@inproceedings{zhu-etal-2026-policy,
    title = "On-Policy Self-Distillation for Efficient Diffusion Language Models with Early-Stage Calibration",
    author = "Zhu, Huaisheng  and
      Liu, MingYu  and
      Liu, Junze  and
      Ge, Zhen  and
      Wang, Tian  and
      Gesi, Jiri  and
      Wang, Dakuo  and
      Zhang, Weiqi  and
      Zhang, Houyu  and
      Guo, Yufan  and
      Li, Xian  and
      Yin, Bing  and
      Sanghavi, Sujay",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.1344/",
    pages = "26954--26965",
    ISBN = "979-8-89176-395-1",
    abstract = "Diffusion Large Language Models (DLLMs) have recently achieved strong performance, e.g., masked diffusion models (MDMs) can surpass autoregressive models (ARMs) in various tasks. However, DLLMs often struggle with inaccurate early-stage predictions due to limited context, which hinders both the model{'}s inference efficiency and the output{'}s overall quality. We propose Calibrated On-Policy Self-Distillation (COPSD) for DLLMs, a simple and efficient method to calibrate early token predictions without requiring demonstration data. COPSD distills an unnormalized target distribution derived from later decoding steps into the original model, enabling more accurate early predictions during inference. Experiments on math, planning, and RLHF tasks show that COPSD improves both effectiveness and efficiency, and further enhances performance when combined with supervised fine-tuning."
}