PaperScope
LIVE · 2026-09-09 05:40 UTC

Disentangling Steering Vectors

Takeru Hiramatsu, Kyohei Atarashi, Koh Takeuchi, Hisashi Kashima

Latestcs.CLcs.LGcs.AIcs.CV
arXiv ID
2609.07037 v1
Category
Submitted
2026-09-07

Abstract

Activation steering has emerged as a lightweight, inference-time approach to control the behavior of Large Language Models (LLMs). However, traditional steering vectors used to intervene in LLMs' activations, such as those derived from the difference-in-means method, tend to entangle multiple semantic and stylistic concepts into a single composite direction, leading to unpredictable steering effects. Our core objective is to disentangle this composite direction into its constituent concepts. To this end, we propose Steering Vector Dissection, a framework to explicitly isolate individual and semantically consistent features from these composite directions. Specifically, we pair positive and negative activations and take their differences to generate a set of instance-level steering vectors, and train a dedicated Sparse Autoencoder (SAE) directly on them. Quantitative evaluations across two datasets, two models, and two intervention depths show that our method yields a set of semantically consistent basis vectors whose steering effects are mutually distinguishable. Furthermore, we show that this disentanglement enables precise control over model behaviors.

Comment: 31 pages, 5 figures

arXiv abs page · PDF · same-day batch