Skip to content

llama : add an attention split separate from the tensor split - #7

Open
Piggidragon wants to merge 3 commits into
mgpu/host-kv-split-by-headfrom
mgpu/attn-split
Open

llama : add an attention split separate from the tensor split#7
Piggidragon wants to merge 3 commits into
mgpu/host-kv-split-by-headfrom
mgpu/attn-split

Commits

  1. Commits on Sep 17, 2026