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, 00:30 UTC
[BugFix] Fix broken import in initialize_ray_cluster() (#27838)
Signed-off-by: Nick Hill <nhill@redhat.com>
Nick Hill committed
Oct 30, 2025 at 16:26 UTC
c9791f18138d1a11bfe68550b10673b493ec9330
1 file changed
+1
-1
vllm/v1/executor/ray_utils.py
+1
-1
@@ -322,7 +322,7 @@ def initialize_ray_cluster(
322
323
# Prevalidate GPU requirements before Ray processing
324
if current_platform.is_cuda() and parallel_config.world_size > 1:
325
- from vllm.utils import cuda_device_count_stateless
325
+ from vllm.utils.torch_utils import cuda_device_count_stateless
326
327
available_gpus = cuda_device_count_stateless()
328
if parallel_config.world_size > available_gpus: