Elamur Research⌕
☰
home / papers / 2026 / D²Plan: Dual-Agent Dynamic Global Planning for Complex Retrieval-Augmented Reasoning

D²Plan: Dual-Agent Dynamic Global Planning for Complex Retrieval-Augmented Reasoning

Paper · 2026 · Kangcheng Luo, Tinglang Wu and Yansong Feng

Summary

Search-augmented LLMs trained with RL can alternate between retrieving and reasoning, but on multi-hop questions their context fills up with retrieved text, much of it irrelevant. The authors identify two resulting failure types. In Ineffective Search Chain Construction, queries drift off target or needed lookups are skipped. In Reasoning Hijacking by Peripheral Evidence, the model treats distractors as evidence. D$^2$Plan addresses both by splitting the work between two agents. A Reasoner writes an explicit plan of ordered sub-questions and revises it as retrieval feedback arrives. A Purifier judges whether retrieved documents are relevant and passes along only a condensed version. Training has two stages:

  • SFT cold-start: both agents learn from teacher-synthesized trajectories, filtered for correct answers and at least one retrieval call.
  • SPlanRL: the Reasoner is trained with plan-oriented rewards for decomposition and plan adaptation, while the Purifier stays frozen.

Across six QA benchmarks, the method beats the strongest baseline by 3.8 points on average. It also lowers error rates and shortens inference, which shows that structured planning combined with evidence filtering makes retrieval-augmented reasoning more robust.

Key points

  • Failure analysis: E1 (search chain errors) accounts for 29% of failures in state-of-the-art methods and E2 (peripheral-evidence hijacking) for 63%.
  • Models and retrieval: experiments use Qwen2.5-3B and Qwen2.5-7B with a local dense retriever over a static Wikipedia corpus.
  • Main result: an average absolute gain of 3.8% over the best baseline across six datasets.
  • Efficiency: on FRAMES, using a Qwen2.5-7B-SFT Purifier cuts total inference time from 432s to 381s (about 12%), because the Reasoner works with a shorter context.
  • Error analysis:
    • Setup: 200 questions each from FRAMES and MuSiQue, labeled by gemini-3-pro-preview, which agreed with manual labels at Cohen's Kappa 0.72.
    • Comparison: D$^2$Plan reduces E1 and E2.1 errors relative to Search-R1 and StepSearch.
    • Remaining errors: in 14 of 20 inspected E2.2 cases, the gold documents were absent from the corpus.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Peking UniversityBeijing · China244 works · 39,359 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksComputer Vision and Pattern Recognition · 2,869 worksInformation Systems · 1,437 works

🧭 Topics 2Research subjects and languages the work is about.Reasoning · 1,426 worksInformation retrieval & search · 473 works

Cite

Cite as Kangcheng Luo, Tinglang Wu and Yansong Feng. 2026. D²Plan: Dual-Agent Dynamic Global Planning for Complex Retrieval-Augmented Reasoning. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.216

@inproceedings{luo2026dual,
  title = {{D²Plan: Dual-Agent Dynamic Global Planning for Complex Retrieval-Augmented Reasoning}},
  author = {Kangcheng Luo, Tinglang Wu and Yansong Feng},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.216},
  url = {https://doi.org/10.18653/v1/2026.acl-long.216}
}

References

33 references, 17 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mingyang Chen, Tianpeng Li, Haoze Sun, Yijie Zhou, Chenzheng Zhu, Haofen Wang, Jeff Z Pan, Wen Zhang, Huajun Chen, Fan Yang, and 1 others. 2025. Learning to reason with search for llms via reinforcement learning. arXiv preprint arXiv:2503.19470. find
  2. Tianzhe Chu, Yuexiang Zhai, Jihan Yang, Shengbang Tong, Saining Xie, Dale Schuurmans, Quoc V Le, Sergey Levine, and Yi Ma. 2025. Sft memorizes, rl generalizes: A comparative study of foundation model post-training. arXiv preprint arXiv:2501.17161.
  3. Yixiong Fang, Tianran Sun, Yuling Shi, and Xiaodong Gu. 2025. Attentionrag: Attention-guided context pruning in retrieval-augmented generation. arXiv preprint arXiv:2503.10720. — [arXiv:2503.10720](https://arxiv.org/abs/2503.10720) find
  4. Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yixin Dai, Jiawei Sun, Haofen Wang, and Haofen Wang. 2023. Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997, 2(1).
  5. Xanh Ho, Anh-Khoa Duong Nguyen, Saku Sugawara, and Akiko Aizawa. 2020. Constructing a multi-hop qa dataset for comprehensive evaluation of reasoning steps. arXiv preprint arXiv:2011.01060.
  6. Cheng-Yu Hsieh, Yung-Sung Chuang, Chun-Liang Li, Zifeng Wang, Long Le, Abhishek Kumar, James Glass, Alexander Ratner, Chen-Yu Lee, Ranjay Krishna, and 1 others. 2024. Found in the middle: Calibrating positional attention bias improves long context utilization. In Findings of the Association for Comp… find
  7. Zhouyu Jiang, Mengshu Sun, Lei Liang, and Zhiqiang Zhang. 2025. Retrieve, summarize, plan: Advancing multi-hop question answering with an iterative approach. In Companion Proceedings of the ACM on Web Conference 2025, pages 1677–1686. find
  8. Bowen Jin, Jinsung Yoon, Priyanka Kargupta, Sercan O Arik, and Jiawei Han. 2025a. An empirical study on reinforcement learning for reasoning-search interleaved llm agents. arXiv preprint arXiv:2505.15117. — [arXiv:2505.15117](https://arxiv.org/abs/2505.15117) find
  9. Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. 2025b. Search-r1: Training llms to reason and leverage search engines with reinforcement learning. arXiv preprint arXiv:2503.09516.
  10. Jiajie Jin, Yutao Zhu, Xinyu Yang, Chenghao Zhang, and Zhicheng Dou. 2024. Flashrag: A modular toolkit for efficient retrieval-augmented generation research. CoRR, abs/2405.13576. find
  11. Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick SH Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense passage retrieval for open-domain question answering. In EMNLP (1), pages 6769–6781.
  12. Satyapriya Krishna, Kalpesh Krishna, Anhad Mohananey, Steven Schwarcz, Adam Stambler, Shyam Upadhyay, and Manaal Faruqui. 2025. Fact, fetch, and reason: A unified evaluation of retrieval-augmented generation. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Associa… find
  13. Tom Kwiatkowski, Jennimaria Palomaki, Olivia Redfield, Michael Collins, Ankur Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, and 1 others. 2019. Natural questions: a benchmark for question answering research. Transactions of the Association for Computational Lin…
  14. Myeonghwa Lee, Seonho An, and Min-Soo Kim. 2024. Planrag: A plan-then-retrieval augmented generation for generative large language models as decision makers. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technolog… find
  15. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, and 1 others. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems, 33:9459–…
  16. Xiaoxi Li, Guanting Dong, Jiajie Jin, Yuyao Zhang, Yujia Zhou, Yutao Zhu, Peitian Zhang, and Zhicheng Dou. 2025. Search-o1: Agentic search-enhanced large reasoning models. arXiv preprint arXiv:2501.05366.
  17. Minhua Lin, Zongyu Wu, Zhichao Xu, Hui Liu, Xianfeng Tang, Qi He, Charu Aggarwal, Xiang Zhang, and Suhang Wang. 2025. A comprehensive survey on reinforcement learning-based agentic search: Foundations, roles, optimizations, evaluations, and applications. arXiv preprint arXiv:2510.16724. — [arXiv:2510.16724](https://arxiv.org/abs/2510.16724) find
  18. Yanming Liu, Xinyue Peng, Xuhong Zhang, Weihao Liu, Jianwei Yin, Jiannan Cao, and Tianyu Du. 2024. Ra-isf: Learning to answer and understand from retrieval augmentation via iterative self-feedback. arXiv preprint arXiv:2403.06840. find
  19. Jinchang Luo, Mingquan Cheng, Fan Wan, Ni Li, Xiaoling Xia, Shuangshuang Tian, Tingcheng Bian, Haiwei Wang, Haohuan Fu, and Yan Tao. 2025a. Globalrag: Enhancing global reasoning in multi-hop question answering via reinforcement learning. arXiv preprint arXiv:2510.20548. — [arXiv:2510.20548](https://arxiv.org/abs/2510.20548) find
  20. Kun Luo, Hongjin Qian, Zheng Liu, Ziyi Xia, Shitao Xiao, Siqi Bao, Jun Zhao, and Kang Liu. 2025b. Infoflow: Reinforcing search agent via reward density optimization. arXiv preprint arXiv:2510.26575. — [arXiv:2510.26575](https://arxiv.org/abs/2510.26575) find
  21. Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, and 25 others. 2025. Qwen2.5 technical report. Preprint, arXiv:2412.15115. find
  22. Zhihong Shao, Yeyun Gong, Yelong Shen, Minlie Huang, Nan Duan, and Weizhu Chen. 2023. Enhancing retrieval-augmented large language models with iterative retrieval-generation synergy. Preprint, arXiv:2305.15294.
  23. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, and 1 others. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300.
  24. Guangming Sheng, Chi Zhang, Zilingfeng Ye, Xibin Wu, Wang Zhang, Ru Zhang, Yanghua Peng, Haibin Lin, and Chuan Wu. 2025. Hybridflow: A flexible and efficient rlhf framework. In Proceedings of the Twentieth European Conference on Computer Systems, pages 1279–1297.
  25. Yaorui Shi, Shihan Li, Chang Wu, Zhiyuan Liu, Junfeng Fang, Hengxing Cai, An Zhang, and Xiang Wang. 2025. Search and refine during think: Autonomous retrieval-augmented reasoning of llms. arXiv eprints, pages arXiv–2505. find
  26. Huatong Song, Jinhao Jiang, Yingqian Min, Jie Chen, Zhipeng Chen, Wayne Xin Zhao, Lei Fang, and Ji-Rong Wen. 2025. R1-searcher: Incentivizing the search capability in llms via reinforcement learning. arXiv preprint arXiv:2503.05592.
  27. Hao Sun, Zile Qiao, Jiayan Guo, Xuanbo Fan, Yingyan Hou, Yong Jiang, Pengjun Xie, Yan Zhang, Fei Huang, and Jingren Zhou. 2025. Zerosearch: Incentivize the search capability of llms without searching. Preprint, arXiv:2505.04588.
  28. Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M. Dai, Anja Hauth, Katie Millican, David Silver, Melvin Johnson, Ioannis Antonoglou, Julian Schrittwieser, Amelia Glaese, Jilin Chen, Emily Pitler, Timothy Lillicrap, Angeliki Lazari… — Gemini: A Family of Highly Capable Multimodal Models
  29. Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022a. Interleaving retrieval with chain-of-thought reasoning for knowledge-intensive multi-step questions. arXiv preprint arXiv:2212.10509.
  30. Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022b. Musique: Multi-hop questions via single-hop question composition. Transactions of the Association for Computational Linguistics, 10:539–554. find
  31. Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. Text embeddings by weakly-supervised contrastive pre-training. arXiv preprint arXiv:2212.03533.
  32. Ziliang Wang, Kang An, Xuhui Zheng, Faqiang Qian, Weikun Zhang, Cijun Ouyang, Jialu Cai, Yuhang Wang, and Yichao Wu. 2025. Erase to improve: Erasable reinforcement learning for search-augmented llms. arXiv preprint arXiv:2510.00861. — [arXiv:2510.00861](https://arxiv.org/abs/2510.00861) find
  33. Jason Wei, Nguyen Karina, Hyung Won Chung, Yunxin Joy Jiao, Spencer Papay, Amelia Glaese, John Schulman, and William Fedus. 2024. Measuring short-form factuality in large language models. Preprint, arXiv:2411.04368. 62nd Annual Meeting of the Association for Computational Linguistics (Volume 3: Syst…

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/luo-2026-d-2-plan-dual-agent/