Stem: Rethinking Causal Information Flow in Sparse Attention

TL;DR

Stem introduces a novel sparsity module for causal attention in LLMs, improving efficiency and accuracy by considering token position and importance, thus addressing the quadratic complexity bottleneck.

Contribution

The paper proposes Stem, a new plug-and-play sparsity method that aligns with information flow, using position-dependent top-k and output-aware metrics for better causal attention.

Findings

Achieves higher accuracy with less computation

Reduces pre-filling latency in large language models

Effectively preserves important tokens during attention

Abstract

The quadratic computational complexity of self-attention remains a fundamental bottleneck for scaling Large Language Models (LLMs) to long contexts, particularly during the pre-filling phase. In this paper, we rethink the causal attention mechanism from the perspective of information flow. Due to causal constraints, tokens at initial positions participate in the aggregation of every subsequent token. However, existing sparse methods typically apply a uniform top-k selection across all token positions within a layer, ignoring the cumulative dependency of token information inherent in causal architectures. To address this, we propose Stem, a novel, plug-and-play sparsity module aligned with information flow. First, Stem employs the Token Position-Decay strategy, applying position-dependent top-k within each layer to retain initial tokens for recursive dependencies. Second, to preserve information-rich tokens, Stem utilizes the Output-Aware Metric. It prioritizes high-impact tokens based on approximate output magnitude. Extensive evaluations demonstrate that Stem achieves superior accuracy with reduced computation and pre-filling latency.