diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4725 @@
+{
+  "best_metric": 0.9316640584246219,
+  "best_model_checkpoint": "SL-CvT/checkpoint-5040",
+  "epoch": 100.0,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.4351,
+      "step": 10
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.4316,
+      "step": 20
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5e-06,
+      "loss": 1.3634,
+      "step": 30
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.3129,
+      "step": 40
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.2931,
+      "step": 50
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 1.2379,
+      "step": 60
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7245696400625978,
+      "eval_f1": 0.6421567406482025,
+      "eval_loss": 1.0716278553009033,
+      "eval_roc_auc": 0.7323469324820772,
+      "eval_runtime": 30.9342,
+      "eval_samples_per_second": 61.97,
+      "eval_steps_per_second": 1.94,
+      "step": 60
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.2069,
+      "step": 70
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.1671,
+      "step": 80
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.5e-06,
+      "loss": 1.1401,
+      "step": 90
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.1452,
+      "step": 100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.0799,
+      "step": 110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1e-05,
+      "loss": 1.0186,
+      "step": 120
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.729264475743349,
+      "eval_f1": 0.6425490906446382,
+      "eval_loss": 0.847660481929779,
+      "eval_roc_auc": 0.7879043691585872,
+      "eval_runtime": 30.8873,
+      "eval_samples_per_second": 62.064,
+      "eval_steps_per_second": 1.943,
+      "step": 120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.0281,
+      "step": 130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.9652,
+      "step": 140
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9686,
+      "step": 150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.9573,
+      "step": 160
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4166666666666668e-05,
+      "loss": 0.9334,
+      "step": 170
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.9433,
+      "step": 180
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.7537819509650495,
+      "eval_f1": 0.705960518775639,
+      "eval_loss": 0.7472825050354004,
+      "eval_roc_auc": 0.845372294497077,
+      "eval_runtime": 31.0251,
+      "eval_samples_per_second": 61.789,
+      "eval_steps_per_second": 1.934,
+      "step": 180
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.935,
+      "step": 190
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.8971,
+      "step": 200
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.75e-05,
+      "loss": 0.8269,
+      "step": 210
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9014,
+      "step": 220
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.9166666666666667e-05,
+      "loss": 0.8513,
+      "step": 230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8644,
+      "step": 240
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.766301512780386,
+      "eval_f1": 0.7188479081954189,
+      "eval_loss": 0.6830809712409973,
+      "eval_roc_auc": 0.8696485953997086,
+      "eval_runtime": 30.9674,
+      "eval_samples_per_second": 61.904,
+      "eval_steps_per_second": 1.938,
+      "step": 240
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.8384,
+      "step": 250
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.8517,
+      "step": 260
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.25e-05,
+      "loss": 0.7973,
+      "step": 270
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.769,
+      "step": 280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.4166666666666667e-05,
+      "loss": 0.8245,
+      "step": 290
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7985,
+      "step": 300
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7798643714136672,
+      "eval_f1": 0.7409260554438973,
+      "eval_loss": 0.641985297203064,
+      "eval_roc_auc": 0.8943028148426463,
+      "eval_runtime": 30.9292,
+      "eval_samples_per_second": 61.98,
+      "eval_steps_per_second": 1.94,
+      "step": 300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 0.7784,
+      "step": 310
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.8006,
+      "step": 320
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.7302,
+      "step": 330
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.8333333333333335e-05,
+      "loss": 0.7941,
+      "step": 340
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.7924,
+      "step": 350
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3e-05,
+      "loss": 0.7322,
+      "step": 360
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.8101199791340636,
+      "eval_f1": 0.7885994269404484,
+      "eval_loss": 0.5712574124336243,
+      "eval_roc_auc": 0.9195657365645341,
+      "eval_runtime": 30.9327,
+      "eval_samples_per_second": 61.973,
+      "eval_steps_per_second": 1.94,
+      "step": 360
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 0.7336,
+      "step": 370
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.663,
+      "step": 380
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.6591,
+      "step": 390
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.7874,
+      "step": 400
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 0.6922,
+      "step": 410
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.5e-05,
+      "loss": 0.725,
+      "step": 420
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.8189880020865936,
+      "eval_f1": 0.7989427228016346,
+      "eval_loss": 0.5310792922973633,
+      "eval_roc_auc": 0.9324356204838778,
+      "eval_runtime": 30.9897,
+      "eval_samples_per_second": 61.859,
+      "eval_steps_per_second": 1.936,
+      "step": 420
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 0.6874,
+      "step": 430
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.6595,
+      "step": 440
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.6524,
+      "step": 450
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.6605,
+      "step": 460
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 0.7416,
+      "step": 470
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4e-05,
+      "loss": 0.6529,
+      "step": 480
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.8116849243609807,
+      "eval_f1": 0.7851628741512076,
+      "eval_loss": 0.5246216654777527,
+      "eval_roc_auc": 0.9404167887041862,
+      "eval_runtime": 30.9255,
+      "eval_samples_per_second": 61.988,
+      "eval_steps_per_second": 1.94,
+      "step": 480
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.0833333333333334e-05,
+      "loss": 0.6527,
+      "step": 490
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.6512,
+      "step": 500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.25e-05,
+      "loss": 0.6601,
+      "step": 510
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.6674,
+      "step": 520
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.4166666666666665e-05,
+      "loss": 0.6243,
+      "step": 530
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6224,
+      "step": 540
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.8440271257172666,
+      "eval_f1": 0.8282011911330186,
+      "eval_loss": 0.4597983658313751,
+      "eval_roc_auc": 0.9516750726473929,
+      "eval_runtime": 30.9543,
+      "eval_samples_per_second": 61.93,
+      "eval_steps_per_second": 1.938,
+      "step": 540
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.6133,
+      "step": 550
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.6199,
+      "step": 560
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.75e-05,
+      "loss": 0.6523,
+      "step": 570
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.6189,
+      "step": 580
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.9166666666666665e-05,
+      "loss": 0.6186,
+      "step": 590
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5e-05,
+      "loss": 0.6315,
+      "step": 600
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.8528951486697965,
+      "eval_f1": 0.8456632918948623,
+      "eval_loss": 0.4362500011920929,
+      "eval_roc_auc": 0.9584596745900972,
+      "eval_runtime": 30.9898,
+      "eval_samples_per_second": 61.859,
+      "eval_steps_per_second": 1.936,
+      "step": 600
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9907407407407406e-05,
+      "loss": 0.596,
+      "step": 610
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.981481481481482e-05,
+      "loss": 0.5767,
+      "step": 620
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.972222222222223e-05,
+      "loss": 0.5942,
+      "step": 630
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.962962962962963e-05,
+      "loss": 0.5511,
+      "step": 640
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.9537037037037035e-05,
+      "loss": 0.5505,
+      "step": 650
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.9444444444444446e-05,
+      "loss": 0.5651,
+      "step": 660
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.8502869066249348,
+      "eval_f1": 0.8323302022733028,
+      "eval_loss": 0.44371965527534485,
+      "eval_roc_auc": 0.9564279884814689,
+      "eval_runtime": 30.9158,
+      "eval_samples_per_second": 62.007,
+      "eval_steps_per_second": 1.941,
+      "step": 660
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.935185185185186e-05,
+      "loss": 0.576,
+      "step": 670
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.925925925925926e-05,
+      "loss": 0.5669,
+      "step": 680
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.9166666666666665e-05,
+      "loss": 0.5954,
+      "step": 690
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.9074074074074075e-05,
+      "loss": 0.5232,
+      "step": 700
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.8981481481481486e-05,
+      "loss": 0.5216,
+      "step": 710
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.574,
+      "step": 720
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.863849765258216,
+      "eval_f1": 0.8530973509302869,
+      "eval_loss": 0.40030428767204285,
+      "eval_roc_auc": 0.9616688906216743,
+      "eval_runtime": 30.8921,
+      "eval_samples_per_second": 62.055,
+      "eval_steps_per_second": 1.942,
+      "step": 720
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.87962962962963e-05,
+      "loss": 0.5462,
+      "step": 730
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.8703703703703704e-05,
+      "loss": 0.5626,
+      "step": 740
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.535,
+      "step": 750
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.851851851851852e-05,
+      "loss": 0.5251,
+      "step": 760
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.842592592592593e-05,
+      "loss": 0.5507,
+      "step": 770
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.5269,
+      "step": 780
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.8721961398017736,
+      "eval_f1": 0.8675936123138328,
+      "eval_loss": 0.3901008367538452,
+      "eval_roc_auc": 0.967124620364712,
+      "eval_runtime": 31.0242,
+      "eval_samples_per_second": 61.79,
+      "eval_steps_per_second": 1.934,
+      "step": 780
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.8240740740740744e-05,
+      "loss": 0.5424,
+      "step": 790
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.5145,
+      "step": 800
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.805555555555556e-05,
+      "loss": 0.5088,
+      "step": 810
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.796296296296296e-05,
+      "loss": 0.524,
+      "step": 820
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.7870370370370373e-05,
+      "loss": 0.5402,
+      "step": 830
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 0.5138,
+      "step": 840
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.8732394366197183,
+      "eval_f1": 0.8607439141106847,
+      "eval_loss": 0.39837101101875305,
+      "eval_roc_auc": 0.9685311128127906,
+      "eval_runtime": 30.8559,
+      "eval_samples_per_second": 62.128,
+      "eval_steps_per_second": 1.945,
+      "step": 840
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.768518518518519e-05,
+      "loss": 0.4656,
+      "step": 850
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.759259259259259e-05,
+      "loss": 0.5275,
+      "step": 860
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.75e-05,
+      "loss": 0.5204,
+      "step": 870
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.740740740740741e-05,
+      "loss": 0.5162,
+      "step": 880
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.731481481481482e-05,
+      "loss": 0.4958,
+      "step": 890
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.4839,
+      "step": 900
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.8768909754825248,
+      "eval_f1": 0.8682586053503393,
+      "eval_loss": 0.376302033662796,
+      "eval_roc_auc": 0.9701314143362457,
+      "eval_runtime": 31.1252,
+      "eval_samples_per_second": 61.59,
+      "eval_steps_per_second": 1.928,
+      "step": 900
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 4.712962962962963e-05,
+      "loss": 0.4959,
+      "step": 910
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.703703703703704e-05,
+      "loss": 0.4924,
+      "step": 920
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.6944444444444446e-05,
+      "loss": 0.5356,
+      "step": 930
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.685185185185185e-05,
+      "loss": 0.5063,
+      "step": 940
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.675925925925926e-05,
+      "loss": 0.4696,
+      "step": 950
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.463,
+      "step": 960
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.8894105372978612,
+      "eval_f1": 0.8836985599474323,
+      "eval_loss": 0.3397548496723175,
+      "eval_roc_auc": 0.9718033457050373,
+      "eval_runtime": 30.8757,
+      "eval_samples_per_second": 62.088,
+      "eval_steps_per_second": 1.943,
+      "step": 960
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.6574074074074076e-05,
+      "loss": 0.4963,
+      "step": 970
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.648148148148148e-05,
+      "loss": 0.4818,
+      "step": 980
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.638888888888889e-05,
+      "loss": 0.4686,
+      "step": 990
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.4525,
+      "step": 1000
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 4.6203703703703705e-05,
+      "loss": 0.4735,
+      "step": 1010
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.6111111111111115e-05,
+      "loss": 0.4767,
+      "step": 1020
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.8914971309337506,
+      "eval_f1": 0.8845954057184805,
+      "eval_loss": 0.3293467164039612,
+      "eval_roc_auc": 0.9738290344332811,
+      "eval_runtime": 30.9209,
+      "eval_samples_per_second": 61.997,
+      "eval_steps_per_second": 1.94,
+      "step": 1020
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 4.601851851851852e-05,
+      "loss": 0.4284,
+      "step": 1030
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 4.592592592592593e-05,
+      "loss": 0.4957,
+      "step": 1040
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.4756,
+      "step": 1050
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 4.5740740740740745e-05,
+      "loss": 0.4693,
+      "step": 1060
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 4.564814814814815e-05,
+      "loss": 0.4691,
+      "step": 1070
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 4.555555555555556e-05,
+      "loss": 0.4985,
+      "step": 1080
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.8862806468440271,
+      "eval_f1": 0.8852238967055369,
+      "eval_loss": 0.33499640226364136,
+      "eval_roc_auc": 0.9763127743838852,
+      "eval_runtime": 31.1019,
+      "eval_samples_per_second": 61.636,
+      "eval_steps_per_second": 1.929,
+      "step": 1080
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 4.546296296296296e-05,
+      "loss": 0.47,
+      "step": 1090
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 4.5370370370370374e-05,
+      "loss": 0.4907,
+      "step": 1100
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 4.527777777777778e-05,
+      "loss": 0.44,
+      "step": 1110
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 4.518518518518519e-05,
+      "loss": 0.4573,
+      "step": 1120
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 4.50925925925926e-05,
+      "loss": 0.4056,
+      "step": 1130
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 4.5e-05,
+      "loss": 0.4657,
+      "step": 1140
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.8951486697965572,
+      "eval_f1": 0.8871889780996989,
+      "eval_loss": 0.33693990111351013,
+      "eval_roc_auc": 0.9746493362037599,
+      "eval_runtime": 30.9425,
+      "eval_samples_per_second": 61.954,
+      "eval_steps_per_second": 1.939,
+      "step": 1140
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.490740740740741e-05,
+      "loss": 0.4504,
+      "step": 1150
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 4.481481481481482e-05,
+      "loss": 0.4723,
+      "step": 1160
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 4.472222222222223e-05,
+      "loss": 0.4148,
+      "step": 1170
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 4.462962962962963e-05,
+      "loss": 0.4446,
+      "step": 1180
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 4.4537037037037036e-05,
+      "loss": 0.4566,
+      "step": 1190
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.4514,
+      "step": 1200
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.8925404277516954,
+      "eval_f1": 0.8880165267578498,
+      "eval_loss": 0.32129156589508057,
+      "eval_roc_auc": 0.9749826486859693,
+      "eval_runtime": 30.9956,
+      "eval_samples_per_second": 61.847,
+      "eval_steps_per_second": 1.936,
+      "step": 1200
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 4.435185185185186e-05,
+      "loss": 0.4075,
+      "step": 1210
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 4.425925925925926e-05,
+      "loss": 0.4775,
+      "step": 1220
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 4.4166666666666665e-05,
+      "loss": 0.4486,
+      "step": 1230
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 4.4074074074074076e-05,
+      "loss": 0.4464,
+      "step": 1240
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 4.3981481481481486e-05,
+      "loss": 0.4287,
+      "step": 1250
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 4.388888888888889e-05,
+      "loss": 0.4207,
+      "step": 1260
+    },
+    {
+      "epoch": 21.0,
+      "eval_accuracy": 0.8977569118414189,
+      "eval_f1": 0.8943328478428479,
+      "eval_loss": 0.31750577688217163,
+      "eval_roc_auc": 0.9770830922893768,
+      "eval_runtime": 30.8486,
+      "eval_samples_per_second": 62.142,
+      "eval_steps_per_second": 1.945,
+      "step": 1260
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 4.3796296296296294e-05,
+      "loss": 0.4354,
+      "step": 1270
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 4.3703703703703705e-05,
+      "loss": 0.4387,
+      "step": 1280
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 4.3611111111111116e-05,
+      "loss": 0.4229,
+      "step": 1290
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 4.351851851851852e-05,
+      "loss": 0.4408,
+      "step": 1300
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 4.342592592592592e-05,
+      "loss": 0.4114,
+      "step": 1310
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.4522,
+      "step": 1320
+    },
+    {
+      "epoch": 22.0,
+      "eval_accuracy": 0.8982785602503912,
+      "eval_f1": 0.89697139555418,
+      "eval_loss": 0.3228977620601654,
+      "eval_roc_auc": 0.9766625249534777,
+      "eval_runtime": 30.8578,
+      "eval_samples_per_second": 62.124,
+      "eval_steps_per_second": 1.944,
+      "step": 1320
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 4.3240740740740745e-05,
+      "loss": 0.3922,
+      "step": 1330
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 4.314814814814815e-05,
+      "loss": 0.3809,
+      "step": 1340
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.4079,
+      "step": 1350
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 4.296296296296296e-05,
+      "loss": 0.412,
+      "step": 1360
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 4.2870370370370374e-05,
+      "loss": 0.3794,
+      "step": 1370
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 4.277777777777778e-05,
+      "loss": 0.4328,
+      "step": 1380
+    },
+    {
+      "epoch": 23.0,
+      "eval_accuracy": 0.8977569118414189,
+      "eval_f1": 0.8947876464589353,
+      "eval_loss": 0.31209608912467957,
+      "eval_roc_auc": 0.979110772862328,
+      "eval_runtime": 31.0256,
+      "eval_samples_per_second": 61.788,
+      "eval_steps_per_second": 1.934,
+      "step": 1380
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 4.268518518518519e-05,
+      "loss": 0.3926,
+      "step": 1390
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 4.259259259259259e-05,
+      "loss": 0.4003,
+      "step": 1400
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 4.25e-05,
+      "loss": 0.375,
+      "step": 1410
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 4.240740740740741e-05,
+      "loss": 0.3676,
+      "step": 1420
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 4.231481481481482e-05,
+      "loss": 0.4068,
+      "step": 1430
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.3942,
+      "step": 1440
+    },
+    {
+      "epoch": 24.0,
+      "eval_accuracy": 0.9029733959311425,
+      "eval_f1": 0.8992700731616186,
+      "eval_loss": 0.3111121356487274,
+      "eval_roc_auc": 0.9764683636725175,
+      "eval_runtime": 30.8717,
+      "eval_samples_per_second": 62.096,
+      "eval_steps_per_second": 1.944,
+      "step": 1440
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 4.212962962962963e-05,
+      "loss": 0.3966,
+      "step": 1450
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 4.203703703703704e-05,
+      "loss": 0.3767,
+      "step": 1460
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 4.194444444444445e-05,
+      "loss": 0.4047,
+      "step": 1470
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 4.185185185185185e-05,
+      "loss": 0.3884,
+      "step": 1480
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 4.175925925925926e-05,
+      "loss": 0.3499,
+      "step": 1490
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4414,
+      "step": 1500
+    },
+    {
+      "epoch": 25.0,
+      "eval_accuracy": 0.9061032863849765,
+      "eval_f1": 0.9031554688660446,
+      "eval_loss": 0.30617934465408325,
+      "eval_roc_auc": 0.9762987886943956,
+      "eval_runtime": 31.0067,
+      "eval_samples_per_second": 61.825,
+      "eval_steps_per_second": 1.935,
+      "step": 1500
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 4.1574074074074076e-05,
+      "loss": 0.4037,
+      "step": 1510
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.148148148148148e-05,
+      "loss": 0.4282,
+      "step": 1520
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 4.138888888888889e-05,
+      "loss": 0.4035,
+      "step": 1530
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 4.12962962962963e-05,
+      "loss": 0.366,
+      "step": 1540
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 4.1203703703703705e-05,
+      "loss": 0.3912,
+      "step": 1550
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.111111111111111e-05,
+      "loss": 0.3608,
+      "step": 1560
+    },
+    {
+      "epoch": 26.0,
+      "eval_accuracy": 0.9014084507042254,
+      "eval_f1": 0.8996931816013317,
+      "eval_loss": 0.309887558221817,
+      "eval_roc_auc": 0.9786590060720087,
+      "eval_runtime": 30.8321,
+      "eval_samples_per_second": 62.175,
+      "eval_steps_per_second": 1.946,
+      "step": 1560
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 4.101851851851852e-05,
+      "loss": 0.3749,
+      "step": 1570
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 4.092592592592593e-05,
+      "loss": 0.3948,
+      "step": 1580
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 4.0833333333333334e-05,
+      "loss": 0.3999,
+      "step": 1590
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 0.3949,
+      "step": 1600
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 4.064814814814815e-05,
+      "loss": 0.4086,
+      "step": 1610
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 4.055555555555556e-05,
+      "loss": 0.3729,
+      "step": 1620
+    },
+    {
+      "epoch": 27.0,
+      "eval_accuracy": 0.908189880020866,
+      "eval_f1": 0.9029396025123122,
+      "eval_loss": 0.305014431476593,
+      "eval_roc_auc": 0.9782605273316775,
+      "eval_runtime": 30.761,
+      "eval_samples_per_second": 62.319,
+      "eval_steps_per_second": 1.951,
+      "step": 1620
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 4.0462962962962963e-05,
+      "loss": 0.4176,
+      "step": 1630
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 4.0370370370370374e-05,
+      "loss": 0.4117,
+      "step": 1640
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.3567,
+      "step": 1650
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 4.018518518518519e-05,
+      "loss": 0.3693,
+      "step": 1660
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 4.009259259259259e-05,
+      "loss": 0.3566,
+      "step": 1670
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4e-05,
+      "loss": 0.393,
+      "step": 1680
+    },
+    {
+      "epoch": 28.0,
+      "eval_accuracy": 0.9107981220657277,
+      "eval_f1": 0.9089869246713798,
+      "eval_loss": 0.2970489263534546,
+      "eval_roc_auc": 0.9796921899115975,
+      "eval_runtime": 31.0331,
+      "eval_samples_per_second": 61.773,
+      "eval_steps_per_second": 1.933,
+      "step": 1680
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 3.990740740740741e-05,
+      "loss": 0.3726,
+      "step": 1690
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 3.981481481481482e-05,
+      "loss": 0.3762,
+      "step": 1700
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.972222222222222e-05,
+      "loss": 0.4289,
+      "step": 1710
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.962962962962963e-05,
+      "loss": 0.3754,
+      "step": 1720
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 3.9537037037037036e-05,
+      "loss": 0.3363,
+      "step": 1730
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 3.944444444444445e-05,
+      "loss": 0.402,
+      "step": 1740
+    },
+    {
+      "epoch": 29.0,
+      "eval_accuracy": 0.9113197704747,
+      "eval_f1": 0.9087183876436168,
+      "eval_loss": 0.2985842525959015,
+      "eval_roc_auc": 0.9793433765347666,
+      "eval_runtime": 30.969,
+      "eval_samples_per_second": 61.901,
+      "eval_steps_per_second": 1.937,
+      "step": 1740
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 3.935185185185186e-05,
+      "loss": 0.3481,
+      "step": 1750
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 3.925925925925926e-05,
+      "loss": 0.4261,
+      "step": 1760
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 0.3484,
+      "step": 1770
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 3.9074074074074076e-05,
+      "loss": 0.3612,
+      "step": 1780
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 3.898148148148149e-05,
+      "loss": 0.3758,
+      "step": 1790
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.3697,
+      "step": 1800
+    },
+    {
+      "epoch": 30.0,
+      "eval_accuracy": 0.9024517475221701,
+      "eval_f1": 0.8967746683628594,
+      "eval_loss": 0.3384450078010559,
+      "eval_roc_auc": 0.9769238753852608,
+      "eval_runtime": 30.8671,
+      "eval_samples_per_second": 62.105,
+      "eval_steps_per_second": 1.944,
+      "step": 1800
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 3.8796296296296295e-05,
+      "loss": 0.3331,
+      "step": 1810
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 3.8703703703703705e-05,
+      "loss": 0.399,
+      "step": 1820
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 3.8611111111111116e-05,
+      "loss": 0.381,
+      "step": 1830
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 0.3682,
+      "step": 1840
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 3.8425925925925924e-05,
+      "loss": 0.3702,
+      "step": 1850
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.3502,
+      "step": 1860
+    },
+    {
+      "epoch": 31.0,
+      "eval_accuracy": 0.9102764736567553,
+      "eval_f1": 0.9058244823258932,
+      "eval_loss": 0.3035435378551483,
+      "eval_roc_auc": 0.978857810039505,
+      "eval_runtime": 30.8727,
+      "eval_samples_per_second": 62.094,
+      "eval_steps_per_second": 1.943,
+      "step": 1860
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 3.8240740740740745e-05,
+      "loss": 0.3545,
+      "step": 1870
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 3.814814814814815e-05,
+      "loss": 0.3626,
+      "step": 1880
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 3.805555555555555e-05,
+      "loss": 0.3393,
+      "step": 1890
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 3.7962962962962964e-05,
+      "loss": 0.3279,
+      "step": 1900
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 3.7870370370370374e-05,
+      "loss": 0.3332,
+      "step": 1910
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.3653,
+      "step": 1920
+    },
+    {
+      "epoch": 32.0,
+      "eval_accuracy": 0.9024517475221701,
+      "eval_f1": 0.9023846727475806,
+      "eval_loss": 0.3126789629459381,
+      "eval_roc_auc": 0.9787706292725965,
+      "eval_runtime": 30.8857,
+      "eval_samples_per_second": 62.068,
+      "eval_steps_per_second": 1.943,
+      "step": 1920
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 3.768518518518518e-05,
+      "loss": 0.3472,
+      "step": 1930
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 3.759259259259259e-05,
+      "loss": 0.3398,
+      "step": 1940
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.3706,
+      "step": 1950
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 3.740740740740741e-05,
+      "loss": 0.3413,
+      "step": 1960
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 3.731481481481482e-05,
+      "loss": 0.3371,
+      "step": 1970
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3.722222222222222e-05,
+      "loss": 0.3898,
+      "step": 1980
+    },
+    {
+      "epoch": 33.0,
+      "eval_accuracy": 0.9061032863849765,
+      "eval_f1": 0.9050367858097491,
+      "eval_loss": 0.32219406962394714,
+      "eval_roc_auc": 0.9778496116752868,
+      "eval_runtime": 31.0225,
+      "eval_samples_per_second": 61.794,
+      "eval_steps_per_second": 1.934,
+      "step": 1980
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 3.712962962962963e-05,
+      "loss": 0.3495,
+      "step": 1990
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.3505,
+      "step": 2000
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 3.694444444444445e-05,
+      "loss": 0.3248,
+      "step": 2010
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 3.685185185185185e-05,
+      "loss": 0.323,
+      "step": 2020
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 3.675925925925926e-05,
+      "loss": 0.3586,
+      "step": 2030
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.317,
+      "step": 2040
+    },
+    {
+      "epoch": 34.0,
+      "eval_accuracy": 0.9139280125195618,
+      "eval_f1": 0.9123929576037129,
+      "eval_loss": 0.30128011107444763,
+      "eval_roc_auc": 0.9798385920819667,
+      "eval_runtime": 31.1045,
+      "eval_samples_per_second": 61.631,
+      "eval_steps_per_second": 1.929,
+      "step": 2040
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.6574074074074076e-05,
+      "loss": 0.3429,
+      "step": 2050
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.648148148148148e-05,
+      "loss": 0.3753,
+      "step": 2060
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.638888888888889e-05,
+      "loss": 0.3698,
+      "step": 2070
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.62962962962963e-05,
+      "loss": 0.3595,
+      "step": 2080
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.6203703703703706e-05,
+      "loss": 0.3423,
+      "step": 2090
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.3166,
+      "step": 2100
+    },
+    {
+      "epoch": 35.0,
+      "eval_accuracy": 0.9134063641105895,
+      "eval_f1": 0.9095292266973257,
+      "eval_loss": 0.3185039162635803,
+      "eval_roc_auc": 0.9775317911265304,
+      "eval_runtime": 31.1126,
+      "eval_samples_per_second": 61.615,
+      "eval_steps_per_second": 1.928,
+      "step": 2100
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 3.601851851851852e-05,
+      "loss": 0.3309,
+      "step": 2110
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 3.592592592592593e-05,
+      "loss": 0.3442,
+      "step": 2120
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 0.3435,
+      "step": 2130
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 3.574074074074074e-05,
+      "loss": 0.3533,
+      "step": 2140
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 3.564814814814815e-05,
+      "loss": 0.3834,
+      "step": 2150
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 0.3771,
+      "step": 2160
+    },
+    {
+      "epoch": 36.0,
+      "eval_accuracy": 0.9066249347939489,
+      "eval_f1": 0.9048923351938147,
+      "eval_loss": 0.3066847622394562,
+      "eval_roc_auc": 0.9782356278578076,
+      "eval_runtime": 31.3435,
+      "eval_samples_per_second": 61.161,
+      "eval_steps_per_second": 1.914,
+      "step": 2160
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3.5462962962962964e-05,
+      "loss": 0.3224,
+      "step": 2170
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 3.537037037037037e-05,
+      "loss": 0.3377,
+      "step": 2180
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 3.527777777777778e-05,
+      "loss": 0.3458,
+      "step": 2190
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 3.518518518518519e-05,
+      "loss": 0.3623,
+      "step": 2200
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 3.509259259259259e-05,
+      "loss": 0.3362,
+      "step": 2210
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.5e-05,
+      "loss": 0.3487,
+      "step": 2220
+    },
+    {
+      "epoch": 37.0,
+      "eval_accuracy": 0.9134063641105895,
+      "eval_f1": 0.9117579891305719,
+      "eval_loss": 0.2947899401187897,
+      "eval_roc_auc": 0.980099185348387,
+      "eval_runtime": 31.0549,
+      "eval_samples_per_second": 61.729,
+      "eval_steps_per_second": 1.932,
+      "step": 2220
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 3.490740740740741e-05,
+      "loss": 0.2907,
+      "step": 2230
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 3.481481481481482e-05,
+      "loss": 0.3329,
+      "step": 2240
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.3353,
+      "step": 2250
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 3.4629629629629626e-05,
+      "loss": 0.3426,
+      "step": 2260
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 3.453703703703704e-05,
+      "loss": 0.3018,
+      "step": 2270
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.444444444444445e-05,
+      "loss": 0.3202,
+      "step": 2280
+    },
+    {
+      "epoch": 38.0,
+      "eval_accuracy": 0.918622848200313,
+      "eval_f1": 0.9167721477242283,
+      "eval_loss": 0.29156479239463806,
+      "eval_roc_auc": 0.978799051349497,
+      "eval_runtime": 30.9244,
+      "eval_samples_per_second": 61.99,
+      "eval_steps_per_second": 1.94,
+      "step": 2280
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 3.435185185185185e-05,
+      "loss": 0.3297,
+      "step": 2290
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 3.425925925925926e-05,
+      "loss": 0.3051,
+      "step": 2300
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 0.315,
+      "step": 2310
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 3.4074074074074077e-05,
+      "loss": 0.3474,
+      "step": 2320
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 3.398148148148148e-05,
+      "loss": 0.3566,
+      "step": 2330
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.388888888888889e-05,
+      "loss": 0.3163,
+      "step": 2340
+    },
+    {
+      "epoch": 39.0,
+      "eval_accuracy": 0.9154929577464789,
+      "eval_f1": 0.9141142757105575,
+      "eval_loss": 0.3148981034755707,
+      "eval_roc_auc": 0.9776534076077666,
+      "eval_runtime": 30.9687,
+      "eval_samples_per_second": 61.901,
+      "eval_steps_per_second": 1.937,
+      "step": 2340
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 3.3796296296296295e-05,
+      "loss": 0.3534,
+      "step": 2350
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 3.3703703703703706e-05,
+      "loss": 0.3158,
+      "step": 2360
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 3.3611111111111116e-05,
+      "loss": 0.3126,
+      "step": 2370
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 3.351851851851852e-05,
+      "loss": 0.3331,
+      "step": 2380
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 3.3425925925925924e-05,
+      "loss": 0.2788,
+      "step": 2390
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.3605,
+      "step": 2400
+    },
+    {
+      "epoch": 40.0,
+      "eval_accuracy": 0.9207094418362024,
+      "eval_f1": 0.919181121459014,
+      "eval_loss": 0.2963573634624481,
+      "eval_roc_auc": 0.9796925781936262,
+      "eval_runtime": 31.0457,
+      "eval_samples_per_second": 61.748,
+      "eval_steps_per_second": 1.933,
+      "step": 2400
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 3.3240740740740746e-05,
+      "loss": 0.318,
+      "step": 2410
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 3.314814814814815e-05,
+      "loss": 0.3283,
+      "step": 2420
+    },
+    {
+      "epoch": 40.5,
+      "learning_rate": 3.3055555555555553e-05,
+      "loss": 0.3141,
+      "step": 2430
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 3.2962962962962964e-05,
+      "loss": 0.3245,
+      "step": 2440
+    },
+    {
+      "epoch": 40.83,
+      "learning_rate": 3.2870370370370375e-05,
+      "loss": 0.3381,
+      "step": 2450
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 3.277777777777778e-05,
+      "loss": 0.3636,
+      "step": 2460
+    },
+    {
+      "epoch": 41.0,
+      "eval_accuracy": 0.9134063641105895,
+      "eval_f1": 0.9111347359382455,
+      "eval_loss": 0.3142088055610657,
+      "eval_roc_auc": 0.9809569663590326,
+      "eval_runtime": 30.9606,
+      "eval_samples_per_second": 61.917,
+      "eval_steps_per_second": 1.938,
+      "step": 2460
+    },
+    {
+      "epoch": 41.17,
+      "learning_rate": 3.268518518518518e-05,
+      "loss": 0.3145,
+      "step": 2470
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 3.25925925925926e-05,
+      "loss": 0.3277,
+      "step": 2480
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.3806,
+      "step": 2490
+    },
+    {
+      "epoch": 41.67,
+      "learning_rate": 3.240740740740741e-05,
+      "loss": 0.3386,
+      "step": 2500
+    },
+    {
+      "epoch": 41.83,
+      "learning_rate": 3.231481481481481e-05,
+      "loss": 0.3237,
+      "step": 2510
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 3.222222222222223e-05,
+      "loss": 0.3454,
+      "step": 2520
+    },
+    {
+      "epoch": 42.0,
+      "eval_accuracy": 0.9113197704747,
+      "eval_f1": 0.9111205125376055,
+      "eval_loss": 0.31330010294914246,
+      "eval_roc_auc": 0.9791633758091489,
+      "eval_runtime": 31.2016,
+      "eval_samples_per_second": 61.439,
+      "eval_steps_per_second": 1.923,
+      "step": 2520
+    },
+    {
+      "epoch": 42.17,
+      "learning_rate": 3.212962962962963e-05,
+      "loss": 0.3132,
+      "step": 2530
+    },
+    {
+      "epoch": 42.33,
+      "learning_rate": 3.203703703703704e-05,
+      "loss": 0.337,
+      "step": 2540
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.3093,
+      "step": 2550
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 3.185185185185185e-05,
+      "loss": 0.3306,
+      "step": 2560
+    },
+    {
+      "epoch": 42.83,
+      "learning_rate": 3.175925925925926e-05,
+      "loss": 0.3781,
+      "step": 2570
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.3561,
+      "step": 2580
+    },
+    {
+      "epoch": 43.0,
+      "eval_accuracy": 0.9076682316118936,
+      "eval_f1": 0.9072816984688045,
+      "eval_loss": 0.3090434968471527,
+      "eval_roc_auc": 0.9804251537873032,
+      "eval_runtime": 30.8687,
+      "eval_samples_per_second": 62.102,
+      "eval_steps_per_second": 1.944,
+      "step": 2580
+    },
+    {
+      "epoch": 43.17,
+      "learning_rate": 3.157407407407408e-05,
+      "loss": 0.3419,
+      "step": 2590
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 3.148148148148148e-05,
+      "loss": 0.3213,
+      "step": 2600
+    },
+    {
+      "epoch": 43.5,
+      "learning_rate": 3.138888888888889e-05,
+      "loss": 0.3134,
+      "step": 2610
+    },
+    {
+      "epoch": 43.67,
+      "learning_rate": 3.1296296296296295e-05,
+      "loss": 0.3419,
+      "step": 2620
+    },
+    {
+      "epoch": 43.83,
+      "learning_rate": 3.1203703703703706e-05,
+      "loss": 0.282,
+      "step": 2630
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.3136,
+      "step": 2640
+    },
+    {
+      "epoch": 44.0,
+      "eval_accuracy": 0.9175795513823682,
+      "eval_f1": 0.9143946754568497,
+      "eval_loss": 0.32355380058288574,
+      "eval_roc_auc": 0.9781668988388087,
+      "eval_runtime": 31.0613,
+      "eval_samples_per_second": 61.717,
+      "eval_steps_per_second": 1.932,
+      "step": 2640
+    },
+    {
+      "epoch": 44.17,
+      "learning_rate": 3.101851851851852e-05,
+      "loss": 0.3187,
+      "step": 2650
+    },
+    {
+      "epoch": 44.33,
+      "learning_rate": 3.0925925925925924e-05,
+      "loss": 0.3254,
+      "step": 2660
+    },
+    {
+      "epoch": 44.5,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 0.292,
+      "step": 2670
+    },
+    {
+      "epoch": 44.67,
+      "learning_rate": 3.074074074074074e-05,
+      "loss": 0.3079,
+      "step": 2680
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 3.064814814814815e-05,
+      "loss": 0.2808,
+      "step": 2690
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 0.3529,
+      "step": 2700
+    },
+    {
+      "epoch": 45.0,
+      "eval_accuracy": 0.92018779342723,
+      "eval_f1": 0.9174792311793412,
+      "eval_loss": 0.30541861057281494,
+      "eval_roc_auc": 0.9800384924019299,
+      "eval_runtime": 31.1166,
+      "eval_samples_per_second": 61.607,
+      "eval_steps_per_second": 1.928,
+      "step": 2700
+    },
+    {
+      "epoch": 45.17,
+      "learning_rate": 3.0462962962962964e-05,
+      "loss": 0.353,
+      "step": 2710
+    },
+    {
+      "epoch": 45.33,
+      "learning_rate": 3.037037037037037e-05,
+      "loss": 0.3488,
+      "step": 2720
+    },
+    {
+      "epoch": 45.5,
+      "learning_rate": 3.0277777777777776e-05,
+      "loss": 0.2973,
+      "step": 2730
+    },
+    {
+      "epoch": 45.67,
+      "learning_rate": 3.018518518518519e-05,
+      "loss": 0.325,
+      "step": 2740
+    },
+    {
+      "epoch": 45.83,
+      "learning_rate": 3.0092592592592593e-05,
+      "loss": 0.3183,
+      "step": 2750
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 3e-05,
+      "loss": 0.2987,
+      "step": 2760
+    },
+    {
+      "epoch": 46.0,
+      "eval_accuracy": 0.9233176838810642,
+      "eval_f1": 0.9221803007650173,
+      "eval_loss": 0.294436514377594,
+      "eval_roc_auc": 0.9801680285600346,
+      "eval_runtime": 30.9755,
+      "eval_samples_per_second": 61.888,
+      "eval_steps_per_second": 1.937,
+      "step": 2760
+    },
+    {
+      "epoch": 46.17,
+      "learning_rate": 2.9907407407407405e-05,
+      "loss": 0.2837,
+      "step": 2770
+    },
+    {
+      "epoch": 46.33,
+      "learning_rate": 2.981481481481482e-05,
+      "loss": 0.3087,
+      "step": 2780
+    },
+    {
+      "epoch": 46.5,
+      "learning_rate": 2.9722222222222223e-05,
+      "loss": 0.2876,
+      "step": 2790
+    },
+    {
+      "epoch": 46.67,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.3132,
+      "step": 2800
+    },
+    {
+      "epoch": 46.83,
+      "learning_rate": 2.953703703703704e-05,
+      "loss": 0.2676,
+      "step": 2810
+    },
+    {
+      "epoch": 47.0,
+      "learning_rate": 2.9444444444444448e-05,
+      "loss": 0.2966,
+      "step": 2820
+    },
+    {
+      "epoch": 47.0,
+      "eval_accuracy": 0.9233176838810642,
+      "eval_f1": 0.920079923314828,
+      "eval_loss": 0.3215111196041107,
+      "eval_roc_auc": 0.9785648638154409,
+      "eval_runtime": 31.0099,
+      "eval_samples_per_second": 61.819,
+      "eval_steps_per_second": 1.935,
+      "step": 2820
+    },
+    {
+      "epoch": 47.17,
+      "learning_rate": 2.9351851851851852e-05,
+      "loss": 0.297,
+      "step": 2830
+    },
+    {
+      "epoch": 47.33,
+      "learning_rate": 2.925925925925926e-05,
+      "loss": 0.2923,
+      "step": 2840
+    },
+    {
+      "epoch": 47.5,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.2761,
+      "step": 2850
+    },
+    {
+      "epoch": 47.67,
+      "learning_rate": 2.9074074074074077e-05,
+      "loss": 0.3353,
+      "step": 2860
+    },
+    {
+      "epoch": 47.83,
+      "learning_rate": 2.898148148148148e-05,
+      "loss": 0.3118,
+      "step": 2870
+    },
+    {
+      "epoch": 48.0,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 0.3203,
+      "step": 2880
+    },
+    {
+      "epoch": 48.0,
+      "eval_accuracy": 0.9243609806990088,
+      "eval_f1": 0.9218866596198343,
+      "eval_loss": 0.31495025753974915,
+      "eval_roc_auc": 0.9797406270362136,
+      "eval_runtime": 30.9782,
+      "eval_samples_per_second": 61.882,
+      "eval_steps_per_second": 1.937,
+      "step": 2880
+    },
+    {
+      "epoch": 48.17,
+      "learning_rate": 2.87962962962963e-05,
+      "loss": 0.2725,
+      "step": 2890
+    },
+    {
+      "epoch": 48.33,
+      "learning_rate": 2.8703703703703706e-05,
+      "loss": 0.3288,
+      "step": 2900
+    },
+    {
+      "epoch": 48.5,
+      "learning_rate": 2.861111111111111e-05,
+      "loss": 0.2848,
+      "step": 2910
+    },
+    {
+      "epoch": 48.67,
+      "learning_rate": 2.851851851851852e-05,
+      "loss": 0.2902,
+      "step": 2920
+    },
+    {
+      "epoch": 48.83,
+      "learning_rate": 2.8425925925925928e-05,
+      "loss": 0.3362,
+      "step": 2930
+    },
+    {
+      "epoch": 49.0,
+      "learning_rate": 2.8333333333333335e-05,
+      "loss": 0.2821,
+      "step": 2940
+    },
+    {
+      "epoch": 49.0,
+      "eval_accuracy": 0.9290558163797601,
+      "eval_f1": 0.9273079215291836,
+      "eval_loss": 0.30717700719833374,
+      "eval_roc_auc": 0.980032505249785,
+      "eval_runtime": 31.1146,
+      "eval_samples_per_second": 61.611,
+      "eval_steps_per_second": 1.928,
+      "step": 2940
+    },
+    {
+      "epoch": 49.17,
+      "learning_rate": 2.824074074074074e-05,
+      "loss": 0.3273,
+      "step": 2950
+    },
+    {
+      "epoch": 49.33,
+      "learning_rate": 2.814814814814815e-05,
+      "loss": 0.3114,
+      "step": 2960
+    },
+    {
+      "epoch": 49.5,
+      "learning_rate": 2.8055555555555557e-05,
+      "loss": 0.3081,
+      "step": 2970
+    },
+    {
+      "epoch": 49.67,
+      "learning_rate": 2.7962962962962965e-05,
+      "loss": 0.281,
+      "step": 2980
+    },
+    {
+      "epoch": 49.83,
+      "learning_rate": 2.7870370370370375e-05,
+      "loss": 0.2993,
+      "step": 2990
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.2852,
+      "step": 3000
+    },
+    {
+      "epoch": 50.0,
+      "eval_accuracy": 0.9175795513823682,
+      "eval_f1": 0.9154771942358105,
+      "eval_loss": 0.32651108503341675,
+      "eval_roc_auc": 0.9792309379704945,
+      "eval_runtime": 31.0614,
+      "eval_samples_per_second": 61.716,
+      "eval_steps_per_second": 1.932,
+      "step": 3000
+    },
+    {
+      "epoch": 50.17,
+      "learning_rate": 2.7685185185185186e-05,
+      "loss": 0.2835,
+      "step": 3010
+    },
+    {
+      "epoch": 50.33,
+      "learning_rate": 2.7592592592592594e-05,
+      "loss": 0.315,
+      "step": 3020
+    },
+    {
+      "epoch": 50.5,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.3096,
+      "step": 3030
+    },
+    {
+      "epoch": 50.67,
+      "learning_rate": 2.7407407407407408e-05,
+      "loss": 0.3351,
+      "step": 3040
+    },
+    {
+      "epoch": 50.83,
+      "learning_rate": 2.7314814814814816e-05,
+      "loss": 0.3037,
+      "step": 3050
+    },
+    {
+      "epoch": 51.0,
+      "learning_rate": 2.7222222222222223e-05,
+      "loss": 0.3544,
+      "step": 3060
+    },
+    {
+      "epoch": 51.0,
+      "eval_accuracy": 0.9149713093375065,
+      "eval_f1": 0.9149966409724043,
+      "eval_loss": 0.31751149892807007,
+      "eval_roc_auc": 0.9801850383933074,
+      "eval_runtime": 30.9664,
+      "eval_samples_per_second": 61.906,
+      "eval_steps_per_second": 1.938,
+      "step": 3060
+    },
+    {
+      "epoch": 51.17,
+      "learning_rate": 2.7129629629629634e-05,
+      "loss": 0.3055,
+      "step": 3070
+    },
+    {
+      "epoch": 51.33,
+      "learning_rate": 2.7037037037037037e-05,
+      "loss": 0.268,
+      "step": 3080
+    },
+    {
+      "epoch": 51.5,
+      "learning_rate": 2.6944444444444445e-05,
+      "loss": 0.3169,
+      "step": 3090
+    },
+    {
+      "epoch": 51.67,
+      "learning_rate": 2.6851851851851855e-05,
+      "loss": 0.3145,
+      "step": 3100
+    },
+    {
+      "epoch": 51.83,
+      "learning_rate": 2.6759259259259263e-05,
+      "loss": 0.2648,
+      "step": 3110
+    },
+    {
+      "epoch": 52.0,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.3327,
+      "step": 3120
+    },
+    {
+      "epoch": 52.0,
+      "eval_accuracy": 0.9243609806990088,
+      "eval_f1": 0.9221864154492979,
+      "eval_loss": 0.31344056129455566,
+      "eval_roc_auc": 0.9802070926178522,
+      "eval_runtime": 30.8335,
+      "eval_samples_per_second": 62.173,
+      "eval_steps_per_second": 1.946,
+      "step": 3120
+    },
+    {
+      "epoch": 52.17,
+      "learning_rate": 2.6574074074074074e-05,
+      "loss": 0.2923,
+      "step": 3130
+    },
+    {
+      "epoch": 52.33,
+      "learning_rate": 2.6481481481481485e-05,
+      "loss": 0.2827,
+      "step": 3140
+    },
+    {
+      "epoch": 52.5,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 0.3051,
+      "step": 3150
+    },
+    {
+      "epoch": 52.67,
+      "learning_rate": 2.6296296296296296e-05,
+      "loss": 0.3418,
+      "step": 3160
+    },
+    {
+      "epoch": 52.83,
+      "learning_rate": 2.6203703703703703e-05,
+      "loss": 0.312,
+      "step": 3170
+    },
+    {
+      "epoch": 53.0,
+      "learning_rate": 2.6111111111111114e-05,
+      "loss": 0.2877,
+      "step": 3180
+    },
+    {
+      "epoch": 53.0,
+      "eval_accuracy": 0.9165362545644236,
+      "eval_f1": 0.9154266697091844,
+      "eval_loss": 0.3222474157810211,
+      "eval_roc_auc": 0.980546397557177,
+      "eval_runtime": 31.0108,
+      "eval_samples_per_second": 61.817,
+      "eval_steps_per_second": 1.935,
+      "step": 3180
+    },
+    {
+      "epoch": 53.17,
+      "learning_rate": 2.601851851851852e-05,
+      "loss": 0.3011,
+      "step": 3190
+    },
+    {
+      "epoch": 53.33,
+      "learning_rate": 2.5925925925925925e-05,
+      "loss": 0.3116,
+      "step": 3200
+    },
+    {
+      "epoch": 53.5,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 0.2903,
+      "step": 3210
+    },
+    {
+      "epoch": 53.67,
+      "learning_rate": 2.5740740740740743e-05,
+      "loss": 0.2932,
+      "step": 3220
+    },
+    {
+      "epoch": 53.83,
+      "learning_rate": 2.564814814814815e-05,
+      "loss": 0.2816,
+      "step": 3230
+    },
+    {
+      "epoch": 54.0,
+      "learning_rate": 2.5555555555555554e-05,
+      "loss": 0.3089,
+      "step": 3240
+    },
+    {
+      "epoch": 54.0,
+      "eval_accuracy": 0.9259259259259259,
+      "eval_f1": 0.924844063747744,
+      "eval_loss": 0.30451875925064087,
+      "eval_roc_auc": 0.9810896700594666,
+      "eval_runtime": 30.7749,
+      "eval_samples_per_second": 62.291,
+      "eval_steps_per_second": 1.95,
+      "step": 3240
+    },
+    {
+      "epoch": 54.17,
+      "learning_rate": 2.5462962962962965e-05,
+      "loss": 0.2764,
+      "step": 3250
+    },
+    {
+      "epoch": 54.33,
+      "learning_rate": 2.5370370370370372e-05,
+      "loss": 0.2805,
+      "step": 3260
+    },
+    {
+      "epoch": 54.5,
+      "learning_rate": 2.527777777777778e-05,
+      "loss": 0.3094,
+      "step": 3270
+    },
+    {
+      "epoch": 54.67,
+      "learning_rate": 2.5185185185185183e-05,
+      "loss": 0.2699,
+      "step": 3280
+    },
+    {
+      "epoch": 54.83,
+      "learning_rate": 2.5092592592592594e-05,
+      "loss": 0.2787,
+      "step": 3290
+    },
+    {
+      "epoch": 55.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2904,
+      "step": 3300
+    },
+    {
+      "epoch": 55.0,
+      "eval_accuracy": 0.918622848200313,
+      "eval_f1": 0.9174621980217676,
+      "eval_loss": 0.3300776183605194,
+      "eval_roc_auc": 0.9786791637818757,
+      "eval_runtime": 31.1955,
+      "eval_samples_per_second": 61.451,
+      "eval_steps_per_second": 1.923,
+      "step": 3300
+    },
+    {
+      "epoch": 55.17,
+      "learning_rate": 2.490740740740741e-05,
+      "loss": 0.2824,
+      "step": 3310
+    },
+    {
+      "epoch": 55.33,
+      "learning_rate": 2.4814814814814816e-05,
+      "loss": 0.2755,
+      "step": 3320
+    },
+    {
+      "epoch": 55.5,
+      "learning_rate": 2.4722222222222223e-05,
+      "loss": 0.3083,
+      "step": 3330
+    },
+    {
+      "epoch": 55.67,
+      "learning_rate": 2.462962962962963e-05,
+      "loss": 0.2926,
+      "step": 3340
+    },
+    {
+      "epoch": 55.83,
+      "learning_rate": 2.4537037037037038e-05,
+      "loss": 0.3447,
+      "step": 3350
+    },
+    {
+      "epoch": 56.0,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 0.2821,
+      "step": 3360
+    },
+    {
+      "epoch": 56.0,
+      "eval_accuracy": 0.9217527386541471,
+      "eval_f1": 0.9206002214325367,
+      "eval_loss": 0.30689331889152527,
+      "eval_roc_auc": 0.9809813323357223,
+      "eval_runtime": 30.8415,
+      "eval_samples_per_second": 62.157,
+      "eval_steps_per_second": 1.945,
+      "step": 3360
+    },
+    {
+      "epoch": 56.17,
+      "learning_rate": 2.4351851851851852e-05,
+      "loss": 0.3106,
+      "step": 3370
+    },
+    {
+      "epoch": 56.33,
+      "learning_rate": 2.425925925925926e-05,
+      "loss": 0.2783,
+      "step": 3380
+    },
+    {
+      "epoch": 56.5,
+      "learning_rate": 2.4166666666666667e-05,
+      "loss": 0.2781,
+      "step": 3390
+    },
+    {
+      "epoch": 56.67,
+      "learning_rate": 2.4074074074074074e-05,
+      "loss": 0.2751,
+      "step": 3400
+    },
+    {
+      "epoch": 56.83,
+      "learning_rate": 2.398148148148148e-05,
+      "loss": 0.2861,
+      "step": 3410
+    },
+    {
+      "epoch": 57.0,
+      "learning_rate": 2.3888888888888892e-05,
+      "loss": 0.321,
+      "step": 3420
+    },
+    {
+      "epoch": 57.0,
+      "eval_accuracy": 0.9269692227438706,
+      "eval_f1": 0.9254350978490243,
+      "eval_loss": 0.32090768218040466,
+      "eval_roc_auc": 0.980025373202588,
+      "eval_runtime": 30.8398,
+      "eval_samples_per_second": 62.16,
+      "eval_steps_per_second": 1.946,
+      "step": 3420
+    },
+    {
+      "epoch": 57.17,
+      "learning_rate": 2.3796296296296296e-05,
+      "loss": 0.3079,
+      "step": 3430
+    },
+    {
+      "epoch": 57.33,
+      "learning_rate": 2.3703703703703707e-05,
+      "loss": 0.3239,
+      "step": 3440
+    },
+    {
+      "epoch": 57.5,
+      "learning_rate": 2.361111111111111e-05,
+      "loss": 0.2967,
+      "step": 3450
+    },
+    {
+      "epoch": 57.67,
+      "learning_rate": 2.351851851851852e-05,
+      "loss": 0.2888,
+      "step": 3460
+    },
+    {
+      "epoch": 57.83,
+      "learning_rate": 2.3425925925925925e-05,
+      "loss": 0.2965,
+      "step": 3470
+    },
+    {
+      "epoch": 58.0,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.2995,
+      "step": 3480
+    },
+    {
+      "epoch": 58.0,
+      "eval_accuracy": 0.9233176838810642,
+      "eval_f1": 0.9202460923868973,
+      "eval_loss": 0.32812780141830444,
+      "eval_roc_auc": 0.9802116570487911,
+      "eval_runtime": 30.7892,
+      "eval_samples_per_second": 62.262,
+      "eval_steps_per_second": 1.949,
+      "step": 3480
+    },
+    {
+      "epoch": 58.17,
+      "learning_rate": 2.324074074074074e-05,
+      "loss": 0.2897,
+      "step": 3490
+    },
+    {
+      "epoch": 58.33,
+      "learning_rate": 2.314814814814815e-05,
+      "loss": 0.279,
+      "step": 3500
+    },
+    {
+      "epoch": 58.5,
+      "learning_rate": 2.3055555555555558e-05,
+      "loss": 0.3187,
+      "step": 3510
+    },
+    {
+      "epoch": 58.67,
+      "learning_rate": 2.2962962962962965e-05,
+      "loss": 0.2889,
+      "step": 3520
+    },
+    {
+      "epoch": 58.83,
+      "learning_rate": 2.2870370370370372e-05,
+      "loss": 0.3007,
+      "step": 3530
+    },
+    {
+      "epoch": 59.0,
+      "learning_rate": 2.277777777777778e-05,
+      "loss": 0.2683,
+      "step": 3540
+    },
+    {
+      "epoch": 59.0,
+      "eval_accuracy": 0.92018779342723,
+      "eval_f1": 0.9173963500365258,
+      "eval_loss": 0.3262675702571869,
+      "eval_roc_auc": 0.9801554792004114,
+      "eval_runtime": 30.9822,
+      "eval_samples_per_second": 61.874,
+      "eval_steps_per_second": 1.937,
+      "step": 3540
+    },
+    {
+      "epoch": 59.17,
+      "learning_rate": 2.2685185185185187e-05,
+      "loss": 0.2798,
+      "step": 3550
+    },
+    {
+      "epoch": 59.33,
+      "learning_rate": 2.2592592592592594e-05,
+      "loss": 0.2968,
+      "step": 3560
+    },
+    {
+      "epoch": 59.5,
+      "learning_rate": 2.25e-05,
+      "loss": 0.293,
+      "step": 3570
+    },
+    {
+      "epoch": 59.67,
+      "learning_rate": 2.240740740740741e-05,
+      "loss": 0.3071,
+      "step": 3580
+    },
+    {
+      "epoch": 59.83,
+      "learning_rate": 2.2314814814814816e-05,
+      "loss": 0.275,
+      "step": 3590
+    },
+    {
+      "epoch": 60.0,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.3021,
+      "step": 3600
+    },
+    {
+      "epoch": 60.0,
+      "eval_accuracy": 0.918622848200313,
+      "eval_f1": 0.9170405607415676,
+      "eval_loss": 0.3483915627002716,
+      "eval_roc_auc": 0.9788368798211406,
+      "eval_runtime": 30.8968,
+      "eval_samples_per_second": 62.045,
+      "eval_steps_per_second": 1.942,
+      "step": 3600
+    },
+    {
+      "epoch": 60.17,
+      "learning_rate": 2.212962962962963e-05,
+      "loss": 0.279,
+      "step": 3610
+    },
+    {
+      "epoch": 60.33,
+      "learning_rate": 2.2037037037037038e-05,
+      "loss": 0.2727,
+      "step": 3620
+    },
+    {
+      "epoch": 60.5,
+      "learning_rate": 2.1944444444444445e-05,
+      "loss": 0.2738,
+      "step": 3630
+    },
+    {
+      "epoch": 60.67,
+      "learning_rate": 2.1851851851851852e-05,
+      "loss": 0.2756,
+      "step": 3640
+    },
+    {
+      "epoch": 60.83,
+      "learning_rate": 2.175925925925926e-05,
+      "loss": 0.2846,
+      "step": 3650
+    },
+    {
+      "epoch": 61.0,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.3262,
+      "step": 3660
+    },
+    {
+      "epoch": 61.0,
+      "eval_accuracy": 0.9165362545644236,
+      "eval_f1": 0.9150909158754031,
+      "eval_loss": 0.327029287815094,
+      "eval_roc_auc": 0.98071548830533,
+      "eval_runtime": 30.9058,
+      "eval_samples_per_second": 62.027,
+      "eval_steps_per_second": 1.941,
+      "step": 3660
+    },
+    {
+      "epoch": 61.17,
+      "learning_rate": 2.1574074074074074e-05,
+      "loss": 0.281,
+      "step": 3670
+    },
+    {
+      "epoch": 61.33,
+      "learning_rate": 2.148148148148148e-05,
+      "loss": 0.2772,
+      "step": 3680
+    },
+    {
+      "epoch": 61.5,
+      "learning_rate": 2.138888888888889e-05,
+      "loss": 0.2932,
+      "step": 3690
+    },
+    {
+      "epoch": 61.67,
+      "learning_rate": 2.1296296296296296e-05,
+      "loss": 0.3086,
+      "step": 3700
+    },
+    {
+      "epoch": 61.83,
+      "learning_rate": 2.1203703703703703e-05,
+      "loss": 0.3077,
+      "step": 3710
+    },
+    {
+      "epoch": 62.0,
+      "learning_rate": 2.111111111111111e-05,
+      "loss": 0.2329,
+      "step": 3720
+    },
+    {
+      "epoch": 62.0,
+      "eval_accuracy": 0.9233176838810642,
+      "eval_f1": 0.9210533466274746,
+      "eval_loss": 0.3280071020126343,
+      "eval_roc_auc": 0.9806818031844223,
+      "eval_runtime": 30.8772,
+      "eval_samples_per_second": 62.085,
+      "eval_steps_per_second": 1.943,
+      "step": 3720
+    },
+    {
+      "epoch": 62.17,
+      "learning_rate": 2.101851851851852e-05,
+      "loss": 0.2588,
+      "step": 3730
+    },
+    {
+      "epoch": 62.33,
+      "learning_rate": 2.0925925925925925e-05,
+      "loss": 0.2897,
+      "step": 3740
+    },
+    {
+      "epoch": 62.5,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.2828,
+      "step": 3750
+    },
+    {
+      "epoch": 62.67,
+      "learning_rate": 2.074074074074074e-05,
+      "loss": 0.3133,
+      "step": 3760
+    },
+    {
+      "epoch": 62.83,
+      "learning_rate": 2.064814814814815e-05,
+      "loss": 0.2802,
+      "step": 3770
+    },
+    {
+      "epoch": 63.0,
+      "learning_rate": 2.0555555555555555e-05,
+      "loss": 0.2935,
+      "step": 3780
+    },
+    {
+      "epoch": 63.0,
+      "eval_accuracy": 0.9264475743348983,
+      "eval_f1": 0.9243945598307481,
+      "eval_loss": 0.32962048053741455,
+      "eval_roc_auc": 0.9806711916360894,
+      "eval_runtime": 30.8281,
+      "eval_samples_per_second": 62.183,
+      "eval_steps_per_second": 1.946,
+      "step": 3780
+    },
+    {
+      "epoch": 63.17,
+      "learning_rate": 2.0462962962962965e-05,
+      "loss": 0.2488,
+      "step": 3790
+    },
+    {
+      "epoch": 63.33,
+      "learning_rate": 2.037037037037037e-05,
+      "loss": 0.2998,
+      "step": 3800
+    },
+    {
+      "epoch": 63.5,
+      "learning_rate": 2.027777777777778e-05,
+      "loss": 0.2742,
+      "step": 3810
+    },
+    {
+      "epoch": 63.67,
+      "learning_rate": 2.0185185185185187e-05,
+      "loss": 0.2898,
+      "step": 3820
+    },
+    {
+      "epoch": 63.83,
+      "learning_rate": 2.0092592592592594e-05,
+      "loss": 0.2845,
+      "step": 3830
+    },
+    {
+      "epoch": 64.0,
+      "learning_rate": 2e-05,
+      "loss": 0.2856,
+      "step": 3840
+    },
+    {
+      "epoch": 64.0,
+      "eval_accuracy": 0.9217527386541471,
+      "eval_f1": 0.9209061464765306,
+      "eval_loss": 0.33229467272758484,
+      "eval_roc_auc": 0.9810524271999177,
+      "eval_runtime": 31.0544,
+      "eval_samples_per_second": 61.73,
+      "eval_steps_per_second": 1.932,
+      "step": 3840
+    },
+    {
+      "epoch": 64.17,
+      "learning_rate": 1.990740740740741e-05,
+      "loss": 0.2797,
+      "step": 3850
+    },
+    {
+      "epoch": 64.33,
+      "learning_rate": 1.9814814814814816e-05,
+      "loss": 0.2596,
+      "step": 3860
+    },
+    {
+      "epoch": 64.5,
+      "learning_rate": 1.9722222222222224e-05,
+      "loss": 0.2541,
+      "step": 3870
+    },
+    {
+      "epoch": 64.67,
+      "learning_rate": 1.962962962962963e-05,
+      "loss": 0.295,
+      "step": 3880
+    },
+    {
+      "epoch": 64.83,
+      "learning_rate": 1.9537037037037038e-05,
+      "loss": 0.2947,
+      "step": 3890
+    },
+    {
+      "epoch": 65.0,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.2829,
+      "step": 3900
+    },
+    {
+      "epoch": 65.0,
+      "eval_accuracy": 0.9217527386541471,
+      "eval_f1": 0.9200417695607408,
+      "eval_loss": 0.3389854431152344,
+      "eval_roc_auc": 0.9802066669139512,
+      "eval_runtime": 30.8727,
+      "eval_samples_per_second": 62.094,
+      "eval_steps_per_second": 1.943,
+      "step": 3900
+    },
+    {
+      "epoch": 65.17,
+      "learning_rate": 1.9351851851851853e-05,
+      "loss": 0.2998,
+      "step": 3910
+    },
+    {
+      "epoch": 65.33,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.285,
+      "step": 3920
+    },
+    {
+      "epoch": 65.5,
+      "learning_rate": 1.9166666666666667e-05,
+      "loss": 0.2674,
+      "step": 3930
+    },
+    {
+      "epoch": 65.67,
+      "learning_rate": 1.9074074074074075e-05,
+      "loss": 0.2454,
+      "step": 3940
+    },
+    {
+      "epoch": 65.83,
+      "learning_rate": 1.8981481481481482e-05,
+      "loss": 0.2881,
+      "step": 3950
+    },
+    {
+      "epoch": 66.0,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.3044,
+      "step": 3960
+    },
+    {
+      "epoch": 66.0,
+      "eval_accuracy": 0.9227960354720918,
+      "eval_f1": 0.9215159958118743,
+      "eval_loss": 0.3324008285999298,
+      "eval_roc_auc": 0.9799450739817069,
+      "eval_runtime": 30.9618,
+      "eval_samples_per_second": 61.915,
+      "eval_steps_per_second": 1.938,
+      "step": 3960
+    },
+    {
+      "epoch": 66.17,
+      "learning_rate": 1.8796296296296296e-05,
+      "loss": 0.2662,
+      "step": 3970
+    },
+    {
+      "epoch": 66.33,
+      "learning_rate": 1.8703703703703704e-05,
+      "loss": 0.3053,
+      "step": 3980
+    },
+    {
+      "epoch": 66.5,
+      "learning_rate": 1.861111111111111e-05,
+      "loss": 0.3118,
+      "step": 3990
+    },
+    {
+      "epoch": 66.67,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.2615,
+      "step": 4000
+    },
+    {
+      "epoch": 66.83,
+      "learning_rate": 1.8425925925925926e-05,
+      "loss": 0.2664,
+      "step": 4010
+    },
+    {
+      "epoch": 67.0,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.2767,
+      "step": 4020
+    },
+    {
+      "epoch": 67.0,
+      "eval_accuracy": 0.9160146061554513,
+      "eval_f1": 0.9149949948229915,
+      "eval_loss": 0.34964630007743835,
+      "eval_roc_auc": 0.9777649341860628,
+      "eval_runtime": 31.1987,
+      "eval_samples_per_second": 61.445,
+      "eval_steps_per_second": 1.923,
+      "step": 4020
+    },
+    {
+      "epoch": 67.17,
+      "learning_rate": 1.824074074074074e-05,
+      "loss": 0.3021,
+      "step": 4030
+    },
+    {
+      "epoch": 67.33,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 0.307,
+      "step": 4040
+    },
+    {
+      "epoch": 67.5,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 0.2269,
+      "step": 4050
+    },
+    {
+      "epoch": 67.67,
+      "learning_rate": 1.7962962962962965e-05,
+      "loss": 0.2514,
+      "step": 4060
+    },
+    {
+      "epoch": 67.83,
+      "learning_rate": 1.787037037037037e-05,
+      "loss": 0.2634,
+      "step": 4070
+    },
+    {
+      "epoch": 68.0,
+      "learning_rate": 1.777777777777778e-05,
+      "loss": 0.2936,
+      "step": 4080
+    },
+    {
+      "epoch": 68.0,
+      "eval_accuracy": 0.927490871152843,
+      "eval_f1": 0.9257053426626348,
+      "eval_loss": 0.3377942442893982,
+      "eval_roc_auc": 0.9790075894232677,
+      "eval_runtime": 30.9839,
+      "eval_samples_per_second": 61.871,
+      "eval_steps_per_second": 1.936,
+      "step": 4080
+    },
+    {
+      "epoch": 68.17,
+      "learning_rate": 1.7685185185185184e-05,
+      "loss": 0.2498,
+      "step": 4090
+    },
+    {
+      "epoch": 68.33,
+      "learning_rate": 1.7592592592592595e-05,
+      "loss": 0.2849,
+      "step": 4100
+    },
+    {
+      "epoch": 68.5,
+      "learning_rate": 1.75e-05,
+      "loss": 0.2601,
+      "step": 4110
+    },
+    {
+      "epoch": 68.67,
+      "learning_rate": 1.740740740740741e-05,
+      "loss": 0.2792,
+      "step": 4120
+    },
+    {
+      "epoch": 68.83,
+      "learning_rate": 1.7314814814814813e-05,
+      "loss": 0.318,
+      "step": 4130
+    },
+    {
+      "epoch": 69.0,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 0.2884,
+      "step": 4140
+    },
+    {
+      "epoch": 69.0,
+      "eval_accuracy": 0.9248826291079812,
+      "eval_f1": 0.9226785544767538,
+      "eval_loss": 0.34928712248802185,
+      "eval_roc_auc": 0.9789693127399084,
+      "eval_runtime": 30.9173,
+      "eval_samples_per_second": 62.004,
+      "eval_steps_per_second": 1.941,
+      "step": 4140
+    },
+    {
+      "epoch": 69.17,
+      "learning_rate": 1.712962962962963e-05,
+      "loss": 0.2657,
+      "step": 4150
+    },
+    {
+      "epoch": 69.33,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.2904,
+      "step": 4160
+    },
+    {
+      "epoch": 69.5,
+      "learning_rate": 1.6944444444444446e-05,
+      "loss": 0.2975,
+      "step": 4170
+    },
+    {
+      "epoch": 69.67,
+      "learning_rate": 1.6851851851851853e-05,
+      "loss": 0.2501,
+      "step": 4180
+    },
+    {
+      "epoch": 69.83,
+      "learning_rate": 1.675925925925926e-05,
+      "loss": 0.2845,
+      "step": 4190
+    },
+    {
+      "epoch": 70.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.2906,
+      "step": 4200
+    },
+    {
+      "epoch": 70.0,
+      "eval_accuracy": 0.927490871152843,
+      "eval_f1": 0.9259332994297357,
+      "eval_loss": 0.3407656252384186,
+      "eval_roc_auc": 0.9793668129056554,
+      "eval_runtime": 31.0324,
+      "eval_samples_per_second": 61.774,
+      "eval_steps_per_second": 1.933,
+      "step": 4200
+    },
+    {
+      "epoch": 70.17,
+      "learning_rate": 1.6574074074074075e-05,
+      "loss": 0.2927,
+      "step": 4210
+    },
+    {
+      "epoch": 70.33,
+      "learning_rate": 1.6481481481481482e-05,
+      "loss": 0.2689,
+      "step": 4220
+    },
+    {
+      "epoch": 70.5,
+      "learning_rate": 1.638888888888889e-05,
+      "loss": 0.2869,
+      "step": 4230
+    },
+    {
+      "epoch": 70.67,
+      "learning_rate": 1.62962962962963e-05,
+      "loss": 0.2907,
+      "step": 4240
+    },
+    {
+      "epoch": 70.83,
+      "learning_rate": 1.6203703703703704e-05,
+      "loss": 0.3168,
+      "step": 4250
+    },
+    {
+      "epoch": 71.0,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 0.2542,
+      "step": 4260
+    },
+    {
+      "epoch": 71.0,
+      "eval_accuracy": 0.9248826291079812,
+      "eval_f1": 0.9233405493627096,
+      "eval_loss": 0.35591500997543335,
+      "eval_roc_auc": 0.9769011325028507,
+      "eval_runtime": 30.9651,
+      "eval_samples_per_second": 61.908,
+      "eval_steps_per_second": 1.938,
+      "step": 4260
+    },
+    {
+      "epoch": 71.17,
+      "learning_rate": 1.601851851851852e-05,
+      "loss": 0.2411,
+      "step": 4270
+    },
+    {
+      "epoch": 71.33,
+      "learning_rate": 1.5925925925925926e-05,
+      "loss": 0.2807,
+      "step": 4280
+    },
+    {
+      "epoch": 71.5,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.3036,
+      "step": 4290
+    },
+    {
+      "epoch": 71.67,
+      "learning_rate": 1.574074074074074e-05,
+      "loss": 0.2893,
+      "step": 4300
+    },
+    {
+      "epoch": 71.83,
+      "learning_rate": 1.5648148148148148e-05,
+      "loss": 0.2895,
+      "step": 4310
+    },
+    {
+      "epoch": 72.0,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.2557,
+      "step": 4320
+    },
+    {
+      "epoch": 72.0,
+      "eval_accuracy": 0.9254042775169535,
+      "eval_f1": 0.9237261497476135,
+      "eval_loss": 0.3480711877346039,
+      "eval_roc_auc": 0.9779474249307349,
+      "eval_runtime": 30.8763,
+      "eval_samples_per_second": 62.086,
+      "eval_steps_per_second": 1.943,
+      "step": 4320
+    },
+    {
+      "epoch": 72.17,
+      "learning_rate": 1.5462962962962962e-05,
+      "loss": 0.3108,
+      "step": 4330
+    },
+    {
+      "epoch": 72.33,
+      "learning_rate": 1.537037037037037e-05,
+      "loss": 0.294,
+      "step": 4340
+    },
+    {
+      "epoch": 72.5,
+      "learning_rate": 1.527777777777778e-05,
+      "loss": 0.2559,
+      "step": 4350
+    },
+    {
+      "epoch": 72.67,
+      "learning_rate": 1.5185185185185186e-05,
+      "loss": 0.2842,
+      "step": 4360
+    },
+    {
+      "epoch": 72.83,
+      "learning_rate": 1.5092592592592595e-05,
+      "loss": 0.2542,
+      "step": 4370
+    },
+    {
+      "epoch": 73.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.2266,
+      "step": 4380
+    },
+    {
+      "epoch": 73.0,
+      "eval_accuracy": 0.9222743870631195,
+      "eval_f1": 0.9208402796789302,
+      "eval_loss": 0.35181114077568054,
+      "eval_roc_auc": 0.9780920215986283,
+      "eval_runtime": 31.1669,
+      "eval_samples_per_second": 61.507,
+      "eval_steps_per_second": 1.925,
+      "step": 4380
+    },
+    {
+      "epoch": 73.17,
+      "learning_rate": 1.490740740740741e-05,
+      "loss": 0.2967,
+      "step": 4390
+    },
+    {
+      "epoch": 73.33,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.2604,
+      "step": 4400
+    },
+    {
+      "epoch": 73.5,
+      "learning_rate": 1.4722222222222224e-05,
+      "loss": 0.2861,
+      "step": 4410
+    },
+    {
+      "epoch": 73.67,
+      "learning_rate": 1.462962962962963e-05,
+      "loss": 0.2847,
+      "step": 4420
+    },
+    {
+      "epoch": 73.83,
+      "learning_rate": 1.4537037037037039e-05,
+      "loss": 0.2558,
+      "step": 4430
+    },
+    {
+      "epoch": 74.0,
+      "learning_rate": 1.4444444444444444e-05,
+      "loss": 0.2771,
+      "step": 4440
+    },
+    {
+      "epoch": 74.0,
+      "eval_accuracy": 0.9254042775169535,
+      "eval_f1": 0.9230920653274585,
+      "eval_loss": 0.35441046953201294,
+      "eval_roc_auc": 0.9775736826942986,
+      "eval_runtime": 31.0663,
+      "eval_samples_per_second": 61.707,
+      "eval_steps_per_second": 1.931,
+      "step": 4440
+    },
+    {
+      "epoch": 74.17,
+      "learning_rate": 1.4351851851851853e-05,
+      "loss": 0.3149,
+      "step": 4450
+    },
+    {
+      "epoch": 74.33,
+      "learning_rate": 1.425925925925926e-05,
+      "loss": 0.2482,
+      "step": 4460
+    },
+    {
+      "epoch": 74.5,
+      "learning_rate": 1.4166666666666668e-05,
+      "loss": 0.2868,
+      "step": 4470
+    },
+    {
+      "epoch": 74.67,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.2285,
+      "step": 4480
+    },
+    {
+      "epoch": 74.83,
+      "learning_rate": 1.3981481481481482e-05,
+      "loss": 0.2915,
+      "step": 4490
+    },
+    {
+      "epoch": 75.0,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.2747,
+      "step": 4500
+    },
+    {
+      "epoch": 75.0,
+      "eval_accuracy": 0.9285341679707877,
+      "eval_f1": 0.9269627662435742,
+      "eval_loss": 0.34690794348716736,
+      "eval_roc_auc": 0.9780339750030952,
+      "eval_runtime": 30.8369,
+      "eval_samples_per_second": 62.166,
+      "eval_steps_per_second": 1.946,
+      "step": 4500
+    },
+    {
+      "epoch": 75.17,
+      "learning_rate": 1.3796296296296297e-05,
+      "loss": 0.3009,
+      "step": 4510
+    },
+    {
+      "epoch": 75.33,
+      "learning_rate": 1.3703703703703704e-05,
+      "loss": 0.3021,
+      "step": 4520
+    },
+    {
+      "epoch": 75.5,
+      "learning_rate": 1.3611111111111111e-05,
+      "loss": 0.2934,
+      "step": 4530
+    },
+    {
+      "epoch": 75.67,
+      "learning_rate": 1.3518518518518519e-05,
+      "loss": 0.2476,
+      "step": 4540
+    },
+    {
+      "epoch": 75.83,
+      "learning_rate": 1.3425925925925928e-05,
+      "loss": 0.2232,
+      "step": 4550
+    },
+    {
+      "epoch": 76.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.2443,
+      "step": 4560
+    },
+    {
+      "epoch": 76.0,
+      "eval_accuracy": 0.9233176838810642,
+      "eval_f1": 0.9216258234683066,
+      "eval_loss": 0.3513363301753998,
+      "eval_roc_auc": 0.9767332205772213,
+      "eval_runtime": 31.1882,
+      "eval_samples_per_second": 61.466,
+      "eval_steps_per_second": 1.924,
+      "step": 4560
+    },
+    {
+      "epoch": 76.17,
+      "learning_rate": 1.3240740740740742e-05,
+      "loss": 0.2741,
+      "step": 4570
+    },
+    {
+      "epoch": 76.33,
+      "learning_rate": 1.3148148148148148e-05,
+      "loss": 0.2708,
+      "step": 4580
+    },
+    {
+      "epoch": 76.5,
+      "learning_rate": 1.3055555555555557e-05,
+      "loss": 0.2803,
+      "step": 4590
+    },
+    {
+      "epoch": 76.67,
+      "learning_rate": 1.2962962962962962e-05,
+      "loss": 0.2612,
+      "step": 4600
+    },
+    {
+      "epoch": 76.83,
+      "learning_rate": 1.2870370370370371e-05,
+      "loss": 0.277,
+      "step": 4610
+    },
+    {
+      "epoch": 77.0,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 0.2859,
+      "step": 4620
+    },
+    {
+      "epoch": 77.0,
+      "eval_accuracy": 0.9254042775169535,
+      "eval_f1": 0.9234239757485955,
+      "eval_loss": 0.3456012010574341,
+      "eval_roc_auc": 0.9771096750183106,
+      "eval_runtime": 31.0209,
+      "eval_samples_per_second": 61.797,
+      "eval_steps_per_second": 1.934,
+      "step": 4620
+    },
+    {
+      "epoch": 77.17,
+      "learning_rate": 1.2685185185185186e-05,
+      "loss": 0.2851,
+      "step": 4630
+    },
+    {
+      "epoch": 77.33,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 0.2572,
+      "step": 4640
+    },
+    {
+      "epoch": 77.5,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2639,
+      "step": 4650
+    },
+    {
+      "epoch": 77.67,
+      "learning_rate": 1.2407407407407408e-05,
+      "loss": 0.2768,
+      "step": 4660
+    },
+    {
+      "epoch": 77.83,
+      "learning_rate": 1.2314814814814815e-05,
+      "loss": 0.3115,
+      "step": 4670
+    },
+    {
+      "epoch": 78.0,
+      "learning_rate": 1.2222222222222222e-05,
+      "loss": 0.2677,
+      "step": 4680
+    },
+    {
+      "epoch": 78.0,
+      "eval_accuracy": 0.9254042775169535,
+      "eval_f1": 0.9238967953597521,
+      "eval_loss": 0.34742477536201477,
+      "eval_roc_auc": 0.9780192714539518,
+      "eval_runtime": 30.9011,
+      "eval_samples_per_second": 62.037,
+      "eval_steps_per_second": 1.942,
+      "step": 4680
+    },
+    {
+      "epoch": 78.17,
+      "learning_rate": 1.212962962962963e-05,
+      "loss": 0.3025,
+      "step": 4690
+    },
+    {
+      "epoch": 78.33,
+      "learning_rate": 1.2037037037037037e-05,
+      "loss": 0.3503,
+      "step": 4700
+    },
+    {
+      "epoch": 78.5,
+      "learning_rate": 1.1944444444444446e-05,
+      "loss": 0.2569,
+      "step": 4710
+    },
+    {
+      "epoch": 78.67,
+      "learning_rate": 1.1851851851851853e-05,
+      "loss": 0.2529,
+      "step": 4720
+    },
+    {
+      "epoch": 78.83,
+      "learning_rate": 1.175925925925926e-05,
+      "loss": 0.2804,
+      "step": 4730
+    },
+    {
+      "epoch": 79.0,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.2492,
+      "step": 4740
+    },
+    {
+      "epoch": 79.0,
+      "eval_accuracy": 0.9254042775169535,
+      "eval_f1": 0.9234698228772668,
+      "eval_loss": 0.35134339332580566,
+      "eval_roc_auc": 0.9777977841330655,
+      "eval_runtime": 30.9883,
+      "eval_samples_per_second": 61.862,
+      "eval_steps_per_second": 1.936,
+      "step": 4740
+    },
+    {
+      "epoch": 79.17,
+      "learning_rate": 1.1574074074074075e-05,
+      "loss": 0.265,
+      "step": 4750
+    },
+    {
+      "epoch": 79.33,
+      "learning_rate": 1.1481481481481482e-05,
+      "loss": 0.2664,
+      "step": 4760
+    },
+    {
+      "epoch": 79.5,
+      "learning_rate": 1.138888888888889e-05,
+      "loss": 0.2669,
+      "step": 4770
+    },
+    {
+      "epoch": 79.67,
+      "learning_rate": 1.1296296296296297e-05,
+      "loss": 0.2859,
+      "step": 4780
+    },
+    {
+      "epoch": 79.83,
+      "learning_rate": 1.1203703703703704e-05,
+      "loss": 0.2518,
+      "step": 4790
+    },
+    {
+      "epoch": 80.0,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.2532,
+      "step": 4800
+    },
+    {
+      "epoch": 80.0,
+      "eval_accuracy": 0.9233176838810642,
+      "eval_f1": 0.9210239732637377,
+      "eval_loss": 0.3524177074432373,
+      "eval_roc_auc": 0.9773028713209709,
+      "eval_runtime": 30.9122,
+      "eval_samples_per_second": 62.014,
+      "eval_steps_per_second": 1.941,
+      "step": 4800
+    },
+    {
+      "epoch": 80.17,
+      "learning_rate": 1.1018518518518519e-05,
+      "loss": 0.2829,
+      "step": 4810
+    },
+    {
+      "epoch": 80.33,
+      "learning_rate": 1.0925925925925926e-05,
+      "loss": 0.2988,
+      "step": 4820
+    },
+    {
+      "epoch": 80.5,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.2675,
+      "step": 4830
+    },
+    {
+      "epoch": 80.67,
+      "learning_rate": 1.074074074074074e-05,
+      "loss": 0.232,
+      "step": 4840
+    },
+    {
+      "epoch": 80.83,
+      "learning_rate": 1.0648148148148148e-05,
+      "loss": 0.3254,
+      "step": 4850
+    },
+    {
+      "epoch": 81.0,
+      "learning_rate": 1.0555555555555555e-05,
+      "loss": 0.2646,
+      "step": 4860
+    },
+    {
+      "epoch": 81.0,
+      "eval_accuracy": 0.9238393322900366,
+      "eval_f1": 0.9239580070656279,
+      "eval_loss": 0.35290443897247314,
+      "eval_roc_auc": 0.9783679016873998,
+      "eval_runtime": 30.9829,
+      "eval_samples_per_second": 61.873,
+      "eval_steps_per_second": 1.937,
+      "step": 4860
+    },
+    {
+      "epoch": 81.17,
+      "learning_rate": 1.0462962962962963e-05,
+      "loss": 0.2492,
+      "step": 4870
+    },
+    {
+      "epoch": 81.33,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.2775,
+      "step": 4880
+    },
+    {
+      "epoch": 81.5,
+      "learning_rate": 1.0277777777777777e-05,
+      "loss": 0.2573,
+      "step": 4890
+    },
+    {
+      "epoch": 81.67,
+      "learning_rate": 1.0185185185185185e-05,
+      "loss": 0.2504,
+      "step": 4900
+    },
+    {
+      "epoch": 81.83,
+      "learning_rate": 1.0092592592592594e-05,
+      "loss": 0.31,
+      "step": 4910
+    },
+    {
+      "epoch": 82.0,
+      "learning_rate": 1e-05,
+      "loss": 0.2842,
+      "step": 4920
+    },
+    {
+      "epoch": 82.0,
+      "eval_accuracy": 0.9280125195618153,
+      "eval_f1": 0.9260333195816027,
+      "eval_loss": 0.34334856271743774,
+      "eval_roc_auc": 0.977665008318065,
+      "eval_runtime": 31.0483,
+      "eval_samples_per_second": 61.742,
+      "eval_steps_per_second": 1.932,
+      "step": 4920
+    },
+    {
+      "epoch": 82.17,
+      "learning_rate": 9.907407407407408e-06,
+      "loss": 0.2765,
+      "step": 4930
+    },
+    {
+      "epoch": 82.33,
+      "learning_rate": 9.814814814814815e-06,
+      "loss": 0.3014,
+      "step": 4940
+    },
+    {
+      "epoch": 82.5,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 0.2825,
+      "step": 4950
+    },
+    {
+      "epoch": 82.67,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.2874,
+      "step": 4960
+    },
+    {
+      "epoch": 82.83,
+      "learning_rate": 9.537037037037037e-06,
+      "loss": 0.2796,
+      "step": 4970
+    },
+    {
+      "epoch": 83.0,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 0.2872,
+      "step": 4980
+    },
+    {
+      "epoch": 83.0,
+      "eval_accuracy": 0.9285341679707877,
+      "eval_f1": 0.9271958003115927,
+      "eval_loss": 0.35838404297828674,
+      "eval_roc_auc": 0.9770701360759796,
+      "eval_runtime": 31.004,
+      "eval_samples_per_second": 61.831,
+      "eval_steps_per_second": 1.935,
+      "step": 4980
+    },
+    {
+      "epoch": 83.17,
+      "learning_rate": 9.351851851851852e-06,
+      "loss": 0.2552,
+      "step": 4990
+    },
+    {
+      "epoch": 83.33,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.2755,
+      "step": 5000
+    },
+    {
+      "epoch": 83.5,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.3196,
+      "step": 5010
+    },
+    {
+      "epoch": 83.67,
+      "learning_rate": 9.074074074074075e-06,
+      "loss": 0.2738,
+      "step": 5020
+    },
+    {
+      "epoch": 83.83,
+      "learning_rate": 8.981481481481483e-06,
+      "loss": 0.2764,
+      "step": 5030
+    },
+    {
+      "epoch": 84.0,
+      "learning_rate": 8.88888888888889e-06,
+      "loss": 0.2678,
+      "step": 5040
+    },
+    {
+      "epoch": 84.0,
+      "eval_accuracy": 0.9316640584246219,
+      "eval_f1": 0.9297928229609359,
+      "eval_loss": 0.3429672122001648,
+      "eval_roc_auc": 0.9777372174983401,
+      "eval_runtime": 30.8771,
+      "eval_samples_per_second": 62.085,
+      "eval_steps_per_second": 1.943,
+      "step": 5040
+    },
+    {
+      "epoch": 84.17,
+      "learning_rate": 8.796296296296297e-06,
+      "loss": 0.2943,
+      "step": 5050
+    },
+    {
+      "epoch": 84.33,
+      "learning_rate": 8.703703703703705e-06,
+      "loss": 0.2491,
+      "step": 5060
+    },
+    {
+      "epoch": 84.5,
+      "learning_rate": 8.611111111111112e-06,
+      "loss": 0.2702,
+      "step": 5070
+    },
+    {
+      "epoch": 84.67,
+      "learning_rate": 8.518518518518519e-06,
+      "loss": 0.2818,
+      "step": 5080
+    },
+    {
+      "epoch": 84.83,
+      "learning_rate": 8.425925925925926e-06,
+      "loss": 0.2681,
+      "step": 5090
+    },
+    {
+      "epoch": 85.0,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.2705,
+      "step": 5100
+    },
+    {
+      "epoch": 85.0,
+      "eval_accuracy": 0.9290558163797601,
+      "eval_f1": 0.9268182369344703,
+      "eval_loss": 0.3533850312232971,
+      "eval_roc_auc": 0.9777106946227319,
+      "eval_runtime": 30.756,
+      "eval_samples_per_second": 62.329,
+      "eval_steps_per_second": 1.951,
+      "step": 5100
+    },
+    {
+      "epoch": 85.17,
+      "learning_rate": 8.240740740740741e-06,
+      "loss": 0.2345,
+      "step": 5110
+    },
+    {
+      "epoch": 85.33,
+      "learning_rate": 8.14814814814815e-06,
+      "loss": 0.2591,
+      "step": 5120
+    },
+    {
+      "epoch": 85.5,
+      "learning_rate": 8.055555555555557e-06,
+      "loss": 0.2629,
+      "step": 5130
+    },
+    {
+      "epoch": 85.67,
+      "learning_rate": 7.962962962962963e-06,
+      "loss": 0.2509,
+      "step": 5140
+    },
+    {
+      "epoch": 85.83,
+      "learning_rate": 7.87037037037037e-06,
+      "loss": 0.2657,
+      "step": 5150
+    },
+    {
+      "epoch": 86.0,
+      "learning_rate": 7.777777777777777e-06,
+      "loss": 0.2605,
+      "step": 5160
+    },
+    {
+      "epoch": 86.0,
+      "eval_accuracy": 0.9295774647887324,
+      "eval_f1": 0.9271674219062209,
+      "eval_loss": 0.35735803842544556,
+      "eval_roc_auc": 0.9776619601955235,
+      "eval_runtime": 31.1184,
+      "eval_samples_per_second": 61.603,
+      "eval_steps_per_second": 1.928,
+      "step": 5160
+    },
+    {
+      "epoch": 86.17,
+      "learning_rate": 7.685185185185185e-06,
+      "loss": 0.274,
+      "step": 5170
+    },
+    {
+      "epoch": 86.33,
+      "learning_rate": 7.592592592592593e-06,
+      "loss": 0.2786,
+      "step": 5180
+    },
+    {
+      "epoch": 86.5,
+      "learning_rate": 7.5e-06,
+      "loss": 0.2748,
+      "step": 5190
+    },
+    {
+      "epoch": 86.67,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.2589,
+      "step": 5200
+    },
+    {
+      "epoch": 86.83,
+      "learning_rate": 7.314814814814815e-06,
+      "loss": 0.2675,
+      "step": 5210
+    },
+    {
+      "epoch": 87.0,
+      "learning_rate": 7.222222222222222e-06,
+      "loss": 0.2572,
+      "step": 5220
+    },
+    {
+      "epoch": 87.0,
+      "eval_accuracy": 0.9290558163797601,
+      "eval_f1": 0.927251726970012,
+      "eval_loss": 0.34258973598480225,
+      "eval_roc_auc": 0.9780606091514321,
+      "eval_runtime": 31.1961,
+      "eval_samples_per_second": 61.45,
+      "eval_steps_per_second": 1.923,
+      "step": 5220
+    },
+    {
+      "epoch": 87.17,
+      "learning_rate": 7.12962962962963e-06,
+      "loss": 0.2674,
+      "step": 5230
+    },
+    {
+      "epoch": 87.33,
+      "learning_rate": 7.0370370370370375e-06,
+      "loss": 0.2552,
+      "step": 5240
+    },
+    {
+      "epoch": 87.5,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.2634,
+      "step": 5250
+    },
+    {
+      "epoch": 87.67,
+      "learning_rate": 6.851851851851852e-06,
+      "loss": 0.2737,
+      "step": 5260
+    },
+    {
+      "epoch": 87.83,
+      "learning_rate": 6.759259259259259e-06,
+      "loss": 0.2788,
+      "step": 5270
+    },
+    {
+      "epoch": 88.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.2646,
+      "step": 5280
+    },
+    {
+      "epoch": 88.0,
+      "eval_accuracy": 0.9243609806990088,
+      "eval_f1": 0.9233615308327541,
+      "eval_loss": 0.3472360372543335,
+      "eval_roc_auc": 0.978910913704422,
+      "eval_runtime": 30.9382,
+      "eval_samples_per_second": 61.962,
+      "eval_steps_per_second": 1.939,
+      "step": 5280
+    },
+    {
+      "epoch": 88.17,
+      "learning_rate": 6.574074074074074e-06,
+      "loss": 0.2603,
+      "step": 5290
+    },
+    {
+      "epoch": 88.33,
+      "learning_rate": 6.481481481481481e-06,
+      "loss": 0.2721,
+      "step": 5300
+    },
+    {
+      "epoch": 88.5,
+      "learning_rate": 6.3888888888888885e-06,
+      "loss": 0.2864,
+      "step": 5310
+    },
+    {
+      "epoch": 88.67,
+      "learning_rate": 6.296296296296296e-06,
+      "loss": 0.2517,
+      "step": 5320
+    },
+    {
+      "epoch": 88.83,
+      "learning_rate": 6.203703703703704e-06,
+      "loss": 0.2494,
+      "step": 5330
+    },
+    {
+      "epoch": 89.0,
+      "learning_rate": 6.111111111111111e-06,
+      "loss": 0.2831,
+      "step": 5340
+    },
+    {
+      "epoch": 89.0,
+      "eval_accuracy": 0.9290558163797601,
+      "eval_f1": 0.9272407593779144,
+      "eval_loss": 0.34334689378738403,
+      "eval_roc_auc": 0.9778616671938408,
+      "eval_runtime": 30.9945,
+      "eval_samples_per_second": 61.85,
+      "eval_steps_per_second": 1.936,
+      "step": 5340
+    },
+    {
+      "epoch": 89.17,
+      "learning_rate": 6.0185185185185185e-06,
+      "loss": 0.2497,
+      "step": 5350
+    },
+    {
+      "epoch": 89.33,
+      "learning_rate": 5.925925925925927e-06,
+      "loss": 0.2433,
+      "step": 5360
+    },
+    {
+      "epoch": 89.5,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.2521,
+      "step": 5370
+    },
+    {
+      "epoch": 89.67,
+      "learning_rate": 5.740740740740741e-06,
+      "loss": 0.2343,
+      "step": 5380
+    },
+    {
+      "epoch": 89.83,
+      "learning_rate": 5.6481481481481485e-06,
+      "loss": 0.2693,
+      "step": 5390
+    },
+    {
+      "epoch": 90.0,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.277,
+      "step": 5400
+    },
+    {
+      "epoch": 90.0,
+      "eval_accuracy": 0.9280125195618153,
+      "eval_f1": 0.9262715742423608,
+      "eval_loss": 0.3440592885017395,
+      "eval_roc_auc": 0.9788712016876208,
+      "eval_runtime": 31.049,
+      "eval_samples_per_second": 61.741,
+      "eval_steps_per_second": 1.932,
+      "step": 5400
+    },
+    {
+      "epoch": 90.17,
+      "learning_rate": 5.462962962962963e-06,
+      "loss": 0.2516,
+      "step": 5410
+    },
+    {
+      "epoch": 90.33,
+      "learning_rate": 5.37037037037037e-06,
+      "loss": 0.278,
+      "step": 5420
+    },
+    {
+      "epoch": 90.5,
+      "learning_rate": 5.277777777777778e-06,
+      "loss": 0.2603,
+      "step": 5430
+    },
+    {
+      "epoch": 90.67,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.244,
+      "step": 5440
+    },
+    {
+      "epoch": 90.83,
+      "learning_rate": 5.092592592592592e-06,
+      "loss": 0.2609,
+      "step": 5450
+    },
+    {
+      "epoch": 91.0,
+      "learning_rate": 5e-06,
+      "loss": 0.2584,
+      "step": 5460
+    },
+    {
+      "epoch": 91.0,
+      "eval_accuracy": 0.9248826291079812,
+      "eval_f1": 0.9236426065868731,
+      "eval_loss": 0.3431851267814636,
+      "eval_roc_auc": 0.9787553845371705,
+      "eval_runtime": 30.977,
+      "eval_samples_per_second": 61.885,
+      "eval_steps_per_second": 1.937,
+      "step": 5460
+    },
+    {
+      "epoch": 91.17,
+      "learning_rate": 4.907407407407408e-06,
+      "loss": 0.2829,
+      "step": 5470
+    },
+    {
+      "epoch": 91.33,
+      "learning_rate": 4.814814814814815e-06,
+      "loss": 0.2694,
+      "step": 5480
+    },
+    {
+      "epoch": 91.5,
+      "learning_rate": 4.722222222222222e-06,
+      "loss": 0.2226,
+      "step": 5490
+    },
+    {
+      "epoch": 91.67,
+      "learning_rate": 4.6296296296296296e-06,
+      "loss": 0.2856,
+      "step": 5500
+    },
+    {
+      "epoch": 91.83,
+      "learning_rate": 4.537037037037038e-06,
+      "loss": 0.2646,
+      "step": 5510
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 4.444444444444445e-06,
+      "loss": 0.2703,
+      "step": 5520
+    },
+    {
+      "epoch": 92.0,
+      "eval_accuracy": 0.9259259259259259,
+      "eval_f1": 0.9247749239486609,
+      "eval_loss": 0.3409390151500702,
+      "eval_roc_auc": 0.9789125551817118,
+      "eval_runtime": 31.0632,
+      "eval_samples_per_second": 61.713,
+      "eval_steps_per_second": 1.932,
+      "step": 5520
+    },
+    {
+      "epoch": 92.17,
+      "learning_rate": 4.351851851851852e-06,
+      "loss": 0.2429,
+      "step": 5530
+    },
+    {
+      "epoch": 92.33,
+      "learning_rate": 4.2592592592592596e-06,
+      "loss": 0.2819,
+      "step": 5540
+    },
+    {
+      "epoch": 92.5,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.2748,
+      "step": 5550
+    },
+    {
+      "epoch": 92.67,
+      "learning_rate": 4.074074074074075e-06,
+      "loss": 0.2526,
+      "step": 5560
+    },
+    {
+      "epoch": 92.83,
+      "learning_rate": 3.9814814814814814e-06,
+      "loss": 0.2468,
+      "step": 5570
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.888888888888889e-06,
+      "loss": 0.2811,
+      "step": 5580
+    },
+    {
+      "epoch": 93.0,
+      "eval_accuracy": 0.9227960354720918,
+      "eval_f1": 0.9214672904997486,
+      "eval_loss": 0.34487679600715637,
+      "eval_roc_auc": 0.9795045467703533,
+      "eval_runtime": 30.8936,
+      "eval_samples_per_second": 62.052,
+      "eval_steps_per_second": 1.942,
+      "step": 5580
+    },
+    {
+      "epoch": 93.17,
+      "learning_rate": 3.7962962962962964e-06,
+      "loss": 0.2401,
+      "step": 5590
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.262,
+      "step": 5600
+    },
+    {
+      "epoch": 93.5,
+      "learning_rate": 3.611111111111111e-06,
+      "loss": 0.2586,
+      "step": 5610
+    },
+    {
+      "epoch": 93.67,
+      "learning_rate": 3.5185185185185187e-06,
+      "loss": 0.2708,
+      "step": 5620
+    },
+    {
+      "epoch": 93.83,
+      "learning_rate": 3.425925925925926e-06,
+      "loss": 0.2627,
+      "step": 5630
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.2786,
+      "step": 5640
+    },
+    {
+      "epoch": 94.0,
+      "eval_accuracy": 0.9280125195618153,
+      "eval_f1": 0.925961217058845,
+      "eval_loss": 0.34653154015541077,
+      "eval_roc_auc": 0.9788772928385914,
+      "eval_runtime": 30.9745,
+      "eval_samples_per_second": 61.89,
+      "eval_steps_per_second": 1.937,
+      "step": 5640
+    },
+    {
+      "epoch": 94.17,
+      "learning_rate": 3.2407407407407406e-06,
+      "loss": 0.2449,
+      "step": 5650
+    },
+    {
+      "epoch": 94.33,
+      "learning_rate": 3.148148148148148e-06,
+      "loss": 0.2742,
+      "step": 5660
+    },
+    {
+      "epoch": 94.5,
+      "learning_rate": 3.0555555555555556e-06,
+      "loss": 0.2185,
+      "step": 5670
+    },
+    {
+      "epoch": 94.67,
+      "learning_rate": 2.9629629629629633e-06,
+      "loss": 0.2374,
+      "step": 5680
+    },
+    {
+      "epoch": 94.83,
+      "learning_rate": 2.8703703703703706e-06,
+      "loss": 0.2448,
+      "step": 5690
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 0.267,
+      "step": 5700
+    },
+    {
+      "epoch": 95.0,
+      "eval_accuracy": 0.927490871152843,
+      "eval_f1": 0.926020550825123,
+      "eval_loss": 0.3472473919391632,
+      "eval_roc_auc": 0.9791268419832962,
+      "eval_runtime": 31.0494,
+      "eval_samples_per_second": 61.74,
+      "eval_steps_per_second": 1.932,
+      "step": 5700
+    },
+    {
+      "epoch": 95.17,
+      "learning_rate": 2.685185185185185e-06,
+      "loss": 0.2583,
+      "step": 5710
+    },
+    {
+      "epoch": 95.33,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 0.2697,
+      "step": 5720
+    },
+    {
+      "epoch": 95.5,
+      "learning_rate": 2.5e-06,
+      "loss": 0.2727,
+      "step": 5730
+    },
+    {
+      "epoch": 95.67,
+      "learning_rate": 2.4074074074074075e-06,
+      "loss": 0.244,
+      "step": 5740
+    },
+    {
+      "epoch": 95.83,
+      "learning_rate": 2.3148148148148148e-06,
+      "loss": 0.2818,
+      "step": 5750
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 2.2222222222222225e-06,
+      "loss": 0.2695,
+      "step": 5760
+    },
+    {
+      "epoch": 96.0,
+      "eval_accuracy": 0.9285341679707877,
+      "eval_f1": 0.926806533960345,
+      "eval_loss": 0.34995850920677185,
+      "eval_roc_auc": 0.9785719343176981,
+      "eval_runtime": 30.9084,
+      "eval_samples_per_second": 62.022,
+      "eval_steps_per_second": 1.941,
+      "step": 5760
+    },
+    {
+      "epoch": 96.17,
+      "learning_rate": 2.1296296296296298e-06,
+      "loss": 0.2907,
+      "step": 5770
+    },
+    {
+      "epoch": 96.33,
+      "learning_rate": 2.0370370370370375e-06,
+      "loss": 0.2722,
+      "step": 5780
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 1.9444444444444444e-06,
+      "loss": 0.2875,
+      "step": 5790
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 0.2591,
+      "step": 5800
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 1.7592592592592594e-06,
+      "loss": 0.2508,
+      "step": 5810
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.279,
+      "step": 5820
+    },
+    {
+      "epoch": 97.0,
+      "eval_accuracy": 0.9269692227438706,
+      "eval_f1": 0.9248843021081147,
+      "eval_loss": 0.35818785429000854,
+      "eval_roc_auc": 0.9782476583971591,
+      "eval_runtime": 30.9227,
+      "eval_samples_per_second": 61.993,
+      "eval_steps_per_second": 1.94,
+      "step": 5820
+    },
+    {
+      "epoch": 97.17,
+      "learning_rate": 1.574074074074074e-06,
+      "loss": 0.2663,
+      "step": 5830
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 1.4814814814814817e-06,
+      "loss": 0.2585,
+      "step": 5840
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 1.388888888888889e-06,
+      "loss": 0.255,
+      "step": 5850
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 1.2962962962962962e-06,
+      "loss": 0.2737,
+      "step": 5860
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 1.2037037037037037e-06,
+      "loss": 0.271,
+      "step": 5870
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 1.1111111111111112e-06,
+      "loss": 0.2774,
+      "step": 5880
+    },
+    {
+      "epoch": 98.0,
+      "eval_accuracy": 0.9269692227438706,
+      "eval_f1": 0.9250939587511013,
+      "eval_loss": 0.3485565483570099,
+      "eval_roc_auc": 0.9790459422116572,
+      "eval_runtime": 31.1331,
+      "eval_samples_per_second": 61.574,
+      "eval_steps_per_second": 1.927,
+      "step": 5880
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 1.0185185185185188e-06,
+      "loss": 0.2523,
+      "step": 5890
+    },
+    {
+      "epoch": 98.33,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 0.304,
+      "step": 5900
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 0.254,
+      "step": 5910
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 7.407407407407408e-07,
+      "loss": 0.2096,
+      "step": 5920
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 6.481481481481481e-07,
+      "loss": 0.2487,
+      "step": 5930
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 5.555555555555556e-07,
+      "loss": 0.2512,
+      "step": 5940
+    },
+    {
+      "epoch": 99.0,
+      "eval_accuracy": 0.9306207616066771,
+      "eval_f1": 0.9287006146320661,
+      "eval_loss": 0.3514097332954407,
+      "eval_roc_auc": 0.9786490635988997,
+      "eval_runtime": 30.7798,
+      "eval_samples_per_second": 62.281,
+      "eval_steps_per_second": 1.949,
+      "step": 5940
+    },
+    {
+      "epoch": 99.17,
+      "learning_rate": 4.6296296296296297e-07,
+      "loss": 0.2616,
+      "step": 5950
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 3.703703703703704e-07,
+      "loss": 0.2575,
+      "step": 5960
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 2.777777777777778e-07,
+      "loss": 0.2419,
+      "step": 5970
+    },
+    {
+      "epoch": 99.67,
+      "learning_rate": 1.851851851851852e-07,
+      "loss": 0.2778,
+      "step": 5980
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 9.25925925925926e-08,
+      "loss": 0.2805,
+      "step": 5990
+    },
+    {
+      "epoch": 100.0,
+      "learning_rate": 0.0,
+      "loss": 0.2218,
+      "step": 6000
+    },
+    {
+      "epoch": 100.0,
+      "eval_accuracy": 0.9285341679707877,
+      "eval_f1": 0.9269190721647167,
+      "eval_loss": 0.34819331765174866,
+      "eval_roc_auc": 0.9788880411666135,
+      "eval_runtime": 31.0647,
+      "eval_samples_per_second": 61.71,
+      "eval_steps_per_second": 1.931,
+      "step": 6000
+    },
+    {
+      "epoch": 100.0,
+      "step": 6000,
+      "total_flos": 1.357738787414016e+19,
+      "train_loss": 0.3802111453215281,
+      "train_runtime": 17260.548,
+      "train_samples_per_second": 44.402,
+      "train_steps_per_second": 0.348
+    }
+  ],
+  "max_steps": 6000,
+  "num_train_epochs": 100,
+  "total_flos": 1.357738787414016e+19,
+  "trial_name": null,
+  "trial_params": null
+}