Why Attention Patterns Exist: A Unifying Temporal Perspective Analysis

• URL: http://arxiv.org/abs/2601.21709v1
• Date: Thu, 29 Jan 2026 13:40:23 GMT
• Title: Why Attention Patterns Exist: A Unifying Temporal Perspective Analysis
• Authors: Qingyue Yang, Jie Wang, Xing Li, Yinqi Bai, Xialiang Tong, Huiling Zhen, Jianye Hao, Mingxuan Yuan, Bin Li,
• Abstract summary: We introduce textbfTemporal Attention Pattern Predictability Analysis (TAPPA)<n>TAPPA characterizes attention patterns as predictable patterns with clear regularities and unpredictable patterns that appear effectively random.<n>We provide a detailed mathematical analysis of three representative cases through the joint effect of queries, keys, and Rotary Positional Embeddings (RoPE)
• Score: 61.597286699809395
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Attention patterns play a crucial role in both training and inference of large language models (LLMs). Prior works have identified individual patterns such as retrieval heads, sink heads, and diagonal traces, yet these observations remain fragmented and lack a unifying explanation. To bridge this gap, we introduce \textbf{Temporal Attention Pattern Predictability Analysis (TAPPA), a unifying framework that explains diverse attention patterns by analyzing their underlying mathematical formulations} from a temporally continuous perspective. TAPPA both deepens the understanding of attention behavior and guides inference acceleration approaches. Specifically, TAPPA characterizes attention patterns as predictable patterns with clear regularities and unpredictable patterns that appear effectively random. Our analysis further reveals that this distinction can be explained by the degree of query self-similarity along the temporal dimension. Focusing on the predictable patterns, we further provide a detailed mathematical analysis of three representative cases through the joint effect of queries, keys, and Rotary Positional Embeddings (RoPE). We validate TAPPA by applying its insights to KV cache compression and LLM pruning tasks. Across these tasks, a simple metric motivated by TAPPA consistently improves performance over baseline methods. The code is available at https://github.com/MIRALab-USTC/LLM-TAPPA.

Related papers

• A Self-explainable Model of Long Time Series by Extracting Informative Structured Causal Patterns [22.54910673667678]
  We propose EXCAP, a unified framework for interpretable time-series modeling.<n>We show that EXCAP provides smooth and stable explanations over time and is robust to perturbations in causal masks.<n>These results show that EXCAP offers a principled and scalable approach to interpretable modeling of long time series with relevance to high-stakes domains such as healthcare and finance.
  arXiv  Detail & Related papers  (2025-12-01T08:33:33Z)
• Learning to Focus: Causal Attention Distillation via Gradient-Guided Token Pruning [62.23671919314693]
  Large language models (LLMs) have demonstrated significant improvements in contextual understanding.<n>However, their ability to attend to truly critical information during long-context reasoning and generation still falls behind the pace.<n>We introduce a two-stage framework called Learning to Focus (LeaF) to mitigate confounding factors.
  arXiv  Detail & Related papers  (2025-06-09T15:16:39Z)
• A Closer Look at Bias and Chain-of-Thought Faithfulness of Large (Vision) Language Models [58.32070787537946]
  Chain-of-thought (CoT) reasoning enhances performance of large language models.<n>We present the first comprehensive study of CoT faithfulness in large vision-language models.
  arXiv  Detail & Related papers  (2025-05-29T18:55:05Z)
• Don't Take Things Out of Context: Attention Intervention for Enhancing Chain-of-Thought Reasoning in Large Language Models [32.71672086718058]
  Few-shot Chain-of-Thought (CoT) significantly enhances the reasoning capabilities of large language models (LLMs)<n>We observe that isolated segments, words, or tokens within CoT demonstrations can unexpectedly disrupt the generation process of LLMs.<n>We propose a Few-shot Attention Intervention method (FAI) that dynamically analyzes the attention patterns of demonstrations to accurately identify these tokens.
  arXiv  Detail & Related papers  (2025-03-14T07:46:33Z)
• Learning Multi-graph Structure for Temporal Knowledge Graph Reasoning [3.3571415078869955]
  This paper proposes an innovative reasoning approach that focuses on Learning Multi-graph Structure (LMS) LMS incorporates an adaptive gate for merging entity representations both along and across timestamps effectively. It also integrates timestamp semantics into graph attention calculations and time-aware decoders.
  arXiv  Detail & Related papers  (2023-12-04T08:23:09Z)
• Generic Temporal Reasoning with Differential Analysis and Explanation [61.96034987217583]
  We introduce a novel task named TODAY that bridges the gap with temporal differential analysis. TODAY evaluates whether systems can correctly understand the effect of incremental changes. We show that TODAY's supervision style and explanation annotations can be used in joint learning.
  arXiv  Detail & Related papers  (2022-12-20T17:40:03Z)
• Self-Attention Neural Bag-of-Features [103.70855797025689]
  We build on the recently introduced 2D-Attention and reformulate the attention learning methodology. We propose a joint feature-temporal attention mechanism that learns a joint 2D attention mask highlighting relevant information.
  arXiv  Detail & Related papers  (2022-01-26T17:54:14Z)
• SparseBERT: Rethinking the Importance Analysis in Self-attention [107.68072039537311]
  Transformer-based models are popular for natural language processing (NLP) tasks due to its powerful capacity. Attention map visualization of a pre-trained model is one direct method for understanding self-attention mechanism. We propose a Differentiable Attention Mask (DAM) algorithm, which can be also applied in guidance of SparseBERT design.
  arXiv  Detail & Related papers  (2021-02-25T14:13:44Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.