Hacker News Reader

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.

not really. embeddings are embeddings. check out llava