8000 Release b5434 · ggml-org/llama.cpp · GitHub
[go: up one dir, main page]

Skip to content
EE69

b5434

Compare
Choose a tag to compare
@github-actions github-actions released this 20 May 14:11
b69f164
CUDA: skip fully masked-out KV in FA vec kernel (#13584)

* CUDA: skip fully masked-out KV in FA vec kernel
0