Search for Truth from Reasoning: A Dynamic Representation Editing Framework for Steering LLM Trajectories
Quick Answer
The paper introduces DynaSteer, a dynamic Representation Editing framework that enhances LLM reasoning by effectively steering trajectories toward truth.
Quick Take
The paper introduces DynaSteer, a dynamic Representation Editing framework that enhances LLM reasoning by effectively steering trajectories toward truth. It identifies critical insights about truth encoding and proposes interventions based on uncertainty principles, achieving significant performance improvements on MATH benchmarks and out-of-domain coding tasks.
Key Points
- DynaSteer employs pattern clustering and Fisher-LDA for truth projection.
- It monitors lookahead entropy to selectively steer reasoning trajectories.
- The framework shows improved performance on MATH benchmarks.
- Experiments confirm DynaSteer's generalization to out-of-domain coding tasks.
- Insights reveal truth is entangled with latent reasoning patterns.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Current approaches to enhance Large Language Model (LLM) reasoning, such as Chain-of-Thought and "Wait" prompts, primarily encourage models to think more, yet often fail to guide them toward Truth. While Representation Editing (RepE) offers a intrinsic control, its application to dynamic reasoning trajectories remains underexplored. In this work, we bridge this gap by investigating the geometry of truth within unfolding reasoning chains. We uncover three critical insights: (1) Truth is encoded at the sentence level and is entangled with latent reasoning patterns; (2) Effective intervention follows an Uncertainty Principle and a Decay Effect, requiring localization to early, high-entropy forks; (3) Naive steering vectors suffer from noise, risking collateral damage to correct trajectories. Based on these findings, we propose DynaSteer, a dynamic RepE framework. DynaSteer employs pattern clustering to disentangle reasoning manifolds and utilizes Fisher-LDA to project purified truth. By dynamically monitoring lookahead entropy, it selectively steers and rolls back trajectories only when necessary. Comprehensive experimental results on several MATH benchmark verify the effectiveness of DynaSteer, and experiments on out-of-domain coding tasks further confirm its generalization ability. Our code is publicly available at this https URL.
| Comments: | Accepted by ICML'26 |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2606.28589 [cs.AI] |
| (or arXiv:2606.28589v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2606.28589 arXiv-issued DOI via DataCite |
Submission history
From: Weibin Liao [view email]
[v1]
Fri, 26 Jun 2026 20:33:55 UTC (3,082 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →The Verification Horizon: No Silver Bullet for Coding Agent Rewards
As coding agents evolve, verifying solutions becomes more challenging than generating them, necessitating a focus on scalable, faithful, and robust verification methods. The study reveals that no fixed reward function can sustain effectiveness as model capabilities advance, emphasizing the need for verification to evolve alongside solution generation.