perf: optimize grad_weight accumulation with addmm#1239

Merged

vaibhavjindal merged 5 commits into

linkedin:mainfrom

maskyuanzh:fix-fused-linear-ce-addmm

Jul 1, 2026

Commits on Jun 24, 2026

Optimize grad_weight accumulation with addmm
maskyuanzh
committed
fix fused linear CE addmm fallback dtype handling
maskyuanzh
committed
Remove redundant input chunk cast
maskyuanzh
committed

Commits on Jun 30, 2026

Merge branch 'main' into fix-fused-linear-ce-addmm
vaibhavjindal
authored

Commits on Jul 1, 2026

Rename addmm out_dtype support constant
maskyuanzh
committed