Stem: Rethinking Causal Information Flow in Sparse Attention
This addresses a fundamental scaling problem for LLM developers and researchers, offering an incremental improvement over existing sparse methods.
The paper tackles the quadratic computational bottleneck of self-attention in Large Language Models for long contexts by proposing Stem, a sparsity module that rethinks causal information flow, resulting in superior accuracy with reduced computation and pre-filling latency.
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.