docs: add docstrings in train script

scripts/train.py

@@ -183,6 +183,12 @@ def evaluate(
 
 
 def log_eval_metrics(episode: int, metrics: dict[str, Any]) -> None:
+    """
+    Logs the evaluation metrics for an episode.
+
+    :param episode: Episode number.
+    :param metrics: Performance measures.
+    """
     logger.info(
         "episode %d: winning rate = %.2f, mean score = %.2f, max tile = %d",
         episode,
@@ -193,6 +199,12 @@ def log_eval_metrics(episode: int, metrics: dict[str, Any]) -> None:
 
 
 def save_best_policy(out_dir: str, policy: NTupleNetworkBasePolicy) -> None:
+    """
+    Saves the best policy.
+
+    :param out_dir: Path to output directory.
+    :param policy: Policy to save.
+    """
     best_model_path = os.path.join(out_dir, "best_n_tuple_network_policy.zip")
     logger.info("new best model saved to %s", best_model_path)
     policy.save(path=best_model_path)
@@ -203,6 +215,13 @@ def save_checkpoint(
     out_dir: str,
     policy: NTupleNetworkBasePolicy,
 ) -> None:
+    """
+    Saves a checkpoint.
+
+    :param episode: Episode number.
+    :param out_dir: Output directory.
+    :param policy: Policy to save.
+    """
     checkpoint_path = os.path.join(out_dir, f"checkpoint_episode_{episode}.zip")
     logger.info("checkpoint saved to %s", checkpoint_path)
     policy.save(path=checkpoint_path)