[ryanackley]

The technology that allows an LLM to "see" images and video is completely different though. It's not what is being trained on common crawl.

[brrrrrm]

not really. embeddings are embeddings. check out llava