A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Bugfix] UnboundLocalError when GptOss reasoning specified (#23054)
Signed-off-by: Federico <65908512+coval3nte@users.noreply.github.com>
Federico committed
Aug 27, 2025 at 02:29 UTC
585e0bde36abdb2ab2967fd42005cbe62459020e
1 file changed
+2
-1
vllm/entrypoints/openai/serving_chat.py
+2
-1
@@ -887,7 +887,8 @@ class OpenAIServingChat(OpenAIServing):
887
delta_message = DeltaMessage(content=delta_text)
888
889
# update the previous values for the next iteration
890
- if tool_choice_auto or self.reasoning_parser:
890
+ if ((tool_choice_auto or self.reasoning_parser)
891
+ and not self.use_harmony):
892
assert previous_texts is not None
893
assert all_previous_token_ids is not None
894
previous_texts[i] = current_text