Towards general conceptual model editing via adversarial representation engineering

• Main Authors: ZHANG, Yihao, WEI, Zeming, SUN, Jun, SUN, Meng
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2024
• Subjects: Software Engineering
• Online Access: https://ink.library.smu.edu.sg/sis_research/9833; https://ink.library.smu.edu.sg/context/sis_research/article/10833/viewcontent/2404.13752v3.pdf
• Institution: Singapore Management University

Since the rapid development of Large Language Models (LLMs) has achieved remarkable success, understanding and rectifying their internal complex mechanisms has become an urgent issue. Recent research has attempted to interpret their behaviors through the lens of inner representation. However, developing practical and efficient methods for applying these representations for general and flexible model editing remains challenging. In this work, we explore how to leverage insights from representation engineering to guide the editing of LLMs by deploying a representation sensor as an editing oracle. We first identify the importance of a robust and reliable sensor during editing, then propose an Adversarial Representation Engineering (ARE) framework to provide a unified and interpretable approach for conceptual model editing without compromising baseline performance. Experiments on multiple tasks demonstrate the effectiveness of ARE in various model editing scenarios. Our code and data are available at https://github.com/ Zhang-Yihao/Adversarial-Representation-Engineering.