Repository navigation

[CPU] Fix accumulation precision in affine quantized matmul - #4803

Merged

zcbenz merged 2 commits into

ml-explore:mainfrom

erwinzhang7:fix-cpu-qmm-bf16-accum

Nov 5, 2026

Merged

[CPU] Fix accumulation precision in affine quantized matmul#4803
zcbenz merged 2 commits into
ml-explore:mainfrom
erwinzhang7:fix-cpu-qmm-bf16-accum

Commits on Nov 3, 2026
- [CPU] Fix accumulation precision in affine quantized matmul
  
  erwinzhang7
  authored and
  zcbenz
  committed
  1e15461
- [CPU] Use MLX allocator for quantized matmul accumulator
  
  erwinzhang7
  authored and
  zcbenz
  committed
  b434bdb