MMaDA-VLA: Large Diffusion Vision-Language-Action Model with Unified Multi-Modal Instruction and Generation Paper • 2603.25406 • Published Mar 26 • 5
