trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.9983931440814136,
  "global_step": 1398,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.15,
      "learning_rate": 6.42346208869814e-06,
      "loss": 2.369,
      "step": 500
    },
    {
      "epoch": 4.29,
      "learning_rate": 2.8469241773962807e-06,
      "loss": 2.0863,
      "step": 1000
    },
    {
      "epoch": 6.0,
      "step": 1398,
      "total_flos": 3.490577523475661e+17,
      "train_loss": 2.174733916406809,
      "train_runtime": 8756.4326,
      "train_samples_per_second": 191.853,
      "train_steps_per_second": 0.16
    }
  ],
  "max_steps": 1398,
  "num_train_epochs": 6,
  "total_flos": 3.490577523475661e+17,
  "trial_name": null,
  "trial_params": null
}