Comment by Jack5500 Comment by Jack5500 2 months ago 1 reply Copy Link View on Hacker News Isn‘t clip superseeded by multimodal llms?
Copy Link Eisenstein 2 months ago Collapse Comment - In this program CLIP is being used to create embeddings. A multimodal LLM does something very similar. In this case the language model is not needed because the embeddings are being used to search directly. Reply View | 0 replies
In this program CLIP is being used to create embeddings. A multimodal LLM does something very similar. In this case the language model is not needed because the embeddings are being used to search directly.