checkpoint-1500/trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.976400136947632,
      "learning_rate": 3.336666666666667e-05,
      "loss": 1.4885,
      "step": 500
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.292872428894043,
      "learning_rate": 1.6700000000000003e-05,
      "loss": 1.1504,
      "step": 1000
    },
    {
      "epoch": 3.0,
      "grad_norm": 7.022700309753418,
      "learning_rate": 3.3333333333333334e-08,
      "loss": 1.0006,
      "step": 1500
    }
  ],
  "logging_steps": 500,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 1020745916282304.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}