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 12, 13:30 UTC
fix assertion for single world use case (uni) (#29429)
Signed-off-by: Lu Fang <fanglu@fb.com> Co-authored-by: Lucia (Lu) Fang <fanglu@meta.com>
Lucia Fang committed
Nov 25, 2025 at 16:14 UTC
d8819c88eb633f1435ef389f87c23a9fd6010917
1 file changed
+3
-2
vllm/config/parallel.py
+3
-2
@@ -593,9 +593,10 @@ class ParallelConfig:
593
"max_parallel_loading_workers is currently "
594
"not supported and will be ignored."
595
)
596
- if self.distributed_executor_backend != "mp" and self.nnodes > 1:
596
+ if self.distributed_executor_backend not in ("mp", "uni") and self.nnodes > 1:
597
raise ValueError(
598
- "nnodes > 1 can only be set when distributed exectuor backend is mp."
598
+ "nnodes > 1 can only be set when distributed executor "
599
+ "backend is mp or uni."
600
)
601
602
@property