mlsanityData to 7 Oct 2026

Extending Activation Steering to Broad Skills and Multiple Behaviours

arXiv 2403.05767 ↗First posted 9 Mar 2024cs.LG cs.AI cs.CLcited by 36

Not on the map. Why?

Abstract (arXiv)

Current large language models have dangerous capabilities, which are likely to become more problematic in the future. Activation steering techniques can be used to reduce risks from these capabilities. In this paper, we investigate the efficacy of activation steering for broad skills and multiple behaviours. First, by comparing the effects of reducing performance on general coding ability and Python-specific ability, we find that steering broader skills is competitive to steering narrower skills. Second, we steer models to become more or less myopic and wealth-seeking, among other behaviours. In our experiments, combining steering vectors for multiple different behaviours into one steering vector is largely unsuccessful. On the other hand, injecting individual steering vectors at different places in a model simultaneously is promising.

Key facts. Cited by 36 papers in the map (data through 7 Oct 2026). Builds on 15 of its 19 references that are in the map. First posted on arXiv 9 Mar 2024.

Builds on

15 of 19 references.

References · 15

  1. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  2. 2017Deep reinforcement learning from human preferencescited by 3,846
  3. 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,346
  4. 2021Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processingcited by 2,222
  5. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,259
  6. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,239
  7. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,009
  8. 2023Steering Language Models With Activation Engineeringcited by 824
  9. 2022Discovering Language Model Behaviors with Model-Written Evaluationscited by 805
  10. 2023Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedbackcited by 743
  11. 2023Fast Machine Unlearning Without Retraining Through Selective Synaptic Dampeningcited by 177
  12. 2023Model evaluation for extreme riskscited by 167
3 more
  1. 2023Copy Suppression: Comprehensively Understanding an Attention Headcited by 73
  2. 2023Improving Activation Steering in Language Models with Mean-Centringcited by 64
  3. 2022Language models are better than humans at next-token predictioncited by 14

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 4; 2025: 14; 2026: 18.

36 papers.

Sort:

By area:

  1. 2025AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoderscited by 199
  2. 2024Improving Instruction-Following in Language Models through Activation Steeringcited by 149
  3. 2024Analyzing the Generalization and Reliability of Steering Vectorscited by 130
  4. 2025SAEs Are Good for Steering -- If You Select the Right Featurescited by 74
  5. 2025Multi-Attribute Steering of Language Models via Targeted Interventioncited by 44
  6. 2024Steering Large Language Models using Conceptors: Improving Addition-Based Activation Engineeringcited by 33
  7. 2025Towards Understanding Distilled Reasoning Models: A Representational Approachcited by 26
  8. 2025Beyond Prompt Engineering: Robust Behavior Control in LLMs via Steering Target Atomscited by 23
  9. 2025Beyond Linear Steering: Unified Multi-Attribute Control for Language Modelscited by 23
  10. 2026Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Modelscited by 22
  11. 2024Can sparse autoencoders be used to decompose and interpret steering vectors?cited by 19
  12. 2025Representation Engineering for Large-Language Models: Survey and Research Challengescited by 18
24 more
  1. 2025Spotlight Your Instructions: Instruction-following with Dynamic Attention Steeringcited by 18
  2. 2025Improved Representation Steering for Language Modelscited by 17
  3. 2025Adaptive Multi-Subspace Representation Steering for Attribute Alignment in Large Language Modelscited by 16
  4. 2025Learning to Steer: Input-dependent Steering for Multimodal LLMscited by 13
  5. 2026Steering Vector Fields for Context-Aware Inference-Time Control in Large Language Modelscited by 10
  6. 2025Does higher interpretability imply better utility? A Pairwise Analysis on Sparse Autoencoderscited by 9
  7. 2026Analysing the Safety Pitfalls of Steering Vectorscited by 9
  8. 2025Guiding Giants: Lightweight Controllers for Weighted Activation Steering in LLMscited by 8
  9. 2026Why Steering Works: Toward a Unified View of Language Model Parameter Dynamicscited by 8
  10. 2026Efficient and accurate steering of Large Language Models through attention-guided feature learningcited by 6
  11. 2026Fine-Grained Activation Steering: Steering Less, Achieving Morecited by 6
  12. 2026Compositional Steering of Large Language Models with Steering Tokenscited by 5
  13. 2025Failure by Interference: Language Models Make Balanced Parentheses Errors When Faulty Mechanisms Overshadow Sound Onescited by 4
  14. 2026The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluationcited by 3
  15. 2026Prompt-Activation Duality: Improving Activation Steering via Attention-Level Interventionscited by 3
  16. 2026FishBack: Pullback Fisher Geometry for Optimal Activation Steering in Transformerscited by 3
  17. 2026ORBIT: Training-Free Multi-Attribute Behavioral Steering via Orthogonal Subspace Rotationcited by 2
  18. 2026OPIUM: Mitigating Steering Externalities and Over-Refusal via Dual Objective Latent Optimizationcited by 2
  19. 2026Spillover-Aware Multi-Value Steering for Pluralistic LLM Alignmentcited by 1
  20. 2026The Master Key Hypothesis: Unlocking Cross-Model Capability Transfer via Linear Subspace Alignmentcited by 0
  21. 2026Towards the Readability of LLM-Generated Codes through Multitask Representation Engineeringcited by 0
  22. 2026Activation-Space Order-Swap Geometry: A Site-Asymmetry Auditcited by 0
  23. 2026Can Activation Steering Capture Multidimensional Authorship Style?cited by 0
  24. 2026Adaptive Multi-Value Control in LLMs via Causal Activation Steeringcited by 0