llama.cpp

5628ec71 - gguf : read / write sample models