mlsanityData to 7 Oct 2026

Efficient and accurate steering of Large Language Models through attention-guided feature learning

arXiv 2602.00333 ↗First posted 30 Jan 2026cs.LGcited by 6

Not on the map. Why?

Abstract (arXiv)

Steering, or direct manipulation of internal activations to guide LLM responses toward specific semantic concepts, is emerging as a promising avenue for both understanding how semantic concepts are stored within LLMs and advancing LLM capabilities. Yet, existing steering methods are remarkably brittle, with seemingly non-steerable concepts becoming completely steerable based on subtle algorithmic choices in how concept-related features are extracted. In this work, we introduce an attention-guided steering framework that overcomes three core challenges associated with steering: (1) automatic selection of relevant token embeddings for extracting concept-related features; (2) accounting for heterogeneity of concept-related features across LLM activations; and (3) identification of layers most relevant for steering. Across a steering benchmark of 512 semantic concepts, our framework substantially improved steering over previous state-of-the-art (nearly doubling the number of successfully steered concepts) across model architectures and sizes (up to 70 billion parameter models). Furthermore, we use our framework to shed light on the distribution of concept-specific features across LLM layers. Overall, our framework opens further avenues for developing efficient, highly-scalable fine-tuning algorithms for industry-scale LLMs.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 46 of its 63 references that are in the map. First posted on arXiv 30 Jan 2026.

Builds on

46 of 63 references.

References · 46

  1. 2017Attention Is All You Needcited by 59,932
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2016Layer Normalizationcited by 5,579
  4. 2024GPT-4o System Cardcited by 4,467
  5. 2016Gaussian Error Linear Units (GELUs)cited by 3,431
  6. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 2,914
  7. 2023Efficient Streaming Language Models with Attention Sinkscited by 2,115
  8. 2023Jailbreaking Black Box Large Language Models in Twenty Queriescited by 1,326
  9. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,259
  10. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,239
  11. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,009
  12. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 960
34 more
  1. 2023Steering Language Models With Activation Engineeringcited by 824
  2. 2019Plug and Play Language Models: A Simple Approach to Controlled Text Generationcited by 793
  3. 2022Discovering Latent Knowledge in Language Models Without Supervisioncited by 733
  4. 2023The Linear Representation Hypothesis and the Geometry of Large Language Modelscited by 679
  5. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 673
  6. 2023Tree of Attacks: Jailbreaking Black-Box LLMs Automaticallycited by 619
  7. 2023Emergent Linear Representations in World Models of Self-Supervised Sequence Modelscited by 313
  8. 2025Persona Vectors: Monitoring and Controlling Character Traits in Language Modelscited by 303
  9. 2024Improving Alignment and Robustness with Circuit Breakerscited by 302
  10. 2023Function Vectors in Large Language Modelscited by 292
  11. 2023In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steeringcited by 240
  12. 2022Extracting Latent Steering Vectors from Pretrained Language Modelscited by 217
  13. 2025AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoderscited by 199
  14. 2024Improving Instruction-Following in Language Models through Activation Steeringcited by 149
  15. 2023Inspecting and Editing Knowledge Representations in Language Modelscited by 140
  16. 2024Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimizationcited by 112
  17. 2023Tell Your Model Where to Attend: Post-hoc Attention Steering for LLMscited by 82
  18. 2024On the Origins of Linear Representations in Large Language Modelscited by 76
  19. 2025LLMs Encode Harmfulness and Refusal Separatelycited by 75
  20. 2024Adaptive Activation Steering: A Tuning-Free LLM Truthfulness Improvement Method for Diverse Hallucinations Categoriescited by 74
  21. 2023Improving Activation Steering in Language Models with Mean-Centringcited by 64
  22. 2024SelfIE: Self-Interpretation of Large Language Model Embeddingscited by 63
  23. 2024Robust LLM safeguarding via refusal feature adversarial trainingcited by 60
  24. 2024Style Vectors for Steering Generative Large Language Modelcited by 56
  25. 2024Estimating Knowledge in Large Language Models Without Generating a Single Tokencited by 39
  26. 2024Extending Activation Steering to Broad Skills and Multiple Behaviourscited by 36
  27. 2025Linear Representations of Political Perspective Emerge in Large Language Modelscited by 30
  28. 2025Toward universal steering and monitoring of AI modelscited by 29
  29. 2024Emoji Attack: Enhancing Jailbreak Attacks Against Judge LLM Detectioncited by 20
  30. 2025Internal Activation Revision: Safeguarding Vision Language Models Without Parameter Updatecited by 15
  31. 2025Instruction Following by Principled Boosting Attention of Large Language Modelscited by 6
  32. 2025Activation Steering Meets Preference Optimization: Defense Against Jailbreaks in Vision Language Modelscited by 5
  33. 2025Attention Misses Visual Risk: Risk-Adaptive Steering for Multimodal Safety Alignmentcited by 4
  34. 2025Depth-Wise Activation Steering for Honest Language Modelscited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 6.

6 papers.

  1. 2026Catching rationalization in the act: detecting motivated reasoning before and after CoT via activation probingcited by 6
  2. 2026FaithSteer-BENCH: A Deployment-Aligned Stress-Testing Benchmark for Inference-Time Steeringcited by 1
  3. 2026Steer-to-Detect: Probing Hidden Representations for Detection of LLM-Generated Textscited by 0
  4. 2026Steered Generation via Gradient-Based Optimization on Sparse Query Featurescited by 0
  5. 2026Fast Multi-dimensional Refusal Subspaces via RFM-AGOPcited by 0
  6. 2026Attention is Case-Sensitivecited by 0