← back
arXivBlake Olson, Yuhang Song, Emmett McQuinn, Yuan ShangguanFri, Sep 11, 2026, 9:59 AM PDT
score 15.5

Teacher guidance speeds up reinforcement learning for diffusion language models

Original: CanvasAnneal: Curriculum Reinforcement Learning for Diffusion Language Models

Source: arxiv.org

Writing ELI5 summary…