mlsanityData to 7 Oct 2026

Controlling Large Language Model Agents with Entropic Activation Steering

arXiv 2406.00244 ↗First posted 1 Jun 2024cs.CLcited by 29

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

The rise of large language models (LLMs) has prompted increasing interest in their use as in-context learning agents. At the core of agentic behavior is the capacity for exploration, or the ability to actively gather information about the environment. But how do LLM agents explore, and how can we control their exploratory behaviors? To answer these questions, we take a representation-level perspective, and introduce Entropic Activation Steering (EAST), an activation steering method for in-context LLM agents. Firstly, we demonstrate that EAST can effectively manipulate an LLM agent's exploration by directly affecting the high-level actions parsed from the outputs of the LLM, in contrast to token-level temperature sampling. Secondly, we reveal how applying this control modulates the uncertainty exhibited in the LLM's thoughts, guiding the agent towards more exploratory actions. Finally, we demonstrate that the steering vectors obtained by EAST generalize across task variants. In total, these results show that LLM agents explicitly encode uncertainty over their actions in their representation space. Our work paves the way for a new understanding of the functioning of LLM agents and to effective control of their decision-making behaviors.

Key facts. Cited by 29 papers in the map (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 32 of its 42 references that are in the map. First posted on arXiv 1 Jun 2024.

Builds on

32 of 42 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024In-context learning agents are asymmetric belief updaters6 Feb
  2. 2023Reason for Future, Act for Now: A Principled Framework for Autonomous LLM Agents with Provable Sample Efficiency29 Sep

Other references · 30

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
  4. 2013Efficient Estimation of Word Representations in Vector Spacecited by 4,346
  5. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,072
  6. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
  7. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,193
  8. 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,081
  9. 2024Mixtral of Expertscited by 1,785
  10. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,259
  11. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,239
  12. 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,081
18 more
  1. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,009
  2. 2023Steering Language Models With Activation Engineeringcited by 824
  3. 2023Eureka: Human-Level Reward Design via Coding Large Language Modelscited by 542
  4. 2022Using cognitive psychology to understand GPT-3cited by 389
  5. 2023In-Context Learning Creates Task Vectorscited by 302
  6. 2023Reward Design with Language Modelscited by 257
  7. 2022Extracting Latent Steering Vectors from Pretrained Language Modelscited by 217
  8. 2024ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RLcited by 212
  9. 2023Large Language Models as General Pattern Machinescited by 196
  10. 2024How to use and interpret activation patchingcited by 191
  11. 2024ReFT: Representation Finetuning for Language Modelscited by 186
  12. 2021Between words and characters: A Brief History of Open-Vocabulary Modeling and Tokenization in NLPcited by 130
  13. 2023Motif: Intrinsic Motivation from Artificial Intelligence Feedbackcited by 99
  14. 2024Can large language models explore in-context?cited by 75
  15. 2023Improving Activation Steering in Language Models with Mean-Centringcited by 64
  16. 2024Emergent World Models and Latent Variable Estimation in Chess-Playing Language Modelscited by 60
  17. 2024Do LLM Agents Have Regret? A Case Study in Online Learning and Gamescited by 33
  18. 2023Understanding and Controlling a Maze-Solving Policy Networkcited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 5; 2025: 12; 2026: 12.

29 papers.

Sort:

By area:

  1. 2024Programming Refusal with Conditional Activation Steeringcited by 191
  2. 2024Improving Instruction-Following in Language Models through Activation Steeringcited by 149
  3. 2024Controlling Language and Diffusion Models by Transporting Activationscited by 59
  4. 2024EVOLvE: Evaluating and Optimizing LLMs For In-Context Explorationcited by 33
  5. 2025SAE-SSV: Supervised Steering in Sparse Representation Spaces for Reliable Control of Language Modelscited by 28
  6. 2024LLMs Are In-Context Bandit Reinforcement Learnerscited by 27
  7. 2025Beyond Prompt Engineering: Robust Behavior Control in LLMs via Steering Target Atomscited by 23
  8. 2025Representation Engineering for Large-Language Models: Survey and Research Challengescited by 18
  9. 2025Learning When to Plan: Efficiently Allocating Test-Time Compute for LLM Agentscited by 18
  10. 2025SafeSteer: Interpretable Safety Steering with Refusal-Evasion in LLMscited by 9
  11. 2025Asymmetric Proximal Policy Optimization: mini-critics boost LLM reasoningcited by 9
  12. 2025Guiding Giants: Lightweight Controllers for Weighted Activation Steering in LLMscited by 8
17 more
  1. 2026Why Steering Works: Toward a Unified View of Language Model Parameter Dynamicscited by 8
  2. 2026Scaling In-Context Online Learning Capability of LLMs via Cross-Episode Meta-RLcited by 6
  3. 2026Fine-Grained Activation Steering: Steering Less, Achieving Morecited by 6
  4. 2026ASA: Backbone-Training-Free Representation Engineering for Tool-Calling Agentscited by 6
  5. 2025Post-Training LLMs as Better Decision-Making Agents: A Regret-Minimization Approachcited by 4
  6. 2026The Unreasonable Effectiveness of Text Embedding Interpolation for Continuous Image Steeringcited by 4
  7. 2025Preventing Rogue Agents Improves Multi-Agent Collaborationcited by 3
  8. 2025When Greedy Wins: Emergent Exploitation Bias in Meta-Bandit LLM Trainingcited by 3
  9. 2025DSO: Direct Steering Optimization for Bias Mitigationcited by 3
  10. 2025Let's Get You Hired: A Job Seeker's Perspective on Multi-Agent Recruitment Systems for Explaining Hiring Decisionscited by 2
  11. 2026Reasoning aligns language models to human cognitioncited by 2
  12. 2026A Component-Based Survey of Interactions between Large Language Models and Multi-Armed Banditscited by 1
  13. 2026DORA Explorer: Improving the Exploration Ability of LLMs Without Trainingcited by 0
  14. 2026Representational Curvature Modulates Behavioral Uncertainty in Large Language Modelscited by 0
  15. 2026Entropy-informed Decoding: Adaptive Information-Driven Branchingcited by 0
  16. 2026Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detectioncited by 0
  17. 2026Absorbed in Inertia: Activation Analysis for Computer-Use Agentscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Empowering Large Language Model Agents through Action Learning
  2. 2024Active Inference for Self-Organizing Multi-LLM Systems: A Bayesian Thermodynamic Approach to Adaptation
  3. 2024From Words to Actions: Unveiling the Theoretical Underpinnings of LLM-Driven Autonomous Systems
  4. 2025Expanding the Action Space of LLMs to Reason Beyond Language
  5. 2026Multi-Head Latent Control: A Unified Interface for LLM Agent Decision Making
  6. 2025From Language to Action: A Review of Large Language Models as Autonomous Agents and Tool Users
  7. 2026Latent Action Reparameterization for Efficient Agent Inference
  8. 2025ATLaS: Agent Tuning via Learning Critical Steps