Support different formats Huggingface phi 3

[birdup000]

When trying to load the phi 3 models on HuggingFace it fails load since the format is different in the repo.

[Josh-XT]

Try changing DEFAULT_MODEL in your .env to this:

DEFAULT_MODEL=mradermacher/Phi-3-mini-128k-instruct-LLaMAfied-GGUF

[birdup000]

Try changing DEFAULT_MODEL in your .env to this:

DEFAULT_MODEL=mradermacher/Phi-3-mini-128k-instruct-LLaMAfied-GGUF

Yes that one loads and works but the other ones do not work since its the format is different but its still GGUF

[Josh-XT]

Those following the standard TheBloke set on Hugging Face will work, but the people who have not followed those standards may have issues. I am planning to try to expand on that to try to find the file in whichever repo but I haven't yet. Currently it is limited to those who are in-standard with TheBloke's repos. I recommend looking at quants from https://huggingface.co/MaziyarPanahi