Successor Heads: Recurring, Interpretable Attention Heads In The Wild
원문 초록
In this work we present successor heads: attention heads that increment tokens with a natural ordering, such as numbers, months, and days. For example, successor heads increment 'Monday' into 'Tuesday'. We explain the successor head behavior with an approach rooted in mechanistic interpretability, the field that aims to explain how models complete tasks in human-understandable terms. Existing research in this area has found interpretable language model components in small toy models. However, results in toy models have not yet led to insights that explain the internals of frontier models and little is currently understood about the internal operations of large language models. In this paper, we analyze the behavior of successor heads in large language models (LLMs) and find that they implement abstract representations that are common to different architectures. They form in LLMs with as few as 31 million parameters, and at least as many as 12 billion parameters, such as GPT-2, Pythia, and Llama-2. We find a set of 'mod-10 features' that underlie how successor heads increment in LLMs across different architectures and sizes. We perform vector arithmetic with these features to edit head behavior and provide insights into numeric representations within LLMs. Additionally, we study the behavior of successor heads on natural language data, identifying interpretable polysemanticity in a Pythia successor head.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Specification-based Security Evaluation of Real-World Agents Tool-Calling under Long-Horizon Context Tasks
Making Language Model a Hierarchical Classifier and Generator
InstructFLIP: Exploring Unified Vision-Language Model for Face Anti-spoofing
Describe Anything Model for Visual Question Answering on Text-rich Images
Assay2Mol: large language model-based drug design using BioAssay context
Kodezi Chronos: A Debugging-First Language Model for Repository-Scale, Memory-Driven Code Understanding
토론
이 논문과 연결된 포럼 글이 아직 없습니다.