EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

accomendate reasoning_effort set in chat_template_kwargs (#9458)

Browse Source
This commit is contained in:
gongwei-130
2025-08-21 13:22:03 -07:00
committed by GitHub
parent 9c8e4f69c3
commit 9ba7253094
1 changed files with 8 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
python/sglang/srt/entrypoints/openai/serving_chat.py
View File

@@ -87,6 +87,14 @@ class OpenAIServingChat(OpenAIServingBase):
self, self,
request: ChatCompletionRequest, request: ChatCompletionRequest,
) -> tuple[GenerateReqInput, ChatCompletionRequest]: ) -> tuple[GenerateReqInput, ChatCompletionRequest]:
reasoning_effort = (
request.chat_template_kwargs.pop("reasoning_effort", None)
if request.chat_template_kwargs
else None
)
if reasoning_effort is not None:
request.reasoning_effort = reasoning_effort
"""Convert OpenAI chat completion request to internal format""" """Convert OpenAI chat completion request to internal format"""
is_multimodal = self.tokenizer_manager.model_config.is_multimodal is_multimodal = self.tokenizer_manager.model_config.is_multimodal