Soft-Di[M]O: Improving One-Step Discrete Image Generation with Soft Embeddings

ArXi:2509.22925v2 Announce Type: replace-cross One-step generators distilled from Masked Diffusion Models (MDMs) compress multiple sampling steps into a single forward pass, enabling efficient text and image synthesis. However, they suffer two key limitations: they inherit modeling bias from the teacher, and their discrete token outputs block gradient flow, preventing post-distillation refinements such as adversarial