Householder Pseudo-Rotation: A Novel Approach to Activation Editing in LLMs with Direction-Magnitude Perspective

TL;DR

This paper introduces Householder Pseudo-Rotation, a new activation editing method for LLMs that considers direction and magnitude, leading to better performance and safety in model behavior modification.

Contribution

The paper proposes a novel activation editing technique based on direction-magnitude perspective, improving upon existing methods by preserving activation norms.

Findings

Enhanced safety benchmark performance

Preserved activation magnitudes during editing

Outperformed existing editing methods

Abstract

Activation Editing, which involves directly editting the internal representations of large language models (LLMs) to alter their behaviors and achieve desired properties, has emerged as a promising area of research. Existing works primarily treat LLMs' activations as points in space and modify them by adding steering vectors. However, this approach is limited in its ability to achieve greater performance improvement while maintaining the necessary consistency of activation magnitudes. To overcome these issues, we propose a novel editing method that views activations in terms of their directions and magnitudes. Our method, named Householder Pseudo-Rotation (HPR), mimics the rotation transformation, thus preserving activation norms and resulting in an improved performance on various safety benchmarks.