fix gen kwargs

src/imitater/model/chat_model.py

@@ -125,7 +125,7 @@ async def _generate(self, messages: List[Dict[str, str]], request_id: str, **gen
         sampling_params = SamplingParams(
             temperature=gen_kwargs.pop("temperature", self._generation_config.temperature),
             top_p=gen_kwargs.pop("top_p", self._generation_config.top_p),
-            max_tokens=gen_kwargs.pop("max_tokens", self._generation_config.max_new_tokens),
+            max_tokens=gen_kwargs.pop("max_tokens") or self._generation_config.max_new_tokens,
             stop=gen_kwargs.pop("stop", None),
             stop_token_ids=self._generation_config.eos_token_id + gen_kwargs.pop("stop_token_ids", []),
         )