Lexicon3D: Probing Visual Foundation Models for Complex 3D Scene Understanding | Read Paper on Bytez