Skip to content

ggml-cuda: Q8_0 weights take dequant-to-BF16 + cuBLAS by default; k-quants stay on MMQ - #39

Open
danielhanchen wants to merge 1 commit into
sync-upstream-a77d362from
cuda-q8-bf16-cublas
Open

ggml-cuda: Q8_0 weights take dequant-to-BF16 + cuBLAS by default; k-quants stay on MMQ#39
danielhanchen wants to merge 1 commit into
sync-upstream-a77d362from
cuda-q8-bf16-cublas

Commits

  1. Commits on Oct 23, 2026