[Model] Explicit interface for vLLM models and support OOT embedding models (#9108)

vllm/utils.py

@@ -1277,6 +1277,15 @@ async def _run_task_with_lock(task: Callable, lock: asyncio.Lock, *args,
         return await task(*args, **kwargs)
 
 
+def supports_kw(callable: Callable[..., object], kw_name: str) -> bool:
+    params = inspect.signature(callable).parameters
+    if kw_name in params:
+        return True
+
+    return any(param.kind == inspect.Parameter.VAR_KEYWORD
+               for param in params.values())
+
+
 def get_allowed_kwarg_only_overrides(
     callable: Callable[..., object],
     overrides: Optional[Dict[str, Any]],