Lavida-O: Elastic Large Masked Diffusion Models for Unified Multimodal Understanding and Generation Paper • 2509.19244 • Published Sep 23 • 11