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] Fix task field initialization when PYTHONOPTIMIZE is enabled (#23718)
Signed-off-by: cndoit18 <cndoit18@outlook.com>
cndoit18 committed
Aug 27, 2025 at 20:42 UTC
8c13820f0b203976eab8e821c102234a73f338cd
1 file changed
+3
-2
vllm/worker/pooling_model_runner.py
+3
-2
@@ -199,8 +199,9 @@ class PoolingModelRunner(
199
200
pooling_params = seq_group_metadata.pooling_params
201
assert pooling_params is not None
202
- assert (task := pooling_params.task) is not None, (
203
- "You did not set `task` in the API")
202
+
203
+ task = pooling_params.task
204
+ assert task is not None, "You did not set `task` in the API"
205
206
model = cast(VllmModelForPooling, self.model)
207
to_update = model.pooler.get_pooling_updates(task)