{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8186236889229982,
  "global_step": 400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.5454545454545457e-07,
      "loss": 1.7447,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 9.090909090909091e-07,
      "loss": 1.7545,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.3636363636363636e-06,
      "loss": 1.729,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 1.6094,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 1.6613,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 1.5168,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.181818181818182e-06,
      "loss": 1.4354,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 1.218,
      "step": 8
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.0909090909090915e-06,
      "loss": 1.2204,
      "step": 9
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 1.0702,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 5e-06,
      "loss": 0.9516,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 1.0377,
      "step": 12
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.90909090909091e-06,
      "loss": 0.9798,
      "step": 13
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.363636363636364e-06,
      "loss": 0.9245,
      "step": 14
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.818181818181818e-06,
      "loss": 0.9279,
      "step": 15
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.9215,
      "step": 16
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.727272727272727e-06,
      "loss": 0.9659,
      "step": 17
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.181818181818183e-06,
      "loss": 0.9475,
      "step": 18
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.636363636363637e-06,
      "loss": 0.9297,
      "step": 19
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.93,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.545454545454547e-06,
      "loss": 0.8521,
      "step": 21
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 0.8539,
      "step": 22
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.0454545454545455e-05,
      "loss": 0.8036,
      "step": 23
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 0.8456,
      "step": 24
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.1363636363636366e-05,
      "loss": 0.819,
      "step": 25
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.181818181818182e-05,
      "loss": 0.8771,
      "step": 26
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.2272727272727274e-05,
      "loss": 0.8038,
      "step": 27
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 0.8301,
      "step": 28
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.3181818181818183e-05,
      "loss": 0.8483,
      "step": 29
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 0.8009,
      "step": 30
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.4090909090909092e-05,
      "loss": 0.7494,
      "step": 31
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 0.8033,
      "step": 32
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.7652,
      "step": 33
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.5454545454545454e-05,
      "loss": 0.8586,
      "step": 34
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.590909090909091e-05,
      "loss": 0.7789,
      "step": 35
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 0.8003,
      "step": 36
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.681818181818182e-05,
      "loss": 0.7443,
      "step": 37
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.7272727272727274e-05,
      "loss": 0.8285,
      "step": 38
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.772727272727273e-05,
      "loss": 0.7628,
      "step": 39
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.7868,
      "step": 40
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8636363636363638e-05,
      "loss": 0.7481,
      "step": 41
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9090909090909094e-05,
      "loss": 0.7855,
      "step": 42
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9545454545454546e-05,
      "loss": 0.7506,
      "step": 43
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 0.7504,
      "step": 44
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9999975526680286e-05,
      "loss": 0.7607,
      "step": 45
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.999990210684092e-05,
      "loss": 0.8022,
      "step": 46
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.999977974084128e-05,
      "loss": 0.8513,
      "step": 47
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9999608429280302e-05,
      "loss": 0.7948,
      "step": 48
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9999388172996495e-05,
      "loss": 0.7338,
      "step": 49
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9999118973067942e-05,
      "loss": 0.7139,
      "step": 50
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.999880083081229e-05,
      "loss": 0.7114,
      "step": 51
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.999843374778673e-05,
      "loss": 0.7345,
      "step": 52
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.999801772578801e-05,
      "loss": 0.6277,
      "step": 53
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9997552766852434e-05,
      "loss": 0.7228,
      "step": 54
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9997038873255802e-05,
      "loss": 0.7264,
      "step": 55
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9996476047513454e-05,
      "loss": 0.7726,
      "step": 56
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.999586429238024e-05,
      "loss": 0.7114,
      "step": 57
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.999520361085049e-05,
      "loss": 0.7448,
      "step": 58
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9994494006158018e-05,
      "loss": 0.7882,
      "step": 59
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9993735481776105e-05,
      "loss": 0.7219,
      "step": 60
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9992928041417468e-05,
      "loss": 0.7117,
      "step": 61
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9992071689034257e-05,
      "loss": 0.6483,
      "step": 62
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9991166428818036e-05,
      "loss": 0.7375,
      "step": 63
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9990212265199738e-05,
      "loss": 0.6829,
      "step": 64
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.998920920284968e-05,
      "loss": 0.7334,
      "step": 65
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9988157246677514e-05,
      "loss": 0.7891,
      "step": 66
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9987056401832213e-05,
      "loss": 0.7708,
      "step": 67
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.998590667370204e-05,
      "loss": 0.787,
      "step": 68
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9984708067914533e-05,
      "loss": 0.7632,
      "step": 69
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9983460590336458e-05,
      "loss": 0.6609,
      "step": 70
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9982164247073803e-05,
      "loss": 0.7458,
      "step": 71
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9980819044471732e-05,
      "loss": 0.6935,
      "step": 72
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9979424989114555e-05,
      "loss": 0.6969,
      "step": 73
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9977982087825714e-05,
      "loss": 0.697,
      "step": 74
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9976490347667713e-05,
      "loss": 0.6652,
      "step": 75
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9974949775942134e-05,
      "loss": 0.7321,
      "step": 76
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9973360380189547e-05,
      "loss": 0.7344,
      "step": 77
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.997172216818951e-05,
      "loss": 0.6714,
      "step": 78
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9970035147960524e-05,
      "loss": 0.7586,
      "step": 79
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.996829932775999e-05,
      "loss": 0.6622,
      "step": 80
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.996651471608415e-05,
      "loss": 0.7297,
      "step": 81
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9964681321668097e-05,
      "loss": 0.6464,
      "step": 82
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.996279915348567e-05,
      "loss": 0.7213,
      "step": 83
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.996086822074945e-05,
      "loss": 0.6702,
      "step": 84
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9958888532910708e-05,
      "loss": 0.7002,
      "step": 85
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9956860099659345e-05,
      "loss": 0.6698,
      "step": 86
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9954782930923864e-05,
      "loss": 0.7386,
      "step": 87
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.995265703687131e-05,
      "loss": 0.7486,
      "step": 88
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.995048242790721e-05,
      "loss": 0.6778,
      "step": 89
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9948259114675553e-05,
      "loss": 0.7191,
      "step": 90
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9945987108058713e-05,
      "loss": 0.6939,
      "step": 91
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9943666419177392e-05,
      "loss": 0.6387,
      "step": 92
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9941297059390586e-05,
      "loss": 0.6718,
      "step": 93
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9938879040295508e-05,
      "loss": 0.683,
      "step": 94
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.993641237372756e-05,
      "loss": 0.7169,
      "step": 95
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9933897071760236e-05,
      "loss": 0.678,
      "step": 96
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9931333146705097e-05,
      "loss": 0.7516,
      "step": 97
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.99287206111117e-05,
      "loss": 0.7245,
      "step": 98
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.992605947776752e-05,
      "loss": 0.6419,
      "step": 99
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9923349759697915e-05,
      "loss": 0.6566,
      "step": 100
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9920591470166047e-05,
      "loss": 0.7252,
      "step": 101
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.991778462267281e-05,
      "loss": 0.6467,
      "step": 102
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9914929230956785e-05,
      "loss": 0.6693,
      "step": 103
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9912025308994146e-05,
      "loss": 0.671,
      "step": 104
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9909072870998625e-05,
      "loss": 0.6422,
      "step": 105
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9906071931421413e-05,
      "loss": 0.7121,
      "step": 106
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9903022504951098e-05,
      "loss": 0.7052,
      "step": 107
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9899924606513593e-05,
      "loss": 0.6915,
      "step": 108
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.989677825127208e-05,
      "loss": 0.7078,
      "step": 109
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9893583454626905e-05,
      "loss": 0.6683,
      "step": 110
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9890340232215517e-05,
      "loss": 0.6911,
      "step": 111
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9887048599912412e-05,
      "loss": 0.6564,
      "step": 112
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.988370857382902e-05,
      "loss": 0.6423,
      "step": 113
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.988032017031364e-05,
      "loss": 0.6684,
      "step": 114
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9876883405951378e-05,
      "loss": 0.725,
      "step": 115
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9873398297564036e-05,
      "loss": 0.6663,
      "step": 116
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9869864862210045e-05,
      "loss": 0.6671,
      "step": 117
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9866283117184393e-05,
      "loss": 0.6233,
      "step": 118
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9862653080018508e-05,
      "loss": 0.6376,
      "step": 119
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9858974768480206e-05,
      "loss": 0.6803,
      "step": 120
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.985524820057359e-05,
      "loss": 0.6626,
      "step": 121
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.985147339453895e-05,
      "loss": 0.6307,
      "step": 122
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9847650368852697e-05,
      "loss": 0.6416,
      "step": 123
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9843779142227258e-05,
      "loss": 0.6992,
      "step": 124
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9839859733610984e-05,
      "loss": 0.6593,
      "step": 125
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9835892162188065e-05,
      "loss": 0.6937,
      "step": 126
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.983187644737843e-05,
      "loss": 0.6826,
      "step": 127
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.982781260883765e-05,
      "loss": 0.6342,
      "step": 128
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9823700666456854e-05,
      "loss": 0.6551,
      "step": 129
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9819540640362612e-05,
      "loss": 0.6724,
      "step": 130
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9815332550916864e-05,
      "loss": 0.687,
      "step": 131
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.981107641871678e-05,
      "loss": 0.6661,
      "step": 132
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.980677226459471e-05,
      "loss": 0.6977,
      "step": 133
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.980242010961803e-05,
      "loss": 0.7067,
      "step": 134
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9798019975089083e-05,
      "loss": 0.685,
      "step": 135
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9793571882545048e-05,
      "loss": 0.6531,
      "step": 136
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9789075853757845e-05,
      "loss": 0.6709,
      "step": 137
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.978453191073402e-05,
      "loss": 0.6399,
      "step": 138
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.977994007571465e-05,
      "loss": 0.6996,
      "step": 139
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9775300371175225e-05,
      "loss": 0.6513,
      "step": 140
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9770612819825534e-05,
      "loss": 0.692,
      "step": 141
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.976587744460957e-05,
      "loss": 0.5954,
      "step": 142
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.97610942687054e-05,
      "loss": 0.5954,
      "step": 143
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.975626331552507e-05,
      "loss": 0.7209,
      "step": 144
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9751384608714466e-05,
      "loss": 0.6777,
      "step": 145
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.974645817215322e-05,
      "loss": 0.6925,
      "step": 146
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9741484029954582e-05,
      "loss": 0.6353,
      "step": 147
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.973646220646531e-05,
      "loss": 0.6056,
      "step": 148
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9731392726265538e-05,
      "loss": 0.7167,
      "step": 149
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.972627561416867e-05,
      "loss": 0.7169,
      "step": 150
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9721110895221252e-05,
      "loss": 0.6759,
      "step": 151
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9715898594702845e-05,
      "loss": 0.663,
      "step": 152
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9710638738125908e-05,
      "loss": 0.6235,
      "step": 153
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9705331351235673e-05,
      "loss": 0.5991,
      "step": 154
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9699976460010016e-05,
      "loss": 0.6922,
      "step": 155
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.969457409065933e-05,
      "loss": 0.6976,
      "step": 156
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9689124269626393e-05,
      "loss": 0.6361,
      "step": 157
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9683627023586252e-05,
      "loss": 0.6306,
      "step": 158
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.967808237944608e-05,
      "loss": 0.6463,
      "step": 159
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.967249036434504e-05,
      "loss": 0.6487,
      "step": 160
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9666851005654174e-05,
      "loss": 0.6386,
      "step": 161
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9661164330976246e-05,
      "loss": 0.6127,
      "step": 162
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.965543036814561e-05,
      "loss": 0.6314,
      "step": 163
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.96496491452281e-05,
      "loss": 0.6329,
      "step": 164
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.964382069052085e-05,
      "loss": 0.6284,
      "step": 165
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.963794503255219e-05,
      "loss": 0.639,
      "step": 166
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9632022200081496e-05,
      "loss": 0.6172,
      "step": 167
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.962605222209903e-05,
      "loss": 0.6282,
      "step": 168
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.962003512782584e-05,
      "loss": 0.6409,
      "step": 169
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9613970946713574e-05,
      "loss": 0.5883,
      "step": 170
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9607859708444364e-05,
      "loss": 0.6287,
      "step": 171
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9601701442930667e-05,
      "loss": 0.6535,
      "step": 172
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.959549618031512e-05,
      "loss": 0.6392,
      "step": 173
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.95892439509704e-05,
      "loss": 0.6127,
      "step": 174
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9582944785499076e-05,
      "loss": 0.6815,
      "step": 175
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9576598714733434e-05,
      "loss": 0.5827,
      "step": 176
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.957020576973536e-05,
      "loss": 0.6628,
      "step": 177
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.956376598179618e-05,
      "loss": 0.6723,
      "step": 178
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9557279382436483e-05,
      "loss": 0.6952,
      "step": 179
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9550746003405996e-05,
      "loss": 0.7105,
      "step": 180
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9544165876683413e-05,
      "loss": 0.6247,
      "step": 181
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9537539034476247e-05,
      "loss": 0.6238,
      "step": 182
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.953086550922066e-05,
      "loss": 0.6857,
      "step": 183
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9524145333581315e-05,
      "loss": 0.6661,
      "step": 184
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9517378540451215e-05,
      "loss": 0.6574,
      "step": 185
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 0.6571,
      "step": 186
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9503705234431477e-05,
      "loss": 0.6454,
      "step": 187
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9496798788468077e-05,
      "loss": 0.6569,
      "step": 188
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9489845858866066e-05,
      "loss": 0.572,
      "step": 189
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9482846479657707e-05,
      "loss": 0.6289,
      "step": 190
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.94758006851026e-05,
      "loss": 0.6365,
      "step": 191
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9468708509687545e-05,
      "loss": 0.6705,
      "step": 192
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.946156998812636e-05,
      "loss": 0.6304,
      "step": 193
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9454385155359704e-05,
      "loss": 0.6822,
      "step": 194
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9447154046554923e-05,
      "loss": 0.6451,
      "step": 195
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.943987669710586e-05,
      "loss": 0.6208,
      "step": 196
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9432553142632704e-05,
      "loss": 0.6142,
      "step": 197
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9425183418981783e-05,
      "loss": 0.5998,
      "step": 198
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9417767562225422e-05,
      "loss": 0.6414,
      "step": 199
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9410305608661745e-05,
      "loss": 0.6145,
      "step": 200
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9402797594814517e-05,
      "loss": 0.5703,
      "step": 201
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.939524355743293e-05,
      "loss": 0.5993,
      "step": 202
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9387643533491462e-05,
      "loss": 0.5606,
      "step": 203
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9379997560189677e-05,
      "loss": 0.5842,
      "step": 204
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.937230567495205e-05,
      "loss": 0.6151,
      "step": 205
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.936456791542776e-05,
      "loss": 0.6354,
      "step": 206
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.935678431949056e-05,
      "loss": 0.6016,
      "step": 207
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.934895492523852e-05,
      "loss": 0.6464,
      "step": 208
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.93410797709939e-05,
      "loss": 0.6306,
      "step": 209
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9333158895302934e-05,
      "loss": 0.712,
      "step": 210
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9325192336935646e-05,
      "loss": 0.6244,
      "step": 211
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.931718013488566e-05,
      "loss": 0.6237,
      "step": 212
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9309122328370013e-05,
      "loss": 0.5864,
      "step": 213
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9301018956828966e-05,
      "loss": 0.6459,
      "step": 214
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.929287005992579e-05,
      "loss": 0.6109,
      "step": 215
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.92846756775466e-05,
      "loss": 0.606,
      "step": 216
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.927643584980015e-05,
      "loss": 0.6363,
      "step": 217
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.926815061701762e-05,
      "loss": 0.6476,
      "step": 218
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9259820019752445e-05,
      "loss": 0.5801,
      "step": 219
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9251444098780095e-05,
      "loss": 0.6469,
      "step": 220
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.924302289509789e-05,
      "loss": 0.5569,
      "step": 221
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9234556449924796e-05,
      "loss": 0.5868,
      "step": 222
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9226044804701214e-05,
      "loss": 0.6507,
      "step": 223
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9217488001088784e-05,
      "loss": 0.6277,
      "step": 224
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9208886080970187e-05,
      "loss": 0.6239,
      "step": 225
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9200239086448934e-05,
      "loss": 0.624,
      "step": 226
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9191547059849155e-05,
      "loss": 0.624,
      "step": 227
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9182810043715394e-05,
      "loss": 0.6085,
      "step": 228
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9174028080812415e-05,
      "loss": 0.6327,
      "step": 229
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9165201214124972e-05,
      "loss": 0.5803,
      "step": 230
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9156329486857615e-05,
      "loss": 0.6179,
      "step": 231
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9147412942434463e-05,
      "loss": 0.5817,
      "step": 232
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9138451624499007e-05,
      "loss": 0.6135,
      "step": 233
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9129445576913886e-05,
      "loss": 0.5763,
      "step": 234
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.912039484376068e-05,
      "loss": 0.5978,
      "step": 235
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.911129946933968e-05,
      "loss": 0.6477,
      "step": 236
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.910215949816969e-05,
      "loss": 0.6291,
      "step": 237
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.90929749749878e-05,
      "loss": 0.6731,
      "step": 238
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9083745944749163e-05,
      "loss": 0.7298,
      "step": 239
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.907447245262678e-05,
      "loss": 0.6642,
      "step": 240
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9065154544011273e-05,
      "loss": 0.6143,
      "step": 241
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9055792264510683e-05,
      "loss": 0.6343,
      "step": 242
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9046385659950213e-05,
      "loss": 0.6532,
      "step": 243
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.903693477637204e-05,
      "loss": 0.6067,
      "step": 244
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.902743966003506e-05,
      "loss": 0.5724,
      "step": 245
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.901790035741467e-05,
      "loss": 0.5935,
      "step": 246
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9008316915202555e-05,
      "loss": 0.646,
      "step": 247
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.899868938030645e-05,
      "loss": 0.6535,
      "step": 248
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.8989017799849896e-05,
      "loss": 0.6427,
      "step": 249
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.8979302221172027e-05,
      "loss": 0.6327,
      "step": 250
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.8969542691827346e-05,
      "loss": 0.6216,
      "step": 251
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8959739259585458e-05,
      "loss": 0.6665,
      "step": 252
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8949891972430878e-05,
      "loss": 0.53,
      "step": 253
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8940000878562758e-05,
      "loss": 0.5817,
      "step": 254
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.893006602639469e-05,
      "loss": 0.6036,
      "step": 255
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8920087464554427e-05,
      "loss": 0.6375,
      "step": 256
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.891006524188368e-05,
      "loss": 0.5278,
      "step": 257
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.889999940743786e-05,
      "loss": 0.5926,
      "step": 258
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8889890010485847e-05,
      "loss": 0.6125,
      "step": 259
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.887973710050974e-05,
      "loss": 0.6059,
      "step": 260
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8869540727204615e-05,
      "loss": 0.6221,
      "step": 261
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8859300940478304e-05,
      "loss": 0.5928,
      "step": 262
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8849017790451114e-05,
      "loss": 0.5739,
      "step": 263
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.883869132745561e-05,
      "loss": 0.5876,
      "step": 264
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8828321602036363e-05,
      "loss": 0.5671,
      "step": 265
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.881790866494969e-05,
      "loss": 0.616,
      "step": 266
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8807452567163413e-05,
      "loss": 0.5381,
      "step": 267
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8796953359856626e-05,
      "loss": 0.5394,
      "step": 268
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.878641109441942e-05,
      "loss": 0.5925,
      "step": 269
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8775825822452636e-05,
      "loss": 0.624,
      "step": 270
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8765197595767623e-05,
      "loss": 0.6122,
      "step": 271
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8754526466385986e-05,
      "loss": 0.6139,
      "step": 272
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.874381248653931e-05,
      "loss": 0.6111,
      "step": 273
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8733055708668928e-05,
      "loss": 0.5524,
      "step": 274
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8722256185425654e-05,
      "loss": 0.5514,
      "step": 275
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.871141396966953e-05,
      "loss": 0.6109,
      "step": 276
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8700529114469545e-05,
      "loss": 0.6452,
      "step": 277
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.868960167310342e-05,
      "loss": 0.5266,
      "step": 278
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.86786316990573e-05,
      "loss": 0.5495,
      "step": 279
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8667619246025527e-05,
      "loss": 0.588,
      "step": 280
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8656564367910358e-05,
      "loss": 0.5112,
      "step": 281
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.86454671188217e-05,
      "loss": 0.6438,
      "step": 282
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8634327553076864e-05,
      "loss": 0.5494,
      "step": 283
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.862314572520028e-05,
      "loss": 0.5849,
      "step": 284
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8611921689923234e-05,
      "loss": 0.5857,
      "step": 285
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.860065550218361e-05,
      "loss": 0.6192,
      "step": 286
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.858934721712561e-05,
      "loss": 0.4881,
      "step": 287
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8577996890099494e-05,
      "loss": 0.541,
      "step": 288
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8566604576661288e-05,
      "loss": 0.52,
      "step": 289
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8555170332572544e-05,
      "loss": 0.5997,
      "step": 290
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8543694213800044e-05,
      "loss": 0.5964,
      "step": 291
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8532176276515538e-05,
      "loss": 0.5221,
      "step": 292
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8520616577095447e-05,
      "loss": 0.5878,
      "step": 293
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.850901517212062e-05,
      "loss": 0.5868,
      "step": 294
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8497372118376043e-05,
      "loss": 0.5946,
      "step": 295
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8485687472850538e-05,
      "loss": 0.6517,
      "step": 296
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.847396129273653e-05,
      "loss": 0.5609,
      "step": 297
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8462193635429723e-05,
      "loss": 0.5922,
      "step": 298
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8450384558528848e-05,
      "loss": 0.5165,
      "step": 299
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8438534119835365e-05,
      "loss": 0.5359,
      "step": 300
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.842664237735319e-05,
      "loss": 0.607,
      "step": 301
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.841470938928841e-05,
      "loss": 0.5162,
      "step": 302
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8402735214048988e-05,
      "loss": 0.5648,
      "step": 303
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8390719910244487e-05,
      "loss": 0.5554,
      "step": 304
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8378663536685784e-05,
      "loss": 0.5435,
      "step": 305
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8366566152384773e-05,
      "loss": 0.5684,
      "step": 306
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8354427816554085e-05,
      "loss": 0.6131,
      "step": 307
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.83422485886068e-05,
      "loss": 0.5604,
      "step": 308
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8330028528156138e-05,
      "loss": 0.5868,
      "step": 309
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8317767695015195e-05,
      "loss": 0.5601,
      "step": 310
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8305466149196624e-05,
      "loss": 0.646,
      "step": 311
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8293123950912362e-05,
      "loss": 0.5813,
      "step": 312
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8280741160573318e-05,
      "loss": 0.6303,
      "step": 313
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.826831783878909e-05,
      "loss": 0.5233,
      "step": 314
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.825585404636766e-05,
      "loss": 0.5722,
      "step": 315
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8243349844315116e-05,
      "loss": 0.6125,
      "step": 316
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8230805293835315e-05,
      "loss": 0.5913,
      "step": 317
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8218220456329618e-05,
      "loss": 0.5453,
      "step": 318
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.820559539339657e-05,
      "loss": 0.6059,
      "step": 319
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8192930166831618e-05,
      "loss": 0.5906,
      "step": 320
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.818022483862679e-05,
      "loss": 0.5697,
      "step": 321
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8167479470970394e-05,
      "loss": 0.5952,
      "step": 322
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8154694126246725e-05,
      "loss": 0.5524,
      "step": 323
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8141868867035745e-05,
      "loss": 0.5772,
      "step": 324
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.812900375611279e-05,
      "loss": 0.5916,
      "step": 325
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8116098856448256e-05,
      "loss": 0.5707,
      "step": 326
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8103154231207286e-05,
      "loss": 0.5902,
      "step": 327
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.6478,
      "step": 328
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8077146057628547e-05,
      "loss": 0.5788,
      "step": 329
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8064082636592043e-05,
      "loss": 0.5773,
      "step": 330
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8050979744581022e-05,
      "loss": 0.5594,
      "step": 331
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8037837445729733e-05,
      "loss": 0.5812,
      "step": 332
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.802465580436532e-05,
      "loss": 0.5748,
      "step": 333
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8011434885007482e-05,
      "loss": 0.5564,
      "step": 334
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.7998174752368178e-05,
      "loss": 0.6678,
      "step": 335
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.79848754713513e-05,
      "loss": 0.6761,
      "step": 336
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.7971537107052364e-05,
      "loss": 0.5973,
      "step": 337
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.7958159724758173e-05,
      "loss": 0.5899,
      "step": 338
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.7944743389946524e-05,
      "loss": 0.5607,
      "step": 339
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7931288168285863e-05,
      "loss": 0.5449,
      "step": 340
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7917794125634983e-05,
      "loss": 0.6624,
      "step": 341
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.790426132804268e-05,
      "loss": 0.6022,
      "step": 342
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7890689841747463e-05,
      "loss": 0.5466,
      "step": 343
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7877079733177185e-05,
      "loss": 0.5806,
      "step": 344
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7863431068948757e-05,
      "loss": 0.5916,
      "step": 345
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.784974391586781e-05,
      "loss": 0.5027,
      "step": 346
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7836018340928348e-05,
      "loss": 0.5486,
      "step": 347
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7822254411312455e-05,
      "loss": 0.5552,
      "step": 348
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.780845219438994e-05,
      "loss": 0.6191,
      "step": 349
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7794611757718013e-05,
      "loss": 0.5775,
      "step": 350
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7780733169040964e-05,
      "loss": 0.5991,
      "step": 351
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7766816496289822e-05,
      "loss": 0.6355,
      "step": 352
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.775286180758202e-05,
      "loss": 0.5947,
      "step": 353
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.773886917122107e-05,
      "loss": 0.6049,
      "step": 354
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7724838655696225e-05,
      "loss": 0.5597,
      "step": 355
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7710770329682145e-05,
      "loss": 0.6115,
      "step": 356
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7696664262038554e-05,
      "loss": 0.5728,
      "step": 357
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7682520521809918e-05,
      "loss": 0.5278,
      "step": 358
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.766833917822509e-05,
      "loss": 0.5627,
      "step": 359
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.765412030069698e-05,
      "loss": 0.6724,
      "step": 360
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7639863958822213e-05,
      "loss": 0.5225,
      "step": 361
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.76255702223808e-05,
      "loss": 0.551,
      "step": 362
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.761123916133577e-05,
      "loss": 0.5401,
      "step": 363
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.759687084583285e-05,
      "loss": 0.6538,
      "step": 364
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.758246534620012e-05,
      "loss": 0.5547,
      "step": 365
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.756802273294766e-05,
      "loss": 0.6067,
      "step": 366
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7553543076767207e-05,
      "loss": 0.5354,
      "step": 367
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.753902644853181e-05,
      "loss": 0.5518,
      "step": 368
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7524472919295488e-05,
      "loss": 0.5829,
      "step": 369
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.750988256029287e-05,
      "loss": 0.5488,
      "step": 370
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7495255442938873e-05,
      "loss": 0.5554,
      "step": 371
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.748059163882831e-05,
      "loss": 0.5965,
      "step": 372
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7465891219735576e-05,
      "loss": 0.5826,
      "step": 373
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7451154257614287e-05,
      "loss": 0.4955,
      "step": 374
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7436380824596915e-05,
      "loss": 0.5438,
      "step": 375
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.742157099299445e-05,
      "loss": 0.5584,
      "step": 376
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7406724835296045e-05,
      "loss": 0.5441,
      "step": 377
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7391842424168652e-05,
      "loss": 0.611,
      "step": 378
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7376923832456665e-05,
      "loss": 0.5766,
      "step": 379
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7361969133181585e-05,
      "loss": 0.5686,
      "step": 380
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7346978399541638e-05,
      "loss": 0.4756,
      "step": 381
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7331951704911426e-05,
      "loss": 0.5605,
      "step": 382
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.731688912284157e-05,
      "loss": 0.5667,
      "step": 383
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7301790727058344e-05,
      "loss": 0.545,
      "step": 384
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7286656591463325e-05,
      "loss": 0.5494,
      "step": 385
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7271486790133022e-05,
      "loss": 0.5661,
      "step": 386
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.725628139731851e-05,
      "loss": 0.5165,
      "step": 387
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7241040487445082e-05,
      "loss": 0.5775,
      "step": 388
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7225764135111867e-05,
      "loss": 0.5586,
      "step": 389
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7210452415091477e-05,
      "loss": 0.515,
      "step": 390
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7195105402329636e-05,
      "loss": 0.5611,
      "step": 391
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7179723171944812e-05,
      "loss": 0.5666,
      "step": 392
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.716430579922786e-05,
      "loss": 0.6011,
      "step": 393
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7148853359641627e-05,
      "loss": 0.588,
      "step": 394
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.713336592882062e-05,
      "loss": 0.5629,
      "step": 395
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7117843582570608e-05,
      "loss": 0.5584,
      "step": 396
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7102286396868254e-05,
      "loss": 0.5626,
      "step": 397
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.708669444786076e-05,
      "loss": 0.5522,
      "step": 398
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.5697,
      "step": 399
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.705540656536953e-05,
      "loss": 0.5854,
      "step": 400
    }
  ],
  "max_steps": 1464,
  "num_train_epochs": 3,
  "total_flos": 1.0973812951403725e+17,
  "trial_name": null,
  "trial_params": null
}