Curse of High Dimensionality Issue in Transformer for Long-context Modeling

Transformer-based large language models (LLMs) excel in natural language processing tasks by capturing long-range dependencies through self-attention mechanisms. However, long-context modeling faces significant computational inefficiencies due to textitredundant attention computations: while attention weights are often textitsparse, all tokens consume textitequal computational resources. In this paper, we reformulate traditional probabilistic sequence modeling as a textitsupervised learning task, enabling the separation of relevant and irrelevant tokens and providing a clearer understanding of redundancy. Based on this reformulation, we theoretically analyze attention sparsity, revealing that only a few tokens significantly contribute to predictions. Building on this, we formulate attention optimization as a linear coding problem and propose a textitgroup coding strategy, theoretically showing its ability to improve robustness against random noise and enhance learning efficiency. Motivated by this, we propose textitDynamic Group Attention (DGA), which leverages the group coding to explicitly reduce redundancy by aggregating less important tokens during attention computation. Empirical results show that our DGA significantly reduces computational costs while maintaining competitive performance.Code is available at https://github.com/bolixinyu/DynamicGroupAttention.