3 ms·
> they failed miserably at finding bounding boxes, usually just making up random coordinates. This makes sense to me. These LLMs likely have no statistics abou
by bob1029 2y ago
> they failed miserably at finding bounding boxes, usually just making up random coordinates.
This makes sense to me. These LLMs likely have no statistics about the spatial relationships of tokens in a 2D raster space.
- nostrebored 2y agoThe spatial awareness is what grounding models try to achieve, e.g. UGround [1] [1] https://huggingface.co/osunlp/UGround-V1-7B?language=python https://huggingface.co/osunlp/UGround-V1-7B?language=python