[#10780][feat] AutoDeploy: Support per-expert scales in FP8 and NVFP4 MoE#11322

galagam merged 8 commits into

NVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

nv-auto-deploy:gagam/handle-non-identical-moe-scales-v2nv-auto-deploy/TensorRT-LLM:gagam/handle-non-identical-moe-scales-v2Copy head branch name to clipboard