Tactic: Adaptive Sparse Attention with Clustering and Distribution Fitting for Long-Context LLMs

TL;DR

Tactic introduces an adaptive sparse attention mechanism for long-context LLMs that dynamically selects tokens based on attention importance, improving efficiency and accuracy over fixed-budget methods.

Contribution

It proposes a novel, calibration-free sparse attention method using clustering and distribution fitting to adaptively select tokens based on attention scores.

Findings

Achieves up to 7.29x speedup in decode attention

Outperforms existing sparse attention algorithms in accuracy

Provides a 1.58x overall inference speedup

Abstract

Long-context models are essential for many applications but face inefficiencies in loading large KV caches during decoding. Prior methods enforce fixed token budgets for sparse attention, assuming a set number of tokens can approximate full attention. However, these methods overlook variations in the importance of attention across heads, layers, and contexts. To address these limitations, we propose Tactic, a sparsity-adaptive and calibration-free sparse attention mechanism that dynamically selects tokens based on their cumulative attention scores rather than a fixed token budget. By setting a target fraction of total attention scores, Tactic ensures that token selection naturally adapts to variations in attention sparsity. To efficiently approximate this selection, Tactic leverages clustering-based sorting and distribution fitting, allowing it to accurately estimate token importance with minimal computational overhead. We show that Tactic outperforms existing sparse attention algorithms, achieving superior accuracy and up to 7.29x decode attention speedup. This improvement translates to an overall 1.58x end-to-end inference speedup, making Tactic a practical and effective solution for long-context LLM inference in accuracy-sensitive applications.