cturan/llama.cpp @ 06658ad7c37f440502de2b9486ce43c47b4ec710

镜像来自 https://github.com/cturan/llama.cpp

Georgi Gerganov 06658ad7c3 metal : separate scale and mask from QKT in FA kernel (#9189)		1 年之前
..
cmake	f3f65429c4 llama : reorganize source code + improve CMake (#8006)	1 年之前
include	e11bd856d5 CPU/CUDA: Gemma 2 FlashAttention support (#8542)	1 年之前
src	06658ad7c3 metal : separate scale and mask from QKT in FA kernel (#9189)	1 年之前
.gitignore	17eb6aa8a9 vulkan : cmake integration (#8119)	1 年之前
CMakeLists.txt	5fd89a70ea Vulkan Optimizations and Fixes (#8959)	1 年之前