trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.96,
  "eval_steps": 500,
  "global_step": 9,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.6469,
      "step": 1
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0001,
      "loss": 0.408,
      "step": 5
    },
    {
      "epoch": 0.96,
      "eval_loss": 0.24174152314662933,
      "eval_runtime": 6.0031,
      "eval_samples_per_second": 16.658,
      "eval_steps_per_second": 2.166,
      "step": 9
    },
    {
      "epoch": 0.96,
      "step": 9,
      "total_flos": 9507448946688.0,
      "train_loss": 0.3598402738571167,
      "train_runtime": 147.976,
      "train_samples_per_second": 2.027,
      "train_steps_per_second": 0.061
    }
  ],
  "logging_steps": 5,
  "max_steps": 9,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 9507448946688.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}