support do_preprocess=False for chat.completions (#3645)

Author: irexyc

lmdeploy/serve/openai/api_server.py

@@ -419,6 +419,8 @@ async def chat_completions_v1(raw_request: Request = None):
             ]
         else:
             tools = [item.function.model_dump() for item in request.tools]
+    # text completion for string input
+    do_preprocess = False if isinstance(request.messages, str) else request.do_preprocess
     result_generator = VariableInterface.async_engine.generate(
         request.messages,
         request.session_id,
@@ -427,7 +429,7 @@ async def chat_completions_v1(raw_request: Request = None):
         stream_response=True,  # always use stream to enable batching
         sequence_start=True,
         sequence_end=True,
-        do_preprocess=not isinstance(request.messages, str),  # text completion for string input
+        do_preprocess=do_preprocess,
         adapter_name=adapter_name,
         enable_thinking=request.enable_thinking,
     )

lmdeploy/serve/openai/protocol.py

@@ -127,6 +127,7 @@ class ChatCompletionRequest(BaseModel):
     user: Optional[str] = None
     response_format: Optional[ResponseFormat] = Field(default=None, examples=[None])  # noqa
     # additional argument of lmdeploy
+    do_preprocess: Optional[bool] = True
     repetition_penalty: Optional[float] = 1.0
     session_id: Optional[int] = -1
     ignore_eos: Optional[bool] = False