update code

Author: kevincheng2

fastdeploy/input/ernie_vl_processor.py

@@ -165,6 +165,7 @@ def _check_mm_limits(self, item):
     def process_request_dict(self, request, max_model_len=None):
         """process the input data"""
 
+        request['enable_thinking'] = self.get_enable_thinking(request.get("enable_thinking"))
         if not request.get("eos_token_ids"):
             request["eos_token_ids"] = self.eos_token_ids
 
@@ -250,7 +251,7 @@ def process_response_dict(self, response_dict, stream, **kwargs):
         Returns:
             Dict: response contain text fields
         """
-        enable_thinking = self.get_enable_thinking(kwargs.get("enable_thinking"))
+        enable_thinking = self.get_enable_thinking(kwargs.pop("enable_thinking", None))
         if stream:
             return self.process_response_dict_streaming(response_dict, enable_thinking=enable_thinking, **kwargs)
         else:

fastdeploy/input/text_processor.py

@@ -445,9 +445,7 @@ def process_response_dict(self, response_dict, **kwargs):
         Returns:
             Dict: response contain text fields
         """
-        enable_thinking = kwargs.pop("enable_thinking", True)
-        if enable_thinking is None:
-            enable_thinking = True
+        enable_thinking = self.get_enable_thinking(kwargs.pop("enable_thinking", None))
         stream = kwargs.get("stream", True)
         if stream:
             return self.process_response_dict_streaming(