Unified Lexical Representation for Interpretable Visual-Language Alignment | Read Paper on Bytez