feat: bf16 GPU residency — device-resident weight upload + axis reductions + Rsqrt (ADR-075 L4)#155

Merged

dndungu merged 3 commits into

mainfrom

feat/bf16-upload-reductions

Jun 17, 2026

Commits on Jun 16, 2026

feat(compute): native bf16 GPU Rsqrt (close non-f32 CPU fallback)
dndungu
committed

Commits on Jun 17, 2026

feat(compute): native bf16 GPU axis reductions (Sum/ReduceSum/ReduceMean)
dndungu
committed
feat(compute): generic bf16 bulk weight upload (device-residency)
dndungu
committed