Attention Heads of Large Language Models: A Survey
This survey addresses the critical challenge of demystifying LLMs as black-box systems for researchers and practitioners, but it is incremental as it synthesizes existing research rather than proposing new methods.
The authors tackled the problem of understanding the reasoning bottlenecks in Large Language Models (LLMs) by systematically exploring the roles and mechanisms of attention heads, introducing a novel four-stage framework to categorize their functions and analyzing experimental methodologies for discovery.
Since the advent of ChatGPT, Large Language Models (LLMs) have excelled in various tasks but remain as black-box systems. Understanding the reasoning bottlenecks of LLMs has become a critical challenge, as these limitations are deeply tied to their internal architecture. Among these, attention heads have emerged as a focal point for investigating the underlying mechanics of LLMs. In this survey, we aim to demystify the internal reasoning processes of LLMs by systematically exploring the roles and mechanisms of attention heads. We first introduce a novel four-stage framework inspired by the human thought process: Knowledge Recalling, In-Context Identification, Latent Reasoning, and Expression Preparation. Using this framework, we comprehensively review existing research to identify and categorize the functions of specific attention heads. Additionally, we analyze the experimental methodologies used to discover these special heads, dividing them into two categories: Modeling-Free and Modeling-Required methods. We further summarize relevant evaluation methods and benchmarks. Finally, we discuss the limitations of current research and propose several potential future directions.