[Fixbug][Perf] Qwen3-omni: code predictor with re-prefill + SDPA and eliminate decode hot-path CPU round-trips#2012

hsliuustc0106 merged 4 commits into

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

LJH-LBJ:refactor/qwen3-omni-code-predictor-v2LJH-LBJ/vllm-omni:refactor/qwen3-omni-code-predictor-v2Copy head branch name to clipboard