Global Evolutionary Steering: Refining Activation Steering Control via Cross-Layer Consistency

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Jiang, Xinyan, Yu, Wenjing, Wang, Di, Hu, Lijie
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917336162762752
author Jiang, Xinyan
Yu, Wenjing
Wang, Di
Hu, Lijie
author_facet Jiang, Xinyan
Yu, Wenjing
Wang, Di
Hu, Lijie
contents Activation engineering enables precise control over Large Language Models (LLMs) without the computational cost of fine-tuning. However, existing methods deriving vectors from static activation differences are susceptible to high-dimensional noise and layer-wise semantic drift, often capturing spurious correlations rather than the target intent. To address this, we propose Global Evolutionary Refined Steering (GER-steer), a training-free framework that grounded in the geometric stability of the network's representation evolution. GER-steer exploits this global signal to rectify raw steering vectors, effectively decoupling robust semantic intent from orthogonal artifacts. Extensive evaluations confirm that GER-steer consistently outperforms baselines, delivering superior efficacy and generalization without layer-specific tuning, establishing a universal solution for reliable model alignment.
format Preprint
id arxiv_https___arxiv_org_abs_2603_12298
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Global Evolutionary Steering: Refining Activation Steering Control via Cross-Layer Consistency
Jiang, Xinyan
Yu, Wenjing
Wang, Di
Hu, Lijie
Machine Learning
Artificial Intelligence
Activation engineering enables precise control over Large Language Models (LLMs) without the computational cost of fine-tuning. However, existing methods deriving vectors from static activation differences are susceptible to high-dimensional noise and layer-wise semantic drift, often capturing spurious correlations rather than the target intent. To address this, we propose Global Evolutionary Refined Steering (GER-steer), a training-free framework that grounded in the geometric stability of the network's representation evolution. GER-steer exploits this global signal to rectify raw steering vectors, effectively decoupling robust semantic intent from orthogonal artifacts. Extensive evaluations confirm that GER-steer consistently outperforms baselines, delivering superior efficacy and generalization without layer-specific tuning, establishing a universal solution for reliable model alignment.
title Global Evolutionary Steering: Refining Activation Steering Control via Cross-Layer Consistency
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2603.12298