This repository contains a legacy .ggml file whose first tensor header uses raw dtype 99, causing the public llama.cpp converter to abort with:
.ggml
99
llama.cpp
ValueError: 99 is not a valid GGMLQuantizationType
-