checkpoint-129/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 129,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.65,
      "grad_norm": 146076.03125,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 1.2315,
      "step": 100
    }
  ],
  "logging_steps": 100,
  "max_steps": 210,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1.7560604267925996e+19,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}