Add comment

Author: Yard1

vllm/engine/arg_utils.py

@@ -41,6 +41,9 @@ class EngineArgs:
     seed: int = 0
     max_model_len: Optional[int] = None
     worker_use_ray: bool = False
+    # Note: Specifying a custom executor backend by passing a class
+    # is intended for expert use only. The API may change without
+    # notice.
     distributed_executor_backend: Optional[Union[str,
                                                  Type[ExecutorBase]]] = None
     pipeline_parallel_size: int = 1
@@ -68,6 +71,9 @@ class EngineArgs:
     max_seq_len_to_capture: int = 8192
     disable_custom_all_reduce: bool = False
     tokenizer_pool_size: int = 0
+    # Note: Specifying a tokenizer pool by passing a class
+    # is intended for expert use only. The API may change without
+    # notice.
     tokenizer_pool_type: Union[str, Type["BaseTokenizerGroup"]] = "ray"
     tokenizer_pool_extra_config: Optional[dict] = None
     enable_lora: bool = False