arXivVladislav Gromadskii, David Li, Samson Gourevitch, Yazid Janati, Eric Moulines, Maxim Panov, Alexander KorotinFri, Oct 2, 2026, 10:29 AM PDT
score 15.2
Method fine-tunes fast masked diffusion models with rewards
Original: IDRF: Inverse-Distilled Reward Fine-tuning of Masked Discrete Diffusion Models
Source: arxiv.org ↗
Writing ELI5 summary…