BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models.

Junnan Li, Dongxu Li, Silvio Savarese, Steven C. H. Hoi

Browse the full ICML paper archive.