Search for Truth from Reasoning: A Dynamic Representation Editing Framework for Steering LLM Trajectories

Researchers introduced DynaSteer, a dynamic representation editing framework that steers large language model reasoning trajectories toward truth by leveraging geometric insights into truth encoding and entropy-based intervention. The method outperforms existing approaches on math benchmarks and generalizes to coding tasks, with code publicly available.

arXiv:2606.28589v1 Announce Type: new Abstract: 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 https://github.com/tianlwang/DynaSteer.