arXiv · cs.AI · Yufei Xu, Fanxu Meng, Fan Jiang, Yuxuan Wang, Ruijie Zhou, Zhaohui Wang, Jiexi Wu, Zhixin Pan, Xiaojuan Tang, Wenjie Pei, Tongxuan Liu, Di Yin, Xing Sun, Muhan Zhang · 5 小时前模型评分42
HISA：面向细粒度稀疏注意力的高效分层索引方法
HISA: Efficient Hierarchical Indexing for Fine-Grained Sparse Attention
导读
HISA 提出一种即插即用的分层索引器，把逐 token 扫描改为块级粗筛加 token 级精筛两阶段流程，无需额外训练即可保持原有 top-sparse 模式。
分类置信度
中 · 待人工复核
- 只有单一媒体来源，需要人工复核。
来源级别：媒体与个人
事件类型：研究发布
证据摘录
- “We propose HISA (Hierarchical Indexed Sparse Attention), a plug-and-play replacement for the indexer that rewrites the search path from a flat token scan into a two-stage hierarchical procedure”
- “HISA preserves the identical token-level top-sparse pattern consumed by the downstream Sparse MLA operator and requires no additional training.”
- “On Needle-in-a-Haystack and LongBench, we directly replace the indexer in DeepSeek-V3.2 and GLM-5 with our HISA indexer, without any finetuning.”
来源：arXiv · cs.AI · arxiv.org