MRO: Enhancing Reasoning in Diffusion Language Models via Multi-Reward Optimization

Recent advances in diffusion language models (DLMs) have presented a promising alternative to traditional autoregressive large language models (LLMs).