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] Apply same sampling parameters for both `n=1` and `n>1` (#26005)
Signed-off-by: Kenichi Maehashi <maehashi@preferred.jp>
Kenichi Maehashi committed
Oct 1, 2025 at 23:37 UTC
3b7c20a6b52f8b9510669131c4ce4d8040099f41
1 file changed
+1
-1
vllm/v1/engine/async_llm.py
+1
-1
@@ -290,7 +290,7 @@ class AsyncLLM(EngineClient):
290
return queue
291
292
# Fan out child requests (for n>1).
293
- parent_request = ParentRequest(request_id, params)
293
+ parent_request = ParentRequest(request_id, request.sampling_params)
294
for idx in range(params.n):
295
request_id, params = parent_request.get_child_info(idx)
296
child_request = request if idx == params.n - 1 else copy(request)