{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.984,
  "eval_steps": 500,
  "global_step": 124,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 5e-06,
      "loss": 2.9684,
      "step": 1
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 3.0316,
      "step": 2
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 2.8996,
      "step": 3
    },
    {
      "epoch": 0.06,
      "learning_rate": 2e-05,
      "loss": 2.7286,
      "step": 4
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9996573249755573e-05,
      "loss": 2.8527,
      "step": 5
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9986295347545738e-05,
      "loss": 2.5942,
      "step": 6
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9969173337331283e-05,
      "loss": 2.4836,
      "step": 7
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9945218953682736e-05,
      "loss": 2.666,
      "step": 8
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9914448613738107e-05,
      "loss": 2.3577,
      "step": 9
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9876883405951378e-05,
      "loss": 2.4005,
      "step": 10
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.983254907563955e-05,
      "loss": 2.3426,
      "step": 11
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9781476007338058e-05,
      "loss": 2.1247,
      "step": 12
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9723699203976768e-05,
      "loss": 2.207,
      "step": 13
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9659258262890683e-05,
      "loss": 2.0202,
      "step": 14
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.958819734868193e-05,
      "loss": 2.1052,
      "step": 15
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 2.1671,
      "step": 16
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9426414910921785e-05,
      "loss": 2.1553,
      "step": 17
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9335804264972018e-05,
      "loss": 2.0372,
      "step": 18
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 1.9472,
      "step": 19
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.913545457642601e-05,
      "loss": 1.8312,
      "step": 20
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.902585284349861e-05,
      "loss": 1.9444,
      "step": 21
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.891006524188368e-05,
      "loss": 1.9848,
      "step": 22
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.8788171126619653e-05,
      "loss": 1.9345,
      "step": 23
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.866025403784439e-05,
      "loss": 2.0124,
      "step": 24
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.8526401643540924e-05,
      "loss": 1.7981,
      "step": 25
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.8386705679454243e-05,
      "loss": 1.8936,
      "step": 26
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.8241261886220155e-05,
      "loss": 1.7496,
      "step": 27
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 1.8839,
      "step": 28
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.7933533402912354e-05,
      "loss": 1.8223,
      "step": 29
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.777145961456971e-05,
      "loss": 1.7236,
      "step": 30
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.7604059656000313e-05,
      "loss": 1.7847,
      "step": 31
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.7431448254773943e-05,
      "loss": 1.7522,
      "step": 32
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.7253743710122877e-05,
      "loss": 1.793,
      "step": 33
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 1.7028,
      "step": 34
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.688354575693754e-05,
      "loss": 1.741,
      "step": 35
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.6691306063588583e-05,
      "loss": 1.6532,
      "step": 36
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6494480483301836e-05,
      "loss": 1.7473,
      "step": 37
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.6293203910498375e-05,
      "loss": 1.6594,
      "step": 38
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.608761429008721e-05,
      "loss": 1.7145,
      "step": 39
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 1.711,
      "step": 40
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.566406236924833e-05,
      "loss": 1.6677,
      "step": 41
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.5446390350150272e-05,
      "loss": 1.6504,
      "step": 42
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.5224985647159489e-05,
      "loss": 1.6619,
      "step": 43
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.6082,
      "step": 44
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.4771587602596085e-05,
      "loss": 1.6774,
      "step": 45
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.4539904997395468e-05,
      "loss": 1.6716,
      "step": 46
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.4305110968082953e-05,
      "loss": 1.5922,
      "step": 47
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.4067366430758004e-05,
      "loss": 1.5776,
      "step": 48
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 1.5135,
      "step": 49
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.3583679495453e-05,
      "loss": 1.6156,
      "step": 50
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.333806859233771e-05,
      "loss": 1.4918,
      "step": 51
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 1.6014,
      "step": 52
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.284015344703923e-05,
      "loss": 1.6738,
      "step": 53
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.2588190451025209e-05,
      "loss": 1.5877,
      "step": 54
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.2334453638559057e-05,
      "loss": 1.5073,
      "step": 55
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.2079116908177592e-05,
      "loss": 1.5979,
      "step": 56
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.1822355254921478e-05,
      "loss": 1.5981,
      "step": 57
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.156434465040231e-05,
      "loss": 1.5809,
      "step": 58
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.130526192220052e-05,
      "loss": 1.6249,
      "step": 59
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.1045284632676535e-05,
      "loss": 1.5176,
      "step": 60
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.0784590957278452e-05,
      "loss": 1.5136,
      "step": 61
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0523359562429441e-05,
      "loss": 1.5118,
      "step": 62
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.0261769483078734e-05,
      "loss": 1.5413,
      "step": 63
    },
    {
      "epoch": 1.02,
      "learning_rate": 1e-05,
      "loss": 1.5132,
      "step": 64
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.738230516921272e-06,
      "loss": 1.4406,
      "step": 65
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.476640437570562e-06,
      "loss": 1.4744,
      "step": 66
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.215409042721553e-06,
      "loss": 1.4483,
      "step": 67
    },
    {
      "epoch": 1.09,
      "learning_rate": 8.954715367323468e-06,
      "loss": 1.4445,
      "step": 68
    },
    {
      "epoch": 1.1,
      "learning_rate": 8.694738077799487e-06,
      "loss": 1.4893,
      "step": 69
    },
    {
      "epoch": 1.12,
      "learning_rate": 8.43565534959769e-06,
      "loss": 1.4632,
      "step": 70
    },
    {
      "epoch": 1.14,
      "learning_rate": 8.177644745078525e-06,
      "loss": 1.4622,
      "step": 71
    },
    {
      "epoch": 1.15,
      "learning_rate": 7.92088309182241e-06,
      "loss": 1.4093,
      "step": 72
    },
    {
      "epoch": 1.17,
      "learning_rate": 7.66554636144095e-06,
      "loss": 1.3794,
      "step": 73
    },
    {
      "epoch": 1.18,
      "learning_rate": 7.411809548974792e-06,
      "loss": 1.467,
      "step": 74
    },
    {
      "epoch": 1.2,
      "learning_rate": 7.159846552960774e-06,
      "loss": 1.4391,
      "step": 75
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.909830056250527e-06,
      "loss": 1.3792,
      "step": 76
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.661931407662292e-06,
      "loss": 1.3963,
      "step": 77
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.4163205045469975e-06,
      "loss": 1.3913,
      "step": 78
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.173165676349103e-06,
      "loss": 1.4604,
      "step": 79
    },
    {
      "epoch": 1.28,
      "learning_rate": 5.932633569242e-06,
      "loss": 1.3834,
      "step": 80
    },
    {
      "epoch": 1.3,
      "learning_rate": 5.694889031917047e-06,
      "loss": 1.3693,
      "step": 81
    },
    {
      "epoch": 1.31,
      "learning_rate": 5.460095002604533e-06,
      "loss": 1.3386,
      "step": 82
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.228412397403916e-06,
      "loss": 1.3589,
      "step": 83
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.000000000000003e-06,
      "loss": 1.3702,
      "step": 84
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.775014352840512e-06,
      "loss": 1.3905,
      "step": 85
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.5536096498497295e-06,
      "loss": 1.4237,
      "step": 86
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.335937630751675e-06,
      "loss": 1.4223,
      "step": 87
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.12214747707527e-06,
      "loss": 1.4153,
      "step": 88
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.912385709912794e-06,
      "loss": 1.4039,
      "step": 89
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.7067960895016277e-06,
      "loss": 1.3521,
      "step": 90
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.505519516698165e-06,
      "loss": 1.445,
      "step": 91
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.308693936411421e-06,
      "loss": 1.3813,
      "step": 92
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.116454243062459e-06,
      "loss": 1.3369,
      "step": 93
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 1.367,
      "step": 94
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.746256289877126e-06,
      "loss": 1.3921,
      "step": 95
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.5685517452260566e-06,
      "loss": 1.3977,
      "step": 96
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.395940343999691e-06,
      "loss": 1.5059,
      "step": 97
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.2285403854302912e-06,
      "loss": 1.3491,
      "step": 98
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.0664665970876496e-06,
      "loss": 1.3816,
      "step": 99
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 1.3663,
      "step": 100
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.7587381137798432e-06,
      "loss": 1.4237,
      "step": 101
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.6132943205457607e-06,
      "loss": 1.4992,
      "step": 102
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.4735983564590784e-06,
      "loss": 1.4245,
      "step": 103
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.339745962155613e-06,
      "loss": 1.4137,
      "step": 104
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.2118288733803474e-06,
      "loss": 1.3232,
      "step": 105
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.0899347581163222e-06,
      "loss": 1.3351,
      "step": 106
    },
    {
      "epoch": 1.71,
      "learning_rate": 9.74147156501396e-07,
      "loss": 1.443,
      "step": 107
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.645454235739903e-07,
      "loss": 1.3864,
      "step": 108
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.612046748871327e-07,
      "loss": 1.4728,
      "step": 109
    },
    {
      "epoch": 1.76,
      "learning_rate": 6.641957350279838e-07,
      "loss": 1.3847,
      "step": 110
    },
    {
      "epoch": 1.78,
      "learning_rate": 5.735850890782158e-07,
      "loss": 1.4876,
      "step": 111
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.894348370484648e-07,
      "loss": 1.4311,
      "step": 112
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.118026513180695e-07,
      "loss": 1.459,
      "step": 113
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.4074173710931804e-07,
      "loss": 1.4378,
      "step": 114
    },
    {
      "epoch": 1.84,
      "learning_rate": 2.7630079602323447e-07,
      "loss": 1.3836,
      "step": 115
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.1852399266194312e-07,
      "loss": 1.346,
      "step": 116
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.6745092436045495e-07,
      "loss": 1.4084,
      "step": 117
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.231165940486234e-07,
      "loss": 1.3656,
      "step": 118
    },
    {
      "epoch": 1.9,
      "learning_rate": 8.555138626189619e-08,
      "loss": 1.473,
      "step": 119
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.4781046317267103e-08,
      "loss": 1.4061,
      "step": 120
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.082666266872036e-08,
      "loss": 1.3331,
      "step": 121
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.370465245426167e-08,
      "loss": 1.4088,
      "step": 122
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.4267502444274013e-09,
      "loss": 1.3246,
      "step": 123
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0,
      "loss": 1.4014,
      "step": 124
    },
    {
      "epoch": 1.98,
      "step": 124,
      "total_flos": 2.688391775571149e+16,
      "train_loss": 1.6652600323000262,
      "train_runtime": 1279.2707,
      "train_samples_per_second": 12.507,
      "train_steps_per_second": 0.097
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 124,
  "num_train_epochs": 2,
  "save_steps": 1000,
  "total_flos": 2.688391775571149e+16,
  "trial_name": null,
  "trial_params": null
}