llama.cpp

2973a65e - gguf-py : allow converting multi-tensor models from read-only locations (#18100)