[BugFix] Fix prefer_model_sampler token history in async scheduling#3681

gcanlin merged 6 commits into

vllm-project:mainvllm-project/vllm-omni:mainfrom

zengchuang-hw:fix-async-sampler-token-historyzengchuang-hw/vllm-omni:fix-async-sampler-token-historyCopy head branch name to clipboard