Fetching the paper…

LION : Empowering Multimodal Large Language Model with Dual-Level Visual Knowledge · Around