[Perf] Optimize the Qwen2.5-Omni Model thinker-to-talker-proj with nn.Linear#825

hsliuustc0106 merged 1 commit intovllm-project:mainvllm-project/vllm-omni:mainfrom

kechengliu97:Qwen2.5-Optimizekechengliu97/vllm-omni:Qwen2.5-OptimizeCopy head branch name to clipboard