per_second": 2.081,
      "step": 800
    },
    {
      "epoch": 0.29433139534883723,
      "grad_norm": 24.75929069519043,
      "learning_rate": 7.060319767441861e-05,
      "loss": 1.0438,
      "step": 810
    },
    {
      "epoch": 0.29796511627906974,
      "grad_norm": 28.923444747924805,
      "learning_rate": 7.023982558139535e-05,
      "loss": 1.0488,
      "step": 820
    },
    {
      "epoch": 0.3015988372093023,
      "grad_norm": 18.869667053222656,
      "learning_rate": 6.987645348837209e-05,
      "loss": 1.0892,
      "step": 830
    },
    {
      "epoch": 0.30523255813953487,
      "grad_norm": 18.0045108795166,
      "learning_rate": 6.951308139534885e-05,
      "loss": 1.0249,
      "step": 840
    },
    {
      "epoch": 0.30886627906976744,
      "grad_norm": 18.63671875,
      "learning_rate": 6.914970930232559e-05,
      "loss": 0.9398,
      "step": 850
    },
    {
      "epoch": 0.30886627906976744,
      "eval_accuracy": 0.6420471958320564,
      "eval_loss": 0.9642446041107178,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.8675,
      "eval_samples_per_second": 68.167,
      "eval_steps_per_second": 2.131,
      "step": 850
    },
    {
      "epoch": 0.3125,
      "grad_norm": 22.47454261779785,
      "learning_rate": 6.878633720930233e-05,
      "loss": 1.0515,
      "step": 860
    },
    {
      "epoch": 0.31613372093023256,
      "grad_norm": 21.17203712463379,
      "learning_rate": 6.842296511627907e-05,
      "loss": 0.973,
      "step": 870
    },
    {
      "epoch": 0.31976744186046513,
      "grad_norm": 20.848777770996094,
      "learning_rate": 6.805959302325582e-05,
      "loss": 0.9193,
      "step": 880
    },
    {
      "epoch": 0.3234011627906977,
      "grad_norm": 25.05883026123047,
      "learning_rate": 6.769622093023256e-05,
      "loss": 0.9351,
      "step": 890
    },
    {
      "epoch": 0.32703488372093026,
      "grad_norm": 22.536279678344727,
      "learning_rate": 6.73328488372093e-05,
      "loss": 1.0203,
      "step": 900
    },
    {
      "epoch": 0.32703488372093026,
      "eval_accuracy": 0.6736132393502912,
      "eval_loss": 0.9405654072761536,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 49.3771,
      "eval_samples_per_second": 66.083,
      "eval_steps_per_second": 2.066,
      "step": 900
    },
    {
      "epoch": 0.33066860465116277,
      "grad_norm": 14.16507339477539,
      "learning_rate": 6.696947674418606e-05,
      "loss": 1.0585,
      "step": 910
    },
    {
      "epoch": 0.33430232558139533,
      "grad_norm": 19.16563606262207,
      "learning_rate": 6.66061046511628e-05,
      "loss": 0.9585,
      "step": 920
    },
    {
      "epoch": 0.3379360465116279,
      "grad_norm": 26.094350814819336,
      "learning_rate": 6.624273255813954e-05,
      "loss": 1.0055,
      "step": 930
    },
    {
      "epoch": 0.34156976744186046,
      "grad_norm": 18.31908416748047,
      "learning_rate": 6.587936046511628e-05,
      "loss": 0.9208,
      "step": 940
    },
    {
      "epoch": 0.345203488372093,
      "grad_norm": 31.46238136291504,
      "learning_rate": 6.551598837209303e-05,
      "loss": 1.0836,
      "step": 950
    },
    {
      "epoch": 0.345203488372093,
      "eval_accuracy": 0.6647257125344774,
      "eval_loss": 0.9281411170959473,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.41,
      "eval_samples_per_second": 64.729,
      "eval_steps_per_second": 2.023,
      "step": 950
    },
    {
      "epoch": 0.3488372093023256,
      "grad_norm": 16.519235610961914,
      "learning_rate": 6.515261627906977e-05,
      "loss": 0.9886,
      "step": 960
    },
    {
      "epoch": 0.35247093023255816,
      "grad_norm": 21.729778289794922,
      "learning_rate": 6.478924418604651e-05,
      "loss": 1.0044,
      "step": 970
    },
    {
      "epoch": 0.3561046511627907,
      "grad_norm": 19.362590789794922,
      "learning_rate": 6.442587209302325e-05,
      "loss": 0.9829,
      "step": 980
    },
    {
      "epoch": 0.35973837209302323,
      "grad_norm": 19.202898025512695,
      "learning_rate": 6.40625e-05,
      "loss": 1.0145,
      "step": 990
    },
    {
      "epoch": 0.3633720930232558,
      "grad_norm": 17.671977996826172,
      "learning_rate": 6.369912790697675e-05,
      "loss": 0.9329,
      "step": 1000
    },
    {
      "epoch": 0.3633720930232558,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8783280849456787,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 49.9909,
      "eval_samples_per_second": 65.272,
      "eval_steps_per_second": 2.04,
      "step": 1000
    },
    {
      "epoch": 0.36700581395348836,
      "grad_norm": 22.96895408630371,
      "learning_rate": 6.333575581395349e-05,
      "loss": 0.9242,
      "step": 1010
    },
    {
      "epoch": 0.3706395348837209,
      "grad_norm": 19.682958602905273,
      "learning_rate": 6.297238372093024e-05,
      "loss": 0.9181,
      "step": 1020
    },
    {
      "epoch": 0.3742732558139535,
      "grad_norm": 21.612184524536133,
      "learning_rate": 6.260901162790698e-05,
      "loss": 0.8911,
      "step": 1030
    },
    {
      "epoch": 0.37790697674418605,
      "grad_norm": 17.922510147094727,
      "learning_rate": 6.224563953488372e-05,
      "loss": 0.8801,
      "step": 1040
    },
    {
      "epoch": 0.3815406976744186,
      "grad_norm": 19.79427719116211,
      "learning_rate": 6.188226744186046e-05,
      "loss": 0.9331,
      "step": 1050
    },
    {
      "epoch": 0.3815406976744186,
      "eval_accuracy": 0.6972111553784861,
      "eval_loss": 0.8626508712768555,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.7203,
      "eval_samples_per_second": 66.974,
      "eval_steps_per_second": 2.094,
      "step": 1050
    },
    {
      "epoch": 0.3851744186046512,
      "grad_norm": 23.669225692749023,
      "learning_rate": 6.151889534883722e-05,
      "loss": 0.9045,
      "step": 1060
    },
    {
      "epoch": 0.38880813953488375,
      "grad_norm": 19.99860191345215,
      "learning_rate": 6.115552325581396e-05,
      "loss": 0.9222,
      "step": 1070
    },
    {
      "epoch": 0.39244186046511625,
      "grad_norm": 43.26332092285156,
      "learning_rate": 6.07921511627907e-05,
      "loss": 0.9973,
      "step": 1080
    },
    {
      "epoch": 0.3960755813953488,
      "grad_norm": 25.616634368896484,
      "learning_rate": 6.042877906976745e-05,
      "loss": 0.9616,
      "step": 1090
    },
    {
      "epoch": 0.3997093023255814,
      "grad_norm": 21.20777702331543,
      "learning_rate": 6.006540697674419e-05,
      "loss": 0.9042,
      "step": 1100
    },
    {
      "epoch": 0.3997093023255814,
      "eval_accuracy": 0.7269384002451732,
      "eval_loss": 0.832632303237915,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.8314,
      "eval_samples_per_second": 68.219,
      "eval_steps_per_second": 2.132,
      "step": 1100
    },
    {
      "epoch": 0.40334302325581395,
      "grad_norm": 25.138526916503906,
      "learning_rate": 5.970203488372094e-05,
      "loss": 0.9166,
      "step": 1110
    },
    {
      "epoch": 0.4069767441860465,
      "grad_norm": 31.42925453186035,
      "learning_rate": 5.933866279069767e-05,
      "loss": 0.8452,
      "step": 1120
    },
    {
      "epoch": 0.4106104651162791,
      "grad_norm": 21.83854866027832,
      "learning_rate": 5.8975290697674425e-05,
      "loss": 0.9061,
      "step": 1130
    },
    {
      "epoch": 0.41424418604651164,
      "grad_norm": 24.000688552856445,
      "learning_rate": 5.861191860465116e-05,
      "loss": 0.9286,
      "step": 1140
    },
    {
      "epoch": 0.4178779069767442,
      "grad_norm": 20.03739356994629,
      "learning_rate": 5.824854651162791e-05,
      "loss": 0.8758,
      "step": 1150
    },
    {
      "epoch": 0.4178779069767442,
      "eval_accuracy": 0.7247931351517009,
      "eval_loss": 0.8062854409217834,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 49.4415,
      "eval_samples_per_second": 65.997,
      "eval_steps_per_second": 2.063,
      "step": 1150
    },
    {
      "epoch": 0.42151162790697677,
      "grad_norm": 22.991212844848633,
      "learning_rate": 5.7885174418604646e-05,
      "loss": 0.8388,
      "step": 1160
    },
    {
      "epoch": 0.4251453488372093,
      "grad_norm": 14.57451343536377,
      "learning_rate": 5.75218023255814e-05,
      "loss": 0.8124,
      "step": 1170
    },
    {
      "epoch": 0.42877906976744184,
      "grad_norm": 16.236392974853516,
      "learning_rate": 5.715843023255815e-05,
      "loss": 0.9013,
      "step": 1180
    },
    {
      "epoch": 0.4324127906976744,
      "grad_norm": 13.36828327178955,
      "learning_rate": 5.679505813953489e-05,
      "loss": 0.8201,
      "step": 1190
    },
    {
      "epoch": 0.436046511627907,
      "grad_norm": 25.648473739624023,
      "learning_rate": 5.6431686046511635e-05,
      "loss": 0.857,
      "step": 1200
    },
    {
      "epoch": 0.436046511627907,
      "eval_accuracy": 0.6993564204719583,
      "eval_loss": 0.8466485738754272,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.354,
      "eval_samples_per_second": 67.482,
      "eval_steps_per_second": 2.109,
      "step": 1200
    },
    {
      "epoch": 0.43968023255813954,
      "grad_norm": 18.110530853271484,
      "learning_rate": 5.606831395348837e-05,
      "loss": 0.9745,
      "step": 1210
    },
    {
      "epoch": 0.4433139534883721,
      "grad_norm": 25.08212661743164,
      "learning_rate": 5.570494186046512e-05,
      "loss": 0.8266,
      "step": 1220
    },
    {
      "epoch": 0.44694767441860467,
      "grad_norm": 17.428245544433594,
      "learning_rate": 5.5341569767441856e-05,
      "loss": 0.8776,
      "step": 1230
    },
    {
      "epoch": 0.45058139534883723,
      "grad_norm": 20.17462158203125,
      "learning_rate": 5.497819767441861e-05,
      "loss": 0.9266,
      "step": 1240
    },
    {
      "epoch": 0.45421511627906974,
      "grad_norm": 19.790435791015625,
      "learning_rate": 5.461482558139536e-05,
      "loss": 0.8733,
      "step": 1250
    },
    {
      "epoch": 0.45421511627906974,
      "eval_accuracy": 0.7223414036163041,
      "eval_loss": 0.787076473236084,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.1791,
      "eval_samples_per_second": 63.756,
      "eval_steps_per_second": 1.993,
      "step": 1250
    },
    {
      "epoch": 0.4578488372093023,
      "grad_norm": 17.204328536987305,
      "learning_rate": 5.42514534883721e-05,
      "loss": 0.8393,
      "step": 1260
    },
    {
      "epoch": 0.46148255813953487,
      "grad_norm": 16.980710983276367,
      "learning_rate": 5.3888081395348845e-05,
      "loss": 0.8185,
      "step": 1270
    },
    {
      "epoch": 0.46511627906976744,
      "grad_norm": 21.34583854675293,
      "learning_rate": 5.352470930232558e-05,
      "loss": 0.8666,
      "step": 1280
    },
    {
      "epoch": 0.46875,
      "grad_norm": 15.561858177185059,
      "learning_rate": 5.316133720930233e-05,
      "loss": 0.8883,
      "step": 1290
    },
    {
      "epoch": 0.47238372093023256,
      "grad_norm": 24.96396255493164,
      "learning_rate": 5.2797965116279066e-05,
      "loss": 0.7988,
      "step": 1300
    },
    {
      "epoch": 0.47238372093023256,
      "eval_accuracy": 0.7486975176218205,
      "eval_loss": 0.7476297616958618,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.5075,
      "eval_samples_per_second": 67.268,
      "eval_steps_per_second": 2.103,
      "step": 1300
    },
    {
      "epoch": 0.47601744186046513,
      "grad_norm": 25.246776580810547,
      "learning_rate": 5.243459302325582e-05,
      "loss": 0.9025,
      "step": 1310
    },
    {
      "epoch": 0.4796511627906977,
      "grad_norm": 18.316572189331055,
      "learning_rate": 5.2071220930232554e-05,
      "loss": 0.7593,
      "step": 1320
    },
    {
      "epoch": 0.48328488372093026,
      "grad_norm": 22.122472763061523,
      "learning_rate": 5.170784883720931e-05,
      "loss": 0.8634,
      "step": 1330
    },
    {
      "epoch": 0.48691860465116277,
      "grad_norm": 24.7249698638916,
      "learning_rate": 5.1344476744186055e-05,
      "loss": 0.7407,
      "step": 1340
    },
    {
      "epoch": 0.49055232558139533,
      "grad_norm": 27.893335342407227,
      "learning_rate": 5.0981104651162795e-05,
      "loss": 0.9492,
      "step": 1350
    },
    {
      "epoch": 0.49055232558139533,
      "eval_accuracy": 0.6944529574011645,
      "eval_loss": 0.8117865920066833,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.5657,
      "eval_samples_per_second": 67.187,
      "eval_steps_per_second": 2.1,
      "step": 1350
    },
    {
      "epoch": 0.4941860465116279,
      "grad_norm": 19.129026412963867,
      "learning_rate": 5.061773255813954e-05,
      "loss": 0.9274,
      "step": 1360
    },
    {
      "epoch": 0.49781976744186046,
      "grad_norm": 18.911165237426758,
      "learning_rate": 5.0254360465116276e-05,
      "loss": 0.807,
      "step": 1370
    },
    {
      "epoch": 0.501453488372093,
      "grad_norm": 19.536846160888672,
      "learning_rate": 4.989098837209303e-05,
      "loss": 0.8404,
      "step": 1380
    },
    {
      "epoch": 0.5050872093023255,
      "grad_norm": 23.16112518310547,
      "learning_rate": 4.952761627906977e-05,
      "loss": 0.812,
      "step": 1390
    },
    {
      "epoch": 0.5087209302325582,
      "grad_norm": 24.2547664642334,
      "learning_rate": 4.916424418604652e-05,
      "loss": 0.8047,
      "step": 1400
    },
    {
      "epoch": 0.5087209302325582,
      "eval_accuracy": 0.7391970579221575,
      "eval_loss": 0.7529688477516174,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.0516,
      "eval_samples_per_second": 65.193,
      "eval_steps_per_second": 2.038,
      "step": 1400
    },
    {
      "epoch": 0.5123546511627907,
      "grad_norm": 21.53682518005371,
      "learning_rate": 4.880087209302326e-05,
      "loss": 0.799,
      "step": 1410
    },
    {
      "epoch": 0.5159883720930233,
      "grad_norm": 23.544010162353516,
      "learning_rate": 4.8437500000000005e-05,
      "loss": 0.7933,
      "step": 1420
    },
    {
      "epoch": 0.5196220930232558,
      "grad_norm": 19.342622756958008,
      "learning_rate": 4.8074127906976745e-05,
      "loss": 0.8245,
      "step": 1430
    },
    {
      "epoch": 0.5232558139534884,
      "grad_norm": 20.08898162841797,
      "learning_rate": 4.771075581395349e-05,
      "loss": 0.8104,
      "step": 1440
    },
    {
      "epoch": 0.5268895348837209,
      "grad_norm": 36.42573165893555,
      "learning_rate": 4.734738372093023e-05,
      "loss": 0.8601,
      "step": 1450
    },
    {
      "epoch": 0.5268895348837209,
      "eval_accuracy": 0.7477781182960466,
      "eval_loss": 0.7395877838134766,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.2379,
      "eval_samples_per_second": 67.644,
      "eval_steps_per_second": 2.115,
      "step": 1450
    },
    {
      "epoch": 0.5305232558139535,
      "grad_norm": 18.966615676879883,
      "learning_rate": 4.6984011627906973e-05,
      "loss": 0.8142,
      "step": 1460
    },
    {
      "epoch": 0.534156976744186,
      "grad_norm": 19.256010055541992,
      "learning_rate": 4.662063953488373e-05,
      "loss": 0.7733,
      "step": 1470
    },
    {
      "epoch": 0.5377906976744186,
      "grad_norm": 17.08919334411621,
      "learning_rate": 4.625726744186047e-05,
      "loss": 0.7369,
      "step": 1480
    },
    {
      "epoch": 0.5414244186046512,
      "grad_norm": 30.637798309326172,
      "learning_rate": 4.5893895348837215e-05,
      "loss": 0.7535,
      "step": 1490
    },
    {
      "epoch": 0.5450581395348837,
      "grad_norm": 16.82142448425293,
      "learning_rate": 4.5530523255813955e-05,
      "loss": 0.7873,
      "step": 1500
    },
    {
      "epoch": 0.5450581395348837,
      "eval_accuracy": 0.7618755746245786,
      "eval_loss": 0.7191266417503357,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 47.9425,
      "eval_samples_per_second": 68.061,
      "eval_steps_per_second": 2.128,
      "step": 1500
    },
    {
      "epoch": 0.5486918604651163,
      "grad_norm": 15.000924110412598,
      "learning_rate": 4.51671511627907e-05,
      "loss": 0.7462,
      "step": 1510
    },
    {
      "epoch": 0.5523255813953488,
      "grad_norm": 19.328227996826172,
      "learning_rate": 4.480377906976744e-05,
      "loss": 0.7997,
      "step": 1520
    },
    {
      "epoch": 0.5559593023255814,
      "grad_norm": 17.77477264404297,
      "learning_rate": 4.444040697674418e-05,
      "loss": 0.7797,
      "step": 1530
    },
    {
      "epoch": 0.559593023255814,
      "grad_norm": 19.144107818603516,
      "learning_rate": 4.407703488372093e-05,
      "loss": 0.7292,
      "step": 1540
    },
    {
      "epoch": 0.5632267441860465,
      "grad_norm": 23.17219352722168,
      "learning_rate": 4.371366279069768e-05,
      "loss": 0.704,
      "step": 1550
    },
    {
      "epoch": 0.5632267441860465,
      "eval_accuracy": 0.7581979773214833,
      "eval_loss": 0.6839959621429443,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.3014,
      "eval_samples_per_second": 67.555,
      "eval_steps_per_second": 2.112,
      "step": 1550
    },
    {
      "epoch": 0.5668604651162791,
      "grad_norm": 17.988855361938477,
      "learning_rate": 4.3350290697674425e-05,
      "loss": 0.8192,
      "step": 1560
    },
    {
      "epoch": 0.5704941860465116,
      "grad_norm": 39.58876037597656,
      "learning_rate": 4.2986918604651165e-05,
      "loss": 0.7845,
      "step": 1570
    },
    {
      "epoch": 0.5741279069767442,
      "grad_norm": 24.698930740356445,
      "learning_rate": 4.262354651162791e-05,
      "loss": 0.7521,
      "step": 1580
    },
    {
      "epoch": 0.5777616279069767,
      "grad_norm": 17.967649459838867,
      "learning_rate": 4.226017441860465e-05,
      "loss": 0.8063,
      "step": 1590
    },
    {
      "epoch": 0.5813953488372093,
      "grad_norm": 25.17499351501465,
      "learning_rate": 4.18968023255814e-05,
      "loss": 0.8536,
      "step": 1600
    },
    {
      "epoch": 0.5813953488372093,
      "eval_accuracy": 0.7581979773214833,
      "eval_loss": 0.6965886950492859,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 48.257,
      "eval_samples_per_second": 67.617,
      "eval_steps_per_second": 2.114,
      "step": 1600
    },
    {
      "epoch": 0.5850290697674418,
      "grad_norm": 27.656429290771484,
      "learning_rate": 4.153343023255814e-05,
      "loss": 0.7501,
      "step": 1610
    },
    {
      "epoch": 0.5886627906976745,
      "grad_norm": 19.678503036499023,
      "learning_rate": 4.117005813953488e-05,
      "loss": 0.7515,
      "step": 1620
    },
    {
      "epoch": 0.592296511627907,
      "grad_norm": 19.597240447998047,
      "learning_rate": 4.080668604651163e-05,
      "loss": 0.7465,
      "step": 1630
    },
    {
      "epoch": 0.5959302325581395,
      "grad_norm": 24.52513885498047,
      "learning_rate": 4.0443313953488375e-05,
      "loss": 0.6374,
      "step": 1640
    },
    {
      "epoch": 0.5995639534883721,
      "grad_norm": 27.33843231201172,
      "learning_rate": 4.007994186046512e-05,
      "loss": 0.7612,
      "step": 1650
    },
    {
      "epoch": 0.5995639534883721,
      "eval_accuracy": 0.7894575543977934,
      "eval_loss": 0.6561300754547119,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 53.5897,
      "eval_samples_per_second": 60.889,
      "eval_steps_per_second": 1.903,
      "step": 1650
    },
    {
      "epoch": 0.6031976744186046,
      "grad_norm": 17.966524124145508,
      "learning_rate": 3.971656976744186e-05,
      "loss": 0.6416,
      "step": 1660
    },
    {
      "epoch": 0.6068313953488372,
      "grad_norm": 24.99458885192871,
      "learning_rate": 3.935319767441861e-05,
      "loss": 0.7307,
      "step": 1670
    },
    {
      "epoch": 0.6104651162790697,
      "grad_norm": 17.375545501708984,
      "learning_rate": 3.898982558139535e-05,
      "loss": 0.6845,
      "step": 1680
    },
    {
      "epoch": 0.6140988372093024,
      "grad_norm": 21.309844970703125,
      "learning_rate": 3.86264534883721e-05,
      "loss": 0.7197,
      "step": 1690
    },
    {
      "epoch": 0.6177325581395349,
      "grad_norm": 19.636783599853516,
      "learning_rate": 3.826308139534884e-05,
      "loss": 0.7701,
      "step": 1700
    },
    {
      "epoch": 0.6177325581395349,
      "eval_accuracy": 0.7931351517008888,
      "eval_loss": 0.6381626725196838,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.3975,
      "eval_samples_per_second": 63.486,
      "eval_steps_per_second": 1.985,
      "step": 1700
    },
    {
      "epoch": 0.6213662790697675,
      "grad_norm": 16.005521774291992,
      "learning_rate": 3.789970930232558e-05,
      "loss": 0.73,
      "step": 1710
    },
    {
      "epoch": 0.625,
      "grad_norm": 22.36250114440918,
      "learning_rate": 3.7536337209302325e-05,
      "loss": 0.7021,
      "step": 1720
    },
    {
      "epoch": 0.6286337209302325,
      "grad_norm": 27.990943908691406,
      "learning_rate": 3.717296511627907e-05,
      "loss": 0.6955,
      "step": 1730
    },
    {
      "epoch": 0.6322674418604651,
      "grad_norm": 16.45665168762207,
      "learning_rate": 3.680959302325582e-05,
      "loss": 0.7612,
      "step": 1740
    },
    {
      "epoch": 0.6359011627906976,
      "grad_norm": 37.34355163574219,
      "learning_rate": 3.644622093023256e-05,
      "loss": 0.7538,
      "step": 1750
    },
    {
      "epoch": 0.6359011627906976,
      "eval_accuracy": 0.7821023597916028,
      "eval_loss": 0.6540101766586304,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.9202,
      "eval_samples_per_second": 64.081,
      "eval_steps_per_second": 2.003,
      "step": 1750
    },
    {
      "epoch": 0.6395348837209303,
      "grad_norm": 22.030466079711914,
      "learning_rate": 3.608284883720931e-05,
      "loss": 0.7048,
      "step": 1760
    },
    {
      "epoch": 0.6431686046511628,
      "grad_norm": 16.459775924682617,
      "learning_rate": 3.571947674418605e-05,
      "loss": 0.7338,
      "step": 1770
    },
    {
      "epoch": 0.6468023255813954,
      "grad_norm": 20.406429290771484,
      "learning_rate": 3.535610465116279e-05,
      "loss": 0.7017,
      "step": 1780
    },
    {
      "epoch": 0.6504360465116279,
      "grad_norm": 32.876670837402344,
      "learning_rate": 3.4992732558139535e-05,
      "loss": 0.7467,
      "step": 1790
    },
    {
      "epoch": 0.6540697674418605,
      "grad_norm": 21.809101104736328,
      "learning_rate": 3.4629360465116276e-05,
      "loss": 0.7496,
      "step": 1800
    },
    {
      "epoch": 0.6540697674418605,
      "eval_accuracy": 0.7735212994177137,
      "eval_loss": 0.6603609323501587,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 54.6689,
      "eval_samples_per_second": 59.687,
      "eval_steps_per_second": 1.866,
      "step": 1800
    },
    {
      "epoch": 0.657703488372093,
      "grad_norm": 17.124296188354492,
      "learning_rate": 3.426598837209303e-05,
      "loss": 0.6712,
      "step": 1810
    },
    {
      "epoch": 0.6613372093023255,
      "grad_norm": 17.997486114501953,
      "learning_rate": 3.390261627906977e-05,
      "loss": 0.6917,
      "step": 1820
    },
    {
      "epoch": 0.6649709302325582,
      "grad_norm": 30.947837829589844,
      "learning_rate": 3.353924418604652e-05,
      "loss": 0.7091,
      "step": 1830
    },
    {
      "epoch": 0.6686046511627907,
      "grad_norm": 24.643381118774414,
      "learning_rate": 3.317587209302326e-05,
      "loss": 0.6695,
      "step": 1840
    },
    {
      "epoch": 0.6722383720930233,
      "grad_norm": 23.651065826416016,
      "learning_rate": 3.2812500000000005e-05,
      "loss": 0.7512,
      "step": 1850
    },
    {
      "epoch": 0.6722383720930233,
      "eval_accuracy": 0.8032485442844008,
      "eval_loss": 0.6195615530014038,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.8657,
      "eval_samples_per_second": 62.913,
      "eval_steps_per_second": 1.967,
      "step": 1850
    },
    {
      "epoch": 0.6758720930232558,
      "grad_norm": 19.081098556518555,
      "learning_rate": 3.2449127906976745e-05,
      "loss": 0.6343,
      "step": 1860
    },
    {
      "epoch": 0.6795058139534884,
      "grad_norm": 23.336320877075195,
      "learning_rate": 3.2085755813953486e-05,
      "loss": 0.7375,
      "step": 1870
    },
    {
      "epoch": 0.6831395348837209,
      "grad_norm": 19.397146224975586,
      "learning_rate": 3.172238372093023e-05,
      "loss": 0.6674,
      "step": 1880
    },
    {
      "epoch": 0.6867732558139535,
      "grad_norm": 21.171167373657227,
      "learning_rate": 3.135901162790697e-05,
      "loss": 0.7242,
      "step": 1890
    },
    {
      "epoch": 0.690406976744186,
      "grad_norm": 24.170560836791992,
      "learning_rate": 3.099563953488373e-05,
      "loss": 0.6773,
      "step": 1900
    },
    {
      "epoch": 0.690406976744186,
      "eval_accuracy": 0.8075390744713454,
      "eval_loss": 0.6085334420204163,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.933,
      "eval_samples_per_second": 64.065,
      "eval_steps_per_second": 2.003,
      "step": 1900
    },
    {
      "epoch": 0.6940406976744186,
      "grad_norm": 19.82205581665039,
      "learning_rate": 3.063226744186047e-05,
      "loss": 0.7194,
      "step": 1910
    },
    {
      "epoch": 0.6976744186046512,
      "grad_norm": 19.169923782348633,
      "learning_rate": 3.026889534883721e-05,
      "loss": 0.6967,
      "step": 1920
    },
    {
      "epoch": 0.7013081395348837,
      "grad_norm": 18.116230010986328,
      "learning_rate": 2.9905523255813955e-05,
      "loss": 0.705,
      "step": 1930
    },
    {
      "epoch": 0.7049418604651163,
      "grad_norm": 16.55962562561035,
      "learning_rate": 2.95421511627907e-05,
      "loss": 0.6809,
      "step": 1940
    },
    {
      "epoch": 0.7085755813953488,
      "grad_norm": 34.48362731933594,
      "learning_rate": 2.9178779069767443e-05,
      "loss": 0.6655,
      "step": 1950
    },
    {
      "epoch": 0.7085755813953488,
      "eval_accuracy": 0.7995709469813056,
      "eval_loss": 0.6168962717056274,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 49.9889,
      "eval_samples_per_second": 65.274,
      "eval_steps_per_second": 2.04,
      "step": 1950
    },
    {
      "epoch": 0.7122093023255814,
      "grad_norm": 27.100082397460938,
      "learning_rate": 2.8815406976744186e-05,
      "loss": 0.6525,
      "step": 1960
    },
    {
      "epoch": 0.715843023255814,
      "grad_norm": 15.424944877624512,
      "learning_rate": 2.845203488372093e-05,
      "loss": 0.7398,
      "step": 1970
    },
    {
      "epoch": 0.7194767441860465,
      "grad_norm": 18.602054595947266,
      "learning_rate": 2.8088662790697677e-05,
      "loss": 0.7341,
      "step": 1980
    },
    {
      "epoch": 0.7231104651162791,
      "grad_norm": 18.60715675354004,
      "learning_rate": 2.772529069767442e-05,
      "loss": 0.62,
      "step": 1990
    },
    {
      "epoch": 0.7267441860465116,
      "grad_norm": 18.766788482666016,
      "learning_rate": 2.7361918604651165e-05,
      "loss": 0.5997,
      "step": 2000
    },
    {
      "epoch": 0.7267441860465116,
      "eval_accuracy": 0.8004903463070794,
      "eval_loss": 0.6034106016159058,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.3413,
      "eval_samples_per_second": 64.818,
      "eval_steps_per_second": 2.026,
      "step": 2000
    },
    {
      "epoch": 0.7303779069767442,
      "grad_norm": 20.883230209350586,
      "learning_rate": 2.699854651162791e-05,
      "loss": 0.6445,
      "step": 2010
    },
    {
      "epoch": 0.7340116279069767,
      "grad_norm": 17.367563247680664,
      "learning_rate": 2.6635174418604652e-05,
      "loss": 0.6727,
      "step": 2020
    },
    {
      "epoch": 0.7376453488372093,
      "grad_norm": 21.541458129882812,
      "learning_rate": 2.6271802325581396e-05,
      "loss": 0.6984,
      "step": 2030
    },
    {
      "epoch": 0.7412790697674418,
      "grad_norm": 25.181142807006836,
      "learning_rate": 2.590843023255814e-05,
      "loss": 0.6668,
      "step": 2040
    },
    {
      "epoch": 0.7449127906976745,
      "grad_norm": 19.112144470214844,
      "learning_rate": 2.5545058139534884e-05,
      "loss": 0.6144,
      "step": 2050
    },
    {
      "epoch": 0.7449127906976745,
      "eval_accuracy": 0.8075390744713454,
      "eval_loss": 0.6008254885673523,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.7754,
      "eval_samples_per_second": 64.263,
      "eval_steps_per_second": 2.009,
      "step": 2050
    },
    {
      "epoch": 0.748546511627907,
      "grad_norm": 23.80943489074707,
      "learning_rate": 2.5181686046511628e-05,
      "loss": 0.6887,
      "step": 2060
    },
    {
      "epoch": 0.7521802325581395,
      "grad_norm": 22.5864200592041,
      "learning_rate": 2.481831395348837e-05,
      "loss": 0.6269,
      "step": 2070
    },
    {
      "epoch": 0.7558139534883721,
      "grad_norm": 20.330726623535156,
      "learning_rate": 2.4454941860465115e-05,
      "loss": 0.6929,
      "step": 2080
    },
    {
      "epoch": 0.7594476744186046,
      "grad_norm": 24.566627502441406,
      "learning_rate": 2.4091569767441862e-05,
      "loss": 0.5906,
      "step": 2090
    },
    {
      "epoch": 0.7630813953488372,
      "grad_norm": 23.221717834472656,
      "learning_rate": 2.3728197674418606e-05,
      "loss": 0.6519,
      "step": 2100
    },
    {
      "epoch": 0.7630813953488372,
      "eval_accuracy": 0.8231688630095004,
      "eval_loss": 0.5798885822296143,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 52.3499,
      "eval_samples_per_second": 62.331,
      "eval_steps_per_second": 1.948,
      "step": 2100
    },
    {
      "epoch": 0.7667151162790697,
      "grad_norm": 25.42754364013672,
      "learning_rate": 2.336482558139535e-05,
      "loss": 0.6025,
      "step": 2110
    },
    {
      "epoch": 0.7703488372093024,
      "grad_norm": 17.46138572692871,
      "learning_rate": 2.3001453488372094e-05,
      "loss": 0.6702,
      "step": 2120
    },
    {
      "epoch": 0.7739825581395349,
      "grad_norm": 19.8096923828125,
      "learning_rate": 2.263808139534884e-05,
      "loss": 0.6543,
      "step": 2130
    },
    {
      "epoch": 0.7776162790697675,
      "grad_norm": 18.788450241088867,
      "learning_rate": 2.2274709302325585e-05,
      "loss": 0.704,
      "step": 2140
    },
    {
      "epoch": 0.78125,
      "grad_norm": 21.987895965576172,
      "learning_rate": 2.1911337209302325e-05,
      "loss": 0.6834,
      "step": 2150
    },
    {
      "epoch": 0.78125,
      "eval_accuracy": 0.8292981918479927,
      "eval_loss": 0.5684976577758789,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.8943,
      "eval_samples_per_second": 62.878,
      "eval_steps_per_second": 1.966,
      "step": 2150
    },
    {
      "epoch": 0.7848837209302325,
      "grad_norm": 22.958097457885742,
      "learning_rate": 2.154796511627907e-05,
      "loss": 0.631,
      "step": 2160
    },
    {
      "epoch": 0.7885174418604651,
      "grad_norm": 20.562345504760742,
      "learning_rate": 2.1184593023255813e-05,
      "loss": 0.6287,
      "step": 2170
    },
    {
      "epoch": 0.7921511627906976,
      "grad_norm": 14.870391845703125,
      "learning_rate": 2.082122093023256e-05,
      "loss": 0.5629,
      "step": 2180
    },
    {
      "epoch": 0.7957848837209303,
      "grad_norm": 23.68880844116211,
      "learning_rate": 2.0457848837209304e-05,
      "loss": 0.6285,
      "step": 2190
    },
    {
      "epoch": 0.7994186046511628,
      "grad_norm": 23.51922035217285,
      "learning_rate": 2.0094476744186047e-05,
      "loss": 0.6258,
      "step": 2200
    },
    {
      "epoch": 0.7994186046511628,
      "eval_accuracy": 0.8240882623352743,
      "eval_loss": 0.5687205791473389,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.3481,
      "eval_samples_per_second": 63.547,
      "eval_steps_per_second": 1.986,
      "step": 2200
    },
    {
      "epoch": 0.8030523255813954,
      "grad_norm": 18.42901611328125,
      "learning_rate": 1.973110465116279e-05,
      "loss": 0.6338,
      "step": 2210
    },
    {
      "epoch": 0.8066860465116279,
      "grad_norm": 24.76369857788086,
      "learning_rate": 1.936773255813954e-05,
      "loss": 0.6247,
      "step": 2220
    },
    {
      "epoch": 0.8103197674418605,
      "grad_norm": 21.530895233154297,
      "learning_rate": 1.9004360465116282e-05,
      "loss": 0.5696,
      "step": 2230
    },
    {
      "epoch": 0.813953488372093,
      "grad_norm": 24.415956497192383,
      "learning_rate": 1.8640988372093023e-05,
      "loss": 0.621,
      "step": 2240
    },
    {
      "epoch": 0.8175872093023255,
      "grad_norm": 24.22767448425293,
      "learning_rate": 1.8277616279069766e-05,
      "loss": 0.6685,
      "step": 2250
    },
    {
      "epoch": 0.8175872093023255,
      "eval_accuracy": 0.8243947287771989,
      "eval_loss": 0.5570237636566162,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 52.4926,
      "eval_samples_per_second": 62.161,
      "eval_steps_per_second": 1.943,
      "step": 2250
    },
    {
      "epoch": 0.8212209302325582,
      "grad_norm": 15.582283973693848,
      "learning_rate": 1.7914244186046513e-05,
      "loss": 0.6792,
      "step": 2260
    },
    {
      "epoch": 0.8248546511627907,
      "grad_norm": 21.79721450805664,
      "learning_rate": 1.7550872093023257e-05,
      "loss": 0.5503,
      "step": 2270
    },
    {
      "epoch": 0.8284883720930233,
      "grad_norm": 17.022436141967773,
      "learning_rate": 1.71875e-05,
      "loss": 0.5733,
      "step": 2280
    },
    {
      "epoch": 0.8321220930232558,
      "grad_norm": 19.0545711517334,
      "learning_rate": 1.6824127906976745e-05,
      "loss": 0.5846,
      "step": 2290
    },
    {
      "epoch": 0.8357558139534884,
      "grad_norm": 23.020610809326172,
      "learning_rate": 1.646075581395349e-05,
      "loss": 0.6844,
      "step": 2300
    },
    {
      "epoch": 0.8357558139534884,
      "eval_accuracy": 0.8253141281029728,
      "eval_loss": 0.5579020977020264,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.8057,
      "eval_samples_per_second": 64.225,
      "eval_steps_per_second": 2.008,
      "step": 2300
    },
    {
      "epoch": 0.8393895348837209,
      "grad_norm": 20.574384689331055,
      "learning_rate": 1.6097383720930236e-05,
      "loss": 0.6167,
      "step": 2310
    },
    {
      "epoch": 0.8430232558139535,
      "grad_norm": 28.64992332458496,
      "learning_rate": 1.5734011627906976e-05,
      "loss": 0.6587,
      "step": 2320
    },
    {
      "epoch": 0.846656976744186,
      "grad_norm": 15.830301284790039,
      "learning_rate": 1.537063953488372e-05,
      "loss": 0.6342,
      "step": 2330
    },
    {
      "epoch": 0.8502906976744186,
      "grad_norm": 18.624250411987305,
      "learning_rate": 1.5007267441860465e-05,
      "loss": 0.6572,
      "step": 2340
    },
    {
      "epoch": 0.8539244186046512,
      "grad_norm": 30.454252243041992,
      "learning_rate": 1.4643895348837211e-05,
      "loss": 0.6446,
      "step": 2350
    },
    {
      "epoch": 0.8539244186046512,
      "eval_accuracy": 0.8243947287771989,
      "eval_loss": 0.5529566407203674,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.0771,
      "eval_samples_per_second": 63.884,
      "eval_steps_per_second": 1.997,
      "step": 2350
    },
    {
      "epoch": 0.8575581395348837,
      "grad_norm": 25.79448127746582,
      "learning_rate": 1.4280523255813955e-05,
      "loss": 0.6686,
      "step": 2360
    },
    {
      "epoch": 0.8611918604651163,
      "grad_norm": 17.485763549804688,
      "learning_rate": 1.3917151162790698e-05,
      "loss": 0.5882,
      "step": 2370
    },
    {
      "epoch": 0.8648255813953488,
      "grad_norm": 19.48297691345215,
      "learning_rate": 1.3553779069767442e-05,
      "loss": 0.6424,
      "step": 2380
    },
    {
      "epoch": 0.8684593023255814,
      "grad_norm": 28.619121551513672,
      "learning_rate": 1.3190406976744188e-05,
      "loss": 0.5695,
      "step": 2390
    },
    {
      "epoch": 0.872093023255814,
      "grad_norm": 16.906198501586914,
      "learning_rate": 1.2827034883720932e-05,
      "loss": 0.6243,
      "step": 2400
    },
    {
      "epoch": 0.872093023255814,
      "eval_accuracy": 0.8277658596383696,
      "eval_loss": 0.5536515712738037,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 50.8517,
      "eval_samples_per_second": 64.167,
      "eval_steps_per_second": 2.006,
      "step": 2400
    },
    {
      "epoch": 0.8757267441860465,
      "grad_norm": 21.633602142333984,
      "learning_rate": 1.2463662790697675e-05,
      "loss": 0.5708,
      "step": 2410
    },
    {
      "epoch": 0.8793604651162791,
      "grad_norm": 28.101240158081055,
      "learning_rate": 1.2100290697674419e-05,
      "loss": 0.6923,
      "step": 2420
    },
    {
      "epoch": 0.8829941860465116,
      "grad_norm": 16.087495803833008,
      "learning_rate": 1.1736918604651163e-05,
      "loss": 0.4912,
      "step": 2430
    },
    {
      "epoch": 0.8866279069767442,
      "grad_norm": 18.913463592529297,
      "learning_rate": 1.1373546511627907e-05,
      "loss": 0.583,
      "step": 2440
    },
    {
      "epoch": 0.8902616279069767,
      "grad_norm": 26.900217056274414,
      "learning_rate": 1.1010174418604652e-05,
      "loss": 0.617,
      "step": 2450
    },
    {
      "epoch": 0.8902616279069767,
      "eval_accuracy": 0.8372663193380325,
      "eval_loss": 0.5327870845794678,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.6187,
      "eval_samples_per_second": 63.214,
      "eval_steps_per_second": 1.976,
      "step": 2450
    },
    {
      "epoch": 0.8938953488372093,
      "grad_norm": 18.022111892700195,
      "learning_rate": 1.0646802325581396e-05,
      "loss": 0.547,
      "step": 2460
    },
    {
      "epoch": 0.8975290697674418,
      "grad_norm": 16.78557777404785,
      "learning_rate": 1.028343023255814e-05,
      "loss": 0.5687,
      "step": 2470
    },
    {
      "epoch": 0.9011627906976745,
      "grad_norm": 20.011295318603516,
      "learning_rate": 9.920058139534884e-06,
      "loss": 0.6218,
      "step": 2480
    },
    {
      "epoch": 0.904796511627907,
      "grad_norm": 22.754287719726562,
      "learning_rate": 9.556686046511629e-06,
      "loss": 0.6328,
      "step": 2490
    },
    {
      "epoch": 0.9084302325581395,
      "grad_norm": 19.715274810791016,
      "learning_rate": 9.193313953488373e-06,
      "loss": 0.5724,
      "step": 2500
    },
    {
      "epoch": 0.9084302325581395,
      "eval_accuracy": 0.840024517315354,
      "eval_loss": 0.5281281471252441,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 52.2053,
      "eval_samples_per_second": 62.503,
      "eval_steps_per_second": 1.954,
      "step": 2500
    },
    {
      "epoch": 0.9120639534883721,
      "grad_norm": 19.64166831970215,
      "learning_rate": 8.829941860465117e-06,
      "loss": 0.5446,
      "step": 2510
    },
    {
      "epoch": 0.9156976744186046,
      "grad_norm": 19.133466720581055,
      "learning_rate": 8.46656976744186e-06,
      "loss": 0.569,
      "step": 2520
    },
    {
      "epoch": 0.9193313953488372,
      "grad_norm": 20.076690673828125,
      "learning_rate": 8.103197674418606e-06,
      "loss": 0.6545,
      "step": 2530
    },
    {
      "epoch": 0.9229651162790697,
      "grad_norm": 18.778846740722656,
      "learning_rate": 7.73982558139535e-06,
      "loss": 0.5915,
      "step": 2540
    },
    {
      "epoch": 0.9265988372093024,
      "grad_norm": 18.11453628540039,
      "learning_rate": 7.376453488372094e-06,
      "loss": 0.4931,
      "step": 2550
    },
    {
      "epoch": 0.9265988372093024,
      "eval_accuracy": 0.8440085810603739,
      "eval_loss": 0.5274325609207153,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 52.3555,
      "eval_samples_per_second": 62.324,
      "eval_steps_per_second": 1.948,
      "step": 2550
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 26.267868041992188,
      "learning_rate": 7.013081395348837e-06,
      "loss": 0.5701,
      "step": 2560
    },
    {
      "epoch": 0.9338662790697675,
      "grad_norm": 34.12333679199219,
      "learning_rate": 6.649709302325581e-06,
      "loss": 0.5949,
      "step": 2570
    },
    {
      "epoch": 0.9375,
      "grad_norm": 18.76346778869629,
      "learning_rate": 6.286337209302326e-06,
      "loss": 0.5934,
      "step": 2580
    },
    {
      "epoch": 0.9411337209302325,
      "grad_norm": 28.602293014526367,
      "learning_rate": 5.92296511627907e-06,
      "loss": 0.5537,
      "step": 2590
    },
    {
      "epoch": 0.9447674418604651,
      "grad_norm": 24.983774185180664,
      "learning_rate": 5.559593023255814e-06,
      "loss": 0.6459,
      "step": 2600
    },
    {
      "epoch": 0.9447674418604651,
      "eval_accuracy": 0.8446215139442231,
      "eval_loss": 0.5270681977272034,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.3715,
      "eval_samples_per_second": 63.518,
      "eval_steps_per_second": 1.986,
      "step": 2600
    },
    {
      "epoch": 0.9484011627906976,
      "grad_norm": 20.338668823242188,
      "learning_rate": 5.196220930232559e-06,
      "loss": 0.5579,
      "step": 2610
    },
    {
      "epoch": 0.9520348837209303,
      "grad_norm": 13.569022178649902,
      "learning_rate": 4.832848837209302e-06,
      "loss": 0.5295,
      "step": 2620
    },
    {
      "epoch": 0.9556686046511628,
      "grad_norm": 17.16480827331543,
      "learning_rate": 4.469476744186047e-06,
      "loss": 0.5583,
      "step": 2630
    },
    {
      "epoch": 0.9593023255813954,
      "grad_norm": 25.31192398071289,
      "learning_rate": 4.106104651162791e-06,
      "loss": 0.5767,
      "step": 2640
    },
    {
      "epoch": 0.9629360465116279,
      "grad_norm": 14.07887077331543,
      "learning_rate": 3.7427325581395346e-06,
      "loss": 0.5954,
      "step": 2650
    },
    {
      "epoch": 0.9629360465116279,
      "eval_accuracy": 0.8446215139442231,
      "eval_loss": 0.5231106281280518,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 52.8556,
      "eval_samples_per_second": 61.734,
      "eval_steps_per_second": 1.93,
      "step": 2650
    },
    {
      "epoch": 0.9665697674418605,
      "grad_norm": 17.648242950439453,
      "learning_rate": 3.379360465116279e-06,
      "loss": 0.5344,
      "step": 2660
    },
    {
      "epoch": 0.970203488372093,
      "grad_norm": 15.185200691223145,
      "learning_rate": 3.0159883720930235e-06,
      "loss": 0.6055,
      "step": 2670
    },
    {
      "epoch": 0.9738372093023255,
      "grad_norm": 23.906478881835938,
      "learning_rate": 2.6526162790697672e-06,
      "loss": 0.5212,
      "step": 2680
    },
    {
      "epoch": 0.9774709302325582,
      "grad_norm": 22.67589569091797,
      "learning_rate": 2.2892441860465114e-06,
      "loss": 0.6287,
      "step": 2690
    },
    {
      "epoch": 0.9811046511627907,
      "grad_norm": 18.85926055908203,
      "learning_rate": 1.9258720930232557e-06,
      "loss": 0.5407,
      "step": 2700
    },
    {
      "epoch": 0.9811046511627907,
      "eval_accuracy": 0.8467667790376954,
      "eval_loss": 0.5167918801307678,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.0533,
      "eval_samples_per_second": 63.914,
      "eval_steps_per_second": 1.998,
      "step": 2700
    },
    {
      "epoch": 0.9847383720930233,
      "grad_norm": 17.184297561645508,
      "learning_rate": 1.5625e-06,
      "loss": 0.653,
      "step": 2710
    },
    {
      "epoch": 0.9883720930232558,
      "grad_norm": 27.768354415893555,
      "learning_rate": 1.1991279069767443e-06,
      "loss": 0.5727,
      "step": 2720
    },
    {
      "epoch": 0.9920058139534884,
      "grad_norm": 22.3969783782959,
      "learning_rate": 8.357558139534884e-07,
      "loss": 0.603,
      "step": 2730
    },
    {
      "epoch": 0.9956395348837209,
      "grad_norm": 17.257413864135742,
      "learning_rate": 4.7238372093023254e-07,
      "loss": 0.5879,
      "step": 2740
    },
    {
      "epoch": 0.9992732558139535,
      "grad_norm": 28.563993453979492,
      "learning_rate": 1.0901162790697675e-07,
      "loss": 0.5729,
      "step": 2750
    },
    {
      "epoch": 0.9992732558139535,
      "eval_accuracy": 0.8479926448053938,
      "eval_loss": 0.5158767700195312,
      "eval_model_preparation_time": 0.0015,
      "eval_runtime": 51.352,
      "eval_samples_per_second": 63.542,
      "eval_steps_per_second": 1.986,
      "step": 2750
    }
  ],
  "logging_steps": 10,
  "max_steps": 2752,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.055112855342285e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}