diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5416 @@
+{
+  "best_metric": 0.7045454545454546,
+  "best_model_checkpoint": "squarerun/checkpoint-435",
+  "epoch": 45.0,
+  "eval_steps": 500,
+  "global_step": 1305,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.06896551724137931,
+      "grad_norm": 2.82375431060791,
+      "learning_rate": 1.5267175572519084e-06,
+      "loss": 1.9171,
+      "step": 2
+    },
+    {
+      "epoch": 0.13793103448275862,
+      "grad_norm": 1.8254162073135376,
+      "learning_rate": 3.053435114503817e-06,
+      "loss": 1.9196,
+      "step": 4
+    },
+    {
+      "epoch": 0.20689655172413793,
+      "grad_norm": 1.877601146697998,
+      "learning_rate": 4.580152671755725e-06,
+      "loss": 1.9126,
+      "step": 6
+    },
+    {
+      "epoch": 0.27586206896551724,
+      "grad_norm": 1.842034935951233,
+      "learning_rate": 6.106870229007634e-06,
+      "loss": 1.9303,
+      "step": 8
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 2.6617424488067627,
+      "learning_rate": 7.633587786259543e-06,
+      "loss": 1.9303,
+      "step": 10
+    },
+    {
+      "epoch": 0.41379310344827586,
+      "grad_norm": 1.9283453226089478,
+      "learning_rate": 9.16030534351145e-06,
+      "loss": 1.957,
+      "step": 12
+    },
+    {
+      "epoch": 0.4827586206896552,
+      "grad_norm": 1.7121816873550415,
+      "learning_rate": 1.0687022900763359e-05,
+      "loss": 1.9194,
+      "step": 14
+    },
+    {
+      "epoch": 0.5517241379310345,
+      "grad_norm": 1.5719634294509888,
+      "learning_rate": 1.2213740458015267e-05,
+      "loss": 1.9612,
+      "step": 16
+    },
+    {
+      "epoch": 0.6206896551724138,
+      "grad_norm": 1.857325553894043,
+      "learning_rate": 1.3740458015267178e-05,
+      "loss": 1.8854,
+      "step": 18
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 2.044182538986206,
+      "learning_rate": 1.5267175572519086e-05,
+      "loss": 1.9012,
+      "step": 20
+    },
+    {
+      "epoch": 0.7586206896551724,
+      "grad_norm": 1.0480695962905884,
+      "learning_rate": 1.6793893129770993e-05,
+      "loss": 1.8995,
+      "step": 22
+    },
+    {
+      "epoch": 0.8275862068965517,
+      "grad_norm": 1.2127116918563843,
+      "learning_rate": 1.83206106870229e-05,
+      "loss": 1.8923,
+      "step": 24
+    },
+    {
+      "epoch": 0.896551724137931,
+      "grad_norm": 2.0899465084075928,
+      "learning_rate": 1.984732824427481e-05,
+      "loss": 1.8761,
+      "step": 26
+    },
+    {
+      "epoch": 0.9655172413793104,
+      "grad_norm": 1.5972628593444824,
+      "learning_rate": 2.1374045801526718e-05,
+      "loss": 1.903,
+      "step": 28
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.17424242424242425,
+      "eval_f1_macro": 0.06584113560857748,
+      "eval_f1_micro": 0.17424242424242425,
+      "eval_f1_weighted": 0.09004420526619257,
+      "eval_loss": 1.8867504596710205,
+      "eval_precision_macro": 0.050235478806907374,
+      "eval_precision_micro": 0.17424242424242425,
+      "eval_precision_weighted": 0.0693056943056943,
+      "eval_recall_macro": 0.1293121693121693,
+      "eval_recall_micro": 0.17424242424242425,
+      "eval_recall_weighted": 0.17424242424242425,
+      "eval_runtime": 2.3295,
+      "eval_samples_per_second": 56.664,
+      "eval_steps_per_second": 7.298,
+      "step": 29
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "grad_norm": 2.178473711013794,
+      "learning_rate": 2.2900763358778628e-05,
+      "loss": 1.891,
+      "step": 30
+    },
+    {
+      "epoch": 1.103448275862069,
+      "grad_norm": 1.646786093711853,
+      "learning_rate": 2.4427480916030535e-05,
+      "loss": 1.8301,
+      "step": 32
+    },
+    {
+      "epoch": 1.1724137931034484,
+      "grad_norm": 1.4257752895355225,
+      "learning_rate": 2.5954198473282442e-05,
+      "loss": 1.8346,
+      "step": 34
+    },
+    {
+      "epoch": 1.2413793103448276,
+      "grad_norm": 1.3215000629425049,
+      "learning_rate": 2.7480916030534355e-05,
+      "loss": 1.905,
+      "step": 36
+    },
+    {
+      "epoch": 1.3103448275862069,
+      "grad_norm": 1.763136625289917,
+      "learning_rate": 2.900763358778626e-05,
+      "loss": 1.8771,
+      "step": 38
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "grad_norm": 1.7233166694641113,
+      "learning_rate": 3.053435114503817e-05,
+      "loss": 1.8977,
+      "step": 40
+    },
+    {
+      "epoch": 1.4482758620689655,
+      "grad_norm": 1.3865277767181396,
+      "learning_rate": 3.2061068702290076e-05,
+      "loss": 1.8815,
+      "step": 42
+    },
+    {
+      "epoch": 1.5172413793103448,
+      "grad_norm": 2.227480411529541,
+      "learning_rate": 3.358778625954199e-05,
+      "loss": 1.8876,
+      "step": 44
+    },
+    {
+      "epoch": 1.5862068965517242,
+      "grad_norm": 1.27533757686615,
+      "learning_rate": 3.511450381679389e-05,
+      "loss": 1.9272,
+      "step": 46
+    },
+    {
+      "epoch": 1.6551724137931034,
+      "grad_norm": 1.4631495475769043,
+      "learning_rate": 3.66412213740458e-05,
+      "loss": 1.8779,
+      "step": 48
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 1.4583027362823486,
+      "learning_rate": 3.816793893129771e-05,
+      "loss": 1.9327,
+      "step": 50
+    },
+    {
+      "epoch": 1.793103448275862,
+      "grad_norm": 1.856370210647583,
+      "learning_rate": 3.969465648854962e-05,
+      "loss": 1.84,
+      "step": 52
+    },
+    {
+      "epoch": 1.8620689655172413,
+      "grad_norm": 2.055604934692383,
+      "learning_rate": 4.122137404580153e-05,
+      "loss": 1.9563,
+      "step": 54
+    },
+    {
+      "epoch": 1.9310344827586206,
+      "grad_norm": 2.059593439102173,
+      "learning_rate": 4.2748091603053435e-05,
+      "loss": 1.8515,
+      "step": 56
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.1090753078460693,
+      "learning_rate": 4.4274809160305345e-05,
+      "loss": 1.8662,
+      "step": 58
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.2196969696969697,
+      "eval_f1_macro": 0.07539682539682539,
+      "eval_f1_micro": 0.2196969696969697,
+      "eval_f1_weighted": 0.10037878787878787,
+      "eval_loss": 1.8739551305770874,
+      "eval_precision_macro": 0.06031746031746032,
+      "eval_precision_micro": 0.2196969696969697,
+      "eval_precision_weighted": 0.07727272727272727,
+      "eval_recall_macro": 0.15797430083144368,
+      "eval_recall_micro": 0.2196969696969697,
+      "eval_recall_weighted": 0.2196969696969697,
+      "eval_runtime": 2.3196,
+      "eval_samples_per_second": 56.907,
+      "eval_steps_per_second": 7.329,
+      "step": 58
+    },
+    {
+      "epoch": 2.0689655172413794,
+      "grad_norm": 1.6586414575576782,
+      "learning_rate": 4.5801526717557256e-05,
+      "loss": 1.9037,
+      "step": 60
+    },
+    {
+      "epoch": 2.1379310344827585,
+      "grad_norm": 1.2022308111190796,
+      "learning_rate": 4.7328244274809166e-05,
+      "loss": 1.8325,
+      "step": 62
+    },
+    {
+      "epoch": 2.206896551724138,
+      "grad_norm": 2.1608262062072754,
+      "learning_rate": 4.885496183206107e-05,
+      "loss": 1.8745,
+      "step": 64
+    },
+    {
+      "epoch": 2.2758620689655173,
+      "grad_norm": 2.147564172744751,
+      "learning_rate": 5.038167938931297e-05,
+      "loss": 1.9165,
+      "step": 66
+    },
+    {
+      "epoch": 2.344827586206897,
+      "grad_norm": 1.5065761804580688,
+      "learning_rate": 5.1908396946564884e-05,
+      "loss": 1.7814,
+      "step": 68
+    },
+    {
+      "epoch": 2.413793103448276,
+      "grad_norm": 1.8903454542160034,
+      "learning_rate": 5.3435114503816794e-05,
+      "loss": 1.8222,
+      "step": 70
+    },
+    {
+      "epoch": 2.4827586206896552,
+      "grad_norm": 1.8370472192764282,
+      "learning_rate": 5.496183206106871e-05,
+      "loss": 1.8853,
+      "step": 72
+    },
+    {
+      "epoch": 2.5517241379310347,
+      "grad_norm": 1.6297768354415894,
+      "learning_rate": 5.648854961832062e-05,
+      "loss": 1.8725,
+      "step": 74
+    },
+    {
+      "epoch": 2.6206896551724137,
+      "grad_norm": 1.4402002096176147,
+      "learning_rate": 5.801526717557252e-05,
+      "loss": 1.8109,
+      "step": 76
+    },
+    {
+      "epoch": 2.689655172413793,
+      "grad_norm": 2.1347367763519287,
+      "learning_rate": 5.954198473282443e-05,
+      "loss": 1.8819,
+      "step": 78
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "grad_norm": 1.732499599456787,
+      "learning_rate": 6.106870229007635e-05,
+      "loss": 1.9278,
+      "step": 80
+    },
+    {
+      "epoch": 2.8275862068965516,
+      "grad_norm": 1.4573041200637817,
+      "learning_rate": 6.259541984732826e-05,
+      "loss": 1.8427,
+      "step": 82
+    },
+    {
+      "epoch": 2.896551724137931,
+      "grad_norm": 1.4883620738983154,
+      "learning_rate": 6.412213740458015e-05,
+      "loss": 1.9268,
+      "step": 84
+    },
+    {
+      "epoch": 2.9655172413793105,
+      "grad_norm": 1.8515830039978027,
+      "learning_rate": 6.564885496183206e-05,
+      "loss": 1.9291,
+      "step": 86
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.20454545454545456,
+      "eval_f1_macro": 0.04851752021563342,
+      "eval_f1_micro": 0.20454545454545456,
+      "eval_f1_weighted": 0.06946826758147512,
+      "eval_loss": 1.8862061500549316,
+      "eval_precision_macro": 0.029220779220779224,
+      "eval_precision_micro": 0.20454545454545456,
+      "eval_precision_weighted": 0.04183884297520662,
+      "eval_recall_macro": 0.14285714285714285,
+      "eval_recall_micro": 0.20454545454545456,
+      "eval_recall_weighted": 0.20454545454545456,
+      "eval_runtime": 2.3118,
+      "eval_samples_per_second": 57.1,
+      "eval_steps_per_second": 7.354,
+      "step": 87
+    },
+    {
+      "epoch": 3.0344827586206895,
+      "grad_norm": 2.0376968383789062,
+      "learning_rate": 6.717557251908397e-05,
+      "loss": 1.9533,
+      "step": 88
+    },
+    {
+      "epoch": 3.103448275862069,
+      "grad_norm": 1.5388916730880737,
+      "learning_rate": 6.870229007633588e-05,
+      "loss": 1.8628,
+      "step": 90
+    },
+    {
+      "epoch": 3.1724137931034484,
+      "grad_norm": 1.5677577257156372,
+      "learning_rate": 7.022900763358778e-05,
+      "loss": 1.9064,
+      "step": 92
+    },
+    {
+      "epoch": 3.2413793103448274,
+      "grad_norm": 2.0428049564361572,
+      "learning_rate": 7.175572519083969e-05,
+      "loss": 1.8747,
+      "step": 94
+    },
+    {
+      "epoch": 3.310344827586207,
+      "grad_norm": 2.1180825233459473,
+      "learning_rate": 7.32824427480916e-05,
+      "loss": 1.8842,
+      "step": 96
+    },
+    {
+      "epoch": 3.3793103448275863,
+      "grad_norm": 1.6793889999389648,
+      "learning_rate": 7.480916030534351e-05,
+      "loss": 1.8041,
+      "step": 98
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "grad_norm": 1.4385968446731567,
+      "learning_rate": 7.633587786259542e-05,
+      "loss": 1.7604,
+      "step": 100
+    },
+    {
+      "epoch": 3.5172413793103448,
+      "grad_norm": 2.31620717048645,
+      "learning_rate": 7.786259541984733e-05,
+      "loss": 1.8775,
+      "step": 102
+    },
+    {
+      "epoch": 3.586206896551724,
+      "grad_norm": 1.4493972063064575,
+      "learning_rate": 7.938931297709924e-05,
+      "loss": 1.7543,
+      "step": 104
+    },
+    {
+      "epoch": 3.655172413793103,
+      "grad_norm": 1.9932034015655518,
+      "learning_rate": 8.091603053435115e-05,
+      "loss": 1.8983,
+      "step": 106
+    },
+    {
+      "epoch": 3.7241379310344827,
+      "grad_norm": 1.7643533945083618,
+      "learning_rate": 8.244274809160306e-05,
+      "loss": 1.8041,
+      "step": 108
+    },
+    {
+      "epoch": 3.793103448275862,
+      "grad_norm": 2.046786069869995,
+      "learning_rate": 8.396946564885496e-05,
+      "loss": 1.8857,
+      "step": 110
+    },
+    {
+      "epoch": 3.862068965517241,
+      "grad_norm": 2.543879985809326,
+      "learning_rate": 8.549618320610687e-05,
+      "loss": 1.9249,
+      "step": 112
+    },
+    {
+      "epoch": 3.9310344827586206,
+      "grad_norm": 2.2258007526397705,
+      "learning_rate": 8.702290076335878e-05,
+      "loss": 1.8182,
+      "step": 114
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.7140247821807861,
+      "learning_rate": 8.854961832061069e-05,
+      "loss": 1.7838,
+      "step": 116
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.26515151515151514,
+      "eval_f1_macro": 0.11706349206349206,
+      "eval_f1_micro": 0.26515151515151514,
+      "eval_f1_weighted": 0.14741161616161616,
+      "eval_loss": 1.8126838207244873,
+      "eval_precision_macro": 0.10915750915750914,
+      "eval_precision_micro": 0.26515151515151514,
+      "eval_precision_weighted": 0.13205128205128203,
+      "eval_recall_macro": 0.19727891156462585,
+      "eval_recall_micro": 0.26515151515151514,
+      "eval_recall_weighted": 0.26515151515151514,
+      "eval_runtime": 2.3137,
+      "eval_samples_per_second": 57.052,
+      "eval_steps_per_second": 7.348,
+      "step": 116
+    },
+    {
+      "epoch": 4.068965517241379,
+      "grad_norm": 2.673126220703125,
+      "learning_rate": 9.007633587786259e-05,
+      "loss": 1.7886,
+      "step": 118
+    },
+    {
+      "epoch": 4.137931034482759,
+      "grad_norm": 1.7165743112564087,
+      "learning_rate": 9.160305343511451e-05,
+      "loss": 1.7106,
+      "step": 120
+    },
+    {
+      "epoch": 4.206896551724138,
+      "grad_norm": 1.7721195220947266,
+      "learning_rate": 9.312977099236642e-05,
+      "loss": 1.7189,
+      "step": 122
+    },
+    {
+      "epoch": 4.275862068965517,
+      "grad_norm": 2.3345961570739746,
+      "learning_rate": 9.465648854961833e-05,
+      "loss": 1.6691,
+      "step": 124
+    },
+    {
+      "epoch": 4.344827586206897,
+      "grad_norm": 4.940219402313232,
+      "learning_rate": 9.618320610687024e-05,
+      "loss": 1.661,
+      "step": 126
+    },
+    {
+      "epoch": 4.413793103448276,
+      "grad_norm": 1.7643016576766968,
+      "learning_rate": 9.770992366412214e-05,
+      "loss": 1.6656,
+      "step": 128
+    },
+    {
+      "epoch": 4.482758620689655,
+      "grad_norm": 2.4930827617645264,
+      "learning_rate": 9.923664122137405e-05,
+      "loss": 1.7436,
+      "step": 130
+    },
+    {
+      "epoch": 4.551724137931035,
+      "grad_norm": 3.006988763809204,
+      "learning_rate": 9.991482112436117e-05,
+      "loss": 1.7548,
+      "step": 132
+    },
+    {
+      "epoch": 4.620689655172414,
+      "grad_norm": 2.7570700645446777,
+      "learning_rate": 9.974446337308349e-05,
+      "loss": 1.8445,
+      "step": 134
+    },
+    {
+      "epoch": 4.689655172413794,
+      "grad_norm": 3.1985325813293457,
+      "learning_rate": 9.957410562180581e-05,
+      "loss": 1.837,
+      "step": 136
+    },
+    {
+      "epoch": 4.758620689655173,
+      "grad_norm": 1.967631220817566,
+      "learning_rate": 9.940374787052812e-05,
+      "loss": 1.7548,
+      "step": 138
+    },
+    {
+      "epoch": 4.827586206896552,
+      "grad_norm": 2.0813746452331543,
+      "learning_rate": 9.923339011925044e-05,
+      "loss": 1.7141,
+      "step": 140
+    },
+    {
+      "epoch": 4.896551724137931,
+      "grad_norm": 1.9808541536331177,
+      "learning_rate": 9.906303236797275e-05,
+      "loss": 1.6873,
+      "step": 142
+    },
+    {
+      "epoch": 4.9655172413793105,
+      "grad_norm": 1.347161889076233,
+      "learning_rate": 9.889267461669507e-05,
+      "loss": 1.7113,
+      "step": 144
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.3484848484848485,
+      "eval_f1_macro": 0.2133114591830188,
+      "eval_f1_micro": 0.3484848484848485,
+      "eval_f1_weighted": 0.2592243730692438,
+      "eval_loss": 1.6978838443756104,
+      "eval_precision_macro": 0.31892757102841135,
+      "eval_precision_micro": 0.3484848484848485,
+      "eval_precision_weighted": 0.363105848399966,
+      "eval_recall_macro": 0.28217687074829934,
+      "eval_recall_micro": 0.3484848484848485,
+      "eval_recall_weighted": 0.3484848484848485,
+      "eval_runtime": 2.3205,
+      "eval_samples_per_second": 56.883,
+      "eval_steps_per_second": 7.326,
+      "step": 145
+    },
+    {
+      "epoch": 5.0344827586206895,
+      "grad_norm": 2.642209529876709,
+      "learning_rate": 9.872231686541739e-05,
+      "loss": 1.9708,
+      "step": 146
+    },
+    {
+      "epoch": 5.103448275862069,
+      "grad_norm": 2.9190890789031982,
+      "learning_rate": 9.85519591141397e-05,
+      "loss": 1.6693,
+      "step": 148
+    },
+    {
+      "epoch": 5.172413793103448,
+      "grad_norm": 2.941225290298462,
+      "learning_rate": 9.838160136286202e-05,
+      "loss": 1.6507,
+      "step": 150
+    },
+    {
+      "epoch": 5.241379310344827,
+      "grad_norm": 2.5989174842834473,
+      "learning_rate": 9.821124361158434e-05,
+      "loss": 1.4749,
+      "step": 152
+    },
+    {
+      "epoch": 5.310344827586207,
+      "grad_norm": 2.6868557929992676,
+      "learning_rate": 9.804088586030665e-05,
+      "loss": 1.5171,
+      "step": 154
+    },
+    {
+      "epoch": 5.379310344827586,
+      "grad_norm": 2.383096218109131,
+      "learning_rate": 9.787052810902897e-05,
+      "loss": 1.5249,
+      "step": 156
+    },
+    {
+      "epoch": 5.448275862068965,
+      "grad_norm": 2.130948543548584,
+      "learning_rate": 9.770017035775129e-05,
+      "loss": 1.5954,
+      "step": 158
+    },
+    {
+      "epoch": 5.517241379310345,
+      "grad_norm": 3.0674099922180176,
+      "learning_rate": 9.75298126064736e-05,
+      "loss": 1.4673,
+      "step": 160
+    },
+    {
+      "epoch": 5.586206896551724,
+      "grad_norm": 2.2742505073547363,
+      "learning_rate": 9.735945485519592e-05,
+      "loss": 1.6837,
+      "step": 162
+    },
+    {
+      "epoch": 5.655172413793103,
+      "grad_norm": 2.3131611347198486,
+      "learning_rate": 9.718909710391823e-05,
+      "loss": 1.4481,
+      "step": 164
+    },
+    {
+      "epoch": 5.724137931034483,
+      "grad_norm": 2.722858190536499,
+      "learning_rate": 9.701873935264055e-05,
+      "loss": 1.4517,
+      "step": 166
+    },
+    {
+      "epoch": 5.793103448275862,
+      "grad_norm": 2.8864994049072266,
+      "learning_rate": 9.684838160136287e-05,
+      "loss": 1.4883,
+      "step": 168
+    },
+    {
+      "epoch": 5.862068965517241,
+      "grad_norm": 3.632519483566284,
+      "learning_rate": 9.667802385008518e-05,
+      "loss": 1.4962,
+      "step": 170
+    },
+    {
+      "epoch": 5.931034482758621,
+      "grad_norm": 3.605977773666382,
+      "learning_rate": 9.65076660988075e-05,
+      "loss": 1.6613,
+      "step": 172
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 2.918729305267334,
+      "learning_rate": 9.633730834752982e-05,
+      "loss": 1.6459,
+      "step": 174
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.3939393939393939,
+      "eval_f1_macro": 0.2713789610398824,
+      "eval_f1_micro": 0.3939393939393939,
+      "eval_f1_weighted": 0.3224658104540408,
+      "eval_loss": 1.5577471256256104,
+      "eval_precision_macro": 0.4295932138037401,
+      "eval_precision_micro": 0.3939393939393939,
+      "eval_precision_weighted": 0.4530558622663886,
+      "eval_recall_macro": 0.3197883597883598,
+      "eval_recall_micro": 0.3939393939393939,
+      "eval_recall_weighted": 0.3939393939393939,
+      "eval_runtime": 2.3137,
+      "eval_samples_per_second": 57.052,
+      "eval_steps_per_second": 7.348,
+      "step": 174
+    },
+    {
+      "epoch": 6.068965517241379,
+      "grad_norm": 3.8798978328704834,
+      "learning_rate": 9.616695059625213e-05,
+      "loss": 1.349,
+      "step": 176
+    },
+    {
+      "epoch": 6.137931034482759,
+      "grad_norm": 2.2578699588775635,
+      "learning_rate": 9.599659284497445e-05,
+      "loss": 1.3423,
+      "step": 178
+    },
+    {
+      "epoch": 6.206896551724138,
+      "grad_norm": 3.3518385887145996,
+      "learning_rate": 9.582623509369678e-05,
+      "loss": 1.4951,
+      "step": 180
+    },
+    {
+      "epoch": 6.275862068965517,
+      "grad_norm": 2.808967351913452,
+      "learning_rate": 9.565587734241908e-05,
+      "loss": 1.601,
+      "step": 182
+    },
+    {
+      "epoch": 6.344827586206897,
+      "grad_norm": 2.637211322784424,
+      "learning_rate": 9.54855195911414e-05,
+      "loss": 1.5365,
+      "step": 184
+    },
+    {
+      "epoch": 6.413793103448276,
+      "grad_norm": 3.144742727279663,
+      "learning_rate": 9.531516183986373e-05,
+      "loss": 1.3149,
+      "step": 186
+    },
+    {
+      "epoch": 6.482758620689655,
+      "grad_norm": 3.037153482437134,
+      "learning_rate": 9.514480408858603e-05,
+      "loss": 1.3832,
+      "step": 188
+    },
+    {
+      "epoch": 6.551724137931035,
+      "grad_norm": 2.0758683681488037,
+      "learning_rate": 9.497444633730836e-05,
+      "loss": 1.3602,
+      "step": 190
+    },
+    {
+      "epoch": 6.620689655172414,
+      "grad_norm": 2.5833723545074463,
+      "learning_rate": 9.480408858603066e-05,
+      "loss": 1.4422,
+      "step": 192
+    },
+    {
+      "epoch": 6.689655172413794,
+      "grad_norm": 2.2444944381713867,
+      "learning_rate": 9.463373083475299e-05,
+      "loss": 1.3833,
+      "step": 194
+    },
+    {
+      "epoch": 6.758620689655173,
+      "grad_norm": 2.526270627975464,
+      "learning_rate": 9.446337308347531e-05,
+      "loss": 1.6393,
+      "step": 196
+    },
+    {
+      "epoch": 6.827586206896552,
+      "grad_norm": 2.5092673301696777,
+      "learning_rate": 9.429301533219762e-05,
+      "loss": 1.3183,
+      "step": 198
+    },
+    {
+      "epoch": 6.896551724137931,
+      "grad_norm": 2.4997923374176025,
+      "learning_rate": 9.412265758091994e-05,
+      "loss": 1.1796,
+      "step": 200
+    },
+    {
+      "epoch": 6.9655172413793105,
+      "grad_norm": 2.6789186000823975,
+      "learning_rate": 9.395229982964226e-05,
+      "loss": 1.4829,
+      "step": 202
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.5227272727272727,
+      "eval_f1_macro": 0.4068715903691273,
+      "eval_f1_micro": 0.5227272727272727,
+      "eval_f1_weighted": 0.46112149344907966,
+      "eval_loss": 1.381365180015564,
+      "eval_precision_macro": 0.37857508594835787,
+      "eval_precision_micro": 0.5227272727272727,
+      "eval_precision_weighted": 0.42157170786203046,
+      "eval_recall_macro": 0.4510506424792139,
+      "eval_recall_micro": 0.5227272727272727,
+      "eval_recall_weighted": 0.5227272727272727,
+      "eval_runtime": 2.3191,
+      "eval_samples_per_second": 56.919,
+      "eval_steps_per_second": 7.331,
+      "step": 203
+    },
+    {
+      "epoch": 7.0344827586206895,
+      "grad_norm": 2.8184642791748047,
+      "learning_rate": 9.378194207836457e-05,
+      "loss": 1.0758,
+      "step": 204
+    },
+    {
+      "epoch": 7.103448275862069,
+      "grad_norm": 3.6033294200897217,
+      "learning_rate": 9.361158432708689e-05,
+      "loss": 1.2313,
+      "step": 206
+    },
+    {
+      "epoch": 7.172413793103448,
+      "grad_norm": 3.935368299484253,
+      "learning_rate": 9.344122657580921e-05,
+      "loss": 1.5238,
+      "step": 208
+    },
+    {
+      "epoch": 7.241379310344827,
+      "grad_norm": 2.5644354820251465,
+      "learning_rate": 9.327086882453152e-05,
+      "loss": 1.4117,
+      "step": 210
+    },
+    {
+      "epoch": 7.310344827586207,
+      "grad_norm": 3.393187999725342,
+      "learning_rate": 9.310051107325384e-05,
+      "loss": 1.5217,
+      "step": 212
+    },
+    {
+      "epoch": 7.379310344827586,
+      "grad_norm": 2.552579879760742,
+      "learning_rate": 9.293015332197616e-05,
+      "loss": 1.145,
+      "step": 214
+    },
+    {
+      "epoch": 7.448275862068965,
+      "grad_norm": 2.249608039855957,
+      "learning_rate": 9.275979557069847e-05,
+      "loss": 1.3357,
+      "step": 216
+    },
+    {
+      "epoch": 7.517241379310345,
+      "grad_norm": 2.600088357925415,
+      "learning_rate": 9.258943781942079e-05,
+      "loss": 1.1833,
+      "step": 218
+    },
+    {
+      "epoch": 7.586206896551724,
+      "grad_norm": 2.311519145965576,
+      "learning_rate": 9.24190800681431e-05,
+      "loss": 1.0488,
+      "step": 220
+    },
+    {
+      "epoch": 7.655172413793103,
+      "grad_norm": 2.6178081035614014,
+      "learning_rate": 9.224872231686542e-05,
+      "loss": 1.2106,
+      "step": 222
+    },
+    {
+      "epoch": 7.724137931034483,
+      "grad_norm": 3.762129545211792,
+      "learning_rate": 9.207836456558774e-05,
+      "loss": 1.1438,
+      "step": 224
+    },
+    {
+      "epoch": 7.793103448275862,
+      "grad_norm": 3.4986026287078857,
+      "learning_rate": 9.190800681431005e-05,
+      "loss": 1.4612,
+      "step": 226
+    },
+    {
+      "epoch": 7.862068965517241,
+      "grad_norm": 2.58974552154541,
+      "learning_rate": 9.173764906303237e-05,
+      "loss": 0.9742,
+      "step": 228
+    },
+    {
+      "epoch": 7.931034482758621,
+      "grad_norm": 2.3729593753814697,
+      "learning_rate": 9.156729131175469e-05,
+      "loss": 1.3228,
+      "step": 230
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 2.6781985759735107,
+      "learning_rate": 9.1396933560477e-05,
+      "loss": 1.2847,
+      "step": 232
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_f1_macro": 0.3675035848097526,
+      "eval_f1_micro": 0.45454545454545453,
+      "eval_f1_weighted": 0.4175505688962959,
+      "eval_loss": 1.3783015012741089,
+      "eval_precision_macro": 0.49916764667595237,
+      "eval_precision_micro": 0.45454545454545453,
+      "eval_precision_weighted": 0.570213988501515,
+      "eval_recall_macro": 0.4079591836734694,
+      "eval_recall_micro": 0.45454545454545453,
+      "eval_recall_weighted": 0.45454545454545453,
+      "eval_runtime": 2.4636,
+      "eval_samples_per_second": 53.581,
+      "eval_steps_per_second": 6.901,
+      "step": 232
+    },
+    {
+      "epoch": 8.068965517241379,
+      "grad_norm": 3.757275104522705,
+      "learning_rate": 9.122657580919932e-05,
+      "loss": 1.0559,
+      "step": 234
+    },
+    {
+      "epoch": 8.137931034482758,
+      "grad_norm": 2.25482439994812,
+      "learning_rate": 9.105621805792164e-05,
+      "loss": 1.181,
+      "step": 236
+    },
+    {
+      "epoch": 8.206896551724139,
+      "grad_norm": 3.7089550495147705,
+      "learning_rate": 9.088586030664395e-05,
+      "loss": 1.3496,
+      "step": 238
+    },
+    {
+      "epoch": 8.275862068965518,
+      "grad_norm": 2.729207754135132,
+      "learning_rate": 9.071550255536627e-05,
+      "loss": 1.1984,
+      "step": 240
+    },
+    {
+      "epoch": 8.344827586206897,
+      "grad_norm": 2.9847185611724854,
+      "learning_rate": 9.054514480408858e-05,
+      "loss": 1.1024,
+      "step": 242
+    },
+    {
+      "epoch": 8.413793103448276,
+      "grad_norm": 3.846353530883789,
+      "learning_rate": 9.03747870528109e-05,
+      "loss": 1.1308,
+      "step": 244
+    },
+    {
+      "epoch": 8.482758620689655,
+      "grad_norm": 3.132477283477783,
+      "learning_rate": 9.020442930153323e-05,
+      "loss": 1.2066,
+      "step": 246
+    },
+    {
+      "epoch": 8.551724137931034,
+      "grad_norm": 3.4364778995513916,
+      "learning_rate": 9.003407155025553e-05,
+      "loss": 1.1284,
+      "step": 248
+    },
+    {
+      "epoch": 8.620689655172415,
+      "grad_norm": 3.002615451812744,
+      "learning_rate": 8.986371379897785e-05,
+      "loss": 1.0977,
+      "step": 250
+    },
+    {
+      "epoch": 8.689655172413794,
+      "grad_norm": 2.997836112976074,
+      "learning_rate": 8.969335604770018e-05,
+      "loss": 0.977,
+      "step": 252
+    },
+    {
+      "epoch": 8.758620689655173,
+      "grad_norm": 4.079631805419922,
+      "learning_rate": 8.952299829642248e-05,
+      "loss": 1.3108,
+      "step": 254
+    },
+    {
+      "epoch": 8.827586206896552,
+      "grad_norm": 4.065947532653809,
+      "learning_rate": 8.93526405451448e-05,
+      "loss": 1.1677,
+      "step": 256
+    },
+    {
+      "epoch": 8.89655172413793,
+      "grad_norm": 4.696402549743652,
+      "learning_rate": 8.918228279386713e-05,
+      "loss": 1.1538,
+      "step": 258
+    },
+    {
+      "epoch": 8.96551724137931,
+      "grad_norm": 2.8291256427764893,
+      "learning_rate": 8.901192504258944e-05,
+      "loss": 0.7746,
+      "step": 260
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.5757575757575758,
+      "eval_f1_macro": 0.4579442109215706,
+      "eval_f1_micro": 0.5757575757575758,
+      "eval_f1_weighted": 0.5297653032901569,
+      "eval_loss": 1.1536388397216797,
+      "eval_precision_macro": 0.5301086941953815,
+      "eval_precision_micro": 0.5757575757575758,
+      "eval_precision_weighted": 0.589626978576458,
+      "eval_recall_macro": 0.4852532123960695,
+      "eval_recall_micro": 0.5757575757575758,
+      "eval_recall_weighted": 0.5757575757575758,
+      "eval_runtime": 2.3555,
+      "eval_samples_per_second": 56.04,
+      "eval_steps_per_second": 7.217,
+      "step": 261
+    },
+    {
+      "epoch": 9.03448275862069,
+      "grad_norm": 3.098844289779663,
+      "learning_rate": 8.884156729131176e-05,
+      "loss": 1.0718,
+      "step": 262
+    },
+    {
+      "epoch": 9.10344827586207,
+      "grad_norm": 2.9349565505981445,
+      "learning_rate": 8.867120954003408e-05,
+      "loss": 0.9468,
+      "step": 264
+    },
+    {
+      "epoch": 9.172413793103448,
+      "grad_norm": 4.122389316558838,
+      "learning_rate": 8.850085178875639e-05,
+      "loss": 1.1746,
+      "step": 266
+    },
+    {
+      "epoch": 9.241379310344827,
+      "grad_norm": 2.8826913833618164,
+      "learning_rate": 8.833049403747871e-05,
+      "loss": 0.8535,
+      "step": 268
+    },
+    {
+      "epoch": 9.310344827586206,
+      "grad_norm": 3.3379621505737305,
+      "learning_rate": 8.816013628620102e-05,
+      "loss": 0.9954,
+      "step": 270
+    },
+    {
+      "epoch": 9.379310344827585,
+      "grad_norm": 5.267432689666748,
+      "learning_rate": 8.798977853492334e-05,
+      "loss": 0.9242,
+      "step": 272
+    },
+    {
+      "epoch": 9.448275862068966,
+      "grad_norm": 3.121428966522217,
+      "learning_rate": 8.781942078364566e-05,
+      "loss": 1.0053,
+      "step": 274
+    },
+    {
+      "epoch": 9.517241379310345,
+      "grad_norm": 3.526177167892456,
+      "learning_rate": 8.764906303236797e-05,
+      "loss": 0.8233,
+      "step": 276
+    },
+    {
+      "epoch": 9.586206896551724,
+      "grad_norm": 2.7245631217956543,
+      "learning_rate": 8.747870528109029e-05,
+      "loss": 0.9055,
+      "step": 278
+    },
+    {
+      "epoch": 9.655172413793103,
+      "grad_norm": 1.7660030126571655,
+      "learning_rate": 8.730834752981261e-05,
+      "loss": 0.5839,
+      "step": 280
+    },
+    {
+      "epoch": 9.724137931034482,
+      "grad_norm": 3.6391730308532715,
+      "learning_rate": 8.713798977853492e-05,
+      "loss": 1.1208,
+      "step": 282
+    },
+    {
+      "epoch": 9.793103448275861,
+      "grad_norm": 4.08744478225708,
+      "learning_rate": 8.696763202725724e-05,
+      "loss": 0.8893,
+      "step": 284
+    },
+    {
+      "epoch": 9.862068965517242,
+      "grad_norm": 4.032042503356934,
+      "learning_rate": 8.679727427597956e-05,
+      "loss": 0.973,
+      "step": 286
+    },
+    {
+      "epoch": 9.931034482758621,
+      "grad_norm": 5.962705135345459,
+      "learning_rate": 8.662691652470187e-05,
+      "loss": 1.0092,
+      "step": 288
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 5.076080799102783,
+      "learning_rate": 8.645655877342419e-05,
+      "loss": 1.0172,
+      "step": 290
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.5909090909090909,
+      "eval_f1_macro": 0.47002148227712137,
+      "eval_f1_micro": 0.5909090909090909,
+      "eval_f1_weighted": 0.5365472393103972,
+      "eval_loss": 1.221096396446228,
+      "eval_precision_macro": 0.572196223316913,
+      "eval_precision_micro": 0.5909090909090909,
+      "eval_precision_weighted": 0.6399457941483804,
+      "eval_recall_macro": 0.518238851095994,
+      "eval_recall_micro": 0.5909090909090909,
+      "eval_recall_weighted": 0.5909090909090909,
+      "eval_runtime": 2.3184,
+      "eval_samples_per_second": 56.936,
+      "eval_steps_per_second": 7.333,
+      "step": 290
+    },
+    {
+      "epoch": 10.068965517241379,
+      "grad_norm": 4.521097183227539,
+      "learning_rate": 8.628620102214651e-05,
+      "loss": 0.9744,
+      "step": 292
+    },
+    {
+      "epoch": 10.137931034482758,
+      "grad_norm": 2.6411921977996826,
+      "learning_rate": 8.611584327086882e-05,
+      "loss": 0.7116,
+      "step": 294
+    },
+    {
+      "epoch": 10.206896551724139,
+      "grad_norm": 4.236311912536621,
+      "learning_rate": 8.594548551959114e-05,
+      "loss": 1.016,
+      "step": 296
+    },
+    {
+      "epoch": 10.275862068965518,
+      "grad_norm": 4.348973751068115,
+      "learning_rate": 8.577512776831345e-05,
+      "loss": 0.7783,
+      "step": 298
+    },
+    {
+      "epoch": 10.344827586206897,
+      "grad_norm": 2.6386284828186035,
+      "learning_rate": 8.560477001703579e-05,
+      "loss": 1.0324,
+      "step": 300
+    },
+    {
+      "epoch": 10.413793103448276,
+      "grad_norm": 3.886791706085205,
+      "learning_rate": 8.543441226575811e-05,
+      "loss": 0.8633,
+      "step": 302
+    },
+    {
+      "epoch": 10.482758620689655,
+      "grad_norm": 5.126855373382568,
+      "learning_rate": 8.526405451448042e-05,
+      "loss": 0.9599,
+      "step": 304
+    },
+    {
+      "epoch": 10.551724137931034,
+      "grad_norm": 3.6911346912384033,
+      "learning_rate": 8.509369676320274e-05,
+      "loss": 1.0509,
+      "step": 306
+    },
+    {
+      "epoch": 10.620689655172415,
+      "grad_norm": 3.802255392074585,
+      "learning_rate": 8.492333901192505e-05,
+      "loss": 0.8213,
+      "step": 308
+    },
+    {
+      "epoch": 10.689655172413794,
+      "grad_norm": 5.074642181396484,
+      "learning_rate": 8.475298126064737e-05,
+      "loss": 1.0766,
+      "step": 310
+    },
+    {
+      "epoch": 10.758620689655173,
+      "grad_norm": 4.2655720710754395,
+      "learning_rate": 8.458262350936969e-05,
+      "loss": 0.9113,
+      "step": 312
+    },
+    {
+      "epoch": 10.827586206896552,
+      "grad_norm": 3.382858991622925,
+      "learning_rate": 8.4412265758092e-05,
+      "loss": 0.7092,
+      "step": 314
+    },
+    {
+      "epoch": 10.89655172413793,
+      "grad_norm": 4.771972179412842,
+      "learning_rate": 8.424190800681432e-05,
+      "loss": 0.547,
+      "step": 316
+    },
+    {
+      "epoch": 10.96551724137931,
+      "grad_norm": 4.0019378662109375,
+      "learning_rate": 8.407155025553664e-05,
+      "loss": 0.7865,
+      "step": 318
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.6136363636363636,
+      "eval_f1_macro": 0.5281776060910356,
+      "eval_f1_micro": 0.6136363636363636,
+      "eval_f1_weighted": 0.596118348928358,
+      "eval_loss": 1.1357340812683105,
+      "eval_precision_macro": 0.5342261904761905,
+      "eval_precision_micro": 0.6136363636363636,
+      "eval_precision_weighted": 0.6009311868686869,
+      "eval_recall_macro": 0.5431594860166289,
+      "eval_recall_micro": 0.6136363636363636,
+      "eval_recall_weighted": 0.6136363636363636,
+      "eval_runtime": 2.3136,
+      "eval_samples_per_second": 57.054,
+      "eval_steps_per_second": 7.348,
+      "step": 319
+    },
+    {
+      "epoch": 11.03448275862069,
+      "grad_norm": 3.1186282634735107,
+      "learning_rate": 8.390119250425895e-05,
+      "loss": 0.6251,
+      "step": 320
+    },
+    {
+      "epoch": 11.10344827586207,
+      "grad_norm": 4.061969757080078,
+      "learning_rate": 8.373083475298127e-05,
+      "loss": 0.709,
+      "step": 322
+    },
+    {
+      "epoch": 11.172413793103448,
+      "grad_norm": 3.080505132675171,
+      "learning_rate": 8.356047700170359e-05,
+      "loss": 0.6536,
+      "step": 324
+    },
+    {
+      "epoch": 11.241379310344827,
+      "grad_norm": 4.702264785766602,
+      "learning_rate": 8.33901192504259e-05,
+      "loss": 0.764,
+      "step": 326
+    },
+    {
+      "epoch": 11.310344827586206,
+      "grad_norm": 3.9699347019195557,
+      "learning_rate": 8.321976149914822e-05,
+      "loss": 0.7632,
+      "step": 328
+    },
+    {
+      "epoch": 11.379310344827585,
+      "grad_norm": 3.679388999938965,
+      "learning_rate": 8.304940374787054e-05,
+      "loss": 1.0315,
+      "step": 330
+    },
+    {
+      "epoch": 11.448275862068966,
+      "grad_norm": 3.994912624359131,
+      "learning_rate": 8.287904599659285e-05,
+      "loss": 1.0187,
+      "step": 332
+    },
+    {
+      "epoch": 11.517241379310345,
+      "grad_norm": 5.398199558258057,
+      "learning_rate": 8.270868824531517e-05,
+      "loss": 0.5242,
+      "step": 334
+    },
+    {
+      "epoch": 11.586206896551724,
+      "grad_norm": 3.0399880409240723,
+      "learning_rate": 8.253833049403748e-05,
+      "loss": 0.7834,
+      "step": 336
+    },
+    {
+      "epoch": 11.655172413793103,
+      "grad_norm": 3.8799502849578857,
+      "learning_rate": 8.23679727427598e-05,
+      "loss": 0.7336,
+      "step": 338
+    },
+    {
+      "epoch": 11.724137931034482,
+      "grad_norm": 2.1970696449279785,
+      "learning_rate": 8.219761499148212e-05,
+      "loss": 0.6121,
+      "step": 340
+    },
+    {
+      "epoch": 11.793103448275861,
+      "grad_norm": 4.472761631011963,
+      "learning_rate": 8.202725724020443e-05,
+      "loss": 0.8251,
+      "step": 342
+    },
+    {
+      "epoch": 11.862068965517242,
+      "grad_norm": 2.3058536052703857,
+      "learning_rate": 8.185689948892675e-05,
+      "loss": 0.7287,
+      "step": 344
+    },
+    {
+      "epoch": 11.931034482758621,
+      "grad_norm": 4.462862014770508,
+      "learning_rate": 8.168654173764907e-05,
+      "loss": 0.7826,
+      "step": 346
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 4.2772440910339355,
+      "learning_rate": 8.151618398637138e-05,
+      "loss": 0.8335,
+      "step": 348
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.6060606060606061,
+      "eval_f1_macro": 0.5314848082363927,
+      "eval_f1_micro": 0.6060606060606061,
+      "eval_f1_weighted": 0.6017190744361463,
+      "eval_loss": 1.152970552444458,
+      "eval_precision_macro": 0.536459399813437,
+      "eval_precision_micro": 0.6060606060606061,
+      "eval_precision_weighted": 0.6208833437094308,
+      "eval_recall_macro": 0.5489417989417988,
+      "eval_recall_micro": 0.6060606060606061,
+      "eval_recall_weighted": 0.6060606060606061,
+      "eval_runtime": 2.3168,
+      "eval_samples_per_second": 56.976,
+      "eval_steps_per_second": 7.338,
+      "step": 348
+    },
+    {
+      "epoch": 12.068965517241379,
+      "grad_norm": 3.217472791671753,
+      "learning_rate": 8.13458262350937e-05,
+      "loss": 0.6884,
+      "step": 350
+    },
+    {
+      "epoch": 12.137931034482758,
+      "grad_norm": 3.676238536834717,
+      "learning_rate": 8.117546848381603e-05,
+      "loss": 0.6886,
+      "step": 352
+    },
+    {
+      "epoch": 12.206896551724139,
+      "grad_norm": 3.851684093475342,
+      "learning_rate": 8.100511073253833e-05,
+      "loss": 0.8202,
+      "step": 354
+    },
+    {
+      "epoch": 12.275862068965518,
+      "grad_norm": 3.2188940048217773,
+      "learning_rate": 8.083475298126066e-05,
+      "loss": 0.6026,
+      "step": 356
+    },
+    {
+      "epoch": 12.344827586206897,
+      "grad_norm": 3.6800220012664795,
+      "learning_rate": 8.066439522998296e-05,
+      "loss": 0.5762,
+      "step": 358
+    },
+    {
+      "epoch": 12.413793103448276,
+      "grad_norm": 4.217052459716797,
+      "learning_rate": 8.049403747870528e-05,
+      "loss": 0.824,
+      "step": 360
+    },
+    {
+      "epoch": 12.482758620689655,
+      "grad_norm": 4.007477760314941,
+      "learning_rate": 8.03236797274276e-05,
+      "loss": 0.5009,
+      "step": 362
+    },
+    {
+      "epoch": 12.551724137931034,
+      "grad_norm": 3.780930995941162,
+      "learning_rate": 8.015332197614991e-05,
+      "loss": 0.6427,
+      "step": 364
+    },
+    {
+      "epoch": 12.620689655172415,
+      "grad_norm": 3.340984582901001,
+      "learning_rate": 7.998296422487224e-05,
+      "loss": 0.6936,
+      "step": 366
+    },
+    {
+      "epoch": 12.689655172413794,
+      "grad_norm": 6.2322258949279785,
+      "learning_rate": 7.981260647359456e-05,
+      "loss": 0.7349,
+      "step": 368
+    },
+    {
+      "epoch": 12.758620689655173,
+      "grad_norm": 4.197908878326416,
+      "learning_rate": 7.964224872231687e-05,
+      "loss": 1.0689,
+      "step": 370
+    },
+    {
+      "epoch": 12.827586206896552,
+      "grad_norm": 3.51385498046875,
+      "learning_rate": 7.947189097103919e-05,
+      "loss": 0.4522,
+      "step": 372
+    },
+    {
+      "epoch": 12.89655172413793,
+      "grad_norm": 4.907188415527344,
+      "learning_rate": 7.930153321976151e-05,
+      "loss": 0.6371,
+      "step": 374
+    },
+    {
+      "epoch": 12.96551724137931,
+      "grad_norm": 4.427891731262207,
+      "learning_rate": 7.913117546848382e-05,
+      "loss": 0.6959,
+      "step": 376
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.6666666666666666,
+      "eval_f1_macro": 0.5637755814516842,
+      "eval_f1_micro": 0.6666666666666666,
+      "eval_f1_weighted": 0.6451496154313497,
+      "eval_loss": 1.1307237148284912,
+      "eval_precision_macro": 0.5912295422147639,
+      "eval_precision_micro": 0.6666666666666666,
+      "eval_precision_weighted": 0.6614738231117541,
+      "eval_recall_macro": 0.577278911564626,
+      "eval_recall_micro": 0.6666666666666666,
+      "eval_recall_weighted": 0.6666666666666666,
+      "eval_runtime": 2.3277,
+      "eval_samples_per_second": 56.707,
+      "eval_steps_per_second": 7.303,
+      "step": 377
+    },
+    {
+      "epoch": 13.03448275862069,
+      "grad_norm": 1.9292876720428467,
+      "learning_rate": 7.896081771720614e-05,
+      "loss": 0.5996,
+      "step": 378
+    },
+    {
+      "epoch": 13.10344827586207,
+      "grad_norm": 4.09467887878418,
+      "learning_rate": 7.879045996592846e-05,
+      "loss": 0.5993,
+      "step": 380
+    },
+    {
+      "epoch": 13.172413793103448,
+      "grad_norm": 2.3077073097229004,
+      "learning_rate": 7.862010221465077e-05,
+      "loss": 0.6907,
+      "step": 382
+    },
+    {
+      "epoch": 13.241379310344827,
+      "grad_norm": 4.537380218505859,
+      "learning_rate": 7.844974446337309e-05,
+      "loss": 0.6659,
+      "step": 384
+    },
+    {
+      "epoch": 13.310344827586206,
+      "grad_norm": 2.6575474739074707,
+      "learning_rate": 7.82793867120954e-05,
+      "loss": 0.4833,
+      "step": 386
+    },
+    {
+      "epoch": 13.379310344827585,
+      "grad_norm": 4.1183929443359375,
+      "learning_rate": 7.810902896081772e-05,
+      "loss": 0.6274,
+      "step": 388
+    },
+    {
+      "epoch": 13.448275862068966,
+      "grad_norm": 2.1658565998077393,
+      "learning_rate": 7.793867120954004e-05,
+      "loss": 0.6313,
+      "step": 390
+    },
+    {
+      "epoch": 13.517241379310345,
+      "grad_norm": 2.3575236797332764,
+      "learning_rate": 7.776831345826235e-05,
+      "loss": 0.5313,
+      "step": 392
+    },
+    {
+      "epoch": 13.586206896551724,
+      "grad_norm": 2.720118999481201,
+      "learning_rate": 7.759795570698467e-05,
+      "loss": 0.3753,
+      "step": 394
+    },
+    {
+      "epoch": 13.655172413793103,
+      "grad_norm": 4.6414475440979,
+      "learning_rate": 7.742759795570699e-05,
+      "loss": 0.8126,
+      "step": 396
+    },
+    {
+      "epoch": 13.724137931034482,
+      "grad_norm": 4.100179195404053,
+      "learning_rate": 7.72572402044293e-05,
+      "loss": 0.6283,
+      "step": 398
+    },
+    {
+      "epoch": 13.793103448275861,
+      "grad_norm": 2.20033860206604,
+      "learning_rate": 7.708688245315162e-05,
+      "loss": 0.6954,
+      "step": 400
+    },
+    {
+      "epoch": 13.862068965517242,
+      "grad_norm": 4.777364253997803,
+      "learning_rate": 7.691652470187394e-05,
+      "loss": 0.6302,
+      "step": 402
+    },
+    {
+      "epoch": 13.931034482758621,
+      "grad_norm": 3.3690972328186035,
+      "learning_rate": 7.674616695059625e-05,
+      "loss": 0.8916,
+      "step": 404
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 5.667202472686768,
+      "learning_rate": 7.657580919931857e-05,
+      "loss": 0.5864,
+      "step": 406
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.5984848484848485,
+      "eval_f1_macro": 0.521144758673933,
+      "eval_f1_micro": 0.5984848484848485,
+      "eval_f1_weighted": 0.5893727222006053,
+      "eval_loss": 1.1956861019134521,
+      "eval_precision_macro": 0.5536575030457123,
+      "eval_precision_micro": 0.5984848484848485,
+      "eval_precision_weighted": 0.6274957584746239,
+      "eval_recall_macro": 0.5389191232048375,
+      "eval_recall_micro": 0.5984848484848485,
+      "eval_recall_weighted": 0.5984848484848485,
+      "eval_runtime": 2.3168,
+      "eval_samples_per_second": 56.975,
+      "eval_steps_per_second": 7.338,
+      "step": 406
+    },
+    {
+      "epoch": 14.068965517241379,
+      "grad_norm": 6.122737884521484,
+      "learning_rate": 7.64054514480409e-05,
+      "loss": 0.5025,
+      "step": 408
+    },
+    {
+      "epoch": 14.137931034482758,
+      "grad_norm": 2.439298629760742,
+      "learning_rate": 7.62350936967632e-05,
+      "loss": 0.4959,
+      "step": 410
+    },
+    {
+      "epoch": 14.206896551724139,
+      "grad_norm": 3.6299631595611572,
+      "learning_rate": 7.606473594548552e-05,
+      "loss": 0.5056,
+      "step": 412
+    },
+    {
+      "epoch": 14.275862068965518,
+      "grad_norm": 4.751211166381836,
+      "learning_rate": 7.589437819420783e-05,
+      "loss": 0.7691,
+      "step": 414
+    },
+    {
+      "epoch": 14.344827586206897,
+      "grad_norm": 4.157205104827881,
+      "learning_rate": 7.572402044293015e-05,
+      "loss": 0.5365,
+      "step": 416
+    },
+    {
+      "epoch": 14.413793103448276,
+      "grad_norm": 4.521823406219482,
+      "learning_rate": 7.555366269165248e-05,
+      "loss": 0.4214,
+      "step": 418
+    },
+    {
+      "epoch": 14.482758620689655,
+      "grad_norm": 1.9958875179290771,
+      "learning_rate": 7.538330494037478e-05,
+      "loss": 0.3496,
+      "step": 420
+    },
+    {
+      "epoch": 14.551724137931034,
+      "grad_norm": 3.086988925933838,
+      "learning_rate": 7.52129471890971e-05,
+      "loss": 0.6931,
+      "step": 422
+    },
+    {
+      "epoch": 14.620689655172415,
+      "grad_norm": 5.060652256011963,
+      "learning_rate": 7.504258943781943e-05,
+      "loss": 0.6347,
+      "step": 424
+    },
+    {
+      "epoch": 14.689655172413794,
+      "grad_norm": 2.613652229309082,
+      "learning_rate": 7.487223168654173e-05,
+      "loss": 0.5666,
+      "step": 426
+    },
+    {
+      "epoch": 14.758620689655173,
+      "grad_norm": 2.4513535499572754,
+      "learning_rate": 7.470187393526406e-05,
+      "loss": 0.4125,
+      "step": 428
+    },
+    {
+      "epoch": 14.827586206896552,
+      "grad_norm": 2.7038187980651855,
+      "learning_rate": 7.453151618398638e-05,
+      "loss": 0.4961,
+      "step": 430
+    },
+    {
+      "epoch": 14.89655172413793,
+      "grad_norm": 4.4557085037231445,
+      "learning_rate": 7.436115843270869e-05,
+      "loss": 0.4675,
+      "step": 432
+    },
+    {
+      "epoch": 14.96551724137931,
+      "grad_norm": 2.6344242095947266,
+      "learning_rate": 7.419080068143101e-05,
+      "loss": 0.6145,
+      "step": 434
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.7045454545454546,
+      "eval_f1_macro": 0.6086358317016884,
+      "eval_f1_micro": 0.7045454545454546,
+      "eval_f1_weighted": 0.6833139383291237,
+      "eval_loss": 0.995664656162262,
+      "eval_precision_macro": 0.6164373139373139,
+      "eval_precision_micro": 0.7045454545454546,
+      "eval_precision_weighted": 0.6790800235497206,
+      "eval_recall_macro": 0.6160166288737718,
+      "eval_recall_micro": 0.7045454545454546,
+      "eval_recall_weighted": 0.7045454545454546,
+      "eval_runtime": 2.3352,
+      "eval_samples_per_second": 56.526,
+      "eval_steps_per_second": 7.28,
+      "step": 435
+    },
+    {
+      "epoch": 15.03448275862069,
+      "grad_norm": 2.6183178424835205,
+      "learning_rate": 7.402044293015332e-05,
+      "loss": 0.5516,
+      "step": 436
+    },
+    {
+      "epoch": 15.10344827586207,
+      "grad_norm": 1.3235374689102173,
+      "learning_rate": 7.385008517887564e-05,
+      "loss": 0.4201,
+      "step": 438
+    },
+    {
+      "epoch": 15.172413793103448,
+      "grad_norm": 0.8775862455368042,
+      "learning_rate": 7.367972742759796e-05,
+      "loss": 0.3188,
+      "step": 440
+    },
+    {
+      "epoch": 15.241379310344827,
+      "grad_norm": 1.7227734327316284,
+      "learning_rate": 7.350936967632027e-05,
+      "loss": 0.3833,
+      "step": 442
+    },
+    {
+      "epoch": 15.310344827586206,
+      "grad_norm": 4.330628871917725,
+      "learning_rate": 7.333901192504259e-05,
+      "loss": 0.5839,
+      "step": 444
+    },
+    {
+      "epoch": 15.379310344827585,
+      "grad_norm": 3.0039358139038086,
+      "learning_rate": 7.316865417376491e-05,
+      "loss": 0.6361,
+      "step": 446
+    },
+    {
+      "epoch": 15.448275862068966,
+      "grad_norm": 3.6399760246276855,
+      "learning_rate": 7.299829642248722e-05,
+      "loss": 0.7401,
+      "step": 448
+    },
+    {
+      "epoch": 15.517241379310345,
+      "grad_norm": 4.385819435119629,
+      "learning_rate": 7.282793867120954e-05,
+      "loss": 0.3982,
+      "step": 450
+    },
+    {
+      "epoch": 15.586206896551724,
+      "grad_norm": 3.1038174629211426,
+      "learning_rate": 7.265758091993186e-05,
+      "loss": 0.5036,
+      "step": 452
+    },
+    {
+      "epoch": 15.655172413793103,
+      "grad_norm": 3.780447244644165,
+      "learning_rate": 7.248722316865417e-05,
+      "loss": 0.3955,
+      "step": 454
+    },
+    {
+      "epoch": 15.724137931034482,
+      "grad_norm": 3.7337963581085205,
+      "learning_rate": 7.231686541737649e-05,
+      "loss": 0.4673,
+      "step": 456
+    },
+    {
+      "epoch": 15.793103448275861,
+      "grad_norm": 2.2187936305999756,
+      "learning_rate": 7.214650766609881e-05,
+      "loss": 0.3166,
+      "step": 458
+    },
+    {
+      "epoch": 15.862068965517242,
+      "grad_norm": 5.6369476318359375,
+      "learning_rate": 7.197614991482112e-05,
+      "loss": 0.5263,
+      "step": 460
+    },
+    {
+      "epoch": 15.931034482758621,
+      "grad_norm": 2.084303855895996,
+      "learning_rate": 7.180579216354344e-05,
+      "loss": 0.3393,
+      "step": 462
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 4.383190631866455,
+      "learning_rate": 7.163543441226575e-05,
+      "loss": 0.5632,
+      "step": 464
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.5984848484848485,
+      "eval_f1_macro": 0.5112417598080162,
+      "eval_f1_micro": 0.5984848484848485,
+      "eval_f1_weighted": 0.5780904923018929,
+      "eval_loss": 1.230219841003418,
+      "eval_precision_macro": 0.5219010225654743,
+      "eval_precision_micro": 0.5984848484848485,
+      "eval_precision_weighted": 0.5853223415062738,
+      "eval_recall_macro": 0.5235525321239607,
+      "eval_recall_micro": 0.5984848484848485,
+      "eval_recall_weighted": 0.5984848484848485,
+      "eval_runtime": 2.3147,
+      "eval_samples_per_second": 57.027,
+      "eval_steps_per_second": 7.344,
+      "step": 464
+    },
+    {
+      "epoch": 16.06896551724138,
+      "grad_norm": 2.8568015098571777,
+      "learning_rate": 7.146507666098807e-05,
+      "loss": 0.453,
+      "step": 466
+    },
+    {
+      "epoch": 16.137931034482758,
+      "grad_norm": 3.751786231994629,
+      "learning_rate": 7.12947189097104e-05,
+      "loss": 0.4951,
+      "step": 468
+    },
+    {
+      "epoch": 16.20689655172414,
+      "grad_norm": 3.3349523544311523,
+      "learning_rate": 7.112436115843271e-05,
+      "loss": 0.3434,
+      "step": 470
+    },
+    {
+      "epoch": 16.275862068965516,
+      "grad_norm": 0.8290226459503174,
+      "learning_rate": 7.095400340715504e-05,
+      "loss": 0.2894,
+      "step": 472
+    },
+    {
+      "epoch": 16.344827586206897,
+      "grad_norm": 2.8990588188171387,
+      "learning_rate": 7.078364565587734e-05,
+      "loss": 0.3078,
+      "step": 474
+    },
+    {
+      "epoch": 16.413793103448278,
+      "grad_norm": 3.8644514083862305,
+      "learning_rate": 7.061328790459967e-05,
+      "loss": 0.5261,
+      "step": 476
+    },
+    {
+      "epoch": 16.482758620689655,
+      "grad_norm": 3.9147894382476807,
+      "learning_rate": 7.044293015332199e-05,
+      "loss": 0.4895,
+      "step": 478
+    },
+    {
+      "epoch": 16.551724137931036,
+      "grad_norm": 6.574154376983643,
+      "learning_rate": 7.02725724020443e-05,
+      "loss": 0.4676,
+      "step": 480
+    },
+    {
+      "epoch": 16.620689655172413,
+      "grad_norm": 2.001779079437256,
+      "learning_rate": 7.010221465076662e-05,
+      "loss": 0.379,
+      "step": 482
+    },
+    {
+      "epoch": 16.689655172413794,
+      "grad_norm": 3.2596888542175293,
+      "learning_rate": 6.993185689948894e-05,
+      "loss": 0.4082,
+      "step": 484
+    },
+    {
+      "epoch": 16.75862068965517,
+      "grad_norm": 3.3688106536865234,
+      "learning_rate": 6.976149914821125e-05,
+      "loss": 0.2854,
+      "step": 486
+    },
+    {
+      "epoch": 16.82758620689655,
+      "grad_norm": 4.847587585449219,
+      "learning_rate": 6.959114139693357e-05,
+      "loss": 0.465,
+      "step": 488
+    },
+    {
+      "epoch": 16.896551724137932,
+      "grad_norm": 1.7020337581634521,
+      "learning_rate": 6.942078364565589e-05,
+      "loss": 0.2685,
+      "step": 490
+    },
+    {
+      "epoch": 16.96551724137931,
+      "grad_norm": 4.529393672943115,
+      "learning_rate": 6.92504258943782e-05,
+      "loss": 0.3392,
+      "step": 492
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.6287878787878788,
+      "eval_f1_macro": 0.5334567769861888,
+      "eval_f1_micro": 0.6287878787878788,
+      "eval_f1_weighted": 0.6043366300496426,
+      "eval_loss": 1.192486047744751,
+      "eval_precision_macro": 0.5902809125181222,
+      "eval_precision_micro": 0.6287878787878788,
+      "eval_precision_weighted": 0.6434826072733183,
+      "eval_recall_macro": 0.535479969765684,
+      "eval_recall_micro": 0.6287878787878788,
+      "eval_recall_weighted": 0.6287878787878788,
+      "eval_runtime": 2.312,
+      "eval_samples_per_second": 57.094,
+      "eval_steps_per_second": 7.353,
+      "step": 493
+    },
+    {
+      "epoch": 17.03448275862069,
+      "grad_norm": 5.458866596221924,
+      "learning_rate": 6.908006814310052e-05,
+      "loss": 0.4056,
+      "step": 494
+    },
+    {
+      "epoch": 17.103448275862068,
+      "grad_norm": 4.587505340576172,
+      "learning_rate": 6.890971039182284e-05,
+      "loss": 0.2633,
+      "step": 496
+    },
+    {
+      "epoch": 17.17241379310345,
+      "grad_norm": 2.994865894317627,
+      "learning_rate": 6.873935264054515e-05,
+      "loss": 0.1985,
+      "step": 498
+    },
+    {
+      "epoch": 17.24137931034483,
+      "grad_norm": 2.954942464828491,
+      "learning_rate": 6.856899488926747e-05,
+      "loss": 0.2269,
+      "step": 500
+    },
+    {
+      "epoch": 17.310344827586206,
+      "grad_norm": 3.353323459625244,
+      "learning_rate": 6.839863713798978e-05,
+      "loss": 0.2764,
+      "step": 502
+    },
+    {
+      "epoch": 17.379310344827587,
+      "grad_norm": 3.8564820289611816,
+      "learning_rate": 6.82282793867121e-05,
+      "loss": 0.4139,
+      "step": 504
+    },
+    {
+      "epoch": 17.448275862068964,
+      "grad_norm": 6.136475086212158,
+      "learning_rate": 6.805792163543442e-05,
+      "loss": 0.4473,
+      "step": 506
+    },
+    {
+      "epoch": 17.517241379310345,
+      "grad_norm": 4.821500301361084,
+      "learning_rate": 6.788756388415673e-05,
+      "loss": 0.4126,
+      "step": 508
+    },
+    {
+      "epoch": 17.586206896551722,
+      "grad_norm": 6.1974382400512695,
+      "learning_rate": 6.771720613287905e-05,
+      "loss": 0.6224,
+      "step": 510
+    },
+    {
+      "epoch": 17.655172413793103,
+      "grad_norm": 0.697867214679718,
+      "learning_rate": 6.754684838160137e-05,
+      "loss": 0.2629,
+      "step": 512
+    },
+    {
+      "epoch": 17.724137931034484,
+      "grad_norm": 6.468963623046875,
+      "learning_rate": 6.737649063032368e-05,
+      "loss": 0.4406,
+      "step": 514
+    },
+    {
+      "epoch": 17.79310344827586,
+      "grad_norm": 5.978774547576904,
+      "learning_rate": 6.7206132879046e-05,
+      "loss": 0.3807,
+      "step": 516
+    },
+    {
+      "epoch": 17.862068965517242,
+      "grad_norm": 2.750267744064331,
+      "learning_rate": 6.703577512776832e-05,
+      "loss": 0.3092,
+      "step": 518
+    },
+    {
+      "epoch": 17.93103448275862,
+      "grad_norm": 3.745131492614746,
+      "learning_rate": 6.686541737649063e-05,
+      "loss": 0.5316,
+      "step": 520
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 1.9382257461547852,
+      "learning_rate": 6.669505962521295e-05,
+      "loss": 0.2998,
+      "step": 522
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.6363636363636364,
+      "eval_f1_macro": 0.5543597280552314,
+      "eval_f1_micro": 0.6363636363636364,
+      "eval_f1_weighted": 0.625135726951402,
+      "eval_loss": 1.1443867683410645,
+      "eval_precision_macro": 0.5519855144855145,
+      "eval_precision_micro": 0.6363636363636364,
+      "eval_precision_weighted": 0.6248355495514586,
+      "eval_recall_macro": 0.566984126984127,
+      "eval_recall_micro": 0.6363636363636364,
+      "eval_recall_weighted": 0.6363636363636364,
+      "eval_runtime": 2.3094,
+      "eval_samples_per_second": 57.157,
+      "eval_steps_per_second": 7.361,
+      "step": 522
+    },
+    {
+      "epoch": 18.06896551724138,
+      "grad_norm": 3.785149335861206,
+      "learning_rate": 6.652470187393528e-05,
+      "loss": 0.285,
+      "step": 524
+    },
+    {
+      "epoch": 18.137931034482758,
+      "grad_norm": 1.203855276107788,
+      "learning_rate": 6.635434412265758e-05,
+      "loss": 0.2495,
+      "step": 526
+    },
+    {
+      "epoch": 18.20689655172414,
+      "grad_norm": 1.5303699970245361,
+      "learning_rate": 6.61839863713799e-05,
+      "loss": 0.2783,
+      "step": 528
+    },
+    {
+      "epoch": 18.275862068965516,
+      "grad_norm": 0.9707581996917725,
+      "learning_rate": 6.601362862010221e-05,
+      "loss": 0.3239,
+      "step": 530
+    },
+    {
+      "epoch": 18.344827586206897,
+      "grad_norm": 5.368850231170654,
+      "learning_rate": 6.584327086882453e-05,
+      "loss": 0.3304,
+      "step": 532
+    },
+    {
+      "epoch": 18.413793103448278,
+      "grad_norm": 2.649095296859741,
+      "learning_rate": 6.567291311754686e-05,
+      "loss": 0.2108,
+      "step": 534
+    },
+    {
+      "epoch": 18.482758620689655,
+      "grad_norm": 6.152352333068848,
+      "learning_rate": 6.550255536626916e-05,
+      "loss": 0.4266,
+      "step": 536
+    },
+    {
+      "epoch": 18.551724137931036,
+      "grad_norm": 2.5450921058654785,
+      "learning_rate": 6.533219761499149e-05,
+      "loss": 0.3052,
+      "step": 538
+    },
+    {
+      "epoch": 18.620689655172413,
+      "grad_norm": 6.302542209625244,
+      "learning_rate": 6.516183986371381e-05,
+      "loss": 0.2785,
+      "step": 540
+    },
+    {
+      "epoch": 18.689655172413794,
+      "grad_norm": 1.360953688621521,
+      "learning_rate": 6.499148211243612e-05,
+      "loss": 0.3293,
+      "step": 542
+    },
+    {
+      "epoch": 18.75862068965517,
+      "grad_norm": 1.7876797914505005,
+      "learning_rate": 6.482112436115844e-05,
+      "loss": 0.445,
+      "step": 544
+    },
+    {
+      "epoch": 18.82758620689655,
+      "grad_norm": 2.4594149589538574,
+      "learning_rate": 6.465076660988076e-05,
+      "loss": 0.2731,
+      "step": 546
+    },
+    {
+      "epoch": 18.896551724137932,
+      "grad_norm": 0.9318203330039978,
+      "learning_rate": 6.448040885860307e-05,
+      "loss": 0.2172,
+      "step": 548
+    },
+    {
+      "epoch": 18.96551724137931,
+      "grad_norm": 1.2655283212661743,
+      "learning_rate": 6.431005110732539e-05,
+      "loss": 0.2706,
+      "step": 550
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.6439393939393939,
+      "eval_f1_macro": 0.5578558652329144,
+      "eval_f1_micro": 0.6439393939393939,
+      "eval_f1_weighted": 0.630778897377258,
+      "eval_loss": 1.1071969270706177,
+      "eval_precision_macro": 0.578964121533781,
+      "eval_precision_micro": 0.6439393939393939,
+      "eval_precision_weighted": 0.6404139552007551,
+      "eval_recall_macro": 0.5571126228269084,
+      "eval_recall_micro": 0.6439393939393939,
+      "eval_recall_weighted": 0.6439393939393939,
+      "eval_runtime": 2.3243,
+      "eval_samples_per_second": 56.79,
+      "eval_steps_per_second": 7.314,
+      "step": 551
+    },
+    {
+      "epoch": 19.03448275862069,
+      "grad_norm": 2.0673553943634033,
+      "learning_rate": 6.41396933560477e-05,
+      "loss": 0.2022,
+      "step": 552
+    },
+    {
+      "epoch": 19.103448275862068,
+      "grad_norm": 2.8956522941589355,
+      "learning_rate": 6.396933560477002e-05,
+      "loss": 0.2003,
+      "step": 554
+    },
+    {
+      "epoch": 19.17241379310345,
+      "grad_norm": 4.309751987457275,
+      "learning_rate": 6.379897785349234e-05,
+      "loss": 0.1702,
+      "step": 556
+    },
+    {
+      "epoch": 19.24137931034483,
+      "grad_norm": 3.166416645050049,
+      "learning_rate": 6.362862010221465e-05,
+      "loss": 0.2946,
+      "step": 558
+    },
+    {
+      "epoch": 19.310344827586206,
+      "grad_norm": 3.587289810180664,
+      "learning_rate": 6.345826235093697e-05,
+      "loss": 0.2177,
+      "step": 560
+    },
+    {
+      "epoch": 19.379310344827587,
+      "grad_norm": 1.5400705337524414,
+      "learning_rate": 6.328790459965929e-05,
+      "loss": 0.1854,
+      "step": 562
+    },
+    {
+      "epoch": 19.448275862068964,
+      "grad_norm": 5.357079029083252,
+      "learning_rate": 6.31175468483816e-05,
+      "loss": 0.3174,
+      "step": 564
+    },
+    {
+      "epoch": 19.517241379310345,
+      "grad_norm": 6.901614189147949,
+      "learning_rate": 6.294718909710392e-05,
+      "loss": 0.2479,
+      "step": 566
+    },
+    {
+      "epoch": 19.586206896551722,
+      "grad_norm": 0.9026078581809998,
+      "learning_rate": 6.277683134582624e-05,
+      "loss": 0.1429,
+      "step": 568
+    },
+    {
+      "epoch": 19.655172413793103,
+      "grad_norm": 1.1682796478271484,
+      "learning_rate": 6.260647359454855e-05,
+      "loss": 0.196,
+      "step": 570
+    },
+    {
+      "epoch": 19.724137931034484,
+      "grad_norm": 5.2128095626831055,
+      "learning_rate": 6.243611584327087e-05,
+      "loss": 0.4045,
+      "step": 572
+    },
+    {
+      "epoch": 19.79310344827586,
+      "grad_norm": 1.158043622970581,
+      "learning_rate": 6.22657580919932e-05,
+      "loss": 0.2009,
+      "step": 574
+    },
+    {
+      "epoch": 19.862068965517242,
+      "grad_norm": 3.482198715209961,
+      "learning_rate": 6.20954003407155e-05,
+      "loss": 0.3158,
+      "step": 576
+    },
+    {
+      "epoch": 19.93103448275862,
+      "grad_norm": 5.8238749504089355,
+      "learning_rate": 6.192504258943782e-05,
+      "loss": 0.2716,
+      "step": 578
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.5987008810043335,
+      "learning_rate": 6.175468483816013e-05,
+      "loss": 0.2012,
+      "step": 580
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.6212121212121212,
+      "eval_f1_macro": 0.5277531023715308,
+      "eval_f1_micro": 0.6212121212121212,
+      "eval_f1_weighted": 0.6012226561830858,
+      "eval_loss": 1.1352587938308716,
+      "eval_precision_macro": 0.5433209647495362,
+      "eval_precision_micro": 0.6212121212121212,
+      "eval_precision_weighted": 0.6063147710874984,
+      "eval_recall_macro": 0.5345502645502646,
+      "eval_recall_micro": 0.6212121212121212,
+      "eval_recall_weighted": 0.6212121212121212,
+      "eval_runtime": 2.3114,
+      "eval_samples_per_second": 57.108,
+      "eval_steps_per_second": 7.355,
+      "step": 580
+    },
+    {
+      "epoch": 20.06896551724138,
+      "grad_norm": 2.505438804626465,
+      "learning_rate": 6.158432708688245e-05,
+      "loss": 0.224,
+      "step": 582
+    },
+    {
+      "epoch": 20.137931034482758,
+      "grad_norm": 2.0416412353515625,
+      "learning_rate": 6.141396933560477e-05,
+      "loss": 0.2043,
+      "step": 584
+    },
+    {
+      "epoch": 20.20689655172414,
+      "grad_norm": 1.9071308374404907,
+      "learning_rate": 6.124361158432708e-05,
+      "loss": 0.2274,
+      "step": 586
+    },
+    {
+      "epoch": 20.275862068965516,
+      "grad_norm": 0.7224813103675842,
+      "learning_rate": 6.10732538330494e-05,
+      "loss": 0.2217,
+      "step": 588
+    },
+    {
+      "epoch": 20.344827586206897,
+      "grad_norm": 2.6817595958709717,
+      "learning_rate": 6.090289608177172e-05,
+      "loss": 0.1622,
+      "step": 590
+    },
+    {
+      "epoch": 20.413793103448278,
+      "grad_norm": 1.356693983078003,
+      "learning_rate": 6.073253833049404e-05,
+      "loss": 0.2742,
+      "step": 592
+    },
+    {
+      "epoch": 20.482758620689655,
+      "grad_norm": 1.0252429246902466,
+      "learning_rate": 6.0562180579216355e-05,
+      "loss": 0.2045,
+      "step": 594
+    },
+    {
+      "epoch": 20.551724137931036,
+      "grad_norm": 4.123605251312256,
+      "learning_rate": 6.039182282793867e-05,
+      "loss": 0.2,
+      "step": 596
+    },
+    {
+      "epoch": 20.620689655172413,
+      "grad_norm": 6.3188252449035645,
+      "learning_rate": 6.0221465076660985e-05,
+      "loss": 0.4003,
+      "step": 598
+    },
+    {
+      "epoch": 20.689655172413794,
+      "grad_norm": 0.8833699226379395,
+      "learning_rate": 6.0051107325383306e-05,
+      "loss": 0.1515,
+      "step": 600
+    },
+    {
+      "epoch": 20.75862068965517,
+      "grad_norm": 4.920917510986328,
+      "learning_rate": 5.988074957410562e-05,
+      "loss": 0.2139,
+      "step": 602
+    },
+    {
+      "epoch": 20.82758620689655,
+      "grad_norm": 4.224968910217285,
+      "learning_rate": 5.9710391822827936e-05,
+      "loss": 0.3153,
+      "step": 604
+    },
+    {
+      "epoch": 20.896551724137932,
+      "grad_norm": 3.5032029151916504,
+      "learning_rate": 5.954003407155026e-05,
+      "loss": 0.3067,
+      "step": 606
+    },
+    {
+      "epoch": 20.96551724137931,
+      "grad_norm": 10.811616897583008,
+      "learning_rate": 5.936967632027257e-05,
+      "loss": 0.532,
+      "step": 608
+    },
+    {
+      "epoch": 21.0,
+      "eval_accuracy": 0.6212121212121212,
+      "eval_f1_macro": 0.5421176283730086,
+      "eval_f1_micro": 0.6212121212121212,
+      "eval_f1_weighted": 0.6079439132523782,
+      "eval_loss": 1.2502676248550415,
+      "eval_precision_macro": 0.5651044277360067,
+      "eval_precision_micro": 0.6212121212121212,
+      "eval_precision_weighted": 0.6253389154704945,
+      "eval_recall_macro": 0.5488057445200303,
+      "eval_recall_micro": 0.6212121212121212,
+      "eval_recall_weighted": 0.6212121212121212,
+      "eval_runtime": 2.3136,
+      "eval_samples_per_second": 57.055,
+      "eval_steps_per_second": 7.348,
+      "step": 609
+    },
+    {
+      "epoch": 21.03448275862069,
+      "grad_norm": 3.4482898712158203,
+      "learning_rate": 5.919931856899489e-05,
+      "loss": 0.153,
+      "step": 610
+    },
+    {
+      "epoch": 21.103448275862068,
+      "grad_norm": 7.616995811462402,
+      "learning_rate": 5.90289608177172e-05,
+      "loss": 0.2165,
+      "step": 612
+    },
+    {
+      "epoch": 21.17241379310345,
+      "grad_norm": 2.2934651374816895,
+      "learning_rate": 5.8858603066439524e-05,
+      "loss": 0.1314,
+      "step": 614
+    },
+    {
+      "epoch": 21.24137931034483,
+      "grad_norm": 3.9521491527557373,
+      "learning_rate": 5.868824531516184e-05,
+      "loss": 0.2012,
+      "step": 616
+    },
+    {
+      "epoch": 21.310344827586206,
+      "grad_norm": 5.459672451019287,
+      "learning_rate": 5.851788756388415e-05,
+      "loss": 0.1458,
+      "step": 618
+    },
+    {
+      "epoch": 21.379310344827587,
+      "grad_norm": 4.220061302185059,
+      "learning_rate": 5.8347529812606475e-05,
+      "loss": 0.386,
+      "step": 620
+    },
+    {
+      "epoch": 21.448275862068964,
+      "grad_norm": 5.224494934082031,
+      "learning_rate": 5.817717206132879e-05,
+      "loss": 0.3038,
+      "step": 622
+    },
+    {
+      "epoch": 21.517241379310345,
+      "grad_norm": 1.2199459075927734,
+      "learning_rate": 5.8006814310051104e-05,
+      "loss": 0.1402,
+      "step": 624
+    },
+    {
+      "epoch": 21.586206896551722,
+      "grad_norm": 0.805578887462616,
+      "learning_rate": 5.783645655877342e-05,
+      "loss": 0.1285,
+      "step": 626
+    },
+    {
+      "epoch": 21.655172413793103,
+      "grad_norm": 1.9446711540222168,
+      "learning_rate": 5.766609880749574e-05,
+      "loss": 0.2583,
+      "step": 628
+    },
+    {
+      "epoch": 21.724137931034484,
+      "grad_norm": 4.205049514770508,
+      "learning_rate": 5.7495741056218056e-05,
+      "loss": 0.222,
+      "step": 630
+    },
+    {
+      "epoch": 21.79310344827586,
+      "grad_norm": 6.98620080947876,
+      "learning_rate": 5.732538330494037e-05,
+      "loss": 0.2813,
+      "step": 632
+    },
+    {
+      "epoch": 21.862068965517242,
+      "grad_norm": 5.716063499450684,
+      "learning_rate": 5.715502555366269e-05,
+      "loss": 0.219,
+      "step": 634
+    },
+    {
+      "epoch": 21.93103448275862,
+      "grad_norm": 5.608608722686768,
+      "learning_rate": 5.6984667802385014e-05,
+      "loss": 0.1247,
+      "step": 636
+    },
+    {
+      "epoch": 22.0,
+      "grad_norm": 1.143032431602478,
+      "learning_rate": 5.6814310051107335e-05,
+      "loss": 0.0963,
+      "step": 638
+    },
+    {
+      "epoch": 22.0,
+      "eval_accuracy": 0.6287878787878788,
+      "eval_f1_macro": 0.5702385325783071,
+      "eval_f1_micro": 0.6287878787878788,
+      "eval_f1_weighted": 0.6226690485974845,
+      "eval_loss": 1.2203162908554077,
+      "eval_precision_macro": 0.5806972789115646,
+      "eval_precision_micro": 0.6287878787878788,
+      "eval_precision_weighted": 0.6326839826839827,
+      "eval_recall_macro": 0.5744897959183674,
+      "eval_recall_micro": 0.6287878787878788,
+      "eval_recall_weighted": 0.6287878787878788,
+      "eval_runtime": 2.3213,
+      "eval_samples_per_second": 56.865,
+      "eval_steps_per_second": 7.323,
+      "step": 638
+    },
+    {
+      "epoch": 22.06896551724138,
+      "grad_norm": 5.11508321762085,
+      "learning_rate": 5.664395229982965e-05,
+      "loss": 0.1649,
+      "step": 640
+    },
+    {
+      "epoch": 22.137931034482758,
+      "grad_norm": 0.7132185697555542,
+      "learning_rate": 5.6473594548551965e-05,
+      "loss": 0.1236,
+      "step": 642
+    },
+    {
+      "epoch": 22.20689655172414,
+      "grad_norm": 0.6817318201065063,
+      "learning_rate": 5.6303236797274287e-05,
+      "loss": 0.0845,
+      "step": 644
+    },
+    {
+      "epoch": 22.275862068965516,
+      "grad_norm": 2.2783679962158203,
+      "learning_rate": 5.61328790459966e-05,
+      "loss": 0.092,
+      "step": 646
+    },
+    {
+      "epoch": 22.344827586206897,
+      "grad_norm": 3.2513129711151123,
+      "learning_rate": 5.5962521294718916e-05,
+      "loss": 0.1328,
+      "step": 648
+    },
+    {
+      "epoch": 22.413793103448278,
+      "grad_norm": 6.269516468048096,
+      "learning_rate": 5.579216354344123e-05,
+      "loss": 0.2076,
+      "step": 650
+    },
+    {
+      "epoch": 22.482758620689655,
+      "grad_norm": 0.8975334763526917,
+      "learning_rate": 5.562180579216355e-05,
+      "loss": 0.1214,
+      "step": 652
+    },
+    {
+      "epoch": 22.551724137931036,
+      "grad_norm": 0.4030892848968506,
+      "learning_rate": 5.545144804088587e-05,
+      "loss": 0.1049,
+      "step": 654
+    },
+    {
+      "epoch": 22.620689655172413,
+      "grad_norm": 1.1087042093276978,
+      "learning_rate": 5.528109028960818e-05,
+      "loss": 0.1951,
+      "step": 656
+    },
+    {
+      "epoch": 22.689655172413794,
+      "grad_norm": 3.3451485633850098,
+      "learning_rate": 5.5110732538330504e-05,
+      "loss": 0.1705,
+      "step": 658
+    },
+    {
+      "epoch": 22.75862068965517,
+      "grad_norm": 7.313813209533691,
+      "learning_rate": 5.494037478705282e-05,
+      "loss": 0.1902,
+      "step": 660
+    },
+    {
+      "epoch": 22.82758620689655,
+      "grad_norm": 1.9803084135055542,
+      "learning_rate": 5.4770017035775133e-05,
+      "loss": 0.1722,
+      "step": 662
+    },
+    {
+      "epoch": 22.896551724137932,
+      "grad_norm": 3.1250195503234863,
+      "learning_rate": 5.459965928449745e-05,
+      "loss": 0.137,
+      "step": 664
+    },
+    {
+      "epoch": 22.96551724137931,
+      "grad_norm": 4.113572120666504,
+      "learning_rate": 5.442930153321977e-05,
+      "loss": 0.1076,
+      "step": 666
+    },
+    {
+      "epoch": 23.0,
+      "eval_accuracy": 0.6136363636363636,
+      "eval_f1_macro": 0.5216254460440507,
+      "eval_f1_micro": 0.6136363636363636,
+      "eval_f1_weighted": 0.5893711827578106,
+      "eval_loss": 1.379845142364502,
+      "eval_precision_macro": 0.5338904716851017,
+      "eval_precision_micro": 0.6136363636363636,
+      "eval_precision_weighted": 0.5970973452876414,
+      "eval_recall_macro": 0.5370143613000756,
+      "eval_recall_micro": 0.6136363636363636,
+      "eval_recall_weighted": 0.6136363636363636,
+      "eval_runtime": 2.3195,
+      "eval_samples_per_second": 56.908,
+      "eval_steps_per_second": 7.329,
+      "step": 667
+    },
+    {
+      "epoch": 23.03448275862069,
+      "grad_norm": 1.8101320266723633,
+      "learning_rate": 5.4258943781942085e-05,
+      "loss": 0.1341,
+      "step": 668
+    },
+    {
+      "epoch": 23.103448275862068,
+      "grad_norm": 0.7451447248458862,
+      "learning_rate": 5.40885860306644e-05,
+      "loss": 0.0805,
+      "step": 670
+    },
+    {
+      "epoch": 23.17241379310345,
+      "grad_norm": 0.24270519614219666,
+      "learning_rate": 5.391822827938672e-05,
+      "loss": 0.0853,
+      "step": 672
+    },
+    {
+      "epoch": 23.24137931034483,
+      "grad_norm": 3.7437033653259277,
+      "learning_rate": 5.3747870528109036e-05,
+      "loss": 0.1616,
+      "step": 674
+    },
+    {
+      "epoch": 23.310344827586206,
+      "grad_norm": 6.19563102722168,
+      "learning_rate": 5.357751277683135e-05,
+      "loss": 0.1714,
+      "step": 676
+    },
+    {
+      "epoch": 23.379310344827587,
+      "grad_norm": 3.476529359817505,
+      "learning_rate": 5.3407155025553665e-05,
+      "loss": 0.2492,
+      "step": 678
+    },
+    {
+      "epoch": 23.448275862068964,
+      "grad_norm": 13.964451789855957,
+      "learning_rate": 5.323679727427599e-05,
+      "loss": 0.2977,
+      "step": 680
+    },
+    {
+      "epoch": 23.517241379310345,
+      "grad_norm": 0.46169885993003845,
+      "learning_rate": 5.30664395229983e-05,
+      "loss": 0.0657,
+      "step": 682
+    },
+    {
+      "epoch": 23.586206896551722,
+      "grad_norm": 0.8046391010284424,
+      "learning_rate": 5.289608177172062e-05,
+      "loss": 0.0722,
+      "step": 684
+    },
+    {
+      "epoch": 23.655172413793103,
+      "grad_norm": 3.1619741916656494,
+      "learning_rate": 5.272572402044293e-05,
+      "loss": 0.3397,
+      "step": 686
+    },
+    {
+      "epoch": 23.724137931034484,
+      "grad_norm": 4.315649032592773,
+      "learning_rate": 5.255536626916525e-05,
+      "loss": 0.198,
+      "step": 688
+    },
+    {
+      "epoch": 23.79310344827586,
+      "grad_norm": 2.2377915382385254,
+      "learning_rate": 5.238500851788757e-05,
+      "loss": 0.1962,
+      "step": 690
+    },
+    {
+      "epoch": 23.862068965517242,
+      "grad_norm": 1.14326012134552,
+      "learning_rate": 5.221465076660988e-05,
+      "loss": 0.1153,
+      "step": 692
+    },
+    {
+      "epoch": 23.93103448275862,
+      "grad_norm": 1.239844799041748,
+      "learning_rate": 5.2044293015332204e-05,
+      "loss": 0.0955,
+      "step": 694
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 0.43866583704948425,
+      "learning_rate": 5.187393526405452e-05,
+      "loss": 0.1773,
+      "step": 696
+    },
+    {
+      "epoch": 24.0,
+      "eval_accuracy": 0.6287878787878788,
+      "eval_f1_macro": 0.5421531023295404,
+      "eval_f1_micro": 0.6287878787878788,
+      "eval_f1_weighted": 0.6169479734448103,
+      "eval_loss": 1.312862515449524,
+      "eval_precision_macro": 0.5581038009609438,
+      "eval_precision_micro": 0.6287878787878788,
+      "eval_precision_weighted": 0.6253027275754549,
+      "eval_recall_macro": 0.5452985638699924,
+      "eval_recall_micro": 0.6287878787878788,
+      "eval_recall_weighted": 0.6287878787878788,
+      "eval_runtime": 2.32,
+      "eval_samples_per_second": 56.896,
+      "eval_steps_per_second": 7.328,
+      "step": 696
+    },
+    {
+      "epoch": 24.06896551724138,
+      "grad_norm": 0.3895813822746277,
+      "learning_rate": 5.1703577512776834e-05,
+      "loss": 0.0867,
+      "step": 698
+    },
+    {
+      "epoch": 24.137931034482758,
+      "grad_norm": 0.483073353767395,
+      "learning_rate": 5.153321976149915e-05,
+      "loss": 0.0728,
+      "step": 700
+    },
+    {
+      "epoch": 24.20689655172414,
+      "grad_norm": 0.414039671421051,
+      "learning_rate": 5.136286201022147e-05,
+      "loss": 0.0723,
+      "step": 702
+    },
+    {
+      "epoch": 24.275862068965516,
+      "grad_norm": 1.4860070943832397,
+      "learning_rate": 5.1192504258943785e-05,
+      "loss": 0.0724,
+      "step": 704
+    },
+    {
+      "epoch": 24.344827586206897,
+      "grad_norm": 3.1332881450653076,
+      "learning_rate": 5.10221465076661e-05,
+      "loss": 0.2335,
+      "step": 706
+    },
+    {
+      "epoch": 24.413793103448278,
+      "grad_norm": 0.35537058115005493,
+      "learning_rate": 5.085178875638842e-05,
+      "loss": 0.0597,
+      "step": 708
+    },
+    {
+      "epoch": 24.482758620689655,
+      "grad_norm": 0.21754102408885956,
+      "learning_rate": 5.0681431005110736e-05,
+      "loss": 0.0556,
+      "step": 710
+    },
+    {
+      "epoch": 24.551724137931036,
+      "grad_norm": 0.2571113109588623,
+      "learning_rate": 5.051107325383305e-05,
+      "loss": 0.0597,
+      "step": 712
+    },
+    {
+      "epoch": 24.620689655172413,
+      "grad_norm": 4.0861124992370605,
+      "learning_rate": 5.0340715502555366e-05,
+      "loss": 0.0977,
+      "step": 714
+    },
+    {
+      "epoch": 24.689655172413794,
+      "grad_norm": 5.946489334106445,
+      "learning_rate": 5.017035775127769e-05,
+      "loss": 0.1114,
+      "step": 716
+    },
+    {
+      "epoch": 24.75862068965517,
+      "grad_norm": 0.2896263301372528,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "step": 718
+    },
+    {
+      "epoch": 24.82758620689655,
+      "grad_norm": 1.2867406606674194,
+      "learning_rate": 4.982964224872232e-05,
+      "loss": 0.0753,
+      "step": 720
+    },
+    {
+      "epoch": 24.896551724137932,
+      "grad_norm": 0.8423182964324951,
+      "learning_rate": 4.965928449744464e-05,
+      "loss": 0.1063,
+      "step": 722
+    },
+    {
+      "epoch": 24.96551724137931,
+      "grad_norm": 0.9143354892730713,
+      "learning_rate": 4.9488926746166954e-05,
+      "loss": 0.0598,
+      "step": 724
+    },
+    {
+      "epoch": 25.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5632663632663633,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6381401426855972,
+      "eval_loss": 1.285531759262085,
+      "eval_precision_macro": 0.5845804988662131,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.6562049062049062,
+      "eval_recall_macro": 0.5713227513227512,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3186,
+      "eval_samples_per_second": 56.931,
+      "eval_steps_per_second": 7.332,
+      "step": 725
+    },
+    {
+      "epoch": 25.03448275862069,
+      "grad_norm": 2.1182336807250977,
+      "learning_rate": 4.931856899488927e-05,
+      "loss": 0.1474,
+      "step": 726
+    },
+    {
+      "epoch": 25.103448275862068,
+      "grad_norm": 4.869687080383301,
+      "learning_rate": 4.914821124361158e-05,
+      "loss": 0.0703,
+      "step": 728
+    },
+    {
+      "epoch": 25.17241379310345,
+      "grad_norm": 0.5248813629150391,
+      "learning_rate": 4.8977853492333905e-05,
+      "loss": 0.0831,
+      "step": 730
+    },
+    {
+      "epoch": 25.24137931034483,
+      "grad_norm": 5.952985763549805,
+      "learning_rate": 4.880749574105622e-05,
+      "loss": 0.1589,
+      "step": 732
+    },
+    {
+      "epoch": 25.310344827586206,
+      "grad_norm": 0.14842434227466583,
+      "learning_rate": 4.8637137989778534e-05,
+      "loss": 0.0472,
+      "step": 734
+    },
+    {
+      "epoch": 25.379310344827587,
+      "grad_norm": 1.5915170907974243,
+      "learning_rate": 4.8466780238500856e-05,
+      "loss": 0.0788,
+      "step": 736
+    },
+    {
+      "epoch": 25.448275862068964,
+      "grad_norm": 0.9509754180908203,
+      "learning_rate": 4.829642248722317e-05,
+      "loss": 0.1258,
+      "step": 738
+    },
+    {
+      "epoch": 25.517241379310345,
+      "grad_norm": 0.5396161079406738,
+      "learning_rate": 4.8126064735945486e-05,
+      "loss": 0.0547,
+      "step": 740
+    },
+    {
+      "epoch": 25.586206896551722,
+      "grad_norm": 0.6499558091163635,
+      "learning_rate": 4.79557069846678e-05,
+      "loss": 0.0606,
+      "step": 742
+    },
+    {
+      "epoch": 25.655172413793103,
+      "grad_norm": 0.24467584490776062,
+      "learning_rate": 4.778534923339012e-05,
+      "loss": 0.0465,
+      "step": 744
+    },
+    {
+      "epoch": 25.724137931034484,
+      "grad_norm": 1.15547513961792,
+      "learning_rate": 4.761499148211244e-05,
+      "loss": 0.0808,
+      "step": 746
+    },
+    {
+      "epoch": 25.79310344827586,
+      "grad_norm": 5.311452388763428,
+      "learning_rate": 4.744463373083475e-05,
+      "loss": 0.2962,
+      "step": 748
+    },
+    {
+      "epoch": 25.862068965517242,
+      "grad_norm": 0.5441858172416687,
+      "learning_rate": 4.727427597955707e-05,
+      "loss": 0.0454,
+      "step": 750
+    },
+    {
+      "epoch": 25.93103448275862,
+      "grad_norm": 6.218692302703857,
+      "learning_rate": 4.710391822827939e-05,
+      "loss": 0.2303,
+      "step": 752
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 1.4410537481307983,
+      "learning_rate": 4.69335604770017e-05,
+      "loss": 0.0632,
+      "step": 754
+    },
+    {
+      "epoch": 26.0,
+      "eval_accuracy": 0.6590909090909091,
+      "eval_f1_macro": 0.6414174726494373,
+      "eval_f1_micro": 0.6590909090909091,
+      "eval_f1_weighted": 0.6643352984418467,
+      "eval_loss": 1.3154751062393188,
+      "eval_precision_macro": 0.6524587989650101,
+      "eval_precision_micro": 0.6590909090909091,
+      "eval_precision_weighted": 0.6925259605348538,
+      "eval_recall_macro": 0.6584580498866214,
+      "eval_recall_micro": 0.6590909090909091,
+      "eval_recall_weighted": 0.6590909090909091,
+      "eval_runtime": 2.3313,
+      "eval_samples_per_second": 56.62,
+      "eval_steps_per_second": 7.292,
+      "step": 754
+    },
+    {
+      "epoch": 26.06896551724138,
+      "grad_norm": 0.5304362177848816,
+      "learning_rate": 4.676320272572402e-05,
+      "loss": 0.0479,
+      "step": 756
+    },
+    {
+      "epoch": 26.137931034482758,
+      "grad_norm": 2.6431336402893066,
+      "learning_rate": 4.659284497444634e-05,
+      "loss": 0.1618,
+      "step": 758
+    },
+    {
+      "epoch": 26.20689655172414,
+      "grad_norm": 0.23049412667751312,
+      "learning_rate": 4.642248722316866e-05,
+      "loss": 0.0411,
+      "step": 760
+    },
+    {
+      "epoch": 26.275862068965516,
+      "grad_norm": 2.1315579414367676,
+      "learning_rate": 4.6252129471890976e-05,
+      "loss": 0.0493,
+      "step": 762
+    },
+    {
+      "epoch": 26.344827586206897,
+      "grad_norm": 0.13269948959350586,
+      "learning_rate": 4.608177172061329e-05,
+      "loss": 0.0437,
+      "step": 764
+    },
+    {
+      "epoch": 26.413793103448278,
+      "grad_norm": 6.615429401397705,
+      "learning_rate": 4.591141396933561e-05,
+      "loss": 0.1554,
+      "step": 766
+    },
+    {
+      "epoch": 26.482758620689655,
+      "grad_norm": 0.19145725667476654,
+      "learning_rate": 4.574105621805793e-05,
+      "loss": 0.0791,
+      "step": 768
+    },
+    {
+      "epoch": 26.551724137931036,
+      "grad_norm": 0.16384397447109222,
+      "learning_rate": 4.557069846678024e-05,
+      "loss": 0.0493,
+      "step": 770
+    },
+    {
+      "epoch": 26.620689655172413,
+      "grad_norm": 9.984197616577148,
+      "learning_rate": 4.5400340715502557e-05,
+      "loss": 0.1315,
+      "step": 772
+    },
+    {
+      "epoch": 26.689655172413794,
+      "grad_norm": 0.5026840567588806,
+      "learning_rate": 4.522998296422488e-05,
+      "loss": 0.1321,
+      "step": 774
+    },
+    {
+      "epoch": 26.75862068965517,
+      "grad_norm": 1.3563400506973267,
+      "learning_rate": 4.505962521294719e-05,
+      "loss": 0.062,
+      "step": 776
+    },
+    {
+      "epoch": 26.82758620689655,
+      "grad_norm": 0.9698802828788757,
+      "learning_rate": 4.488926746166951e-05,
+      "loss": 0.2455,
+      "step": 778
+    },
+    {
+      "epoch": 26.896551724137932,
+      "grad_norm": 0.633067786693573,
+      "learning_rate": 4.471890971039183e-05,
+      "loss": 0.2064,
+      "step": 780
+    },
+    {
+      "epoch": 26.96551724137931,
+      "grad_norm": 2.6660585403442383,
+      "learning_rate": 4.4548551959114144e-05,
+      "loss": 0.0644,
+      "step": 782
+    },
+    {
+      "epoch": 27.0,
+      "eval_accuracy": 0.6439393939393939,
+      "eval_f1_macro": 0.5587869494846239,
+      "eval_f1_micro": 0.6439393939393939,
+      "eval_f1_weighted": 0.6314992501039013,
+      "eval_loss": 1.3210983276367188,
+      "eval_precision_macro": 0.5744595234850107,
+      "eval_precision_micro": 0.6439393939393939,
+      "eval_precision_weighted": 0.6356558291102383,
+      "eval_recall_macro": 0.559546485260771,
+      "eval_recall_micro": 0.6439393939393939,
+      "eval_recall_weighted": 0.6439393939393939,
+      "eval_runtime": 2.313,
+      "eval_samples_per_second": 57.068,
+      "eval_steps_per_second": 7.35,
+      "step": 783
+    },
+    {
+      "epoch": 27.03448275862069,
+      "grad_norm": 1.29978346824646,
+      "learning_rate": 4.437819420783646e-05,
+      "loss": 0.0725,
+      "step": 784
+    },
+    {
+      "epoch": 27.103448275862068,
+      "grad_norm": 0.27794161438941956,
+      "learning_rate": 4.4207836456558774e-05,
+      "loss": 0.0392,
+      "step": 786
+    },
+    {
+      "epoch": 27.17241379310345,
+      "grad_norm": 2.312304735183716,
+      "learning_rate": 4.4037478705281095e-05,
+      "loss": 0.0548,
+      "step": 788
+    },
+    {
+      "epoch": 27.24137931034483,
+      "grad_norm": 0.5731463432312012,
+      "learning_rate": 4.386712095400341e-05,
+      "loss": 0.0438,
+      "step": 790
+    },
+    {
+      "epoch": 27.310344827586206,
+      "grad_norm": 0.17765074968338013,
+      "learning_rate": 4.3696763202725725e-05,
+      "loss": 0.0423,
+      "step": 792
+    },
+    {
+      "epoch": 27.379310344827587,
+      "grad_norm": 7.317860126495361,
+      "learning_rate": 4.352640545144805e-05,
+      "loss": 0.1978,
+      "step": 794
+    },
+    {
+      "epoch": 27.448275862068964,
+      "grad_norm": 0.3569648265838623,
+      "learning_rate": 4.335604770017036e-05,
+      "loss": 0.0662,
+      "step": 796
+    },
+    {
+      "epoch": 27.517241379310345,
+      "grad_norm": 0.5732914805412292,
+      "learning_rate": 4.3185689948892676e-05,
+      "loss": 0.0781,
+      "step": 798
+    },
+    {
+      "epoch": 27.586206896551722,
+      "grad_norm": 10.172492980957031,
+      "learning_rate": 4.301533219761499e-05,
+      "loss": 0.1123,
+      "step": 800
+    },
+    {
+      "epoch": 27.655172413793103,
+      "grad_norm": 0.29396459460258484,
+      "learning_rate": 4.284497444633731e-05,
+      "loss": 0.0433,
+      "step": 802
+    },
+    {
+      "epoch": 27.724137931034484,
+      "grad_norm": 0.17927728593349457,
+      "learning_rate": 4.267461669505963e-05,
+      "loss": 0.075,
+      "step": 804
+    },
+    {
+      "epoch": 27.79310344827586,
+      "grad_norm": 0.27556878328323364,
+      "learning_rate": 4.250425894378194e-05,
+      "loss": 0.0457,
+      "step": 806
+    },
+    {
+      "epoch": 27.862068965517242,
+      "grad_norm": 1.3352630138397217,
+      "learning_rate": 4.2333901192504264e-05,
+      "loss": 0.0662,
+      "step": 808
+    },
+    {
+      "epoch": 27.93103448275862,
+      "grad_norm": 0.1427721381187439,
+      "learning_rate": 4.216354344122658e-05,
+      "loss": 0.055,
+      "step": 810
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 6.838861465454102,
+      "learning_rate": 4.1993185689948894e-05,
+      "loss": 0.1495,
+      "step": 812
+    },
+    {
+      "epoch": 28.0,
+      "eval_accuracy": 0.6363636363636364,
+      "eval_f1_macro": 0.5539424598362265,
+      "eval_f1_micro": 0.6363636363636364,
+      "eval_f1_weighted": 0.6244950107428886,
+      "eval_loss": 1.4195839166641235,
+      "eval_precision_macro": 0.5650270364556079,
+      "eval_precision_micro": 0.6363636363636364,
+      "eval_precision_weighted": 0.6269573019573019,
+      "eval_recall_macro": 0.5555555555555556,
+      "eval_recall_micro": 0.6363636363636364,
+      "eval_recall_weighted": 0.6363636363636364,
+      "eval_runtime": 2.3019,
+      "eval_samples_per_second": 57.344,
+      "eval_steps_per_second": 7.385,
+      "step": 812
+    },
+    {
+      "epoch": 28.06896551724138,
+      "grad_norm": 0.4316509962081909,
+      "learning_rate": 4.182282793867121e-05,
+      "loss": 0.0652,
+      "step": 814
+    },
+    {
+      "epoch": 28.137931034482758,
+      "grad_norm": 0.5784430503845215,
+      "learning_rate": 4.165247018739353e-05,
+      "loss": 0.0404,
+      "step": 816
+    },
+    {
+      "epoch": 28.20689655172414,
+      "grad_norm": 0.12339165061712265,
+      "learning_rate": 4.1482112436115845e-05,
+      "loss": 0.0342,
+      "step": 818
+    },
+    {
+      "epoch": 28.275862068965516,
+      "grad_norm": 9.590283393859863,
+      "learning_rate": 4.131175468483816e-05,
+      "loss": 0.09,
+      "step": 820
+    },
+    {
+      "epoch": 28.344827586206897,
+      "grad_norm": 5.919569492340088,
+      "learning_rate": 4.1141396933560474e-05,
+      "loss": 0.1994,
+      "step": 822
+    },
+    {
+      "epoch": 28.413793103448278,
+      "grad_norm": 0.18216995894908905,
+      "learning_rate": 4.0971039182282796e-05,
+      "loss": 0.0384,
+      "step": 824
+    },
+    {
+      "epoch": 28.482758620689655,
+      "grad_norm": 0.36890658736228943,
+      "learning_rate": 4.080068143100511e-05,
+      "loss": 0.0993,
+      "step": 826
+    },
+    {
+      "epoch": 28.551724137931036,
+      "grad_norm": 0.1248667761683464,
+      "learning_rate": 4.0630323679727426e-05,
+      "loss": 0.0398,
+      "step": 828
+    },
+    {
+      "epoch": 28.620689655172413,
+      "grad_norm": 0.8702082633972168,
+      "learning_rate": 4.045996592844975e-05,
+      "loss": 0.0409,
+      "step": 830
+    },
+    {
+      "epoch": 28.689655172413794,
+      "grad_norm": 0.16621750593185425,
+      "learning_rate": 4.028960817717206e-05,
+      "loss": 0.0386,
+      "step": 832
+    },
+    {
+      "epoch": 28.75862068965517,
+      "grad_norm": 2.846982717514038,
+      "learning_rate": 4.011925042589438e-05,
+      "loss": 0.0558,
+      "step": 834
+    },
+    {
+      "epoch": 28.82758620689655,
+      "grad_norm": 0.125545471906662,
+      "learning_rate": 3.994889267461669e-05,
+      "loss": 0.0348,
+      "step": 836
+    },
+    {
+      "epoch": 28.896551724137932,
+      "grad_norm": 0.17237234115600586,
+      "learning_rate": 3.977853492333901e-05,
+      "loss": 0.038,
+      "step": 838
+    },
+    {
+      "epoch": 28.96551724137931,
+      "grad_norm": 0.919359028339386,
+      "learning_rate": 3.960817717206133e-05,
+      "loss": 0.0413,
+      "step": 840
+    },
+    {
+      "epoch": 29.0,
+      "eval_accuracy": 0.6136363636363636,
+      "eval_f1_macro": 0.5377616118217622,
+      "eval_f1_micro": 0.6136363636363636,
+      "eval_f1_weighted": 0.6102201958063201,
+      "eval_loss": 1.4026981592178345,
+      "eval_precision_macro": 0.5405016036594984,
+      "eval_precision_micro": 0.6136363636363636,
+      "eval_precision_weighted": 0.6100277712717904,
+      "eval_recall_macro": 0.5379591836734694,
+      "eval_recall_micro": 0.6136363636363636,
+      "eval_recall_weighted": 0.6136363636363636,
+      "eval_runtime": 2.3087,
+      "eval_samples_per_second": 57.176,
+      "eval_steps_per_second": 7.364,
+      "step": 841
+    },
+    {
+      "epoch": 29.03448275862069,
+      "grad_norm": 0.18945536017417908,
+      "learning_rate": 3.943781942078364e-05,
+      "loss": 0.0357,
+      "step": 842
+    },
+    {
+      "epoch": 29.103448275862068,
+      "grad_norm": 0.18712615966796875,
+      "learning_rate": 3.9267461669505964e-05,
+      "loss": 0.037,
+      "step": 844
+    },
+    {
+      "epoch": 29.17241379310345,
+      "grad_norm": 0.21111753582954407,
+      "learning_rate": 3.9097103918228286e-05,
+      "loss": 0.0404,
+      "step": 846
+    },
+    {
+      "epoch": 29.24137931034483,
+      "grad_norm": 0.13927434384822845,
+      "learning_rate": 3.89267461669506e-05,
+      "loss": 0.0379,
+      "step": 848
+    },
+    {
+      "epoch": 29.310344827586206,
+      "grad_norm": 0.3238166272640228,
+      "learning_rate": 3.8756388415672916e-05,
+      "loss": 0.0359,
+      "step": 850
+    },
+    {
+      "epoch": 29.379310344827587,
+      "grad_norm": 0.4770512878894806,
+      "learning_rate": 3.858603066439524e-05,
+      "loss": 0.0624,
+      "step": 852
+    },
+    {
+      "epoch": 29.448275862068964,
+      "grad_norm": 12.409387588500977,
+      "learning_rate": 3.841567291311755e-05,
+      "loss": 0.1086,
+      "step": 854
+    },
+    {
+      "epoch": 29.517241379310345,
+      "grad_norm": 0.1156606376171112,
+      "learning_rate": 3.824531516183987e-05,
+      "loss": 0.0345,
+      "step": 856
+    },
+    {
+      "epoch": 29.586206896551722,
+      "grad_norm": 0.11645504087209702,
+      "learning_rate": 3.807495741056218e-05,
+      "loss": 0.1033,
+      "step": 858
+    },
+    {
+      "epoch": 29.655172413793103,
+      "grad_norm": 0.11167976260185242,
+      "learning_rate": 3.79045996592845e-05,
+      "loss": 0.042,
+      "step": 860
+    },
+    {
+      "epoch": 29.724137931034484,
+      "grad_norm": 0.1662347912788391,
+      "learning_rate": 3.773424190800682e-05,
+      "loss": 0.0336,
+      "step": 862
+    },
+    {
+      "epoch": 29.79310344827586,
+      "grad_norm": 0.13040611147880554,
+      "learning_rate": 3.756388415672913e-05,
+      "loss": 0.0335,
+      "step": 864
+    },
+    {
+      "epoch": 29.862068965517242,
+      "grad_norm": 2.854461193084717,
+      "learning_rate": 3.7393526405451454e-05,
+      "loss": 0.0475,
+      "step": 866
+    },
+    {
+      "epoch": 29.93103448275862,
+      "grad_norm": 0.21949215233325958,
+      "learning_rate": 3.722316865417377e-05,
+      "loss": 0.0622,
+      "step": 868
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.12153808027505875,
+      "learning_rate": 3.7052810902896084e-05,
+      "loss": 0.0323,
+      "step": 870
+    },
+    {
+      "epoch": 30.0,
+      "eval_accuracy": 0.6363636363636364,
+      "eval_f1_macro": 0.5640770593951866,
+      "eval_f1_micro": 0.6363636363636364,
+      "eval_f1_weighted": 0.6329471182412358,
+      "eval_loss": 1.4301841259002686,
+      "eval_precision_macro": 0.5688986640599544,
+      "eval_precision_micro": 0.6363636363636364,
+      "eval_precision_weighted": 0.6429970526378151,
+      "eval_recall_macro": 0.5711866969009826,
+      "eval_recall_micro": 0.6363636363636364,
+      "eval_recall_weighted": 0.6363636363636364,
+      "eval_runtime": 2.3127,
+      "eval_samples_per_second": 57.076,
+      "eval_steps_per_second": 7.351,
+      "step": 870
+    },
+    {
+      "epoch": 30.06896551724138,
+      "grad_norm": 10.862177848815918,
+      "learning_rate": 3.68824531516184e-05,
+      "loss": 0.0783,
+      "step": 872
+    },
+    {
+      "epoch": 30.137931034482758,
+      "grad_norm": 0.12320105731487274,
+      "learning_rate": 3.671209540034072e-05,
+      "loss": 0.0354,
+      "step": 874
+    },
+    {
+      "epoch": 30.20689655172414,
+      "grad_norm": 0.14022612571716309,
+      "learning_rate": 3.6541737649063035e-05,
+      "loss": 0.0315,
+      "step": 876
+    },
+    {
+      "epoch": 30.275862068965516,
+      "grad_norm": 2.5534183979034424,
+      "learning_rate": 3.637137989778535e-05,
+      "loss": 0.0403,
+      "step": 878
+    },
+    {
+      "epoch": 30.344827586206897,
+      "grad_norm": 0.13770627975463867,
+      "learning_rate": 3.6201022146507665e-05,
+      "loss": 0.0552,
+      "step": 880
+    },
+    {
+      "epoch": 30.413793103448278,
+      "grad_norm": 0.3032757639884949,
+      "learning_rate": 3.6030664395229987e-05,
+      "loss": 0.0618,
+      "step": 882
+    },
+    {
+      "epoch": 30.482758620689655,
+      "grad_norm": 9.125004768371582,
+      "learning_rate": 3.58603066439523e-05,
+      "loss": 0.0986,
+      "step": 884
+    },
+    {
+      "epoch": 30.551724137931036,
+      "grad_norm": 0.39664426445961,
+      "learning_rate": 3.5689948892674616e-05,
+      "loss": 0.0303,
+      "step": 886
+    },
+    {
+      "epoch": 30.620689655172413,
+      "grad_norm": 0.8094646334648132,
+      "learning_rate": 3.551959114139694e-05,
+      "loss": 0.0347,
+      "step": 888
+    },
+    {
+      "epoch": 30.689655172413794,
+      "grad_norm": 0.10856208205223083,
+      "learning_rate": 3.534923339011925e-05,
+      "loss": 0.0661,
+      "step": 890
+    },
+    {
+      "epoch": 30.75862068965517,
+      "grad_norm": 0.09460315108299255,
+      "learning_rate": 3.517887563884157e-05,
+      "loss": 0.0547,
+      "step": 892
+    },
+    {
+      "epoch": 30.82758620689655,
+      "grad_norm": 2.669100284576416,
+      "learning_rate": 3.500851788756388e-05,
+      "loss": 0.0503,
+      "step": 894
+    },
+    {
+      "epoch": 30.896551724137932,
+      "grad_norm": 0.212706059217453,
+      "learning_rate": 3.4838160136286204e-05,
+      "loss": 0.0637,
+      "step": 896
+    },
+    {
+      "epoch": 30.96551724137931,
+      "grad_norm": 0.41738227009773254,
+      "learning_rate": 3.466780238500852e-05,
+      "loss": 0.0452,
+      "step": 898
+    },
+    {
+      "epoch": 31.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5705624622730447,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6412393051725056,
+      "eval_loss": 1.457693338394165,
+      "eval_precision_macro": 0.5835023111808825,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.647774968087468,
+      "eval_recall_macro": 0.5738321995464852,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3087,
+      "eval_samples_per_second": 57.175,
+      "eval_steps_per_second": 7.363,
+      "step": 899
+    },
+    {
+      "epoch": 31.03448275862069,
+      "grad_norm": 0.5998754501342773,
+      "learning_rate": 3.4497444633730833e-05,
+      "loss": 0.0327,
+      "step": 900
+    },
+    {
+      "epoch": 31.103448275862068,
+      "grad_norm": 0.10836658626794815,
+      "learning_rate": 3.4327086882453155e-05,
+      "loss": 0.0317,
+      "step": 902
+    },
+    {
+      "epoch": 31.17241379310345,
+      "grad_norm": 0.3383055329322815,
+      "learning_rate": 3.415672913117547e-05,
+      "loss": 0.1491,
+      "step": 904
+    },
+    {
+      "epoch": 31.24137931034483,
+      "grad_norm": 0.13469484448432922,
+      "learning_rate": 3.3986371379897785e-05,
+      "loss": 0.0619,
+      "step": 906
+    },
+    {
+      "epoch": 31.310344827586206,
+      "grad_norm": 0.15399225056171417,
+      "learning_rate": 3.38160136286201e-05,
+      "loss": 0.0576,
+      "step": 908
+    },
+    {
+      "epoch": 31.379310344827587,
+      "grad_norm": 0.2508799731731415,
+      "learning_rate": 3.364565587734242e-05,
+      "loss": 0.0283,
+      "step": 910
+    },
+    {
+      "epoch": 31.448275862068964,
+      "grad_norm": 0.09723825007677078,
+      "learning_rate": 3.3475298126064736e-05,
+      "loss": 0.0289,
+      "step": 912
+    },
+    {
+      "epoch": 31.517241379310345,
+      "grad_norm": 0.11320387572050095,
+      "learning_rate": 3.330494037478705e-05,
+      "loss": 0.041,
+      "step": 914
+    },
+    {
+      "epoch": 31.586206896551722,
+      "grad_norm": 0.09735345840454102,
+      "learning_rate": 3.313458262350937e-05,
+      "loss": 0.0296,
+      "step": 916
+    },
+    {
+      "epoch": 31.655172413793103,
+      "grad_norm": 0.1624167114496231,
+      "learning_rate": 3.296422487223169e-05,
+      "loss": 0.0284,
+      "step": 918
+    },
+    {
+      "epoch": 31.724137931034484,
+      "grad_norm": 0.15523399412631989,
+      "learning_rate": 3.2793867120954e-05,
+      "loss": 0.0316,
+      "step": 920
+    },
+    {
+      "epoch": 31.79310344827586,
+      "grad_norm": 0.6281076073646545,
+      "learning_rate": 3.262350936967632e-05,
+      "loss": 0.0301,
+      "step": 922
+    },
+    {
+      "epoch": 31.862068965517242,
+      "grad_norm": 0.10810775309801102,
+      "learning_rate": 3.245315161839864e-05,
+      "loss": 0.1419,
+      "step": 924
+    },
+    {
+      "epoch": 31.93103448275862,
+      "grad_norm": 0.09449290484189987,
+      "learning_rate": 3.228279386712095e-05,
+      "loss": 0.1476,
+      "step": 926
+    },
+    {
+      "epoch": 32.0,
+      "grad_norm": 0.11448001116514206,
+      "learning_rate": 3.2112436115843275e-05,
+      "loss": 0.0285,
+      "step": 928
+    },
+    {
+      "epoch": 32.0,
+      "eval_accuracy": 0.6439393939393939,
+      "eval_f1_macro": 0.5596847319551131,
+      "eval_f1_micro": 0.6439393939393939,
+      "eval_f1_weighted": 0.6300303506636783,
+      "eval_loss": 1.422427773475647,
+      "eval_precision_macro": 0.5618261922409388,
+      "eval_precision_micro": 0.6439393939393939,
+      "eval_precision_weighted": 0.6250287835331823,
+      "eval_recall_macro": 0.5657142857142857,
+      "eval_recall_micro": 0.6439393939393939,
+      "eval_recall_weighted": 0.6439393939393939,
+      "eval_runtime": 2.3052,
+      "eval_samples_per_second": 57.263,
+      "eval_steps_per_second": 7.375,
+      "step": 928
+    },
+    {
+      "epoch": 32.06896551724138,
+      "grad_norm": 0.09805119037628174,
+      "learning_rate": 3.194207836456559e-05,
+      "loss": 0.0305,
+      "step": 930
+    },
+    {
+      "epoch": 32.13793103448276,
+      "grad_norm": 0.6063154339790344,
+      "learning_rate": 3.177172061328791e-05,
+      "loss": 0.0317,
+      "step": 932
+    },
+    {
+      "epoch": 32.206896551724135,
+      "grad_norm": 0.10352925211191177,
+      "learning_rate": 3.1601362862010226e-05,
+      "loss": 0.0272,
+      "step": 934
+    },
+    {
+      "epoch": 32.275862068965516,
+      "grad_norm": 12.472454071044922,
+      "learning_rate": 3.143100511073254e-05,
+      "loss": 0.1287,
+      "step": 936
+    },
+    {
+      "epoch": 32.3448275862069,
+      "grad_norm": 2.9856293201446533,
+      "learning_rate": 3.1260647359454856e-05,
+      "loss": 0.0404,
+      "step": 938
+    },
+    {
+      "epoch": 32.41379310344828,
+      "grad_norm": 0.22983527183532715,
+      "learning_rate": 3.109028960817718e-05,
+      "loss": 0.0285,
+      "step": 940
+    },
+    {
+      "epoch": 32.48275862068966,
+      "grad_norm": 0.1563962697982788,
+      "learning_rate": 3.091993185689949e-05,
+      "loss": 0.0262,
+      "step": 942
+    },
+    {
+      "epoch": 32.55172413793103,
+      "grad_norm": 0.10210330039262772,
+      "learning_rate": 3.074957410562181e-05,
+      "loss": 0.0286,
+      "step": 944
+    },
+    {
+      "epoch": 32.62068965517241,
+      "grad_norm": 0.10098184645175934,
+      "learning_rate": 3.057921635434413e-05,
+      "loss": 0.0273,
+      "step": 946
+    },
+    {
+      "epoch": 32.689655172413794,
+      "grad_norm": 0.19312156736850739,
+      "learning_rate": 3.0408858603066443e-05,
+      "loss": 0.0272,
+      "step": 948
+    },
+    {
+      "epoch": 32.758620689655174,
+      "grad_norm": 0.8814791440963745,
+      "learning_rate": 3.0238500851788758e-05,
+      "loss": 0.0335,
+      "step": 950
+    },
+    {
+      "epoch": 32.827586206896555,
+      "grad_norm": 0.38704535365104675,
+      "learning_rate": 3.0068143100511076e-05,
+      "loss": 0.03,
+      "step": 952
+    },
+    {
+      "epoch": 32.89655172413793,
+      "grad_norm": 0.36842238903045654,
+      "learning_rate": 2.989778534923339e-05,
+      "loss": 0.0466,
+      "step": 954
+    },
+    {
+      "epoch": 32.96551724137931,
+      "grad_norm": 0.12097663432359695,
+      "learning_rate": 2.972742759795571e-05,
+      "loss": 0.0241,
+      "step": 956
+    },
+    {
+      "epoch": 33.0,
+      "eval_accuracy": 0.6363636363636364,
+      "eval_f1_macro": 0.5541615185524713,
+      "eval_f1_micro": 0.6363636363636364,
+      "eval_f1_weighted": 0.6252499952281971,
+      "eval_loss": 1.4512511491775513,
+      "eval_precision_macro": 0.5700210921075582,
+      "eval_precision_micro": 0.6363636363636364,
+      "eval_precision_weighted": 0.6309470561115298,
+      "eval_recall_macro": 0.5532501889644746,
+      "eval_recall_micro": 0.6363636363636364,
+      "eval_recall_weighted": 0.6363636363636364,
+      "eval_runtime": 2.3075,
+      "eval_samples_per_second": 57.205,
+      "eval_steps_per_second": 7.367,
+      "step": 957
+    },
+    {
+      "epoch": 33.03448275862069,
+      "grad_norm": 0.1205965206027031,
+      "learning_rate": 2.9557069846678027e-05,
+      "loss": 0.0294,
+      "step": 958
+    },
+    {
+      "epoch": 33.10344827586207,
+      "grad_norm": 0.09090107679367065,
+      "learning_rate": 2.9386712095400342e-05,
+      "loss": 0.0296,
+      "step": 960
+    },
+    {
+      "epoch": 33.172413793103445,
+      "grad_norm": 0.08838998526334763,
+      "learning_rate": 2.921635434412266e-05,
+      "loss": 0.026,
+      "step": 962
+    },
+    {
+      "epoch": 33.241379310344826,
+      "grad_norm": 0.2968568801879883,
+      "learning_rate": 2.9045996592844975e-05,
+      "loss": 0.0405,
+      "step": 964
+    },
+    {
+      "epoch": 33.310344827586206,
+      "grad_norm": 0.11646933853626251,
+      "learning_rate": 2.8875638841567293e-05,
+      "loss": 0.0276,
+      "step": 966
+    },
+    {
+      "epoch": 33.37931034482759,
+      "grad_norm": 0.09644908457994461,
+      "learning_rate": 2.8705281090289608e-05,
+      "loss": 0.0257,
+      "step": 968
+    },
+    {
+      "epoch": 33.44827586206897,
+      "grad_norm": 0.258109986782074,
+      "learning_rate": 2.8534923339011926e-05,
+      "loss": 0.0279,
+      "step": 970
+    },
+    {
+      "epoch": 33.51724137931034,
+      "grad_norm": 1.4819831848144531,
+      "learning_rate": 2.8364565587734245e-05,
+      "loss": 0.0681,
+      "step": 972
+    },
+    {
+      "epoch": 33.58620689655172,
+      "grad_norm": 0.12361834198236465,
+      "learning_rate": 2.819420783645656e-05,
+      "loss": 0.03,
+      "step": 974
+    },
+    {
+      "epoch": 33.6551724137931,
+      "grad_norm": 0.12781710922718048,
+      "learning_rate": 2.8023850085178878e-05,
+      "loss": 0.025,
+      "step": 976
+    },
+    {
+      "epoch": 33.724137931034484,
+      "grad_norm": 0.10559111833572388,
+      "learning_rate": 2.7853492333901192e-05,
+      "loss": 0.0261,
+      "step": 978
+    },
+    {
+      "epoch": 33.793103448275865,
+      "grad_norm": 0.08984573930501938,
+      "learning_rate": 2.768313458262351e-05,
+      "loss": 0.0244,
+      "step": 980
+    },
+    {
+      "epoch": 33.86206896551724,
+      "grad_norm": 0.08784312754869461,
+      "learning_rate": 2.7512776831345826e-05,
+      "loss": 0.0283,
+      "step": 982
+    },
+    {
+      "epoch": 33.93103448275862,
+      "grad_norm": 0.09832680970430374,
+      "learning_rate": 2.7342419080068144e-05,
+      "loss": 0.0278,
+      "step": 984
+    },
+    {
+      "epoch": 34.0,
+      "grad_norm": 0.07693750411272049,
+      "learning_rate": 2.717206132879046e-05,
+      "loss": 0.0224,
+      "step": 986
+    },
+    {
+      "epoch": 34.0,
+      "eval_accuracy": 0.6742424242424242,
+      "eval_f1_macro": 0.5794868578213456,
+      "eval_f1_micro": 0.6742424242424242,
+      "eval_f1_weighted": 0.65445260813248,
+      "eval_loss": 1.4700735807418823,
+      "eval_precision_macro": 0.5855705141419428,
+      "eval_precision_micro": 0.6742424242424242,
+      "eval_precision_weighted": 0.652266504539232,
+      "eval_recall_macro": 0.5901965230536659,
+      "eval_recall_micro": 0.6742424242424242,
+      "eval_recall_weighted": 0.6742424242424242,
+      "eval_runtime": 2.4834,
+      "eval_samples_per_second": 53.154,
+      "eval_steps_per_second": 6.846,
+      "step": 986
+    },
+    {
+      "epoch": 34.06896551724138,
+      "grad_norm": 0.08656442165374756,
+      "learning_rate": 2.7001703577512777e-05,
+      "loss": 0.0231,
+      "step": 988
+    },
+    {
+      "epoch": 34.13793103448276,
+      "grad_norm": 0.10006435215473175,
+      "learning_rate": 2.6831345826235095e-05,
+      "loss": 0.0233,
+      "step": 990
+    },
+    {
+      "epoch": 34.206896551724135,
+      "grad_norm": 0.10205938667058945,
+      "learning_rate": 2.666098807495741e-05,
+      "loss": 0.0269,
+      "step": 992
+    },
+    {
+      "epoch": 34.275862068965516,
+      "grad_norm": 0.14712445437908173,
+      "learning_rate": 2.6490630323679728e-05,
+      "loss": 0.0263,
+      "step": 994
+    },
+    {
+      "epoch": 34.3448275862069,
+      "grad_norm": 0.476298987865448,
+      "learning_rate": 2.6320272572402043e-05,
+      "loss": 0.0529,
+      "step": 996
+    },
+    {
+      "epoch": 34.41379310344828,
+      "grad_norm": 0.08625518530607224,
+      "learning_rate": 2.614991482112436e-05,
+      "loss": 0.0242,
+      "step": 998
+    },
+    {
+      "epoch": 34.48275862068966,
+      "grad_norm": 0.1275705099105835,
+      "learning_rate": 2.5979557069846676e-05,
+      "loss": 0.0264,
+      "step": 1000
+    },
+    {
+      "epoch": 34.55172413793103,
+      "grad_norm": 0.07826410233974457,
+      "learning_rate": 2.5809199318568994e-05,
+      "loss": 0.0227,
+      "step": 1002
+    },
+    {
+      "epoch": 34.62068965517241,
+      "grad_norm": 0.3700888454914093,
+      "learning_rate": 2.5638841567291312e-05,
+      "loss": 0.0303,
+      "step": 1004
+    },
+    {
+      "epoch": 34.689655172413794,
+      "grad_norm": 0.15820983052253723,
+      "learning_rate": 2.5468483816013627e-05,
+      "loss": 0.0253,
+      "step": 1006
+    },
+    {
+      "epoch": 34.758620689655174,
+      "grad_norm": 0.11389223486185074,
+      "learning_rate": 2.5298126064735945e-05,
+      "loss": 0.0247,
+      "step": 1008
+    },
+    {
+      "epoch": 34.827586206896555,
+      "grad_norm": 0.08679822832345963,
+      "learning_rate": 2.512776831345826e-05,
+      "loss": 0.0273,
+      "step": 1010
+    },
+    {
+      "epoch": 34.89655172413793,
+      "grad_norm": 0.47994059324264526,
+      "learning_rate": 2.495741056218058e-05,
+      "loss": 0.0523,
+      "step": 1012
+    },
+    {
+      "epoch": 34.96551724137931,
+      "grad_norm": 0.08631986379623413,
+      "learning_rate": 2.4787052810902896e-05,
+      "loss": 0.0228,
+      "step": 1014
+    },
+    {
+      "epoch": 35.0,
+      "eval_accuracy": 0.6590909090909091,
+      "eval_f1_macro": 0.5772129682250426,
+      "eval_f1_micro": 0.6590909090909091,
+      "eval_f1_weighted": 0.6488555790270587,
+      "eval_loss": 1.4697364568710327,
+      "eval_precision_macro": 0.5869569578494528,
+      "eval_precision_micro": 0.6590909090909091,
+      "eval_precision_weighted": 0.649717549185095,
+      "eval_recall_macro": 0.577384731670446,
+      "eval_recall_micro": 0.6590909090909091,
+      "eval_recall_weighted": 0.6590909090909091,
+      "eval_runtime": 2.3103,
+      "eval_samples_per_second": 57.135,
+      "eval_steps_per_second": 7.358,
+      "step": 1015
+    },
+    {
+      "epoch": 35.03448275862069,
+      "grad_norm": 0.16488300263881683,
+      "learning_rate": 2.4616695059625215e-05,
+      "loss": 0.0273,
+      "step": 1016
+    },
+    {
+      "epoch": 35.10344827586207,
+      "grad_norm": 0.5682753920555115,
+      "learning_rate": 2.444633730834753e-05,
+      "loss": 0.0278,
+      "step": 1018
+    },
+    {
+      "epoch": 35.172413793103445,
+      "grad_norm": 0.4378216564655304,
+      "learning_rate": 2.4275979557069848e-05,
+      "loss": 0.0508,
+      "step": 1020
+    },
+    {
+      "epoch": 35.241379310344826,
+      "grad_norm": 0.14175258576869965,
+      "learning_rate": 2.4105621805792162e-05,
+      "loss": 0.0235,
+      "step": 1022
+    },
+    {
+      "epoch": 35.310344827586206,
+      "grad_norm": 0.08108456432819366,
+      "learning_rate": 2.393526405451448e-05,
+      "loss": 0.0218,
+      "step": 1024
+    },
+    {
+      "epoch": 35.37931034482759,
+      "grad_norm": 0.08066286891698837,
+      "learning_rate": 2.37649063032368e-05,
+      "loss": 0.0212,
+      "step": 1026
+    },
+    {
+      "epoch": 35.44827586206897,
+      "grad_norm": 0.14611414074897766,
+      "learning_rate": 2.3594548551959114e-05,
+      "loss": 0.0252,
+      "step": 1028
+    },
+    {
+      "epoch": 35.51724137931034,
+      "grad_norm": 2.4684391021728516,
+      "learning_rate": 2.3424190800681432e-05,
+      "loss": 0.0292,
+      "step": 1030
+    },
+    {
+      "epoch": 35.58620689655172,
+      "grad_norm": 0.43363311886787415,
+      "learning_rate": 2.3253833049403747e-05,
+      "loss": 0.0469,
+      "step": 1032
+    },
+    {
+      "epoch": 35.6551724137931,
+      "grad_norm": 0.0832495167851448,
+      "learning_rate": 2.3083475298126068e-05,
+      "loss": 0.0259,
+      "step": 1034
+    },
+    {
+      "epoch": 35.724137931034484,
+      "grad_norm": 1.9775596857070923,
+      "learning_rate": 2.2913117546848383e-05,
+      "loss": 0.0303,
+      "step": 1036
+    },
+    {
+      "epoch": 35.793103448275865,
+      "grad_norm": 0.08929072320461273,
+      "learning_rate": 2.27427597955707e-05,
+      "loss": 0.0238,
+      "step": 1038
+    },
+    {
+      "epoch": 35.86206896551724,
+      "grad_norm": 0.0937528908252716,
+      "learning_rate": 2.2572402044293016e-05,
+      "loss": 0.0258,
+      "step": 1040
+    },
+    {
+      "epoch": 35.93103448275862,
+      "grad_norm": 0.07003093510866165,
+      "learning_rate": 2.2402044293015334e-05,
+      "loss": 0.0218,
+      "step": 1042
+    },
+    {
+      "epoch": 36.0,
+      "grad_norm": 0.11906813085079193,
+      "learning_rate": 2.223168654173765e-05,
+      "loss": 0.0231,
+      "step": 1044
+    },
+    {
+      "epoch": 36.0,
+      "eval_accuracy": 0.6590909090909091,
+      "eval_f1_macro": 0.5744938259848144,
+      "eval_f1_micro": 0.6590909090909091,
+      "eval_f1_weighted": 0.6491010450687431,
+      "eval_loss": 1.5315077304840088,
+      "eval_precision_macro": 0.5783361246596541,
+      "eval_precision_micro": 0.6590909090909091,
+      "eval_precision_weighted": 0.648294506666833,
+      "eval_recall_macro": 0.5788133030990174,
+      "eval_recall_micro": 0.6590909090909091,
+      "eval_recall_weighted": 0.6590909090909091,
+      "eval_runtime": 2.314,
+      "eval_samples_per_second": 57.045,
+      "eval_steps_per_second": 7.347,
+      "step": 1044
+    },
+    {
+      "epoch": 36.06896551724138,
+      "grad_norm": 1.3290965557098389,
+      "learning_rate": 2.2061328790459967e-05,
+      "loss": 0.028,
+      "step": 1046
+    },
+    {
+      "epoch": 36.13793103448276,
+      "grad_norm": 0.1269829422235489,
+      "learning_rate": 2.1890971039182286e-05,
+      "loss": 0.0232,
+      "step": 1048
+    },
+    {
+      "epoch": 36.206896551724135,
+      "grad_norm": 0.14340566098690033,
+      "learning_rate": 2.17206132879046e-05,
+      "loss": 0.0223,
+      "step": 1050
+    },
+    {
+      "epoch": 36.275862068965516,
+      "grad_norm": 4.602632999420166,
+      "learning_rate": 2.155025553662692e-05,
+      "loss": 0.1708,
+      "step": 1052
+    },
+    {
+      "epoch": 36.3448275862069,
+      "grad_norm": 0.09221786260604858,
+      "learning_rate": 2.1379897785349233e-05,
+      "loss": 0.0343,
+      "step": 1054
+    },
+    {
+      "epoch": 36.41379310344828,
+      "grad_norm": 0.11092764139175415,
+      "learning_rate": 2.120954003407155e-05,
+      "loss": 0.0253,
+      "step": 1056
+    },
+    {
+      "epoch": 36.48275862068966,
+      "grad_norm": 0.09004894644021988,
+      "learning_rate": 2.1039182282793866e-05,
+      "loss": 0.0223,
+      "step": 1058
+    },
+    {
+      "epoch": 36.55172413793103,
+      "grad_norm": 0.08809762448072433,
+      "learning_rate": 2.0868824531516185e-05,
+      "loss": 0.0229,
+      "step": 1060
+    },
+    {
+      "epoch": 36.62068965517241,
+      "grad_norm": 0.08181875944137573,
+      "learning_rate": 2.0698466780238503e-05,
+      "loss": 0.0224,
+      "step": 1062
+    },
+    {
+      "epoch": 36.689655172413794,
+      "grad_norm": 0.11032210290431976,
+      "learning_rate": 2.0528109028960818e-05,
+      "loss": 0.022,
+      "step": 1064
+    },
+    {
+      "epoch": 36.758620689655174,
+      "grad_norm": 0.07693520188331604,
+      "learning_rate": 2.0357751277683136e-05,
+      "loss": 0.0206,
+      "step": 1066
+    },
+    {
+      "epoch": 36.827586206896555,
+      "grad_norm": 0.07412053644657135,
+      "learning_rate": 2.018739352640545e-05,
+      "loss": 0.0212,
+      "step": 1068
+    },
+    {
+      "epoch": 36.89655172413793,
+      "grad_norm": 4.513420104980469,
+      "learning_rate": 2.001703577512777e-05,
+      "loss": 0.0467,
+      "step": 1070
+    },
+    {
+      "epoch": 36.96551724137931,
+      "grad_norm": 0.08167243748903275,
+      "learning_rate": 1.9846678023850084e-05,
+      "loss": 0.0457,
+      "step": 1072
+    },
+    {
+      "epoch": 37.0,
+      "eval_accuracy": 0.6439393939393939,
+      "eval_f1_macro": 0.5532375287737503,
+      "eval_f1_micro": 0.6439393939393939,
+      "eval_f1_weighted": 0.6276644798060326,
+      "eval_loss": 1.520954966545105,
+      "eval_precision_macro": 0.5641351058808365,
+      "eval_precision_micro": 0.6439393939393939,
+      "eval_precision_weighted": 0.6317489292950739,
+      "eval_recall_macro": 0.5605820105820106,
+      "eval_recall_micro": 0.6439393939393939,
+      "eval_recall_weighted": 0.6439393939393939,
+      "eval_runtime": 2.3121,
+      "eval_samples_per_second": 57.091,
+      "eval_steps_per_second": 7.353,
+      "step": 1073
+    },
+    {
+      "epoch": 37.03448275862069,
+      "grad_norm": 0.3906444311141968,
+      "learning_rate": 1.9676320272572402e-05,
+      "loss": 0.0419,
+      "step": 1074
+    },
+    {
+      "epoch": 37.10344827586207,
+      "grad_norm": 0.36884042620658875,
+      "learning_rate": 1.950596252129472e-05,
+      "loss": 0.0413,
+      "step": 1076
+    },
+    {
+      "epoch": 37.172413793103445,
+      "grad_norm": 0.07680459320545197,
+      "learning_rate": 1.9335604770017038e-05,
+      "loss": 0.02,
+      "step": 1078
+    },
+    {
+      "epoch": 37.241379310344826,
+      "grad_norm": 0.07265330106019974,
+      "learning_rate": 1.9165247018739353e-05,
+      "loss": 0.0228,
+      "step": 1080
+    },
+    {
+      "epoch": 37.310344827586206,
+      "grad_norm": 0.15032625198364258,
+      "learning_rate": 1.899488926746167e-05,
+      "loss": 0.0223,
+      "step": 1082
+    },
+    {
+      "epoch": 37.37931034482759,
+      "grad_norm": 1.7580552101135254,
+      "learning_rate": 1.882453151618399e-05,
+      "loss": 0.0259,
+      "step": 1084
+    },
+    {
+      "epoch": 37.44827586206897,
+      "grad_norm": 0.07701607793569565,
+      "learning_rate": 1.8654173764906304e-05,
+      "loss": 0.0211,
+      "step": 1086
+    },
+    {
+      "epoch": 37.51724137931034,
+      "grad_norm": 0.13334742188453674,
+      "learning_rate": 1.8483816013628622e-05,
+      "loss": 0.0281,
+      "step": 1088
+    },
+    {
+      "epoch": 37.58620689655172,
+      "grad_norm": 0.08021260797977448,
+      "learning_rate": 1.8313458262350937e-05,
+      "loss": 0.0199,
+      "step": 1090
+    },
+    {
+      "epoch": 37.6551724137931,
+      "grad_norm": 0.06848295032978058,
+      "learning_rate": 1.8143100511073255e-05,
+      "loss": 0.021,
+      "step": 1092
+    },
+    {
+      "epoch": 37.724137931034484,
+      "grad_norm": 0.09606592357158661,
+      "learning_rate": 1.797274275979557e-05,
+      "loss": 0.0498,
+      "step": 1094
+    },
+    {
+      "epoch": 37.793103448275865,
+      "grad_norm": 0.09107226878404617,
+      "learning_rate": 1.780238500851789e-05,
+      "loss": 0.022,
+      "step": 1096
+    },
+    {
+      "epoch": 37.86206896551724,
+      "grad_norm": 0.07415739446878433,
+      "learning_rate": 1.7632027257240207e-05,
+      "loss": 0.02,
+      "step": 1098
+    },
+    {
+      "epoch": 37.93103448275862,
+      "grad_norm": 0.10790450870990753,
+      "learning_rate": 1.746166950596252e-05,
+      "loss": 0.0219,
+      "step": 1100
+    },
+    {
+      "epoch": 38.0,
+      "grad_norm": 0.07875606417655945,
+      "learning_rate": 1.729131175468484e-05,
+      "loss": 0.0197,
+      "step": 1102
+    },
+    {
+      "epoch": 38.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5636123898806141,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6385926013380662,
+      "eval_loss": 1.4955607652664185,
+      "eval_precision_macro": 0.5590277777777778,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.6296033249158249,
+      "eval_recall_macro": 0.5713756613756613,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3054,
+      "eval_samples_per_second": 57.256,
+      "eval_steps_per_second": 7.374,
+      "step": 1102
+    },
+    {
+      "epoch": 38.06896551724138,
+      "grad_norm": 0.10304208844900131,
+      "learning_rate": 1.7120954003407155e-05,
+      "loss": 0.0228,
+      "step": 1104
+    },
+    {
+      "epoch": 38.13793103448276,
+      "grad_norm": 0.07519730180501938,
+      "learning_rate": 1.6950596252129473e-05,
+      "loss": 0.0199,
+      "step": 1106
+    },
+    {
+      "epoch": 38.206896551724135,
+      "grad_norm": 0.07089941203594208,
+      "learning_rate": 1.6780238500851788e-05,
+      "loss": 0.0216,
+      "step": 1108
+    },
+    {
+      "epoch": 38.275862068965516,
+      "grad_norm": 0.06765555590391159,
+      "learning_rate": 1.6609880749574106e-05,
+      "loss": 0.043,
+      "step": 1110
+    },
+    {
+      "epoch": 38.3448275862069,
+      "grad_norm": 0.07931018620729446,
+      "learning_rate": 1.643952299829642e-05,
+      "loss": 0.0237,
+      "step": 1112
+    },
+    {
+      "epoch": 38.41379310344828,
+      "grad_norm": 0.46399685740470886,
+      "learning_rate": 1.626916524701874e-05,
+      "loss": 0.0493,
+      "step": 1114
+    },
+    {
+      "epoch": 38.48275862068966,
+      "grad_norm": 0.07535680383443832,
+      "learning_rate": 1.6098807495741057e-05,
+      "loss": 0.0206,
+      "step": 1116
+    },
+    {
+      "epoch": 38.55172413793103,
+      "grad_norm": 0.07040387392044067,
+      "learning_rate": 1.5928449744463375e-05,
+      "loss": 0.0207,
+      "step": 1118
+    },
+    {
+      "epoch": 38.62068965517241,
+      "grad_norm": 0.06553125381469727,
+      "learning_rate": 1.5758091993185693e-05,
+      "loss": 0.0203,
+      "step": 1120
+    },
+    {
+      "epoch": 38.689655172413794,
+      "grad_norm": 0.10302378237247467,
+      "learning_rate": 1.5587734241908008e-05,
+      "loss": 0.02,
+      "step": 1122
+    },
+    {
+      "epoch": 38.758620689655174,
+      "grad_norm": 4.081165790557861,
+      "learning_rate": 1.5417376490630326e-05,
+      "loss": 0.0323,
+      "step": 1124
+    },
+    {
+      "epoch": 38.827586206896555,
+      "grad_norm": 0.435150682926178,
+      "learning_rate": 1.5247018739352641e-05,
+      "loss": 0.044,
+      "step": 1126
+    },
+    {
+      "epoch": 38.89655172413793,
+      "grad_norm": 0.07893764227628708,
+      "learning_rate": 1.507666098807496e-05,
+      "loss": 0.0192,
+      "step": 1128
+    },
+    {
+      "epoch": 38.96551724137931,
+      "grad_norm": 0.0766209065914154,
+      "learning_rate": 1.4906303236797276e-05,
+      "loss": 0.0219,
+      "step": 1130
+    },
+    {
+      "epoch": 39.0,
+      "eval_accuracy": 0.6590909090909091,
+      "eval_f1_macro": 0.5981099582690218,
+      "eval_f1_micro": 0.6590909090909091,
+      "eval_f1_weighted": 0.6540264969624213,
+      "eval_loss": 1.4909534454345703,
+      "eval_precision_macro": 0.6063481453187336,
+      "eval_precision_micro": 0.6590909090909091,
+      "eval_precision_weighted": 0.6553739264300761,
+      "eval_recall_macro": 0.5969538926681783,
+      "eval_recall_micro": 0.6590909090909091,
+      "eval_recall_weighted": 0.6590909090909091,
+      "eval_runtime": 2.3146,
+      "eval_samples_per_second": 57.029,
+      "eval_steps_per_second": 7.345,
+      "step": 1131
+    },
+    {
+      "epoch": 39.03448275862069,
+      "grad_norm": 0.0769796147942543,
+      "learning_rate": 1.4735945485519592e-05,
+      "loss": 0.0194,
+      "step": 1132
+    },
+    {
+      "epoch": 39.10344827586207,
+      "grad_norm": 0.07249811291694641,
+      "learning_rate": 1.4565587734241909e-05,
+      "loss": 0.0198,
+      "step": 1134
+    },
+    {
+      "epoch": 39.172413793103445,
+      "grad_norm": 0.07959507405757904,
+      "learning_rate": 1.4395229982964225e-05,
+      "loss": 0.022,
+      "step": 1136
+    },
+    {
+      "epoch": 39.241379310344826,
+      "grad_norm": 0.44026121497154236,
+      "learning_rate": 1.4224872231686542e-05,
+      "loss": 0.0457,
+      "step": 1138
+    },
+    {
+      "epoch": 39.310344827586206,
+      "grad_norm": 0.072527214884758,
+      "learning_rate": 1.4054514480408858e-05,
+      "loss": 0.0208,
+      "step": 1140
+    },
+    {
+      "epoch": 39.37931034482759,
+      "grad_norm": 0.22599418461322784,
+      "learning_rate": 1.3884156729131177e-05,
+      "loss": 0.0199,
+      "step": 1142
+    },
+    {
+      "epoch": 39.44827586206897,
+      "grad_norm": 0.07740206271409988,
+      "learning_rate": 1.3713798977853493e-05,
+      "loss": 0.0195,
+      "step": 1144
+    },
+    {
+      "epoch": 39.51724137931034,
+      "grad_norm": 0.07508723437786102,
+      "learning_rate": 1.354344122657581e-05,
+      "loss": 0.0188,
+      "step": 1146
+    },
+    {
+      "epoch": 39.58620689655172,
+      "grad_norm": 0.07625211030244827,
+      "learning_rate": 1.3373083475298126e-05,
+      "loss": 0.0231,
+      "step": 1148
+    },
+    {
+      "epoch": 39.6551724137931,
+      "grad_norm": 0.0625428780913353,
+      "learning_rate": 1.3202725724020443e-05,
+      "loss": 0.0184,
+      "step": 1150
+    },
+    {
+      "epoch": 39.724137931034484,
+      "grad_norm": 0.06486790627241135,
+      "learning_rate": 1.303236797274276e-05,
+      "loss": 0.0185,
+      "step": 1152
+    },
+    {
+      "epoch": 39.793103448275865,
+      "grad_norm": 0.09792531281709671,
+      "learning_rate": 1.2862010221465076e-05,
+      "loss": 0.0223,
+      "step": 1154
+    },
+    {
+      "epoch": 39.86206896551724,
+      "grad_norm": 0.5140519142150879,
+      "learning_rate": 1.2691652470187392e-05,
+      "loss": 0.052,
+      "step": 1156
+    },
+    {
+      "epoch": 39.93103448275862,
+      "grad_norm": 0.10591938346624374,
+      "learning_rate": 1.252129471890971e-05,
+      "loss": 0.0198,
+      "step": 1158
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 0.08880186080932617,
+      "learning_rate": 1.2350936967632027e-05,
+      "loss": 0.0212,
+      "step": 1160
+    },
+    {
+      "epoch": 40.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5911823960353372,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6461534176607706,
+      "eval_loss": 1.5049692392349243,
+      "eval_precision_macro": 0.5996689473237461,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.6472166962333723,
+      "eval_recall_macro": 0.5898110355253213,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3355,
+      "eval_samples_per_second": 56.519,
+      "eval_steps_per_second": 7.279,
+      "step": 1160
+    },
+    {
+      "epoch": 40.06896551724138,
+      "grad_norm": 0.06325220316648483,
+      "learning_rate": 1.2180579216354345e-05,
+      "loss": 0.0191,
+      "step": 1162
+    },
+    {
+      "epoch": 40.13793103448276,
+      "grad_norm": 0.0681692361831665,
+      "learning_rate": 1.2010221465076662e-05,
+      "loss": 0.0181,
+      "step": 1164
+    },
+    {
+      "epoch": 40.206896551724135,
+      "grad_norm": 0.14537657797336578,
+      "learning_rate": 1.1839863713798978e-05,
+      "loss": 0.0198,
+      "step": 1166
+    },
+    {
+      "epoch": 40.275862068965516,
+      "grad_norm": 0.060531288385391235,
+      "learning_rate": 1.1669505962521295e-05,
+      "loss": 0.0177,
+      "step": 1168
+    },
+    {
+      "epoch": 40.3448275862069,
+      "grad_norm": 0.07529687881469727,
+      "learning_rate": 1.1499148211243613e-05,
+      "loss": 0.0212,
+      "step": 1170
+    },
+    {
+      "epoch": 40.41379310344828,
+      "grad_norm": 0.37628257274627686,
+      "learning_rate": 1.132879045996593e-05,
+      "loss": 0.0398,
+      "step": 1172
+    },
+    {
+      "epoch": 40.48275862068966,
+      "grad_norm": 0.15280403196811676,
+      "learning_rate": 1.1158432708688246e-05,
+      "loss": 0.0208,
+      "step": 1174
+    },
+    {
+      "epoch": 40.55172413793103,
+      "grad_norm": 0.06886907666921616,
+      "learning_rate": 1.0988074957410562e-05,
+      "loss": 0.0213,
+      "step": 1176
+    },
+    {
+      "epoch": 40.62068965517241,
+      "grad_norm": 0.481803834438324,
+      "learning_rate": 1.0817717206132879e-05,
+      "loss": 0.0478,
+      "step": 1178
+    },
+    {
+      "epoch": 40.689655172413794,
+      "grad_norm": 0.06786196678876877,
+      "learning_rate": 1.0647359454855197e-05,
+      "loss": 0.02,
+      "step": 1180
+    },
+    {
+      "epoch": 40.758620689655174,
+      "grad_norm": 0.07694634050130844,
+      "learning_rate": 1.0477001703577514e-05,
+      "loss": 0.0196,
+      "step": 1182
+    },
+    {
+      "epoch": 40.827586206896555,
+      "grad_norm": 0.06034290790557861,
+      "learning_rate": 1.030664395229983e-05,
+      "loss": 0.0209,
+      "step": 1184
+    },
+    {
+      "epoch": 40.89655172413793,
+      "grad_norm": 0.06212831288576126,
+      "learning_rate": 1.0136286201022147e-05,
+      "loss": 0.0189,
+      "step": 1186
+    },
+    {
+      "epoch": 40.96551724137931,
+      "grad_norm": 0.07599721103906631,
+      "learning_rate": 9.965928449744463e-06,
+      "loss": 0.0212,
+      "step": 1188
+    },
+    {
+      "epoch": 41.0,
+      "eval_accuracy": 0.6590909090909091,
+      "eval_f1_macro": 0.5976809954751132,
+      "eval_f1_micro": 0.6590909090909091,
+      "eval_f1_weighted": 0.6537291752365282,
+      "eval_loss": 1.5091404914855957,
+      "eval_precision_macro": 0.6079748081296069,
+      "eval_precision_micro": 0.6590909090909091,
+      "eval_precision_weighted": 0.6557976985643745,
+      "eval_recall_macro": 0.5955253212396069,
+      "eval_recall_micro": 0.6590909090909091,
+      "eval_recall_weighted": 0.6590909090909091,
+      "eval_runtime": 2.3163,
+      "eval_samples_per_second": 56.988,
+      "eval_steps_per_second": 7.339,
+      "step": 1189
+    },
+    {
+      "epoch": 41.03448275862069,
+      "grad_norm": 0.07553800195455551,
+      "learning_rate": 9.795570698466781e-06,
+      "loss": 0.0202,
+      "step": 1190
+    },
+    {
+      "epoch": 41.10344827586207,
+      "grad_norm": 0.08795283734798431,
+      "learning_rate": 9.625212947189098e-06,
+      "loss": 0.0205,
+      "step": 1192
+    },
+    {
+      "epoch": 41.172413793103445,
+      "grad_norm": 0.08366087824106216,
+      "learning_rate": 9.454855195911414e-06,
+      "loss": 0.0202,
+      "step": 1194
+    },
+    {
+      "epoch": 41.241379310344826,
+      "grad_norm": 0.5255555510520935,
+      "learning_rate": 9.28449744463373e-06,
+      "loss": 0.0414,
+      "step": 1196
+    },
+    {
+      "epoch": 41.310344827586206,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.114139693356049e-06,
+      "loss": 0.0384,
+      "step": 1198
+    },
+    {
+      "epoch": 41.37931034482759,
+      "grad_norm": 0.07240649312734604,
+      "learning_rate": 8.943781942078366e-06,
+      "loss": 0.0203,
+      "step": 1200
+    },
+    {
+      "epoch": 41.44827586206897,
+      "grad_norm": 0.07140922546386719,
+      "learning_rate": 8.773424190800682e-06,
+      "loss": 0.0179,
+      "step": 1202
+    },
+    {
+      "epoch": 41.51724137931034,
+      "grad_norm": 1.3324774503707886,
+      "learning_rate": 8.603066439522999e-06,
+      "loss": 0.025,
+      "step": 1204
+    },
+    {
+      "epoch": 41.58620689655172,
+      "grad_norm": 0.0719638466835022,
+      "learning_rate": 8.432708688245315e-06,
+      "loss": 0.0196,
+      "step": 1206
+    },
+    {
+      "epoch": 41.6551724137931,
+      "grad_norm": 0.06998977810144424,
+      "learning_rate": 8.262350936967632e-06,
+      "loss": 0.0208,
+      "step": 1208
+    },
+    {
+      "epoch": 41.724137931034484,
+      "grad_norm": 0.06412280350923538,
+      "learning_rate": 8.091993185689948e-06,
+      "loss": 0.0207,
+      "step": 1210
+    },
+    {
+      "epoch": 41.793103448275865,
+      "grad_norm": 0.06555226445198059,
+      "learning_rate": 7.921635434412266e-06,
+      "loss": 0.0202,
+      "step": 1212
+    },
+    {
+      "epoch": 41.86206896551724,
+      "grad_norm": 0.06935158371925354,
+      "learning_rate": 7.751277683134583e-06,
+      "loss": 0.018,
+      "step": 1214
+    },
+    {
+      "epoch": 41.93103448275862,
+      "grad_norm": 0.1250106245279312,
+      "learning_rate": 7.5809199318569e-06,
+      "loss": 0.019,
+      "step": 1216
+    },
+    {
+      "epoch": 42.0,
+      "grad_norm": 0.06154052913188934,
+      "learning_rate": 7.410562180579217e-06,
+      "loss": 0.0202,
+      "step": 1218
+    },
+    {
+      "epoch": 42.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5654648815113407,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6411283473548656,
+      "eval_loss": 1.4960678815841675,
+      "eval_precision_macro": 0.5707833560774738,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.6411209251895527,
+      "eval_recall_macro": 0.5694935752078608,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3136,
+      "eval_samples_per_second": 57.055,
+      "eval_steps_per_second": 7.348,
+      "step": 1218
+    },
+    {
+      "epoch": 42.06896551724138,
+      "grad_norm": 0.10921443998813629,
+      "learning_rate": 7.240204429301534e-06,
+      "loss": 0.0216,
+      "step": 1220
+    },
+    {
+      "epoch": 42.13793103448276,
+      "grad_norm": 0.08505412936210632,
+      "learning_rate": 7.0698466780238505e-06,
+      "loss": 0.0181,
+      "step": 1222
+    },
+    {
+      "epoch": 42.206896551724135,
+      "grad_norm": 0.06461058557033539,
+      "learning_rate": 6.899488926746167e-06,
+      "loss": 0.0437,
+      "step": 1224
+    },
+    {
+      "epoch": 42.275862068965516,
+      "grad_norm": 0.0641995221376419,
+      "learning_rate": 6.7291311754684835e-06,
+      "loss": 0.0181,
+      "step": 1226
+    },
+    {
+      "epoch": 42.3448275862069,
+      "grad_norm": 0.07062099874019623,
+      "learning_rate": 6.558773424190801e-06,
+      "loss": 0.0195,
+      "step": 1228
+    },
+    {
+      "epoch": 42.41379310344828,
+      "grad_norm": 0.06934709846973419,
+      "learning_rate": 6.388415672913117e-06,
+      "loss": 0.0197,
+      "step": 1230
+    },
+    {
+      "epoch": 42.48275862068966,
+      "grad_norm": 0.07409219443798065,
+      "learning_rate": 6.218057921635435e-06,
+      "loss": 0.0195,
+      "step": 1232
+    },
+    {
+      "epoch": 42.55172413793103,
+      "grad_norm": 0.07015547901391983,
+      "learning_rate": 6.047700170357751e-06,
+      "loss": 0.0478,
+      "step": 1234
+    },
+    {
+      "epoch": 42.62068965517241,
+      "grad_norm": 0.06233002990484238,
+      "learning_rate": 5.877342419080068e-06,
+      "loss": 0.0181,
+      "step": 1236
+    },
+    {
+      "epoch": 42.689655172413794,
+      "grad_norm": 0.07605413347482681,
+      "learning_rate": 5.706984667802386e-06,
+      "loss": 0.0195,
+      "step": 1238
+    },
+    {
+      "epoch": 42.758620689655174,
+      "grad_norm": 0.06724195927381516,
+      "learning_rate": 5.5366269165247025e-06,
+      "loss": 0.0197,
+      "step": 1240
+    },
+    {
+      "epoch": 42.827586206896555,
+      "grad_norm": 0.0727328211069107,
+      "learning_rate": 5.366269165247019e-06,
+      "loss": 0.0199,
+      "step": 1242
+    },
+    {
+      "epoch": 42.89655172413793,
+      "grad_norm": 0.061808932572603226,
+      "learning_rate": 5.1959114139693355e-06,
+      "loss": 0.0181,
+      "step": 1244
+    },
+    {
+      "epoch": 42.96551724137931,
+      "grad_norm": 0.0680130198597908,
+      "learning_rate": 5.025553662691653e-06,
+      "loss": 0.0216,
+      "step": 1246
+    },
+    {
+      "epoch": 43.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5654648815113407,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6411283473548656,
+      "eval_loss": 1.491672396659851,
+      "eval_precision_macro": 0.5707833560774738,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.6411209251895527,
+      "eval_recall_macro": 0.5694935752078608,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3103,
+      "eval_samples_per_second": 57.136,
+      "eval_steps_per_second": 7.358,
+      "step": 1247
+    },
+    {
+      "epoch": 43.03448275862069,
+      "grad_norm": 0.06223515793681145,
+      "learning_rate": 4.85519591141397e-06,
+      "loss": 0.0179,
+      "step": 1248
+    },
+    {
+      "epoch": 43.10344827586207,
+      "grad_norm": 0.0637139230966568,
+      "learning_rate": 4.684838160136287e-06,
+      "loss": 0.0196,
+      "step": 1250
+    },
+    {
+      "epoch": 43.172413793103445,
+      "grad_norm": 0.06648910790681839,
+      "learning_rate": 4.514480408858603e-06,
+      "loss": 0.0376,
+      "step": 1252
+    },
+    {
+      "epoch": 43.241379310344826,
+      "grad_norm": 0.08096196502447128,
+      "learning_rate": 4.34412265758092e-06,
+      "loss": 0.045,
+      "step": 1254
+    },
+    {
+      "epoch": 43.310344827586206,
+      "grad_norm": 0.0885854959487915,
+      "learning_rate": 4.173764906303237e-06,
+      "loss": 0.0206,
+      "step": 1256
+    },
+    {
+      "epoch": 43.37931034482759,
+      "grad_norm": 0.0679527223110199,
+      "learning_rate": 4.0034071550255544e-06,
+      "loss": 0.0185,
+      "step": 1258
+    },
+    {
+      "epoch": 43.44827586206897,
+      "grad_norm": 0.07362633943557739,
+      "learning_rate": 3.833049403747871e-06,
+      "loss": 0.02,
+      "step": 1260
+    },
+    {
+      "epoch": 43.51724137931034,
+      "grad_norm": 0.0694928914308548,
+      "learning_rate": 3.6626916524701875e-06,
+      "loss": 0.0196,
+      "step": 1262
+    },
+    {
+      "epoch": 43.58620689655172,
+      "grad_norm": 0.06688851863145828,
+      "learning_rate": 3.4923339011925044e-06,
+      "loss": 0.0179,
+      "step": 1264
+    },
+    {
+      "epoch": 43.6551724137931,
+      "grad_norm": 0.11848729103803635,
+      "learning_rate": 3.321976149914821e-06,
+      "loss": 0.0219,
+      "step": 1266
+    },
+    {
+      "epoch": 43.724137931034484,
+      "grad_norm": 0.9978265762329102,
+      "learning_rate": 3.151618398637138e-06,
+      "loss": 0.0194,
+      "step": 1268
+    },
+    {
+      "epoch": 43.793103448275865,
+      "grad_norm": 0.06428433954715729,
+      "learning_rate": 2.9812606473594548e-06,
+      "loss": 0.0182,
+      "step": 1270
+    },
+    {
+      "epoch": 43.86206896551724,
+      "grad_norm": 0.20028172433376312,
+      "learning_rate": 2.810902896081772e-06,
+      "loss": 0.0195,
+      "step": 1272
+    },
+    {
+      "epoch": 43.93103448275862,
+      "grad_norm": 0.07131318747997284,
+      "learning_rate": 2.6405451448040886e-06,
+      "loss": 0.0183,
+      "step": 1274
+    },
+    {
+      "epoch": 44.0,
+      "grad_norm": 0.13960784673690796,
+      "learning_rate": 2.4701873935264056e-06,
+      "loss": 0.0199,
+      "step": 1276
+    },
+    {
+      "epoch": 44.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5674399966365533,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6423054871360109,
+      "eval_loss": 1.485514760017395,
+      "eval_precision_macro": 0.5694260552951398,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.6400870203536964,
+      "eval_recall_macro": 0.5717157974300832,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3079,
+      "eval_samples_per_second": 57.194,
+      "eval_steps_per_second": 7.366,
+      "step": 1276
+    },
+    {
+      "epoch": 44.06896551724138,
+      "grad_norm": 0.09822448343038559,
+      "learning_rate": 2.2998296422487225e-06,
+      "loss": 0.0178,
+      "step": 1278
+    },
+    {
+      "epoch": 44.13793103448276,
+      "grad_norm": 0.0626293197274208,
+      "learning_rate": 2.1294718909710394e-06,
+      "loss": 0.0185,
+      "step": 1280
+    },
+    {
+      "epoch": 44.206896551724135,
+      "grad_norm": 0.08184079080820084,
+      "learning_rate": 1.9591141396933563e-06,
+      "loss": 0.0178,
+      "step": 1282
+    },
+    {
+      "epoch": 44.275862068965516,
+      "grad_norm": 0.06804870814085007,
+      "learning_rate": 1.788756388415673e-06,
+      "loss": 0.0178,
+      "step": 1284
+    },
+    {
+      "epoch": 44.3448275862069,
+      "grad_norm": 0.07142641395330429,
+      "learning_rate": 1.6183986371379898e-06,
+      "loss": 0.0215,
+      "step": 1286
+    },
+    {
+      "epoch": 44.41379310344828,
+      "grad_norm": 0.06901556253433228,
+      "learning_rate": 1.4480408858603067e-06,
+      "loss": 0.0186,
+      "step": 1288
+    },
+    {
+      "epoch": 44.48275862068966,
+      "grad_norm": 0.07577712833881378,
+      "learning_rate": 1.2776831345826234e-06,
+      "loss": 0.0207,
+      "step": 1290
+    },
+    {
+      "epoch": 44.55172413793103,
+      "grad_norm": 0.0839504599571228,
+      "learning_rate": 1.1073253833049404e-06,
+      "loss": 0.0213,
+      "step": 1292
+    },
+    {
+      "epoch": 44.62068965517241,
+      "grad_norm": 0.11831686645746231,
+      "learning_rate": 9.369676320272573e-07,
+      "loss": 0.0206,
+      "step": 1294
+    },
+    {
+      "epoch": 44.689655172413794,
+      "grad_norm": 0.4428146183490753,
+      "learning_rate": 7.666098807495741e-07,
+      "loss": 0.0697,
+      "step": 1296
+    },
+    {
+      "epoch": 44.758620689655174,
+      "grad_norm": 0.05690554529428482,
+      "learning_rate": 5.962521294718911e-07,
+      "loss": 0.0182,
+      "step": 1298
+    },
+    {
+      "epoch": 44.827586206896555,
+      "grad_norm": 0.06728281080722809,
+      "learning_rate": 4.2589437819420783e-07,
+      "loss": 0.0169,
+      "step": 1300
+    },
+    {
+      "epoch": 44.89655172413793,
+      "grad_norm": 0.06453272700309753,
+      "learning_rate": 2.555366269165247e-07,
+      "loss": 0.0189,
+      "step": 1302
+    },
+    {
+      "epoch": 44.96551724137931,
+      "grad_norm": 0.06359589844942093,
+      "learning_rate": 8.517887563884157e-08,
+      "loss": 0.027,
+      "step": 1304
+    },
+    {
+      "epoch": 45.0,
+      "eval_accuracy": 0.6515151515151515,
+      "eval_f1_macro": 0.5674399966365533,
+      "eval_f1_micro": 0.6515151515151515,
+      "eval_f1_weighted": 0.6423054871360109,
+      "eval_loss": 1.4831533432006836,
+      "eval_precision_macro": 0.5694260552951398,
+      "eval_precision_micro": 0.6515151515151515,
+      "eval_precision_weighted": 0.6400870203536964,
+      "eval_recall_macro": 0.5717157974300832,
+      "eval_recall_micro": 0.6515151515151515,
+      "eval_recall_weighted": 0.6515151515151515,
+      "eval_runtime": 2.3557,
+      "eval_samples_per_second": 56.034,
+      "eval_steps_per_second": 7.217,
+      "step": 1305
+    },
+    {
+      "epoch": 45.0,
+      "step": 1305,
+      "total_flos": 1.611130661124526e+18,
+      "train_loss": 0.5009927850430724,
+      "train_runtime": 1277.117,
+      "train_samples_per_second": 16.279,
+      "train_steps_per_second": 1.022
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1305,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 45,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.611130661124526e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}