trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5,
  "eval_steps": 500,
  "global_step": 388,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.12886597938144329,
      "grad_norm": 0.5468710660934448,
      "learning_rate": 1.484536082474227e-05,
      "loss": 2.3122,
      "step": 100
    },
    {
      "epoch": 0.25773195876288657,
      "grad_norm": 0.37763309478759766,
      "learning_rate": 9.690721649484536e-06,
      "loss": 1.9019,
      "step": 200
    },
    {
      "epoch": 0.3865979381443299,
      "grad_norm": 0.471895694732666,
      "learning_rate": 4.536082474226804e-06,
      "loss": 1.6506,
      "step": 300
    }
  ],
  "logging_steps": 100,
  "max_steps": 388,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 188,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.98491633614848e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}