{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 12897,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.12,
      "learning_rate": 9.612312940994031e-06,
      "loss": 0.5637,
      "step": 500
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.22462588198806e-06,
      "loss": 0.4,
      "step": 1000
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.83693882298209e-06,
      "loss": 0.3797,
      "step": 1500
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.44925176397612e-06,
      "loss": 0.3645,
      "step": 2000
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.061564704970149e-06,
      "loss": 0.3484,
      "step": 2500
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.673877645964178e-06,
      "loss": 0.3472,
      "step": 3000
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.286190586958209e-06,
      "loss": 0.3356,
      "step": 3500
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.898503527952237e-06,
      "loss": 0.3341,
      "step": 4000
    },
    {
      "epoch": 1.05,
      "learning_rate": 6.5108164689462675e-06,
      "loss": 0.3156,
      "step": 4500
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.123129409940297e-06,
      "loss": 0.3061,
      "step": 5000
    },
    {
      "epoch": 1.28,
      "learning_rate": 5.735442350934326e-06,
      "loss": 0.3022,
      "step": 5500
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.347755291928356e-06,
      "loss": 0.2965,
      "step": 6000
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.960068232922385e-06,
      "loss": 0.2934,
      "step": 6500
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.572381173916415e-06,
      "loss": 0.2979,
      "step": 7000
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.184694114910445e-06,
      "loss": 0.2915,
      "step": 7500
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.797007055904474e-06,
      "loss": 0.2914,
      "step": 8000
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.4093199968985036e-06,
      "loss": 0.2909,
      "step": 8500
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.0216329378925334e-06,
      "loss": 0.2711,
      "step": 9000
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.6339458788865628e-06,
      "loss": 0.2674,
      "step": 9500
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.2462588198805926e-06,
      "loss": 0.2683,
      "step": 10000
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.858571760874622e-06,
      "loss": 0.2631,
      "step": 10500
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.4708847018686518e-06,
      "loss": 0.2637,
      "step": 11000
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.0831976428626812e-06,
      "loss": 0.2643,
      "step": 11500
    },
    {
      "epoch": 2.79,
      "learning_rate": 6.955105838567109e-07,
      "loss": 0.2689,
      "step": 12000
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.078235248507405e-07,
      "loss": 0.2695,
      "step": 12500
    },
    {
      "epoch": 3.0,
      "step": 12897,
      "total_flos": 3.159950420124426e+17,
      "train_runtime": 16177.7165,
      "train_samples_per_second": 0.797
    }
  ],
  "max_steps": 12897,
  "num_train_epochs": 3,
  "total_flos": 3.159950420124426e+17,
  "trial_name": null,
  "trial_params": null
}