EvoThink is proposed, a framework that reduces redundant verification and encourages the exploration of new reasoning paths that not only substantially reduces inference-time token usage but also improves the reasoning capability of LRMs.
Abstract
Large Reasoning Models (LRMs) often suffer from overthinking due to redundant verification steps. Existing approaches for mitigating overthinking, such as fast-slow thinking switching and reasoning trajectory compression, fail to make a fine-grained distinction between beneficial and redundant steps within the LRM's reasoning process, and may thus impair reasoning capability in their pursuit of efficiency. To simultaneously improve reasoning efficiency and capability, we propose EvoThink, a framework that reduces redundant verification and encourages the exploration of new reasoning paths. EvoThink comprises two key components: Self-Pruning Training (SPT), an unsupervised method that iteratively prunes redundant reasoning steps and self-trains on the concise trajectories; and Aha-Moment Preference Optimization (AMPO), which, inspired by genetic algorithms, identifies valuable failed reasoning attempts, synthesizes from-wrong-to-right aha-moment data, and optimizes the model to internalize this reasoning pattern. Extensive evaluations across mathematical reasoning and code generation benchmarks demonstrate that EvoThink not only substantially reduces inference-time token usage but also improves the reasoning capability of LRMs.
This work proposes ChainPrune, a novel reasoning path semantic structural optimization method to efficiently and controllably synthesize self-generated high-quality training data and incorporates a DPO-based preference learning method combined with supervised loss, effectively mitigating false reward suppression.
Weihang Pan, Zhengxu Yu, Yuxiang Zhang et al.· 1 citation
OS-Pruner is a lightweight plug-in framework that formulates chain-of-thought pruning as an optimal stopping problem that achieves 20-60\% reduction in generation length with minimal accuracy sacrifice on diverse reasoning benchmarks and base models.
Mohammed Ehab, Aymane El Gadarri, Vivek F. Farias et al.· arXiv.org· 0 citations
This work conducts a fine-grained analysis of hallucinations arising in LLM reasoning and finds that the reasoning traces are particularly prone to Context-Sensitive Factual Hallucinations: cases where the model actually has the relevant knowledge, yet makes factual errors due to contextual interference during reasoning.
While Knowledge Graphs (KGs) can ground Large Language Models (LLMs) in factual knowledge, existing LLM-KG integration methods for complex reasoning are plagued by computational inefficiency and semantic inconsistency. The tight coupling of LLM inference and KG traversal leads to prohibitive costs, while spurious reasoning paths often misguide learning-based agents, causing reward hacking. To this end, we propose EAT (Exploration-and-Thinking), a novel agentic framework that synergizes LLMs with Reinforcement Learning (RL) for efficient and faithful reasoning on KGs. EAT's core innovations are twofold: (1) an adaptive retrieval-augmented generation mechanism that decouples language comprehension from structured exploration, dramatically improving efficiency; and (2) an LLM-guided reward shaping strategy that explicitly penalizes semantically inconsistent paths and promotes logically valid trajectories grounded in the KG. Extensive experiments on benchmarks like WebQSP, CWQ, and GrailQA show that EAT achieves state-of-the-art performance. Notably, it surpasses the reasoning capability of GPT-4o while utilizing a significantly smaller 8B-parameter LLM.
Yi Xia, Gang Zhou, Jing Chen et al.· Annual International ACM SIG...· 0 citations
Recent methods using Reinforcement Learning (RL) have improved Tool-Integrated Reasoning (TIR) by training large language models to learn end-to-end policies for multi-step tool usage, enabling them to solve complex tasks more effectively. Despite these advances, existing methods often suffer from overthinking at both the action and reasoning levels: models tend to invoke tools redundantly and generate excessively long reasoning trajectories, resulting in high computational cost. To address this, in this paper, we propose LightTIR, a dual-penalty reward framework, to achieve efficient TIR. For action efficiency, LightTIR estimates the marginal utility of each tool call through prefix-aligned counterfactual trajectories, encouraging calls that contribute meaningful information while penalizing low-utility or redundant invocations. For reasoning efficiency, LightTIR introduces a length-aware regularization term, adaptively penalizing intermediate reasoning steps that exceed the minimal effective trajectory required for correct prediction. Extensive experiments demonstrate that LightTIR can reduce redundancy and trajectory expansion while maintaining answer correctness, achieving more efficient RL-based TIR. Code is available at https://github.com/ekventitas/LightTIR.
Yichen Xiao, Siyu Gong, Linan Yue· Proceedings of the 32nd ACM...· 0 citations
Inference-time search with large language models (LLMs) often concentrates on a small set of structurally or semantically similar trajectories, leaving alternatives underexplored---a failure mode we call \textit{reasoning basin collapse}. We introduce BASIN, a training-free, structure-aware selection method that groups reasoning states into basins and penalizes repeated visits to the same strategy, thereby reallocating search across genuinely distinct reasoning paths under a fixed compute budget. Under matched inference budgets, BASIN improves over Tree of Thoughts (ToT) by up to $+22$pp on Game of 24 and $+6.7$pp on MuSR. A quality-aware variant, QA-BASIN, further improves robustness by preserving high-quality basins when unconditional diversification over-explores. To explain when basin-aware selection helps, we introduce the redundancy gap $\Delta$, which measures how differently search concentrates for correct versus incorrect predictions: standard ToT often operates near $\Delta \approx 0$, while BASIN consistently shifts $\Delta$ positive. More broadly, BASIN suggests structure-aware selection as a simple and general approach to improving inference-time reasoning. Code can be found at https://github.com/GitHubLuCheng/basin.
Lu Cheng· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
