trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 1525,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.639344262295082,
      "grad_norm": 6.152276992797852,
      "learning_rate": 1.3455737704918036e-05,
      "loss": 0.2563,
      "step": 500
    },
    {
      "epoch": 3.278688524590164,
      "grad_norm": 1.6254922151565552,
      "learning_rate": 6.8983606557377056e-06,
      "loss": 0.081,
      "step": 1000
    },
    {
      "epoch": 4.918032786885246,
      "grad_norm": 0.03872436285018921,
      "learning_rate": 3.409836065573771e-07,
      "loss": 0.0242,
      "step": 1500
    }
  ],
  "logging_steps": 500,
  "max_steps": 1525,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4774349102423040.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}