{
  "best_metric": 0.972972972972973,
  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-375",
  "epoch": 49.02,
  "global_step": 3750,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 2.4049,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.666666666666667e-06,
      "loss": 2.3782,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.000000000000001e-06,
      "loss": 2.3345,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.333333333333334e-06,
      "loss": 2.3762,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.666666666666667e-06,
      "loss": 2.3277,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.000000000000001e-06,
      "loss": 2.2711,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.333333333333334e-06,
      "loss": 2.369,
      "step": 70
    },
    {
      "epoch": 0.02,
      "eval_accuracy": 0.2972972972972973,
      "eval_loss": 2.2216224670410156,
      "eval_runtime": 2.6845,
      "eval_samples_per_second": 13.783,
      "eval_steps_per_second": 3.725,
      "step": 75
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 2.2595,
      "step": 80
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.2e-05,
      "loss": 2.2602,
      "step": 90
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 2.1993,
      "step": 100
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.4666666666666668e-05,
      "loss": 2.2198,
      "step": 110
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 2.1327,
      "step": 120
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 2.0154,
      "step": 130
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.866666666666667e-05,
      "loss": 2.0248,
      "step": 140
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 1.8283,
      "step": 150
    },
    {
      "epoch": 1.02,
      "eval_accuracy": 0.4864864864864865,
      "eval_loss": 1.7583507299423218,
      "eval_runtime": 2.7247,
      "eval_samples_per_second": 13.58,
      "eval_steps_per_second": 3.67,
      "step": 150
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 1.5075,
      "step": 160
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.2666666666666668e-05,
      "loss": 1.4104,
      "step": 170
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4e-05,
      "loss": 1.1892,
      "step": 180
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.5333333333333337e-05,
      "loss": 0.9929,
      "step": 190
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.8859,
      "step": 200
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.7763,
      "step": 210
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9333333333333336e-05,
      "loss": 0.8729,
      "step": 220
    },
    {
      "epoch": 2.02,
      "eval_accuracy": 0.7027027027027027,
      "eval_loss": 1.0192049741744995,
      "eval_runtime": 2.6893,
      "eval_samples_per_second": 13.758,
      "eval_steps_per_second": 3.718,
      "step": 225
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.066666666666667e-05,
      "loss": 0.7605,
      "step": 230
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.709,
      "step": 240
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.4213,
      "step": 250
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.466666666666667e-05,
      "loss": 0.6716,
      "step": 260
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.6e-05,
      "loss": 0.7687,
      "step": 270
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.733333333333334e-05,
      "loss": 0.4992,
      "step": 280
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.866666666666667e-05,
      "loss": 0.5896,
      "step": 290
    },
    {
      "epoch": 3.02,
      "learning_rate": 4e-05,
      "loss": 0.4077,
      "step": 300
    },
    {
      "epoch": 3.02,
      "eval_accuracy": 0.8378378378378378,
      "eval_loss": 0.48492956161499023,
      "eval_runtime": 2.6873,
      "eval_samples_per_second": 13.768,
      "eval_steps_per_second": 3.721,
      "step": 300
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.133333333333333e-05,
      "loss": 0.2703,
      "step": 310
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.266666666666667e-05,
      "loss": 0.3431,
      "step": 320
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.4284,
      "step": 330
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.5333333333333335e-05,
      "loss": 0.5786,
      "step": 340
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.5347,
      "step": 350
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.8e-05,
      "loss": 0.2638,
      "step": 360
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.933333333333334e-05,
      "loss": 0.3742,
      "step": 370
    },
    {
      "epoch": 4.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.1343977451324463,
      "eval_runtime": 2.7118,
      "eval_samples_per_second": 13.644,
      "eval_steps_per_second": 3.688,
      "step": 375
    },
    {
      "epoch": 5.0,
      "learning_rate": 4.9925925925925926e-05,
      "loss": 0.2036,
      "step": 380
    },
    {
      "epoch": 5.0,
      "learning_rate": 4.977777777777778e-05,
      "loss": 0.1899,
      "step": 390
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.962962962962963e-05,
      "loss": 0.3761,
      "step": 400
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.9481481481481485e-05,
      "loss": 0.3704,
      "step": 410
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.933333333333334e-05,
      "loss": 0.2038,
      "step": 420
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.918518518518519e-05,
      "loss": 0.4751,
      "step": 430
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.903703703703704e-05,
      "loss": 0.1951,
      "step": 440
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.094,
      "step": 450
    },
    {
      "epoch": 5.02,
      "eval_accuracy": 0.8918918918918919,
      "eval_loss": 0.24485082924365997,
      "eval_runtime": 2.7174,
      "eval_samples_per_second": 13.616,
      "eval_steps_per_second": 3.68,
      "step": 450
    },
    {
      "epoch": 6.0,
      "learning_rate": 4.874074074074074e-05,
      "loss": 0.1336,
      "step": 460
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.8592592592592596e-05,
      "loss": 0.0813,
      "step": 470
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.844444444444445e-05,
      "loss": 0.1722,
      "step": 480
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.82962962962963e-05,
      "loss": 0.5579,
      "step": 490
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.814814814814815e-05,
      "loss": 0.3787,
      "step": 500
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.8e-05,
      "loss": 0.5287,
      "step": 510
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.7851851851851854e-05,
      "loss": 0.1005,
      "step": 520
    },
    {
      "epoch": 6.02,
      "eval_accuracy": 0.7837837837837838,
      "eval_loss": 1.0794074535369873,
      "eval_runtime": 2.7101,
      "eval_samples_per_second": 13.653,
      "eval_steps_per_second": 3.69,
      "step": 525
    },
    {
      "epoch": 7.0,
      "learning_rate": 4.770370370370371e-05,
      "loss": 0.5154,
      "step": 530
    },
    {
      "epoch": 7.0,
      "learning_rate": 4.755555555555556e-05,
      "loss": 0.1985,
      "step": 540
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.740740740740741e-05,
      "loss": 0.2665,
      "step": 550
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.7259259259259266e-05,
      "loss": 0.0321,
      "step": 560
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.711111111111111e-05,
      "loss": 0.159,
      "step": 570
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.6962962962962966e-05,
      "loss": 0.3711,
      "step": 580
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.681481481481482e-05,
      "loss": 0.0089,
      "step": 590
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.0053,
      "step": 600
    },
    {
      "epoch": 7.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.23636393249034882,
      "eval_runtime": 2.6811,
      "eval_samples_per_second": 13.8,
      "eval_steps_per_second": 3.73,
      "step": 600
    },
    {
      "epoch": 8.0,
      "learning_rate": 4.6518518518518525e-05,
      "loss": 0.0056,
      "step": 610
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.637037037037038e-05,
      "loss": 0.3093,
      "step": 620
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.6222222222222224e-05,
      "loss": 0.3216,
      "step": 630
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.607407407407408e-05,
      "loss": 0.1304,
      "step": 640
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.592592592592593e-05,
      "loss": 0.148,
      "step": 650
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.577777777777778e-05,
      "loss": 0.132,
      "step": 660
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.5629629629629636e-05,
      "loss": 0.0807,
      "step": 670
    },
    {
      "epoch": 8.02,
      "eval_accuracy": 0.8378378378378378,
      "eval_loss": 0.6658951640129089,
      "eval_runtime": 2.6902,
      "eval_samples_per_second": 13.754,
      "eval_steps_per_second": 3.717,
      "step": 675
    },
    {
      "epoch": 9.0,
      "learning_rate": 4.548148148148149e-05,
      "loss": 0.0775,
      "step": 680
    },
    {
      "epoch": 9.0,
      "learning_rate": 4.5333333333333335e-05,
      "loss": 0.1351,
      "step": 690
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.518518518518519e-05,
      "loss": 0.1521,
      "step": 700
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.503703703703704e-05,
      "loss": 0.4525,
      "step": 710
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.4888888888888894e-05,
      "loss": 0.4013,
      "step": 720
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.474074074074075e-05,
      "loss": 0.0261,
      "step": 730
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.4592592592592594e-05,
      "loss": 0.0091,
      "step": 740
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.0031,
      "step": 750
    },
    {
      "epoch": 9.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.4496133029460907,
      "eval_runtime": 2.7338,
      "eval_samples_per_second": 13.534,
      "eval_steps_per_second": 3.658,
      "step": 750
    },
    {
      "epoch": 10.0,
      "learning_rate": 4.42962962962963e-05,
      "loss": 0.0035,
      "step": 760
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.414814814814815e-05,
      "loss": 0.0199,
      "step": 770
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0028,
      "step": 780
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.385185185185185e-05,
      "loss": 0.0021,
      "step": 790
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.3703703703703705e-05,
      "loss": 0.0035,
      "step": 800
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.355555555555556e-05,
      "loss": 0.0025,
      "step": 810
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.340740740740741e-05,
      "loss": 0.0203,
      "step": 820
    },
    {
      "epoch": 10.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.3398858308792114,
      "eval_runtime": 2.8361,
      "eval_samples_per_second": 13.046,
      "eval_steps_per_second": 3.526,
      "step": 825
    },
    {
      "epoch": 11.0,
      "learning_rate": 4.325925925925926e-05,
      "loss": 0.0042,
      "step": 830
    },
    {
      "epoch": 11.0,
      "learning_rate": 4.311111111111111e-05,
      "loss": 0.0024,
      "step": 840
    },
    {
      "epoch": 11.01,
      "learning_rate": 4.296296296296296e-05,
      "loss": 0.0023,
      "step": 850
    },
    {
      "epoch": 11.01,
      "learning_rate": 4.2814814814814816e-05,
      "loss": 0.0266,
      "step": 860
    },
    {
      "epoch": 11.01,
      "learning_rate": 4.266666666666667e-05,
      "loss": 0.018,
      "step": 870
    },
    {
      "epoch": 11.01,
      "learning_rate": 4.2518518518518515e-05,
      "loss": 0.0028,
      "step": 880
    },
    {
      "epoch": 11.02,
      "learning_rate": 4.237037037037037e-05,
      "loss": 0.2435,
      "step": 890
    },
    {
      "epoch": 11.02,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.0093,
      "step": 900
    },
    {
      "epoch": 11.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.3724738359451294,
      "eval_runtime": 2.729,
      "eval_samples_per_second": 13.558,
      "eval_steps_per_second": 3.664,
      "step": 900
    },
    {
      "epoch": 12.0,
      "learning_rate": 4.2074074074074075e-05,
      "loss": 0.1554,
      "step": 910
    },
    {
      "epoch": 12.01,
      "learning_rate": 4.192592592592593e-05,
      "loss": 0.0049,
      "step": 920
    },
    {
      "epoch": 12.01,
      "learning_rate": 4.177777777777778e-05,
      "loss": 0.1019,
      "step": 930
    },
    {
      "epoch": 12.01,
      "learning_rate": 4.162962962962963e-05,
      "loss": 0.2167,
      "step": 940
    },
    {
      "epoch": 12.01,
      "learning_rate": 4.148148148148148e-05,
      "loss": 0.1541,
      "step": 950
    },
    {
      "epoch": 12.02,
      "learning_rate": 4.133333333333333e-05,
      "loss": 0.0032,
      "step": 960
    },
    {
      "epoch": 12.02,
      "learning_rate": 4.1185185185185186e-05,
      "loss": 0.0022,
      "step": 970
    },
    {
      "epoch": 12.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.5498412847518921,
      "eval_runtime": 2.7424,
      "eval_samples_per_second": 13.492,
      "eval_steps_per_second": 3.647,
      "step": 975
    },
    {
      "epoch": 13.0,
      "learning_rate": 4.103703703703704e-05,
      "loss": 0.1514,
      "step": 980
    },
    {
      "epoch": 13.0,
      "learning_rate": 4.088888888888889e-05,
      "loss": 0.0019,
      "step": 990
    },
    {
      "epoch": 13.01,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.0018,
      "step": 1000
    },
    {
      "epoch": 13.01,
      "learning_rate": 4.059259259259259e-05,
      "loss": 0.0017,
      "step": 1010
    },
    {
      "epoch": 13.01,
      "learning_rate": 4.0444444444444444e-05,
      "loss": 0.0017,
      "step": 1020
    },
    {
      "epoch": 13.01,
      "learning_rate": 4.02962962962963e-05,
      "loss": 0.0015,
      "step": 1030
    },
    {
      "epoch": 13.02,
      "learning_rate": 4.014814814814815e-05,
      "loss": 0.0015,
      "step": 1040
    },
    {
      "epoch": 13.02,
      "learning_rate": 4e-05,
      "loss": 0.0017,
      "step": 1050
    },
    {
      "epoch": 13.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.16978278756141663,
      "eval_runtime": 2.7539,
      "eval_samples_per_second": 13.436,
      "eval_steps_per_second": 3.631,
      "step": 1050
    },
    {
      "epoch": 14.0,
      "learning_rate": 3.985185185185185e-05,
      "loss": 0.0015,
      "step": 1060
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.97037037037037e-05,
      "loss": 0.0014,
      "step": 1070
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.9555555555555556e-05,
      "loss": 0.0014,
      "step": 1080
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.940740740740741e-05,
      "loss": 0.0013,
      "step": 1090
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.925925925925926e-05,
      "loss": 0.0012,
      "step": 1100
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.9111111111111115e-05,
      "loss": 0.0013,
      "step": 1110
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.896296296296296e-05,
      "loss": 0.0014,
      "step": 1120
    },
    {
      "epoch": 14.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.19233068823814392,
      "eval_runtime": 2.8803,
      "eval_samples_per_second": 12.846,
      "eval_steps_per_second": 3.472,
      "step": 1125
    },
    {
      "epoch": 15.0,
      "learning_rate": 3.8814814814814814e-05,
      "loss": 0.0012,
      "step": 1130
    },
    {
      "epoch": 15.0,
      "learning_rate": 3.866666666666667e-05,
      "loss": 0.0013,
      "step": 1140
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.851851851851852e-05,
      "loss": 0.0206,
      "step": 1150
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.837037037037037e-05,
      "loss": 0.0014,
      "step": 1160
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.8222222222222226e-05,
      "loss": 0.2069,
      "step": 1170
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.807407407407408e-05,
      "loss": 0.0014,
      "step": 1180
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.7925925925925925e-05,
      "loss": 0.0012,
      "step": 1190
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.777777777777778e-05,
      "loss": 0.0014,
      "step": 1200
    },
    {
      "epoch": 15.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.15707702934741974,
      "eval_runtime": 2.8394,
      "eval_samples_per_second": 13.031,
      "eval_steps_per_second": 3.522,
      "step": 1200
    },
    {
      "epoch": 16.0,
      "learning_rate": 3.762962962962963e-05,
      "loss": 0.0012,
      "step": 1210
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.7481481481481484e-05,
      "loss": 0.0018,
      "step": 1220
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.733333333333334e-05,
      "loss": 0.1871,
      "step": 1230
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.718518518518519e-05,
      "loss": 0.0012,
      "step": 1240
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.0014,
      "step": 1250
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.688888888888889e-05,
      "loss": 0.132,
      "step": 1260
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.674074074074074e-05,
      "loss": 0.0474,
      "step": 1270
    },
    {
      "epoch": 16.02,
      "eval_accuracy": 0.8918918918918919,
      "eval_loss": 0.5192855596542358,
      "eval_runtime": 2.8857,
      "eval_samples_per_second": 12.822,
      "eval_steps_per_second": 3.465,
      "step": 1275
    },
    {
      "epoch": 17.0,
      "learning_rate": 3.6592592592592596e-05,
      "loss": 0.0019,
      "step": 1280
    },
    {
      "epoch": 17.0,
      "learning_rate": 3.644444444444445e-05,
      "loss": 0.0013,
      "step": 1290
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.62962962962963e-05,
      "loss": 0.1087,
      "step": 1300
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.614814814814815e-05,
      "loss": 0.081,
      "step": 1310
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.6e-05,
      "loss": 0.0019,
      "step": 1320
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.5851851851851854e-05,
      "loss": 0.0012,
      "step": 1330
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.570370370370371e-05,
      "loss": 0.0012,
      "step": 1340
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.555555555555556e-05,
      "loss": 0.0011,
      "step": 1350
    },
    {
      "epoch": 17.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.14079852402210236,
      "eval_runtime": 3.0095,
      "eval_samples_per_second": 12.294,
      "eval_steps_per_second": 3.323,
      "step": 1350
    },
    {
      "epoch": 18.0,
      "learning_rate": 3.540740740740741e-05,
      "loss": 0.0009,
      "step": 1360
    },
    {
      "epoch": 18.01,
      "learning_rate": 3.525925925925926e-05,
      "loss": 0.0011,
      "step": 1370
    },
    {
      "epoch": 18.01,
      "learning_rate": 3.511111111111111e-05,
      "loss": 0.0277,
      "step": 1380
    },
    {
      "epoch": 18.01,
      "learning_rate": 3.4962962962962965e-05,
      "loss": 0.0011,
      "step": 1390
    },
    {
      "epoch": 18.01,
      "learning_rate": 3.481481481481482e-05,
      "loss": 0.0011,
      "step": 1400
    },
    {
      "epoch": 18.02,
      "learning_rate": 3.466666666666667e-05,
      "loss": 0.1377,
      "step": 1410
    },
    {
      "epoch": 18.02,
      "learning_rate": 3.4518518518518524e-05,
      "loss": 0.001,
      "step": 1420
    },
    {
      "epoch": 18.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.34057652950286865,
      "eval_runtime": 2.8674,
      "eval_samples_per_second": 12.904,
      "eval_steps_per_second": 3.488,
      "step": 1425
    },
    {
      "epoch": 19.0,
      "learning_rate": 3.437037037037037e-05,
      "loss": 0.002,
      "step": 1430
    },
    {
      "epoch": 19.0,
      "learning_rate": 3.4222222222222224e-05,
      "loss": 0.0018,
      "step": 1440
    },
    {
      "epoch": 19.01,
      "learning_rate": 3.4074074074074077e-05,
      "loss": 0.2287,
      "step": 1450
    },
    {
      "epoch": 19.01,
      "learning_rate": 3.392592592592593e-05,
      "loss": 0.102,
      "step": 1460
    },
    {
      "epoch": 19.01,
      "learning_rate": 3.377777777777778e-05,
      "loss": 0.0785,
      "step": 1470
    },
    {
      "epoch": 19.01,
      "learning_rate": 3.3629629629629636e-05,
      "loss": 0.002,
      "step": 1480
    },
    {
      "epoch": 19.02,
      "learning_rate": 3.348148148148148e-05,
      "loss": 0.0357,
      "step": 1490
    },
    {
      "epoch": 19.02,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.0034,
      "step": 1500
    },
    {
      "epoch": 19.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.25160789489746094,
      "eval_runtime": 2.8519,
      "eval_samples_per_second": 12.974,
      "eval_steps_per_second": 3.506,
      "step": 1500
    },
    {
      "epoch": 20.0,
      "learning_rate": 3.318518518518519e-05,
      "loss": 0.0609,
      "step": 1510
    },
    {
      "epoch": 20.01,
      "learning_rate": 3.303703703703704e-05,
      "loss": 0.0013,
      "step": 1520
    },
    {
      "epoch": 20.01,
      "learning_rate": 3.2888888888888894e-05,
      "loss": 0.0012,
      "step": 1530
    },
    {
      "epoch": 20.01,
      "learning_rate": 3.274074074074075e-05,
      "loss": 0.0014,
      "step": 1540
    },
    {
      "epoch": 20.01,
      "learning_rate": 3.25925925925926e-05,
      "loss": 0.1104,
      "step": 1550
    },
    {
      "epoch": 20.02,
      "learning_rate": 3.2444444444444446e-05,
      "loss": 0.0011,
      "step": 1560
    },
    {
      "epoch": 20.02,
      "learning_rate": 3.22962962962963e-05,
      "loss": 0.0029,
      "step": 1570
    },
    {
      "epoch": 20.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.29616308212280273,
      "eval_runtime": 2.9222,
      "eval_samples_per_second": 12.662,
      "eval_steps_per_second": 3.422,
      "step": 1575
    },
    {
      "epoch": 21.0,
      "learning_rate": 3.214814814814815e-05,
      "loss": 0.0014,
      "step": 1580
    },
    {
      "epoch": 21.0,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.0769,
      "step": 1590
    },
    {
      "epoch": 21.01,
      "learning_rate": 3.185185185185185e-05,
      "loss": 0.001,
      "step": 1600
    },
    {
      "epoch": 21.01,
      "learning_rate": 3.1703703703703705e-05,
      "loss": 0.001,
      "step": 1610
    },
    {
      "epoch": 21.01,
      "learning_rate": 3.155555555555556e-05,
      "loss": 0.1219,
      "step": 1620
    },
    {
      "epoch": 21.01,
      "learning_rate": 3.140740740740741e-05,
      "loss": 0.001,
      "step": 1630
    },
    {
      "epoch": 21.02,
      "learning_rate": 3.1259259259259264e-05,
      "loss": 0.0013,
      "step": 1640
    },
    {
      "epoch": 21.02,
      "learning_rate": 3.111111111111111e-05,
      "loss": 0.0008,
      "step": 1650
    },
    {
      "epoch": 21.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.4023502469062805,
      "eval_runtime": 2.9031,
      "eval_samples_per_second": 12.745,
      "eval_steps_per_second": 3.445,
      "step": 1650
    },
    {
      "epoch": 22.0,
      "learning_rate": 3.096296296296296e-05,
      "loss": 0.001,
      "step": 1660
    },
    {
      "epoch": 22.01,
      "learning_rate": 3.0814814814814816e-05,
      "loss": 0.0009,
      "step": 1670
    },
    {
      "epoch": 22.01,
      "learning_rate": 3.066666666666667e-05,
      "loss": 0.0022,
      "step": 1680
    },
    {
      "epoch": 22.01,
      "learning_rate": 3.0518518518518515e-05,
      "loss": 0.0009,
      "step": 1690
    },
    {
      "epoch": 22.01,
      "learning_rate": 3.037037037037037e-05,
      "loss": 0.0009,
      "step": 1700
    },
    {
      "epoch": 22.02,
      "learning_rate": 3.0222222222222225e-05,
      "loss": 0.0008,
      "step": 1710
    },
    {
      "epoch": 22.02,
      "learning_rate": 3.0074074074074078e-05,
      "loss": 0.0008,
      "step": 1720
    },
    {
      "epoch": 22.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.4643724262714386,
      "eval_runtime": 2.9019,
      "eval_samples_per_second": 12.75,
      "eval_steps_per_second": 3.446,
      "step": 1725
    },
    {
      "epoch": 23.0,
      "learning_rate": 2.992592592592593e-05,
      "loss": 0.0009,
      "step": 1730
    },
    {
      "epoch": 23.0,
      "learning_rate": 2.9777777777777777e-05,
      "loss": 0.0008,
      "step": 1740
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.962962962962963e-05,
      "loss": 0.0336,
      "step": 1750
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.9481481481481483e-05,
      "loss": 0.0044,
      "step": 1760
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.9333333333333336e-05,
      "loss": 0.0008,
      "step": 1770
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.918518518518519e-05,
      "loss": 0.001,
      "step": 1780
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.9037037037037042e-05,
      "loss": 0.0008,
      "step": 1790
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.8888888888888888e-05,
      "loss": 0.1521,
      "step": 1800
    },
    {
      "epoch": 23.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.48252156376838684,
      "eval_runtime": 2.9421,
      "eval_samples_per_second": 12.576,
      "eval_steps_per_second": 3.399,
      "step": 1800
    },
    {
      "epoch": 24.0,
      "learning_rate": 2.874074074074074e-05,
      "loss": 0.0023,
      "step": 1810
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.8592592592592594e-05,
      "loss": 0.0008,
      "step": 1820
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.8444444444444447e-05,
      "loss": 0.0074,
      "step": 1830
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.8296296296296297e-05,
      "loss": 0.0009,
      "step": 1840
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.814814814814815e-05,
      "loss": 0.0007,
      "step": 1850
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.0007,
      "step": 1860
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.7851851851851853e-05,
      "loss": 0.001,
      "step": 1870
    },
    {
      "epoch": 24.02,
      "eval_accuracy": 0.918918918918919,
      "eval_loss": 0.6339796185493469,
      "eval_runtime": 2.8461,
      "eval_samples_per_second": 13.0,
      "eval_steps_per_second": 3.514,
      "step": 1875
    },
    {
      "epoch": 25.0,
      "learning_rate": 2.7703703703703706e-05,
      "loss": 0.0007,
      "step": 1880
    },
    {
      "epoch": 25.0,
      "learning_rate": 2.7555555555555555e-05,
      "loss": 0.0007,
      "step": 1890
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.7407407407407408e-05,
      "loss": 0.0008,
      "step": 1900
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.725925925925926e-05,
      "loss": 0.001,
      "step": 1910
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.7111111111111114e-05,
      "loss": 0.0007,
      "step": 1920
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.696296296296296e-05,
      "loss": 0.0007,
      "step": 1930
    },
    {
      "epoch": 25.02,
      "learning_rate": 2.6814814814814814e-05,
      "loss": 0.0007,
      "step": 1940
    },
    {
      "epoch": 25.02,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.0245,
      "step": 1950
    },
    {
      "epoch": 25.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.3778836727142334,
      "eval_runtime": 2.8022,
      "eval_samples_per_second": 13.204,
      "eval_steps_per_second": 3.569,
      "step": 1950
    },
    {
      "epoch": 26.0,
      "learning_rate": 2.651851851851852e-05,
      "loss": 0.0007,
      "step": 1960
    },
    {
      "epoch": 26.01,
      "learning_rate": 2.6370370370370373e-05,
      "loss": 0.0007,
      "step": 1970
    },
    {
      "epoch": 26.01,
      "learning_rate": 2.6222222222222226e-05,
      "loss": 0.0007,
      "step": 1980
    },
    {
      "epoch": 26.01,
      "learning_rate": 2.6074074074074072e-05,
      "loss": 0.0007,
      "step": 1990
    },
    {
      "epoch": 26.01,
      "learning_rate": 2.5925925925925925e-05,
      "loss": 0.0947,
      "step": 2000
    },
    {
      "epoch": 26.02,
      "learning_rate": 2.5777777777777778e-05,
      "loss": 0.001,
      "step": 2010
    },
    {
      "epoch": 26.02,
      "learning_rate": 2.562962962962963e-05,
      "loss": 0.0007,
      "step": 2020
    },
    {
      "epoch": 26.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.3375699818134308,
      "eval_runtime": 2.7928,
      "eval_samples_per_second": 13.248,
      "eval_steps_per_second": 3.581,
      "step": 2025
    },
    {
      "epoch": 27.0,
      "learning_rate": 2.5481481481481484e-05,
      "loss": 0.0006,
      "step": 2030
    },
    {
      "epoch": 27.0,
      "learning_rate": 2.5333333333333337e-05,
      "loss": 0.0012,
      "step": 2040
    },
    {
      "epoch": 27.01,
      "learning_rate": 2.5185185185185183e-05,
      "loss": 0.0007,
      "step": 2050
    },
    {
      "epoch": 27.01,
      "learning_rate": 2.5037037037037036e-05,
      "loss": 0.0007,
      "step": 2060
    },
    {
      "epoch": 27.01,
      "learning_rate": 2.488888888888889e-05,
      "loss": 0.0927,
      "step": 2070
    },
    {
      "epoch": 27.01,
      "learning_rate": 2.4740740740740742e-05,
      "loss": 0.0817,
      "step": 2080
    },
    {
      "epoch": 27.02,
      "learning_rate": 2.4592592592592595e-05,
      "loss": 0.2909,
      "step": 2090
    },
    {
      "epoch": 27.02,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.0011,
      "step": 2100
    },
    {
      "epoch": 27.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.28334298729896545,
      "eval_runtime": 2.7397,
      "eval_samples_per_second": 13.505,
      "eval_steps_per_second": 3.65,
      "step": 2100
    },
    {
      "epoch": 28.0,
      "learning_rate": 2.4296296296296298e-05,
      "loss": 0.0007,
      "step": 2110
    },
    {
      "epoch": 28.01,
      "learning_rate": 2.414814814814815e-05,
      "loss": 0.0007,
      "step": 2120
    },
    {
      "epoch": 28.01,
      "learning_rate": 2.4e-05,
      "loss": 0.0007,
      "step": 2130
    },
    {
      "epoch": 28.01,
      "learning_rate": 2.3851851851851854e-05,
      "loss": 0.0007,
      "step": 2140
    },
    {
      "epoch": 28.01,
      "learning_rate": 2.3703703703703707e-05,
      "loss": 0.0007,
      "step": 2150
    },
    {
      "epoch": 28.02,
      "learning_rate": 2.3555555555555556e-05,
      "loss": 0.0006,
      "step": 2160
    },
    {
      "epoch": 28.02,
      "learning_rate": 2.340740740740741e-05,
      "loss": 0.0008,
      "step": 2170
    },
    {
      "epoch": 28.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.15925714373588562,
      "eval_runtime": 2.6329,
      "eval_samples_per_second": 14.053,
      "eval_steps_per_second": 3.798,
      "step": 2175
    },
    {
      "epoch": 29.0,
      "learning_rate": 2.3259259259259262e-05,
      "loss": 0.0006,
      "step": 2180
    },
    {
      "epoch": 29.0,
      "learning_rate": 2.3111111111111112e-05,
      "loss": 0.0015,
      "step": 2190
    },
    {
      "epoch": 29.01,
      "learning_rate": 2.2962962962962965e-05,
      "loss": 0.0008,
      "step": 2200
    },
    {
      "epoch": 29.01,
      "learning_rate": 2.2814814814814818e-05,
      "loss": 0.0007,
      "step": 2210
    },
    {
      "epoch": 29.01,
      "learning_rate": 2.2666666666666668e-05,
      "loss": 0.0006,
      "step": 2220
    },
    {
      "epoch": 29.01,
      "learning_rate": 2.251851851851852e-05,
      "loss": 0.0006,
      "step": 2230
    },
    {
      "epoch": 29.02,
      "learning_rate": 2.2370370370370374e-05,
      "loss": 0.0953,
      "step": 2240
    },
    {
      "epoch": 29.02,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.0008,
      "step": 2250
    },
    {
      "epoch": 29.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.08563826233148575,
      "eval_runtime": 2.7114,
      "eval_samples_per_second": 13.646,
      "eval_steps_per_second": 3.688,
      "step": 2250
    },
    {
      "epoch": 30.0,
      "learning_rate": 2.2074074074074076e-05,
      "loss": 0.0006,
      "step": 2260
    },
    {
      "epoch": 30.01,
      "learning_rate": 2.1925925925925926e-05,
      "loss": 0.0006,
      "step": 2270
    },
    {
      "epoch": 30.01,
      "learning_rate": 2.177777777777778e-05,
      "loss": 0.0006,
      "step": 2280
    },
    {
      "epoch": 30.01,
      "learning_rate": 2.162962962962963e-05,
      "loss": 0.0006,
      "step": 2290
    },
    {
      "epoch": 30.01,
      "learning_rate": 2.148148148148148e-05,
      "loss": 0.0007,
      "step": 2300
    },
    {
      "epoch": 30.02,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 0.0006,
      "step": 2310
    },
    {
      "epoch": 30.02,
      "learning_rate": 2.1185185185185184e-05,
      "loss": 0.0005,
      "step": 2320
    },
    {
      "epoch": 30.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.1049196645617485,
      "eval_runtime": 2.6802,
      "eval_samples_per_second": 13.805,
      "eval_steps_per_second": 3.731,
      "step": 2325
    },
    {
      "epoch": 31.0,
      "learning_rate": 2.1037037037037037e-05,
      "loss": 0.0006,
      "step": 2330
    },
    {
      "epoch": 31.0,
      "learning_rate": 2.088888888888889e-05,
      "loss": 0.0005,
      "step": 2340
    },
    {
      "epoch": 31.01,
      "learning_rate": 2.074074074074074e-05,
      "loss": 0.0006,
      "step": 2350
    },
    {
      "epoch": 31.01,
      "learning_rate": 2.0592592592592593e-05,
      "loss": 0.0005,
      "step": 2360
    },
    {
      "epoch": 31.01,
      "learning_rate": 2.0444444444444446e-05,
      "loss": 0.0006,
      "step": 2370
    },
    {
      "epoch": 31.01,
      "learning_rate": 2.0296296296296296e-05,
      "loss": 0.0005,
      "step": 2380
    },
    {
      "epoch": 31.02,
      "learning_rate": 2.014814814814815e-05,
      "loss": 0.0005,
      "step": 2390
    },
    {
      "epoch": 31.02,
      "learning_rate": 2e-05,
      "loss": 0.0005,
      "step": 2400
    },
    {
      "epoch": 31.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.11323297023773193,
      "eval_runtime": 3.2205,
      "eval_samples_per_second": 11.489,
      "eval_steps_per_second": 3.105,
      "step": 2400
    },
    {
      "epoch": 32.0,
      "learning_rate": 1.985185185185185e-05,
      "loss": 0.0005,
      "step": 2410
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.9703703703703704e-05,
      "loss": 0.0005,
      "step": 2420
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 0.0005,
      "step": 2430
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.9407407407407407e-05,
      "loss": 0.0005,
      "step": 2440
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.0005,
      "step": 2450
    },
    {
      "epoch": 32.02,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 0.0005,
      "step": 2460
    },
    {
      "epoch": 32.02,
      "learning_rate": 1.8962962962962963e-05,
      "loss": 0.0005,
      "step": 2470
    },
    {
      "epoch": 32.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.11638977378606796,
      "eval_runtime": 2.5685,
      "eval_samples_per_second": 14.405,
      "eval_steps_per_second": 3.893,
      "step": 2475
    },
    {
      "epoch": 33.0,
      "learning_rate": 1.8814814814814816e-05,
      "loss": 0.0005,
      "step": 2480
    },
    {
      "epoch": 33.0,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.0005,
      "step": 2490
    },
    {
      "epoch": 33.01,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.0005,
      "step": 2500
    },
    {
      "epoch": 33.01,
      "learning_rate": 1.837037037037037e-05,
      "loss": 0.0005,
      "step": 2510
    },
    {
      "epoch": 33.01,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 0.0005,
      "step": 2520
    },
    {
      "epoch": 33.01,
      "learning_rate": 1.8074074074074074e-05,
      "loss": 0.0005,
      "step": 2530
    },
    {
      "epoch": 33.02,
      "learning_rate": 1.7925925925925927e-05,
      "loss": 0.0005,
      "step": 2540
    },
    {
      "epoch": 33.02,
      "learning_rate": 1.777777777777778e-05,
      "loss": 0.0005,
      "step": 2550
    },
    {
      "epoch": 33.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.12427014112472534,
      "eval_runtime": 2.5792,
      "eval_samples_per_second": 14.345,
      "eval_steps_per_second": 3.877,
      "step": 2550
    },
    {
      "epoch": 34.0,
      "learning_rate": 1.762962962962963e-05,
      "loss": 0.0005,
      "step": 2560
    },
    {
      "epoch": 34.01,
      "learning_rate": 1.7481481481481483e-05,
      "loss": 0.0005,
      "step": 2570
    },
    {
      "epoch": 34.01,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.0005,
      "step": 2580
    },
    {
      "epoch": 34.01,
      "learning_rate": 1.7185185185185185e-05,
      "loss": 0.0005,
      "step": 2590
    },
    {
      "epoch": 34.01,
      "learning_rate": 1.7037037037037038e-05,
      "loss": 0.0005,
      "step": 2600
    },
    {
      "epoch": 34.02,
      "learning_rate": 1.688888888888889e-05,
      "loss": 0.0005,
      "step": 2610
    },
    {
      "epoch": 34.02,
      "learning_rate": 1.674074074074074e-05,
      "loss": 0.0005,
      "step": 2620
    },
    {
      "epoch": 34.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.1306069791316986,
      "eval_runtime": 2.6024,
      "eval_samples_per_second": 14.217,
      "eval_steps_per_second": 3.843,
      "step": 2625
    },
    {
      "epoch": 35.0,
      "learning_rate": 1.6592592592592594e-05,
      "loss": 0.0005,
      "step": 2630
    },
    {
      "epoch": 35.0,
      "learning_rate": 1.6444444444444447e-05,
      "loss": 0.0005,
      "step": 2640
    },
    {
      "epoch": 35.01,
      "learning_rate": 1.62962962962963e-05,
      "loss": 0.0005,
      "step": 2650
    },
    {
      "epoch": 35.01,
      "learning_rate": 1.614814814814815e-05,
      "loss": 0.0007,
      "step": 2660
    },
    {
      "epoch": 35.01,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.0005,
      "step": 2670
    },
    {
      "epoch": 35.01,
      "learning_rate": 1.5851851851851852e-05,
      "loss": 0.0005,
      "step": 2680
    },
    {
      "epoch": 35.02,
      "learning_rate": 1.5703703703703705e-05,
      "loss": 0.0005,
      "step": 2690
    },
    {
      "epoch": 35.02,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.0005,
      "step": 2700
    },
    {
      "epoch": 35.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.39192795753479004,
      "eval_runtime": 2.5502,
      "eval_samples_per_second": 14.509,
      "eval_steps_per_second": 3.921,
      "step": 2700
    },
    {
      "epoch": 36.0,
      "learning_rate": 1.5407407407407408e-05,
      "loss": 0.0005,
      "step": 2710
    },
    {
      "epoch": 36.01,
      "learning_rate": 1.5259259259259258e-05,
      "loss": 0.0005,
      "step": 2720
    },
    {
      "epoch": 36.01,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 0.0005,
      "step": 2730
    },
    {
      "epoch": 36.01,
      "learning_rate": 1.4962962962962965e-05,
      "loss": 0.0005,
      "step": 2740
    },
    {
      "epoch": 36.01,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 0.0005,
      "step": 2750
    },
    {
      "epoch": 36.02,
      "learning_rate": 1.4666666666666668e-05,
      "loss": 0.0005,
      "step": 2760
    },
    {
      "epoch": 36.02,
      "learning_rate": 1.4518518518518521e-05,
      "loss": 0.0004,
      "step": 2770
    },
    {
      "epoch": 36.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.36302804946899414,
      "eval_runtime": 2.4982,
      "eval_samples_per_second": 14.811,
      "eval_steps_per_second": 4.003,
      "step": 2775
    },
    {
      "epoch": 37.0,
      "learning_rate": 1.437037037037037e-05,
      "loss": 0.0005,
      "step": 2780
    },
    {
      "epoch": 37.0,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 0.0005,
      "step": 2790
    },
    {
      "epoch": 37.01,
      "learning_rate": 1.4074074074074075e-05,
      "loss": 0.0004,
      "step": 2800
    },
    {
      "epoch": 37.01,
      "learning_rate": 1.3925925925925926e-05,
      "loss": 0.0005,
      "step": 2810
    },
    {
      "epoch": 37.01,
      "learning_rate": 1.3777777777777778e-05,
      "loss": 0.0004,
      "step": 2820
    },
    {
      "epoch": 37.01,
      "learning_rate": 1.362962962962963e-05,
      "loss": 0.0005,
      "step": 2830
    },
    {
      "epoch": 37.02,
      "learning_rate": 1.348148148148148e-05,
      "loss": 0.0005,
      "step": 2840
    },
    {
      "epoch": 37.02,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0004,
      "step": 2850
    },
    {
      "epoch": 37.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.2762458324432373,
      "eval_runtime": 2.5203,
      "eval_samples_per_second": 14.681,
      "eval_steps_per_second": 3.968,
      "step": 2850
    },
    {
      "epoch": 38.0,
      "learning_rate": 1.3185185185185186e-05,
      "loss": 0.0004,
      "step": 2860
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.3037037037037036e-05,
      "loss": 0.0005,
      "step": 2870
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.2888888888888889e-05,
      "loss": 0.0004,
      "step": 2880
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.2740740740740742e-05,
      "loss": 0.0004,
      "step": 2890
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.2592592592592592e-05,
      "loss": 0.0004,
      "step": 2900
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.2444444444444445e-05,
      "loss": 0.0004,
      "step": 2910
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.2296296296296298e-05,
      "loss": 0.0005,
      "step": 2920
    },
    {
      "epoch": 38.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.23680266737937927,
      "eval_runtime": 2.5362,
      "eval_samples_per_second": 14.589,
      "eval_steps_per_second": 3.943,
      "step": 2925
    },
    {
      "epoch": 39.0,
      "learning_rate": 1.2148148148148149e-05,
      "loss": 0.0004,
      "step": 2930
    },
    {
      "epoch": 39.0,
      "learning_rate": 1.2e-05,
      "loss": 0.0004,
      "step": 2940
    },
    {
      "epoch": 39.01,
      "learning_rate": 1.1851851851851853e-05,
      "loss": 0.0005,
      "step": 2950
    },
    {
      "epoch": 39.01,
      "learning_rate": 1.1703703703703705e-05,
      "loss": 0.0005,
      "step": 2960
    },
    {
      "epoch": 39.01,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 0.0004,
      "step": 2970
    },
    {
      "epoch": 39.01,
      "learning_rate": 1.1407407407407409e-05,
      "loss": 0.0004,
      "step": 2980
    },
    {
      "epoch": 39.02,
      "learning_rate": 1.125925925925926e-05,
      "loss": 0.0004,
      "step": 2990
    },
    {
      "epoch": 39.02,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.0004,
      "step": 3000
    },
    {
      "epoch": 39.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.1935373693704605,
      "eval_runtime": 2.5511,
      "eval_samples_per_second": 14.504,
      "eval_steps_per_second": 3.92,
      "step": 3000
    },
    {
      "epoch": 40.0,
      "learning_rate": 1.0962962962962963e-05,
      "loss": 0.0004,
      "step": 3010
    },
    {
      "epoch": 40.01,
      "learning_rate": 1.0814814814814814e-05,
      "loss": 0.0004,
      "step": 3020
    },
    {
      "epoch": 40.01,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.0004,
      "step": 3030
    },
    {
      "epoch": 40.01,
      "learning_rate": 1.0518518518518519e-05,
      "loss": 0.0004,
      "step": 3040
    },
    {
      "epoch": 40.01,
      "learning_rate": 1.037037037037037e-05,
      "loss": 0.0004,
      "step": 3050
    },
    {
      "epoch": 40.02,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 0.0004,
      "step": 3060
    },
    {
      "epoch": 40.02,
      "learning_rate": 1.0074074074074074e-05,
      "loss": 0.0004,
      "step": 3070
    },
    {
      "epoch": 40.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.19308657944202423,
      "eval_runtime": 2.5348,
      "eval_samples_per_second": 14.597,
      "eval_steps_per_second": 3.945,
      "step": 3075
    },
    {
      "epoch": 41.0,
      "learning_rate": 9.925925925925926e-06,
      "loss": 0.0004,
      "step": 3080
    },
    {
      "epoch": 41.0,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.0085,
      "step": 3090
    },
    {
      "epoch": 41.01,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.0004,
      "step": 3100
    },
    {
      "epoch": 41.01,
      "learning_rate": 9.481481481481481e-06,
      "loss": 0.0004,
      "step": 3110
    },
    {
      "epoch": 41.01,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.0004,
      "step": 3120
    },
    {
      "epoch": 41.01,
      "learning_rate": 9.185185185185186e-06,
      "loss": 0.0004,
      "step": 3130
    },
    {
      "epoch": 41.02,
      "learning_rate": 9.037037037037037e-06,
      "loss": 0.0005,
      "step": 3140
    },
    {
      "epoch": 41.02,
      "learning_rate": 8.88888888888889e-06,
      "loss": 0.0004,
      "step": 3150
    },
    {
      "epoch": 41.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.21387839317321777,
      "eval_runtime": 2.5103,
      "eval_samples_per_second": 14.739,
      "eval_steps_per_second": 3.984,
      "step": 3150
    },
    {
      "epoch": 42.0,
      "learning_rate": 8.740740740740741e-06,
      "loss": 0.0004,
      "step": 3160
    },
    {
      "epoch": 42.01,
      "learning_rate": 8.592592592592593e-06,
      "loss": 0.0005,
      "step": 3170
    },
    {
      "epoch": 42.01,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.0004,
      "step": 3180
    },
    {
      "epoch": 42.01,
      "learning_rate": 8.296296296296297e-06,
      "loss": 0.0004,
      "step": 3190
    },
    {
      "epoch": 42.01,
      "learning_rate": 8.14814814814815e-06,
      "loss": 0.0004,
      "step": 3200
    },
    {
      "epoch": 42.02,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0004,
      "step": 3210
    },
    {
      "epoch": 42.02,
      "learning_rate": 7.851851851851853e-06,
      "loss": 0.0004,
      "step": 3220
    },
    {
      "epoch": 42.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.1899683177471161,
      "eval_runtime": 2.6252,
      "eval_samples_per_second": 14.094,
      "eval_steps_per_second": 3.809,
      "step": 3225
    },
    {
      "epoch": 43.0,
      "learning_rate": 7.703703703703704e-06,
      "loss": 0.0015,
      "step": 3230
    },
    {
      "epoch": 43.0,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.0004,
      "step": 3240
    },
    {
      "epoch": 43.01,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.0004,
      "step": 3250
    },
    {
      "epoch": 43.01,
      "learning_rate": 7.2592592592592605e-06,
      "loss": 0.0004,
      "step": 3260
    },
    {
      "epoch": 43.01,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.0004,
      "step": 3270
    },
    {
      "epoch": 43.01,
      "learning_rate": 6.962962962962963e-06,
      "loss": 0.0004,
      "step": 3280
    },
    {
      "epoch": 43.02,
      "learning_rate": 6.814814814814815e-06,
      "loss": 0.0004,
      "step": 3290
    },
    {
      "epoch": 43.02,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0006,
      "step": 3300
    },
    {
      "epoch": 43.02,
      "eval_accuracy": 0.972972972972973,
      "eval_loss": 0.1750660091638565,
      "eval_runtime": 2.5683,
      "eval_samples_per_second": 14.407,
      "eval_steps_per_second": 3.894,
      "step": 3300
    },
    {
      "epoch": 44.0,
      "learning_rate": 6.518518518518518e-06,
      "loss": 0.0004,
      "step": 3310
    },
    {
      "epoch": 44.01,
      "learning_rate": 6.370370370370371e-06,
      "loss": 0.0004,
      "step": 3320
    },
    {
      "epoch": 44.01,
      "learning_rate": 6.222222222222222e-06,
      "loss": 0.0004,
      "step": 3330
    },
    {
      "epoch": 44.01,
      "learning_rate": 6.0740740740740745e-06,
      "loss": 0.0004,
      "step": 3340
    },
    {
      "epoch": 44.01,
      "learning_rate": 5.925925925925927e-06,
      "loss": 0.0004,
      "step": 3350
    },
    {
      "epoch": 44.02,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.0004,
      "step": 3360
    },
    {
      "epoch": 44.02,
      "learning_rate": 5.62962962962963e-06,
      "loss": 0.0004,
      "step": 3370
    },
    {
      "epoch": 44.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.29775792360305786,
      "eval_runtime": 2.6767,
      "eval_samples_per_second": 13.823,
      "eval_steps_per_second": 3.736,
      "step": 3375
    },
    {
      "epoch": 45.0,
      "learning_rate": 5.4814814814814815e-06,
      "loss": 0.0004,
      "step": 3380
    },
    {
      "epoch": 45.0,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0004,
      "step": 3390
    },
    {
      "epoch": 45.01,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.0004,
      "step": 3400
    },
    {
      "epoch": 45.01,
      "learning_rate": 5.037037037037037e-06,
      "loss": 0.0004,
      "step": 3410
    },
    {
      "epoch": 45.01,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.0004,
      "step": 3420
    },
    {
      "epoch": 45.01,
      "learning_rate": 4.740740740740741e-06,
      "loss": 0.0004,
      "step": 3430
    },
    {
      "epoch": 45.02,
      "learning_rate": 4.592592592592593e-06,
      "loss": 0.0004,
      "step": 3440
    },
    {
      "epoch": 45.02,
      "learning_rate": 4.444444444444445e-06,
      "loss": 0.0004,
      "step": 3450
    },
    {
      "epoch": 45.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.2776608467102051,
      "eval_runtime": 2.5843,
      "eval_samples_per_second": 14.317,
      "eval_steps_per_second": 3.87,
      "step": 3450
    },
    {
      "epoch": 46.0,
      "learning_rate": 4.296296296296296e-06,
      "loss": 0.0004,
      "step": 3460
    },
    {
      "epoch": 46.01,
      "learning_rate": 4.1481481481481485e-06,
      "loss": 0.0004,
      "step": 3470
    },
    {
      "epoch": 46.01,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0004,
      "step": 3480
    },
    {
      "epoch": 46.01,
      "learning_rate": 3.851851851851852e-06,
      "loss": 0.0004,
      "step": 3490
    },
    {
      "epoch": 46.01,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.0004,
      "step": 3500
    },
    {
      "epoch": 46.02,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0004,
      "step": 3510
    },
    {
      "epoch": 46.02,
      "learning_rate": 3.4074074074074077e-06,
      "loss": 0.0004,
      "step": 3520
    },
    {
      "epoch": 46.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.2706151008605957,
      "eval_runtime": 2.6031,
      "eval_samples_per_second": 14.214,
      "eval_steps_per_second": 3.842,
      "step": 3525
    },
    {
      "epoch": 47.0,
      "learning_rate": 3.259259259259259e-06,
      "loss": 0.0004,
      "step": 3530
    },
    {
      "epoch": 47.0,
      "learning_rate": 3.111111111111111e-06,
      "loss": 0.0004,
      "step": 3540
    },
    {
      "epoch": 47.01,
      "learning_rate": 2.9629629629629633e-06,
      "loss": 0.0004,
      "step": 3550
    },
    {
      "epoch": 47.01,
      "learning_rate": 2.814814814814815e-06,
      "loss": 0.0004,
      "step": 3560
    },
    {
      "epoch": 47.01,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0004,
      "step": 3570
    },
    {
      "epoch": 47.01,
      "learning_rate": 2.5185185185185186e-06,
      "loss": 0.0004,
      "step": 3580
    },
    {
      "epoch": 47.02,
      "learning_rate": 2.3703703703703703e-06,
      "loss": 0.0004,
      "step": 3590
    },
    {
      "epoch": 47.02,
      "learning_rate": 2.2222222222222225e-06,
      "loss": 0.0004,
      "step": 3600
    },
    {
      "epoch": 47.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.26377302408218384,
      "eval_runtime": 2.5538,
      "eval_samples_per_second": 14.488,
      "eval_steps_per_second": 3.916,
      "step": 3600
    },
    {
      "epoch": 48.0,
      "learning_rate": 2.0740740740740742e-06,
      "loss": 0.0004,
      "step": 3610
    },
    {
      "epoch": 48.01,
      "learning_rate": 1.925925925925926e-06,
      "loss": 0.0003,
      "step": 3620
    },
    {
      "epoch": 48.01,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.0004,
      "step": 3630
    },
    {
      "epoch": 48.01,
      "learning_rate": 1.6296296296296295e-06,
      "loss": 0.0005,
      "step": 3640
    },
    {
      "epoch": 48.01,
      "learning_rate": 1.4814814814814817e-06,
      "loss": 0.0004,
      "step": 3650
    },
    {
      "epoch": 48.02,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0004,
      "step": 3660
    },
    {
      "epoch": 48.02,
      "learning_rate": 1.1851851851851852e-06,
      "loss": 0.0004,
      "step": 3670
    },
    {
      "epoch": 48.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.21225357055664062,
      "eval_runtime": 2.6586,
      "eval_samples_per_second": 13.917,
      "eval_steps_per_second": 3.761,
      "step": 3675
    },
    {
      "epoch": 49.0,
      "learning_rate": 1.0370370370370371e-06,
      "loss": 0.0004,
      "step": 3680
    },
    {
      "epoch": 49.0,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0004,
      "step": 3690
    },
    {
      "epoch": 49.01,
      "learning_rate": 7.407407407407408e-07,
      "loss": 0.0004,
      "step": 3700
    },
    {
      "epoch": 49.01,
      "learning_rate": 5.925925925925926e-07,
      "loss": 0.0004,
      "step": 3710
    },
    {
      "epoch": 49.01,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0004,
      "step": 3720
    },
    {
      "epoch": 49.01,
      "learning_rate": 2.962962962962963e-07,
      "loss": 0.0004,
      "step": 3730
    },
    {
      "epoch": 49.02,
      "learning_rate": 1.4814814814814815e-07,
      "loss": 0.0004,
      "step": 3740
    },
    {
      "epoch": 49.02,
      "learning_rate": 0.0,
      "loss": 0.0004,
      "step": 3750
    },
    {
      "epoch": 49.02,
      "eval_accuracy": 0.9459459459459459,
      "eval_loss": 0.21062754094600677,
      "eval_runtime": 2.7099,
      "eval_samples_per_second": 13.653,
      "eval_steps_per_second": 3.69,
      "step": 3750
    },
    {
      "epoch": 49.02,
      "step": 3750,
      "total_flos": 1.86923023515648e+19,
      "train_loss": 0.16172246270999313,
      "train_runtime": 3621.2802,
      "train_samples_per_second": 4.142,
      "train_steps_per_second": 1.036
    },
    {
      "epoch": 49.02,
      "eval_accuracy": 0.8850574712643678,
      "eval_loss": 0.38414710760116577,
      "eval_runtime": 8.6129,
      "eval_samples_per_second": 10.101,
      "eval_steps_per_second": 2.554,
      "step": 3750
    },
    {
      "epoch": 49.02,
      "eval_accuracy": 0.8850574712643678,
      "eval_loss": 0.38414719700813293,
      "eval_runtime": 6.1302,
      "eval_samples_per_second": 14.192,
      "eval_steps_per_second": 3.589,
      "step": 3750
    }
  ],
  "max_steps": 3750,
  "num_train_epochs": 9223372036854775807,
  "total_flos": 1.86923023515648e+19,
  "trial_name": null,
  "trial_params": null
}