skip cuda operations when running qwen 3.5 moe on other backend#19095

Gasoonjia merged 1 commit into

mainpytorch/executorch:mainfrom

moe-metal-fixpytorch/executorch:moe-metal-fixCopy head branch name to clipboard