[Bugfix] Ensure download_weights_from_hf(..) inside loader is using t…

…he revision parameter (vllm-project#4217)

vllm/model_executor/model_loader/loader.py

@@ -172,7 +172,7 @@ def _prepare_weights(self, model_name_or_path: str,
         if not is_local:
             hf_folder = download_weights_from_hf(model_name_or_path,
                                                  self.load_config.download_dir,
-                                                 allow_patterns)
+                                                 allow_patterns, revision)
         else:
             hf_folder = model_name_or_path