trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 49.96321070234114,
  "eval_steps": 500,
  "global_step": 900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 27.74916387959866,
      "grad_norm": 20.473485946655273,
      "learning_rate": 1.4035087719298246e-05,
      "loss": 0.2915,
      "step": 500
    },
    {
      "epoch": 27.74916387959866,
      "eval_loss": 0.39518147706985474,
      "eval_runtime": 11.6206,
      "eval_samples_per_second": 17.211,
      "eval_steps_per_second": 4.303,
      "step": 500
    }
  ],
  "logging_steps": 500,
  "max_steps": 900,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "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": 7.230035243040768e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}