Anyone know how to convert a torch model (.pt) to GGUF for use with ollama?

[arpieb]

Exploring this interesting embedding model:

https://github.com/exlab-research/Banyan

but am unable to load it into ollama for real-world testing as it appears every “torch to GGUF” article out there starts off with “download a model from HF, use llama.cpp’s convert scripts” - which is not what I need.

Seems like every other article is re fine-tuning some existing model staying within the HF ecosystem. Kinda stunts efforts to use novel non-HF models in common LLM tooling...

Thanks in advance!