{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.259780907668231,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 9.992088607594937e-05,
      "loss": 3.0081,
      "step": 1
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.984177215189874e-05,
      "loss": 2.8615,
      "step": 2
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.97626582278481e-05,
      "loss": 2.8922,
      "step": 3
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.968354430379747e-05,
      "loss": 2.7182,
      "step": 4
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.960443037974683e-05,
      "loss": 2.8407,
      "step": 5
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.952531645569621e-05,
      "loss": 2.7972,
      "step": 6
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.944620253164557e-05,
      "loss": 2.7558,
      "step": 7
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.936708860759493e-05,
      "loss": 2.8176,
      "step": 8
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.928797468354431e-05,
      "loss": 2.7444,
      "step": 9
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.920886075949367e-05,
      "loss": 2.8396,
      "step": 10
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.912974683544304e-05,
      "loss": 2.7709,
      "step": 11
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.90506329113924e-05,
      "loss": 2.9788,
      "step": 12
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.897151898734177e-05,
      "loss": 2.8087,
      "step": 13
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.889240506329115e-05,
      "loss": 2.6894,
      "step": 14
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.881329113924051e-05,
      "loss": 2.7998,
      "step": 15
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.873417721518988e-05,
      "loss": 2.747,
      "step": 16
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.865506329113925e-05,
      "loss": 2.788,
      "step": 17
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.857594936708862e-05,
      "loss": 2.7536,
      "step": 18
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.849683544303798e-05,
      "loss": 2.7604,
      "step": 19
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.841772151898735e-05,
      "loss": 2.6864,
      "step": 20
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.833860759493672e-05,
      "loss": 2.7934,
      "step": 21
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.825949367088608e-05,
      "loss": 2.6659,
      "step": 22
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.818037974683544e-05,
      "loss": 2.7234,
      "step": 23
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.810126582278482e-05,
      "loss": 2.7003,
      "step": 24
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.802215189873418e-05,
      "loss": 2.828,
      "step": 25
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.794303797468355e-05,
      "loss": 2.668,
      "step": 26
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.786392405063292e-05,
      "loss": 2.7377,
      "step": 27
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.778481012658228e-05,
      "loss": 2.7635,
      "step": 28
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.770569620253165e-05,
      "loss": 2.7134,
      "step": 29
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.762658227848101e-05,
      "loss": 2.6454,
      "step": 30
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.754746835443039e-05,
      "loss": 2.6397,
      "step": 31
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.746835443037975e-05,
      "loss": 2.7034,
      "step": 32
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.738924050632911e-05,
      "loss": 2.7257,
      "step": 33
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.731012658227849e-05,
      "loss": 2.6564,
      "step": 34
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.723101265822785e-05,
      "loss": 2.7504,
      "step": 35
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.715189873417721e-05,
      "loss": 2.597,
      "step": 36
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.707278481012658e-05,
      "loss": 2.6232,
      "step": 37
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.699367088607595e-05,
      "loss": 2.6682,
      "step": 38
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.691455696202532e-05,
      "loss": 2.6603,
      "step": 39
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.683544303797469e-05,
      "loss": 2.6109,
      "step": 40
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.675632911392405e-05,
      "loss": 2.6726,
      "step": 41
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.667721518987343e-05,
      "loss": 2.7147,
      "step": 42
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.65981012658228e-05,
      "loss": 2.6925,
      "step": 43
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.651898734177216e-05,
      "loss": 2.7408,
      "step": 44
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.643987341772153e-05,
      "loss": 2.7109,
      "step": 45
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.63607594936709e-05,
      "loss": 2.732,
      "step": 46
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.628164556962026e-05,
      "loss": 2.7063,
      "step": 47
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.620253164556962e-05,
      "loss": 2.7112,
      "step": 48
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.6123417721519e-05,
      "loss": 2.5878,
      "step": 49
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.604430379746836e-05,
      "loss": 2.6371,
      "step": 50
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.596518987341772e-05,
      "loss": 2.6441,
      "step": 51
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.58860759493671e-05,
      "loss": 2.6282,
      "step": 52
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.580696202531646e-05,
      "loss": 2.608,
      "step": 53
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.572784810126582e-05,
      "loss": 2.6978,
      "step": 54
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.564873417721519e-05,
      "loss": 2.7785,
      "step": 55
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.556962025316456e-05,
      "loss": 2.6797,
      "step": 56
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.549050632911393e-05,
      "loss": 2.7262,
      "step": 57
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.541139240506329e-05,
      "loss": 2.6339,
      "step": 58
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.533227848101267e-05,
      "loss": 2.6378,
      "step": 59
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.525316455696203e-05,
      "loss": 2.7182,
      "step": 60
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.517405063291139e-05,
      "loss": 2.6166,
      "step": 61
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.509493670886075e-05,
      "loss": 2.7077,
      "step": 62
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.501582278481013e-05,
      "loss": 2.6853,
      "step": 63
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.493670886075949e-05,
      "loss": 2.7928,
      "step": 64
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.485759493670886e-05,
      "loss": 2.7323,
      "step": 65
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.477848101265823e-05,
      "loss": 2.6468,
      "step": 66
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.469936708860761e-05,
      "loss": 2.6338,
      "step": 67
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.462025316455697e-05,
      "loss": 2.6622,
      "step": 68
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.454113924050633e-05,
      "loss": 2.679,
      "step": 69
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.446202531645571e-05,
      "loss": 2.7418,
      "step": 70
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.438291139240507e-05,
      "loss": 2.6329,
      "step": 71
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.430379746835444e-05,
      "loss": 2.7766,
      "step": 72
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.42246835443038e-05,
      "loss": 2.6382,
      "step": 73
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.414556962025317e-05,
      "loss": 2.7261,
      "step": 74
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.406645569620254e-05,
      "loss": 2.732,
      "step": 75
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.39873417721519e-05,
      "loss": 2.7585,
      "step": 76
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.390822784810128e-05,
      "loss": 2.7085,
      "step": 77
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.382911392405064e-05,
      "loss": 2.7088,
      "step": 78
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.375e-05,
      "loss": 2.6059,
      "step": 79
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.367088607594936e-05,
      "loss": 2.6854,
      "step": 80
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.359177215189874e-05,
      "loss": 2.5747,
      "step": 81
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.35126582278481e-05,
      "loss": 2.695,
      "step": 82
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.343354430379747e-05,
      "loss": 2.6193,
      "step": 83
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.335443037974684e-05,
      "loss": 2.5892,
      "step": 84
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.32753164556962e-05,
      "loss": 2.5214,
      "step": 85
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.319620253164557e-05,
      "loss": 2.6025,
      "step": 86
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.311708860759493e-05,
      "loss": 2.6703,
      "step": 87
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.303797468354431e-05,
      "loss": 2.6365,
      "step": 88
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.295886075949367e-05,
      "loss": 2.5981,
      "step": 89
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.287974683544303e-05,
      "loss": 2.6546,
      "step": 90
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.280063291139241e-05,
      "loss": 2.6226,
      "step": 91
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.272151898734177e-05,
      "loss": 2.5896,
      "step": 92
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.264240506329115e-05,
      "loss": 2.6591,
      "step": 93
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.256329113924051e-05,
      "loss": 2.5707,
      "step": 94
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.248417721518989e-05,
      "loss": 2.781,
      "step": 95
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.240506329113925e-05,
      "loss": 2.5795,
      "step": 96
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.232594936708861e-05,
      "loss": 2.6313,
      "step": 97
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.224683544303798e-05,
      "loss": 2.683,
      "step": 98
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.216772151898735e-05,
      "loss": 2.6623,
      "step": 99
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.208860759493671e-05,
      "loss": 2.7411,
      "step": 100
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.200949367088608e-05,
      "loss": 2.6245,
      "step": 101
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.193037974683545e-05,
      "loss": 2.6614,
      "step": 102
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.185126582278482e-05,
      "loss": 2.6445,
      "step": 103
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.177215189873418e-05,
      "loss": 2.5834,
      "step": 104
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.169303797468354e-05,
      "loss": 2.6261,
      "step": 105
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.161392405063292e-05,
      "loss": 2.6625,
      "step": 106
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.153481012658228e-05,
      "loss": 2.6882,
      "step": 107
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.145569620253164e-05,
      "loss": 2.6707,
      "step": 108
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.137658227848102e-05,
      "loss": 2.6243,
      "step": 109
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.129746835443038e-05,
      "loss": 2.6536,
      "step": 110
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.121835443037975e-05,
      "loss": 2.6165,
      "step": 111
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.113924050632912e-05,
      "loss": 2.6707,
      "step": 112
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.106012658227849e-05,
      "loss": 2.5652,
      "step": 113
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.098101265822785e-05,
      "loss": 2.6854,
      "step": 114
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.090189873417721e-05,
      "loss": 2.6967,
      "step": 115
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.082278481012659e-05,
      "loss": 2.609,
      "step": 116
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.074367088607595e-05,
      "loss": 2.6294,
      "step": 117
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.066455696202531e-05,
      "loss": 2.6093,
      "step": 118
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.058544303797469e-05,
      "loss": 2.6536,
      "step": 119
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.050632911392407e-05,
      "loss": 2.6064,
      "step": 120
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.042721518987343e-05,
      "loss": 2.5646,
      "step": 121
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.034810126582279e-05,
      "loss": 2.68,
      "step": 122
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.026898734177215e-05,
      "loss": 2.6302,
      "step": 123
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.018987341772153e-05,
      "loss": 2.6241,
      "step": 124
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.011075949367089e-05,
      "loss": 2.5816,
      "step": 125
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.003164556962026e-05,
      "loss": 2.6094,
      "step": 126
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.995253164556963e-05,
      "loss": 2.5592,
      "step": 127
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.9873417721519e-05,
      "loss": 2.7118,
      "step": 128
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.979430379746836e-05,
      "loss": 2.6414,
      "step": 129
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.971518987341772e-05,
      "loss": 2.644,
      "step": 130
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.96360759493671e-05,
      "loss": 2.6476,
      "step": 131
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.955696202531646e-05,
      "loss": 2.6763,
      "step": 132
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.947784810126582e-05,
      "loss": 2.6349,
      "step": 133
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.93987341772152e-05,
      "loss": 2.6594,
      "step": 134
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.931962025316456e-05,
      "loss": 2.7204,
      "step": 135
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.924050632911392e-05,
      "loss": 2.6037,
      "step": 136
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.91613924050633e-05,
      "loss": 2.5948,
      "step": 137
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.908227848101266e-05,
      "loss": 2.653,
      "step": 138
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.900316455696203e-05,
      "loss": 2.6549,
      "step": 139
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.892405063291139e-05,
      "loss": 2.602,
      "step": 140
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.884493670886076e-05,
      "loss": 2.5651,
      "step": 141
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.876582278481013e-05,
      "loss": 2.6517,
      "step": 142
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.868670886075949e-05,
      "loss": 2.6373,
      "step": 143
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.860759493670887e-05,
      "loss": 2.6907,
      "step": 144
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.852848101265824e-05,
      "loss": 2.5885,
      "step": 145
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.84493670886076e-05,
      "loss": 2.579,
      "step": 146
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.837025316455697e-05,
      "loss": 2.7174,
      "step": 147
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.829113924050633e-05,
      "loss": 2.6545,
      "step": 148
    },
    {
      "epoch": 1.87,
      "learning_rate": 8.821202531645571e-05,
      "loss": 2.659,
      "step": 149
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.813291139240507e-05,
      "loss": 2.6381,
      "step": 150
    },
    {
      "epoch": 1.89,
      "learning_rate": 8.805379746835443e-05,
      "loss": 2.595,
      "step": 151
    },
    {
      "epoch": 1.9,
      "learning_rate": 8.797468354430381e-05,
      "loss": 2.6264,
      "step": 152
    },
    {
      "epoch": 1.92,
      "learning_rate": 8.789556962025317e-05,
      "loss": 2.6035,
      "step": 153
    },
    {
      "epoch": 1.93,
      "learning_rate": 8.781645569620253e-05,
      "loss": 2.6478,
      "step": 154
    },
    {
      "epoch": 1.94,
      "learning_rate": 8.773734177215191e-05,
      "loss": 2.6257,
      "step": 155
    },
    {
      "epoch": 1.95,
      "learning_rate": 8.765822784810127e-05,
      "loss": 2.5814,
      "step": 156
    },
    {
      "epoch": 1.97,
      "learning_rate": 8.757911392405064e-05,
      "loss": 2.585,
      "step": 157
    },
    {
      "epoch": 1.98,
      "learning_rate": 8.75e-05,
      "loss": 2.6877,
      "step": 158
    },
    {
      "epoch": 1.99,
      "learning_rate": 8.742088607594938e-05,
      "loss": 2.5594,
      "step": 159
    },
    {
      "epoch": 2.0,
      "learning_rate": 8.734177215189874e-05,
      "loss": 2.665,
      "step": 160
    },
    {
      "epoch": 2.02,
      "learning_rate": 8.72626582278481e-05,
      "loss": 2.6302,
      "step": 161
    },
    {
      "epoch": 2.03,
      "learning_rate": 8.718354430379748e-05,
      "loss": 2.5499,
      "step": 162
    },
    {
      "epoch": 2.04,
      "learning_rate": 8.710443037974684e-05,
      "loss": 2.5304,
      "step": 163
    },
    {
      "epoch": 2.05,
      "learning_rate": 8.70253164556962e-05,
      "loss": 2.5099,
      "step": 164
    },
    {
      "epoch": 2.07,
      "learning_rate": 8.694620253164557e-05,
      "loss": 2.5894,
      "step": 165
    },
    {
      "epoch": 2.08,
      "learning_rate": 8.686708860759494e-05,
      "loss": 2.5542,
      "step": 166
    },
    {
      "epoch": 2.09,
      "learning_rate": 8.67879746835443e-05,
      "loss": 2.5983,
      "step": 167
    },
    {
      "epoch": 2.1,
      "learning_rate": 8.670886075949367e-05,
      "loss": 2.6068,
      "step": 168
    },
    {
      "epoch": 2.12,
      "learning_rate": 8.662974683544304e-05,
      "loss": 2.6554,
      "step": 169
    },
    {
      "epoch": 2.13,
      "learning_rate": 8.65506329113924e-05,
      "loss": 2.6135,
      "step": 170
    },
    {
      "epoch": 2.14,
      "learning_rate": 8.647151898734177e-05,
      "loss": 2.5338,
      "step": 171
    },
    {
      "epoch": 2.15,
      "learning_rate": 8.639240506329115e-05,
      "loss": 2.5368,
      "step": 172
    },
    {
      "epoch": 2.17,
      "learning_rate": 8.631329113924052e-05,
      "loss": 2.5368,
      "step": 173
    },
    {
      "epoch": 2.18,
      "learning_rate": 8.623417721518988e-05,
      "loss": 2.5696,
      "step": 174
    },
    {
      "epoch": 2.19,
      "learning_rate": 8.615506329113925e-05,
      "loss": 2.5253,
      "step": 175
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.607594936708861e-05,
      "loss": 2.4822,
      "step": 176
    },
    {
      "epoch": 2.22,
      "learning_rate": 8.599683544303799e-05,
      "loss": 2.5394,
      "step": 177
    },
    {
      "epoch": 2.23,
      "learning_rate": 8.591772151898735e-05,
      "loss": 2.6708,
      "step": 178
    },
    {
      "epoch": 2.24,
      "learning_rate": 8.583860759493671e-05,
      "loss": 2.5803,
      "step": 179
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.575949367088609e-05,
      "loss": 2.5357,
      "step": 180
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.568037974683545e-05,
      "loss": 2.5887,
      "step": 181
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.560126582278481e-05,
      "loss": 2.6092,
      "step": 182
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.552215189873418e-05,
      "loss": 2.5351,
      "step": 183
    },
    {
      "epoch": 2.3,
      "learning_rate": 8.544303797468355e-05,
      "loss": 2.518,
      "step": 184
    },
    {
      "epoch": 2.32,
      "learning_rate": 8.536392405063292e-05,
      "loss": 2.6434,
      "step": 185
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.528481012658228e-05,
      "loss": 2.5951,
      "step": 186
    },
    {
      "epoch": 2.34,
      "learning_rate": 8.520569620253165e-05,
      "loss": 2.6177,
      "step": 187
    },
    {
      "epoch": 2.35,
      "learning_rate": 8.512658227848102e-05,
      "loss": 2.5993,
      "step": 188
    },
    {
      "epoch": 2.37,
      "learning_rate": 8.504746835443038e-05,
      "loss": 2.487,
      "step": 189
    },
    {
      "epoch": 2.38,
      "learning_rate": 8.496835443037974e-05,
      "loss": 2.5761,
      "step": 190
    },
    {
      "epoch": 2.39,
      "learning_rate": 8.488924050632912e-05,
      "loss": 2.599,
      "step": 191
    },
    {
      "epoch": 2.4,
      "learning_rate": 8.481012658227848e-05,
      "loss": 2.5772,
      "step": 192
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.473101265822784e-05,
      "loss": 2.5415,
      "step": 193
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.465189873417722e-05,
      "loss": 2.5598,
      "step": 194
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.457278481012658e-05,
      "loss": 2.5532,
      "step": 195
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.449367088607595e-05,
      "loss": 2.539,
      "step": 196
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.441455696202531e-05,
      "loss": 2.5363,
      "step": 197
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.43354430379747e-05,
      "loss": 2.5749,
      "step": 198
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.425632911392406e-05,
      "loss": 2.5694,
      "step": 199
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.417721518987342e-05,
      "loss": 2.5436,
      "step": 200
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.409810126582279e-05,
      "loss": 2.5848,
      "step": 201
    },
    {
      "epoch": 2.53,
      "learning_rate": 8.401898734177216e-05,
      "loss": 2.6092,
      "step": 202
    },
    {
      "epoch": 2.54,
      "learning_rate": 8.393987341772153e-05,
      "loss": 2.6622,
      "step": 203
    },
    {
      "epoch": 2.55,
      "learning_rate": 8.386075949367089e-05,
      "loss": 2.5721,
      "step": 204
    },
    {
      "epoch": 2.57,
      "learning_rate": 8.378164556962027e-05,
      "loss": 2.6728,
      "step": 205
    },
    {
      "epoch": 2.58,
      "learning_rate": 8.370253164556963e-05,
      "loss": 2.6081,
      "step": 206
    },
    {
      "epoch": 2.59,
      "learning_rate": 8.362341772151899e-05,
      "loss": 2.6626,
      "step": 207
    },
    {
      "epoch": 2.6,
      "learning_rate": 8.354430379746835e-05,
      "loss": 2.6292,
      "step": 208
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.346518987341773e-05,
      "loss": 2.5524,
      "step": 209
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.33860759493671e-05,
      "loss": 2.5729,
      "step": 210
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.330696202531646e-05,
      "loss": 2.5565,
      "step": 211
    },
    {
      "epoch": 2.65,
      "learning_rate": 8.322784810126583e-05,
      "loss": 2.6039,
      "step": 212
    },
    {
      "epoch": 2.67,
      "learning_rate": 8.31487341772152e-05,
      "loss": 2.5557,
      "step": 213
    },
    {
      "epoch": 2.68,
      "learning_rate": 8.306962025316456e-05,
      "loss": 2.5331,
      "step": 214
    },
    {
      "epoch": 2.69,
      "learning_rate": 8.299050632911392e-05,
      "loss": 2.6137,
      "step": 215
    },
    {
      "epoch": 2.7,
      "learning_rate": 8.29113924050633e-05,
      "loss": 2.5428,
      "step": 216
    },
    {
      "epoch": 2.72,
      "learning_rate": 8.283227848101266e-05,
      "loss": 2.5655,
      "step": 217
    },
    {
      "epoch": 2.73,
      "learning_rate": 8.275316455696202e-05,
      "loss": 2.5869,
      "step": 218
    },
    {
      "epoch": 2.74,
      "learning_rate": 8.26740506329114e-05,
      "loss": 2.6131,
      "step": 219
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.259493670886076e-05,
      "loss": 2.5545,
      "step": 220
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.251582278481012e-05,
      "loss": 2.6437,
      "step": 221
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.243670886075949e-05,
      "loss": 2.5611,
      "step": 222
    },
    {
      "epoch": 2.79,
      "learning_rate": 8.235759493670886e-05,
      "loss": 2.5757,
      "step": 223
    },
    {
      "epoch": 2.8,
      "learning_rate": 8.227848101265824e-05,
      "loss": 2.6145,
      "step": 224
    },
    {
      "epoch": 2.82,
      "learning_rate": 8.21993670886076e-05,
      "loss": 2.573,
      "step": 225
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.212025316455697e-05,
      "loss": 2.6449,
      "step": 226
    },
    {
      "epoch": 2.84,
      "learning_rate": 8.204113924050634e-05,
      "loss": 2.5274,
      "step": 227
    },
    {
      "epoch": 2.85,
      "learning_rate": 8.19620253164557e-05,
      "loss": 2.5667,
      "step": 228
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.188291139240507e-05,
      "loss": 2.5355,
      "step": 229
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.180379746835444e-05,
      "loss": 2.58,
      "step": 230
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.17246835443038e-05,
      "loss": 2.6295,
      "step": 231
    },
    {
      "epoch": 2.9,
      "learning_rate": 8.164556962025317e-05,
      "loss": 2.5535,
      "step": 232
    },
    {
      "epoch": 2.92,
      "learning_rate": 8.156645569620253e-05,
      "loss": 2.6072,
      "step": 233
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.148734177215191e-05,
      "loss": 2.5914,
      "step": 234
    },
    {
      "epoch": 2.94,
      "learning_rate": 8.140822784810127e-05,
      "loss": 2.5315,
      "step": 235
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.132911392405063e-05,
      "loss": 2.5623,
      "step": 236
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.125000000000001e-05,
      "loss": 2.5596,
      "step": 237
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.117088607594937e-05,
      "loss": 2.563,
      "step": 238
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.109177215189874e-05,
      "loss": 2.5054,
      "step": 239
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.10126582278481e-05,
      "loss": 2.5953,
      "step": 240
    },
    {
      "epoch": 3.02,
      "learning_rate": 8.093354430379747e-05,
      "loss": 2.4561,
      "step": 241
    },
    {
      "epoch": 3.03,
      "learning_rate": 8.085443037974684e-05,
      "loss": 2.5268,
      "step": 242
    },
    {
      "epoch": 3.04,
      "learning_rate": 8.07753164556962e-05,
      "loss": 2.5851,
      "step": 243
    },
    {
      "epoch": 3.05,
      "learning_rate": 8.069620253164558e-05,
      "loss": 2.5286,
      "step": 244
    },
    {
      "epoch": 3.07,
      "learning_rate": 8.061708860759494e-05,
      "loss": 2.5408,
      "step": 245
    },
    {
      "epoch": 3.08,
      "learning_rate": 8.05379746835443e-05,
      "loss": 2.5685,
      "step": 246
    },
    {
      "epoch": 3.09,
      "learning_rate": 8.045886075949366e-05,
      "loss": 2.453,
      "step": 247
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.037974683544304e-05,
      "loss": 2.5055,
      "step": 248
    },
    {
      "epoch": 3.12,
      "learning_rate": 8.03006329113924e-05,
      "loss": 2.4933,
      "step": 249
    },
    {
      "epoch": 3.13,
      "learning_rate": 8.022151898734177e-05,
      "loss": 2.5857,
      "step": 250
    },
    {
      "epoch": 3.14,
      "learning_rate": 8.014240506329114e-05,
      "loss": 2.4926,
      "step": 251
    },
    {
      "epoch": 3.15,
      "learning_rate": 8.006329113924052e-05,
      "loss": 2.4991,
      "step": 252
    },
    {
      "epoch": 3.17,
      "learning_rate": 7.998417721518988e-05,
      "loss": 2.492,
      "step": 253
    },
    {
      "epoch": 3.18,
      "learning_rate": 7.990506329113924e-05,
      "loss": 2.5084,
      "step": 254
    },
    {
      "epoch": 3.19,
      "learning_rate": 7.982594936708862e-05,
      "loss": 2.5279,
      "step": 255
    },
    {
      "epoch": 3.21,
      "learning_rate": 7.974683544303798e-05,
      "loss": 2.4119,
      "step": 256
    },
    {
      "epoch": 3.22,
      "learning_rate": 7.966772151898735e-05,
      "loss": 2.496,
      "step": 257
    },
    {
      "epoch": 3.23,
      "learning_rate": 7.958860759493671e-05,
      "loss": 2.5748,
      "step": 258
    },
    {
      "epoch": 3.24,
      "learning_rate": 7.950949367088609e-05,
      "loss": 2.4665,
      "step": 259
    },
    {
      "epoch": 3.26,
      "learning_rate": 7.943037974683545e-05,
      "loss": 2.4504,
      "step": 260
    },
    {
      "epoch": 3.27,
      "learning_rate": 7.935126582278481e-05,
      "loss": 2.5703,
      "step": 261
    },
    {
      "epoch": 3.28,
      "learning_rate": 7.927215189873419e-05,
      "loss": 2.5075,
      "step": 262
    },
    {
      "epoch": 3.29,
      "learning_rate": 7.919303797468355e-05,
      "loss": 2.5279,
      "step": 263
    },
    {
      "epoch": 3.31,
      "learning_rate": 7.911392405063291e-05,
      "loss": 2.4899,
      "step": 264
    },
    {
      "epoch": 3.32,
      "learning_rate": 7.903481012658228e-05,
      "loss": 2.4601,
      "step": 265
    },
    {
      "epoch": 3.33,
      "learning_rate": 7.895569620253165e-05,
      "loss": 2.6405,
      "step": 266
    },
    {
      "epoch": 3.34,
      "learning_rate": 7.887658227848101e-05,
      "loss": 2.4824,
      "step": 267
    },
    {
      "epoch": 3.36,
      "learning_rate": 7.879746835443038e-05,
      "loss": 2.5388,
      "step": 268
    },
    {
      "epoch": 3.37,
      "learning_rate": 7.871835443037975e-05,
      "loss": 2.5799,
      "step": 269
    },
    {
      "epoch": 3.38,
      "learning_rate": 7.863924050632912e-05,
      "loss": 2.5451,
      "step": 270
    },
    {
      "epoch": 3.39,
      "learning_rate": 7.856012658227848e-05,
      "loss": 2.4391,
      "step": 271
    },
    {
      "epoch": 3.41,
      "learning_rate": 7.848101265822784e-05,
      "loss": 2.4616,
      "step": 272
    },
    {
      "epoch": 3.42,
      "learning_rate": 7.840189873417722e-05,
      "loss": 2.5493,
      "step": 273
    },
    {
      "epoch": 3.43,
      "learning_rate": 7.832278481012658e-05,
      "loss": 2.5831,
      "step": 274
    },
    {
      "epoch": 3.44,
      "learning_rate": 7.824367088607594e-05,
      "loss": 2.5137,
      "step": 275
    },
    {
      "epoch": 3.46,
      "learning_rate": 7.816455696202532e-05,
      "loss": 2.4994,
      "step": 276
    },
    {
      "epoch": 3.47,
      "learning_rate": 7.80854430379747e-05,
      "loss": 2.5692,
      "step": 277
    },
    {
      "epoch": 3.48,
      "learning_rate": 7.800632911392406e-05,
      "loss": 2.5474,
      "step": 278
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.792721518987342e-05,
      "loss": 2.551,
      "step": 279
    },
    {
      "epoch": 3.51,
      "learning_rate": 7.78481012658228e-05,
      "loss": 2.5591,
      "step": 280
    },
    {
      "epoch": 3.52,
      "learning_rate": 7.776898734177216e-05,
      "loss": 2.5553,
      "step": 281
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.768987341772152e-05,
      "loss": 2.4976,
      "step": 282
    },
    {
      "epoch": 3.54,
      "learning_rate": 7.761075949367089e-05,
      "loss": 2.4964,
      "step": 283
    },
    {
      "epoch": 3.56,
      "learning_rate": 7.753164556962026e-05,
      "loss": 2.4731,
      "step": 284
    },
    {
      "epoch": 3.57,
      "learning_rate": 7.745253164556963e-05,
      "loss": 2.5036,
      "step": 285
    },
    {
      "epoch": 3.58,
      "learning_rate": 7.737341772151899e-05,
      "loss": 2.5074,
      "step": 286
    },
    {
      "epoch": 3.59,
      "learning_rate": 7.729430379746836e-05,
      "loss": 2.5307,
      "step": 287
    },
    {
      "epoch": 3.61,
      "learning_rate": 7.721518987341773e-05,
      "loss": 2.5479,
      "step": 288
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.713607594936709e-05,
      "loss": 2.5085,
      "step": 289
    },
    {
      "epoch": 3.63,
      "learning_rate": 7.705696202531645e-05,
      "loss": 2.5537,
      "step": 290
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.697784810126583e-05,
      "loss": 2.6198,
      "step": 291
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.689873417721519e-05,
      "loss": 2.5547,
      "step": 292
    },
    {
      "epoch": 3.67,
      "learning_rate": 7.681962025316455e-05,
      "loss": 2.4875,
      "step": 293
    },
    {
      "epoch": 3.68,
      "learning_rate": 7.674050632911393e-05,
      "loss": 2.4859,
      "step": 294
    },
    {
      "epoch": 3.69,
      "learning_rate": 7.66613924050633e-05,
      "loss": 2.5,
      "step": 295
    },
    {
      "epoch": 3.71,
      "learning_rate": 7.658227848101266e-05,
      "loss": 2.4782,
      "step": 296
    },
    {
      "epoch": 3.72,
      "learning_rate": 7.650316455696202e-05,
      "loss": 2.5015,
      "step": 297
    },
    {
      "epoch": 3.73,
      "learning_rate": 7.64240506329114e-05,
      "loss": 2.4838,
      "step": 298
    },
    {
      "epoch": 3.74,
      "learning_rate": 7.634493670886076e-05,
      "loss": 2.4972,
      "step": 299
    },
    {
      "epoch": 3.76,
      "learning_rate": 7.626582278481012e-05,
      "loss": 2.5914,
      "step": 300
    },
    {
      "epoch": 3.77,
      "learning_rate": 7.61867088607595e-05,
      "loss": 2.6018,
      "step": 301
    },
    {
      "epoch": 3.78,
      "learning_rate": 7.610759493670886e-05,
      "loss": 2.5574,
      "step": 302
    },
    {
      "epoch": 3.79,
      "learning_rate": 7.602848101265824e-05,
      "loss": 2.5567,
      "step": 303
    },
    {
      "epoch": 3.81,
      "learning_rate": 7.59493670886076e-05,
      "loss": 2.519,
      "step": 304
    },
    {
      "epoch": 3.82,
      "learning_rate": 7.587025316455698e-05,
      "loss": 2.6013,
      "step": 305
    },
    {
      "epoch": 3.83,
      "learning_rate": 7.579113924050634e-05,
      "loss": 2.5606,
      "step": 306
    },
    {
      "epoch": 3.84,
      "learning_rate": 7.57120253164557e-05,
      "loss": 2.5331,
      "step": 307
    },
    {
      "epoch": 3.86,
      "learning_rate": 7.563291139240506e-05,
      "loss": 2.5034,
      "step": 308
    },
    {
      "epoch": 3.87,
      "learning_rate": 7.555379746835444e-05,
      "loss": 2.5237,
      "step": 309
    },
    {
      "epoch": 3.88,
      "learning_rate": 7.54746835443038e-05,
      "loss": 2.4697,
      "step": 310
    },
    {
      "epoch": 3.89,
      "learning_rate": 7.539556962025317e-05,
      "loss": 2.5559,
      "step": 311
    },
    {
      "epoch": 3.91,
      "learning_rate": 7.531645569620254e-05,
      "loss": 2.5485,
      "step": 312
    },
    {
      "epoch": 3.92,
      "learning_rate": 7.52373417721519e-05,
      "loss": 2.5404,
      "step": 313
    },
    {
      "epoch": 3.93,
      "learning_rate": 7.515822784810127e-05,
      "loss": 2.5336,
      "step": 314
    },
    {
      "epoch": 3.94,
      "learning_rate": 7.507911392405063e-05,
      "loss": 2.4819,
      "step": 315
    },
    {
      "epoch": 3.96,
      "learning_rate": 7.500000000000001e-05,
      "loss": 2.5207,
      "step": 316
    },
    {
      "epoch": 3.97,
      "learning_rate": 7.492088607594937e-05,
      "loss": 2.5366,
      "step": 317
    },
    {
      "epoch": 3.98,
      "learning_rate": 7.484177215189873e-05,
      "loss": 2.5074,
      "step": 318
    },
    {
      "epoch": 3.99,
      "learning_rate": 7.476265822784811e-05,
      "loss": 2.5523,
      "step": 319
    },
    {
      "epoch": 4.01,
      "learning_rate": 7.468354430379747e-05,
      "loss": 2.5471,
      "step": 320
    },
    {
      "epoch": 4.02,
      "learning_rate": 7.460443037974683e-05,
      "loss": 2.4375,
      "step": 321
    },
    {
      "epoch": 4.03,
      "learning_rate": 7.45253164556962e-05,
      "loss": 2.5254,
      "step": 322
    },
    {
      "epoch": 4.04,
      "learning_rate": 7.444620253164557e-05,
      "loss": 2.4685,
      "step": 323
    },
    {
      "epoch": 4.06,
      "learning_rate": 7.436708860759494e-05,
      "loss": 2.3927,
      "step": 324
    },
    {
      "epoch": 4.07,
      "learning_rate": 7.42879746835443e-05,
      "loss": 2.467,
      "step": 325
    },
    {
      "epoch": 4.08,
      "learning_rate": 7.420886075949368e-05,
      "loss": 2.4507,
      "step": 326
    },
    {
      "epoch": 4.09,
      "learning_rate": 7.412974683544304e-05,
      "loss": 2.5232,
      "step": 327
    },
    {
      "epoch": 4.11,
      "learning_rate": 7.40506329113924e-05,
      "loss": 2.5789,
      "step": 328
    },
    {
      "epoch": 4.12,
      "learning_rate": 7.397151898734176e-05,
      "loss": 2.3973,
      "step": 329
    },
    {
      "epoch": 4.13,
      "learning_rate": 7.389240506329115e-05,
      "loss": 2.5065,
      "step": 330
    },
    {
      "epoch": 4.14,
      "learning_rate": 7.381329113924052e-05,
      "loss": 2.4783,
      "step": 331
    },
    {
      "epoch": 4.16,
      "learning_rate": 7.373417721518988e-05,
      "loss": 2.4642,
      "step": 332
    },
    {
      "epoch": 4.17,
      "learning_rate": 7.365506329113924e-05,
      "loss": 2.4137,
      "step": 333
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.357594936708862e-05,
      "loss": 2.4881,
      "step": 334
    },
    {
      "epoch": 4.19,
      "learning_rate": 7.349683544303798e-05,
      "loss": 2.5382,
      "step": 335
    },
    {
      "epoch": 4.21,
      "learning_rate": 7.341772151898734e-05,
      "loss": 2.4962,
      "step": 336
    },
    {
      "epoch": 4.22,
      "learning_rate": 7.333860759493672e-05,
      "loss": 2.4236,
      "step": 337
    },
    {
      "epoch": 4.23,
      "learning_rate": 7.325949367088608e-05,
      "loss": 2.4238,
      "step": 338
    },
    {
      "epoch": 4.24,
      "learning_rate": 7.318037974683545e-05,
      "loss": 2.4651,
      "step": 339
    },
    {
      "epoch": 4.26,
      "learning_rate": 7.310126582278481e-05,
      "loss": 2.434,
      "step": 340
    },
    {
      "epoch": 4.27,
      "learning_rate": 7.302215189873418e-05,
      "loss": 2.4445,
      "step": 341
    },
    {
      "epoch": 4.28,
      "learning_rate": 7.294303797468355e-05,
      "loss": 2.4645,
      "step": 342
    },
    {
      "epoch": 4.29,
      "learning_rate": 7.286392405063291e-05,
      "loss": 2.4203,
      "step": 343
    },
    {
      "epoch": 4.31,
      "learning_rate": 7.278481012658229e-05,
      "loss": 2.4255,
      "step": 344
    },
    {
      "epoch": 4.32,
      "learning_rate": 7.270569620253165e-05,
      "loss": 2.5024,
      "step": 345
    },
    {
      "epoch": 4.33,
      "learning_rate": 7.262658227848101e-05,
      "loss": 2.4631,
      "step": 346
    },
    {
      "epoch": 4.34,
      "learning_rate": 7.254746835443037e-05,
      "loss": 2.4118,
      "step": 347
    },
    {
      "epoch": 4.36,
      "learning_rate": 7.246835443037975e-05,
      "loss": 2.4434,
      "step": 348
    },
    {
      "epoch": 4.37,
      "learning_rate": 7.238924050632911e-05,
      "loss": 2.4452,
      "step": 349
    },
    {
      "epoch": 4.38,
      "learning_rate": 7.231012658227848e-05,
      "loss": 2.4091,
      "step": 350
    },
    {
      "epoch": 4.39,
      "learning_rate": 7.223101265822785e-05,
      "loss": 2.4788,
      "step": 351
    },
    {
      "epoch": 4.41,
      "learning_rate": 7.215189873417722e-05,
      "loss": 2.4385,
      "step": 352
    },
    {
      "epoch": 4.42,
      "learning_rate": 7.207278481012658e-05,
      "loss": 2.4585,
      "step": 353
    },
    {
      "epoch": 4.43,
      "learning_rate": 7.199367088607595e-05,
      "loss": 2.5625,
      "step": 354
    },
    {
      "epoch": 4.44,
      "learning_rate": 7.191455696202532e-05,
      "loss": 2.4731,
      "step": 355
    },
    {
      "epoch": 4.46,
      "learning_rate": 7.18354430379747e-05,
      "loss": 2.5434,
      "step": 356
    },
    {
      "epoch": 4.47,
      "learning_rate": 7.175632911392406e-05,
      "loss": 2.4586,
      "step": 357
    },
    {
      "epoch": 4.48,
      "learning_rate": 7.167721518987342e-05,
      "loss": 2.4995,
      "step": 358
    },
    {
      "epoch": 4.49,
      "learning_rate": 7.15981012658228e-05,
      "loss": 2.4497,
      "step": 359
    },
    {
      "epoch": 4.51,
      "learning_rate": 7.151898734177216e-05,
      "loss": 2.4311,
      "step": 360
    },
    {
      "epoch": 4.52,
      "learning_rate": 7.143987341772152e-05,
      "loss": 2.5488,
      "step": 361
    },
    {
      "epoch": 4.53,
      "learning_rate": 7.13607594936709e-05,
      "loss": 2.4446,
      "step": 362
    },
    {
      "epoch": 4.54,
      "learning_rate": 7.128164556962026e-05,
      "loss": 2.5131,
      "step": 363
    },
    {
      "epoch": 4.56,
      "learning_rate": 7.120253164556962e-05,
      "loss": 2.4211,
      "step": 364
    },
    {
      "epoch": 4.57,
      "learning_rate": 7.112341772151899e-05,
      "loss": 2.5215,
      "step": 365
    },
    {
      "epoch": 4.58,
      "learning_rate": 7.104430379746836e-05,
      "loss": 2.3838,
      "step": 366
    },
    {
      "epoch": 4.59,
      "learning_rate": 7.096518987341772e-05,
      "loss": 2.4798,
      "step": 367
    },
    {
      "epoch": 4.61,
      "learning_rate": 7.088607594936709e-05,
      "loss": 2.464,
      "step": 368
    },
    {
      "epoch": 4.62,
      "learning_rate": 7.080696202531646e-05,
      "loss": 2.4536,
      "step": 369
    },
    {
      "epoch": 4.63,
      "learning_rate": 7.072784810126583e-05,
      "loss": 2.5556,
      "step": 370
    },
    {
      "epoch": 4.64,
      "learning_rate": 7.064873417721519e-05,
      "loss": 2.5225,
      "step": 371
    },
    {
      "epoch": 4.66,
      "learning_rate": 7.056962025316457e-05,
      "loss": 2.5068,
      "step": 372
    },
    {
      "epoch": 4.67,
      "learning_rate": 7.049050632911393e-05,
      "loss": 2.5099,
      "step": 373
    },
    {
      "epoch": 4.68,
      "learning_rate": 7.041139240506329e-05,
      "loss": 2.4494,
      "step": 374
    },
    {
      "epoch": 4.69,
      "learning_rate": 7.033227848101265e-05,
      "loss": 2.4185,
      "step": 375
    },
    {
      "epoch": 4.71,
      "learning_rate": 7.025316455696203e-05,
      "loss": 2.4423,
      "step": 376
    },
    {
      "epoch": 4.72,
      "learning_rate": 7.017405063291139e-05,
      "loss": 2.534,
      "step": 377
    },
    {
      "epoch": 4.73,
      "learning_rate": 7.009493670886076e-05,
      "loss": 2.4981,
      "step": 378
    },
    {
      "epoch": 4.74,
      "learning_rate": 7.001582278481013e-05,
      "loss": 2.4773,
      "step": 379
    },
    {
      "epoch": 4.76,
      "learning_rate": 6.99367088607595e-05,
      "loss": 2.4654,
      "step": 380
    },
    {
      "epoch": 4.77,
      "learning_rate": 6.985759493670886e-05,
      "loss": 2.5398,
      "step": 381
    },
    {
      "epoch": 4.78,
      "learning_rate": 6.977848101265823e-05,
      "loss": 2.5319,
      "step": 382
    },
    {
      "epoch": 4.79,
      "learning_rate": 6.96993670886076e-05,
      "loss": 2.4655,
      "step": 383
    },
    {
      "epoch": 4.81,
      "learning_rate": 6.962025316455697e-05,
      "loss": 2.4246,
      "step": 384
    },
    {
      "epoch": 4.82,
      "learning_rate": 6.954113924050634e-05,
      "loss": 2.4681,
      "step": 385
    },
    {
      "epoch": 4.83,
      "learning_rate": 6.94620253164557e-05,
      "loss": 2.5336,
      "step": 386
    },
    {
      "epoch": 4.85,
      "learning_rate": 6.938291139240507e-05,
      "loss": 2.4236,
      "step": 387
    },
    {
      "epoch": 4.86,
      "learning_rate": 6.930379746835444e-05,
      "loss": 2.4617,
      "step": 388
    },
    {
      "epoch": 4.87,
      "learning_rate": 6.92246835443038e-05,
      "loss": 2.4018,
      "step": 389
    },
    {
      "epoch": 4.88,
      "learning_rate": 6.914556962025316e-05,
      "loss": 2.4973,
      "step": 390
    },
    {
      "epoch": 4.9,
      "learning_rate": 6.906645569620254e-05,
      "loss": 2.511,
      "step": 391
    },
    {
      "epoch": 4.91,
      "learning_rate": 6.89873417721519e-05,
      "loss": 2.5253,
      "step": 392
    },
    {
      "epoch": 4.92,
      "learning_rate": 6.890822784810127e-05,
      "loss": 2.3835,
      "step": 393
    },
    {
      "epoch": 4.93,
      "learning_rate": 6.882911392405064e-05,
      "loss": 2.4304,
      "step": 394
    },
    {
      "epoch": 4.95,
      "learning_rate": 6.875e-05,
      "loss": 2.5319,
      "step": 395
    },
    {
      "epoch": 4.96,
      "learning_rate": 6.867088607594937e-05,
      "loss": 2.4936,
      "step": 396
    },
    {
      "epoch": 4.97,
      "learning_rate": 6.859177215189874e-05,
      "loss": 2.4887,
      "step": 397
    },
    {
      "epoch": 4.98,
      "learning_rate": 6.85126582278481e-05,
      "loss": 2.5686,
      "step": 398
    },
    {
      "epoch": 5.0,
      "learning_rate": 6.843354430379747e-05,
      "loss": 2.3999,
      "step": 399
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.835443037974683e-05,
      "loss": 2.3567,
      "step": 400
    },
    {
      "epoch": 5.02,
      "learning_rate": 6.827531645569621e-05,
      "loss": 2.3757,
      "step": 401
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.819620253164557e-05,
      "loss": 2.3118,
      "step": 402
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.811708860759493e-05,
      "loss": 2.4044,
      "step": 403
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.803797468354431e-05,
      "loss": 2.5191,
      "step": 404
    },
    {
      "epoch": 5.07,
      "learning_rate": 6.795886075949367e-05,
      "loss": 2.3796,
      "step": 405
    },
    {
      "epoch": 5.08,
      "learning_rate": 6.787974683544304e-05,
      "loss": 2.3092,
      "step": 406
    },
    {
      "epoch": 5.1,
      "learning_rate": 6.78006329113924e-05,
      "loss": 2.4262,
      "step": 407
    },
    {
      "epoch": 5.11,
      "learning_rate": 6.772151898734177e-05,
      "loss": 2.4117,
      "step": 408
    },
    {
      "epoch": 5.12,
      "learning_rate": 6.764240506329115e-05,
      "loss": 2.4613,
      "step": 409
    },
    {
      "epoch": 5.13,
      "learning_rate": 6.756329113924051e-05,
      "loss": 2.4062,
      "step": 410
    },
    {
      "epoch": 5.15,
      "learning_rate": 6.748417721518988e-05,
      "loss": 2.4383,
      "step": 411
    },
    {
      "epoch": 5.16,
      "learning_rate": 6.740506329113925e-05,
      "loss": 2.4523,
      "step": 412
    },
    {
      "epoch": 5.17,
      "learning_rate": 6.732594936708862e-05,
      "loss": 2.4081,
      "step": 413
    },
    {
      "epoch": 5.18,
      "learning_rate": 6.724683544303798e-05,
      "loss": 2.4348,
      "step": 414
    },
    {
      "epoch": 5.2,
      "learning_rate": 6.716772151898735e-05,
      "loss": 2.4608,
      "step": 415
    },
    {
      "epoch": 5.21,
      "learning_rate": 6.708860759493672e-05,
      "loss": 2.4999,
      "step": 416
    },
    {
      "epoch": 5.22,
      "learning_rate": 6.700949367088608e-05,
      "loss": 2.3906,
      "step": 417
    },
    {
      "epoch": 5.23,
      "learning_rate": 6.693037974683544e-05,
      "loss": 2.3757,
      "step": 418
    },
    {
      "epoch": 5.25,
      "learning_rate": 6.685126582278482e-05,
      "loss": 2.4686,
      "step": 419
    },
    {
      "epoch": 5.26,
      "learning_rate": 6.677215189873418e-05,
      "loss": 2.4821,
      "step": 420
    },
    {
      "epoch": 5.27,
      "learning_rate": 6.669303797468354e-05,
      "loss": 2.423,
      "step": 421
    },
    {
      "epoch": 5.28,
      "learning_rate": 6.661392405063292e-05,
      "loss": 2.3645,
      "step": 422
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.653481012658228e-05,
      "loss": 2.452,
      "step": 423
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.645569620253165e-05,
      "loss": 2.3124,
      "step": 424
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.637658227848101e-05,
      "loss": 2.4703,
      "step": 425
    },
    {
      "epoch": 5.33,
      "learning_rate": 6.629746835443039e-05,
      "loss": 2.3629,
      "step": 426
    },
    {
      "epoch": 5.35,
      "learning_rate": 6.621835443037975e-05,
      "loss": 2.433,
      "step": 427
    },
    {
      "epoch": 5.36,
      "learning_rate": 6.613924050632911e-05,
      "loss": 2.3075,
      "step": 428
    },
    {
      "epoch": 5.37,
      "learning_rate": 6.606012658227849e-05,
      "loss": 2.4005,
      "step": 429
    },
    {
      "epoch": 5.38,
      "learning_rate": 6.598101265822785e-05,
      "loss": 2.3672,
      "step": 430
    },
    {
      "epoch": 5.4,
      "learning_rate": 6.590189873417721e-05,
      "loss": 2.4682,
      "step": 431
    },
    {
      "epoch": 5.41,
      "learning_rate": 6.582278481012658e-05,
      "loss": 2.4498,
      "step": 432
    },
    {
      "epoch": 5.42,
      "learning_rate": 6.574367088607595e-05,
      "loss": 2.4776,
      "step": 433
    },
    {
      "epoch": 5.43,
      "learning_rate": 6.566455696202531e-05,
      "loss": 2.387,
      "step": 434
    },
    {
      "epoch": 5.45,
      "learning_rate": 6.558544303797469e-05,
      "loss": 2.4102,
      "step": 435
    },
    {
      "epoch": 5.46,
      "learning_rate": 6.550632911392405e-05,
      "loss": 2.4719,
      "step": 436
    },
    {
      "epoch": 5.47,
      "learning_rate": 6.542721518987343e-05,
      "loss": 2.4796,
      "step": 437
    },
    {
      "epoch": 5.48,
      "learning_rate": 6.534810126582279e-05,
      "loss": 2.3951,
      "step": 438
    },
    {
      "epoch": 5.5,
      "learning_rate": 6.526898734177216e-05,
      "loss": 2.4206,
      "step": 439
    },
    {
      "epoch": 5.51,
      "learning_rate": 6.518987341772153e-05,
      "loss": 2.4221,
      "step": 440
    },
    {
      "epoch": 5.52,
      "learning_rate": 6.51107594936709e-05,
      "loss": 2.4705,
      "step": 441
    },
    {
      "epoch": 5.53,
      "learning_rate": 6.503164556962026e-05,
      "loss": 2.4581,
      "step": 442
    },
    {
      "epoch": 5.55,
      "learning_rate": 6.495253164556962e-05,
      "loss": 2.35,
      "step": 443
    },
    {
      "epoch": 5.56,
      "learning_rate": 6.4873417721519e-05,
      "loss": 2.3569,
      "step": 444
    },
    {
      "epoch": 5.57,
      "learning_rate": 6.479430379746836e-05,
      "loss": 2.4375,
      "step": 445
    },
    {
      "epoch": 5.58,
      "learning_rate": 6.471518987341772e-05,
      "loss": 2.4266,
      "step": 446
    },
    {
      "epoch": 5.6,
      "learning_rate": 6.46360759493671e-05,
      "loss": 2.5028,
      "step": 447
    },
    {
      "epoch": 5.61,
      "learning_rate": 6.455696202531646e-05,
      "loss": 2.3958,
      "step": 448
    },
    {
      "epoch": 5.62,
      "learning_rate": 6.447784810126582e-05,
      "loss": 2.3978,
      "step": 449
    },
    {
      "epoch": 5.63,
      "learning_rate": 6.439873417721519e-05,
      "loss": 2.4629,
      "step": 450
    },
    {
      "epoch": 5.65,
      "learning_rate": 6.431962025316456e-05,
      "loss": 2.4269,
      "step": 451
    },
    {
      "epoch": 5.66,
      "learning_rate": 6.424050632911393e-05,
      "loss": 2.3565,
      "step": 452
    },
    {
      "epoch": 5.67,
      "learning_rate": 6.416139240506329e-05,
      "loss": 2.4155,
      "step": 453
    },
    {
      "epoch": 5.68,
      "learning_rate": 6.408227848101266e-05,
      "loss": 2.4556,
      "step": 454
    },
    {
      "epoch": 5.7,
      "learning_rate": 6.400316455696203e-05,
      "loss": 2.4244,
      "step": 455
    },
    {
      "epoch": 5.71,
      "learning_rate": 6.392405063291139e-05,
      "loss": 2.2978,
      "step": 456
    },
    {
      "epoch": 5.72,
      "learning_rate": 6.384493670886075e-05,
      "loss": 2.4742,
      "step": 457
    },
    {
      "epoch": 5.73,
      "learning_rate": 6.376582278481013e-05,
      "loss": 2.3629,
      "step": 458
    },
    {
      "epoch": 5.75,
      "learning_rate": 6.368670886075949e-05,
      "loss": 2.4468,
      "step": 459
    },
    {
      "epoch": 5.76,
      "learning_rate": 6.360759493670885e-05,
      "loss": 2.4993,
      "step": 460
    },
    {
      "epoch": 5.77,
      "learning_rate": 6.352848101265823e-05,
      "loss": 2.4546,
      "step": 461
    },
    {
      "epoch": 5.78,
      "learning_rate": 6.344936708860761e-05,
      "loss": 2.442,
      "step": 462
    },
    {
      "epoch": 5.8,
      "learning_rate": 6.337025316455697e-05,
      "loss": 2.4744,
      "step": 463
    },
    {
      "epoch": 5.81,
      "learning_rate": 6.329113924050633e-05,
      "loss": 2.5044,
      "step": 464
    },
    {
      "epoch": 5.82,
      "learning_rate": 6.321202531645571e-05,
      "loss": 2.4399,
      "step": 465
    },
    {
      "epoch": 5.83,
      "learning_rate": 6.313291139240507e-05,
      "loss": 2.3402,
      "step": 466
    },
    {
      "epoch": 5.85,
      "learning_rate": 6.305379746835443e-05,
      "loss": 2.432,
      "step": 467
    },
    {
      "epoch": 5.86,
      "learning_rate": 6.29746835443038e-05,
      "loss": 2.3745,
      "step": 468
    },
    {
      "epoch": 5.87,
      "learning_rate": 6.289556962025317e-05,
      "loss": 2.4026,
      "step": 469
    },
    {
      "epoch": 5.88,
      "learning_rate": 6.281645569620254e-05,
      "loss": 2.3891,
      "step": 470
    },
    {
      "epoch": 5.9,
      "learning_rate": 6.27373417721519e-05,
      "loss": 2.4781,
      "step": 471
    },
    {
      "epoch": 5.91,
      "learning_rate": 6.265822784810128e-05,
      "loss": 2.468,
      "step": 472
    },
    {
      "epoch": 5.92,
      "learning_rate": 6.257911392405064e-05,
      "loss": 2.4899,
      "step": 473
    },
    {
      "epoch": 5.93,
      "learning_rate": 6.25e-05,
      "loss": 2.4882,
      "step": 474
    },
    {
      "epoch": 5.95,
      "learning_rate": 6.242088607594936e-05,
      "loss": 2.3793,
      "step": 475
    },
    {
      "epoch": 5.96,
      "learning_rate": 6.234177215189874e-05,
      "loss": 2.4381,
      "step": 476
    },
    {
      "epoch": 5.97,
      "learning_rate": 6.22626582278481e-05,
      "loss": 2.4239,
      "step": 477
    },
    {
      "epoch": 5.98,
      "learning_rate": 6.218354430379747e-05,
      "loss": 2.4914,
      "step": 478
    },
    {
      "epoch": 6.0,
      "learning_rate": 6.210443037974684e-05,
      "loss": 2.4158,
      "step": 479
    },
    {
      "epoch": 6.01,
      "learning_rate": 6.20253164556962e-05,
      "loss": 2.3808,
      "step": 480
    },
    {
      "epoch": 6.02,
      "learning_rate": 6.194620253164557e-05,
      "loss": 2.3322,
      "step": 481
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.186708860759493e-05,
      "loss": 2.3791,
      "step": 482
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.178797468354431e-05,
      "loss": 2.3428,
      "step": 483
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.170886075949367e-05,
      "loss": 2.3487,
      "step": 484
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.162974683544303e-05,
      "loss": 2.324,
      "step": 485
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.155063291139241e-05,
      "loss": 2.439,
      "step": 486
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.147151898734177e-05,
      "loss": 2.4255,
      "step": 487
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.139240506329115e-05,
      "loss": 2.3432,
      "step": 488
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.131329113924051e-05,
      "loss": 2.391,
      "step": 489
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.123417721518989e-05,
      "loss": 2.3704,
      "step": 490
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.115506329113925e-05,
      "loss": 2.2892,
      "step": 491
    },
    {
      "epoch": 6.16,
      "learning_rate": 6.107594936708861e-05,
      "loss": 2.3995,
      "step": 492
    },
    {
      "epoch": 6.17,
      "learning_rate": 6.099683544303798e-05,
      "loss": 2.3633,
      "step": 493
    },
    {
      "epoch": 6.18,
      "learning_rate": 6.0917721518987345e-05,
      "loss": 2.3074,
      "step": 494
    },
    {
      "epoch": 6.2,
      "learning_rate": 6.0838607594936714e-05,
      "loss": 2.3454,
      "step": 495
    },
    {
      "epoch": 6.21,
      "learning_rate": 6.0759493670886084e-05,
      "loss": 2.3856,
      "step": 496
    },
    {
      "epoch": 6.22,
      "learning_rate": 6.0680379746835447e-05,
      "loss": 2.3991,
      "step": 497
    },
    {
      "epoch": 6.23,
      "learning_rate": 6.0601265822784816e-05,
      "loss": 2.3348,
      "step": 498
    },
    {
      "epoch": 6.25,
      "learning_rate": 6.052215189873418e-05,
      "loss": 2.3506,
      "step": 499
    },
    {
      "epoch": 6.26,
      "learning_rate": 6.044303797468355e-05,
      "loss": 2.3444,
      "step": 500
    }
  ],
  "max_steps": 1264,
  "num_train_epochs": 16,
  "total_flos": 1.3096315313163264e+17,
  "trial_name": null,
  "trial_params": null
}