[Bugfix] Fix max_num_batched_token not captured in cuda graph #40734

vllm-bot merged 7 commits into

vllm-project:mainvllm-project/vllm:mainfrom

wzhao18:wzhao/fix-cudagraph-max-sizewzhao18/vllm:wzhao/fix-cudagraph-max-sizeCopy head branch name to clipboard