Elamur Research⌕
☰
home / papers / 2026 / Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning

Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning

Paper · 2026 · Bo Li, Mingda Wang, Gexiang Fang et al.

Summary

Typical RAG pipelines fetch documents once, using the initial query, and then generate from that fixed context. This fails when information needs only become clear during step-by-step reasoning. Earlier adaptive methods rely on hand-crafted uncertainty triggers or on separate controllers and multi-stage pipelines. The authors propose GRIP (Generation-guided Retrieval with Information Planning), which turns retrieval decisions into ordinary decoding. The model emits special control tokens to request evidence, write a reformulated query, report partial progress, or finish. The retriever stays external; only the decisions about using it are generated. Training data come from QA datasets and are split into four supervision types, filtered with EM and CoverEM, using BM25 over Wikipedia. GPT-4o-mini produces the multi-hop trajectories. Supervised fine-tuning is followed by reinforcement learning that rewards both answer quality and control behavior. Across five QA benchmarks, GRIP outperforms powerful retrieval-augmented baselines and performs comparably to GPT-4o, even though it is a far smaller model. It also largely preserves general abilities, which makes retrieval timing and stopping learnable and interpretable.

Key points

  • Four control tokens drive the loop: [RETRIEVE], [INTERMEDIARY], [ANSWER], [SOLVED]. The [INTERMEDIARY]→[RETRIEVE] branch continues the evidence search, and [ANSWER]→[SOLVED] ends it. The default budget is three retrieval rounds, adjustable at inference time.
  • There are four supervision types: Type-α (direct answer, identified by consistent EM answers from the LLaMA3-8B-Instruct backbone), Type-β (retrieval needed), Type-γ (multi-hop planning with a new query), and Type-θ (answer completion).
  • A teacher-replacement study on Type-γ data shows that open-source teachers still yield the learned retrieval policy, so the gains do not hinge on proprietary distillation. A weaker, backbone-scale teacher (LLaMA3-8B-Instruct) lowers quality more.
  • General capability is mostly preserved. GRIP scores 65.73 vs 66.56 for Instruct on MMLU and 53.8 vs 54.6 on MBPP. The SFT-RAG baseline drops further, to 62.93 and 47.8.
  • On CNN/DailyMail summarization judged pairwise by GPT-4o, GRIP wins 51.0% and loses 11.5% against Instruct. It is strongly preferred over SFT-RAG (90.5% wins) and R1-Searcher (59.5% wins).

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 · China315 works · 43,267 citations🏢Hugging FaceNew York · United States14 works · 20,431 citations🎓Hebei University of TechnologyTianjin · China2 works · 8 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Vision and Pattern Recognition · 3,486 worksInformation Systems · 1,730 works

🧭 Topics 1Research subjects and languages the work is about.Information retrieval & search · 552 works

Cite

Cite as Bo Li, Mingda Wang, Gexiang Fang, Shikun Zhang and Wei Ye. 2026. Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning. 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.196

@inproceedings{li2026retrieval,
  title = {{Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning}},
  author = {Bo Li, Mingda Wang, Gexiang Fang, Shikun Zhang and Wei Ye},
  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.196},
  url = {https://doi.org/10.18653/v1/2026.acl-long.196}
}

References

45 references, 28 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2020. Measuring massive multitask language understanding. arXiv preprint arXiv:2009.03300.
  2. Karl Moritz Hermann, Tomas Kocisky, Edward Grefenstette, Lasse Espeholt, Will Kay, Mustafa Suleyman, and Phil Blunsom. 2015. Teaching machines to read and comprehend. Advances in neural information processing systems, 28.
  3. Xanh Ho, A. Nguyen, Saku Sugawara, and Akiko Aizawa. 2020. Constructing a multi-hop qa dataset for comprehensive evaluation of reasoning steps. ArXiv, abs/2011.01060.
  4. Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi. 2024. Self-rag: Learning to retrieve, generate, and critique through self-reflection.
  5. Yi Jiang, Sendong Zhao, Jianbo Li, Haochun Wang, and Bing Qin. 2025. Gainrag: Preference alignment in retrieval-augmented generation through gain signal synthesis. arXiv preprint arXiv:2505.18710. find
  6. Jacob Austin, Augustus Odena, Maxwell Nye, Maarten Bosma, Henryk Michalewski, David Dohan, Ellen Jiang, Carrie Cai, Michael Terry, Quoc Le, and 1 others. 2021. Program synthesis with large language models. arXiv preprint arXiv:2108.07732.
  7. Zhengbao Jiang, Frank F Xu, Luyu Gao, Zhiqing Sun, Qian Liu, Jane Dwivedi-Yu, Yiming Yang, Jamie Callan, and Graham Neubig. 2023. Active retrieval augmented generation. arXiv preprint arXiv:2305.06983.
  8. Jonathan Berant, Andrew K. Chou, Roy Frostig, and Percy Liang. 2013. Semantic parsing on freebase from question-answer pairs. In Conference on Empirical Methods in Natural Language Processing.
  9. Yapei Chang, Yekyung Kim, Michael Krumdick, Amir Zadeh, Chuan Li, Chris Tanner, and Mohit Iyyer. 2025. Bleuberi: Bleu is a surprisingly effective reward for instruction following. ArXiv, abs/2505.11080. find
  10. Wenqi Fan, Yujuan Ding, Liangbo Ning, Shijie Wang, Hengyun Li, Dawei Yin, Tat-Seng Chua, and Qing Li. 2024. A survey on rag meeting llms: Towards retrieval-augmented large language models. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 6491– 6501.
  11. Bowen Jin, Hansi Zeng, Zhenrui Yue, Dong Wang, Hamed Zamani, and Jiawei Han. 2025. Search-r1: Training llms to reason and leverage search engines with reinforcement learning. ArXiv, abs/2503.09516.
  12. Mandar Joshi, Eunsol Choi, Daniel S. Weld, and Luke Zettlemoyer. 2017. Triviaqa: A large scale distantly supervised challenge dataset for reading comprehension. ArXiv, abs/1705.03551.
  13. Nikhil Kandpal, Haikang Deng, Adam Roberts, Eric Wallace, and Colin Raffel. 2023. Large language models struggle to learn long-tail knowledge. In International Conference on Machine Learning, pages 15696–15707. PMLR.
  14. Feiteng Fang, Yuelin Bai, Shiwen Ni, Min Yang, Xiaojun Chen, and Ruifeng Xu. 2024. Enhancing noise robustness of retrieval-augmented language models with adaptive adversarial training. In Annual Meeting of the Association for Computational Linguistics.
  15. Tom Kwiatkowski, Jennimaria Palomaki, Olivia Redfield, Michael Collins, Ankur P. Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, Kristina Toutanova, Llion Jones, Matthew Kelcey, Ming-Wei Chang, Andrew M. Dai, Jakob Uszkoreit, Quoc V. Le, and Slav Petrov. 2019. Na… — Natural Questions: A Benchmark for Question Answering Research
  16. Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, and Haofen Wang. 2023. Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997.
  17. 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 North American Chapter of the Association for Computational Linguistics. find
  18. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, and 1 others. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783.
  19. Bo Li, Tian Tian, Zhenghua Xu, Hao Cheng, Shikun Zhang, and Wei Ye. 2026a. Modeling uncertainty trends for timely retrieval in dynamic rag. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 40, pages 31527–31535. find
  20. Bo Li, Zhenghua Xu, and Rui Xie. 2026b. Language drift in multilingual retrieval-augmented generation: Characterization and decoding-time mitigation. In Fortieth AAAI Conference on Artificial Intelligence, Thirty-Eighth Conference on Innovative Applications of Artificial Intelligence, Sixteenth Symp… find
  21. Kun Luo, Zheng Liu, Shitao Xiao, Tong Zhou, Yubo Chen, Jun Zhao, and Kang Liu. 2024. Landmark embedding: A chunking-free embedding method for retrieval augmented long-context large language models. In Annual Meeting of the Association for Computational Linguistics. find
  22. Alex Troy Mallen, Akari Asai, Victor Zhong, Rajarshi Das, Hannaneh Hajishirzi, and Daniel Khashabi. 2022. When not to trust language models: Investigating effectiveness of parametric and non-parametric memories. In Annual Meeting of the Association for Computational Linguistics. find
  23. Ori Ram, Yoav Levine, Itay Dalmedigos, Dor Muhlgay, Amnon Shashua, Kevin Leyton-Brown, and Yoav Shoham. 2023. In-context retrieval-augmented language models. Transactions of the Association for Computational Linguistics, 11:1316–1331.
  24. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Jun-Mei Song, Mingchuan Zhang, Y. K. Li, Yu Wu, and Daya Guo. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. ArXiv, abs/2402.03300.
  25. Weijia Shi, Sewon Min, Michihiro Yasunaga, Minjoon Seo, Rich James, Mike Lewis, Luke Zettlemoyer, and Wen tau Yih. 2023. Replug: Retrieval-augmented black-box language models. In North American Chapter of the Association for Computational Linguistics.
  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, abs/2503.05592.
  27. Weihang Su, Yichen Tang, Qingyao Ai, Zhijing Wu, and Yiqun Liu. 2024. Dragin: Dynamic retrieval augmented generation based on the real-time information needs of large language models. arXiv preprint arXiv:2403.10081.
  28. 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. ArXiv, abs/2505.04588.
  29. Liang Wang, Nan Yang, and Furu Wei. 2023. Query2doc: Query expansion with large language models. In Conference on Empirical Methods in Natural Language Processing.
  30. Zhepei Wei, Wei-Lin Chen, and Yu Meng. 2024. Instructrag: Instructing retrieval-augmented generation via self-synthesized rationales. arXiv preprint arXiv:2406.13629. — [arXiv:2406.13629](https://arxiv.org/abs/2406.13629) find
  31. Haoyi Xiong, Jiang Bian, Yuchen Li, Xuhong Li, Mengnan Du, Shuaiqiang Wang, Dawei Yin, and Sumi Helal. 2024. When search engine services meet large language models: visions and challenges. IEEE Transactions on Services Computing. find
  32. Shicheng Xu, Liang Pang, Mo Yu, Fandong Meng, Huawei Shen, Xueqi Cheng, and Jie Zhou. 2024. Unsupervised information refinement training of large language models for retrieval-augmented generation. ArXiv, abs/2402.18150. find
  33. Shijia Xu, Zhou Wu, Xiaolong Jia, Yu Wang, Kai Liu, and April Xiaowen Dong. 2026. Self-correcting rag: Enhancing faithfulness via mmkp context selection and nli-guided mcts. Preprint, arXiv:2604.10734.
  34. Shi-Qi Yan and Zhen-Hua Ling. 2025. Rpo: Retrieval preference optimization for robust retrieval-augmented generation. ArXiv, abs/2501.13726. find
  35. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, and 41 others. 2025a. Qwen3 technical report. arXiv preprint arXiv:2505.09388.
  36. 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, Junyang Lin, Kai Dang, and 22 others. 2024. Qwen2.5 technical report. arXiv preprint arXi… find
  37. Yuming Yang, Jiang Zhong, Li Jin, Jingwang Huang, Jingpeng Gao, Qing Liu, Yang Bai, Jingyuan Zhang, Rui Jiang, and Kaiwen Wei. 2025b. Benchmarking multimodal rag through a chart-based document question-answering generation framework. Preprint, arXiv:2502.14864. — [arXiv:2502.14864](https://arxiv.org/abs/2502.14864) find
  38. Fuda Ye, Shuangyin Li, Yongqi Zhang, and Lei Chen. 2024. R2ag: Incorporating retrieval information into retrieval augmented generation. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 11584–11596. find
  39. Ori Yoran, Tomer Wolfson, Ori Ram, and Jonathan Berant. Making retrieval-augmented language models robust to irrelevant context. In ICLR 2024 Workshop on Large Language Model (LLM) Agents.
  40. Qiying Yu, Zheng Zhang, Ruofei Zhu, Yufeng Yuan, Xiaochen Zuo, Yu Yue, Tiantian Fan, Gaohong Liu, Lingjun Liu, Xin Liu, Haibin Lin, Zhiqi Lin, Bole Ma, Guangming Sheng, Yuxuan Tong, Chi Zhang, Mofan Zhang, Wang Zhang, Hang Zhu, and 16 others. 2025. Dapo: An open-source llm reinforcement learning sys…
  41. W. Yu, Zhihan Zhang, Zhenwen Liang, Meng Jiang, and Ashish Sabharwal. 2023. Improving language models via plug-and-play retrieval feedback. ArXiv, abs/2305.14002.
  42. Qianchi Zhang, Hainan Zhang, Liang Pang, Yongxin Tong, Hongwei Zheng, and Zhiming Zheng. 2025. Less is more: Compact clue selection for efficient retrieval-augmented generation reasoning. arXiv preprint arXiv:2502.11811. — [arXiv:2502.11811](https://arxiv.org/abs/2502.11811) find
  43. Qianchi Zhang, Hainan Zhang, Liang Pang, Hongwei Zheng, and Zhiming Zheng. 2026. Stable-rag: Mitigating retrieval-permutation-induced hallucinations in retrieval-augmented generation. arXiv preprint arXiv:2601.02993.
  44. Xiaofan Zhou, Huy Nguyen, Bo Yu, Chenxi Liu, and Lu Cheng. 2026. Adaptive stopping for multi-turn llm reasoning. arXiv preprint arXiv:2604.01413. — [arXiv:2604.01413](https://arxiv.org/abs/2604.01413) find
  45. Wei Zhu, Zhiwen Tang, and Kun Yue. 2026. Symphony: Synergistic multi-agent planning with heterogeneous language model assembly. arXiv preprint arXiv:2601.22623. — [arXiv:2601.22623](https://arxiv.org/abs/2601.22623) find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/li-2026-retrieval-as-generation-unified-framework/