Activation steering offers training-free and dynamic control of target behaviors in large language models, but existing methods can impair general capabilities and require domain-specific intervention designs. We introduce Capability-Preserving Neuron Steering (CANS), a unified framework that balances target-behavior c...
Yu-Chen Wu, Cheng-Yi Xie, Chen-Peng Wang et al.· 0 citations
Large Language Models (LLMs) frequently encounter conflicts between internal parametric knowledge and external contextual knowledge, yet the mechanisms governing this selection remain unclear. We show that the choice between context and parametric knowledge is linearly represented and can be controlled along a single d...
Yibin Lou, Yi Hu, Jian-Hui Chen et al.· 0 citations
Activation steering provides a lightweight and flexible way to control large language model (LLM) behavior. However, effective steering requires more than inducing the intended behavior: it should also limit unintended changes and remain robust across inputs and training data. Existing evaluations cover these dimension...
Hao-Tian Yang, Hui-Kang Jiang, Yu-Chen Wu et al.· 1 citation
Overall, representation engineering does not generally replace behavioral safeguards, but offers practical advantages under specific conditions and can provide complementary safety benefits.
Influence-guided response rewriting is introduced, which uses IF to identify intervention targets and replaces their responses with behavior-aligned or behavior-opposed supervision while keeping instructions fixed, motivating intervention-aware evaluation of TDA methods.
Yu-Zhang Luo, Chen-Peng Wang, Jian-Hui Chen et al.· 0 citations
This work introduces a diagnostic protocol using a minimal, target-label-free additive correction, showing that many apparent zero-shot reasoning deficits are expression failures masking intact internal logic, urging a narrower interpretation of benchmark evaluations.
