1 paper · 1 filter
Han Li, Xinyu Peng, Yaoming Wang +7
We introduce OneCAT, a unified multimodal model that seamlessly integrates understanding, generation, and editing within a novel, pure decoder-only transformer architecture. Our fr…