prr
prr copied to clipboard
[FEAT] Huggingface models
implement a universal way to use inference endpoints from huggingface potential issue - different prompts needed for different models (instruct vs chat, etc)