[Opt] Optimize deepstack buffer handling for multimodal Qwen3 models - #40145

[Opt] Optimize deepstack buffer handling for multimodal Qwen3 models#40145

Isotr0py merged 1 commit into

vllm-project:mainvllm-project/vllm:mainfrom

labAxiaoming:qwen3vl_optlabAxiaoming/vllm:qwen3vl_optCopy head branch name to clipboard