diff --git "a/.ipynb_checkpoints/trainer_state-checkpoint.json" "b/.ipynb_checkpoints/trainer_state-checkpoint.json"
new file mode 100644--- /dev/null
+++ "b/.ipynb_checkpoints/trainer_state-checkpoint.json"
@@ -0,0 +1,7198 @@
+{
+  "best_metric": 0.9261042886093471,
+  "best_model_checkpoint": "videomae-surf-analytics-runpod7/checkpoint-8463",
+  "epoch": 5.002391304347826,
+  "eval_steps": 500,
+  "global_step": 8857,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0004347826086956522,
+      "grad_norm": 10.079814910888672,
+      "learning_rate": 2.173913043478261e-07,
+      "loss": 1.3951,
+      "step": 10
+    },
+    {
+      "epoch": 0.0008695652173913044,
+      "grad_norm": 9.322056770324707,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 1.3872,
+      "step": 20
+    },
+    {
+      "epoch": 0.0013043478260869566,
+      "grad_norm": 6.732794761657715,
+      "learning_rate": 6.521739130434782e-07,
+      "loss": 1.3724,
+      "step": 30
+    },
+    {
+      "epoch": 0.0017391304347826088,
+      "grad_norm": 12.304842948913574,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 1.3522,
+      "step": 40
+    },
+    {
+      "epoch": 0.002173913043478261,
+      "grad_norm": 8.10595703125,
+      "learning_rate": 1.0869565217391306e-06,
+      "loss": 1.3382,
+      "step": 50
+    },
+    {
+      "epoch": 0.0026086956521739132,
+      "grad_norm": 6.890604019165039,
+      "learning_rate": 1.3043478260869564e-06,
+      "loss": 1.3565,
+      "step": 60
+    },
+    {
+      "epoch": 0.003043478260869565,
+      "grad_norm": 11.639082908630371,
+      "learning_rate": 1.5217391304347827e-06,
+      "loss": 1.2803,
+      "step": 70
+    },
+    {
+      "epoch": 0.0034782608695652175,
+      "grad_norm": 9.563755989074707,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 1.2934,
+      "step": 80
+    },
+    {
+      "epoch": 0.00391304347826087,
+      "grad_norm": 5.474796295166016,
+      "learning_rate": 1.956521739130435e-06,
+      "loss": 1.2803,
+      "step": 90
+    },
+    {
+      "epoch": 0.004043478260869565,
+      "eval_accuracy": 0.4426229508196721,
+      "eval_f1": 0.28327868852459015,
+      "eval_loss": 1.2739324569702148,
+      "eval_runtime": 99.1883,
+      "eval_samples_per_second": 1.23,
+      "eval_steps_per_second": 0.161,
+      "step": 93
+    },
+    {
+      "epoch": 1.000304347826087,
+      "grad_norm": 4.489353179931641,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 1.2057,
+      "step": 100
+    },
+    {
+      "epoch": 1.0007391304347826,
+      "grad_norm": 4.060410976409912,
+      "learning_rate": 2.391304347826087e-06,
+      "loss": 1.2485,
+      "step": 110
+    },
+    {
+      "epoch": 1.0011739130434782,
+      "grad_norm": 6.812442302703857,
+      "learning_rate": 2.608695652173913e-06,
+      "loss": 1.1412,
+      "step": 120
+    },
+    {
+      "epoch": 1.001608695652174,
+      "grad_norm": 6.631100177764893,
+      "learning_rate": 2.8260869565217393e-06,
+      "loss": 1.1666,
+      "step": 130
+    },
+    {
+      "epoch": 1.0020434782608696,
+      "grad_norm": 5.362147331237793,
+      "learning_rate": 3.0434782608695654e-06,
+      "loss": 1.1275,
+      "step": 140
+    },
+    {
+      "epoch": 1.0024782608695653,
+      "grad_norm": 6.68129301071167,
+      "learning_rate": 3.2608695652173914e-06,
+      "loss": 1.1033,
+      "step": 150
+    },
+    {
+      "epoch": 1.002913043478261,
+      "grad_norm": 13.178942680358887,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 1.1418,
+      "step": 160
+    },
+    {
+      "epoch": 1.0033478260869566,
+      "grad_norm": 13.012588500976562,
+      "learning_rate": 3.695652173913044e-06,
+      "loss": 1.265,
+      "step": 170
+    },
+    {
+      "epoch": 1.0037826086956523,
+      "grad_norm": 5.870707988739014,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 1.0844,
+      "step": 180
+    },
+    {
+      "epoch": 1.0040434782608696,
+      "eval_accuracy": 0.45081967213114754,
+      "eval_f1": 0.285264425303325,
+      "eval_loss": 1.186285376548767,
+      "eval_runtime": 74.0993,
+      "eval_samples_per_second": 1.646,
+      "eval_steps_per_second": 0.216,
+      "step": 186
+    },
+    {
+      "epoch": 2.0001739130434784,
+      "grad_norm": 6.740423679351807,
+      "learning_rate": 4.130434782608695e-06,
+      "loss": 1.1175,
+      "step": 190
+    },
+    {
+      "epoch": 2.000608695652174,
+      "grad_norm": 8.124707221984863,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 1.0575,
+      "step": 200
+    },
+    {
+      "epoch": 2.0010434782608697,
+      "grad_norm": 7.8553266525268555,
+      "learning_rate": 4.565217391304348e-06,
+      "loss": 1.0517,
+      "step": 210
+    },
+    {
+      "epoch": 2.001478260869565,
+      "grad_norm": 9.870964050292969,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 1.1031,
+      "step": 220
+    },
+    {
+      "epoch": 2.001913043478261,
+      "grad_norm": 8.00858211517334,
+      "learning_rate": 5e-06,
+      "loss": 0.9564,
+      "step": 230
+    },
+    {
+      "epoch": 2.0023478260869565,
+      "grad_norm": 8.030350685119629,
+      "learning_rate": 5.217391304347826e-06,
+      "loss": 1.0014,
+      "step": 240
+    },
+    {
+      "epoch": 2.0027826086956524,
+      "grad_norm": 6.9377522468566895,
+      "learning_rate": 5.4347826086956525e-06,
+      "loss": 0.9769,
+      "step": 250
+    },
+    {
+      "epoch": 2.003217391304348,
+      "grad_norm": 6.949226379394531,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 1.0181,
+      "step": 260
+    },
+    {
+      "epoch": 2.0036521739130433,
+      "grad_norm": 8.180449485778809,
+      "learning_rate": 5.869565217391305e-06,
+      "loss": 0.9224,
+      "step": 270
+    },
+    {
+      "epoch": 2.0040434782608694,
+      "eval_accuracy": 0.5245901639344263,
+      "eval_f1": 0.4366849047835448,
+      "eval_loss": 1.0750652551651,
+      "eval_runtime": 69.4086,
+      "eval_samples_per_second": 1.758,
+      "eval_steps_per_second": 0.231,
+      "step": 279
+    },
+    {
+      "epoch": 3.0000434782608694,
+      "grad_norm": 6.714990139007568,
+      "learning_rate": 6.086956521739131e-06,
+      "loss": 0.9395,
+      "step": 280
+    },
+    {
+      "epoch": 3.0004782608695653,
+      "grad_norm": 9.589141845703125,
+      "learning_rate": 6.304347826086957e-06,
+      "loss": 0.9219,
+      "step": 290
+    },
+    {
+      "epoch": 3.0009130434782607,
+      "grad_norm": 17.355304718017578,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 0.764,
+      "step": 300
+    },
+    {
+      "epoch": 3.0013478260869566,
+      "grad_norm": 14.70441722869873,
+      "learning_rate": 6.739130434782609e-06,
+      "loss": 0.9942,
+      "step": 310
+    },
+    {
+      "epoch": 3.001782608695652,
+      "grad_norm": 8.38525104522705,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 0.9428,
+      "step": 320
+    },
+    {
+      "epoch": 3.002217391304348,
+      "grad_norm": 9.241342544555664,
+      "learning_rate": 7.173913043478261e-06,
+      "loss": 0.8697,
+      "step": 330
+    },
+    {
+      "epoch": 3.0026521739130434,
+      "grad_norm": 20.59735107421875,
+      "learning_rate": 7.391304347826088e-06,
+      "loss": 0.7409,
+      "step": 340
+    },
+    {
+      "epoch": 3.0030869565217393,
+      "grad_norm": 9.710801124572754,
+      "learning_rate": 7.608695652173914e-06,
+      "loss": 0.8818,
+      "step": 350
+    },
+    {
+      "epoch": 3.0035217391304347,
+      "grad_norm": 8.560890197753906,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 0.8719,
+      "step": 360
+    },
+    {
+      "epoch": 3.0039565217391306,
+      "grad_norm": 6.683012962341309,
+      "learning_rate": 8.043478260869565e-06,
+      "loss": 0.7226,
+      "step": 370
+    },
+    {
+      "epoch": 3.0040434782608694,
+      "eval_accuracy": 0.5655737704918032,
+      "eval_f1": 0.4909767680959837,
+      "eval_loss": 1.003716230392456,
+      "eval_runtime": 71.8801,
+      "eval_samples_per_second": 1.697,
+      "eval_steps_per_second": 0.223,
+      "step": 372
+    },
+    {
+      "epoch": 4.000347826086957,
+      "grad_norm": 10.244637489318848,
+      "learning_rate": 8.26086956521739e-06,
+      "loss": 0.678,
+      "step": 380
+    },
+    {
+      "epoch": 4.000782608695652,
+      "grad_norm": 19.61251449584961,
+      "learning_rate": 8.478260869565217e-06,
+      "loss": 0.6654,
+      "step": 390
+    },
+    {
+      "epoch": 4.001217391304348,
+      "grad_norm": 15.357687950134277,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 0.7421,
+      "step": 400
+    },
+    {
+      "epoch": 4.001652173913043,
+      "grad_norm": 10.771465301513672,
+      "learning_rate": 8.91304347826087e-06,
+      "loss": 0.5394,
+      "step": 410
+    },
+    {
+      "epoch": 4.002086956521739,
+      "grad_norm": 26.172359466552734,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 0.7181,
+      "step": 420
+    },
+    {
+      "epoch": 4.002521739130435,
+      "grad_norm": 24.619651794433594,
+      "learning_rate": 9.347826086956523e-06,
+      "loss": 0.605,
+      "step": 430
+    },
+    {
+      "epoch": 4.00295652173913,
+      "grad_norm": 23.711095809936523,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 0.7164,
+      "step": 440
+    },
+    {
+      "epoch": 4.003391304347826,
+      "grad_norm": 11.760732650756836,
+      "learning_rate": 9.782608695652175e-06,
+      "loss": 0.5269,
+      "step": 450
+    },
+    {
+      "epoch": 4.003826086956522,
+      "grad_norm": 21.210412979125977,
+      "learning_rate": 1e-05,
+      "loss": 0.7025,
+      "step": 460
+    },
+    {
+      "epoch": 4.004043478260869,
+      "eval_accuracy": 0.5983606557377049,
+      "eval_f1": 0.5695439501100179,
+      "eval_loss": 0.9121292233467102,
+      "eval_runtime": 78.8043,
+      "eval_samples_per_second": 1.548,
+      "eval_steps_per_second": 0.203,
+      "step": 465
+    },
+    {
+      "epoch": 5.000217391304348,
+      "grad_norm": 6.867463111877441,
+      "learning_rate": 1.0217391304347827e-05,
+      "loss": 0.4847,
+      "step": 470
+    },
+    {
+      "epoch": 5.000652173913044,
+      "grad_norm": 3.2302322387695312,
+      "learning_rate": 1.0434782608695651e-05,
+      "loss": 0.3333,
+      "step": 480
+    },
+    {
+      "epoch": 5.001086956521739,
+      "grad_norm": 21.763275146484375,
+      "learning_rate": 1.0652173913043479e-05,
+      "loss": 0.494,
+      "step": 490
+    },
+    {
+      "epoch": 5.0015217391304345,
+      "grad_norm": 13.490067481994629,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 0.3955,
+      "step": 500
+    },
+    {
+      "epoch": 5.001956521739131,
+      "grad_norm": 20.92444610595703,
+      "learning_rate": 1.1086956521739131e-05,
+      "loss": 0.627,
+      "step": 510
+    },
+    {
+      "epoch": 5.002391304347826,
+      "grad_norm": 23.297204971313477,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 0.5445,
+      "step": 520
+    },
+    {
+      "epoch": 5.002826086956522,
+      "grad_norm": 30.925203323364258,
+      "learning_rate": 1.1521739130434783e-05,
+      "loss": 0.4394,
+      "step": 530
+    },
+    {
+      "epoch": 5.003260869565217,
+      "grad_norm": 35.236183166503906,
+      "learning_rate": 1.173913043478261e-05,
+      "loss": 0.6552,
+      "step": 540
+    },
+    {
+      "epoch": 5.003695652173913,
+      "grad_norm": 4.368950843811035,
+      "learning_rate": 1.1956521739130435e-05,
+      "loss": 0.5447,
+      "step": 550
+    },
+    {
+      "epoch": 5.004043478260869,
+      "eval_accuracy": 0.6065573770491803,
+      "eval_f1": 0.5667313624233182,
+      "eval_loss": 0.9521207809448242,
+      "eval_runtime": 75.9971,
+      "eval_samples_per_second": 1.605,
+      "eval_steps_per_second": 0.211,
+      "step": 558
+    },
+    {
+      "epoch": 6.000086956521739,
+      "grad_norm": 6.621151447296143,
+      "learning_rate": 1.2173913043478261e-05,
+      "loss": 0.4113,
+      "step": 560
+    },
+    {
+      "epoch": 6.000521739130435,
+      "grad_norm": 20.69222640991211,
+      "learning_rate": 1.2391304347826088e-05,
+      "loss": 0.3668,
+      "step": 570
+    },
+    {
+      "epoch": 6.0009565217391305,
+      "grad_norm": 11.611593246459961,
+      "learning_rate": 1.2608695652173914e-05,
+      "loss": 0.3163,
+      "step": 580
+    },
+    {
+      "epoch": 6.001391304347826,
+      "grad_norm": 24.46592140197754,
+      "learning_rate": 1.2826086956521741e-05,
+      "loss": 0.4861,
+      "step": 590
+    },
+    {
+      "epoch": 6.001826086956521,
+      "grad_norm": 4.075656890869141,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 0.2725,
+      "step": 600
+    },
+    {
+      "epoch": 6.002260869565218,
+      "grad_norm": 46.16673278808594,
+      "learning_rate": 1.3260869565217394e-05,
+      "loss": 0.5276,
+      "step": 610
+    },
+    {
+      "epoch": 6.002695652173913,
+      "grad_norm": 25.51651382446289,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 0.3851,
+      "step": 620
+    },
+    {
+      "epoch": 6.003130434782609,
+      "grad_norm": 13.07934284210205,
+      "learning_rate": 1.3695652173913042e-05,
+      "loss": 0.4227,
+      "step": 630
+    },
+    {
+      "epoch": 6.003565217391304,
+      "grad_norm": 48.23696517944336,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 0.4125,
+      "step": 640
+    },
+    {
+      "epoch": 6.004,
+      "grad_norm": 14.747217178344727,
+      "learning_rate": 1.4130434782608694e-05,
+      "loss": 0.4752,
+      "step": 650
+    },
+    {
+      "epoch": 6.004043478260869,
+      "eval_accuracy": 0.7131147540983607,
+      "eval_f1": 0.6802356113831525,
+      "eval_loss": 0.8963390588760376,
+      "eval_runtime": 68.2476,
+      "eval_samples_per_second": 1.788,
+      "eval_steps_per_second": 0.234,
+      "step": 651
+    },
+    {
+      "epoch": 7.0003913043478265,
+      "grad_norm": 19.88422393798828,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 0.2777,
+      "step": 660
+    },
+    {
+      "epoch": 7.000826086956522,
+      "grad_norm": 13.482111930847168,
+      "learning_rate": 1.4565217391304348e-05,
+      "loss": 0.4554,
+      "step": 670
+    },
+    {
+      "epoch": 7.001260869565217,
+      "grad_norm": 11.722796440124512,
+      "learning_rate": 1.4782608695652176e-05,
+      "loss": 0.2043,
+      "step": 680
+    },
+    {
+      "epoch": 7.001695652173913,
+      "grad_norm": 16.32537269592285,
+      "learning_rate": 1.5e-05,
+      "loss": 0.1863,
+      "step": 690
+    },
+    {
+      "epoch": 7.002130434782608,
+      "grad_norm": 3.987351179122925,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 0.1437,
+      "step": 700
+    },
+    {
+      "epoch": 7.002565217391305,
+      "grad_norm": 1.536703109741211,
+      "learning_rate": 1.5434782608695654e-05,
+      "loss": 0.2714,
+      "step": 710
+    },
+    {
+      "epoch": 7.003,
+      "grad_norm": 31.207611083984375,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 0.5409,
+      "step": 720
+    },
+    {
+      "epoch": 7.003434782608696,
+      "grad_norm": 10.564144134521484,
+      "learning_rate": 1.5869565217391306e-05,
+      "loss": 0.347,
+      "step": 730
+    },
+    {
+      "epoch": 7.003869565217391,
+      "grad_norm": 15.847654342651367,
+      "learning_rate": 1.608695652173913e-05,
+      "loss": 0.3256,
+      "step": 740
+    },
+    {
+      "epoch": 7.004043478260869,
+      "eval_accuracy": 0.7213114754098361,
+      "eval_f1": 0.7152792066865421,
+      "eval_loss": 0.7381582260131836,
+      "eval_runtime": 85.386,
+      "eval_samples_per_second": 1.429,
+      "eval_steps_per_second": 0.187,
+      "step": 744
+    },
+    {
+      "epoch": 8.000260869565217,
+      "grad_norm": 4.257699489593506,
+      "learning_rate": 1.630434782608696e-05,
+      "loss": 0.1054,
+      "step": 750
+    },
+    {
+      "epoch": 8.000695652173913,
+      "grad_norm": 2.503180742263794,
+      "learning_rate": 1.652173913043478e-05,
+      "loss": 0.182,
+      "step": 760
+    },
+    {
+      "epoch": 8.001130434782608,
+      "grad_norm": 13.347521781921387,
+      "learning_rate": 1.673913043478261e-05,
+      "loss": 0.1134,
+      "step": 770
+    },
+    {
+      "epoch": 8.001565217391304,
+      "grad_norm": 7.592967510223389,
+      "learning_rate": 1.6956521739130433e-05,
+      "loss": 0.2086,
+      "step": 780
+    },
+    {
+      "epoch": 8.002,
+      "grad_norm": 9.522619247436523,
+      "learning_rate": 1.7173913043478263e-05,
+      "loss": 0.2546,
+      "step": 790
+    },
+    {
+      "epoch": 8.002434782608695,
+      "grad_norm": 22.83403778076172,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 0.4554,
+      "step": 800
+    },
+    {
+      "epoch": 8.002869565217392,
+      "grad_norm": 25.185588836669922,
+      "learning_rate": 1.7608695652173915e-05,
+      "loss": 0.3155,
+      "step": 810
+    },
+    {
+      "epoch": 8.003304347826086,
+      "grad_norm": 4.784695625305176,
+      "learning_rate": 1.782608695652174e-05,
+      "loss": 0.2442,
+      "step": 820
+    },
+    {
+      "epoch": 8.003739130434782,
+      "grad_norm": 49.901519775390625,
+      "learning_rate": 1.8043478260869567e-05,
+      "loss": 0.1971,
+      "step": 830
+    },
+    {
+      "epoch": 8.00404347826087,
+      "eval_accuracy": 0.6885245901639344,
+      "eval_f1": 0.652401417303907,
+      "eval_loss": 1.1315979957580566,
+      "eval_runtime": 75.6618,
+      "eval_samples_per_second": 1.612,
+      "eval_steps_per_second": 0.211,
+      "step": 837
+    },
+    {
+      "epoch": 9.000130434782609,
+      "grad_norm": 11.014081001281738,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 0.2153,
+      "step": 840
+    },
+    {
+      "epoch": 9.000565217391305,
+      "grad_norm": 3.1919617652893066,
+      "learning_rate": 1.8478260869565216e-05,
+      "loss": 0.143,
+      "step": 850
+    },
+    {
+      "epoch": 9.001,
+      "grad_norm": 50.34938049316406,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 0.1954,
+      "step": 860
+    },
+    {
+      "epoch": 9.001434782608696,
+      "grad_norm": 42.44783401489258,
+      "learning_rate": 1.8913043478260868e-05,
+      "loss": 0.2195,
+      "step": 870
+    },
+    {
+      "epoch": 9.001869565217392,
+      "grad_norm": 27.059328079223633,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 0.1425,
+      "step": 880
+    },
+    {
+      "epoch": 9.002304347826087,
+      "grad_norm": 58.7094841003418,
+      "learning_rate": 1.9347826086956523e-05,
+      "loss": 0.3012,
+      "step": 890
+    },
+    {
+      "epoch": 9.002739130434783,
+      "grad_norm": 20.4552001953125,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 0.445,
+      "step": 900
+    },
+    {
+      "epoch": 9.003173913043478,
+      "grad_norm": 0.22613933682441711,
+      "learning_rate": 1.9782608695652176e-05,
+      "loss": 0.3575,
+      "step": 910
+    },
+    {
+      "epoch": 9.003608695652174,
+      "grad_norm": 24.949710845947266,
+      "learning_rate": 2e-05,
+      "loss": 0.2652,
+      "step": 920
+    },
+    {
+      "epoch": 9.00404347826087,
+      "grad_norm": 26.43189811706543,
+      "learning_rate": 2.0217391304347828e-05,
+      "loss": 0.5144,
+      "step": 930
+    },
+    {
+      "epoch": 9.00404347826087,
+      "eval_accuracy": 0.7459016393442623,
+      "eval_f1": 0.7200532157059749,
+      "eval_loss": 0.8559982180595398,
+      "eval_runtime": 77.4935,
+      "eval_samples_per_second": 1.574,
+      "eval_steps_per_second": 0.206,
+      "step": 930
+    },
+    {
+      "epoch": 10.000434782608696,
+      "grad_norm": 1.3086036443710327,
+      "learning_rate": 2.0434782608695654e-05,
+      "loss": 0.2612,
+      "step": 940
+    },
+    {
+      "epoch": 10.000869565217391,
+      "grad_norm": 8.613253593444824,
+      "learning_rate": 2.065217391304348e-05,
+      "loss": 0.0451,
+      "step": 950
+    },
+    {
+      "epoch": 10.001304347826087,
+      "grad_norm": 0.1866026222705841,
+      "learning_rate": 2.0869565217391303e-05,
+      "loss": 0.1809,
+      "step": 960
+    },
+    {
+      "epoch": 10.001739130434782,
+      "grad_norm": 5.5913825035095215,
+      "learning_rate": 2.1086956521739132e-05,
+      "loss": 0.2038,
+      "step": 970
+    },
+    {
+      "epoch": 10.002173913043478,
+      "grad_norm": 0.5955354571342468,
+      "learning_rate": 2.1304347826086958e-05,
+      "loss": 0.2476,
+      "step": 980
+    },
+    {
+      "epoch": 10.002608695652174,
+      "grad_norm": 2.7775936126708984,
+      "learning_rate": 2.1521739130434784e-05,
+      "loss": 0.0573,
+      "step": 990
+    },
+    {
+      "epoch": 10.003043478260869,
+      "grad_norm": 0.2440994530916214,
+      "learning_rate": 2.173913043478261e-05,
+      "loss": 0.0826,
+      "step": 1000
+    },
+    {
+      "epoch": 10.003478260869565,
+      "grad_norm": 0.09992615133523941,
+      "learning_rate": 2.1956521739130436e-05,
+      "loss": 0.2261,
+      "step": 1010
+    },
+    {
+      "epoch": 10.003913043478262,
+      "grad_norm": 62.008052825927734,
+      "learning_rate": 2.2173913043478262e-05,
+      "loss": 0.2414,
+      "step": 1020
+    },
+    {
+      "epoch": 10.00404347826087,
+      "eval_accuracy": 0.7704918032786885,
+      "eval_f1": 0.7633172613307619,
+      "eval_loss": 0.86429762840271,
+      "eval_runtime": 71.2234,
+      "eval_samples_per_second": 1.713,
+      "eval_steps_per_second": 0.225,
+      "step": 1023
+    },
+    {
+      "epoch": 11.000304347826088,
+      "grad_norm": 10.307206153869629,
+      "learning_rate": 2.239130434782609e-05,
+      "loss": 0.2326,
+      "step": 1030
+    },
+    {
+      "epoch": 11.000739130434782,
+      "grad_norm": 0.6196430921554565,
+      "learning_rate": 2.2608695652173914e-05,
+      "loss": 0.0381,
+      "step": 1040
+    },
+    {
+      "epoch": 11.001173913043479,
+      "grad_norm": 40.027503967285156,
+      "learning_rate": 2.282608695652174e-05,
+      "loss": 0.1084,
+      "step": 1050
+    },
+    {
+      "epoch": 11.001608695652173,
+      "grad_norm": 14.861847877502441,
+      "learning_rate": 2.3043478260869567e-05,
+      "loss": 0.092,
+      "step": 1060
+    },
+    {
+      "epoch": 11.00204347826087,
+      "grad_norm": 4.125478744506836,
+      "learning_rate": 2.3260869565217393e-05,
+      "loss": 0.1853,
+      "step": 1070
+    },
+    {
+      "epoch": 11.002478260869566,
+      "grad_norm": 6.733792781829834,
+      "learning_rate": 2.347826086956522e-05,
+      "loss": 0.3253,
+      "step": 1080
+    },
+    {
+      "epoch": 11.00291304347826,
+      "grad_norm": 78.53083038330078,
+      "learning_rate": 2.3695652173913045e-05,
+      "loss": 0.3464,
+      "step": 1090
+    },
+    {
+      "epoch": 11.003347826086957,
+      "grad_norm": 23.05046844482422,
+      "learning_rate": 2.391304347826087e-05,
+      "loss": 0.194,
+      "step": 1100
+    },
+    {
+      "epoch": 11.003782608695651,
+      "grad_norm": 0.8519552946090698,
+      "learning_rate": 2.4130434782608697e-05,
+      "loss": 0.0972,
+      "step": 1110
+    },
+    {
+      "epoch": 11.00404347826087,
+      "eval_accuracy": 0.6967213114754098,
+      "eval_f1": 0.6836640572123361,
+      "eval_loss": 1.2529149055480957,
+      "eval_runtime": 76.6021,
+      "eval_samples_per_second": 1.593,
+      "eval_steps_per_second": 0.209,
+      "step": 1116
+    },
+    {
+      "epoch": 12.000173913043477,
+      "grad_norm": 0.48925498127937317,
+      "learning_rate": 2.4347826086956523e-05,
+      "loss": 0.0541,
+      "step": 1120
+    },
+    {
+      "epoch": 12.000608695652174,
+      "grad_norm": 0.18016597628593445,
+      "learning_rate": 2.456521739130435e-05,
+      "loss": 0.0193,
+      "step": 1130
+    },
+    {
+      "epoch": 12.00104347826087,
+      "grad_norm": 61.650970458984375,
+      "learning_rate": 2.4782608695652175e-05,
+      "loss": 0.1264,
+      "step": 1140
+    },
+    {
+      "epoch": 12.001478260869565,
+      "grad_norm": 0.5460256934165955,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1422,
+      "step": 1150
+    },
+    {
+      "epoch": 12.001913043478261,
+      "grad_norm": 1.2666577100753784,
+      "learning_rate": 2.5217391304347827e-05,
+      "loss": 0.166,
+      "step": 1160
+    },
+    {
+      "epoch": 12.002347826086957,
+      "grad_norm": 38.8865966796875,
+      "learning_rate": 2.543478260869565e-05,
+      "loss": 0.2384,
+      "step": 1170
+    },
+    {
+      "epoch": 12.002782608695652,
+      "grad_norm": 7.161799907684326,
+      "learning_rate": 2.5652173913043483e-05,
+      "loss": 0.1389,
+      "step": 1180
+    },
+    {
+      "epoch": 12.003217391304348,
+      "grad_norm": 31.820159912109375,
+      "learning_rate": 2.5869565217391305e-05,
+      "loss": 0.3968,
+      "step": 1190
+    },
+    {
+      "epoch": 12.003652173913043,
+      "grad_norm": 18.057905197143555,
+      "learning_rate": 2.608695652173913e-05,
+      "loss": 0.3371,
+      "step": 1200
+    },
+    {
+      "epoch": 12.00404347826087,
+      "eval_accuracy": 0.7213114754098361,
+      "eval_f1": 0.7123017501384928,
+      "eval_loss": 1.0838274955749512,
+      "eval_runtime": 75.3169,
+      "eval_samples_per_second": 1.62,
+      "eval_steps_per_second": 0.212,
+      "step": 1209
+    },
+    {
+      "epoch": 13.000043478260869,
+      "grad_norm": 50.66958236694336,
+      "learning_rate": 2.6304347826086954e-05,
+      "loss": 0.4032,
+      "step": 1210
+    },
+    {
+      "epoch": 13.000478260869565,
+      "grad_norm": 4.794472694396973,
+      "learning_rate": 2.6521739130434787e-05,
+      "loss": 0.169,
+      "step": 1220
+    },
+    {
+      "epoch": 13.000913043478262,
+      "grad_norm": 3.5523149967193604,
+      "learning_rate": 2.673913043478261e-05,
+      "loss": 0.0234,
+      "step": 1230
+    },
+    {
+      "epoch": 13.001347826086956,
+      "grad_norm": 0.033308062702417374,
+      "learning_rate": 2.6956521739130436e-05,
+      "loss": 0.0155,
+      "step": 1240
+    },
+    {
+      "epoch": 13.001782608695652,
+      "grad_norm": 2.988459587097168,
+      "learning_rate": 2.7173913043478262e-05,
+      "loss": 0.0085,
+      "step": 1250
+    },
+    {
+      "epoch": 13.002217391304347,
+      "grad_norm": 78.12406921386719,
+      "learning_rate": 2.7391304347826085e-05,
+      "loss": 0.2082,
+      "step": 1260
+    },
+    {
+      "epoch": 13.002652173913043,
+      "grad_norm": 23.308473587036133,
+      "learning_rate": 2.7608695652173917e-05,
+      "loss": 0.0724,
+      "step": 1270
+    },
+    {
+      "epoch": 13.00308695652174,
+      "grad_norm": 0.23175761103630066,
+      "learning_rate": 2.782608695652174e-05,
+      "loss": 0.3893,
+      "step": 1280
+    },
+    {
+      "epoch": 13.003521739130434,
+      "grad_norm": 5.151690483093262,
+      "learning_rate": 2.8043478260869566e-05,
+      "loss": 0.1066,
+      "step": 1290
+    },
+    {
+      "epoch": 13.00395652173913,
+      "grad_norm": 0.07266629487276077,
+      "learning_rate": 2.826086956521739e-05,
+      "loss": 0.1133,
+      "step": 1300
+    },
+    {
+      "epoch": 13.00404347826087,
+      "eval_accuracy": 0.7295081967213115,
+      "eval_f1": 0.7128166457266807,
+      "eval_loss": 1.2047704458236694,
+      "eval_runtime": 70.4266,
+      "eval_samples_per_second": 1.732,
+      "eval_steps_per_second": 0.227,
+      "step": 1302
+    },
+    {
+      "epoch": 14.000347826086957,
+      "grad_norm": 25.666845321655273,
+      "learning_rate": 2.847826086956522e-05,
+      "loss": 0.2143,
+      "step": 1310
+    },
+    {
+      "epoch": 14.000782608695653,
+      "grad_norm": 0.48047909140586853,
+      "learning_rate": 2.8695652173913044e-05,
+      "loss": 0.111,
+      "step": 1320
+    },
+    {
+      "epoch": 14.001217391304348,
+      "grad_norm": 1.6062359809875488,
+      "learning_rate": 2.891304347826087e-05,
+      "loss": 0.1202,
+      "step": 1330
+    },
+    {
+      "epoch": 14.001652173913044,
+      "grad_norm": 46.08817672729492,
+      "learning_rate": 2.9130434782608696e-05,
+      "loss": 0.3498,
+      "step": 1340
+    },
+    {
+      "epoch": 14.002086956521739,
+      "grad_norm": 0.1339997947216034,
+      "learning_rate": 2.9347826086956526e-05,
+      "loss": 0.366,
+      "step": 1350
+    },
+    {
+      "epoch": 14.002521739130435,
+      "grad_norm": 49.66046905517578,
+      "learning_rate": 2.9565217391304352e-05,
+      "loss": 0.2443,
+      "step": 1360
+    },
+    {
+      "epoch": 14.002956521739131,
+      "grad_norm": 0.571479082107544,
+      "learning_rate": 2.9782608695652175e-05,
+      "loss": 0.0953,
+      "step": 1370
+    },
+    {
+      "epoch": 14.003391304347826,
+      "grad_norm": 0.11195419728755951,
+      "learning_rate": 3e-05,
+      "loss": 0.0104,
+      "step": 1380
+    },
+    {
+      "epoch": 14.003826086956522,
+      "grad_norm": 0.2546157240867615,
+      "learning_rate": 3.0217391304347827e-05,
+      "loss": 0.1091,
+      "step": 1390
+    },
+    {
+      "epoch": 14.00404347826087,
+      "eval_accuracy": 0.7868852459016393,
+      "eval_f1": 0.772974027172348,
+      "eval_loss": 1.0909240245819092,
+      "eval_runtime": 78.9629,
+      "eval_samples_per_second": 1.545,
+      "eval_steps_per_second": 0.203,
+      "step": 1395
+    },
+    {
+      "epoch": 15.000217391304348,
+      "grad_norm": 24.337543487548828,
+      "learning_rate": 3.0434782608695656e-05,
+      "loss": 0.1526,
+      "step": 1400
+    },
+    {
+      "epoch": 15.000652173913043,
+      "grad_norm": 0.14164993166923523,
+      "learning_rate": 3.065217391304348e-05,
+      "loss": 0.0855,
+      "step": 1410
+    },
+    {
+      "epoch": 15.001086956521739,
+      "grad_norm": 0.027599574998021126,
+      "learning_rate": 3.086956521739131e-05,
+      "loss": 0.1999,
+      "step": 1420
+    },
+    {
+      "epoch": 15.001521739130435,
+      "grad_norm": 0.1311451494693756,
+      "learning_rate": 3.108695652173913e-05,
+      "loss": 0.1114,
+      "step": 1430
+    },
+    {
+      "epoch": 15.00195652173913,
+      "grad_norm": 0.4929811358451843,
+      "learning_rate": 3.130434782608696e-05,
+      "loss": 0.269,
+      "step": 1440
+    },
+    {
+      "epoch": 15.002391304347826,
+      "grad_norm": 73.08831024169922,
+      "learning_rate": 3.152173913043479e-05,
+      "loss": 0.149,
+      "step": 1450
+    },
+    {
+      "epoch": 15.00282608695652,
+      "grad_norm": 0.11047495901584625,
+      "learning_rate": 3.173913043478261e-05,
+      "loss": 0.224,
+      "step": 1460
+    },
+    {
+      "epoch": 15.003260869565217,
+      "grad_norm": 0.08704402297735214,
+      "learning_rate": 3.195652173913043e-05,
+      "loss": 0.2719,
+      "step": 1470
+    },
+    {
+      "epoch": 15.003695652173914,
+      "grad_norm": 0.2108268141746521,
+      "learning_rate": 3.217391304347826e-05,
+      "loss": 0.0642,
+      "step": 1480
+    },
+    {
+      "epoch": 15.00404347826087,
+      "eval_accuracy": 0.8278688524590164,
+      "eval_f1": 0.8271381313811401,
+      "eval_loss": 0.8966262936592102,
+      "eval_runtime": 75.9595,
+      "eval_samples_per_second": 1.606,
+      "eval_steps_per_second": 0.211,
+      "step": 1488
+    },
+    {
+      "epoch": 16.000086956521738,
+      "grad_norm": 0.016246311366558075,
+      "learning_rate": 3.239130434782609e-05,
+      "loss": 0.057,
+      "step": 1490
+    },
+    {
+      "epoch": 16.000521739130434,
+      "grad_norm": 0.024972323328256607,
+      "learning_rate": 3.260869565217392e-05,
+      "loss": 0.0157,
+      "step": 1500
+    },
+    {
+      "epoch": 16.00095652173913,
+      "grad_norm": 0.2207304686307907,
+      "learning_rate": 3.282608695652174e-05,
+      "loss": 0.0099,
+      "step": 1510
+    },
+    {
+      "epoch": 16.001391304347827,
+      "grad_norm": 0.05993523448705673,
+      "learning_rate": 3.304347826086956e-05,
+      "loss": 0.2193,
+      "step": 1520
+    },
+    {
+      "epoch": 16.001826086956523,
+      "grad_norm": 1.4549189805984497,
+      "learning_rate": 3.3260869565217395e-05,
+      "loss": 0.2023,
+      "step": 1530
+    },
+    {
+      "epoch": 16.002260869565216,
+      "grad_norm": 0.024738606065511703,
+      "learning_rate": 3.347826086956522e-05,
+      "loss": 0.2535,
+      "step": 1540
+    },
+    {
+      "epoch": 16.002695652173912,
+      "grad_norm": 43.513267517089844,
+      "learning_rate": 3.369565217391305e-05,
+      "loss": 0.1875,
+      "step": 1550
+    },
+    {
+      "epoch": 16.00313043478261,
+      "grad_norm": 6.210598945617676,
+      "learning_rate": 3.3913043478260867e-05,
+      "loss": 0.0425,
+      "step": 1560
+    },
+    {
+      "epoch": 16.003565217391305,
+      "grad_norm": 0.13376347720623016,
+      "learning_rate": 3.413043478260869e-05,
+      "loss": 0.1634,
+      "step": 1570
+    },
+    {
+      "epoch": 16.004,
+      "grad_norm": 66.50971984863281,
+      "learning_rate": 3.4347826086956526e-05,
+      "loss": 0.1608,
+      "step": 1580
+    },
+    {
+      "epoch": 16.00404347826087,
+      "eval_accuracy": 0.7950819672131147,
+      "eval_f1": 0.7918367666938827,
+      "eval_loss": 0.7916240096092224,
+      "eval_runtime": 68.1076,
+      "eval_samples_per_second": 1.791,
+      "eval_steps_per_second": 0.235,
+      "step": 1581
+    },
+    {
+      "epoch": 17.000391304347826,
+      "grad_norm": 52.1787223815918,
+      "learning_rate": 3.456521739130435e-05,
+      "loss": 0.0693,
+      "step": 1590
+    },
+    {
+      "epoch": 17.000826086956522,
+      "grad_norm": 0.028341565281152725,
+      "learning_rate": 3.478260869565218e-05,
+      "loss": 0.168,
+      "step": 1600
+    },
+    {
+      "epoch": 17.00126086956522,
+      "grad_norm": 0.06816719472408295,
+      "learning_rate": 3.5e-05,
+      "loss": 0.1966,
+      "step": 1610
+    },
+    {
+      "epoch": 17.001695652173915,
+      "grad_norm": 0.06487961858510971,
+      "learning_rate": 3.521739130434783e-05,
+      "loss": 0.2377,
+      "step": 1620
+    },
+    {
+      "epoch": 17.002130434782607,
+      "grad_norm": 1.7196918725967407,
+      "learning_rate": 3.5434782608695656e-05,
+      "loss": 0.0641,
+      "step": 1630
+    },
+    {
+      "epoch": 17.002565217391304,
+      "grad_norm": 1.357792615890503,
+      "learning_rate": 3.565217391304348e-05,
+      "loss": 0.1772,
+      "step": 1640
+    },
+    {
+      "epoch": 17.003,
+      "grad_norm": 48.746490478515625,
+      "learning_rate": 3.58695652173913e-05,
+      "loss": 0.2143,
+      "step": 1650
+    },
+    {
+      "epoch": 17.003434782608696,
+      "grad_norm": 0.2187844216823578,
+      "learning_rate": 3.6086956521739134e-05,
+      "loss": 0.2618,
+      "step": 1660
+    },
+    {
+      "epoch": 17.003869565217393,
+      "grad_norm": 27.896387100219727,
+      "learning_rate": 3.630434782608696e-05,
+      "loss": 0.0274,
+      "step": 1670
+    },
+    {
+      "epoch": 17.00404347826087,
+      "eval_accuracy": 0.8360655737704918,
+      "eval_f1": 0.8350123233242202,
+      "eval_loss": 0.7110782265663147,
+      "eval_runtime": 69.761,
+      "eval_samples_per_second": 1.749,
+      "eval_steps_per_second": 0.229,
+      "step": 1674
+    },
+    {
+      "epoch": 18.000260869565217,
+      "grad_norm": 0.09103922545909882,
+      "learning_rate": 3.6521739130434786e-05,
+      "loss": 0.0427,
+      "step": 1680
+    },
+    {
+      "epoch": 18.000695652173913,
+      "grad_norm": 0.02076614834368229,
+      "learning_rate": 3.673913043478261e-05,
+      "loss": 0.011,
+      "step": 1690
+    },
+    {
+      "epoch": 18.00113043478261,
+      "grad_norm": 1.5786951780319214,
+      "learning_rate": 3.695652173913043e-05,
+      "loss": 0.0256,
+      "step": 1700
+    },
+    {
+      "epoch": 18.001565217391306,
+      "grad_norm": 0.04359204322099686,
+      "learning_rate": 3.7173913043478264e-05,
+      "loss": 0.1148,
+      "step": 1710
+    },
+    {
+      "epoch": 18.002,
+      "grad_norm": 0.9095045924186707,
+      "learning_rate": 3.739130434782609e-05,
+      "loss": 0.3361,
+      "step": 1720
+    },
+    {
+      "epoch": 18.002434782608695,
+      "grad_norm": 69.62400817871094,
+      "learning_rate": 3.7608695652173917e-05,
+      "loss": 0.2205,
+      "step": 1730
+    },
+    {
+      "epoch": 18.00286956521739,
+      "grad_norm": 30.217559814453125,
+      "learning_rate": 3.7826086956521736e-05,
+      "loss": 0.1105,
+      "step": 1740
+    },
+    {
+      "epoch": 18.003304347826088,
+      "grad_norm": 0.47550705075263977,
+      "learning_rate": 3.804347826086957e-05,
+      "loss": 0.1594,
+      "step": 1750
+    },
+    {
+      "epoch": 18.003739130434784,
+      "grad_norm": 68.50760650634766,
+      "learning_rate": 3.8260869565217395e-05,
+      "loss": 0.2955,
+      "step": 1760
+    },
+    {
+      "epoch": 18.00404347826087,
+      "eval_accuracy": 0.7950819672131147,
+      "eval_f1": 0.7968699869172604,
+      "eval_loss": 0.9873971343040466,
+      "eval_runtime": 68.4296,
+      "eval_samples_per_second": 1.783,
+      "eval_steps_per_second": 0.234,
+      "step": 1767
+    },
+    {
+      "epoch": 19.00013043478261,
+      "grad_norm": 6.308517932891846,
+      "learning_rate": 3.847826086956522e-05,
+      "loss": 0.4724,
+      "step": 1770
+    },
+    {
+      "epoch": 19.000565217391305,
+      "grad_norm": 30.987382888793945,
+      "learning_rate": 3.869565217391305e-05,
+      "loss": 0.2195,
+      "step": 1780
+    },
+    {
+      "epoch": 19.001,
+      "grad_norm": 0.02650444395840168,
+      "learning_rate": 3.8913043478260866e-05,
+      "loss": 0.003,
+      "step": 1790
+    },
+    {
+      "epoch": 19.001434782608694,
+      "grad_norm": 0.2237127423286438,
+      "learning_rate": 3.91304347826087e-05,
+      "loss": 0.2009,
+      "step": 1800
+    },
+    {
+      "epoch": 19.00186956521739,
+      "grad_norm": 0.05298967659473419,
+      "learning_rate": 3.9347826086956525e-05,
+      "loss": 0.1038,
+      "step": 1810
+    },
+    {
+      "epoch": 19.002304347826087,
+      "grad_norm": 62.364654541015625,
+      "learning_rate": 3.956521739130435e-05,
+      "loss": 0.0839,
+      "step": 1820
+    },
+    {
+      "epoch": 19.002739130434783,
+      "grad_norm": 0.050453417003154755,
+      "learning_rate": 3.978260869565217e-05,
+      "loss": 0.0338,
+      "step": 1830
+    },
+    {
+      "epoch": 19.00317391304348,
+      "grad_norm": 31.080060958862305,
+      "learning_rate": 4e-05,
+      "loss": 0.0454,
+      "step": 1840
+    },
+    {
+      "epoch": 19.003608695652172,
+      "grad_norm": 62.11554718017578,
+      "learning_rate": 4.021739130434783e-05,
+      "loss": 0.1027,
+      "step": 1850
+    },
+    {
+      "epoch": 19.00404347826087,
+      "grad_norm": 0.8470473885536194,
+      "learning_rate": 4.0434782608695655e-05,
+      "loss": 0.3371,
+      "step": 1860
+    },
+    {
+      "epoch": 19.00404347826087,
+      "eval_accuracy": 0.7377049180327869,
+      "eval_f1": 0.725067323090087,
+      "eval_loss": 1.4002968072891235,
+      "eval_runtime": 82.0023,
+      "eval_samples_per_second": 1.488,
+      "eval_steps_per_second": 0.195,
+      "step": 1860
+    },
+    {
+      "epoch": 20.000434782608696,
+      "grad_norm": 0.7842612862586975,
+      "learning_rate": 4.065217391304348e-05,
+      "loss": 0.2897,
+      "step": 1870
+    },
+    {
+      "epoch": 20.000869565217393,
+      "grad_norm": 45.9306526184082,
+      "learning_rate": 4.086956521739131e-05,
+      "loss": 0.4222,
+      "step": 1880
+    },
+    {
+      "epoch": 20.001304347826085,
+      "grad_norm": 0.01409738790243864,
+      "learning_rate": 4.1086956521739134e-05,
+      "loss": 0.2394,
+      "step": 1890
+    },
+    {
+      "epoch": 20.001739130434782,
+      "grad_norm": 0.834911048412323,
+      "learning_rate": 4.130434782608696e-05,
+      "loss": 0.0976,
+      "step": 1900
+    },
+    {
+      "epoch": 20.002173913043478,
+      "grad_norm": 1.8054020404815674,
+      "learning_rate": 4.1521739130434786e-05,
+      "loss": 0.1883,
+      "step": 1910
+    },
+    {
+      "epoch": 20.002608695652174,
+      "grad_norm": 0.029708171263337135,
+      "learning_rate": 4.1739130434782605e-05,
+      "loss": 0.2925,
+      "step": 1920
+    },
+    {
+      "epoch": 20.00304347826087,
+      "grad_norm": 0.054145630449056625,
+      "learning_rate": 4.195652173913044e-05,
+      "loss": 0.4088,
+      "step": 1930
+    },
+    {
+      "epoch": 20.003478260869564,
+      "grad_norm": 12.527750015258789,
+      "learning_rate": 4.2173913043478264e-05,
+      "loss": 0.3098,
+      "step": 1940
+    },
+    {
+      "epoch": 20.00391304347826,
+      "grad_norm": 0.0675775408744812,
+      "learning_rate": 4.239130434782609e-05,
+      "loss": 0.1977,
+      "step": 1950
+    },
+    {
+      "epoch": 20.00404347826087,
+      "eval_accuracy": 0.8278688524590164,
+      "eval_f1": 0.8285237447031244,
+      "eval_loss": 0.729884147644043,
+      "eval_runtime": 78.9289,
+      "eval_samples_per_second": 1.546,
+      "eval_steps_per_second": 0.203,
+      "step": 1953
+    },
+    {
+      "epoch": 21.000304347826088,
+      "grad_norm": 0.1842467486858368,
+      "learning_rate": 4.2608695652173916e-05,
+      "loss": 0.0141,
+      "step": 1960
+    },
+    {
+      "epoch": 21.000739130434784,
+      "grad_norm": 35.704742431640625,
+      "learning_rate": 4.282608695652174e-05,
+      "loss": 0.1253,
+      "step": 1970
+    },
+    {
+      "epoch": 21.001173913043477,
+      "grad_norm": 25.044652938842773,
+      "learning_rate": 4.304347826086957e-05,
+      "loss": 0.2175,
+      "step": 1980
+    },
+    {
+      "epoch": 21.001608695652173,
+      "grad_norm": 9.421120643615723,
+      "learning_rate": 4.3260869565217394e-05,
+      "loss": 0.0545,
+      "step": 1990
+    },
+    {
+      "epoch": 21.00204347826087,
+      "grad_norm": 0.5154250264167786,
+      "learning_rate": 4.347826086956522e-05,
+      "loss": 0.3332,
+      "step": 2000
+    },
+    {
+      "epoch": 21.002478260869566,
+      "grad_norm": 0.18789710104465485,
+      "learning_rate": 4.3695652173913046e-05,
+      "loss": 0.4334,
+      "step": 2010
+    },
+    {
+      "epoch": 21.002913043478262,
+      "grad_norm": 0.4829055666923523,
+      "learning_rate": 4.391304347826087e-05,
+      "loss": 0.2281,
+      "step": 2020
+    },
+    {
+      "epoch": 21.003347826086955,
+      "grad_norm": 35.912315368652344,
+      "learning_rate": 4.41304347826087e-05,
+      "loss": 0.3521,
+      "step": 2030
+    },
+    {
+      "epoch": 21.00378260869565,
+      "grad_norm": 0.1704353392124176,
+      "learning_rate": 4.4347826086956525e-05,
+      "loss": 0.1155,
+      "step": 2040
+    },
+    {
+      "epoch": 21.00404347826087,
+      "eval_accuracy": 0.8524590163934426,
+      "eval_f1": 0.8524527538533494,
+      "eval_loss": 0.7051700353622437,
+      "eval_runtime": 79.7194,
+      "eval_samples_per_second": 1.53,
+      "eval_steps_per_second": 0.201,
+      "step": 2046
+    },
+    {
+      "epoch": 22.00017391304348,
+      "grad_norm": 0.10614888370037079,
+      "learning_rate": 4.456521739130435e-05,
+      "loss": 0.1375,
+      "step": 2050
+    },
+    {
+      "epoch": 22.000608695652176,
+      "grad_norm": 0.028057333081960678,
+      "learning_rate": 4.478260869565218e-05,
+      "loss": 0.0373,
+      "step": 2060
+    },
+    {
+      "epoch": 22.00104347826087,
+      "grad_norm": 81.05805206298828,
+      "learning_rate": 4.5e-05,
+      "loss": 0.0963,
+      "step": 2070
+    },
+    {
+      "epoch": 22.001478260869565,
+      "grad_norm": 15.839224815368652,
+      "learning_rate": 4.521739130434783e-05,
+      "loss": 0.1109,
+      "step": 2080
+    },
+    {
+      "epoch": 22.00191304347826,
+      "grad_norm": 17.285728454589844,
+      "learning_rate": 4.5434782608695655e-05,
+      "loss": 0.086,
+      "step": 2090
+    },
+    {
+      "epoch": 22.002347826086957,
+      "grad_norm": 0.011755176819860935,
+      "learning_rate": 4.565217391304348e-05,
+      "loss": 0.005,
+      "step": 2100
+    },
+    {
+      "epoch": 22.002782608695654,
+      "grad_norm": 0.04565587639808655,
+      "learning_rate": 4.586956521739131e-05,
+      "loss": 0.2197,
+      "step": 2110
+    },
+    {
+      "epoch": 22.003217391304347,
+      "grad_norm": 0.1595648229122162,
+      "learning_rate": 4.608695652173913e-05,
+      "loss": 0.2304,
+      "step": 2120
+    },
+    {
+      "epoch": 22.003652173913043,
+      "grad_norm": 0.09167420864105225,
+      "learning_rate": 4.630434782608696e-05,
+      "loss": 0.0134,
+      "step": 2130
+    },
+    {
+      "epoch": 22.00404347826087,
+      "eval_accuracy": 0.8770491803278688,
+      "eval_f1": 0.8773614363778297,
+      "eval_loss": 0.7146317958831787,
+      "eval_runtime": 78.4014,
+      "eval_samples_per_second": 1.556,
+      "eval_steps_per_second": 0.204,
+      "step": 2139
+    },
+    {
+      "epoch": 23.00004347826087,
+      "grad_norm": 0.05441008135676384,
+      "learning_rate": 4.6521739130434785e-05,
+      "loss": 0.1697,
+      "step": 2140
+    },
+    {
+      "epoch": 23.000478260869563,
+      "grad_norm": 63.973304748535156,
+      "learning_rate": 4.673913043478261e-05,
+      "loss": 0.3334,
+      "step": 2150
+    },
+    {
+      "epoch": 23.00091304347826,
+      "grad_norm": 27.009883880615234,
+      "learning_rate": 4.695652173913044e-05,
+      "loss": 0.1678,
+      "step": 2160
+    },
+    {
+      "epoch": 23.001347826086956,
+      "grad_norm": 0.037991661578416824,
+      "learning_rate": 4.7173913043478264e-05,
+      "loss": 0.1246,
+      "step": 2170
+    },
+    {
+      "epoch": 23.001782608695652,
+      "grad_norm": 0.040783315896987915,
+      "learning_rate": 4.739130434782609e-05,
+      "loss": 0.0249,
+      "step": 2180
+    },
+    {
+      "epoch": 23.00221739130435,
+      "grad_norm": 18.87450408935547,
+      "learning_rate": 4.7608695652173916e-05,
+      "loss": 0.1833,
+      "step": 2190
+    },
+    {
+      "epoch": 23.002652173913045,
+      "grad_norm": 0.08050432801246643,
+      "learning_rate": 4.782608695652174e-05,
+      "loss": 0.2014,
+      "step": 2200
+    },
+    {
+      "epoch": 23.003086956521738,
+      "grad_norm": 0.021646007895469666,
+      "learning_rate": 4.804347826086957e-05,
+      "loss": 0.1677,
+      "step": 2210
+    },
+    {
+      "epoch": 23.003521739130434,
+      "grad_norm": 33.0234375,
+      "learning_rate": 4.8260869565217394e-05,
+      "loss": 0.2297,
+      "step": 2220
+    },
+    {
+      "epoch": 23.00395652173913,
+      "grad_norm": 0.19216619431972504,
+      "learning_rate": 4.847826086956522e-05,
+      "loss": 0.1011,
+      "step": 2230
+    },
+    {
+      "epoch": 23.00404347826087,
+      "eval_accuracy": 0.8278688524590164,
+      "eval_f1": 0.8276948820520866,
+      "eval_loss": 0.8542459607124329,
+      "eval_runtime": 70.3843,
+      "eval_samples_per_second": 1.733,
+      "eval_steps_per_second": 0.227,
+      "step": 2232
+    },
+    {
+      "epoch": 24.000347826086955,
+      "grad_norm": 19.964126586914062,
+      "learning_rate": 4.8695652173913046e-05,
+      "loss": 0.395,
+      "step": 2240
+    },
+    {
+      "epoch": 24.00078260869565,
+      "grad_norm": 1.8568761348724365,
+      "learning_rate": 4.891304347826087e-05,
+      "loss": 0.2026,
+      "step": 2250
+    },
+    {
+      "epoch": 24.001217391304348,
+      "grad_norm": 29.76133918762207,
+      "learning_rate": 4.91304347826087e-05,
+      "loss": 0.1282,
+      "step": 2260
+    },
+    {
+      "epoch": 24.001652173913044,
+      "grad_norm": 0.4110662043094635,
+      "learning_rate": 4.9347826086956524e-05,
+      "loss": 0.0744,
+      "step": 2270
+    },
+    {
+      "epoch": 24.00208695652174,
+      "grad_norm": 0.05301794409751892,
+      "learning_rate": 4.956521739130435e-05,
+      "loss": 0.1104,
+      "step": 2280
+    },
+    {
+      "epoch": 24.002521739130433,
+      "grad_norm": 100.25152587890625,
+      "learning_rate": 4.9782608695652176e-05,
+      "loss": 0.0886,
+      "step": 2290
+    },
+    {
+      "epoch": 24.00295652173913,
+      "grad_norm": 4.144306182861328,
+      "learning_rate": 5e-05,
+      "loss": 0.4572,
+      "step": 2300
+    },
+    {
+      "epoch": 24.003391304347826,
+      "grad_norm": 84.53608703613281,
+      "learning_rate": 4.997584541062802e-05,
+      "loss": 0.4282,
+      "step": 2310
+    },
+    {
+      "epoch": 24.003826086956522,
+      "grad_norm": 1.2512717247009277,
+      "learning_rate": 4.9951690821256045e-05,
+      "loss": 0.0834,
+      "step": 2320
+    },
+    {
+      "epoch": 24.00404347826087,
+      "eval_accuracy": 0.7131147540983607,
+      "eval_f1": 0.71432835887995,
+      "eval_loss": 1.6143077611923218,
+      "eval_runtime": 78.2242,
+      "eval_samples_per_second": 1.56,
+      "eval_steps_per_second": 0.205,
+      "step": 2325
+    },
+    {
+      "epoch": 25.000217391304346,
+      "grad_norm": 0.014534312300384045,
+      "learning_rate": 4.9927536231884056e-05,
+      "loss": 0.0735,
+      "step": 2330
+    },
+    {
+      "epoch": 25.000652173913043,
+      "grad_norm": 2.242990493774414,
+      "learning_rate": 4.990338164251208e-05,
+      "loss": 0.0942,
+      "step": 2340
+    },
+    {
+      "epoch": 25.00108695652174,
+      "grad_norm": 0.02222440391778946,
+      "learning_rate": 4.98792270531401e-05,
+      "loss": 0.0775,
+      "step": 2350
+    },
+    {
+      "epoch": 25.001521739130435,
+      "grad_norm": 0.014810199849307537,
+      "learning_rate": 4.985507246376812e-05,
+      "loss": 0.1185,
+      "step": 2360
+    },
+    {
+      "epoch": 25.00195652173913,
+      "grad_norm": 53.17665100097656,
+      "learning_rate": 4.983091787439614e-05,
+      "loss": 0.5145,
+      "step": 2370
+    },
+    {
+      "epoch": 25.002391304347825,
+      "grad_norm": 0.023323476314544678,
+      "learning_rate": 4.980676328502415e-05,
+      "loss": 0.0293,
+      "step": 2380
+    },
+    {
+      "epoch": 25.00282608695652,
+      "grad_norm": 1.3914483785629272,
+      "learning_rate": 4.9782608695652176e-05,
+      "loss": 0.1876,
+      "step": 2390
+    },
+    {
+      "epoch": 25.003260869565217,
+      "grad_norm": 0.20789887011051178,
+      "learning_rate": 4.9758454106280194e-05,
+      "loss": 0.1446,
+      "step": 2400
+    },
+    {
+      "epoch": 25.003695652173914,
+      "grad_norm": 6.419387340545654,
+      "learning_rate": 4.973429951690822e-05,
+      "loss": 0.3776,
+      "step": 2410
+    },
+    {
+      "epoch": 25.00404347826087,
+      "eval_accuracy": 0.7868852459016393,
+      "eval_f1": 0.7833783048095553,
+      "eval_loss": 1.0066475868225098,
+      "eval_runtime": 75.7936,
+      "eval_samples_per_second": 1.61,
+      "eval_steps_per_second": 0.211,
+      "step": 2418
+    },
+    {
+      "epoch": 26.000086956521738,
+      "grad_norm": 0.11995876580476761,
+      "learning_rate": 4.9710144927536237e-05,
+      "loss": 0.1378,
+      "step": 2420
+    },
+    {
+      "epoch": 26.000521739130434,
+      "grad_norm": 0.10857793688774109,
+      "learning_rate": 4.9685990338164254e-05,
+      "loss": 0.0582,
+      "step": 2430
+    },
+    {
+      "epoch": 26.00095652173913,
+      "grad_norm": 3.5968430042266846,
+      "learning_rate": 4.966183574879227e-05,
+      "loss": 0.0596,
+      "step": 2440
+    },
+    {
+      "epoch": 26.001391304347827,
+      "grad_norm": 0.012462865561246872,
+      "learning_rate": 4.963768115942029e-05,
+      "loss": 0.036,
+      "step": 2450
+    },
+    {
+      "epoch": 26.001826086956523,
+      "grad_norm": 88.11896514892578,
+      "learning_rate": 4.9613526570048315e-05,
+      "loss": 0.1124,
+      "step": 2460
+    },
+    {
+      "epoch": 26.002260869565216,
+      "grad_norm": 0.010418935678899288,
+      "learning_rate": 4.958937198067633e-05,
+      "loss": 0.2141,
+      "step": 2470
+    },
+    {
+      "epoch": 26.002695652173912,
+      "grad_norm": 0.8756917119026184,
+      "learning_rate": 4.956521739130435e-05,
+      "loss": 0.0481,
+      "step": 2480
+    },
+    {
+      "epoch": 26.00313043478261,
+      "grad_norm": 0.014706839807331562,
+      "learning_rate": 4.954106280193237e-05,
+      "loss": 0.0433,
+      "step": 2490
+    },
+    {
+      "epoch": 26.003565217391305,
+      "grad_norm": 33.63422393798828,
+      "learning_rate": 4.9516908212560386e-05,
+      "loss": 0.1711,
+      "step": 2500
+    },
+    {
+      "epoch": 26.004,
+      "grad_norm": 8.496404647827148,
+      "learning_rate": 4.949275362318841e-05,
+      "loss": 0.0745,
+      "step": 2510
+    },
+    {
+      "epoch": 26.00404347826087,
+      "eval_accuracy": 0.8442622950819673,
+      "eval_f1": 0.8448557228692234,
+      "eval_loss": 0.794857382774353,
+      "eval_runtime": 80.5247,
+      "eval_samples_per_second": 1.515,
+      "eval_steps_per_second": 0.199,
+      "step": 2511
+    },
+    {
+      "epoch": 27.000391304347826,
+      "grad_norm": 30.50170135498047,
+      "learning_rate": 4.946859903381643e-05,
+      "loss": 0.0833,
+      "step": 2520
+    },
+    {
+      "epoch": 27.000826086956522,
+      "grad_norm": 0.010464085265994072,
+      "learning_rate": 4.9444444444444446e-05,
+      "loss": 0.0171,
+      "step": 2530
+    },
+    {
+      "epoch": 27.00126086956522,
+      "grad_norm": 0.013533636927604675,
+      "learning_rate": 4.9420289855072464e-05,
+      "loss": 0.1177,
+      "step": 2540
+    },
+    {
+      "epoch": 27.001695652173915,
+      "grad_norm": 0.06204245612025261,
+      "learning_rate": 4.939613526570048e-05,
+      "loss": 0.0623,
+      "step": 2550
+    },
+    {
+      "epoch": 27.002130434782607,
+      "grad_norm": 0.007505207788199186,
+      "learning_rate": 4.9371980676328506e-05,
+      "loss": 0.0204,
+      "step": 2560
+    },
+    {
+      "epoch": 27.002565217391304,
+      "grad_norm": 0.03377118706703186,
+      "learning_rate": 4.9347826086956524e-05,
+      "loss": 0.0159,
+      "step": 2570
+    },
+    {
+      "epoch": 27.003,
+      "grad_norm": 0.009533570148050785,
+      "learning_rate": 4.932367149758454e-05,
+      "loss": 0.2513,
+      "step": 2580
+    },
+    {
+      "epoch": 27.003434782608696,
+      "grad_norm": 0.007220500614494085,
+      "learning_rate": 4.929951690821256e-05,
+      "loss": 0.0012,
+      "step": 2590
+    },
+    {
+      "epoch": 27.003869565217393,
+      "grad_norm": 0.008370584808290005,
+      "learning_rate": 4.9275362318840584e-05,
+      "loss": 0.0012,
+      "step": 2600
+    },
+    {
+      "epoch": 27.00404347826087,
+      "eval_accuracy": 0.8524590163934426,
+      "eval_f1": 0.8519739016181949,
+      "eval_loss": 0.8901496529579163,
+      "eval_runtime": 78.3188,
+      "eval_samples_per_second": 1.558,
+      "eval_steps_per_second": 0.204,
+      "step": 2604
+    },
+    {
+      "epoch": 28.000260869565217,
+      "grad_norm": 10.66296672821045,
+      "learning_rate": 4.92512077294686e-05,
+      "loss": 0.0684,
+      "step": 2610
+    },
+    {
+      "epoch": 28.000695652173913,
+      "grad_norm": 0.016988499090075493,
+      "learning_rate": 4.922705314009662e-05,
+      "loss": 0.0504,
+      "step": 2620
+    },
+    {
+      "epoch": 28.00113043478261,
+      "grad_norm": 11.786551475524902,
+      "learning_rate": 4.920289855072464e-05,
+      "loss": 0.1493,
+      "step": 2630
+    },
+    {
+      "epoch": 28.001565217391306,
+      "grad_norm": 0.027907541021704674,
+      "learning_rate": 4.9178743961352656e-05,
+      "loss": 0.0218,
+      "step": 2640
+    },
+    {
+      "epoch": 28.002,
+      "grad_norm": 0.006081975065171719,
+      "learning_rate": 4.915458937198068e-05,
+      "loss": 0.0492,
+      "step": 2650
+    },
+    {
+      "epoch": 28.002434782608695,
+      "grad_norm": 0.011573576368391514,
+      "learning_rate": 4.91304347826087e-05,
+      "loss": 0.0006,
+      "step": 2660
+    },
+    {
+      "epoch": 28.00286956521739,
+      "grad_norm": 6.206803321838379,
+      "learning_rate": 4.910628019323672e-05,
+      "loss": 0.1155,
+      "step": 2670
+    },
+    {
+      "epoch": 28.003304347826088,
+      "grad_norm": 0.031189128756523132,
+      "learning_rate": 4.9082125603864734e-05,
+      "loss": 0.0613,
+      "step": 2680
+    },
+    {
+      "epoch": 28.003739130434784,
+      "grad_norm": 0.06879521161317825,
+      "learning_rate": 4.905797101449275e-05,
+      "loss": 0.0433,
+      "step": 2690
+    },
+    {
+      "epoch": 28.00404347826087,
+      "eval_accuracy": 0.8360655737704918,
+      "eval_f1": 0.8360363588217401,
+      "eval_loss": 0.9370259642601013,
+      "eval_runtime": 75.9173,
+      "eval_samples_per_second": 1.607,
+      "eval_steps_per_second": 0.211,
+      "step": 2697
+    },
+    {
+      "epoch": 29.00013043478261,
+      "grad_norm": 0.008160380646586418,
+      "learning_rate": 4.9033816425120776e-05,
+      "loss": 0.0016,
+      "step": 2700
+    },
+    {
+      "epoch": 29.000565217391305,
+      "grad_norm": 0.0150864627212286,
+      "learning_rate": 4.9009661835748794e-05,
+      "loss": 0.0037,
+      "step": 2710
+    },
+    {
+      "epoch": 29.001,
+      "grad_norm": 6.826006889343262,
+      "learning_rate": 4.898550724637682e-05,
+      "loss": 0.0263,
+      "step": 2720
+    },
+    {
+      "epoch": 29.001434782608694,
+      "grad_norm": 0.010020852088928223,
+      "learning_rate": 4.8961352657004836e-05,
+      "loss": 0.0041,
+      "step": 2730
+    },
+    {
+      "epoch": 29.00186956521739,
+      "grad_norm": 0.0070611899718642235,
+      "learning_rate": 4.893719806763285e-05,
+      "loss": 0.058,
+      "step": 2740
+    },
+    {
+      "epoch": 29.002304347826087,
+      "grad_norm": 0.006664105225354433,
+      "learning_rate": 4.891304347826087e-05,
+      "loss": 0.0311,
+      "step": 2750
+    },
+    {
+      "epoch": 29.002739130434783,
+      "grad_norm": 0.011251281015574932,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.0891,
+      "step": 2760
+    },
+    {
+      "epoch": 29.00317391304348,
+      "grad_norm": 0.0483035147190094,
+      "learning_rate": 4.8864734299516914e-05,
+      "loss": 0.0673,
+      "step": 2770
+    },
+    {
+      "epoch": 29.003608695652172,
+      "grad_norm": 0.01135986763983965,
+      "learning_rate": 4.884057971014493e-05,
+      "loss": 0.1729,
+      "step": 2780
+    },
+    {
+      "epoch": 29.00404347826087,
+      "grad_norm": 0.04626130685210228,
+      "learning_rate": 4.881642512077295e-05,
+      "loss": 0.0708,
+      "step": 2790
+    },
+    {
+      "epoch": 29.00404347826087,
+      "eval_accuracy": 0.8114754098360656,
+      "eval_f1": 0.8062107421774495,
+      "eval_loss": 1.0806177854537964,
+      "eval_runtime": 75.2291,
+      "eval_samples_per_second": 1.622,
+      "eval_steps_per_second": 0.213,
+      "step": 2790
+    },
+    {
+      "epoch": 30.000434782608696,
+      "grad_norm": 0.009849685244262218,
+      "learning_rate": 4.879227053140097e-05,
+      "loss": 0.0661,
+      "step": 2800
+    },
+    {
+      "epoch": 30.000869565217393,
+      "grad_norm": 0.009964972734451294,
+      "learning_rate": 4.8768115942028986e-05,
+      "loss": 0.1059,
+      "step": 2810
+    },
+    {
+      "epoch": 30.001304347826085,
+      "grad_norm": 0.016774578019976616,
+      "learning_rate": 4.874396135265701e-05,
+      "loss": 0.0907,
+      "step": 2820
+    },
+    {
+      "epoch": 30.001739130434782,
+      "grad_norm": 0.040092673152685165,
+      "learning_rate": 4.871980676328503e-05,
+      "loss": 0.0298,
+      "step": 2830
+    },
+    {
+      "epoch": 30.002173913043478,
+      "grad_norm": 0.12641261518001556,
+      "learning_rate": 4.8695652173913046e-05,
+      "loss": 0.0006,
+      "step": 2840
+    },
+    {
+      "epoch": 30.002608695652174,
+      "grad_norm": 0.009654521942138672,
+      "learning_rate": 4.8671497584541064e-05,
+      "loss": 0.0342,
+      "step": 2850
+    },
+    {
+      "epoch": 30.00304347826087,
+      "grad_norm": 5.2498040199279785,
+      "learning_rate": 4.864734299516908e-05,
+      "loss": 0.249,
+      "step": 2860
+    },
+    {
+      "epoch": 30.003478260869564,
+      "grad_norm": 0.485901802778244,
+      "learning_rate": 4.8623188405797106e-05,
+      "loss": 0.0121,
+      "step": 2870
+    },
+    {
+      "epoch": 30.00391304347826,
+      "grad_norm": 0.030848659574985504,
+      "learning_rate": 4.8599033816425124e-05,
+      "loss": 0.0335,
+      "step": 2880
+    },
+    {
+      "epoch": 30.00404347826087,
+      "eval_accuracy": 0.819672131147541,
+      "eval_f1": 0.8112946016110332,
+      "eval_loss": 1.1136860847473145,
+      "eval_runtime": 74.2045,
+      "eval_samples_per_second": 1.644,
+      "eval_steps_per_second": 0.216,
+      "step": 2883
+    },
+    {
+      "epoch": 31.000304347826088,
+      "grad_norm": 2.0275466442108154,
+      "learning_rate": 4.857487922705314e-05,
+      "loss": 0.0145,
+      "step": 2890
+    },
+    {
+      "epoch": 31.000739130434784,
+      "grad_norm": 10.900162696838379,
+      "learning_rate": 4.855072463768116e-05,
+      "loss": 0.0999,
+      "step": 2900
+    },
+    {
+      "epoch": 31.001173913043477,
+      "grad_norm": 0.12738248705863953,
+      "learning_rate": 4.8526570048309184e-05,
+      "loss": 0.0026,
+      "step": 2910
+    },
+    {
+      "epoch": 31.001608695652173,
+      "grad_norm": 29.855815887451172,
+      "learning_rate": 4.85024154589372e-05,
+      "loss": 0.2289,
+      "step": 2920
+    },
+    {
+      "epoch": 31.00204347826087,
+      "grad_norm": 0.00635722279548645,
+      "learning_rate": 4.847826086956522e-05,
+      "loss": 0.0039,
+      "step": 2930
+    },
+    {
+      "epoch": 31.002478260869566,
+      "grad_norm": 0.007796896155923605,
+      "learning_rate": 4.845410628019324e-05,
+      "loss": 0.1027,
+      "step": 2940
+    },
+    {
+      "epoch": 31.002913043478262,
+      "grad_norm": 0.011087978258728981,
+      "learning_rate": 4.8429951690821256e-05,
+      "loss": 0.1863,
+      "step": 2950
+    },
+    {
+      "epoch": 31.003347826086955,
+      "grad_norm": 0.16981664299964905,
+      "learning_rate": 4.840579710144928e-05,
+      "loss": 0.1017,
+      "step": 2960
+    },
+    {
+      "epoch": 31.00378260869565,
+      "grad_norm": 0.005533864255994558,
+      "learning_rate": 4.83816425120773e-05,
+      "loss": 0.0571,
+      "step": 2970
+    },
+    {
+      "epoch": 31.00404347826087,
+      "eval_accuracy": 0.7540983606557377,
+      "eval_f1": 0.7606644114840837,
+      "eval_loss": 1.6491549015045166,
+      "eval_runtime": 72.4132,
+      "eval_samples_per_second": 1.685,
+      "eval_steps_per_second": 0.221,
+      "step": 2976
+    },
+    {
+      "epoch": 32.000173913043476,
+      "grad_norm": 4.148243427276611,
+      "learning_rate": 4.8357487922705316e-05,
+      "loss": 0.2991,
+      "step": 2980
+    },
+    {
+      "epoch": 32.000608695652176,
+      "grad_norm": 0.16210371255874634,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.0832,
+      "step": 2990
+    },
+    {
+      "epoch": 32.00104347826087,
+      "grad_norm": 0.11444944888353348,
+      "learning_rate": 4.830917874396135e-05,
+      "loss": 0.131,
+      "step": 3000
+    },
+    {
+      "epoch": 32.00147826086957,
+      "grad_norm": 20.465253829956055,
+      "learning_rate": 4.8285024154589376e-05,
+      "loss": 0.0559,
+      "step": 3010
+    },
+    {
+      "epoch": 32.00191304347826,
+      "grad_norm": 0.008609439246356487,
+      "learning_rate": 4.8260869565217394e-05,
+      "loss": 0.1028,
+      "step": 3020
+    },
+    {
+      "epoch": 32.002347826086954,
+      "grad_norm": 0.005297601222991943,
+      "learning_rate": 4.823671497584542e-05,
+      "loss": 0.0911,
+      "step": 3030
+    },
+    {
+      "epoch": 32.002782608695654,
+      "grad_norm": 0.009243907406926155,
+      "learning_rate": 4.821256038647343e-05,
+      "loss": 0.4324,
+      "step": 3040
+    },
+    {
+      "epoch": 32.00321739130435,
+      "grad_norm": 0.07699494808912277,
+      "learning_rate": 4.818840579710145e-05,
+      "loss": 0.1528,
+      "step": 3050
+    },
+    {
+      "epoch": 32.003652173913046,
+      "grad_norm": 0.5034440159797668,
+      "learning_rate": 4.816425120772947e-05,
+      "loss": 0.0965,
+      "step": 3060
+    },
+    {
+      "epoch": 32.00404347826087,
+      "eval_accuracy": 0.8770491803278688,
+      "eval_f1": 0.877946484907584,
+      "eval_loss": 0.7584701180458069,
+      "eval_runtime": 68.4338,
+      "eval_samples_per_second": 1.783,
+      "eval_steps_per_second": 0.234,
+      "step": 3069
+    },
+    {
+      "epoch": 4.347826086956522e-05,
+      "grad_norm": 0.03973979130387306,
+      "learning_rate": 4.814009661835749e-05,
+      "loss": 0.0007,
+      "step": 3070
+    },
+    {
+      "epoch": 0.0004782608695652174,
+      "grad_norm": 0.011583452112972736,
+      "learning_rate": 4.8115942028985514e-05,
+      "loss": 0.0774,
+      "step": 3080
+    },
+    {
+      "epoch": 0.0009130434782608696,
+      "grad_norm": 0.13716067373752594,
+      "learning_rate": 4.8091787439613525e-05,
+      "loss": 0.0874,
+      "step": 3090
+    },
+    {
+      "epoch": 0.0013478260869565217,
+      "grad_norm": 52.338008880615234,
+      "learning_rate": 4.806763285024155e-05,
+      "loss": 0.0513,
+      "step": 3100
+    },
+    {
+      "epoch": 0.0017826086956521739,
+      "grad_norm": 14.776463508605957,
+      "learning_rate": 4.804347826086957e-05,
+      "loss": 0.1144,
+      "step": 3110
+    },
+    {
+      "epoch": 0.002217391304347826,
+      "grad_norm": 0.02545876055955887,
+      "learning_rate": 4.8019323671497586e-05,
+      "loss": 0.1638,
+      "step": 3120
+    },
+    {
+      "epoch": 0.0026521739130434784,
+      "grad_norm": 0.31789082288742065,
+      "learning_rate": 4.799516908212561e-05,
+      "loss": 0.1676,
+      "step": 3130
+    },
+    {
+      "epoch": 0.0030869565217391303,
+      "grad_norm": 2.230436325073242,
+      "learning_rate": 4.797101449275362e-05,
+      "loss": 0.0492,
+      "step": 3140
+    },
+    {
+      "epoch": 0.0035217391304347826,
+      "grad_norm": 0.008552751503884792,
+      "learning_rate": 4.7946859903381646e-05,
+      "loss": 0.1069,
+      "step": 3150
+    },
+    {
+      "epoch": 0.003956521739130435,
+      "grad_norm": 0.006410131696611643,
+      "learning_rate": 4.7922705314009664e-05,
+      "loss": 0.0019,
+      "step": 3160
+    },
+    {
+      "epoch": 0.004043478260869565,
+      "eval_accuracy": 0.8688524590163934,
+      "eval_f1": 0.8677470707315438,
+      "eval_loss": 0.8008872866630554,
+      "eval_runtime": 75.7619,
+      "eval_samples_per_second": 1.61,
+      "eval_steps_per_second": 0.211,
+      "step": 3162
+    },
+    {
+      "epoch": 1.0003478260869565,
+      "grad_norm": 0.0073925890028476715,
+      "learning_rate": 4.789855072463768e-05,
+      "loss": 0.0007,
+      "step": 3170
+    },
+    {
+      "epoch": 1.0007826086956522,
+      "grad_norm": 0.009194503538310528,
+      "learning_rate": 4.7874396135265706e-05,
+      "loss": 0.0828,
+      "step": 3180
+    },
+    {
+      "epoch": 1.0012173913043478,
+      "grad_norm": 0.01020804513245821,
+      "learning_rate": 4.785024154589372e-05,
+      "loss": 0.0298,
+      "step": 3190
+    },
+    {
+      "epoch": 1.0016521739130435,
+      "grad_norm": 46.36565017700195,
+      "learning_rate": 4.782608695652174e-05,
+      "loss": 0.0086,
+      "step": 3200
+    },
+    {
+      "epoch": 1.0020869565217392,
+      "grad_norm": 0.007284100167453289,
+      "learning_rate": 4.780193236714976e-05,
+      "loss": 0.1071,
+      "step": 3210
+    },
+    {
+      "epoch": 1.0025217391304349,
+      "grad_norm": 3.947695016860962,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 0.0892,
+      "step": 3220
+    },
+    {
+      "epoch": 1.0029565217391305,
+      "grad_norm": 49.097679138183594,
+      "learning_rate": 4.77536231884058e-05,
+      "loss": 0.1349,
+      "step": 3230
+    },
+    {
+      "epoch": 1.0033913043478262,
+      "grad_norm": 0.18935832381248474,
+      "learning_rate": 4.772946859903382e-05,
+      "loss": 0.062,
+      "step": 3240
+    },
+    {
+      "epoch": 1.0038260869565216,
+      "grad_norm": 2.587529182434082,
+      "learning_rate": 4.770531400966184e-05,
+      "loss": 0.0724,
+      "step": 3250
+    },
+    {
+      "epoch": 1.0040434782608696,
+      "eval_accuracy": 0.8770491803278688,
+      "eval_f1": 0.8773590184659386,
+      "eval_loss": 0.7587926983833313,
+      "eval_runtime": 73.9063,
+      "eval_samples_per_second": 1.651,
+      "eval_steps_per_second": 0.216,
+      "step": 3255
+    },
+    {
+      "epoch": 2.0002173913043477,
+      "grad_norm": 0.12760405242443085,
+      "learning_rate": 4.7681159420289855e-05,
+      "loss": 0.0737,
+      "step": 3260
+    },
+    {
+      "epoch": 2.0006521739130436,
+      "grad_norm": 0.626366376876831,
+      "learning_rate": 4.765700483091788e-05,
+      "loss": 0.04,
+      "step": 3270
+    },
+    {
+      "epoch": 2.001086956521739,
+      "grad_norm": 0.5861339569091797,
+      "learning_rate": 4.76328502415459e-05,
+      "loss": 0.0014,
+      "step": 3280
+    },
+    {
+      "epoch": 2.001521739130435,
+      "grad_norm": 0.016094276681542397,
+      "learning_rate": 4.7608695652173916e-05,
+      "loss": 0.0006,
+      "step": 3290
+    },
+    {
+      "epoch": 2.0019565217391304,
+      "grad_norm": 0.03283320367336273,
+      "learning_rate": 4.7584541062801933e-05,
+      "loss": 0.0004,
+      "step": 3300
+    },
+    {
+      "epoch": 2.0023913043478263,
+      "grad_norm": 0.007951553910970688,
+      "learning_rate": 4.756038647342995e-05,
+      "loss": 0.0869,
+      "step": 3310
+    },
+    {
+      "epoch": 2.0028260869565218,
+      "grad_norm": 78.94805145263672,
+      "learning_rate": 4.7536231884057976e-05,
+      "loss": 0.0195,
+      "step": 3320
+    },
+    {
+      "epoch": 2.003260869565217,
+      "grad_norm": 0.00863903108984232,
+      "learning_rate": 4.7512077294685994e-05,
+      "loss": 0.0031,
+      "step": 3330
+    },
+    {
+      "epoch": 2.003695652173913,
+      "grad_norm": 0.010919756256043911,
+      "learning_rate": 4.748792270531401e-05,
+      "loss": 0.0539,
+      "step": 3340
+    },
+    {
+      "epoch": 2.0040434782608694,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8845549202267897,
+      "eval_loss": 0.784960150718689,
+      "eval_runtime": 71.8592,
+      "eval_samples_per_second": 1.698,
+      "eval_steps_per_second": 0.223,
+      "step": 3348
+    },
+    {
+      "epoch": 3.000086956521739,
+      "grad_norm": 2.7393057346343994,
+      "learning_rate": 4.746376811594203e-05,
+      "loss": 0.0407,
+      "step": 3350
+    },
+    {
+      "epoch": 3.0005217391304346,
+      "grad_norm": 0.034644246101379395,
+      "learning_rate": 4.743961352657005e-05,
+      "loss": 0.0011,
+      "step": 3360
+    },
+    {
+      "epoch": 3.0009565217391305,
+      "grad_norm": 0.042398907244205475,
+      "learning_rate": 4.741545893719807e-05,
+      "loss": 0.0005,
+      "step": 3370
+    },
+    {
+      "epoch": 3.001391304347826,
+      "grad_norm": 0.005558915436267853,
+      "learning_rate": 4.739130434782609e-05,
+      "loss": 0.0004,
+      "step": 3380
+    },
+    {
+      "epoch": 3.001826086956522,
+      "grad_norm": 0.0052604591473937035,
+      "learning_rate": 4.736714975845411e-05,
+      "loss": 0.0087,
+      "step": 3390
+    },
+    {
+      "epoch": 3.0022608695652173,
+      "grad_norm": 0.0026659213472157717,
+      "learning_rate": 4.7342995169082125e-05,
+      "loss": 0.0153,
+      "step": 3400
+    },
+    {
+      "epoch": 3.002695652173913,
+      "grad_norm": 0.005264274310320616,
+      "learning_rate": 4.731884057971015e-05,
+      "loss": 0.0119,
+      "step": 3410
+    },
+    {
+      "epoch": 3.0031304347826087,
+      "grad_norm": 0.007553906179964542,
+      "learning_rate": 4.729468599033817e-05,
+      "loss": 0.0531,
+      "step": 3420
+    },
+    {
+      "epoch": 3.0035652173913046,
+      "grad_norm": 2.2136473655700684,
+      "learning_rate": 4.7270531400966185e-05,
+      "loss": 0.0227,
+      "step": 3430
+    },
+    {
+      "epoch": 3.004,
+      "grad_norm": 0.020709699019789696,
+      "learning_rate": 4.72463768115942e-05,
+      "loss": 0.1208,
+      "step": 3440
+    },
+    {
+      "epoch": 3.0040434782608694,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8853364802271643,
+      "eval_loss": 0.7922825813293457,
+      "eval_runtime": 76.251,
+      "eval_samples_per_second": 1.6,
+      "eval_steps_per_second": 0.21,
+      "step": 3441
+    },
+    {
+      "epoch": 4.0003913043478265,
+      "grad_norm": 0.029359057545661926,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.0104,
+      "step": 3450
+    },
+    {
+      "epoch": 4.000826086956522,
+      "grad_norm": 0.0061401985585689545,
+      "learning_rate": 4.7198067632850246e-05,
+      "loss": 0.0329,
+      "step": 3460
+    },
+    {
+      "epoch": 4.001260869565217,
+      "grad_norm": 0.004194505978375673,
+      "learning_rate": 4.7173913043478264e-05,
+      "loss": 0.0005,
+      "step": 3470
+    },
+    {
+      "epoch": 4.001695652173913,
+      "grad_norm": 0.016542797908186913,
+      "learning_rate": 4.714975845410628e-05,
+      "loss": 0.0005,
+      "step": 3480
+    },
+    {
+      "epoch": 4.002130434782608,
+      "grad_norm": 0.009181750938296318,
+      "learning_rate": 4.71256038647343e-05,
+      "loss": 0.0004,
+      "step": 3490
+    },
+    {
+      "epoch": 4.002565217391305,
+      "grad_norm": 0.0038063046522438526,
+      "learning_rate": 4.710144927536232e-05,
+      "loss": 0.0102,
+      "step": 3500
+    },
+    {
+      "epoch": 4.003,
+      "grad_norm": 0.012156417593359947,
+      "learning_rate": 4.707729468599034e-05,
+      "loss": 0.0101,
+      "step": 3510
+    },
+    {
+      "epoch": 4.003434782608696,
+      "grad_norm": 0.00890531949698925,
+      "learning_rate": 4.705314009661836e-05,
+      "loss": 0.0217,
+      "step": 3520
+    },
+    {
+      "epoch": 4.003869565217391,
+      "grad_norm": 0.004342174157500267,
+      "learning_rate": 4.7028985507246384e-05,
+      "loss": 0.0467,
+      "step": 3530
+    },
+    {
+      "epoch": 4.004043478260869,
+      "eval_accuracy": 0.860655737704918,
+      "eval_f1": 0.8612302252359274,
+      "eval_loss": 1.0240317583084106,
+      "eval_runtime": 63.4449,
+      "eval_samples_per_second": 1.923,
+      "eval_steps_per_second": 0.252,
+      "step": 3534
+    },
+    {
+      "epoch": 5.000260869565217,
+      "grad_norm": 0.025819310918450356,
+      "learning_rate": 4.7004830917874395e-05,
+      "loss": 0.0004,
+      "step": 3540
+    },
+    {
+      "epoch": 5.000695652173913,
+      "grad_norm": 0.010917126201093197,
+      "learning_rate": 4.698067632850241e-05,
+      "loss": 0.0163,
+      "step": 3550
+    },
+    {
+      "epoch": 5.001130434782609,
+      "grad_norm": 0.0050573805347085,
+      "learning_rate": 4.695652173913044e-05,
+      "loss": 0.2046,
+      "step": 3560
+    },
+    {
+      "epoch": 5.001565217391304,
+      "grad_norm": 22.131898880004883,
+      "learning_rate": 4.6932367149758455e-05,
+      "loss": 0.0945,
+      "step": 3570
+    },
+    {
+      "epoch": 5.002,
+      "grad_norm": 0.013935157097876072,
+      "learning_rate": 4.690821256038648e-05,
+      "loss": 0.0026,
+      "step": 3580
+    },
+    {
+      "epoch": 5.002434782608695,
+      "grad_norm": 0.006178060546517372,
+      "learning_rate": 4.68840579710145e-05,
+      "loss": 0.0018,
+      "step": 3590
+    },
+    {
+      "epoch": 5.002869565217392,
+      "grad_norm": 0.006322579458355904,
+      "learning_rate": 4.6859903381642516e-05,
+      "loss": 0.0309,
+      "step": 3600
+    },
+    {
+      "epoch": 5.003304347826087,
+      "grad_norm": 0.006985423155128956,
+      "learning_rate": 4.683574879227053e-05,
+      "loss": 0.0285,
+      "step": 3610
+    },
+    {
+      "epoch": 5.0037391304347825,
+      "grad_norm": 0.004971032030880451,
+      "learning_rate": 4.681159420289855e-05,
+      "loss": 0.0764,
+      "step": 3620
+    },
+    {
+      "epoch": 5.004043478260869,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8863337496985733,
+      "eval_loss": 0.7494808435440063,
+      "eval_runtime": 68.5855,
+      "eval_samples_per_second": 1.779,
+      "eval_steps_per_second": 0.233,
+      "step": 3627
+    },
+    {
+      "epoch": 6.0001304347826085,
+      "grad_norm": 0.26816919445991516,
+      "learning_rate": 4.6787439613526576e-05,
+      "loss": 0.0008,
+      "step": 3630
+    },
+    {
+      "epoch": 6.000565217391304,
+      "grad_norm": 0.011856114491820335,
+      "learning_rate": 4.6763285024154594e-05,
+      "loss": 0.08,
+      "step": 3640
+    },
+    {
+      "epoch": 6.001,
+      "grad_norm": 0.006870527286082506,
+      "learning_rate": 4.673913043478261e-05,
+      "loss": 0.0005,
+      "step": 3650
+    },
+    {
+      "epoch": 6.001434782608696,
+      "grad_norm": 0.20965741574764252,
+      "learning_rate": 4.671497584541063e-05,
+      "loss": 0.1102,
+      "step": 3660
+    },
+    {
+      "epoch": 6.001869565217391,
+      "grad_norm": 0.018843989819288254,
+      "learning_rate": 4.669082125603865e-05,
+      "loss": 0.0741,
+      "step": 3670
+    },
+    {
+      "epoch": 6.002304347826087,
+      "grad_norm": 97.60299682617188,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.049,
+      "step": 3680
+    },
+    {
+      "epoch": 6.002739130434783,
+      "grad_norm": 0.006012632045894861,
+      "learning_rate": 4.664251207729469e-05,
+      "loss": 0.0635,
+      "step": 3690
+    },
+    {
+      "epoch": 6.0031739130434785,
+      "grad_norm": 10.848343849182129,
+      "learning_rate": 4.661835748792271e-05,
+      "loss": 0.0984,
+      "step": 3700
+    },
+    {
+      "epoch": 6.003608695652174,
+      "grad_norm": 0.004343053791671991,
+      "learning_rate": 4.6594202898550725e-05,
+      "loss": 0.0887,
+      "step": 3710
+    },
+    {
+      "epoch": 6.004043478260869,
+      "grad_norm": 0.006757623050361872,
+      "learning_rate": 4.657004830917875e-05,
+      "loss": 0.0006,
+      "step": 3720
+    },
+    {
+      "epoch": 6.004043478260869,
+      "eval_accuracy": 0.8360655737704918,
+      "eval_f1": 0.83469500668468,
+      "eval_loss": 0.9808096289634705,
+      "eval_runtime": 64.2569,
+      "eval_samples_per_second": 1.899,
+      "eval_steps_per_second": 0.249,
+      "step": 3720
+    },
+    {
+      "epoch": 7.0004347826086954,
+      "grad_norm": 0.014435753226280212,
+      "learning_rate": 4.654589371980677e-05,
+      "loss": 0.0005,
+      "step": 3730
+    },
+    {
+      "epoch": 7.000869565217391,
+      "grad_norm": 0.0036905386950820684,
+      "learning_rate": 4.6521739130434785e-05,
+      "loss": 0.0005,
+      "step": 3740
+    },
+    {
+      "epoch": 7.001304347826087,
+      "grad_norm": 0.005206993781030178,
+      "learning_rate": 4.64975845410628e-05,
+      "loss": 0.0003,
+      "step": 3750
+    },
+    {
+      "epoch": 7.001739130434783,
+      "grad_norm": 0.015529815107584,
+      "learning_rate": 4.647342995169082e-05,
+      "loss": 0.0177,
+      "step": 3760
+    },
+    {
+      "epoch": 7.002173913043478,
+      "grad_norm": 0.02242157608270645,
+      "learning_rate": 4.6449275362318846e-05,
+      "loss": 0.0003,
+      "step": 3770
+    },
+    {
+      "epoch": 7.002608695652174,
+      "grad_norm": 0.004425575491040945,
+      "learning_rate": 4.642512077294686e-05,
+      "loss": 0.0123,
+      "step": 3780
+    },
+    {
+      "epoch": 7.00304347826087,
+      "grad_norm": 0.0038235390093177557,
+      "learning_rate": 4.640096618357488e-05,
+      "loss": 0.0101,
+      "step": 3790
+    },
+    {
+      "epoch": 7.003478260869565,
+      "grad_norm": 0.051797546446323395,
+      "learning_rate": 4.63768115942029e-05,
+      "loss": 0.0724,
+      "step": 3800
+    },
+    {
+      "epoch": 7.003913043478261,
+      "grad_norm": 0.021821072325110435,
+      "learning_rate": 4.635265700483092e-05,
+      "loss": 0.1386,
+      "step": 3810
+    },
+    {
+      "epoch": 7.004043478260869,
+      "eval_accuracy": 0.8770491803278688,
+      "eval_f1": 0.8770793838312594,
+      "eval_loss": 0.7652427554130554,
+      "eval_runtime": 71.3373,
+      "eval_samples_per_second": 1.71,
+      "eval_steps_per_second": 0.224,
+      "step": 3813
+    },
+    {
+      "epoch": 8.000304347826088,
+      "grad_norm": 0.0032370842527598143,
+      "learning_rate": 4.632850241545894e-05,
+      "loss": 0.0129,
+      "step": 3820
+    },
+    {
+      "epoch": 8.000739130434782,
+      "grad_norm": 1.7167001962661743,
+      "learning_rate": 4.630434782608696e-05,
+      "loss": 0.021,
+      "step": 3830
+    },
+    {
+      "epoch": 8.001173913043479,
+      "grad_norm": 0.003859913209453225,
+      "learning_rate": 4.628019323671498e-05,
+      "loss": 0.0015,
+      "step": 3840
+    },
+    {
+      "epoch": 8.001608695652173,
+      "grad_norm": 0.0031530093401670456,
+      "learning_rate": 4.6256038647342995e-05,
+      "loss": 0.0003,
+      "step": 3850
+    },
+    {
+      "epoch": 8.00204347826087,
+      "grad_norm": 0.007489124778658152,
+      "learning_rate": 4.623188405797101e-05,
+      "loss": 0.0081,
+      "step": 3860
+    },
+    {
+      "epoch": 8.002478260869566,
+      "grad_norm": 2.5608975887298584,
+      "learning_rate": 4.620772946859904e-05,
+      "loss": 0.1232,
+      "step": 3870
+    },
+    {
+      "epoch": 8.00291304347826,
+      "grad_norm": 63.46572494506836,
+      "learning_rate": 4.6183574879227055e-05,
+      "loss": 0.0507,
+      "step": 3880
+    },
+    {
+      "epoch": 8.003347826086957,
+      "grad_norm": 124.65734100341797,
+      "learning_rate": 4.615942028985508e-05,
+      "loss": 0.1479,
+      "step": 3890
+    },
+    {
+      "epoch": 8.003782608695651,
+      "grad_norm": 57.03828430175781,
+      "learning_rate": 4.613526570048309e-05,
+      "loss": 0.1791,
+      "step": 3900
+    },
+    {
+      "epoch": 8.00404347826087,
+      "eval_accuracy": 0.8360655737704918,
+      "eval_f1": 0.8328396085819532,
+      "eval_loss": 1.0916365385055542,
+      "eval_runtime": 76.5479,
+      "eval_samples_per_second": 1.594,
+      "eval_steps_per_second": 0.209,
+      "step": 3906
+    },
+    {
+      "epoch": 9.000173913043477,
+      "grad_norm": 0.017659667879343033,
+      "learning_rate": 4.6111111111111115e-05,
+      "loss": 0.0024,
+      "step": 3910
+    },
+    {
+      "epoch": 9.000608695652174,
+      "grad_norm": 0.004569427575916052,
+      "learning_rate": 4.608695652173913e-05,
+      "loss": 0.1475,
+      "step": 3920
+    },
+    {
+      "epoch": 9.00104347826087,
+      "grad_norm": 0.4839405417442322,
+      "learning_rate": 4.606280193236715e-05,
+      "loss": 0.0576,
+      "step": 3930
+    },
+    {
+      "epoch": 9.001478260869565,
+      "grad_norm": 0.005257844924926758,
+      "learning_rate": 4.6038647342995176e-05,
+      "loss": 0.0811,
+      "step": 3940
+    },
+    {
+      "epoch": 9.001913043478261,
+      "grad_norm": 0.005183045752346516,
+      "learning_rate": 4.601449275362319e-05,
+      "loss": 0.2326,
+      "step": 3950
+    },
+    {
+      "epoch": 9.002347826086957,
+      "grad_norm": 0.20619826018810272,
+      "learning_rate": 4.599033816425121e-05,
+      "loss": 0.084,
+      "step": 3960
+    },
+    {
+      "epoch": 9.002782608695652,
+      "grad_norm": 0.10698182135820389,
+      "learning_rate": 4.596618357487923e-05,
+      "loss": 0.0156,
+      "step": 3970
+    },
+    {
+      "epoch": 9.003217391304348,
+      "grad_norm": 0.004087397828698158,
+      "learning_rate": 4.594202898550725e-05,
+      "loss": 0.0312,
+      "step": 3980
+    },
+    {
+      "epoch": 9.003652173913043,
+      "grad_norm": 0.004443726502358913,
+      "learning_rate": 4.591787439613527e-05,
+      "loss": 0.0169,
+      "step": 3990
+    },
+    {
+      "epoch": 9.00404347826087,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8848651800896989,
+      "eval_loss": 0.6546049118041992,
+      "eval_runtime": 74.6104,
+      "eval_samples_per_second": 1.635,
+      "eval_steps_per_second": 0.214,
+      "step": 3999
+    },
+    {
+      "epoch": 10.000043478260869,
+      "grad_norm": 0.004015814512968063,
+      "learning_rate": 4.589371980676328e-05,
+      "loss": 0.086,
+      "step": 4000
+    },
+    {
+      "epoch": 10.000478260869565,
+      "grad_norm": 0.0028713741339743137,
+      "learning_rate": 4.586956521739131e-05,
+      "loss": 0.0137,
+      "step": 4010
+    },
+    {
+      "epoch": 10.000913043478262,
+      "grad_norm": 0.0047510224394500256,
+      "learning_rate": 4.5845410628019325e-05,
+      "loss": 0.0065,
+      "step": 4020
+    },
+    {
+      "epoch": 10.001347826086956,
+      "grad_norm": 0.0044855643063783646,
+      "learning_rate": 4.582125603864735e-05,
+      "loss": 0.0504,
+      "step": 4030
+    },
+    {
+      "epoch": 10.001782608695652,
+      "grad_norm": 0.07487785071134567,
+      "learning_rate": 4.579710144927537e-05,
+      "loss": 0.0837,
+      "step": 4040
+    },
+    {
+      "epoch": 10.002217391304347,
+      "grad_norm": 0.025662241503596306,
+      "learning_rate": 4.577294685990338e-05,
+      "loss": 0.0422,
+      "step": 4050
+    },
+    {
+      "epoch": 10.002652173913043,
+      "grad_norm": 3.2152881622314453,
+      "learning_rate": 4.57487922705314e-05,
+      "loss": 0.0009,
+      "step": 4060
+    },
+    {
+      "epoch": 10.00308695652174,
+      "grad_norm": 0.009620684199035168,
+      "learning_rate": 4.572463768115942e-05,
+      "loss": 0.0598,
+      "step": 4070
+    },
+    {
+      "epoch": 10.003521739130434,
+      "grad_norm": 0.0028383173048496246,
+      "learning_rate": 4.5700483091787445e-05,
+      "loss": 0.1004,
+      "step": 4080
+    },
+    {
+      "epoch": 10.00395652173913,
+      "grad_norm": 0.23549234867095947,
+      "learning_rate": 4.567632850241546e-05,
+      "loss": 0.0228,
+      "step": 4090
+    },
+    {
+      "epoch": 10.00404347826087,
+      "eval_accuracy": 0.8524590163934426,
+      "eval_f1": 0.8534775589608828,
+      "eval_loss": 0.9656055569648743,
+      "eval_runtime": 77.7343,
+      "eval_samples_per_second": 1.569,
+      "eval_steps_per_second": 0.206,
+      "step": 4092
+    },
+    {
+      "epoch": 11.000347826086957,
+      "grad_norm": 0.001953911269083619,
+      "learning_rate": 4.565217391304348e-05,
+      "loss": 0.0454,
+      "step": 4100
+    },
+    {
+      "epoch": 11.000782608695653,
+      "grad_norm": 0.09437835961580276,
+      "learning_rate": 4.56280193236715e-05,
+      "loss": 0.0004,
+      "step": 4110
+    },
+    {
+      "epoch": 11.001217391304348,
+      "grad_norm": 0.009714662097394466,
+      "learning_rate": 4.560386473429952e-05,
+      "loss": 0.0106,
+      "step": 4120
+    },
+    {
+      "epoch": 11.001652173913044,
+      "grad_norm": 0.002876744605600834,
+      "learning_rate": 4.557971014492754e-05,
+      "loss": 0.0918,
+      "step": 4130
+    },
+    {
+      "epoch": 11.002086956521739,
+      "grad_norm": 0.0034505492076277733,
+      "learning_rate": 4.555555555555556e-05,
+      "loss": 0.0166,
+      "step": 4140
+    },
+    {
+      "epoch": 11.002521739130435,
+      "grad_norm": 0.0048155467957258224,
+      "learning_rate": 4.553140096618358e-05,
+      "loss": 0.3045,
+      "step": 4150
+    },
+    {
+      "epoch": 11.002956521739131,
+      "grad_norm": 1.3779292106628418,
+      "learning_rate": 4.5507246376811595e-05,
+      "loss": 0.0084,
+      "step": 4160
+    },
+    {
+      "epoch": 11.003391304347826,
+      "grad_norm": 0.0718054547905922,
+      "learning_rate": 4.548309178743961e-05,
+      "loss": 0.2187,
+      "step": 4170
+    },
+    {
+      "epoch": 11.003826086956522,
+      "grad_norm": 0.044493287801742554,
+      "learning_rate": 4.545893719806764e-05,
+      "loss": 0.0307,
+      "step": 4180
+    },
+    {
+      "epoch": 11.00404347826087,
+      "eval_accuracy": 0.7950819672131147,
+      "eval_f1": 0.7963983516128617,
+      "eval_loss": 1.334099292755127,
+      "eval_runtime": 79.1262,
+      "eval_samples_per_second": 1.542,
+      "eval_steps_per_second": 0.202,
+      "step": 4185
+    },
+    {
+      "epoch": 12.000217391304348,
+      "grad_norm": 12.314984321594238,
+      "learning_rate": 4.5434782608695655e-05,
+      "loss": 0.2449,
+      "step": 4190
+    },
+    {
+      "epoch": 12.000652173913043,
+      "grad_norm": 0.4841020107269287,
+      "learning_rate": 4.541062801932367e-05,
+      "loss": 0.0824,
+      "step": 4200
+    },
+    {
+      "epoch": 12.001086956521739,
+      "grad_norm": 31.790552139282227,
+      "learning_rate": 4.538647342995169e-05,
+      "loss": 0.0731,
+      "step": 4210
+    },
+    {
+      "epoch": 12.001521739130435,
+      "grad_norm": 0.0036116011906415224,
+      "learning_rate": 4.5362318840579715e-05,
+      "loss": 0.0276,
+      "step": 4220
+    },
+    {
+      "epoch": 12.00195652173913,
+      "grad_norm": 0.016497021540999413,
+      "learning_rate": 4.533816425120773e-05,
+      "loss": 0.096,
+      "step": 4230
+    },
+    {
+      "epoch": 12.002391304347826,
+      "grad_norm": 0.016652707010507584,
+      "learning_rate": 4.531400966183575e-05,
+      "loss": 0.4404,
+      "step": 4240
+    },
+    {
+      "epoch": 12.00282608695652,
+      "grad_norm": 0.03973345831036568,
+      "learning_rate": 4.528985507246377e-05,
+      "loss": 0.0343,
+      "step": 4250
+    },
+    {
+      "epoch": 12.003260869565217,
+      "grad_norm": 133.8014373779297,
+      "learning_rate": 4.5265700483091786e-05,
+      "loss": 0.497,
+      "step": 4260
+    },
+    {
+      "epoch": 12.003695652173914,
+      "grad_norm": 2.6162045001983643,
+      "learning_rate": 4.524154589371981e-05,
+      "loss": 0.5106,
+      "step": 4270
+    },
+    {
+      "epoch": 12.00404347826087,
+      "eval_accuracy": 0.7950819672131147,
+      "eval_f1": 0.7876538622910378,
+      "eval_loss": 1.2499514818191528,
+      "eval_runtime": 68.1294,
+      "eval_samples_per_second": 1.791,
+      "eval_steps_per_second": 0.235,
+      "step": 4278
+    },
+    {
+      "epoch": 13.00008695652174,
+      "grad_norm": 41.279598236083984,
+      "learning_rate": 4.521739130434783e-05,
+      "loss": 0.1115,
+      "step": 4280
+    },
+    {
+      "epoch": 13.000521739130434,
+      "grad_norm": 1.6680902242660522,
+      "learning_rate": 4.519323671497585e-05,
+      "loss": 0.018,
+      "step": 4290
+    },
+    {
+      "epoch": 13.00095652173913,
+      "grad_norm": 0.009661109186708927,
+      "learning_rate": 4.5169082125603865e-05,
+      "loss": 0.1721,
+      "step": 4300
+    },
+    {
+      "epoch": 13.001391304347827,
+      "grad_norm": 0.02459769882261753,
+      "learning_rate": 4.514492753623188e-05,
+      "loss": 0.0452,
+      "step": 4310
+    },
+    {
+      "epoch": 13.001826086956521,
+      "grad_norm": 0.026296913623809814,
+      "learning_rate": 4.512077294685991e-05,
+      "loss": 0.1139,
+      "step": 4320
+    },
+    {
+      "epoch": 13.002260869565218,
+      "grad_norm": 0.31147295236587524,
+      "learning_rate": 4.5096618357487925e-05,
+      "loss": 0.0273,
+      "step": 4330
+    },
+    {
+      "epoch": 13.002695652173912,
+      "grad_norm": 12.396149635314941,
+      "learning_rate": 4.507246376811595e-05,
+      "loss": 0.3046,
+      "step": 4340
+    },
+    {
+      "epoch": 13.003130434782609,
+      "grad_norm": 0.010565920732915401,
+      "learning_rate": 4.504830917874396e-05,
+      "loss": 0.0204,
+      "step": 4350
+    },
+    {
+      "epoch": 13.003565217391305,
+      "grad_norm": 0.004811236634850502,
+      "learning_rate": 4.502415458937198e-05,
+      "loss": 0.0167,
+      "step": 4360
+    },
+    {
+      "epoch": 13.004,
+      "grad_norm": 0.008650255389511585,
+      "learning_rate": 4.5e-05,
+      "loss": 0.0108,
+      "step": 4370
+    },
+    {
+      "epoch": 13.00404347826087,
+      "eval_accuracy": 0.8524590163934426,
+      "eval_f1": 0.8495038417270663,
+      "eval_loss": 0.8027352690696716,
+      "eval_runtime": 94.8246,
+      "eval_samples_per_second": 1.287,
+      "eval_steps_per_second": 0.169,
+      "step": 4371
+    },
+    {
+      "epoch": 14.000391304347826,
+      "grad_norm": 2.1783447265625,
+      "learning_rate": 4.497584541062802e-05,
+      "loss": 0.0578,
+      "step": 4380
+    },
+    {
+      "epoch": 14.000826086956522,
+      "grad_norm": 0.008437794633209705,
+      "learning_rate": 4.4951690821256045e-05,
+      "loss": 0.0542,
+      "step": 4390
+    },
+    {
+      "epoch": 14.001260869565217,
+      "grad_norm": 0.006100878585129976,
+      "learning_rate": 4.492753623188406e-05,
+      "loss": 0.0004,
+      "step": 4400
+    },
+    {
+      "epoch": 14.001695652173913,
+      "grad_norm": 0.005631875246763229,
+      "learning_rate": 4.4903381642512074e-05,
+      "loss": 0.0932,
+      "step": 4410
+    },
+    {
+      "epoch": 14.00213043478261,
+      "grad_norm": 52.9527587890625,
+      "learning_rate": 4.48792270531401e-05,
+      "loss": 0.1454,
+      "step": 4420
+    },
+    {
+      "epoch": 14.002565217391304,
+      "grad_norm": 0.0049645500257611275,
+      "learning_rate": 4.4855072463768117e-05,
+      "loss": 0.0007,
+      "step": 4430
+    },
+    {
+      "epoch": 14.003,
+      "grad_norm": 0.005069791804999113,
+      "learning_rate": 4.483091787439614e-05,
+      "loss": 0.0104,
+      "step": 4440
+    },
+    {
+      "epoch": 14.003434782608696,
+      "grad_norm": 0.06516723334789276,
+      "learning_rate": 4.480676328502416e-05,
+      "loss": 0.1654,
+      "step": 4450
+    },
+    {
+      "epoch": 14.003869565217391,
+      "grad_norm": 0.01129752118140459,
+      "learning_rate": 4.478260869565218e-05,
+      "loss": 0.0873,
+      "step": 4460
+    },
+    {
+      "epoch": 14.00404347826087,
+      "eval_accuracy": 0.9016393442622951,
+      "eval_f1": 0.9017991182732418,
+      "eval_loss": 0.680787980556488,
+      "eval_runtime": 78.5271,
+      "eval_samples_per_second": 1.554,
+      "eval_steps_per_second": 0.204,
+      "step": 4464
+    },
+    {
+      "epoch": 15.000260869565217,
+      "grad_norm": 0.0042943209409713745,
+      "learning_rate": 4.4758454106280195e-05,
+      "loss": 0.0165,
+      "step": 4470
+    },
+    {
+      "epoch": 15.000695652173913,
+      "grad_norm": 0.0025938786566257477,
+      "learning_rate": 4.473429951690821e-05,
+      "loss": 0.0602,
+      "step": 4480
+    },
+    {
+      "epoch": 15.001130434782608,
+      "grad_norm": 0.02354370430111885,
+      "learning_rate": 4.471014492753624e-05,
+      "loss": 0.0003,
+      "step": 4490
+    },
+    {
+      "epoch": 15.001565217391304,
+      "grad_norm": 0.0034618491772562265,
+      "learning_rate": 4.4685990338164255e-05,
+      "loss": 0.0157,
+      "step": 4500
+    },
+    {
+      "epoch": 15.002,
+      "grad_norm": 0.002460119780153036,
+      "learning_rate": 4.466183574879227e-05,
+      "loss": 0.0834,
+      "step": 4510
+    },
+    {
+      "epoch": 15.002434782608695,
+      "grad_norm": 0.008845089003443718,
+      "learning_rate": 4.463768115942029e-05,
+      "loss": 0.1572,
+      "step": 4520
+    },
+    {
+      "epoch": 15.002869565217392,
+      "grad_norm": 0.004269142635166645,
+      "learning_rate": 4.4613526570048315e-05,
+      "loss": 0.3163,
+      "step": 4530
+    },
+    {
+      "epoch": 15.003304347826086,
+      "grad_norm": 0.04755193367600441,
+      "learning_rate": 4.458937198067633e-05,
+      "loss": 0.1112,
+      "step": 4540
+    },
+    {
+      "epoch": 15.003739130434782,
+      "grad_norm": 0.022059939801692963,
+      "learning_rate": 4.456521739130435e-05,
+      "loss": 0.0149,
+      "step": 4550
+    },
+    {
+      "epoch": 15.00404347826087,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8867104809727759,
+      "eval_loss": 0.6811984777450562,
+      "eval_runtime": 79.1497,
+      "eval_samples_per_second": 1.541,
+      "eval_steps_per_second": 0.202,
+      "step": 4557
+    },
+    {
+      "epoch": 16.00013043478261,
+      "grad_norm": 0.004382013343274593,
+      "learning_rate": 4.454106280193237e-05,
+      "loss": 0.0005,
+      "step": 4560
+    },
+    {
+      "epoch": 16.000565217391305,
+      "grad_norm": 0.007960804738104343,
+      "learning_rate": 4.4516908212560386e-05,
+      "loss": 0.0917,
+      "step": 4570
+    },
+    {
+      "epoch": 16.001,
+      "grad_norm": 22.968700408935547,
+      "learning_rate": 4.449275362318841e-05,
+      "loss": 0.1009,
+      "step": 4580
+    },
+    {
+      "epoch": 16.001434782608694,
+      "grad_norm": 0.004100767895579338,
+      "learning_rate": 4.446859903381643e-05,
+      "loss": 0.006,
+      "step": 4590
+    },
+    {
+      "epoch": 16.00186956521739,
+      "grad_norm": 41.614383697509766,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.0491,
+      "step": 4600
+    },
+    {
+      "epoch": 16.002304347826087,
+      "grad_norm": 75.32587432861328,
+      "learning_rate": 4.4420289855072464e-05,
+      "loss": 0.2419,
+      "step": 4610
+    },
+    {
+      "epoch": 16.002739130434783,
+      "grad_norm": 0.010296987369656563,
+      "learning_rate": 4.439613526570048e-05,
+      "loss": 0.0824,
+      "step": 4620
+    },
+    {
+      "epoch": 16.00317391304348,
+      "grad_norm": 0.010856476612389088,
+      "learning_rate": 4.437198067632851e-05,
+      "loss": 0.1039,
+      "step": 4630
+    },
+    {
+      "epoch": 16.003608695652172,
+      "grad_norm": 0.015950683504343033,
+      "learning_rate": 4.4347826086956525e-05,
+      "loss": 0.0357,
+      "step": 4640
+    },
+    {
+      "epoch": 16.00404347826087,
+      "grad_norm": 0.006432201247662306,
+      "learning_rate": 4.432367149758454e-05,
+      "loss": 0.0146,
+      "step": 4650
+    },
+    {
+      "epoch": 16.00404347826087,
+      "eval_accuracy": 0.9016393442622951,
+      "eval_f1": 0.9015790225488998,
+      "eval_loss": 0.6003494262695312,
+      "eval_runtime": 64.7244,
+      "eval_samples_per_second": 1.885,
+      "eval_steps_per_second": 0.247,
+      "step": 4650
+    },
+    {
+      "epoch": 17.000434782608696,
+      "grad_norm": 0.01721499115228653,
+      "learning_rate": 4.429951690821256e-05,
+      "loss": 0.0075,
+      "step": 4660
+    },
+    {
+      "epoch": 17.000869565217393,
+      "grad_norm": 0.010429782792925835,
+      "learning_rate": 4.427536231884058e-05,
+      "loss": 0.0034,
+      "step": 4670
+    },
+    {
+      "epoch": 17.001304347826085,
+      "grad_norm": 4.780106544494629,
+      "learning_rate": 4.42512077294686e-05,
+      "loss": 0.0837,
+      "step": 4680
+    },
+    {
+      "epoch": 17.001739130434782,
+      "grad_norm": 0.015756214037537575,
+      "learning_rate": 4.422705314009662e-05,
+      "loss": 0.0114,
+      "step": 4690
+    },
+    {
+      "epoch": 17.002173913043478,
+      "grad_norm": 0.3344515860080719,
+      "learning_rate": 4.4202898550724645e-05,
+      "loss": 0.0015,
+      "step": 4700
+    },
+    {
+      "epoch": 17.002608695652174,
+      "grad_norm": 2.3182766437530518,
+      "learning_rate": 4.4178743961352656e-05,
+      "loss": 0.0464,
+      "step": 4710
+    },
+    {
+      "epoch": 17.00304347826087,
+      "grad_norm": 0.008929559960961342,
+      "learning_rate": 4.4154589371980674e-05,
+      "loss": 0.0879,
+      "step": 4720
+    },
+    {
+      "epoch": 17.003478260869564,
+      "grad_norm": 0.004621615167707205,
+      "learning_rate": 4.41304347826087e-05,
+      "loss": 0.0379,
+      "step": 4730
+    },
+    {
+      "epoch": 17.00391304347826,
+      "grad_norm": 0.012686613015830517,
+      "learning_rate": 4.4106280193236716e-05,
+      "loss": 0.0994,
+      "step": 4740
+    },
+    {
+      "epoch": 17.00404347826087,
+      "eval_accuracy": 0.860655737704918,
+      "eval_f1": 0.8632714330882123,
+      "eval_loss": 0.8046178221702576,
+      "eval_runtime": 85.1421,
+      "eval_samples_per_second": 1.433,
+      "eval_steps_per_second": 0.188,
+      "step": 4743
+    },
+    {
+      "epoch": 18.000304347826088,
+      "grad_norm": 0.003794121090322733,
+      "learning_rate": 4.408212560386474e-05,
+      "loss": 0.0121,
+      "step": 4750
+    },
+    {
+      "epoch": 18.000739130434784,
+      "grad_norm": 0.002671042224392295,
+      "learning_rate": 4.405797101449275e-05,
+      "loss": 0.0112,
+      "step": 4760
+    },
+    {
+      "epoch": 18.001173913043477,
+      "grad_norm": 0.0030148469377309084,
+      "learning_rate": 4.403381642512078e-05,
+      "loss": 0.0222,
+      "step": 4770
+    },
+    {
+      "epoch": 18.001608695652173,
+      "grad_norm": 0.004704237449914217,
+      "learning_rate": 4.4009661835748794e-05,
+      "loss": 0.152,
+      "step": 4780
+    },
+    {
+      "epoch": 18.00204347826087,
+      "grad_norm": 0.003907118458300829,
+      "learning_rate": 4.398550724637681e-05,
+      "loss": 0.0005,
+      "step": 4790
+    },
+    {
+      "epoch": 18.002478260869566,
+      "grad_norm": 0.0029985394794493914,
+      "learning_rate": 4.396135265700484e-05,
+      "loss": 0.0003,
+      "step": 4800
+    },
+    {
+      "epoch": 18.002913043478262,
+      "grad_norm": 2.0193307399749756,
+      "learning_rate": 4.393719806763285e-05,
+      "loss": 0.1093,
+      "step": 4810
+    },
+    {
+      "epoch": 18.003347826086955,
+      "grad_norm": 0.007247489411383867,
+      "learning_rate": 4.391304347826087e-05,
+      "loss": 0.0969,
+      "step": 4820
+    },
+    {
+      "epoch": 18.00378260869565,
+      "grad_norm": 35.1983642578125,
+      "learning_rate": 4.388888888888889e-05,
+      "loss": 0.1715,
+      "step": 4830
+    },
+    {
+      "epoch": 18.00404347826087,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8853454924681606,
+      "eval_loss": 0.5631609559059143,
+      "eval_runtime": 81.4165,
+      "eval_samples_per_second": 1.498,
+      "eval_steps_per_second": 0.197,
+      "step": 4836
+    },
+    {
+      "epoch": 19.00017391304348,
+      "grad_norm": 0.24528348445892334,
+      "learning_rate": 4.3864734299516915e-05,
+      "loss": 0.0645,
+      "step": 4840
+    },
+    {
+      "epoch": 19.000608695652176,
+      "grad_norm": 0.018249379470944405,
+      "learning_rate": 4.384057971014493e-05,
+      "loss": 0.0146,
+      "step": 4850
+    },
+    {
+      "epoch": 19.00104347826087,
+      "grad_norm": 17.096920013427734,
+      "learning_rate": 4.3816425120772944e-05,
+      "loss": 0.0072,
+      "step": 4860
+    },
+    {
+      "epoch": 19.001478260869565,
+      "grad_norm": 0.0039555905386805534,
+      "learning_rate": 4.379227053140097e-05,
+      "loss": 0.0008,
+      "step": 4870
+    },
+    {
+      "epoch": 19.00191304347826,
+      "grad_norm": 0.050936609506607056,
+      "learning_rate": 4.3768115942028986e-05,
+      "loss": 0.0004,
+      "step": 4880
+    },
+    {
+      "epoch": 19.002347826086957,
+      "grad_norm": 0.0030154080595821142,
+      "learning_rate": 4.374396135265701e-05,
+      "loss": 0.003,
+      "step": 4890
+    },
+    {
+      "epoch": 19.002782608695654,
+      "grad_norm": 0.003285268321633339,
+      "learning_rate": 4.371980676328503e-05,
+      "loss": 0.0207,
+      "step": 4900
+    },
+    {
+      "epoch": 19.003217391304347,
+      "grad_norm": 0.0029449728317558765,
+      "learning_rate": 4.3695652173913046e-05,
+      "loss": 0.0097,
+      "step": 4910
+    },
+    {
+      "epoch": 19.003652173913043,
+      "grad_norm": 0.003369753947481513,
+      "learning_rate": 4.3671497584541064e-05,
+      "loss": 0.0176,
+      "step": 4920
+    },
+    {
+      "epoch": 19.00404347826087,
+      "eval_accuracy": 0.8934426229508197,
+      "eval_f1": 0.8942685404457407,
+      "eval_loss": 0.5749318599700928,
+      "eval_runtime": 93.0357,
+      "eval_samples_per_second": 1.311,
+      "eval_steps_per_second": 0.172,
+      "step": 4929
+    },
+    {
+      "epoch": 20.00004347826087,
+      "grad_norm": 0.03470902889966965,
+      "learning_rate": 4.364734299516908e-05,
+      "loss": 0.018,
+      "step": 4930
+    },
+    {
+      "epoch": 20.000478260869563,
+      "grad_norm": 0.003587034996598959,
+      "learning_rate": 4.362318840579711e-05,
+      "loss": 0.0015,
+      "step": 4940
+    },
+    {
+      "epoch": 20.00091304347826,
+      "grad_norm": 0.005226669833064079,
+      "learning_rate": 4.3599033816425124e-05,
+      "loss": 0.0002,
+      "step": 4950
+    },
+    {
+      "epoch": 20.001347826086956,
+      "grad_norm": 0.0028250955510884523,
+      "learning_rate": 4.357487922705314e-05,
+      "loss": 0.0108,
+      "step": 4960
+    },
+    {
+      "epoch": 20.001782608695652,
+      "grad_norm": 0.002752603031694889,
+      "learning_rate": 4.355072463768116e-05,
+      "loss": 0.0002,
+      "step": 4970
+    },
+    {
+      "epoch": 20.00221739130435,
+      "grad_norm": 0.003101842477917671,
+      "learning_rate": 4.352657004830918e-05,
+      "loss": 0.0122,
+      "step": 4980
+    },
+    {
+      "epoch": 20.002652173913045,
+      "grad_norm": 1.1627310514450073,
+      "learning_rate": 4.35024154589372e-05,
+      "loss": 0.0164,
+      "step": 4990
+    },
+    {
+      "epoch": 20.003086956521738,
+      "grad_norm": 0.0022625508718192577,
+      "learning_rate": 4.347826086956522e-05,
+      "loss": 0.0273,
+      "step": 5000
+    },
+    {
+      "epoch": 20.003521739130434,
+      "grad_norm": 0.0464153066277504,
+      "learning_rate": 4.345410628019324e-05,
+      "loss": 0.0002,
+      "step": 5010
+    },
+    {
+      "epoch": 20.00395652173913,
+      "grad_norm": 0.0038320214953273535,
+      "learning_rate": 4.3429951690821256e-05,
+      "loss": 0.0002,
+      "step": 5020
+    },
+    {
+      "epoch": 20.00404347826087,
+      "eval_accuracy": 0.9180327868852459,
+      "eval_f1": 0.9177113694429735,
+      "eval_loss": 0.4604657292366028,
+      "eval_runtime": 88.8434,
+      "eval_samples_per_second": 1.373,
+      "eval_steps_per_second": 0.18,
+      "step": 5022
+    },
+    {
+      "epoch": 21.000347826086955,
+      "grad_norm": 41.3768310546875,
+      "learning_rate": 4.3405797101449274e-05,
+      "loss": 0.1367,
+      "step": 5030
+    },
+    {
+      "epoch": 21.00078260869565,
+      "grad_norm": 0.003354653250426054,
+      "learning_rate": 4.33816425120773e-05,
+      "loss": 0.0002,
+      "step": 5040
+    },
+    {
+      "epoch": 21.001217391304348,
+      "grad_norm": 0.0035812542773783207,
+      "learning_rate": 4.3357487922705316e-05,
+      "loss": 0.0102,
+      "step": 5050
+    },
+    {
+      "epoch": 21.001652173913044,
+      "grad_norm": 0.0032342160120606422,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.0198,
+      "step": 5060
+    },
+    {
+      "epoch": 21.00208695652174,
+      "grad_norm": 0.0021403057035058737,
+      "learning_rate": 4.330917874396135e-05,
+      "loss": 0.0005,
+      "step": 5070
+    },
+    {
+      "epoch": 21.002521739130433,
+      "grad_norm": 1.876338005065918,
+      "learning_rate": 4.3285024154589376e-05,
+      "loss": 0.02,
+      "step": 5080
+    },
+    {
+      "epoch": 21.00295652173913,
+      "grad_norm": 0.0027030061464756727,
+      "learning_rate": 4.3260869565217394e-05,
+      "loss": 0.0136,
+      "step": 5090
+    },
+    {
+      "epoch": 21.003391304347826,
+      "grad_norm": 0.0033732217270880938,
+      "learning_rate": 4.323671497584541e-05,
+      "loss": 0.1237,
+      "step": 5100
+    },
+    {
+      "epoch": 21.003826086956522,
+      "grad_norm": 0.004835965111851692,
+      "learning_rate": 4.321256038647343e-05,
+      "loss": 0.0002,
+      "step": 5110
+    },
+    {
+      "epoch": 21.00404347826087,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.885472695453409,
+      "eval_loss": 0.6781102418899536,
+      "eval_runtime": 84.1219,
+      "eval_samples_per_second": 1.45,
+      "eval_steps_per_second": 0.19,
+      "step": 5115
+    },
+    {
+      "epoch": 0.0002173913043478261,
+      "grad_norm": 0.019052619114518166,
+      "learning_rate": 4.318840579710145e-05,
+      "loss": 0.0002,
+      "step": 5120
+    },
+    {
+      "epoch": 0.0006521739130434783,
+      "grad_norm": 0.008575594052672386,
+      "learning_rate": 4.316425120772947e-05,
+      "loss": 0.0003,
+      "step": 5130
+    },
+    {
+      "epoch": 0.0010869565217391304,
+      "grad_norm": 0.0024844331201165915,
+      "learning_rate": 4.314009661835749e-05,
+      "loss": 0.0002,
+      "step": 5140
+    },
+    {
+      "epoch": 0.0015217391304347826,
+      "grad_norm": 0.003250346751883626,
+      "learning_rate": 4.3115942028985515e-05,
+      "loss": 0.005,
+      "step": 5150
+    },
+    {
+      "epoch": 0.001956521739130435,
+      "grad_norm": 0.016990575939416885,
+      "learning_rate": 4.3091787439613526e-05,
+      "loss": 0.0092,
+      "step": 5160
+    },
+    {
+      "epoch": 0.002391304347826087,
+      "grad_norm": 0.009739736095070839,
+      "learning_rate": 4.3067632850241544e-05,
+      "loss": 0.0109,
+      "step": 5170
+    },
+    {
+      "epoch": 0.002826086956521739,
+      "grad_norm": 0.00987186562269926,
+      "learning_rate": 4.304347826086957e-05,
+      "loss": 0.0241,
+      "step": 5180
+    },
+    {
+      "epoch": 0.003260869565217391,
+      "grad_norm": 0.0024082553572952747,
+      "learning_rate": 4.3019323671497586e-05,
+      "loss": 0.0089,
+      "step": 5190
+    },
+    {
+      "epoch": 0.0036956521739130435,
+      "grad_norm": 0.018611734732985497,
+      "learning_rate": 4.299516908212561e-05,
+      "loss": 0.0447,
+      "step": 5200
+    },
+    {
+      "epoch": 0.004043478260869565,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.884764867622235,
+      "eval_loss": 0.8523582816123962,
+      "eval_runtime": 79.2349,
+      "eval_samples_per_second": 1.54,
+      "eval_steps_per_second": 0.202,
+      "step": 5208
+    },
+    {
+      "epoch": 1.0000869565217392,
+      "grad_norm": 0.0037880963645875454,
+      "learning_rate": 4.297101449275363e-05,
+      "loss": 0.0003,
+      "step": 5210
+    },
+    {
+      "epoch": 1.0005217391304349,
+      "grad_norm": 0.002207667101174593,
+      "learning_rate": 4.294685990338164e-05,
+      "loss": 0.0002,
+      "step": 5220
+    },
+    {
+      "epoch": 1.0009565217391305,
+      "grad_norm": 0.003156916704028845,
+      "learning_rate": 4.2922705314009664e-05,
+      "loss": 0.0118,
+      "step": 5230
+    },
+    {
+      "epoch": 1.0013913043478262,
+      "grad_norm": 0.0022045038640499115,
+      "learning_rate": 4.289855072463768e-05,
+      "loss": 0.0133,
+      "step": 5240
+    },
+    {
+      "epoch": 1.0018260869565216,
+      "grad_norm": 0.00293562444858253,
+      "learning_rate": 4.2874396135265707e-05,
+      "loss": 0.0092,
+      "step": 5250
+    },
+    {
+      "epoch": 1.0022608695652173,
+      "grad_norm": 0.0026238299906253815,
+      "learning_rate": 4.2850241545893724e-05,
+      "loss": 0.0002,
+      "step": 5260
+    },
+    {
+      "epoch": 1.002695652173913,
+      "grad_norm": 0.002277608262374997,
+      "learning_rate": 4.282608695652174e-05,
+      "loss": 0.0002,
+      "step": 5270
+    },
+    {
+      "epoch": 1.0031304347826087,
+      "grad_norm": 0.002632198855280876,
+      "learning_rate": 4.280193236714976e-05,
+      "loss": 0.0244,
+      "step": 5280
+    },
+    {
+      "epoch": 1.0035652173913043,
+      "grad_norm": 0.0037972743157297373,
+      "learning_rate": 4.277777777777778e-05,
+      "loss": 0.0152,
+      "step": 5290
+    },
+    {
+      "epoch": 1.004,
+      "grad_norm": 0.002417921321466565,
+      "learning_rate": 4.27536231884058e-05,
+      "loss": 0.0002,
+      "step": 5300
+    },
+    {
+      "epoch": 1.0040434782608696,
+      "eval_accuracy": 0.9180327868852459,
+      "eval_f1": 0.9179627294381393,
+      "eval_loss": 0.7144317030906677,
+      "eval_runtime": 62.0825,
+      "eval_samples_per_second": 1.965,
+      "eval_steps_per_second": 0.258,
+      "step": 5301
+    },
+    {
+      "epoch": 2.000391304347826,
+      "grad_norm": 0.0018766126595437527,
+      "learning_rate": 4.272946859903382e-05,
+      "loss": 0.0002,
+      "step": 5310
+    },
+    {
+      "epoch": 2.0008260869565215,
+      "grad_norm": 0.006529956124722958,
+      "learning_rate": 4.270531400966184e-05,
+      "loss": 0.0178,
+      "step": 5320
+    },
+    {
+      "epoch": 2.0012608695652174,
+      "grad_norm": 0.002090044552460313,
+      "learning_rate": 4.2681159420289856e-05,
+      "loss": 0.0005,
+      "step": 5330
+    },
+    {
+      "epoch": 2.001695652173913,
+      "grad_norm": 0.004191617481410503,
+      "learning_rate": 4.2657004830917874e-05,
+      "loss": 0.0002,
+      "step": 5340
+    },
+    {
+      "epoch": 2.0021304347826088,
+      "grad_norm": 0.0032148694153875113,
+      "learning_rate": 4.26328502415459e-05,
+      "loss": 0.2159,
+      "step": 5350
+    },
+    {
+      "epoch": 2.002565217391304,
+      "grad_norm": 0.0023662829771637917,
+      "learning_rate": 4.2608695652173916e-05,
+      "loss": 0.0307,
+      "step": 5360
+    },
+    {
+      "epoch": 2.003,
+      "grad_norm": 0.002081876154989004,
+      "learning_rate": 4.2584541062801934e-05,
+      "loss": 0.0354,
+      "step": 5370
+    },
+    {
+      "epoch": 2.0034347826086956,
+      "grad_norm": 0.002491084625944495,
+      "learning_rate": 4.256038647342995e-05,
+      "loss": 0.0328,
+      "step": 5380
+    },
+    {
+      "epoch": 2.0038695652173915,
+      "grad_norm": 0.0021801667753607035,
+      "learning_rate": 4.2536231884057976e-05,
+      "loss": 0.0047,
+      "step": 5390
+    },
+    {
+      "epoch": 2.0040434782608694,
+      "eval_accuracy": 0.8114754098360656,
+      "eval_f1": 0.8120128926178732,
+      "eval_loss": 1.1922979354858398,
+      "eval_runtime": 71.8577,
+      "eval_samples_per_second": 1.698,
+      "eval_steps_per_second": 0.223,
+      "step": 5394
+    },
+    {
+      "epoch": 3.0002608695652175,
+      "grad_norm": 0.004960095044225454,
+      "learning_rate": 4.2512077294685994e-05,
+      "loss": 0.2896,
+      "step": 5400
+    },
+    {
+      "epoch": 3.000695652173913,
+      "grad_norm": 0.2444988638162613,
+      "learning_rate": 4.248792270531401e-05,
+      "loss": 0.1418,
+      "step": 5410
+    },
+    {
+      "epoch": 3.001130434782609,
+      "grad_norm": 0.33257365226745605,
+      "learning_rate": 4.246376811594203e-05,
+      "loss": 0.0054,
+      "step": 5420
+    },
+    {
+      "epoch": 3.0015652173913043,
+      "grad_norm": 0.028887081891298294,
+      "learning_rate": 4.243961352657005e-05,
+      "loss": 0.0013,
+      "step": 5430
+    },
+    {
+      "epoch": 3.002,
+      "grad_norm": 0.00819989200681448,
+      "learning_rate": 4.241545893719807e-05,
+      "loss": 0.0003,
+      "step": 5440
+    },
+    {
+      "epoch": 3.0024347826086957,
+      "grad_norm": 0.005405559204518795,
+      "learning_rate": 4.239130434782609e-05,
+      "loss": 0.0362,
+      "step": 5450
+    },
+    {
+      "epoch": 3.002869565217391,
+      "grad_norm": 0.005664165131747723,
+      "learning_rate": 4.236714975845411e-05,
+      "loss": 0.1342,
+      "step": 5460
+    },
+    {
+      "epoch": 3.003304347826087,
+      "grad_norm": 0.012249620631337166,
+      "learning_rate": 4.2342995169082126e-05,
+      "loss": 0.0077,
+      "step": 5470
+    },
+    {
+      "epoch": 3.0037391304347825,
+      "grad_norm": 0.010283100418746471,
+      "learning_rate": 4.2318840579710143e-05,
+      "loss": 0.0298,
+      "step": 5480
+    },
+    {
+      "epoch": 3.0040434782608694,
+      "eval_accuracy": 0.860655737704918,
+      "eval_f1": 0.8604635470785563,
+      "eval_loss": 0.9435864090919495,
+      "eval_runtime": 62.0517,
+      "eval_samples_per_second": 1.966,
+      "eval_steps_per_second": 0.258,
+      "step": 5487
+    },
+    {
+      "epoch": 4.0001304347826085,
+      "grad_norm": 1.3064463138580322,
+      "learning_rate": 4.229468599033817e-05,
+      "loss": 0.0081,
+      "step": 5490
+    },
+    {
+      "epoch": 4.000565217391304,
+      "grad_norm": 0.008303687907755375,
+      "learning_rate": 4.2270531400966186e-05,
+      "loss": 0.0299,
+      "step": 5500
+    },
+    {
+      "epoch": 4.001,
+      "grad_norm": 0.008156073279678822,
+      "learning_rate": 4.224637681159421e-05,
+      "loss": 0.0087,
+      "step": 5510
+    },
+    {
+      "epoch": 4.001434782608696,
+      "grad_norm": 0.003047351958230138,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.0008,
+      "step": 5520
+    },
+    {
+      "epoch": 4.001869565217391,
+      "grad_norm": 0.0039140209555625916,
+      "learning_rate": 4.219806763285024e-05,
+      "loss": 0.0339,
+      "step": 5530
+    },
+    {
+      "epoch": 4.002304347826087,
+      "grad_norm": 0.015015006996691227,
+      "learning_rate": 4.2173913043478264e-05,
+      "loss": 0.0004,
+      "step": 5540
+    },
+    {
+      "epoch": 4.002739130434783,
+      "grad_norm": 0.00695813400670886,
+      "learning_rate": 4.214975845410628e-05,
+      "loss": 0.0869,
+      "step": 5550
+    },
+    {
+      "epoch": 4.0031739130434785,
+      "grad_norm": 180.4748077392578,
+      "learning_rate": 4.2125603864734306e-05,
+      "loss": 0.0392,
+      "step": 5560
+    },
+    {
+      "epoch": 4.003608695652174,
+      "grad_norm": 0.003250324632972479,
+      "learning_rate": 4.210144927536232e-05,
+      "loss": 0.0073,
+      "step": 5570
+    },
+    {
+      "epoch": 4.004043478260869,
+      "grad_norm": 0.002948058070614934,
+      "learning_rate": 4.207729468599034e-05,
+      "loss": 0.0244,
+      "step": 5580
+    },
+    {
+      "epoch": 4.004043478260869,
+      "eval_accuracy": 0.8770491803278688,
+      "eval_f1": 0.8768068873079699,
+      "eval_loss": 0.8491593599319458,
+      "eval_runtime": 63.8759,
+      "eval_samples_per_second": 1.91,
+      "eval_steps_per_second": 0.25,
+      "step": 5580
+    },
+    {
+      "epoch": 5.0004347826086954,
+      "grad_norm": 1.8423364162445068,
+      "learning_rate": 4.205314009661836e-05,
+      "loss": 0.055,
+      "step": 5590
+    },
+    {
+      "epoch": 5.000869565217391,
+      "grad_norm": 0.005326602142304182,
+      "learning_rate": 4.202898550724638e-05,
+      "loss": 0.0004,
+      "step": 5600
+    },
+    {
+      "epoch": 5.001304347826087,
+      "grad_norm": 0.0026577089447528124,
+      "learning_rate": 4.20048309178744e-05,
+      "loss": 0.0002,
+      "step": 5610
+    },
+    {
+      "epoch": 5.001739130434783,
+      "grad_norm": 0.002887171460315585,
+      "learning_rate": 4.198067632850241e-05,
+      "loss": 0.0037,
+      "step": 5620
+    },
+    {
+      "epoch": 5.002173913043478,
+      "grad_norm": 0.0028708172030746937,
+      "learning_rate": 4.195652173913044e-05,
+      "loss": 0.0003,
+      "step": 5630
+    },
+    {
+      "epoch": 5.002608695652174,
+      "grad_norm": 0.3837863504886627,
+      "learning_rate": 4.1932367149758456e-05,
+      "loss": 0.0188,
+      "step": 5640
+    },
+    {
+      "epoch": 5.00304347826087,
+      "grad_norm": 0.0025473798159509897,
+      "learning_rate": 4.1908212560386474e-05,
+      "loss": 0.0156,
+      "step": 5650
+    },
+    {
+      "epoch": 5.003478260869565,
+      "grad_norm": 0.001516752177849412,
+      "learning_rate": 4.18840579710145e-05,
+      "loss": 0.0002,
+      "step": 5660
+    },
+    {
+      "epoch": 5.003913043478261,
+      "grad_norm": 0.0037251082248985767,
+      "learning_rate": 4.185990338164251e-05,
+      "loss": 0.1255,
+      "step": 5670
+    },
+    {
+      "epoch": 5.004043478260869,
+      "eval_accuracy": 0.8934426229508197,
+      "eval_f1": 0.893593266973388,
+      "eval_loss": 0.8068659901618958,
+      "eval_runtime": 71.6747,
+      "eval_samples_per_second": 1.702,
+      "eval_steps_per_second": 0.223,
+      "step": 5673
+    },
+    {
+      "epoch": 6.000304347826087,
+      "grad_norm": 0.0021381359547376633,
+      "learning_rate": 4.1835748792270534e-05,
+      "loss": 0.0126,
+      "step": 5680
+    },
+    {
+      "epoch": 6.000739130434782,
+      "grad_norm": 0.0025412938557565212,
+      "learning_rate": 4.181159420289855e-05,
+      "loss": 0.0066,
+      "step": 5690
+    },
+    {
+      "epoch": 6.001173913043479,
+      "grad_norm": 1.9342097043991089,
+      "learning_rate": 4.1787439613526576e-05,
+      "loss": 0.0237,
+      "step": 5700
+    },
+    {
+      "epoch": 6.001608695652174,
+      "grad_norm": 0.003369554877281189,
+      "learning_rate": 4.1763285024154594e-05,
+      "loss": 0.0775,
+      "step": 5710
+    },
+    {
+      "epoch": 6.00204347826087,
+      "grad_norm": 0.0038899381179362535,
+      "learning_rate": 4.1739130434782605e-05,
+      "loss": 0.0123,
+      "step": 5720
+    },
+    {
+      "epoch": 6.002478260869565,
+      "grad_norm": 0.002373365219682455,
+      "learning_rate": 4.171497584541063e-05,
+      "loss": 0.0174,
+      "step": 5730
+    },
+    {
+      "epoch": 6.0029130434782605,
+      "grad_norm": 0.03260812163352966,
+      "learning_rate": 4.169082125603865e-05,
+      "loss": 0.0002,
+      "step": 5740
+    },
+    {
+      "epoch": 6.003347826086957,
+      "grad_norm": 39.89988708496094,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.021,
+      "step": 5750
+    },
+    {
+      "epoch": 6.003782608695652,
+      "grad_norm": 0.002141584875062108,
+      "learning_rate": 4.164251207729469e-05,
+      "loss": 0.0002,
+      "step": 5760
+    },
+    {
+      "epoch": 6.004043478260869,
+      "eval_accuracy": 0.8688524590163934,
+      "eval_f1": 0.8688620747444277,
+      "eval_loss": 0.9789605736732483,
+      "eval_runtime": 61.9663,
+      "eval_samples_per_second": 1.969,
+      "eval_steps_per_second": 0.258,
+      "step": 5766
+    },
+    {
+      "epoch": 7.000173913043478,
+      "grad_norm": 0.0036677520256489515,
+      "learning_rate": 4.161835748792271e-05,
+      "loss": 0.0002,
+      "step": 5770
+    },
+    {
+      "epoch": 7.000608695652174,
+      "grad_norm": 0.0025961585342884064,
+      "learning_rate": 4.1594202898550726e-05,
+      "loss": 0.0003,
+      "step": 5780
+    },
+    {
+      "epoch": 7.001043478260869,
+      "grad_norm": 0.003315119771286845,
+      "learning_rate": 4.157004830917874e-05,
+      "loss": 0.0002,
+      "step": 5790
+    },
+    {
+      "epoch": 7.001478260869566,
+      "grad_norm": 0.015529535710811615,
+      "learning_rate": 4.154589371980677e-05,
+      "loss": 0.0002,
+      "step": 5800
+    },
+    {
+      "epoch": 7.001913043478261,
+      "grad_norm": 0.0024377082008868456,
+      "learning_rate": 4.1521739130434786e-05,
+      "loss": 0.0154,
+      "step": 5810
+    },
+    {
+      "epoch": 7.0023478260869565,
+      "grad_norm": 0.002965396037325263,
+      "learning_rate": 4.1497584541062804e-05,
+      "loss": 0.0002,
+      "step": 5820
+    },
+    {
+      "epoch": 7.002782608695652,
+      "grad_norm": 0.0019290961790829897,
+      "learning_rate": 4.147342995169082e-05,
+      "loss": 0.0131,
+      "step": 5830
+    },
+    {
+      "epoch": 7.003217391304347,
+      "grad_norm": 0.0026897243224084377,
+      "learning_rate": 4.144927536231884e-05,
+      "loss": 0.0002,
+      "step": 5840
+    },
+    {
+      "epoch": 7.003652173913044,
+      "grad_norm": 0.0020241059828549623,
+      "learning_rate": 4.1425120772946864e-05,
+      "loss": 0.0225,
+      "step": 5850
+    },
+    {
+      "epoch": 7.004043478260869,
+      "eval_accuracy": 0.8934426229508197,
+      "eval_f1": 0.8935846750223277,
+      "eval_loss": 0.83548903465271,
+      "eval_runtime": 69.454,
+      "eval_samples_per_second": 1.757,
+      "eval_steps_per_second": 0.23,
+      "step": 5859
+    },
+    {
+      "epoch": 8.000043478260869,
+      "grad_norm": 0.001890950370579958,
+      "learning_rate": 4.140096618357488e-05,
+      "loss": 0.1447,
+      "step": 5860
+    },
+    {
+      "epoch": 8.000478260869565,
+      "grad_norm": 0.006349007599055767,
+      "learning_rate": 4.13768115942029e-05,
+      "loss": 0.1237,
+      "step": 5870
+    },
+    {
+      "epoch": 8.000913043478262,
+      "grad_norm": 1.3416682481765747,
+      "learning_rate": 4.135265700483092e-05,
+      "loss": 0.0205,
+      "step": 5880
+    },
+    {
+      "epoch": 8.001347826086956,
+      "grad_norm": 0.01573265716433525,
+      "learning_rate": 4.132850241545894e-05,
+      "loss": 0.0002,
+      "step": 5890
+    },
+    {
+      "epoch": 8.001782608695652,
+      "grad_norm": 0.9573756456375122,
+      "learning_rate": 4.130434782608696e-05,
+      "loss": 0.0054,
+      "step": 5900
+    },
+    {
+      "epoch": 8.002217391304347,
+      "grad_norm": 0.002642852021381259,
+      "learning_rate": 4.128019323671498e-05,
+      "loss": 0.0002,
+      "step": 5910
+    },
+    {
+      "epoch": 8.002652173913043,
+      "grad_norm": 0.03945804014801979,
+      "learning_rate": 4.1256038647342995e-05,
+      "loss": 0.0368,
+      "step": 5920
+    },
+    {
+      "epoch": 8.00308695652174,
+      "grad_norm": 0.0025761283468455076,
+      "learning_rate": 4.123188405797101e-05,
+      "loss": 0.0001,
+      "step": 5930
+    },
+    {
+      "epoch": 8.003521739130434,
+      "grad_norm": 0.004890764597803354,
+      "learning_rate": 4.120772946859904e-05,
+      "loss": 0.0002,
+      "step": 5940
+    },
+    {
+      "epoch": 8.00395652173913,
+      "grad_norm": 0.0019304666202515364,
+      "learning_rate": 4.1183574879227056e-05,
+      "loss": 0.0002,
+      "step": 5950
+    },
+    {
+      "epoch": 8.00404347826087,
+      "eval_accuracy": 0.8278688524590164,
+      "eval_f1": 0.827731349212412,
+      "eval_loss": 1.2655301094055176,
+      "eval_runtime": 71.2771,
+      "eval_samples_per_second": 1.712,
+      "eval_steps_per_second": 0.224,
+      "step": 5952
+    },
+    {
+      "epoch": 9.000347826086957,
+      "grad_norm": 0.0022518360055983067,
+      "learning_rate": 4.115942028985507e-05,
+      "loss": 0.0055,
+      "step": 5960
+    },
+    {
+      "epoch": 9.000782608695653,
+      "grad_norm": 0.002156842267140746,
+      "learning_rate": 4.113526570048309e-05,
+      "loss": 0.0135,
+      "step": 5970
+    },
+    {
+      "epoch": 9.001217391304348,
+      "grad_norm": 0.006422512233257294,
+      "learning_rate": 4.111111111111111e-05,
+      "loss": 0.0061,
+      "step": 5980
+    },
+    {
+      "epoch": 9.001652173913044,
+      "grad_norm": 0.0017242766916751862,
+      "learning_rate": 4.1086956521739134e-05,
+      "loss": 0.0002,
+      "step": 5990
+    },
+    {
+      "epoch": 9.002086956521739,
+      "grad_norm": 0.0017920706886798143,
+      "learning_rate": 4.106280193236715e-05,
+      "loss": 0.0001,
+      "step": 6000
+    },
+    {
+      "epoch": 9.002521739130435,
+      "grad_norm": 0.0021232902072370052,
+      "learning_rate": 4.1038647342995176e-05,
+      "loss": 0.0001,
+      "step": 6010
+    },
+    {
+      "epoch": 9.002956521739131,
+      "grad_norm": 0.001960981637239456,
+      "learning_rate": 4.101449275362319e-05,
+      "loss": 0.0002,
+      "step": 6020
+    },
+    {
+      "epoch": 9.003391304347826,
+      "grad_norm": 0.0018294297624379396,
+      "learning_rate": 4.0990338164251205e-05,
+      "loss": 0.0181,
+      "step": 6030
+    },
+    {
+      "epoch": 9.003826086956522,
+      "grad_norm": 0.002742977347224951,
+      "learning_rate": 4.096618357487923e-05,
+      "loss": 0.0153,
+      "step": 6040
+    },
+    {
+      "epoch": 9.00404347826087,
+      "eval_accuracy": 0.8278688524590164,
+      "eval_f1": 0.827731349212412,
+      "eval_loss": 1.2649469375610352,
+      "eval_runtime": 75.5533,
+      "eval_samples_per_second": 1.615,
+      "eval_steps_per_second": 0.212,
+      "step": 6045
+    },
+    {
+      "epoch": 10.000217391304348,
+      "grad_norm": 0.001949600875377655,
+      "learning_rate": 4.094202898550725e-05,
+      "loss": 0.0073,
+      "step": 6050
+    },
+    {
+      "epoch": 10.000652173913043,
+      "grad_norm": 0.0014879930531606078,
+      "learning_rate": 4.091787439613527e-05,
+      "loss": 0.0192,
+      "step": 6060
+    },
+    {
+      "epoch": 10.001086956521739,
+      "grad_norm": 0.0032951009925454855,
+      "learning_rate": 4.089371980676329e-05,
+      "loss": 0.0266,
+      "step": 6070
+    },
+    {
+      "epoch": 10.001521739130435,
+      "grad_norm": 0.0022343776654452085,
+      "learning_rate": 4.086956521739131e-05,
+      "loss": 0.0002,
+      "step": 6080
+    },
+    {
+      "epoch": 10.00195652173913,
+      "grad_norm": 0.0020602564327418804,
+      "learning_rate": 4.0845410628019325e-05,
+      "loss": 0.0175,
+      "step": 6090
+    },
+    {
+      "epoch": 10.002391304347826,
+      "grad_norm": 0.001147582079283893,
+      "learning_rate": 4.082125603864734e-05,
+      "loss": 0.0001,
+      "step": 6100
+    },
+    {
+      "epoch": 10.00282608695652,
+      "grad_norm": 0.00242516677826643,
+      "learning_rate": 4.079710144927537e-05,
+      "loss": 0.0002,
+      "step": 6110
+    },
+    {
+      "epoch": 10.003260869565217,
+      "grad_norm": 0.24815811216831207,
+      "learning_rate": 4.0772946859903386e-05,
+      "loss": 0.0488,
+      "step": 6120
+    },
+    {
+      "epoch": 10.003695652173914,
+      "grad_norm": 0.0025010844692587852,
+      "learning_rate": 4.0748792270531403e-05,
+      "loss": 0.0002,
+      "step": 6130
+    },
+    {
+      "epoch": 10.00404347826087,
+      "eval_accuracy": 0.8524590163934426,
+      "eval_f1": 0.8526902679666354,
+      "eval_loss": 1.144180417060852,
+      "eval_runtime": 72.1738,
+      "eval_samples_per_second": 1.69,
+      "eval_steps_per_second": 0.222,
+      "step": 6138
+    },
+    {
+      "epoch": 11.00008695652174,
+      "grad_norm": 1.7668468952178955,
+      "learning_rate": 4.072463768115942e-05,
+      "loss": 0.1222,
+      "step": 6140
+    },
+    {
+      "epoch": 11.000521739130434,
+      "grad_norm": 32.476314544677734,
+      "learning_rate": 4.070048309178744e-05,
+      "loss": 0.0861,
+      "step": 6150
+    },
+    {
+      "epoch": 11.00095652173913,
+      "grad_norm": 0.011918846517801285,
+      "learning_rate": 4.0676328502415464e-05,
+      "loss": 0.0074,
+      "step": 6160
+    },
+    {
+      "epoch": 11.001391304347827,
+      "grad_norm": 0.0021514969412237406,
+      "learning_rate": 4.065217391304348e-05,
+      "loss": 0.1079,
+      "step": 6170
+    },
+    {
+      "epoch": 11.001826086956521,
+      "grad_norm": 0.006829989142715931,
+      "learning_rate": 4.06280193236715e-05,
+      "loss": 0.0009,
+      "step": 6180
+    },
+    {
+      "epoch": 11.002260869565218,
+      "grad_norm": 0.00387940788641572,
+      "learning_rate": 4.060386473429952e-05,
+      "loss": 0.0529,
+      "step": 6190
+    },
+    {
+      "epoch": 11.002695652173912,
+      "grad_norm": 0.001859953161329031,
+      "learning_rate": 4.057971014492754e-05,
+      "loss": 0.1167,
+      "step": 6200
+    },
+    {
+      "epoch": 11.003130434782609,
+      "grad_norm": 5.892602443695068,
+      "learning_rate": 4.055555555555556e-05,
+      "loss": 0.0117,
+      "step": 6210
+    },
+    {
+      "epoch": 11.003565217391305,
+      "grad_norm": 0.0074269757606089115,
+      "learning_rate": 4.053140096618358e-05,
+      "loss": 0.0002,
+      "step": 6220
+    },
+    {
+      "epoch": 11.004,
+      "grad_norm": 33.7471809387207,
+      "learning_rate": 4.0507246376811595e-05,
+      "loss": 0.1902,
+      "step": 6230
+    },
+    {
+      "epoch": 11.00404347826087,
+      "eval_accuracy": 0.8524590163934426,
+      "eval_f1": 0.8515994026118366,
+      "eval_loss": 1.0411537885665894,
+      "eval_runtime": 66.8169,
+      "eval_samples_per_second": 1.826,
+      "eval_steps_per_second": 0.239,
+      "step": 6231
+    },
+    {
+      "epoch": 12.000391304347826,
+      "grad_norm": 0.05204038694500923,
+      "learning_rate": 4.048309178743961e-05,
+      "loss": 0.0344,
+      "step": 6240
+    },
+    {
+      "epoch": 12.000826086956522,
+      "grad_norm": 1.3173763751983643,
+      "learning_rate": 4.045893719806764e-05,
+      "loss": 0.1976,
+      "step": 6250
+    },
+    {
+      "epoch": 12.001260869565217,
+      "grad_norm": 0.0014710385585203767,
+      "learning_rate": 4.0434782608695655e-05,
+      "loss": 0.0094,
+      "step": 6260
+    },
+    {
+      "epoch": 12.001695652173913,
+      "grad_norm": 0.0017820164794102311,
+      "learning_rate": 4.041062801932367e-05,
+      "loss": 0.0094,
+      "step": 6270
+    },
+    {
+      "epoch": 12.00213043478261,
+      "grad_norm": 16.647350311279297,
+      "learning_rate": 4.038647342995169e-05,
+      "loss": 0.1017,
+      "step": 6280
+    },
+    {
+      "epoch": 12.002565217391304,
+      "grad_norm": 9.301661491394043,
+      "learning_rate": 4.036231884057971e-05,
+      "loss": 0.1693,
+      "step": 6290
+    },
+    {
+      "epoch": 12.003,
+      "grad_norm": 16.365354537963867,
+      "learning_rate": 4.0338164251207733e-05,
+      "loss": 0.1188,
+      "step": 6300
+    },
+    {
+      "epoch": 12.003434782608696,
+      "grad_norm": 8.132902145385742,
+      "learning_rate": 4.031400966183575e-05,
+      "loss": 0.0691,
+      "step": 6310
+    },
+    {
+      "epoch": 12.003869565217391,
+      "grad_norm": 0.017481571063399315,
+      "learning_rate": 4.028985507246377e-05,
+      "loss": 0.0622,
+      "step": 6320
+    },
+    {
+      "epoch": 12.00404347826087,
+      "eval_accuracy": 0.8114754098360656,
+      "eval_f1": 0.8078045336889167,
+      "eval_loss": 1.100170612335205,
+      "eval_runtime": 80.9477,
+      "eval_samples_per_second": 1.507,
+      "eval_steps_per_second": 0.198,
+      "step": 6324
+    },
+    {
+      "epoch": 13.000260869565217,
+      "grad_norm": 62.808135986328125,
+      "learning_rate": 4.026570048309179e-05,
+      "loss": 0.0311,
+      "step": 6330
+    },
+    {
+      "epoch": 13.000695652173913,
+      "grad_norm": 0.0037931206170469522,
+      "learning_rate": 4.0241545893719805e-05,
+      "loss": 0.0109,
+      "step": 6340
+    },
+    {
+      "epoch": 13.001130434782608,
+      "grad_norm": 27.289682388305664,
+      "learning_rate": 4.021739130434783e-05,
+      "loss": 0.0662,
+      "step": 6350
+    },
+    {
+      "epoch": 13.001565217391304,
+      "grad_norm": 0.018042033538222313,
+      "learning_rate": 4.019323671497585e-05,
+      "loss": 0.0022,
+      "step": 6360
+    },
+    {
+      "epoch": 13.002,
+      "grad_norm": 0.010825947858393192,
+      "learning_rate": 4.016908212560387e-05,
+      "loss": 0.0002,
+      "step": 6370
+    },
+    {
+      "epoch": 13.002434782608695,
+      "grad_norm": 61.149078369140625,
+      "learning_rate": 4.014492753623188e-05,
+      "loss": 0.2017,
+      "step": 6380
+    },
+    {
+      "epoch": 13.002869565217392,
+      "grad_norm": 0.00819044467061758,
+      "learning_rate": 4.012077294685991e-05,
+      "loss": 0.0005,
+      "step": 6390
+    },
+    {
+      "epoch": 13.003304347826086,
+      "grad_norm": 7.115654468536377,
+      "learning_rate": 4.0096618357487925e-05,
+      "loss": 0.0981,
+      "step": 6400
+    },
+    {
+      "epoch": 13.003739130434782,
+      "grad_norm": 0.06668082624673843,
+      "learning_rate": 4.007246376811594e-05,
+      "loss": 0.1367,
+      "step": 6410
+    },
+    {
+      "epoch": 13.00404347826087,
+      "eval_accuracy": 0.8442622950819673,
+      "eval_f1": 0.8402451615360002,
+      "eval_loss": 1.031615138053894,
+      "eval_runtime": 71.7599,
+      "eval_samples_per_second": 1.7,
+      "eval_steps_per_second": 0.223,
+      "step": 6417
+    },
+    {
+      "epoch": 14.000130434782609,
+      "grad_norm": 1.688230037689209,
+      "learning_rate": 4.004830917874397e-05,
+      "loss": 0.1085,
+      "step": 6420
+    },
+    {
+      "epoch": 14.000565217391305,
+      "grad_norm": 0.002879357198253274,
+      "learning_rate": 4.002415458937198e-05,
+      "loss": 0.0148,
+      "step": 6430
+    },
+    {
+      "epoch": 14.001,
+      "grad_norm": 0.005582920741289854,
+      "learning_rate": 4e-05,
+      "loss": 0.0532,
+      "step": 6440
+    },
+    {
+      "epoch": 14.001434782608696,
+      "grad_norm": 0.01366361603140831,
+      "learning_rate": 3.997584541062802e-05,
+      "loss": 0.0036,
+      "step": 6450
+    },
+    {
+      "epoch": 14.001869565217392,
+      "grad_norm": 0.009635832160711288,
+      "learning_rate": 3.995169082125604e-05,
+      "loss": 0.0027,
+      "step": 6460
+    },
+    {
+      "epoch": 14.002304347826087,
+      "grad_norm": 6.955427646636963,
+      "learning_rate": 3.9927536231884064e-05,
+      "loss": 0.3633,
+      "step": 6470
+    },
+    {
+      "epoch": 14.002739130434783,
+      "grad_norm": 25.112009048461914,
+      "learning_rate": 3.9903381642512075e-05,
+      "loss": 0.161,
+      "step": 6480
+    },
+    {
+      "epoch": 14.003173913043478,
+      "grad_norm": 0.009927201084792614,
+      "learning_rate": 3.98792270531401e-05,
+      "loss": 0.0014,
+      "step": 6490
+    },
+    {
+      "epoch": 14.003608695652174,
+      "grad_norm": 0.014570885337889194,
+      "learning_rate": 3.985507246376812e-05,
+      "loss": 0.02,
+      "step": 6500
+    },
+    {
+      "epoch": 14.00404347826087,
+      "grad_norm": 1.4303860664367676,
+      "learning_rate": 3.983091787439614e-05,
+      "loss": 0.0165,
+      "step": 6510
+    },
+    {
+      "epoch": 14.00404347826087,
+      "eval_accuracy": 0.819672131147541,
+      "eval_f1": 0.82006315755217,
+      "eval_loss": 1.0104899406433105,
+      "eval_runtime": 79.2274,
+      "eval_samples_per_second": 1.54,
+      "eval_steps_per_second": 0.202,
+      "step": 6510
+    },
+    {
+      "epoch": 15.000434782608696,
+      "grad_norm": 0.014703233726322651,
+      "learning_rate": 3.980676328502416e-05,
+      "loss": 0.1467,
+      "step": 6520
+    },
+    {
+      "epoch": 15.000869565217391,
+      "grad_norm": 0.021894462406635284,
+      "learning_rate": 3.978260869565217e-05,
+      "loss": 0.1916,
+      "step": 6530
+    },
+    {
+      "epoch": 15.001304347826087,
+      "grad_norm": 48.06509780883789,
+      "learning_rate": 3.9758454106280195e-05,
+      "loss": 0.1599,
+      "step": 6540
+    },
+    {
+      "epoch": 15.001739130434782,
+      "grad_norm": 26.76535987854004,
+      "learning_rate": 3.973429951690821e-05,
+      "loss": 0.5953,
+      "step": 6550
+    },
+    {
+      "epoch": 15.002173913043478,
+      "grad_norm": 9.520294189453125,
+      "learning_rate": 3.971014492753624e-05,
+      "loss": 0.1125,
+      "step": 6560
+    },
+    {
+      "epoch": 15.002608695652174,
+      "grad_norm": 1.8650518655776978,
+      "learning_rate": 3.9685990338164255e-05,
+      "loss": 0.0568,
+      "step": 6570
+    },
+    {
+      "epoch": 15.003043478260869,
+      "grad_norm": 0.007623857352882624,
+      "learning_rate": 3.966183574879227e-05,
+      "loss": 0.0069,
+      "step": 6580
+    },
+    {
+      "epoch": 15.003478260869565,
+      "grad_norm": 0.0030073132365942,
+      "learning_rate": 3.963768115942029e-05,
+      "loss": 0.0094,
+      "step": 6590
+    },
+    {
+      "epoch": 15.003913043478262,
+      "grad_norm": 32.13124084472656,
+      "learning_rate": 3.961352657004831e-05,
+      "loss": 0.166,
+      "step": 6600
+    },
+    {
+      "epoch": 15.00404347826087,
+      "eval_accuracy": 0.8688524590163934,
+      "eval_f1": 0.8678181563427466,
+      "eval_loss": 0.8523256778717041,
+      "eval_runtime": 75.8405,
+      "eval_samples_per_second": 1.609,
+      "eval_steps_per_second": 0.211,
+      "step": 6603
+    },
+    {
+      "epoch": 16.000304347826088,
+      "grad_norm": 0.024344058707356453,
+      "learning_rate": 3.958937198067633e-05,
+      "loss": 0.0005,
+      "step": 6610
+    },
+    {
+      "epoch": 16.000739130434784,
+      "grad_norm": 0.008748427033424377,
+      "learning_rate": 3.956521739130435e-05,
+      "loss": 0.2414,
+      "step": 6620
+    },
+    {
+      "epoch": 16.001173913043477,
+      "grad_norm": 26.980012893676758,
+      "learning_rate": 3.954106280193237e-05,
+      "loss": 0.1468,
+      "step": 6630
+    },
+    {
+      "epoch": 16.001608695652173,
+      "grad_norm": 0.027768012136220932,
+      "learning_rate": 3.951690821256039e-05,
+      "loss": 0.001,
+      "step": 6640
+    },
+    {
+      "epoch": 16.00204347826087,
+      "grad_norm": 0.05727599933743477,
+      "learning_rate": 3.9492753623188405e-05,
+      "loss": 0.0021,
+      "step": 6650
+    },
+    {
+      "epoch": 16.002478260869566,
+      "grad_norm": 0.025578580796718597,
+      "learning_rate": 3.946859903381643e-05,
+      "loss": 0.0005,
+      "step": 6660
+    },
+    {
+      "epoch": 16.002913043478262,
+      "grad_norm": 34.755123138427734,
+      "learning_rate": 3.944444444444445e-05,
+      "loss": 0.0842,
+      "step": 6670
+    },
+    {
+      "epoch": 16.003347826086955,
+      "grad_norm": 0.7231902480125427,
+      "learning_rate": 3.9420289855072465e-05,
+      "loss": 0.0845,
+      "step": 6680
+    },
+    {
+      "epoch": 16.00378260869565,
+      "grad_norm": 0.006564811337739229,
+      "learning_rate": 3.939613526570048e-05,
+      "loss": 0.0898,
+      "step": 6690
+    },
+    {
+      "epoch": 16.00404347826087,
+      "eval_accuracy": 0.9016393442622951,
+      "eval_f1": 0.9015288061359007,
+      "eval_loss": 0.6139894127845764,
+      "eval_runtime": 69.4275,
+      "eval_samples_per_second": 1.757,
+      "eval_steps_per_second": 0.23,
+      "step": 6696
+    },
+    {
+      "epoch": 17.00017391304348,
+      "grad_norm": 0.0037501673214137554,
+      "learning_rate": 3.937198067632851e-05,
+      "loss": 0.0117,
+      "step": 6700
+    },
+    {
+      "epoch": 17.000608695652176,
+      "grad_norm": 0.00506971450522542,
+      "learning_rate": 3.9347826086956525e-05,
+      "loss": 0.0186,
+      "step": 6710
+    },
+    {
+      "epoch": 17.00104347826087,
+      "grad_norm": 0.003616392845287919,
+      "learning_rate": 3.932367149758454e-05,
+      "loss": 0.0002,
+      "step": 6720
+    },
+    {
+      "epoch": 17.001478260869565,
+      "grad_norm": 0.002944325562566519,
+      "learning_rate": 3.929951690821256e-05,
+      "loss": 0.0869,
+      "step": 6730
+    },
+    {
+      "epoch": 17.00191304347826,
+      "grad_norm": 0.018326986581087112,
+      "learning_rate": 3.927536231884058e-05,
+      "loss": 0.0846,
+      "step": 6740
+    },
+    {
+      "epoch": 17.002347826086957,
+      "grad_norm": 28.93506622314453,
+      "learning_rate": 3.92512077294686e-05,
+      "loss": 0.0637,
+      "step": 6750
+    },
+    {
+      "epoch": 17.002782608695654,
+      "grad_norm": 0.0321519635617733,
+      "learning_rate": 3.922705314009662e-05,
+      "loss": 0.103,
+      "step": 6760
+    },
+    {
+      "epoch": 17.003217391304347,
+      "grad_norm": 0.00872319471091032,
+      "learning_rate": 3.920289855072464e-05,
+      "loss": 0.1741,
+      "step": 6770
+    },
+    {
+      "epoch": 17.003652173913043,
+      "grad_norm": 0.005549349822103977,
+      "learning_rate": 3.9178743961352657e-05,
+      "loss": 0.0384,
+      "step": 6780
+    },
+    {
+      "epoch": 17.00404347826087,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8832644966100266,
+      "eval_loss": 0.6065811514854431,
+      "eval_runtime": 73.4098,
+      "eval_samples_per_second": 1.662,
+      "eval_steps_per_second": 0.218,
+      "step": 6789
+    },
+    {
+      "epoch": 18.00004347826087,
+      "grad_norm": 0.0036791262682527304,
+      "learning_rate": 3.9154589371980674e-05,
+      "loss": 0.0881,
+      "step": 6790
+    },
+    {
+      "epoch": 18.000478260869563,
+      "grad_norm": 0.02601425163447857,
+      "learning_rate": 3.91304347826087e-05,
+      "loss": 0.0205,
+      "step": 6800
+    },
+    {
+      "epoch": 18.00091304347826,
+      "grad_norm": 1.111623764038086,
+      "learning_rate": 3.910628019323672e-05,
+      "loss": 0.0746,
+      "step": 6810
+    },
+    {
+      "epoch": 18.001347826086956,
+      "grad_norm": 0.08039382100105286,
+      "learning_rate": 3.908212560386474e-05,
+      "loss": 0.0017,
+      "step": 6820
+    },
+    {
+      "epoch": 18.001782608695652,
+      "grad_norm": 0.0041095721535384655,
+      "learning_rate": 3.905797101449275e-05,
+      "loss": 0.0002,
+      "step": 6830
+    },
+    {
+      "epoch": 18.00221739130435,
+      "grad_norm": 0.02693447656929493,
+      "learning_rate": 3.903381642512077e-05,
+      "loss": 0.0433,
+      "step": 6840
+    },
+    {
+      "epoch": 18.002652173913045,
+      "grad_norm": 14.348990440368652,
+      "learning_rate": 3.9009661835748795e-05,
+      "loss": 0.0977,
+      "step": 6850
+    },
+    {
+      "epoch": 18.003086956521738,
+      "grad_norm": 0.0028455532155930996,
+      "learning_rate": 3.898550724637681e-05,
+      "loss": 0.0241,
+      "step": 6860
+    },
+    {
+      "epoch": 18.003521739130434,
+      "grad_norm": 0.03243542090058327,
+      "learning_rate": 3.896135265700484e-05,
+      "loss": 0.0003,
+      "step": 6870
+    },
+    {
+      "epoch": 18.00395652173913,
+      "grad_norm": 0.002338533289730549,
+      "learning_rate": 3.8937198067632855e-05,
+      "loss": 0.0206,
+      "step": 6880
+    },
+    {
+      "epoch": 18.00404347826087,
+      "eval_accuracy": 0.9098360655737705,
+      "eval_f1": 0.9110839096946348,
+      "eval_loss": 0.5883862972259521,
+      "eval_runtime": 78.6246,
+      "eval_samples_per_second": 1.552,
+      "eval_steps_per_second": 0.203,
+      "step": 6882
+    },
+    {
+      "epoch": 0.00034782608695652176,
+      "grad_norm": 0.007827730849385262,
+      "learning_rate": 3.8913043478260866e-05,
+      "loss": 0.0003,
+      "step": 6890
+    },
+    {
+      "epoch": 0.000782608695652174,
+      "grad_norm": 0.004137350246310234,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.0008,
+      "step": 6900
+    },
+    {
+      "epoch": 0.001217391304347826,
+      "grad_norm": 0.0026433924213051796,
+      "learning_rate": 3.886473429951691e-05,
+      "loss": 0.0002,
+      "step": 6910
+    },
+    {
+      "epoch": 0.0016521739130434783,
+      "grad_norm": 0.001748176058754325,
+      "learning_rate": 3.884057971014493e-05,
+      "loss": 0.0018,
+      "step": 6920
+    },
+    {
+      "epoch": 0.0020869565217391303,
+      "grad_norm": 0.009172141551971436,
+      "learning_rate": 3.881642512077295e-05,
+      "loss": 0.0194,
+      "step": 6930
+    },
+    {
+      "epoch": 0.0025217391304347826,
+      "grad_norm": 0.002410092856734991,
+      "learning_rate": 3.879227053140097e-05,
+      "loss": 0.0002,
+      "step": 6940
+    },
+    {
+      "epoch": 0.002956521739130435,
+      "grad_norm": 0.003946972079575062,
+      "learning_rate": 3.876811594202899e-05,
+      "loss": 0.017,
+      "step": 6950
+    },
+    {
+      "epoch": 0.003391304347826087,
+      "grad_norm": 14.169075965881348,
+      "learning_rate": 3.8743961352657004e-05,
+      "loss": 0.0713,
+      "step": 6960
+    },
+    {
+      "epoch": 0.0038260869565217392,
+      "grad_norm": 0.0025988216511905193,
+      "learning_rate": 3.871980676328503e-05,
+      "loss": 0.0009,
+      "step": 6970
+    },
+    {
+      "epoch": 0.004043478260869565,
+      "eval_accuracy": 0.9016393442622951,
+      "eval_f1": 0.9011749047870193,
+      "eval_loss": 0.6782234907150269,
+      "eval_runtime": 88.3548,
+      "eval_samples_per_second": 1.381,
+      "eval_steps_per_second": 0.181,
+      "step": 6975
+    },
+    {
+      "epoch": 1.0002173913043477,
+      "grad_norm": 0.5530427694320679,
+      "learning_rate": 3.869565217391305e-05,
+      "loss": 0.0023,
+      "step": 6980
+    },
+    {
+      "epoch": 1.0006521739130434,
+      "grad_norm": 0.007848013192415237,
+      "learning_rate": 3.8671497584541065e-05,
+      "loss": 0.0118,
+      "step": 6990
+    },
+    {
+      "epoch": 1.001086956521739,
+      "grad_norm": 0.001887777354568243,
+      "learning_rate": 3.864734299516908e-05,
+      "loss": 0.0071,
+      "step": 7000
+    },
+    {
+      "epoch": 1.0015217391304347,
+      "grad_norm": 0.003241229336708784,
+      "learning_rate": 3.862318840579711e-05,
+      "loss": 0.1008,
+      "step": 7010
+    },
+    {
+      "epoch": 1.0019565217391304,
+      "grad_norm": 0.0033748249989002943,
+      "learning_rate": 3.8599033816425125e-05,
+      "loss": 0.0148,
+      "step": 7020
+    },
+    {
+      "epoch": 1.002391304347826,
+      "grad_norm": 0.0038462060038000345,
+      "learning_rate": 3.857487922705314e-05,
+      "loss": 0.0002,
+      "step": 7030
+    },
+    {
+      "epoch": 1.0028260869565218,
+      "grad_norm": 0.0021478806156665087,
+      "learning_rate": 3.855072463768116e-05,
+      "loss": 0.0003,
+      "step": 7040
+    },
+    {
+      "epoch": 1.0032608695652174,
+      "grad_norm": 0.006713178474456072,
+      "learning_rate": 3.852657004830918e-05,
+      "loss": 0.0368,
+      "step": 7050
+    },
+    {
+      "epoch": 1.003695652173913,
+      "grad_norm": 13.128242492675781,
+      "learning_rate": 3.85024154589372e-05,
+      "loss": 0.088,
+      "step": 7060
+    },
+    {
+      "epoch": 1.0040434782608696,
+      "eval_accuracy": 0.8360655737704918,
+      "eval_f1": 0.8393668317566647,
+      "eval_loss": 0.8946360945701599,
+      "eval_runtime": 84.9121,
+      "eval_samples_per_second": 1.437,
+      "eval_steps_per_second": 0.188,
+      "step": 7068
+    },
+    {
+      "epoch": 2.000086956521739,
+      "grad_norm": 0.02620559372007847,
+      "learning_rate": 3.847826086956522e-05,
+      "loss": 0.0704,
+      "step": 7070
+    },
+    {
+      "epoch": 2.0005217391304346,
+      "grad_norm": 0.0025947631802409887,
+      "learning_rate": 3.845410628019324e-05,
+      "loss": 0.0662,
+      "step": 7080
+    },
+    {
+      "epoch": 2.0009565217391305,
+      "grad_norm": 0.007165535818785429,
+      "learning_rate": 3.8429951690821256e-05,
+      "loss": 0.0202,
+      "step": 7090
+    },
+    {
+      "epoch": 2.001391304347826,
+      "grad_norm": 0.0037356368266046047,
+      "learning_rate": 3.8405797101449274e-05,
+      "loss": 0.2776,
+      "step": 7100
+    },
+    {
+      "epoch": 2.001826086956522,
+      "grad_norm": 0.0042053996585309505,
+      "learning_rate": 3.83816425120773e-05,
+      "loss": 0.0702,
+      "step": 7110
+    },
+    {
+      "epoch": 2.0022608695652173,
+      "grad_norm": 0.0028894192073494196,
+      "learning_rate": 3.835748792270532e-05,
+      "loss": 0.0299,
+      "step": 7120
+    },
+    {
+      "epoch": 2.002695652173913,
+      "grad_norm": 0.002774425782263279,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.0154,
+      "step": 7130
+    },
+    {
+      "epoch": 2.0031304347826087,
+      "grad_norm": 0.0035831418354064226,
+      "learning_rate": 3.830917874396135e-05,
+      "loss": 0.0061,
+      "step": 7140
+    },
+    {
+      "epoch": 2.0035652173913046,
+      "grad_norm": 0.0023740469478070736,
+      "learning_rate": 3.828502415458937e-05,
+      "loss": 0.0294,
+      "step": 7150
+    },
+    {
+      "epoch": 2.004,
+      "grad_norm": 0.007240784354507923,
+      "learning_rate": 3.8260869565217395e-05,
+      "loss": 0.0002,
+      "step": 7160
+    },
+    {
+      "epoch": 2.0040434782608694,
+      "eval_accuracy": 0.9016393442622951,
+      "eval_f1": 0.9007072239914696,
+      "eval_loss": 0.6232790946960449,
+      "eval_runtime": 74.029,
+      "eval_samples_per_second": 1.648,
+      "eval_steps_per_second": 0.216,
+      "step": 7161
+    },
+    {
+      "epoch": 3.000391304347826,
+      "grad_norm": 0.002939334139227867,
+      "learning_rate": 3.823671497584541e-05,
+      "loss": 0.0094,
+      "step": 7170
+    },
+    {
+      "epoch": 3.0008260869565215,
+      "grad_norm": 0.003376334672793746,
+      "learning_rate": 3.821256038647344e-05,
+      "loss": 0.0002,
+      "step": 7180
+    },
+    {
+      "epoch": 3.0012608695652174,
+      "grad_norm": 0.0019673022907227278,
+      "learning_rate": 3.818840579710145e-05,
+      "loss": 0.0002,
+      "step": 7190
+    },
+    {
+      "epoch": 3.001695652173913,
+      "grad_norm": 0.002067855792120099,
+      "learning_rate": 3.8164251207729466e-05,
+      "loss": 0.0003,
+      "step": 7200
+    },
+    {
+      "epoch": 3.0021304347826088,
+      "grad_norm": 0.0016294581582769752,
+      "learning_rate": 3.814009661835749e-05,
+      "loss": 0.0114,
+      "step": 7210
+    },
+    {
+      "epoch": 3.002565217391304,
+      "grad_norm": 0.0030627830419689417,
+      "learning_rate": 3.811594202898551e-05,
+      "loss": 0.0105,
+      "step": 7220
+    },
+    {
+      "epoch": 3.003,
+      "grad_norm": 0.0019431186374276876,
+      "learning_rate": 3.809178743961353e-05,
+      "loss": 0.0248,
+      "step": 7230
+    },
+    {
+      "epoch": 3.0034347826086956,
+      "grad_norm": 0.005873087793588638,
+      "learning_rate": 3.8067632850241544e-05,
+      "loss": 0.0144,
+      "step": 7240
+    },
+    {
+      "epoch": 3.0038695652173915,
+      "grad_norm": 0.005152291618287563,
+      "learning_rate": 3.804347826086957e-05,
+      "loss": 0.0118,
+      "step": 7250
+    },
+    {
+      "epoch": 3.0040434782608694,
+      "eval_accuracy": 0.8524590163934426,
+      "eval_f1": 0.8544169462202248,
+      "eval_loss": 0.8239415884017944,
+      "eval_runtime": 84.6909,
+      "eval_samples_per_second": 1.441,
+      "eval_steps_per_second": 0.189,
+      "step": 7254
+    },
+    {
+      "epoch": 4.000260869565217,
+      "grad_norm": 0.002831781283020973,
+      "learning_rate": 3.8019323671497586e-05,
+      "loss": 0.0087,
+      "step": 7260
+    },
+    {
+      "epoch": 4.000695652173913,
+      "grad_norm": 0.0033079604618251324,
+      "learning_rate": 3.7995169082125604e-05,
+      "loss": 0.0111,
+      "step": 7270
+    },
+    {
+      "epoch": 4.001130434782609,
+      "grad_norm": 0.0020894971676170826,
+      "learning_rate": 3.797101449275363e-05,
+      "loss": 0.0001,
+      "step": 7280
+    },
+    {
+      "epoch": 4.001565217391304,
+      "grad_norm": 0.00628332793712616,
+      "learning_rate": 3.794685990338164e-05,
+      "loss": 0.0008,
+      "step": 7290
+    },
+    {
+      "epoch": 4.002,
+      "grad_norm": 0.0022865217179059982,
+      "learning_rate": 3.7922705314009665e-05,
+      "loss": 0.0003,
+      "step": 7300
+    },
+    {
+      "epoch": 4.002434782608695,
+      "grad_norm": 0.003050363389775157,
+      "learning_rate": 3.789855072463768e-05,
+      "loss": 0.0001,
+      "step": 7310
+    },
+    {
+      "epoch": 4.002869565217392,
+      "grad_norm": 0.0020351973362267017,
+      "learning_rate": 3.787439613526571e-05,
+      "loss": 0.0001,
+      "step": 7320
+    },
+    {
+      "epoch": 4.003304347826087,
+      "grad_norm": 0.002275005681440234,
+      "learning_rate": 3.7850241545893725e-05,
+      "loss": 0.0282,
+      "step": 7330
+    },
+    {
+      "epoch": 4.0037391304347825,
+      "grad_norm": 0.056021079421043396,
+      "learning_rate": 3.7826086956521736e-05,
+      "loss": 0.01,
+      "step": 7340
+    },
+    {
+      "epoch": 4.004043478260869,
+      "eval_accuracy": 0.8770491803278688,
+      "eval_f1": 0.875608828010926,
+      "eval_loss": 0.8939600586891174,
+      "eval_runtime": 73.7752,
+      "eval_samples_per_second": 1.654,
+      "eval_steps_per_second": 0.217,
+      "step": 7347
+    },
+    {
+      "epoch": 5.0001304347826085,
+      "grad_norm": 0.0015969000523909926,
+      "learning_rate": 3.780193236714976e-05,
+      "loss": 0.0001,
+      "step": 7350
+    },
+    {
+      "epoch": 5.000565217391304,
+      "grad_norm": 0.002829205244779587,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.0204,
+      "step": 7360
+    },
+    {
+      "epoch": 5.001,
+      "grad_norm": 0.0035543395206332207,
+      "learning_rate": 3.77536231884058e-05,
+      "loss": 0.0002,
+      "step": 7370
+    },
+    {
+      "epoch": 5.001434782608696,
+      "grad_norm": 0.0014324752846732736,
+      "learning_rate": 3.772946859903382e-05,
+      "loss": 0.0001,
+      "step": 7380
+    },
+    {
+      "epoch": 5.001869565217391,
+      "grad_norm": 0.009829861111938953,
+      "learning_rate": 3.770531400966184e-05,
+      "loss": 0.0003,
+      "step": 7390
+    },
+    {
+      "epoch": 5.002304347826087,
+      "grad_norm": 0.0021648143883794546,
+      "learning_rate": 3.7681159420289856e-05,
+      "loss": 0.0214,
+      "step": 7400
+    },
+    {
+      "epoch": 5.002739130434783,
+      "grad_norm": 0.002256903564557433,
+      "learning_rate": 3.7657004830917874e-05,
+      "loss": 0.0179,
+      "step": 7410
+    },
+    {
+      "epoch": 5.0031739130434785,
+      "grad_norm": 0.008477986790239811,
+      "learning_rate": 3.76328502415459e-05,
+      "loss": 0.0156,
+      "step": 7420
+    },
+    {
+      "epoch": 5.003608695652174,
+      "grad_norm": 1.539802074432373,
+      "learning_rate": 3.7608695652173917e-05,
+      "loss": 0.0102,
+      "step": 7430
+    },
+    {
+      "epoch": 5.004043478260869,
+      "grad_norm": 0.001704772817902267,
+      "learning_rate": 3.7584541062801934e-05,
+      "loss": 0.0922,
+      "step": 7440
+    },
+    {
+      "epoch": 5.004043478260869,
+      "eval_accuracy": 0.8934426229508197,
+      "eval_f1": 0.890414307004471,
+      "eval_loss": 0.7101210951805115,
+      "eval_runtime": 74.1203,
+      "eval_samples_per_second": 1.646,
+      "eval_steps_per_second": 0.216,
+      "step": 7440
+    },
+    {
+      "epoch": 6.0004347826086954,
+      "grad_norm": 0.0019552623853087425,
+      "learning_rate": 3.756038647342995e-05,
+      "loss": 0.0084,
+      "step": 7450
+    },
+    {
+      "epoch": 6.000869565217391,
+      "grad_norm": 0.014820483513176441,
+      "learning_rate": 3.753623188405797e-05,
+      "loss": 0.0081,
+      "step": 7460
+    },
+    {
+      "epoch": 6.001304347826087,
+      "grad_norm": 0.0021731434389948845,
+      "learning_rate": 3.7512077294685995e-05,
+      "loss": 0.0126,
+      "step": 7470
+    },
+    {
+      "epoch": 6.001739130434783,
+      "grad_norm": 0.0026017893105745316,
+      "learning_rate": 3.748792270531401e-05,
+      "loss": 0.0001,
+      "step": 7480
+    },
+    {
+      "epoch": 6.002173913043478,
+      "grad_norm": 16.81439781188965,
+      "learning_rate": 3.746376811594203e-05,
+      "loss": 0.0209,
+      "step": 7490
+    },
+    {
+      "epoch": 6.002608695652174,
+      "grad_norm": 56.37519454956055,
+      "learning_rate": 3.743961352657005e-05,
+      "loss": 0.0623,
+      "step": 7500
+    },
+    {
+      "epoch": 6.00304347826087,
+      "grad_norm": 0.0025837034918367863,
+      "learning_rate": 3.7415458937198066e-05,
+      "loss": 0.0009,
+      "step": 7510
+    },
+    {
+      "epoch": 6.003478260869565,
+      "grad_norm": 0.0026143037248402834,
+      "learning_rate": 3.739130434782609e-05,
+      "loss": 0.0022,
+      "step": 7520
+    },
+    {
+      "epoch": 6.003913043478261,
+      "grad_norm": 0.0012122008483856916,
+      "learning_rate": 3.736714975845411e-05,
+      "loss": 0.0004,
+      "step": 7530
+    },
+    {
+      "epoch": 6.004043478260869,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8842763274645694,
+      "eval_loss": 0.8199717998504639,
+      "eval_runtime": 75.9271,
+      "eval_samples_per_second": 1.607,
+      "eval_steps_per_second": 0.211,
+      "step": 7533
+    },
+    {
+      "epoch": 7.000304347826087,
+      "grad_norm": 0.0016479407204315066,
+      "learning_rate": 3.7342995169082126e-05,
+      "loss": 0.1146,
+      "step": 7540
+    },
+    {
+      "epoch": 7.000739130434782,
+      "grad_norm": 0.006008865777403116,
+      "learning_rate": 3.7318840579710144e-05,
+      "loss": 0.0003,
+      "step": 7550
+    },
+    {
+      "epoch": 7.001173913043479,
+      "grad_norm": 0.013335944153368473,
+      "learning_rate": 3.729468599033817e-05,
+      "loss": 0.0013,
+      "step": 7560
+    },
+    {
+      "epoch": 7.001608695652174,
+      "grad_norm": 0.010156466625630856,
+      "learning_rate": 3.7270531400966186e-05,
+      "loss": 0.001,
+      "step": 7570
+    },
+    {
+      "epoch": 7.00204347826087,
+      "grad_norm": 97.3908462524414,
+      "learning_rate": 3.7246376811594204e-05,
+      "loss": 0.1149,
+      "step": 7580
+    },
+    {
+      "epoch": 7.002478260869565,
+      "grad_norm": 1.504181146621704,
+      "learning_rate": 3.722222222222222e-05,
+      "loss": 0.0102,
+      "step": 7590
+    },
+    {
+      "epoch": 7.0029130434782605,
+      "grad_norm": 95.67487335205078,
+      "learning_rate": 3.719806763285024e-05,
+      "loss": 0.1669,
+      "step": 7600
+    },
+    {
+      "epoch": 7.003347826086957,
+      "grad_norm": 1.208970069885254,
+      "learning_rate": 3.7173913043478264e-05,
+      "loss": 0.0376,
+      "step": 7610
+    },
+    {
+      "epoch": 7.003782608695652,
+      "grad_norm": 0.00800069235265255,
+      "learning_rate": 3.714975845410628e-05,
+      "loss": 0.1436,
+      "step": 7620
+    },
+    {
+      "epoch": 7.004043478260869,
+      "eval_accuracy": 0.8278688524590164,
+      "eval_f1": 0.8165147652045807,
+      "eval_loss": 1.2253801822662354,
+      "eval_runtime": 75.6061,
+      "eval_samples_per_second": 1.614,
+      "eval_steps_per_second": 0.212,
+      "step": 7626
+    },
+    {
+      "epoch": 8.000173913043477,
+      "grad_norm": 0.7469026446342468,
+      "learning_rate": 3.71256038647343e-05,
+      "loss": 0.1145,
+      "step": 7630
+    },
+    {
+      "epoch": 8.000608695652174,
+      "grad_norm": 0.007953154854476452,
+      "learning_rate": 3.710144927536232e-05,
+      "loss": 0.1984,
+      "step": 7640
+    },
+    {
+      "epoch": 8.00104347826087,
+      "grad_norm": 0.0036359576042741537,
+      "learning_rate": 3.7077294685990336e-05,
+      "loss": 0.0086,
+      "step": 7650
+    },
+    {
+      "epoch": 8.001478260869565,
+      "grad_norm": 0.003452168544754386,
+      "learning_rate": 3.705314009661836e-05,
+      "loss": 0.0541,
+      "step": 7660
+    },
+    {
+      "epoch": 8.001913043478261,
+      "grad_norm": 0.0031451256945729256,
+      "learning_rate": 3.702898550724638e-05,
+      "loss": 0.0092,
+      "step": 7670
+    },
+    {
+      "epoch": 8.002347826086957,
+      "grad_norm": 0.003350870218127966,
+      "learning_rate": 3.70048309178744e-05,
+      "loss": 0.0102,
+      "step": 7680
+    },
+    {
+      "epoch": 8.002782608695652,
+      "grad_norm": 99.97783660888672,
+      "learning_rate": 3.698067632850242e-05,
+      "loss": 0.1118,
+      "step": 7690
+    },
+    {
+      "epoch": 8.003217391304348,
+      "grad_norm": 0.0027307348791509867,
+      "learning_rate": 3.695652173913043e-05,
+      "loss": 0.0789,
+      "step": 7700
+    },
+    {
+      "epoch": 8.003652173913043,
+      "grad_norm": 0.002151375636458397,
+      "learning_rate": 3.6932367149758456e-05,
+      "loss": 0.0002,
+      "step": 7710
+    },
+    {
+      "epoch": 8.00404347826087,
+      "eval_accuracy": 0.860655737704918,
+      "eval_f1": 0.8544375353306953,
+      "eval_loss": 0.8904465436935425,
+      "eval_runtime": 74.2202,
+      "eval_samples_per_second": 1.644,
+      "eval_steps_per_second": 0.216,
+      "step": 7719
+    },
+    {
+      "epoch": 9.000043478260869,
+      "grad_norm": 0.0031512887217104435,
+      "learning_rate": 3.6908212560386474e-05,
+      "loss": 0.0003,
+      "step": 7720
+    },
+    {
+      "epoch": 9.000478260869565,
+      "grad_norm": 0.004215599969029427,
+      "learning_rate": 3.68840579710145e-05,
+      "loss": 0.0434,
+      "step": 7730
+    },
+    {
+      "epoch": 9.000913043478262,
+      "grad_norm": 0.0027098949067294598,
+      "learning_rate": 3.6859903381642516e-05,
+      "loss": 0.0012,
+      "step": 7740
+    },
+    {
+      "epoch": 9.001347826086956,
+      "grad_norm": 0.0018039434216916561,
+      "learning_rate": 3.6835748792270534e-05,
+      "loss": 0.0084,
+      "step": 7750
+    },
+    {
+      "epoch": 9.001782608695652,
+      "grad_norm": 0.002414983231574297,
+      "learning_rate": 3.681159420289855e-05,
+      "loss": 0.0079,
+      "step": 7760
+    },
+    {
+      "epoch": 9.002217391304347,
+      "grad_norm": 0.001327994978055358,
+      "learning_rate": 3.678743961352657e-05,
+      "loss": 0.0002,
+      "step": 7770
+    },
+    {
+      "epoch": 9.002652173913043,
+      "grad_norm": 0.002577988663688302,
+      "learning_rate": 3.6763285024154594e-05,
+      "loss": 0.0805,
+      "step": 7780
+    },
+    {
+      "epoch": 9.00308695652174,
+      "grad_norm": 0.002542161149904132,
+      "learning_rate": 3.673913043478261e-05,
+      "loss": 0.0002,
+      "step": 7790
+    },
+    {
+      "epoch": 9.003521739130434,
+      "grad_norm": 0.002315797843039036,
+      "learning_rate": 3.671497584541063e-05,
+      "loss": 0.0132,
+      "step": 7800
+    },
+    {
+      "epoch": 9.00395652173913,
+      "grad_norm": 12.634377479553223,
+      "learning_rate": 3.669082125603865e-05,
+      "loss": 0.1179,
+      "step": 7810
+    },
+    {
+      "epoch": 9.00404347826087,
+      "eval_accuracy": 0.9016393442622951,
+      "eval_f1": 0.9009970695079303,
+      "eval_loss": 0.6947475671768188,
+      "eval_runtime": 78.5379,
+      "eval_samples_per_second": 1.553,
+      "eval_steps_per_second": 0.204,
+      "step": 7812
+    },
+    {
+      "epoch": 10.000347826086957,
+      "grad_norm": 0.004218494985252619,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.0091,
+      "step": 7820
+    },
+    {
+      "epoch": 10.000782608695653,
+      "grad_norm": 0.07844866067171097,
+      "learning_rate": 3.664251207729469e-05,
+      "loss": 0.0356,
+      "step": 7830
+    },
+    {
+      "epoch": 10.001217391304348,
+      "grad_norm": 0.0030379490926861763,
+      "learning_rate": 3.661835748792271e-05,
+      "loss": 0.0255,
+      "step": 7840
+    },
+    {
+      "epoch": 10.001652173913044,
+      "grad_norm": 0.01330595277249813,
+      "learning_rate": 3.6594202898550726e-05,
+      "loss": 0.0002,
+      "step": 7850
+    },
+    {
+      "epoch": 10.002086956521739,
+      "grad_norm": 0.0024776251520961523,
+      "learning_rate": 3.6570048309178744e-05,
+      "loss": 0.0133,
+      "step": 7860
+    },
+    {
+      "epoch": 10.002521739130435,
+      "grad_norm": 0.005490115378051996,
+      "learning_rate": 3.654589371980677e-05,
+      "loss": 0.0002,
+      "step": 7870
+    },
+    {
+      "epoch": 10.002956521739131,
+      "grad_norm": 0.0017077559605240822,
+      "learning_rate": 3.6521739130434786e-05,
+      "loss": 0.0002,
+      "step": 7880
+    },
+    {
+      "epoch": 10.003391304347826,
+      "grad_norm": 0.0023673507384955883,
+      "learning_rate": 3.6497584541062804e-05,
+      "loss": 0.0071,
+      "step": 7890
+    },
+    {
+      "epoch": 10.003826086956522,
+      "grad_norm": 0.047657690942287445,
+      "learning_rate": 3.647342995169082e-05,
+      "loss": 0.0033,
+      "step": 7900
+    },
+    {
+      "epoch": 10.00404347826087,
+      "eval_accuracy": 0.9016393442622951,
+      "eval_f1": 0.9009244537168961,
+      "eval_loss": 0.6869687438011169,
+      "eval_runtime": 73.4337,
+      "eval_samples_per_second": 1.661,
+      "eval_steps_per_second": 0.218,
+      "step": 7905
+    },
+    {
+      "epoch": 11.000217391304348,
+      "grad_norm": 0.0020451026502996683,
+      "learning_rate": 3.644927536231884e-05,
+      "loss": 0.009,
+      "step": 7910
+    },
+    {
+      "epoch": 11.000652173913043,
+      "grad_norm": 0.03150273114442825,
+      "learning_rate": 3.6425120772946864e-05,
+      "loss": 0.0003,
+      "step": 7920
+    },
+    {
+      "epoch": 11.001086956521739,
+      "grad_norm": 0.0021130284294486046,
+      "learning_rate": 3.640096618357488e-05,
+      "loss": 0.0102,
+      "step": 7930
+    },
+    {
+      "epoch": 11.001521739130435,
+      "grad_norm": 0.0027534847613424063,
+      "learning_rate": 3.63768115942029e-05,
+      "loss": 0.0001,
+      "step": 7940
+    },
+    {
+      "epoch": 11.00195652173913,
+      "grad_norm": 0.0018295743502676487,
+      "learning_rate": 3.635265700483092e-05,
+      "loss": 0.0022,
+      "step": 7950
+    },
+    {
+      "epoch": 11.002391304347826,
+      "grad_norm": 0.001954080071300268,
+      "learning_rate": 3.6328502415458936e-05,
+      "loss": 0.0287,
+      "step": 7960
+    },
+    {
+      "epoch": 11.00282608695652,
+      "grad_norm": 0.001990045653656125,
+      "learning_rate": 3.630434782608696e-05,
+      "loss": 0.0001,
+      "step": 7970
+    },
+    {
+      "epoch": 11.003260869565217,
+      "grad_norm": 0.0021924017928540707,
+      "learning_rate": 3.628019323671498e-05,
+      "loss": 0.1093,
+      "step": 7980
+    },
+    {
+      "epoch": 11.003695652173914,
+      "grad_norm": 0.0014019764494150877,
+      "learning_rate": 3.6256038647343e-05,
+      "loss": 0.0003,
+      "step": 7990
+    },
+    {
+      "epoch": 11.00404347826087,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8834037386396851,
+      "eval_loss": 0.6801308393478394,
+      "eval_runtime": 76.3228,
+      "eval_samples_per_second": 1.598,
+      "eval_steps_per_second": 0.21,
+      "step": 7998
+    },
+    {
+      "epoch": 12.00008695652174,
+      "grad_norm": 0.0022465840447694063,
+      "learning_rate": 3.6231884057971014e-05,
+      "loss": 0.0086,
+      "step": 8000
+    },
+    {
+      "epoch": 12.000521739130434,
+      "grad_norm": 0.009984511882066727,
+      "learning_rate": 3.620772946859903e-05,
+      "loss": 0.0171,
+      "step": 8010
+    },
+    {
+      "epoch": 12.00095652173913,
+      "grad_norm": 0.0018829620676115155,
+      "learning_rate": 3.6183574879227056e-05,
+      "loss": 0.0082,
+      "step": 8020
+    },
+    {
+      "epoch": 12.001391304347827,
+      "grad_norm": 0.0022015601862221956,
+      "learning_rate": 3.6159420289855074e-05,
+      "loss": 0.0018,
+      "step": 8030
+    },
+    {
+      "epoch": 12.001826086956521,
+      "grad_norm": 0.0011975386878475547,
+      "learning_rate": 3.61352657004831e-05,
+      "loss": 0.0168,
+      "step": 8040
+    },
+    {
+      "epoch": 12.002260869565218,
+      "grad_norm": 0.0013748151250183582,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.0003,
+      "step": 8050
+    },
+    {
+      "epoch": 12.002695652173912,
+      "grad_norm": 0.0027548738289624453,
+      "learning_rate": 3.6086956521739134e-05,
+      "loss": 0.0103,
+      "step": 8060
+    },
+    {
+      "epoch": 12.003130434782609,
+      "grad_norm": 0.001284057623706758,
+      "learning_rate": 3.606280193236715e-05,
+      "loss": 0.1057,
+      "step": 8070
+    },
+    {
+      "epoch": 12.003565217391305,
+      "grad_norm": 0.0024062602315098047,
+      "learning_rate": 3.603864734299517e-05,
+      "loss": 0.0391,
+      "step": 8080
+    },
+    {
+      "epoch": 12.004,
+      "grad_norm": 1.8473665714263916,
+      "learning_rate": 3.6014492753623194e-05,
+      "loss": 0.0566,
+      "step": 8090
+    },
+    {
+      "epoch": 12.00404347826087,
+      "eval_accuracy": 0.8852459016393442,
+      "eval_f1": 0.8826751117734725,
+      "eval_loss": 0.7470003366470337,
+      "eval_runtime": 76.8169,
+      "eval_samples_per_second": 1.588,
+      "eval_steps_per_second": 0.208,
+      "step": 8091
+    },
+    {
+      "epoch": 13.000391304347826,
+      "grad_norm": 0.006008814089000225,
+      "learning_rate": 3.5990338164251205e-05,
+      "loss": 0.01,
+      "step": 8100
+    },
+    {
+      "epoch": 13.000826086956522,
+      "grad_norm": 0.003681233851239085,
+      "learning_rate": 3.596618357487923e-05,
+      "loss": 0.0084,
+      "step": 8110
+    },
+    {
+      "epoch": 13.001260869565217,
+      "grad_norm": 0.002310408977791667,
+      "learning_rate": 3.594202898550725e-05,
+      "loss": 0.1267,
+      "step": 8120
+    },
+    {
+      "epoch": 13.001695652173913,
+      "grad_norm": 0.009503014385700226,
+      "learning_rate": 3.5917874396135266e-05,
+      "loss": 0.0066,
+      "step": 8130
+    },
+    {
+      "epoch": 13.00213043478261,
+      "grad_norm": 0.009119764901697636,
+      "learning_rate": 3.589371980676329e-05,
+      "loss": 0.0108,
+      "step": 8140
+    },
+    {
+      "epoch": 13.002565217391304,
+      "grad_norm": 0.00435229716822505,
+      "learning_rate": 3.58695652173913e-05,
+      "loss": 0.0003,
+      "step": 8150
+    },
+    {
+      "epoch": 13.003,
+      "grad_norm": 0.0040861936286091805,
+      "learning_rate": 3.5845410628019326e-05,
+      "loss": 0.0109,
+      "step": 8160
+    },
+    {
+      "epoch": 13.003434782608696,
+      "grad_norm": 0.0029671050142496824,
+      "learning_rate": 3.5821256038647344e-05,
+      "loss": 0.0003,
+      "step": 8170
+    },
+    {
+      "epoch": 13.003869565217391,
+      "grad_norm": 0.0019406734500080347,
+      "learning_rate": 3.579710144927537e-05,
+      "loss": 0.0111,
+      "step": 8180
+    },
+    {
+      "epoch": 13.00404347826087,
+      "eval_accuracy": 0.8770491803278688,
+      "eval_f1": 0.8738998934662604,
+      "eval_loss": 0.738959789276123,
+      "eval_runtime": 77.6141,
+      "eval_samples_per_second": 1.572,
+      "eval_steps_per_second": 0.206,
+      "step": 8184
+    },
+    {
+      "epoch": 14.000260869565217,
+      "grad_norm": 0.0053750560618937016,
+      "learning_rate": 3.5772946859903386e-05,
+      "loss": 0.0188,
+      "step": 8190
+    },
+    {
+      "epoch": 14.000695652173913,
+      "grad_norm": 0.0030118555296212435,
+      "learning_rate": 3.57487922705314e-05,
+      "loss": 0.0217,
+      "step": 8200
+    },
+    {
+      "epoch": 14.001130434782608,
+      "grad_norm": 0.0030062675941735506,
+      "learning_rate": 3.572463768115942e-05,
+      "loss": 0.0091,
+      "step": 8210
+    },
+    {
+      "epoch": 14.001565217391304,
+      "grad_norm": 0.002661097561940551,
+      "learning_rate": 3.570048309178744e-05,
+      "loss": 0.0002,
+      "step": 8220
+    },
+    {
+      "epoch": 14.002,
+      "grad_norm": 0.0018812001217156649,
+      "learning_rate": 3.5676328502415464e-05,
+      "loss": 0.0124,
+      "step": 8230
+    },
+    {
+      "epoch": 14.002434782608695,
+      "grad_norm": 0.003940836526453495,
+      "learning_rate": 3.565217391304348e-05,
+      "loss": 0.0001,
+      "step": 8240
+    },
+    {
+      "epoch": 14.002869565217392,
+      "grad_norm": 0.0017722449265420437,
+      "learning_rate": 3.56280193236715e-05,
+      "loss": 0.0001,
+      "step": 8250
+    },
+    {
+      "epoch": 14.003304347826086,
+      "grad_norm": 0.0024237651377916336,
+      "learning_rate": 3.560386473429952e-05,
+      "loss": 0.0001,
+      "step": 8260
+    },
+    {
+      "epoch": 14.003739130434782,
+      "grad_norm": 0.0013878210447728634,
+      "learning_rate": 3.5579710144927535e-05,
+      "loss": 0.0096,
+      "step": 8270
+    },
+    {
+      "epoch": 14.00404347826087,
+      "eval_accuracy": 0.9262295081967213,
+      "eval_f1": 0.9249199343461638,
+      "eval_loss": 0.449712336063385,
+      "eval_runtime": 71.2976,
+      "eval_samples_per_second": 1.711,
+      "eval_steps_per_second": 0.224,
+      "step": 8277
+    },
+    {
+      "epoch": 15.000130434782609,
+      "grad_norm": 0.0015423373552039266,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 0.0096,
+      "step": 8280
+    },
+    {
+      "epoch": 15.000565217391305,
+      "grad_norm": 0.0017125405138358474,
+      "learning_rate": 3.553140096618358e-05,
+      "loss": 0.0094,
+      "step": 8290
+    },
+    {
+      "epoch": 15.001,
+      "grad_norm": 0.0017512089107185602,
+      "learning_rate": 3.5507246376811596e-05,
+      "loss": 0.0001,
+      "step": 8300
+    },
+    {
+      "epoch": 15.001434782608696,
+      "grad_norm": 0.0016162145184352994,
+      "learning_rate": 3.5483091787439613e-05,
+      "loss": 0.0001,
+      "step": 8310
+    },
+    {
+      "epoch": 15.001869565217392,
+      "grad_norm": 0.001804832834750414,
+      "learning_rate": 3.545893719806763e-05,
+      "loss": 0.0098,
+      "step": 8320
+    },
+    {
+      "epoch": 15.002304347826087,
+      "grad_norm": 0.0017595053650438786,
+      "learning_rate": 3.5434782608695656e-05,
+      "loss": 0.0001,
+      "step": 8330
+    },
+    {
+      "epoch": 15.002739130434783,
+      "grad_norm": 0.0015612575225532055,
+      "learning_rate": 3.5410628019323674e-05,
+      "loss": 0.0186,
+      "step": 8340
+    },
+    {
+      "epoch": 15.003173913043478,
+      "grad_norm": 0.01624816469848156,
+      "learning_rate": 3.538647342995169e-05,
+      "loss": 0.0138,
+      "step": 8350
+    },
+    {
+      "epoch": 15.003608695652174,
+      "grad_norm": 0.0014068948803469539,
+      "learning_rate": 3.536231884057971e-05,
+      "loss": 0.0308,
+      "step": 8360
+    },
+    {
+      "epoch": 15.00404347826087,
+      "grad_norm": 0.003260946599766612,
+      "learning_rate": 3.5338164251207734e-05,
+      "loss": 0.0001,
+      "step": 8370
+    },
+    {
+      "epoch": 15.00404347826087,
+      "eval_accuracy": 0.9262295081967213,
+      "eval_f1": 0.9258667205923087,
+      "eval_loss": 0.5235211253166199,
+      "eval_runtime": 86.922,
+      "eval_samples_per_second": 1.404,
+      "eval_steps_per_second": 0.184,
+      "step": 8370
+    },
+    {
+      "epoch": 0.0010869565217391304,
+      "grad_norm": 0.0017507346346974373,
+      "learning_rate": 4.951690821256039e-06,
+      "loss": 0.0001,
+      "step": 8380
+    },
+    {
+      "epoch": 0.002173913043478261,
+      "grad_norm": 0.0017071341862902045,
+      "learning_rate": 4.891304347826087e-06,
+      "loss": 0.0001,
+      "step": 8390
+    },
+    {
+      "epoch": 0.003260869565217391,
+      "grad_norm": 0.0015891263028606772,
+      "learning_rate": 4.830917874396135e-06,
+      "loss": 0.0001,
+      "step": 8400
+    },
+    {
+      "epoch": 0.004347826086956522,
+      "grad_norm": 0.002104247687384486,
+      "learning_rate": 4.770531400966183e-06,
+      "loss": 0.0099,
+      "step": 8410
+    },
+    {
+      "epoch": 0.005434782608695652,
+      "grad_norm": 0.0007910887943580747,
+      "learning_rate": 4.710144927536232e-06,
+      "loss": 0.008,
+      "step": 8420
+    },
+    {
+      "epoch": 0.006521739130434782,
+      "grad_norm": 0.0013656615046784282,
+      "learning_rate": 4.64975845410628e-06,
+      "loss": 0.0101,
+      "step": 8430
+    },
+    {
+      "epoch": 0.007608695652173913,
+      "grad_norm": 1.222392201423645,
+      "learning_rate": 4.589371980676329e-06,
+      "loss": 0.0146,
+      "step": 8440
+    },
+    {
+      "epoch": 0.008695652173913044,
+      "grad_norm": 0.0019674005452543497,
+      "learning_rate": 4.528985507246377e-06,
+      "loss": 0.0108,
+      "step": 8450
+    },
+    {
+      "epoch": 0.009782608695652175,
+      "grad_norm": 0.0012134442804381251,
+      "learning_rate": 4.468599033816425e-06,
+      "loss": 0.0002,
+      "step": 8460
+    },
+    {
+      "epoch": 0.010108695652173913,
+      "eval_accuracy": 0.9262295081967213,
+      "eval_f1": 0.9261042886093471,
+      "eval_loss": 0.5340837240219116,
+      "eval_runtime": 87.7431,
+      "eval_samples_per_second": 1.39,
+      "eval_steps_per_second": 0.182,
+      "step": 8463
+    },
+    {
+      "epoch": 1.0007608695652175,
+      "grad_norm": 0.001567196799442172,
+      "learning_rate": 4.408212560386473e-06,
+      "loss": 0.0001,
+      "step": 8470
+    },
+    {
+      "epoch": 1.0018478260869565,
+      "grad_norm": 0.0016586320707574487,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 0.0096,
+      "step": 8480
+    },
+    {
+      "epoch": 1.0029347826086956,
+      "grad_norm": 0.0023030065931379795,
+      "learning_rate": 4.28743961352657e-06,
+      "loss": 0.0077,
+      "step": 8490
+    },
+    {
+      "epoch": 1.0040217391304347,
+      "grad_norm": 0.002070185961201787,
+      "learning_rate": 4.227053140096619e-06,
+      "loss": 0.0001,
+      "step": 8500
+    },
+    {
+      "epoch": 1.005108695652174,
+      "grad_norm": 0.0012701443629339337,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.0106,
+      "step": 8510
+    },
+    {
+      "epoch": 1.006195652173913,
+      "grad_norm": 0.0018321698298677802,
+      "learning_rate": 4.106280193236716e-06,
+      "loss": 0.0001,
+      "step": 8520
+    },
+    {
+      "epoch": 1.0072826086956521,
+      "grad_norm": 0.0011002240935340524,
+      "learning_rate": 4.045893719806764e-06,
+      "loss": 0.0002,
+      "step": 8530
+    },
+    {
+      "epoch": 1.0083695652173914,
+      "grad_norm": 0.0006837508990429342,
+      "learning_rate": 3.9855072463768115e-06,
+      "loss": 0.0253,
+      "step": 8540
+    },
+    {
+      "epoch": 1.0094565217391305,
+      "grad_norm": 0.0011974646477028728,
+      "learning_rate": 3.92512077294686e-06,
+      "loss": 0.0001,
+      "step": 8550
+    },
+    {
+      "epoch": 1.0101086956521739,
+      "eval_accuracy": 0.9180327868852459,
+      "eval_f1": 0.9178693897416935,
+      "eval_loss": 0.5411162376403809,
+      "eval_runtime": 71.3448,
+      "eval_samples_per_second": 1.71,
+      "eval_steps_per_second": 0.224,
+      "step": 8556
+    },
+    {
+      "epoch": 2.0004347826086954,
+      "grad_norm": 0.00192980095744133,
+      "learning_rate": 3.864734299516908e-06,
+      "loss": 0.0747,
+      "step": 8560
+    },
+    {
+      "epoch": 2.001521739130435,
+      "grad_norm": 0.0019794933032244444,
+      "learning_rate": 3.804347826086957e-06,
+      "loss": 0.0176,
+      "step": 8570
+    },
+    {
+      "epoch": 2.002608695652174,
+      "grad_norm": 0.0013387734070420265,
+      "learning_rate": 3.743961352657005e-06,
+      "loss": 0.0028,
+      "step": 8580
+    },
+    {
+      "epoch": 2.003695652173913,
+      "grad_norm": 0.0012565733632072806,
+      "learning_rate": 3.683574879227053e-06,
+      "loss": 0.0001,
+      "step": 8590
+    },
+    {
+      "epoch": 2.004782608695652,
+      "grad_norm": 0.0024499129503965378,
+      "learning_rate": 3.6231884057971017e-06,
+      "loss": 0.0001,
+      "step": 8600
+    },
+    {
+      "epoch": 2.0058695652173912,
+      "grad_norm": 0.0025332611985504627,
+      "learning_rate": 3.5628019323671496e-06,
+      "loss": 0.0001,
+      "step": 8610
+    },
+    {
+      "epoch": 2.0069565217391303,
+      "grad_norm": 0.0016111568547785282,
+      "learning_rate": 3.5024154589371984e-06,
+      "loss": 0.0075,
+      "step": 8620
+    },
+    {
+      "epoch": 2.0080434782608694,
+      "grad_norm": 0.0015177864115685225,
+      "learning_rate": 3.4420289855072464e-06,
+      "loss": 0.0068,
+      "step": 8630
+    },
+    {
+      "epoch": 2.009130434782609,
+      "grad_norm": 0.002059792634099722,
+      "learning_rate": 3.3816425120772947e-06,
+      "loss": 0.0211,
+      "step": 8640
+    },
+    {
+      "epoch": 2.010108695652174,
+      "eval_accuracy": 0.9180327868852459,
+      "eval_f1": 0.9178693897416935,
+      "eval_loss": 0.5265507698059082,
+      "eval_runtime": 77.2441,
+      "eval_samples_per_second": 1.579,
+      "eval_steps_per_second": 0.207,
+      "step": 8649
+    },
+    {
+      "epoch": 3.000108695652174,
+      "grad_norm": 0.0017185317119583488,
+      "learning_rate": 3.3212560386473435e-06,
+      "loss": 0.0001,
+      "step": 8650
+    },
+    {
+      "epoch": 3.001195652173913,
+      "grad_norm": 0.0017326638335362077,
+      "learning_rate": 3.2608695652173914e-06,
+      "loss": 0.0079,
+      "step": 8660
+    },
+    {
+      "epoch": 3.002282608695652,
+      "grad_norm": 0.002113594673573971,
+      "learning_rate": 3.2004830917874402e-06,
+      "loss": 0.0001,
+      "step": 8670
+    },
+    {
+      "epoch": 3.0033695652173913,
+      "grad_norm": 0.0021217581816017628,
+      "learning_rate": 3.140096618357488e-06,
+      "loss": 0.0001,
+      "step": 8680
+    },
+    {
+      "epoch": 3.0044565217391304,
+      "grad_norm": 0.0013151352759450674,
+      "learning_rate": 3.0797101449275365e-06,
+      "loss": 0.0001,
+      "step": 8690
+    },
+    {
+      "epoch": 3.0055434782608694,
+      "grad_norm": 0.001686801202595234,
+      "learning_rate": 3.0193236714975845e-06,
+      "loss": 0.0098,
+      "step": 8700
+    },
+    {
+      "epoch": 3.0066304347826085,
+      "grad_norm": 0.001321412273682654,
+      "learning_rate": 2.958937198067633e-06,
+      "loss": 0.0105,
+      "step": 8710
+    },
+    {
+      "epoch": 3.007717391304348,
+      "grad_norm": 0.0011213988764211535,
+      "learning_rate": 2.898550724637681e-06,
+      "loss": 0.01,
+      "step": 8720
+    },
+    {
+      "epoch": 3.008804347826087,
+      "grad_norm": 0.0020139431580901146,
+      "learning_rate": 2.8381642512077295e-06,
+      "loss": 0.0076,
+      "step": 8730
+    },
+    {
+      "epoch": 3.009891304347826,
+      "grad_norm": 0.0018427580362185836,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 0.0077,
+      "step": 8740
+    },
+    {
+      "epoch": 3.010108695652174,
+      "eval_accuracy": 0.9180327868852459,
+      "eval_f1": 0.9178693897416935,
+      "eval_loss": 0.5282787680625916,
+      "eval_runtime": 69.3627,
+      "eval_samples_per_second": 1.759,
+      "eval_steps_per_second": 0.231,
+      "step": 8742
+    },
+    {
+      "epoch": 4.000869565217391,
+      "grad_norm": 0.0015537792351096869,
+      "learning_rate": 2.7173913043478263e-06,
+      "loss": 0.0104,
+      "step": 8750
+    },
+    {
+      "epoch": 4.001956521739131,
+      "grad_norm": 0.0009913562098518014,
+      "learning_rate": 2.6570048309178746e-06,
+      "loss": 0.008,
+      "step": 8760
+    },
+    {
+      "epoch": 4.00304347826087,
+      "grad_norm": 0.0024362721014767885,
+      "learning_rate": 2.596618357487923e-06,
+      "loss": 0.0001,
+      "step": 8770
+    },
+    {
+      "epoch": 4.004130434782609,
+      "grad_norm": 0.0016615530475974083,
+      "learning_rate": 2.5362318840579714e-06,
+      "loss": 0.0001,
+      "step": 8780
+    },
+    {
+      "epoch": 4.005217391304348,
+      "grad_norm": 0.0013662370620295405,
+      "learning_rate": 2.4758454106280193e-06,
+      "loss": 0.0001,
+      "step": 8790
+    },
+    {
+      "epoch": 4.006304347826087,
+      "grad_norm": 0.0018021605210378766,
+      "learning_rate": 2.4154589371980677e-06,
+      "loss": 0.0001,
+      "step": 8800
+    },
+    {
+      "epoch": 4.007391304347826,
+      "grad_norm": 2.3952181339263916,
+      "learning_rate": 2.355072463768116e-06,
+      "loss": 0.0146,
+      "step": 8810
+    },
+    {
+      "epoch": 4.008478260869565,
+      "grad_norm": 0.0014442868996411562,
+      "learning_rate": 2.2946859903381644e-06,
+      "loss": 0.0098,
+      "step": 8820
+    },
+    {
+      "epoch": 4.009565217391304,
+      "grad_norm": 0.0016332181403413415,
+      "learning_rate": 2.2342995169082123e-06,
+      "loss": 0.0101,
+      "step": 8830
+    },
+    {
+      "epoch": 4.010108695652174,
+      "eval_accuracy": 0.9180327868852459,
+      "eval_f1": 0.9178693897416935,
+      "eval_loss": 0.5279106497764587,
+      "eval_runtime": 68.8675,
+      "eval_samples_per_second": 1.772,
+      "eval_steps_per_second": 0.232,
+      "step": 8835
+    },
+    {
+      "epoch": 5.0005434782608695,
+      "grad_norm": 0.001660835579968989,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 0.0001,
+      "step": 8840
+    },
+    {
+      "epoch": 5.001630434782609,
+      "grad_norm": 0.0013907772954553366,
+      "learning_rate": 2.1135265700483095e-06,
+      "loss": 0.015,
+      "step": 8850
+    },
+    {
+      "epoch": 5.002391304347826,
+      "eval_accuracy": 0.8777292576419214,
+      "eval_f1": 0.8759548256004502,
+      "eval_loss": 0.9210804104804993,
+      "eval_runtime": 135.0706,
+      "eval_samples_per_second": 1.695,
+      "eval_steps_per_second": 0.215,
+      "step": 8857
+    },
+    {
+      "epoch": 5.002391304347826,
+      "eval_accuracy": 0.9180327868852459,
+      "eval_f1": 0.9178693897416935,
+      "eval_loss": 0.5277689099311829,
+      "eval_runtime": 72.002,
+      "eval_samples_per_second": 1.694,
+      "eval_steps_per_second": 0.222,
+      "step": 8857
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9200,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.786781173362655e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}