{
  "best_metric": 0.9624402458456636,
  "best_model_checkpoint": "resnet-50-finetuned-eurosat/checkpoint-2745",
  "epoch": 4.997724169321803,
  "global_step": 2745,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1.818181818181818e-06,
      "loss": 3.6408,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.636363636363636e-06,
      "loss": 3.6412,
      "step": 20
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 3.641,
      "step": 30
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.272727272727272e-06,
      "loss": 3.6404,
      "step": 40
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.090909090909091e-06,
      "loss": 3.6347,
      "step": 50
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 3.6377,
      "step": 60
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.2727272727272727e-05,
      "loss": 3.6316,
      "step": 70
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.4545454545454545e-05,
      "loss": 3.6339,
      "step": 80
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 3.6301,
      "step": 90
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 3.626,
      "step": 100
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 3.621,
      "step": 110
    },
    {
      "epoch": 0.22,
      "learning_rate": 2.1818181818181818e-05,
      "loss": 3.6206,
      "step": 120
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.3636363636363637e-05,
      "loss": 3.6118,
      "step": 130
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.5454545454545454e-05,
      "loss": 3.6068,
      "step": 140
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 3.598,
      "step": 150
    },
    {
      "epoch": 0.29,
      "learning_rate": 2.909090909090909e-05,
      "loss": 3.5991,
      "step": 160
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.090909090909091e-05,
      "loss": 3.5924,
      "step": 170
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.272727272727273e-05,
      "loss": 3.5819,
      "step": 180
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.454545454545455e-05,
      "loss": 3.5726,
      "step": 190
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 3.5638,
      "step": 200
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.818181818181819e-05,
      "loss": 3.5568,
      "step": 210
    },
    {
      "epoch": 0.4,
      "learning_rate": 4e-05,
      "loss": 3.5451,
      "step": 220
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.181818181818182e-05,
      "loss": 3.5333,
      "step": 230
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.3636363636363636e-05,
      "loss": 3.5256,
      "step": 240
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.545454545454546e-05,
      "loss": 3.5079,
      "step": 250
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.7272727272727275e-05,
      "loss": 3.4865,
      "step": 260
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.909090909090909e-05,
      "loss": 3.4667,
      "step": 270
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.989878542510122e-05,
      "loss": 3.4516,
      "step": 280
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.9696356275303645e-05,
      "loss": 3.427,
      "step": 290
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.9493927125506076e-05,
      "loss": 3.3999,
      "step": 300
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.9291497975708506e-05,
      "loss": 3.3732,
      "step": 310
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.908906882591093e-05,
      "loss": 3.3437,
      "step": 320
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.888663967611336e-05,
      "loss": 3.2873,
      "step": 330
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.868421052631579e-05,
      "loss": 3.2593,
      "step": 340
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.848178137651822e-05,
      "loss": 3.2006,
      "step": 350
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.8279352226720646e-05,
      "loss": 3.1593,
      "step": 360
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8076923076923084e-05,
      "loss": 3.1129,
      "step": 370
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.787449392712551e-05,
      "loss": 3.0435,
      "step": 380
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.767206477732794e-05,
      "loss": 2.9742,
      "step": 390
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.746963562753037e-05,
      "loss": 2.9306,
      "step": 400
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.726720647773279e-05,
      "loss": 2.8627,
      "step": 410
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.7064777327935223e-05,
      "loss": 2.8301,
      "step": 420
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.6862348178137654e-05,
      "loss": 2.7201,
      "step": 430
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.6659919028340085e-05,
      "loss": 2.6538,
      "step": 440
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.6457489878542516e-05,
      "loss": 2.5858,
      "step": 450
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.625506072874494e-05,
      "loss": 2.4747,
      "step": 460
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.605263157894737e-05,
      "loss": 2.41,
      "step": 470
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.58502024291498e-05,
      "loss": 2.366,
      "step": 480
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.564777327935223e-05,
      "loss": 2.3017,
      "step": 490
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.5445344129554655e-05,
      "loss": 2.2318,
      "step": 500
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.524291497975709e-05,
      "loss": 2.1774,
      "step": 510
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.504048582995952e-05,
      "loss": 2.089,
      "step": 520
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.483805668016194e-05,
      "loss": 2.0213,
      "step": 530
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.463562753036438e-05,
      "loss": 1.9386,
      "step": 540
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7653084452538129,
      "eval_loss": 1.521803379058838,
      "eval_runtime": 116.2062,
      "eval_samples_per_second": 151.214,
      "eval_steps_per_second": 4.733,
      "step": 549
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.44331983805668e-05,
      "loss": 1.9215,
      "step": 550
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.423076923076923e-05,
      "loss": 1.8288,
      "step": 560
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.402834008097166e-05,
      "loss": 1.761,
      "step": 570
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.3825910931174094e-05,
      "loss": 1.7306,
      "step": 580
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.362348178137652e-05,
      "loss": 1.6337,
      "step": 590
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.342105263157895e-05,
      "loss": 1.5869,
      "step": 600
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.321862348178138e-05,
      "loss": 1.5347,
      "step": 610
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.30161943319838e-05,
      "loss": 1.4658,
      "step": 620
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.281376518218624e-05,
      "loss": 1.444,
      "step": 630
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.2611336032388664e-05,
      "loss": 1.3742,
      "step": 640
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.2408906882591095e-05,
      "loss": 1.3474,
      "step": 650
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.2206477732793526e-05,
      "loss": 1.3503,
      "step": 660
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.200404858299595e-05,
      "loss": 1.2557,
      "step": 670
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.180161943319838e-05,
      "loss": 1.2054,
      "step": 680
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.159919028340081e-05,
      "loss": 1.253,
      "step": 690
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.139676113360324e-05,
      "loss": 1.1814,
      "step": 700
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.1194331983805666e-05,
      "loss": 1.1699,
      "step": 710
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.09919028340081e-05,
      "loss": 1.1464,
      "step": 720
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.078947368421053e-05,
      "loss": 1.1289,
      "step": 730
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.058704453441296e-05,
      "loss": 1.0347,
      "step": 740
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.038461538461539e-05,
      "loss": 1.0617,
      "step": 750
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.018218623481781e-05,
      "loss": 1.0069,
      "step": 760
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.997975708502025e-05,
      "loss": 0.9519,
      "step": 770
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.9777327935222674e-05,
      "loss": 0.9229,
      "step": 780
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.9574898785425104e-05,
      "loss": 0.9309,
      "step": 790
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.9372469635627535e-05,
      "loss": 0.8647,
      "step": 800
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.917004048582996e-05,
      "loss": 0.8625,
      "step": 810
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.896761133603239e-05,
      "loss": 0.9097,
      "step": 820
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.876518218623482e-05,
      "loss": 0.8751,
      "step": 830
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.856275303643725e-05,
      "loss": 0.8613,
      "step": 840
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.8360323886639675e-05,
      "loss": 0.8289,
      "step": 850
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.815789473684211e-05,
      "loss": 0.8145,
      "step": 860
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.7955465587044536e-05,
      "loss": 0.8031,
      "step": 870
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.775303643724697e-05,
      "loss": 0.844,
      "step": 880
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.75506072874494e-05,
      "loss": 0.8124,
      "step": 890
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.734817813765182e-05,
      "loss": 0.7609,
      "step": 900
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.714574898785425e-05,
      "loss": 0.7603,
      "step": 910
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.694331983805668e-05,
      "loss": 0.7959,
      "step": 920
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.6740890688259113e-05,
      "loss": 0.719,
      "step": 930
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.653846153846154e-05,
      "loss": 0.7447,
      "step": 940
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.633603238866397e-05,
      "loss": 0.7588,
      "step": 950
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.61336032388664e-05,
      "loss": 0.6941,
      "step": 960
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.593117408906882e-05,
      "loss": 0.709,
      "step": 970
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.572874493927126e-05,
      "loss": 0.7446,
      "step": 980
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.5526315789473684e-05,
      "loss": 0.6729,
      "step": 990
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.5323886639676115e-05,
      "loss": 0.6262,
      "step": 1000
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.5121457489878545e-05,
      "loss": 0.6725,
      "step": 1010
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.491902834008097e-05,
      "loss": 0.6287,
      "step": 1020
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.47165991902834e-05,
      "loss": 0.6443,
      "step": 1030
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.451417004048583e-05,
      "loss": 0.6444,
      "step": 1040
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.431174089068826e-05,
      "loss": 0.6219,
      "step": 1050
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.410931174089069e-05,
      "loss": 0.6271,
      "step": 1060
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.390688259109312e-05,
      "loss": 0.6186,
      "step": 1070
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.3704453441295546e-05,
      "loss": 0.5921,
      "step": 1080
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.350202429149798e-05,
      "loss": 0.5662,
      "step": 1090
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9288071932620078,
      "eval_loss": 0.33246490359306335,
      "eval_runtime": 116.3356,
      "eval_samples_per_second": 151.046,
      "eval_steps_per_second": 4.728,
      "step": 1098
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.329959514170041e-05,
      "loss": 0.6063,
      "step": 1100
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.309716599190283e-05,
      "loss": 0.5733,
      "step": 1110
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.289473684210527e-05,
      "loss": 0.5662,
      "step": 1120
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.269230769230769e-05,
      "loss": 0.5893,
      "step": 1130
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.2489878542510124e-05,
      "loss": 0.5498,
      "step": 1140
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.2287449392712554e-05,
      "loss": 0.5416,
      "step": 1150
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.208502024291498e-05,
      "loss": 0.5271,
      "step": 1160
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.188259109311741e-05,
      "loss": 0.5506,
      "step": 1170
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.168016194331984e-05,
      "loss": 0.5227,
      "step": 1180
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.147773279352227e-05,
      "loss": 0.5379,
      "step": 1190
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.1275303643724694e-05,
      "loss": 0.4872,
      "step": 1200
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.107287449392713e-05,
      "loss": 0.5166,
      "step": 1210
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.0870445344129556e-05,
      "loss": 0.5416,
      "step": 1220
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.0668016194331986e-05,
      "loss": 0.5171,
      "step": 1230
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0465587044534417e-05,
      "loss": 0.4827,
      "step": 1240
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.0263157894736844e-05,
      "loss": 0.5535,
      "step": 1250
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.006072874493927e-05,
      "loss": 0.5095,
      "step": 1260
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.9858299595141702e-05,
      "loss": 0.4874,
      "step": 1270
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.965587044534413e-05,
      "loss": 0.4798,
      "step": 1280
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.9453441295546557e-05,
      "loss": 0.487,
      "step": 1290
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.925101214574899e-05,
      "loss": 0.4855,
      "step": 1300
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.9048582995951418e-05,
      "loss": 0.4818,
      "step": 1310
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.8846153846153845e-05,
      "loss": 0.4359,
      "step": 1320
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.8643724696356276e-05,
      "loss": 0.4553,
      "step": 1330
    },
    {
      "epoch": 2.44,
      "learning_rate": 2.8441295546558703e-05,
      "loss": 0.4566,
      "step": 1340
    },
    {
      "epoch": 2.46,
      "learning_rate": 2.823886639676113e-05,
      "loss": 0.468,
      "step": 1350
    },
    {
      "epoch": 2.48,
      "learning_rate": 2.8036437246963565e-05,
      "loss": 0.4448,
      "step": 1360
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.7834008097165992e-05,
      "loss": 0.4629,
      "step": 1370
    },
    {
      "epoch": 2.51,
      "learning_rate": 2.7631578947368426e-05,
      "loss": 0.4876,
      "step": 1380
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.7429149797570853e-05,
      "loss": 0.4499,
      "step": 1390
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.722672064777328e-05,
      "loss": 0.4449,
      "step": 1400
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.702429149797571e-05,
      "loss": 0.404,
      "step": 1410
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.682186234817814e-05,
      "loss": 0.4521,
      "step": 1420
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.6619433198380566e-05,
      "loss": 0.4075,
      "step": 1430
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.6417004048583e-05,
      "loss": 0.4528,
      "step": 1440
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.6214574898785427e-05,
      "loss": 0.4615,
      "step": 1450
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.6012145748987855e-05,
      "loss": 0.4654,
      "step": 1460
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.5809716599190285e-05,
      "loss": 0.425,
      "step": 1470
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.5607287449392713e-05,
      "loss": 0.409,
      "step": 1480
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.540485829959514e-05,
      "loss": 0.4284,
      "step": 1490
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.5202429149797574e-05,
      "loss": 0.4122,
      "step": 1500
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.5e-05,
      "loss": 0.4273,
      "step": 1510
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.4797570850202432e-05,
      "loss": 0.3865,
      "step": 1520
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.459514170040486e-05,
      "loss": 0.378,
      "step": 1530
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.439271255060729e-05,
      "loss": 0.4051,
      "step": 1540
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.4190283400809717e-05,
      "loss": 0.4127,
      "step": 1550
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.3987854251012144e-05,
      "loss": 0.4141,
      "step": 1560
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3785425101214575e-05,
      "loss": 0.3842,
      "step": 1570
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.3582995951417006e-05,
      "loss": 0.423,
      "step": 1580
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.3380566801619436e-05,
      "loss": 0.3714,
      "step": 1590
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.3178137651821864e-05,
      "loss": 0.3949,
      "step": 1600
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.2975708502024294e-05,
      "loss": 0.398,
      "step": 1610
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.2773279352226722e-05,
      "loss": 0.3838,
      "step": 1620
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.257085020242915e-05,
      "loss": 0.365,
      "step": 1630
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.236842105263158e-05,
      "loss": 0.4104,
      "step": 1640
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9561233781015251,
      "eval_loss": 0.18532179296016693,
      "eval_runtime": 113.1925,
      "eval_samples_per_second": 155.24,
      "eval_steps_per_second": 4.859,
      "step": 1647
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.216599190283401e-05,
      "loss": 0.4087,
      "step": 1650
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.1963562753036438e-05,
      "loss": 0.3953,
      "step": 1660
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.1761133603238868e-05,
      "loss": 0.3972,
      "step": 1670
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.15587044534413e-05,
      "loss": 0.4105,
      "step": 1680
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.1356275303643726e-05,
      "loss": 0.3761,
      "step": 1690
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.1153846153846154e-05,
      "loss": 0.3555,
      "step": 1700
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.0951417004048584e-05,
      "loss": 0.3801,
      "step": 1710
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.074898785425101e-05,
      "loss": 0.402,
      "step": 1720
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.0546558704453442e-05,
      "loss": 0.3929,
      "step": 1730
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.0344129554655873e-05,
      "loss": 0.3884,
      "step": 1740
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.0141700404858304e-05,
      "loss": 0.3853,
      "step": 1750
    },
    {
      "epoch": 3.2,
      "learning_rate": 1.9939271255060727e-05,
      "loss": 0.3869,
      "step": 1760
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.9736842105263158e-05,
      "loss": 0.3758,
      "step": 1770
    },
    {
      "epoch": 3.24,
      "learning_rate": 1.953441295546559e-05,
      "loss": 0.3961,
      "step": 1780
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.9331983805668016e-05,
      "loss": 0.3389,
      "step": 1790
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.9129554655870447e-05,
      "loss": 0.3624,
      "step": 1800
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.8927125506072877e-05,
      "loss": 0.3905,
      "step": 1810
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.8724696356275305e-05,
      "loss": 0.3685,
      "step": 1820
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.8522267206477732e-05,
      "loss": 0.3582,
      "step": 1830
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.8319838056680163e-05,
      "loss": 0.3785,
      "step": 1840
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.811740890688259e-05,
      "loss": 0.3758,
      "step": 1850
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.791497975708502e-05,
      "loss": 0.386,
      "step": 1860
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.771255060728745e-05,
      "loss": 0.3565,
      "step": 1870
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.751012145748988e-05,
      "loss": 0.3423,
      "step": 1880
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.730769230769231e-05,
      "loss": 0.3859,
      "step": 1890
    },
    {
      "epoch": 3.46,
      "learning_rate": 1.7105263157894737e-05,
      "loss": 0.3606,
      "step": 1900
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.6902834008097167e-05,
      "loss": 0.3829,
      "step": 1910
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.6700404858299595e-05,
      "loss": 0.3409,
      "step": 1920
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.6497975708502025e-05,
      "loss": 0.358,
      "step": 1930
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.6295546558704456e-05,
      "loss": 0.3182,
      "step": 1940
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.6093117408906883e-05,
      "loss": 0.3595,
      "step": 1950
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.5890688259109314e-05,
      "loss": 0.3536,
      "step": 1960
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.568825910931174e-05,
      "loss": 0.3818,
      "step": 1970
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.548582995951417e-05,
      "loss": 0.3401,
      "step": 1980
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.52834008097166e-05,
      "loss": 0.3554,
      "step": 1990
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.508097165991903e-05,
      "loss": 0.3465,
      "step": 2000
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.4878542510121457e-05,
      "loss": 0.3482,
      "step": 2010
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.4676113360323888e-05,
      "loss": 0.3492,
      "step": 2020
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.4473684210526317e-05,
      "loss": 0.3366,
      "step": 2030
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.4271255060728744e-05,
      "loss": 0.3392,
      "step": 2040
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.4068825910931175e-05,
      "loss": 0.3593,
      "step": 2050
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.3866396761133604e-05,
      "loss": 0.3961,
      "step": 2060
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.3663967611336034e-05,
      "loss": 0.3996,
      "step": 2070
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.3461538461538462e-05,
      "loss": 0.3178,
      "step": 2080
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.3259109311740892e-05,
      "loss": 0.3641,
      "step": 2090
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.3056680161943321e-05,
      "loss": 0.3176,
      "step": 2100
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.2854251012145749e-05,
      "loss": 0.3189,
      "step": 2110
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.265182186234818e-05,
      "loss": 0.3663,
      "step": 2120
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.2449392712550607e-05,
      "loss": 0.3105,
      "step": 2130
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.2246963562753037e-05,
      "loss": 0.3479,
      "step": 2140
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.2044534412955466e-05,
      "loss": 0.3301,
      "step": 2150
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.1842105263157895e-05,
      "loss": 0.3344,
      "step": 2160
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.1639676113360324e-05,
      "loss": 0.3169,
      "step": 2170
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.1437246963562753e-05,
      "loss": 0.3486,
      "step": 2180
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.1234817813765184e-05,
      "loss": 0.3551,
      "step": 2190
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9622695196904165,
      "eval_loss": 0.14936527609825134,
      "eval_runtime": 112.1506,
      "eval_samples_per_second": 156.682,
      "eval_steps_per_second": 4.904,
      "step": 2197
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.1032388663967611e-05,
      "loss": 0.3307,
      "step": 2200
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.082995951417004e-05,
      "loss": 0.3035,
      "step": 2210
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.062753036437247e-05,
      "loss": 0.3192,
      "step": 2220
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.04251012145749e-05,
      "loss": 0.3357,
      "step": 2230
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.0222672064777327e-05,
      "loss": 0.3081,
      "step": 2240
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.0020242914979758e-05,
      "loss": 0.3343,
      "step": 2250
    },
    {
      "epoch": 4.11,
      "learning_rate": 9.817813765182187e-06,
      "loss": 0.3438,
      "step": 2260
    },
    {
      "epoch": 4.13,
      "learning_rate": 9.615384615384616e-06,
      "loss": 0.2939,
      "step": 2270
    },
    {
      "epoch": 4.15,
      "learning_rate": 9.412955465587045e-06,
      "loss": 0.2954,
      "step": 2280
    },
    {
      "epoch": 4.17,
      "learning_rate": 9.210526315789474e-06,
      "loss": 0.3206,
      "step": 2290
    },
    {
      "epoch": 4.19,
      "learning_rate": 9.008097165991904e-06,
      "loss": 0.3116,
      "step": 2300
    },
    {
      "epoch": 4.21,
      "learning_rate": 8.805668016194332e-06,
      "loss": 0.3201,
      "step": 2310
    },
    {
      "epoch": 4.22,
      "learning_rate": 8.60323886639676e-06,
      "loss": 0.3439,
      "step": 2320
    },
    {
      "epoch": 4.24,
      "learning_rate": 8.400809716599191e-06,
      "loss": 0.3503,
      "step": 2330
    },
    {
      "epoch": 4.26,
      "learning_rate": 8.19838056680162e-06,
      "loss": 0.317,
      "step": 2340
    },
    {
      "epoch": 4.28,
      "learning_rate": 7.99595141700405e-06,
      "loss": 0.3113,
      "step": 2350
    },
    {
      "epoch": 4.3,
      "learning_rate": 7.793522267206478e-06,
      "loss": 0.311,
      "step": 2360
    },
    {
      "epoch": 4.31,
      "learning_rate": 7.591093117408906e-06,
      "loss": 0.3083,
      "step": 2370
    },
    {
      "epoch": 4.33,
      "learning_rate": 7.388663967611337e-06,
      "loss": 0.3165,
      "step": 2380
    },
    {
      "epoch": 4.35,
      "learning_rate": 7.186234817813765e-06,
      "loss": 0.3547,
      "step": 2390
    },
    {
      "epoch": 4.37,
      "learning_rate": 6.983805668016195e-06,
      "loss": 0.3071,
      "step": 2400
    },
    {
      "epoch": 4.39,
      "learning_rate": 6.781376518218624e-06,
      "loss": 0.3028,
      "step": 2410
    },
    {
      "epoch": 4.41,
      "learning_rate": 6.578947368421053e-06,
      "loss": 0.3215,
      "step": 2420
    },
    {
      "epoch": 4.42,
      "learning_rate": 6.376518218623483e-06,
      "loss": 0.3259,
      "step": 2430
    },
    {
      "epoch": 4.44,
      "learning_rate": 6.174089068825911e-06,
      "loss": 0.3515,
      "step": 2440
    },
    {
      "epoch": 4.46,
      "learning_rate": 5.971659919028341e-06,
      "loss": 0.3158,
      "step": 2450
    },
    {
      "epoch": 4.48,
      "learning_rate": 5.76923076923077e-06,
      "loss": 0.3067,
      "step": 2460
    },
    {
      "epoch": 4.5,
      "learning_rate": 5.566801619433199e-06,
      "loss": 0.324,
      "step": 2470
    },
    {
      "epoch": 4.52,
      "learning_rate": 5.364372469635628e-06,
      "loss": 0.3271,
      "step": 2480
    },
    {
      "epoch": 4.53,
      "learning_rate": 5.161943319838057e-06,
      "loss": 0.3367,
      "step": 2490
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.9595141700404865e-06,
      "loss": 0.3166,
      "step": 2500
    },
    {
      "epoch": 4.57,
      "learning_rate": 4.757085020242915e-06,
      "loss": 0.3264,
      "step": 2510
    },
    {
      "epoch": 4.59,
      "learning_rate": 4.5546558704453445e-06,
      "loss": 0.3129,
      "step": 2520
    },
    {
      "epoch": 4.61,
      "learning_rate": 4.3522267206477735e-06,
      "loss": 0.3114,
      "step": 2530
    },
    {
      "epoch": 4.62,
      "learning_rate": 4.149797570850203e-06,
      "loss": 0.2751,
      "step": 2540
    },
    {
      "epoch": 4.64,
      "learning_rate": 3.9473684210526315e-06,
      "loss": 0.3191,
      "step": 2550
    },
    {
      "epoch": 4.66,
      "learning_rate": 3.744939271255061e-06,
      "loss": 0.3171,
      "step": 2560
    },
    {
      "epoch": 4.68,
      "learning_rate": 3.5425101214574903e-06,
      "loss": 0.2981,
      "step": 2570
    },
    {
      "epoch": 4.7,
      "learning_rate": 3.3400809716599197e-06,
      "loss": 0.3169,
      "step": 2580
    },
    {
      "epoch": 4.72,
      "learning_rate": 3.1376518218623482e-06,
      "loss": 0.304,
      "step": 2590
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.9352226720647772e-06,
      "loss": 0.3021,
      "step": 2600
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.7327935222672066e-06,
      "loss": 0.3106,
      "step": 2610
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.5303643724696356e-06,
      "loss": 0.3066,
      "step": 2620
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.327935222672065e-06,
      "loss": 0.3117,
      "step": 2630
    },
    {
      "epoch": 4.81,
      "learning_rate": 2.125506072874494e-06,
      "loss": 0.3079,
      "step": 2640
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 0.3316,
      "step": 2650
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.7206477732793522e-06,
      "loss": 0.3103,
      "step": 2660
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5182186234817814e-06,
      "loss": 0.3034,
      "step": 2670
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.3157894736842106e-06,
      "loss": 0.294,
      "step": 2680
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.1133603238866398e-06,
      "loss": 0.3151,
      "step": 2690
    },
    {
      "epoch": 4.92,
      "learning_rate": 9.109311740890688e-07,
      "loss": 0.3544,
      "step": 2700
    },
    {
      "epoch": 4.93,
      "learning_rate": 7.08502024291498e-07,
      "loss": 0.3365,
      "step": 2710
    },
    {
      "epoch": 4.95,
      "learning_rate": 5.060728744939271e-07,
      "loss": 0.3181,
      "step": 2720
    },
    {
      "epoch": 4.97,
      "learning_rate": 3.036437246963563e-07,
      "loss": 0.303,
      "step": 2730
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.0121457489878543e-07,
      "loss": 0.3174,
      "step": 2740
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9624402458456636,
      "eval_loss": 0.1446385681629181,
      "eval_runtime": 113.1571,
      "eval_samples_per_second": 155.288,
      "eval_steps_per_second": 4.86,
      "step": 2745
    },
    {
      "epoch": 5.0,
      "step": 2745,
      "total_flos": 7.483912529986142e+18,
      "train_loss": 1.0602939221811207,
      "train_runtime": 5306.8618,
      "train_samples_per_second": 66.23,
      "train_steps_per_second": 0.517
    }
  ],
  "max_steps": 2745,
  "num_train_epochs": 5,
  "total_flos": 7.483912529986142e+18,
  "trial_name": null,
  "trial_params": null
}