llama.cpp

feat: Extra debugging support for model conversion

#15877

Merged