trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4468864468864469,
  "eval_steps": 500,
  "global_step": 122,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.9996684675143132e-05,
      "loss": 1.1215,
      "step": 1
    },
    {
      "epoch": 0.45,
      "eval_loss": 0.9631918668746948,
      "eval_runtime": 88.1248,
      "eval_samples_per_second": 175.104,
      "eval_steps_per_second": 0.352,
      "step": 122
    },
    {
      "epoch": 0.45,
      "step": 122,
      "total_flos": 204354543943680.0,
      "train_loss": 1.0108161842236754,
      "train_runtime": 1465.8388,
      "train_samples_per_second": 42.613,
      "train_steps_per_second": 0.083
    }
  ],
  "logging_steps": 500,
  "max_steps": 122,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 204354543943680.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}