Very neat! I've been stalking this project ever since I saw it get mentioned on the Candle repo, I'm curious to see where this goes next.

Any plans for multimodal models like llava?