trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 48.78048780487805,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 12.2,
      "learning_rate": 4.390243902439025e-05,
      "loss": 0.8905,
      "step": 500
    },
    {
      "epoch": 24.39,
      "learning_rate": 3.780487804878049e-05,
      "loss": 0.0511,
      "step": 1000
    },
    {
      "epoch": 36.59,
      "learning_rate": 3.170731707317073e-05,
      "loss": 0.0304,
      "step": 1500
    },
    {
      "epoch": 48.78,
      "learning_rate": 2.5609756097560977e-05,
      "loss": 0.0265,
      "step": 2000
    }
  ],
  "max_steps": 4100,
  "num_train_epochs": 100,
  "total_flos": 2065509794709504.0,
  "trial_name": null,
  "trial_params": null
}