{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 83.33333333333333,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.39,
      "learning_rate": 3e-06,
      "loss": 7.0999,
      "step": 50
    },
    {
      "epoch": 2.78,
      "learning_rate": 6e-06,
      "loss": 3.9223,
      "step": 100
    },
    {
      "epoch": 4.17,
      "learning_rate": 9e-06,
      "loss": 2.5274,
      "step": 150
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.2e-05,
      "loss": 2.1012,
      "step": 200
    },
    {
      "epoch": 6.94,
      "learning_rate": 1.5e-05,
      "loss": 1.9323,
      "step": 250
    },
    {
      "epoch": 8.33,
      "learning_rate": 1.8e-05,
      "loss": 1.7811,
      "step": 300
    },
    {
      "epoch": 9.72,
      "learning_rate": 2.1e-05,
      "loss": 1.7308,
      "step": 350
    },
    {
      "epoch": 11.11,
      "learning_rate": 2.4e-05,
      "loss": 1.6515,
      "step": 400
    },
    {
      "epoch": 12.5,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 1.6299,
      "step": 450
    },
    {
      "epoch": 13.89,
      "learning_rate": 3e-05,
      "loss": 1.6019,
      "step": 500
    },
    {
      "epoch": 15.28,
      "learning_rate": 2.9516129032258067e-05,
      "loss": 1.5727,
      "step": 550
    },
    {
      "epoch": 16.67,
      "learning_rate": 2.903225806451613e-05,
      "loss": 1.5442,
      "step": 600
    },
    {
      "epoch": 18.06,
      "learning_rate": 2.8548387096774196e-05,
      "loss": 1.5382,
      "step": 650
    },
    {
      "epoch": 19.44,
      "learning_rate": 2.806451612903226e-05,
      "loss": 1.5224,
      "step": 700
    },
    {
      "epoch": 20.83,
      "learning_rate": 2.758064516129032e-05,
      "loss": 1.5323,
      "step": 750
    },
    {
      "epoch": 22.22,
      "learning_rate": 2.7096774193548387e-05,
      "loss": 1.5036,
      "step": 800
    },
    {
      "epoch": 23.61,
      "learning_rate": 2.6612903225806453e-05,
      "loss": 1.4981,
      "step": 850
    },
    {
      "epoch": 25.0,
      "learning_rate": 2.6129032258064516e-05,
      "loss": 1.5055,
      "step": 900
    },
    {
      "epoch": 26.39,
      "learning_rate": 2.5645161290322582e-05,
      "loss": 1.4892,
      "step": 950
    },
    {
      "epoch": 27.78,
      "learning_rate": 2.5161290322580648e-05,
      "loss": 1.4892,
      "step": 1000
    },
    {
      "epoch": 29.17,
      "learning_rate": 2.467741935483871e-05,
      "loss": 1.4802,
      "step": 1050
    },
    {
      "epoch": 30.56,
      "learning_rate": 2.4193548387096773e-05,
      "loss": 1.4804,
      "step": 1100
    },
    {
      "epoch": 31.94,
      "learning_rate": 2.370967741935484e-05,
      "loss": 1.4692,
      "step": 1150
    },
    {
      "epoch": 33.33,
      "learning_rate": 2.3225806451612902e-05,
      "loss": 1.478,
      "step": 1200
    },
    {
      "epoch": 34.72,
      "learning_rate": 2.274193548387097e-05,
      "loss": 1.4692,
      "step": 1250
    },
    {
      "epoch": 36.11,
      "learning_rate": 2.2258064516129034e-05,
      "loss": 1.465,
      "step": 1300
    },
    {
      "epoch": 37.5,
      "learning_rate": 2.1774193548387097e-05,
      "loss": 1.4653,
      "step": 1350
    },
    {
      "epoch": 38.89,
      "learning_rate": 2.1290322580645163e-05,
      "loss": 1.4636,
      "step": 1400
    },
    {
      "epoch": 40.28,
      "learning_rate": 2.080645161290323e-05,
      "loss": 1.467,
      "step": 1450
    },
    {
      "epoch": 41.67,
      "learning_rate": 2.032258064516129e-05,
      "loss": 1.4605,
      "step": 1500
    },
    {
      "epoch": 43.06,
      "learning_rate": 1.9838709677419355e-05,
      "loss": 1.4617,
      "step": 1550
    },
    {
      "epoch": 44.44,
      "learning_rate": 1.935483870967742e-05,
      "loss": 1.4632,
      "step": 1600
    },
    {
      "epoch": 45.83,
      "learning_rate": 1.8870967741935484e-05,
      "loss": 1.4641,
      "step": 1650
    },
    {
      "epoch": 47.22,
      "learning_rate": 1.838709677419355e-05,
      "loss": 1.4528,
      "step": 1700
    },
    {
      "epoch": 48.61,
      "learning_rate": 1.7903225806451616e-05,
      "loss": 1.4566,
      "step": 1750
    },
    {
      "epoch": 50.0,
      "learning_rate": 1.741935483870968e-05,
      "loss": 1.4522,
      "step": 1800
    },
    {
      "epoch": 51.39,
      "learning_rate": 1.6935483870967744e-05,
      "loss": 1.4598,
      "step": 1850
    },
    {
      "epoch": 52.78,
      "learning_rate": 1.6451612903225807e-05,
      "loss": 1.454,
      "step": 1900
    },
    {
      "epoch": 54.17,
      "learning_rate": 1.596774193548387e-05,
      "loss": 1.4568,
      "step": 1950
    },
    {
      "epoch": 55.56,
      "learning_rate": 1.5483870967741936e-05,
      "loss": 1.4528,
      "step": 2000
    },
    {
      "epoch": 56.94,
      "learning_rate": 1.5e-05,
      "loss": 1.4471,
      "step": 2050
    },
    {
      "epoch": 58.33,
      "learning_rate": 1.4516129032258065e-05,
      "loss": 1.4471,
      "step": 2100
    },
    {
      "epoch": 59.72,
      "learning_rate": 1.403225806451613e-05,
      "loss": 1.4554,
      "step": 2150
    },
    {
      "epoch": 61.11,
      "learning_rate": 1.3548387096774194e-05,
      "loss": 1.4435,
      "step": 2200
    },
    {
      "epoch": 62.5,
      "learning_rate": 1.3064516129032258e-05,
      "loss": 1.4456,
      "step": 2250
    },
    {
      "epoch": 63.89,
      "learning_rate": 1.2580645161290324e-05,
      "loss": 1.4452,
      "step": 2300
    },
    {
      "epoch": 65.28,
      "learning_rate": 1.2096774193548387e-05,
      "loss": 1.4496,
      "step": 2350
    },
    {
      "epoch": 66.67,
      "learning_rate": 1.1612903225806451e-05,
      "loss": 1.4431,
      "step": 2400
    },
    {
      "epoch": 68.06,
      "learning_rate": 1.1129032258064517e-05,
      "loss": 1.4433,
      "step": 2450
    },
    {
      "epoch": 69.44,
      "learning_rate": 1.0645161290322582e-05,
      "loss": 1.4437,
      "step": 2500
    },
    {
      "epoch": 70.83,
      "learning_rate": 1.0161290322580644e-05,
      "loss": 1.4414,
      "step": 2550
    },
    {
      "epoch": 72.22,
      "learning_rate": 9.67741935483871e-06,
      "loss": 1.4413,
      "step": 2600
    },
    {
      "epoch": 73.61,
      "learning_rate": 9.193548387096775e-06,
      "loss": 1.4414,
      "step": 2650
    },
    {
      "epoch": 75.0,
      "learning_rate": 8.70967741935484e-06,
      "loss": 1.4396,
      "step": 2700
    },
    {
      "epoch": 76.39,
      "learning_rate": 8.225806451612904e-06,
      "loss": 1.4415,
      "step": 2750
    },
    {
      "epoch": 77.78,
      "learning_rate": 7.741935483870968e-06,
      "loss": 1.443,
      "step": 2800
    },
    {
      "epoch": 79.17,
      "learning_rate": 7.258064516129032e-06,
      "loss": 1.4432,
      "step": 2850
    },
    {
      "epoch": 80.56,
      "learning_rate": 6.774193548387097e-06,
      "loss": 1.4389,
      "step": 2900
    },
    {
      "epoch": 81.94,
      "learning_rate": 6.290322580645162e-06,
      "loss": 1.4383,
      "step": 2950
    },
    {
      "epoch": 83.33,
      "learning_rate": 5.8064516129032256e-06,
      "loss": 1.4387,
      "step": 3000
    }
  ],
  "max_steps": 3600,
  "num_train_epochs": 100,
  "total_flos": 7215621246812160.0,
  "trial_name": null,
  "trial_params": null
}