Taxonomy, Opportunities, and Challenges of Representation Engineering for Large Language Models
This work addresses the challenge of controlling LLM behavior for researchers and practitioners, but it is incremental as it surveys and organizes existing literature rather than presenting new experimental results.
The paper tackles the problem of controlling large language models (LLMs) by introducing Representation Engineering (RepE) as a novel paradigm that directly manipulates internal representations, offering more effective, interpretable, data-efficient, and flexible control compared to traditional methods. It provides the first comprehensive survey of RepE, reviewing existing methods, applications, and challenges, and proposes a unified framework and best practices guide.
Representation Engineering (RepE) is a novel paradigm for controlling the behavior of LLMs. Unlike traditional approaches that modify inputs or fine-tune the model, RepE directly manipulates the model's internal representations. As a result, it may offer more effective, interpretable, data-efficient, and flexible control over models' behavior. We present the first comprehensive survey of RepE for LLMs, reviewing the rapidly growing literature to address key questions: What RepE methods exist and how do they differ? For what concepts and problems has RepE been applied? What are the strengths and weaknesses of RepE compared to other methods? To answer these, we propose a unified framework describing RepE as a pipeline comprising representation identification, operationalization, and control. We posit that while RepE methods offer significant potential, challenges remain, including managing multiple concepts, ensuring reliability, and preserving models' performance. Towards improving RepE, we identify opportunities for experimental and methodological improvements and construct a guide for best practices.