trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 2738,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.37,
      "learning_rate": 8.692476260043827e-06,
      "loss": 1.2919,
      "step": 500
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0,
      "loss": 1.4435,
      "step": 1000
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0,
      "loss": 1.4318,
      "step": 1500
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0,
      "loss": 1.4235,
      "step": 2000
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0,
      "loss": 1.4192,
      "step": 2500
    },
    {
      "epoch": 2.0,
      "step": 2738,
      "total_flos": 1.7783709700718592e+16,
      "train_loss": 1.4037482273500275,
      "train_runtime": 3969.1352,
      "train_samples_per_second": 44.14,
      "train_steps_per_second": 0.69
    }
  ],
  "logging_steps": 500,
  "max_steps": 2738,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 1.7783709700718592e+16,
  "trial_name": null,
  "trial_params": null
}