Research
AutoLexSteer: Automatic Contrast Construction for Lexical Activation Steering
arXiv:2609.06879v1 Announce Type: cross Abstract: Steering vectors have rapidly emerged as a popular and effective method for guiding the output of LLMs in very specific ways. But constructing accurat
arXiv:2609.06879v1 Announce Type: cross 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.
Related
- SHIFT: Gate-Modulated Activation Steering for Knowledge Conflict Mitigation in Retrieval-Augmented Generation
- TimpaTeks: Automatic In-place Text Sequence Modification via Diffusion Language Model Steering
- Steering Where to Listen: Instruction-Based Activation Steering Redirects Temporal Attention in Large Audio-Language Models
Source: arXiv cs.AI | 2026-09-10