llama.cpp

4f0547e4 - gguf : add support for 64-bit (no backwards comp yet)