@inproceedings{sun-etal-2026-rectified,
    title = "Rectified Sparse Attention for Efficient Long-Sequence Generation",
    author = "Sun, Yutao  and
      Ye, Tianzhu  and
      Dong, Li  and
      Xia, Yuqing  and
      Chen, Jian  and
      Gao, Yizhao  and
      Cao, Shijie  and
      Wang, Jianyong  and
      Wei, Furu",
    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.348/",
    pages = "7023--7034",
    ISBN = "979-8-89176-395-1",
    abstract = "Efficient long-sequence generation is a critical challenge for Large Language Models. While recent sparse decoding methods improve efficiency, they suffer from KV cache misalignment, where approximation errors accumulate and degrade generation quality. In this work, we propose Rectified Sparse Attention (ReSA), a simple yet effective method that combines block-sparse attention with periodic dense rectification. By refreshing the KV cache at fixed intervals using a dense forward pass, ReSA bounds error accumulation and preserves alignment with the pretraining distribution. Experiments across math reasoning, language modeling, and retrieval tasks demonstrate that ReSA achieves near-lossless generation quality with significantly improved efficiency. Notably, ReSA delivers up to 3.77x end-to-end speedup under decoding at 256K sequence length, making it a practical solution for scalable long-context inference."
}