1 paper
Jiahan Li, Huangyuan Su, Shengbang Tong +9
Most unified multimodal models treat images and text asymmetrically: a pretrained language model generates text, an attached diffusion head generates images, and outputs that conta…