mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-06-30 01:27:42 +02:00
0eb874d374
This trims down some of the shader variant explosion and reduces binary size.