Unity by Diversity: Improved Representation Learning for Multimodal VAEs | Read Paper on Bytez