Open yoavf opened 1 day ago
Some of the models accept images (and possibly other media) as input for text generation.
How should this be implemented? How would drivers declare support for this?
Its definitely something I want to support. I've got an item on the roadmap. Initially, I'll be looking to the Vercel SDK (ref) and probably looking to implement something similar.
Some of the models accept images (and possibly other media) as input for text generation.
How should this be implemented? How would drivers declare support for this?