mlsanityData to 7 Oct 2026

In-Distribution Steering: Balancing Control and Coherence in Language Model Generation

arXiv 2510.13285 ↗First posted 15 Oct 2025cs.CLcited by 8

Not on the map. Why?

Abstract (arXiv)

Activation steering methods control large language model (LLM) behavior by modifying internal activations at inference time. However, most existing activation steering methods rely on a fixed steering strength, leading to either insufficient control or unadapted intervention that degrades text plausibility and coherence. We introduce In-Distribution Steering (IDS), a novel method that adapts steering strength based on the input data distribution in representation space. IDS dynamically adjusts interventions according to how far a given input lies within the distribution, enabling adaptive intervention and generation stability during text generation. Experiments demonstrate that IDS achieves strong accuracy on classification tasks while producing coherent text without collapse, making IDS particularly well suited for real-world applications.

Key facts. Cited by 8 papers in the map (data through 7 Oct 2026). Builds on 19 of its 37 references that are in the map. First posted on arXiv 15 Oct 2025.

Builds on

19 of 37 references.

References · 19

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  2. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  3. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  4. 2017Deep reinforcement learning from human preferencescited by 3,846
  5. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  6. 2016Understanding intermediate layers using linear classifier probescited by 1,279
  7. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,259
  8. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,239
  9. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,009
  10. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 960
  11. 2023Steering Language Models With Activation Engineeringcited by 824
  12. 2023AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedbackcited by 738
7 more
  1. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 673
  2. 2023Large Language Model Alignment: A Surveycited by 241
  3. 2024Reinforcement Learning for LLM Post-Training: A Surveycited by 91
  4. 2023Interpreting the Curse of Dimensionality from Distance Concentration and Manifold Effectcited by 14
  5. 2025The Blessing and Curse of Dimensionality in Safety Alignmentcited by 9
  6. 2024Self-AMPLIFY: Improving Small Language Models with Self Post Hoc Explanationscited by 8
  7. 2025Towards Achieving Concept Completeness for Textual Concept Bottleneck Modelscited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 8.

8 papers.

  1. 2026Local Linearity of LLMs Enables Activation Steering via Model-Based Linear Optimal Controlcited by 8
  2. 2026Steering at the Source: Style Modulation Heads for Robust Persona Controlcited by 4
  3. 2026Prefill-Time Intervention for Mitigating Hallucination in Large Vision-Language Modelscited by 3
  4. 2026Activation Steering for Aligned Open-ended Generation without Sacrificing Coherencecited by 2
  5. 2026Tethered Reasoning: Decoupling Entropy from Hallucination in Quantized LLMs via Manifold Steeringcited by 1
  6. 2026Steer Like the LLM: Activation Steering that Mimics Promptingcited by 1
  7. 2026Steering Interference Reflects the Model's Defaults, Not the Behavior Directionscited by 0
  8. 2026Replay-Gated Neural Execution: Decoupling Persistent Behavioral Specifications from Neural Realizations in Frozen Language Modelscited by 0