{
  "best_metric": 0.7839129072681704,
  "best_model_checkpoint": "vit-base-patch16-224-in21k-finetuned/checkpoint-3192",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 4788,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.5625e-06,
      "loss": 1.5128,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.125e-06,
      "loss": 1.3411,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 1.2153,
      "step": 30
    },
    {
      "epoch": 0.05,
      "learning_rate": 6.25e-06,
      "loss": 1.2697,
      "step": 40
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.8125e-06,
      "loss": 1.1394,
      "step": 50
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.0215,
      "step": 60
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.09375e-05,
      "loss": 0.9762,
      "step": 70
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.25e-05,
      "loss": 0.8893,
      "step": 80
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.4062500000000001e-05,
      "loss": 0.9721,
      "step": 90
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.5625e-05,
      "loss": 0.9281,
      "step": 100
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.71875e-05,
      "loss": 0.9426,
      "step": 110
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.8554,
      "step": 120
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.0312500000000002e-05,
      "loss": 0.808,
      "step": 130
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.1875e-05,
      "loss": 0.8681,
      "step": 140
    },
    {
      "epoch": 0.19,
      "learning_rate": 2.34375e-05,
      "loss": 0.897,
      "step": 150
    },
    {
      "epoch": 0.2,
      "learning_rate": 2.5e-05,
      "loss": 0.861,
      "step": 160
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.6562500000000002e-05,
      "loss": 0.9079,
      "step": 170
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.9046,
      "step": 180
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.96875e-05,
      "loss": 0.8711,
      "step": 190
    },
    {
      "epoch": 0.25,
      "learning_rate": 3.125e-05,
      "loss": 0.8316,
      "step": 200
    },
    {
      "epoch": 0.26,
      "learning_rate": 3.2812500000000005e-05,
      "loss": 0.829,
      "step": 210
    },
    {
      "epoch": 0.28,
      "learning_rate": 3.4375e-05,
      "loss": 0.8234,
      "step": 220
    },
    {
      "epoch": 0.29,
      "learning_rate": 3.59375e-05,
      "loss": 0.8658,
      "step": 230
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.8653,
      "step": 240
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.90625e-05,
      "loss": 0.8014,
      "step": 250
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 0.7934,
      "step": 260
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.21875e-05,
      "loss": 0.8023,
      "step": 270
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.375e-05,
      "loss": 0.8294,
      "step": 280
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.5312500000000004e-05,
      "loss": 0.7873,
      "step": 290
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.6875e-05,
      "loss": 0.8071,
      "step": 300
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.8437500000000005e-05,
      "loss": 0.8086,
      "step": 310
    },
    {
      "epoch": 0.4,
      "learning_rate": 5e-05,
      "loss": 0.8296,
      "step": 320
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.9825905292479115e-05,
      "loss": 0.8096,
      "step": 330
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.965181058495822e-05,
      "loss": 0.861,
      "step": 340
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9477715877437334e-05,
      "loss": 0.8047,
      "step": 350
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.930362116991643e-05,
      "loss": 0.8182,
      "step": 360
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.9129526462395546e-05,
      "loss": 0.8412,
      "step": 370
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.895543175487465e-05,
      "loss": 0.8438,
      "step": 380
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.8781337047353764e-05,
      "loss": 0.7776,
      "step": 390
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.860724233983287e-05,
      "loss": 0.8141,
      "step": 400
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.843314763231198e-05,
      "loss": 0.9028,
      "step": 410
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.825905292479109e-05,
      "loss": 0.8347,
      "step": 420
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.80849582172702e-05,
      "loss": 0.8383,
      "step": 430
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.79108635097493e-05,
      "loss": 0.8052,
      "step": 440
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.7736768802228414e-05,
      "loss": 0.8401,
      "step": 450
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.756267409470752e-05,
      "loss": 0.7737,
      "step": 460
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.738857938718663e-05,
      "loss": 0.7741,
      "step": 470
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.721448467966574e-05,
      "loss": 0.8272,
      "step": 480
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.704038997214485e-05,
      "loss": 0.7843,
      "step": 490
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.686629526462396e-05,
      "loss": 0.8069,
      "step": 500
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.669220055710306e-05,
      "loss": 0.7725,
      "step": 510
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.6518105849582176e-05,
      "loss": 0.8299,
      "step": 520
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.634401114206128e-05,
      "loss": 0.7864,
      "step": 530
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.6169916434540394e-05,
      "loss": 0.8226,
      "step": 540
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.59958217270195e-05,
      "loss": 0.785,
      "step": 550
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.582172701949861e-05,
      "loss": 0.8298,
      "step": 560
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.564763231197772e-05,
      "loss": 0.7736,
      "step": 570
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.547353760445683e-05,
      "loss": 0.8041,
      "step": 580
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.529944289693593e-05,
      "loss": 0.7805,
      "step": 590
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.5125348189415044e-05,
      "loss": 0.8247,
      "step": 600
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.495125348189415e-05,
      "loss": 0.7941,
      "step": 610
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.477715877437326e-05,
      "loss": 0.7644,
      "step": 620
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.460306406685237e-05,
      "loss": 0.77,
      "step": 630
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.442896935933148e-05,
      "loss": 0.778,
      "step": 640
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.425487465181059e-05,
      "loss": 0.8061,
      "step": 650
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.40807799442897e-05,
      "loss": 0.7544,
      "step": 660
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.39066852367688e-05,
      "loss": 0.8129,
      "step": 670
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.373259052924791e-05,
      "loss": 0.7776,
      "step": 680
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.355849582172702e-05,
      "loss": 0.8078,
      "step": 690
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.338440111420613e-05,
      "loss": 0.7845,
      "step": 700
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.321030640668524e-05,
      "loss": 0.7826,
      "step": 710
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.303621169916435e-05,
      "loss": 0.7844,
      "step": 720
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.286211699164346e-05,
      "loss": 0.7905,
      "step": 730
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.268802228412256e-05,
      "loss": 0.7858,
      "step": 740
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.2513927576601673e-05,
      "loss": 0.822,
      "step": 750
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.233983286908078e-05,
      "loss": 0.7246,
      "step": 760
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.216573816155989e-05,
      "loss": 0.797,
      "step": 770
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.1991643454039e-05,
      "loss": 0.7749,
      "step": 780
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.181754874651811e-05,
      "loss": 0.7964,
      "step": 790
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7270911654135338,
      "eval_loss": 0.7869051694869995,
      "eval_runtime": 240.7482,
      "eval_samples_per_second": 106.069,
      "eval_steps_per_second": 3.315,
      "step": 798
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.164345403899722e-05,
      "loss": 0.7452,
      "step": 800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.146935933147633e-05,
      "loss": 0.6894,
      "step": 810
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.129526462395543e-05,
      "loss": 0.7115,
      "step": 820
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.112116991643454e-05,
      "loss": 0.6841,
      "step": 830
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.094707520891365e-05,
      "loss": 0.6697,
      "step": 840
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.077298050139276e-05,
      "loss": 0.6796,
      "step": 850
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.0598885793871866e-05,
      "loss": 0.7095,
      "step": 860
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.042479108635098e-05,
      "loss": 0.667,
      "step": 870
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.0250696378830085e-05,
      "loss": 0.678,
      "step": 880
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.007660167130919e-05,
      "loss": 0.6771,
      "step": 890
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.9902506963788303e-05,
      "loss": 0.7135,
      "step": 900
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.972841225626741e-05,
      "loss": 0.6982,
      "step": 910
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.955431754874652e-05,
      "loss": 0.7658,
      "step": 920
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.938022284122563e-05,
      "loss": 0.6841,
      "step": 930
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.920612813370474e-05,
      "loss": 0.7158,
      "step": 940
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.903203342618385e-05,
      "loss": 0.7176,
      "step": 950
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.885793871866296e-05,
      "loss": 0.7032,
      "step": 960
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.868384401114206e-05,
      "loss": 0.7131,
      "step": 970
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.850974930362117e-05,
      "loss": 0.7125,
      "step": 980
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.833565459610028e-05,
      "loss": 0.7089,
      "step": 990
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.816155988857939e-05,
      "loss": 0.7246,
      "step": 1000
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.7987465181058496e-05,
      "loss": 0.6763,
      "step": 1010
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.781337047353761e-05,
      "loss": 0.6693,
      "step": 1020
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.7639275766016715e-05,
      "loss": 0.6912,
      "step": 1030
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.746518105849583e-05,
      "loss": 0.6866,
      "step": 1040
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.7291086350974927e-05,
      "loss": 0.6532,
      "step": 1050
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.711699164345404e-05,
      "loss": 0.7283,
      "step": 1060
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.6942896935933145e-05,
      "loss": 0.7032,
      "step": 1070
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.676880222841226e-05,
      "loss": 0.6854,
      "step": 1080
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.659470752089137e-05,
      "loss": 0.6769,
      "step": 1090
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.642061281337048e-05,
      "loss": 0.7349,
      "step": 1100
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.624651810584959e-05,
      "loss": 0.7274,
      "step": 1110
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.607242339832869e-05,
      "loss": 0.6858,
      "step": 1120
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.58983286908078e-05,
      "loss": 0.7229,
      "step": 1130
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.572423398328691e-05,
      "loss": 0.6924,
      "step": 1140
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.555013927576602e-05,
      "loss": 0.686,
      "step": 1150
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.5376044568245126e-05,
      "loss": 0.7268,
      "step": 1160
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.520194986072424e-05,
      "loss": 0.7107,
      "step": 1170
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.5027855153203345e-05,
      "loss": 0.7095,
      "step": 1180
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.485376044568246e-05,
      "loss": 0.6764,
      "step": 1190
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.4679665738161556e-05,
      "loss": 0.7076,
      "step": 1200
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.450557103064067e-05,
      "loss": 0.7194,
      "step": 1210
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.4331476323119775e-05,
      "loss": 0.7164,
      "step": 1220
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.415738161559889e-05,
      "loss": 0.6958,
      "step": 1230
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.3983286908077994e-05,
      "loss": 0.662,
      "step": 1240
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.3809192200557107e-05,
      "loss": 0.6475,
      "step": 1250
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.363509749303621e-05,
      "loss": 0.7169,
      "step": 1260
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.346100278551532e-05,
      "loss": 0.6948,
      "step": 1270
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.328690807799443e-05,
      "loss": 0.7413,
      "step": 1280
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.311281337047354e-05,
      "loss": 0.6824,
      "step": 1290
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.293871866295265e-05,
      "loss": 0.7024,
      "step": 1300
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.2764623955431756e-05,
      "loss": 0.6748,
      "step": 1310
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.259052924791087e-05,
      "loss": 0.7112,
      "step": 1320
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.2416434540389975e-05,
      "loss": 0.6878,
      "step": 1330
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.224233983286909e-05,
      "loss": 0.7114,
      "step": 1340
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.2068245125348186e-05,
      "loss": 0.6924,
      "step": 1350
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.18941504178273e-05,
      "loss": 0.7354,
      "step": 1360
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.1720055710306405e-05,
      "loss": 0.6757,
      "step": 1370
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.154596100278552e-05,
      "loss": 0.6675,
      "step": 1380
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.1371866295264624e-05,
      "loss": 0.6923,
      "step": 1390
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.1197771587743737e-05,
      "loss": 0.665,
      "step": 1400
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.102367688022284e-05,
      "loss": 0.6473,
      "step": 1410
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.0849582172701955e-05,
      "loss": 0.6194,
      "step": 1420
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.0675487465181054e-05,
      "loss": 0.6535,
      "step": 1430
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.050139275766017e-05,
      "loss": 0.6704,
      "step": 1440
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.0327298050139273e-05,
      "loss": 0.713,
      "step": 1450
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.0153203342618386e-05,
      "loss": 0.7435,
      "step": 1460
    },
    {
      "epoch": 1.84,
      "learning_rate": 2.99791086350975e-05,
      "loss": 0.7073,
      "step": 1470
    },
    {
      "epoch": 1.85,
      "learning_rate": 2.98050139275766e-05,
      "loss": 0.6691,
      "step": 1480
    },
    {
      "epoch": 1.87,
      "learning_rate": 2.9630919220055714e-05,
      "loss": 0.6859,
      "step": 1490
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.945682451253482e-05,
      "loss": 0.7689,
      "step": 1500
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.9282729805013932e-05,
      "loss": 0.6777,
      "step": 1510
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.9108635097493035e-05,
      "loss": 0.6909,
      "step": 1520
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.8934540389972148e-05,
      "loss": 0.7016,
      "step": 1530
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8760445682451254e-05,
      "loss": 0.701,
      "step": 1540
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8586350974930366e-05,
      "loss": 0.7016,
      "step": 1550
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.841225626740947e-05,
      "loss": 0.6674,
      "step": 1560
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.8238161559888582e-05,
      "loss": 0.6663,
      "step": 1570
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.8064066852367688e-05,
      "loss": 0.6621,
      "step": 1580
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.78899721448468e-05,
      "loss": 0.6567,
      "step": 1590
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.738016917293233,
      "eval_loss": 0.7538765668869019,
      "eval_runtime": 243.4161,
      "eval_samples_per_second": 104.907,
      "eval_steps_per_second": 3.278,
      "step": 1596
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.69923416105825e-05,
      "loss": 0.651,
      "step": 1600
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.687630540728708e-05,
      "loss": 0.5874,
      "step": 1610
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.6760269203991646e-05,
      "loss": 0.6426,
      "step": 1620
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.664423300069622e-05,
      "loss": 0.6418,
      "step": 1630
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.652819679740079e-05,
      "loss": 0.6023,
      "step": 1640
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.641216059410536e-05,
      "loss": 0.6012,
      "step": 1650
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.6296124390809935e-05,
      "loss": 0.6276,
      "step": 1660
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.6180088187514504e-05,
      "loss": 0.5936,
      "step": 1670
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.606405198421908e-05,
      "loss": 0.6586,
      "step": 1680
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.594801578092365e-05,
      "loss": 0.6888,
      "step": 1690
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.5831979577628224e-05,
      "loss": 0.6365,
      "step": 1700
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.571594337433279e-05,
      "loss": 0.627,
      "step": 1710
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.559990717103737e-05,
      "loss": 0.6088,
      "step": 1720
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.548387096774194e-05,
      "loss": 0.6739,
      "step": 1730
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.536783476444651e-05,
      "loss": 0.6618,
      "step": 1740
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.5251798561151075e-05,
      "loss": 0.5868,
      "step": 1750
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.513576235785565e-05,
      "loss": 0.6162,
      "step": 1760
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.501972615456022e-05,
      "loss": 0.6023,
      "step": 1770
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.4903689951264796e-05,
      "loss": 0.6005,
      "step": 1780
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.4787653747969365e-05,
      "loss": 0.6084,
      "step": 1790
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.467161754467394e-05,
      "loss": 0.5963,
      "step": 1800
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.455558134137851e-05,
      "loss": 0.6816,
      "step": 1810
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.4439545138083085e-05,
      "loss": 0.6564,
      "step": 1820
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.4323508934787654e-05,
      "loss": 0.6332,
      "step": 1830
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.420747273149223e-05,
      "loss": 0.6295,
      "step": 1840
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.40914365281968e-05,
      "loss": 0.6467,
      "step": 1850
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.397540032490137e-05,
      "loss": 0.622,
      "step": 1860
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.385936412160594e-05,
      "loss": 0.662,
      "step": 1870
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.374332791831051e-05,
      "loss": 0.6514,
      "step": 1880
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.362729171501509e-05,
      "loss": 0.6493,
      "step": 1890
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.351125551171966e-05,
      "loss": 0.6167,
      "step": 1900
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.339521930842423e-05,
      "loss": 0.6575,
      "step": 1910
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.32791831051288e-05,
      "loss": 0.5828,
      "step": 1920
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.316314690183338e-05,
      "loss": 0.6323,
      "step": 1930
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.3047110698537946e-05,
      "loss": 0.6285,
      "step": 1940
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.293107449524252e-05,
      "loss": 0.5822,
      "step": 1950
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.2815038291947084e-05,
      "loss": 0.6441,
      "step": 1960
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.269900208865166e-05,
      "loss": 0.6513,
      "step": 1970
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.258296588535623e-05,
      "loss": 0.6371,
      "step": 1980
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.2466929682060804e-05,
      "loss": 0.6149,
      "step": 1990
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.235089347876537e-05,
      "loss": 0.6348,
      "step": 2000
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.223485727546995e-05,
      "loss": 0.6046,
      "step": 2010
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.211882107217452e-05,
      "loss": 0.6119,
      "step": 2020
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.200278486887909e-05,
      "loss": 0.6382,
      "step": 2030
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.188674866558366e-05,
      "loss": 0.6272,
      "step": 2040
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.177071246228824e-05,
      "loss": 0.6405,
      "step": 2050
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.165467625899281e-05,
      "loss": 0.6679,
      "step": 2060
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.1538640055697376e-05,
      "loss": 0.6367,
      "step": 2070
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.142260385240195e-05,
      "loss": 0.6656,
      "step": 2080
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.130656764910652e-05,
      "loss": 0.6427,
      "step": 2090
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.1190531445811096e-05,
      "loss": 0.6316,
      "step": 2100
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.1074495242515665e-05,
      "loss": 0.6047,
      "step": 2110
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.095845903922024e-05,
      "loss": 0.6562,
      "step": 2120
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.084242283592481e-05,
      "loss": 0.6164,
      "step": 2130
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.0726386632629385e-05,
      "loss": 0.638,
      "step": 2140
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.0610350429333954e-05,
      "loss": 0.6587,
      "step": 2150
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.0494314226038523e-05,
      "loss": 0.6784,
      "step": 2160
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.0378278022743096e-05,
      "loss": 0.6727,
      "step": 2170
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.0262241819447668e-05,
      "loss": 0.6163,
      "step": 2180
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.014620561615224e-05,
      "loss": 0.6304,
      "step": 2190
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0030169412856812e-05,
      "loss": 0.6053,
      "step": 2200
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.9914133209561385e-05,
      "loss": 0.6296,
      "step": 2210
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.9798097006265957e-05,
      "loss": 0.643,
      "step": 2220
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.968206080297053e-05,
      "loss": 0.6033,
      "step": 2230
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.95660245996751e-05,
      "loss": 0.6418,
      "step": 2240
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9449988396379674e-05,
      "loss": 0.6206,
      "step": 2250
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.933395219308424e-05,
      "loss": 0.6655,
      "step": 2260
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.9217915989788812e-05,
      "loss": 0.5923,
      "step": 2270
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.9101879786493384e-05,
      "loss": 0.6219,
      "step": 2280
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.8985843583197956e-05,
      "loss": 0.6688,
      "step": 2290
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.886980737990253e-05,
      "loss": 0.6476,
      "step": 2300
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.87537711766071e-05,
      "loss": 0.6278,
      "step": 2310
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.8637734973311677e-05,
      "loss": 0.5953,
      "step": 2320
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.852169877001625e-05,
      "loss": 0.6612,
      "step": 2330
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.840566256672082e-05,
      "loss": 0.6813,
      "step": 2340
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.8289626363425394e-05,
      "loss": 0.6529,
      "step": 2350
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8173590160129966e-05,
      "loss": 0.5897,
      "step": 2360
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.805755395683453e-05,
      "loss": 0.647,
      "step": 2370
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.7941517753539104e-05,
      "loss": 0.6369,
      "step": 2380
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.7825481550243676e-05,
      "loss": 0.6842,
      "step": 2390
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7837171052631579,
      "eval_loss": 0.6286582946777344,
      "eval_runtime": 244.0222,
      "eval_samples_per_second": 104.646,
      "eval_steps_per_second": 3.27,
      "step": 2394
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.770944534694825e-05,
      "loss": 0.5777,
      "step": 2400
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.759340914365282e-05,
      "loss": 0.5252,
      "step": 2410
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.7477372940357393e-05,
      "loss": 0.5019,
      "step": 2420
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.7361336737061965e-05,
      "loss": 0.5629,
      "step": 2430
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.7245300533766538e-05,
      "loss": 0.4852,
      "step": 2440
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.712926433047111e-05,
      "loss": 0.5356,
      "step": 2450
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.7013228127175682e-05,
      "loss": 0.5717,
      "step": 2460
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.6897191923880248e-05,
      "loss": 0.5351,
      "step": 2470
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.678115572058482e-05,
      "loss": 0.4641,
      "step": 2480
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.6665119517289396e-05,
      "loss": 0.5437,
      "step": 2490
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.6549083313993968e-05,
      "loss": 0.5385,
      "step": 2500
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.643304711069854e-05,
      "loss": 0.5425,
      "step": 2510
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.6317010907403113e-05,
      "loss": 0.5478,
      "step": 2520
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.6200974704107685e-05,
      "loss": 0.5799,
      "step": 2530
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.6084938500812257e-05,
      "loss": 0.5189,
      "step": 2540
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.596890229751683e-05,
      "loss": 0.5384,
      "step": 2550
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.5852866094221402e-05,
      "loss": 0.5281,
      "step": 2560
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.5736829890925968e-05,
      "loss": 0.4708,
      "step": 2570
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.562079368763054e-05,
      "loss": 0.5123,
      "step": 2580
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.5504757484335112e-05,
      "loss": 0.4915,
      "step": 2590
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.5388721281039684e-05,
      "loss": 0.5612,
      "step": 2600
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.5272685077744257e-05,
      "loss": 0.5616,
      "step": 2610
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.515664887444883e-05,
      "loss": 0.5495,
      "step": 2620
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.50406126711534e-05,
      "loss": 0.5311,
      "step": 2630
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.4924576467857974e-05,
      "loss": 0.4424,
      "step": 2640
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.4808540264562543e-05,
      "loss": 0.5303,
      "step": 2650
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.4692504061267115e-05,
      "loss": 0.5391,
      "step": 2660
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.4576467857971687e-05,
      "loss": 0.4972,
      "step": 2670
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.446043165467626e-05,
      "loss": 0.5716,
      "step": 2680
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.4344395451380832e-05,
      "loss": 0.4912,
      "step": 2690
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.4228359248085404e-05,
      "loss": 0.5043,
      "step": 2700
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.4112323044789976e-05,
      "loss": 0.5148,
      "step": 2710
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.399628684149455e-05,
      "loss": 0.5139,
      "step": 2720
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.388025063819912e-05,
      "loss": 0.514,
      "step": 2730
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.3764214434903693e-05,
      "loss": 0.5302,
      "step": 2740
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.3648178231608262e-05,
      "loss": 0.5554,
      "step": 2750
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.3532142028312835e-05,
      "loss": 0.5057,
      "step": 2760
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.3416105825017407e-05,
      "loss": 0.5366,
      "step": 2770
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.330006962172198e-05,
      "loss": 0.5234,
      "step": 2780
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.318403341842655e-05,
      "loss": 0.5847,
      "step": 2790
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.306799721513112e-05,
      "loss": 0.5515,
      "step": 2800
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.2951961011835693e-05,
      "loss": 0.526,
      "step": 2810
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.2835924808540265e-05,
      "loss": 0.5571,
      "step": 2820
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.2719888605244837e-05,
      "loss": 0.5698,
      "step": 2830
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.260385240194941e-05,
      "loss": 0.5039,
      "step": 2840
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.248781619865398e-05,
      "loss": 0.5164,
      "step": 2850
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.237177999535855e-05,
      "loss": 0.5741,
      "step": 2860
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.2255743792063123e-05,
      "loss": 0.5243,
      "step": 2870
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.2139707588767696e-05,
      "loss": 0.5214,
      "step": 2880
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.2023671385472268e-05,
      "loss": 0.5029,
      "step": 2890
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.190763518217684e-05,
      "loss": 0.5528,
      "step": 2900
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.1791598978881413e-05,
      "loss": 0.5115,
      "step": 2910
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.1675562775585985e-05,
      "loss": 0.5399,
      "step": 2920
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.1559526572290557e-05,
      "loss": 0.5467,
      "step": 2930
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.144349036899513e-05,
      "loss": 0.5583,
      "step": 2940
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.1327454165699702e-05,
      "loss": 0.5826,
      "step": 2950
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.121141796240427e-05,
      "loss": 0.5042,
      "step": 2960
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.1095381759108843e-05,
      "loss": 0.4947,
      "step": 2970
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.0979345555813415e-05,
      "loss": 0.5185,
      "step": 2980
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.0863309352517988e-05,
      "loss": 0.539,
      "step": 2990
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.074727314922256e-05,
      "loss": 0.5194,
      "step": 3000
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.063123694592713e-05,
      "loss": 0.5049,
      "step": 3010
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.05152007426317e-05,
      "loss": 0.4813,
      "step": 3020
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.0399164539336273e-05,
      "loss": 0.4671,
      "step": 3030
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.0283128336040846e-05,
      "loss": 0.5214,
      "step": 3040
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.0167092132745418e-05,
      "loss": 0.5476,
      "step": 3050
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.0051055929449987e-05,
      "loss": 0.5231,
      "step": 3060
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.993501972615456e-05,
      "loss": 0.5671,
      "step": 3070
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.981898352285913e-05,
      "loss": 0.5647,
      "step": 3080
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.9702947319563704e-05,
      "loss": 0.5219,
      "step": 3090
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.9586911116268276e-05,
      "loss": 0.5142,
      "step": 3100
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.947087491297285e-05,
      "loss": 0.5069,
      "step": 3110
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.935483870967742e-05,
      "loss": 0.5581,
      "step": 3120
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.9238802506381993e-05,
      "loss": 0.4728,
      "step": 3130
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.9122766303086565e-05,
      "loss": 0.5513,
      "step": 3140
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.9006730099791138e-05,
      "loss": 0.5409,
      "step": 3150
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.8890693896495707e-05,
      "loss": 0.5164,
      "step": 3160
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.877465769320028e-05,
      "loss": 0.5415,
      "step": 3170
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.865862148990485e-05,
      "loss": 0.52,
      "step": 3180
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.8542585286609424e-05,
      "loss": 0.5242,
      "step": 3190
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7839129072681704,
      "eval_loss": 0.62819904088974,
      "eval_runtime": 245.126,
      "eval_samples_per_second": 104.175,
      "eval_steps_per_second": 3.255,
      "step": 3192
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.8426549083313996e-05,
      "loss": 0.45,
      "step": 3200
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.8310512880018565e-05,
      "loss": 0.4147,
      "step": 3210
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.8194476676723137e-05,
      "loss": 0.415,
      "step": 3220
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.807844047342771e-05,
      "loss": 0.4206,
      "step": 3230
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.7962404270132282e-05,
      "loss": 0.4103,
      "step": 3240
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.7846368066836854e-05,
      "loss": 0.4269,
      "step": 3250
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.7730331863541423e-05,
      "loss": 0.4498,
      "step": 3260
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.7614295660245995e-05,
      "loss": 0.4054,
      "step": 3270
    },
    {
      "epoch": 4.11,
      "learning_rate": 1.7498259456950568e-05,
      "loss": 0.4487,
      "step": 3280
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.738222325365514e-05,
      "loss": 0.4052,
      "step": 3290
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.7266187050359716e-05,
      "loss": 0.4106,
      "step": 3300
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.7150150847064285e-05,
      "loss": 0.4004,
      "step": 3310
    },
    {
      "epoch": 4.16,
      "learning_rate": 1.7034114643768857e-05,
      "loss": 0.433,
      "step": 3320
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.691807844047343e-05,
      "loss": 0.3979,
      "step": 3330
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.6802042237178e-05,
      "loss": 0.4445,
      "step": 3340
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.6686006033882574e-05,
      "loss": 0.4357,
      "step": 3350
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.6569969830587146e-05,
      "loss": 0.4279,
      "step": 3360
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.6453933627291715e-05,
      "loss": 0.4089,
      "step": 3370
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.6337897423996287e-05,
      "loss": 0.4039,
      "step": 3380
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.622186122070086e-05,
      "loss": 0.4216,
      "step": 3390
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.6105825017405432e-05,
      "loss": 0.4535,
      "step": 3400
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.5989788814110004e-05,
      "loss": 0.4646,
      "step": 3410
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.5873752610814573e-05,
      "loss": 0.4324,
      "step": 3420
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.5757716407519145e-05,
      "loss": 0.4402,
      "step": 3430
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.5641680204223718e-05,
      "loss": 0.4123,
      "step": 3440
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.552564400092829e-05,
      "loss": 0.4742,
      "step": 3450
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.5409607797632862e-05,
      "loss": 0.4177,
      "step": 3460
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.5293571594337435e-05,
      "loss": 0.4215,
      "step": 3470
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.5177535391042005e-05,
      "loss": 0.3837,
      "step": 3480
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.5061499187746578e-05,
      "loss": 0.4275,
      "step": 3490
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.494546298445115e-05,
      "loss": 0.4526,
      "step": 3500
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.4829426781155722e-05,
      "loss": 0.3635,
      "step": 3510
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.4713390577860293e-05,
      "loss": 0.4344,
      "step": 3520
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.4597354374564865e-05,
      "loss": 0.4282,
      "step": 3530
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.4481318171269437e-05,
      "loss": 0.4838,
      "step": 3540
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.436528196797401e-05,
      "loss": 0.3926,
      "step": 3550
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.4249245764678582e-05,
      "loss": 0.4181,
      "step": 3560
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.4133209561383151e-05,
      "loss": 0.4614,
      "step": 3570
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.4017173358087723e-05,
      "loss": 0.4338,
      "step": 3580
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.3901137154792296e-05,
      "loss": 0.4306,
      "step": 3590
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.3785100951496868e-05,
      "loss": 0.4294,
      "step": 3600
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.366906474820144e-05,
      "loss": 0.4147,
      "step": 3610
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.3553028544906011e-05,
      "loss": 0.4619,
      "step": 3620
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.3436992341610583e-05,
      "loss": 0.4146,
      "step": 3630
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.3320956138315155e-05,
      "loss": 0.4718,
      "step": 3640
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.3204919935019728e-05,
      "loss": 0.3824,
      "step": 3650
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.30888837317243e-05,
      "loss": 0.4173,
      "step": 3660
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.2972847528428869e-05,
      "loss": 0.4221,
      "step": 3670
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.2856811325133441e-05,
      "loss": 0.4027,
      "step": 3680
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.2740775121838014e-05,
      "loss": 0.4,
      "step": 3690
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.2624738918542586e-05,
      "loss": 0.3792,
      "step": 3700
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.2508702715247158e-05,
      "loss": 0.4146,
      "step": 3710
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.2392666511951729e-05,
      "loss": 0.4145,
      "step": 3720
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.2276630308656301e-05,
      "loss": 0.4651,
      "step": 3730
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.2160594105360874e-05,
      "loss": 0.4459,
      "step": 3740
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.2044557902065446e-05,
      "loss": 0.4244,
      "step": 3750
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.1928521698770016e-05,
      "loss": 0.476,
      "step": 3760
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.1812485495474589e-05,
      "loss": 0.4,
      "step": 3770
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.1696449292179161e-05,
      "loss": 0.4452,
      "step": 3780
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.1580413088883732e-05,
      "loss": 0.3968,
      "step": 3790
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.1464376885588304e-05,
      "loss": 0.4265,
      "step": 3800
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.1348340682292875e-05,
      "loss": 0.414,
      "step": 3810
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.1232304478997447e-05,
      "loss": 0.4283,
      "step": 3820
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.1116268275702021e-05,
      "loss": 0.4335,
      "step": 3830
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.1000232072406592e-05,
      "loss": 0.4735,
      "step": 3840
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.0884195869111164e-05,
      "loss": 0.4222,
      "step": 3850
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.0768159665815734e-05,
      "loss": 0.4328,
      "step": 3860
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.0652123462520307e-05,
      "loss": 0.4262,
      "step": 3870
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.0536087259224879e-05,
      "loss": 0.4021,
      "step": 3880
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.042005105592945e-05,
      "loss": 0.4266,
      "step": 3890
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.0304014852634022e-05,
      "loss": 0.4273,
      "step": 3900
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.0187978649338594e-05,
      "loss": 0.4609,
      "step": 3910
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.0071942446043167e-05,
      "loss": 0.4008,
      "step": 3920
    },
    {
      "epoch": 4.92,
      "learning_rate": 9.955906242747739e-06,
      "loss": 0.3794,
      "step": 3930
    },
    {
      "epoch": 4.94,
      "learning_rate": 9.83987003945231e-06,
      "loss": 0.3763,
      "step": 3940
    },
    {
      "epoch": 4.95,
      "learning_rate": 9.723833836156882e-06,
      "loss": 0.4246,
      "step": 3950
    },
    {
      "epoch": 4.96,
      "learning_rate": 9.607797632861454e-06,
      "loss": 0.4331,
      "step": 3960
    },
    {
      "epoch": 4.97,
      "learning_rate": 9.491761429566025e-06,
      "loss": 0.4279,
      "step": 3970
    },
    {
      "epoch": 4.99,
      "learning_rate": 9.375725226270597e-06,
      "loss": 0.4245,
      "step": 3980
    },
    {
      "epoch": 5.0,
      "learning_rate": 9.259689022975168e-06,
      "loss": 0.4321,
      "step": 3990
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7823073308270677,
      "eval_loss": 0.6423132419586182,
      "eval_runtime": 243.4032,
      "eval_samples_per_second": 104.912,
      "eval_steps_per_second": 3.279,
      "step": 3990
    },
    {
      "epoch": 5.01,
      "learning_rate": 9.14365281967974e-06,
      "loss": 0.3513,
      "step": 4000
    },
    {
      "epoch": 5.03,
      "learning_rate": 9.027616616384312e-06,
      "loss": 0.337,
      "step": 4010
    },
    {
      "epoch": 5.04,
      "learning_rate": 8.911580413088885e-06,
      "loss": 0.3227,
      "step": 4020
    },
    {
      "epoch": 5.05,
      "learning_rate": 8.795544209793457e-06,
      "loss": 0.3443,
      "step": 4030
    },
    {
      "epoch": 5.06,
      "learning_rate": 8.679508006498028e-06,
      "loss": 0.3443,
      "step": 4040
    },
    {
      "epoch": 5.08,
      "learning_rate": 8.5634718032026e-06,
      "loss": 0.3033,
      "step": 4050
    },
    {
      "epoch": 5.09,
      "learning_rate": 8.447435599907172e-06,
      "loss": 0.3805,
      "step": 4060
    },
    {
      "epoch": 5.1,
      "learning_rate": 8.331399396611743e-06,
      "loss": 0.2883,
      "step": 4070
    },
    {
      "epoch": 5.11,
      "learning_rate": 8.215363193316315e-06,
      "loss": 0.3702,
      "step": 4080
    },
    {
      "epoch": 5.13,
      "learning_rate": 8.099326990020886e-06,
      "loss": 0.3468,
      "step": 4090
    },
    {
      "epoch": 5.14,
      "learning_rate": 7.983290786725458e-06,
      "loss": 0.3937,
      "step": 4100
    },
    {
      "epoch": 5.15,
      "learning_rate": 7.86725458343003e-06,
      "loss": 0.4071,
      "step": 4110
    },
    {
      "epoch": 5.16,
      "learning_rate": 7.751218380134603e-06,
      "loss": 0.3627,
      "step": 4120
    },
    {
      "epoch": 5.18,
      "learning_rate": 7.635182176839175e-06,
      "loss": 0.397,
      "step": 4130
    },
    {
      "epoch": 5.19,
      "learning_rate": 7.519145973543746e-06,
      "loss": 0.338,
      "step": 4140
    },
    {
      "epoch": 5.2,
      "learning_rate": 7.403109770248318e-06,
      "loss": 0.3559,
      "step": 4150
    },
    {
      "epoch": 5.21,
      "learning_rate": 7.28707356695289e-06,
      "loss": 0.3805,
      "step": 4160
    },
    {
      "epoch": 5.23,
      "learning_rate": 7.171037363657461e-06,
      "loss": 0.3679,
      "step": 4170
    },
    {
      "epoch": 5.24,
      "learning_rate": 7.055001160362033e-06,
      "loss": 0.3594,
      "step": 4180
    },
    {
      "epoch": 5.25,
      "learning_rate": 6.938964957066605e-06,
      "loss": 0.3311,
      "step": 4190
    },
    {
      "epoch": 5.26,
      "learning_rate": 6.822928753771177e-06,
      "loss": 0.3348,
      "step": 4200
    },
    {
      "epoch": 5.28,
      "learning_rate": 6.706892550475749e-06,
      "loss": 0.3552,
      "step": 4210
    },
    {
      "epoch": 5.29,
      "learning_rate": 6.59085634718032e-06,
      "loss": 0.34,
      "step": 4220
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.474820143884892e-06,
      "loss": 0.3524,
      "step": 4230
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.358783940589464e-06,
      "loss": 0.3401,
      "step": 4240
    },
    {
      "epoch": 5.33,
      "learning_rate": 6.242747737294036e-06,
      "loss": 0.3467,
      "step": 4250
    },
    {
      "epoch": 5.34,
      "learning_rate": 6.126711533998608e-06,
      "loss": 0.3115,
      "step": 4260
    },
    {
      "epoch": 5.35,
      "learning_rate": 6.01067533070318e-06,
      "loss": 0.344,
      "step": 4270
    },
    {
      "epoch": 5.36,
      "learning_rate": 5.894639127407751e-06,
      "loss": 0.3388,
      "step": 4280
    },
    {
      "epoch": 5.38,
      "learning_rate": 5.778602924112323e-06,
      "loss": 0.3483,
      "step": 4290
    },
    {
      "epoch": 5.39,
      "learning_rate": 5.662566720816896e-06,
      "loss": 0.355,
      "step": 4300
    },
    {
      "epoch": 5.4,
      "learning_rate": 5.546530517521467e-06,
      "loss": 0.3636,
      "step": 4310
    },
    {
      "epoch": 5.41,
      "learning_rate": 5.430494314226039e-06,
      "loss": 0.3675,
      "step": 4320
    },
    {
      "epoch": 5.43,
      "learning_rate": 5.31445811093061e-06,
      "loss": 0.3346,
      "step": 4330
    },
    {
      "epoch": 5.44,
      "learning_rate": 5.198421907635182e-06,
      "loss": 0.3191,
      "step": 4340
    },
    {
      "epoch": 5.45,
      "learning_rate": 5.082385704339755e-06,
      "loss": 0.3095,
      "step": 4350
    },
    {
      "epoch": 5.46,
      "learning_rate": 4.966349501044326e-06,
      "loss": 0.357,
      "step": 4360
    },
    {
      "epoch": 5.48,
      "learning_rate": 4.850313297748898e-06,
      "loss": 0.3123,
      "step": 4370
    },
    {
      "epoch": 5.49,
      "learning_rate": 4.734277094453469e-06,
      "loss": 0.347,
      "step": 4380
    },
    {
      "epoch": 5.5,
      "learning_rate": 4.618240891158042e-06,
      "loss": 0.3338,
      "step": 4390
    },
    {
      "epoch": 5.51,
      "learning_rate": 4.502204687862614e-06,
      "loss": 0.3698,
      "step": 4400
    },
    {
      "epoch": 5.53,
      "learning_rate": 4.386168484567185e-06,
      "loss": 0.3654,
      "step": 4410
    },
    {
      "epoch": 5.54,
      "learning_rate": 4.270132281271757e-06,
      "loss": 0.3458,
      "step": 4420
    },
    {
      "epoch": 5.55,
      "learning_rate": 4.154096077976329e-06,
      "loss": 0.3096,
      "step": 4430
    },
    {
      "epoch": 5.56,
      "learning_rate": 4.038059874680901e-06,
      "loss": 0.3391,
      "step": 4440
    },
    {
      "epoch": 5.58,
      "learning_rate": 3.922023671385473e-06,
      "loss": 0.3683,
      "step": 4450
    },
    {
      "epoch": 5.59,
      "learning_rate": 3.805987468090044e-06,
      "loss": 0.3855,
      "step": 4460
    },
    {
      "epoch": 5.6,
      "learning_rate": 3.689951264794616e-06,
      "loss": 0.353,
      "step": 4470
    },
    {
      "epoch": 5.61,
      "learning_rate": 3.5739150614991875e-06,
      "loss": 0.3303,
      "step": 4480
    },
    {
      "epoch": 5.63,
      "learning_rate": 3.45787885820376e-06,
      "loss": 0.2987,
      "step": 4490
    },
    {
      "epoch": 5.64,
      "learning_rate": 3.3418426549083317e-06,
      "loss": 0.3412,
      "step": 4500
    },
    {
      "epoch": 5.65,
      "learning_rate": 3.225806451612903e-06,
      "loss": 0.3066,
      "step": 4510
    },
    {
      "epoch": 5.66,
      "learning_rate": 3.1097702483174755e-06,
      "loss": 0.3252,
      "step": 4520
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.993734045022047e-06,
      "loss": 0.3243,
      "step": 4530
    },
    {
      "epoch": 5.69,
      "learning_rate": 2.877697841726619e-06,
      "loss": 0.3888,
      "step": 4540
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.7616616384311907e-06,
      "loss": 0.3635,
      "step": 4550
    },
    {
      "epoch": 5.71,
      "learning_rate": 2.645625435135762e-06,
      "loss": 0.2875,
      "step": 4560
    },
    {
      "epoch": 5.73,
      "learning_rate": 2.5295892318403345e-06,
      "loss": 0.3443,
      "step": 4570
    },
    {
      "epoch": 5.74,
      "learning_rate": 2.413553028544906e-06,
      "loss": 0.3334,
      "step": 4580
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.297516825249478e-06,
      "loss": 0.337,
      "step": 4590
    },
    {
      "epoch": 5.76,
      "learning_rate": 2.1814806219540497e-06,
      "loss": 0.3204,
      "step": 4600
    },
    {
      "epoch": 5.78,
      "learning_rate": 2.0654444186586216e-06,
      "loss": 0.318,
      "step": 4610
    },
    {
      "epoch": 5.79,
      "learning_rate": 1.9494082153631935e-06,
      "loss": 0.3113,
      "step": 4620
    },
    {
      "epoch": 5.8,
      "learning_rate": 1.833372012067765e-06,
      "loss": 0.3623,
      "step": 4630
    },
    {
      "epoch": 5.81,
      "learning_rate": 1.717335808772337e-06,
      "loss": 0.3489,
      "step": 4640
    },
    {
      "epoch": 5.83,
      "learning_rate": 1.6012996054769087e-06,
      "loss": 0.3477,
      "step": 4650
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.4852634021814808e-06,
      "loss": 0.3124,
      "step": 4660
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.3692271988860525e-06,
      "loss": 0.3658,
      "step": 4670
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.2531909955906242e-06,
      "loss": 0.3652,
      "step": 4680
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.137154792295196e-06,
      "loss": 0.3415,
      "step": 4690
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.021118588999768e-06,
      "loss": 0.3479,
      "step": 4700
    },
    {
      "epoch": 5.9,
      "learning_rate": 9.050823857043398e-07,
      "loss": 0.3746,
      "step": 4710
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.890461824089116e-07,
      "loss": 0.3311,
      "step": 4720
    },
    {
      "epoch": 5.93,
      "learning_rate": 6.730099791134834e-07,
      "loss": 0.3282,
      "step": 4730
    },
    {
      "epoch": 5.94,
      "learning_rate": 5.569737758180553e-07,
      "loss": 0.3269,
      "step": 4740
    },
    {
      "epoch": 5.95,
      "learning_rate": 4.4093757252262707e-07,
      "loss": 0.3035,
      "step": 4750
    },
    {
      "epoch": 5.96,
      "learning_rate": 3.249013692271989e-07,
      "loss": 0.3208,
      "step": 4760
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.0886516593177073e-07,
      "loss": 0.3553,
      "step": 4770
    },
    {
      "epoch": 5.99,
      "learning_rate": 9.282896263634255e-08,
      "loss": 0.3129,
      "step": 4780
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7837562656641605,
      "eval_loss": 0.6533188819885254,
      "eval_runtime": 249.7809,
      "eval_samples_per_second": 102.234,
      "eval_steps_per_second": 3.195,
      "step": 4788
    },
    {
      "epoch": 6.0,
      "step": 4788,
      "total_flos": 4.749417900430118e+19,
      "train_loss": 0.321556950521748,
      "train_runtime": 10205.5984,
      "train_samples_per_second": 60.052,
      "train_steps_per_second": 0.469
    }
  ],
  "logging_steps": 10,
  "max_steps": 4788,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 4.749417900430118e+19,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}