trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1654,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.6,
      "learning_rate": 2.0931076178960096e-05,
      "loss": 2.1458,
      "step": 500
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.1862152357920194e-05,
      "loss": 1.7773,
      "step": 1000
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.7932285368802904e-06,
      "loss": 1.4845,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "step": 1654,
      "total_flos": 3887703795096576.0,
      "train_loss": 1.7669375141701935,
      "train_runtime": 10839.0571,
      "train_samples_per_second": 1.83,
      "train_steps_per_second": 0.153
    }
  ],
  "logging_steps": 500,
  "max_steps": 1654,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 3887703795096576.0,
  "trial_name": null,
  "trial_params": null
}