news

May 21, 2025 MMaDA is released, and I serve as the core code contributor. We built a novel unified multimodal understanding and generation model, purely with a discrete diffusion backbone.