1 paper
Tianyu Yang, Shir Simon, Zhenzhen Li +2
Multimodal retrieval-augmented generation (mRAG) aims to answer image-text queries with external knowledge, but most existing systems still retrieve directly from raw multimodal in…