{"slug": "uncertainty-gated-selection-for-block-sparse-attention", "title": "Uncertainty-gated selection for block-sparse attention", "summary": "Researchers propose a value-of-information router for block-sparse attention that dynamically doubles the kept key blocks when the top-k selection is uncertain, improving recall on LongBench-v2 from 0.47 to 0.75 and preserving dense accuracy on RULER NIAH within 2 percentage points, while running at 0.62–0.80x dense wall time on 128K contexts across Qwen2.5, Mistral-Nemo, and Qwen3.6 models.", "body_md": "arXiv:2607.07724v1 Announce Type: new\nAbstract: Block-sparse attention scales long-context language models by replacing the O(N^2) softmax with a per-query top-k selection over key blocks. This cutoff is myopic: when the k-th and (k+1)-th blocks are nearly tied in score, the selector commits without spending extra budget, and a dropped block carrying answer evidence is unrecoverable downstream. We propose a value-of-information router that measures, for each query, how decisively the top-k cut was made, and doubles the kept set for the queries where that gap is smallest; the rule is backbone-agnostic and stacks with existing block-scoring methods such as Quest. On LongBench-v2 medium at n=215 (the entire dataset subset), router-on-Quest reaches paired recall 0.75 vs. top-k 0.47 -- +28 pp over the SSA-style baseline (McNemar p<0.01) -- and lands within 2 pp of dense on RULER NIAH multikey at the same context. The lift reproduces on four models from three architectures (Qwen2.5, Mistral-Nemo, Qwen3.6). At 128K, the router preserves 0.81 and 0.89 of dense accuracy on Qwen2.5-7B-1M and Qwen3.6 (vs. SSA-style top-k at 0.09 on the former) while the fused selection-plus-kernel pipeline runs at 0.62x and 0.80x dense wall time.", "url": "https://wpnews.pro/news/uncertainty-gated-selection-for-block-sparse-attention", "canonical_source": "https://arxiv.org/abs/2607.07724", "published_at": "2026-07-10 04:00:00+00:00", "updated_at": "2026-07-10 04:16:15.334305+00:00", "lang": "en", "topics": ["large-language-models", "machine-learning", "artificial-intelligence", "ai-research"], "entities": ["Qwen2.5", "Mistral-Nemo", "Qwen3.6", "LongBench-v2", "RULER", "Quest"], "alternates": {"html": "https://wpnews.pro/news/uncertainty-gated-selection-for-block-sparse-attention", "markdown": "https://wpnews.pro/news/uncertainty-gated-selection-for-block-sparse-attention.md", "text": "https://wpnews.pro/news/uncertainty-gated-selection-for-block-sparse-attention.txt", "jsonld": "https://wpnews.pro/news/uncertainty-gated-selection-for-block-sparse-attention.jsonld"}}