trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 6,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "completion_length": 219.8839340209961,
      "epoch": 1.0,
      "grad_norm": 0.7357717259237131,
      "kl": 0.0,
      "learning_rate": 4.5225424859373684e-07,
      "loss": 0.0,
      "reward": 1.5958333015441895,
      "reward_std": 0.535225123167038,
      "rewards/answer_reward_func": 0.7104167342185974,
      "rewards/format_reward_func": 0.8854166865348816,
      "step": 2
    },
    {
      "completion_length": 264.921142578125,
      "epoch": 2.0,
      "grad_norm": 0.9424382874363444,
      "kl": 0.00024127960205078125,
      "learning_rate": 1.7274575140626315e-07,
      "loss": 0.0,
      "reward": 1.6104167699813843,
      "reward_std": 0.5595069080591202,
      "rewards/answer_reward_func": 0.7145833671092987,
      "rewards/format_reward_func": 0.8958333432674408,
      "step": 4
    },
    {
      "completion_length": 248.2396011352539,
      "epoch": 3.0,
      "grad_norm": 1.041831469599061,
      "kl": 0.00026798248291015625,
      "learning_rate": 0.0,
      "loss": 0.0,
      "reward": 1.558333396911621,
      "reward_std": 0.6721908450126648,
      "rewards/answer_reward_func": 0.7041667401790619,
      "rewards/format_reward_func": 0.8541666865348816,
      "step": 6
    },
    {
      "epoch": 3.0,
      "step": 6,
      "total_flos": 0.0,
      "train_loss": 1.7912438347404228e-07,
      "train_runtime": 441.0039,
      "train_samples_per_second": 0.048,
      "train_steps_per_second": 0.014
    }
  ],
  "logging_steps": 2,
  "max_steps": 6,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 32,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}