Decoupling Visual Encoding for Unified Multimodal Understanding and Generationarxiv.org2 points·jinqueeny··0 commentsOpen articleSaveView on HN