fix(w8a16): correct MMA A-fragment register mapping + MoE improvements#125

Merged

m96-chan merged 21 commits intomainfrom

feature/v0.2.18

Dec 30, 2025

Commits on Dec 28, 2025

perf(fp8): add cached scale buffer variants (v5-v8) for FP8xFP8 GEMM

m96-chan
and
claude
committed
feat(gemv): add accurate FP8/FP8 GEMV kernel (#123)

m96-chan
and
claude
committed
perf(gemv): optimize accurate FP8 kernel to match fast version speed

m96-chan
and
claude
committed
Merge main: fix GEMV error rates and add rel-error-check skill
m96-chan
committed
fix(test): correct W8A16 scale format and update error rate

m96-chan
and
claude
committed
perf(gemv): add optimized BF16 GEMV kernel with B[N,K] layout

m96-chan
and
claude
committed
feat(llm): use optimized BF16 GEMV in LinearBF16 layer

m96-chan
and
claude
committed

Commits on Dec 29, 2025