{
  "best_metric": 0.8043478260869565,
  "best_model_checkpoint": "BEiT-DMAE-XDA-REVAL-80\\checkpoint-1620",
  "epoch": 80.0,
  "eval_steps": 500,
  "global_step": 4800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 1.6266,
      "step": 10
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 1.6316,
      "step": 20
    },
    {
      "epoch": 0.5,
      "learning_rate": 3.125e-06,
      "loss": 1.6026,
      "step": 30
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.166666666666667e-06,
      "loss": 1.6124,
      "step": 40
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.208333333333334e-06,
      "loss": 1.5523,
      "step": 50
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.25e-06,
      "loss": 1.5404,
      "step": 60
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.45652173913043476,
      "eval_loss": 1.2860267162322998,
      "eval_runtime": 0.6757,
      "eval_samples_per_second": 68.082,
      "eval_steps_per_second": 4.44,
      "step": 60
    },
    {
      "epoch": 1.17,
      "learning_rate": 7.2916666666666674e-06,
      "loss": 1.4915,
      "step": 70
    },
    {
      "epoch": 1.33,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.4587,
      "step": 80
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.4275,
      "step": 90
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 1.3967,
      "step": 100
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.1458333333333333e-05,
      "loss": 1.3448,
      "step": 110
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.25e-05,
      "loss": 1.2921,
      "step": 120
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.4782608695652174,
      "eval_loss": 1.1579527854919434,
      "eval_runtime": 0.6727,
      "eval_samples_per_second": 68.385,
      "eval_steps_per_second": 4.46,
      "step": 120
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3541666666666666e-05,
      "loss": 1.1931,
      "step": 130
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.4583333333333335e-05,
      "loss": 1.1388,
      "step": 140
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.5625e-05,
      "loss": 1.0504,
      "step": 150
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.9868,
      "step": 160
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7708333333333335e-05,
      "loss": 0.9146,
      "step": 170
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.8862,
      "step": 180
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.4782608695652174,
      "eval_loss": 0.9702665209770203,
      "eval_runtime": 0.6477,
      "eval_samples_per_second": 71.025,
      "eval_steps_per_second": 4.632,
      "step": 180
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.9791666666666665e-05,
      "loss": 0.7701,
      "step": 190
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.743,
      "step": 200
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.1875e-05,
      "loss": 0.7122,
      "step": 210
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.2916666666666667e-05,
      "loss": 0.6751,
      "step": 220
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.3958333333333334e-05,
      "loss": 0.5876,
      "step": 230
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.5e-05,
      "loss": 0.5719,
      "step": 240
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 0.8845036625862122,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 70.916,
      "eval_steps_per_second": 4.625,
      "step": 240
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.604166666666667e-05,
      "loss": 0.523,
      "step": 250
    },
    {
      "epoch": 4.33,
      "learning_rate": 2.7083333333333332e-05,
      "loss": 0.4376,
      "step": 260
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.4403,
      "step": 270
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.3861,
      "step": 280
    },
    {
      "epoch": 4.83,
      "learning_rate": 3.0208333333333334e-05,
      "loss": 0.333,
      "step": 290
    },
    {
      "epoch": 5.0,
      "learning_rate": 3.125e-05,
      "loss": 0.3656,
      "step": 300
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 0.7957310080528259,
      "eval_runtime": 0.6312,
      "eval_samples_per_second": 72.882,
      "eval_steps_per_second": 4.753,
      "step": 300
    },
    {
      "epoch": 5.17,
      "learning_rate": 3.229166666666667e-05,
      "loss": 0.3111,
      "step": 310
    },
    {
      "epoch": 5.33,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.3207,
      "step": 320
    },
    {
      "epoch": 5.5,
      "learning_rate": 3.4375e-05,
      "loss": 0.3406,
      "step": 330
    },
    {
      "epoch": 5.67,
      "learning_rate": 3.541666666666667e-05,
      "loss": 0.3338,
      "step": 340
    },
    {
      "epoch": 5.83,
      "learning_rate": 3.6458333333333336e-05,
      "loss": 0.2383,
      "step": 350
    },
    {
      "epoch": 6.0,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.2146,
      "step": 360
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 0.8631888628005981,
      "eval_runtime": 0.6427,
      "eval_samples_per_second": 71.579,
      "eval_steps_per_second": 4.668,
      "step": 360
    },
    {
      "epoch": 6.17,
      "learning_rate": 3.854166666666667e-05,
      "loss": 0.2466,
      "step": 370
    },
    {
      "epoch": 6.33,
      "learning_rate": 3.958333333333333e-05,
      "loss": 0.2325,
      "step": 380
    },
    {
      "epoch": 6.5,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 0.2371,
      "step": 390
    },
    {
      "epoch": 6.67,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.2506,
      "step": 400
    },
    {
      "epoch": 6.83,
      "learning_rate": 4.270833333333333e-05,
      "loss": 0.2066,
      "step": 410
    },
    {
      "epoch": 7.0,
      "learning_rate": 4.375e-05,
      "loss": 0.182,
      "step": 420
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 1.0533522367477417,
      "eval_runtime": 0.6637,
      "eval_samples_per_second": 69.313,
      "eval_steps_per_second": 4.52,
      "step": 420
    },
    {
      "epoch": 7.17,
      "learning_rate": 4.4791666666666673e-05,
      "loss": 0.2197,
      "step": 430
    },
    {
      "epoch": 7.33,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.2128,
      "step": 440
    },
    {
      "epoch": 7.5,
      "learning_rate": 4.6875e-05,
      "loss": 0.1654,
      "step": 450
    },
    {
      "epoch": 7.67,
      "learning_rate": 4.791666666666667e-05,
      "loss": 0.2006,
      "step": 460
    },
    {
      "epoch": 7.83,
      "learning_rate": 4.8958333333333335e-05,
      "loss": 0.1728,
      "step": 470
    },
    {
      "epoch": 8.0,
      "learning_rate": 5e-05,
      "loss": 0.1809,
      "step": 480
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.115868091583252,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 70.916,
      "eval_steps_per_second": 4.625,
      "step": 480
    },
    {
      "epoch": 8.17,
      "learning_rate": 4.988425925925926e-05,
      "loss": 0.1869,
      "step": 490
    },
    {
      "epoch": 8.33,
      "learning_rate": 4.976851851851852e-05,
      "loss": 0.1638,
      "step": 500
    },
    {
      "epoch": 8.5,
      "learning_rate": 4.965277777777778e-05,
      "loss": 0.2048,
      "step": 510
    },
    {
      "epoch": 8.67,
      "learning_rate": 4.9537037037037035e-05,
      "loss": 0.1325,
      "step": 520
    },
    {
      "epoch": 8.83,
      "learning_rate": 4.94212962962963e-05,
      "loss": 0.2008,
      "step": 530
    },
    {
      "epoch": 9.0,
      "learning_rate": 4.930555555555556e-05,
      "loss": 0.1687,
      "step": 540
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 0.796657919883728,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 70.916,
      "eval_steps_per_second": 4.625,
      "step": 540
    },
    {
      "epoch": 9.17,
      "learning_rate": 4.9189814814814815e-05,
      "loss": 0.1808,
      "step": 550
    },
    {
      "epoch": 9.33,
      "learning_rate": 4.9074074074074075e-05,
      "loss": 0.1238,
      "step": 560
    },
    {
      "epoch": 9.5,
      "learning_rate": 4.8958333333333335e-05,
      "loss": 0.1646,
      "step": 570
    },
    {
      "epoch": 9.67,
      "learning_rate": 4.8842592592592595e-05,
      "loss": 0.1544,
      "step": 580
    },
    {
      "epoch": 9.83,
      "learning_rate": 4.8726851851851855e-05,
      "loss": 0.1674,
      "step": 590
    },
    {
      "epoch": 10.0,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 0.1798,
      "step": 600
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 1.1139856576919556,
      "eval_runtime": 0.6532,
      "eval_samples_per_second": 70.427,
      "eval_steps_per_second": 4.593,
      "step": 600
    },
    {
      "epoch": 10.17,
      "learning_rate": 4.8495370370370375e-05,
      "loss": 0.1186,
      "step": 610
    },
    {
      "epoch": 10.33,
      "learning_rate": 4.837962962962963e-05,
      "loss": 0.2239,
      "step": 620
    },
    {
      "epoch": 10.5,
      "learning_rate": 4.8263888888888895e-05,
      "loss": 0.138,
      "step": 630
    },
    {
      "epoch": 10.67,
      "learning_rate": 4.814814814814815e-05,
      "loss": 0.1222,
      "step": 640
    },
    {
      "epoch": 10.83,
      "learning_rate": 4.803240740740741e-05,
      "loss": 0.1348,
      "step": 650
    },
    {
      "epoch": 11.0,
      "learning_rate": 4.791666666666667e-05,
      "loss": 0.1961,
      "step": 660
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.5869565217391305,
      "eval_loss": 1.640942096710205,
      "eval_runtime": 0.6597,
      "eval_samples_per_second": 69.733,
      "eval_steps_per_second": 4.548,
      "step": 660
    },
    {
      "epoch": 11.17,
      "learning_rate": 4.780092592592593e-05,
      "loss": 0.1497,
      "step": 670
    },
    {
      "epoch": 11.33,
      "learning_rate": 4.768518518518519e-05,
      "loss": 0.1203,
      "step": 680
    },
    {
      "epoch": 11.5,
      "learning_rate": 4.756944444444444e-05,
      "loss": 0.1132,
      "step": 690
    },
    {
      "epoch": 11.67,
      "learning_rate": 4.745370370370371e-05,
      "loss": 0.144,
      "step": 700
    },
    {
      "epoch": 11.83,
      "learning_rate": 4.733796296296297e-05,
      "loss": 0.1301,
      "step": 710
    },
    {
      "epoch": 12.0,
      "learning_rate": 4.722222222222222e-05,
      "loss": 0.1475,
      "step": 720
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 1.4891263246536255,
      "eval_runtime": 0.6527,
      "eval_samples_per_second": 70.482,
      "eval_steps_per_second": 4.597,
      "step": 720
    },
    {
      "epoch": 12.17,
      "learning_rate": 4.710648148148149e-05,
      "loss": 0.1248,
      "step": 730
    },
    {
      "epoch": 12.33,
      "learning_rate": 4.699074074074074e-05,
      "loss": 0.1667,
      "step": 740
    },
    {
      "epoch": 12.5,
      "learning_rate": 4.6875e-05,
      "loss": 0.0992,
      "step": 750
    },
    {
      "epoch": 12.67,
      "learning_rate": 4.675925925925926e-05,
      "loss": 0.0981,
      "step": 760
    },
    {
      "epoch": 12.83,
      "learning_rate": 4.664351851851852e-05,
      "loss": 0.1579,
      "step": 770
    },
    {
      "epoch": 13.0,
      "learning_rate": 4.652777777777778e-05,
      "loss": 0.0964,
      "step": 780
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 1.611473798751831,
      "eval_runtime": 0.7197,
      "eval_samples_per_second": 63.918,
      "eval_steps_per_second": 4.169,
      "step": 780
    },
    {
      "epoch": 13.17,
      "learning_rate": 4.6412037037037034e-05,
      "loss": 0.1322,
      "step": 790
    },
    {
      "epoch": 13.33,
      "learning_rate": 4.62962962962963e-05,
      "loss": 0.1037,
      "step": 800
    },
    {
      "epoch": 13.5,
      "learning_rate": 4.618055555555556e-05,
      "loss": 0.1435,
      "step": 810
    },
    {
      "epoch": 13.67,
      "learning_rate": 4.6064814814814814e-05,
      "loss": 0.1106,
      "step": 820
    },
    {
      "epoch": 13.83,
      "learning_rate": 4.594907407407408e-05,
      "loss": 0.0681,
      "step": 830
    },
    {
      "epoch": 14.0,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.0797,
      "step": 840
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 1.4461338520050049,
      "eval_runtime": 0.6512,
      "eval_samples_per_second": 70.644,
      "eval_steps_per_second": 4.607,
      "step": 840
    },
    {
      "epoch": 14.17,
      "learning_rate": 4.5717592592592594e-05,
      "loss": 0.1439,
      "step": 850
    },
    {
      "epoch": 14.33,
      "learning_rate": 4.5601851851851854e-05,
      "loss": 0.1083,
      "step": 860
    },
    {
      "epoch": 14.5,
      "learning_rate": 4.5486111111111114e-05,
      "loss": 0.0731,
      "step": 870
    },
    {
      "epoch": 14.67,
      "learning_rate": 4.5370370370370374e-05,
      "loss": 0.1432,
      "step": 880
    },
    {
      "epoch": 14.83,
      "learning_rate": 4.525462962962963e-05,
      "loss": 0.1108,
      "step": 890
    },
    {
      "epoch": 15.0,
      "learning_rate": 4.5138888888888894e-05,
      "loss": 0.069,
      "step": 900
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 1.3022737503051758,
      "eval_runtime": 0.6407,
      "eval_samples_per_second": 71.802,
      "eval_steps_per_second": 4.683,
      "step": 900
    },
    {
      "epoch": 15.17,
      "learning_rate": 4.502314814814815e-05,
      "loss": 0.0899,
      "step": 910
    },
    {
      "epoch": 15.33,
      "learning_rate": 4.490740740740741e-05,
      "loss": 0.0851,
      "step": 920
    },
    {
      "epoch": 15.5,
      "learning_rate": 4.4791666666666673e-05,
      "loss": 0.0925,
      "step": 930
    },
    {
      "epoch": 15.67,
      "learning_rate": 4.467592592592593e-05,
      "loss": 0.098,
      "step": 940
    },
    {
      "epoch": 15.83,
      "learning_rate": 4.456018518518519e-05,
      "loss": 0.1312,
      "step": 950
    },
    {
      "epoch": 16.0,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.0937,
      "step": 960
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 1.2890084981918335,
      "eval_runtime": 0.6457,
      "eval_samples_per_second": 71.246,
      "eval_steps_per_second": 4.646,
      "step": 960
    },
    {
      "epoch": 16.17,
      "learning_rate": 4.4328703703703707e-05,
      "loss": 0.0771,
      "step": 970
    },
    {
      "epoch": 16.33,
      "learning_rate": 4.4212962962962966e-05,
      "loss": 0.133,
      "step": 980
    },
    {
      "epoch": 16.5,
      "learning_rate": 4.4097222222222226e-05,
      "loss": 0.1166,
      "step": 990
    },
    {
      "epoch": 16.67,
      "learning_rate": 4.3981481481481486e-05,
      "loss": 0.1091,
      "step": 1000
    },
    {
      "epoch": 16.83,
      "learning_rate": 4.386574074074074e-05,
      "loss": 0.0724,
      "step": 1010
    },
    {
      "epoch": 17.0,
      "learning_rate": 4.375e-05,
      "loss": 0.105,
      "step": 1020
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.1600866317749023,
      "eval_runtime": 0.6507,
      "eval_samples_per_second": 70.697,
      "eval_steps_per_second": 4.611,
      "step": 1020
    },
    {
      "epoch": 17.17,
      "learning_rate": 4.3634259259259266e-05,
      "loss": 0.0936,
      "step": 1030
    },
    {
      "epoch": 17.33,
      "learning_rate": 4.351851851851852e-05,
      "loss": 0.0849,
      "step": 1040
    },
    {
      "epoch": 17.5,
      "learning_rate": 4.340277777777778e-05,
      "loss": 0.0885,
      "step": 1050
    },
    {
      "epoch": 17.67,
      "learning_rate": 4.328703703703704e-05,
      "loss": 0.0968,
      "step": 1060
    },
    {
      "epoch": 17.83,
      "learning_rate": 4.31712962962963e-05,
      "loss": 0.0744,
      "step": 1070
    },
    {
      "epoch": 18.0,
      "learning_rate": 4.305555555555556e-05,
      "loss": 0.0909,
      "step": 1080
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.0927455425262451,
      "eval_runtime": 0.6392,
      "eval_samples_per_second": 71.969,
      "eval_steps_per_second": 4.694,
      "step": 1080
    },
    {
      "epoch": 18.17,
      "learning_rate": 4.293981481481482e-05,
      "loss": 0.0831,
      "step": 1090
    },
    {
      "epoch": 18.33,
      "learning_rate": 4.282407407407408e-05,
      "loss": 0.0895,
      "step": 1100
    },
    {
      "epoch": 18.5,
      "learning_rate": 4.270833333333333e-05,
      "loss": 0.0535,
      "step": 1110
    },
    {
      "epoch": 18.67,
      "learning_rate": 4.259259259259259e-05,
      "loss": 0.0682,
      "step": 1120
    },
    {
      "epoch": 18.83,
      "learning_rate": 4.247685185185186e-05,
      "loss": 0.0797,
      "step": 1130
    },
    {
      "epoch": 19.0,
      "learning_rate": 4.236111111111111e-05,
      "loss": 0.0561,
      "step": 1140
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.4283857345581055,
      "eval_runtime": 0.6427,
      "eval_samples_per_second": 71.579,
      "eval_steps_per_second": 4.668,
      "step": 1140
    },
    {
      "epoch": 19.17,
      "learning_rate": 4.224537037037037e-05,
      "loss": 0.0697,
      "step": 1150
    },
    {
      "epoch": 19.33,
      "learning_rate": 4.212962962962963e-05,
      "loss": 0.0563,
      "step": 1160
    },
    {
      "epoch": 19.5,
      "learning_rate": 4.201388888888889e-05,
      "loss": 0.0542,
      "step": 1170
    },
    {
      "epoch": 19.67,
      "learning_rate": 4.1898148148148145e-05,
      "loss": 0.0639,
      "step": 1180
    },
    {
      "epoch": 19.83,
      "learning_rate": 4.178240740740741e-05,
      "loss": 0.0731,
      "step": 1190
    },
    {
      "epoch": 20.0,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.102,
      "step": 1200
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 1.4660907983779907,
      "eval_runtime": 0.6512,
      "eval_samples_per_second": 70.643,
      "eval_steps_per_second": 4.607,
      "step": 1200
    },
    {
      "epoch": 20.17,
      "learning_rate": 4.1550925925925925e-05,
      "loss": 0.0861,
      "step": 1210
    },
    {
      "epoch": 20.33,
      "learning_rate": 4.1435185185185185e-05,
      "loss": 0.067,
      "step": 1220
    },
    {
      "epoch": 20.5,
      "learning_rate": 4.1319444444444445e-05,
      "loss": 0.0625,
      "step": 1230
    },
    {
      "epoch": 20.67,
      "learning_rate": 4.1203703703703705e-05,
      "loss": 0.0837,
      "step": 1240
    },
    {
      "epoch": 20.83,
      "learning_rate": 4.1087962962962965e-05,
      "loss": 0.0823,
      "step": 1250
    },
    {
      "epoch": 21.0,
      "learning_rate": 4.0972222222222225e-05,
      "loss": 0.0632,
      "step": 1260
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.273358941078186,
      "eval_runtime": 0.6627,
      "eval_samples_per_second": 69.418,
      "eval_steps_per_second": 4.527,
      "step": 1260
    },
    {
      "epoch": 21.17,
      "learning_rate": 4.0856481481481485e-05,
      "loss": 0.0545,
      "step": 1270
    },
    {
      "epoch": 21.33,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.044,
      "step": 1280
    },
    {
      "epoch": 21.5,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 0.0677,
      "step": 1290
    },
    {
      "epoch": 21.67,
      "learning_rate": 4.0509259259259265e-05,
      "loss": 0.0743,
      "step": 1300
    },
    {
      "epoch": 21.83,
      "learning_rate": 4.039351851851852e-05,
      "loss": 0.0862,
      "step": 1310
    },
    {
      "epoch": 22.0,
      "learning_rate": 4.027777777777778e-05,
      "loss": 0.0396,
      "step": 1320
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.6521739130434783,
      "eval_loss": 1.816393256187439,
      "eval_runtime": 0.6617,
      "eval_samples_per_second": 69.522,
      "eval_steps_per_second": 4.534,
      "step": 1320
    },
    {
      "epoch": 22.17,
      "learning_rate": 4.016203703703704e-05,
      "loss": 0.0703,
      "step": 1330
    },
    {
      "epoch": 22.33,
      "learning_rate": 4.00462962962963e-05,
      "loss": 0.0894,
      "step": 1340
    },
    {
      "epoch": 22.5,
      "learning_rate": 3.993055555555556e-05,
      "loss": 0.0432,
      "step": 1350
    },
    {
      "epoch": 22.67,
      "learning_rate": 3.981481481481482e-05,
      "loss": 0.0335,
      "step": 1360
    },
    {
      "epoch": 22.83,
      "learning_rate": 3.969907407407408e-05,
      "loss": 0.0638,
      "step": 1370
    },
    {
      "epoch": 23.0,
      "learning_rate": 3.958333333333333e-05,
      "loss": 0.0831,
      "step": 1380
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.510336995124817,
      "eval_runtime": 0.6607,
      "eval_samples_per_second": 69.628,
      "eval_steps_per_second": 4.541,
      "step": 1380
    },
    {
      "epoch": 23.17,
      "learning_rate": 3.94675925925926e-05,
      "loss": 0.0428,
      "step": 1390
    },
    {
      "epoch": 23.33,
      "learning_rate": 3.935185185185186e-05,
      "loss": 0.0747,
      "step": 1400
    },
    {
      "epoch": 23.5,
      "learning_rate": 3.923611111111111e-05,
      "loss": 0.0657,
      "step": 1410
    },
    {
      "epoch": 23.67,
      "learning_rate": 3.912037037037037e-05,
      "loss": 0.0571,
      "step": 1420
    },
    {
      "epoch": 23.83,
      "learning_rate": 3.900462962962963e-05,
      "loss": 0.0695,
      "step": 1430
    },
    {
      "epoch": 24.0,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.0696,
      "step": 1440
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 1.6660609245300293,
      "eval_runtime": 0.6457,
      "eval_samples_per_second": 71.246,
      "eval_steps_per_second": 4.646,
      "step": 1440
    },
    {
      "epoch": 24.17,
      "learning_rate": 3.877314814814815e-05,
      "loss": 0.0566,
      "step": 1450
    },
    {
      "epoch": 24.33,
      "learning_rate": 3.865740740740741e-05,
      "loss": 0.0515,
      "step": 1460
    },
    {
      "epoch": 24.5,
      "learning_rate": 3.854166666666667e-05,
      "loss": 0.0406,
      "step": 1470
    },
    {
      "epoch": 24.67,
      "learning_rate": 3.8425925925925924e-05,
      "loss": 0.0325,
      "step": 1480
    },
    {
      "epoch": 24.83,
      "learning_rate": 3.831018518518519e-05,
      "loss": 0.078,
      "step": 1490
    },
    {
      "epoch": 25.0,
      "learning_rate": 3.8194444444444444e-05,
      "loss": 0.0787,
      "step": 1500
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.5281383991241455,
      "eval_runtime": 0.6512,
      "eval_samples_per_second": 70.643,
      "eval_steps_per_second": 4.607,
      "step": 1500
    },
    {
      "epoch": 25.17,
      "learning_rate": 3.8078703703703704e-05,
      "loss": 0.0755,
      "step": 1510
    },
    {
      "epoch": 25.33,
      "learning_rate": 3.7962962962962964e-05,
      "loss": 0.0389,
      "step": 1520
    },
    {
      "epoch": 25.5,
      "learning_rate": 3.7847222222222224e-05,
      "loss": 0.0868,
      "step": 1530
    },
    {
      "epoch": 25.67,
      "learning_rate": 3.7731481481481484e-05,
      "loss": 0.0682,
      "step": 1540
    },
    {
      "epoch": 25.83,
      "learning_rate": 3.7615740740740744e-05,
      "loss": 0.0869,
      "step": 1550
    },
    {
      "epoch": 26.0,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.0318,
      "step": 1560
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.4043877124786377,
      "eval_runtime": 0.6557,
      "eval_samples_per_second": 70.159,
      "eval_steps_per_second": 4.576,
      "step": 1560
    },
    {
      "epoch": 26.17,
      "learning_rate": 3.7384259259259263e-05,
      "loss": 0.0504,
      "step": 1570
    },
    {
      "epoch": 26.33,
      "learning_rate": 3.726851851851852e-05,
      "loss": 0.0629,
      "step": 1580
    },
    {
      "epoch": 26.5,
      "learning_rate": 3.715277777777778e-05,
      "loss": 0.0742,
      "step": 1590
    },
    {
      "epoch": 26.67,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.0928,
      "step": 1600
    },
    {
      "epoch": 26.83,
      "learning_rate": 3.6921296296296297e-05,
      "loss": 0.0637,
      "step": 1610
    },
    {
      "epoch": 27.0,
      "learning_rate": 3.6805555555555556e-05,
      "loss": 0.056,
      "step": 1620
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 0.8043478260869565,
      "eval_loss": 1.250475525856018,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 70.916,
      "eval_steps_per_second": 4.625,
      "step": 1620
    },
    {
      "epoch": 27.17,
      "learning_rate": 3.6689814814814816e-05,
      "loss": 0.0799,
      "step": 1630
    },
    {
      "epoch": 27.33,
      "learning_rate": 3.6574074074074076e-05,
      "loss": 0.074,
      "step": 1640
    },
    {
      "epoch": 27.5,
      "learning_rate": 3.6458333333333336e-05,
      "loss": 0.0469,
      "step": 1650
    },
    {
      "epoch": 27.67,
      "learning_rate": 3.6342592592592596e-05,
      "loss": 0.0507,
      "step": 1660
    },
    {
      "epoch": 27.83,
      "learning_rate": 3.6226851851851856e-05,
      "loss": 0.0441,
      "step": 1670
    },
    {
      "epoch": 28.0,
      "learning_rate": 3.611111111111111e-05,
      "loss": 0.0379,
      "step": 1680
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.447351336479187,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 70.916,
      "eval_steps_per_second": 4.625,
      "step": 1680
    },
    {
      "epoch": 28.17,
      "learning_rate": 3.5995370370370376e-05,
      "loss": 0.0317,
      "step": 1690
    },
    {
      "epoch": 28.33,
      "learning_rate": 3.587962962962963e-05,
      "loss": 0.1132,
      "step": 1700
    },
    {
      "epoch": 28.5,
      "learning_rate": 3.576388888888889e-05,
      "loss": 0.0617,
      "step": 1710
    },
    {
      "epoch": 28.67,
      "learning_rate": 3.564814814814815e-05,
      "loss": 0.0568,
      "step": 1720
    },
    {
      "epoch": 28.83,
      "learning_rate": 3.553240740740741e-05,
      "loss": 0.0402,
      "step": 1730
    },
    {
      "epoch": 29.0,
      "learning_rate": 3.541666666666667e-05,
      "loss": 0.0475,
      "step": 1740
    },
    {
      "epoch": 29.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 1.6855353116989136,
      "eval_runtime": 0.6807,
      "eval_samples_per_second": 67.582,
      "eval_steps_per_second": 4.407,
      "step": 1740
    },
    {
      "epoch": 29.17,
      "learning_rate": 3.530092592592593e-05,
      "loss": 0.0575,
      "step": 1750
    },
    {
      "epoch": 29.33,
      "learning_rate": 3.518518518518519e-05,
      "loss": 0.0494,
      "step": 1760
    },
    {
      "epoch": 29.5,
      "learning_rate": 3.506944444444444e-05,
      "loss": 0.0822,
      "step": 1770
    },
    {
      "epoch": 29.67,
      "learning_rate": 3.49537037037037e-05,
      "loss": 0.0557,
      "step": 1780
    },
    {
      "epoch": 29.83,
      "learning_rate": 3.483796296296297e-05,
      "loss": 0.0533,
      "step": 1790
    },
    {
      "epoch": 30.0,
      "learning_rate": 3.472222222222222e-05,
      "loss": 0.0315,
      "step": 1800
    },
    {
      "epoch": 30.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.3772252798080444,
      "eval_runtime": 0.6393,
      "eval_samples_per_second": 71.954,
      "eval_steps_per_second": 4.693,
      "step": 1800
    },
    {
      "epoch": 30.17,
      "learning_rate": 3.460648148148148e-05,
      "loss": 0.0586,
      "step": 1810
    },
    {
      "epoch": 30.33,
      "learning_rate": 3.449074074074074e-05,
      "loss": 0.0198,
      "step": 1820
    },
    {
      "epoch": 30.5,
      "learning_rate": 3.4375e-05,
      "loss": 0.0627,
      "step": 1830
    },
    {
      "epoch": 30.67,
      "learning_rate": 3.425925925925926e-05,
      "loss": 0.0556,
      "step": 1840
    },
    {
      "epoch": 30.83,
      "learning_rate": 3.414351851851852e-05,
      "loss": 0.0448,
      "step": 1850
    },
    {
      "epoch": 31.0,
      "learning_rate": 3.402777777777778e-05,
      "loss": 0.0661,
      "step": 1860
    },
    {
      "epoch": 31.0,
      "eval_accuracy": 0.6521739130434783,
      "eval_loss": 1.7189699411392212,
      "eval_runtime": 0.6492,
      "eval_samples_per_second": 70.86,
      "eval_steps_per_second": 4.621,
      "step": 1860
    },
    {
      "epoch": 31.17,
      "learning_rate": 3.3912037037037035e-05,
      "loss": 0.0799,
      "step": 1870
    },
    {
      "epoch": 31.33,
      "learning_rate": 3.3796296296296295e-05,
      "loss": 0.0693,
      "step": 1880
    },
    {
      "epoch": 31.5,
      "learning_rate": 3.368055555555556e-05,
      "loss": 0.0417,
      "step": 1890
    },
    {
      "epoch": 31.67,
      "learning_rate": 3.3564814814814815e-05,
      "loss": 0.0418,
      "step": 1900
    },
    {
      "epoch": 31.83,
      "learning_rate": 3.3449074074074075e-05,
      "loss": 0.0545,
      "step": 1910
    },
    {
      "epoch": 32.0,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.0401,
      "step": 1920
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.8043478260869565,
      "eval_loss": 1.2324845790863037,
      "eval_runtime": 0.6582,
      "eval_samples_per_second": 69.892,
      "eval_steps_per_second": 4.558,
      "step": 1920
    },
    {
      "epoch": 32.17,
      "learning_rate": 3.3217592592592595e-05,
      "loss": 0.0265,
      "step": 1930
    },
    {
      "epoch": 32.33,
      "learning_rate": 3.3101851851851855e-05,
      "loss": 0.0369,
      "step": 1940
    },
    {
      "epoch": 32.5,
      "learning_rate": 3.2986111111111115e-05,
      "loss": 0.0711,
      "step": 1950
    },
    {
      "epoch": 32.67,
      "learning_rate": 3.2870370370370375e-05,
      "loss": 0.054,
      "step": 1960
    },
    {
      "epoch": 32.83,
      "learning_rate": 3.275462962962963e-05,
      "loss": 0.0231,
      "step": 1970
    },
    {
      "epoch": 33.0,
      "learning_rate": 3.263888888888889e-05,
      "loss": 0.0503,
      "step": 1980
    },
    {
      "epoch": 33.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.6231039762496948,
      "eval_runtime": 0.6402,
      "eval_samples_per_second": 71.856,
      "eval_steps_per_second": 4.686,
      "step": 1980
    },
    {
      "epoch": 33.17,
      "learning_rate": 3.252314814814815e-05,
      "loss": 0.062,
      "step": 1990
    },
    {
      "epoch": 33.33,
      "learning_rate": 3.240740740740741e-05,
      "loss": 0.0261,
      "step": 2000
    },
    {
      "epoch": 33.5,
      "learning_rate": 3.229166666666667e-05,
      "loss": 0.0464,
      "step": 2010
    },
    {
      "epoch": 33.67,
      "learning_rate": 3.217592592592593e-05,
      "loss": 0.0397,
      "step": 2020
    },
    {
      "epoch": 33.83,
      "learning_rate": 3.206018518518519e-05,
      "loss": 0.047,
      "step": 2030
    },
    {
      "epoch": 34.0,
      "learning_rate": 3.194444444444444e-05,
      "loss": 0.0585,
      "step": 2040
    },
    {
      "epoch": 34.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.4189562797546387,
      "eval_runtime": 0.6462,
      "eval_samples_per_second": 71.189,
      "eval_steps_per_second": 4.643,
      "step": 2040
    },
    {
      "epoch": 34.17,
      "learning_rate": 3.182870370370371e-05,
      "loss": 0.0213,
      "step": 2050
    },
    {
      "epoch": 34.33,
      "learning_rate": 3.171296296296297e-05,
      "loss": 0.0595,
      "step": 2060
    },
    {
      "epoch": 34.5,
      "learning_rate": 3.159722222222222e-05,
      "loss": 0.0157,
      "step": 2070
    },
    {
      "epoch": 34.67,
      "learning_rate": 3.148148148148148e-05,
      "loss": 0.0343,
      "step": 2080
    },
    {
      "epoch": 34.83,
      "learning_rate": 3.136574074074074e-05,
      "loss": 0.0448,
      "step": 2090
    },
    {
      "epoch": 35.0,
      "learning_rate": 3.125e-05,
      "loss": 0.0338,
      "step": 2100
    },
    {
      "epoch": 35.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.3640186786651611,
      "eval_runtime": 0.6419,
      "eval_samples_per_second": 71.663,
      "eval_steps_per_second": 4.674,
      "step": 2100
    },
    {
      "epoch": 35.17,
      "learning_rate": 3.113425925925926e-05,
      "loss": 0.0435,
      "step": 2110
    },
    {
      "epoch": 35.33,
      "learning_rate": 3.101851851851852e-05,
      "loss": 0.0262,
      "step": 2120
    },
    {
      "epoch": 35.5,
      "learning_rate": 3.090277777777778e-05,
      "loss": 0.0379,
      "step": 2130
    },
    {
      "epoch": 35.67,
      "learning_rate": 3.0787037037037034e-05,
      "loss": 0.0291,
      "step": 2140
    },
    {
      "epoch": 35.83,
      "learning_rate": 3.06712962962963e-05,
      "loss": 0.0348,
      "step": 2150
    },
    {
      "epoch": 36.0,
      "learning_rate": 3.055555555555556e-05,
      "loss": 0.0343,
      "step": 2160
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.6304347826086957,
      "eval_loss": 2.122363567352295,
      "eval_runtime": 0.6472,
      "eval_samples_per_second": 71.078,
      "eval_steps_per_second": 4.635,
      "step": 2160
    },
    {
      "epoch": 36.17,
      "learning_rate": 3.0439814814814817e-05,
      "loss": 0.036,
      "step": 2170
    },
    {
      "epoch": 36.33,
      "learning_rate": 3.0324074074074077e-05,
      "loss": 0.0277,
      "step": 2180
    },
    {
      "epoch": 36.5,
      "learning_rate": 3.0208333333333334e-05,
      "loss": 0.0197,
      "step": 2190
    },
    {
      "epoch": 36.67,
      "learning_rate": 3.0092592592592593e-05,
      "loss": 0.0339,
      "step": 2200
    },
    {
      "epoch": 36.83,
      "learning_rate": 2.9976851851851857e-05,
      "loss": 0.047,
      "step": 2210
    },
    {
      "epoch": 37.0,
      "learning_rate": 2.9861111111111113e-05,
      "loss": 0.064,
      "step": 2220
    },
    {
      "epoch": 37.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 2.013063669204712,
      "eval_runtime": 0.6437,
      "eval_samples_per_second": 71.466,
      "eval_steps_per_second": 4.661,
      "step": 2220
    },
    {
      "epoch": 37.17,
      "learning_rate": 2.9745370370370373e-05,
      "loss": 0.0633,
      "step": 2230
    },
    {
      "epoch": 37.33,
      "learning_rate": 2.962962962962963e-05,
      "loss": 0.0267,
      "step": 2240
    },
    {
      "epoch": 37.5,
      "learning_rate": 2.951388888888889e-05,
      "loss": 0.0384,
      "step": 2250
    },
    {
      "epoch": 37.67,
      "learning_rate": 2.9398148148148146e-05,
      "loss": 0.0494,
      "step": 2260
    },
    {
      "epoch": 37.83,
      "learning_rate": 2.928240740740741e-05,
      "loss": 0.0597,
      "step": 2270
    },
    {
      "epoch": 38.0,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.0253,
      "step": 2280
    },
    {
      "epoch": 38.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.8281474113464355,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 70.915,
      "eval_steps_per_second": 4.625,
      "step": 2280
    },
    {
      "epoch": 38.17,
      "learning_rate": 2.9050925925925926e-05,
      "loss": 0.0454,
      "step": 2290
    },
    {
      "epoch": 38.33,
      "learning_rate": 2.8935185185185186e-05,
      "loss": 0.0351,
      "step": 2300
    },
    {
      "epoch": 38.5,
      "learning_rate": 2.8819444444444443e-05,
      "loss": 0.0311,
      "step": 2310
    },
    {
      "epoch": 38.67,
      "learning_rate": 2.8703703703703706e-05,
      "loss": 0.0289,
      "step": 2320
    },
    {
      "epoch": 38.83,
      "learning_rate": 2.8587962962962966e-05,
      "loss": 0.025,
      "step": 2330
    },
    {
      "epoch": 39.0,
      "learning_rate": 2.8472222222222223e-05,
      "loss": 0.0435,
      "step": 2340
    },
    {
      "epoch": 39.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 1.4019962549209595,
      "eval_runtime": 0.6529,
      "eval_samples_per_second": 70.45,
      "eval_steps_per_second": 4.595,
      "step": 2340
    },
    {
      "epoch": 39.17,
      "learning_rate": 2.8356481481481483e-05,
      "loss": 0.0295,
      "step": 2350
    },
    {
      "epoch": 39.33,
      "learning_rate": 2.824074074074074e-05,
      "loss": 0.0529,
      "step": 2360
    },
    {
      "epoch": 39.5,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.0506,
      "step": 2370
    },
    {
      "epoch": 39.67,
      "learning_rate": 2.8009259259259263e-05,
      "loss": 0.0249,
      "step": 2380
    },
    {
      "epoch": 39.83,
      "learning_rate": 2.789351851851852e-05,
      "loss": 0.0243,
      "step": 2390
    },
    {
      "epoch": 40.0,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.0344,
      "step": 2400
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.4518897533416748,
      "eval_runtime": 0.6552,
      "eval_samples_per_second": 70.211,
      "eval_steps_per_second": 4.579,
      "step": 2400
    },
    {
      "epoch": 40.17,
      "learning_rate": 2.7662037037037036e-05,
      "loss": 0.0305,
      "step": 2410
    },
    {
      "epoch": 40.33,
      "learning_rate": 2.75462962962963e-05,
      "loss": 0.0306,
      "step": 2420
    },
    {
      "epoch": 40.5,
      "learning_rate": 2.743055555555556e-05,
      "loss": 0.0224,
      "step": 2430
    },
    {
      "epoch": 40.67,
      "learning_rate": 2.7314814814814816e-05,
      "loss": 0.0268,
      "step": 2440
    },
    {
      "epoch": 40.83,
      "learning_rate": 2.7199074074074076e-05,
      "loss": 0.0358,
      "step": 2450
    },
    {
      "epoch": 41.0,
      "learning_rate": 2.7083333333333332e-05,
      "loss": 0.0114,
      "step": 2460
    },
    {
      "epoch": 41.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 1.8910354375839233,
      "eval_runtime": 0.6351,
      "eval_samples_per_second": 72.427,
      "eval_steps_per_second": 4.724,
      "step": 2460
    },
    {
      "epoch": 41.17,
      "learning_rate": 2.6967592592592595e-05,
      "loss": 0.0333,
      "step": 2470
    },
    {
      "epoch": 41.33,
      "learning_rate": 2.6851851851851855e-05,
      "loss": 0.0153,
      "step": 2480
    },
    {
      "epoch": 41.5,
      "learning_rate": 2.6736111111111112e-05,
      "loss": 0.0685,
      "step": 2490
    },
    {
      "epoch": 41.67,
      "learning_rate": 2.6620370370370372e-05,
      "loss": 0.0556,
      "step": 2500
    },
    {
      "epoch": 41.83,
      "learning_rate": 2.650462962962963e-05,
      "loss": 0.0298,
      "step": 2510
    },
    {
      "epoch": 42.0,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 0.0366,
      "step": 2520
    },
    {
      "epoch": 42.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.3172496557235718,
      "eval_runtime": 0.6682,
      "eval_samples_per_second": 68.845,
      "eval_steps_per_second": 4.49,
      "step": 2520
    },
    {
      "epoch": 42.17,
      "learning_rate": 2.627314814814815e-05,
      "loss": 0.0293,
      "step": 2530
    },
    {
      "epoch": 42.33,
      "learning_rate": 2.615740740740741e-05,
      "loss": 0.0303,
      "step": 2540
    },
    {
      "epoch": 42.5,
      "learning_rate": 2.604166666666667e-05,
      "loss": 0.0531,
      "step": 2550
    },
    {
      "epoch": 42.67,
      "learning_rate": 2.5925925925925925e-05,
      "loss": 0.0318,
      "step": 2560
    },
    {
      "epoch": 42.83,
      "learning_rate": 2.5810185185185188e-05,
      "loss": 0.039,
      "step": 2570
    },
    {
      "epoch": 43.0,
      "learning_rate": 2.5694444444444445e-05,
      "loss": 0.0331,
      "step": 2580
    },
    {
      "epoch": 43.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.3785505294799805,
      "eval_runtime": 0.6582,
      "eval_samples_per_second": 69.889,
      "eval_steps_per_second": 4.558,
      "step": 2580
    },
    {
      "epoch": 43.17,
      "learning_rate": 2.5578703703703705e-05,
      "loss": 0.0468,
      "step": 2590
    },
    {
      "epoch": 43.33,
      "learning_rate": 2.5462962962962965e-05,
      "loss": 0.0301,
      "step": 2600
    },
    {
      "epoch": 43.5,
      "learning_rate": 2.534722222222222e-05,
      "loss": 0.0245,
      "step": 2610
    },
    {
      "epoch": 43.67,
      "learning_rate": 2.5231481481481485e-05,
      "loss": 0.0351,
      "step": 2620
    },
    {
      "epoch": 43.83,
      "learning_rate": 2.511574074074074e-05,
      "loss": 0.0488,
      "step": 2630
    },
    {
      "epoch": 44.0,
      "learning_rate": 2.5e-05,
      "loss": 0.0298,
      "step": 2640
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.3052302598953247,
      "eval_runtime": 0.6581,
      "eval_samples_per_second": 69.903,
      "eval_steps_per_second": 4.559,
      "step": 2640
    },
    {
      "epoch": 44.17,
      "learning_rate": 2.488425925925926e-05,
      "loss": 0.0425,
      "step": 2650
    },
    {
      "epoch": 44.33,
      "learning_rate": 2.4768518518518518e-05,
      "loss": 0.0152,
      "step": 2660
    },
    {
      "epoch": 44.5,
      "learning_rate": 2.465277777777778e-05,
      "loss": 0.0408,
      "step": 2670
    },
    {
      "epoch": 44.67,
      "learning_rate": 2.4537037037037038e-05,
      "loss": 0.0383,
      "step": 2680
    },
    {
      "epoch": 44.83,
      "learning_rate": 2.4421296296296298e-05,
      "loss": 0.0254,
      "step": 2690
    },
    {
      "epoch": 45.0,
      "learning_rate": 2.4305555555555558e-05,
      "loss": 0.0426,
      "step": 2700
    },
    {
      "epoch": 45.0,
      "eval_accuracy": 0.782608695652174,
      "eval_loss": 1.2308094501495361,
      "eval_runtime": 0.6512,
      "eval_samples_per_second": 70.642,
      "eval_steps_per_second": 4.607,
      "step": 2700
    },
    {
      "epoch": 45.17,
      "learning_rate": 2.4189814814814814e-05,
      "loss": 0.0168,
      "step": 2710
    },
    {
      "epoch": 45.33,
      "learning_rate": 2.4074074074074074e-05,
      "loss": 0.0217,
      "step": 2720
    },
    {
      "epoch": 45.5,
      "learning_rate": 2.3958333333333334e-05,
      "loss": 0.0137,
      "step": 2730
    },
    {
      "epoch": 45.67,
      "learning_rate": 2.3842592592592594e-05,
      "loss": 0.0484,
      "step": 2740
    },
    {
      "epoch": 45.83,
      "learning_rate": 2.3726851851851854e-05,
      "loss": 0.0166,
      "step": 2750
    },
    {
      "epoch": 46.0,
      "learning_rate": 2.361111111111111e-05,
      "loss": 0.0259,
      "step": 2760
    },
    {
      "epoch": 46.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.6801015138626099,
      "eval_runtime": 0.6367,
      "eval_samples_per_second": 72.252,
      "eval_steps_per_second": 4.712,
      "step": 2760
    },
    {
      "epoch": 46.17,
      "learning_rate": 2.349537037037037e-05,
      "loss": 0.0351,
      "step": 2770
    },
    {
      "epoch": 46.33,
      "learning_rate": 2.337962962962963e-05,
      "loss": 0.0408,
      "step": 2780
    },
    {
      "epoch": 46.5,
      "learning_rate": 2.326388888888889e-05,
      "loss": 0.0293,
      "step": 2790
    },
    {
      "epoch": 46.67,
      "learning_rate": 2.314814814814815e-05,
      "loss": 0.0174,
      "step": 2800
    },
    {
      "epoch": 46.83,
      "learning_rate": 2.3032407407407407e-05,
      "loss": 0.0299,
      "step": 2810
    },
    {
      "epoch": 47.0,
      "learning_rate": 2.2916666666666667e-05,
      "loss": 0.0181,
      "step": 2820
    },
    {
      "epoch": 47.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.607580304145813,
      "eval_runtime": 0.6397,
      "eval_samples_per_second": 71.911,
      "eval_steps_per_second": 4.69,
      "step": 2820
    },
    {
      "epoch": 47.17,
      "learning_rate": 2.2800925925925927e-05,
      "loss": 0.0217,
      "step": 2830
    },
    {
      "epoch": 47.33,
      "learning_rate": 2.2685185185185187e-05,
      "loss": 0.0179,
      "step": 2840
    },
    {
      "epoch": 47.5,
      "learning_rate": 2.2569444444444447e-05,
      "loss": 0.0278,
      "step": 2850
    },
    {
      "epoch": 47.67,
      "learning_rate": 2.2453703703703703e-05,
      "loss": 0.0183,
      "step": 2860
    },
    {
      "epoch": 47.83,
      "learning_rate": 2.2337962962962963e-05,
      "loss": 0.0125,
      "step": 2870
    },
    {
      "epoch": 48.0,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.0266,
      "step": 2880
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.782608695652174,
      "eval_loss": 1.4670273065567017,
      "eval_runtime": 0.6509,
      "eval_samples_per_second": 70.668,
      "eval_steps_per_second": 4.609,
      "step": 2880
    },
    {
      "epoch": 48.17,
      "learning_rate": 2.2106481481481483e-05,
      "loss": 0.0129,
      "step": 2890
    },
    {
      "epoch": 48.33,
      "learning_rate": 2.1990740740740743e-05,
      "loss": 0.0147,
      "step": 2900
    },
    {
      "epoch": 48.5,
      "learning_rate": 2.1875e-05,
      "loss": 0.0176,
      "step": 2910
    },
    {
      "epoch": 48.67,
      "learning_rate": 2.175925925925926e-05,
      "loss": 0.0325,
      "step": 2920
    },
    {
      "epoch": 48.83,
      "learning_rate": 2.164351851851852e-05,
      "loss": 0.0262,
      "step": 2930
    },
    {
      "epoch": 49.0,
      "learning_rate": 2.152777777777778e-05,
      "loss": 0.0239,
      "step": 2940
    },
    {
      "epoch": 49.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.4582098722457886,
      "eval_runtime": 0.6492,
      "eval_samples_per_second": 70.858,
      "eval_steps_per_second": 4.621,
      "step": 2940
    },
    {
      "epoch": 49.17,
      "learning_rate": 2.141203703703704e-05,
      "loss": 0.0077,
      "step": 2950
    },
    {
      "epoch": 49.33,
      "learning_rate": 2.1296296296296296e-05,
      "loss": 0.0385,
      "step": 2960
    },
    {
      "epoch": 49.5,
      "learning_rate": 2.1180555555555556e-05,
      "loss": 0.0226,
      "step": 2970
    },
    {
      "epoch": 49.67,
      "learning_rate": 2.1064814814814816e-05,
      "loss": 0.0233,
      "step": 2980
    },
    {
      "epoch": 49.83,
      "learning_rate": 2.0949074074074073e-05,
      "loss": 0.0268,
      "step": 2990
    },
    {
      "epoch": 50.0,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.0174,
      "step": 3000
    },
    {
      "epoch": 50.0,
      "eval_accuracy": 0.782608695652174,
      "eval_loss": 1.477756381034851,
      "eval_runtime": 0.6387,
      "eval_samples_per_second": 72.027,
      "eval_steps_per_second": 4.697,
      "step": 3000
    },
    {
      "epoch": 50.17,
      "learning_rate": 2.0717592592592593e-05,
      "loss": 0.0418,
      "step": 3010
    },
    {
      "epoch": 50.33,
      "learning_rate": 2.0601851851851853e-05,
      "loss": 0.0216,
      "step": 3020
    },
    {
      "epoch": 50.5,
      "learning_rate": 2.0486111111111113e-05,
      "loss": 0.0254,
      "step": 3030
    },
    {
      "epoch": 50.67,
      "learning_rate": 2.037037037037037e-05,
      "loss": 0.0468,
      "step": 3040
    },
    {
      "epoch": 50.83,
      "learning_rate": 2.0254629629629632e-05,
      "loss": 0.0112,
      "step": 3050
    },
    {
      "epoch": 51.0,
      "learning_rate": 2.013888888888889e-05,
      "loss": 0.0365,
      "step": 3060
    },
    {
      "epoch": 51.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.7034292221069336,
      "eval_runtime": 0.6423,
      "eval_samples_per_second": 71.62,
      "eval_steps_per_second": 4.671,
      "step": 3060
    },
    {
      "epoch": 51.17,
      "learning_rate": 2.002314814814815e-05,
      "loss": 0.0141,
      "step": 3070
    },
    {
      "epoch": 51.33,
      "learning_rate": 1.990740740740741e-05,
      "loss": 0.0079,
      "step": 3080
    },
    {
      "epoch": 51.5,
      "learning_rate": 1.9791666666666665e-05,
      "loss": 0.0387,
      "step": 3090
    },
    {
      "epoch": 51.67,
      "learning_rate": 1.967592592592593e-05,
      "loss": 0.0192,
      "step": 3100
    },
    {
      "epoch": 51.83,
      "learning_rate": 1.9560185185185185e-05,
      "loss": 0.026,
      "step": 3110
    },
    {
      "epoch": 52.0,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 0.0124,
      "step": 3120
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 2.001340389251709,
      "eval_runtime": 0.6569,
      "eval_samples_per_second": 70.022,
      "eval_steps_per_second": 4.567,
      "step": 3120
    },
    {
      "epoch": 52.17,
      "learning_rate": 1.9328703703703705e-05,
      "loss": 0.0223,
      "step": 3130
    },
    {
      "epoch": 52.33,
      "learning_rate": 1.9212962962962962e-05,
      "loss": 0.013,
      "step": 3140
    },
    {
      "epoch": 52.5,
      "learning_rate": 1.9097222222222222e-05,
      "loss": 0.016,
      "step": 3150
    },
    {
      "epoch": 52.67,
      "learning_rate": 1.8981481481481482e-05,
      "loss": 0.0339,
      "step": 3160
    },
    {
      "epoch": 52.83,
      "learning_rate": 1.8865740740740742e-05,
      "loss": 0.0343,
      "step": 3170
    },
    {
      "epoch": 53.0,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.0299,
      "step": 3180
    },
    {
      "epoch": 53.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.8081084489822388,
      "eval_runtime": 0.6411,
      "eval_samples_per_second": 71.746,
      "eval_steps_per_second": 4.679,
      "step": 3180
    },
    {
      "epoch": 53.17,
      "learning_rate": 1.863425925925926e-05,
      "loss": 0.0211,
      "step": 3190
    },
    {
      "epoch": 53.33,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.0255,
      "step": 3200
    },
    {
      "epoch": 53.5,
      "learning_rate": 1.8402777777777778e-05,
      "loss": 0.0115,
      "step": 3210
    },
    {
      "epoch": 53.67,
      "learning_rate": 1.8287037037037038e-05,
      "loss": 0.0132,
      "step": 3220
    },
    {
      "epoch": 53.83,
      "learning_rate": 1.8171296296296298e-05,
      "loss": 0.0075,
      "step": 3230
    },
    {
      "epoch": 54.0,
      "learning_rate": 1.8055555555555555e-05,
      "loss": 0.0042,
      "step": 3240
    },
    {
      "epoch": 54.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.6417256593704224,
      "eval_runtime": 0.6626,
      "eval_samples_per_second": 69.423,
      "eval_steps_per_second": 4.528,
      "step": 3240
    },
    {
      "epoch": 54.17,
      "learning_rate": 1.7939814814814815e-05,
      "loss": 0.0106,
      "step": 3250
    },
    {
      "epoch": 54.33,
      "learning_rate": 1.7824074074074075e-05,
      "loss": 0.0107,
      "step": 3260
    },
    {
      "epoch": 54.5,
      "learning_rate": 1.7708333333333335e-05,
      "loss": 0.0331,
      "step": 3270
    },
    {
      "epoch": 54.67,
      "learning_rate": 1.7592592592592595e-05,
      "loss": 0.0111,
      "step": 3280
    },
    {
      "epoch": 54.83,
      "learning_rate": 1.747685185185185e-05,
      "loss": 0.0167,
      "step": 3290
    },
    {
      "epoch": 55.0,
      "learning_rate": 1.736111111111111e-05,
      "loss": 0.0305,
      "step": 3300
    },
    {
      "epoch": 55.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.7992557287216187,
      "eval_runtime": 0.6422,
      "eval_samples_per_second": 71.633,
      "eval_steps_per_second": 4.672,
      "step": 3300
    },
    {
      "epoch": 55.17,
      "learning_rate": 1.724537037037037e-05,
      "loss": 0.0227,
      "step": 3310
    },
    {
      "epoch": 55.33,
      "learning_rate": 1.712962962962963e-05,
      "loss": 0.0159,
      "step": 3320
    },
    {
      "epoch": 55.5,
      "learning_rate": 1.701388888888889e-05,
      "loss": 0.017,
      "step": 3330
    },
    {
      "epoch": 55.67,
      "learning_rate": 1.6898148148148148e-05,
      "loss": 0.0196,
      "step": 3340
    },
    {
      "epoch": 55.83,
      "learning_rate": 1.6782407407407408e-05,
      "loss": 0.009,
      "step": 3350
    },
    {
      "epoch": 56.0,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.026,
      "step": 3360
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 2.234051465988159,
      "eval_runtime": 0.6412,
      "eval_samples_per_second": 71.742,
      "eval_steps_per_second": 4.679,
      "step": 3360
    },
    {
      "epoch": 56.17,
      "learning_rate": 1.6550925925925927e-05,
      "loss": 0.0452,
      "step": 3370
    },
    {
      "epoch": 56.33,
      "learning_rate": 1.6435185185185187e-05,
      "loss": 0.0197,
      "step": 3380
    },
    {
      "epoch": 56.5,
      "learning_rate": 1.6319444444444444e-05,
      "loss": 0.0029,
      "step": 3390
    },
    {
      "epoch": 56.67,
      "learning_rate": 1.6203703703703704e-05,
      "loss": 0.0075,
      "step": 3400
    },
    {
      "epoch": 56.83,
      "learning_rate": 1.6087962962962964e-05,
      "loss": 0.024,
      "step": 3410
    },
    {
      "epoch": 57.0,
      "learning_rate": 1.597222222222222e-05,
      "loss": 0.0207,
      "step": 3420
    },
    {
      "epoch": 57.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.673930287361145,
      "eval_runtime": 0.6482,
      "eval_samples_per_second": 70.969,
      "eval_steps_per_second": 4.628,
      "step": 3420
    },
    {
      "epoch": 57.17,
      "learning_rate": 1.5856481481481484e-05,
      "loss": 0.0278,
      "step": 3430
    },
    {
      "epoch": 57.33,
      "learning_rate": 1.574074074074074e-05,
      "loss": 0.0369,
      "step": 3440
    },
    {
      "epoch": 57.5,
      "learning_rate": 1.5625e-05,
      "loss": 0.0124,
      "step": 3450
    },
    {
      "epoch": 57.67,
      "learning_rate": 1.550925925925926e-05,
      "loss": 0.0369,
      "step": 3460
    },
    {
      "epoch": 57.83,
      "learning_rate": 1.5393518518518517e-05,
      "loss": 0.0132,
      "step": 3470
    },
    {
      "epoch": 58.0,
      "learning_rate": 1.527777777777778e-05,
      "loss": 0.0063,
      "step": 3480
    },
    {
      "epoch": 58.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.800796627998352,
      "eval_runtime": 0.6675,
      "eval_samples_per_second": 68.918,
      "eval_steps_per_second": 4.495,
      "step": 3480
    },
    {
      "epoch": 58.17,
      "learning_rate": 1.5162037037037038e-05,
      "loss": 0.0097,
      "step": 3490
    },
    {
      "epoch": 58.33,
      "learning_rate": 1.5046296296296297e-05,
      "loss": 0.0379,
      "step": 3500
    },
    {
      "epoch": 58.5,
      "learning_rate": 1.4930555555555557e-05,
      "loss": 0.012,
      "step": 3510
    },
    {
      "epoch": 58.67,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 0.0101,
      "step": 3520
    },
    {
      "epoch": 58.83,
      "learning_rate": 1.4699074074074073e-05,
      "loss": 0.0165,
      "step": 3530
    },
    {
      "epoch": 59.0,
      "learning_rate": 1.4583333333333335e-05,
      "loss": 0.0052,
      "step": 3540
    },
    {
      "epoch": 59.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.1552655696868896,
      "eval_runtime": 0.6661,
      "eval_samples_per_second": 69.056,
      "eval_steps_per_second": 4.504,
      "step": 3540
    },
    {
      "epoch": 59.17,
      "learning_rate": 1.4467592592592593e-05,
      "loss": 0.0184,
      "step": 3550
    },
    {
      "epoch": 59.33,
      "learning_rate": 1.4351851851851853e-05,
      "loss": 0.0113,
      "step": 3560
    },
    {
      "epoch": 59.5,
      "learning_rate": 1.4236111111111111e-05,
      "loss": 0.0154,
      "step": 3570
    },
    {
      "epoch": 59.67,
      "learning_rate": 1.412037037037037e-05,
      "loss": 0.0217,
      "step": 3580
    },
    {
      "epoch": 59.83,
      "learning_rate": 1.4004629629629631e-05,
      "loss": 0.0184,
      "step": 3590
    },
    {
      "epoch": 60.0,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.0061,
      "step": 3600
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.1446449756622314,
      "eval_runtime": 0.6282,
      "eval_samples_per_second": 73.231,
      "eval_steps_per_second": 4.776,
      "step": 3600
    },
    {
      "epoch": 60.17,
      "learning_rate": 1.377314814814815e-05,
      "loss": 0.0107,
      "step": 3610
    },
    {
      "epoch": 60.33,
      "learning_rate": 1.3657407407407408e-05,
      "loss": 0.0071,
      "step": 3620
    },
    {
      "epoch": 60.5,
      "learning_rate": 1.3541666666666666e-05,
      "loss": 0.0219,
      "step": 3630
    },
    {
      "epoch": 60.67,
      "learning_rate": 1.3425925925925928e-05,
      "loss": 0.014,
      "step": 3640
    },
    {
      "epoch": 60.83,
      "learning_rate": 1.3310185185185186e-05,
      "loss": 0.0122,
      "step": 3650
    },
    {
      "epoch": 61.0,
      "learning_rate": 1.3194444444444446e-05,
      "loss": 0.0017,
      "step": 3660
    },
    {
      "epoch": 61.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.213233709335327,
      "eval_runtime": 0.6447,
      "eval_samples_per_second": 71.354,
      "eval_steps_per_second": 4.654,
      "step": 3660
    },
    {
      "epoch": 61.17,
      "learning_rate": 1.3078703703703704e-05,
      "loss": 0.0326,
      "step": 3670
    },
    {
      "epoch": 61.33,
      "learning_rate": 1.2962962962962962e-05,
      "loss": 0.0153,
      "step": 3680
    },
    {
      "epoch": 61.5,
      "learning_rate": 1.2847222222222222e-05,
      "loss": 0.0226,
      "step": 3690
    },
    {
      "epoch": 61.67,
      "learning_rate": 1.2731481481481482e-05,
      "loss": 0.0181,
      "step": 3700
    },
    {
      "epoch": 61.83,
      "learning_rate": 1.2615740740740742e-05,
      "loss": 0.0319,
      "step": 3710
    },
    {
      "epoch": 62.0,
      "learning_rate": 1.25e-05,
      "loss": 0.0143,
      "step": 3720
    },
    {
      "epoch": 62.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.7295066118240356,
      "eval_runtime": 0.8582,
      "eval_samples_per_second": 53.6,
      "eval_steps_per_second": 3.496,
      "step": 3720
    },
    {
      "epoch": 62.17,
      "learning_rate": 1.2384259259259259e-05,
      "loss": 0.027,
      "step": 3730
    },
    {
      "epoch": 62.33,
      "learning_rate": 1.2268518518518519e-05,
      "loss": 0.0297,
      "step": 3740
    },
    {
      "epoch": 62.5,
      "learning_rate": 1.2152777777777779e-05,
      "loss": 0.002,
      "step": 3750
    },
    {
      "epoch": 62.67,
      "learning_rate": 1.2037037037037037e-05,
      "loss": 0.02,
      "step": 3760
    },
    {
      "epoch": 62.83,
      "learning_rate": 1.1921296296296297e-05,
      "loss": 0.0136,
      "step": 3770
    },
    {
      "epoch": 63.0,
      "learning_rate": 1.1805555555555555e-05,
      "loss": 0.004,
      "step": 3780
    },
    {
      "epoch": 63.0,
      "eval_accuracy": 0.8043478260869565,
      "eval_loss": 1.5038210153579712,
      "eval_runtime": 0.6432,
      "eval_samples_per_second": 71.522,
      "eval_steps_per_second": 4.664,
      "step": 3780
    },
    {
      "epoch": 63.17,
      "learning_rate": 1.1689814814814815e-05,
      "loss": 0.0227,
      "step": 3790
    },
    {
      "epoch": 63.33,
      "learning_rate": 1.1574074074074075e-05,
      "loss": 0.0044,
      "step": 3800
    },
    {
      "epoch": 63.5,
      "learning_rate": 1.1458333333333333e-05,
      "loss": 0.0157,
      "step": 3810
    },
    {
      "epoch": 63.67,
      "learning_rate": 1.1342592592592593e-05,
      "loss": 0.0144,
      "step": 3820
    },
    {
      "epoch": 63.83,
      "learning_rate": 1.1226851851851852e-05,
      "loss": 0.0204,
      "step": 3830
    },
    {
      "epoch": 64.0,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.0164,
      "step": 3840
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.573388695716858,
      "eval_runtime": 0.6452,
      "eval_samples_per_second": 71.301,
      "eval_steps_per_second": 4.65,
      "step": 3840
    },
    {
      "epoch": 64.17,
      "learning_rate": 1.0995370370370372e-05,
      "loss": 0.0255,
      "step": 3850
    },
    {
      "epoch": 64.33,
      "learning_rate": 1.087962962962963e-05,
      "loss": 0.0055,
      "step": 3860
    },
    {
      "epoch": 64.5,
      "learning_rate": 1.076388888888889e-05,
      "loss": 0.0247,
      "step": 3870
    },
    {
      "epoch": 64.67,
      "learning_rate": 1.0648148148148148e-05,
      "loss": 0.007,
      "step": 3880
    },
    {
      "epoch": 64.83,
      "learning_rate": 1.0532407407407408e-05,
      "loss": 0.0071,
      "step": 3890
    },
    {
      "epoch": 65.0,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 0.0105,
      "step": 3900
    },
    {
      "epoch": 65.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.6072806119918823,
      "eval_runtime": 0.6397,
      "eval_samples_per_second": 71.913,
      "eval_steps_per_second": 4.69,
      "step": 3900
    },
    {
      "epoch": 65.17,
      "learning_rate": 1.0300925925925926e-05,
      "loss": 0.0015,
      "step": 3910
    },
    {
      "epoch": 65.33,
      "learning_rate": 1.0185185185185185e-05,
      "loss": 0.0091,
      "step": 3920
    },
    {
      "epoch": 65.5,
      "learning_rate": 1.0069444444444445e-05,
      "loss": 0.0121,
      "step": 3930
    },
    {
      "epoch": 65.67,
      "learning_rate": 9.953703703703704e-06,
      "loss": 0.0158,
      "step": 3940
    },
    {
      "epoch": 65.83,
      "learning_rate": 9.837962962962964e-06,
      "loss": 0.0179,
      "step": 3950
    },
    {
      "epoch": 66.0,
      "learning_rate": 9.722222222222223e-06,
      "loss": 0.0161,
      "step": 3960
    },
    {
      "epoch": 66.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.6312615871429443,
      "eval_runtime": 0.6482,
      "eval_samples_per_second": 70.969,
      "eval_steps_per_second": 4.628,
      "step": 3960
    },
    {
      "epoch": 66.17,
      "learning_rate": 9.606481481481481e-06,
      "loss": 0.0124,
      "step": 3970
    },
    {
      "epoch": 66.33,
      "learning_rate": 9.490740740740741e-06,
      "loss": 0.0043,
      "step": 3980
    },
    {
      "epoch": 66.5,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.0056,
      "step": 3990
    },
    {
      "epoch": 66.67,
      "learning_rate": 9.259259259259259e-06,
      "loss": 0.0096,
      "step": 4000
    },
    {
      "epoch": 66.83,
      "learning_rate": 9.143518518518519e-06,
      "loss": 0.0072,
      "step": 4010
    },
    {
      "epoch": 67.0,
      "learning_rate": 9.027777777777777e-06,
      "loss": 0.0073,
      "step": 4020
    },
    {
      "epoch": 67.0,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.7684271335601807,
      "eval_runtime": 0.6827,
      "eval_samples_per_second": 67.383,
      "eval_steps_per_second": 4.395,
      "step": 4020
    },
    {
      "epoch": 67.17,
      "learning_rate": 8.912037037037037e-06,
      "loss": 0.027,
      "step": 4030
    },
    {
      "epoch": 67.33,
      "learning_rate": 8.796296296296297e-06,
      "loss": 0.0121,
      "step": 4040
    },
    {
      "epoch": 67.5,
      "learning_rate": 8.680555555555556e-06,
      "loss": 0.0067,
      "step": 4050
    },
    {
      "epoch": 67.67,
      "learning_rate": 8.564814814814816e-06,
      "loss": 0.0058,
      "step": 4060
    },
    {
      "epoch": 67.83,
      "learning_rate": 8.449074074074074e-06,
      "loss": 0.0031,
      "step": 4070
    },
    {
      "epoch": 68.0,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.004,
      "step": 4080
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.542335867881775,
      "eval_runtime": 0.6793,
      "eval_samples_per_second": 67.72,
      "eval_steps_per_second": 4.417,
      "step": 4080
    },
    {
      "epoch": 68.17,
      "learning_rate": 8.217592592592594e-06,
      "loss": 0.0066,
      "step": 4090
    },
    {
      "epoch": 68.33,
      "learning_rate": 8.101851851851852e-06,
      "loss": 0.0148,
      "step": 4100
    },
    {
      "epoch": 68.5,
      "learning_rate": 7.98611111111111e-06,
      "loss": 0.003,
      "step": 4110
    },
    {
      "epoch": 68.67,
      "learning_rate": 7.87037037037037e-06,
      "loss": 0.0122,
      "step": 4120
    },
    {
      "epoch": 68.83,
      "learning_rate": 7.75462962962963e-06,
      "loss": 0.0043,
      "step": 4130
    },
    {
      "epoch": 69.0,
      "learning_rate": 7.63888888888889e-06,
      "loss": 0.0096,
      "step": 4140
    },
    {
      "epoch": 69.0,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.8457850217819214,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 70.913,
      "eval_steps_per_second": 4.625,
      "step": 4140
    },
    {
      "epoch": 69.17,
      "learning_rate": 7.523148148148148e-06,
      "loss": 0.0206,
      "step": 4150
    },
    {
      "epoch": 69.33,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.0092,
      "step": 4160
    },
    {
      "epoch": 69.5,
      "learning_rate": 7.2916666666666674e-06,
      "loss": 0.0028,
      "step": 4170
    },
    {
      "epoch": 69.67,
      "learning_rate": 7.1759259259259266e-06,
      "loss": 0.0089,
      "step": 4180
    },
    {
      "epoch": 69.83,
      "learning_rate": 7.060185185185185e-06,
      "loss": 0.0109,
      "step": 4190
    },
    {
      "epoch": 70.0,
      "learning_rate": 6.944444444444445e-06,
      "loss": 0.0279,
      "step": 4200
    },
    {
      "epoch": 70.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 2.0680949687957764,
      "eval_runtime": 0.6597,
      "eval_samples_per_second": 69.724,
      "eval_steps_per_second": 4.547,
      "step": 4200
    },
    {
      "epoch": 70.17,
      "learning_rate": 6.828703703703704e-06,
      "loss": 0.0092,
      "step": 4210
    },
    {
      "epoch": 70.33,
      "learning_rate": 6.712962962962964e-06,
      "loss": 0.0027,
      "step": 4220
    },
    {
      "epoch": 70.5,
      "learning_rate": 6.597222222222223e-06,
      "loss": 0.0115,
      "step": 4230
    },
    {
      "epoch": 70.67,
      "learning_rate": 6.481481481481481e-06,
      "loss": 0.0004,
      "step": 4240
    },
    {
      "epoch": 70.83,
      "learning_rate": 6.365740740740741e-06,
      "loss": 0.0049,
      "step": 4250
    },
    {
      "epoch": 71.0,
      "learning_rate": 6.25e-06,
      "loss": 0.0067,
      "step": 4260
    },
    {
      "epoch": 71.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 2.0441176891326904,
      "eval_runtime": 0.6642,
      "eval_samples_per_second": 69.255,
      "eval_steps_per_second": 4.517,
      "step": 4260
    },
    {
      "epoch": 71.17,
      "learning_rate": 6.134259259259259e-06,
      "loss": 0.0012,
      "step": 4270
    },
    {
      "epoch": 71.33,
      "learning_rate": 6.0185185185185185e-06,
      "loss": 0.004,
      "step": 4280
    },
    {
      "epoch": 71.5,
      "learning_rate": 5.902777777777778e-06,
      "loss": 0.0152,
      "step": 4290
    },
    {
      "epoch": 71.67,
      "learning_rate": 5.787037037037038e-06,
      "loss": 0.0093,
      "step": 4300
    },
    {
      "epoch": 71.83,
      "learning_rate": 5.671296296296297e-06,
      "loss": 0.008,
      "step": 4310
    },
    {
      "epoch": 72.0,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.0145,
      "step": 4320
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.1306214332580566,
      "eval_runtime": 0.6912,
      "eval_samples_per_second": 66.553,
      "eval_steps_per_second": 4.34,
      "step": 4320
    },
    {
      "epoch": 72.17,
      "learning_rate": 5.439814814814815e-06,
      "loss": 0.0152,
      "step": 4330
    },
    {
      "epoch": 72.33,
      "learning_rate": 5.324074074074074e-06,
      "loss": 0.0081,
      "step": 4340
    },
    {
      "epoch": 72.5,
      "learning_rate": 5.208333333333334e-06,
      "loss": 0.0042,
      "step": 4350
    },
    {
      "epoch": 72.67,
      "learning_rate": 5.092592592592592e-06,
      "loss": 0.0133,
      "step": 4360
    },
    {
      "epoch": 72.83,
      "learning_rate": 4.976851851851852e-06,
      "loss": 0.0036,
      "step": 4370
    },
    {
      "epoch": 73.0,
      "learning_rate": 4.861111111111111e-06,
      "loss": 0.0084,
      "step": 4380
    },
    {
      "epoch": 73.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.0877416133880615,
      "eval_runtime": 0.6527,
      "eval_samples_per_second": 70.481,
      "eval_steps_per_second": 4.597,
      "step": 4380
    },
    {
      "epoch": 73.17,
      "learning_rate": 4.7453703703703705e-06,
      "loss": 0.0334,
      "step": 4390
    },
    {
      "epoch": 73.33,
      "learning_rate": 4.6296296296296296e-06,
      "loss": 0.014,
      "step": 4400
    },
    {
      "epoch": 73.5,
      "learning_rate": 4.513888888888889e-06,
      "loss": 0.0048,
      "step": 4410
    },
    {
      "epoch": 73.67,
      "learning_rate": 4.398148148148149e-06,
      "loss": 0.0063,
      "step": 4420
    },
    {
      "epoch": 73.83,
      "learning_rate": 4.282407407407408e-06,
      "loss": 0.0006,
      "step": 4430
    },
    {
      "epoch": 74.0,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.0068,
      "step": 4440
    },
    {
      "epoch": 74.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.140406608581543,
      "eval_runtime": 0.6517,
      "eval_samples_per_second": 70.581,
      "eval_steps_per_second": 4.603,
      "step": 4440
    },
    {
      "epoch": 74.17,
      "learning_rate": 4.050925925925926e-06,
      "loss": 0.0152,
      "step": 4450
    },
    {
      "epoch": 74.33,
      "learning_rate": 3.935185185185185e-06,
      "loss": 0.0127,
      "step": 4460
    },
    {
      "epoch": 74.5,
      "learning_rate": 3.819444444444445e-06,
      "loss": 0.0097,
      "step": 4470
    },
    {
      "epoch": 74.67,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.0245,
      "step": 4480
    },
    {
      "epoch": 74.83,
      "learning_rate": 3.5879629629629633e-06,
      "loss": 0.0229,
      "step": 4490
    },
    {
      "epoch": 75.0,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 0.0044,
      "step": 4500
    },
    {
      "epoch": 75.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.1603734493255615,
      "eval_runtime": 0.6857,
      "eval_samples_per_second": 67.088,
      "eval_steps_per_second": 4.375,
      "step": 4500
    },
    {
      "epoch": 75.17,
      "learning_rate": 3.356481481481482e-06,
      "loss": 0.0067,
      "step": 4510
    },
    {
      "epoch": 75.33,
      "learning_rate": 3.2407407407407406e-06,
      "loss": 0.0062,
      "step": 4520
    },
    {
      "epoch": 75.5,
      "learning_rate": 3.125e-06,
      "loss": 0.0044,
      "step": 4530
    },
    {
      "epoch": 75.67,
      "learning_rate": 3.0092592592592593e-06,
      "loss": 0.0057,
      "step": 4540
    },
    {
      "epoch": 75.83,
      "learning_rate": 2.893518518518519e-06,
      "loss": 0.0095,
      "step": 4550
    },
    {
      "epoch": 76.0,
      "learning_rate": 2.777777777777778e-06,
      "loss": 0.0143,
      "step": 4560
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.6956521739130435,
      "eval_loss": 2.183516502380371,
      "eval_runtime": 0.6527,
      "eval_samples_per_second": 70.481,
      "eval_steps_per_second": 4.597,
      "step": 4560
    },
    {
      "epoch": 76.17,
      "learning_rate": 2.662037037037037e-06,
      "loss": 0.009,
      "step": 4570
    },
    {
      "epoch": 76.33,
      "learning_rate": 2.546296296296296e-06,
      "loss": 0.0083,
      "step": 4580
    },
    {
      "epoch": 76.5,
      "learning_rate": 2.4305555555555557e-06,
      "loss": 0.0339,
      "step": 4590
    },
    {
      "epoch": 76.67,
      "learning_rate": 2.3148148148148148e-06,
      "loss": 0.0054,
      "step": 4600
    },
    {
      "epoch": 76.83,
      "learning_rate": 2.1990740740740743e-06,
      "loss": 0.0064,
      "step": 4610
    },
    {
      "epoch": 77.0,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.0105,
      "step": 4620
    },
    {
      "epoch": 77.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 2.25715970993042,
      "eval_runtime": 0.7667,
      "eval_samples_per_second": 59.998,
      "eval_steps_per_second": 3.913,
      "step": 4620
    },
    {
      "epoch": 77.17,
      "learning_rate": 1.9675925925925925e-06,
      "loss": 0.0017,
      "step": 4630
    },
    {
      "epoch": 77.33,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.0059,
      "step": 4640
    },
    {
      "epoch": 77.5,
      "learning_rate": 1.7361111111111112e-06,
      "loss": 0.0052,
      "step": 4650
    },
    {
      "epoch": 77.67,
      "learning_rate": 1.6203703703703703e-06,
      "loss": 0.0003,
      "step": 4660
    },
    {
      "epoch": 77.83,
      "learning_rate": 1.5046296296296296e-06,
      "loss": 0.0021,
      "step": 4670
    },
    {
      "epoch": 78.0,
      "learning_rate": 1.388888888888889e-06,
      "loss": 0.0036,
      "step": 4680
    },
    {
      "epoch": 78.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 2.235708236694336,
      "eval_runtime": 0.6312,
      "eval_samples_per_second": 72.882,
      "eval_steps_per_second": 4.753,
      "step": 4680
    },
    {
      "epoch": 78.17,
      "learning_rate": 1.273148148148148e-06,
      "loss": 0.0183,
      "step": 4690
    },
    {
      "epoch": 78.33,
      "learning_rate": 1.1574074074074074e-06,
      "loss": 0.0258,
      "step": 4700
    },
    {
      "epoch": 78.5,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 0.0023,
      "step": 4710
    },
    {
      "epoch": 78.67,
      "learning_rate": 9.259259259259259e-07,
      "loss": 0.0066,
      "step": 4720
    },
    {
      "epoch": 78.83,
      "learning_rate": 8.101851851851852e-07,
      "loss": 0.0071,
      "step": 4730
    },
    {
      "epoch": 79.0,
      "learning_rate": 6.944444444444445e-07,
      "loss": 0.0205,
      "step": 4740
    },
    {
      "epoch": 79.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 2.2577764987945557,
      "eval_runtime": 0.6629,
      "eval_samples_per_second": 69.397,
      "eval_steps_per_second": 4.526,
      "step": 4740
    },
    {
      "epoch": 79.17,
      "learning_rate": 5.787037037037037e-07,
      "loss": 0.0103,
      "step": 4750
    },
    {
      "epoch": 79.33,
      "learning_rate": 4.6296296296296297e-07,
      "loss": 0.004,
      "step": 4760
    },
    {
      "epoch": 79.5,
      "learning_rate": 3.4722222222222224e-07,
      "loss": 0.0037,
      "step": 4770
    },
    {
      "epoch": 79.67,
      "learning_rate": 2.3148148148148148e-07,
      "loss": 0.0118,
      "step": 4780
    },
    {
      "epoch": 79.83,
      "learning_rate": 1.1574074074074074e-07,
      "loss": 0.0032,
      "step": 4790
    },
    {
      "epoch": 80.0,
      "learning_rate": 0.0,
      "loss": 0.0095,
      "step": 4800
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.6739130434782609,
      "eval_loss": 2.2530040740966797,
      "eval_runtime": 0.6247,
      "eval_samples_per_second": 73.637,
      "eval_steps_per_second": 4.802,
      "step": 4800
    },
    {
      "epoch": 80.0,
      "step": 4800,
      "total_flos": 2.3727626141751706e+19,
      "train_loss": 0.11213935243470284,
      "train_runtime": 6818.2753,
      "train_samples_per_second": 44.926,
      "train_steps_per_second": 0.704
    }
  ],
  "logging_steps": 10,
  "max_steps": 4800,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 80,
  "save_steps": 500,
  "total_flos": 2.3727626141751706e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}