{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 4806,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "learning_rate": 0.00013305613305613308,
      "loss": 1.0196,
      "step": 320
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019264739884393065,
      "loss": 0.8615,
      "step": 640
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00017784971098265897,
      "loss": 0.8536,
      "step": 960
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001630520231213873,
      "loss": 0.8487,
      "step": 1280
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001482543352601156,
      "loss": 0.8546,
      "step": 1600
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00013345664739884393,
      "loss": 0.826,
      "step": 1920
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011865895953757227,
      "loss": 0.8328,
      "step": 2240
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00010386127167630057,
      "loss": 0.8243,
      "step": 2560
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.906358381502891e-05,
      "loss": 0.824,
      "step": 2880
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.426589595375722e-05,
      "loss": 0.8275,
      "step": 3200
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.946820809248556e-05,
      "loss": 0.8031,
      "step": 3520
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.467052023121387e-05,
      "loss": 0.7995,
      "step": 3840
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.98728323699422e-05,
      "loss": 0.8027,
      "step": 4160
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.5075144508670521e-05,
      "loss": 0.8074,
      "step": 4480
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.7745664739884395e-07,
      "loss": 0.801,
      "step": 4800
    }
  ],
  "max_steps": 4806,
  "num_train_epochs": 3,
  "total_flos": 3.989512556547932e+17,
  "trial_name": null,
  "trial_params": null
}