Skip to content

feat(vllm): support refit-safe MXFP8 linear backends - #3638

Open
seonjinn wants to merge 51 commits into
NVIDIA-NeMo:mainfrom
seonjinn:sna/mxfp8-linear-backend-refit
Open

seonjinn wants to merge 51 commits into
NVIDIA-NeMo:mainfrom
seonjinn:sna/mxfp8-linear-backend-refit