{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.944,
  "global_step": 138,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.9901,
      "step": 1
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.9263,
      "step": 2
    },
    {
      "epoch": 0.06,
      "learning_rate": 6e-06,
      "loss": 0.9251,
      "step": 3
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.9197,
      "step": 4
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 0.8951,
      "step": 5
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.998605186060138e-06,
      "loss": 0.8878,
      "step": 6
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.99442152244292e-06,
      "loss": 0.8363,
      "step": 7
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.98745134332128e-06,
      "loss": 0.8379,
      "step": 8
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.97769853753642e-06,
      "loss": 0.2429,
      "step": 9
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.965168546428122e-06,
      "loss": 0.8244,
      "step": 10
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.949868360798893e-06,
      "loss": 0.8101,
      "step": 11
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.931806517013612e-06,
      "loss": 0.863,
      "step": 12
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.910993092236878e-06,
      "loss": 0.8458,
      "step": 13
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.887439698810694e-06,
      "loss": 0.806,
      "step": 14
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.861159477775653e-06,
      "loss": 0.8507,
      "step": 15
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.832167091539215e-06,
      "loss": 0.7992,
      "step": 16
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.800478715695165e-06,
      "loss": 0.8719,
      "step": 17
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.766112029998847e-06,
      "loss": 0.2583,
      "step": 18
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.729086208503174e-06,
      "loss": 0.8628,
      "step": 19
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.689421908860928e-06,
      "loss": 0.8408,
      "step": 20
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.64714126079933e-06,
      "loss": 0.8208,
      "step": 21
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.602267853773301e-06,
      "loss": 0.8414,
      "step": 22
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.554826723804304e-06,
      "loss": 0.7868,
      "step": 23
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.504844339512096e-06,
      "loss": 0.8611,
      "step": 24
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.452348587347224e-06,
      "loss": 0.8196,
      "step": 25
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.397368756032445e-06,
      "loss": 0.8339,
      "step": 26
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.339935520221816e-06,
      "loss": 0.7913,
      "step": 27
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.280080923386501e-06,
      "loss": 0.8374,
      "step": 28
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.217838359936914e-06,
      "loss": 0.8664,
      "step": 29
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.153242556591115e-06,
      "loss": 0.8206,
      "step": 30
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.08632955299989e-06,
      "loss": 0.8471,
      "step": 31
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.017136681639307e-06,
      "loss": 0.773,
      "step": 32
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.94570254698197e-06,
      "loss": 0.7712,
      "step": 33
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.872067003958597e-06,
      "loss": 0.7854,
      "step": 34
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.796271135721944e-06,
      "loss": 0.7561,
      "step": 35
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.71835723072545e-06,
      "loss": 0.8138,
      "step": 36
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.638368759129433e-06,
      "loss": 0.7854,
      "step": 37
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.556350348547978e-06,
      "loss": 0.8172,
      "step": 38
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.472347759150044e-06,
      "loss": 0.7749,
      "step": 39
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.386407858128707e-06,
      "loss": 0.8892,
      "step": 40
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.298578593552737e-06,
      "loss": 0.86,
      "step": 41
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.208908967615159e-06,
      "loss": 0.773,
      "step": 42
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.117449009293668e-06,
      "loss": 0.8129,
      "step": 43
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.024249746438189e-06,
      "loss": 0.8166,
      "step": 44
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.929363177301124e-06,
      "loss": 0.8203,
      "step": 45
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.832842241526212e-06,
      "loss": 0.754,
      "step": 46
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.734740790612137e-06,
      "loss": 0.3353,
      "step": 47
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.635113557867395e-06,
      "loss": 0.6187,
      "step": 48
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.5340161278732e-06,
      "loss": 0.5845,
      "step": 49
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.431504905471407e-06,
      "loss": 0.6143,
      "step": 50
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.327637084294818e-06,
      "loss": 0.6156,
      "step": 51
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.22247061485738e-06,
      "loss": 0.5726,
      "step": 52
    },
    {
      "epoch": 1.13,
      "learning_rate": 7.1160641722221255e-06,
      "loss": 0.5536,
      "step": 53
    },
    {
      "epoch": 1.15,
      "learning_rate": 7.008477123264849e-06,
      "loss": 0.5926,
      "step": 54
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.8997694935518e-06,
      "loss": 0.4939,
      "step": 55
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.7900019338499005e-06,
      "loss": 0.1926,
      "step": 56
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.6792356862881144e-06,
      "loss": 0.56,
      "step": 57
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.567532550188908e-06,
      "loss": 0.5608,
      "step": 58
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.454954847588824e-06,
      "loss": 0.5033,
      "step": 59
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.341565388467425e-06,
      "loss": 0.4969,
      "step": 60
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.227427435703997e-06,
      "loss": 0.547,
      "step": 61
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.112604669781572e-06,
      "loss": 0.535,
      "step": 62
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.997161153257963e-06,
      "loss": 0.5331,
      "step": 63
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.88116129502361e-06,
      "loss": 0.5558,
      "step": 64
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.764669814366231e-06,
      "loss": 0.5304,
      "step": 65
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.647751704862263e-06,
      "loss": 0.5342,
      "step": 66
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.530472198115291e-06,
      "loss": 0.5577,
      "step": 67
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.412896727361663e-06,
      "loss": 0.5478,
      "step": 68
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.2950908909636144e-06,
      "loss": 0.4553,
      "step": 69
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.177120415810271e-06,
      "loss": 0.5004,
      "step": 70
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.059051120646924e-06,
      "loss": 0.5255,
      "step": 71
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.940948879353078e-06,
      "loss": 0.5618,
      "step": 72
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.822879584189732e-06,
      "loss": 0.5546,
      "step": 73
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.704909109036387e-06,
      "loss": 0.5701,
      "step": 74
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.587103272638339e-06,
      "loss": 0.5303,
      "step": 75
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.46952780188471e-06,
      "loss": 0.5049,
      "step": 76
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.352248295137739e-06,
      "loss": 0.5607,
      "step": 77
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.23533018563377e-06,
      "loss": 0.5129,
      "step": 78
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.118838704976392e-06,
      "loss": 0.5127,
      "step": 79
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.002838846742039e-06,
      "loss": 0.5433,
      "step": 80
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.887395330218429e-06,
      "loss": 0.5042,
      "step": 81
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7725725642960047e-06,
      "loss": 0.5336,
      "step": 82
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.658434611532578e-06,
      "loss": 0.5502,
      "step": 83
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.545045152411178e-06,
      "loss": 0.5162,
      "step": 84
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4324674498110956e-06,
      "loss": 0.175,
      "step": 85
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.3207643137118872e-06,
      "loss": 0.5485,
      "step": 86
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.2099980661501016e-06,
      "loss": 0.5243,
      "step": 87
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1002305064482006e-06,
      "loss": 0.5174,
      "step": 88
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.991522876735154e-06,
      "loss": 0.5354,
      "step": 89
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.8839358277778758e-06,
      "loss": 0.5243,
      "step": 90
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.7775293851426233e-06,
      "loss": 0.5518,
      "step": 91
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.6723629157051844e-06,
      "loss": 0.5625,
      "step": 92
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.5684950945285937e-06,
      "loss": 0.5083,
      "step": 93
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4659838721268005e-06,
      "loss": 0.2397,
      "step": 94
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.364886442132606e-06,
      "loss": 0.4048,
      "step": 95
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.265259209387867e-06,
      "loss": 0.3924,
      "step": 96
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.16715775847379e-06,
      "loss": 0.3762,
      "step": 97
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.0706368226988772e-06,
      "loss": 0.3902,
      "step": 98
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.9757502535618137e-06,
      "loss": 0.4002,
      "step": 99
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.8825509907063328e-06,
      "loss": 0.4317,
      "step": 100
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.7910910323848435e-06,
      "loss": 0.4067,
      "step": 101
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7014214064472646e-06,
      "loss": 0.4105,
      "step": 102
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6135921418712959e-06,
      "loss": 0.3521,
      "step": 103
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.5276522408499567e-06,
      "loss": 0.369,
      "step": 104
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.4436496514520253e-06,
      "loss": 0.3684,
      "step": 105
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.361631240870569e-06,
      "loss": 0.3558,
      "step": 106
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.281642769274552e-06,
      "loss": 0.3689,
      "step": 107
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2037288642780575e-06,
      "loss": 0.3308,
      "step": 108
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1279329960414047e-06,
      "loss": 0.3509,
      "step": 109
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0542974530180327e-06,
      "loss": 0.3442,
      "step": 110
    },
    {
      "epoch": 2.37,
      "learning_rate": 9.82863318360695e-07,
      "loss": 0.3694,
      "step": 111
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.136704470001101e-07,
      "loss": 0.3714,
      "step": 112
    },
    {
      "epoch": 2.41,
      "learning_rate": 8.46757443408886e-07,
      "loss": 0.3559,
      "step": 113
    },
    {
      "epoch": 2.43,
      "learning_rate": 7.821616400630866e-07,
      "loss": 0.375,
      "step": 114
    },
    {
      "epoch": 2.45,
      "learning_rate": 7.199190766135001e-07,
      "loss": 0.3558,
      "step": 115
    },
    {
      "epoch": 2.47,
      "learning_rate": 6.600644797781847e-07,
      "loss": 0.3836,
      "step": 116
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.026312439675553e-07,
      "loss": 0.3547,
      "step": 117
    },
    {
      "epoch": 2.52,
      "learning_rate": 5.476514126527771e-07,
      "loss": 0.3729,
      "step": 118
    },
    {
      "epoch": 2.54,
      "learning_rate": 4.951556604879049e-07,
      "loss": 0.3592,
      "step": 119
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.4517327619569784e-07,
      "loss": 0.3452,
      "step": 120
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.9773214622669974e-07,
      "loss": 0.1724,
      "step": 121
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.528587392006716e-07,
      "loss": 0.3604,
      "step": 122
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.105780911390738e-07,
      "loss": 0.3588,
      "step": 123
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.7091379149682683e-07,
      "loss": 0.289,
      "step": 124
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.3388797000115427e-07,
      "loss": 0.1545,
      "step": 125
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.9952128430483718e-07,
      "loss": 0.387,
      "step": 126
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.6783290846078714e-07,
      "loss": 0.3806,
      "step": 127
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.388405222243472e-07,
      "loss": 0.3519,
      "step": 128
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.1256030118930727e-07,
      "loss": 0.3362,
      "step": 129
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.900690776312282e-08,
      "loss": 0.3886,
      "step": 130
    },
    {
      "epoch": 2.79,
      "learning_rate": 6.819348298638839e-08,
      "loss": 0.3294,
      "step": 131
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.013163920110864e-08,
      "loss": 0.3649,
      "step": 132
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.483145357187967e-08,
      "loss": 0.3504,
      "step": 133
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.230146246358256e-08,
      "loss": 0.3427,
      "step": 134
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.2548656678721404e-08,
      "loss": 0.3589,
      "step": 135
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.578477557081074e-09,
      "loss": 0.3562,
      "step": 136
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.3948139398628492e-09,
      "loss": 0.3825,
      "step": 137
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0,
      "loss": 0.3726,
      "step": 138
    },
    {
      "epoch": 2.94,
      "step": 138,
      "total_flos": 280987049787392.0,
      "train_loss": 0.5631573870778084,
      "train_runtime": 3550.0224,
      "train_samples_per_second": 5.07,
      "train_steps_per_second": 0.039
    }
  ],
  "max_steps": 138,
  "num_train_epochs": 3,
  "total_flos": 280987049787392.0,
  "trial_name": null,
  "trial_params": null
}