llama.cpp will happily download models from hugging face, btw.

Huggingface now owns llama.cpp, btw.

https://huggingface.co/blog/ggml-joins-hf