Mimic In-Context Learning for Multimodal Tasks

Yuchu Jiang, Jiale Fu, Chenduo Hao +4

Recently, In-context Learning (ICL) has become a significant inference paradigm in Large Multimodal Models (LMMs), utilizing a few in-context demonstrations (ICDs) to prompt LMMs f…