HiUni: Harmonizing Understanding and Generation in Shared Visual Space
Unified multimodal models aim to support visual understanding and generation within a single framework. However, prevailing approaches often separate their visual representations or backbone computati
- Upvotes18