Update llm.py

llm.py

@@ -82,7 +82,7 @@ def build_messages(
 @torch.inference_mode()
 def stream_response(
         messages: list[dict],
-        enable_thinking: bool=True,
+        enable_thinking: bool=False,
         max_new_tokens: int=1_024
     ) -> tuple[str, str]:
     """Stream the model's response to the chat interface.