Skip to content

llama : add an attention split separate from the tensor split - #69

Closed
Piggidragon wants to merge 14 commits into
GenerelSchwerz:llama/devfrom
Piggidragon:mgpu/attn-split
Closed

Piggidragon wants to merge 14 commits into
GenerelSchwerz:llama/devfrom
Piggidragon:mgpu/attn-split