English(EN) SparseGDN: Chunk Sparse in Linear Attention

SparseGDN框架提升长上下文线性注意力效率

作者 PulseAugur 编辑部 · [1 个来源] · 2026-11-02 05:00

研究人员开发了SparseGDN，一个旨在提高序列建模中线性注意力机制效率的新框架。该方法通过引入结构化块内近似和依赖输入的循环事件选择，解决了长上下文预填充的计算成本问题。该框架的GPU实现选择性地物化更新并重建token输出，从而显著加快了预填充操作的速度。在Qwen3.5模型上的评估表明，在保留相当一部分循环事件的同时，性能得到了显著提升。 AI

影响该框架可能导致大型语言模型更高效的训练和推理，特别是那些需要长上下文窗口的模型。

排序理由该集群包含一篇学术论文，详细介绍了用于提高AI模型效率的新技术框架。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CL 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 1 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇学术论文，详细介绍了用于提高AI模型效率的新技术框架。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, infra; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 1 days old; Coverage has settled into its steady-state source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.CL TIER_1 English(EN) · Jiedong Zhuang, Peng Zhao, Jialin Tang, Haoji Hu · 2026-11-02 05:00

SparseGDN: Chunk Sparse in Linear Attention

arXiv:2610.41420v1 Announce Type: cross Abstract: Gated Delta Networks (GDNs) offer linear-time sequence modeling, yet long-context prefill remains costly due to dense intra-chunk computation, recurrent-state updates, and intermediate state storage. We introduce SparseGDN, a hard…

本文如何被排名

报道来源 [1]

SparseGDN: Chunk Sparse in Linear Attention

相关实体

相关话题