UniIR
First page
Paper summary
UniIR is a unified instruction-guided multimodal retriever that handles eight retrieval tasks across modalities with a single model.
Ask this paper
01
Instruction-guided: A single retriever conditioned on natural-language instructions determines which retrieval task to perform, rather than one retriever per task.
02
Eight tasks: Handles image-to-text, text-to-image, composed-image retrieval, video retrieval, and other multimodal variants under one umbrella.
03
Zero-shot generalization: Generalizes to unseen retrieval tasks not explicitly trained on, approaching a truly general multimodal retrieval model.
04
M-BEIR benchmark: Ships with a new multimodal retrieval benchmark (M-BEIR) designed to standardize evaluation across tasks and modalities.