checkpoint-48/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 48,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.5385825634002686,
      "learning_rate": 0.00017142857142857143,
      "loss": 1.637,
      "step": 24
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.1714123785495758,
      "eval_runtime": 394.108,
      "eval_samples_per_second": 1.462,
      "eval_steps_per_second": 1.462,
      "step": 24
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.946814775466919,
      "learning_rate": 0.00014285714285714287,
      "loss": 0.1624,
      "step": 48
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.14965203404426575,
      "eval_runtime": 392.7243,
      "eval_samples_per_second": 1.467,
      "eval_steps_per_second": 1.467,
      "step": 48
    }
  ],
  "logging_steps": 500,
  "max_steps": 168,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 500,
  "total_flos": 1.96410372784128e+18,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}