[BugFix][Multi-GPU] Allocate and launch MoE outputs on the input device - #30

[BugFix][Multi-GPU] Allocate and launch MoE outputs on the input device#30

PerryLink wants to merge 2 commits into

deepseek-ai:maindeepseek-ai/TileKernels:mainfrom

PerryLink:fix/moe-output-devicePerryLink/TileKernels:fix/moe-output-deviceCopy head branch name to clipboard