Agentic Chain-of-Thought Steering for Efficient and Controllable LLM Reasoning

arXiv:2606.03965 · cs.CL, cs.AI · Submitted 2026-06-02 · Read on arXiv

cs.CL, cs.AI

Submitted: 2026-06-02

Updated: 2026-08-30

Code: https://github.com/Andree-9/ACTS

Terminology

• Steering — 98× in this paper · explained in 2 episodes
• Trace — 46× in this paper · explained in 2 episodes
• Accuracy — 25× in this paper · explained in 1 episode
• Benchmark — 16× in this paper · explained in 2 episodes
• Reinforcement Learning — 14× in this paper · explained in 4 episodes
• Reward Shaping — 9× in this paper · explained in 2 episodes
• Reasoning Traces — 7× in this paper · explained in 3 episodes
• Agentic Chain-of-Thought Steering — 5× in this paper
• Verification — 4× in this paper · explained in 2 episodes
• GRPO — 3× in this paper · explained in 6 episodes
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Pruning — 2× in this paper · explained in 4 episodes
• Adaptive Reasoning — 1× in this paper · explained in 2 episodes
• Agentic Reasoning — 1× in this paper · explained in 1 episode
• Calibration — 1× in this paper · explained in 7 episodes
• Failure Modes — 1× in this paper · explained in 3 episodes
• Inference-Time Steering — 1× in this paper · explained in 2 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Markov Decision Process (MDP) — 1× in this paper · explained in 11 episodes
• Model-Agnostic — 1× in this paper · explained in 2 episodes
• Online Reinforcement Learning — 1× in this paper · explained in 2 episodes

Sources

• RL of Thoughts: Navigating LLM Reasoning with Inference-time Reinforcement Learning
• Neural Chain-of-Thought Search: Searching the Optimal Reasoning Path to Enhance Large Language Models
• SmartThinker: Learning to Compress and Preserve Reasoning by Step-Level Length Control
• Understanding R1-Zero-Like Training: A Critical Perspective
• OpenRLHF: An Easy-to-use, Scalable and High-performance RLHF Framework
• O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning
• FlashThink: An Early Exit Method For Efficient Reasoning
• Reasoning Models Can Be Effective Without Thinking
• Steering LLM Thinking with Budget Guidance
• Concise Thoughts: Impact of Output Length on LLM Reasoning and Cost
• Towards System 2 Reasoning in LLMs: Learning How to Think With Meta Chain-of-Thought
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Chain of Draft: Thinking Faster by Writing Less
• ThinkRouter: Efficient Reasoning via Routing Thinking between Latent and Discrete Spaces
• Meta-Reasoner: Dynamic Guidance for Optimized Inference-time Reasoning in Large Language Models
• Qwen3 Technical Report
• Understanding Reasoning in Thinking Language Models via Steering Vectors
• Understanding and Steering the Cognitive Behaviors of Reasoning Models at Test-Time
• BudgetThinker: Empowering Budget-aware LLM Reasoning with Control Tokens

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering