{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 206,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.996,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.0034,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.9937,
      "step": 3
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.994,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.9761,
      "step": 5
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.9838,
      "step": 6
    },
    {
      "epoch": 0.03,
      "learning_rate": 2e-05,
      "loss": 0.9728,
      "step": 7
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9998753895176576e-05,
      "loss": 0.9452,
      "step": 8
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999501589126174e-05,
      "loss": 0.95,
      "step": 9
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9988786919844437e-05,
      "loss": 0.953,
      "step": 10
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9980068533314937e-05,
      "loss": 0.9442,
      "step": 11
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9968862904477936e-05,
      "loss": 0.8945,
      "step": 12
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.995517282601106e-05,
      "loss": 0.9333,
      "step": 13
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.993900170976888e-05,
      "loss": 0.9008,
      "step": 14
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.992035358593258e-05,
      "loss": 0.8652,
      "step": 15
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9899233102005573e-05,
      "loss": 0.8659,
      "step": 16
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.987564552165524e-05,
      "loss": 0.8572,
      "step": 17
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.984959672340111e-05,
      "loss": 0.8595,
      "step": 18
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9821093199149806e-05,
      "loss": 0.8946,
      "step": 19
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9790142052577148e-05,
      "loss": 0.8551,
      "step": 20
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9756750997357738e-05,
      "loss": 0.8567,
      "step": 21
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.972092835524257e-05,
      "loss": 0.8264,
      "step": 22
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9682683053985073e-05,
      "loss": 0.8544,
      "step": 23
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9642024625116117e-05,
      "loss": 0.807,
      "step": 24
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.959896320156857e-05,
      "loss": 0.816,
      "step": 25
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.955350951515195e-05,
      "loss": 0.8022,
      "step": 26
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.950567489387783e-05,
      "loss": 0.7919,
      "step": 27
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.945547125913667e-05,
      "loss": 0.7854,
      "step": 28
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9402911122726756e-05,
      "loss": 0.7748,
      "step": 29
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9348007583735985e-05,
      "loss": 0.7631,
      "step": 30
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9290774325277305e-05,
      "loss": 0.7694,
      "step": 31
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.923122561107861e-05,
      "loss": 0.7693,
      "step": 32
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.916937628192789e-05,
      "loss": 0.7977,
      "step": 33
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9105241751974624e-05,
      "loss": 0.7403,
      "step": 34
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.903883800488824e-05,
      "loss": 0.7152,
      "step": 35
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8970181589874637e-05,
      "loss": 0.7462,
      "step": 36
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8899289617551803e-05,
      "loss": 0.7399,
      "step": 37
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.882617975568547e-05,
      "loss": 0.6994,
      "step": 38
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.875087022478594e-05,
      "loss": 0.6904,
      "step": 39
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.867337979356715e-05,
      "loss": 0.7154,
      "step": 40
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8593727774269122e-05,
      "loss": 0.683,
      "step": 41
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.851193401784495e-05,
      "loss": 0.6527,
      "step": 42
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.842801890901351e-05,
      "loss": 0.7025,
      "step": 43
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.834200336117918e-05,
      "loss": 0.6714,
      "step": 44
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8253908811219764e-05,
      "loss": 0.7061,
      "step": 45
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8163757214143993e-05,
      "loss": 0.6749,
      "step": 46
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8071571037619856e-05,
      "loss": 0.6753,
      "step": 47
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7977373256375194e-05,
      "loss": 0.6536,
      "step": 48
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7881187346471924e-05,
      "loss": 0.6698,
      "step": 49
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.77830372794553e-05,
      "loss": 0.6557,
      "step": 50
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7682947516379706e-05,
      "loss": 0.6635,
      "step": 51
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7580943001712457e-05,
      "loss": 0.6786,
      "step": 52
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7477049157117093e-05,
      "loss": 0.634,
      "step": 53
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.737129187511779e-05,
      "loss": 0.6512,
      "step": 54
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7263697512646397e-05,
      "loss": 0.6311,
      "step": 55
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7154292884473712e-05,
      "loss": 0.657,
      "step": 56
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7043105256526723e-05,
      "loss": 0.6448,
      "step": 57
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.693016233909332e-05,
      "loss": 0.6516,
      "step": 58
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.681549227991634e-05,
      "loss": 0.6134,
      "step": 59
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6699123657178553e-05,
      "loss": 0.6768,
      "step": 60
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.658108547238038e-05,
      "loss": 0.6706,
      "step": 61
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.64614071431121e-05,
      "loss": 0.664,
      "step": 62
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.634011849572239e-05,
      "loss": 0.6412,
      "step": 63
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6217249757884954e-05,
      "loss": 0.6058,
      "step": 64
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.609283155106517e-05,
      "loss": 0.6355,
      "step": 65
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.596689488288856e-05,
      "loss": 0.6553,
      "step": 66
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5839471139413065e-05,
      "loss": 0.6181,
      "step": 67
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.571059207730695e-05,
      "loss": 0.6601,
      "step": 68
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.55802898159344e-05,
      "loss": 0.6285,
      "step": 69
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5448596829350706e-05,
      "loss": 0.6398,
      "step": 70
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5315545938209016e-05,
      "loss": 0.6388,
      "step": 71
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5181170301580776e-05,
      "loss": 0.6174,
      "step": 72
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5045503408691776e-05,
      "loss": 0.6202,
      "step": 73
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4908579070575936e-05,
      "loss": 0.6329,
      "step": 74
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4770431411648898e-05,
      "loss": 0.6149,
      "step": 75
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4631094861203478e-05,
      "loss": 0.6222,
      "step": 76
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4490604144829204e-05,
      "loss": 0.5883,
      "step": 77
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4348994275757933e-05,
      "loss": 0.6191,
      "step": 78
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4206300546137844e-05,
      "loss": 0.6168,
      "step": 79
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4062558518237893e-05,
      "loss": 0.6291,
      "step": 80
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3917804015584932e-05,
      "loss": 0.6263,
      "step": 81
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3772073114035762e-05,
      "loss": 0.6145,
      "step": 82
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3625402132786247e-05,
      "loss": 0.6068,
      "step": 83
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3477827625319826e-05,
      "loss": 0.6052,
      "step": 84
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3329386370297615e-05,
      "loss": 0.62,
      "step": 85
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3180115362392383e-05,
      "loss": 0.6165,
      "step": 86
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3030051803068729e-05,
      "loss": 0.6146,
      "step": 87
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2879233091311667e-05,
      "loss": 0.6137,
      "step": 88
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2727696814306034e-05,
      "loss": 0.5891,
      "step": 89
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2575480738068971e-05,
      "loss": 0.6241,
      "step": 90
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2422622798037833e-05,
      "loss": 0.57,
      "step": 91
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2269161089615902e-05,
      "loss": 0.6129,
      "step": 92
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2115133858678192e-05,
      "loss": 0.6138,
      "step": 93
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1960579492039783e-05,
      "loss": 0.585,
      "step": 94
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1805536507889021e-05,
      "loss": 0.5916,
      "step": 95
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1650043546187994e-05,
      "loss": 0.5805,
      "step": 96
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1494139359042612e-05,
      "loss": 0.593,
      "step": 97
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1337862801044792e-05,
      "loss": 0.6136,
      "step": 98
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1181252819589081e-05,
      "loss": 0.5979,
      "step": 99
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.1024348445166133e-05,
      "loss": 0.5984,
      "step": 100
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.086718878163551e-05,
      "loss": 0.5988,
      "step": 101
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.070981299648016e-05,
      "loss": 0.5895,
      "step": 102
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0552260311045082e-05,
      "loss": 0.5827,
      "step": 103
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0394569990762528e-05,
      "loss": 0.5592,
      "step": 104
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0236781335366239e-05,
      "loss": 0.5977,
      "step": 105
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0078933669097135e-05,
      "loss": 0.6285,
      "step": 106
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.92106633090287e-06,
      "loss": 0.5707,
      "step": 107
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.763218664633763e-06,
      "loss": 0.5734,
      "step": 108
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.605430009237474e-06,
      "loss": 0.5634,
      "step": 109
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.44773968895492e-06,
      "loss": 0.5805,
      "step": 110
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.290187003519841e-06,
      "loss": 0.6077,
      "step": 111
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.132811218364494e-06,
      "loss": 0.5553,
      "step": 112
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.975651554833869e-06,
      "loss": 0.5861,
      "step": 113
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.81874718041092e-06,
      "loss": 0.5703,
      "step": 114
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.662137198955211e-06,
      "loss": 0.6203,
      "step": 115
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.50586064095739e-06,
      "loss": 0.5881,
      "step": 116
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.349956453812009e-06,
      "loss": 0.5663,
      "step": 117
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.194463492110982e-06,
      "loss": 0.5783,
      "step": 118
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.03942050796022e-06,
      "loss": 0.5991,
      "step": 119
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.884866141321811e-06,
      "loss": 0.5875,
      "step": 120
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.730838910384098e-06,
      "loss": 0.5777,
      "step": 121
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.57737720196217e-06,
      "loss": 0.5558,
      "step": 122
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.424519261931036e-06,
      "loss": 0.5659,
      "step": 123
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.27230318569397e-06,
      "loss": 0.5724,
      "step": 124
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.1207669086883366e-06,
      "loss": 0.5906,
      "step": 125
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.969948196931272e-06,
      "loss": 0.6066,
      "step": 126
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.819884637607619e-06,
      "loss": 0.5358,
      "step": 127
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.670613629702391e-06,
      "loss": 0.5714,
      "step": 128
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.522172374680177e-06,
      "loss": 0.5963,
      "step": 129
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.374597867213756e-06,
      "loss": 0.5721,
      "step": 130
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.2279268859642396e-06,
      "loss": 0.5789,
      "step": 131
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.082195984415069e-06,
      "loss": 0.5486,
      "step": 132
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.937441481762112e-06,
      "loss": 0.5891,
      "step": 133
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.793699453862161e-06,
      "loss": 0.5825,
      "step": 134
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.651005724242072e-06,
      "loss": 0.5871,
      "step": 135
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.509395855170798e-06,
      "loss": 0.5727,
      "step": 136
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.368905138796523e-06,
      "loss": 0.5596,
      "step": 137
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.2295685883511086e-06,
      "loss": 0.564,
      "step": 138
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.091420929424065e-06,
      "loss": 0.5774,
      "step": 139
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.954496591308227e-06,
      "loss": 0.5657,
      "step": 140
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.818829698419225e-06,
      "loss": 0.5987,
      "step": 141
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.684454061790987e-06,
      "loss": 0.581,
      "step": 142
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.551403170649299e-06,
      "loss": 0.5937,
      "step": 143
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.4197101840656e-06,
      "loss": 0.5584,
      "step": 144
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.289407922693053e-06,
      "loss": 0.5957,
      "step": 145
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.1605288605869365e-06,
      "loss": 0.576,
      "step": 146
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.033105117111441e-06,
      "loss": 0.5939,
      "step": 147
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.907168448934836e-06,
      "loss": 0.5769,
      "step": 148
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7827502421150497e-06,
      "loss": 0.5406,
      "step": 149
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6598815042776135e-06,
      "loss": 0.5554,
      "step": 150
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5385928568879012e-06,
      "loss": 0.5752,
      "step": 151
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.4189145276196244e-06,
      "loss": 0.5839,
      "step": 152
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.300876342821451e-06,
      "loss": 0.5926,
      "step": 153
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1845077200836638e-06,
      "loss": 0.5547,
      "step": 154
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0698376609066828e-06,
      "loss": 0.584,
      "step": 155
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9568947434732777e-06,
      "loss": 0.5641,
      "step": 156
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8457071155262885e-06,
      "loss": 0.5679,
      "step": 157
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.7363024873536093e-06,
      "loss": 0.5801,
      "step": 158
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.628708124882212e-06,
      "loss": 0.5783,
      "step": 159
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.52295084288291e-06,
      "loss": 0.5805,
      "step": 160
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.419056998287547e-06,
      "loss": 0.5629,
      "step": 161
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.3170524836202936e-06,
      "loss": 0.5662,
      "step": 162
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.216962720544703e-06,
      "loss": 0.5781,
      "step": 163
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.118812653528077e-06,
      "loss": 0.5541,
      "step": 164
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.022626743624807e-06,
      "loss": 0.5714,
      "step": 165
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.928428962380148e-06,
      "loss": 0.5708,
      "step": 166
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8362427858560094e-06,
      "loss": 0.5806,
      "step": 167
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.74609118878024e-06,
      "loss": 0.5739,
      "step": 168
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6579966388208257e-06,
      "loss": 0.5445,
      "step": 169
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5719810909864941e-06,
      "loss": 0.5699,
      "step": 170
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4880659821550547e-06,
      "loss": 0.5833,
      "step": 171
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4062722257308803e-06,
      "loss": 0.5601,
      "step": 172
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3266202064328548e-06,
      "loss": 0.5845,
      "step": 173
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.249129775214064e-06,
      "loss": 0.5776,
      "step": 174
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1738202443145307e-06,
      "loss": 0.5798,
      "step": 175
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.100710382448198e-06,
      "loss": 0.5707,
      "step": 176
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.029818410125365e-06,
      "loss": 0.5948,
      "step": 177
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.611619951117657e-07,
      "loss": 0.5705,
      "step": 178
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.94758248025378e-07,
      "loss": 0.5589,
      "step": 179
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.306237180721121e-07,
      "loss": 0.5739,
      "step": 180
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.687743889213939e-07,
      "loss": 0.591,
      "step": 181
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.092256747226944e-07,
      "loss": 0.5556,
      "step": 182
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.519924162640168e-07,
      "loss": 0.5899,
      "step": 183
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.970888772732453e-07,
      "loss": 0.5464,
      "step": 184
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.445287408633304e-07,
      "loss": 0.5811,
      "step": 185
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.943251061221721e-07,
      "loss": 0.5808,
      "step": 186
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.464904848480522e-07,
      "loss": 0.5603,
      "step": 187
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.0103679843142895e-07,
      "loss": 0.5533,
      "step": 188
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.5797537488388326e-07,
      "loss": 0.5364,
      "step": 189
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.1731694601492834e-07,
      "loss": 0.5921,
      "step": 190
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.790716447574304e-07,
      "loss": 0.5815,
      "step": 191
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.4324900264226405e-07,
      "loss": 0.5596,
      "step": 192
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.098579474228546e-07,
      "loss": 0.5692,
      "step": 193
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7890680085019597e-07,
      "loss": 0.551,
      "step": 194
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.504032765988961e-07,
      "loss": 0.5526,
      "step": 195
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2435447834476254e-07,
      "loss": 0.5647,
      "step": 196
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.0076689799442874e-07,
      "loss": 0.5546,
      "step": 197
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.964641406742135e-08,
      "loss": 0.5609,
      "step": 198
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.099829023112236e-08,
      "loss": 0.5782,
      "step": 199
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.482717398894165e-08,
      "loss": 0.5829,
      "step": 200
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.1137095522068006e-08,
      "loss": 0.5691,
      "step": 201
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.993146668506585e-08,
      "loss": 0.5723,
      "step": 202
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.1213080155564327e-08,
      "loss": 0.557,
      "step": 203
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.984108738261828e-09,
      "loss": 0.58,
      "step": 204
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.246104823426908e-09,
      "loss": 0.5601,
      "step": 205
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 0.5731,
      "step": 206
    },
    {
      "epoch": 1.0,
      "step": 206,
      "total_flos": 5.847863731316326e+16,
      "train_loss": 0.6453299609202783,
      "train_runtime": 605.87,
      "train_samples_per_second": 10.844,
      "train_steps_per_second": 0.34
    }
  ],
  "max_steps": 206,
  "num_train_epochs": 1,
  "total_flos": 5.847863731316326e+16,
  "trial_name": null,
  "trial_params": null
}