2025/05/26 by Dan Peng, Zhihui Fu, Peng, Dan +7 · 2 citations
Computer Science · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Digital Rights Management and Security #FOS: Computer and information sciences #Machine Learning (cs.LG) #Service-Oriented Architecture and Web Services
paper · pdf · doi:10.48550/arxiv.2505.19578
openalex publication_date 2025/05/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Sparse attention methods exploit the inherent sparsity in attention to speed up the prefilling phase of long-context inference, mitigating the quadratic complexity of full attention computation. While existing sparse attention methods rely on predefined patterns or inaccurate estimations to approximate attention behavior, they often fail to fully capture the true dynamics of attention, resulting in reduced efficiency and compromised accuracy. Instead, we propose a highly accurate sparse attention mechanism that shares similar yet precise attention patterns across heads, enabling a more realistic capture of the dynamic behavior of attention. Our approach is grounded in two key observations: (1) attention patterns demonstrate strong inter-head similarity, and (2) this similarity remains remarkably consistent across diverse inputs. By strategically sharing computed accurate patterns across attention heads, our method effectively captures actual patterns while requiring full attention computation for only a small subset of heads. Comprehensive evaluations demonstrate that our approach achieves superior or comparable speedup relative to state-of-the-art methods while delivering the best overall accuracy.