fix

Author: luukunn

fastdeploy/engine/request.py

@@ -72,7 +72,6 @@ def __init__(
         structural_tag: Optional[Any] = None,
         guided_json_object: Optional[bool] = None,
         enable_thinking: Optional[bool] = False,
-        model_status: Optional[str] = None,
         trace_carrier: dict = dict(),
         dp_rank: Optional[int] = None,
         chat_template: Optional[str] = None,

fastdeploy/input/ernie4_5_processor.py

@@ -156,11 +156,9 @@ def process_request(self, request, max_model_len=None, **kwargs):
         if self.reasoning_parser and self.reasoning_parser.__class__.__name__ == "ErnieX1ReasoningParser":
             request.enable_thinking = True
         if self.reasoning_parser:
-            self.model_status_dict[request.request_id] = self.reasoning_parser.get_model_status(
-                request.prompt_token_ids
-            )
-            if self.model_status_dict[request.request_id] == "think_start":
-                request.enable_thinking = True
+            model_status = self.reasoning_parser.get_model_status(request.prompt_token_ids)
+            self.model_status_dict[request.request_id] = model_status
+            request.enable_thinking = model_status == "think_start"
 
         data_processor_logger.info(f"Processed request: {request}")
         return request

fastdeploy/input/text_processor.py

@@ -267,11 +267,9 @@ def process_request(self, request, max_model_len=None, **kwargs):
         if request.get("top_p") < _SAMPLING_EPS:
             request.set("top_p", _SAMPLING_EPS)
         if self.reasoning_parser:
-            self.model_status_dict[request.request_id] = self.reasoning_parser.get_model_status(
-                request.prompt_token_ids
-            )
-            if self.model_status_dict[request.request_id] == "think_start":
-                request.enable_thinking = True
+            model_status = self.reasoning_parser.get_model_status(request.prompt_token_ids)
+            self.model_status_dict[request.request_id] = model_status
+            request.enable_thinking = model_status == "think_start"
 
         data_processor_logger.info(f"Processed request: {request}")
         return request