checkpoint-1500/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.23539104337079975,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 1.3674,
      "step": 1000
    }
  ],
  "max_steps": 3000,
  "num_train_epochs": 1,
  "total_flos": 2.430441850634404e+17,
  "trial_name": null,
  "trial_params": null
}