trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9991356957649092,
  "global_step": 867,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.58,
      "learning_rate": 2.116493656286044e-05,
      "loss": 3.1243,
      "step": 500
    },
    {
      "epoch": 1.0,
      "step": 867,
      "total_flos": 6.28807388626944e+16,
      "train_loss": 3.1041302004487457,
      "train_runtime": 5384.0575,
      "train_samples_per_second": 1.289,
      "train_steps_per_second": 0.161
    }
  ],
  "max_steps": 867,
  "num_train_epochs": 1,
  "total_flos": 6.28807388626944e+16,
  "trial_name": null,
  "trial_params": null
}