llm2vec
llm2vec copied to clipboard
Interesting idea
Could you do something like this with Qwen 2.5 VL (or internVL2.5) to make a multimodal vector embedding model? I'm dumb so couldn't do it myself, but I'm sure smart people like you all could!
Nvm somebody already thought about it...
https://github.com/TIGER-AI-Lab/VLM2Vec