checkpoint-500/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8620689655172413,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.86,
      "grad_norm": 2.777045488357544,
      "learning_rate": 3.563218390804598e-05,
      "loss": 3.2654,
      "max_memory_allocated (GB)": 9.66,
      "memory_allocated (GB)": 7.96,
      "step": 500,
      "total_memory_available (GB)": 94.62
    }
  ],
  "logging_steps": 500,
  "max_steps": 1740,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1045168128000000.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}