trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 30.0,
  "global_step": 1380,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 10.87,
      "learning_rate": 3.188405797101449e-05,
      "loss": 2.0841,
      "step": 500
    },
    {
      "epoch": 21.74,
      "learning_rate": 1.3768115942028985e-05,
      "loss": 1.6236,
      "step": 1000
    },
    {
      "epoch": 30.0,
      "step": 1380,
      "total_flos": 2906450714787840.0,
      "train_loss": 1.7510387918223504,
      "train_runtime": 776.791,
      "train_samples_per_second": 14.212,
      "train_steps_per_second": 1.777
    }
  ],
  "max_steps": 1380,
  "num_train_epochs": 30,
  "total_flos": 2906450714787840.0,
  "trial_name": null,
  "trial_params": null
}