llama.cpp

bec4772f - Add Q2_0 quantization: type definition and CPU backend (#24448)