llama.cpp
graph : make FA compatible with MLA + add initial Metal kernels
#12953

Merged

graph : make FA compatible with MLA + add initial Metal kernels #12953

ggerganov merged 5 commits into master from gg/mla

github-actions added testing

github-actions added Nvidia GPU

github-actions added Vulkan

github-actions added ggml

github-actions added Apple Metal

graph : make mla compatible with FA

e3308567

metal : add exp FA kernels for DeepSeek models

9b64dccf

llama : minor naming updates

9cc85ddb

ggml : disable FA for DS head sizes

43c762bb

tests : add FA tests for MLA shapes

facdf870

ggerganov force pushed from 2cbc16d1 to facdf870 1 year ago

ggerganov merged 2f74c354 into master 1 year ago

ggerganov deleted the gg/mla branch 1 year ago

Reviewers

No reviews

Assignees

No one assigned

Labels

testing Nvidia GPU Vulkan ggml Apple Metal

Milestone

No milestone