ACL Findings2026
Graph-Based Chain-of-Thought Pruning for Reducing Redundant Reflections in Reasoning LLMs
Hongyuan Yuan, Xinran He, Run Shao, Bolei He, Xianwei Xue, Mengke Chen, Qiutong Pan, Haiwei Wang, Haifeng Li
Abstract
Extending CoT through RL has been widely used to enhance the reasoning capabilities of LLMs. However, due to the sparsity of reward signals, it can also induce undesirable thinking patterns such as overthinking, i.e., generating redundant intermediate reasoning content. In this work, we argue that a major source of such redundancy is inefficient reflection, which often manifests in two problematic patterns: Indiscriminate Reflection, where the model performs broad, low-impact checks throughout reasoning, and Repetitive Reflection, where it repeatedly re-verifies an already established conclusion. To address this, we introduce a graph-based CoT optimization framework. Specifically, we convert each linear CoT into a directed acyclic graph (DAG) with explicit dependency edges, and design a dual pruning strategy: branch-level pruning removes weakly contributing reflection branches, while depth-level pruning eliminates late-stage reverification. We distill this behavior via a threestage pipeline: (1) SFT to initialize the policy on pruned concise traces, (2) DPO to prefer correct but less redundant trajectories, and (3) GRPO with length penalty to jointly optimize answer correctness and efficiency. Experiments show that our approach reduces the average reasoning tokens by 42% while maintaining or improving accuracy.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
1 top-tier paper has cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Large Language Models are Zero-Shot Reasoners cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
- Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs cites DAPO: An Open-Source LLM Reinforcement Learning System at Scale
- Training Language Models to Reason Efficiently cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Cited by top-tier papers1
Builds on30
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
- Large Language Models are Zero-Shot ReasonersTakeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo et al.NeurIPS 2022 · 8,168 citations
- DAPO: An Open-Source LLM Reinforcement Learning System at ScaleQiying Yu, Zheng Zhang, Ruofei Zhu, Yufeng Yuan et al.NeurIPS 2025 · 2,828 citations
- Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMsXumeng Wen, Zihan Liu, Shun Zheng, Shengyu Ye et al.ICLR 2026 · 279 citations
- Training Language Models to Reason EfficientlyDaman Arora, Andrea ZanetteNeurIPS 2025 · 270 citations
Related papers
- Your Reasoning Model Knows What Counts: Self-Guided Chain-of-Thought Pruning for Efficient ReasoningZi-Ao Ma, Xian-Ling Mao, Tian Lan, Chen Xu et al.ACL 2026
- On the Bias of Next-Token Predictors Toward Systematically Inefficient Reasoning: A Shortest-Path Case StudyRiccardo Alberghi, Elizaveta Demyanenko, Luca Biggio, Luca SagliettiNeurIPS 2025 · 2 citations
- Making Slow Thinking Faster: Compressing LLM Chain-of-Thought via Step EntropyZeju Li, Jianyuan Zhong, Ziyang Zheng, Xiangyu Wen et al.ICLR 2026 · 35 citations
- Efficient Reasoning for Large Reasoning Language Models via Certainty-Guided Reflection SuppressionJiameng Huang, Baijiong Lin, Guhao Feng, Jierun Chen et al.AAAI 2026 · 21 citations
- GoT-R1: Internalizing Graph-of-Thought via Structural Reinforcement for High-Density ReasoningZuchao Li, Qiwei Li, Yao Yao, Hai Zhao et al.ACL Findings 2026
