{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.960769520935496,
  "global_step": 21000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.09,
      "learning_rate": 1.952847981893625e-05,
      "loss": 1.3455,
      "step": 500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.90569596378725e-05,
      "loss": 0.6266,
      "step": 1000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.8585439456808755e-05,
      "loss": 0.6164,
      "step": 1500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.8113919275745004e-05,
      "loss": 0.6065,
      "step": 2000
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.7642399094681254e-05,
      "loss": 0.6048,
      "step": 2500
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.7170878913617504e-05,
      "loss": 0.602,
      "step": 3000
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.6699358732553754e-05,
      "loss": 0.5918,
      "step": 3500
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.6227838551490004e-05,
      "loss": 0.5913,
      "step": 4000
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.5756318370426257e-05,
      "loss": 0.5869,
      "step": 4500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.5284798189362507e-05,
      "loss": 0.5885,
      "step": 5000
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.4813278008298757e-05,
      "loss": 0.5795,
      "step": 5500
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4341757827235009e-05,
      "loss": 0.5732,
      "step": 6000
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3870237646171257e-05,
      "loss": 0.5744,
      "step": 6500
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.3398717465107507e-05,
      "loss": 0.5726,
      "step": 7000
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.2927197284043758e-05,
      "loss": 0.5695,
      "step": 7500
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.2455677102980008e-05,
      "loss": 0.5672,
      "step": 8000
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.198415692191626e-05,
      "loss": 0.5706,
      "step": 8500
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.151263674085251e-05,
      "loss": 0.5673,
      "step": 9000
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.1041116559788761e-05,
      "loss": 0.5702,
      "step": 9500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.056959637872501e-05,
      "loss": 0.5596,
      "step": 10000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.0098076197661262e-05,
      "loss": 0.5589,
      "step": 10500
    },
    {
      "epoch": 2.07,
      "learning_rate": 9.62655601659751e-06,
      "loss": 0.5565,
      "step": 11000
    },
    {
      "epoch": 2.17,
      "learning_rate": 9.155035835533762e-06,
      "loss": 0.5544,
      "step": 11500
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.683515654470012e-06,
      "loss": 0.5525,
      "step": 12000
    },
    {
      "epoch": 2.36,
      "learning_rate": 8.211995473406262e-06,
      "loss": 0.5543,
      "step": 12500
    },
    {
      "epoch": 2.45,
      "learning_rate": 7.740475292342513e-06,
      "loss": 0.5525,
      "step": 13000
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.268955111278764e-06,
      "loss": 0.5532,
      "step": 13500
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.797434930215013e-06,
      "loss": 0.5512,
      "step": 14000
    },
    {
      "epoch": 2.73,
      "learning_rate": 6.325914749151264e-06,
      "loss": 0.5517,
      "step": 14500
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.854394568087515e-06,
      "loss": 0.5506,
      "step": 15000
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.382874387023765e-06,
      "loss": 0.5508,
      "step": 15500
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.911354205960015e-06,
      "loss": 0.5508,
      "step": 16000
    },
    {
      "epoch": 3.11,
      "learning_rate": 4.439834024896266e-06,
      "loss": 0.5435,
      "step": 16500
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.968313843832517e-06,
      "loss": 0.5448,
      "step": 17000
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.4967936627687665e-06,
      "loss": 0.5431,
      "step": 17500
    },
    {
      "epoch": 3.39,
      "learning_rate": 3.025273481705017e-06,
      "loss": 0.5436,
      "step": 18000
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.553753300641268e-06,
      "loss": 0.5407,
      "step": 18500
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.082233119577518e-06,
      "loss": 0.5449,
      "step": 19000
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.6107129385137685e-06,
      "loss": 0.5436,
      "step": 19500
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.139192757450019e-06,
      "loss": 0.5468,
      "step": 20000
    },
    {
      "epoch": 3.87,
      "learning_rate": 6.676725763862695e-07,
      "loss": 0.5423,
      "step": 20500
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.9615239532251984e-07,
      "loss": 0.5506,
      "step": 21000
    }
  ],
  "max_steps": 21208,
  "num_train_epochs": 4,
  "total_flos": 3.75634128609792e+16,
  "trial_name": null,
  "trial_params": null
}