Activation Steering

Follow this topic Ask about this topic

Activation steering studies how edits to a model’s hidden activations can control its behavior at inference time, enabling changes without retraining the full model. Common approaches add or remove contrastive steering vectors, target sparse autoencoder features, neurons or attention heads, and use conditional or feedback-based controllers to adapt interventions to the input or generation state. Researchers apply these methods to language tasks such as reasoning, refusal, factuality and hallucination mitigation, as well as image and video generation, speech, and embodied agents; evaluations include benchmarks such as GSM8K, TruthfulQA, POPE and LIBERO. A central challenge is that readable or decodable features are not necessarily controllable, and effective edits can vary by context and architecture or cause collateral changes to fluency, utility and safety.

Latest papers

376 papers77 papers in the last four weeks, up 93% on the four weeks before. 0.5% of all new papers.

A Geometric Account of Activation Steering through Angle-Norm Decomposition
Jun 4, 2026Georgii Aparin, Tatiana GaintsevaLanguage Model Steering Language Modeling
Huawei Noah’s Ark Lab · Queen Mary University of London
Do Models Share Safety Representations? Cross-Model Steering for Safe Visual Generation
Jun 3, 2026Tobia Poppi, Silvia Cappelletti, Sara Sarto +5Image Generation Activation Steering
University of Modena and Reggio Emilia · University of Pisa · Amazon Prime Video
Fast & Faithful Function Vectors
Jun 3, 2026Minh An Pham, Anton Segeler, Thomas Wiegand +4Task Vectors LLM Interpretability
Fraunhofer Heinrich-Hertz-Institute, Berlin, Germany · Technische Universität Berlin, Berlin, Germany · BIFOLD – Berlin Institute for the Foundations of Learning and Data, Berlin, Germany +1
Activation Steering of Video Generation Models via Reduced-Order Linear Optimal Control
Jun 3, 2026Jihoon Hong, Alice Chan, Qiyue Dai +2T2V Generation Video Generation
Georgia Institute of Technology Atlanta, GA 30308
Rotate2Think: Geometric Priming via Orthogonal Rotation to Improve Language Model Reasoning
Jun 2, 2026Aditya Sharma, Christopher J. Pal, Amal ZouaqOrthogonal Procrustes Alignment Neural Representation Geometry
Polytechnique Montréal · Mila - Quebec AI Institute · LAMA-WeST Lab +1
Steer Where It Matters: Token-Level Visual-Sensitivity Steering for LVLMs Hallucination Mitigation
Jun 2, 2026Ruipeng Zhang, Zhihao Li, C. L. Philip Chen +1VLM Hallucination Large Vision-Language Models
Guangdong Provincial Key Laboratory of Computational AI Models and Cognitive Intelligence, School of Computer Science & Engineering, South China University of Technology, Guangzhou, China · Engineering Research Center of the Ministry of Education on Health Intelligent Perception and Paralleled Digital-Human, Guangzhou, China · Pazhou Lab, Guangzhou, China.
Hallucinations as Orthogonal Noise: Inference-Time Manifold Alignment via Dynamic Contextual Orthogonalization
Jun 2, 2026Mingkuan Zhao, Wentao Hu, Tianchen Huang +6LLM Hallucination Mitigation Hallucination in Language Models
Xi’an Jiaotong University · Xingchen AGI Lab, China Telecom AI Technology (Beijing) Co., Ltd. · University of Science and Technology of China +2
Low-Resource Safety Failures Are Action Failures, Not Representation Failures
May 31, 2026Rashad Aziz, Ikhlasul Akmal Hanif, Fajri KotoLanguage Model Calibration LLM Safety
Mohamed bin Zayed University of Artificial Intelligence
Subliminal Learning Is Steering Vector Distillation
May 31, 2026Camila Blank, Agam Bhatia, Senthooran Rajamanoharan +2Language Model Steering Subliminal Learning
Stanford University
Latent Reward Steering: An Adaptive Inference-Time Framework that Implicitly Promotes Cognitive Behaviors in Reasoning LLMs
May 30, 2026Jiakang Li, Guanyu Zhu, Can Jin +8Reward Modeling Language Model Steering
Rutgers University · South China Agricultural University · Columbia University +4
Closed-Loop Neural Activation Control in Vision-Language-Action Models
May 29, 2026Abhijith Babu, Ramneet Kaur, Nathaniel D. Bastian +5Closed-Loop Control of VLA Models Vision-Language-Action Models
Florida International University · SRI International · United States Military Academy +2
FBHM: Functional Benchmarking and Steering of VLMs for Hateful Meme Detection
May 29, 2026Paramananda Bhaskar, Naquee Rizwan, Daksh Jogchand +2VLM Evaluation VLM Adaptation
1Indian Institute of Technology (IIT), Kharagpur · 2Microsoft
Latent Space Disentanglement via Activation Steering for Interpretable Attribute Control in Symbolic Music Generation
May 29, 2026Ioannis Prokopiou, Pantelis Vikatos, Maximos Kaliakatsos-Papakostas +2Disentangled Representation Learning Controllable Music Generation
Athens University of Economics and Business Orfium Athens, Greece · Innovation Lab Orfium Athens, Greece · Department of Music Technology and Acoustics Hellenic Mediterranean University Rethymno, Greece +2
AnchorSteer: Self-Discovered Concept Injection for Structure-Preserving Music Editing
May 29, 2026Chih-Heng Chang, Keng-Seng Ho, Chih-Yu Tsai +3Audio Editing Controllable Music Generation
National Taiwan University Taipei, Taiwan
Cross-Lingual Steering for Figurative Language Generation
May 28, 2026Linfeng Liu, Tiffany Zhan, Louie Hong Yao +2Multilingual Language Models Language Model Steering
Department of Computer Science, University of Cincinnati · School of Computer Science, Carnegie Mellon University · Independent Researcher
UniSteer: Text-Guided Flow Matching in Activation Space for Versatile LLM Steering
May 28, 2026Yingdong Shi, Ruiming Zhang, Changming Li +4Flow Matching Language Model Steering
ShanghaiTech University
Mitigating Hallucination in Vision-Language Models through Barrier-Regulated Adaptive Closed-form Steering
May 28, 2026Soumyadeep Jana, Pulkit Mittal, Sanasam Ranbir SinghHallucination in Language Models Large Vision-Language Models
Indian Institute of Technology Guwahati, India
Relational Rank Geometry in Transformers: Detecting and Steering Hidden-State Relation Frames
May 28, 2026Mazen KobroslyTransformer Interpretability Neural Representation Geometry
Independent Researcher
DenseSteer: Steering Small Language Models towards Dense Math Reasoning
May 28, 2026Yang Ouyang, Shuhang Lin, Jung-Eun KimSmall Language Models Activation Steering
1North Carolina State University · 2Rutgers University
Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detection
May 27, 2026Vijeta Deshpande, Tootiya Giyahchi, Veena Padmanabhan +2Synthetic Data Generation Synthetic Data Generation for Language Models
UMass Lowell · Amazon · Amazon, CMU +1
Mitigating Adaptive Attacks against Reasoning Models with Activation Consistency Training
May 27, 2026Avidan Shah, Jannik Brinkmann, Rico AngellPrompt Injection Defense Adversarial Robustness of Language Models
New York University · Technical University Clausthal
Rethinking Visual Neglect: Steering via Context-Preference for MLLM Hallucination Mitigation
May 27, 2026Jingwen Wu, Xijun Zhang, Ge SongLLM Hallucination Mitigation Hallucination in Language Models
School of Computer and Electronic Information, Nanjing Normal University, China
Beyond a Single Direction: Chain-of-Thought Disrupts Simple Steering of Refusal
May 26, 2026Kia-Jüng Yang, Dominik Meier, Jiachen Zhao +2Reasoning Trace Analysis LLM Refusal Behavior
University of Göttingen, Germany · LKA NRW, Germany · Northeastern University, Boston, MA, USA.
Cultural Value Alignment Via Latent Activation Steering in Large Language Models
May 25, 2026Trung Duc Anh Dang, Sarah MasudCultural Alignment Cultural Bias in Language Models
University of Copenhagen
Analogies between Transformer Layers and Power Method
May 25, 2026Chenglong Li, Claudio AltafiniTransformer Interpretability Transformer
Universal Boosts, Specific Suppressors: Sparse Autoencoder Steering of Medical Vision-Language Models
May 24, 2026Farhad Nooralahzadeh, Benjamin Gundersen, Nicolas Deperrois +7Radiology Report Generation Healthcare
University of Zurich and University Hospital of Zurich, Switzerland · Zurich University of Applied Sciences, Switzerland · Kobe University, Japan +3
Steering Beyond the Support: Adversarial Training on Unsupervised Jailbroken Activation Simulation
May 23, 2026Luoyu Chen, Weiqi Wang, Zhiyi Tian +5Adversarial Training Adversarial Robustness of Language Models
University of Technology Sydney, Sydney, Australia · Xi’an Jiaotong University, Xi’an, China · School of Cyber Science and Engineering, Southeast University, Nanjing, China.
Causal Physics Steering in Video World Models via Concept Activation Vectors
May 23, 2026Nahid AlamTransformer Interpretability Activation Steering
Oreon Labs, Cohere Labs Community
As X, Do Y: How Persona and Task Combine in Instruction-Tuned LLMs
May 22, 2026Eric XuLLM Prompting LLM Interpretability
Independent Researcher
Steered Generation via Gradient-Based Optimization on Sparse Query Features
May 21, 2026Sumanta Bhattacharyya, Pedram RooshenasConstrained Motion Planning Language Model Steering
Department of Computer Science University of Illinois Chicago

Jun 4, 2026J/K move · Enter open · S save

A Geometric Account of Activation Steering through Angle-Norm Decomposition

Georgii Aparin, Tatiana Gaintseva

Huawei Noah’s Ark Lab · Queen Mary University of London

Language Model Steering Language Modeling Causal Interventions in Language Models Activation Steering

Open paper AskPDF arXiv

Linear activation steering has gained popularity as a simple and empirically effective way to control language model behavior. More recently, spherical steering paradigms have been proposed to address limitations of additive interventions, often motivated by the assumption that hidden-state norm does not carry concept-relevant information. In this work, we revisit this assumption through a controlled empirical study designed to disentangle the roles of angular and radial components. We show that steering methods differ mainly in how they couple two geometric effects: changing a token's angular alignment with a concept direction and changing its hidden-state norm. Across seven language models, we find that concepts are represented primarily in angular structure, supporting the motivation for spherical methods, but that norm remains important for the stability and downstream effects of steering. Our results explain why interventions with similar concept-level effects can behave differently, and suggest that activation steering should be parameterized by interpretable angular and radial components of the intervention, rather than by a single additive coefficient that entangles these two effects.