{
  "best_metric": 1.0249524116516113,
  "best_model_checkpoint": "AST-vocal-disorder-classification/checkpoint-570",
  "epoch": 10.0,
  "eval_steps": 1,
  "global_step": 570,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.017543859649122806,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 2.988,
      "step": 1
    },
    {
      "epoch": 0.03508771929824561,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 2.6954,
      "step": 2
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 2.2632,
      "step": 3
    },
    {
      "epoch": 0.07017543859649122,
      "grad_norm": 165.02626037597656,
      "learning_rate": 2e-08,
      "loss": 3.195,
      "step": 4
    },
    {
      "epoch": 0.08771929824561403,
      "grad_norm": 168.1376495361328,
      "learning_rate": 4e-08,
      "loss": 2.8908,
      "step": 5
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": Infinity,
      "learning_rate": 4e-08,
      "loss": 2.7552,
      "step": 6
    },
    {
      "epoch": 0.12280701754385964,
      "grad_norm": 158.0127716064453,
      "learning_rate": 6e-08,
      "loss": 3.1245,
      "step": 7
    },
    {
      "epoch": 0.14035087719298245,
      "grad_norm": 150.22560119628906,
      "learning_rate": 8e-08,
      "loss": 2.8198,
      "step": 8
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 133.4413299560547,
      "learning_rate": 1e-07,
      "loss": 2.4031,
      "step": 9
    },
    {
      "epoch": 0.17543859649122806,
      "grad_norm": 99.31004333496094,
      "learning_rate": 1.2e-07,
      "loss": 2.457,
      "step": 10
    },
    {
      "epoch": 0.19298245614035087,
      "grad_norm": 118.67373657226562,
      "learning_rate": 1.4e-07,
      "loss": 2.5123,
      "step": 11
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 132.25393676757812,
      "learning_rate": 1.6e-07,
      "loss": 2.5434,
      "step": 12
    },
    {
      "epoch": 0.22807017543859648,
      "grad_norm": 94.40389251708984,
      "learning_rate": 1.8e-07,
      "loss": 2.3049,
      "step": 13
    },
    {
      "epoch": 0.24561403508771928,
      "grad_norm": 125.89042663574219,
      "learning_rate": 2e-07,
      "loss": 2.5478,
      "step": 14
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 142.48435974121094,
      "learning_rate": 2.1999999999999998e-07,
      "loss": 2.8247,
      "step": 15
    },
    {
      "epoch": 0.2807017543859649,
      "grad_norm": 126.35132598876953,
      "learning_rate": 2.4e-07,
      "loss": 2.4798,
      "step": 16
    },
    {
      "epoch": 0.2982456140350877,
      "grad_norm": 151.13067626953125,
      "learning_rate": 2.6e-07,
      "loss": 2.7472,
      "step": 17
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 91.00186920166016,
      "learning_rate": 2.8e-07,
      "loss": 2.2366,
      "step": 18
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 81.35576629638672,
      "learning_rate": 3e-07,
      "loss": 2.3093,
      "step": 19
    },
    {
      "epoch": 0.3508771929824561,
      "grad_norm": 164.45582580566406,
      "learning_rate": 3.2e-07,
      "loss": 2.9604,
      "step": 20
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 118.8172378540039,
      "learning_rate": 3.4000000000000003e-07,
      "loss": 2.3687,
      "step": 21
    },
    {
      "epoch": 0.38596491228070173,
      "grad_norm": 136.17398071289062,
      "learning_rate": 3.6e-07,
      "loss": 2.3948,
      "step": 22
    },
    {
      "epoch": 0.40350877192982454,
      "grad_norm": 118.8042221069336,
      "learning_rate": 3.7999999999999996e-07,
      "loss": 2.2426,
      "step": 23
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 107.0354995727539,
      "learning_rate": 4e-07,
      "loss": 2.3865,
      "step": 24
    },
    {
      "epoch": 0.43859649122807015,
      "grad_norm": 77.19119262695312,
      "learning_rate": 4.1999999999999995e-07,
      "loss": 1.9904,
      "step": 25
    },
    {
      "epoch": 0.45614035087719296,
      "grad_norm": 140.62896728515625,
      "learning_rate": 4.3999999999999997e-07,
      "loss": 2.7652,
      "step": 26
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 75.05490112304688,
      "learning_rate": 4.6e-07,
      "loss": 2.075,
      "step": 27
    },
    {
      "epoch": 0.49122807017543857,
      "grad_norm": 117.8460922241211,
      "learning_rate": 4.8e-07,
      "loss": 2.5994,
      "step": 28
    },
    {
      "epoch": 0.5087719298245614,
      "grad_norm": 113.06562042236328,
      "learning_rate": 5e-07,
      "loss": 1.8539,
      "step": 29
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 110.89696502685547,
      "learning_rate": 5.2e-07,
      "loss": 2.1113,
      "step": 30
    },
    {
      "epoch": 0.543859649122807,
      "grad_norm": 98.19458770751953,
      "learning_rate": 5.4e-07,
      "loss": 2.1587,
      "step": 31
    },
    {
      "epoch": 0.5614035087719298,
      "grad_norm": 66.95838928222656,
      "learning_rate": 5.6e-07,
      "loss": 1.8982,
      "step": 32
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 52.46982192993164,
      "learning_rate": 5.8e-07,
      "loss": 1.6926,
      "step": 33
    },
    {
      "epoch": 0.5964912280701754,
      "grad_norm": 79.77543640136719,
      "learning_rate": 6e-07,
      "loss": 1.9471,
      "step": 34
    },
    {
      "epoch": 0.6140350877192983,
      "grad_norm": 116.57611846923828,
      "learning_rate": 6.2e-07,
      "loss": 1.799,
      "step": 35
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 91.27916717529297,
      "learning_rate": 6.4e-07,
      "loss": 2.2582,
      "step": 36
    },
    {
      "epoch": 0.6491228070175439,
      "grad_norm": 79.36409759521484,
      "learning_rate": 6.6e-07,
      "loss": 2.1673,
      "step": 37
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 65.35518646240234,
      "learning_rate": 6.800000000000001e-07,
      "loss": 1.7037,
      "step": 38
    },
    {
      "epoch": 0.6842105263157895,
      "grad_norm": 42.132991790771484,
      "learning_rate": 7e-07,
      "loss": 1.4278,
      "step": 39
    },
    {
      "epoch": 0.7017543859649122,
      "grad_norm": 51.01911544799805,
      "learning_rate": 7.2e-07,
      "loss": 1.7739,
      "step": 40
    },
    {
      "epoch": 0.7192982456140351,
      "grad_norm": 65.60211181640625,
      "learning_rate": 7.4e-07,
      "loss": 1.7796,
      "step": 41
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 48.78630828857422,
      "learning_rate": 7.599999999999999e-07,
      "loss": 1.5501,
      "step": 42
    },
    {
      "epoch": 0.7543859649122807,
      "grad_norm": 29.828662872314453,
      "learning_rate": 7.799999999999999e-07,
      "loss": 1.56,
      "step": 43
    },
    {
      "epoch": 0.7719298245614035,
      "grad_norm": 50.522647857666016,
      "learning_rate": 8e-07,
      "loss": 1.4566,
      "step": 44
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 52.197025299072266,
      "learning_rate": 8.199999999999999e-07,
      "loss": 1.5521,
      "step": 45
    },
    {
      "epoch": 0.8070175438596491,
      "grad_norm": 41.581092834472656,
      "learning_rate": 8.399999999999999e-07,
      "loss": 1.6864,
      "step": 46
    },
    {
      "epoch": 0.8245614035087719,
      "grad_norm": 61.20100021362305,
      "learning_rate": 8.599999999999999e-07,
      "loss": 1.3111,
      "step": 47
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 39.471736907958984,
      "learning_rate": 8.799999999999999e-07,
      "loss": 1.6241,
      "step": 48
    },
    {
      "epoch": 0.8596491228070176,
      "grad_norm": 39.578338623046875,
      "learning_rate": 9e-07,
      "loss": 1.4799,
      "step": 49
    },
    {
      "epoch": 0.8771929824561403,
      "grad_norm": 57.32265090942383,
      "learning_rate": 9.2e-07,
      "loss": 1.8162,
      "step": 50
    },
    {
      "epoch": 0.8947368421052632,
      "grad_norm": 37.478973388671875,
      "learning_rate": 9.399999999999999e-07,
      "loss": 1.5859,
      "step": 51
    },
    {
      "epoch": 0.9122807017543859,
      "grad_norm": 45.59309387207031,
      "learning_rate": 9.6e-07,
      "loss": 1.3365,
      "step": 52
    },
    {
      "epoch": 0.9298245614035088,
      "grad_norm": 47.914791107177734,
      "learning_rate": 9.8e-07,
      "loss": 1.7568,
      "step": 53
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 31.730653762817383,
      "learning_rate": 1e-06,
      "loss": 1.6654,
      "step": 54
    },
    {
      "epoch": 0.9649122807017544,
      "grad_norm": 47.1168327331543,
      "learning_rate": 9.98076923076923e-07,
      "loss": 1.2154,
      "step": 55
    },
    {
      "epoch": 0.9824561403508771,
      "grad_norm": 34.07067108154297,
      "learning_rate": 9.961538461538461e-07,
      "loss": 1.296,
      "step": 56
    },
    {
      "epoch": 1.0,
      "grad_norm": 32.9833869934082,
      "learning_rate": 9.942307692307691e-07,
      "loss": 1.5427,
      "step": 57
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.3831417624521073,
      "eval_loss": 1.5033824443817139,
      "eval_roc_auc": 0.5521317875556301,
      "eval_runtime": 24.2654,
      "eval_samples_per_second": 10.756,
      "eval_steps_per_second": 0.701,
      "step": 57
    },
    {
      "epoch": 1.0175438596491229,
      "grad_norm": 21.84532356262207,
      "learning_rate": 9.923076923076923e-07,
      "loss": 1.3469,
      "step": 58
    },
    {
      "epoch": 1.0350877192982457,
      "grad_norm": 54.183753967285156,
      "learning_rate": 9.903846153846153e-07,
      "loss": 1.5014,
      "step": 59
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 45.070030212402344,
      "learning_rate": 9.884615384615385e-07,
      "loss": 1.6736,
      "step": 60
    },
    {
      "epoch": 1.0701754385964912,
      "grad_norm": 54.34022903442383,
      "learning_rate": 9.865384615384615e-07,
      "loss": 1.7181,
      "step": 61
    },
    {
      "epoch": 1.087719298245614,
      "grad_norm": 61.96958923339844,
      "learning_rate": 9.846153846153847e-07,
      "loss": 1.7463,
      "step": 62
    },
    {
      "epoch": 1.1052631578947367,
      "grad_norm": 38.20405960083008,
      "learning_rate": 9.826923076923076e-07,
      "loss": 1.516,
      "step": 63
    },
    {
      "epoch": 1.1228070175438596,
      "grad_norm": 31.333251953125,
      "learning_rate": 9.807692307692306e-07,
      "loss": 1.6652,
      "step": 64
    },
    {
      "epoch": 1.1403508771929824,
      "grad_norm": 33.82837677001953,
      "learning_rate": 9.788461538461538e-07,
      "loss": 1.3185,
      "step": 65
    },
    {
      "epoch": 1.1578947368421053,
      "grad_norm": 44.01963424682617,
      "learning_rate": 9.769230769230768e-07,
      "loss": 1.2599,
      "step": 66
    },
    {
      "epoch": 1.1754385964912282,
      "grad_norm": 41.35686492919922,
      "learning_rate": 9.75e-07,
      "loss": 1.4831,
      "step": 67
    },
    {
      "epoch": 1.1929824561403508,
      "grad_norm": 37.442283630371094,
      "learning_rate": 9.73076923076923e-07,
      "loss": 1.6914,
      "step": 68
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 45.21449661254883,
      "learning_rate": 9.711538461538462e-07,
      "loss": 1.0162,
      "step": 69
    },
    {
      "epoch": 1.2280701754385965,
      "grad_norm": 37.826507568359375,
      "learning_rate": 9.692307692307691e-07,
      "loss": 1.5321,
      "step": 70
    },
    {
      "epoch": 1.2456140350877192,
      "grad_norm": 81.76914978027344,
      "learning_rate": 9.673076923076923e-07,
      "loss": 1.7058,
      "step": 71
    },
    {
      "epoch": 1.263157894736842,
      "grad_norm": 29.697542190551758,
      "learning_rate": 9.653846153846153e-07,
      "loss": 1.1953,
      "step": 72
    },
    {
      "epoch": 1.280701754385965,
      "grad_norm": 25.966896057128906,
      "learning_rate": 9.634615384615385e-07,
      "loss": 1.2336,
      "step": 73
    },
    {
      "epoch": 1.2982456140350878,
      "grad_norm": 27.649991989135742,
      "learning_rate": 9.615384615384615e-07,
      "loss": 1.4547,
      "step": 74
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 85.1219482421875,
      "learning_rate": 9.596153846153847e-07,
      "loss": 1.8594,
      "step": 75
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 45.845516204833984,
      "learning_rate": 9.576923076923077e-07,
      "loss": 1.4723,
      "step": 76
    },
    {
      "epoch": 1.3508771929824561,
      "grad_norm": 33.39055633544922,
      "learning_rate": 9.557692307692309e-07,
      "loss": 1.3105,
      "step": 77
    },
    {
      "epoch": 1.368421052631579,
      "grad_norm": 32.90497970581055,
      "learning_rate": 9.538461538461538e-07,
      "loss": 1.551,
      "step": 78
    },
    {
      "epoch": 1.3859649122807016,
      "grad_norm": 18.763742446899414,
      "learning_rate": 9.519230769230768e-07,
      "loss": 1.1766,
      "step": 79
    },
    {
      "epoch": 1.4035087719298245,
      "grad_norm": 38.28258514404297,
      "learning_rate": 9.499999999999999e-07,
      "loss": 1.3663,
      "step": 80
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 44.95620346069336,
      "learning_rate": 9.48076923076923e-07,
      "loss": 1.4312,
      "step": 81
    },
    {
      "epoch": 1.4385964912280702,
      "grad_norm": 46.334720611572266,
      "learning_rate": 9.461538461538461e-07,
      "loss": 1.5499,
      "step": 82
    },
    {
      "epoch": 1.456140350877193,
      "grad_norm": 24.238452911376953,
      "learning_rate": 9.442307692307692e-07,
      "loss": 1.4905,
      "step": 83
    },
    {
      "epoch": 1.4736842105263157,
      "grad_norm": 33.23431396484375,
      "learning_rate": 9.423076923076923e-07,
      "loss": 1.3647,
      "step": 84
    },
    {
      "epoch": 1.4912280701754386,
      "grad_norm": 46.839908599853516,
      "learning_rate": 9.403846153846153e-07,
      "loss": 1.7358,
      "step": 85
    },
    {
      "epoch": 1.5087719298245614,
      "grad_norm": 27.96128273010254,
      "learning_rate": 9.384615384615384e-07,
      "loss": 1.4824,
      "step": 86
    },
    {
      "epoch": 1.526315789473684,
      "grad_norm": 47.32505416870117,
      "learning_rate": 9.365384615384615e-07,
      "loss": 1.5127,
      "step": 87
    },
    {
      "epoch": 1.543859649122807,
      "grad_norm": 31.273122787475586,
      "learning_rate": 9.346153846153846e-07,
      "loss": 1.1443,
      "step": 88
    },
    {
      "epoch": 1.5614035087719298,
      "grad_norm": 47.530086517333984,
      "learning_rate": 9.326923076923077e-07,
      "loss": 1.4594,
      "step": 89
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 52.47103500366211,
      "learning_rate": 9.307692307692308e-07,
      "loss": 1.3066,
      "step": 90
    },
    {
      "epoch": 1.5964912280701755,
      "grad_norm": 48.503936767578125,
      "learning_rate": 9.288461538461539e-07,
      "loss": 1.2926,
      "step": 91
    },
    {
      "epoch": 1.6140350877192984,
      "grad_norm": 37.38385009765625,
      "learning_rate": 9.26923076923077e-07,
      "loss": 1.2452,
      "step": 92
    },
    {
      "epoch": 1.631578947368421,
      "grad_norm": 69.00371551513672,
      "learning_rate": 9.25e-07,
      "loss": 1.7103,
      "step": 93
    },
    {
      "epoch": 1.6491228070175439,
      "grad_norm": 20.759923934936523,
      "learning_rate": 9.230769230769231e-07,
      "loss": 1.4222,
      "step": 94
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 17.494487762451172,
      "learning_rate": 9.211538461538461e-07,
      "loss": 1.3884,
      "step": 95
    },
    {
      "epoch": 1.6842105263157894,
      "grad_norm": 72.35301971435547,
      "learning_rate": 9.192307692307692e-07,
      "loss": 1.7956,
      "step": 96
    },
    {
      "epoch": 1.7017543859649122,
      "grad_norm": 58.388648986816406,
      "learning_rate": 9.173076923076922e-07,
      "loss": 1.3706,
      "step": 97
    },
    {
      "epoch": 1.719298245614035,
      "grad_norm": 32.08078384399414,
      "learning_rate": 9.153846153846153e-07,
      "loss": 1.3232,
      "step": 98
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 43.424861907958984,
      "learning_rate": 9.134615384615383e-07,
      "loss": 1.0919,
      "step": 99
    },
    {
      "epoch": 1.7543859649122808,
      "grad_norm": 43.55315017700195,
      "learning_rate": 9.115384615384614e-07,
      "loss": 1.3153,
      "step": 100
    },
    {
      "epoch": 1.7719298245614035,
      "grad_norm": 12.423308372497559,
      "learning_rate": 9.096153846153845e-07,
      "loss": 1.2841,
      "step": 101
    },
    {
      "epoch": 1.7894736842105263,
      "grad_norm": 32.44642639160156,
      "learning_rate": 9.076923076923076e-07,
      "loss": 1.2104,
      "step": 102
    },
    {
      "epoch": 1.807017543859649,
      "grad_norm": 74.81303405761719,
      "learning_rate": 9.057692307692307e-07,
      "loss": 0.9296,
      "step": 103
    },
    {
      "epoch": 1.8245614035087718,
      "grad_norm": 62.9661979675293,
      "learning_rate": 9.038461538461538e-07,
      "loss": 1.1964,
      "step": 104
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 52.237205505371094,
      "learning_rate": 9.019230769230769e-07,
      "loss": 1.3407,
      "step": 105
    },
    {
      "epoch": 1.8596491228070176,
      "grad_norm": 26.808528900146484,
      "learning_rate": 9e-07,
      "loss": 1.1248,
      "step": 106
    },
    {
      "epoch": 1.8771929824561404,
      "grad_norm": 28.834348678588867,
      "learning_rate": 8.98076923076923e-07,
      "loss": 1.1314,
      "step": 107
    },
    {
      "epoch": 1.8947368421052633,
      "grad_norm": 58.954833984375,
      "learning_rate": 8.961538461538461e-07,
      "loss": 1.1691,
      "step": 108
    },
    {
      "epoch": 1.912280701754386,
      "grad_norm": 24.54899024963379,
      "learning_rate": 8.942307692307692e-07,
      "loss": 1.3596,
      "step": 109
    },
    {
      "epoch": 1.9298245614035088,
      "grad_norm": 38.63195037841797,
      "learning_rate": 8.923076923076923e-07,
      "loss": 1.1972,
      "step": 110
    },
    {
      "epoch": 1.9473684210526314,
      "grad_norm": 40.493141174316406,
      "learning_rate": 8.903846153846153e-07,
      "loss": 1.4167,
      "step": 111
    },
    {
      "epoch": 1.9649122807017543,
      "grad_norm": 25.149085998535156,
      "learning_rate": 8.884615384615384e-07,
      "loss": 1.1808,
      "step": 112
    },
    {
      "epoch": 1.9824561403508771,
      "grad_norm": 26.5919132232666,
      "learning_rate": 8.865384615384615e-07,
      "loss": 1.2026,
      "step": 113
    },
    {
      "epoch": 2.0,
      "grad_norm": 27.22132682800293,
      "learning_rate": 8.846153846153846e-07,
      "loss": 1.1829,
      "step": 114
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.44061302681992337,
      "eval_loss": 1.3389430046081543,
      "eval_roc_auc": 0.7114645372622986,
      "eval_runtime": 22.9802,
      "eval_samples_per_second": 11.358,
      "eval_steps_per_second": 0.74,
      "step": 114
    },
    {
      "epoch": 2.017543859649123,
      "grad_norm": 45.17961883544922,
      "learning_rate": 8.826923076923076e-07,
      "loss": 1.2122,
      "step": 115
    },
    {
      "epoch": 2.0350877192982457,
      "grad_norm": 39.010257720947266,
      "learning_rate": 8.807692307692307e-07,
      "loss": 1.143,
      "step": 116
    },
    {
      "epoch": 2.0526315789473686,
      "grad_norm": 45.62030029296875,
      "learning_rate": 8.788461538461538e-07,
      "loss": 1.4336,
      "step": 117
    },
    {
      "epoch": 2.0701754385964914,
      "grad_norm": 39.72404098510742,
      "learning_rate": 8.769230769230769e-07,
      "loss": 1.2491,
      "step": 118
    },
    {
      "epoch": 2.087719298245614,
      "grad_norm": 60.91448211669922,
      "learning_rate": 8.75e-07,
      "loss": 1.3879,
      "step": 119
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 22.340192794799805,
      "learning_rate": 8.730769230769231e-07,
      "loss": 1.1815,
      "step": 120
    },
    {
      "epoch": 2.1228070175438596,
      "grad_norm": 51.66433334350586,
      "learning_rate": 8.711538461538462e-07,
      "loss": 1.3663,
      "step": 121
    },
    {
      "epoch": 2.1403508771929824,
      "grad_norm": 34.69876480102539,
      "learning_rate": 8.692307692307692e-07,
      "loss": 0.8601,
      "step": 122
    },
    {
      "epoch": 2.1578947368421053,
      "grad_norm": 31.769758224487305,
      "learning_rate": 8.673076923076923e-07,
      "loss": 1.067,
      "step": 123
    },
    {
      "epoch": 2.175438596491228,
      "grad_norm": 15.30595588684082,
      "learning_rate": 8.653846153846154e-07,
      "loss": 1.2825,
      "step": 124
    },
    {
      "epoch": 2.192982456140351,
      "grad_norm": 50.3326416015625,
      "learning_rate": 8.634615384615385e-07,
      "loss": 1.2839,
      "step": 125
    },
    {
      "epoch": 2.2105263157894735,
      "grad_norm": 39.54274368286133,
      "learning_rate": 8.615384615384616e-07,
      "loss": 1.2719,
      "step": 126
    },
    {
      "epoch": 2.2280701754385963,
      "grad_norm": 40.03786087036133,
      "learning_rate": 8.596153846153846e-07,
      "loss": 1.2902,
      "step": 127
    },
    {
      "epoch": 2.245614035087719,
      "grad_norm": 25.278501510620117,
      "learning_rate": 8.576923076923076e-07,
      "loss": 0.9576,
      "step": 128
    },
    {
      "epoch": 2.263157894736842,
      "grad_norm": 74.27449035644531,
      "learning_rate": 8.557692307692306e-07,
      "loss": 1.6626,
      "step": 129
    },
    {
      "epoch": 2.280701754385965,
      "grad_norm": 49.429100036621094,
      "learning_rate": 8.538461538461537e-07,
      "loss": 1.1279,
      "step": 130
    },
    {
      "epoch": 2.2982456140350878,
      "grad_norm": 66.35237121582031,
      "learning_rate": 8.519230769230768e-07,
      "loss": 1.3382,
      "step": 131
    },
    {
      "epoch": 2.3157894736842106,
      "grad_norm": 54.39750671386719,
      "learning_rate": 8.499999999999999e-07,
      "loss": 1.1413,
      "step": 132
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 39.203617095947266,
      "learning_rate": 8.48076923076923e-07,
      "loss": 1.646,
      "step": 133
    },
    {
      "epoch": 2.3508771929824563,
      "grad_norm": 40.577457427978516,
      "learning_rate": 8.461538461538461e-07,
      "loss": 1.5547,
      "step": 134
    },
    {
      "epoch": 2.3684210526315788,
      "grad_norm": 61.2403564453125,
      "learning_rate": 8.442307692307692e-07,
      "loss": 1.4927,
      "step": 135
    },
    {
      "epoch": 2.3859649122807016,
      "grad_norm": 27.469097137451172,
      "learning_rate": 8.423076923076923e-07,
      "loss": 1.2704,
      "step": 136
    },
    {
      "epoch": 2.4035087719298245,
      "grad_norm": 41.40801239013672,
      "learning_rate": 8.403846153846153e-07,
      "loss": 1.12,
      "step": 137
    },
    {
      "epoch": 2.4210526315789473,
      "grad_norm": 31.15001106262207,
      "learning_rate": 8.384615384615384e-07,
      "loss": 1.1188,
      "step": 138
    },
    {
      "epoch": 2.43859649122807,
      "grad_norm": 48.32929611206055,
      "learning_rate": 8.365384615384615e-07,
      "loss": 0.9936,
      "step": 139
    },
    {
      "epoch": 2.456140350877193,
      "grad_norm": 43.33560562133789,
      "learning_rate": 8.346153846153846e-07,
      "loss": 1.5216,
      "step": 140
    },
    {
      "epoch": 2.473684210526316,
      "grad_norm": 34.28059387207031,
      "learning_rate": 8.326923076923077e-07,
      "loss": 1.328,
      "step": 141
    },
    {
      "epoch": 2.4912280701754383,
      "grad_norm": 55.07752227783203,
      "learning_rate": 8.307692307692308e-07,
      "loss": 1.3941,
      "step": 142
    },
    {
      "epoch": 2.5087719298245617,
      "grad_norm": 33.26469421386719,
      "learning_rate": 8.288461538461539e-07,
      "loss": 1.0795,
      "step": 143
    },
    {
      "epoch": 2.526315789473684,
      "grad_norm": 41.03183364868164,
      "learning_rate": 8.269230769230768e-07,
      "loss": 1.1216,
      "step": 144
    },
    {
      "epoch": 2.543859649122807,
      "grad_norm": 27.419574737548828,
      "learning_rate": 8.249999999999999e-07,
      "loss": 1.2125,
      "step": 145
    },
    {
      "epoch": 2.56140350877193,
      "grad_norm": 34.500667572021484,
      "learning_rate": 8.23076923076923e-07,
      "loss": 0.9645,
      "step": 146
    },
    {
      "epoch": 2.5789473684210527,
      "grad_norm": 33.50322341918945,
      "learning_rate": 8.211538461538461e-07,
      "loss": 1.1995,
      "step": 147
    },
    {
      "epoch": 2.5964912280701755,
      "grad_norm": 73.93108367919922,
      "learning_rate": 8.192307692307692e-07,
      "loss": 1.7082,
      "step": 148
    },
    {
      "epoch": 2.6140350877192984,
      "grad_norm": 14.201051712036133,
      "learning_rate": 8.173076923076923e-07,
      "loss": 1.1467,
      "step": 149
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 23.20082664489746,
      "learning_rate": 8.153846153846154e-07,
      "loss": 1.2548,
      "step": 150
    },
    {
      "epoch": 2.6491228070175437,
      "grad_norm": 37.417213439941406,
      "learning_rate": 8.134615384615385e-07,
      "loss": 0.9575,
      "step": 151
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 69.82878112792969,
      "learning_rate": 8.115384615384615e-07,
      "loss": 1.3022,
      "step": 152
    },
    {
      "epoch": 2.6842105263157894,
      "grad_norm": 45.68886184692383,
      "learning_rate": 8.096153846153846e-07,
      "loss": 1.1804,
      "step": 153
    },
    {
      "epoch": 2.7017543859649122,
      "grad_norm": 51.7285270690918,
      "learning_rate": 8.076923076923077e-07,
      "loss": 1.3008,
      "step": 154
    },
    {
      "epoch": 2.719298245614035,
      "grad_norm": 51.00165939331055,
      "learning_rate": 8.057692307692308e-07,
      "loss": 1.2146,
      "step": 155
    },
    {
      "epoch": 2.736842105263158,
      "grad_norm": 56.5025634765625,
      "learning_rate": 8.038461538461538e-07,
      "loss": 1.7964,
      "step": 156
    },
    {
      "epoch": 2.754385964912281,
      "grad_norm": 29.800735473632812,
      "learning_rate": 8.019230769230769e-07,
      "loss": 1.0893,
      "step": 157
    },
    {
      "epoch": 2.7719298245614032,
      "grad_norm": 35.997886657714844,
      "learning_rate": 8e-07,
      "loss": 1.2403,
      "step": 158
    },
    {
      "epoch": 2.7894736842105265,
      "grad_norm": 65.59085083007812,
      "learning_rate": 7.98076923076923e-07,
      "loss": 1.0296,
      "step": 159
    },
    {
      "epoch": 2.807017543859649,
      "grad_norm": 57.628726959228516,
      "learning_rate": 7.96153846153846e-07,
      "loss": 1.3895,
      "step": 160
    },
    {
      "epoch": 2.824561403508772,
      "grad_norm": 51.50771713256836,
      "learning_rate": 7.942307692307691e-07,
      "loss": 0.7671,
      "step": 161
    },
    {
      "epoch": 2.8421052631578947,
      "grad_norm": 17.941665649414062,
      "learning_rate": 7.923076923076922e-07,
      "loss": 1.0704,
      "step": 162
    },
    {
      "epoch": 2.8596491228070176,
      "grad_norm": 52.35181427001953,
      "learning_rate": 7.903846153846153e-07,
      "loss": 1.13,
      "step": 163
    },
    {
      "epoch": 2.8771929824561404,
      "grad_norm": 42.02775955200195,
      "learning_rate": 7.884615384615384e-07,
      "loss": 0.8805,
      "step": 164
    },
    {
      "epoch": 2.8947368421052633,
      "grad_norm": 58.180484771728516,
      "learning_rate": 7.865384615384615e-07,
      "loss": 1.1904,
      "step": 165
    },
    {
      "epoch": 2.912280701754386,
      "grad_norm": 30.56048583984375,
      "learning_rate": 7.846153846153846e-07,
      "loss": 1.1506,
      "step": 166
    },
    {
      "epoch": 2.9298245614035086,
      "grad_norm": 39.470184326171875,
      "learning_rate": 7.826923076923076e-07,
      "loss": 0.741,
      "step": 167
    },
    {
      "epoch": 2.9473684210526314,
      "grad_norm": 72.42061614990234,
      "learning_rate": 7.807692307692307e-07,
      "loss": 1.3581,
      "step": 168
    },
    {
      "epoch": 2.9649122807017543,
      "grad_norm": 42.25473403930664,
      "learning_rate": 7.788461538461538e-07,
      "loss": 1.3318,
      "step": 169
    },
    {
      "epoch": 2.982456140350877,
      "grad_norm": 28.385595321655273,
      "learning_rate": 7.769230769230769e-07,
      "loss": 1.0435,
      "step": 170
    },
    {
      "epoch": 3.0,
      "grad_norm": 44.694984436035156,
      "learning_rate": 7.75e-07,
      "loss": 1.2793,
      "step": 171
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5019157088122606,
      "eval_loss": 1.2615339756011963,
      "eval_roc_auc": 0.7588809381350943,
      "eval_runtime": 24.6917,
      "eval_samples_per_second": 10.57,
      "eval_steps_per_second": 0.688,
      "step": 171
    },
    {
      "epoch": 3.017543859649123,
      "grad_norm": 42.023494720458984,
      "learning_rate": 7.730769230769231e-07,
      "loss": 1.1126,
      "step": 172
    },
    {
      "epoch": 3.0350877192982457,
      "grad_norm": 79.69966125488281,
      "learning_rate": 7.711538461538462e-07,
      "loss": 1.2264,
      "step": 173
    },
    {
      "epoch": 3.0526315789473686,
      "grad_norm": 33.83009719848633,
      "learning_rate": 7.692307692307693e-07,
      "loss": 1.0632,
      "step": 174
    },
    {
      "epoch": 3.0701754385964914,
      "grad_norm": 42.74213409423828,
      "learning_rate": 7.673076923076923e-07,
      "loss": 0.7554,
      "step": 175
    },
    {
      "epoch": 3.087719298245614,
      "grad_norm": 29.90761375427246,
      "learning_rate": 7.653846153846153e-07,
      "loss": 1.1182,
      "step": 176
    },
    {
      "epoch": 3.1052631578947367,
      "grad_norm": 34.67997741699219,
      "learning_rate": 7.634615384615384e-07,
      "loss": 1.2303,
      "step": 177
    },
    {
      "epoch": 3.1228070175438596,
      "grad_norm": 28.055559158325195,
      "learning_rate": 7.615384615384615e-07,
      "loss": 0.9286,
      "step": 178
    },
    {
      "epoch": 3.1403508771929824,
      "grad_norm": 48.56884765625,
      "learning_rate": 7.596153846153846e-07,
      "loss": 0.9129,
      "step": 179
    },
    {
      "epoch": 3.1578947368421053,
      "grad_norm": 23.18881607055664,
      "learning_rate": 7.576923076923077e-07,
      "loss": 1.1422,
      "step": 180
    },
    {
      "epoch": 3.175438596491228,
      "grad_norm": 19.716148376464844,
      "learning_rate": 7.557692307692308e-07,
      "loss": 1.1811,
      "step": 181
    },
    {
      "epoch": 3.192982456140351,
      "grad_norm": 23.179607391357422,
      "learning_rate": 7.538461538461538e-07,
      "loss": 0.9284,
      "step": 182
    },
    {
      "epoch": 3.2105263157894735,
      "grad_norm": 21.30498504638672,
      "learning_rate": 7.519230769230769e-07,
      "loss": 0.909,
      "step": 183
    },
    {
      "epoch": 3.2280701754385963,
      "grad_norm": 39.3275260925293,
      "learning_rate": 7.5e-07,
      "loss": 1.1318,
      "step": 184
    },
    {
      "epoch": 3.245614035087719,
      "grad_norm": 52.91360855102539,
      "learning_rate": 7.48076923076923e-07,
      "loss": 1.2176,
      "step": 185
    },
    {
      "epoch": 3.263157894736842,
      "grad_norm": 40.860450744628906,
      "learning_rate": 7.461538461538461e-07,
      "loss": 1.2245,
      "step": 186
    },
    {
      "epoch": 3.280701754385965,
      "grad_norm": 28.36117935180664,
      "learning_rate": 7.442307692307692e-07,
      "loss": 1.0225,
      "step": 187
    },
    {
      "epoch": 3.2982456140350878,
      "grad_norm": 40.48318099975586,
      "learning_rate": 7.423076923076923e-07,
      "loss": 1.289,
      "step": 188
    },
    {
      "epoch": 3.3157894736842106,
      "grad_norm": 27.249305725097656,
      "learning_rate": 7.403846153846153e-07,
      "loss": 1.2829,
      "step": 189
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 23.647472381591797,
      "learning_rate": 7.384615384615384e-07,
      "loss": 1.223,
      "step": 190
    },
    {
      "epoch": 3.3508771929824563,
      "grad_norm": 21.608871459960938,
      "learning_rate": 7.365384615384615e-07,
      "loss": 1.0349,
      "step": 191
    },
    {
      "epoch": 3.3684210526315788,
      "grad_norm": 49.232421875,
      "learning_rate": 7.346153846153846e-07,
      "loss": 1.1577,
      "step": 192
    },
    {
      "epoch": 3.3859649122807016,
      "grad_norm": 35.15854263305664,
      "learning_rate": 7.326923076923076e-07,
      "loss": 0.8111,
      "step": 193
    },
    {
      "epoch": 3.4035087719298245,
      "grad_norm": 33.47184371948242,
      "learning_rate": 7.307692307692307e-07,
      "loss": 1.4127,
      "step": 194
    },
    {
      "epoch": 3.4210526315789473,
      "grad_norm": 38.76816940307617,
      "learning_rate": 7.288461538461538e-07,
      "loss": 1.2329,
      "step": 195
    },
    {
      "epoch": 3.43859649122807,
      "grad_norm": 64.3719253540039,
      "learning_rate": 7.269230769230769e-07,
      "loss": 1.4848,
      "step": 196
    },
    {
      "epoch": 3.456140350877193,
      "grad_norm": 26.278711318969727,
      "learning_rate": 7.249999999999999e-07,
      "loss": 1.1606,
      "step": 197
    },
    {
      "epoch": 3.473684210526316,
      "grad_norm": 33.774658203125,
      "learning_rate": 7.23076923076923e-07,
      "loss": 1.1549,
      "step": 198
    },
    {
      "epoch": 3.4912280701754383,
      "grad_norm": 27.449785232543945,
      "learning_rate": 7.211538461538461e-07,
      "loss": 0.7551,
      "step": 199
    },
    {
      "epoch": 3.5087719298245617,
      "grad_norm": 18.257455825805664,
      "learning_rate": 7.192307692307692e-07,
      "loss": 0.9322,
      "step": 200
    },
    {
      "epoch": 3.526315789473684,
      "grad_norm": 54.30673599243164,
      "learning_rate": 7.173076923076923e-07,
      "loss": 0.7284,
      "step": 201
    },
    {
      "epoch": 3.543859649122807,
      "grad_norm": 56.77223587036133,
      "learning_rate": 7.153846153846154e-07,
      "loss": 1.0322,
      "step": 202
    },
    {
      "epoch": 3.56140350877193,
      "grad_norm": 23.362255096435547,
      "learning_rate": 7.134615384615385e-07,
      "loss": 0.9139,
      "step": 203
    },
    {
      "epoch": 3.5789473684210527,
      "grad_norm": 46.12820816040039,
      "learning_rate": 7.115384615384616e-07,
      "loss": 0.9289,
      "step": 204
    },
    {
      "epoch": 3.5964912280701755,
      "grad_norm": 38.374629974365234,
      "learning_rate": 7.096153846153846e-07,
      "loss": 0.9759,
      "step": 205
    },
    {
      "epoch": 3.6140350877192984,
      "grad_norm": 40.8470458984375,
      "learning_rate": 7.076923076923077e-07,
      "loss": 0.9255,
      "step": 206
    },
    {
      "epoch": 3.6315789473684212,
      "grad_norm": 51.18167495727539,
      "learning_rate": 7.057692307692308e-07,
      "loss": 1.6224,
      "step": 207
    },
    {
      "epoch": 3.6491228070175437,
      "grad_norm": 65.96550750732422,
      "learning_rate": 7.038461538461539e-07,
      "loss": 1.4234,
      "step": 208
    },
    {
      "epoch": 3.6666666666666665,
      "grad_norm": 46.82773208618164,
      "learning_rate": 7.019230769230769e-07,
      "loss": 1.4326,
      "step": 209
    },
    {
      "epoch": 3.6842105263157894,
      "grad_norm": 29.189741134643555,
      "learning_rate": 7e-07,
      "loss": 0.9554,
      "step": 210
    },
    {
      "epoch": 3.7017543859649122,
      "grad_norm": 21.677045822143555,
      "learning_rate": 6.980769230769231e-07,
      "loss": 1.2323,
      "step": 211
    },
    {
      "epoch": 3.719298245614035,
      "grad_norm": 55.265682220458984,
      "learning_rate": 6.961538461538461e-07,
      "loss": 1.211,
      "step": 212
    },
    {
      "epoch": 3.736842105263158,
      "grad_norm": 31.52195167541504,
      "learning_rate": 6.942307692307691e-07,
      "loss": 0.9928,
      "step": 213
    },
    {
      "epoch": 3.754385964912281,
      "grad_norm": 70.80876159667969,
      "learning_rate": 6.923076923076922e-07,
      "loss": 1.4359,
      "step": 214
    },
    {
      "epoch": 3.7719298245614032,
      "grad_norm": 40.67531967163086,
      "learning_rate": 6.903846153846153e-07,
      "loss": 1.1827,
      "step": 215
    },
    {
      "epoch": 3.7894736842105265,
      "grad_norm": 33.12160873413086,
      "learning_rate": 6.884615384615384e-07,
      "loss": 1.0607,
      "step": 216
    },
    {
      "epoch": 3.807017543859649,
      "grad_norm": 47.94017791748047,
      "learning_rate": 6.865384615384615e-07,
      "loss": 0.8461,
      "step": 217
    },
    {
      "epoch": 3.824561403508772,
      "grad_norm": 46.735321044921875,
      "learning_rate": 6.846153846153846e-07,
      "loss": 0.8335,
      "step": 218
    },
    {
      "epoch": 3.8421052631578947,
      "grad_norm": 44.989681243896484,
      "learning_rate": 6.826923076923076e-07,
      "loss": 1.0917,
      "step": 219
    },
    {
      "epoch": 3.8596491228070176,
      "grad_norm": 33.04579162597656,
      "learning_rate": 6.807692307692307e-07,
      "loss": 0.8094,
      "step": 220
    },
    {
      "epoch": 3.8771929824561404,
      "grad_norm": 26.442249298095703,
      "learning_rate": 6.788461538461538e-07,
      "loss": 1.3672,
      "step": 221
    },
    {
      "epoch": 3.8947368421052633,
      "grad_norm": 47.97966003417969,
      "learning_rate": 6.769230769230769e-07,
      "loss": 1.1112,
      "step": 222
    },
    {
      "epoch": 3.912280701754386,
      "grad_norm": 49.454437255859375,
      "learning_rate": 6.75e-07,
      "loss": 1.2617,
      "step": 223
    },
    {
      "epoch": 3.9298245614035086,
      "grad_norm": 39.12055587768555,
      "learning_rate": 6.730769230769231e-07,
      "loss": 1.3417,
      "step": 224
    },
    {
      "epoch": 3.9473684210526314,
      "grad_norm": 38.112548828125,
      "learning_rate": 6.711538461538461e-07,
      "loss": 1.1137,
      "step": 225
    },
    {
      "epoch": 3.9649122807017543,
      "grad_norm": 18.497282028198242,
      "learning_rate": 6.692307692307692e-07,
      "loss": 0.636,
      "step": 226
    },
    {
      "epoch": 3.982456140350877,
      "grad_norm": 39.13663101196289,
      "learning_rate": 6.673076923076922e-07,
      "loss": 1.1667,
      "step": 227
    },
    {
      "epoch": 4.0,
      "grad_norm": 27.05844497680664,
      "learning_rate": 6.653846153846153e-07,
      "loss": 0.9226,
      "step": 228
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.5363984674329502,
      "eval_loss": 1.1785928010940552,
      "eval_roc_auc": 0.7825664232388221,
      "eval_runtime": 24.813,
      "eval_samples_per_second": 10.519,
      "eval_steps_per_second": 0.685,
      "step": 228
    },
    {
      "epoch": 4.017543859649122,
      "grad_norm": 40.03508758544922,
      "learning_rate": 6.634615384615384e-07,
      "loss": 0.6566,
      "step": 229
    },
    {
      "epoch": 4.035087719298246,
      "grad_norm": 31.557764053344727,
      "learning_rate": 6.615384615384615e-07,
      "loss": 0.9575,
      "step": 230
    },
    {
      "epoch": 4.052631578947368,
      "grad_norm": 25.278390884399414,
      "learning_rate": 6.596153846153846e-07,
      "loss": 0.7892,
      "step": 231
    },
    {
      "epoch": 4.0701754385964914,
      "grad_norm": 30.69614028930664,
      "learning_rate": 6.576923076923077e-07,
      "loss": 1.0433,
      "step": 232
    },
    {
      "epoch": 4.087719298245614,
      "grad_norm": 58.29319381713867,
      "learning_rate": 6.557692307692308e-07,
      "loss": 1.5216,
      "step": 233
    },
    {
      "epoch": 4.105263157894737,
      "grad_norm": 42.853153228759766,
      "learning_rate": 6.538461538461538e-07,
      "loss": 1.22,
      "step": 234
    },
    {
      "epoch": 4.12280701754386,
      "grad_norm": 44.47975540161133,
      "learning_rate": 6.519230769230769e-07,
      "loss": 1.0883,
      "step": 235
    },
    {
      "epoch": 4.140350877192983,
      "grad_norm": 32.618743896484375,
      "learning_rate": 6.5e-07,
      "loss": 0.8169,
      "step": 236
    },
    {
      "epoch": 4.157894736842105,
      "grad_norm": 45.881961822509766,
      "learning_rate": 6.480769230769231e-07,
      "loss": 1.0063,
      "step": 237
    },
    {
      "epoch": 4.175438596491228,
      "grad_norm": 34.217777252197266,
      "learning_rate": 6.461538461538462e-07,
      "loss": 1.3464,
      "step": 238
    },
    {
      "epoch": 4.192982456140351,
      "grad_norm": 26.423919677734375,
      "learning_rate": 6.442307692307693e-07,
      "loss": 1.001,
      "step": 239
    },
    {
      "epoch": 4.2105263157894735,
      "grad_norm": 29.074453353881836,
      "learning_rate": 6.423076923076924e-07,
      "loss": 1.0627,
      "step": 240
    },
    {
      "epoch": 4.228070175438597,
      "grad_norm": 38.35462188720703,
      "learning_rate": 6.403846153846154e-07,
      "loss": 0.8004,
      "step": 241
    },
    {
      "epoch": 4.245614035087719,
      "grad_norm": 21.466026306152344,
      "learning_rate": 6.384615384615383e-07,
      "loss": 1.1149,
      "step": 242
    },
    {
      "epoch": 4.2631578947368425,
      "grad_norm": 34.41309356689453,
      "learning_rate": 6.365384615384614e-07,
      "loss": 1.1782,
      "step": 243
    },
    {
      "epoch": 4.280701754385965,
      "grad_norm": 34.77060317993164,
      "learning_rate": 6.346153846153845e-07,
      "loss": 0.9231,
      "step": 244
    },
    {
      "epoch": 4.298245614035087,
      "grad_norm": 26.94576072692871,
      "learning_rate": 6.326923076923076e-07,
      "loss": 1.1394,
      "step": 245
    },
    {
      "epoch": 4.315789473684211,
      "grad_norm": 19.103694915771484,
      "learning_rate": 6.307692307692307e-07,
      "loss": 0.939,
      "step": 246
    },
    {
      "epoch": 4.333333333333333,
      "grad_norm": 40.02509307861328,
      "learning_rate": 6.288461538461538e-07,
      "loss": 1.1394,
      "step": 247
    },
    {
      "epoch": 4.350877192982456,
      "grad_norm": 62.130611419677734,
      "learning_rate": 6.269230769230769e-07,
      "loss": 1.1179,
      "step": 248
    },
    {
      "epoch": 4.368421052631579,
      "grad_norm": 45.432708740234375,
      "learning_rate": 6.249999999999999e-07,
      "loss": 0.8211,
      "step": 249
    },
    {
      "epoch": 4.385964912280702,
      "grad_norm": 52.313697814941406,
      "learning_rate": 6.23076923076923e-07,
      "loss": 1.1714,
      "step": 250
    },
    {
      "epoch": 4.4035087719298245,
      "grad_norm": 24.41583824157715,
      "learning_rate": 6.211538461538461e-07,
      "loss": 1.0351,
      "step": 251
    },
    {
      "epoch": 4.421052631578947,
      "grad_norm": 45.19395446777344,
      "learning_rate": 6.192307692307692e-07,
      "loss": 0.6095,
      "step": 252
    },
    {
      "epoch": 4.43859649122807,
      "grad_norm": 31.14399528503418,
      "learning_rate": 6.173076923076923e-07,
      "loss": 1.138,
      "step": 253
    },
    {
      "epoch": 4.456140350877193,
      "grad_norm": 48.5388298034668,
      "learning_rate": 6.153846153846154e-07,
      "loss": 1.2723,
      "step": 254
    },
    {
      "epoch": 4.473684210526316,
      "grad_norm": 63.80923843383789,
      "learning_rate": 6.134615384615385e-07,
      "loss": 1.272,
      "step": 255
    },
    {
      "epoch": 4.491228070175438,
      "grad_norm": 24.565670013427734,
      "learning_rate": 6.115384615384616e-07,
      "loss": 0.8632,
      "step": 256
    },
    {
      "epoch": 4.508771929824562,
      "grad_norm": 27.799039840698242,
      "learning_rate": 6.096153846153846e-07,
      "loss": 0.9728,
      "step": 257
    },
    {
      "epoch": 4.526315789473684,
      "grad_norm": 32.81442642211914,
      "learning_rate": 6.076923076923076e-07,
      "loss": 1.0024,
      "step": 258
    },
    {
      "epoch": 4.543859649122807,
      "grad_norm": 19.44063949584961,
      "learning_rate": 6.057692307692307e-07,
      "loss": 1.1237,
      "step": 259
    },
    {
      "epoch": 4.56140350877193,
      "grad_norm": 50.00696563720703,
      "learning_rate": 6.038461538461538e-07,
      "loss": 0.9283,
      "step": 260
    },
    {
      "epoch": 4.578947368421053,
      "grad_norm": 35.66808319091797,
      "learning_rate": 6.019230769230769e-07,
      "loss": 0.7856,
      "step": 261
    },
    {
      "epoch": 4.5964912280701755,
      "grad_norm": 36.49198913574219,
      "learning_rate": 6e-07,
      "loss": 1.0933,
      "step": 262
    },
    {
      "epoch": 4.614035087719298,
      "grad_norm": 29.842639923095703,
      "learning_rate": 5.980769230769231e-07,
      "loss": 0.9233,
      "step": 263
    },
    {
      "epoch": 4.631578947368421,
      "grad_norm": 80.13423156738281,
      "learning_rate": 5.961538461538461e-07,
      "loss": 1.2123,
      "step": 264
    },
    {
      "epoch": 4.649122807017544,
      "grad_norm": 43.099300384521484,
      "learning_rate": 5.942307692307692e-07,
      "loss": 1.0768,
      "step": 265
    },
    {
      "epoch": 4.666666666666667,
      "grad_norm": 32.86907196044922,
      "learning_rate": 5.923076923076923e-07,
      "loss": 0.8827,
      "step": 266
    },
    {
      "epoch": 4.684210526315789,
      "grad_norm": 51.32353973388672,
      "learning_rate": 5.903846153846154e-07,
      "loss": 1.2962,
      "step": 267
    },
    {
      "epoch": 4.701754385964913,
      "grad_norm": 38.76735305786133,
      "learning_rate": 5.884615384615385e-07,
      "loss": 0.5998,
      "step": 268
    },
    {
      "epoch": 4.719298245614035,
      "grad_norm": 29.20305824279785,
      "learning_rate": 5.865384615384616e-07,
      "loss": 0.8401,
      "step": 269
    },
    {
      "epoch": 4.7368421052631575,
      "grad_norm": 37.6125373840332,
      "learning_rate": 5.846153846153847e-07,
      "loss": 1.0015,
      "step": 270
    },
    {
      "epoch": 4.754385964912281,
      "grad_norm": 60.04852294921875,
      "learning_rate": 5.826923076923078e-07,
      "loss": 1.052,
      "step": 271
    },
    {
      "epoch": 4.771929824561403,
      "grad_norm": 48.63616943359375,
      "learning_rate": 5.807692307692307e-07,
      "loss": 1.0901,
      "step": 272
    },
    {
      "epoch": 4.7894736842105265,
      "grad_norm": 42.764347076416016,
      "learning_rate": 5.788461538461538e-07,
      "loss": 0.5775,
      "step": 273
    },
    {
      "epoch": 4.807017543859649,
      "grad_norm": 38.956268310546875,
      "learning_rate": 5.769230769230768e-07,
      "loss": 1.13,
      "step": 274
    },
    {
      "epoch": 4.824561403508772,
      "grad_norm": 39.21760940551758,
      "learning_rate": 5.749999999999999e-07,
      "loss": 1.0837,
      "step": 275
    },
    {
      "epoch": 4.842105263157895,
      "grad_norm": 28.150279998779297,
      "learning_rate": 5.73076923076923e-07,
      "loss": 0.7626,
      "step": 276
    },
    {
      "epoch": 4.859649122807017,
      "grad_norm": 37.51382827758789,
      "learning_rate": 5.711538461538461e-07,
      "loss": 0.8416,
      "step": 277
    },
    {
      "epoch": 4.87719298245614,
      "grad_norm": 62.94125747680664,
      "learning_rate": 5.692307692307692e-07,
      "loss": 0.9175,
      "step": 278
    },
    {
      "epoch": 4.894736842105263,
      "grad_norm": 26.954492568969727,
      "learning_rate": 5.673076923076922e-07,
      "loss": 1.233,
      "step": 279
    },
    {
      "epoch": 4.912280701754386,
      "grad_norm": 35.33868408203125,
      "learning_rate": 5.653846153846153e-07,
      "loss": 0.8353,
      "step": 280
    },
    {
      "epoch": 4.9298245614035086,
      "grad_norm": 43.24800491333008,
      "learning_rate": 5.634615384615384e-07,
      "loss": 0.876,
      "step": 281
    },
    {
      "epoch": 4.947368421052632,
      "grad_norm": 36.65079879760742,
      "learning_rate": 5.615384615384615e-07,
      "loss": 1.1462,
      "step": 282
    },
    {
      "epoch": 4.964912280701754,
      "grad_norm": 20.44588851928711,
      "learning_rate": 5.596153846153846e-07,
      "loss": 0.9175,
      "step": 283
    },
    {
      "epoch": 4.982456140350877,
      "grad_norm": 29.738140106201172,
      "learning_rate": 5.576923076923077e-07,
      "loss": 0.878,
      "step": 284
    },
    {
      "epoch": 5.0,
      "grad_norm": 69.77880859375,
      "learning_rate": 5.557692307692308e-07,
      "loss": 1.176,
      "step": 285
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.5363984674329502,
      "eval_loss": 1.1245133876800537,
      "eval_roc_auc": 0.8024073042505275,
      "eval_runtime": 24.2409,
      "eval_samples_per_second": 10.767,
      "eval_steps_per_second": 0.701,
      "step": 285
    },
    {
      "epoch": 5.017543859649122,
      "grad_norm": 43.9262580871582,
      "learning_rate": 5.538461538461539e-07,
      "loss": 0.9786,
      "step": 286
    },
    {
      "epoch": 5.035087719298246,
      "grad_norm": 42.07478713989258,
      "learning_rate": 5.519230769230769e-07,
      "loss": 0.8725,
      "step": 287
    },
    {
      "epoch": 5.052631578947368,
      "grad_norm": 18.638660430908203,
      "learning_rate": 5.5e-07,
      "loss": 0.5895,
      "step": 288
    },
    {
      "epoch": 5.0701754385964914,
      "grad_norm": 29.218503952026367,
      "learning_rate": 5.480769230769231e-07,
      "loss": 0.8934,
      "step": 289
    },
    {
      "epoch": 5.087719298245614,
      "grad_norm": 72.49881744384766,
      "learning_rate": 5.461538461538461e-07,
      "loss": 1.0794,
      "step": 290
    },
    {
      "epoch": 5.105263157894737,
      "grad_norm": 47.327239990234375,
      "learning_rate": 5.442307692307692e-07,
      "loss": 0.9808,
      "step": 291
    },
    {
      "epoch": 5.12280701754386,
      "grad_norm": 49.39778137207031,
      "learning_rate": 5.423076923076923e-07,
      "loss": 1.0015,
      "step": 292
    },
    {
      "epoch": 5.140350877192983,
      "grad_norm": 36.040836334228516,
      "learning_rate": 5.403846153846154e-07,
      "loss": 1.0248,
      "step": 293
    },
    {
      "epoch": 5.157894736842105,
      "grad_norm": 33.901615142822266,
      "learning_rate": 5.384615384615384e-07,
      "loss": 0.908,
      "step": 294
    },
    {
      "epoch": 5.175438596491228,
      "grad_norm": 32.479549407958984,
      "learning_rate": 5.365384615384615e-07,
      "loss": 0.7743,
      "step": 295
    },
    {
      "epoch": 5.192982456140351,
      "grad_norm": 73.39112854003906,
      "learning_rate": 5.346153846153846e-07,
      "loss": 0.911,
      "step": 296
    },
    {
      "epoch": 5.2105263157894735,
      "grad_norm": 50.840206146240234,
      "learning_rate": 5.326923076923077e-07,
      "loss": 1.1771,
      "step": 297
    },
    {
      "epoch": 5.228070175438597,
      "grad_norm": 29.241947174072266,
      "learning_rate": 5.307692307692308e-07,
      "loss": 1.0472,
      "step": 298
    },
    {
      "epoch": 5.245614035087719,
      "grad_norm": 32.0538215637207,
      "learning_rate": 5.288461538461539e-07,
      "loss": 0.9781,
      "step": 299
    },
    {
      "epoch": 5.2631578947368425,
      "grad_norm": 39.43052291870117,
      "learning_rate": 5.269230769230769e-07,
      "loss": 0.9577,
      "step": 300
    },
    {
      "epoch": 5.280701754385965,
      "grad_norm": 23.0631160736084,
      "learning_rate": 5.25e-07,
      "loss": 1.0764,
      "step": 301
    },
    {
      "epoch": 5.298245614035087,
      "grad_norm": 46.16770553588867,
      "learning_rate": 5.23076923076923e-07,
      "loss": 0.792,
      "step": 302
    },
    {
      "epoch": 5.315789473684211,
      "grad_norm": 51.57275390625,
      "learning_rate": 5.211538461538461e-07,
      "loss": 0.9721,
      "step": 303
    },
    {
      "epoch": 5.333333333333333,
      "grad_norm": 30.06395721435547,
      "learning_rate": 5.192307692307692e-07,
      "loss": 1.0893,
      "step": 304
    },
    {
      "epoch": 5.350877192982456,
      "grad_norm": 71.85147094726562,
      "learning_rate": 5.173076923076923e-07,
      "loss": 1.2034,
      "step": 305
    },
    {
      "epoch": 5.368421052631579,
      "grad_norm": 45.4450569152832,
      "learning_rate": 5.153846153846153e-07,
      "loss": 0.7588,
      "step": 306
    },
    {
      "epoch": 5.385964912280702,
      "grad_norm": 40.09158706665039,
      "learning_rate": 5.134615384615384e-07,
      "loss": 0.9968,
      "step": 307
    },
    {
      "epoch": 5.4035087719298245,
      "grad_norm": 62.48497772216797,
      "learning_rate": 5.115384615384615e-07,
      "loss": 1.2783,
      "step": 308
    },
    {
      "epoch": 5.421052631578947,
      "grad_norm": 59.85103988647461,
      "learning_rate": 5.096153846153845e-07,
      "loss": 1.2234,
      "step": 309
    },
    {
      "epoch": 5.43859649122807,
      "grad_norm": 41.75504684448242,
      "learning_rate": 5.076923076923076e-07,
      "loss": 1.0972,
      "step": 310
    },
    {
      "epoch": 5.456140350877193,
      "grad_norm": 25.908830642700195,
      "learning_rate": 5.057692307692307e-07,
      "loss": 1.0729,
      "step": 311
    },
    {
      "epoch": 5.473684210526316,
      "grad_norm": 52.866825103759766,
      "learning_rate": 5.038461538461538e-07,
      "loss": 0.7172,
      "step": 312
    },
    {
      "epoch": 5.491228070175438,
      "grad_norm": 62.79011917114258,
      "learning_rate": 5.019230769230769e-07,
      "loss": 0.8971,
      "step": 313
    },
    {
      "epoch": 5.508771929824562,
      "grad_norm": 61.6699333190918,
      "learning_rate": 5e-07,
      "loss": 1.1901,
      "step": 314
    },
    {
      "epoch": 5.526315789473684,
      "grad_norm": 39.02582550048828,
      "learning_rate": 4.980769230769231e-07,
      "loss": 0.9623,
      "step": 315
    },
    {
      "epoch": 5.543859649122807,
      "grad_norm": 20.786083221435547,
      "learning_rate": 4.961538461538462e-07,
      "loss": 0.9772,
      "step": 316
    },
    {
      "epoch": 5.56140350877193,
      "grad_norm": 29.874135971069336,
      "learning_rate": 4.942307692307692e-07,
      "loss": 0.5292,
      "step": 317
    },
    {
      "epoch": 5.578947368421053,
      "grad_norm": 73.50117492675781,
      "learning_rate": 4.923076923076923e-07,
      "loss": 1.4202,
      "step": 318
    },
    {
      "epoch": 5.5964912280701755,
      "grad_norm": 28.74661636352539,
      "learning_rate": 4.903846153846153e-07,
      "loss": 0.6696,
      "step": 319
    },
    {
      "epoch": 5.614035087719298,
      "grad_norm": 38.96842956542969,
      "learning_rate": 4.884615384615384e-07,
      "loss": 0.9159,
      "step": 320
    },
    {
      "epoch": 5.631578947368421,
      "grad_norm": 36.87344741821289,
      "learning_rate": 4.865384615384615e-07,
      "loss": 0.9728,
      "step": 321
    },
    {
      "epoch": 5.649122807017544,
      "grad_norm": 38.103668212890625,
      "learning_rate": 4.846153846153846e-07,
      "loss": 0.9663,
      "step": 322
    },
    {
      "epoch": 5.666666666666667,
      "grad_norm": 24.839292526245117,
      "learning_rate": 4.826923076923077e-07,
      "loss": 0.8722,
      "step": 323
    },
    {
      "epoch": 5.684210526315789,
      "grad_norm": 27.795520782470703,
      "learning_rate": 4.807692307692307e-07,
      "loss": 0.7313,
      "step": 324
    },
    {
      "epoch": 5.701754385964913,
      "grad_norm": 48.86296844482422,
      "learning_rate": 4.788461538461538e-07,
      "loss": 0.7879,
      "step": 325
    },
    {
      "epoch": 5.719298245614035,
      "grad_norm": 43.693058013916016,
      "learning_rate": 4.769230769230769e-07,
      "loss": 0.944,
      "step": 326
    },
    {
      "epoch": 5.7368421052631575,
      "grad_norm": 26.41690444946289,
      "learning_rate": 4.7499999999999995e-07,
      "loss": 0.6766,
      "step": 327
    },
    {
      "epoch": 5.754385964912281,
      "grad_norm": 51.20581817626953,
      "learning_rate": 4.7307692307692304e-07,
      "loss": 0.8956,
      "step": 328
    },
    {
      "epoch": 5.771929824561403,
      "grad_norm": 42.70225143432617,
      "learning_rate": 4.711538461538461e-07,
      "loss": 0.9747,
      "step": 329
    },
    {
      "epoch": 5.7894736842105265,
      "grad_norm": 32.69301986694336,
      "learning_rate": 4.692307692307692e-07,
      "loss": 0.5307,
      "step": 330
    },
    {
      "epoch": 5.807017543859649,
      "grad_norm": 47.880863189697266,
      "learning_rate": 4.673076923076923e-07,
      "loss": 0.7341,
      "step": 331
    },
    {
      "epoch": 5.824561403508772,
      "grad_norm": 37.469017028808594,
      "learning_rate": 4.653846153846154e-07,
      "loss": 0.8207,
      "step": 332
    },
    {
      "epoch": 5.842105263157895,
      "grad_norm": 44.53699493408203,
      "learning_rate": 4.634615384615385e-07,
      "loss": 1.0366,
      "step": 333
    },
    {
      "epoch": 5.859649122807017,
      "grad_norm": 34.84064483642578,
      "learning_rate": 4.6153846153846156e-07,
      "loss": 0.9788,
      "step": 334
    },
    {
      "epoch": 5.87719298245614,
      "grad_norm": 35.885215759277344,
      "learning_rate": 4.596153846153846e-07,
      "loss": 0.8453,
      "step": 335
    },
    {
      "epoch": 5.894736842105263,
      "grad_norm": 33.525081634521484,
      "learning_rate": 4.5769230769230763e-07,
      "loss": 0.6823,
      "step": 336
    },
    {
      "epoch": 5.912280701754386,
      "grad_norm": 46.2805290222168,
      "learning_rate": 4.557692307692307e-07,
      "loss": 0.9438,
      "step": 337
    },
    {
      "epoch": 5.9298245614035086,
      "grad_norm": 56.342018127441406,
      "learning_rate": 4.538461538461538e-07,
      "loss": 1.2671,
      "step": 338
    },
    {
      "epoch": 5.947368421052632,
      "grad_norm": 32.246089935302734,
      "learning_rate": 4.519230769230769e-07,
      "loss": 0.9635,
      "step": 339
    },
    {
      "epoch": 5.964912280701754,
      "grad_norm": 39.31554412841797,
      "learning_rate": 4.5e-07,
      "loss": 0.8637,
      "step": 340
    },
    {
      "epoch": 5.982456140350877,
      "grad_norm": 32.075775146484375,
      "learning_rate": 4.4807692307692307e-07,
      "loss": 1.0645,
      "step": 341
    },
    {
      "epoch": 6.0,
      "grad_norm": 33.591068267822266,
      "learning_rate": 4.4615384615384615e-07,
      "loss": 0.7403,
      "step": 342
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.5708812260536399,
      "eval_loss": 1.0752440690994263,
      "eval_roc_auc": 0.8091941568824714,
      "eval_runtime": 25.1571,
      "eval_samples_per_second": 10.375,
      "eval_steps_per_second": 0.676,
      "step": 342
    },
    {
      "epoch": 6.017543859649122,
      "grad_norm": 23.306591033935547,
      "learning_rate": 4.442307692307692e-07,
      "loss": 0.5686,
      "step": 343
    },
    {
      "epoch": 6.035087719298246,
      "grad_norm": 44.798919677734375,
      "learning_rate": 4.423076923076923e-07,
      "loss": 1.203,
      "step": 344
    },
    {
      "epoch": 6.052631578947368,
      "grad_norm": 23.386505126953125,
      "learning_rate": 4.4038461538461536e-07,
      "loss": 0.7083,
      "step": 345
    },
    {
      "epoch": 6.0701754385964914,
      "grad_norm": 49.338531494140625,
      "learning_rate": 4.3846153846153845e-07,
      "loss": 0.9921,
      "step": 346
    },
    {
      "epoch": 6.087719298245614,
      "grad_norm": 26.851451873779297,
      "learning_rate": 4.3653846153846154e-07,
      "loss": 0.7737,
      "step": 347
    },
    {
      "epoch": 6.105263157894737,
      "grad_norm": 41.42471694946289,
      "learning_rate": 4.346153846153846e-07,
      "loss": 1.1449,
      "step": 348
    },
    {
      "epoch": 6.12280701754386,
      "grad_norm": 30.84017562866211,
      "learning_rate": 4.326923076923077e-07,
      "loss": 1.0377,
      "step": 349
    },
    {
      "epoch": 6.140350877192983,
      "grad_norm": 21.94278335571289,
      "learning_rate": 4.307692307692308e-07,
      "loss": 0.7842,
      "step": 350
    },
    {
      "epoch": 6.157894736842105,
      "grad_norm": 54.80387496948242,
      "learning_rate": 4.288461538461538e-07,
      "loss": 0.6877,
      "step": 351
    },
    {
      "epoch": 6.175438596491228,
      "grad_norm": 52.03696060180664,
      "learning_rate": 4.2692307692307687e-07,
      "loss": 0.819,
      "step": 352
    },
    {
      "epoch": 6.192982456140351,
      "grad_norm": 47.98991775512695,
      "learning_rate": 4.2499999999999995e-07,
      "loss": 1.2738,
      "step": 353
    },
    {
      "epoch": 6.2105263157894735,
      "grad_norm": 31.361557006835938,
      "learning_rate": 4.2307692307692304e-07,
      "loss": 1.0351,
      "step": 354
    },
    {
      "epoch": 6.228070175438597,
      "grad_norm": 58.75925827026367,
      "learning_rate": 4.2115384615384613e-07,
      "loss": 0.9005,
      "step": 355
    },
    {
      "epoch": 6.245614035087719,
      "grad_norm": 61.34735107421875,
      "learning_rate": 4.192307692307692e-07,
      "loss": 1.335,
      "step": 356
    },
    {
      "epoch": 6.2631578947368425,
      "grad_norm": 38.18897247314453,
      "learning_rate": 4.173076923076923e-07,
      "loss": 0.8518,
      "step": 357
    },
    {
      "epoch": 6.280701754385965,
      "grad_norm": 47.511199951171875,
      "learning_rate": 4.153846153846154e-07,
      "loss": 1.0651,
      "step": 358
    },
    {
      "epoch": 6.298245614035087,
      "grad_norm": 52.5938720703125,
      "learning_rate": 4.134615384615384e-07,
      "loss": 1.2098,
      "step": 359
    },
    {
      "epoch": 6.315789473684211,
      "grad_norm": 46.33355712890625,
      "learning_rate": 4.115384615384615e-07,
      "loss": 0.6994,
      "step": 360
    },
    {
      "epoch": 6.333333333333333,
      "grad_norm": 43.15740203857422,
      "learning_rate": 4.096153846153846e-07,
      "loss": 1.0389,
      "step": 361
    },
    {
      "epoch": 6.350877192982456,
      "grad_norm": 33.72315216064453,
      "learning_rate": 4.076923076923077e-07,
      "loss": 0.862,
      "step": 362
    },
    {
      "epoch": 6.368421052631579,
      "grad_norm": 50.00696563720703,
      "learning_rate": 4.0576923076923077e-07,
      "loss": 0.9237,
      "step": 363
    },
    {
      "epoch": 6.385964912280702,
      "grad_norm": 22.04112434387207,
      "learning_rate": 4.0384615384615386e-07,
      "loss": 0.7235,
      "step": 364
    },
    {
      "epoch": 6.4035087719298245,
      "grad_norm": 56.314048767089844,
      "learning_rate": 4.019230769230769e-07,
      "loss": 1.048,
      "step": 365
    },
    {
      "epoch": 6.421052631578947,
      "grad_norm": 26.58925437927246,
      "learning_rate": 4e-07,
      "loss": 0.6484,
      "step": 366
    },
    {
      "epoch": 6.43859649122807,
      "grad_norm": 55.187747955322266,
      "learning_rate": 3.98076923076923e-07,
      "loss": 0.9495,
      "step": 367
    },
    {
      "epoch": 6.456140350877193,
      "grad_norm": 45.539772033691406,
      "learning_rate": 3.961538461538461e-07,
      "loss": 0.6283,
      "step": 368
    },
    {
      "epoch": 6.473684210526316,
      "grad_norm": 40.78237533569336,
      "learning_rate": 3.942307692307692e-07,
      "loss": 0.5378,
      "step": 369
    },
    {
      "epoch": 6.491228070175438,
      "grad_norm": 24.31471824645996,
      "learning_rate": 3.923076923076923e-07,
      "loss": 0.7294,
      "step": 370
    },
    {
      "epoch": 6.508771929824562,
      "grad_norm": 30.035335540771484,
      "learning_rate": 3.9038461538461536e-07,
      "loss": 0.8559,
      "step": 371
    },
    {
      "epoch": 6.526315789473684,
      "grad_norm": 31.007173538208008,
      "learning_rate": 3.8846153846153845e-07,
      "loss": 0.802,
      "step": 372
    },
    {
      "epoch": 6.543859649122807,
      "grad_norm": 43.71822738647461,
      "learning_rate": 3.8653846153846154e-07,
      "loss": 0.748,
      "step": 373
    },
    {
      "epoch": 6.56140350877193,
      "grad_norm": 21.99559783935547,
      "learning_rate": 3.8461538461538463e-07,
      "loss": 0.6553,
      "step": 374
    },
    {
      "epoch": 6.578947368421053,
      "grad_norm": 44.91862869262695,
      "learning_rate": 3.8269230769230766e-07,
      "loss": 0.8155,
      "step": 375
    },
    {
      "epoch": 6.5964912280701755,
      "grad_norm": 28.355440139770508,
      "learning_rate": 3.8076923076923075e-07,
      "loss": 1.1146,
      "step": 376
    },
    {
      "epoch": 6.614035087719298,
      "grad_norm": 28.48116111755371,
      "learning_rate": 3.7884615384615384e-07,
      "loss": 1.1448,
      "step": 377
    },
    {
      "epoch": 6.631578947368421,
      "grad_norm": 50.27273941040039,
      "learning_rate": 3.769230769230769e-07,
      "loss": 1.0349,
      "step": 378
    },
    {
      "epoch": 6.649122807017544,
      "grad_norm": 32.425384521484375,
      "learning_rate": 3.75e-07,
      "loss": 0.8526,
      "step": 379
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 73.90445709228516,
      "learning_rate": 3.7307692307692304e-07,
      "loss": 1.2825,
      "step": 380
    },
    {
      "epoch": 6.684210526315789,
      "grad_norm": 48.79448699951172,
      "learning_rate": 3.7115384615384613e-07,
      "loss": 0.6351,
      "step": 381
    },
    {
      "epoch": 6.701754385964913,
      "grad_norm": 24.40697479248047,
      "learning_rate": 3.692307692307692e-07,
      "loss": 0.7081,
      "step": 382
    },
    {
      "epoch": 6.719298245614035,
      "grad_norm": 47.5269660949707,
      "learning_rate": 3.673076923076923e-07,
      "loss": 0.671,
      "step": 383
    },
    {
      "epoch": 6.7368421052631575,
      "grad_norm": 37.11399841308594,
      "learning_rate": 3.6538461538461534e-07,
      "loss": 0.6574,
      "step": 384
    },
    {
      "epoch": 6.754385964912281,
      "grad_norm": 43.93232345581055,
      "learning_rate": 3.6346153846153843e-07,
      "loss": 1.0039,
      "step": 385
    },
    {
      "epoch": 6.771929824561403,
      "grad_norm": 35.62379455566406,
      "learning_rate": 3.615384615384615e-07,
      "loss": 0.7748,
      "step": 386
    },
    {
      "epoch": 6.7894736842105265,
      "grad_norm": 68.19344329833984,
      "learning_rate": 3.596153846153846e-07,
      "loss": 1.033,
      "step": 387
    },
    {
      "epoch": 6.807017543859649,
      "grad_norm": 39.3653678894043,
      "learning_rate": 3.576923076923077e-07,
      "loss": 0.5758,
      "step": 388
    },
    {
      "epoch": 6.824561403508772,
      "grad_norm": 30.772722244262695,
      "learning_rate": 3.557692307692308e-07,
      "loss": 0.8695,
      "step": 389
    },
    {
      "epoch": 6.842105263157895,
      "grad_norm": 32.83871078491211,
      "learning_rate": 3.5384615384615386e-07,
      "loss": 1.0549,
      "step": 390
    },
    {
      "epoch": 6.859649122807017,
      "grad_norm": 62.99473190307617,
      "learning_rate": 3.5192307692307695e-07,
      "loss": 0.9173,
      "step": 391
    },
    {
      "epoch": 6.87719298245614,
      "grad_norm": 42.774044036865234,
      "learning_rate": 3.5e-07,
      "loss": 1.0925,
      "step": 392
    },
    {
      "epoch": 6.894736842105263,
      "grad_norm": 45.01066589355469,
      "learning_rate": 3.4807692307692307e-07,
      "loss": 1.0266,
      "step": 393
    },
    {
      "epoch": 6.912280701754386,
      "grad_norm": 30.733928680419922,
      "learning_rate": 3.461538461538461e-07,
      "loss": 0.7034,
      "step": 394
    },
    {
      "epoch": 6.9298245614035086,
      "grad_norm": 19.22307014465332,
      "learning_rate": 3.442307692307692e-07,
      "loss": 0.9147,
      "step": 395
    },
    {
      "epoch": 6.947368421052632,
      "grad_norm": 68.61734008789062,
      "learning_rate": 3.423076923076923e-07,
      "loss": 1.1881,
      "step": 396
    },
    {
      "epoch": 6.964912280701754,
      "grad_norm": 33.16205978393555,
      "learning_rate": 3.4038461538461537e-07,
      "loss": 0.7483,
      "step": 397
    },
    {
      "epoch": 6.982456140350877,
      "grad_norm": 19.65161895751953,
      "learning_rate": 3.3846153846153845e-07,
      "loss": 0.6954,
      "step": 398
    },
    {
      "epoch": 7.0,
      "grad_norm": 92.69136810302734,
      "learning_rate": 3.3653846153846154e-07,
      "loss": 1.1487,
      "step": 399
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.5632183908045977,
      "eval_loss": 1.069606065750122,
      "eval_roc_auc": 0.8184030717743787,
      "eval_runtime": 24.405,
      "eval_samples_per_second": 10.695,
      "eval_steps_per_second": 0.697,
      "step": 399
    },
    {
      "epoch": 7.017543859649122,
      "grad_norm": 87.78544616699219,
      "learning_rate": 3.346153846153846e-07,
      "loss": 0.976,
      "step": 400
    },
    {
      "epoch": 7.035087719298246,
      "grad_norm": 27.48891258239746,
      "learning_rate": 3.3269230769230766e-07,
      "loss": 0.5747,
      "step": 401
    },
    {
      "epoch": 7.052631578947368,
      "grad_norm": 45.76131057739258,
      "learning_rate": 3.3076923076923075e-07,
      "loss": 0.9605,
      "step": 402
    },
    {
      "epoch": 7.0701754385964914,
      "grad_norm": 20.498069763183594,
      "learning_rate": 3.2884615384615384e-07,
      "loss": 0.5053,
      "step": 403
    },
    {
      "epoch": 7.087719298245614,
      "grad_norm": 22.58955192565918,
      "learning_rate": 3.269230769230769e-07,
      "loss": 0.5368,
      "step": 404
    },
    {
      "epoch": 7.105263157894737,
      "grad_norm": 22.225399017333984,
      "learning_rate": 3.25e-07,
      "loss": 0.777,
      "step": 405
    },
    {
      "epoch": 7.12280701754386,
      "grad_norm": 39.7629508972168,
      "learning_rate": 3.230769230769231e-07,
      "loss": 0.7354,
      "step": 406
    },
    {
      "epoch": 7.140350877192983,
      "grad_norm": 58.76552200317383,
      "learning_rate": 3.211538461538462e-07,
      "loss": 1.1802,
      "step": 407
    },
    {
      "epoch": 7.157894736842105,
      "grad_norm": 22.695573806762695,
      "learning_rate": 3.1923076923076917e-07,
      "loss": 0.8024,
      "step": 408
    },
    {
      "epoch": 7.175438596491228,
      "grad_norm": 33.9383430480957,
      "learning_rate": 3.1730769230769225e-07,
      "loss": 1.0508,
      "step": 409
    },
    {
      "epoch": 7.192982456140351,
      "grad_norm": 30.876842498779297,
      "learning_rate": 3.1538461538461534e-07,
      "loss": 0.7214,
      "step": 410
    },
    {
      "epoch": 7.2105263157894735,
      "grad_norm": 36.76365280151367,
      "learning_rate": 3.1346153846153843e-07,
      "loss": 0.7907,
      "step": 411
    },
    {
      "epoch": 7.228070175438597,
      "grad_norm": 65.87890625,
      "learning_rate": 3.115384615384615e-07,
      "loss": 1.2976,
      "step": 412
    },
    {
      "epoch": 7.245614035087719,
      "grad_norm": 43.14383316040039,
      "learning_rate": 3.096153846153846e-07,
      "loss": 1.1367,
      "step": 413
    },
    {
      "epoch": 7.2631578947368425,
      "grad_norm": 27.98732566833496,
      "learning_rate": 3.076923076923077e-07,
      "loss": 1.0924,
      "step": 414
    },
    {
      "epoch": 7.280701754385965,
      "grad_norm": 48.689537048339844,
      "learning_rate": 3.057692307692308e-07,
      "loss": 1.0298,
      "step": 415
    },
    {
      "epoch": 7.298245614035087,
      "grad_norm": 27.990142822265625,
      "learning_rate": 3.038461538461538e-07,
      "loss": 0.9779,
      "step": 416
    },
    {
      "epoch": 7.315789473684211,
      "grad_norm": 29.33123016357422,
      "learning_rate": 3.019230769230769e-07,
      "loss": 0.7683,
      "step": 417
    },
    {
      "epoch": 7.333333333333333,
      "grad_norm": 26.44052505493164,
      "learning_rate": 3e-07,
      "loss": 0.8771,
      "step": 418
    },
    {
      "epoch": 7.350877192982456,
      "grad_norm": 28.419343948364258,
      "learning_rate": 2.980769230769231e-07,
      "loss": 0.9028,
      "step": 419
    },
    {
      "epoch": 7.368421052631579,
      "grad_norm": 38.75981521606445,
      "learning_rate": 2.9615384615384616e-07,
      "loss": 0.9058,
      "step": 420
    },
    {
      "epoch": 7.385964912280702,
      "grad_norm": 27.943565368652344,
      "learning_rate": 2.9423076923076925e-07,
      "loss": 0.5811,
      "step": 421
    },
    {
      "epoch": 7.4035087719298245,
      "grad_norm": 38.048736572265625,
      "learning_rate": 2.9230769230769234e-07,
      "loss": 1.0206,
      "step": 422
    },
    {
      "epoch": 7.421052631578947,
      "grad_norm": 30.790170669555664,
      "learning_rate": 2.9038461538461537e-07,
      "loss": 0.8268,
      "step": 423
    },
    {
      "epoch": 7.43859649122807,
      "grad_norm": 52.75239562988281,
      "learning_rate": 2.884615384615384e-07,
      "loss": 1.0024,
      "step": 424
    },
    {
      "epoch": 7.456140350877193,
      "grad_norm": 30.304975509643555,
      "learning_rate": 2.865384615384615e-07,
      "loss": 0.7764,
      "step": 425
    },
    {
      "epoch": 7.473684210526316,
      "grad_norm": 44.08205795288086,
      "learning_rate": 2.846153846153846e-07,
      "loss": 0.6279,
      "step": 426
    },
    {
      "epoch": 7.491228070175438,
      "grad_norm": 29.00324058532715,
      "learning_rate": 2.8269230769230767e-07,
      "loss": 0.7732,
      "step": 427
    },
    {
      "epoch": 7.508771929824562,
      "grad_norm": 44.4395637512207,
      "learning_rate": 2.8076923076923075e-07,
      "loss": 0.5693,
      "step": 428
    },
    {
      "epoch": 7.526315789473684,
      "grad_norm": 28.682804107666016,
      "learning_rate": 2.7884615384615384e-07,
      "loss": 0.8483,
      "step": 429
    },
    {
      "epoch": 7.543859649122807,
      "grad_norm": 71.88604736328125,
      "learning_rate": 2.7692307692307693e-07,
      "loss": 1.0121,
      "step": 430
    },
    {
      "epoch": 7.56140350877193,
      "grad_norm": 37.93716049194336,
      "learning_rate": 2.75e-07,
      "loss": 0.9846,
      "step": 431
    },
    {
      "epoch": 7.578947368421053,
      "grad_norm": 68.40493774414062,
      "learning_rate": 2.7307692307692305e-07,
      "loss": 0.7968,
      "step": 432
    },
    {
      "epoch": 7.5964912280701755,
      "grad_norm": 34.130680084228516,
      "learning_rate": 2.7115384615384614e-07,
      "loss": 0.6804,
      "step": 433
    },
    {
      "epoch": 7.614035087719298,
      "grad_norm": 66.54443359375,
      "learning_rate": 2.692307692307692e-07,
      "loss": 0.8493,
      "step": 434
    },
    {
      "epoch": 7.631578947368421,
      "grad_norm": 30.927793502807617,
      "learning_rate": 2.673076923076923e-07,
      "loss": 0.8261,
      "step": 435
    },
    {
      "epoch": 7.649122807017544,
      "grad_norm": 31.034412384033203,
      "learning_rate": 2.653846153846154e-07,
      "loss": 1.0382,
      "step": 436
    },
    {
      "epoch": 7.666666666666667,
      "grad_norm": 28.199138641357422,
      "learning_rate": 2.6346153846153843e-07,
      "loss": 0.6613,
      "step": 437
    },
    {
      "epoch": 7.684210526315789,
      "grad_norm": 34.426605224609375,
      "learning_rate": 2.615384615384615e-07,
      "loss": 0.8676,
      "step": 438
    },
    {
      "epoch": 7.701754385964913,
      "grad_norm": 26.49811553955078,
      "learning_rate": 2.596153846153846e-07,
      "loss": 0.6587,
      "step": 439
    },
    {
      "epoch": 7.719298245614035,
      "grad_norm": 28.70757293701172,
      "learning_rate": 2.5769230769230764e-07,
      "loss": 0.7954,
      "step": 440
    },
    {
      "epoch": 7.7368421052631575,
      "grad_norm": 38.85836410522461,
      "learning_rate": 2.5576923076923073e-07,
      "loss": 0.8062,
      "step": 441
    },
    {
      "epoch": 7.754385964912281,
      "grad_norm": 26.44515609741211,
      "learning_rate": 2.538461538461538e-07,
      "loss": 1.0693,
      "step": 442
    },
    {
      "epoch": 7.771929824561403,
      "grad_norm": 55.86748504638672,
      "learning_rate": 2.519230769230769e-07,
      "loss": 0.9166,
      "step": 443
    },
    {
      "epoch": 7.7894736842105265,
      "grad_norm": 28.77703857421875,
      "learning_rate": 2.5e-07,
      "loss": 0.8507,
      "step": 444
    },
    {
      "epoch": 7.807017543859649,
      "grad_norm": 39.333343505859375,
      "learning_rate": 2.480769230769231e-07,
      "loss": 1.02,
      "step": 445
    },
    {
      "epoch": 7.824561403508772,
      "grad_norm": 44.16427230834961,
      "learning_rate": 2.4615384615384616e-07,
      "loss": 1.0713,
      "step": 446
    },
    {
      "epoch": 7.842105263157895,
      "grad_norm": 22.932966232299805,
      "learning_rate": 2.442307692307692e-07,
      "loss": 0.4415,
      "step": 447
    },
    {
      "epoch": 7.859649122807017,
      "grad_norm": 48.378841400146484,
      "learning_rate": 2.423076923076923e-07,
      "loss": 0.7085,
      "step": 448
    },
    {
      "epoch": 7.87719298245614,
      "grad_norm": 50.131019592285156,
      "learning_rate": 2.4038461538461537e-07,
      "loss": 0.9735,
      "step": 449
    },
    {
      "epoch": 7.894736842105263,
      "grad_norm": 20.94889259338379,
      "learning_rate": 2.3846153846153846e-07,
      "loss": 0.6935,
      "step": 450
    },
    {
      "epoch": 7.912280701754386,
      "grad_norm": 30.711423873901367,
      "learning_rate": 2.3653846153846152e-07,
      "loss": 0.7382,
      "step": 451
    },
    {
      "epoch": 7.9298245614035086,
      "grad_norm": 37.16843795776367,
      "learning_rate": 2.346153846153846e-07,
      "loss": 1.0025,
      "step": 452
    },
    {
      "epoch": 7.947368421052632,
      "grad_norm": 61.31568908691406,
      "learning_rate": 2.326923076923077e-07,
      "loss": 0.7435,
      "step": 453
    },
    {
      "epoch": 7.964912280701754,
      "grad_norm": 30.279325485229492,
      "learning_rate": 2.3076923076923078e-07,
      "loss": 0.8862,
      "step": 454
    },
    {
      "epoch": 7.982456140350877,
      "grad_norm": 22.893749237060547,
      "learning_rate": 2.2884615384615382e-07,
      "loss": 0.8411,
      "step": 455
    },
    {
      "epoch": 8.0,
      "grad_norm": 33.30168533325195,
      "learning_rate": 2.269230769230769e-07,
      "loss": 0.521,
      "step": 456
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.5747126436781609,
      "eval_loss": 1.0445302724838257,
      "eval_roc_auc": 0.821241054744927,
      "eval_runtime": 24.8722,
      "eval_samples_per_second": 10.494,
      "eval_steps_per_second": 0.683,
      "step": 456
    },
    {
      "epoch": 8.017543859649123,
      "grad_norm": 32.153465270996094,
      "learning_rate": 2.25e-07,
      "loss": 0.8686,
      "step": 457
    },
    {
      "epoch": 8.035087719298245,
      "grad_norm": 26.060014724731445,
      "learning_rate": 2.2307692307692308e-07,
      "loss": 0.7634,
      "step": 458
    },
    {
      "epoch": 8.052631578947368,
      "grad_norm": 60.5893440246582,
      "learning_rate": 2.2115384615384614e-07,
      "loss": 1.3411,
      "step": 459
    },
    {
      "epoch": 8.070175438596491,
      "grad_norm": 28.305891036987305,
      "learning_rate": 2.1923076923076922e-07,
      "loss": 0.6241,
      "step": 460
    },
    {
      "epoch": 8.087719298245615,
      "grad_norm": 20.124664306640625,
      "learning_rate": 2.173076923076923e-07,
      "loss": 0.591,
      "step": 461
    },
    {
      "epoch": 8.105263157894736,
      "grad_norm": 87.66210174560547,
      "learning_rate": 2.153846153846154e-07,
      "loss": 1.2101,
      "step": 462
    },
    {
      "epoch": 8.12280701754386,
      "grad_norm": 35.75629425048828,
      "learning_rate": 2.1346153846153843e-07,
      "loss": 0.8357,
      "step": 463
    },
    {
      "epoch": 8.140350877192983,
      "grad_norm": 58.722328186035156,
      "learning_rate": 2.1153846153846152e-07,
      "loss": 1.0812,
      "step": 464
    },
    {
      "epoch": 8.157894736842104,
      "grad_norm": 45.29573059082031,
      "learning_rate": 2.096153846153846e-07,
      "loss": 0.7323,
      "step": 465
    },
    {
      "epoch": 8.175438596491228,
      "grad_norm": 29.968141555786133,
      "learning_rate": 2.076923076923077e-07,
      "loss": 0.6938,
      "step": 466
    },
    {
      "epoch": 8.192982456140351,
      "grad_norm": 37.96227264404297,
      "learning_rate": 2.0576923076923076e-07,
      "loss": 0.9054,
      "step": 467
    },
    {
      "epoch": 8.210526315789474,
      "grad_norm": 66.0616455078125,
      "learning_rate": 2.0384615384615384e-07,
      "loss": 0.9688,
      "step": 468
    },
    {
      "epoch": 8.228070175438596,
      "grad_norm": 70.00723266601562,
      "learning_rate": 2.0192307692307693e-07,
      "loss": 1.1553,
      "step": 469
    },
    {
      "epoch": 8.24561403508772,
      "grad_norm": 39.04280090332031,
      "learning_rate": 2e-07,
      "loss": 0.9341,
      "step": 470
    },
    {
      "epoch": 8.263157894736842,
      "grad_norm": 34.20655059814453,
      "learning_rate": 1.9807692307692305e-07,
      "loss": 0.9593,
      "step": 471
    },
    {
      "epoch": 8.280701754385966,
      "grad_norm": 63.928348541259766,
      "learning_rate": 1.9615384615384614e-07,
      "loss": 0.9055,
      "step": 472
    },
    {
      "epoch": 8.298245614035087,
      "grad_norm": 71.03302001953125,
      "learning_rate": 1.9423076923076923e-07,
      "loss": 0.921,
      "step": 473
    },
    {
      "epoch": 8.31578947368421,
      "grad_norm": 37.35957717895508,
      "learning_rate": 1.9230769230769231e-07,
      "loss": 0.6336,
      "step": 474
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 17.601655960083008,
      "learning_rate": 1.9038461538461537e-07,
      "loss": 0.4762,
      "step": 475
    },
    {
      "epoch": 8.350877192982455,
      "grad_norm": 45.87922286987305,
      "learning_rate": 1.8846153846153846e-07,
      "loss": 0.7384,
      "step": 476
    },
    {
      "epoch": 8.368421052631579,
      "grad_norm": 35.179988861083984,
      "learning_rate": 1.8653846153846152e-07,
      "loss": 0.6582,
      "step": 477
    },
    {
      "epoch": 8.385964912280702,
      "grad_norm": 29.4314022064209,
      "learning_rate": 1.846153846153846e-07,
      "loss": 0.6557,
      "step": 478
    },
    {
      "epoch": 8.403508771929825,
      "grad_norm": 39.96945571899414,
      "learning_rate": 1.8269230769230767e-07,
      "loss": 0.9508,
      "step": 479
    },
    {
      "epoch": 8.421052631578947,
      "grad_norm": 41.48187255859375,
      "learning_rate": 1.8076923076923076e-07,
      "loss": 0.6461,
      "step": 480
    },
    {
      "epoch": 8.43859649122807,
      "grad_norm": 46.457733154296875,
      "learning_rate": 1.7884615384615384e-07,
      "loss": 0.7903,
      "step": 481
    },
    {
      "epoch": 8.456140350877194,
      "grad_norm": 60.275028228759766,
      "learning_rate": 1.7692307692307693e-07,
      "loss": 0.9536,
      "step": 482
    },
    {
      "epoch": 8.473684210526315,
      "grad_norm": 32.316471099853516,
      "learning_rate": 1.75e-07,
      "loss": 0.7564,
      "step": 483
    },
    {
      "epoch": 8.491228070175438,
      "grad_norm": 30.793420791625977,
      "learning_rate": 1.7307692307692305e-07,
      "loss": 0.5605,
      "step": 484
    },
    {
      "epoch": 8.508771929824562,
      "grad_norm": 19.355497360229492,
      "learning_rate": 1.7115384615384614e-07,
      "loss": 0.5652,
      "step": 485
    },
    {
      "epoch": 8.526315789473685,
      "grad_norm": 24.5403995513916,
      "learning_rate": 1.6923076923076923e-07,
      "loss": 0.6776,
      "step": 486
    },
    {
      "epoch": 8.543859649122806,
      "grad_norm": 35.900184631347656,
      "learning_rate": 1.673076923076923e-07,
      "loss": 0.7501,
      "step": 487
    },
    {
      "epoch": 8.56140350877193,
      "grad_norm": 28.108530044555664,
      "learning_rate": 1.6538461538461538e-07,
      "loss": 0.5605,
      "step": 488
    },
    {
      "epoch": 8.578947368421053,
      "grad_norm": 39.321144104003906,
      "learning_rate": 1.6346153846153846e-07,
      "loss": 0.9753,
      "step": 489
    },
    {
      "epoch": 8.596491228070175,
      "grad_norm": 24.50644874572754,
      "learning_rate": 1.6153846153846155e-07,
      "loss": 0.9194,
      "step": 490
    },
    {
      "epoch": 8.614035087719298,
      "grad_norm": 50.84926986694336,
      "learning_rate": 1.5961538461538458e-07,
      "loss": 0.9425,
      "step": 491
    },
    {
      "epoch": 8.631578947368421,
      "grad_norm": 38.82880401611328,
      "learning_rate": 1.5769230769230767e-07,
      "loss": 0.8447,
      "step": 492
    },
    {
      "epoch": 8.649122807017545,
      "grad_norm": 48.61659240722656,
      "learning_rate": 1.5576923076923076e-07,
      "loss": 0.9379,
      "step": 493
    },
    {
      "epoch": 8.666666666666666,
      "grad_norm": 41.58673095703125,
      "learning_rate": 1.5384615384615385e-07,
      "loss": 0.8007,
      "step": 494
    },
    {
      "epoch": 8.68421052631579,
      "grad_norm": 49.88103485107422,
      "learning_rate": 1.519230769230769e-07,
      "loss": 1.2287,
      "step": 495
    },
    {
      "epoch": 8.701754385964913,
      "grad_norm": 32.42145538330078,
      "learning_rate": 1.5e-07,
      "loss": 0.655,
      "step": 496
    },
    {
      "epoch": 8.719298245614034,
      "grad_norm": 31.83061981201172,
      "learning_rate": 1.4807692307692308e-07,
      "loss": 0.7932,
      "step": 497
    },
    {
      "epoch": 8.736842105263158,
      "grad_norm": 45.73976516723633,
      "learning_rate": 1.4615384615384617e-07,
      "loss": 0.7517,
      "step": 498
    },
    {
      "epoch": 8.75438596491228,
      "grad_norm": 24.301603317260742,
      "learning_rate": 1.442307692307692e-07,
      "loss": 0.6368,
      "step": 499
    },
    {
      "epoch": 8.771929824561404,
      "grad_norm": 43.589569091796875,
      "learning_rate": 1.423076923076923e-07,
      "loss": 0.7572,
      "step": 500
    },
    {
      "epoch": 8.789473684210526,
      "grad_norm": 55.28351974487305,
      "learning_rate": 1.4038461538461538e-07,
      "loss": 0.8073,
      "step": 501
    },
    {
      "epoch": 8.807017543859649,
      "grad_norm": 29.774131774902344,
      "learning_rate": 1.3846153846153846e-07,
      "loss": 0.7737,
      "step": 502
    },
    {
      "epoch": 8.824561403508772,
      "grad_norm": 62.9353141784668,
      "learning_rate": 1.3653846153846152e-07,
      "loss": 0.7529,
      "step": 503
    },
    {
      "epoch": 8.842105263157894,
      "grad_norm": 38.0306510925293,
      "learning_rate": 1.346153846153846e-07,
      "loss": 0.7406,
      "step": 504
    },
    {
      "epoch": 8.859649122807017,
      "grad_norm": 41.864105224609375,
      "learning_rate": 1.326923076923077e-07,
      "loss": 0.9119,
      "step": 505
    },
    {
      "epoch": 8.87719298245614,
      "grad_norm": 67.60685729980469,
      "learning_rate": 1.3076923076923076e-07,
      "loss": 0.828,
      "step": 506
    },
    {
      "epoch": 8.894736842105264,
      "grad_norm": 37.95588302612305,
      "learning_rate": 1.2884615384615382e-07,
      "loss": 0.8381,
      "step": 507
    },
    {
      "epoch": 8.912280701754385,
      "grad_norm": 40.29434585571289,
      "learning_rate": 1.269230769230769e-07,
      "loss": 0.8085,
      "step": 508
    },
    {
      "epoch": 8.929824561403509,
      "grad_norm": 68.67707061767578,
      "learning_rate": 1.25e-07,
      "loss": 1.0056,
      "step": 509
    },
    {
      "epoch": 8.947368421052632,
      "grad_norm": 45.547428131103516,
      "learning_rate": 1.2307692307692308e-07,
      "loss": 0.7981,
      "step": 510
    },
    {
      "epoch": 8.964912280701755,
      "grad_norm": 39.541439056396484,
      "learning_rate": 1.2115384615384614e-07,
      "loss": 0.6487,
      "step": 511
    },
    {
      "epoch": 8.982456140350877,
      "grad_norm": 28.73076820373535,
      "learning_rate": 1.1923076923076923e-07,
      "loss": 0.5069,
      "step": 512
    },
    {
      "epoch": 9.0,
      "grad_norm": 31.97428321838379,
      "learning_rate": 1.173076923076923e-07,
      "loss": 0.9922,
      "step": 513
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.5977011494252874,
      "eval_loss": 1.0299025774002075,
      "eval_roc_auc": 0.8216890808633837,
      "eval_runtime": 34.5614,
      "eval_samples_per_second": 7.552,
      "eval_steps_per_second": 0.492,
      "step": 513
    },
    {
      "epoch": 9.017543859649123,
      "grad_norm": 48.0136833190918,
      "learning_rate": 1.1538461538461539e-07,
      "loss": 1.0235,
      "step": 514
    },
    {
      "epoch": 9.035087719298245,
      "grad_norm": 28.533130645751953,
      "learning_rate": 1.1346153846153845e-07,
      "loss": 0.5459,
      "step": 515
    },
    {
      "epoch": 9.052631578947368,
      "grad_norm": 41.17901611328125,
      "learning_rate": 1.1153846153846154e-07,
      "loss": 0.5279,
      "step": 516
    },
    {
      "epoch": 9.070175438596491,
      "grad_norm": 32.64225387573242,
      "learning_rate": 1.0961538461538461e-07,
      "loss": 1.0252,
      "step": 517
    },
    {
      "epoch": 9.087719298245615,
      "grad_norm": 32.45205307006836,
      "learning_rate": 1.076923076923077e-07,
      "loss": 0.6554,
      "step": 518
    },
    {
      "epoch": 9.105263157894736,
      "grad_norm": 35.420379638671875,
      "learning_rate": 1.0576923076923076e-07,
      "loss": 0.6954,
      "step": 519
    },
    {
      "epoch": 9.12280701754386,
      "grad_norm": 53.34769058227539,
      "learning_rate": 1.0384615384615385e-07,
      "loss": 0.7513,
      "step": 520
    },
    {
      "epoch": 9.140350877192983,
      "grad_norm": 28.763874053955078,
      "learning_rate": 1.0192307692307692e-07,
      "loss": 0.5913,
      "step": 521
    },
    {
      "epoch": 9.157894736842104,
      "grad_norm": 24.717885971069336,
      "learning_rate": 1e-07,
      "loss": 0.6692,
      "step": 522
    },
    {
      "epoch": 9.175438596491228,
      "grad_norm": 25.547611236572266,
      "learning_rate": 9.807692307692307e-08,
      "loss": 0.75,
      "step": 523
    },
    {
      "epoch": 9.192982456140351,
      "grad_norm": 41.97018051147461,
      "learning_rate": 9.615384615384616e-08,
      "loss": 0.6604,
      "step": 524
    },
    {
      "epoch": 9.210526315789474,
      "grad_norm": 80.47174835205078,
      "learning_rate": 9.423076923076923e-08,
      "loss": 1.2781,
      "step": 525
    },
    {
      "epoch": 9.228070175438596,
      "grad_norm": 27.064115524291992,
      "learning_rate": 9.23076923076923e-08,
      "loss": 0.7449,
      "step": 526
    },
    {
      "epoch": 9.24561403508772,
      "grad_norm": 34.857425689697266,
      "learning_rate": 9.038461538461538e-08,
      "loss": 0.5484,
      "step": 527
    },
    {
      "epoch": 9.263157894736842,
      "grad_norm": 63.48345184326172,
      "learning_rate": 8.846153846153847e-08,
      "loss": 0.8263,
      "step": 528
    },
    {
      "epoch": 9.280701754385966,
      "grad_norm": 36.6185417175293,
      "learning_rate": 8.653846153846153e-08,
      "loss": 0.8279,
      "step": 529
    },
    {
      "epoch": 9.298245614035087,
      "grad_norm": 45.01316833496094,
      "learning_rate": 8.461538461538461e-08,
      "loss": 0.9115,
      "step": 530
    },
    {
      "epoch": 9.31578947368421,
      "grad_norm": 19.511629104614258,
      "learning_rate": 8.269230769230769e-08,
      "loss": 0.4552,
      "step": 531
    },
    {
      "epoch": 9.333333333333334,
      "grad_norm": 42.41179275512695,
      "learning_rate": 8.076923076923077e-08,
      "loss": 1.0983,
      "step": 532
    },
    {
      "epoch": 9.350877192982455,
      "grad_norm": 26.47096061706543,
      "learning_rate": 7.884615384615384e-08,
      "loss": 0.5845,
      "step": 533
    },
    {
      "epoch": 9.368421052631579,
      "grad_norm": 30.966955184936523,
      "learning_rate": 7.692307692307692e-08,
      "loss": 0.6616,
      "step": 534
    },
    {
      "epoch": 9.385964912280702,
      "grad_norm": 40.17842102050781,
      "learning_rate": 7.5e-08,
      "loss": 0.854,
      "step": 535
    },
    {
      "epoch": 9.403508771929825,
      "grad_norm": 37.50959396362305,
      "learning_rate": 7.307692307692308e-08,
      "loss": 0.9193,
      "step": 536
    },
    {
      "epoch": 9.421052631578947,
      "grad_norm": 32.44236755371094,
      "learning_rate": 7.115384615384614e-08,
      "loss": 0.855,
      "step": 537
    },
    {
      "epoch": 9.43859649122807,
      "grad_norm": 37.18986511230469,
      "learning_rate": 6.923076923076923e-08,
      "loss": 0.9304,
      "step": 538
    },
    {
      "epoch": 9.456140350877194,
      "grad_norm": 36.91654586791992,
      "learning_rate": 6.73076923076923e-08,
      "loss": 0.8258,
      "step": 539
    },
    {
      "epoch": 9.473684210526315,
      "grad_norm": 27.073163986206055,
      "learning_rate": 6.538461538461538e-08,
      "loss": 0.4538,
      "step": 540
    },
    {
      "epoch": 9.491228070175438,
      "grad_norm": 33.02398681640625,
      "learning_rate": 6.346153846153845e-08,
      "loss": 0.8084,
      "step": 541
    },
    {
      "epoch": 9.508771929824562,
      "grad_norm": 25.321035385131836,
      "learning_rate": 6.153846153846154e-08,
      "loss": 0.6776,
      "step": 542
    },
    {
      "epoch": 9.526315789473685,
      "grad_norm": 35.03627395629883,
      "learning_rate": 5.961538461538461e-08,
      "loss": 0.7051,
      "step": 543
    },
    {
      "epoch": 9.543859649122806,
      "grad_norm": 32.04972457885742,
      "learning_rate": 5.7692307692307695e-08,
      "loss": 1.1416,
      "step": 544
    },
    {
      "epoch": 9.56140350877193,
      "grad_norm": 32.8387451171875,
      "learning_rate": 5.576923076923077e-08,
      "loss": 0.7075,
      "step": 545
    },
    {
      "epoch": 9.578947368421053,
      "grad_norm": 53.04689407348633,
      "learning_rate": 5.384615384615385e-08,
      "loss": 0.8701,
      "step": 546
    },
    {
      "epoch": 9.596491228070175,
      "grad_norm": 23.254966735839844,
      "learning_rate": 5.1923076923076924e-08,
      "loss": 0.4444,
      "step": 547
    },
    {
      "epoch": 9.614035087719298,
      "grad_norm": 42.885440826416016,
      "learning_rate": 5e-08,
      "loss": 0.8811,
      "step": 548
    },
    {
      "epoch": 9.631578947368421,
      "grad_norm": 29.11221694946289,
      "learning_rate": 4.807692307692308e-08,
      "loss": 0.7966,
      "step": 549
    },
    {
      "epoch": 9.649122807017545,
      "grad_norm": 62.373077392578125,
      "learning_rate": 4.615384615384615e-08,
      "loss": 0.9317,
      "step": 550
    },
    {
      "epoch": 9.666666666666666,
      "grad_norm": 43.64496612548828,
      "learning_rate": 4.423076923076923e-08,
      "loss": 0.7716,
      "step": 551
    },
    {
      "epoch": 9.68421052631579,
      "grad_norm": 30.535717010498047,
      "learning_rate": 4.230769230769231e-08,
      "loss": 1.0664,
      "step": 552
    },
    {
      "epoch": 9.701754385964913,
      "grad_norm": 24.165536880493164,
      "learning_rate": 4.038461538461539e-08,
      "loss": 0.8169,
      "step": 553
    },
    {
      "epoch": 9.719298245614034,
      "grad_norm": 47.13362503051758,
      "learning_rate": 3.846153846153846e-08,
      "loss": 0.7841,
      "step": 554
    },
    {
      "epoch": 9.736842105263158,
      "grad_norm": 28.94150161743164,
      "learning_rate": 3.653846153846154e-08,
      "loss": 0.7046,
      "step": 555
    },
    {
      "epoch": 9.75438596491228,
      "grad_norm": 56.8846549987793,
      "learning_rate": 3.4615384615384616e-08,
      "loss": 0.8988,
      "step": 556
    },
    {
      "epoch": 9.771929824561404,
      "grad_norm": 30.389190673828125,
      "learning_rate": 3.269230769230769e-08,
      "loss": 0.7905,
      "step": 557
    },
    {
      "epoch": 9.789473684210526,
      "grad_norm": 30.133394241333008,
      "learning_rate": 3.076923076923077e-08,
      "loss": 0.5083,
      "step": 558
    },
    {
      "epoch": 9.807017543859649,
      "grad_norm": 46.20474624633789,
      "learning_rate": 2.8846153846153848e-08,
      "loss": 0.9206,
      "step": 559
    },
    {
      "epoch": 9.824561403508772,
      "grad_norm": 38.54856872558594,
      "learning_rate": 2.6923076923076925e-08,
      "loss": 0.9944,
      "step": 560
    },
    {
      "epoch": 9.842105263157894,
      "grad_norm": 73.813232421875,
      "learning_rate": 2.5e-08,
      "loss": 0.9167,
      "step": 561
    },
    {
      "epoch": 9.859649122807017,
      "grad_norm": 42.26284408569336,
      "learning_rate": 2.3076923076923076e-08,
      "loss": 0.8989,
      "step": 562
    },
    {
      "epoch": 9.87719298245614,
      "grad_norm": 36.5334587097168,
      "learning_rate": 2.1153846153846153e-08,
      "loss": 0.6554,
      "step": 563
    },
    {
      "epoch": 9.894736842105264,
      "grad_norm": 30.85855484008789,
      "learning_rate": 1.923076923076923e-08,
      "loss": 0.4981,
      "step": 564
    },
    {
      "epoch": 9.912280701754385,
      "grad_norm": 38.74631881713867,
      "learning_rate": 1.7307692307692308e-08,
      "loss": 0.8997,
      "step": 565
    },
    {
      "epoch": 9.929824561403509,
      "grad_norm": 50.60679244995117,
      "learning_rate": 1.5384615384615385e-08,
      "loss": 0.9265,
      "step": 566
    },
    {
      "epoch": 9.947368421052632,
      "grad_norm": 36.15625762939453,
      "learning_rate": 1.3461538461538462e-08,
      "loss": 1.0373,
      "step": 567
    },
    {
      "epoch": 9.964912280701755,
      "grad_norm": 64.9268798828125,
      "learning_rate": 1.1538461538461538e-08,
      "loss": 1.0077,
      "step": 568
    },
    {
      "epoch": 9.982456140350877,
      "grad_norm": 28.213552474975586,
      "learning_rate": 9.615384615384615e-09,
      "loss": 1.102,
      "step": 569
    },
    {
      "epoch": 10.0,
      "grad_norm": 47.85600280761719,
      "learning_rate": 7.692307692307693e-09,
      "loss": 0.555,
      "step": 570
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.5977011494252874,
      "eval_loss": 1.0249524116516113,
      "eval_roc_auc": 0.8248013488853114,
      "eval_runtime": 23.6967,
      "eval_samples_per_second": 11.014,
      "eval_steps_per_second": 0.717,
      "step": 570
    }
  ],
  "logging_steps": 1,
  "max_steps": 570,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.204020265084518e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}