4 ms·
This is very similar to how LLMs are taught to understand images in llava style models (the image embeddings are encoded into the existing language token stream
by tempusalaria 2y ago
This is very similar to how LLMs are taught to understand images in llava style models (the image embeddings are encoded into the existing language token stream)