summaryrefslogtreecommitdiff
path: root/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu
AgeCommit message (Expand)Author
2024-06-05CUDA: refactor mmq, dmmv, mmvq (#7716)Johannes Gäßler
2024-06-01CUDA: quantized KV support for FA vec (#7527)Johannes Gäßler