last-checkpoint/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9995002498750625,
  "eval_steps": 250,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "learning_rate": 0.028000000000000004,
      "loss": 136488665.088,
      "step": 250
    },
    {
      "epoch": 0.5,
      "eval_runtime": 1439.7269,
      "eval_samples_per_second": 6.949,
      "eval_steps_per_second": 0.29,
      "step": 250
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.028000000000000004,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_runtime": 1440.1676,
      "eval_samples_per_second": 6.947,
      "eval_steps_per_second": 0.29,
      "step": 500
    }
  ],
  "logging_steps": 250,
  "max_steps": 1000,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 4.1525270544384e+17,
  "trial_name": null,
  "trial_params": null
}