#multimodal diffusion model

MMaDA is a novel unified multimodal diffusion model that excels in textual reasoning, visual understanding, and image generation, outperforming existing systems across multiple benchmarks.