LIVE: Learnable In-Context Vector for Visual Question Answering | Read Paper on Bytez