diff --git "a/checkpoint-1026/trainer_state.json" "b/checkpoint-1026/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1026/trainer_state.json"
@@ -0,0 +1,7395 @@
+{
+  "best_metric": 0.8051581382751465,
+  "best_model_checkpoint": "wav2vec2-vocal-disable-classification/checkpoint-513",
+  "epoch": 18.0,
+  "eval_steps": 1,
+  "global_step": 1026,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.017543859649122806,
+      "grad_norm": 1.56613290309906,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.6056,
+      "step": 1
+    },
+    {
+      "epoch": 0.03508771929824561,
+      "grad_norm": 1.8094905614852905,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.6101,
+      "step": 2
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 1.138840913772583,
+      "learning_rate": 1.8e-06,
+      "loss": 1.609,
+      "step": 3
+    },
+    {
+      "epoch": 0.07017543859649122,
+      "grad_norm": 3.695634603500366,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.6097,
+      "step": 4
+    },
+    {
+      "epoch": 0.08771929824561403,
+      "grad_norm": 1.9893696308135986,
+      "learning_rate": 3e-06,
+      "loss": 1.6028,
+      "step": 5
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 2.095263957977295,
+      "learning_rate": 3.6e-06,
+      "loss": 1.6016,
+      "step": 6
+    },
+    {
+      "epoch": 0.12280701754385964,
+      "grad_norm": 1.7659621238708496,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.5994,
+      "step": 7
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 2.041477680206299,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.5951,
+      "step": 8
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 1.6106765270233154,
+      "learning_rate": 5.4e-06,
+      "loss": 1.5957,
+      "step": 9
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "grad_norm": 1.5744407176971436,
+      "learning_rate": 6e-06,
+      "loss": 1.5951,
+      "step": 10
+    },
+    {
+      "epoch": 0.19298245614035087,
+      "grad_norm": 1.0527958869934082,
+      "learning_rate": 6.6e-06,
+      "loss": 1.5994,
+      "step": 11
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 1.5314522981643677,
+      "learning_rate": 7.2e-06,
+      "loss": 1.5766,
+      "step": 12
+    },
+    {
+      "epoch": 0.22807017543859648,
+      "grad_norm": 1.0843933820724487,
+      "learning_rate": 7.8e-06,
+      "loss": 1.5976,
+      "step": 13
+    },
+    {
+      "epoch": 0.24561403508771928,
+      "grad_norm": 1.8420531749725342,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.5803,
+      "step": 14
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 2.5668458938598633,
+      "learning_rate": 9e-06,
+      "loss": 1.5839,
+      "step": 15
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 1.5706558227539062,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.5522,
+      "step": 16
+    },
+    {
+      "epoch": 0.2982456140350877,
+      "grad_norm": 1.722334384918213,
+      "learning_rate": 1.02e-05,
+      "loss": 1.5391,
+      "step": 17
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.317153811454773,
+      "learning_rate": 1.08e-05,
+      "loss": 1.6246,
+      "step": 18
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 1.3377695083618164,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.6166,
+      "step": 19
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "grad_norm": 3.216266632080078,
+      "learning_rate": 1.2e-05,
+      "loss": 1.5151,
+      "step": 20
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.3238260746002197,
+      "learning_rate": 1.26e-05,
+      "loss": 1.5195,
+      "step": 21
+    },
+    {
+      "epoch": 0.38596491228070173,
+      "grad_norm": 2.946779251098633,
+      "learning_rate": 1.32e-05,
+      "loss": 1.503,
+      "step": 22
+    },
+    {
+      "epoch": 0.40350877192982454,
+      "grad_norm": 1.427848219871521,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.5538,
+      "step": 23
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 3.5313634872436523,
+      "learning_rate": 1.44e-05,
+      "loss": 1.545,
+      "step": 24
+    },
+    {
+      "epoch": 0.43859649122807015,
+      "grad_norm": 1.3384391069412231,
+      "learning_rate": 1.5e-05,
+      "loss": 1.5406,
+      "step": 25
+    },
+    {
+      "epoch": 0.45614035087719296,
+      "grad_norm": 1.635528802871704,
+      "learning_rate": 1.56e-05,
+      "loss": 1.4864,
+      "step": 26
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.3669475317001343,
+      "learning_rate": 1.62e-05,
+      "loss": 1.5609,
+      "step": 27
+    },
+    {
+      "epoch": 0.49122807017543857,
+      "grad_norm": 1.532422661781311,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.4829,
+      "step": 28
+    },
+    {
+      "epoch": 0.5087719298245614,
+      "grad_norm": 14.834467887878418,
+      "learning_rate": 1.74e-05,
+      "loss": 1.4049,
+      "step": 29
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.5839215517044067,
+      "learning_rate": 1.8e-05,
+      "loss": 1.4429,
+      "step": 30
+    },
+    {
+      "epoch": 0.543859649122807,
+      "grad_norm": 8.556641578674316,
+      "learning_rate": 1.86e-05,
+      "loss": 1.5076,
+      "step": 31
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "grad_norm": 2.6420958042144775,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.6325,
+      "step": 32
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.5453814268112183,
+      "learning_rate": 1.98e-05,
+      "loss": 1.5331,
+      "step": 33
+    },
+    {
+      "epoch": 0.5964912280701754,
+      "grad_norm": 2.1342880725860596,
+      "learning_rate": 2.04e-05,
+      "loss": 1.5563,
+      "step": 34
+    },
+    {
+      "epoch": 0.6140350877192983,
+      "grad_norm": 7.298281669616699,
+      "learning_rate": 2.1e-05,
+      "loss": 1.3024,
+      "step": 35
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 2.423535108566284,
+      "learning_rate": 2.16e-05,
+      "loss": 1.5851,
+      "step": 36
+    },
+    {
+      "epoch": 0.6491228070175439,
+      "grad_norm": 2.8099899291992188,
+      "learning_rate": 2.22e-05,
+      "loss": 1.664,
+      "step": 37
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 3.804816961288452,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 1.4642,
+      "step": 38
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 5.346778392791748,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 1.5016,
+      "step": 39
+    },
+    {
+      "epoch": 0.7017543859649122,
+      "grad_norm": 5.4332733154296875,
+      "learning_rate": 2.4e-05,
+      "loss": 1.4937,
+      "step": 40
+    },
+    {
+      "epoch": 0.7192982456140351,
+      "grad_norm": 3.2267627716064453,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 1.4857,
+      "step": 41
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 9.248668670654297,
+      "learning_rate": 2.52e-05,
+      "loss": 1.4574,
+      "step": 42
+    },
+    {
+      "epoch": 0.7543859649122807,
+      "grad_norm": 2.983009099960327,
+      "learning_rate": 2.58e-05,
+      "loss": 1.4802,
+      "step": 43
+    },
+    {
+      "epoch": 0.7719298245614035,
+      "grad_norm": 9.322035789489746,
+      "learning_rate": 2.64e-05,
+      "loss": 1.3646,
+      "step": 44
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 3.1577627658843994,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.3842,
+      "step": 45
+    },
+    {
+      "epoch": 0.8070175438596491,
+      "grad_norm": 3.0866541862487793,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 1.478,
+      "step": 46
+    },
+    {
+      "epoch": 0.8245614035087719,
+      "grad_norm": 2.434828281402588,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 1.2678,
+      "step": 47
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 19.617595672607422,
+      "learning_rate": 2.88e-05,
+      "loss": 1.5149,
+      "step": 48
+    },
+    {
+      "epoch": 0.8596491228070176,
+      "grad_norm": 3.927741527557373,
+      "learning_rate": 2.94e-05,
+      "loss": 1.3689,
+      "step": 49
+    },
+    {
+      "epoch": 0.8771929824561403,
+      "grad_norm": 4.532490253448486,
+      "learning_rate": 3e-05,
+      "loss": 1.5975,
+      "step": 50
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 4.9621500968933105,
+      "learning_rate": 2.9972477064220187e-05,
+      "loss": 1.4361,
+      "step": 51
+    },
+    {
+      "epoch": 0.9122807017543859,
+      "grad_norm": 6.061293125152588,
+      "learning_rate": 2.994495412844037e-05,
+      "loss": 1.4083,
+      "step": 52
+    },
+    {
+      "epoch": 0.9298245614035088,
+      "grad_norm": 4.741385459899902,
+      "learning_rate": 2.991743119266055e-05,
+      "loss": 1.4497,
+      "step": 53
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 3.2169480323791504,
+      "learning_rate": 2.9889908256880734e-05,
+      "loss": 1.4465,
+      "step": 54
+    },
+    {
+      "epoch": 0.9649122807017544,
+      "grad_norm": 7.039127349853516,
+      "learning_rate": 2.986238532110092e-05,
+      "loss": 1.2431,
+      "step": 55
+    },
+    {
+      "epoch": 0.9824561403508771,
+      "grad_norm": 6.163113594055176,
+      "learning_rate": 2.9834862385321102e-05,
+      "loss": 1.2167,
+      "step": 56
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4295806884765625,
+      "learning_rate": 2.9807339449541285e-05,
+      "loss": 1.3198,
+      "step": 57
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.4482758620689655,
+      "eval_loss": 1.3703213930130005,
+      "eval_roc_auc": 0.7240004822764974,
+      "eval_runtime": 9.1707,
+      "eval_samples_per_second": 28.46,
+      "eval_steps_per_second": 1.854,
+      "step": 57
+    },
+    {
+      "epoch": 1.0175438596491229,
+      "grad_norm": 2.1605288982391357,
+      "learning_rate": 2.9779816513761467e-05,
+      "loss": 1.3161,
+      "step": 58
+    },
+    {
+      "epoch": 1.0350877192982457,
+      "grad_norm": 2.469620943069458,
+      "learning_rate": 2.9752293577981653e-05,
+      "loss": 1.4468,
+      "step": 59
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 5.3216376304626465,
+      "learning_rate": 2.9724770642201836e-05,
+      "loss": 1.3517,
+      "step": 60
+    },
+    {
+      "epoch": 1.0701754385964912,
+      "grad_norm": 5.191178798675537,
+      "learning_rate": 2.9697247706422018e-05,
+      "loss": 1.3841,
+      "step": 61
+    },
+    {
+      "epoch": 1.087719298245614,
+      "grad_norm": 2.3336193561553955,
+      "learning_rate": 2.9669724770642204e-05,
+      "loss": 1.3057,
+      "step": 62
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 42.809783935546875,
+      "learning_rate": 2.9642201834862386e-05,
+      "loss": 1.2346,
+      "step": 63
+    },
+    {
+      "epoch": 1.1228070175438596,
+      "grad_norm": 6.859128952026367,
+      "learning_rate": 2.9614678899082572e-05,
+      "loss": 1.4394,
+      "step": 64
+    },
+    {
+      "epoch": 1.1403508771929824,
+      "grad_norm": 9.006386756896973,
+      "learning_rate": 2.958715596330275e-05,
+      "loss": 1.2954,
+      "step": 65
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 2.543628454208374,
+      "learning_rate": 2.9559633027522937e-05,
+      "loss": 1.0557,
+      "step": 66
+    },
+    {
+      "epoch": 1.1754385964912282,
+      "grad_norm": 2.736093044281006,
+      "learning_rate": 2.953211009174312e-05,
+      "loss": 1.2225,
+      "step": 67
+    },
+    {
+      "epoch": 1.1929824561403508,
+      "grad_norm": 4.434782028198242,
+      "learning_rate": 2.9504587155963306e-05,
+      "loss": 1.1875,
+      "step": 68
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 8.687559127807617,
+      "learning_rate": 2.9477064220183485e-05,
+      "loss": 0.937,
+      "step": 69
+    },
+    {
+      "epoch": 1.2280701754385965,
+      "grad_norm": 3.4958977699279785,
+      "learning_rate": 2.944954128440367e-05,
+      "loss": 1.2851,
+      "step": 70
+    },
+    {
+      "epoch": 1.2456140350877192,
+      "grad_norm": 12.490559577941895,
+      "learning_rate": 2.9422018348623853e-05,
+      "loss": 1.5578,
+      "step": 71
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 3.466176748275757,
+      "learning_rate": 2.939449541284404e-05,
+      "loss": 0.9698,
+      "step": 72
+    },
+    {
+      "epoch": 1.280701754385965,
+      "grad_norm": 9.730321884155273,
+      "learning_rate": 2.936697247706422e-05,
+      "loss": 1.2044,
+      "step": 73
+    },
+    {
+      "epoch": 1.2982456140350878,
+      "grad_norm": 12.908202171325684,
+      "learning_rate": 2.9339449541284404e-05,
+      "loss": 1.3778,
+      "step": 74
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 7.409069538116455,
+      "learning_rate": 2.931192660550459e-05,
+      "loss": 1.6065,
+      "step": 75
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 12.185145378112793,
+      "learning_rate": 2.9284403669724772e-05,
+      "loss": 1.2463,
+      "step": 76
+    },
+    {
+      "epoch": 1.3508771929824561,
+      "grad_norm": 16.505674362182617,
+      "learning_rate": 2.9256880733944955e-05,
+      "loss": 1.0027,
+      "step": 77
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 9.078436851501465,
+      "learning_rate": 2.9229357798165137e-05,
+      "loss": 1.2037,
+      "step": 78
+    },
+    {
+      "epoch": 1.3859649122807016,
+      "grad_norm": 23.13191032409668,
+      "learning_rate": 2.9201834862385323e-05,
+      "loss": 1.0528,
+      "step": 79
+    },
+    {
+      "epoch": 1.4035087719298245,
+      "grad_norm": 7.651644229888916,
+      "learning_rate": 2.9174311926605506e-05,
+      "loss": 1.0373,
+      "step": 80
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 15.649460792541504,
+      "learning_rate": 2.9146788990825688e-05,
+      "loss": 1.3094,
+      "step": 81
+    },
+    {
+      "epoch": 1.4385964912280702,
+      "grad_norm": 14.335933685302734,
+      "learning_rate": 2.9119266055045874e-05,
+      "loss": 1.229,
+      "step": 82
+    },
+    {
+      "epoch": 1.456140350877193,
+      "grad_norm": 13.704193115234375,
+      "learning_rate": 2.9091743119266056e-05,
+      "loss": 1.2517,
+      "step": 83
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 6.872323036193848,
+      "learning_rate": 2.9064220183486242e-05,
+      "loss": 1.0614,
+      "step": 84
+    },
+    {
+      "epoch": 1.4912280701754386,
+      "grad_norm": 58.63323211669922,
+      "learning_rate": 2.903669724770642e-05,
+      "loss": 1.2651,
+      "step": 85
+    },
+    {
+      "epoch": 1.5087719298245614,
+      "grad_norm": 16.493391036987305,
+      "learning_rate": 2.9009174311926607e-05,
+      "loss": 1.0977,
+      "step": 86
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 76.30412292480469,
+      "learning_rate": 2.898165137614679e-05,
+      "loss": 1.3534,
+      "step": 87
+    },
+    {
+      "epoch": 1.543859649122807,
+      "grad_norm": 35.07970428466797,
+      "learning_rate": 2.8954128440366975e-05,
+      "loss": 1.1015,
+      "step": 88
+    },
+    {
+      "epoch": 1.5614035087719298,
+      "grad_norm": 41.934513092041016,
+      "learning_rate": 2.8926605504587155e-05,
+      "loss": 1.252,
+      "step": 89
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 6.429330825805664,
+      "learning_rate": 2.889908256880734e-05,
+      "loss": 0.9942,
+      "step": 90
+    },
+    {
+      "epoch": 1.5964912280701755,
+      "grad_norm": 4.915737152099609,
+      "learning_rate": 2.8871559633027523e-05,
+      "loss": 1.0583,
+      "step": 91
+    },
+    {
+      "epoch": 1.6140350877192984,
+      "grad_norm": 5.171088695526123,
+      "learning_rate": 2.884403669724771e-05,
+      "loss": 0.9167,
+      "step": 92
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 4.840869426727295,
+      "learning_rate": 2.881651376146789e-05,
+      "loss": 1.3165,
+      "step": 93
+    },
+    {
+      "epoch": 1.6491228070175439,
+      "grad_norm": 8.278257369995117,
+      "learning_rate": 2.8788990825688074e-05,
+      "loss": 1.0294,
+      "step": 94
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 7.380605697631836,
+      "learning_rate": 2.876146788990826e-05,
+      "loss": 1.16,
+      "step": 95
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 27.351318359375,
+      "learning_rate": 2.8733944954128442e-05,
+      "loss": 1.6427,
+      "step": 96
+    },
+    {
+      "epoch": 1.7017543859649122,
+      "grad_norm": 5.766673564910889,
+      "learning_rate": 2.8706422018348625e-05,
+      "loss": 1.1461,
+      "step": 97
+    },
+    {
+      "epoch": 1.719298245614035,
+      "grad_norm": 6.586304664611816,
+      "learning_rate": 2.8678899082568807e-05,
+      "loss": 1.1368,
+      "step": 98
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 13.888910293579102,
+      "learning_rate": 2.8651376146788993e-05,
+      "loss": 0.8426,
+      "step": 99
+    },
+    {
+      "epoch": 1.7543859649122808,
+      "grad_norm": 14.992203712463379,
+      "learning_rate": 2.8623853211009175e-05,
+      "loss": 1.1494,
+      "step": 100
+    },
+    {
+      "epoch": 1.7719298245614035,
+      "grad_norm": 8.557571411132812,
+      "learning_rate": 2.8596330275229358e-05,
+      "loss": 1.1022,
+      "step": 101
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 19.743364334106445,
+      "learning_rate": 2.856880733944954e-05,
+      "loss": 1.0737,
+      "step": 102
+    },
+    {
+      "epoch": 1.807017543859649,
+      "grad_norm": 13.77326488494873,
+      "learning_rate": 2.8541284403669726e-05,
+      "loss": 0.8547,
+      "step": 103
+    },
+    {
+      "epoch": 1.8245614035087718,
+      "grad_norm": 4.906980514526367,
+      "learning_rate": 2.8513761467889912e-05,
+      "loss": 0.8529,
+      "step": 104
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 2.8242599964141846,
+      "learning_rate": 2.848623853211009e-05,
+      "loss": 1.2058,
+      "step": 105
+    },
+    {
+      "epoch": 1.8596491228070176,
+      "grad_norm": 3.802095890045166,
+      "learning_rate": 2.8458715596330277e-05,
+      "loss": 0.9236,
+      "step": 106
+    },
+    {
+      "epoch": 1.8771929824561404,
+      "grad_norm": 4.197514057159424,
+      "learning_rate": 2.843119266055046e-05,
+      "loss": 0.8856,
+      "step": 107
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 13.207415580749512,
+      "learning_rate": 2.8403669724770645e-05,
+      "loss": 0.769,
+      "step": 108
+    },
+    {
+      "epoch": 1.912280701754386,
+      "grad_norm": 4.843390941619873,
+      "learning_rate": 2.8376146788990824e-05,
+      "loss": 1.1706,
+      "step": 109
+    },
+    {
+      "epoch": 1.9298245614035088,
+      "grad_norm": 13.065381050109863,
+      "learning_rate": 2.834862385321101e-05,
+      "loss": 1.0655,
+      "step": 110
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 15.791166305541992,
+      "learning_rate": 2.8321100917431193e-05,
+      "loss": 1.2378,
+      "step": 111
+    },
+    {
+      "epoch": 1.9649122807017543,
+      "grad_norm": 11.551010131835938,
+      "learning_rate": 2.829357798165138e-05,
+      "loss": 0.9768,
+      "step": 112
+    },
+    {
+      "epoch": 1.9824561403508771,
+      "grad_norm": 14.838546752929688,
+      "learning_rate": 2.826605504587156e-05,
+      "loss": 1.0073,
+      "step": 113
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.385023355484009,
+      "learning_rate": 2.8238532110091744e-05,
+      "loss": 0.9486,
+      "step": 114
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.6091954022988506,
+      "eval_loss": 1.0561790466308594,
+      "eval_roc_auc": 0.8297140297597192,
+      "eval_runtime": 10.2393,
+      "eval_samples_per_second": 25.49,
+      "eval_steps_per_second": 1.66,
+      "step": 114
+    },
+    {
+      "epoch": 2.017543859649123,
+      "grad_norm": 13.100618362426758,
+      "learning_rate": 2.821100917431193e-05,
+      "loss": 0.9394,
+      "step": 115
+    },
+    {
+      "epoch": 2.0350877192982457,
+      "grad_norm": 22.331451416015625,
+      "learning_rate": 2.8183486238532112e-05,
+      "loss": 1.0603,
+      "step": 116
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 4.956656455993652,
+      "learning_rate": 2.8155963302752294e-05,
+      "loss": 0.9566,
+      "step": 117
+    },
+    {
+      "epoch": 2.0701754385964914,
+      "grad_norm": 19.40747833251953,
+      "learning_rate": 2.8128440366972477e-05,
+      "loss": 1.1504,
+      "step": 118
+    },
+    {
+      "epoch": 2.087719298245614,
+      "grad_norm": 13.512128829956055,
+      "learning_rate": 2.8100917431192663e-05,
+      "loss": 1.194,
+      "step": 119
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 6.616446495056152,
+      "learning_rate": 2.8073394495412845e-05,
+      "loss": 1.0321,
+      "step": 120
+    },
+    {
+      "epoch": 2.1228070175438596,
+      "grad_norm": 2.3998453617095947,
+      "learning_rate": 2.8045871559633028e-05,
+      "loss": 1.0636,
+      "step": 121
+    },
+    {
+      "epoch": 2.1403508771929824,
+      "grad_norm": 3.214811086654663,
+      "learning_rate": 2.801834862385321e-05,
+      "loss": 0.7859,
+      "step": 122
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 2.0244710445404053,
+      "learning_rate": 2.7990825688073396e-05,
+      "loss": 0.8801,
+      "step": 123
+    },
+    {
+      "epoch": 2.175438596491228,
+      "grad_norm": 3.1682400703430176,
+      "learning_rate": 2.7963302752293582e-05,
+      "loss": 0.9413,
+      "step": 124
+    },
+    {
+      "epoch": 2.192982456140351,
+      "grad_norm": 3.509514570236206,
+      "learning_rate": 2.793577981651376e-05,
+      "loss": 0.8083,
+      "step": 125
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 6.175449848175049,
+      "learning_rate": 2.7908256880733947e-05,
+      "loss": 1.0884,
+      "step": 126
+    },
+    {
+      "epoch": 2.2280701754385963,
+      "grad_norm": 4.469012260437012,
+      "learning_rate": 2.788073394495413e-05,
+      "loss": 0.8281,
+      "step": 127
+    },
+    {
+      "epoch": 2.245614035087719,
+      "grad_norm": 18.924325942993164,
+      "learning_rate": 2.7853211009174315e-05,
+      "loss": 0.9676,
+      "step": 128
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 24.372833251953125,
+      "learning_rate": 2.7825688073394494e-05,
+      "loss": 1.1955,
+      "step": 129
+    },
+    {
+      "epoch": 2.280701754385965,
+      "grad_norm": 1.3634321689605713,
+      "learning_rate": 2.779816513761468e-05,
+      "loss": 0.9889,
+      "step": 130
+    },
+    {
+      "epoch": 2.2982456140350878,
+      "grad_norm": 17.57159996032715,
+      "learning_rate": 2.7770642201834863e-05,
+      "loss": 1.1324,
+      "step": 131
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 16.63871192932129,
+      "learning_rate": 2.774311926605505e-05,
+      "loss": 1.1951,
+      "step": 132
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 16.466339111328125,
+      "learning_rate": 2.7715596330275228e-05,
+      "loss": 0.9375,
+      "step": 133
+    },
+    {
+      "epoch": 2.3508771929824563,
+      "grad_norm": 13.808109283447266,
+      "learning_rate": 2.7688073394495413e-05,
+      "loss": 1.0418,
+      "step": 134
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 5.8455729484558105,
+      "learning_rate": 2.76605504587156e-05,
+      "loss": 1.0906,
+      "step": 135
+    },
+    {
+      "epoch": 2.3859649122807016,
+      "grad_norm": 18.494245529174805,
+      "learning_rate": 2.7633027522935782e-05,
+      "loss": 0.9786,
+      "step": 136
+    },
+    {
+      "epoch": 2.4035087719298245,
+      "grad_norm": 21.47800636291504,
+      "learning_rate": 2.7605504587155964e-05,
+      "loss": 0.8138,
+      "step": 137
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 8.400023460388184,
+      "learning_rate": 2.7577981651376147e-05,
+      "loss": 1.027,
+      "step": 138
+    },
+    {
+      "epoch": 2.43859649122807,
+      "grad_norm": 7.198583602905273,
+      "learning_rate": 2.7550458715596333e-05,
+      "loss": 0.7104,
+      "step": 139
+    },
+    {
+      "epoch": 2.456140350877193,
+      "grad_norm": 23.315410614013672,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 1.2115,
+      "step": 140
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 18.734041213989258,
+      "learning_rate": 2.7495412844036697e-05,
+      "loss": 0.9739,
+      "step": 141
+    },
+    {
+      "epoch": 2.4912280701754383,
+      "grad_norm": 10.015871047973633,
+      "learning_rate": 2.746788990825688e-05,
+      "loss": 0.9796,
+      "step": 142
+    },
+    {
+      "epoch": 2.5087719298245617,
+      "grad_norm": 11.979524612426758,
+      "learning_rate": 2.7440366972477066e-05,
+      "loss": 0.8025,
+      "step": 143
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 9.438429832458496,
+      "learning_rate": 2.741284403669725e-05,
+      "loss": 0.8838,
+      "step": 144
+    },
+    {
+      "epoch": 2.543859649122807,
+      "grad_norm": 43.801063537597656,
+      "learning_rate": 2.738532110091743e-05,
+      "loss": 0.9144,
+      "step": 145
+    },
+    {
+      "epoch": 2.56140350877193,
+      "grad_norm": 19.495887756347656,
+      "learning_rate": 2.7357798165137617e-05,
+      "loss": 0.8573,
+      "step": 146
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 21.922138214111328,
+      "learning_rate": 2.73302752293578e-05,
+      "loss": 0.9611,
+      "step": 147
+    },
+    {
+      "epoch": 2.5964912280701755,
+      "grad_norm": 16.30131721496582,
+      "learning_rate": 2.7302752293577985e-05,
+      "loss": 1.3519,
+      "step": 148
+    },
+    {
+      "epoch": 2.6140350877192984,
+      "grad_norm": 43.10526657104492,
+      "learning_rate": 2.7275229357798164e-05,
+      "loss": 1.0194,
+      "step": 149
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 18.148771286010742,
+      "learning_rate": 2.724770642201835e-05,
+      "loss": 1.0759,
+      "step": 150
+    },
+    {
+      "epoch": 2.6491228070175437,
+      "grad_norm": 28.365219116210938,
+      "learning_rate": 2.7220183486238532e-05,
+      "loss": 0.9998,
+      "step": 151
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 10.812620162963867,
+      "learning_rate": 2.7192660550458718e-05,
+      "loss": 0.9368,
+      "step": 152
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 32.16633605957031,
+      "learning_rate": 2.7165137614678897e-05,
+      "loss": 1.0304,
+      "step": 153
+    },
+    {
+      "epoch": 2.7017543859649122,
+      "grad_norm": 63.48967742919922,
+      "learning_rate": 2.7137614678899083e-05,
+      "loss": 1.0795,
+      "step": 154
+    },
+    {
+      "epoch": 2.719298245614035,
+      "grad_norm": 4.874389171600342,
+      "learning_rate": 2.711009174311927e-05,
+      "loss": 1.062,
+      "step": 155
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 18.235118865966797,
+      "learning_rate": 2.708256880733945e-05,
+      "loss": 1.3184,
+      "step": 156
+    },
+    {
+      "epoch": 2.754385964912281,
+      "grad_norm": 22.853351593017578,
+      "learning_rate": 2.7055045871559634e-05,
+      "loss": 1.0108,
+      "step": 157
+    },
+    {
+      "epoch": 2.7719298245614032,
+      "grad_norm": 50.805885314941406,
+      "learning_rate": 2.7027522935779817e-05,
+      "loss": 0.914,
+      "step": 158
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 5.718901634216309,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.7322,
+      "step": 159
+    },
+    {
+      "epoch": 2.807017543859649,
+      "grad_norm": 60.18952178955078,
+      "learning_rate": 2.6972477064220185e-05,
+      "loss": 1.088,
+      "step": 160
+    },
+    {
+      "epoch": 2.824561403508772,
+      "grad_norm": 5.959671974182129,
+      "learning_rate": 2.6944954128440367e-05,
+      "loss": 0.5275,
+      "step": 161
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 2.963286876678467,
+      "learning_rate": 2.691743119266055e-05,
+      "loss": 0.8271,
+      "step": 162
+    },
+    {
+      "epoch": 2.8596491228070176,
+      "grad_norm": 18.080677032470703,
+      "learning_rate": 2.6889908256880736e-05,
+      "loss": 0.7593,
+      "step": 163
+    },
+    {
+      "epoch": 2.8771929824561404,
+      "grad_norm": 3.0679006576538086,
+      "learning_rate": 2.6862385321100918e-05,
+      "loss": 0.582,
+      "step": 164
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 27.97750473022461,
+      "learning_rate": 2.68348623853211e-05,
+      "loss": 1.0477,
+      "step": 165
+    },
+    {
+      "epoch": 2.912280701754386,
+      "grad_norm": 9.784193992614746,
+      "learning_rate": 2.6807339449541286e-05,
+      "loss": 0.7083,
+      "step": 166
+    },
+    {
+      "epoch": 2.9298245614035086,
+      "grad_norm": 7.2379045486450195,
+      "learning_rate": 2.677981651376147e-05,
+      "loss": 0.6767,
+      "step": 167
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 17.974803924560547,
+      "learning_rate": 2.6752293577981655e-05,
+      "loss": 1.117,
+      "step": 168
+    },
+    {
+      "epoch": 2.9649122807017543,
+      "grad_norm": 8.451967239379883,
+      "learning_rate": 2.6724770642201834e-05,
+      "loss": 1.0163,
+      "step": 169
+    },
+    {
+      "epoch": 2.982456140350877,
+      "grad_norm": 16.834774017333984,
+      "learning_rate": 2.669724770642202e-05,
+      "loss": 0.642,
+      "step": 170
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 10.743306159973145,
+      "learning_rate": 2.6669724770642202e-05,
+      "loss": 0.8485,
+      "step": 171
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.5938697318007663,
+      "eval_loss": 1.042966604232788,
+      "eval_roc_auc": 0.8370073235789949,
+      "eval_runtime": 9.8627,
+      "eval_samples_per_second": 26.463,
+      "eval_steps_per_second": 1.724,
+      "step": 171
+    },
+    {
+      "epoch": 3.017543859649123,
+      "grad_norm": 7.450799942016602,
+      "learning_rate": 2.6642201834862388e-05,
+      "loss": 0.7936,
+      "step": 172
+    },
+    {
+      "epoch": 3.0350877192982457,
+      "grad_norm": 13.938980102539062,
+      "learning_rate": 2.6614678899082567e-05,
+      "loss": 0.9981,
+      "step": 173
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 10.046002388000488,
+      "learning_rate": 2.6587155963302753e-05,
+      "loss": 0.9664,
+      "step": 174
+    },
+    {
+      "epoch": 3.0701754385964914,
+      "grad_norm": 2.47877836227417,
+      "learning_rate": 2.6559633027522936e-05,
+      "loss": 0.4564,
+      "step": 175
+    },
+    {
+      "epoch": 3.087719298245614,
+      "grad_norm": 4.682964324951172,
+      "learning_rate": 2.653211009174312e-05,
+      "loss": 0.7715,
+      "step": 176
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 31.68276023864746,
+      "learning_rate": 2.6504587155963304e-05,
+      "loss": 1.0283,
+      "step": 177
+    },
+    {
+      "epoch": 3.1228070175438596,
+      "grad_norm": 6.329732418060303,
+      "learning_rate": 2.6477064220183486e-05,
+      "loss": 0.6996,
+      "step": 178
+    },
+    {
+      "epoch": 3.1403508771929824,
+      "grad_norm": 14.793739318847656,
+      "learning_rate": 2.6449541284403672e-05,
+      "loss": 0.5858,
+      "step": 179
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 123.91106414794922,
+      "learning_rate": 2.6422018348623855e-05,
+      "loss": 0.8345,
+      "step": 180
+    },
+    {
+      "epoch": 3.175438596491228,
+      "grad_norm": 10.042440414428711,
+      "learning_rate": 2.6394495412844037e-05,
+      "loss": 0.8356,
+      "step": 181
+    },
+    {
+      "epoch": 3.192982456140351,
+      "grad_norm": 15.220785140991211,
+      "learning_rate": 2.636697247706422e-05,
+      "loss": 0.5512,
+      "step": 182
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 9.333229064941406,
+      "learning_rate": 2.6339449541284406e-05,
+      "loss": 0.6247,
+      "step": 183
+    },
+    {
+      "epoch": 3.2280701754385963,
+      "grad_norm": 13.827474594116211,
+      "learning_rate": 2.6311926605504588e-05,
+      "loss": 1.0353,
+      "step": 184
+    },
+    {
+      "epoch": 3.245614035087719,
+      "grad_norm": 4.803915500640869,
+      "learning_rate": 2.628440366972477e-05,
+      "loss": 0.8724,
+      "step": 185
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 9.242655754089355,
+      "learning_rate": 2.6256880733944953e-05,
+      "loss": 0.7395,
+      "step": 186
+    },
+    {
+      "epoch": 3.280701754385965,
+      "grad_norm": 82.34434509277344,
+      "learning_rate": 2.622935779816514e-05,
+      "loss": 0.7797,
+      "step": 187
+    },
+    {
+      "epoch": 3.2982456140350878,
+      "grad_norm": 29.72227668762207,
+      "learning_rate": 2.6201834862385325e-05,
+      "loss": 0.9626,
+      "step": 188
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 30.35354232788086,
+      "learning_rate": 2.6174311926605504e-05,
+      "loss": 1.0387,
+      "step": 189
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 9.3469820022583,
+      "learning_rate": 2.614678899082569e-05,
+      "loss": 0.9555,
+      "step": 190
+    },
+    {
+      "epoch": 3.3508771929824563,
+      "grad_norm": 10.0740327835083,
+      "learning_rate": 2.6119266055045872e-05,
+      "loss": 0.7608,
+      "step": 191
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 16.590267181396484,
+      "learning_rate": 2.6091743119266058e-05,
+      "loss": 0.8894,
+      "step": 192
+    },
+    {
+      "epoch": 3.3859649122807016,
+      "grad_norm": 6.453067779541016,
+      "learning_rate": 2.6064220183486237e-05,
+      "loss": 0.6409,
+      "step": 193
+    },
+    {
+      "epoch": 3.4035087719298245,
+      "grad_norm": 18.96304702758789,
+      "learning_rate": 2.6036697247706423e-05,
+      "loss": 1.1674,
+      "step": 194
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 4.519999027252197,
+      "learning_rate": 2.6009174311926605e-05,
+      "loss": 0.9088,
+      "step": 195
+    },
+    {
+      "epoch": 3.43859649122807,
+      "grad_norm": 14.560205459594727,
+      "learning_rate": 2.598165137614679e-05,
+      "loss": 1.0394,
+      "step": 196
+    },
+    {
+      "epoch": 3.456140350877193,
+      "grad_norm": 9.643004417419434,
+      "learning_rate": 2.5954128440366974e-05,
+      "loss": 0.7166,
+      "step": 197
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 6.686179161071777,
+      "learning_rate": 2.5926605504587156e-05,
+      "loss": 0.9138,
+      "step": 198
+    },
+    {
+      "epoch": 3.4912280701754383,
+      "grad_norm": 6.065926551818848,
+      "learning_rate": 2.5899082568807342e-05,
+      "loss": 0.5985,
+      "step": 199
+    },
+    {
+      "epoch": 3.5087719298245617,
+      "grad_norm": 8.297146797180176,
+      "learning_rate": 2.5871559633027525e-05,
+      "loss": 0.8459,
+      "step": 200
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 28.6617431640625,
+      "learning_rate": 2.5844036697247707e-05,
+      "loss": 0.4553,
+      "step": 201
+    },
+    {
+      "epoch": 3.543859649122807,
+      "grad_norm": 8.133190155029297,
+      "learning_rate": 2.581651376146789e-05,
+      "loss": 0.8616,
+      "step": 202
+    },
+    {
+      "epoch": 3.56140350877193,
+      "grad_norm": 7.848640441894531,
+      "learning_rate": 2.5788990825688075e-05,
+      "loss": 1.0041,
+      "step": 203
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 12.316092491149902,
+      "learning_rate": 2.5761467889908258e-05,
+      "loss": 0.8715,
+      "step": 204
+    },
+    {
+      "epoch": 3.5964912280701755,
+      "grad_norm": 7.93389892578125,
+      "learning_rate": 2.573394495412844e-05,
+      "loss": 0.7744,
+      "step": 205
+    },
+    {
+      "epoch": 3.6140350877192984,
+      "grad_norm": 9.929866790771484,
+      "learning_rate": 2.5706422018348623e-05,
+      "loss": 0.6355,
+      "step": 206
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 11.27926254272461,
+      "learning_rate": 2.567889908256881e-05,
+      "loss": 1.0616,
+      "step": 207
+    },
+    {
+      "epoch": 3.6491228070175437,
+      "grad_norm": 12.538276672363281,
+      "learning_rate": 2.5651376146788994e-05,
+      "loss": 0.8681,
+      "step": 208
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 14.14212703704834,
+      "learning_rate": 2.5623853211009174e-05,
+      "loss": 0.8615,
+      "step": 209
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 12.915414810180664,
+      "learning_rate": 2.559633027522936e-05,
+      "loss": 0.905,
+      "step": 210
+    },
+    {
+      "epoch": 3.7017543859649122,
+      "grad_norm": 6.382973670959473,
+      "learning_rate": 2.5568807339449542e-05,
+      "loss": 0.8229,
+      "step": 211
+    },
+    {
+      "epoch": 3.719298245614035,
+      "grad_norm": 8.958497047424316,
+      "learning_rate": 2.5541284403669728e-05,
+      "loss": 0.8367,
+      "step": 212
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 13.200749397277832,
+      "learning_rate": 2.5513761467889907e-05,
+      "loss": 0.7738,
+      "step": 213
+    },
+    {
+      "epoch": 3.754385964912281,
+      "grad_norm": 25.975767135620117,
+      "learning_rate": 2.5486238532110093e-05,
+      "loss": 1.1707,
+      "step": 214
+    },
+    {
+      "epoch": 3.7719298245614032,
+      "grad_norm": 46.98295211791992,
+      "learning_rate": 2.5458715596330275e-05,
+      "loss": 0.815,
+      "step": 215
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 6.342257499694824,
+      "learning_rate": 2.543119266055046e-05,
+      "loss": 0.9045,
+      "step": 216
+    },
+    {
+      "epoch": 3.807017543859649,
+      "grad_norm": 11.122331619262695,
+      "learning_rate": 2.540366972477064e-05,
+      "loss": 0.4915,
+      "step": 217
+    },
+    {
+      "epoch": 3.824561403508772,
+      "grad_norm": 11.39393424987793,
+      "learning_rate": 2.5376146788990826e-05,
+      "loss": 0.8584,
+      "step": 218
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 16.47450828552246,
+      "learning_rate": 2.5348623853211012e-05,
+      "loss": 0.9062,
+      "step": 219
+    },
+    {
+      "epoch": 3.8596491228070176,
+      "grad_norm": 8.9651517868042,
+      "learning_rate": 2.5321100917431194e-05,
+      "loss": 0.6332,
+      "step": 220
+    },
+    {
+      "epoch": 3.8771929824561404,
+      "grad_norm": 14.864767074584961,
+      "learning_rate": 2.5293577981651377e-05,
+      "loss": 0.7327,
+      "step": 221
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 13.695717811584473,
+      "learning_rate": 2.526605504587156e-05,
+      "loss": 0.8439,
+      "step": 222
+    },
+    {
+      "epoch": 3.912280701754386,
+      "grad_norm": 10.765061378479004,
+      "learning_rate": 2.5238532110091745e-05,
+      "loss": 1.0394,
+      "step": 223
+    },
+    {
+      "epoch": 3.9298245614035086,
+      "grad_norm": 15.805130958557129,
+      "learning_rate": 2.5211009174311928e-05,
+      "loss": 1.3535,
+      "step": 224
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 19.387144088745117,
+      "learning_rate": 2.518348623853211e-05,
+      "loss": 0.7274,
+      "step": 225
+    },
+    {
+      "epoch": 3.9649122807017543,
+      "grad_norm": 37.748497009277344,
+      "learning_rate": 2.5155963302752293e-05,
+      "loss": 0.7626,
+      "step": 226
+    },
+    {
+      "epoch": 3.982456140350877,
+      "grad_norm": 16.551725387573242,
+      "learning_rate": 2.512844036697248e-05,
+      "loss": 0.9978,
+      "step": 227
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 25.66643524169922,
+      "learning_rate": 2.510091743119266e-05,
+      "loss": 0.7405,
+      "step": 228
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.6245210727969349,
+      "eval_loss": 0.9546777606010437,
+      "eval_roc_auc": 0.8586184139970431,
+      "eval_runtime": 9.814,
+      "eval_samples_per_second": 26.595,
+      "eval_steps_per_second": 1.732,
+      "step": 228
+    },
+    {
+      "epoch": 4.017543859649122,
+      "grad_norm": 4.81406307220459,
+      "learning_rate": 2.5073394495412843e-05,
+      "loss": 0.5165,
+      "step": 229
+    },
+    {
+      "epoch": 4.035087719298246,
+      "grad_norm": 9.769265174865723,
+      "learning_rate": 2.504587155963303e-05,
+      "loss": 0.6222,
+      "step": 230
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 8.312880516052246,
+      "learning_rate": 2.5018348623853212e-05,
+      "loss": 0.6313,
+      "step": 231
+    },
+    {
+      "epoch": 4.0701754385964914,
+      "grad_norm": 5.13502311706543,
+      "learning_rate": 2.4990825688073398e-05,
+      "loss": 0.7425,
+      "step": 232
+    },
+    {
+      "epoch": 4.087719298245614,
+      "grad_norm": 19.665512084960938,
+      "learning_rate": 2.4963302752293577e-05,
+      "loss": 1.0384,
+      "step": 233
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 29.155986785888672,
+      "learning_rate": 2.4935779816513763e-05,
+      "loss": 1.0976,
+      "step": 234
+    },
+    {
+      "epoch": 4.12280701754386,
+      "grad_norm": 11.831360816955566,
+      "learning_rate": 2.4908256880733945e-05,
+      "loss": 0.7877,
+      "step": 235
+    },
+    {
+      "epoch": 4.140350877192983,
+      "grad_norm": 19.49347686767578,
+      "learning_rate": 2.488073394495413e-05,
+      "loss": 0.6279,
+      "step": 236
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 6.922645092010498,
+      "learning_rate": 2.485321100917431e-05,
+      "loss": 0.7153,
+      "step": 237
+    },
+    {
+      "epoch": 4.175438596491228,
+      "grad_norm": 3.6549301147460938,
+      "learning_rate": 2.4825688073394496e-05,
+      "loss": 0.873,
+      "step": 238
+    },
+    {
+      "epoch": 4.192982456140351,
+      "grad_norm": 10.724910736083984,
+      "learning_rate": 2.4798165137614682e-05,
+      "loss": 0.7728,
+      "step": 239
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 3.7120726108551025,
+      "learning_rate": 2.4770642201834864e-05,
+      "loss": 0.7132,
+      "step": 240
+    },
+    {
+      "epoch": 4.228070175438597,
+      "grad_norm": 29.73471450805664,
+      "learning_rate": 2.4743119266055047e-05,
+      "loss": 0.5634,
+      "step": 241
+    },
+    {
+      "epoch": 4.245614035087719,
+      "grad_norm": 5.868150234222412,
+      "learning_rate": 2.471559633027523e-05,
+      "loss": 0.6967,
+      "step": 242
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 9.828577995300293,
+      "learning_rate": 2.4688073394495415e-05,
+      "loss": 0.8814,
+      "step": 243
+    },
+    {
+      "epoch": 4.280701754385965,
+      "grad_norm": 19.289398193359375,
+      "learning_rate": 2.4660550458715597e-05,
+      "loss": 0.6907,
+      "step": 244
+    },
+    {
+      "epoch": 4.298245614035087,
+      "grad_norm": 18.95272445678711,
+      "learning_rate": 2.463302752293578e-05,
+      "loss": 0.7413,
+      "step": 245
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 9.175381660461426,
+      "learning_rate": 2.4605504587155962e-05,
+      "loss": 0.5549,
+      "step": 246
+    },
+    {
+      "epoch": 4.333333333333333,
+      "grad_norm": 11.017579078674316,
+      "learning_rate": 2.4577981651376148e-05,
+      "loss": 0.5834,
+      "step": 247
+    },
+    {
+      "epoch": 4.350877192982456,
+      "grad_norm": 4.641144752502441,
+      "learning_rate": 2.455045871559633e-05,
+      "loss": 0.7171,
+      "step": 248
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 30.48017120361328,
+      "learning_rate": 2.4522935779816513e-05,
+      "loss": 0.5314,
+      "step": 249
+    },
+    {
+      "epoch": 4.385964912280702,
+      "grad_norm": 5.151516914367676,
+      "learning_rate": 2.44954128440367e-05,
+      "loss": 0.822,
+      "step": 250
+    },
+    {
+      "epoch": 4.4035087719298245,
+      "grad_norm": 4.152652263641357,
+      "learning_rate": 2.446788990825688e-05,
+      "loss": 0.8574,
+      "step": 251
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 26.325111389160156,
+      "learning_rate": 2.4440366972477067e-05,
+      "loss": 0.5536,
+      "step": 252
+    },
+    {
+      "epoch": 4.43859649122807,
+      "grad_norm": 3.4374260902404785,
+      "learning_rate": 2.4412844036697247e-05,
+      "loss": 0.7328,
+      "step": 253
+    },
+    {
+      "epoch": 4.456140350877193,
+      "grad_norm": 7.868347644805908,
+      "learning_rate": 2.4385321100917432e-05,
+      "loss": 0.8698,
+      "step": 254
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 14.990694046020508,
+      "learning_rate": 2.4357798165137615e-05,
+      "loss": 0.9928,
+      "step": 255
+    },
+    {
+      "epoch": 4.491228070175438,
+      "grad_norm": 55.8467903137207,
+      "learning_rate": 2.43302752293578e-05,
+      "loss": 0.8542,
+      "step": 256
+    },
+    {
+      "epoch": 4.508771929824562,
+      "grad_norm": 15.516489028930664,
+      "learning_rate": 2.430275229357798e-05,
+      "loss": 0.935,
+      "step": 257
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 31.769515991210938,
+      "learning_rate": 2.4275229357798166e-05,
+      "loss": 0.5695,
+      "step": 258
+    },
+    {
+      "epoch": 4.543859649122807,
+      "grad_norm": 6.501729488372803,
+      "learning_rate": 2.4247706422018348e-05,
+      "loss": 0.7384,
+      "step": 259
+    },
+    {
+      "epoch": 4.56140350877193,
+      "grad_norm": 13.194774627685547,
+      "learning_rate": 2.4220183486238534e-05,
+      "loss": 0.7311,
+      "step": 260
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 9.523067474365234,
+      "learning_rate": 2.4192660550458717e-05,
+      "loss": 0.8254,
+      "step": 261
+    },
+    {
+      "epoch": 4.5964912280701755,
+      "grad_norm": 40.60873794555664,
+      "learning_rate": 2.41651376146789e-05,
+      "loss": 0.7756,
+      "step": 262
+    },
+    {
+      "epoch": 4.614035087719298,
+      "grad_norm": 20.81313133239746,
+      "learning_rate": 2.4137614678899085e-05,
+      "loss": 0.7474,
+      "step": 263
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 30.596458435058594,
+      "learning_rate": 2.4110091743119267e-05,
+      "loss": 0.8261,
+      "step": 264
+    },
+    {
+      "epoch": 4.649122807017544,
+      "grad_norm": 36.82753372192383,
+      "learning_rate": 2.408256880733945e-05,
+      "loss": 0.7422,
+      "step": 265
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 7.798766136169434,
+      "learning_rate": 2.4055045871559632e-05,
+      "loss": 0.7433,
+      "step": 266
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 10.442023277282715,
+      "learning_rate": 2.4027522935779818e-05,
+      "loss": 0.9382,
+      "step": 267
+    },
+    {
+      "epoch": 4.701754385964913,
+      "grad_norm": 7.129739761352539,
+      "learning_rate": 2.4e-05,
+      "loss": 0.4544,
+      "step": 268
+    },
+    {
+      "epoch": 4.719298245614035,
+      "grad_norm": 6.29140567779541,
+      "learning_rate": 2.3972477064220183e-05,
+      "loss": 0.5828,
+      "step": 269
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 14.20146369934082,
+      "learning_rate": 2.3944954128440366e-05,
+      "loss": 0.7544,
+      "step": 270
+    },
+    {
+      "epoch": 4.754385964912281,
+      "grad_norm": 25.83724594116211,
+      "learning_rate": 2.391743119266055e-05,
+      "loss": 0.8725,
+      "step": 271
+    },
+    {
+      "epoch": 4.771929824561403,
+      "grad_norm": 19.661861419677734,
+      "learning_rate": 2.3889908256880737e-05,
+      "loss": 0.9125,
+      "step": 272
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 7.706132888793945,
+      "learning_rate": 2.3862385321100916e-05,
+      "loss": 0.4359,
+      "step": 273
+    },
+    {
+      "epoch": 4.807017543859649,
+      "grad_norm": 13.709455490112305,
+      "learning_rate": 2.3834862385321102e-05,
+      "loss": 0.6747,
+      "step": 274
+    },
+    {
+      "epoch": 4.824561403508772,
+      "grad_norm": 27.501550674438477,
+      "learning_rate": 2.3807339449541285e-05,
+      "loss": 0.5231,
+      "step": 275
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 3.0259737968444824,
+      "learning_rate": 2.377981651376147e-05,
+      "loss": 0.5482,
+      "step": 276
+    },
+    {
+      "epoch": 4.859649122807017,
+      "grad_norm": 22.053667068481445,
+      "learning_rate": 2.375229357798165e-05,
+      "loss": 0.6844,
+      "step": 277
+    },
+    {
+      "epoch": 4.87719298245614,
+      "grad_norm": 21.116025924682617,
+      "learning_rate": 2.3724770642201836e-05,
+      "loss": 0.5924,
+      "step": 278
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 45.17483901977539,
+      "learning_rate": 2.3697247706422018e-05,
+      "loss": 0.9959,
+      "step": 279
+    },
+    {
+      "epoch": 4.912280701754386,
+      "grad_norm": 2.0751166343688965,
+      "learning_rate": 2.3669724770642204e-05,
+      "loss": 0.4215,
+      "step": 280
+    },
+    {
+      "epoch": 4.9298245614035086,
+      "grad_norm": 17.698915481567383,
+      "learning_rate": 2.3642201834862386e-05,
+      "loss": 0.8212,
+      "step": 281
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 11.810864448547363,
+      "learning_rate": 2.361467889908257e-05,
+      "loss": 0.7421,
+      "step": 282
+    },
+    {
+      "epoch": 4.964912280701754,
+      "grad_norm": 25.751745223999023,
+      "learning_rate": 2.3587155963302755e-05,
+      "loss": 0.8161,
+      "step": 283
+    },
+    {
+      "epoch": 4.982456140350877,
+      "grad_norm": 9.799457550048828,
+      "learning_rate": 2.3559633027522937e-05,
+      "loss": 0.6857,
+      "step": 284
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 10.592196464538574,
+      "learning_rate": 2.353211009174312e-05,
+      "loss": 0.8668,
+      "step": 285
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7011494252873564,
+      "eval_loss": 0.8508906364440918,
+      "eval_roc_auc": 0.8698378210083979,
+      "eval_runtime": 9.9199,
+      "eval_samples_per_second": 26.311,
+      "eval_steps_per_second": 1.714,
+      "step": 285
+    },
+    {
+      "epoch": 5.017543859649122,
+      "grad_norm": 29.6278133392334,
+      "learning_rate": 2.3504587155963302e-05,
+      "loss": 0.8195,
+      "step": 286
+    },
+    {
+      "epoch": 5.035087719298246,
+      "grad_norm": 8.088125228881836,
+      "learning_rate": 2.3477064220183488e-05,
+      "loss": 0.4163,
+      "step": 287
+    },
+    {
+      "epoch": 5.052631578947368,
+      "grad_norm": 2.1910946369171143,
+      "learning_rate": 2.344954128440367e-05,
+      "loss": 0.4273,
+      "step": 288
+    },
+    {
+      "epoch": 5.0701754385964914,
+      "grad_norm": 27.934329986572266,
+      "learning_rate": 2.3422018348623853e-05,
+      "loss": 0.7804,
+      "step": 289
+    },
+    {
+      "epoch": 5.087719298245614,
+      "grad_norm": 28.2197265625,
+      "learning_rate": 2.3394495412844035e-05,
+      "loss": 0.5328,
+      "step": 290
+    },
+    {
+      "epoch": 5.105263157894737,
+      "grad_norm": 20.080032348632812,
+      "learning_rate": 2.336697247706422e-05,
+      "loss": 0.9274,
+      "step": 291
+    },
+    {
+      "epoch": 5.12280701754386,
+      "grad_norm": 8.307439804077148,
+      "learning_rate": 2.3339449541284407e-05,
+      "loss": 0.7575,
+      "step": 292
+    },
+    {
+      "epoch": 5.140350877192983,
+      "grad_norm": 15.012224197387695,
+      "learning_rate": 2.3311926605504586e-05,
+      "loss": 0.6881,
+      "step": 293
+    },
+    {
+      "epoch": 5.157894736842105,
+      "grad_norm": 19.6754093170166,
+      "learning_rate": 2.3284403669724772e-05,
+      "loss": 0.5243,
+      "step": 294
+    },
+    {
+      "epoch": 5.175438596491228,
+      "grad_norm": 2.4565722942352295,
+      "learning_rate": 2.3256880733944955e-05,
+      "loss": 0.5858,
+      "step": 295
+    },
+    {
+      "epoch": 5.192982456140351,
+      "grad_norm": 12.843107223510742,
+      "learning_rate": 2.322935779816514e-05,
+      "loss": 0.7095,
+      "step": 296
+    },
+    {
+      "epoch": 5.2105263157894735,
+      "grad_norm": 3.9343743324279785,
+      "learning_rate": 2.320183486238532e-05,
+      "loss": 0.6592,
+      "step": 297
+    },
+    {
+      "epoch": 5.228070175438597,
+      "grad_norm": 21.421518325805664,
+      "learning_rate": 2.3174311926605505e-05,
+      "loss": 0.6792,
+      "step": 298
+    },
+    {
+      "epoch": 5.245614035087719,
+      "grad_norm": 29.099624633789062,
+      "learning_rate": 2.3146788990825688e-05,
+      "loss": 0.8901,
+      "step": 299
+    },
+    {
+      "epoch": 5.2631578947368425,
+      "grad_norm": 28.176443099975586,
+      "learning_rate": 2.3119266055045874e-05,
+      "loss": 0.677,
+      "step": 300
+    },
+    {
+      "epoch": 5.280701754385965,
+      "grad_norm": 49.84172439575195,
+      "learning_rate": 2.3091743119266053e-05,
+      "loss": 0.6627,
+      "step": 301
+    },
+    {
+      "epoch": 5.298245614035087,
+      "grad_norm": 11.457085609436035,
+      "learning_rate": 2.306422018348624e-05,
+      "loss": 0.3993,
+      "step": 302
+    },
+    {
+      "epoch": 5.315789473684211,
+      "grad_norm": 187.01239013671875,
+      "learning_rate": 2.3036697247706425e-05,
+      "loss": 0.6346,
+      "step": 303
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 3.9368183612823486,
+      "learning_rate": 2.3009174311926607e-05,
+      "loss": 0.8296,
+      "step": 304
+    },
+    {
+      "epoch": 5.350877192982456,
+      "grad_norm": 100.2872085571289,
+      "learning_rate": 2.298165137614679e-05,
+      "loss": 0.8781,
+      "step": 305
+    },
+    {
+      "epoch": 5.368421052631579,
+      "grad_norm": 24.209882736206055,
+      "learning_rate": 2.2954128440366972e-05,
+      "loss": 0.5137,
+      "step": 306
+    },
+    {
+      "epoch": 5.385964912280702,
+      "grad_norm": 3.6952366828918457,
+      "learning_rate": 2.2926605504587158e-05,
+      "loss": 0.6469,
+      "step": 307
+    },
+    {
+      "epoch": 5.4035087719298245,
+      "grad_norm": 29.038558959960938,
+      "learning_rate": 2.289908256880734e-05,
+      "loss": 0.8628,
+      "step": 308
+    },
+    {
+      "epoch": 5.421052631578947,
+      "grad_norm": 40.4047737121582,
+      "learning_rate": 2.2871559633027523e-05,
+      "loss": 1.065,
+      "step": 309
+    },
+    {
+      "epoch": 5.43859649122807,
+      "grad_norm": 43.313846588134766,
+      "learning_rate": 2.2844036697247705e-05,
+      "loss": 0.5773,
+      "step": 310
+    },
+    {
+      "epoch": 5.456140350877193,
+      "grad_norm": 7.541952133178711,
+      "learning_rate": 2.281651376146789e-05,
+      "loss": 0.7124,
+      "step": 311
+    },
+    {
+      "epoch": 5.473684210526316,
+      "grad_norm": 9.793838500976562,
+      "learning_rate": 2.2788990825688074e-05,
+      "loss": 0.5441,
+      "step": 312
+    },
+    {
+      "epoch": 5.491228070175438,
+      "grad_norm": 17.21137809753418,
+      "learning_rate": 2.2761467889908256e-05,
+      "loss": 0.5927,
+      "step": 313
+    },
+    {
+      "epoch": 5.508771929824562,
+      "grad_norm": 23.36025047302246,
+      "learning_rate": 2.2733944954128442e-05,
+      "loss": 0.8349,
+      "step": 314
+    },
+    {
+      "epoch": 5.526315789473684,
+      "grad_norm": 10.170068740844727,
+      "learning_rate": 2.2706422018348624e-05,
+      "loss": 0.7946,
+      "step": 315
+    },
+    {
+      "epoch": 5.543859649122807,
+      "grad_norm": 7.996306419372559,
+      "learning_rate": 2.267889908256881e-05,
+      "loss": 0.7086,
+      "step": 316
+    },
+    {
+      "epoch": 5.56140350877193,
+      "grad_norm": 4.8672027587890625,
+      "learning_rate": 2.265137614678899e-05,
+      "loss": 0.4384,
+      "step": 317
+    },
+    {
+      "epoch": 5.578947368421053,
+      "grad_norm": 14.543272972106934,
+      "learning_rate": 2.2623853211009175e-05,
+      "loss": 0.9153,
+      "step": 318
+    },
+    {
+      "epoch": 5.5964912280701755,
+      "grad_norm": 28.248517990112305,
+      "learning_rate": 2.2596330275229358e-05,
+      "loss": 0.4361,
+      "step": 319
+    },
+    {
+      "epoch": 5.614035087719298,
+      "grad_norm": 10.02695083618164,
+      "learning_rate": 2.2568807339449544e-05,
+      "loss": 0.5584,
+      "step": 320
+    },
+    {
+      "epoch": 5.631578947368421,
+      "grad_norm": 255.10897827148438,
+      "learning_rate": 2.2541284403669723e-05,
+      "loss": 0.812,
+      "step": 321
+    },
+    {
+      "epoch": 5.649122807017544,
+      "grad_norm": 29.070514678955078,
+      "learning_rate": 2.251376146788991e-05,
+      "loss": 0.7201,
+      "step": 322
+    },
+    {
+      "epoch": 5.666666666666667,
+      "grad_norm": 70.94801330566406,
+      "learning_rate": 2.2486238532110094e-05,
+      "loss": 0.7983,
+      "step": 323
+    },
+    {
+      "epoch": 5.684210526315789,
+      "grad_norm": 9.982966423034668,
+      "learning_rate": 2.2458715596330277e-05,
+      "loss": 0.3869,
+      "step": 324
+    },
+    {
+      "epoch": 5.701754385964913,
+      "grad_norm": 44.79756546020508,
+      "learning_rate": 2.243119266055046e-05,
+      "loss": 0.5092,
+      "step": 325
+    },
+    {
+      "epoch": 5.719298245614035,
+      "grad_norm": 14.900089263916016,
+      "learning_rate": 2.2403669724770642e-05,
+      "loss": 0.7798,
+      "step": 326
+    },
+    {
+      "epoch": 5.7368421052631575,
+      "grad_norm": 5.473045825958252,
+      "learning_rate": 2.2376146788990828e-05,
+      "loss": 0.527,
+      "step": 327
+    },
+    {
+      "epoch": 5.754385964912281,
+      "grad_norm": 5.292675495147705,
+      "learning_rate": 2.234862385321101e-05,
+      "loss": 0.599,
+      "step": 328
+    },
+    {
+      "epoch": 5.771929824561403,
+      "grad_norm": 7.409575462341309,
+      "learning_rate": 2.2321100917431193e-05,
+      "loss": 0.6566,
+      "step": 329
+    },
+    {
+      "epoch": 5.7894736842105265,
+      "grad_norm": 16.337003707885742,
+      "learning_rate": 2.2293577981651375e-05,
+      "loss": 0.421,
+      "step": 330
+    },
+    {
+      "epoch": 5.807017543859649,
+      "grad_norm": 10.473221778869629,
+      "learning_rate": 2.226605504587156e-05,
+      "loss": 0.6791,
+      "step": 331
+    },
+    {
+      "epoch": 5.824561403508772,
+      "grad_norm": 24.943050384521484,
+      "learning_rate": 2.2238532110091743e-05,
+      "loss": 1.0399,
+      "step": 332
+    },
+    {
+      "epoch": 5.842105263157895,
+      "grad_norm": 3.1360652446746826,
+      "learning_rate": 2.2211009174311926e-05,
+      "loss": 0.4585,
+      "step": 333
+    },
+    {
+      "epoch": 5.859649122807017,
+      "grad_norm": 5.742803573608398,
+      "learning_rate": 2.2183486238532112e-05,
+      "loss": 0.7424,
+      "step": 334
+    },
+    {
+      "epoch": 5.87719298245614,
+      "grad_norm": 10.02749252319336,
+      "learning_rate": 2.2155963302752294e-05,
+      "loss": 0.5505,
+      "step": 335
+    },
+    {
+      "epoch": 5.894736842105263,
+      "grad_norm": 11.549918174743652,
+      "learning_rate": 2.212844036697248e-05,
+      "loss": 0.7165,
+      "step": 336
+    },
+    {
+      "epoch": 5.912280701754386,
+      "grad_norm": 11.704695701599121,
+      "learning_rate": 2.210091743119266e-05,
+      "loss": 0.5897,
+      "step": 337
+    },
+    {
+      "epoch": 5.9298245614035086,
+      "grad_norm": 8.045721054077148,
+      "learning_rate": 2.2073394495412845e-05,
+      "loss": 0.5606,
+      "step": 338
+    },
+    {
+      "epoch": 5.947368421052632,
+      "grad_norm": 6.550736904144287,
+      "learning_rate": 2.2045871559633028e-05,
+      "loss": 0.4498,
+      "step": 339
+    },
+    {
+      "epoch": 5.964912280701754,
+      "grad_norm": 7.8976335525512695,
+      "learning_rate": 2.2018348623853213e-05,
+      "loss": 0.6578,
+      "step": 340
+    },
+    {
+      "epoch": 5.982456140350877,
+      "grad_norm": 19.857566833496094,
+      "learning_rate": 2.1990825688073392e-05,
+      "loss": 0.6241,
+      "step": 341
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 389.2672119140625,
+      "learning_rate": 2.196330275229358e-05,
+      "loss": 0.5834,
+      "step": 342
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.6436781609195402,
+      "eval_loss": 0.9111022353172302,
+      "eval_roc_auc": 0.8661937516704385,
+      "eval_runtime": 8.4523,
+      "eval_samples_per_second": 30.879,
+      "eval_steps_per_second": 2.011,
+      "step": 342
+    },
+    {
+      "epoch": 6.017543859649122,
+      "grad_norm": 6.004997253417969,
+      "learning_rate": 2.193577981651376e-05,
+      "loss": 0.4383,
+      "step": 343
+    },
+    {
+      "epoch": 6.035087719298246,
+      "grad_norm": 32.85868453979492,
+      "learning_rate": 2.1908256880733947e-05,
+      "loss": 0.8127,
+      "step": 344
+    },
+    {
+      "epoch": 6.052631578947368,
+      "grad_norm": 22.06620216369629,
+      "learning_rate": 2.188073394495413e-05,
+      "loss": 0.5588,
+      "step": 345
+    },
+    {
+      "epoch": 6.0701754385964914,
+      "grad_norm": 12.496269226074219,
+      "learning_rate": 2.185321100917431e-05,
+      "loss": 0.7202,
+      "step": 346
+    },
+    {
+      "epoch": 6.087719298245614,
+      "grad_norm": 39.94806671142578,
+      "learning_rate": 2.1825688073394497e-05,
+      "loss": 0.4362,
+      "step": 347
+    },
+    {
+      "epoch": 6.105263157894737,
+      "grad_norm": 5.034066677093506,
+      "learning_rate": 2.179816513761468e-05,
+      "loss": 0.6235,
+      "step": 348
+    },
+    {
+      "epoch": 6.12280701754386,
+      "grad_norm": 98.4699935913086,
+      "learning_rate": 2.1770642201834862e-05,
+      "loss": 0.8505,
+      "step": 349
+    },
+    {
+      "epoch": 6.140350877192983,
+      "grad_norm": 22.5294132232666,
+      "learning_rate": 2.1743119266055045e-05,
+      "loss": 0.6741,
+      "step": 350
+    },
+    {
+      "epoch": 6.157894736842105,
+      "grad_norm": 33.31329345703125,
+      "learning_rate": 2.171559633027523e-05,
+      "loss": 0.6783,
+      "step": 351
+    },
+    {
+      "epoch": 6.175438596491228,
+      "grad_norm": 2.9709410667419434,
+      "learning_rate": 2.1688073394495413e-05,
+      "loss": 0.3269,
+      "step": 352
+    },
+    {
+      "epoch": 6.192982456140351,
+      "grad_norm": 14.349902153015137,
+      "learning_rate": 2.1660550458715596e-05,
+      "loss": 0.5928,
+      "step": 353
+    },
+    {
+      "epoch": 6.2105263157894735,
+      "grad_norm": 6.400623321533203,
+      "learning_rate": 2.163302752293578e-05,
+      "loss": 0.5708,
+      "step": 354
+    },
+    {
+      "epoch": 6.228070175438597,
+      "grad_norm": 40.32659912109375,
+      "learning_rate": 2.1605504587155964e-05,
+      "loss": 0.4459,
+      "step": 355
+    },
+    {
+      "epoch": 6.245614035087719,
+      "grad_norm": 16.54581069946289,
+      "learning_rate": 2.157798165137615e-05,
+      "loss": 1.0475,
+      "step": 356
+    },
+    {
+      "epoch": 6.2631578947368425,
+      "grad_norm": 7.048498630523682,
+      "learning_rate": 2.155045871559633e-05,
+      "loss": 0.6128,
+      "step": 357
+    },
+    {
+      "epoch": 6.280701754385965,
+      "grad_norm": 18.782760620117188,
+      "learning_rate": 2.1522935779816515e-05,
+      "loss": 0.6483,
+      "step": 358
+    },
+    {
+      "epoch": 6.298245614035087,
+      "grad_norm": 13.411852836608887,
+      "learning_rate": 2.1495412844036697e-05,
+      "loss": 0.6957,
+      "step": 359
+    },
+    {
+      "epoch": 6.315789473684211,
+      "grad_norm": 3.7218034267425537,
+      "learning_rate": 2.1467889908256883e-05,
+      "loss": 0.268,
+      "step": 360
+    },
+    {
+      "epoch": 6.333333333333333,
+      "grad_norm": 5.883011817932129,
+      "learning_rate": 2.1440366972477062e-05,
+      "loss": 0.5415,
+      "step": 361
+    },
+    {
+      "epoch": 6.350877192982456,
+      "grad_norm": 6.700387477874756,
+      "learning_rate": 2.1412844036697248e-05,
+      "loss": 0.4707,
+      "step": 362
+    },
+    {
+      "epoch": 6.368421052631579,
+      "grad_norm": 12.57919692993164,
+      "learning_rate": 2.138532110091743e-05,
+      "loss": 0.5245,
+      "step": 363
+    },
+    {
+      "epoch": 6.385964912280702,
+      "grad_norm": 23.14158058166504,
+      "learning_rate": 2.1357798165137616e-05,
+      "loss": 0.4275,
+      "step": 364
+    },
+    {
+      "epoch": 6.4035087719298245,
+      "grad_norm": 20.404285430908203,
+      "learning_rate": 2.13302752293578e-05,
+      "loss": 0.7325,
+      "step": 365
+    },
+    {
+      "epoch": 6.421052631578947,
+      "grad_norm": 7.9764580726623535,
+      "learning_rate": 2.130275229357798e-05,
+      "loss": 0.469,
+      "step": 366
+    },
+    {
+      "epoch": 6.43859649122807,
+      "grad_norm": 11.24958324432373,
+      "learning_rate": 2.1275229357798167e-05,
+      "loss": 0.7704,
+      "step": 367
+    },
+    {
+      "epoch": 6.456140350877193,
+      "grad_norm": 3.0611751079559326,
+      "learning_rate": 2.124770642201835e-05,
+      "loss": 0.3323,
+      "step": 368
+    },
+    {
+      "epoch": 6.473684210526316,
+      "grad_norm": 40.933162689208984,
+      "learning_rate": 2.1220183486238532e-05,
+      "loss": 0.5066,
+      "step": 369
+    },
+    {
+      "epoch": 6.491228070175438,
+      "grad_norm": 9.126625061035156,
+      "learning_rate": 2.1192660550458715e-05,
+      "loss": 0.4907,
+      "step": 370
+    },
+    {
+      "epoch": 6.508771929824562,
+      "grad_norm": 9.465023040771484,
+      "learning_rate": 2.11651376146789e-05,
+      "loss": 0.5455,
+      "step": 371
+    },
+    {
+      "epoch": 6.526315789473684,
+      "grad_norm": 24.881996154785156,
+      "learning_rate": 2.1137614678899083e-05,
+      "loss": 0.6284,
+      "step": 372
+    },
+    {
+      "epoch": 6.543859649122807,
+      "grad_norm": 32.151466369628906,
+      "learning_rate": 2.1110091743119266e-05,
+      "loss": 0.4083,
+      "step": 373
+    },
+    {
+      "epoch": 6.56140350877193,
+      "grad_norm": 12.968864440917969,
+      "learning_rate": 2.1082568807339448e-05,
+      "loss": 0.5154,
+      "step": 374
+    },
+    {
+      "epoch": 6.578947368421053,
+      "grad_norm": 64.264892578125,
+      "learning_rate": 2.1055045871559634e-05,
+      "loss": 0.2809,
+      "step": 375
+    },
+    {
+      "epoch": 6.5964912280701755,
+      "grad_norm": 19.623029708862305,
+      "learning_rate": 2.102752293577982e-05,
+      "loss": 0.7864,
+      "step": 376
+    },
+    {
+      "epoch": 6.614035087719298,
+      "grad_norm": 7.40114688873291,
+      "learning_rate": 2.1e-05,
+      "loss": 0.5327,
+      "step": 377
+    },
+    {
+      "epoch": 6.631578947368421,
+      "grad_norm": 15.018710136413574,
+      "learning_rate": 2.0972477064220185e-05,
+      "loss": 0.5948,
+      "step": 378
+    },
+    {
+      "epoch": 6.649122807017544,
+      "grad_norm": 6.5761847496032715,
+      "learning_rate": 2.0944954128440367e-05,
+      "loss": 0.5606,
+      "step": 379
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 21.176044464111328,
+      "learning_rate": 2.0917431192660553e-05,
+      "loss": 0.8495,
+      "step": 380
+    },
+    {
+      "epoch": 6.684210526315789,
+      "grad_norm": 15.794927597045898,
+      "learning_rate": 2.0889908256880732e-05,
+      "loss": 0.2922,
+      "step": 381
+    },
+    {
+      "epoch": 6.701754385964913,
+      "grad_norm": 21.976320266723633,
+      "learning_rate": 2.0862385321100918e-05,
+      "loss": 0.4193,
+      "step": 382
+    },
+    {
+      "epoch": 6.719298245614035,
+      "grad_norm": 127.4170150756836,
+      "learning_rate": 2.08348623853211e-05,
+      "loss": 0.7539,
+      "step": 383
+    },
+    {
+      "epoch": 6.7368421052631575,
+      "grad_norm": 23.519792556762695,
+      "learning_rate": 2.0807339449541286e-05,
+      "loss": 0.8324,
+      "step": 384
+    },
+    {
+      "epoch": 6.754385964912281,
+      "grad_norm": 34.533729553222656,
+      "learning_rate": 2.0779816513761465e-05,
+      "loss": 0.635,
+      "step": 385
+    },
+    {
+      "epoch": 6.771929824561403,
+      "grad_norm": 48.43922805786133,
+      "learning_rate": 2.075229357798165e-05,
+      "loss": 0.7622,
+      "step": 386
+    },
+    {
+      "epoch": 6.7894736842105265,
+      "grad_norm": 9.6075439453125,
+      "learning_rate": 2.0724770642201837e-05,
+      "loss": 0.4885,
+      "step": 387
+    },
+    {
+      "epoch": 6.807017543859649,
+      "grad_norm": 25.955398559570312,
+      "learning_rate": 2.069724770642202e-05,
+      "loss": 0.2685,
+      "step": 388
+    },
+    {
+      "epoch": 6.824561403508772,
+      "grad_norm": 17.847700119018555,
+      "learning_rate": 2.0669724770642202e-05,
+      "loss": 0.3958,
+      "step": 389
+    },
+    {
+      "epoch": 6.842105263157895,
+      "grad_norm": 13.096657752990723,
+      "learning_rate": 2.0642201834862385e-05,
+      "loss": 0.9942,
+      "step": 390
+    },
+    {
+      "epoch": 6.859649122807017,
+      "grad_norm": 11.541616439819336,
+      "learning_rate": 2.061467889908257e-05,
+      "loss": 0.5704,
+      "step": 391
+    },
+    {
+      "epoch": 6.87719298245614,
+      "grad_norm": 3.565450668334961,
+      "learning_rate": 2.0587155963302753e-05,
+      "loss": 0.6809,
+      "step": 392
+    },
+    {
+      "epoch": 6.894736842105263,
+      "grad_norm": 27.333837509155273,
+      "learning_rate": 2.0559633027522935e-05,
+      "loss": 0.6584,
+      "step": 393
+    },
+    {
+      "epoch": 6.912280701754386,
+      "grad_norm": 15.244169235229492,
+      "learning_rate": 2.0532110091743118e-05,
+      "loss": 0.6753,
+      "step": 394
+    },
+    {
+      "epoch": 6.9298245614035086,
+      "grad_norm": 18.803529739379883,
+      "learning_rate": 2.0504587155963304e-05,
+      "loss": 0.5975,
+      "step": 395
+    },
+    {
+      "epoch": 6.947368421052632,
+      "grad_norm": 7.317320346832275,
+      "learning_rate": 2.047706422018349e-05,
+      "loss": 0.9889,
+      "step": 396
+    },
+    {
+      "epoch": 6.964912280701754,
+      "grad_norm": 19.616382598876953,
+      "learning_rate": 2.044954128440367e-05,
+      "loss": 0.6709,
+      "step": 397
+    },
+    {
+      "epoch": 6.982456140350877,
+      "grad_norm": 46.652469635009766,
+      "learning_rate": 2.0422018348623855e-05,
+      "loss": 0.5268,
+      "step": 398
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 11.762057304382324,
+      "learning_rate": 2.0394495412844037e-05,
+      "loss": 0.9838,
+      "step": 399
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.6819923371647509,
+      "eval_loss": 0.8196430802345276,
+      "eval_roc_auc": 0.8813437942252584,
+      "eval_runtime": 9.6385,
+      "eval_samples_per_second": 27.079,
+      "eval_steps_per_second": 1.764,
+      "step": 399
+    },
+    {
+      "epoch": 7.017543859649122,
+      "grad_norm": 14.474690437316895,
+      "learning_rate": 2.0366972477064223e-05,
+      "loss": 0.3408,
+      "step": 400
+    },
+    {
+      "epoch": 7.035087719298246,
+      "grad_norm": 19.78782081604004,
+      "learning_rate": 2.0339449541284402e-05,
+      "loss": 0.4244,
+      "step": 401
+    },
+    {
+      "epoch": 7.052631578947368,
+      "grad_norm": 10.25345230102539,
+      "learning_rate": 2.0311926605504588e-05,
+      "loss": 0.6707,
+      "step": 402
+    },
+    {
+      "epoch": 7.0701754385964914,
+      "grad_norm": 2.85701060295105,
+      "learning_rate": 2.028440366972477e-05,
+      "loss": 0.2397,
+      "step": 403
+    },
+    {
+      "epoch": 7.087719298245614,
+      "grad_norm": 6.676547527313232,
+      "learning_rate": 2.0256880733944956e-05,
+      "loss": 0.3686,
+      "step": 404
+    },
+    {
+      "epoch": 7.105263157894737,
+      "grad_norm": 165.5542755126953,
+      "learning_rate": 2.0229357798165135e-05,
+      "loss": 0.5689,
+      "step": 405
+    },
+    {
+      "epoch": 7.12280701754386,
+      "grad_norm": 14.226078033447266,
+      "learning_rate": 2.020183486238532e-05,
+      "loss": 0.3423,
+      "step": 406
+    },
+    {
+      "epoch": 7.140350877192983,
+      "grad_norm": 5.127176761627197,
+      "learning_rate": 2.0174311926605507e-05,
+      "loss": 0.6358,
+      "step": 407
+    },
+    {
+      "epoch": 7.157894736842105,
+      "grad_norm": 21.368282318115234,
+      "learning_rate": 2.014678899082569e-05,
+      "loss": 0.5103,
+      "step": 408
+    },
+    {
+      "epoch": 7.175438596491228,
+      "grad_norm": 3.635981321334839,
+      "learning_rate": 2.0119266055045872e-05,
+      "loss": 0.3426,
+      "step": 409
+    },
+    {
+      "epoch": 7.192982456140351,
+      "grad_norm": 8.354199409484863,
+      "learning_rate": 2.0091743119266054e-05,
+      "loss": 0.4433,
+      "step": 410
+    },
+    {
+      "epoch": 7.2105263157894735,
+      "grad_norm": 125.07445526123047,
+      "learning_rate": 2.006422018348624e-05,
+      "loss": 0.5754,
+      "step": 411
+    },
+    {
+      "epoch": 7.228070175438597,
+      "grad_norm": 18.86847496032715,
+      "learning_rate": 2.0036697247706423e-05,
+      "loss": 0.6004,
+      "step": 412
+    },
+    {
+      "epoch": 7.245614035087719,
+      "grad_norm": 13.290031433105469,
+      "learning_rate": 2.0009174311926605e-05,
+      "loss": 0.6924,
+      "step": 413
+    },
+    {
+      "epoch": 7.2631578947368425,
+      "grad_norm": 7.521588325500488,
+      "learning_rate": 1.9981651376146788e-05,
+      "loss": 0.4925,
+      "step": 414
+    },
+    {
+      "epoch": 7.280701754385965,
+      "grad_norm": 14.16098690032959,
+      "learning_rate": 1.9954128440366974e-05,
+      "loss": 0.6182,
+      "step": 415
+    },
+    {
+      "epoch": 7.298245614035087,
+      "grad_norm": 12.972576141357422,
+      "learning_rate": 1.9926605504587156e-05,
+      "loss": 0.4994,
+      "step": 416
+    },
+    {
+      "epoch": 7.315789473684211,
+      "grad_norm": 7.475390911102295,
+      "learning_rate": 1.989908256880734e-05,
+      "loss": 0.5975,
+      "step": 417
+    },
+    {
+      "epoch": 7.333333333333333,
+      "grad_norm": 19.82378387451172,
+      "learning_rate": 1.9871559633027524e-05,
+      "loss": 0.4584,
+      "step": 418
+    },
+    {
+      "epoch": 7.350877192982456,
+      "grad_norm": 7.333564281463623,
+      "learning_rate": 1.9844036697247707e-05,
+      "loss": 0.5651,
+      "step": 419
+    },
+    {
+      "epoch": 7.368421052631579,
+      "grad_norm": 12.829773902893066,
+      "learning_rate": 1.9816513761467893e-05,
+      "loss": 0.4198,
+      "step": 420
+    },
+    {
+      "epoch": 7.385964912280702,
+      "grad_norm": 3.876220464706421,
+      "learning_rate": 1.9788990825688072e-05,
+      "loss": 0.2843,
+      "step": 421
+    },
+    {
+      "epoch": 7.4035087719298245,
+      "grad_norm": 10.361215591430664,
+      "learning_rate": 1.9761467889908258e-05,
+      "loss": 0.4935,
+      "step": 422
+    },
+    {
+      "epoch": 7.421052631578947,
+      "grad_norm": 59.4860954284668,
+      "learning_rate": 1.973394495412844e-05,
+      "loss": 0.5763,
+      "step": 423
+    },
+    {
+      "epoch": 7.43859649122807,
+      "grad_norm": 10.748738288879395,
+      "learning_rate": 1.9706422018348626e-05,
+      "loss": 0.7249,
+      "step": 424
+    },
+    {
+      "epoch": 7.456140350877193,
+      "grad_norm": 19.900938034057617,
+      "learning_rate": 1.9678899082568805e-05,
+      "loss": 0.3448,
+      "step": 425
+    },
+    {
+      "epoch": 7.473684210526316,
+      "grad_norm": 9.298110961914062,
+      "learning_rate": 1.965137614678899e-05,
+      "loss": 0.424,
+      "step": 426
+    },
+    {
+      "epoch": 7.491228070175438,
+      "grad_norm": 6.225700855255127,
+      "learning_rate": 1.9623853211009173e-05,
+      "loss": 0.5622,
+      "step": 427
+    },
+    {
+      "epoch": 7.508771929824562,
+      "grad_norm": 7.592733860015869,
+      "learning_rate": 1.959633027522936e-05,
+      "loss": 0.327,
+      "step": 428
+    },
+    {
+      "epoch": 7.526315789473684,
+      "grad_norm": 15.926929473876953,
+      "learning_rate": 1.9568807339449542e-05,
+      "loss": 0.5397,
+      "step": 429
+    },
+    {
+      "epoch": 7.543859649122807,
+      "grad_norm": 5.76393461227417,
+      "learning_rate": 1.9541284403669724e-05,
+      "loss": 0.6615,
+      "step": 430
+    },
+    {
+      "epoch": 7.56140350877193,
+      "grad_norm": 9.405485153198242,
+      "learning_rate": 1.951376146788991e-05,
+      "loss": 0.5491,
+      "step": 431
+    },
+    {
+      "epoch": 7.578947368421053,
+      "grad_norm": 6.3222479820251465,
+      "learning_rate": 1.9486238532110093e-05,
+      "loss": 0.5539,
+      "step": 432
+    },
+    {
+      "epoch": 7.5964912280701755,
+      "grad_norm": 15.603589057922363,
+      "learning_rate": 1.9458715596330275e-05,
+      "loss": 0.352,
+      "step": 433
+    },
+    {
+      "epoch": 7.614035087719298,
+      "grad_norm": 54.22720718383789,
+      "learning_rate": 1.9431192660550458e-05,
+      "loss": 0.5936,
+      "step": 434
+    },
+    {
+      "epoch": 7.631578947368421,
+      "grad_norm": 7.901265621185303,
+      "learning_rate": 1.9403669724770643e-05,
+      "loss": 0.6319,
+      "step": 435
+    },
+    {
+      "epoch": 7.649122807017544,
+      "grad_norm": 16.199270248413086,
+      "learning_rate": 1.9376146788990826e-05,
+      "loss": 0.5468,
+      "step": 436
+    },
+    {
+      "epoch": 7.666666666666667,
+      "grad_norm": 84.76512908935547,
+      "learning_rate": 1.934862385321101e-05,
+      "loss": 0.5745,
+      "step": 437
+    },
+    {
+      "epoch": 7.684210526315789,
+      "grad_norm": 8.624265670776367,
+      "learning_rate": 1.9321100917431194e-05,
+      "loss": 0.5141,
+      "step": 438
+    },
+    {
+      "epoch": 7.701754385964913,
+      "grad_norm": 4.701018333435059,
+      "learning_rate": 1.9293577981651377e-05,
+      "loss": 0.2706,
+      "step": 439
+    },
+    {
+      "epoch": 7.719298245614035,
+      "grad_norm": 3.979341745376587,
+      "learning_rate": 1.9266055045871563e-05,
+      "loss": 0.4677,
+      "step": 440
+    },
+    {
+      "epoch": 7.7368421052631575,
+      "grad_norm": 10.237481117248535,
+      "learning_rate": 1.923853211009174e-05,
+      "loss": 0.6246,
+      "step": 441
+    },
+    {
+      "epoch": 7.754385964912281,
+      "grad_norm": 16.742145538330078,
+      "learning_rate": 1.9211009174311927e-05,
+      "loss": 0.5573,
+      "step": 442
+    },
+    {
+      "epoch": 7.771929824561403,
+      "grad_norm": 7.850118637084961,
+      "learning_rate": 1.918348623853211e-05,
+      "loss": 0.3705,
+      "step": 443
+    },
+    {
+      "epoch": 7.7894736842105265,
+      "grad_norm": 7.423269271850586,
+      "learning_rate": 1.9155963302752296e-05,
+      "loss": 0.5847,
+      "step": 444
+    },
+    {
+      "epoch": 7.807017543859649,
+      "grad_norm": 3.7411139011383057,
+      "learning_rate": 1.9128440366972475e-05,
+      "loss": 0.3861,
+      "step": 445
+    },
+    {
+      "epoch": 7.824561403508772,
+      "grad_norm": 131.05946350097656,
+      "learning_rate": 1.910091743119266e-05,
+      "loss": 0.5242,
+      "step": 446
+    },
+    {
+      "epoch": 7.842105263157895,
+      "grad_norm": 12.686647415161133,
+      "learning_rate": 1.9073394495412843e-05,
+      "loss": 0.2422,
+      "step": 447
+    },
+    {
+      "epoch": 7.859649122807017,
+      "grad_norm": 7.073160171508789,
+      "learning_rate": 1.904587155963303e-05,
+      "loss": 0.2323,
+      "step": 448
+    },
+    {
+      "epoch": 7.87719298245614,
+      "grad_norm": 4.771604061126709,
+      "learning_rate": 1.901834862385321e-05,
+      "loss": 0.2788,
+      "step": 449
+    },
+    {
+      "epoch": 7.894736842105263,
+      "grad_norm": 11.123143196105957,
+      "learning_rate": 1.8990825688073394e-05,
+      "loss": 0.3865,
+      "step": 450
+    },
+    {
+      "epoch": 7.912280701754386,
+      "grad_norm": 19.361055374145508,
+      "learning_rate": 1.896330275229358e-05,
+      "loss": 0.4511,
+      "step": 451
+    },
+    {
+      "epoch": 7.9298245614035086,
+      "grad_norm": 8.789932250976562,
+      "learning_rate": 1.8935779816513762e-05,
+      "loss": 0.3365,
+      "step": 452
+    },
+    {
+      "epoch": 7.947368421052632,
+      "grad_norm": 12.437773704528809,
+      "learning_rate": 1.8908256880733945e-05,
+      "loss": 0.3662,
+      "step": 453
+    },
+    {
+      "epoch": 7.964912280701754,
+      "grad_norm": 17.096485137939453,
+      "learning_rate": 1.8880733944954127e-05,
+      "loss": 0.5647,
+      "step": 454
+    },
+    {
+      "epoch": 7.982456140350877,
+      "grad_norm": 20.25951385498047,
+      "learning_rate": 1.8853211009174313e-05,
+      "loss": 0.4497,
+      "step": 455
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 14.128129005432129,
+      "learning_rate": 1.8825688073394496e-05,
+      "loss": 0.3699,
+      "step": 456
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.6743295019157088,
+      "eval_loss": 0.9114387035369873,
+      "eval_roc_auc": 0.8783069858814981,
+      "eval_runtime": 9.7288,
+      "eval_samples_per_second": 26.827,
+      "eval_steps_per_second": 1.747,
+      "step": 456
+    },
+    {
+      "epoch": 8.017543859649123,
+      "grad_norm": 14.421273231506348,
+      "learning_rate": 1.8798165137614678e-05,
+      "loss": 0.4197,
+      "step": 457
+    },
+    {
+      "epoch": 8.035087719298245,
+      "grad_norm": 7.326334476470947,
+      "learning_rate": 1.877064220183486e-05,
+      "loss": 0.3367,
+      "step": 458
+    },
+    {
+      "epoch": 8.052631578947368,
+      "grad_norm": 9.997421264648438,
+      "learning_rate": 1.8743119266055047e-05,
+      "loss": 0.4693,
+      "step": 459
+    },
+    {
+      "epoch": 8.070175438596491,
+      "grad_norm": 16.052030563354492,
+      "learning_rate": 1.8715596330275232e-05,
+      "loss": 0.3885,
+      "step": 460
+    },
+    {
+      "epoch": 8.087719298245615,
+      "grad_norm": 5.081548690795898,
+      "learning_rate": 1.868807339449541e-05,
+      "loss": 0.3361,
+      "step": 461
+    },
+    {
+      "epoch": 8.105263157894736,
+      "grad_norm": 12.288408279418945,
+      "learning_rate": 1.8660550458715597e-05,
+      "loss": 0.4407,
+      "step": 462
+    },
+    {
+      "epoch": 8.12280701754386,
+      "grad_norm": 20.765512466430664,
+      "learning_rate": 1.863302752293578e-05,
+      "loss": 0.3498,
+      "step": 463
+    },
+    {
+      "epoch": 8.140350877192983,
+      "grad_norm": 16.51128578186035,
+      "learning_rate": 1.8605504587155966e-05,
+      "loss": 0.5519,
+      "step": 464
+    },
+    {
+      "epoch": 8.157894736842104,
+      "grad_norm": 9.807088851928711,
+      "learning_rate": 1.8577981651376145e-05,
+      "loss": 0.4753,
+      "step": 465
+    },
+    {
+      "epoch": 8.175438596491228,
+      "grad_norm": 27.620849609375,
+      "learning_rate": 1.855045871559633e-05,
+      "loss": 0.4292,
+      "step": 466
+    },
+    {
+      "epoch": 8.192982456140351,
+      "grad_norm": 3.228712320327759,
+      "learning_rate": 1.8522935779816513e-05,
+      "loss": 0.2354,
+      "step": 467
+    },
+    {
+      "epoch": 8.210526315789474,
+      "grad_norm": 19.046974182128906,
+      "learning_rate": 1.84954128440367e-05,
+      "loss": 0.7982,
+      "step": 468
+    },
+    {
+      "epoch": 8.228070175438596,
+      "grad_norm": 35.34560775756836,
+      "learning_rate": 1.8467889908256878e-05,
+      "loss": 0.6045,
+      "step": 469
+    },
+    {
+      "epoch": 8.24561403508772,
+      "grad_norm": 19.82710838317871,
+      "learning_rate": 1.8440366972477064e-05,
+      "loss": 0.2748,
+      "step": 470
+    },
+    {
+      "epoch": 8.263157894736842,
+      "grad_norm": 107.26939392089844,
+      "learning_rate": 1.841284403669725e-05,
+      "loss": 0.5046,
+      "step": 471
+    },
+    {
+      "epoch": 8.280701754385966,
+      "grad_norm": 9.757340431213379,
+      "learning_rate": 1.8385321100917432e-05,
+      "loss": 0.2989,
+      "step": 472
+    },
+    {
+      "epoch": 8.298245614035087,
+      "grad_norm": 42.89109802246094,
+      "learning_rate": 1.8357798165137615e-05,
+      "loss": 0.4882,
+      "step": 473
+    },
+    {
+      "epoch": 8.31578947368421,
+      "grad_norm": 27.338111877441406,
+      "learning_rate": 1.8330275229357797e-05,
+      "loss": 0.4523,
+      "step": 474
+    },
+    {
+      "epoch": 8.333333333333334,
+      "grad_norm": 3.047417163848877,
+      "learning_rate": 1.8302752293577983e-05,
+      "loss": 0.2369,
+      "step": 475
+    },
+    {
+      "epoch": 8.350877192982455,
+      "grad_norm": 27.468341827392578,
+      "learning_rate": 1.8275229357798166e-05,
+      "loss": 0.4376,
+      "step": 476
+    },
+    {
+      "epoch": 8.368421052631579,
+      "grad_norm": 4.080908298492432,
+      "learning_rate": 1.8247706422018348e-05,
+      "loss": 0.3097,
+      "step": 477
+    },
+    {
+      "epoch": 8.385964912280702,
+      "grad_norm": 15.838717460632324,
+      "learning_rate": 1.822018348623853e-05,
+      "loss": 0.424,
+      "step": 478
+    },
+    {
+      "epoch": 8.403508771929825,
+      "grad_norm": 9.841987609863281,
+      "learning_rate": 1.8192660550458716e-05,
+      "loss": 0.5209,
+      "step": 479
+    },
+    {
+      "epoch": 8.421052631578947,
+      "grad_norm": 11.688602447509766,
+      "learning_rate": 1.8165137614678902e-05,
+      "loss": 0.2871,
+      "step": 480
+    },
+    {
+      "epoch": 8.43859649122807,
+      "grad_norm": 42.60127639770508,
+      "learning_rate": 1.813761467889908e-05,
+      "loss": 0.7017,
+      "step": 481
+    },
+    {
+      "epoch": 8.456140350877194,
+      "grad_norm": 21.921663284301758,
+      "learning_rate": 1.8110091743119267e-05,
+      "loss": 0.6333,
+      "step": 482
+    },
+    {
+      "epoch": 8.473684210526315,
+      "grad_norm": 14.230140686035156,
+      "learning_rate": 1.808256880733945e-05,
+      "loss": 0.4494,
+      "step": 483
+    },
+    {
+      "epoch": 8.491228070175438,
+      "grad_norm": 4.411128997802734,
+      "learning_rate": 1.8055045871559636e-05,
+      "loss": 0.2,
+      "step": 484
+    },
+    {
+      "epoch": 8.508771929824562,
+      "grad_norm": 4.632619857788086,
+      "learning_rate": 1.8027522935779815e-05,
+      "loss": 0.3362,
+      "step": 485
+    },
+    {
+      "epoch": 8.526315789473685,
+      "grad_norm": 11.009140014648438,
+      "learning_rate": 1.8e-05,
+      "loss": 0.2433,
+      "step": 486
+    },
+    {
+      "epoch": 8.543859649122806,
+      "grad_norm": 35.23042297363281,
+      "learning_rate": 1.7972477064220183e-05,
+      "loss": 0.4174,
+      "step": 487
+    },
+    {
+      "epoch": 8.56140350877193,
+      "grad_norm": 6.151736259460449,
+      "learning_rate": 1.794495412844037e-05,
+      "loss": 0.3024,
+      "step": 488
+    },
+    {
+      "epoch": 8.578947368421053,
+      "grad_norm": 25.69590950012207,
+      "learning_rate": 1.7917431192660548e-05,
+      "loss": 0.7038,
+      "step": 489
+    },
+    {
+      "epoch": 8.596491228070175,
+      "grad_norm": 20.76173973083496,
+      "learning_rate": 1.7889908256880734e-05,
+      "loss": 0.6203,
+      "step": 490
+    },
+    {
+      "epoch": 8.614035087719298,
+      "grad_norm": 17.6284122467041,
+      "learning_rate": 1.786238532110092e-05,
+      "loss": 0.3738,
+      "step": 491
+    },
+    {
+      "epoch": 8.631578947368421,
+      "grad_norm": 6.477032661437988,
+      "learning_rate": 1.7834862385321102e-05,
+      "loss": 0.4933,
+      "step": 492
+    },
+    {
+      "epoch": 8.649122807017545,
+      "grad_norm": 11.045804023742676,
+      "learning_rate": 1.7807339449541285e-05,
+      "loss": 0.4582,
+      "step": 493
+    },
+    {
+      "epoch": 8.666666666666666,
+      "grad_norm": 7.45139741897583,
+      "learning_rate": 1.7779816513761467e-05,
+      "loss": 0.2789,
+      "step": 494
+    },
+    {
+      "epoch": 8.68421052631579,
+      "grad_norm": 6.092661380767822,
+      "learning_rate": 1.7752293577981653e-05,
+      "loss": 0.5297,
+      "step": 495
+    },
+    {
+      "epoch": 8.701754385964913,
+      "grad_norm": 9.822498321533203,
+      "learning_rate": 1.7724770642201835e-05,
+      "loss": 0.4994,
+      "step": 496
+    },
+    {
+      "epoch": 8.719298245614034,
+      "grad_norm": 25.944725036621094,
+      "learning_rate": 1.7697247706422018e-05,
+      "loss": 0.5637,
+      "step": 497
+    },
+    {
+      "epoch": 8.736842105263158,
+      "grad_norm": 7.313636302947998,
+      "learning_rate": 1.76697247706422e-05,
+      "loss": 0.3809,
+      "step": 498
+    },
+    {
+      "epoch": 8.75438596491228,
+      "grad_norm": 28.630786895751953,
+      "learning_rate": 1.7642201834862386e-05,
+      "loss": 0.2553,
+      "step": 499
+    },
+    {
+      "epoch": 8.771929824561404,
+      "grad_norm": 13.779499053955078,
+      "learning_rate": 1.761467889908257e-05,
+      "loss": 0.6088,
+      "step": 500
+    },
+    {
+      "epoch": 8.789473684210526,
+      "grad_norm": 13.866922378540039,
+      "learning_rate": 1.758715596330275e-05,
+      "loss": 0.3418,
+      "step": 501
+    },
+    {
+      "epoch": 8.807017543859649,
+      "grad_norm": 9.200101852416992,
+      "learning_rate": 1.7559633027522937e-05,
+      "loss": 0.5261,
+      "step": 502
+    },
+    {
+      "epoch": 8.824561403508772,
+      "grad_norm": 4.6678853034973145,
+      "learning_rate": 1.753211009174312e-05,
+      "loss": 0.2657,
+      "step": 503
+    },
+    {
+      "epoch": 8.842105263157894,
+      "grad_norm": 13.394104957580566,
+      "learning_rate": 1.7504587155963305e-05,
+      "loss": 0.4874,
+      "step": 504
+    },
+    {
+      "epoch": 8.859649122807017,
+      "grad_norm": 8.163538932800293,
+      "learning_rate": 1.7477064220183484e-05,
+      "loss": 0.4425,
+      "step": 505
+    },
+    {
+      "epoch": 8.87719298245614,
+      "grad_norm": 11.24685001373291,
+      "learning_rate": 1.744954128440367e-05,
+      "loss": 0.4143,
+      "step": 506
+    },
+    {
+      "epoch": 8.894736842105264,
+      "grad_norm": 15.51749324798584,
+      "learning_rate": 1.7422018348623853e-05,
+      "loss": 0.2075,
+      "step": 507
+    },
+    {
+      "epoch": 8.912280701754385,
+      "grad_norm": 18.03982162475586,
+      "learning_rate": 1.739449541284404e-05,
+      "loss": 0.3895,
+      "step": 508
+    },
+    {
+      "epoch": 8.929824561403509,
+      "grad_norm": 30.238689422607422,
+      "learning_rate": 1.7366972477064218e-05,
+      "loss": 0.502,
+      "step": 509
+    },
+    {
+      "epoch": 8.947368421052632,
+      "grad_norm": 17.355592727661133,
+      "learning_rate": 1.7339449541284404e-05,
+      "loss": 0.3032,
+      "step": 510
+    },
+    {
+      "epoch": 8.964912280701755,
+      "grad_norm": 8.36916732788086,
+      "learning_rate": 1.7311926605504586e-05,
+      "loss": 0.4566,
+      "step": 511
+    },
+    {
+      "epoch": 8.982456140350877,
+      "grad_norm": 4.937262535095215,
+      "learning_rate": 1.7284403669724772e-05,
+      "loss": 0.2615,
+      "step": 512
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 61.639869689941406,
+      "learning_rate": 1.7256880733944954e-05,
+      "loss": 0.6482,
+      "step": 513
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.7126436781609196,
+      "eval_loss": 0.8051581382751465,
+      "eval_roc_auc": 0.8928142130921077,
+      "eval_runtime": 9.5241,
+      "eval_samples_per_second": 27.404,
+      "eval_steps_per_second": 1.785,
+      "step": 513
+    },
+    {
+      "epoch": 9.017543859649123,
+      "grad_norm": 9.234489440917969,
+      "learning_rate": 1.7229357798165137e-05,
+      "loss": 0.2302,
+      "step": 514
+    },
+    {
+      "epoch": 9.035087719298245,
+      "grad_norm": 9.468729019165039,
+      "learning_rate": 1.7201834862385323e-05,
+      "loss": 0.4173,
+      "step": 515
+    },
+    {
+      "epoch": 9.052631578947368,
+      "grad_norm": 10.059374809265137,
+      "learning_rate": 1.7174311926605505e-05,
+      "loss": 0.232,
+      "step": 516
+    },
+    {
+      "epoch": 9.070175438596491,
+      "grad_norm": 35.99481201171875,
+      "learning_rate": 1.7146788990825688e-05,
+      "loss": 0.42,
+      "step": 517
+    },
+    {
+      "epoch": 9.087719298245615,
+      "grad_norm": 2.6732192039489746,
+      "learning_rate": 1.711926605504587e-05,
+      "loss": 0.2202,
+      "step": 518
+    },
+    {
+      "epoch": 9.105263157894736,
+      "grad_norm": 26.779964447021484,
+      "learning_rate": 1.7091743119266056e-05,
+      "loss": 0.3977,
+      "step": 519
+    },
+    {
+      "epoch": 9.12280701754386,
+      "grad_norm": 8.011734962463379,
+      "learning_rate": 1.706422018348624e-05,
+      "loss": 0.2616,
+      "step": 520
+    },
+    {
+      "epoch": 9.140350877192983,
+      "grad_norm": 9.349709510803223,
+      "learning_rate": 1.703669724770642e-05,
+      "loss": 0.1727,
+      "step": 521
+    },
+    {
+      "epoch": 9.157894736842104,
+      "grad_norm": 3.374070405960083,
+      "learning_rate": 1.7009174311926607e-05,
+      "loss": 0.3442,
+      "step": 522
+    },
+    {
+      "epoch": 9.175438596491228,
+      "grad_norm": 7.063957691192627,
+      "learning_rate": 1.698165137614679e-05,
+      "loss": 0.3712,
+      "step": 523
+    },
+    {
+      "epoch": 9.192982456140351,
+      "grad_norm": 8.70645523071289,
+      "learning_rate": 1.6954128440366975e-05,
+      "loss": 0.371,
+      "step": 524
+    },
+    {
+      "epoch": 9.210526315789474,
+      "grad_norm": 6.593190670013428,
+      "learning_rate": 1.6926605504587154e-05,
+      "loss": 0.4862,
+      "step": 525
+    },
+    {
+      "epoch": 9.228070175438596,
+      "grad_norm": 5.061346054077148,
+      "learning_rate": 1.689908256880734e-05,
+      "loss": 0.2712,
+      "step": 526
+    },
+    {
+      "epoch": 9.24561403508772,
+      "grad_norm": 9.419418334960938,
+      "learning_rate": 1.6871559633027523e-05,
+      "loss": 0.2264,
+      "step": 527
+    },
+    {
+      "epoch": 9.263157894736842,
+      "grad_norm": 10.859743118286133,
+      "learning_rate": 1.684403669724771e-05,
+      "loss": 0.3662,
+      "step": 528
+    },
+    {
+      "epoch": 9.280701754385966,
+      "grad_norm": 30.619741439819336,
+      "learning_rate": 1.6816513761467888e-05,
+      "loss": 0.3531,
+      "step": 529
+    },
+    {
+      "epoch": 9.298245614035087,
+      "grad_norm": 3.2730002403259277,
+      "learning_rate": 1.6788990825688073e-05,
+      "loss": 0.4001,
+      "step": 530
+    },
+    {
+      "epoch": 9.31578947368421,
+      "grad_norm": 8.583318710327148,
+      "learning_rate": 1.6761467889908256e-05,
+      "loss": 0.2039,
+      "step": 531
+    },
+    {
+      "epoch": 9.333333333333334,
+      "grad_norm": 19.599987030029297,
+      "learning_rate": 1.6733944954128442e-05,
+      "loss": 0.5786,
+      "step": 532
+    },
+    {
+      "epoch": 9.350877192982455,
+      "grad_norm": 3.10512638092041,
+      "learning_rate": 1.6706422018348624e-05,
+      "loss": 0.2436,
+      "step": 533
+    },
+    {
+      "epoch": 9.368421052631579,
+      "grad_norm": 10.77072525024414,
+      "learning_rate": 1.6678899082568807e-05,
+      "loss": 0.2794,
+      "step": 534
+    },
+    {
+      "epoch": 9.385964912280702,
+      "grad_norm": 14.997283935546875,
+      "learning_rate": 1.6651376146788993e-05,
+      "loss": 0.4462,
+      "step": 535
+    },
+    {
+      "epoch": 9.403508771929825,
+      "grad_norm": 14.47448444366455,
+      "learning_rate": 1.6623853211009175e-05,
+      "loss": 0.3414,
+      "step": 536
+    },
+    {
+      "epoch": 9.421052631578947,
+      "grad_norm": 26.79741096496582,
+      "learning_rate": 1.6596330275229358e-05,
+      "loss": 0.3425,
+      "step": 537
+    },
+    {
+      "epoch": 9.43859649122807,
+      "grad_norm": 16.75027847290039,
+      "learning_rate": 1.656880733944954e-05,
+      "loss": 0.6174,
+      "step": 538
+    },
+    {
+      "epoch": 9.456140350877194,
+      "grad_norm": 16.61779022216797,
+      "learning_rate": 1.6541284403669726e-05,
+      "loss": 0.4949,
+      "step": 539
+    },
+    {
+      "epoch": 9.473684210526315,
+      "grad_norm": 14.869440078735352,
+      "learning_rate": 1.651376146788991e-05,
+      "loss": 0.337,
+      "step": 540
+    },
+    {
+      "epoch": 9.491228070175438,
+      "grad_norm": 9.696203231811523,
+      "learning_rate": 1.648623853211009e-05,
+      "loss": 0.3742,
+      "step": 541
+    },
+    {
+      "epoch": 9.508771929824562,
+      "grad_norm": 4.401230335235596,
+      "learning_rate": 1.6458715596330273e-05,
+      "loss": 0.2271,
+      "step": 542
+    },
+    {
+      "epoch": 9.526315789473685,
+      "grad_norm": 5.773757457733154,
+      "learning_rate": 1.643119266055046e-05,
+      "loss": 0.3476,
+      "step": 543
+    },
+    {
+      "epoch": 9.543859649122806,
+      "grad_norm": 18.14839744567871,
+      "learning_rate": 1.6403669724770645e-05,
+      "loss": 0.4136,
+      "step": 544
+    },
+    {
+      "epoch": 9.56140350877193,
+      "grad_norm": 12.958759307861328,
+      "learning_rate": 1.6376146788990824e-05,
+      "loss": 0.2327,
+      "step": 545
+    },
+    {
+      "epoch": 9.578947368421053,
+      "grad_norm": 5.4245734214782715,
+      "learning_rate": 1.634862385321101e-05,
+      "loss": 0.2524,
+      "step": 546
+    },
+    {
+      "epoch": 9.596491228070175,
+      "grad_norm": 9.484623908996582,
+      "learning_rate": 1.6321100917431192e-05,
+      "loss": 0.1879,
+      "step": 547
+    },
+    {
+      "epoch": 9.614035087719298,
+      "grad_norm": 5.631322860717773,
+      "learning_rate": 1.6293577981651378e-05,
+      "loss": 0.3271,
+      "step": 548
+    },
+    {
+      "epoch": 9.631578947368421,
+      "grad_norm": 3.2394251823425293,
+      "learning_rate": 1.6266055045871557e-05,
+      "loss": 0.338,
+      "step": 549
+    },
+    {
+      "epoch": 9.649122807017545,
+      "grad_norm": 20.733617782592773,
+      "learning_rate": 1.6238532110091743e-05,
+      "loss": 0.6739,
+      "step": 550
+    },
+    {
+      "epoch": 9.666666666666666,
+      "grad_norm": 5.4612555503845215,
+      "learning_rate": 1.6211009174311926e-05,
+      "loss": 0.2177,
+      "step": 551
+    },
+    {
+      "epoch": 9.68421052631579,
+      "grad_norm": 13.240394592285156,
+      "learning_rate": 1.618348623853211e-05,
+      "loss": 0.5177,
+      "step": 552
+    },
+    {
+      "epoch": 9.701754385964913,
+      "grad_norm": 12.377360343933105,
+      "learning_rate": 1.6155963302752294e-05,
+      "loss": 0.382,
+      "step": 553
+    },
+    {
+      "epoch": 9.719298245614034,
+      "grad_norm": 27.4326114654541,
+      "learning_rate": 1.6128440366972477e-05,
+      "loss": 0.2683,
+      "step": 554
+    },
+    {
+      "epoch": 9.736842105263158,
+      "grad_norm": 102.49446105957031,
+      "learning_rate": 1.6100917431192662e-05,
+      "loss": 0.1924,
+      "step": 555
+    },
+    {
+      "epoch": 9.75438596491228,
+      "grad_norm": 28.67510414123535,
+      "learning_rate": 1.6073394495412845e-05,
+      "loss": 0.4392,
+      "step": 556
+    },
+    {
+      "epoch": 9.771929824561404,
+      "grad_norm": 7.125996112823486,
+      "learning_rate": 1.6045871559633027e-05,
+      "loss": 0.4484,
+      "step": 557
+    },
+    {
+      "epoch": 9.789473684210526,
+      "grad_norm": 6.322452545166016,
+      "learning_rate": 1.601834862385321e-05,
+      "loss": 0.2461,
+      "step": 558
+    },
+    {
+      "epoch": 9.807017543859649,
+      "grad_norm": 14.071425437927246,
+      "learning_rate": 1.5990825688073396e-05,
+      "loss": 0.3559,
+      "step": 559
+    },
+    {
+      "epoch": 9.824561403508772,
+      "grad_norm": 17.456308364868164,
+      "learning_rate": 1.5963302752293578e-05,
+      "loss": 0.3976,
+      "step": 560
+    },
+    {
+      "epoch": 9.842105263157894,
+      "grad_norm": 12.360490798950195,
+      "learning_rate": 1.593577981651376e-05,
+      "loss": 0.6526,
+      "step": 561
+    },
+    {
+      "epoch": 9.859649122807017,
+      "grad_norm": 16.147422790527344,
+      "learning_rate": 1.5908256880733943e-05,
+      "loss": 0.5249,
+      "step": 562
+    },
+    {
+      "epoch": 9.87719298245614,
+      "grad_norm": 3.189755439758301,
+      "learning_rate": 1.588073394495413e-05,
+      "loss": 0.216,
+      "step": 563
+    },
+    {
+      "epoch": 9.894736842105264,
+      "grad_norm": 3.7975523471832275,
+      "learning_rate": 1.5853211009174315e-05,
+      "loss": 0.2947,
+      "step": 564
+    },
+    {
+      "epoch": 9.912280701754385,
+      "grad_norm": 9.276900291442871,
+      "learning_rate": 1.5825688073394494e-05,
+      "loss": 0.3288,
+      "step": 565
+    },
+    {
+      "epoch": 9.929824561403509,
+      "grad_norm": 6.910634994506836,
+      "learning_rate": 1.579816513761468e-05,
+      "loss": 0.3269,
+      "step": 566
+    },
+    {
+      "epoch": 9.947368421052632,
+      "grad_norm": 15.361982345581055,
+      "learning_rate": 1.5770642201834862e-05,
+      "loss": 0.5478,
+      "step": 567
+    },
+    {
+      "epoch": 9.964912280701755,
+      "grad_norm": 8.738931655883789,
+      "learning_rate": 1.5743119266055048e-05,
+      "loss": 0.3404,
+      "step": 568
+    },
+    {
+      "epoch": 9.982456140350877,
+      "grad_norm": 6.036810398101807,
+      "learning_rate": 1.5715596330275227e-05,
+      "loss": 0.4008,
+      "step": 569
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 9.815285682678223,
+      "learning_rate": 1.5688073394495413e-05,
+      "loss": 0.2739,
+      "step": 570
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.6781609195402298,
+      "eval_loss": 0.8435059189796448,
+      "eval_roc_auc": 0.8950196286952365,
+      "eval_runtime": 10.5896,
+      "eval_samples_per_second": 24.647,
+      "eval_steps_per_second": 1.605,
+      "step": 570
+    },
+    {
+      "epoch": 10.017543859649123,
+      "grad_norm": 6.685360431671143,
+      "learning_rate": 1.5660550458715596e-05,
+      "loss": 0.2748,
+      "step": 571
+    },
+    {
+      "epoch": 10.035087719298245,
+      "grad_norm": 5.618095874786377,
+      "learning_rate": 1.563302752293578e-05,
+      "loss": 0.2501,
+      "step": 572
+    },
+    {
+      "epoch": 10.052631578947368,
+      "grad_norm": 12.535326957702637,
+      "learning_rate": 1.560550458715596e-05,
+      "loss": 0.3087,
+      "step": 573
+    },
+    {
+      "epoch": 10.070175438596491,
+      "grad_norm": 8.183959007263184,
+      "learning_rate": 1.5577981651376146e-05,
+      "loss": 0.2843,
+      "step": 574
+    },
+    {
+      "epoch": 10.087719298245615,
+      "grad_norm": 1.980502724647522,
+      "learning_rate": 1.5550458715596332e-05,
+      "loss": 0.1799,
+      "step": 575
+    },
+    {
+      "epoch": 10.105263157894736,
+      "grad_norm": 13.456244468688965,
+      "learning_rate": 1.5522935779816515e-05,
+      "loss": 0.3431,
+      "step": 576
+    },
+    {
+      "epoch": 10.12280701754386,
+      "grad_norm": 10.74759578704834,
+      "learning_rate": 1.5495412844036697e-05,
+      "loss": 0.1717,
+      "step": 577
+    },
+    {
+      "epoch": 10.140350877192983,
+      "grad_norm": 26.72271728515625,
+      "learning_rate": 1.546788990825688e-05,
+      "loss": 0.235,
+      "step": 578
+    },
+    {
+      "epoch": 10.157894736842104,
+      "grad_norm": 1.6154626607894897,
+      "learning_rate": 1.5440366972477066e-05,
+      "loss": 0.1152,
+      "step": 579
+    },
+    {
+      "epoch": 10.175438596491228,
+      "grad_norm": 11.566985130310059,
+      "learning_rate": 1.5412844036697248e-05,
+      "loss": 0.1322,
+      "step": 580
+    },
+    {
+      "epoch": 10.192982456140351,
+      "grad_norm": 5.070148944854736,
+      "learning_rate": 1.538532110091743e-05,
+      "loss": 0.2487,
+      "step": 581
+    },
+    {
+      "epoch": 10.210526315789474,
+      "grad_norm": 17.369050979614258,
+      "learning_rate": 1.5357798165137613e-05,
+      "loss": 0.2958,
+      "step": 582
+    },
+    {
+      "epoch": 10.228070175438596,
+      "grad_norm": 8.392364501953125,
+      "learning_rate": 1.53302752293578e-05,
+      "loss": 0.3091,
+      "step": 583
+    },
+    {
+      "epoch": 10.24561403508772,
+      "grad_norm": 2.859358787536621,
+      "learning_rate": 1.530275229357798e-05,
+      "loss": 0.1137,
+      "step": 584
+    },
+    {
+      "epoch": 10.263157894736842,
+      "grad_norm": 23.28866195678711,
+      "learning_rate": 1.5275229357798164e-05,
+      "loss": 0.2434,
+      "step": 585
+    },
+    {
+      "epoch": 10.280701754385966,
+      "grad_norm": 18.40946388244629,
+      "learning_rate": 1.5247706422018348e-05,
+      "loss": 0.6074,
+      "step": 586
+    },
+    {
+      "epoch": 10.298245614035087,
+      "grad_norm": 12.528573036193848,
+      "learning_rate": 1.5220183486238532e-05,
+      "loss": 0.1412,
+      "step": 587
+    },
+    {
+      "epoch": 10.31578947368421,
+      "grad_norm": 10.554133415222168,
+      "learning_rate": 1.5192660550458716e-05,
+      "loss": 0.4182,
+      "step": 588
+    },
+    {
+      "epoch": 10.333333333333334,
+      "grad_norm": 4.216406345367432,
+      "learning_rate": 1.5165137614678897e-05,
+      "loss": 0.2399,
+      "step": 589
+    },
+    {
+      "epoch": 10.350877192982455,
+      "grad_norm": 9.16854476928711,
+      "learning_rate": 1.5137614678899083e-05,
+      "loss": 0.2882,
+      "step": 590
+    },
+    {
+      "epoch": 10.368421052631579,
+      "grad_norm": 2.488682746887207,
+      "learning_rate": 1.5110091743119267e-05,
+      "loss": 0.0998,
+      "step": 591
+    },
+    {
+      "epoch": 10.385964912280702,
+      "grad_norm": 30.423564910888672,
+      "learning_rate": 1.5082568807339451e-05,
+      "loss": 0.3046,
+      "step": 592
+    },
+    {
+      "epoch": 10.403508771929825,
+      "grad_norm": 20.648027420043945,
+      "learning_rate": 1.5055045871559632e-05,
+      "loss": 0.4733,
+      "step": 593
+    },
+    {
+      "epoch": 10.421052631578947,
+      "grad_norm": 16.398672103881836,
+      "learning_rate": 1.5027522935779816e-05,
+      "loss": 0.2178,
+      "step": 594
+    },
+    {
+      "epoch": 10.43859649122807,
+      "grad_norm": 6.614181995391846,
+      "learning_rate": 1.5e-05,
+      "loss": 0.2064,
+      "step": 595
+    },
+    {
+      "epoch": 10.456140350877194,
+      "grad_norm": 30.626367568969727,
+      "learning_rate": 1.4972477064220185e-05,
+      "loss": 0.3677,
+      "step": 596
+    },
+    {
+      "epoch": 10.473684210526315,
+      "grad_norm": 3.0452020168304443,
+      "learning_rate": 1.4944954128440367e-05,
+      "loss": 0.1095,
+      "step": 597
+    },
+    {
+      "epoch": 10.491228070175438,
+      "grad_norm": 13.619409561157227,
+      "learning_rate": 1.4917431192660551e-05,
+      "loss": 0.4249,
+      "step": 598
+    },
+    {
+      "epoch": 10.508771929824562,
+      "grad_norm": 9.04359245300293,
+      "learning_rate": 1.4889908256880734e-05,
+      "loss": 0.2014,
+      "step": 599
+    },
+    {
+      "epoch": 10.526315789473685,
+      "grad_norm": 17.97020149230957,
+      "learning_rate": 1.4862385321100918e-05,
+      "loss": 0.5679,
+      "step": 600
+    },
+    {
+      "epoch": 10.543859649122806,
+      "grad_norm": 7.490416526794434,
+      "learning_rate": 1.4834862385321102e-05,
+      "loss": 0.4832,
+      "step": 601
+    },
+    {
+      "epoch": 10.56140350877193,
+      "grad_norm": 12.512399673461914,
+      "learning_rate": 1.4807339449541286e-05,
+      "loss": 0.3462,
+      "step": 602
+    },
+    {
+      "epoch": 10.578947368421053,
+      "grad_norm": 6.70708703994751,
+      "learning_rate": 1.4779816513761469e-05,
+      "loss": 0.1513,
+      "step": 603
+    },
+    {
+      "epoch": 10.596491228070175,
+      "grad_norm": 8.195464134216309,
+      "learning_rate": 1.4752293577981653e-05,
+      "loss": 0.2896,
+      "step": 604
+    },
+    {
+      "epoch": 10.614035087719298,
+      "grad_norm": 35.68895721435547,
+      "learning_rate": 1.4724770642201835e-05,
+      "loss": 0.2295,
+      "step": 605
+    },
+    {
+      "epoch": 10.631578947368421,
+      "grad_norm": 6.922648906707764,
+      "learning_rate": 1.469724770642202e-05,
+      "loss": 0.5227,
+      "step": 606
+    },
+    {
+      "epoch": 10.649122807017545,
+      "grad_norm": 6.483743190765381,
+      "learning_rate": 1.4669724770642202e-05,
+      "loss": 0.22,
+      "step": 607
+    },
+    {
+      "epoch": 10.666666666666666,
+      "grad_norm": 5.109847545623779,
+      "learning_rate": 1.4642201834862386e-05,
+      "loss": 0.1755,
+      "step": 608
+    },
+    {
+      "epoch": 10.68421052631579,
+      "grad_norm": 3.8107926845550537,
+      "learning_rate": 1.4614678899082569e-05,
+      "loss": 0.1058,
+      "step": 609
+    },
+    {
+      "epoch": 10.701754385964913,
+      "grad_norm": 9.708571434020996,
+      "learning_rate": 1.4587155963302753e-05,
+      "loss": 0.3024,
+      "step": 610
+    },
+    {
+      "epoch": 10.719298245614034,
+      "grad_norm": 6.205732345581055,
+      "learning_rate": 1.4559633027522937e-05,
+      "loss": 0.1044,
+      "step": 611
+    },
+    {
+      "epoch": 10.736842105263158,
+      "grad_norm": 9.928633689880371,
+      "learning_rate": 1.4532110091743121e-05,
+      "loss": 0.1499,
+      "step": 612
+    },
+    {
+      "epoch": 10.75438596491228,
+      "grad_norm": 19.466793060302734,
+      "learning_rate": 1.4504587155963304e-05,
+      "loss": 0.5,
+      "step": 613
+    },
+    {
+      "epoch": 10.771929824561404,
+      "grad_norm": 11.241743087768555,
+      "learning_rate": 1.4477064220183488e-05,
+      "loss": 0.3349,
+      "step": 614
+    },
+    {
+      "epoch": 10.789473684210526,
+      "grad_norm": 8.29288101196289,
+      "learning_rate": 1.444954128440367e-05,
+      "loss": 0.1464,
+      "step": 615
+    },
+    {
+      "epoch": 10.807017543859649,
+      "grad_norm": 31.766202926635742,
+      "learning_rate": 1.4422018348623854e-05,
+      "loss": 0.2376,
+      "step": 616
+    },
+    {
+      "epoch": 10.824561403508772,
+      "grad_norm": 17.91666030883789,
+      "learning_rate": 1.4394495412844037e-05,
+      "loss": 0.3159,
+      "step": 617
+    },
+    {
+      "epoch": 10.842105263157894,
+      "grad_norm": 19.936201095581055,
+      "learning_rate": 1.4366972477064221e-05,
+      "loss": 0.2585,
+      "step": 618
+    },
+    {
+      "epoch": 10.859649122807017,
+      "grad_norm": 72.99993896484375,
+      "learning_rate": 1.4339449541284404e-05,
+      "loss": 0.5076,
+      "step": 619
+    },
+    {
+      "epoch": 10.87719298245614,
+      "grad_norm": 8.826135635375977,
+      "learning_rate": 1.4311926605504588e-05,
+      "loss": 0.2219,
+      "step": 620
+    },
+    {
+      "epoch": 10.894736842105264,
+      "grad_norm": 26.077892303466797,
+      "learning_rate": 1.428440366972477e-05,
+      "loss": 0.4843,
+      "step": 621
+    },
+    {
+      "epoch": 10.912280701754385,
+      "grad_norm": 20.25221061706543,
+      "learning_rate": 1.4256880733944956e-05,
+      "loss": 0.5897,
+      "step": 622
+    },
+    {
+      "epoch": 10.929824561403509,
+      "grad_norm": 31.701025009155273,
+      "learning_rate": 1.4229357798165138e-05,
+      "loss": 0.3267,
+      "step": 623
+    },
+    {
+      "epoch": 10.947368421052632,
+      "grad_norm": 8.264379501342773,
+      "learning_rate": 1.4201834862385323e-05,
+      "loss": 0.2787,
+      "step": 624
+    },
+    {
+      "epoch": 10.964912280701755,
+      "grad_norm": 48.150184631347656,
+      "learning_rate": 1.4174311926605505e-05,
+      "loss": 0.4824,
+      "step": 625
+    },
+    {
+      "epoch": 10.982456140350877,
+      "grad_norm": 10.574882507324219,
+      "learning_rate": 1.414678899082569e-05,
+      "loss": 0.3676,
+      "step": 626
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 24.72840118408203,
+      "learning_rate": 1.4119266055045872e-05,
+      "loss": 0.581,
+      "step": 627
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.7432950191570882,
+      "eval_loss": 0.8437221646308899,
+      "eval_roc_auc": 0.9043893455592809,
+      "eval_runtime": 9.7269,
+      "eval_samples_per_second": 26.833,
+      "eval_steps_per_second": 1.748,
+      "step": 627
+    },
+    {
+      "epoch": 11.017543859649123,
+      "grad_norm": 6.984513759613037,
+      "learning_rate": 1.4091743119266056e-05,
+      "loss": 0.2926,
+      "step": 628
+    },
+    {
+      "epoch": 11.035087719298245,
+      "grad_norm": 31.709365844726562,
+      "learning_rate": 1.4064220183486238e-05,
+      "loss": 0.1972,
+      "step": 629
+    },
+    {
+      "epoch": 11.052631578947368,
+      "grad_norm": 4.533780574798584,
+      "learning_rate": 1.4036697247706423e-05,
+      "loss": 0.1809,
+      "step": 630
+    },
+    {
+      "epoch": 11.070175438596491,
+      "grad_norm": 8.72680377960205,
+      "learning_rate": 1.4009174311926605e-05,
+      "loss": 0.2892,
+      "step": 631
+    },
+    {
+      "epoch": 11.087719298245615,
+      "grad_norm": 1.0965291261672974,
+      "learning_rate": 1.3981651376146791e-05,
+      "loss": 0.1061,
+      "step": 632
+    },
+    {
+      "epoch": 11.105263157894736,
+      "grad_norm": 7.810534477233887,
+      "learning_rate": 1.3954128440366973e-05,
+      "loss": 0.2574,
+      "step": 633
+    },
+    {
+      "epoch": 11.12280701754386,
+      "grad_norm": 5.813023567199707,
+      "learning_rate": 1.3926605504587158e-05,
+      "loss": 0.1923,
+      "step": 634
+    },
+    {
+      "epoch": 11.140350877192983,
+      "grad_norm": 27.594276428222656,
+      "learning_rate": 1.389908256880734e-05,
+      "loss": 0.3515,
+      "step": 635
+    },
+    {
+      "epoch": 11.157894736842104,
+      "grad_norm": 11.405938148498535,
+      "learning_rate": 1.3871559633027524e-05,
+      "loss": 0.2625,
+      "step": 636
+    },
+    {
+      "epoch": 11.175438596491228,
+      "grad_norm": 4.616410732269287,
+      "learning_rate": 1.3844036697247707e-05,
+      "loss": 0.1946,
+      "step": 637
+    },
+    {
+      "epoch": 11.192982456140351,
+      "grad_norm": 5.800034046173096,
+      "learning_rate": 1.3816513761467891e-05,
+      "loss": 0.2544,
+      "step": 638
+    },
+    {
+      "epoch": 11.210526315789474,
+      "grad_norm": 9.26241397857666,
+      "learning_rate": 1.3788990825688073e-05,
+      "loss": 0.416,
+      "step": 639
+    },
+    {
+      "epoch": 11.228070175438596,
+      "grad_norm": 10.779268264770508,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 0.2791,
+      "step": 640
+    },
+    {
+      "epoch": 11.24561403508772,
+      "grad_norm": 55.15119552612305,
+      "learning_rate": 1.373394495412844e-05,
+      "loss": 0.3073,
+      "step": 641
+    },
+    {
+      "epoch": 11.263157894736842,
+      "grad_norm": 5.921196460723877,
+      "learning_rate": 1.3706422018348624e-05,
+      "loss": 0.1119,
+      "step": 642
+    },
+    {
+      "epoch": 11.280701754385966,
+      "grad_norm": 14.839341163635254,
+      "learning_rate": 1.3678899082568808e-05,
+      "loss": 0.3157,
+      "step": 643
+    },
+    {
+      "epoch": 11.298245614035087,
+      "grad_norm": 6.483922481536865,
+      "learning_rate": 1.3651376146788992e-05,
+      "loss": 0.1668,
+      "step": 644
+    },
+    {
+      "epoch": 11.31578947368421,
+      "grad_norm": 2.650290012359619,
+      "learning_rate": 1.3623853211009175e-05,
+      "loss": 0.1149,
+      "step": 645
+    },
+    {
+      "epoch": 11.333333333333334,
+      "grad_norm": 2.6592800617218018,
+      "learning_rate": 1.3596330275229359e-05,
+      "loss": 0.0868,
+      "step": 646
+    },
+    {
+      "epoch": 11.350877192982455,
+      "grad_norm": 22.124906539916992,
+      "learning_rate": 1.3568807339449542e-05,
+      "loss": 0.4327,
+      "step": 647
+    },
+    {
+      "epoch": 11.368421052631579,
+      "grad_norm": 20.250707626342773,
+      "learning_rate": 1.3541284403669726e-05,
+      "loss": 0.3964,
+      "step": 648
+    },
+    {
+      "epoch": 11.385964912280702,
+      "grad_norm": 5.71112585067749,
+      "learning_rate": 1.3513761467889908e-05,
+      "loss": 0.1463,
+      "step": 649
+    },
+    {
+      "epoch": 11.403508771929825,
+      "grad_norm": 17.264068603515625,
+      "learning_rate": 1.3486238532110092e-05,
+      "loss": 0.5867,
+      "step": 650
+    },
+    {
+      "epoch": 11.421052631578947,
+      "grad_norm": 21.228487014770508,
+      "learning_rate": 1.3458715596330275e-05,
+      "loss": 0.2334,
+      "step": 651
+    },
+    {
+      "epoch": 11.43859649122807,
+      "grad_norm": 8.839598655700684,
+      "learning_rate": 1.3431192660550459e-05,
+      "loss": 0.2421,
+      "step": 652
+    },
+    {
+      "epoch": 11.456140350877194,
+      "grad_norm": 18.84731101989746,
+      "learning_rate": 1.3403669724770643e-05,
+      "loss": 0.3786,
+      "step": 653
+    },
+    {
+      "epoch": 11.473684210526315,
+      "grad_norm": 32.75419998168945,
+      "learning_rate": 1.3376146788990827e-05,
+      "loss": 0.276,
+      "step": 654
+    },
+    {
+      "epoch": 11.491228070175438,
+      "grad_norm": 10.089400291442871,
+      "learning_rate": 1.334862385321101e-05,
+      "loss": 0.2987,
+      "step": 655
+    },
+    {
+      "epoch": 11.508771929824562,
+      "grad_norm": 13.225504875183105,
+      "learning_rate": 1.3321100917431194e-05,
+      "loss": 0.1948,
+      "step": 656
+    },
+    {
+      "epoch": 11.526315789473685,
+      "grad_norm": 17.269506454467773,
+      "learning_rate": 1.3293577981651377e-05,
+      "loss": 0.1486,
+      "step": 657
+    },
+    {
+      "epoch": 11.543859649122806,
+      "grad_norm": 5.897825241088867,
+      "learning_rate": 1.326605504587156e-05,
+      "loss": 0.2797,
+      "step": 658
+    },
+    {
+      "epoch": 11.56140350877193,
+      "grad_norm": 4.05665397644043,
+      "learning_rate": 1.3238532110091743e-05,
+      "loss": 0.1859,
+      "step": 659
+    },
+    {
+      "epoch": 11.578947368421053,
+      "grad_norm": 12.295641899108887,
+      "learning_rate": 1.3211009174311927e-05,
+      "loss": 0.3204,
+      "step": 660
+    },
+    {
+      "epoch": 11.596491228070175,
+      "grad_norm": 15.206548690795898,
+      "learning_rate": 1.318348623853211e-05,
+      "loss": 0.5305,
+      "step": 661
+    },
+    {
+      "epoch": 11.614035087719298,
+      "grad_norm": 19.5164852142334,
+      "learning_rate": 1.3155963302752294e-05,
+      "loss": 0.316,
+      "step": 662
+    },
+    {
+      "epoch": 11.631578947368421,
+      "grad_norm": 4.328670501708984,
+      "learning_rate": 1.3128440366972476e-05,
+      "loss": 0.1426,
+      "step": 663
+    },
+    {
+      "epoch": 11.649122807017545,
+      "grad_norm": 19.915075302124023,
+      "learning_rate": 1.3100917431192662e-05,
+      "loss": 0.5241,
+      "step": 664
+    },
+    {
+      "epoch": 11.666666666666666,
+      "grad_norm": 9.99632740020752,
+      "learning_rate": 1.3073394495412845e-05,
+      "loss": 0.3287,
+      "step": 665
+    },
+    {
+      "epoch": 11.68421052631579,
+      "grad_norm": 8.295022964477539,
+      "learning_rate": 1.3045871559633029e-05,
+      "loss": 0.1386,
+      "step": 666
+    },
+    {
+      "epoch": 11.701754385964913,
+      "grad_norm": 11.660796165466309,
+      "learning_rate": 1.3018348623853211e-05,
+      "loss": 0.3504,
+      "step": 667
+    },
+    {
+      "epoch": 11.719298245614034,
+      "grad_norm": 19.916318893432617,
+      "learning_rate": 1.2990825688073396e-05,
+      "loss": 0.4487,
+      "step": 668
+    },
+    {
+      "epoch": 11.736842105263158,
+      "grad_norm": 6.137487411499023,
+      "learning_rate": 1.2963302752293578e-05,
+      "loss": 0.248,
+      "step": 669
+    },
+    {
+      "epoch": 11.75438596491228,
+      "grad_norm": 16.032516479492188,
+      "learning_rate": 1.2935779816513762e-05,
+      "loss": 0.5295,
+      "step": 670
+    },
+    {
+      "epoch": 11.771929824561404,
+      "grad_norm": 0.6119496822357178,
+      "learning_rate": 1.2908256880733945e-05,
+      "loss": 0.0604,
+      "step": 671
+    },
+    {
+      "epoch": 11.789473684210526,
+      "grad_norm": 5.562976837158203,
+      "learning_rate": 1.2880733944954129e-05,
+      "loss": 0.2119,
+      "step": 672
+    },
+    {
+      "epoch": 11.807017543859649,
+      "grad_norm": 4.243768215179443,
+      "learning_rate": 1.2853211009174311e-05,
+      "loss": 0.2239,
+      "step": 673
+    },
+    {
+      "epoch": 11.824561403508772,
+      "grad_norm": 20.970874786376953,
+      "learning_rate": 1.2825688073394497e-05,
+      "loss": 0.3221,
+      "step": 674
+    },
+    {
+      "epoch": 11.842105263157894,
+      "grad_norm": 2.212523937225342,
+      "learning_rate": 1.279816513761468e-05,
+      "loss": 0.0897,
+      "step": 675
+    },
+    {
+      "epoch": 11.859649122807017,
+      "grad_norm": 28.7296199798584,
+      "learning_rate": 1.2770642201834864e-05,
+      "loss": 0.362,
+      "step": 676
+    },
+    {
+      "epoch": 11.87719298245614,
+      "grad_norm": 13.14705753326416,
+      "learning_rate": 1.2743119266055046e-05,
+      "loss": 0.3464,
+      "step": 677
+    },
+    {
+      "epoch": 11.894736842105264,
+      "grad_norm": 9.211715698242188,
+      "learning_rate": 1.271559633027523e-05,
+      "loss": 0.3259,
+      "step": 678
+    },
+    {
+      "epoch": 11.912280701754385,
+      "grad_norm": 6.302637100219727,
+      "learning_rate": 1.2688073394495413e-05,
+      "loss": 0.1719,
+      "step": 679
+    },
+    {
+      "epoch": 11.929824561403509,
+      "grad_norm": 0.7406455874443054,
+      "learning_rate": 1.2660550458715597e-05,
+      "loss": 0.0792,
+      "step": 680
+    },
+    {
+      "epoch": 11.947368421052632,
+      "grad_norm": 15.276371002197266,
+      "learning_rate": 1.263302752293578e-05,
+      "loss": 0.1284,
+      "step": 681
+    },
+    {
+      "epoch": 11.964912280701755,
+      "grad_norm": 113.07421875,
+      "learning_rate": 1.2605504587155964e-05,
+      "loss": 0.4433,
+      "step": 682
+    },
+    {
+      "epoch": 11.982456140350877,
+      "grad_norm": 4.4429931640625,
+      "learning_rate": 1.2577981651376146e-05,
+      "loss": 0.1152,
+      "step": 683
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 18.602813720703125,
+      "learning_rate": 1.255045871559633e-05,
+      "loss": 0.3169,
+      "step": 684
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.7049808429118773,
+      "eval_loss": 0.9107945561408997,
+      "eval_roc_auc": 0.8936780938450124,
+      "eval_runtime": 8.7978,
+      "eval_samples_per_second": 29.666,
+      "eval_steps_per_second": 1.932,
+      "step": 684
+    },
+    {
+      "epoch": 12.017543859649123,
+      "grad_norm": 0.6636399626731873,
+      "learning_rate": 1.2522935779816515e-05,
+      "loss": 0.0605,
+      "step": 685
+    },
+    {
+      "epoch": 12.035087719298245,
+      "grad_norm": 18.400907516479492,
+      "learning_rate": 1.2495412844036699e-05,
+      "loss": 0.1937,
+      "step": 686
+    },
+    {
+      "epoch": 12.052631578947368,
+      "grad_norm": 8.080551147460938,
+      "learning_rate": 1.2467889908256881e-05,
+      "loss": 0.3127,
+      "step": 687
+    },
+    {
+      "epoch": 12.070175438596491,
+      "grad_norm": 8.157322883605957,
+      "learning_rate": 1.2440366972477065e-05,
+      "loss": 0.1947,
+      "step": 688
+    },
+    {
+      "epoch": 12.087719298245615,
+      "grad_norm": 6.911866188049316,
+      "learning_rate": 1.2412844036697248e-05,
+      "loss": 0.2021,
+      "step": 689
+    },
+    {
+      "epoch": 12.105263157894736,
+      "grad_norm": 1.3366920948028564,
+      "learning_rate": 1.2385321100917432e-05,
+      "loss": 0.0808,
+      "step": 690
+    },
+    {
+      "epoch": 12.12280701754386,
+      "grad_norm": 2.7027173042297363,
+      "learning_rate": 1.2357798165137615e-05,
+      "loss": 0.1324,
+      "step": 691
+    },
+    {
+      "epoch": 12.140350877192983,
+      "grad_norm": 6.3235697746276855,
+      "learning_rate": 1.2330275229357799e-05,
+      "loss": 0.1185,
+      "step": 692
+    },
+    {
+      "epoch": 12.157894736842104,
+      "grad_norm": 9.077238082885742,
+      "learning_rate": 1.2302752293577981e-05,
+      "loss": 0.1614,
+      "step": 693
+    },
+    {
+      "epoch": 12.175438596491228,
+      "grad_norm": 2.126521348953247,
+      "learning_rate": 1.2275229357798165e-05,
+      "loss": 0.181,
+      "step": 694
+    },
+    {
+      "epoch": 12.192982456140351,
+      "grad_norm": 7.8555097579956055,
+      "learning_rate": 1.224770642201835e-05,
+      "loss": 0.1929,
+      "step": 695
+    },
+    {
+      "epoch": 12.210526315789474,
+      "grad_norm": 10.955324172973633,
+      "learning_rate": 1.2220183486238534e-05,
+      "loss": 0.2502,
+      "step": 696
+    },
+    {
+      "epoch": 12.228070175438596,
+      "grad_norm": 24.433555603027344,
+      "learning_rate": 1.2192660550458716e-05,
+      "loss": 0.5622,
+      "step": 697
+    },
+    {
+      "epoch": 12.24561403508772,
+      "grad_norm": 15.262462615966797,
+      "learning_rate": 1.21651376146789e-05,
+      "loss": 0.3855,
+      "step": 698
+    },
+    {
+      "epoch": 12.263157894736842,
+      "grad_norm": 4.2611918449401855,
+      "learning_rate": 1.2137614678899083e-05,
+      "loss": 0.1054,
+      "step": 699
+    },
+    {
+      "epoch": 12.280701754385966,
+      "grad_norm": 3.4223506450653076,
+      "learning_rate": 1.2110091743119267e-05,
+      "loss": 0.2222,
+      "step": 700
+    },
+    {
+      "epoch": 12.298245614035087,
+      "grad_norm": 5.349019527435303,
+      "learning_rate": 1.208256880733945e-05,
+      "loss": 0.2028,
+      "step": 701
+    },
+    {
+      "epoch": 12.31578947368421,
+      "grad_norm": 15.387149810791016,
+      "learning_rate": 1.2055045871559634e-05,
+      "loss": 0.1211,
+      "step": 702
+    },
+    {
+      "epoch": 12.333333333333334,
+      "grad_norm": 1.1758196353912354,
+      "learning_rate": 1.2027522935779816e-05,
+      "loss": 0.0588,
+      "step": 703
+    },
+    {
+      "epoch": 12.350877192982455,
+      "grad_norm": 7.5386176109313965,
+      "learning_rate": 1.2e-05,
+      "loss": 0.1079,
+      "step": 704
+    },
+    {
+      "epoch": 12.368421052631579,
+      "grad_norm": 3.1408779621124268,
+      "learning_rate": 1.1972477064220183e-05,
+      "loss": 0.1088,
+      "step": 705
+    },
+    {
+      "epoch": 12.385964912280702,
+      "grad_norm": 2.542567253112793,
+      "learning_rate": 1.1944954128440369e-05,
+      "loss": 0.2409,
+      "step": 706
+    },
+    {
+      "epoch": 12.403508771929825,
+      "grad_norm": 5.91030216217041,
+      "learning_rate": 1.1917431192660551e-05,
+      "loss": 0.1228,
+      "step": 707
+    },
+    {
+      "epoch": 12.421052631578947,
+      "grad_norm": 144.98297119140625,
+      "learning_rate": 1.1889908256880735e-05,
+      "loss": 0.4235,
+      "step": 708
+    },
+    {
+      "epoch": 12.43859649122807,
+      "grad_norm": 0.39158356189727783,
+      "learning_rate": 1.1862385321100918e-05,
+      "loss": 0.0468,
+      "step": 709
+    },
+    {
+      "epoch": 12.456140350877194,
+      "grad_norm": 16.603181838989258,
+      "learning_rate": 1.1834862385321102e-05,
+      "loss": 0.2334,
+      "step": 710
+    },
+    {
+      "epoch": 12.473684210526315,
+      "grad_norm": 19.475147247314453,
+      "learning_rate": 1.1807339449541284e-05,
+      "loss": 0.3784,
+      "step": 711
+    },
+    {
+      "epoch": 12.491228070175438,
+      "grad_norm": 10.970072746276855,
+      "learning_rate": 1.1779816513761469e-05,
+      "loss": 0.3142,
+      "step": 712
+    },
+    {
+      "epoch": 12.508771929824562,
+      "grad_norm": 591.7606811523438,
+      "learning_rate": 1.1752293577981651e-05,
+      "loss": 0.2733,
+      "step": 713
+    },
+    {
+      "epoch": 12.526315789473685,
+      "grad_norm": 16.299694061279297,
+      "learning_rate": 1.1724770642201835e-05,
+      "loss": 0.3991,
+      "step": 714
+    },
+    {
+      "epoch": 12.543859649122806,
+      "grad_norm": 6.645303249359131,
+      "learning_rate": 1.1697247706422018e-05,
+      "loss": 0.1417,
+      "step": 715
+    },
+    {
+      "epoch": 12.56140350877193,
+      "grad_norm": 11.64770221710205,
+      "learning_rate": 1.1669724770642204e-05,
+      "loss": 0.3619,
+      "step": 716
+    },
+    {
+      "epoch": 12.578947368421053,
+      "grad_norm": 4.1909685134887695,
+      "learning_rate": 1.1642201834862386e-05,
+      "loss": 0.1734,
+      "step": 717
+    },
+    {
+      "epoch": 12.596491228070175,
+      "grad_norm": 21.624954223632812,
+      "learning_rate": 1.161467889908257e-05,
+      "loss": 0.3642,
+      "step": 718
+    },
+    {
+      "epoch": 12.614035087719298,
+      "grad_norm": 14.657827377319336,
+      "learning_rate": 1.1587155963302753e-05,
+      "loss": 0.1945,
+      "step": 719
+    },
+    {
+      "epoch": 12.631578947368421,
+      "grad_norm": 13.738436698913574,
+      "learning_rate": 1.1559633027522937e-05,
+      "loss": 0.179,
+      "step": 720
+    },
+    {
+      "epoch": 12.649122807017545,
+      "grad_norm": 14.478985786437988,
+      "learning_rate": 1.153211009174312e-05,
+      "loss": 0.3448,
+      "step": 721
+    },
+    {
+      "epoch": 12.666666666666666,
+      "grad_norm": 9.730142593383789,
+      "learning_rate": 1.1504587155963303e-05,
+      "loss": 0.1897,
+      "step": 722
+    },
+    {
+      "epoch": 12.68421052631579,
+      "grad_norm": 163.12696838378906,
+      "learning_rate": 1.1477064220183486e-05,
+      "loss": 0.2439,
+      "step": 723
+    },
+    {
+      "epoch": 12.701754385964913,
+      "grad_norm": 9.576773643493652,
+      "learning_rate": 1.144954128440367e-05,
+      "loss": 0.2353,
+      "step": 724
+    },
+    {
+      "epoch": 12.719298245614034,
+      "grad_norm": 10.299495697021484,
+      "learning_rate": 1.1422018348623853e-05,
+      "loss": 0.242,
+      "step": 725
+    },
+    {
+      "epoch": 12.736842105263158,
+      "grad_norm": 33.32149887084961,
+      "learning_rate": 1.1394495412844037e-05,
+      "loss": 0.7456,
+      "step": 726
+    },
+    {
+      "epoch": 12.75438596491228,
+      "grad_norm": 6.180983066558838,
+      "learning_rate": 1.1366972477064221e-05,
+      "loss": 0.1535,
+      "step": 727
+    },
+    {
+      "epoch": 12.771929824561404,
+      "grad_norm": 7.370275020599365,
+      "learning_rate": 1.1339449541284405e-05,
+      "loss": 0.1143,
+      "step": 728
+    },
+    {
+      "epoch": 12.789473684210526,
+      "grad_norm": 14.886811256408691,
+      "learning_rate": 1.1311926605504588e-05,
+      "loss": 0.2212,
+      "step": 729
+    },
+    {
+      "epoch": 12.807017543859649,
+      "grad_norm": 14.88480281829834,
+      "learning_rate": 1.1284403669724772e-05,
+      "loss": 0.2585,
+      "step": 730
+    },
+    {
+      "epoch": 12.824561403508772,
+      "grad_norm": 28.494150161743164,
+      "learning_rate": 1.1256880733944954e-05,
+      "loss": 0.3074,
+      "step": 731
+    },
+    {
+      "epoch": 12.842105263157894,
+      "grad_norm": 5.734357833862305,
+      "learning_rate": 1.1229357798165138e-05,
+      "loss": 0.2147,
+      "step": 732
+    },
+    {
+      "epoch": 12.859649122807017,
+      "grad_norm": 8.859356880187988,
+      "learning_rate": 1.1201834862385321e-05,
+      "loss": 0.3344,
+      "step": 733
+    },
+    {
+      "epoch": 12.87719298245614,
+      "grad_norm": 6.154085159301758,
+      "learning_rate": 1.1174311926605505e-05,
+      "loss": 0.1326,
+      "step": 734
+    },
+    {
+      "epoch": 12.894736842105264,
+      "grad_norm": 86.77043151855469,
+      "learning_rate": 1.1146788990825688e-05,
+      "loss": 0.2653,
+      "step": 735
+    },
+    {
+      "epoch": 12.912280701754385,
+      "grad_norm": 2.9679932594299316,
+      "learning_rate": 1.1119266055045872e-05,
+      "loss": 0.1537,
+      "step": 736
+    },
+    {
+      "epoch": 12.929824561403509,
+      "grad_norm": 6.805034160614014,
+      "learning_rate": 1.1091743119266056e-05,
+      "loss": 0.1669,
+      "step": 737
+    },
+    {
+      "epoch": 12.947368421052632,
+      "grad_norm": 16.86161231994629,
+      "learning_rate": 1.106422018348624e-05,
+      "loss": 0.261,
+      "step": 738
+    },
+    {
+      "epoch": 12.964912280701755,
+      "grad_norm": 4.945237636566162,
+      "learning_rate": 1.1036697247706423e-05,
+      "loss": 0.2191,
+      "step": 739
+    },
+    {
+      "epoch": 12.982456140350877,
+      "grad_norm": 70.42694091796875,
+      "learning_rate": 1.1009174311926607e-05,
+      "loss": 0.5588,
+      "step": 740
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 33.98917007446289,
+      "learning_rate": 1.098165137614679e-05,
+      "loss": 0.5302,
+      "step": 741
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.7049808429118773,
+      "eval_loss": 0.9366002678871155,
+      "eval_roc_auc": 0.8886698826983219,
+      "eval_runtime": 8.6062,
+      "eval_samples_per_second": 30.327,
+      "eval_steps_per_second": 1.975,
+      "step": 741
+    },
+    {
+      "epoch": 13.017543859649123,
+      "grad_norm": 16.30794334411621,
+      "learning_rate": 1.0954128440366973e-05,
+      "loss": 0.1746,
+      "step": 742
+    },
+    {
+      "epoch": 13.035087719298245,
+      "grad_norm": 19.023889541625977,
+      "learning_rate": 1.0926605504587156e-05,
+      "loss": 0.3532,
+      "step": 743
+    },
+    {
+      "epoch": 13.052631578947368,
+      "grad_norm": 58.92171859741211,
+      "learning_rate": 1.089908256880734e-05,
+      "loss": 0.2303,
+      "step": 744
+    },
+    {
+      "epoch": 13.070175438596491,
+      "grad_norm": 2.3900561332702637,
+      "learning_rate": 1.0871559633027522e-05,
+      "loss": 0.0868,
+      "step": 745
+    },
+    {
+      "epoch": 13.087719298245615,
+      "grad_norm": 9.594512939453125,
+      "learning_rate": 1.0844036697247707e-05,
+      "loss": 0.2155,
+      "step": 746
+    },
+    {
+      "epoch": 13.105263157894736,
+      "grad_norm": 6.165768623352051,
+      "learning_rate": 1.081651376146789e-05,
+      "loss": 0.1699,
+      "step": 747
+    },
+    {
+      "epoch": 13.12280701754386,
+      "grad_norm": 11.738020896911621,
+      "learning_rate": 1.0788990825688075e-05,
+      "loss": 0.3146,
+      "step": 748
+    },
+    {
+      "epoch": 13.140350877192983,
+      "grad_norm": 245.3873748779297,
+      "learning_rate": 1.0761467889908257e-05,
+      "loss": 0.3095,
+      "step": 749
+    },
+    {
+      "epoch": 13.157894736842104,
+      "grad_norm": 6.468425750732422,
+      "learning_rate": 1.0733944954128442e-05,
+      "loss": 0.1565,
+      "step": 750
+    },
+    {
+      "epoch": 13.175438596491228,
+      "grad_norm": 51.99348449707031,
+      "learning_rate": 1.0706422018348624e-05,
+      "loss": 0.5008,
+      "step": 751
+    },
+    {
+      "epoch": 13.192982456140351,
+      "grad_norm": 2.463118076324463,
+      "learning_rate": 1.0678899082568808e-05,
+      "loss": 0.1401,
+      "step": 752
+    },
+    {
+      "epoch": 13.210526315789474,
+      "grad_norm": 10.347915649414062,
+      "learning_rate": 1.065137614678899e-05,
+      "loss": 0.2108,
+      "step": 753
+    },
+    {
+      "epoch": 13.228070175438596,
+      "grad_norm": 14.886469841003418,
+      "learning_rate": 1.0623853211009175e-05,
+      "loss": 0.1522,
+      "step": 754
+    },
+    {
+      "epoch": 13.24561403508772,
+      "grad_norm": 4.025850296020508,
+      "learning_rate": 1.0596330275229357e-05,
+      "loss": 0.0933,
+      "step": 755
+    },
+    {
+      "epoch": 13.263157894736842,
+      "grad_norm": 8.510875701904297,
+      "learning_rate": 1.0568807339449542e-05,
+      "loss": 0.4154,
+      "step": 756
+    },
+    {
+      "epoch": 13.280701754385966,
+      "grad_norm": 3.559708595275879,
+      "learning_rate": 1.0541284403669724e-05,
+      "loss": 0.067,
+      "step": 757
+    },
+    {
+      "epoch": 13.298245614035087,
+      "grad_norm": 5.087008476257324,
+      "learning_rate": 1.051376146788991e-05,
+      "loss": 0.1601,
+      "step": 758
+    },
+    {
+      "epoch": 13.31578947368421,
+      "grad_norm": 8.079976081848145,
+      "learning_rate": 1.0486238532110092e-05,
+      "loss": 0.1577,
+      "step": 759
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 17.081035614013672,
+      "learning_rate": 1.0458715596330277e-05,
+      "loss": 0.2366,
+      "step": 760
+    },
+    {
+      "epoch": 13.350877192982455,
+      "grad_norm": 1.349025011062622,
+      "learning_rate": 1.0431192660550459e-05,
+      "loss": 0.0553,
+      "step": 761
+    },
+    {
+      "epoch": 13.368421052631579,
+      "grad_norm": 12.248740196228027,
+      "learning_rate": 1.0403669724770643e-05,
+      "loss": 0.1111,
+      "step": 762
+    },
+    {
+      "epoch": 13.385964912280702,
+      "grad_norm": 5.697534561157227,
+      "learning_rate": 1.0376146788990826e-05,
+      "loss": 0.1188,
+      "step": 763
+    },
+    {
+      "epoch": 13.403508771929825,
+      "grad_norm": 9.995718955993652,
+      "learning_rate": 1.034862385321101e-05,
+      "loss": 0.4405,
+      "step": 764
+    },
+    {
+      "epoch": 13.421052631578947,
+      "grad_norm": 3.141644239425659,
+      "learning_rate": 1.0321100917431192e-05,
+      "loss": 0.1329,
+      "step": 765
+    },
+    {
+      "epoch": 13.43859649122807,
+      "grad_norm": 31.985647201538086,
+      "learning_rate": 1.0293577981651376e-05,
+      "loss": 0.1126,
+      "step": 766
+    },
+    {
+      "epoch": 13.456140350877194,
+      "grad_norm": 9.96912956237793,
+      "learning_rate": 1.0266055045871559e-05,
+      "loss": 0.51,
+      "step": 767
+    },
+    {
+      "epoch": 13.473684210526315,
+      "grad_norm": 12.615479469299316,
+      "learning_rate": 1.0238532110091745e-05,
+      "loss": 0.286,
+      "step": 768
+    },
+    {
+      "epoch": 13.491228070175438,
+      "grad_norm": 6.572673797607422,
+      "learning_rate": 1.0211009174311927e-05,
+      "loss": 0.1562,
+      "step": 769
+    },
+    {
+      "epoch": 13.508771929824562,
+      "grad_norm": 1.261430025100708,
+      "learning_rate": 1.0183486238532111e-05,
+      "loss": 0.078,
+      "step": 770
+    },
+    {
+      "epoch": 13.526315789473685,
+      "grad_norm": 11.288647651672363,
+      "learning_rate": 1.0155963302752294e-05,
+      "loss": 0.2406,
+      "step": 771
+    },
+    {
+      "epoch": 13.543859649122806,
+      "grad_norm": 6.564558506011963,
+      "learning_rate": 1.0128440366972478e-05,
+      "loss": 0.2341,
+      "step": 772
+    },
+    {
+      "epoch": 13.56140350877193,
+      "grad_norm": 71.70243072509766,
+      "learning_rate": 1.010091743119266e-05,
+      "loss": 0.5161,
+      "step": 773
+    },
+    {
+      "epoch": 13.578947368421053,
+      "grad_norm": 1.2909406423568726,
+      "learning_rate": 1.0073394495412845e-05,
+      "loss": 0.0777,
+      "step": 774
+    },
+    {
+      "epoch": 13.596491228070175,
+      "grad_norm": 9.205297470092773,
+      "learning_rate": 1.0045871559633027e-05,
+      "loss": 0.2522,
+      "step": 775
+    },
+    {
+      "epoch": 13.614035087719298,
+      "grad_norm": 19.708072662353516,
+      "learning_rate": 1.0018348623853211e-05,
+      "loss": 0.4493,
+      "step": 776
+    },
+    {
+      "epoch": 13.631578947368421,
+      "grad_norm": 5.593297958374023,
+      "learning_rate": 9.990825688073394e-06,
+      "loss": 0.1821,
+      "step": 777
+    },
+    {
+      "epoch": 13.649122807017545,
+      "grad_norm": 0.7555585503578186,
+      "learning_rate": 9.963302752293578e-06,
+      "loss": 0.066,
+      "step": 778
+    },
+    {
+      "epoch": 13.666666666666666,
+      "grad_norm": 9.295495986938477,
+      "learning_rate": 9.935779816513762e-06,
+      "loss": 0.3145,
+      "step": 779
+    },
+    {
+      "epoch": 13.68421052631579,
+      "grad_norm": 12.652939796447754,
+      "learning_rate": 9.908256880733946e-06,
+      "loss": 0.2695,
+      "step": 780
+    },
+    {
+      "epoch": 13.701754385964913,
+      "grad_norm": 0.517791211605072,
+      "learning_rate": 9.880733944954129e-06,
+      "loss": 0.047,
+      "step": 781
+    },
+    {
+      "epoch": 13.719298245614034,
+      "grad_norm": 2.591848850250244,
+      "learning_rate": 9.853211009174313e-06,
+      "loss": 0.0483,
+      "step": 782
+    },
+    {
+      "epoch": 13.736842105263158,
+      "grad_norm": 13.37531566619873,
+      "learning_rate": 9.825688073394495e-06,
+      "loss": 0.3604,
+      "step": 783
+    },
+    {
+      "epoch": 13.75438596491228,
+      "grad_norm": 9.913411140441895,
+      "learning_rate": 9.79816513761468e-06,
+      "loss": 0.1038,
+      "step": 784
+    },
+    {
+      "epoch": 13.771929824561404,
+      "grad_norm": 84.06814575195312,
+      "learning_rate": 9.770642201834862e-06,
+      "loss": 0.2635,
+      "step": 785
+    },
+    {
+      "epoch": 13.789473684210526,
+      "grad_norm": 12.877340316772461,
+      "learning_rate": 9.743119266055046e-06,
+      "loss": 0.318,
+      "step": 786
+    },
+    {
+      "epoch": 13.807017543859649,
+      "grad_norm": 0.8507923483848572,
+      "learning_rate": 9.715596330275229e-06,
+      "loss": 0.049,
+      "step": 787
+    },
+    {
+      "epoch": 13.824561403508772,
+      "grad_norm": 47.35422134399414,
+      "learning_rate": 9.688073394495413e-06,
+      "loss": 0.4669,
+      "step": 788
+    },
+    {
+      "epoch": 13.842105263157894,
+      "grad_norm": 4.622620582580566,
+      "learning_rate": 9.660550458715597e-06,
+      "loss": 0.1298,
+      "step": 789
+    },
+    {
+      "epoch": 13.859649122807017,
+      "grad_norm": 12.158206939697266,
+      "learning_rate": 9.633027522935781e-06,
+      "loss": 0.3268,
+      "step": 790
+    },
+    {
+      "epoch": 13.87719298245614,
+      "grad_norm": 6.32510232925415,
+      "learning_rate": 9.605504587155964e-06,
+      "loss": 0.1676,
+      "step": 791
+    },
+    {
+      "epoch": 13.894736842105264,
+      "grad_norm": 7.81455135345459,
+      "learning_rate": 9.577981651376148e-06,
+      "loss": 0.327,
+      "step": 792
+    },
+    {
+      "epoch": 13.912280701754385,
+      "grad_norm": 6.204614639282227,
+      "learning_rate": 9.55045871559633e-06,
+      "loss": 0.2074,
+      "step": 793
+    },
+    {
+      "epoch": 13.929824561403509,
+      "grad_norm": 3.006291389465332,
+      "learning_rate": 9.522935779816515e-06,
+      "loss": 0.1931,
+      "step": 794
+    },
+    {
+      "epoch": 13.947368421052632,
+      "grad_norm": 27.143718719482422,
+      "learning_rate": 9.495412844036697e-06,
+      "loss": 0.2277,
+      "step": 795
+    },
+    {
+      "epoch": 13.964912280701755,
+      "grad_norm": 14.346508026123047,
+      "learning_rate": 9.467889908256881e-06,
+      "loss": 0.3415,
+      "step": 796
+    },
+    {
+      "epoch": 13.982456140350877,
+      "grad_norm": 23.73762321472168,
+      "learning_rate": 9.440366972477064e-06,
+      "loss": 0.3348,
+      "step": 797
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 340.23724365234375,
+      "learning_rate": 9.412844036697248e-06,
+      "loss": 0.1826,
+      "step": 798
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.685823754789272,
+      "eval_loss": 1.013038992881775,
+      "eval_roc_auc": 0.8966899641922588,
+      "eval_runtime": 8.463,
+      "eval_samples_per_second": 30.84,
+      "eval_steps_per_second": 2.009,
+      "step": 798
+    },
+    {
+      "epoch": 14.017543859649123,
+      "grad_norm": 9.037409782409668,
+      "learning_rate": 9.38532110091743e-06,
+      "loss": 0.2681,
+      "step": 799
+    },
+    {
+      "epoch": 14.035087719298245,
+      "grad_norm": 14.613738059997559,
+      "learning_rate": 9.357798165137616e-06,
+      "loss": 0.2714,
+      "step": 800
+    },
+    {
+      "epoch": 14.052631578947368,
+      "grad_norm": 18.193498611450195,
+      "learning_rate": 9.330275229357799e-06,
+      "loss": 0.2295,
+      "step": 801
+    },
+    {
+      "epoch": 14.070175438596491,
+      "grad_norm": 19.435922622680664,
+      "learning_rate": 9.302752293577983e-06,
+      "loss": 0.568,
+      "step": 802
+    },
+    {
+      "epoch": 14.087719298245615,
+      "grad_norm": 5.796917915344238,
+      "learning_rate": 9.275229357798165e-06,
+      "loss": 0.1761,
+      "step": 803
+    },
+    {
+      "epoch": 14.105263157894736,
+      "grad_norm": 16.154891967773438,
+      "learning_rate": 9.24770642201835e-06,
+      "loss": 0.2004,
+      "step": 804
+    },
+    {
+      "epoch": 14.12280701754386,
+      "grad_norm": 50.181121826171875,
+      "learning_rate": 9.220183486238532e-06,
+      "loss": 0.286,
+      "step": 805
+    },
+    {
+      "epoch": 14.140350877192983,
+      "grad_norm": 9.757275581359863,
+      "learning_rate": 9.192660550458716e-06,
+      "loss": 0.1081,
+      "step": 806
+    },
+    {
+      "epoch": 14.157894736842104,
+      "grad_norm": 10.20756721496582,
+      "learning_rate": 9.165137614678899e-06,
+      "loss": 0.1053,
+      "step": 807
+    },
+    {
+      "epoch": 14.175438596491228,
+      "grad_norm": 147.6493682861328,
+      "learning_rate": 9.137614678899083e-06,
+      "loss": 0.3689,
+      "step": 808
+    },
+    {
+      "epoch": 14.192982456140351,
+      "grad_norm": 5.954428195953369,
+      "learning_rate": 9.110091743119265e-06,
+      "loss": 0.1967,
+      "step": 809
+    },
+    {
+      "epoch": 14.210526315789474,
+      "grad_norm": 0.8736594915390015,
+      "learning_rate": 9.082568807339451e-06,
+      "loss": 0.0672,
+      "step": 810
+    },
+    {
+      "epoch": 14.228070175438596,
+      "grad_norm": 19.074871063232422,
+      "learning_rate": 9.055045871559634e-06,
+      "loss": 0.2219,
+      "step": 811
+    },
+    {
+      "epoch": 14.24561403508772,
+      "grad_norm": 24.449487686157227,
+      "learning_rate": 9.027522935779818e-06,
+      "loss": 0.3973,
+      "step": 812
+    },
+    {
+      "epoch": 14.263157894736842,
+      "grad_norm": 15.369214057922363,
+      "learning_rate": 9e-06,
+      "loss": 0.3232,
+      "step": 813
+    },
+    {
+      "epoch": 14.280701754385966,
+      "grad_norm": 9.815387725830078,
+      "learning_rate": 8.972477064220184e-06,
+      "loss": 0.2076,
+      "step": 814
+    },
+    {
+      "epoch": 14.298245614035087,
+      "grad_norm": 4.816423416137695,
+      "learning_rate": 8.944954128440367e-06,
+      "loss": 0.0842,
+      "step": 815
+    },
+    {
+      "epoch": 14.31578947368421,
+      "grad_norm": 18.863229751586914,
+      "learning_rate": 8.917431192660551e-06,
+      "loss": 0.1569,
+      "step": 816
+    },
+    {
+      "epoch": 14.333333333333334,
+      "grad_norm": 11.182153701782227,
+      "learning_rate": 8.889908256880734e-06,
+      "loss": 0.1107,
+      "step": 817
+    },
+    {
+      "epoch": 14.350877192982455,
+      "grad_norm": 1.1255884170532227,
+      "learning_rate": 8.862385321100918e-06,
+      "loss": 0.0429,
+      "step": 818
+    },
+    {
+      "epoch": 14.368421052631579,
+      "grad_norm": 8.27351188659668,
+      "learning_rate": 8.8348623853211e-06,
+      "loss": 0.0553,
+      "step": 819
+    },
+    {
+      "epoch": 14.385964912280702,
+      "grad_norm": 2.460218667984009,
+      "learning_rate": 8.807339449541284e-06,
+      "loss": 0.0525,
+      "step": 820
+    },
+    {
+      "epoch": 14.403508771929825,
+      "grad_norm": 9.96479606628418,
+      "learning_rate": 8.779816513761469e-06,
+      "loss": 0.2107,
+      "step": 821
+    },
+    {
+      "epoch": 14.421052631578947,
+      "grad_norm": 19.677907943725586,
+      "learning_rate": 8.752293577981653e-06,
+      "loss": 0.2648,
+      "step": 822
+    },
+    {
+      "epoch": 14.43859649122807,
+      "grad_norm": 10.419135093688965,
+      "learning_rate": 8.724770642201835e-06,
+      "loss": 0.1504,
+      "step": 823
+    },
+    {
+      "epoch": 14.456140350877194,
+      "grad_norm": 15.515992164611816,
+      "learning_rate": 8.69724770642202e-06,
+      "loss": 0.1876,
+      "step": 824
+    },
+    {
+      "epoch": 14.473684210526315,
+      "grad_norm": 4.553437232971191,
+      "learning_rate": 8.669724770642202e-06,
+      "loss": 0.0726,
+      "step": 825
+    },
+    {
+      "epoch": 14.491228070175438,
+      "grad_norm": 6.339976787567139,
+      "learning_rate": 8.642201834862386e-06,
+      "loss": 0.2749,
+      "step": 826
+    },
+    {
+      "epoch": 14.508771929824562,
+      "grad_norm": 23.903335571289062,
+      "learning_rate": 8.614678899082568e-06,
+      "loss": 0.1898,
+      "step": 827
+    },
+    {
+      "epoch": 14.526315789473685,
+      "grad_norm": 3.4404420852661133,
+      "learning_rate": 8.587155963302753e-06,
+      "loss": 0.145,
+      "step": 828
+    },
+    {
+      "epoch": 14.543859649122806,
+      "grad_norm": 0.5167752504348755,
+      "learning_rate": 8.559633027522935e-06,
+      "loss": 0.0594,
+      "step": 829
+    },
+    {
+      "epoch": 14.56140350877193,
+      "grad_norm": 13.959207534790039,
+      "learning_rate": 8.53211009174312e-06,
+      "loss": 0.3462,
+      "step": 830
+    },
+    {
+      "epoch": 14.578947368421053,
+      "grad_norm": 5.944933891296387,
+      "learning_rate": 8.504587155963303e-06,
+      "loss": 0.0515,
+      "step": 831
+    },
+    {
+      "epoch": 14.596491228070175,
+      "grad_norm": 8.689725875854492,
+      "learning_rate": 8.477064220183488e-06,
+      "loss": 0.1593,
+      "step": 832
+    },
+    {
+      "epoch": 14.614035087719298,
+      "grad_norm": 26.8358097076416,
+      "learning_rate": 8.44954128440367e-06,
+      "loss": 0.1423,
+      "step": 833
+    },
+    {
+      "epoch": 14.631578947368421,
+      "grad_norm": 3.620288372039795,
+      "learning_rate": 8.422018348623854e-06,
+      "loss": 0.0726,
+      "step": 834
+    },
+    {
+      "epoch": 14.649122807017545,
+      "grad_norm": 6.725648880004883,
+      "learning_rate": 8.394495412844037e-06,
+      "loss": 0.0989,
+      "step": 835
+    },
+    {
+      "epoch": 14.666666666666666,
+      "grad_norm": 6.595190048217773,
+      "learning_rate": 8.366972477064221e-06,
+      "loss": 0.3077,
+      "step": 836
+    },
+    {
+      "epoch": 14.68421052631579,
+      "grad_norm": 14.732500076293945,
+      "learning_rate": 8.339449541284403e-06,
+      "loss": 0.1057,
+      "step": 837
+    },
+    {
+      "epoch": 14.701754385964913,
+      "grad_norm": 23.38099479675293,
+      "learning_rate": 8.311926605504588e-06,
+      "loss": 0.2858,
+      "step": 838
+    },
+    {
+      "epoch": 14.719298245614034,
+      "grad_norm": 6.091468334197998,
+      "learning_rate": 8.28440366972477e-06,
+      "loss": 0.1929,
+      "step": 839
+    },
+    {
+      "epoch": 14.736842105263158,
+      "grad_norm": 19.15097999572754,
+      "learning_rate": 8.256880733944954e-06,
+      "loss": 0.1127,
+      "step": 840
+    },
+    {
+      "epoch": 14.75438596491228,
+      "grad_norm": 7.495136737823486,
+      "learning_rate": 8.229357798165137e-06,
+      "loss": 0.1402,
+      "step": 841
+    },
+    {
+      "epoch": 14.771929824561404,
+      "grad_norm": 10.095050811767578,
+      "learning_rate": 8.201834862385323e-06,
+      "loss": 0.1918,
+      "step": 842
+    },
+    {
+      "epoch": 14.789473684210526,
+      "grad_norm": 10.647483825683594,
+      "learning_rate": 8.174311926605505e-06,
+      "loss": 0.4263,
+      "step": 843
+    },
+    {
+      "epoch": 14.807017543859649,
+      "grad_norm": 4.994744300842285,
+      "learning_rate": 8.146788990825689e-06,
+      "loss": 0.1223,
+      "step": 844
+    },
+    {
+      "epoch": 14.824561403508772,
+      "grad_norm": 1.3206747770309448,
+      "learning_rate": 8.119266055045872e-06,
+      "loss": 0.0626,
+      "step": 845
+    },
+    {
+      "epoch": 14.842105263157894,
+      "grad_norm": 10.279633522033691,
+      "learning_rate": 8.091743119266056e-06,
+      "loss": 0.1556,
+      "step": 846
+    },
+    {
+      "epoch": 14.859649122807017,
+      "grad_norm": 6.295610427856445,
+      "learning_rate": 8.064220183486238e-06,
+      "loss": 0.1125,
+      "step": 847
+    },
+    {
+      "epoch": 14.87719298245614,
+      "grad_norm": 6.005034446716309,
+      "learning_rate": 8.036697247706422e-06,
+      "loss": 0.1204,
+      "step": 848
+    },
+    {
+      "epoch": 14.894736842105264,
+      "grad_norm": 88.43120574951172,
+      "learning_rate": 8.009174311926605e-06,
+      "loss": 0.4208,
+      "step": 849
+    },
+    {
+      "epoch": 14.912280701754385,
+      "grad_norm": 4.66910982131958,
+      "learning_rate": 7.981651376146789e-06,
+      "loss": 0.0938,
+      "step": 850
+    },
+    {
+      "epoch": 14.929824561403509,
+      "grad_norm": 10.302392959594727,
+      "learning_rate": 7.954128440366972e-06,
+      "loss": 0.2364,
+      "step": 851
+    },
+    {
+      "epoch": 14.947368421052632,
+      "grad_norm": 12.315834045410156,
+      "learning_rate": 7.926605504587157e-06,
+      "loss": 0.3843,
+      "step": 852
+    },
+    {
+      "epoch": 14.964912280701755,
+      "grad_norm": 2.329768180847168,
+      "learning_rate": 7.89908256880734e-06,
+      "loss": 0.2024,
+      "step": 853
+    },
+    {
+      "epoch": 14.982456140350877,
+      "grad_norm": 5.507812976837158,
+      "learning_rate": 7.871559633027524e-06,
+      "loss": 0.2056,
+      "step": 854
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 15.961007118225098,
+      "learning_rate": 7.844036697247707e-06,
+      "loss": 0.2684,
+      "step": 855
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.6781609195402298,
+      "eval_loss": 1.0226866006851196,
+      "eval_roc_auc": 0.8921742047234581,
+      "eval_runtime": 9.4145,
+      "eval_samples_per_second": 27.723,
+      "eval_steps_per_second": 1.806,
+      "step": 855
+    },
+    {
+      "epoch": 15.017543859649123,
+      "grad_norm": 15.226679801940918,
+      "learning_rate": 7.81651376146789e-06,
+      "loss": 0.2926,
+      "step": 856
+    },
+    {
+      "epoch": 15.035087719298245,
+      "grad_norm": 8.864625930786133,
+      "learning_rate": 7.788990825688073e-06,
+      "loss": 0.1756,
+      "step": 857
+    },
+    {
+      "epoch": 15.052631578947368,
+      "grad_norm": 1.6936672925949097,
+      "learning_rate": 7.761467889908257e-06,
+      "loss": 0.0536,
+      "step": 858
+    },
+    {
+      "epoch": 15.070175438596491,
+      "grad_norm": 15.099090576171875,
+      "learning_rate": 7.73394495412844e-06,
+      "loss": 0.1459,
+      "step": 859
+    },
+    {
+      "epoch": 15.087719298245615,
+      "grad_norm": 5.978076457977295,
+      "learning_rate": 7.706422018348624e-06,
+      "loss": 0.116,
+      "step": 860
+    },
+    {
+      "epoch": 15.105263157894736,
+      "grad_norm": 136.03578186035156,
+      "learning_rate": 7.678899082568806e-06,
+      "loss": 0.1331,
+      "step": 861
+    },
+    {
+      "epoch": 15.12280701754386,
+      "grad_norm": 0.6831678152084351,
+      "learning_rate": 7.65137614678899e-06,
+      "loss": 0.0381,
+      "step": 862
+    },
+    {
+      "epoch": 15.140350877192983,
+      "grad_norm": 16.589181900024414,
+      "learning_rate": 7.623853211009174e-06,
+      "loss": 0.2011,
+      "step": 863
+    },
+    {
+      "epoch": 15.157894736842104,
+      "grad_norm": 4.638211250305176,
+      "learning_rate": 7.596330275229358e-06,
+      "loss": 0.0889,
+      "step": 864
+    },
+    {
+      "epoch": 15.175438596491228,
+      "grad_norm": 10.425081253051758,
+      "learning_rate": 7.5688073394495415e-06,
+      "loss": 0.1062,
+      "step": 865
+    },
+    {
+      "epoch": 15.192982456140351,
+      "grad_norm": 5.109747886657715,
+      "learning_rate": 7.541284403669726e-06,
+      "loss": 0.1663,
+      "step": 866
+    },
+    {
+      "epoch": 15.210526315789474,
+      "grad_norm": 8.572009086608887,
+      "learning_rate": 7.513761467889908e-06,
+      "loss": 0.1877,
+      "step": 867
+    },
+    {
+      "epoch": 15.228070175438596,
+      "grad_norm": 7.750594139099121,
+      "learning_rate": 7.486238532110092e-06,
+      "loss": 0.1766,
+      "step": 868
+    },
+    {
+      "epoch": 15.24561403508772,
+      "grad_norm": 13.109859466552734,
+      "learning_rate": 7.458715596330276e-06,
+      "loss": 0.1765,
+      "step": 869
+    },
+    {
+      "epoch": 15.263157894736842,
+      "grad_norm": 3.056985855102539,
+      "learning_rate": 7.431192660550459e-06,
+      "loss": 0.0814,
+      "step": 870
+    },
+    {
+      "epoch": 15.280701754385966,
+      "grad_norm": 0.44665461778640747,
+      "learning_rate": 7.403669724770643e-06,
+      "loss": 0.0525,
+      "step": 871
+    },
+    {
+      "epoch": 15.298245614035087,
+      "grad_norm": 15.339385032653809,
+      "learning_rate": 7.376146788990826e-06,
+      "loss": 0.1396,
+      "step": 872
+    },
+    {
+      "epoch": 15.31578947368421,
+      "grad_norm": 6.135118007659912,
+      "learning_rate": 7.34862385321101e-06,
+      "loss": 0.1651,
+      "step": 873
+    },
+    {
+      "epoch": 15.333333333333334,
+      "grad_norm": 5.053482532501221,
+      "learning_rate": 7.321100917431193e-06,
+      "loss": 0.1773,
+      "step": 874
+    },
+    {
+      "epoch": 15.350877192982455,
+      "grad_norm": 2.368586301803589,
+      "learning_rate": 7.293577981651376e-06,
+      "loss": 0.1684,
+      "step": 875
+    },
+    {
+      "epoch": 15.368421052631579,
+      "grad_norm": 39.954166412353516,
+      "learning_rate": 7.2660550458715605e-06,
+      "loss": 0.3675,
+      "step": 876
+    },
+    {
+      "epoch": 15.385964912280702,
+      "grad_norm": 0.6676788330078125,
+      "learning_rate": 7.238532110091744e-06,
+      "loss": 0.0364,
+      "step": 877
+    },
+    {
+      "epoch": 15.403508771929825,
+      "grad_norm": 7.863750457763672,
+      "learning_rate": 7.211009174311927e-06,
+      "loss": 0.2284,
+      "step": 878
+    },
+    {
+      "epoch": 15.421052631578947,
+      "grad_norm": 2.7634315490722656,
+      "learning_rate": 7.1834862385321105e-06,
+      "loss": 0.0924,
+      "step": 879
+    },
+    {
+      "epoch": 15.43859649122807,
+      "grad_norm": 19.595252990722656,
+      "learning_rate": 7.155963302752294e-06,
+      "loss": 0.1652,
+      "step": 880
+    },
+    {
+      "epoch": 15.456140350877194,
+      "grad_norm": 43.90449905395508,
+      "learning_rate": 7.128440366972478e-06,
+      "loss": 0.3822,
+      "step": 881
+    },
+    {
+      "epoch": 15.473684210526315,
+      "grad_norm": 12.782305717468262,
+      "learning_rate": 7.100917431192661e-06,
+      "loss": 0.1843,
+      "step": 882
+    },
+    {
+      "epoch": 15.491228070175438,
+      "grad_norm": 0.46928420662879944,
+      "learning_rate": 7.073394495412845e-06,
+      "loss": 0.0541,
+      "step": 883
+    },
+    {
+      "epoch": 15.508771929824562,
+      "grad_norm": 7.558313846588135,
+      "learning_rate": 7.045871559633028e-06,
+      "loss": 0.228,
+      "step": 884
+    },
+    {
+      "epoch": 15.526315789473685,
+      "grad_norm": 1.2508295774459839,
+      "learning_rate": 7.018348623853211e-06,
+      "loss": 0.0703,
+      "step": 885
+    },
+    {
+      "epoch": 15.543859649122806,
+      "grad_norm": 14.253005981445312,
+      "learning_rate": 6.9908256880733955e-06,
+      "loss": 0.0983,
+      "step": 886
+    },
+    {
+      "epoch": 15.56140350877193,
+      "grad_norm": 9.677652359008789,
+      "learning_rate": 6.963302752293579e-06,
+      "loss": 0.186,
+      "step": 887
+    },
+    {
+      "epoch": 15.578947368421053,
+      "grad_norm": 24.92579460144043,
+      "learning_rate": 6.935779816513762e-06,
+      "loss": 0.2001,
+      "step": 888
+    },
+    {
+      "epoch": 15.596491228070175,
+      "grad_norm": 40.403839111328125,
+      "learning_rate": 6.9082568807339454e-06,
+      "loss": 0.09,
+      "step": 889
+    },
+    {
+      "epoch": 15.614035087719298,
+      "grad_norm": 11.924145698547363,
+      "learning_rate": 6.880733944954129e-06,
+      "loss": 0.2185,
+      "step": 890
+    },
+    {
+      "epoch": 15.631578947368421,
+      "grad_norm": 1.4456658363342285,
+      "learning_rate": 6.853211009174312e-06,
+      "loss": 0.2724,
+      "step": 891
+    },
+    {
+      "epoch": 15.649122807017545,
+      "grad_norm": 10.164271354675293,
+      "learning_rate": 6.825688073394496e-06,
+      "loss": 0.3432,
+      "step": 892
+    },
+    {
+      "epoch": 15.666666666666666,
+      "grad_norm": 15.096166610717773,
+      "learning_rate": 6.7981651376146796e-06,
+      "loss": 0.2307,
+      "step": 893
+    },
+    {
+      "epoch": 15.68421052631579,
+      "grad_norm": 12.728583335876465,
+      "learning_rate": 6.770642201834863e-06,
+      "loss": 0.242,
+      "step": 894
+    },
+    {
+      "epoch": 15.701754385964913,
+      "grad_norm": 28.29001235961914,
+      "learning_rate": 6.743119266055046e-06,
+      "loss": 0.1087,
+      "step": 895
+    },
+    {
+      "epoch": 15.719298245614034,
+      "grad_norm": 17.925107955932617,
+      "learning_rate": 6.7155963302752295e-06,
+      "loss": 0.4188,
+      "step": 896
+    },
+    {
+      "epoch": 15.736842105263158,
+      "grad_norm": 6.433342933654785,
+      "learning_rate": 6.688073394495414e-06,
+      "loss": 0.0936,
+      "step": 897
+    },
+    {
+      "epoch": 15.75438596491228,
+      "grad_norm": 0.8694811463356018,
+      "learning_rate": 6.660550458715597e-06,
+      "loss": 0.0473,
+      "step": 898
+    },
+    {
+      "epoch": 15.771929824561404,
+      "grad_norm": 30.822620391845703,
+      "learning_rate": 6.63302752293578e-06,
+      "loss": 0.0913,
+      "step": 899
+    },
+    {
+      "epoch": 15.789473684210526,
+      "grad_norm": 22.46412467956543,
+      "learning_rate": 6.605504587155964e-06,
+      "loss": 0.1278,
+      "step": 900
+    },
+    {
+      "epoch": 15.807017543859649,
+      "grad_norm": 19.16153335571289,
+      "learning_rate": 6.577981651376147e-06,
+      "loss": 0.2673,
+      "step": 901
+    },
+    {
+      "epoch": 15.824561403508772,
+      "grad_norm": 50.9233283996582,
+      "learning_rate": 6.550458715596331e-06,
+      "loss": 0.557,
+      "step": 902
+    },
+    {
+      "epoch": 15.842105263157894,
+      "grad_norm": 13.619710922241211,
+      "learning_rate": 6.5229357798165145e-06,
+      "loss": 0.2187,
+      "step": 903
+    },
+    {
+      "epoch": 15.859649122807017,
+      "grad_norm": 3.181476593017578,
+      "learning_rate": 6.495412844036698e-06,
+      "loss": 0.1298,
+      "step": 904
+    },
+    {
+      "epoch": 15.87719298245614,
+      "grad_norm": 12.634454727172852,
+      "learning_rate": 6.467889908256881e-06,
+      "loss": 0.1284,
+      "step": 905
+    },
+    {
+      "epoch": 15.894736842105264,
+      "grad_norm": 21.935789108276367,
+      "learning_rate": 6.4403669724770645e-06,
+      "loss": 0.5293,
+      "step": 906
+    },
+    {
+      "epoch": 15.912280701754385,
+      "grad_norm": 20.671627044677734,
+      "learning_rate": 6.412844036697249e-06,
+      "loss": 0.1588,
+      "step": 907
+    },
+    {
+      "epoch": 15.929824561403509,
+      "grad_norm": 7.02576208114624,
+      "learning_rate": 6.385321100917432e-06,
+      "loss": 0.1236,
+      "step": 908
+    },
+    {
+      "epoch": 15.947368421052632,
+      "grad_norm": 10.424530982971191,
+      "learning_rate": 6.357798165137615e-06,
+      "loss": 0.4814,
+      "step": 909
+    },
+    {
+      "epoch": 15.964912280701755,
+      "grad_norm": 11.366877555847168,
+      "learning_rate": 6.330275229357799e-06,
+      "loss": 0.3608,
+      "step": 910
+    },
+    {
+      "epoch": 15.982456140350877,
+      "grad_norm": 0.31679338216781616,
+      "learning_rate": 6.302752293577982e-06,
+      "loss": 0.0309,
+      "step": 911
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 25.50751495361328,
+      "learning_rate": 6.275229357798165e-06,
+      "loss": 0.3512,
+      "step": 912
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.7088122605363985,
+      "eval_loss": 1.0425840616226196,
+      "eval_roc_auc": 0.8961770097130725,
+      "eval_runtime": 9.7922,
+      "eval_samples_per_second": 26.654,
+      "eval_steps_per_second": 1.736,
+      "step": 912
+    },
+    {
+      "epoch": 16.017543859649123,
+      "grad_norm": 5.3255228996276855,
+      "learning_rate": 6.247706422018349e-06,
+      "loss": 0.0517,
+      "step": 913
+    },
+    {
+      "epoch": 16.035087719298247,
+      "grad_norm": 9.271040916442871,
+      "learning_rate": 6.220183486238533e-06,
+      "loss": 0.1322,
+      "step": 914
+    },
+    {
+      "epoch": 16.05263157894737,
+      "grad_norm": 4.202192783355713,
+      "learning_rate": 6.192660550458716e-06,
+      "loss": 0.1875,
+      "step": 915
+    },
+    {
+      "epoch": 16.07017543859649,
+      "grad_norm": 5.035487174987793,
+      "learning_rate": 6.165137614678899e-06,
+      "loss": 0.2728,
+      "step": 916
+    },
+    {
+      "epoch": 16.087719298245613,
+      "grad_norm": 4.098052024841309,
+      "learning_rate": 6.137614678899083e-06,
+      "loss": 0.1016,
+      "step": 917
+    },
+    {
+      "epoch": 16.105263157894736,
+      "grad_norm": 0.5254385471343994,
+      "learning_rate": 6.110091743119267e-06,
+      "loss": 0.0444,
+      "step": 918
+    },
+    {
+      "epoch": 16.12280701754386,
+      "grad_norm": 3.7572407722473145,
+      "learning_rate": 6.08256880733945e-06,
+      "loss": 0.0953,
+      "step": 919
+    },
+    {
+      "epoch": 16.140350877192983,
+      "grad_norm": 9.17808723449707,
+      "learning_rate": 6.0550458715596335e-06,
+      "loss": 0.0714,
+      "step": 920
+    },
+    {
+      "epoch": 16.157894736842106,
+      "grad_norm": 13.068717002868652,
+      "learning_rate": 6.027522935779817e-06,
+      "loss": 0.5088,
+      "step": 921
+    },
+    {
+      "epoch": 16.17543859649123,
+      "grad_norm": 18.120952606201172,
+      "learning_rate": 6e-06,
+      "loss": 0.3557,
+      "step": 922
+    },
+    {
+      "epoch": 16.19298245614035,
+      "grad_norm": 17.2962589263916,
+      "learning_rate": 5.972477064220184e-06,
+      "loss": 0.2254,
+      "step": 923
+    },
+    {
+      "epoch": 16.210526315789473,
+      "grad_norm": 12.725507736206055,
+      "learning_rate": 5.944954128440368e-06,
+      "loss": 0.2201,
+      "step": 924
+    },
+    {
+      "epoch": 16.228070175438596,
+      "grad_norm": 9.663554191589355,
+      "learning_rate": 5.917431192660551e-06,
+      "loss": 0.4431,
+      "step": 925
+    },
+    {
+      "epoch": 16.24561403508772,
+      "grad_norm": 10.163387298583984,
+      "learning_rate": 5.889908256880734e-06,
+      "loss": 0.1505,
+      "step": 926
+    },
+    {
+      "epoch": 16.263157894736842,
+      "grad_norm": 4.416619300842285,
+      "learning_rate": 5.862385321100918e-06,
+      "loss": 0.1077,
+      "step": 927
+    },
+    {
+      "epoch": 16.280701754385966,
+      "grad_norm": 9.514927864074707,
+      "learning_rate": 5.834862385321102e-06,
+      "loss": 0.209,
+      "step": 928
+    },
+    {
+      "epoch": 16.29824561403509,
+      "grad_norm": 13.716423988342285,
+      "learning_rate": 5.807339449541285e-06,
+      "loss": 0.2191,
+      "step": 929
+    },
+    {
+      "epoch": 16.31578947368421,
+      "grad_norm": 6.016623020172119,
+      "learning_rate": 5.7798165137614684e-06,
+      "loss": 0.2304,
+      "step": 930
+    },
+    {
+      "epoch": 16.333333333333332,
+      "grad_norm": 3.8412370681762695,
+      "learning_rate": 5.752293577981652e-06,
+      "loss": 0.1962,
+      "step": 931
+    },
+    {
+      "epoch": 16.350877192982455,
+      "grad_norm": 2.5499162673950195,
+      "learning_rate": 5.724770642201835e-06,
+      "loss": 0.053,
+      "step": 932
+    },
+    {
+      "epoch": 16.36842105263158,
+      "grad_norm": 0.8069320917129517,
+      "learning_rate": 5.697247706422018e-06,
+      "loss": 0.0544,
+      "step": 933
+    },
+    {
+      "epoch": 16.385964912280702,
+      "grad_norm": 10.786516189575195,
+      "learning_rate": 5.6697247706422026e-06,
+      "loss": 0.2443,
+      "step": 934
+    },
+    {
+      "epoch": 16.403508771929825,
+      "grad_norm": 5.2906389236450195,
+      "learning_rate": 5.642201834862386e-06,
+      "loss": 0.0895,
+      "step": 935
+    },
+    {
+      "epoch": 16.42105263157895,
+      "grad_norm": 7.044477939605713,
+      "learning_rate": 5.614678899082569e-06,
+      "loss": 0.1128,
+      "step": 936
+    },
+    {
+      "epoch": 16.43859649122807,
+      "grad_norm": 4.098491668701172,
+      "learning_rate": 5.5871559633027525e-06,
+      "loss": 0.0885,
+      "step": 937
+    },
+    {
+      "epoch": 16.45614035087719,
+      "grad_norm": 24.7549991607666,
+      "learning_rate": 5.559633027522936e-06,
+      "loss": 0.2579,
+      "step": 938
+    },
+    {
+      "epoch": 16.473684210526315,
+      "grad_norm": 9.953288078308105,
+      "learning_rate": 5.53211009174312e-06,
+      "loss": 0.2695,
+      "step": 939
+    },
+    {
+      "epoch": 16.49122807017544,
+      "grad_norm": 1.9676154851913452,
+      "learning_rate": 5.504587155963303e-06,
+      "loss": 0.0861,
+      "step": 940
+    },
+    {
+      "epoch": 16.50877192982456,
+      "grad_norm": 39.79118728637695,
+      "learning_rate": 5.477064220183487e-06,
+      "loss": 0.2086,
+      "step": 941
+    },
+    {
+      "epoch": 16.526315789473685,
+      "grad_norm": 8.34636402130127,
+      "learning_rate": 5.44954128440367e-06,
+      "loss": 0.1239,
+      "step": 942
+    },
+    {
+      "epoch": 16.54385964912281,
+      "grad_norm": 0.8310889601707458,
+      "learning_rate": 5.422018348623853e-06,
+      "loss": 0.0438,
+      "step": 943
+    },
+    {
+      "epoch": 16.56140350877193,
+      "grad_norm": 2.904680013656616,
+      "learning_rate": 5.3944954128440375e-06,
+      "loss": 0.0959,
+      "step": 944
+    },
+    {
+      "epoch": 16.57894736842105,
+      "grad_norm": 16.282615661621094,
+      "learning_rate": 5.366972477064221e-06,
+      "loss": 0.2871,
+      "step": 945
+    },
+    {
+      "epoch": 16.596491228070175,
+      "grad_norm": 7.053973197937012,
+      "learning_rate": 5.339449541284404e-06,
+      "loss": 0.2315,
+      "step": 946
+    },
+    {
+      "epoch": 16.614035087719298,
+      "grad_norm": 10.810872077941895,
+      "learning_rate": 5.3119266055045874e-06,
+      "loss": 0.1121,
+      "step": 947
+    },
+    {
+      "epoch": 16.63157894736842,
+      "grad_norm": 5.931017875671387,
+      "learning_rate": 5.284403669724771e-06,
+      "loss": 0.0757,
+      "step": 948
+    },
+    {
+      "epoch": 16.649122807017545,
+      "grad_norm": 3.6563878059387207,
+      "learning_rate": 5.256880733944955e-06,
+      "loss": 0.0865,
+      "step": 949
+    },
+    {
+      "epoch": 16.666666666666668,
+      "grad_norm": 1.590165138244629,
+      "learning_rate": 5.229357798165138e-06,
+      "loss": 0.0639,
+      "step": 950
+    },
+    {
+      "epoch": 16.68421052631579,
+      "grad_norm": 2.1470823287963867,
+      "learning_rate": 5.201834862385322e-06,
+      "loss": 0.0781,
+      "step": 951
+    },
+    {
+      "epoch": 16.70175438596491,
+      "grad_norm": 8.965104103088379,
+      "learning_rate": 5.174311926605505e-06,
+      "loss": 0.1548,
+      "step": 952
+    },
+    {
+      "epoch": 16.719298245614034,
+      "grad_norm": 27.018918991088867,
+      "learning_rate": 5.146788990825688e-06,
+      "loss": 0.4119,
+      "step": 953
+    },
+    {
+      "epoch": 16.736842105263158,
+      "grad_norm": 7.252739906311035,
+      "learning_rate": 5.119266055045872e-06,
+      "loss": 0.2262,
+      "step": 954
+    },
+    {
+      "epoch": 16.75438596491228,
+      "grad_norm": 14.698616981506348,
+      "learning_rate": 5.091743119266056e-06,
+      "loss": 0.1485,
+      "step": 955
+    },
+    {
+      "epoch": 16.771929824561404,
+      "grad_norm": 5.847207546234131,
+      "learning_rate": 5.064220183486239e-06,
+      "loss": 0.1799,
+      "step": 956
+    },
+    {
+      "epoch": 16.789473684210527,
+      "grad_norm": 5.336987018585205,
+      "learning_rate": 5.036697247706422e-06,
+      "loss": 0.0802,
+      "step": 957
+    },
+    {
+      "epoch": 16.80701754385965,
+      "grad_norm": 8.911815643310547,
+      "learning_rate": 5.009174311926606e-06,
+      "loss": 0.109,
+      "step": 958
+    },
+    {
+      "epoch": 16.82456140350877,
+      "grad_norm": 0.5730437636375427,
+      "learning_rate": 4.981651376146789e-06,
+      "loss": 0.0404,
+      "step": 959
+    },
+    {
+      "epoch": 16.842105263157894,
+      "grad_norm": 1.8662869930267334,
+      "learning_rate": 4.954128440366973e-06,
+      "loss": 0.0672,
+      "step": 960
+    },
+    {
+      "epoch": 16.859649122807017,
+      "grad_norm": 10.673025131225586,
+      "learning_rate": 4.9266055045871565e-06,
+      "loss": 0.3202,
+      "step": 961
+    },
+    {
+      "epoch": 16.87719298245614,
+      "grad_norm": 2.2482078075408936,
+      "learning_rate": 4.89908256880734e-06,
+      "loss": 0.0618,
+      "step": 962
+    },
+    {
+      "epoch": 16.894736842105264,
+      "grad_norm": 15.744501113891602,
+      "learning_rate": 4.871559633027523e-06,
+      "loss": 0.5188,
+      "step": 963
+    },
+    {
+      "epoch": 16.912280701754387,
+      "grad_norm": 15.033669471740723,
+      "learning_rate": 4.8440366972477065e-06,
+      "loss": 0.1138,
+      "step": 964
+    },
+    {
+      "epoch": 16.92982456140351,
+      "grad_norm": 13.679003715515137,
+      "learning_rate": 4.816513761467891e-06,
+      "loss": 0.3009,
+      "step": 965
+    },
+    {
+      "epoch": 16.94736842105263,
+      "grad_norm": 27.240638732910156,
+      "learning_rate": 4.788990825688074e-06,
+      "loss": 0.1779,
+      "step": 966
+    },
+    {
+      "epoch": 16.964912280701753,
+      "grad_norm": 8.780830383300781,
+      "learning_rate": 4.761467889908257e-06,
+      "loss": 0.2533,
+      "step": 967
+    },
+    {
+      "epoch": 16.982456140350877,
+      "grad_norm": 2.302766799926758,
+      "learning_rate": 4.733944954128441e-06,
+      "loss": 0.0429,
+      "step": 968
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 6.349617958068848,
+      "learning_rate": 4.706422018348624e-06,
+      "loss": 0.1867,
+      "step": 969
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.6896551724137931,
+      "eval_loss": 1.066833734512329,
+      "eval_roc_auc": 0.8918422445130159,
+      "eval_runtime": 8.5336,
+      "eval_samples_per_second": 30.585,
+      "eval_steps_per_second": 1.992,
+      "step": 969
+    },
+    {
+      "epoch": 17.017543859649123,
+      "grad_norm": 0.8767920732498169,
+      "learning_rate": 4.678899082568808e-06,
+      "loss": 0.0473,
+      "step": 970
+    },
+    {
+      "epoch": 17.035087719298247,
+      "grad_norm": 0.24234896898269653,
+      "learning_rate": 4.651376146788991e-06,
+      "loss": 0.0386,
+      "step": 971
+    },
+    {
+      "epoch": 17.05263157894737,
+      "grad_norm": 0.7192365527153015,
+      "learning_rate": 4.623853211009175e-06,
+      "loss": 0.0648,
+      "step": 972
+    },
+    {
+      "epoch": 17.07017543859649,
+      "grad_norm": 13.74758529663086,
+      "learning_rate": 4.596330275229358e-06,
+      "loss": 0.1321,
+      "step": 973
+    },
+    {
+      "epoch": 17.087719298245613,
+      "grad_norm": 54.41605758666992,
+      "learning_rate": 4.568807339449541e-06,
+      "loss": 0.1375,
+      "step": 974
+    },
+    {
+      "epoch": 17.105263157894736,
+      "grad_norm": 1.3205739259719849,
+      "learning_rate": 4.5412844036697256e-06,
+      "loss": 0.056,
+      "step": 975
+    },
+    {
+      "epoch": 17.12280701754386,
+      "grad_norm": 15.213194847106934,
+      "learning_rate": 4.513761467889909e-06,
+      "loss": 0.1278,
+      "step": 976
+    },
+    {
+      "epoch": 17.140350877192983,
+      "grad_norm": 5.64646577835083,
+      "learning_rate": 4.486238532110092e-06,
+      "loss": 0.2903,
+      "step": 977
+    },
+    {
+      "epoch": 17.157894736842106,
+      "grad_norm": 24.47264289855957,
+      "learning_rate": 4.4587155963302755e-06,
+      "loss": 0.1468,
+      "step": 978
+    },
+    {
+      "epoch": 17.17543859649123,
+      "grad_norm": 11.426804542541504,
+      "learning_rate": 4.431192660550459e-06,
+      "loss": 0.0811,
+      "step": 979
+    },
+    {
+      "epoch": 17.19298245614035,
+      "grad_norm": 62.90053939819336,
+      "learning_rate": 4.403669724770642e-06,
+      "loss": 0.1826,
+      "step": 980
+    },
+    {
+      "epoch": 17.210526315789473,
+      "grad_norm": 6.824471950531006,
+      "learning_rate": 4.376146788990826e-06,
+      "loss": 0.203,
+      "step": 981
+    },
+    {
+      "epoch": 17.228070175438596,
+      "grad_norm": 40.639400482177734,
+      "learning_rate": 4.34862385321101e-06,
+      "loss": 0.4493,
+      "step": 982
+    },
+    {
+      "epoch": 17.24561403508772,
+      "grad_norm": 4.320133686065674,
+      "learning_rate": 4.321100917431193e-06,
+      "loss": 0.0976,
+      "step": 983
+    },
+    {
+      "epoch": 17.263157894736842,
+      "grad_norm": 15.716089248657227,
+      "learning_rate": 4.293577981651376e-06,
+      "loss": 0.1551,
+      "step": 984
+    },
+    {
+      "epoch": 17.280701754385966,
+      "grad_norm": 3.049983501434326,
+      "learning_rate": 4.26605504587156e-06,
+      "loss": 0.0604,
+      "step": 985
+    },
+    {
+      "epoch": 17.29824561403509,
+      "grad_norm": 1.962673306465149,
+      "learning_rate": 4.238532110091744e-06,
+      "loss": 0.042,
+      "step": 986
+    },
+    {
+      "epoch": 17.31578947368421,
+      "grad_norm": 19.601146697998047,
+      "learning_rate": 4.211009174311927e-06,
+      "loss": 0.4181,
+      "step": 987
+    },
+    {
+      "epoch": 17.333333333333332,
+      "grad_norm": 16.026573181152344,
+      "learning_rate": 4.1834862385321104e-06,
+      "loss": 0.0682,
+      "step": 988
+    },
+    {
+      "epoch": 17.350877192982455,
+      "grad_norm": 13.569026947021484,
+      "learning_rate": 4.155963302752294e-06,
+      "loss": 0.1055,
+      "step": 989
+    },
+    {
+      "epoch": 17.36842105263158,
+      "grad_norm": 0.9620193243026733,
+      "learning_rate": 4.128440366972477e-06,
+      "loss": 0.0294,
+      "step": 990
+    },
+    {
+      "epoch": 17.385964912280702,
+      "grad_norm": 0.5341245532035828,
+      "learning_rate": 4.100917431192661e-06,
+      "loss": 0.0755,
+      "step": 991
+    },
+    {
+      "epoch": 17.403508771929825,
+      "grad_norm": 2.195127248764038,
+      "learning_rate": 4.0733944954128446e-06,
+      "loss": 0.0427,
+      "step": 992
+    },
+    {
+      "epoch": 17.42105263157895,
+      "grad_norm": 45.747459411621094,
+      "learning_rate": 4.045871559633028e-06,
+      "loss": 0.2021,
+      "step": 993
+    },
+    {
+      "epoch": 17.43859649122807,
+      "grad_norm": 18.44023895263672,
+      "learning_rate": 4.018348623853211e-06,
+      "loss": 0.2546,
+      "step": 994
+    },
+    {
+      "epoch": 17.45614035087719,
+      "grad_norm": 14.940963745117188,
+      "learning_rate": 3.9908256880733945e-06,
+      "loss": 0.0749,
+      "step": 995
+    },
+    {
+      "epoch": 17.473684210526315,
+      "grad_norm": 9.153297424316406,
+      "learning_rate": 3.963302752293579e-06,
+      "loss": 0.1652,
+      "step": 996
+    },
+    {
+      "epoch": 17.49122807017544,
+      "grad_norm": 19.046146392822266,
+      "learning_rate": 3.935779816513762e-06,
+      "loss": 0.2192,
+      "step": 997
+    },
+    {
+      "epoch": 17.50877192982456,
+      "grad_norm": 24.5616397857666,
+      "learning_rate": 3.908256880733945e-06,
+      "loss": 0.0883,
+      "step": 998
+    },
+    {
+      "epoch": 17.526315789473685,
+      "grad_norm": 5.023523330688477,
+      "learning_rate": 3.880733944954129e-06,
+      "loss": 0.2193,
+      "step": 999
+    },
+    {
+      "epoch": 17.54385964912281,
+      "grad_norm": 7.122688293457031,
+      "learning_rate": 3.853211009174312e-06,
+      "loss": 0.1682,
+      "step": 1000
+    },
+    {
+      "epoch": 17.56140350877193,
+      "grad_norm": 4.198907375335693,
+      "learning_rate": 3.825688073394495e-06,
+      "loss": 0.0674,
+      "step": 1001
+    },
+    {
+      "epoch": 17.57894736842105,
+      "grad_norm": 6.622169017791748,
+      "learning_rate": 3.798165137614679e-06,
+      "loss": 0.0591,
+      "step": 1002
+    },
+    {
+      "epoch": 17.596491228070175,
+      "grad_norm": 5.111667156219482,
+      "learning_rate": 3.770642201834863e-06,
+      "loss": 0.104,
+      "step": 1003
+    },
+    {
+      "epoch": 17.614035087719298,
+      "grad_norm": 7.502628326416016,
+      "learning_rate": 3.743119266055046e-06,
+      "loss": 0.0722,
+      "step": 1004
+    },
+    {
+      "epoch": 17.63157894736842,
+      "grad_norm": 0.4268607795238495,
+      "learning_rate": 3.7155963302752295e-06,
+      "loss": 0.036,
+      "step": 1005
+    },
+    {
+      "epoch": 17.649122807017545,
+      "grad_norm": 5.627065181732178,
+      "learning_rate": 3.688073394495413e-06,
+      "loss": 0.2015,
+      "step": 1006
+    },
+    {
+      "epoch": 17.666666666666668,
+      "grad_norm": 10.972799301147461,
+      "learning_rate": 3.6605504587155965e-06,
+      "loss": 0.0586,
+      "step": 1007
+    },
+    {
+      "epoch": 17.68421052631579,
+      "grad_norm": 14.32602596282959,
+      "learning_rate": 3.6330275229357803e-06,
+      "loss": 0.2551,
+      "step": 1008
+    },
+    {
+      "epoch": 17.70175438596491,
+      "grad_norm": 30.01763153076172,
+      "learning_rate": 3.6055045871559636e-06,
+      "loss": 0.1269,
+      "step": 1009
+    },
+    {
+      "epoch": 17.719298245614034,
+      "grad_norm": 1.7071233987808228,
+      "learning_rate": 3.577981651376147e-06,
+      "loss": 0.0459,
+      "step": 1010
+    },
+    {
+      "epoch": 17.736842105263158,
+      "grad_norm": 67.83972930908203,
+      "learning_rate": 3.5504587155963307e-06,
+      "loss": 0.2978,
+      "step": 1011
+    },
+    {
+      "epoch": 17.75438596491228,
+      "grad_norm": 1.1684902906417847,
+      "learning_rate": 3.522935779816514e-06,
+      "loss": 0.0464,
+      "step": 1012
+    },
+    {
+      "epoch": 17.771929824561404,
+      "grad_norm": 6.896801948547363,
+      "learning_rate": 3.4954128440366977e-06,
+      "loss": 0.0724,
+      "step": 1013
+    },
+    {
+      "epoch": 17.789473684210527,
+      "grad_norm": 49.41828918457031,
+      "learning_rate": 3.467889908256881e-06,
+      "loss": 0.2514,
+      "step": 1014
+    },
+    {
+      "epoch": 17.80701754385965,
+      "grad_norm": 1.8706743717193604,
+      "learning_rate": 3.4403669724770644e-06,
+      "loss": 0.0441,
+      "step": 1015
+    },
+    {
+      "epoch": 17.82456140350877,
+      "grad_norm": 4.06084680557251,
+      "learning_rate": 3.412844036697248e-06,
+      "loss": 0.3851,
+      "step": 1016
+    },
+    {
+      "epoch": 17.842105263157894,
+      "grad_norm": 4.851496696472168,
+      "learning_rate": 3.3853211009174314e-06,
+      "loss": 0.1896,
+      "step": 1017
+    },
+    {
+      "epoch": 17.859649122807017,
+      "grad_norm": 22.698795318603516,
+      "learning_rate": 3.3577981651376148e-06,
+      "loss": 0.1371,
+      "step": 1018
+    },
+    {
+      "epoch": 17.87719298245614,
+      "grad_norm": 1.251635193824768,
+      "learning_rate": 3.3302752293577985e-06,
+      "loss": 0.0426,
+      "step": 1019
+    },
+    {
+      "epoch": 17.894736842105264,
+      "grad_norm": 4.991325378417969,
+      "learning_rate": 3.302752293577982e-06,
+      "loss": 0.2134,
+      "step": 1020
+    },
+    {
+      "epoch": 17.912280701754387,
+      "grad_norm": 39.215309143066406,
+      "learning_rate": 3.2752293577981656e-06,
+      "loss": 0.3352,
+      "step": 1021
+    },
+    {
+      "epoch": 17.92982456140351,
+      "grad_norm": 25.196969985961914,
+      "learning_rate": 3.247706422018349e-06,
+      "loss": 0.4699,
+      "step": 1022
+    },
+    {
+      "epoch": 17.94736842105263,
+      "grad_norm": 15.807039260864258,
+      "learning_rate": 3.2201834862385322e-06,
+      "loss": 0.3251,
+      "step": 1023
+    },
+    {
+      "epoch": 17.964912280701753,
+      "grad_norm": 0.42076656222343445,
+      "learning_rate": 3.192660550458716e-06,
+      "loss": 0.041,
+      "step": 1024
+    },
+    {
+      "epoch": 17.982456140350877,
+      "grad_norm": 16.122833251953125,
+      "learning_rate": 3.1651376146788993e-06,
+      "loss": 0.297,
+      "step": 1025
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 0.5662638545036316,
+      "learning_rate": 3.1376146788990826e-06,
+      "loss": 0.0226,
+      "step": 1026
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.7126436781609196,
+      "eval_loss": 1.0820069313049316,
+      "eval_roc_auc": 0.8867955974179313,
+      "eval_runtime": 8.3763,
+      "eval_samples_per_second": 31.159,
+      "eval_steps_per_second": 2.03,
+      "step": 1026
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1140,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.46617895235264e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}