{
  "best_metric": 0.00020612729713320732,
  "best_model_checkpoint": "./ap_train_outputs_tobikoi/checkpoint-7830",
  "epoch": 150.0,
  "eval_steps": 500,
  "global_step": 8100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.19,
      "learning_rate": 1.997530864197531e-05,
      "loss": 0.657,
      "step": 10
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.995061728395062e-05,
      "loss": 0.5358,
      "step": 20
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9925925925925928e-05,
      "loss": 0.5326,
      "step": 30
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.9901234567901237e-05,
      "loss": 0.5757,
      "step": 40
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9876543209876546e-05,
      "loss": 0.4014,
      "step": 50
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7631578947368421,
      "eval_loss": 0.3551758825778961,
      "eval_runtime": 3.0112,
      "eval_samples_per_second": 25.239,
      "eval_steps_per_second": 3.321,
      "step": 54
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.9851851851851855e-05,
      "loss": 0.3156,
      "step": 60
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.9827160493827163e-05,
      "loss": 0.3332,
      "step": 70
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.9802469135802472e-05,
      "loss": 0.2956,
      "step": 80
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.977777777777778e-05,
      "loss": 0.2752,
      "step": 90
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9753086419753087e-05,
      "loss": 0.2253,
      "step": 100
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9736842105263158,
      "eval_loss": 0.1712142825126648,
      "eval_runtime": 2.945,
      "eval_samples_per_second": 25.806,
      "eval_steps_per_second": 3.396,
      "step": 108
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.9728395061728395e-05,
      "loss": 0.1685,
      "step": 110
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.9703703703703704e-05,
      "loss": 0.2022,
      "step": 120
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9679012345679013e-05,
      "loss": 0.1469,
      "step": 130
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.9654320987654322e-05,
      "loss": 0.1177,
      "step": 140
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.962962962962963e-05,
      "loss": 0.081,
      "step": 150
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.960493827160494e-05,
      "loss": 0.0768,
      "step": 160
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.07630854845046997,
      "eval_runtime": 3.0191,
      "eval_samples_per_second": 25.173,
      "eval_steps_per_second": 3.312,
      "step": 162
    },
    {
      "epoch": 3.15,
      "learning_rate": 1.958024691358025e-05,
      "loss": 0.1135,
      "step": 170
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 0.0699,
      "step": 180
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.9530864197530866e-05,
      "loss": 0.0533,
      "step": 190
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.9506172839506175e-05,
      "loss": 0.0447,
      "step": 200
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.9481481481481484e-05,
      "loss": 0.0694,
      "step": 210
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.061516497284173965,
      "eval_runtime": 2.9805,
      "eval_samples_per_second": 25.499,
      "eval_steps_per_second": 3.355,
      "step": 216
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.9456790123456793e-05,
      "loss": 0.1095,
      "step": 220
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.94320987654321e-05,
      "loss": 0.0489,
      "step": 230
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.9407407407407407e-05,
      "loss": 0.0254,
      "step": 240
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.9382716049382716e-05,
      "loss": 0.0304,
      "step": 250
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.9358024691358025e-05,
      "loss": 0.0258,
      "step": 260
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.9333333333333333e-05,
      "loss": 0.0433,
      "step": 270
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.05043227970600128,
      "eval_runtime": 3.0544,
      "eval_samples_per_second": 24.882,
      "eval_steps_per_second": 3.274,
      "step": 270
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.9308641975308646e-05,
      "loss": 0.0204,
      "step": 280
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.9283950617283955e-05,
      "loss": 0.0231,
      "step": 290
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.0463,
      "step": 300
    },
    {
      "epoch": 5.74,
      "learning_rate": 1.923456790123457e-05,
      "loss": 0.0181,
      "step": 310
    },
    {
      "epoch": 5.93,
      "learning_rate": 1.9209876543209878e-05,
      "loss": 0.1045,
      "step": 320
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.03228908032178879,
      "eval_runtime": 2.9647,
      "eval_samples_per_second": 25.635,
      "eval_steps_per_second": 3.373,
      "step": 324
    },
    {
      "epoch": 6.11,
      "learning_rate": 1.9185185185185186e-05,
      "loss": 0.0568,
      "step": 330
    },
    {
      "epoch": 6.3,
      "learning_rate": 1.9160493827160495e-05,
      "loss": 0.0373,
      "step": 340
    },
    {
      "epoch": 6.48,
      "learning_rate": 1.9135802469135804e-05,
      "loss": 0.0157,
      "step": 350
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 0.023,
      "step": 360
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.9086419753086422e-05,
      "loss": 0.0148,
      "step": 370
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.04355274513363838,
      "eval_runtime": 3.0476,
      "eval_samples_per_second": 24.938,
      "eval_steps_per_second": 3.281,
      "step": 378
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.906172839506173e-05,
      "loss": 0.0364,
      "step": 380
    },
    {
      "epoch": 7.22,
      "learning_rate": 1.903703703703704e-05,
      "loss": 0.0515,
      "step": 390
    },
    {
      "epoch": 7.41,
      "learning_rate": 1.901234567901235e-05,
      "loss": 0.0271,
      "step": 400
    },
    {
      "epoch": 7.59,
      "learning_rate": 1.8987654320987657e-05,
      "loss": 0.0134,
      "step": 410
    },
    {
      "epoch": 7.78,
      "learning_rate": 1.8962962962962966e-05,
      "loss": 0.0111,
      "step": 420
    },
    {
      "epoch": 7.96,
      "learning_rate": 1.8938271604938275e-05,
      "loss": 0.0156,
      "step": 430
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.027104511857032776,
      "eval_runtime": 3.0509,
      "eval_samples_per_second": 24.911,
      "eval_steps_per_second": 3.278,
      "step": 432
    },
    {
      "epoch": 8.15,
      "learning_rate": 1.891358024691358e-05,
      "loss": 0.0141,
      "step": 440
    },
    {
      "epoch": 8.33,
      "learning_rate": 1.888888888888889e-05,
      "loss": 0.0118,
      "step": 450
    },
    {
      "epoch": 8.52,
      "learning_rate": 1.8864197530864198e-05,
      "loss": 0.0109,
      "step": 460
    },
    {
      "epoch": 8.7,
      "learning_rate": 1.8839506172839507e-05,
      "loss": 0.009,
      "step": 470
    },
    {
      "epoch": 8.89,
      "learning_rate": 1.8814814814814816e-05,
      "loss": 0.0109,
      "step": 480
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.05106474086642265,
      "eval_runtime": 3.0435,
      "eval_samples_per_second": 24.971,
      "eval_steps_per_second": 3.286,
      "step": 486
    },
    {
      "epoch": 9.07,
      "learning_rate": 1.8790123456790124e-05,
      "loss": 0.0117,
      "step": 490
    },
    {
      "epoch": 9.26,
      "learning_rate": 1.8765432098765433e-05,
      "loss": 0.0244,
      "step": 500
    },
    {
      "epoch": 9.44,
      "learning_rate": 1.8740740740740742e-05,
      "loss": 0.1521,
      "step": 510
    },
    {
      "epoch": 9.63,
      "learning_rate": 1.871604938271605e-05,
      "loss": 0.0103,
      "step": 520
    },
    {
      "epoch": 9.81,
      "learning_rate": 1.869135802469136e-05,
      "loss": 0.0103,
      "step": 530
    },
    {
      "epoch": 10.0,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.0142,
      "step": 540
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.05626223608851433,
      "eval_runtime": 2.9574,
      "eval_samples_per_second": 25.698,
      "eval_steps_per_second": 3.381,
      "step": 540
    },
    {
      "epoch": 10.19,
      "learning_rate": 1.8641975308641977e-05,
      "loss": 0.008,
      "step": 550
    },
    {
      "epoch": 10.37,
      "learning_rate": 1.8617283950617286e-05,
      "loss": 0.0198,
      "step": 560
    },
    {
      "epoch": 10.56,
      "learning_rate": 1.8592592592592592e-05,
      "loss": 0.0093,
      "step": 570
    },
    {
      "epoch": 10.74,
      "learning_rate": 1.85679012345679e-05,
      "loss": 0.0084,
      "step": 580
    },
    {
      "epoch": 10.93,
      "learning_rate": 1.8543209876543213e-05,
      "loss": 0.0307,
      "step": 590
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06329117715358734,
      "eval_runtime": 2.9789,
      "eval_samples_per_second": 25.513,
      "eval_steps_per_second": 3.357,
      "step": 594
    },
    {
      "epoch": 11.11,
      "learning_rate": 1.851851851851852e-05,
      "loss": 0.0108,
      "step": 600
    },
    {
      "epoch": 11.3,
      "learning_rate": 1.849382716049383e-05,
      "loss": 0.0074,
      "step": 610
    },
    {
      "epoch": 11.48,
      "learning_rate": 1.846913580246914e-05,
      "loss": 0.0078,
      "step": 620
    },
    {
      "epoch": 11.67,
      "learning_rate": 1.8444444444444448e-05,
      "loss": 0.0068,
      "step": 630
    },
    {
      "epoch": 11.85,
      "learning_rate": 1.8419753086419754e-05,
      "loss": 0.0092,
      "step": 640
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.04296250268816948,
      "eval_runtime": 3.0671,
      "eval_samples_per_second": 24.779,
      "eval_steps_per_second": 3.26,
      "step": 648
    },
    {
      "epoch": 12.04,
      "learning_rate": 1.8395061728395062e-05,
      "loss": 0.092,
      "step": 650
    },
    {
      "epoch": 12.22,
      "learning_rate": 1.837037037037037e-05,
      "loss": 0.0068,
      "step": 660
    },
    {
      "epoch": 12.41,
      "learning_rate": 1.834567901234568e-05,
      "loss": 0.0508,
      "step": 670
    },
    {
      "epoch": 12.59,
      "learning_rate": 1.832098765432099e-05,
      "loss": 0.0167,
      "step": 680
    },
    {
      "epoch": 12.78,
      "learning_rate": 1.8296296296296298e-05,
      "loss": 0.0064,
      "step": 690
    },
    {
      "epoch": 12.96,
      "learning_rate": 1.8271604938271607e-05,
      "loss": 0.007,
      "step": 700
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.050767723470926285,
      "eval_runtime": 2.9682,
      "eval_samples_per_second": 25.605,
      "eval_steps_per_second": 3.369,
      "step": 702
    },
    {
      "epoch": 13.15,
      "learning_rate": 1.8246913580246915e-05,
      "loss": 0.0063,
      "step": 710
    },
    {
      "epoch": 13.33,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 0.006,
      "step": 720
    },
    {
      "epoch": 13.52,
      "learning_rate": 1.8197530864197533e-05,
      "loss": 0.0082,
      "step": 730
    },
    {
      "epoch": 13.7,
      "learning_rate": 1.8172839506172842e-05,
      "loss": 0.0056,
      "step": 740
    },
    {
      "epoch": 13.89,
      "learning_rate": 1.814814814814815e-05,
      "loss": 0.0059,
      "step": 750
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.05981181189417839,
      "eval_runtime": 2.9705,
      "eval_samples_per_second": 25.585,
      "eval_steps_per_second": 3.366,
      "step": 756
    },
    {
      "epoch": 14.07,
      "learning_rate": 1.812345679012346e-05,
      "loss": 0.0089,
      "step": 760
    },
    {
      "epoch": 14.26,
      "learning_rate": 1.8098765432098765e-05,
      "loss": 0.0058,
      "step": 770
    },
    {
      "epoch": 14.44,
      "learning_rate": 1.8074074074074074e-05,
      "loss": 0.0059,
      "step": 780
    },
    {
      "epoch": 14.63,
      "learning_rate": 1.8049382716049383e-05,
      "loss": 0.0434,
      "step": 790
    },
    {
      "epoch": 14.81,
      "learning_rate": 1.802469135802469e-05,
      "loss": 0.0054,
      "step": 800
    },
    {
      "epoch": 15.0,
      "learning_rate": 1.8e-05,
      "loss": 0.0057,
      "step": 810
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06393761932849884,
      "eval_runtime": 2.9181,
      "eval_samples_per_second": 26.044,
      "eval_steps_per_second": 3.427,
      "step": 810
    },
    {
      "epoch": 15.19,
      "learning_rate": 1.797530864197531e-05,
      "loss": 0.0052,
      "step": 820
    },
    {
      "epoch": 15.37,
      "learning_rate": 1.7950617283950618e-05,
      "loss": 0.0051,
      "step": 830
    },
    {
      "epoch": 15.56,
      "learning_rate": 1.7925925925925927e-05,
      "loss": 0.0249,
      "step": 840
    },
    {
      "epoch": 15.74,
      "learning_rate": 1.7901234567901236e-05,
      "loss": 0.0274,
      "step": 850
    },
    {
      "epoch": 15.93,
      "learning_rate": 1.7876543209876545e-05,
      "loss": 0.0513,
      "step": 860
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.05791584402322769,
      "eval_runtime": 2.9732,
      "eval_samples_per_second": 25.561,
      "eval_steps_per_second": 3.363,
      "step": 864
    },
    {
      "epoch": 16.11,
      "learning_rate": 1.7851851851851853e-05,
      "loss": 0.0049,
      "step": 870
    },
    {
      "epoch": 16.3,
      "learning_rate": 1.7827160493827162e-05,
      "loss": 0.0048,
      "step": 880
    },
    {
      "epoch": 16.48,
      "learning_rate": 1.780246913580247e-05,
      "loss": 0.0133,
      "step": 890
    },
    {
      "epoch": 16.67,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.0101,
      "step": 900
    },
    {
      "epoch": 16.85,
      "learning_rate": 1.775308641975309e-05,
      "loss": 0.0259,
      "step": 910
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.07065781950950623,
      "eval_runtime": 2.9923,
      "eval_samples_per_second": 25.399,
      "eval_steps_per_second": 3.342,
      "step": 918
    },
    {
      "epoch": 17.04,
      "learning_rate": 1.7728395061728398e-05,
      "loss": 0.0122,
      "step": 920
    },
    {
      "epoch": 17.22,
      "learning_rate": 1.7703703703703706e-05,
      "loss": 0.0056,
      "step": 930
    },
    {
      "epoch": 17.41,
      "learning_rate": 1.7679012345679015e-05,
      "loss": 0.037,
      "step": 940
    },
    {
      "epoch": 17.59,
      "learning_rate": 1.7654320987654324e-05,
      "loss": 0.0046,
      "step": 950
    },
    {
      "epoch": 17.78,
      "learning_rate": 1.7629629629629633e-05,
      "loss": 0.0041,
      "step": 960
    },
    {
      "epoch": 17.96,
      "learning_rate": 1.760493827160494e-05,
      "loss": 0.0111,
      "step": 970
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.061096400022506714,
      "eval_runtime": 2.9964,
      "eval_samples_per_second": 25.364,
      "eval_steps_per_second": 3.337,
      "step": 972
    },
    {
      "epoch": 18.15,
      "learning_rate": 1.7580246913580247e-05,
      "loss": 0.0044,
      "step": 980
    },
    {
      "epoch": 18.33,
      "learning_rate": 1.7555555555555556e-05,
      "loss": 0.0612,
      "step": 990
    },
    {
      "epoch": 18.52,
      "learning_rate": 1.7530864197530865e-05,
      "loss": 0.0389,
      "step": 1000
    },
    {
      "epoch": 18.7,
      "learning_rate": 1.7506172839506174e-05,
      "loss": 0.0707,
      "step": 1010
    },
    {
      "epoch": 18.89,
      "learning_rate": 1.7481481481481483e-05,
      "loss": 0.014,
      "step": 1020
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06202290579676628,
      "eval_runtime": 3.0244,
      "eval_samples_per_second": 25.129,
      "eval_steps_per_second": 3.306,
      "step": 1026
    },
    {
      "epoch": 19.07,
      "learning_rate": 1.745679012345679e-05,
      "loss": 0.0042,
      "step": 1030
    },
    {
      "epoch": 19.26,
      "learning_rate": 1.74320987654321e-05,
      "loss": 0.0052,
      "step": 1040
    },
    {
      "epoch": 19.44,
      "learning_rate": 1.740740740740741e-05,
      "loss": 0.0505,
      "step": 1050
    },
    {
      "epoch": 19.63,
      "learning_rate": 1.7382716049382718e-05,
      "loss": 0.0042,
      "step": 1060
    },
    {
      "epoch": 19.81,
      "learning_rate": 1.7358024691358027e-05,
      "loss": 0.0038,
      "step": 1070
    },
    {
      "epoch": 20.0,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.004,
      "step": 1080
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.005762842949479818,
      "eval_runtime": 3.075,
      "eval_samples_per_second": 24.715,
      "eval_steps_per_second": 3.252,
      "step": 1080
    },
    {
      "epoch": 20.19,
      "learning_rate": 1.7308641975308644e-05,
      "loss": 0.0039,
      "step": 1090
    },
    {
      "epoch": 20.37,
      "learning_rate": 1.728395061728395e-05,
      "loss": 0.0037,
      "step": 1100
    },
    {
      "epoch": 20.56,
      "learning_rate": 1.725925925925926e-05,
      "loss": 0.0037,
      "step": 1110
    },
    {
      "epoch": 20.74,
      "learning_rate": 1.7234567901234568e-05,
      "loss": 0.0054,
      "step": 1120
    },
    {
      "epoch": 20.93,
      "learning_rate": 1.7209876543209876e-05,
      "loss": 0.0036,
      "step": 1130
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.004358741920441389,
      "eval_runtime": 3.0437,
      "eval_samples_per_second": 24.97,
      "eval_steps_per_second": 3.286,
      "step": 1134
    },
    {
      "epoch": 21.11,
      "learning_rate": 1.7185185185185185e-05,
      "loss": 0.0039,
      "step": 1140
    },
    {
      "epoch": 21.3,
      "learning_rate": 1.7160493827160498e-05,
      "loss": 0.0418,
      "step": 1150
    },
    {
      "epoch": 21.48,
      "learning_rate": 1.7135802469135806e-05,
      "loss": 0.0104,
      "step": 1160
    },
    {
      "epoch": 21.67,
      "learning_rate": 1.7111111111111112e-05,
      "loss": 0.1184,
      "step": 1170
    },
    {
      "epoch": 21.85,
      "learning_rate": 1.708641975308642e-05,
      "loss": 0.0545,
      "step": 1180
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.011370152235031128,
      "eval_runtime": 2.9933,
      "eval_samples_per_second": 25.39,
      "eval_steps_per_second": 3.341,
      "step": 1188
    },
    {
      "epoch": 22.04,
      "learning_rate": 1.706172839506173e-05,
      "loss": 0.0295,
      "step": 1190
    },
    {
      "epoch": 22.22,
      "learning_rate": 1.7037037037037038e-05,
      "loss": 0.0633,
      "step": 1200
    },
    {
      "epoch": 22.41,
      "learning_rate": 1.7012345679012347e-05,
      "loss": 0.007,
      "step": 1210
    },
    {
      "epoch": 22.59,
      "learning_rate": 1.6987654320987656e-05,
      "loss": 0.0092,
      "step": 1220
    },
    {
      "epoch": 22.78,
      "learning_rate": 1.6962962962962965e-05,
      "loss": 0.0176,
      "step": 1230
    },
    {
      "epoch": 22.96,
      "learning_rate": 1.6938271604938274e-05,
      "loss": 0.0131,
      "step": 1240
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06211669743061066,
      "eval_runtime": 3.0535,
      "eval_samples_per_second": 24.889,
      "eval_steps_per_second": 3.275,
      "step": 1242
    },
    {
      "epoch": 23.15,
      "learning_rate": 1.6913580246913582e-05,
      "loss": 0.4038,
      "step": 1250
    },
    {
      "epoch": 23.33,
      "learning_rate": 1.688888888888889e-05,
      "loss": 0.0658,
      "step": 1260
    },
    {
      "epoch": 23.52,
      "learning_rate": 1.68641975308642e-05,
      "loss": 0.0316,
      "step": 1270
    },
    {
      "epoch": 23.7,
      "learning_rate": 1.683950617283951e-05,
      "loss": 0.0705,
      "step": 1280
    },
    {
      "epoch": 23.89,
      "learning_rate": 1.6814814814814818e-05,
      "loss": 0.0651,
      "step": 1290
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06920474767684937,
      "eval_runtime": 3.011,
      "eval_samples_per_second": 25.241,
      "eval_steps_per_second": 3.321,
      "step": 1296
    },
    {
      "epoch": 24.07,
      "learning_rate": 1.6790123456790123e-05,
      "loss": 0.0035,
      "step": 1300
    },
    {
      "epoch": 24.26,
      "learning_rate": 1.6765432098765432e-05,
      "loss": 0.0147,
      "step": 1310
    },
    {
      "epoch": 24.44,
      "learning_rate": 1.674074074074074e-05,
      "loss": 0.0045,
      "step": 1320
    },
    {
      "epoch": 24.63,
      "learning_rate": 1.671604938271605e-05,
      "loss": 0.0375,
      "step": 1330
    },
    {
      "epoch": 24.81,
      "learning_rate": 1.669135802469136e-05,
      "loss": 0.0278,
      "step": 1340
    },
    {
      "epoch": 25.0,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.0047,
      "step": 1350
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.003432216588407755,
      "eval_runtime": 2.9828,
      "eval_samples_per_second": 25.48,
      "eval_steps_per_second": 3.353,
      "step": 1350
    },
    {
      "epoch": 25.19,
      "learning_rate": 1.6641975308641976e-05,
      "loss": 0.071,
      "step": 1360
    },
    {
      "epoch": 25.37,
      "learning_rate": 1.6617283950617285e-05,
      "loss": 0.003,
      "step": 1370
    },
    {
      "epoch": 25.56,
      "learning_rate": 1.6592592592592594e-05,
      "loss": 0.0045,
      "step": 1380
    },
    {
      "epoch": 25.74,
      "learning_rate": 1.6567901234567903e-05,
      "loss": 0.0033,
      "step": 1390
    },
    {
      "epoch": 25.93,
      "learning_rate": 1.654320987654321e-05,
      "loss": 0.0374,
      "step": 1400
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.003061347408220172,
      "eval_runtime": 2.9183,
      "eval_samples_per_second": 26.042,
      "eval_steps_per_second": 3.427,
      "step": 1404
    },
    {
      "epoch": 26.11,
      "learning_rate": 1.651851851851852e-05,
      "loss": 0.0077,
      "step": 1410
    },
    {
      "epoch": 26.3,
      "learning_rate": 1.649382716049383e-05,
      "loss": 0.0047,
      "step": 1420
    },
    {
      "epoch": 26.48,
      "learning_rate": 1.6469135802469135e-05,
      "loss": 0.0088,
      "step": 1430
    },
    {
      "epoch": 26.67,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 0.0043,
      "step": 1440
    },
    {
      "epoch": 26.85,
      "learning_rate": 1.6419753086419752e-05,
      "loss": 0.0482,
      "step": 1450
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.004511620849370956,
      "eval_runtime": 2.9675,
      "eval_samples_per_second": 25.611,
      "eval_steps_per_second": 3.37,
      "step": 1458
    },
    {
      "epoch": 27.04,
      "learning_rate": 1.6395061728395065e-05,
      "loss": 0.0029,
      "step": 1460
    },
    {
      "epoch": 27.22,
      "learning_rate": 1.6370370370370374e-05,
      "loss": 0.0028,
      "step": 1470
    },
    {
      "epoch": 27.41,
      "learning_rate": 1.6345679012345682e-05,
      "loss": 0.0027,
      "step": 1480
    },
    {
      "epoch": 27.59,
      "learning_rate": 1.632098765432099e-05,
      "loss": 0.0027,
      "step": 1490
    },
    {
      "epoch": 27.78,
      "learning_rate": 1.6296296296296297e-05,
      "loss": 0.0028,
      "step": 1500
    },
    {
      "epoch": 27.96,
      "learning_rate": 1.6271604938271605e-05,
      "loss": 0.0026,
      "step": 1510
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0028049517422914505,
      "eval_runtime": 2.9403,
      "eval_samples_per_second": 25.848,
      "eval_steps_per_second": 3.401,
      "step": 1512
    },
    {
      "epoch": 28.15,
      "learning_rate": 1.6246913580246914e-05,
      "loss": 0.0025,
      "step": 1520
    },
    {
      "epoch": 28.33,
      "learning_rate": 1.6222222222222223e-05,
      "loss": 0.0028,
      "step": 1530
    },
    {
      "epoch": 28.52,
      "learning_rate": 1.6197530864197532e-05,
      "loss": 0.0025,
      "step": 1540
    },
    {
      "epoch": 28.7,
      "learning_rate": 1.617283950617284e-05,
      "loss": 0.0024,
      "step": 1550
    },
    {
      "epoch": 28.89,
      "learning_rate": 1.614814814814815e-05,
      "loss": 0.0038,
      "step": 1560
    },
    {
      "epoch": 29.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0024565160274505615,
      "eval_runtime": 3.0445,
      "eval_samples_per_second": 24.963,
      "eval_steps_per_second": 3.285,
      "step": 1566
    },
    {
      "epoch": 29.07,
      "learning_rate": 1.612345679012346e-05,
      "loss": 0.0023,
      "step": 1570
    },
    {
      "epoch": 29.26,
      "learning_rate": 1.6098765432098767e-05,
      "loss": 0.0024,
      "step": 1580
    },
    {
      "epoch": 29.44,
      "learning_rate": 1.6074074074074076e-05,
      "loss": 0.0293,
      "step": 1590
    },
    {
      "epoch": 29.63,
      "learning_rate": 1.6049382716049385e-05,
      "loss": 0.0024,
      "step": 1600
    },
    {
      "epoch": 29.81,
      "learning_rate": 1.6024691358024694e-05,
      "loss": 0.0023,
      "step": 1610
    },
    {
      "epoch": 30.0,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.0027,
      "step": 1620
    },
    {
      "epoch": 30.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0023441906087100506,
      "eval_runtime": 2.9794,
      "eval_samples_per_second": 25.508,
      "eval_steps_per_second": 3.356,
      "step": 1620
    },
    {
      "epoch": 30.19,
      "learning_rate": 1.5975308641975308e-05,
      "loss": 0.0022,
      "step": 1630
    },
    {
      "epoch": 30.37,
      "learning_rate": 1.5950617283950617e-05,
      "loss": 0.0023,
      "step": 1640
    },
    {
      "epoch": 30.56,
      "learning_rate": 1.5925925925925926e-05,
      "loss": 0.0715,
      "step": 1650
    },
    {
      "epoch": 30.74,
      "learning_rate": 1.5901234567901235e-05,
      "loss": 0.0029,
      "step": 1660
    },
    {
      "epoch": 30.93,
      "learning_rate": 1.5876543209876543e-05,
      "loss": 0.0145,
      "step": 1670
    },
    {
      "epoch": 31.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06983212381601334,
      "eval_runtime": 3.0396,
      "eval_samples_per_second": 25.003,
      "eval_steps_per_second": 3.29,
      "step": 1674
    },
    {
      "epoch": 31.11,
      "learning_rate": 1.5851851851851852e-05,
      "loss": 0.0025,
      "step": 1680
    },
    {
      "epoch": 31.3,
      "learning_rate": 1.582716049382716e-05,
      "loss": 0.0025,
      "step": 1690
    },
    {
      "epoch": 31.48,
      "learning_rate": 1.580246913580247e-05,
      "loss": 0.0655,
      "step": 1700
    },
    {
      "epoch": 31.67,
      "learning_rate": 1.577777777777778e-05,
      "loss": 0.0022,
      "step": 1710
    },
    {
      "epoch": 31.85,
      "learning_rate": 1.5753086419753088e-05,
      "loss": 0.0022,
      "step": 1720
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.025487201288342476,
      "eval_runtime": 3.0285,
      "eval_samples_per_second": 25.095,
      "eval_steps_per_second": 3.302,
      "step": 1728
    },
    {
      "epoch": 32.04,
      "learning_rate": 1.5728395061728396e-05,
      "loss": 0.0932,
      "step": 1730
    },
    {
      "epoch": 32.22,
      "learning_rate": 1.5703703703703705e-05,
      "loss": 0.0023,
      "step": 1740
    },
    {
      "epoch": 32.41,
      "learning_rate": 1.5679012345679014e-05,
      "loss": 0.0404,
      "step": 1750
    },
    {
      "epoch": 32.59,
      "learning_rate": 1.565432098765432e-05,
      "loss": 0.0065,
      "step": 1760
    },
    {
      "epoch": 32.78,
      "learning_rate": 1.5629629629629632e-05,
      "loss": 0.0058,
      "step": 1770
    },
    {
      "epoch": 32.96,
      "learning_rate": 1.560493827160494e-05,
      "loss": 0.0025,
      "step": 1780
    },
    {
      "epoch": 33.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.009514715522527695,
      "eval_runtime": 3.0612,
      "eval_samples_per_second": 24.827,
      "eval_steps_per_second": 3.267,
      "step": 1782
    },
    {
      "epoch": 33.15,
      "learning_rate": 1.558024691358025e-05,
      "loss": 0.002,
      "step": 1790
    },
    {
      "epoch": 33.33,
      "learning_rate": 1.555555555555556e-05,
      "loss": 0.0021,
      "step": 1800
    },
    {
      "epoch": 33.52,
      "learning_rate": 1.5530864197530867e-05,
      "loss": 0.0824,
      "step": 1810
    },
    {
      "epoch": 33.7,
      "learning_rate": 1.5506172839506176e-05,
      "loss": 0.0024,
      "step": 1820
    },
    {
      "epoch": 33.89,
      "learning_rate": 1.548148148148148e-05,
      "loss": 0.0022,
      "step": 1830
    },
    {
      "epoch": 34.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.07251408696174622,
      "eval_runtime": 3.1052,
      "eval_samples_per_second": 24.475,
      "eval_steps_per_second": 3.22,
      "step": 1836
    },
    {
      "epoch": 34.07,
      "learning_rate": 1.545679012345679e-05,
      "loss": 0.0021,
      "step": 1840
    },
    {
      "epoch": 34.26,
      "learning_rate": 1.54320987654321e-05,
      "loss": 0.0022,
      "step": 1850
    },
    {
      "epoch": 34.44,
      "learning_rate": 1.5407407407407408e-05,
      "loss": 0.0035,
      "step": 1860
    },
    {
      "epoch": 34.63,
      "learning_rate": 1.5382716049382717e-05,
      "loss": 0.0021,
      "step": 1870
    },
    {
      "epoch": 34.81,
      "learning_rate": 1.5358024691358026e-05,
      "loss": 0.0019,
      "step": 1880
    },
    {
      "epoch": 35.0,
      "learning_rate": 1.5333333333333334e-05,
      "loss": 0.0019,
      "step": 1890
    },
    {
      "epoch": 35.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.05920861288905144,
      "eval_runtime": 3.006,
      "eval_samples_per_second": 25.283,
      "eval_steps_per_second": 3.327,
      "step": 1890
    },
    {
      "epoch": 35.19,
      "learning_rate": 1.5308641975308643e-05,
      "loss": 0.002,
      "step": 1900
    },
    {
      "epoch": 35.37,
      "learning_rate": 1.5283950617283952e-05,
      "loss": 0.0019,
      "step": 1910
    },
    {
      "epoch": 35.56,
      "learning_rate": 1.525925925925926e-05,
      "loss": 0.0073,
      "step": 1920
    },
    {
      "epoch": 35.74,
      "learning_rate": 1.523456790123457e-05,
      "loss": 0.0018,
      "step": 1930
    },
    {
      "epoch": 35.93,
      "learning_rate": 1.5209876543209879e-05,
      "loss": 0.0159,
      "step": 1940
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.0747406929731369,
      "eval_runtime": 3.0538,
      "eval_samples_per_second": 24.887,
      "eval_steps_per_second": 3.275,
      "step": 1944
    },
    {
      "epoch": 36.11,
      "learning_rate": 1.5185185185185187e-05,
      "loss": 0.0019,
      "step": 1950
    },
    {
      "epoch": 36.3,
      "learning_rate": 1.5160493827160495e-05,
      "loss": 0.0028,
      "step": 1960
    },
    {
      "epoch": 36.48,
      "learning_rate": 1.5135802469135803e-05,
      "loss": 0.0191,
      "step": 1970
    },
    {
      "epoch": 36.67,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 0.0018,
      "step": 1980
    },
    {
      "epoch": 36.85,
      "learning_rate": 1.5086419753086421e-05,
      "loss": 0.0018,
      "step": 1990
    },
    {
      "epoch": 37.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.02436704747378826,
      "eval_runtime": 3.1044,
      "eval_samples_per_second": 24.482,
      "eval_steps_per_second": 3.221,
      "step": 1998
    },
    {
      "epoch": 37.04,
      "learning_rate": 1.506172839506173e-05,
      "loss": 0.0019,
      "step": 2000
    },
    {
      "epoch": 37.22,
      "learning_rate": 1.5037037037037039e-05,
      "loss": 0.0018,
      "step": 2010
    },
    {
      "epoch": 37.41,
      "learning_rate": 1.5012345679012348e-05,
      "loss": 0.0025,
      "step": 2020
    },
    {
      "epoch": 37.59,
      "learning_rate": 1.4987654320987655e-05,
      "loss": 0.0017,
      "step": 2030
    },
    {
      "epoch": 37.78,
      "learning_rate": 1.4962962962962964e-05,
      "loss": 0.0017,
      "step": 2040
    },
    {
      "epoch": 37.96,
      "learning_rate": 1.4938271604938272e-05,
      "loss": 0.0016,
      "step": 2050
    },
    {
      "epoch": 38.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0018851511413231492,
      "eval_runtime": 3.0108,
      "eval_samples_per_second": 25.242,
      "eval_steps_per_second": 3.321,
      "step": 2052
    },
    {
      "epoch": 38.15,
      "learning_rate": 1.4913580246913581e-05,
      "loss": 0.0815,
      "step": 2060
    },
    {
      "epoch": 38.33,
      "learning_rate": 1.488888888888889e-05,
      "loss": 0.03,
      "step": 2070
    },
    {
      "epoch": 38.52,
      "learning_rate": 1.4864197530864199e-05,
      "loss": 0.0017,
      "step": 2080
    },
    {
      "epoch": 38.7,
      "learning_rate": 1.4839506172839508e-05,
      "loss": 0.0021,
      "step": 2090
    },
    {
      "epoch": 38.89,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 0.0017,
      "step": 2100
    },
    {
      "epoch": 39.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0017620852449908853,
      "eval_runtime": 3.0074,
      "eval_samples_per_second": 25.271,
      "eval_steps_per_second": 3.325,
      "step": 2106
    },
    {
      "epoch": 39.07,
      "learning_rate": 1.4790123456790124e-05,
      "loss": 0.0016,
      "step": 2110
    },
    {
      "epoch": 39.26,
      "learning_rate": 1.4765432098765433e-05,
      "loss": 0.0293,
      "step": 2120
    },
    {
      "epoch": 39.44,
      "learning_rate": 1.4740740740740741e-05,
      "loss": 0.0715,
      "step": 2130
    },
    {
      "epoch": 39.63,
      "learning_rate": 1.471604938271605e-05,
      "loss": 0.0355,
      "step": 2140
    },
    {
      "epoch": 39.81,
      "learning_rate": 1.469135802469136e-05,
      "loss": 0.0334,
      "step": 2150
    },
    {
      "epoch": 40.0,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 0.053,
      "step": 2160
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0022870118264108896,
      "eval_runtime": 3.0403,
      "eval_samples_per_second": 24.997,
      "eval_steps_per_second": 3.289,
      "step": 2160
    },
    {
      "epoch": 40.19,
      "learning_rate": 1.4641975308641975e-05,
      "loss": 0.0024,
      "step": 2170
    },
    {
      "epoch": 40.37,
      "learning_rate": 1.4617283950617284e-05,
      "loss": 0.0018,
      "step": 2180
    },
    {
      "epoch": 40.56,
      "learning_rate": 1.4592592592592594e-05,
      "loss": 0.0016,
      "step": 2190
    },
    {
      "epoch": 40.74,
      "learning_rate": 1.4567901234567903e-05,
      "loss": 0.0064,
      "step": 2200
    },
    {
      "epoch": 40.93,
      "learning_rate": 1.4543209876543212e-05,
      "loss": 0.0016,
      "step": 2210
    },
    {
      "epoch": 41.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.006081284489482641,
      "eval_runtime": 3.0753,
      "eval_samples_per_second": 24.713,
      "eval_steps_per_second": 3.252,
      "step": 2214
    },
    {
      "epoch": 41.11,
      "learning_rate": 1.4518518518518521e-05,
      "loss": 0.0771,
      "step": 2220
    },
    {
      "epoch": 41.3,
      "learning_rate": 1.4493827160493828e-05,
      "loss": 0.0015,
      "step": 2230
    },
    {
      "epoch": 41.48,
      "learning_rate": 1.4469135802469137e-05,
      "loss": 0.0016,
      "step": 2240
    },
    {
      "epoch": 41.67,
      "learning_rate": 1.4444444444444446e-05,
      "loss": 0.0015,
      "step": 2250
    },
    {
      "epoch": 41.85,
      "learning_rate": 1.4419753086419755e-05,
      "loss": 0.0015,
      "step": 2260
    },
    {
      "epoch": 42.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.010232136584818363,
      "eval_runtime": 3.0562,
      "eval_samples_per_second": 24.868,
      "eval_steps_per_second": 3.272,
      "step": 2268
    },
    {
      "epoch": 42.04,
      "learning_rate": 1.4395061728395063e-05,
      "loss": 0.0015,
      "step": 2270
    },
    {
      "epoch": 42.22,
      "learning_rate": 1.4370370370370372e-05,
      "loss": 0.0016,
      "step": 2280
    },
    {
      "epoch": 42.41,
      "learning_rate": 1.4345679012345681e-05,
      "loss": 0.0015,
      "step": 2290
    },
    {
      "epoch": 42.59,
      "learning_rate": 1.4320987654320988e-05,
      "loss": 0.0014,
      "step": 2300
    },
    {
      "epoch": 42.78,
      "learning_rate": 1.4296296296296297e-05,
      "loss": 0.0014,
      "step": 2310
    },
    {
      "epoch": 42.96,
      "learning_rate": 1.4271604938271606e-05,
      "loss": 0.0015,
      "step": 2320
    },
    {
      "epoch": 43.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0018882902804762125,
      "eval_runtime": 3.0482,
      "eval_samples_per_second": 24.932,
      "eval_steps_per_second": 3.281,
      "step": 2322
    },
    {
      "epoch": 43.15,
      "learning_rate": 1.4246913580246915e-05,
      "loss": 0.0083,
      "step": 2330
    },
    {
      "epoch": 43.33,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 0.0014,
      "step": 2340
    },
    {
      "epoch": 43.52,
      "learning_rate": 1.4197530864197532e-05,
      "loss": 0.002,
      "step": 2350
    },
    {
      "epoch": 43.7,
      "learning_rate": 1.417283950617284e-05,
      "loss": 0.0014,
      "step": 2360
    },
    {
      "epoch": 43.89,
      "learning_rate": 1.4148148148148148e-05,
      "loss": 0.0015,
      "step": 2370
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.006185897625982761,
      "eval_runtime": 3.0918,
      "eval_samples_per_second": 24.581,
      "eval_steps_per_second": 3.234,
      "step": 2376
    },
    {
      "epoch": 44.07,
      "learning_rate": 1.4123456790123457e-05,
      "loss": 0.0014,
      "step": 2380
    },
    {
      "epoch": 44.26,
      "learning_rate": 1.4098765432098766e-05,
      "loss": 0.1322,
      "step": 2390
    },
    {
      "epoch": 44.44,
      "learning_rate": 1.4074074074074075e-05,
      "loss": 0.0126,
      "step": 2400
    },
    {
      "epoch": 44.63,
      "learning_rate": 1.4049382716049384e-05,
      "loss": 0.0013,
      "step": 2410
    },
    {
      "epoch": 44.81,
      "learning_rate": 1.4024691358024694e-05,
      "loss": 0.0062,
      "step": 2420
    },
    {
      "epoch": 45.0,
      "learning_rate": 1.4e-05,
      "loss": 0.0014,
      "step": 2430
    },
    {
      "epoch": 45.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.001433242461644113,
      "eval_runtime": 3.0606,
      "eval_samples_per_second": 24.832,
      "eval_steps_per_second": 3.267,
      "step": 2430
    },
    {
      "epoch": 45.19,
      "learning_rate": 1.3975308641975309e-05,
      "loss": 0.0013,
      "step": 2440
    },
    {
      "epoch": 45.37,
      "learning_rate": 1.3950617283950617e-05,
      "loss": 0.0013,
      "step": 2450
    },
    {
      "epoch": 45.56,
      "learning_rate": 1.3925925925925928e-05,
      "loss": 0.0013,
      "step": 2460
    },
    {
      "epoch": 45.74,
      "learning_rate": 1.3901234567901237e-05,
      "loss": 0.0014,
      "step": 2470
    },
    {
      "epoch": 45.93,
      "learning_rate": 1.3876543209876546e-05,
      "loss": 0.0015,
      "step": 2480
    },
    {
      "epoch": 46.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0014952166238799691,
      "eval_runtime": 3.073,
      "eval_samples_per_second": 24.731,
      "eval_steps_per_second": 3.254,
      "step": 2484
    },
    {
      "epoch": 46.11,
      "learning_rate": 1.3851851851851851e-05,
      "loss": 0.0012,
      "step": 2490
    },
    {
      "epoch": 46.3,
      "learning_rate": 1.3827160493827162e-05,
      "loss": 0.0015,
      "step": 2500
    },
    {
      "epoch": 46.48,
      "learning_rate": 1.380246913580247e-05,
      "loss": 0.0014,
      "step": 2510
    },
    {
      "epoch": 46.67,
      "learning_rate": 1.377777777777778e-05,
      "loss": 0.0013,
      "step": 2520
    },
    {
      "epoch": 46.85,
      "learning_rate": 1.3753086419753088e-05,
      "loss": 0.0013,
      "step": 2530
    },
    {
      "epoch": 47.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06722357869148254,
      "eval_runtime": 3.0794,
      "eval_samples_per_second": 24.68,
      "eval_steps_per_second": 3.247,
      "step": 2538
    },
    {
      "epoch": 47.04,
      "learning_rate": 1.3728395061728397e-05,
      "loss": 0.0013,
      "step": 2540
    },
    {
      "epoch": 47.22,
      "learning_rate": 1.3703703703703706e-05,
      "loss": 0.0724,
      "step": 2550
    },
    {
      "epoch": 47.41,
      "learning_rate": 1.3679012345679013e-05,
      "loss": 0.0012,
      "step": 2560
    },
    {
      "epoch": 47.59,
      "learning_rate": 1.3654320987654322e-05,
      "loss": 0.0012,
      "step": 2570
    },
    {
      "epoch": 47.78,
      "learning_rate": 1.362962962962963e-05,
      "loss": 0.0012,
      "step": 2580
    },
    {
      "epoch": 47.96,
      "learning_rate": 1.360493827160494e-05,
      "loss": 0.0012,
      "step": 2590
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0014614396495744586,
      "eval_runtime": 3.0742,
      "eval_samples_per_second": 24.722,
      "eval_steps_per_second": 3.253,
      "step": 2592
    },
    {
      "epoch": 48.15,
      "learning_rate": 1.3580246913580248e-05,
      "loss": 0.0012,
      "step": 2600
    },
    {
      "epoch": 48.33,
      "learning_rate": 1.3555555555555557e-05,
      "loss": 0.0012,
      "step": 2610
    },
    {
      "epoch": 48.52,
      "learning_rate": 1.3530864197530866e-05,
      "loss": 0.0014,
      "step": 2620
    },
    {
      "epoch": 48.7,
      "learning_rate": 1.3506172839506173e-05,
      "loss": 0.018,
      "step": 2630
    },
    {
      "epoch": 48.89,
      "learning_rate": 1.3481481481481482e-05,
      "loss": 0.0012,
      "step": 2640
    },
    {
      "epoch": 49.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.06998171657323837,
      "eval_runtime": 3.0377,
      "eval_samples_per_second": 25.019,
      "eval_steps_per_second": 3.292,
      "step": 2646
    },
    {
      "epoch": 49.07,
      "learning_rate": 1.345679012345679e-05,
      "loss": 0.0488,
      "step": 2650
    },
    {
      "epoch": 49.26,
      "learning_rate": 1.34320987654321e-05,
      "loss": 0.0267,
      "step": 2660
    },
    {
      "epoch": 49.44,
      "learning_rate": 1.3407407407407408e-05,
      "loss": 0.0012,
      "step": 2670
    },
    {
      "epoch": 49.63,
      "learning_rate": 1.3382716049382717e-05,
      "loss": 0.0012,
      "step": 2680
    },
    {
      "epoch": 49.81,
      "learning_rate": 1.3358024691358024e-05,
      "loss": 0.0012,
      "step": 2690
    },
    {
      "epoch": 50.0,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0012,
      "step": 2700
    },
    {
      "epoch": 50.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.057919345796108246,
      "eval_runtime": 3.0575,
      "eval_samples_per_second": 24.857,
      "eval_steps_per_second": 3.271,
      "step": 2700
    },
    {
      "epoch": 50.19,
      "learning_rate": 1.3308641975308642e-05,
      "loss": 0.0016,
      "step": 2710
    },
    {
      "epoch": 50.37,
      "learning_rate": 1.3283950617283951e-05,
      "loss": 0.0011,
      "step": 2720
    },
    {
      "epoch": 50.56,
      "learning_rate": 1.325925925925926e-05,
      "loss": 0.0011,
      "step": 2730
    },
    {
      "epoch": 50.74,
      "learning_rate": 1.323456790123457e-05,
      "loss": 0.0013,
      "step": 2740
    },
    {
      "epoch": 50.93,
      "learning_rate": 1.320987654320988e-05,
      "loss": 0.0011,
      "step": 2750
    },
    {
      "epoch": 51.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.05705736204981804,
      "eval_runtime": 3.0295,
      "eval_samples_per_second": 25.087,
      "eval_steps_per_second": 3.301,
      "step": 2754
    },
    {
      "epoch": 51.11,
      "learning_rate": 1.3185185185185185e-05,
      "loss": 0.0011,
      "step": 2760
    },
    {
      "epoch": 51.3,
      "learning_rate": 1.3160493827160493e-05,
      "loss": 0.0011,
      "step": 2770
    },
    {
      "epoch": 51.48,
      "learning_rate": 1.3135802469135804e-05,
      "loss": 0.0011,
      "step": 2780
    },
    {
      "epoch": 51.67,
      "learning_rate": 1.3111111111111113e-05,
      "loss": 0.0011,
      "step": 2790
    },
    {
      "epoch": 51.85,
      "learning_rate": 1.3086419753086422e-05,
      "loss": 0.001,
      "step": 2800
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.0670037791132927,
      "eval_runtime": 3.067,
      "eval_samples_per_second": 24.78,
      "eval_steps_per_second": 3.261,
      "step": 2808
    },
    {
      "epoch": 52.04,
      "learning_rate": 1.306172839506173e-05,
      "loss": 0.001,
      "step": 2810
    },
    {
      "epoch": 52.22,
      "learning_rate": 1.303703703703704e-05,
      "loss": 0.0011,
      "step": 2820
    },
    {
      "epoch": 52.41,
      "learning_rate": 1.3012345679012346e-05,
      "loss": 0.001,
      "step": 2830
    },
    {
      "epoch": 52.59,
      "learning_rate": 1.2987654320987655e-05,
      "loss": 0.001,
      "step": 2840
    },
    {
      "epoch": 52.78,
      "learning_rate": 1.2962962962962964e-05,
      "loss": 0.0012,
      "step": 2850
    },
    {
      "epoch": 52.96,
      "learning_rate": 1.2938271604938273e-05,
      "loss": 0.001,
      "step": 2860
    },
    {
      "epoch": 53.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.07296162843704224,
      "eval_runtime": 3.0567,
      "eval_samples_per_second": 24.863,
      "eval_steps_per_second": 3.271,
      "step": 2862
    },
    {
      "epoch": 53.15,
      "learning_rate": 1.2913580246913582e-05,
      "loss": 0.001,
      "step": 2870
    },
    {
      "epoch": 53.33,
      "learning_rate": 1.288888888888889e-05,
      "loss": 0.001,
      "step": 2880
    },
    {
      "epoch": 53.52,
      "learning_rate": 1.2864197530864198e-05,
      "loss": 0.001,
      "step": 2890
    },
    {
      "epoch": 53.7,
      "learning_rate": 1.2839506172839507e-05,
      "loss": 0.057,
      "step": 2900
    },
    {
      "epoch": 53.89,
      "learning_rate": 1.2814814814814815e-05,
      "loss": 0.0013,
      "step": 2910
    },
    {
      "epoch": 54.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.013546849600970745,
      "eval_runtime": 2.9939,
      "eval_samples_per_second": 25.385,
      "eval_steps_per_second": 3.34,
      "step": 2916
    },
    {
      "epoch": 54.07,
      "learning_rate": 1.2790123456790124e-05,
      "loss": 0.001,
      "step": 2920
    },
    {
      "epoch": 54.26,
      "learning_rate": 1.2765432098765433e-05,
      "loss": 0.0012,
      "step": 2930
    },
    {
      "epoch": 54.44,
      "learning_rate": 1.2740740740740742e-05,
      "loss": 0.0011,
      "step": 2940
    },
    {
      "epoch": 54.63,
      "learning_rate": 1.271604938271605e-05,
      "loss": 0.0009,
      "step": 2950
    },
    {
      "epoch": 54.81,
      "learning_rate": 1.2691358024691358e-05,
      "loss": 0.001,
      "step": 2960
    },
    {
      "epoch": 55.0,
      "learning_rate": 1.2666666666666667e-05,
      "loss": 0.001,
      "step": 2970
    },
    {
      "epoch": 55.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.08355703204870224,
      "eval_runtime": 3.0103,
      "eval_samples_per_second": 25.247,
      "eval_steps_per_second": 3.322,
      "step": 2970
    },
    {
      "epoch": 55.19,
      "learning_rate": 1.2641975308641976e-05,
      "loss": 0.0009,
      "step": 2980
    },
    {
      "epoch": 55.37,
      "learning_rate": 1.2617283950617284e-05,
      "loss": 0.0476,
      "step": 2990
    },
    {
      "epoch": 55.56,
      "learning_rate": 1.2592592592592593e-05,
      "loss": 0.0009,
      "step": 3000
    },
    {
      "epoch": 55.74,
      "learning_rate": 1.2567901234567904e-05,
      "loss": 0.001,
      "step": 3010
    },
    {
      "epoch": 55.93,
      "learning_rate": 1.2543209876543213e-05,
      "loss": 0.0009,
      "step": 3020
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0010229795007035136,
      "eval_runtime": 3.0546,
      "eval_samples_per_second": 24.881,
      "eval_steps_per_second": 3.274,
      "step": 3024
    },
    {
      "epoch": 56.11,
      "learning_rate": 1.2518518518518518e-05,
      "loss": 0.0616,
      "step": 3030
    },
    {
      "epoch": 56.3,
      "learning_rate": 1.2493827160493827e-05,
      "loss": 0.0185,
      "step": 3040
    },
    {
      "epoch": 56.48,
      "learning_rate": 1.2469135802469137e-05,
      "loss": 0.0011,
      "step": 3050
    },
    {
      "epoch": 56.67,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 0.001,
      "step": 3060
    },
    {
      "epoch": 56.85,
      "learning_rate": 1.2419753086419755e-05,
      "loss": 0.0009,
      "step": 3070
    },
    {
      "epoch": 57.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.012184691615402699,
      "eval_runtime": 2.9929,
      "eval_samples_per_second": 25.393,
      "eval_steps_per_second": 3.341,
      "step": 3078
    },
    {
      "epoch": 57.04,
      "learning_rate": 1.2395061728395064e-05,
      "loss": 0.0009,
      "step": 3080
    },
    {
      "epoch": 57.22,
      "learning_rate": 1.2370370370370371e-05,
      "loss": 0.0009,
      "step": 3090
    },
    {
      "epoch": 57.41,
      "learning_rate": 1.234567901234568e-05,
      "loss": 0.0719,
      "step": 3100
    },
    {
      "epoch": 57.59,
      "learning_rate": 1.2320987654320989e-05,
      "loss": 0.0009,
      "step": 3110
    },
    {
      "epoch": 57.78,
      "learning_rate": 1.2296296296296298e-05,
      "loss": 0.0009,
      "step": 3120
    },
    {
      "epoch": 57.96,
      "learning_rate": 1.2271604938271606e-05,
      "loss": 0.001,
      "step": 3130
    },
    {
      "epoch": 58.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.010523747652769089,
      "eval_runtime": 3.0537,
      "eval_samples_per_second": 24.887,
      "eval_steps_per_second": 3.275,
      "step": 3132
    },
    {
      "epoch": 58.15,
      "learning_rate": 1.2246913580246915e-05,
      "loss": 0.0124,
      "step": 3140
    },
    {
      "epoch": 58.33,
      "learning_rate": 1.2222222222222224e-05,
      "loss": 0.0255,
      "step": 3150
    },
    {
      "epoch": 58.52,
      "learning_rate": 1.2197530864197531e-05,
      "loss": 0.0778,
      "step": 3160
    },
    {
      "epoch": 58.7,
      "learning_rate": 1.217283950617284e-05,
      "loss": 0.0009,
      "step": 3170
    },
    {
      "epoch": 58.89,
      "learning_rate": 1.2148148148148149e-05,
      "loss": 0.0017,
      "step": 3180
    },
    {
      "epoch": 59.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.007372140884399414,
      "eval_runtime": 2.9827,
      "eval_samples_per_second": 25.481,
      "eval_steps_per_second": 3.353,
      "step": 3186
    },
    {
      "epoch": 59.07,
      "learning_rate": 1.2123456790123458e-05,
      "loss": 0.0009,
      "step": 3190
    },
    {
      "epoch": 59.26,
      "learning_rate": 1.2098765432098767e-05,
      "loss": 0.0009,
      "step": 3200
    },
    {
      "epoch": 59.44,
      "learning_rate": 1.2074074074074075e-05,
      "loss": 0.0009,
      "step": 3210
    },
    {
      "epoch": 59.63,
      "learning_rate": 1.2049382716049384e-05,
      "loss": 0.0009,
      "step": 3220
    },
    {
      "epoch": 59.81,
      "learning_rate": 1.2024691358024691e-05,
      "loss": 0.0009,
      "step": 3230
    },
    {
      "epoch": 60.0,
      "learning_rate": 1.2e-05,
      "loss": 0.0009,
      "step": 3240
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.001023871824145317,
      "eval_runtime": 3.0602,
      "eval_samples_per_second": 24.835,
      "eval_steps_per_second": 3.268,
      "step": 3240
    },
    {
      "epoch": 60.19,
      "learning_rate": 1.1975308641975309e-05,
      "loss": 0.0009,
      "step": 3250
    },
    {
      "epoch": 60.37,
      "learning_rate": 1.1950617283950618e-05,
      "loss": 0.0008,
      "step": 3260
    },
    {
      "epoch": 60.56,
      "learning_rate": 1.1925925925925927e-05,
      "loss": 0.0009,
      "step": 3270
    },
    {
      "epoch": 60.74,
      "learning_rate": 1.1901234567901236e-05,
      "loss": 0.0008,
      "step": 3280
    },
    {
      "epoch": 60.93,
      "learning_rate": 1.1876543209876543e-05,
      "loss": 0.0009,
      "step": 3290
    },
    {
      "epoch": 61.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0009232511511072516,
      "eval_runtime": 3.0495,
      "eval_samples_per_second": 24.922,
      "eval_steps_per_second": 3.279,
      "step": 3294
    },
    {
      "epoch": 61.11,
      "learning_rate": 1.1851851851851852e-05,
      "loss": 0.0009,
      "step": 3300
    },
    {
      "epoch": 61.3,
      "learning_rate": 1.182716049382716e-05,
      "loss": 0.0008,
      "step": 3310
    },
    {
      "epoch": 61.48,
      "learning_rate": 1.180246913580247e-05,
      "loss": 0.0008,
      "step": 3320
    },
    {
      "epoch": 61.67,
      "learning_rate": 1.177777777777778e-05,
      "loss": 0.0008,
      "step": 3330
    },
    {
      "epoch": 61.85,
      "learning_rate": 1.1753086419753089e-05,
      "loss": 0.0381,
      "step": 3340
    },
    {
      "epoch": 62.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0020482551772147417,
      "eval_runtime": 3.0331,
      "eval_samples_per_second": 25.057,
      "eval_steps_per_second": 3.297,
      "step": 3348
    },
    {
      "epoch": 62.04,
      "learning_rate": 1.1728395061728398e-05,
      "loss": 0.0008,
      "step": 3350
    },
    {
      "epoch": 62.22,
      "learning_rate": 1.1703703703703703e-05,
      "loss": 0.0963,
      "step": 3360
    },
    {
      "epoch": 62.41,
      "learning_rate": 1.1679012345679013e-05,
      "loss": 0.0008,
      "step": 3370
    },
    {
      "epoch": 62.59,
      "learning_rate": 1.1654320987654322e-05,
      "loss": 0.0009,
      "step": 3380
    },
    {
      "epoch": 62.78,
      "learning_rate": 1.1629629629629631e-05,
      "loss": 0.0012,
      "step": 3390
    },
    {
      "epoch": 62.96,
      "learning_rate": 1.160493827160494e-05,
      "loss": 0.0008,
      "step": 3400
    },
    {
      "epoch": 63.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0008004654664546251,
      "eval_runtime": 2.9607,
      "eval_samples_per_second": 25.669,
      "eval_steps_per_second": 3.378,
      "step": 3402
    },
    {
      "epoch": 63.15,
      "learning_rate": 1.1580246913580249e-05,
      "loss": 0.0008,
      "step": 3410
    },
    {
      "epoch": 63.33,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 0.0008,
      "step": 3420
    },
    {
      "epoch": 63.52,
      "learning_rate": 1.1530864197530865e-05,
      "loss": 0.0008,
      "step": 3430
    },
    {
      "epoch": 63.7,
      "learning_rate": 1.1506172839506174e-05,
      "loss": 0.0007,
      "step": 3440
    },
    {
      "epoch": 63.89,
      "learning_rate": 1.1481481481481482e-05,
      "loss": 0.0099,
      "step": 3450
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0007782263564877212,
      "eval_runtime": 2.9249,
      "eval_samples_per_second": 25.984,
      "eval_steps_per_second": 3.419,
      "step": 3456
    },
    {
      "epoch": 64.07,
      "learning_rate": 1.1456790123456791e-05,
      "loss": 0.2271,
      "step": 3460
    },
    {
      "epoch": 64.26,
      "learning_rate": 1.14320987654321e-05,
      "loss": 0.0007,
      "step": 3470
    },
    {
      "epoch": 64.44,
      "learning_rate": 1.1407407407407409e-05,
      "loss": 0.0008,
      "step": 3480
    },
    {
      "epoch": 64.63,
      "learning_rate": 1.1382716049382716e-05,
      "loss": 0.0007,
      "step": 3490
    },
    {
      "epoch": 64.81,
      "learning_rate": 1.1358024691358025e-05,
      "loss": 0.0008,
      "step": 3500
    },
    {
      "epoch": 65.0,
      "learning_rate": 1.1333333333333334e-05,
      "loss": 0.0007,
      "step": 3510
    },
    {
      "epoch": 65.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.07570363581180573,
      "eval_runtime": 3.0318,
      "eval_samples_per_second": 25.068,
      "eval_steps_per_second": 3.298,
      "step": 3510
    },
    {
      "epoch": 65.19,
      "learning_rate": 1.1308641975308643e-05,
      "loss": 0.0009,
      "step": 3520
    },
    {
      "epoch": 65.37,
      "learning_rate": 1.1283950617283951e-05,
      "loss": 0.0009,
      "step": 3530
    },
    {
      "epoch": 65.56,
      "learning_rate": 1.125925925925926e-05,
      "loss": 0.0007,
      "step": 3540
    },
    {
      "epoch": 65.74,
      "learning_rate": 1.123456790123457e-05,
      "loss": 0.0007,
      "step": 3550
    },
    {
      "epoch": 65.93,
      "learning_rate": 1.1209876543209876e-05,
      "loss": 0.0008,
      "step": 3560
    },
    {
      "epoch": 66.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.07635856419801712,
      "eval_runtime": 3.1106,
      "eval_samples_per_second": 24.432,
      "eval_steps_per_second": 3.215,
      "step": 3564
    },
    {
      "epoch": 66.11,
      "learning_rate": 1.1185185185185185e-05,
      "loss": 0.0752,
      "step": 3570
    },
    {
      "epoch": 66.3,
      "learning_rate": 1.1160493827160494e-05,
      "loss": 0.0008,
      "step": 3580
    },
    {
      "epoch": 66.48,
      "learning_rate": 1.1135802469135803e-05,
      "loss": 0.0788,
      "step": 3590
    },
    {
      "epoch": 66.67,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 0.0007,
      "step": 3600
    },
    {
      "epoch": 66.85,
      "learning_rate": 1.1086419753086422e-05,
      "loss": 0.0007,
      "step": 3610
    },
    {
      "epoch": 67.0,
      "eval_accuracy": 0.9736842105263158,
      "eval_loss": 0.12568235397338867,
      "eval_runtime": 2.964,
      "eval_samples_per_second": 25.641,
      "eval_steps_per_second": 3.374,
      "step": 3618
    },
    {
      "epoch": 67.04,
      "learning_rate": 1.1061728395061728e-05,
      "loss": 0.001,
      "step": 3620
    },
    {
      "epoch": 67.22,
      "learning_rate": 1.1037037037037036e-05,
      "loss": 0.0007,
      "step": 3630
    },
    {
      "epoch": 67.41,
      "learning_rate": 1.1012345679012347e-05,
      "loss": 0.0011,
      "step": 3640
    },
    {
      "epoch": 67.59,
      "learning_rate": 1.0987654320987656e-05,
      "loss": 0.0007,
      "step": 3650
    },
    {
      "epoch": 67.78,
      "learning_rate": 1.0962962962962965e-05,
      "loss": 0.0007,
      "step": 3660
    },
    {
      "epoch": 67.96,
      "learning_rate": 1.0938271604938273e-05,
      "loss": 0.0007,
      "step": 3670
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.009799230843782425,
      "eval_runtime": 3.0392,
      "eval_samples_per_second": 25.007,
      "eval_steps_per_second": 3.29,
      "step": 3672
    },
    {
      "epoch": 68.15,
      "learning_rate": 1.0913580246913582e-05,
      "loss": 0.0841,
      "step": 3680
    },
    {
      "epoch": 68.33,
      "learning_rate": 1.088888888888889e-05,
      "loss": 0.0007,
      "step": 3690
    },
    {
      "epoch": 68.52,
      "learning_rate": 1.0864197530864198e-05,
      "loss": 0.0007,
      "step": 3700
    },
    {
      "epoch": 68.7,
      "learning_rate": 1.0839506172839507e-05,
      "loss": 0.0382,
      "step": 3710
    },
    {
      "epoch": 68.89,
      "learning_rate": 1.0814814814814816e-05,
      "loss": 0.0736,
      "step": 3720
    },
    {
      "epoch": 69.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0007917585899122059,
      "eval_runtime": 3.0348,
      "eval_samples_per_second": 25.043,
      "eval_steps_per_second": 3.295,
      "step": 3726
    },
    {
      "epoch": 69.07,
      "learning_rate": 1.0790123456790125e-05,
      "loss": 0.0007,
      "step": 3730
    },
    {
      "epoch": 69.26,
      "learning_rate": 1.0765432098765434e-05,
      "loss": 0.0007,
      "step": 3740
    },
    {
      "epoch": 69.44,
      "learning_rate": 1.0740740740740742e-05,
      "loss": 0.0007,
      "step": 3750
    },
    {
      "epoch": 69.63,
      "learning_rate": 1.071604938271605e-05,
      "loss": 0.0008,
      "step": 3760
    },
    {
      "epoch": 69.81,
      "learning_rate": 1.0691358024691358e-05,
      "loss": 0.0014,
      "step": 3770
    },
    {
      "epoch": 70.0,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.0007,
      "step": 3780
    },
    {
      "epoch": 70.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.060479313135147095,
      "eval_runtime": 3.0758,
      "eval_samples_per_second": 24.709,
      "eval_steps_per_second": 3.251,
      "step": 3780
    },
    {
      "epoch": 70.19,
      "learning_rate": 1.0641975308641976e-05,
      "loss": 0.026,
      "step": 3790
    },
    {
      "epoch": 70.37,
      "learning_rate": 1.0617283950617285e-05,
      "loss": 0.0022,
      "step": 3800
    },
    {
      "epoch": 70.56,
      "learning_rate": 1.0592592592592594e-05,
      "loss": 0.0007,
      "step": 3810
    },
    {
      "epoch": 70.74,
      "learning_rate": 1.0567901234567901e-05,
      "loss": 0.0067,
      "step": 3820
    },
    {
      "epoch": 70.93,
      "learning_rate": 1.054320987654321e-05,
      "loss": 0.0006,
      "step": 3830
    },
    {
      "epoch": 71.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0011679435847327113,
      "eval_runtime": 3.0887,
      "eval_samples_per_second": 24.606,
      "eval_steps_per_second": 3.238,
      "step": 3834
    },
    {
      "epoch": 71.11,
      "learning_rate": 1.0518518518518519e-05,
      "loss": 0.0007,
      "step": 3840
    },
    {
      "epoch": 71.3,
      "learning_rate": 1.0493827160493827e-05,
      "loss": 0.0007,
      "step": 3850
    },
    {
      "epoch": 71.48,
      "learning_rate": 1.0469135802469136e-05,
      "loss": 0.0591,
      "step": 3860
    },
    {
      "epoch": 71.67,
      "learning_rate": 1.0444444444444445e-05,
      "loss": 0.0737,
      "step": 3870
    },
    {
      "epoch": 71.85,
      "learning_rate": 1.0419753086419756e-05,
      "loss": 0.001,
      "step": 3880
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.9736842105263158,
      "eval_loss": 0.16656266152858734,
      "eval_runtime": 3.0805,
      "eval_samples_per_second": 24.672,
      "eval_steps_per_second": 3.246,
      "step": 3888
    },
    {
      "epoch": 72.04,
      "learning_rate": 1.0395061728395061e-05,
      "loss": 0.0008,
      "step": 3890
    },
    {
      "epoch": 72.22,
      "learning_rate": 1.037037037037037e-05,
      "loss": 0.0015,
      "step": 3900
    },
    {
      "epoch": 72.41,
      "learning_rate": 1.0345679012345679e-05,
      "loss": 0.0065,
      "step": 3910
    },
    {
      "epoch": 72.59,
      "learning_rate": 1.032098765432099e-05,
      "loss": 0.0007,
      "step": 3920
    },
    {
      "epoch": 72.78,
      "learning_rate": 1.0296296296296298e-05,
      "loss": 0.0008,
      "step": 3930
    },
    {
      "epoch": 72.96,
      "learning_rate": 1.0271604938271607e-05,
      "loss": 0.0042,
      "step": 3940
    },
    {
      "epoch": 73.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0007003291393630207,
      "eval_runtime": 3.0627,
      "eval_samples_per_second": 24.815,
      "eval_steps_per_second": 3.265,
      "step": 3942
    },
    {
      "epoch": 73.15,
      "learning_rate": 1.0246913580246916e-05,
      "loss": 0.0007,
      "step": 3950
    },
    {
      "epoch": 73.33,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 0.0141,
      "step": 3960
    },
    {
      "epoch": 73.52,
      "learning_rate": 1.0197530864197532e-05,
      "loss": 0.0006,
      "step": 3970
    },
    {
      "epoch": 73.7,
      "learning_rate": 1.017283950617284e-05,
      "loss": 0.0006,
      "step": 3980
    },
    {
      "epoch": 73.89,
      "learning_rate": 1.014814814814815e-05,
      "loss": 0.0006,
      "step": 3990
    },
    {
      "epoch": 74.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0006768173188902438,
      "eval_runtime": 3.0459,
      "eval_samples_per_second": 24.952,
      "eval_steps_per_second": 3.283,
      "step": 3996
    },
    {
      "epoch": 74.07,
      "learning_rate": 1.0123456790123458e-05,
      "loss": 0.0007,
      "step": 4000
    },
    {
      "epoch": 74.26,
      "learning_rate": 1.0098765432098767e-05,
      "loss": 0.0006,
      "step": 4010
    },
    {
      "epoch": 74.44,
      "learning_rate": 1.0074074074074074e-05,
      "loss": 0.0006,
      "step": 4020
    },
    {
      "epoch": 74.63,
      "learning_rate": 1.0049382716049383e-05,
      "loss": 0.0006,
      "step": 4030
    },
    {
      "epoch": 74.81,
      "learning_rate": 1.0024691358024692e-05,
      "loss": 0.0006,
      "step": 4040
    },
    {
      "epoch": 75.0,
      "learning_rate": 1e-05,
      "loss": 0.0007,
      "step": 4050
    },
    {
      "epoch": 75.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0006513882544822991,
      "eval_runtime": 3.0655,
      "eval_samples_per_second": 24.792,
      "eval_steps_per_second": 3.262,
      "step": 4050
    },
    {
      "epoch": 75.19,
      "learning_rate": 9.97530864197531e-06,
      "loss": 0.001,
      "step": 4060
    },
    {
      "epoch": 75.37,
      "learning_rate": 9.950617283950618e-06,
      "loss": 0.0006,
      "step": 4070
    },
    {
      "epoch": 75.56,
      "learning_rate": 9.925925925925927e-06,
      "loss": 0.0006,
      "step": 4080
    },
    {
      "epoch": 75.74,
      "learning_rate": 9.901234567901236e-06,
      "loss": 0.0006,
      "step": 4090
    },
    {
      "epoch": 75.93,
      "learning_rate": 9.876543209876543e-06,
      "loss": 0.0006,
      "step": 4100
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.03311971202492714,
      "eval_runtime": 2.9918,
      "eval_samples_per_second": 25.403,
      "eval_steps_per_second": 3.343,
      "step": 4104
    },
    {
      "epoch": 76.11,
      "learning_rate": 9.851851851851852e-06,
      "loss": 0.0006,
      "step": 4110
    },
    {
      "epoch": 76.3,
      "learning_rate": 9.827160493827161e-06,
      "loss": 0.0006,
      "step": 4120
    },
    {
      "epoch": 76.48,
      "learning_rate": 9.80246913580247e-06,
      "loss": 0.0006,
      "step": 4130
    },
    {
      "epoch": 76.67,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.0006,
      "step": 4140
    },
    {
      "epoch": 76.85,
      "learning_rate": 9.753086419753087e-06,
      "loss": 0.0006,
      "step": 4150
    },
    {
      "epoch": 77.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.01688443496823311,
      "eval_runtime": 3.0085,
      "eval_samples_per_second": 25.262,
      "eval_steps_per_second": 3.324,
      "step": 4158
    },
    {
      "epoch": 77.04,
      "learning_rate": 9.728395061728396e-06,
      "loss": 0.0005,
      "step": 4160
    },
    {
      "epoch": 77.22,
      "learning_rate": 9.703703703703703e-06,
      "loss": 0.0006,
      "step": 4170
    },
    {
      "epoch": 77.41,
      "learning_rate": 9.679012345679012e-06,
      "loss": 0.0005,
      "step": 4180
    },
    {
      "epoch": 77.59,
      "learning_rate": 9.654320987654323e-06,
      "loss": 0.0005,
      "step": 4190
    },
    {
      "epoch": 77.78,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.0005,
      "step": 4200
    },
    {
      "epoch": 77.96,
      "learning_rate": 9.604938271604939e-06,
      "loss": 0.0345,
      "step": 4210
    },
    {
      "epoch": 78.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0006252001621760428,
      "eval_runtime": 3.0532,
      "eval_samples_per_second": 24.892,
      "eval_steps_per_second": 3.275,
      "step": 4212
    },
    {
      "epoch": 78.15,
      "learning_rate": 9.580246913580248e-06,
      "loss": 0.0769,
      "step": 4220
    },
    {
      "epoch": 78.33,
      "learning_rate": 9.555555555555556e-06,
      "loss": 0.0006,
      "step": 4230
    },
    {
      "epoch": 78.52,
      "learning_rate": 9.530864197530865e-06,
      "loss": 0.0006,
      "step": 4240
    },
    {
      "epoch": 78.7,
      "learning_rate": 9.506172839506174e-06,
      "loss": 0.0035,
      "step": 4250
    },
    {
      "epoch": 78.89,
      "learning_rate": 9.481481481481483e-06,
      "loss": 0.0005,
      "step": 4260
    },
    {
      "epoch": 79.0,
      "eval_accuracy": 0.9868421052631579,
      "eval_loss": 0.07624808698892593,
      "eval_runtime": 3.0922,
      "eval_samples_per_second": 24.578,
      "eval_steps_per_second": 3.234,
      "step": 4266
    },
    {
      "epoch": 79.07,
      "learning_rate": 9.45679012345679e-06,
      "loss": 0.0011,
      "step": 4270
    },
    {
      "epoch": 79.26,
      "learning_rate": 9.432098765432099e-06,
      "loss": 0.0005,
      "step": 4280
    },
    {
      "epoch": 79.44,
      "learning_rate": 9.407407407407408e-06,
      "loss": 0.0005,
      "step": 4290
    },
    {
      "epoch": 79.63,
      "learning_rate": 9.382716049382717e-06,
      "loss": 0.0005,
      "step": 4300
    },
    {
      "epoch": 79.81,
      "learning_rate": 9.358024691358025e-06,
      "loss": 0.0006,
      "step": 4310
    },
    {
      "epoch": 80.0,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.0005,
      "step": 4320
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0006611607386730611,
      "eval_runtime": 3.0596,
      "eval_samples_per_second": 24.84,
      "eval_steps_per_second": 3.268,
      "step": 4320
    },
    {
      "epoch": 80.19,
      "learning_rate": 9.308641975308643e-06,
      "loss": 0.0005,
      "step": 4330
    },
    {
      "epoch": 80.37,
      "learning_rate": 9.28395061728395e-06,
      "loss": 0.0027,
      "step": 4340
    },
    {
      "epoch": 80.56,
      "learning_rate": 9.25925925925926e-06,
      "loss": 0.0005,
      "step": 4350
    },
    {
      "epoch": 80.74,
      "learning_rate": 9.23456790123457e-06,
      "loss": 0.0006,
      "step": 4360
    },
    {
      "epoch": 80.93,
      "learning_rate": 9.209876543209877e-06,
      "loss": 0.0005,
      "step": 4370
    },
    {
      "epoch": 81.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0005299604381434619,
      "eval_runtime": 3.0798,
      "eval_samples_per_second": 24.677,
      "eval_steps_per_second": 3.247,
      "step": 4374
    },
    {
      "epoch": 81.11,
      "learning_rate": 9.185185185185186e-06,
      "loss": 0.0005,
      "step": 4380
    },
    {
      "epoch": 81.3,
      "learning_rate": 9.160493827160494e-06,
      "loss": 0.0008,
      "step": 4390
    },
    {
      "epoch": 81.48,
      "learning_rate": 9.135802469135803e-06,
      "loss": 0.0163,
      "step": 4400
    },
    {
      "epoch": 81.67,
      "learning_rate": 9.111111111111112e-06,
      "loss": 0.0005,
      "step": 4410
    },
    {
      "epoch": 81.85,
      "learning_rate": 9.086419753086421e-06,
      "loss": 0.0005,
      "step": 4420
    },
    {
      "epoch": 82.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.000567257811781019,
      "eval_runtime": 3.0454,
      "eval_samples_per_second": 24.956,
      "eval_steps_per_second": 3.284,
      "step": 4428
    },
    {
      "epoch": 82.04,
      "learning_rate": 9.06172839506173e-06,
      "loss": 0.0005,
      "step": 4430
    },
    {
      "epoch": 82.22,
      "learning_rate": 9.037037037037037e-06,
      "loss": 0.0005,
      "step": 4440
    },
    {
      "epoch": 82.41,
      "learning_rate": 9.012345679012346e-06,
      "loss": 0.0005,
      "step": 4450
    },
    {
      "epoch": 82.59,
      "learning_rate": 8.987654320987655e-06,
      "loss": 0.0005,
      "step": 4460
    },
    {
      "epoch": 82.78,
      "learning_rate": 8.962962962962963e-06,
      "loss": 0.0007,
      "step": 4470
    },
    {
      "epoch": 82.96,
      "learning_rate": 8.938271604938272e-06,
      "loss": 0.0005,
      "step": 4480
    },
    {
      "epoch": 83.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00053614896023646,
      "eval_runtime": 3.1232,
      "eval_samples_per_second": 24.334,
      "eval_steps_per_second": 3.202,
      "step": 4482
    },
    {
      "epoch": 83.15,
      "learning_rate": 8.913580246913581e-06,
      "loss": 0.0005,
      "step": 4490
    },
    {
      "epoch": 83.33,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.0005,
      "step": 4500
    },
    {
      "epoch": 83.52,
      "learning_rate": 8.864197530864199e-06,
      "loss": 0.0005,
      "step": 4510
    },
    {
      "epoch": 83.7,
      "learning_rate": 8.839506172839508e-06,
      "loss": 0.0005,
      "step": 4520
    },
    {
      "epoch": 83.89,
      "learning_rate": 8.814814814814817e-06,
      "loss": 0.0005,
      "step": 4530
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0005005816929042339,
      "eval_runtime": 3.0191,
      "eval_samples_per_second": 25.173,
      "eval_steps_per_second": 3.312,
      "step": 4536
    },
    {
      "epoch": 84.07,
      "learning_rate": 8.790123456790124e-06,
      "loss": 0.0005,
      "step": 4540
    },
    {
      "epoch": 84.26,
      "learning_rate": 8.765432098765432e-06,
      "loss": 0.0005,
      "step": 4550
    },
    {
      "epoch": 84.44,
      "learning_rate": 8.740740740740741e-06,
      "loss": 0.0042,
      "step": 4560
    },
    {
      "epoch": 84.63,
      "learning_rate": 8.71604938271605e-06,
      "loss": 0.0004,
      "step": 4570
    },
    {
      "epoch": 84.81,
      "learning_rate": 8.691358024691359e-06,
      "loss": 0.0074,
      "step": 4580
    },
    {
      "epoch": 85.0,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.0047,
      "step": 4590
    },
    {
      "epoch": 85.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0006987836095504463,
      "eval_runtime": 3.0927,
      "eval_samples_per_second": 24.574,
      "eval_steps_per_second": 3.233,
      "step": 4590
    },
    {
      "epoch": 85.19,
      "learning_rate": 8.641975308641975e-06,
      "loss": 0.0005,
      "step": 4600
    },
    {
      "epoch": 85.37,
      "learning_rate": 8.617283950617284e-06,
      "loss": 0.0004,
      "step": 4610
    },
    {
      "epoch": 85.56,
      "learning_rate": 8.592592592592593e-06,
      "loss": 0.0005,
      "step": 4620
    },
    {
      "epoch": 85.74,
      "learning_rate": 8.567901234567903e-06,
      "loss": 0.06,
      "step": 4630
    },
    {
      "epoch": 85.93,
      "learning_rate": 8.54320987654321e-06,
      "loss": 0.0005,
      "step": 4640
    },
    {
      "epoch": 86.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00046328367898240685,
      "eval_runtime": 3.024,
      "eval_samples_per_second": 25.132,
      "eval_steps_per_second": 3.307,
      "step": 4644
    },
    {
      "epoch": 86.11,
      "learning_rate": 8.518518518518519e-06,
      "loss": 0.0004,
      "step": 4650
    },
    {
      "epoch": 86.3,
      "learning_rate": 8.493827160493828e-06,
      "loss": 0.0004,
      "step": 4660
    },
    {
      "epoch": 86.48,
      "learning_rate": 8.469135802469137e-06,
      "loss": 0.0005,
      "step": 4670
    },
    {
      "epoch": 86.67,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.0005,
      "step": 4680
    },
    {
      "epoch": 86.85,
      "learning_rate": 8.419753086419754e-06,
      "loss": 0.0005,
      "step": 4690
    },
    {
      "epoch": 87.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0005323333316482604,
      "eval_runtime": 3.0578,
      "eval_samples_per_second": 24.855,
      "eval_steps_per_second": 3.27,
      "step": 4698
    },
    {
      "epoch": 87.04,
      "learning_rate": 8.395061728395062e-06,
      "loss": 0.0007,
      "step": 4700
    },
    {
      "epoch": 87.22,
      "learning_rate": 8.37037037037037e-06,
      "loss": 0.0006,
      "step": 4710
    },
    {
      "epoch": 87.41,
      "learning_rate": 8.34567901234568e-06,
      "loss": 0.0004,
      "step": 4720
    },
    {
      "epoch": 87.59,
      "learning_rate": 8.320987654320988e-06,
      "loss": 0.0553,
      "step": 4730
    },
    {
      "epoch": 87.78,
      "learning_rate": 8.296296296296297e-06,
      "loss": 0.0005,
      "step": 4740
    },
    {
      "epoch": 87.96,
      "learning_rate": 8.271604938271606e-06,
      "loss": 0.0004,
      "step": 4750
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0004464307858143002,
      "eval_runtime": 3.0308,
      "eval_samples_per_second": 25.076,
      "eval_steps_per_second": 3.299,
      "step": 4752
    },
    {
      "epoch": 88.15,
      "learning_rate": 8.246913580246915e-06,
      "loss": 0.0004,
      "step": 4760
    },
    {
      "epoch": 88.33,
      "learning_rate": 8.222222222222222e-06,
      "loss": 0.0004,
      "step": 4770
    },
    {
      "epoch": 88.52,
      "learning_rate": 8.197530864197532e-06,
      "loss": 0.0004,
      "step": 4780
    },
    {
      "epoch": 88.7,
      "learning_rate": 8.172839506172841e-06,
      "loss": 0.0004,
      "step": 4790
    },
    {
      "epoch": 88.89,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.0004,
      "step": 4800
    },
    {
      "epoch": 89.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0004430774424690753,
      "eval_runtime": 3.0165,
      "eval_samples_per_second": 25.195,
      "eval_steps_per_second": 3.315,
      "step": 4806
    },
    {
      "epoch": 89.07,
      "learning_rate": 8.123456790123457e-06,
      "loss": 0.0004,
      "step": 4810
    },
    {
      "epoch": 89.26,
      "learning_rate": 8.098765432098766e-06,
      "loss": 0.0004,
      "step": 4820
    },
    {
      "epoch": 89.44,
      "learning_rate": 8.074074074074075e-06,
      "loss": 0.0609,
      "step": 4830
    },
    {
      "epoch": 89.63,
      "learning_rate": 8.049382716049384e-06,
      "loss": 0.0007,
      "step": 4840
    },
    {
      "epoch": 89.81,
      "learning_rate": 8.024691358024692e-06,
      "loss": 0.0004,
      "step": 4850
    },
    {
      "epoch": 90.0,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0005,
      "step": 4860
    },
    {
      "epoch": 90.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00045739198685623705,
      "eval_runtime": 3.0947,
      "eval_samples_per_second": 24.558,
      "eval_steps_per_second": 3.231,
      "step": 4860
    },
    {
      "epoch": 90.19,
      "learning_rate": 7.975308641975308e-06,
      "loss": 0.0004,
      "step": 4870
    },
    {
      "epoch": 90.37,
      "learning_rate": 7.950617283950617e-06,
      "loss": 0.0564,
      "step": 4880
    },
    {
      "epoch": 90.56,
      "learning_rate": 7.925925925925926e-06,
      "loss": 0.0004,
      "step": 4890
    },
    {
      "epoch": 90.74,
      "learning_rate": 7.901234567901235e-06,
      "loss": 0.0004,
      "step": 4900
    },
    {
      "epoch": 90.93,
      "learning_rate": 7.876543209876544e-06,
      "loss": 0.0004,
      "step": 4910
    },
    {
      "epoch": 91.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0004953287425450981,
      "eval_runtime": 3.047,
      "eval_samples_per_second": 24.942,
      "eval_steps_per_second": 3.282,
      "step": 4914
    },
    {
      "epoch": 91.11,
      "learning_rate": 7.851851851851853e-06,
      "loss": 0.0009,
      "step": 4920
    },
    {
      "epoch": 91.3,
      "learning_rate": 7.82716049382716e-06,
      "loss": 0.0004,
      "step": 4930
    },
    {
      "epoch": 91.48,
      "learning_rate": 7.80246913580247e-06,
      "loss": 0.0004,
      "step": 4940
    },
    {
      "epoch": 91.67,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.0004,
      "step": 4950
    },
    {
      "epoch": 91.85,
      "learning_rate": 7.753086419753088e-06,
      "loss": 0.0067,
      "step": 4960
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00039551357622258365,
      "eval_runtime": 2.9279,
      "eval_samples_per_second": 25.957,
      "eval_steps_per_second": 3.415,
      "step": 4968
    },
    {
      "epoch": 92.04,
      "learning_rate": 7.728395061728395e-06,
      "loss": 0.0008,
      "step": 4970
    },
    {
      "epoch": 92.22,
      "learning_rate": 7.703703703703704e-06,
      "loss": 0.0004,
      "step": 4980
    },
    {
      "epoch": 92.41,
      "learning_rate": 7.679012345679013e-06,
      "loss": 0.0004,
      "step": 4990
    },
    {
      "epoch": 92.59,
      "learning_rate": 7.654320987654322e-06,
      "loss": 0.0004,
      "step": 5000
    },
    {
      "epoch": 92.78,
      "learning_rate": 7.62962962962963e-06,
      "loss": 0.0004,
      "step": 5010
    },
    {
      "epoch": 92.96,
      "learning_rate": 7.604938271604939e-06,
      "loss": 0.0004,
      "step": 5020
    },
    {
      "epoch": 93.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0003839218115899712,
      "eval_runtime": 3.0585,
      "eval_samples_per_second": 24.849,
      "eval_steps_per_second": 3.27,
      "step": 5022
    },
    {
      "epoch": 93.15,
      "learning_rate": 7.580246913580247e-06,
      "loss": 0.0004,
      "step": 5030
    },
    {
      "epoch": 93.33,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.0433,
      "step": 5040
    },
    {
      "epoch": 93.52,
      "learning_rate": 7.530864197530865e-06,
      "loss": 0.0092,
      "step": 5050
    },
    {
      "epoch": 93.7,
      "learning_rate": 7.506172839506174e-06,
      "loss": 0.0004,
      "step": 5060
    },
    {
      "epoch": 93.89,
      "learning_rate": 7.481481481481482e-06,
      "loss": 0.0004,
      "step": 5070
    },
    {
      "epoch": 94.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0003877544077113271,
      "eval_runtime": 3.07,
      "eval_samples_per_second": 24.756,
      "eval_steps_per_second": 3.257,
      "step": 5076
    },
    {
      "epoch": 94.07,
      "learning_rate": 7.456790123456791e-06,
      "loss": 0.0004,
      "step": 5080
    },
    {
      "epoch": 94.26,
      "learning_rate": 7.4320987654320995e-06,
      "loss": 0.0004,
      "step": 5090
    },
    {
      "epoch": 94.44,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.0004,
      "step": 5100
    },
    {
      "epoch": 94.63,
      "learning_rate": 7.382716049382716e-06,
      "loss": 0.0004,
      "step": 5110
    },
    {
      "epoch": 94.81,
      "learning_rate": 7.358024691358025e-06,
      "loss": 0.0035,
      "step": 5120
    },
    {
      "epoch": 95.0,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.0004,
      "step": 5130
    },
    {
      "epoch": 95.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.000382541969884187,
      "eval_runtime": 3.0815,
      "eval_samples_per_second": 24.663,
      "eval_steps_per_second": 3.245,
      "step": 5130
    },
    {
      "epoch": 95.19,
      "learning_rate": 7.308641975308642e-06,
      "loss": 0.0004,
      "step": 5140
    },
    {
      "epoch": 95.37,
      "learning_rate": 7.283950617283952e-06,
      "loss": 0.0004,
      "step": 5150
    },
    {
      "epoch": 95.56,
      "learning_rate": 7.2592592592592605e-06,
      "loss": 0.0004,
      "step": 5160
    },
    {
      "epoch": 95.74,
      "learning_rate": 7.2345679012345685e-06,
      "loss": 0.0004,
      "step": 5170
    },
    {
      "epoch": 95.93,
      "learning_rate": 7.209876543209877e-06,
      "loss": 0.0004,
      "step": 5180
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00039759272476658225,
      "eval_runtime": 3.0665,
      "eval_samples_per_second": 24.784,
      "eval_steps_per_second": 3.261,
      "step": 5184
    },
    {
      "epoch": 96.11,
      "learning_rate": 7.185185185185186e-06,
      "loss": 0.0004,
      "step": 5190
    },
    {
      "epoch": 96.3,
      "learning_rate": 7.160493827160494e-06,
      "loss": 0.0004,
      "step": 5200
    },
    {
      "epoch": 96.48,
      "learning_rate": 7.135802469135803e-06,
      "loss": 0.0004,
      "step": 5210
    },
    {
      "epoch": 96.67,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.0003,
      "step": 5220
    },
    {
      "epoch": 96.85,
      "learning_rate": 7.08641975308642e-06,
      "loss": 0.0004,
      "step": 5230
    },
    {
      "epoch": 97.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00038783540367148817,
      "eval_runtime": 3.0953,
      "eval_samples_per_second": 24.553,
      "eval_steps_per_second": 3.231,
      "step": 5238
    },
    {
      "epoch": 97.04,
      "learning_rate": 7.061728395061729e-06,
      "loss": 0.0003,
      "step": 5240
    },
    {
      "epoch": 97.22,
      "learning_rate": 7.0370370370370375e-06,
      "loss": 0.0003,
      "step": 5250
    },
    {
      "epoch": 97.41,
      "learning_rate": 7.012345679012347e-06,
      "loss": 0.0004,
      "step": 5260
    },
    {
      "epoch": 97.59,
      "learning_rate": 6.987654320987654e-06,
      "loss": 0.0004,
      "step": 5270
    },
    {
      "epoch": 97.78,
      "learning_rate": 6.962962962962964e-06,
      "loss": 0.0004,
      "step": 5280
    },
    {
      "epoch": 97.96,
      "learning_rate": 6.938271604938273e-06,
      "loss": 0.0004,
      "step": 5290
    },
    {
      "epoch": 98.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0003907608625013381,
      "eval_runtime": 3.0948,
      "eval_samples_per_second": 24.557,
      "eval_steps_per_second": 3.231,
      "step": 5292
    },
    {
      "epoch": 98.15,
      "learning_rate": 6.913580246913581e-06,
      "loss": 0.0003,
      "step": 5300
    },
    {
      "epoch": 98.33,
      "learning_rate": 6.88888888888889e-06,
      "loss": 0.0003,
      "step": 5310
    },
    {
      "epoch": 98.52,
      "learning_rate": 6.8641975308641985e-06,
      "loss": 0.0004,
      "step": 5320
    },
    {
      "epoch": 98.7,
      "learning_rate": 6.8395061728395065e-06,
      "loss": 0.0003,
      "step": 5330
    },
    {
      "epoch": 98.89,
      "learning_rate": 6.814814814814815e-06,
      "loss": 0.0003,
      "step": 5340
    },
    {
      "epoch": 99.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0004004727234132588,
      "eval_runtime": 3.0696,
      "eval_samples_per_second": 24.759,
      "eval_steps_per_second": 3.258,
      "step": 5346
    },
    {
      "epoch": 99.07,
      "learning_rate": 6.790123456790124e-06,
      "loss": 0.0004,
      "step": 5350
    },
    {
      "epoch": 99.26,
      "learning_rate": 6.765432098765433e-06,
      "loss": 0.0003,
      "step": 5360
    },
    {
      "epoch": 99.44,
      "learning_rate": 6.740740740740741e-06,
      "loss": 0.0003,
      "step": 5370
    },
    {
      "epoch": 99.63,
      "learning_rate": 6.71604938271605e-06,
      "loss": 0.0003,
      "step": 5380
    },
    {
      "epoch": 99.81,
      "learning_rate": 6.691358024691359e-06,
      "loss": 0.0017,
      "step": 5390
    },
    {
      "epoch": 100.0,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0003,
      "step": 5400
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00034208386205136776,
      "eval_runtime": 3.0385,
      "eval_samples_per_second": 25.012,
      "eval_steps_per_second": 3.291,
      "step": 5400
    },
    {
      "epoch": 100.19,
      "learning_rate": 6.6419753086419755e-06,
      "loss": 0.0003,
      "step": 5410
    },
    {
      "epoch": 100.37,
      "learning_rate": 6.617283950617285e-06,
      "loss": 0.0003,
      "step": 5420
    },
    {
      "epoch": 100.56,
      "learning_rate": 6.592592592592592e-06,
      "loss": 0.0003,
      "step": 5430
    },
    {
      "epoch": 100.74,
      "learning_rate": 6.567901234567902e-06,
      "loss": 0.0034,
      "step": 5440
    },
    {
      "epoch": 100.93,
      "learning_rate": 6.543209876543211e-06,
      "loss": 0.0003,
      "step": 5450
    },
    {
      "epoch": 101.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00037925393553450704,
      "eval_runtime": 2.9647,
      "eval_samples_per_second": 25.635,
      "eval_steps_per_second": 3.373,
      "step": 5454
    },
    {
      "epoch": 101.11,
      "learning_rate": 6.51851851851852e-06,
      "loss": 0.0003,
      "step": 5460
    },
    {
      "epoch": 101.3,
      "learning_rate": 6.493827160493828e-06,
      "loss": 0.0922,
      "step": 5470
    },
    {
      "epoch": 101.48,
      "learning_rate": 6.4691358024691365e-06,
      "loss": 0.0003,
      "step": 5480
    },
    {
      "epoch": 101.67,
      "learning_rate": 6.444444444444445e-06,
      "loss": 0.0003,
      "step": 5490
    },
    {
      "epoch": 101.85,
      "learning_rate": 6.419753086419753e-06,
      "loss": 0.0004,
      "step": 5500
    },
    {
      "epoch": 102.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00046697567449882627,
      "eval_runtime": 2.9966,
      "eval_samples_per_second": 25.362,
      "eval_steps_per_second": 3.337,
      "step": 5508
    },
    {
      "epoch": 102.04,
      "learning_rate": 6.395061728395062e-06,
      "loss": 0.0004,
      "step": 5510
    },
    {
      "epoch": 102.22,
      "learning_rate": 6.370370370370371e-06,
      "loss": 0.0003,
      "step": 5520
    },
    {
      "epoch": 102.41,
      "learning_rate": 6.345679012345679e-06,
      "loss": 0.0003,
      "step": 5530
    },
    {
      "epoch": 102.59,
      "learning_rate": 6.320987654320988e-06,
      "loss": 0.0003,
      "step": 5540
    },
    {
      "epoch": 102.78,
      "learning_rate": 6.296296296296297e-06,
      "loss": 0.0003,
      "step": 5550
    },
    {
      "epoch": 102.96,
      "learning_rate": 6.271604938271606e-06,
      "loss": 0.0004,
      "step": 5560
    },
    {
      "epoch": 103.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00045954357483424246,
      "eval_runtime": 2.9804,
      "eval_samples_per_second": 25.5,
      "eval_steps_per_second": 3.355,
      "step": 5562
    },
    {
      "epoch": 103.15,
      "learning_rate": 6.2469135802469135e-06,
      "loss": 0.0003,
      "step": 5570
    },
    {
      "epoch": 103.33,
      "learning_rate": 6.222222222222223e-06,
      "loss": 0.0003,
      "step": 5580
    },
    {
      "epoch": 103.52,
      "learning_rate": 6.197530864197532e-06,
      "loss": 0.0003,
      "step": 5590
    },
    {
      "epoch": 103.7,
      "learning_rate": 6.17283950617284e-06,
      "loss": 0.0003,
      "step": 5600
    },
    {
      "epoch": 103.89,
      "learning_rate": 6.148148148148149e-06,
      "loss": 0.0004,
      "step": 5610
    },
    {
      "epoch": 104.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0004455417220015079,
      "eval_runtime": 3.0351,
      "eval_samples_per_second": 25.04,
      "eval_steps_per_second": 3.295,
      "step": 5616
    },
    {
      "epoch": 104.07,
      "learning_rate": 6.123456790123458e-06,
      "loss": 0.0003,
      "step": 5620
    },
    {
      "epoch": 104.26,
      "learning_rate": 6.098765432098766e-06,
      "loss": 0.0003,
      "step": 5630
    },
    {
      "epoch": 104.44,
      "learning_rate": 6.0740740740740745e-06,
      "loss": 0.0241,
      "step": 5640
    },
    {
      "epoch": 104.63,
      "learning_rate": 6.049382716049383e-06,
      "loss": 0.0003,
      "step": 5650
    },
    {
      "epoch": 104.81,
      "learning_rate": 6.024691358024692e-06,
      "loss": 0.0003,
      "step": 5660
    },
    {
      "epoch": 105.0,
      "learning_rate": 6e-06,
      "loss": 0.0006,
      "step": 5670
    },
    {
      "epoch": 105.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0003152435238007456,
      "eval_runtime": 3.0259,
      "eval_samples_per_second": 25.117,
      "eval_steps_per_second": 3.305,
      "step": 5670
    },
    {
      "epoch": 105.19,
      "learning_rate": 5.975308641975309e-06,
      "loss": 0.0003,
      "step": 5680
    },
    {
      "epoch": 105.37,
      "learning_rate": 5.950617283950618e-06,
      "loss": 0.0003,
      "step": 5690
    },
    {
      "epoch": 105.56,
      "learning_rate": 5.925925925925926e-06,
      "loss": 0.0003,
      "step": 5700
    },
    {
      "epoch": 105.74,
      "learning_rate": 5.901234567901235e-06,
      "loss": 0.0003,
      "step": 5710
    },
    {
      "epoch": 105.93,
      "learning_rate": 5.876543209876544e-06,
      "loss": 0.0005,
      "step": 5720
    },
    {
      "epoch": 106.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00031142152147367597,
      "eval_runtime": 3.045,
      "eval_samples_per_second": 24.959,
      "eval_steps_per_second": 3.284,
      "step": 5724
    },
    {
      "epoch": 106.11,
      "learning_rate": 5.8518518518518515e-06,
      "loss": 0.0003,
      "step": 5730
    },
    {
      "epoch": 106.3,
      "learning_rate": 5.827160493827161e-06,
      "loss": 0.0003,
      "step": 5740
    },
    {
      "epoch": 106.48,
      "learning_rate": 5.80246913580247e-06,
      "loss": 0.0003,
      "step": 5750
    },
    {
      "epoch": 106.67,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.0032,
      "step": 5760
    },
    {
      "epoch": 106.85,
      "learning_rate": 5.753086419753087e-06,
      "loss": 0.0003,
      "step": 5770
    },
    {
      "epoch": 107.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0003071220126003027,
      "eval_runtime": 3.0131,
      "eval_samples_per_second": 25.224,
      "eval_steps_per_second": 3.319,
      "step": 5778
    },
    {
      "epoch": 107.04,
      "learning_rate": 5.728395061728396e-06,
      "loss": 0.0003,
      "step": 5780
    },
    {
      "epoch": 107.22,
      "learning_rate": 5.7037037037037045e-06,
      "loss": 0.0033,
      "step": 5790
    },
    {
      "epoch": 107.41,
      "learning_rate": 5.6790123456790125e-06,
      "loss": 0.0003,
      "step": 5800
    },
    {
      "epoch": 107.59,
      "learning_rate": 5.654320987654321e-06,
      "loss": 0.0003,
      "step": 5810
    },
    {
      "epoch": 107.78,
      "learning_rate": 5.62962962962963e-06,
      "loss": 0.0003,
      "step": 5820
    },
    {
      "epoch": 107.96,
      "learning_rate": 5.604938271604938e-06,
      "loss": 0.0003,
      "step": 5830
    },
    {
      "epoch": 108.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002943011641036719,
      "eval_runtime": 3.044,
      "eval_samples_per_second": 24.967,
      "eval_steps_per_second": 3.285,
      "step": 5832
    },
    {
      "epoch": 108.15,
      "learning_rate": 5.580246913580247e-06,
      "loss": 0.0003,
      "step": 5840
    },
    {
      "epoch": 108.33,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.0004,
      "step": 5850
    },
    {
      "epoch": 108.52,
      "learning_rate": 5.530864197530864e-06,
      "loss": 0.0003,
      "step": 5860
    },
    {
      "epoch": 108.7,
      "learning_rate": 5.5061728395061735e-06,
      "loss": 0.0003,
      "step": 5870
    },
    {
      "epoch": 108.89,
      "learning_rate": 5.481481481481482e-06,
      "loss": 0.0003,
      "step": 5880
    },
    {
      "epoch": 109.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00028869544621557,
      "eval_runtime": 3.0963,
      "eval_samples_per_second": 24.546,
      "eval_steps_per_second": 3.23,
      "step": 5886
    },
    {
      "epoch": 109.07,
      "learning_rate": 5.456790123456791e-06,
      "loss": 0.0003,
      "step": 5890
    },
    {
      "epoch": 109.26,
      "learning_rate": 5.432098765432099e-06,
      "loss": 0.0003,
      "step": 5900
    },
    {
      "epoch": 109.44,
      "learning_rate": 5.407407407407408e-06,
      "loss": 0.0003,
      "step": 5910
    },
    {
      "epoch": 109.63,
      "learning_rate": 5.382716049382717e-06,
      "loss": 0.0003,
      "step": 5920
    },
    {
      "epoch": 109.81,
      "learning_rate": 5.358024691358025e-06,
      "loss": 0.0003,
      "step": 5930
    },
    {
      "epoch": 110.0,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0003,
      "step": 5940
    },
    {
      "epoch": 110.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002840006782207638,
      "eval_runtime": 2.9775,
      "eval_samples_per_second": 25.525,
      "eval_steps_per_second": 3.359,
      "step": 5940
    },
    {
      "epoch": 110.19,
      "learning_rate": 5.3086419753086425e-06,
      "loss": 0.0004,
      "step": 5950
    },
    {
      "epoch": 110.37,
      "learning_rate": 5.2839506172839505e-06,
      "loss": 0.0003,
      "step": 5960
    },
    {
      "epoch": 110.56,
      "learning_rate": 5.259259259259259e-06,
      "loss": 0.0003,
      "step": 5970
    },
    {
      "epoch": 110.74,
      "learning_rate": 5.234567901234568e-06,
      "loss": 0.0003,
      "step": 5980
    },
    {
      "epoch": 110.93,
      "learning_rate": 5.209876543209878e-06,
      "loss": 0.0003,
      "step": 5990
    },
    {
      "epoch": 111.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002830446173902601,
      "eval_runtime": 3.079,
      "eval_samples_per_second": 24.683,
      "eval_steps_per_second": 3.248,
      "step": 5994
    },
    {
      "epoch": 111.11,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.0003,
      "step": 6000
    },
    {
      "epoch": 111.3,
      "learning_rate": 5.160493827160495e-06,
      "loss": 0.0003,
      "step": 6010
    },
    {
      "epoch": 111.48,
      "learning_rate": 5.1358024691358035e-06,
      "loss": 0.0016,
      "step": 6020
    },
    {
      "epoch": 111.67,
      "learning_rate": 5.1111111111111115e-06,
      "loss": 0.0072,
      "step": 6030
    },
    {
      "epoch": 111.85,
      "learning_rate": 5.08641975308642e-06,
      "loss": 0.0003,
      "step": 6040
    },
    {
      "epoch": 112.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00027556964778341353,
      "eval_runtime": 3.0731,
      "eval_samples_per_second": 24.731,
      "eval_steps_per_second": 3.254,
      "step": 6048
    },
    {
      "epoch": 112.04,
      "learning_rate": 5.061728395061729e-06,
      "loss": 0.0005,
      "step": 6050
    },
    {
      "epoch": 112.22,
      "learning_rate": 5.037037037037037e-06,
      "loss": 0.0003,
      "step": 6060
    },
    {
      "epoch": 112.41,
      "learning_rate": 5.012345679012346e-06,
      "loss": 0.0003,
      "step": 6070
    },
    {
      "epoch": 112.59,
      "learning_rate": 4.987654320987655e-06,
      "loss": 0.0003,
      "step": 6080
    },
    {
      "epoch": 112.78,
      "learning_rate": 4.962962962962964e-06,
      "loss": 0.0003,
      "step": 6090
    },
    {
      "epoch": 112.96,
      "learning_rate": 4.938271604938272e-06,
      "loss": 0.0003,
      "step": 6100
    },
    {
      "epoch": 113.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00026999213150702417,
      "eval_runtime": 3.0277,
      "eval_samples_per_second": 25.102,
      "eval_steps_per_second": 3.303,
      "step": 6102
    },
    {
      "epoch": 113.15,
      "learning_rate": 4.9135802469135805e-06,
      "loss": 0.0003,
      "step": 6110
    },
    {
      "epoch": 113.33,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.0003,
      "step": 6120
    },
    {
      "epoch": 113.52,
      "learning_rate": 4.864197530864198e-06,
      "loss": 0.0003,
      "step": 6130
    },
    {
      "epoch": 113.7,
      "learning_rate": 4.839506172839506e-06,
      "loss": 0.0728,
      "step": 6140
    },
    {
      "epoch": 113.89,
      "learning_rate": 4.814814814814815e-06,
      "loss": 0.0003,
      "step": 6150
    },
    {
      "epoch": 114.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00028843365726061165,
      "eval_runtime": 2.9883,
      "eval_samples_per_second": 25.433,
      "eval_steps_per_second": 3.346,
      "step": 6156
    },
    {
      "epoch": 114.07,
      "learning_rate": 4.790123456790124e-06,
      "loss": 0.0003,
      "step": 6160
    },
    {
      "epoch": 114.26,
      "learning_rate": 4.765432098765433e-06,
      "loss": 0.0003,
      "step": 6170
    },
    {
      "epoch": 114.44,
      "learning_rate": 4.7407407407407415e-06,
      "loss": 0.0003,
      "step": 6180
    },
    {
      "epoch": 114.63,
      "learning_rate": 4.7160493827160495e-06,
      "loss": 0.0003,
      "step": 6190
    },
    {
      "epoch": 114.81,
      "learning_rate": 4.691358024691358e-06,
      "loss": 0.0003,
      "step": 6200
    },
    {
      "epoch": 115.0,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.0003,
      "step": 6210
    },
    {
      "epoch": 115.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002904118155129254,
      "eval_runtime": 2.9961,
      "eval_samples_per_second": 25.367,
      "eval_steps_per_second": 3.338,
      "step": 6210
    },
    {
      "epoch": 115.19,
      "learning_rate": 4.641975308641975e-06,
      "loss": 0.0003,
      "step": 6220
    },
    {
      "epoch": 115.37,
      "learning_rate": 4.617283950617285e-06,
      "loss": 0.0003,
      "step": 6230
    },
    {
      "epoch": 115.56,
      "learning_rate": 4.592592592592593e-06,
      "loss": 0.0003,
      "step": 6240
    },
    {
      "epoch": 115.74,
      "learning_rate": 4.567901234567902e-06,
      "loss": 0.0003,
      "step": 6250
    },
    {
      "epoch": 115.93,
      "learning_rate": 4.5432098765432105e-06,
      "loss": 0.0003,
      "step": 6260
    },
    {
      "epoch": 116.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002848489675670862,
      "eval_runtime": 2.9812,
      "eval_samples_per_second": 25.493,
      "eval_steps_per_second": 3.354,
      "step": 6264
    },
    {
      "epoch": 116.11,
      "learning_rate": 4.5185185185185185e-06,
      "loss": 0.0003,
      "step": 6270
    },
    {
      "epoch": 116.3,
      "learning_rate": 4.493827160493827e-06,
      "loss": 0.0012,
      "step": 6280
    },
    {
      "epoch": 116.48,
      "learning_rate": 4.469135802469136e-06,
      "loss": 0.0003,
      "step": 6290
    },
    {
      "epoch": 116.67,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.0003,
      "step": 6300
    },
    {
      "epoch": 116.85,
      "learning_rate": 4.419753086419754e-06,
      "loss": 0.0003,
      "step": 6310
    },
    {
      "epoch": 117.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00026769962278194726,
      "eval_runtime": 2.9936,
      "eval_samples_per_second": 25.388,
      "eval_steps_per_second": 3.34,
      "step": 6318
    },
    {
      "epoch": 117.04,
      "learning_rate": 4.395061728395062e-06,
      "loss": 0.0003,
      "step": 6320
    },
    {
      "epoch": 117.22,
      "learning_rate": 4.370370370370371e-06,
      "loss": 0.0002,
      "step": 6330
    },
    {
      "epoch": 117.41,
      "learning_rate": 4.3456790123456795e-06,
      "loss": 0.0003,
      "step": 6340
    },
    {
      "epoch": 117.59,
      "learning_rate": 4.3209876543209875e-06,
      "loss": 0.0002,
      "step": 6350
    },
    {
      "epoch": 117.78,
      "learning_rate": 4.296296296296296e-06,
      "loss": 0.0003,
      "step": 6360
    },
    {
      "epoch": 117.96,
      "learning_rate": 4.271604938271605e-06,
      "loss": 0.0003,
      "step": 6370
    },
    {
      "epoch": 118.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00026247106143273413,
      "eval_runtime": 3.0703,
      "eval_samples_per_second": 24.754,
      "eval_steps_per_second": 3.257,
      "step": 6372
    },
    {
      "epoch": 118.15,
      "learning_rate": 4.246913580246914e-06,
      "loss": 0.0002,
      "step": 6380
    },
    {
      "epoch": 118.33,
      "learning_rate": 4.222222222222223e-06,
      "loss": 0.0493,
      "step": 6390
    },
    {
      "epoch": 118.52,
      "learning_rate": 4.197530864197531e-06,
      "loss": 0.0003,
      "step": 6400
    },
    {
      "epoch": 118.7,
      "learning_rate": 4.17283950617284e-06,
      "loss": 0.0003,
      "step": 6410
    },
    {
      "epoch": 118.89,
      "learning_rate": 4.1481481481481485e-06,
      "loss": 0.0002,
      "step": 6420
    },
    {
      "epoch": 119.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00024906551698222756,
      "eval_runtime": 3.0184,
      "eval_samples_per_second": 25.179,
      "eval_steps_per_second": 3.313,
      "step": 6426
    },
    {
      "epoch": 119.07,
      "learning_rate": 4.123456790123457e-06,
      "loss": 0.0003,
      "step": 6430
    },
    {
      "epoch": 119.26,
      "learning_rate": 4.098765432098766e-06,
      "loss": 0.0804,
      "step": 6440
    },
    {
      "epoch": 119.44,
      "learning_rate": 4.074074074074074e-06,
      "loss": 0.0002,
      "step": 6450
    },
    {
      "epoch": 119.63,
      "learning_rate": 4.049382716049383e-06,
      "loss": 0.0005,
      "step": 6460
    },
    {
      "epoch": 119.81,
      "learning_rate": 4.024691358024692e-06,
      "loss": 0.0003,
      "step": 6470
    },
    {
      "epoch": 120.0,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0002,
      "step": 6480
    },
    {
      "epoch": 120.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002486244193278253,
      "eval_runtime": 2.9769,
      "eval_samples_per_second": 25.53,
      "eval_steps_per_second": 3.359,
      "step": 6480
    },
    {
      "epoch": 120.19,
      "learning_rate": 3.975308641975309e-06,
      "loss": 0.0002,
      "step": 6490
    },
    {
      "epoch": 120.37,
      "learning_rate": 3.9506172839506175e-06,
      "loss": 0.0248,
      "step": 6500
    },
    {
      "epoch": 120.56,
      "learning_rate": 3.925925925925926e-06,
      "loss": 0.0018,
      "step": 6510
    },
    {
      "epoch": 120.74,
      "learning_rate": 3.901234567901235e-06,
      "loss": 0.0002,
      "step": 6520
    },
    {
      "epoch": 120.93,
      "learning_rate": 3.876543209876544e-06,
      "loss": 0.0002,
      "step": 6530
    },
    {
      "epoch": 121.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00024542087339796126,
      "eval_runtime": 3.0455,
      "eval_samples_per_second": 24.955,
      "eval_steps_per_second": 3.283,
      "step": 6534
    },
    {
      "epoch": 121.11,
      "learning_rate": 3.851851851851852e-06,
      "loss": 0.0003,
      "step": 6540
    },
    {
      "epoch": 121.3,
      "learning_rate": 3.827160493827161e-06,
      "loss": 0.0002,
      "step": 6550
    },
    {
      "epoch": 121.48,
      "learning_rate": 3.8024691358024697e-06,
      "loss": 0.0002,
      "step": 6560
    },
    {
      "epoch": 121.67,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.0003,
      "step": 6570
    },
    {
      "epoch": 121.85,
      "learning_rate": 3.753086419753087e-06,
      "loss": 0.0003,
      "step": 6580
    },
    {
      "epoch": 122.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002419725788058713,
      "eval_runtime": 3.1297,
      "eval_samples_per_second": 24.283,
      "eval_steps_per_second": 3.195,
      "step": 6588
    },
    {
      "epoch": 122.04,
      "learning_rate": 3.7283950617283953e-06,
      "loss": 0.0002,
      "step": 6590
    },
    {
      "epoch": 122.22,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.0004,
      "step": 6600
    },
    {
      "epoch": 122.41,
      "learning_rate": 3.6790123456790126e-06,
      "loss": 0.0002,
      "step": 6610
    },
    {
      "epoch": 122.59,
      "learning_rate": 3.654320987654321e-06,
      "loss": 0.0002,
      "step": 6620
    },
    {
      "epoch": 122.78,
      "learning_rate": 3.6296296296296302e-06,
      "loss": 0.0221,
      "step": 6630
    },
    {
      "epoch": 122.96,
      "learning_rate": 3.6049382716049387e-06,
      "loss": 0.0002,
      "step": 6640
    },
    {
      "epoch": 123.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00023867662821430713,
      "eval_runtime": 3.0953,
      "eval_samples_per_second": 24.553,
      "eval_steps_per_second": 3.231,
      "step": 6642
    },
    {
      "epoch": 123.15,
      "learning_rate": 3.580246913580247e-06,
      "loss": 0.0002,
      "step": 6650
    },
    {
      "epoch": 123.33,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0002,
      "step": 6660
    },
    {
      "epoch": 123.52,
      "learning_rate": 3.5308641975308643e-06,
      "loss": 0.0002,
      "step": 6670
    },
    {
      "epoch": 123.7,
      "learning_rate": 3.5061728395061736e-06,
      "loss": 0.0002,
      "step": 6680
    },
    {
      "epoch": 123.89,
      "learning_rate": 3.481481481481482e-06,
      "loss": 0.0002,
      "step": 6690
    },
    {
      "epoch": 124.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00023590722412336618,
      "eval_runtime": 3.1057,
      "eval_samples_per_second": 24.471,
      "eval_steps_per_second": 3.22,
      "step": 6696
    },
    {
      "epoch": 124.07,
      "learning_rate": 3.4567901234567904e-06,
      "loss": 0.0002,
      "step": 6700
    },
    {
      "epoch": 124.26,
      "learning_rate": 3.4320987654320992e-06,
      "loss": 0.0002,
      "step": 6710
    },
    {
      "epoch": 124.44,
      "learning_rate": 3.4074074074074077e-06,
      "loss": 0.0014,
      "step": 6720
    },
    {
      "epoch": 124.63,
      "learning_rate": 3.3827160493827165e-06,
      "loss": 0.0003,
      "step": 6730
    },
    {
      "epoch": 124.81,
      "learning_rate": 3.358024691358025e-06,
      "loss": 0.0002,
      "step": 6740
    },
    {
      "epoch": 125.0,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0002,
      "step": 6750
    },
    {
      "epoch": 125.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00023197440896183252,
      "eval_runtime": 3.1053,
      "eval_samples_per_second": 24.474,
      "eval_steps_per_second": 3.22,
      "step": 6750
    },
    {
      "epoch": 125.19,
      "learning_rate": 3.3086419753086426e-06,
      "loss": 0.0002,
      "step": 6760
    },
    {
      "epoch": 125.37,
      "learning_rate": 3.283950617283951e-06,
      "loss": 0.0002,
      "step": 6770
    },
    {
      "epoch": 125.56,
      "learning_rate": 3.25925925925926e-06,
      "loss": 0.0002,
      "step": 6780
    },
    {
      "epoch": 125.74,
      "learning_rate": 3.2345679012345682e-06,
      "loss": 0.0003,
      "step": 6790
    },
    {
      "epoch": 125.93,
      "learning_rate": 3.2098765432098767e-06,
      "loss": 0.0002,
      "step": 6800
    },
    {
      "epoch": 126.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002292316930834204,
      "eval_runtime": 3.1458,
      "eval_samples_per_second": 24.159,
      "eval_steps_per_second": 3.179,
      "step": 6804
    },
    {
      "epoch": 126.11,
      "learning_rate": 3.1851851851851855e-06,
      "loss": 0.0002,
      "step": 6810
    },
    {
      "epoch": 126.3,
      "learning_rate": 3.160493827160494e-06,
      "loss": 0.0002,
      "step": 6820
    },
    {
      "epoch": 126.48,
      "learning_rate": 3.135802469135803e-06,
      "loss": 0.0003,
      "step": 6830
    },
    {
      "epoch": 126.67,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0002,
      "step": 6840
    },
    {
      "epoch": 126.85,
      "learning_rate": 3.08641975308642e-06,
      "loss": 0.0712,
      "step": 6850
    },
    {
      "epoch": 127.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00022830172383692116,
      "eval_runtime": 3.0547,
      "eval_samples_per_second": 24.879,
      "eval_steps_per_second": 3.274,
      "step": 6858
    },
    {
      "epoch": 127.04,
      "learning_rate": 3.061728395061729e-06,
      "loss": 0.0002,
      "step": 6860
    },
    {
      "epoch": 127.22,
      "learning_rate": 3.0370370370370372e-06,
      "loss": 0.0002,
      "step": 6870
    },
    {
      "epoch": 127.41,
      "learning_rate": 3.012345679012346e-06,
      "loss": 0.0002,
      "step": 6880
    },
    {
      "epoch": 127.59,
      "learning_rate": 2.9876543209876545e-06,
      "loss": 0.0002,
      "step": 6890
    },
    {
      "epoch": 127.78,
      "learning_rate": 2.962962962962963e-06,
      "loss": 0.0002,
      "step": 6900
    },
    {
      "epoch": 127.96,
      "learning_rate": 2.938271604938272e-06,
      "loss": 0.0002,
      "step": 6910
    },
    {
      "epoch": 128.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00022887364320922643,
      "eval_runtime": 3.0745,
      "eval_samples_per_second": 24.72,
      "eval_steps_per_second": 3.253,
      "step": 6912
    },
    {
      "epoch": 128.15,
      "learning_rate": 2.9135802469135806e-06,
      "loss": 0.0002,
      "step": 6920
    },
    {
      "epoch": 128.33,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.0002,
      "step": 6930
    },
    {
      "epoch": 128.52,
      "learning_rate": 2.864197530864198e-06,
      "loss": 0.0002,
      "step": 6940
    },
    {
      "epoch": 128.7,
      "learning_rate": 2.8395061728395062e-06,
      "loss": 0.0002,
      "step": 6950
    },
    {
      "epoch": 128.89,
      "learning_rate": 2.814814814814815e-06,
      "loss": 0.0002,
      "step": 6960
    },
    {
      "epoch": 129.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002265323855681345,
      "eval_runtime": 3.0539,
      "eval_samples_per_second": 24.886,
      "eval_steps_per_second": 3.274,
      "step": 6966
    },
    {
      "epoch": 129.07,
      "learning_rate": 2.7901234567901235e-06,
      "loss": 0.0002,
      "step": 6970
    },
    {
      "epoch": 129.26,
      "learning_rate": 2.765432098765432e-06,
      "loss": 0.0002,
      "step": 6980
    },
    {
      "epoch": 129.44,
      "learning_rate": 2.740740740740741e-06,
      "loss": 0.0002,
      "step": 6990
    },
    {
      "epoch": 129.63,
      "learning_rate": 2.7160493827160496e-06,
      "loss": 0.0002,
      "step": 7000
    },
    {
      "epoch": 129.81,
      "learning_rate": 2.6913580246913584e-06,
      "loss": 0.0002,
      "step": 7010
    },
    {
      "epoch": 130.0,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0002,
      "step": 7020
    },
    {
      "epoch": 130.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00022419424203690141,
      "eval_runtime": 3.1051,
      "eval_samples_per_second": 24.476,
      "eval_steps_per_second": 3.221,
      "step": 7020
    },
    {
      "epoch": 130.19,
      "learning_rate": 2.6419753086419752e-06,
      "loss": 0.081,
      "step": 7030
    },
    {
      "epoch": 130.37,
      "learning_rate": 2.617283950617284e-06,
      "loss": 0.0002,
      "step": 7040
    },
    {
      "epoch": 130.56,
      "learning_rate": 2.5925925925925925e-06,
      "loss": 0.0002,
      "step": 7050
    },
    {
      "epoch": 130.74,
      "learning_rate": 2.5679012345679018e-06,
      "loss": 0.0002,
      "step": 7060
    },
    {
      "epoch": 130.93,
      "learning_rate": 2.54320987654321e-06,
      "loss": 0.0002,
      "step": 7070
    },
    {
      "epoch": 131.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00022435725259128958,
      "eval_runtime": 3.0635,
      "eval_samples_per_second": 24.808,
      "eval_steps_per_second": 3.264,
      "step": 7074
    },
    {
      "epoch": 131.11,
      "learning_rate": 2.5185185185185186e-06,
      "loss": 0.0002,
      "step": 7080
    },
    {
      "epoch": 131.3,
      "learning_rate": 2.4938271604938274e-06,
      "loss": 0.0002,
      "step": 7090
    },
    {
      "epoch": 131.48,
      "learning_rate": 2.469135802469136e-06,
      "loss": 0.0002,
      "step": 7100
    },
    {
      "epoch": 131.67,
      "learning_rate": 2.4444444444444447e-06,
      "loss": 0.0002,
      "step": 7110
    },
    {
      "epoch": 131.85,
      "learning_rate": 2.419753086419753e-06,
      "loss": 0.0002,
      "step": 7120
    },
    {
      "epoch": 132.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00022238449309952557,
      "eval_runtime": 3.099,
      "eval_samples_per_second": 24.524,
      "eval_steps_per_second": 3.227,
      "step": 7128
    },
    {
      "epoch": 132.04,
      "learning_rate": 2.395061728395062e-06,
      "loss": 0.0002,
      "step": 7130
    },
    {
      "epoch": 132.22,
      "learning_rate": 2.3703703703703707e-06,
      "loss": 0.0002,
      "step": 7140
    },
    {
      "epoch": 132.41,
      "learning_rate": 2.345679012345679e-06,
      "loss": 0.0002,
      "step": 7150
    },
    {
      "epoch": 132.59,
      "learning_rate": 2.3209876543209876e-06,
      "loss": 0.0002,
      "step": 7160
    },
    {
      "epoch": 132.78,
      "learning_rate": 2.2962962962962964e-06,
      "loss": 0.0002,
      "step": 7170
    },
    {
      "epoch": 132.96,
      "learning_rate": 2.2716049382716052e-06,
      "loss": 0.0002,
      "step": 7180
    },
    {
      "epoch": 133.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00022056240413803607,
      "eval_runtime": 2.9993,
      "eval_samples_per_second": 25.34,
      "eval_steps_per_second": 3.334,
      "step": 7182
    },
    {
      "epoch": 133.15,
      "learning_rate": 2.2469135802469137e-06,
      "loss": 0.0002,
      "step": 7190
    },
    {
      "epoch": 133.33,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0002,
      "step": 7200
    },
    {
      "epoch": 133.52,
      "learning_rate": 2.197530864197531e-06,
      "loss": 0.0002,
      "step": 7210
    },
    {
      "epoch": 133.7,
      "learning_rate": 2.1728395061728397e-06,
      "loss": 0.0002,
      "step": 7220
    },
    {
      "epoch": 133.89,
      "learning_rate": 2.148148148148148e-06,
      "loss": 0.0002,
      "step": 7230
    },
    {
      "epoch": 134.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00021875585662201047,
      "eval_runtime": 3.0391,
      "eval_samples_per_second": 25.007,
      "eval_steps_per_second": 3.29,
      "step": 7236
    },
    {
      "epoch": 134.07,
      "learning_rate": 2.123456790123457e-06,
      "loss": 0.0002,
      "step": 7240
    },
    {
      "epoch": 134.26,
      "learning_rate": 2.0987654320987654e-06,
      "loss": 0.0002,
      "step": 7250
    },
    {
      "epoch": 134.44,
      "learning_rate": 2.0740740740740742e-06,
      "loss": 0.0002,
      "step": 7260
    },
    {
      "epoch": 134.63,
      "learning_rate": 2.049382716049383e-06,
      "loss": 0.0002,
      "step": 7270
    },
    {
      "epoch": 134.81,
      "learning_rate": 2.0246913580246915e-06,
      "loss": 0.0002,
      "step": 7280
    },
    {
      "epoch": 135.0,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0002,
      "step": 7290
    },
    {
      "epoch": 135.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002170340158045292,
      "eval_runtime": 3.0595,
      "eval_samples_per_second": 24.84,
      "eval_steps_per_second": 3.268,
      "step": 7290
    },
    {
      "epoch": 135.19,
      "learning_rate": 1.9753086419753087e-06,
      "loss": 0.0002,
      "step": 7300
    },
    {
      "epoch": 135.37,
      "learning_rate": 1.9506172839506176e-06,
      "loss": 0.0002,
      "step": 7310
    },
    {
      "epoch": 135.56,
      "learning_rate": 1.925925925925926e-06,
      "loss": 0.0002,
      "step": 7320
    },
    {
      "epoch": 135.74,
      "learning_rate": 1.9012345679012348e-06,
      "loss": 0.0002,
      "step": 7330
    },
    {
      "epoch": 135.93,
      "learning_rate": 1.8765432098765435e-06,
      "loss": 0.0003,
      "step": 7340
    },
    {
      "epoch": 136.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00021536705025937408,
      "eval_runtime": 3.0723,
      "eval_samples_per_second": 24.737,
      "eval_steps_per_second": 3.255,
      "step": 7344
    },
    {
      "epoch": 136.11,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.0002,
      "step": 7350
    },
    {
      "epoch": 136.3,
      "learning_rate": 1.8271604938271605e-06,
      "loss": 0.0002,
      "step": 7360
    },
    {
      "epoch": 136.48,
      "learning_rate": 1.8024691358024693e-06,
      "loss": 0.0002,
      "step": 7370
    },
    {
      "epoch": 136.67,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.0002,
      "step": 7380
    },
    {
      "epoch": 136.85,
      "learning_rate": 1.7530864197530868e-06,
      "loss": 0.0002,
      "step": 7390
    },
    {
      "epoch": 137.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00021387105516623706,
      "eval_runtime": 3.1056,
      "eval_samples_per_second": 24.472,
      "eval_steps_per_second": 3.22,
      "step": 7398
    },
    {
      "epoch": 137.04,
      "learning_rate": 1.7283950617283952e-06,
      "loss": 0.0002,
      "step": 7400
    },
    {
      "epoch": 137.22,
      "learning_rate": 1.7037037037037038e-06,
      "loss": 0.0002,
      "step": 7410
    },
    {
      "epoch": 137.41,
      "learning_rate": 1.6790123456790125e-06,
      "loss": 0.0002,
      "step": 7420
    },
    {
      "epoch": 137.59,
      "learning_rate": 1.6543209876543213e-06,
      "loss": 0.0002,
      "step": 7430
    },
    {
      "epoch": 137.78,
      "learning_rate": 1.62962962962963e-06,
      "loss": 0.0002,
      "step": 7440
    },
    {
      "epoch": 137.96,
      "learning_rate": 1.6049382716049383e-06,
      "loss": 0.0002,
      "step": 7450
    },
    {
      "epoch": 138.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00021247855329420418,
      "eval_runtime": 3.0565,
      "eval_samples_per_second": 24.865,
      "eval_steps_per_second": 3.272,
      "step": 7452
    },
    {
      "epoch": 138.15,
      "learning_rate": 1.580246913580247e-06,
      "loss": 0.0002,
      "step": 7460
    },
    {
      "epoch": 138.33,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 0.0002,
      "step": 7470
    },
    {
      "epoch": 138.52,
      "learning_rate": 1.5308641975308644e-06,
      "loss": 0.0002,
      "step": 7480
    },
    {
      "epoch": 138.7,
      "learning_rate": 1.506172839506173e-06,
      "loss": 0.0002,
      "step": 7490
    },
    {
      "epoch": 138.89,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.0028,
      "step": 7500
    },
    {
      "epoch": 139.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002113178779836744,
      "eval_runtime": 3.0392,
      "eval_samples_per_second": 25.007,
      "eval_steps_per_second": 3.29,
      "step": 7506
    },
    {
      "epoch": 139.07,
      "learning_rate": 1.4567901234567903e-06,
      "loss": 0.0002,
      "step": 7510
    },
    {
      "epoch": 139.26,
      "learning_rate": 1.432098765432099e-06,
      "loss": 0.0002,
      "step": 7520
    },
    {
      "epoch": 139.44,
      "learning_rate": 1.4074074074074075e-06,
      "loss": 0.0002,
      "step": 7530
    },
    {
      "epoch": 139.63,
      "learning_rate": 1.382716049382716e-06,
      "loss": 0.0002,
      "step": 7540
    },
    {
      "epoch": 139.81,
      "learning_rate": 1.3580246913580248e-06,
      "loss": 0.0002,
      "step": 7550
    },
    {
      "epoch": 140.0,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0006,
      "step": 7560
    },
    {
      "epoch": 140.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.000210436453926377,
      "eval_runtime": 3.0044,
      "eval_samples_per_second": 25.296,
      "eval_steps_per_second": 3.328,
      "step": 7560
    },
    {
      "epoch": 140.19,
      "learning_rate": 1.308641975308642e-06,
      "loss": 0.0002,
      "step": 7570
    },
    {
      "epoch": 140.37,
      "learning_rate": 1.2839506172839509e-06,
      "loss": 0.0002,
      "step": 7580
    },
    {
      "epoch": 140.56,
      "learning_rate": 1.2592592592592593e-06,
      "loss": 0.0002,
      "step": 7590
    },
    {
      "epoch": 140.74,
      "learning_rate": 1.234567901234568e-06,
      "loss": 0.0002,
      "step": 7600
    },
    {
      "epoch": 140.93,
      "learning_rate": 1.2098765432098765e-06,
      "loss": 0.0002,
      "step": 7610
    },
    {
      "epoch": 141.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020950206089764833,
      "eval_runtime": 3.0835,
      "eval_samples_per_second": 24.647,
      "eval_steps_per_second": 3.243,
      "step": 7614
    },
    {
      "epoch": 141.11,
      "learning_rate": 1.1851851851851854e-06,
      "loss": 0.0002,
      "step": 7620
    },
    {
      "epoch": 141.3,
      "learning_rate": 1.1604938271604938e-06,
      "loss": 0.0002,
      "step": 7630
    },
    {
      "epoch": 141.48,
      "learning_rate": 1.1358024691358026e-06,
      "loss": 0.0002,
      "step": 7640
    },
    {
      "epoch": 141.67,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.0002,
      "step": 7650
    },
    {
      "epoch": 141.85,
      "learning_rate": 1.0864197530864199e-06,
      "loss": 0.0002,
      "step": 7660
    },
    {
      "epoch": 142.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020854859030805528,
      "eval_runtime": 3.1029,
      "eval_samples_per_second": 24.493,
      "eval_steps_per_second": 3.223,
      "step": 7668
    },
    {
      "epoch": 142.04,
      "learning_rate": 1.0617283950617285e-06,
      "loss": 0.0002,
      "step": 7670
    },
    {
      "epoch": 142.22,
      "learning_rate": 1.0370370370370371e-06,
      "loss": 0.0002,
      "step": 7680
    },
    {
      "epoch": 142.41,
      "learning_rate": 1.0123456790123457e-06,
      "loss": 0.0002,
      "step": 7690
    },
    {
      "epoch": 142.59,
      "learning_rate": 9.876543209876544e-07,
      "loss": 0.0002,
      "step": 7700
    },
    {
      "epoch": 142.78,
      "learning_rate": 9.62962962962963e-07,
      "loss": 0.0002,
      "step": 7710
    },
    {
      "epoch": 142.96,
      "learning_rate": 9.382716049382717e-07,
      "loss": 0.0004,
      "step": 7720
    },
    {
      "epoch": 143.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020763746579177678,
      "eval_runtime": 3.038,
      "eval_samples_per_second": 25.017,
      "eval_steps_per_second": 3.292,
      "step": 7722
    },
    {
      "epoch": 143.15,
      "learning_rate": 9.135802469135802e-07,
      "loss": 0.0002,
      "step": 7730
    },
    {
      "epoch": 143.33,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0002,
      "step": 7740
    },
    {
      "epoch": 143.52,
      "learning_rate": 8.641975308641976e-07,
      "loss": 0.0002,
      "step": 7750
    },
    {
      "epoch": 143.7,
      "learning_rate": 8.395061728395062e-07,
      "loss": 0.0002,
      "step": 7760
    },
    {
      "epoch": 143.89,
      "learning_rate": 8.14814814814815e-07,
      "loss": 0.0002,
      "step": 7770
    },
    {
      "epoch": 144.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020678593136835843,
      "eval_runtime": 3.0151,
      "eval_samples_per_second": 25.206,
      "eval_steps_per_second": 3.317,
      "step": 7776
    },
    {
      "epoch": 144.07,
      "learning_rate": 7.901234567901235e-07,
      "loss": 0.0002,
      "step": 7780
    },
    {
      "epoch": 144.26,
      "learning_rate": 7.654320987654322e-07,
      "loss": 0.0002,
      "step": 7790
    },
    {
      "epoch": 144.44,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.0002,
      "step": 7800
    },
    {
      "epoch": 144.63,
      "learning_rate": 7.160493827160495e-07,
      "loss": 0.0002,
      "step": 7810
    },
    {
      "epoch": 144.81,
      "learning_rate": 6.91358024691358e-07,
      "loss": 0.0002,
      "step": 7820
    },
    {
      "epoch": 145.0,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.0002,
      "step": 7830
    },
    {
      "epoch": 145.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020612729713320732,
      "eval_runtime": 3.0352,
      "eval_samples_per_second": 25.04,
      "eval_steps_per_second": 3.295,
      "step": 7830
    },
    {
      "epoch": 145.19,
      "learning_rate": 6.419753086419754e-07,
      "loss": 0.0002,
      "step": 7840
    },
    {
      "epoch": 145.37,
      "learning_rate": 6.17283950617284e-07,
      "loss": 0.0002,
      "step": 7850
    },
    {
      "epoch": 145.56,
      "learning_rate": 5.925925925925927e-07,
      "loss": 0.0002,
      "step": 7860
    },
    {
      "epoch": 145.74,
      "learning_rate": 5.679012345679013e-07,
      "loss": 0.0002,
      "step": 7870
    },
    {
      "epoch": 145.93,
      "learning_rate": 5.432098765432099e-07,
      "loss": 0.1028,
      "step": 7880
    },
    {
      "epoch": 146.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020691766985692084,
      "eval_runtime": 2.9867,
      "eval_samples_per_second": 25.446,
      "eval_steps_per_second": 3.348,
      "step": 7884
    },
    {
      "epoch": 146.11,
      "learning_rate": 5.185185185185186e-07,
      "loss": 0.0002,
      "step": 7890
    },
    {
      "epoch": 146.3,
      "learning_rate": 4.938271604938272e-07,
      "loss": 0.0002,
      "step": 7900
    },
    {
      "epoch": 146.48,
      "learning_rate": 4.6913580246913586e-07,
      "loss": 0.0002,
      "step": 7910
    },
    {
      "epoch": 146.67,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0002,
      "step": 7920
    },
    {
      "epoch": 146.85,
      "learning_rate": 4.197530864197531e-07,
      "loss": 0.0002,
      "step": 7930
    },
    {
      "epoch": 147.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020708548254333436,
      "eval_runtime": 3.038,
      "eval_samples_per_second": 25.017,
      "eval_steps_per_second": 3.292,
      "step": 7938
    },
    {
      "epoch": 147.04,
      "learning_rate": 3.9506172839506174e-07,
      "loss": 0.0002,
      "step": 7940
    },
    {
      "epoch": 147.22,
      "learning_rate": 3.7037037037037036e-07,
      "loss": 0.0002,
      "step": 7950
    },
    {
      "epoch": 147.41,
      "learning_rate": 3.45679012345679e-07,
      "loss": 0.0002,
      "step": 7960
    },
    {
      "epoch": 147.59,
      "learning_rate": 3.209876543209877e-07,
      "loss": 0.0002,
      "step": 7970
    },
    {
      "epoch": 147.78,
      "learning_rate": 2.9629629629629634e-07,
      "loss": 0.0002,
      "step": 7980
    },
    {
      "epoch": 147.96,
      "learning_rate": 2.7160493827160497e-07,
      "loss": 0.0002,
      "step": 7990
    },
    {
      "epoch": 148.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0002069223701255396,
      "eval_runtime": 3.0851,
      "eval_samples_per_second": 24.635,
      "eval_steps_per_second": 3.241,
      "step": 7992
    },
    {
      "epoch": 148.15,
      "learning_rate": 2.469135802469136e-07,
      "loss": 0.0002,
      "step": 8000
    },
    {
      "epoch": 148.33,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.0002,
      "step": 8010
    },
    {
      "epoch": 148.52,
      "learning_rate": 1.9753086419753087e-07,
      "loss": 0.0002,
      "step": 8020
    },
    {
      "epoch": 148.7,
      "learning_rate": 1.728395061728395e-07,
      "loss": 0.0002,
      "step": 8030
    },
    {
      "epoch": 148.89,
      "learning_rate": 1.4814814814814817e-07,
      "loss": 0.0002,
      "step": 8040
    },
    {
      "epoch": 149.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020683615002781153,
      "eval_runtime": 3.076,
      "eval_samples_per_second": 24.708,
      "eval_steps_per_second": 3.251,
      "step": 8046
    },
    {
      "epoch": 149.07,
      "learning_rate": 1.234567901234568e-07,
      "loss": 0.0002,
      "step": 8050
    },
    {
      "epoch": 149.26,
      "learning_rate": 9.876543209876543e-08,
      "loss": 0.0002,
      "step": 8060
    },
    {
      "epoch": 149.44,
      "learning_rate": 7.407407407407409e-08,
      "loss": 0.0002,
      "step": 8070
    },
    {
      "epoch": 149.63,
      "learning_rate": 4.938271604938272e-08,
      "loss": 0.0002,
      "step": 8080
    },
    {
      "epoch": 149.81,
      "learning_rate": 2.469135802469136e-08,
      "loss": 0.0002,
      "step": 8090
    },
    {
      "epoch": 150.0,
      "learning_rate": 0.0,
      "loss": 0.0002,
      "step": 8100
    },
    {
      "epoch": 150.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.00020679850422311574,
      "eval_runtime": 3.0877,
      "eval_samples_per_second": 24.614,
      "eval_steps_per_second": 3.239,
      "step": 8100
    },
    {
      "epoch": 150.0,
      "step": 8100,
      "total_flos": 4.94011433793024e+18,
      "train_loss": 0.0,
      "train_runtime": 0.0622,
      "train_samples_per_second": 1025154.241,
      "train_steps_per_second": 130254.892
    }
  ],
  "logging_steps": 10,
  "max_steps": 8100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 150,
  "save_steps": 500,
  "total_flos": 4.94011433793024e+18,
  "trial_name": null,
  "trial_params": null
}