{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 126.66666666666667,
  "global_step": 3800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.67,
      "learning_rate": 4.9542483660130725e-05,
      "loss": 4.085,
      "step": 50
    },
    {
      "epoch": 3.33,
      "learning_rate": 4.888888888888889e-05,
      "loss": 2.4616,
      "step": 100
    },
    {
      "epoch": 5.0,
      "learning_rate": 4.823529411764706e-05,
      "loss": 1.4793,
      "step": 150
    },
    {
      "epoch": 6.67,
      "learning_rate": 4.7581699346405226e-05,
      "loss": 0.7873,
      "step": 200
    },
    {
      "epoch": 8.33,
      "learning_rate": 4.69281045751634e-05,
      "loss": 0.4859,
      "step": 250
    },
    {
      "epoch": 10.0,
      "learning_rate": 4.6274509803921576e-05,
      "loss": 0.3261,
      "step": 300
    },
    {
      "epoch": 11.67,
      "learning_rate": 4.562091503267974e-05,
      "loss": 0.2257,
      "step": 350
    },
    {
      "epoch": 13.33,
      "learning_rate": 4.496732026143791e-05,
      "loss": 0.1711,
      "step": 400
    },
    {
      "epoch": 15.0,
      "learning_rate": 4.431372549019608e-05,
      "loss": 0.126,
      "step": 450
    },
    {
      "epoch": 16.67,
      "learning_rate": 4.366013071895425e-05,
      "loss": 0.0819,
      "step": 500
    },
    {
      "epoch": 18.33,
      "learning_rate": 4.300653594771242e-05,
      "loss": 0.0699,
      "step": 550
    },
    {
      "epoch": 20.0,
      "learning_rate": 4.235294117647059e-05,
      "loss": 0.0568,
      "step": 600
    },
    {
      "epoch": 21.67,
      "learning_rate": 4.169934640522876e-05,
      "loss": 0.0473,
      "step": 650
    },
    {
      "epoch": 23.33,
      "learning_rate": 4.104575163398693e-05,
      "loss": 0.0428,
      "step": 700
    },
    {
      "epoch": 25.0,
      "learning_rate": 4.03921568627451e-05,
      "loss": 0.0382,
      "step": 750
    },
    {
      "epoch": 26.67,
      "learning_rate": 3.973856209150327e-05,
      "loss": 0.033,
      "step": 800
    },
    {
      "epoch": 28.33,
      "learning_rate": 3.908496732026144e-05,
      "loss": 0.0306,
      "step": 850
    },
    {
      "epoch": 30.0,
      "learning_rate": 3.8431372549019614e-05,
      "loss": 0.0272,
      "step": 900
    },
    {
      "epoch": 31.67,
      "learning_rate": 3.777777777777778e-05,
      "loss": 0.0225,
      "step": 950
    },
    {
      "epoch": 33.33,
      "learning_rate": 3.712418300653595e-05,
      "loss": 0.0211,
      "step": 1000
    },
    {
      "epoch": 35.0,
      "learning_rate": 3.6470588235294114e-05,
      "loss": 0.0196,
      "step": 1050
    },
    {
      "epoch": 36.67,
      "learning_rate": 3.581699346405229e-05,
      "loss": 0.0167,
      "step": 1100
    },
    {
      "epoch": 38.33,
      "learning_rate": 3.5163398692810464e-05,
      "loss": 0.0158,
      "step": 1150
    },
    {
      "epoch": 40.0,
      "learning_rate": 3.450980392156863e-05,
      "loss": 0.0168,
      "step": 1200
    },
    {
      "epoch": 41.67,
      "learning_rate": 3.38562091503268e-05,
      "loss": 0.0101,
      "step": 1250
    },
    {
      "epoch": 43.33,
      "learning_rate": 3.3202614379084965e-05,
      "loss": 0.0128,
      "step": 1300
    },
    {
      "epoch": 45.0,
      "learning_rate": 3.254901960784314e-05,
      "loss": 0.0077,
      "step": 1350
    },
    {
      "epoch": 46.67,
      "learning_rate": 3.189542483660131e-05,
      "loss": 0.0065,
      "step": 1400
    },
    {
      "epoch": 48.33,
      "learning_rate": 3.124183006535948e-05,
      "loss": 0.0053,
      "step": 1450
    },
    {
      "epoch": 50.0,
      "learning_rate": 3.058823529411765e-05,
      "loss": 0.0064,
      "step": 1500
    },
    {
      "epoch": 51.67,
      "learning_rate": 2.9934640522875816e-05,
      "loss": 0.0059,
      "step": 1550
    },
    {
      "epoch": 53.33,
      "learning_rate": 2.928104575163399e-05,
      "loss": 0.0044,
      "step": 1600
    },
    {
      "epoch": 55.0,
      "learning_rate": 2.8627450980392155e-05,
      "loss": 0.0057,
      "step": 1650
    },
    {
      "epoch": 56.67,
      "learning_rate": 2.7973856209150327e-05,
      "loss": 0.0043,
      "step": 1700
    },
    {
      "epoch": 58.33,
      "learning_rate": 2.7320261437908495e-05,
      "loss": 0.0042,
      "step": 1750
    },
    {
      "epoch": 60.0,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.0024,
      "step": 1800
    },
    {
      "epoch": 61.67,
      "learning_rate": 2.601307189542484e-05,
      "loss": 0.003,
      "step": 1850
    },
    {
      "epoch": 63.33,
      "learning_rate": 2.5359477124183006e-05,
      "loss": 0.0037,
      "step": 1900
    },
    {
      "epoch": 65.0,
      "learning_rate": 2.4705882352941178e-05,
      "loss": 0.0035,
      "step": 1950
    },
    {
      "epoch": 66.67,
      "learning_rate": 2.405228758169935e-05,
      "loss": 0.0019,
      "step": 2000
    },
    {
      "epoch": 68.33,
      "learning_rate": 2.3398692810457517e-05,
      "loss": 0.0027,
      "step": 2050
    },
    {
      "epoch": 70.0,
      "learning_rate": 2.2745098039215685e-05,
      "loss": 0.003,
      "step": 2100
    },
    {
      "epoch": 71.67,
      "learning_rate": 2.2091503267973857e-05,
      "loss": 0.0028,
      "step": 2150
    },
    {
      "epoch": 73.33,
      "learning_rate": 2.1437908496732025e-05,
      "loss": 0.0014,
      "step": 2200
    },
    {
      "epoch": 75.0,
      "learning_rate": 2.0784313725490197e-05,
      "loss": 0.0011,
      "step": 2250
    },
    {
      "epoch": 76.67,
      "learning_rate": 2.0130718954248368e-05,
      "loss": 0.0017,
      "step": 2300
    },
    {
      "epoch": 78.33,
      "learning_rate": 1.9477124183006536e-05,
      "loss": 0.0007,
      "step": 2350
    },
    {
      "epoch": 80.0,
      "learning_rate": 1.8823529411764708e-05,
      "loss": 0.0006,
      "step": 2400
    },
    {
      "epoch": 81.67,
      "learning_rate": 1.8169934640522876e-05,
      "loss": 0.0005,
      "step": 2450
    },
    {
      "epoch": 83.33,
      "learning_rate": 1.7516339869281044e-05,
      "loss": 0.0005,
      "step": 2500
    },
    {
      "epoch": 85.0,
      "learning_rate": 1.686274509803922e-05,
      "loss": 0.0005,
      "step": 2550
    },
    {
      "epoch": 86.67,
      "learning_rate": 1.6209150326797387e-05,
      "loss": 0.0005,
      "step": 2600
    },
    {
      "epoch": 88.33,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.0005,
      "step": 2650
    },
    {
      "epoch": 90.0,
      "learning_rate": 1.4901960784313726e-05,
      "loss": 0.0006,
      "step": 2700
    },
    {
      "epoch": 91.67,
      "learning_rate": 1.4248366013071896e-05,
      "loss": 0.0017,
      "step": 2750
    },
    {
      "epoch": 93.33,
      "learning_rate": 1.3594771241830064e-05,
      "loss": 0.0009,
      "step": 2800
    },
    {
      "epoch": 95.0,
      "learning_rate": 1.2941176470588238e-05,
      "loss": 0.0008,
      "step": 2850
    },
    {
      "epoch": 96.67,
      "learning_rate": 1.2287581699346406e-05,
      "loss": 0.0005,
      "step": 2900
    },
    {
      "epoch": 98.33,
      "learning_rate": 1.1633986928104575e-05,
      "loss": 0.0004,
      "step": 2950
    },
    {
      "epoch": 100.0,
      "learning_rate": 1.0980392156862745e-05,
      "loss": 0.0005,
      "step": 3000
    },
    {
      "epoch": 101.67,
      "learning_rate": 1.0326797385620915e-05,
      "loss": 0.0004,
      "step": 3050
    },
    {
      "epoch": 103.33,
      "learning_rate": 9.673202614379087e-06,
      "loss": 0.0004,
      "step": 3100
    },
    {
      "epoch": 105.0,
      "learning_rate": 9.019607843137255e-06,
      "loss": 0.0004,
      "step": 3150
    },
    {
      "epoch": 106.67,
      "learning_rate": 8.366013071895424e-06,
      "loss": 0.0004,
      "step": 3200
    },
    {
      "epoch": 108.33,
      "learning_rate": 7.712418300653596e-06,
      "loss": 0.0004,
      "step": 3250
    },
    {
      "epoch": 110.0,
      "learning_rate": 7.058823529411765e-06,
      "loss": 0.0003,
      "step": 3300
    },
    {
      "epoch": 111.67,
      "learning_rate": 6.405228758169935e-06,
      "loss": 0.0003,
      "step": 3350
    },
    {
      "epoch": 113.33,
      "learning_rate": 5.7516339869281045e-06,
      "loss": 0.0003,
      "step": 3400
    },
    {
      "epoch": 115.0,
      "learning_rate": 5.098039215686274e-06,
      "loss": 0.0004,
      "step": 3450
    },
    {
      "epoch": 116.67,
      "learning_rate": 4.444444444444445e-06,
      "loss": 0.0004,
      "step": 3500
    },
    {
      "epoch": 118.33,
      "learning_rate": 3.7908496732026144e-06,
      "loss": 0.0003,
      "step": 3550
    },
    {
      "epoch": 120.0,
      "learning_rate": 3.1372549019607846e-06,
      "loss": 0.0003,
      "step": 3600
    },
    {
      "epoch": 121.67,
      "learning_rate": 2.4836601307189544e-06,
      "loss": 0.0003,
      "step": 3650
    },
    {
      "epoch": 123.33,
      "learning_rate": 1.8300653594771244e-06,
      "loss": 0.0003,
      "step": 3700
    },
    {
      "epoch": 125.0,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 0.0003,
      "step": 3750
    },
    {
      "epoch": 126.67,
      "learning_rate": 5.22875816993464e-07,
      "loss": 0.0003,
      "step": 3800
    }
  ],
  "max_steps": 3840,
  "num_train_epochs": 128,
  "total_flos": 29937460838400000,
  "trial_name": null,
  "trial_params": null
}