AutoLexSteer: Automatic Contrast Construction for Lexical Activation Steering

2026-09-06

AI Summary

为了解决构建准确引导向量的难题，本文提出了AutoLexSteer，一种全自动过程，利用WordNet中紧密相关的词族来构建引导向量。

Abstract

Steering vectors have rapidly emerged as a popular and effective method for guiding the output of LLMs in very specific ways. But constructing accurate steering vectors is a difficult manual process due to the opacity of embeddings. We introduce Hangman, a novel type of steering vector that operates using word senses, as well as AutoLexSteer, the first fully automated process for building steering vectors. AutoLexSteer employs families of closely-related words extracted from WordNet to specify both the steering source to be avoided and the desired steering target. The steering vectors are quite precise, can be used to steer at the level of words and sets of word senses (meanings), and are able to steer certain LLM behaviors like sycophancy. The dataset and code can be found at https://github.com/ShuheWang1998/autolexsteer.

Problem

• steering vectors
• embeddings
• manual process

Innovation

• steering vector
• AutoLexSteer
• word senses
• automated process
• WordNet