Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat
Sign Up
Cited by 35

The DOI matches its Crossref record; 2 of the 2 references it cites were found in a registry or an index.

DeepResearcher: Scaling Deep Research via Reinforcement Learning in Real-world Environments

Zheng, Yuxiang, Fu, Dayuan, Hu, Xiangkun, Cai, Xiaojie, Ye, Lyumanshan, Lu, Pengrui, Liu, Pengfei·2025·Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
DOI 10.18653/v1/2025.emnlp-main.22
Open this paper in CoChat — free account

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.18653/v1/2025.emnlp-main.22

Cites

2 of 2 references verified0 citation errors · 0 not checkable

Statistics

Figures

Image 1
CC-BY · From the PDF
Figure 2 Training dynamics of F1, turns and response length. In this figure, we find the performance gradually scaling with reinforcement learning. The numbers of tool calls and responses also increase.
Figure 2 Training dynamics of F1, turns and response length. In this figure, we find the performance gradually scaling with reinforcement learning. The numbers of tool calls and responses also increase.CC-BY · From the PDF
Figure 3 Planning and Cross Validation Behaviors. Left : DeepResearcher is capable of making plans and dynamically adjusting it throughout the reasoning process. Notably, the model can merge steps when appropriate. Right : DeepResearcher identifies the correct answer during its first tool call. However, rather than immediately committing to this result, it proceeds to verify its accuracy through subsequent steps. This rigorous behavior enhances the reliability of the model's responses, ensuring greater robustness in its final answers. Note: In the actual model, the tool call and tool response format is a JSON string, rather than a tag. However, for clarity and ease of presentation, we have used tags in the figure to represent the output.
Figure 3 Planning and Cross Validation Behaviors. Left : DeepResearcher is capable of making plans and dynamically adjusting it throughout the reasoning process. Notably, the model can merge steps when appropriate. Right : DeepResearcher identifies the correct answer during its first tool call. However, rather than immediately committing to this result, it proceeds to verify its accuracy through subsequent steps. This rigorous behavior enhances the reliability of the model's responses, ensuring greater robustness in its final answers. Note: In the actual model, the tool call and tool response format is a JSON string, rather than a tag. However, for clarity and ease of presentation, we have used tags in the figure to represent the output.CC-BY · From the PDF
Figure 4 Reflection and Honesty Behavior. The search and browse are 2 apis in json format in the real inference stage. Left: When the retrieved information does not fully align with the question, DeepResearcher recognizes this discrepancy based on environmental feedback and refines its search query in subsequent tool calls. This proves its reflection ability. Right: DeepResearcher is capable of recognizing when it has not found the correct answer and appropriately declines to provide a response to be honesty. Note: In the actual model, the tool call and tool response format is a JSON string, rather than a tag. However, for clarity and ease of presentation, we have used tags in the figure to represent the output.
Figure 4 Reflection and Honesty Behavior. The search and browse are 2 apis in json format in the real inference stage. Left: When the retrieved information does not fully align with the question, DeepResearcher recognizes this discrepancy based on environmental feedback and refines its search query in subsequent tool calls. This proves its reflection ability. Right: DeepResearcher is capable of recognizing when it has not found the correct answer and appropriately declines to provide a response to be honesty. Note: In the actual model, the tool call and tool response format is a JSON string, rather than a tag. However, for clarity and ease of presentation, we have used tags in the figure to represent the output.CC-BY · From the PDF
Image 5
CC-BY · From the PDF
Figure 5 F1 score during training
Figure 5 F1 score during trainingCC-BY · From the PDF
Figure 6 DeepResearcher performs the best on all 7 datasets measured by reliable model-based evaluation.
Figure 6 DeepResearcher performs the best on all 7 datasets measured by reliable model-based evaluation.CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 31

  1. Salaheddin Alzubi, Creston Brooks, Purva Chiniya, Edoardo Contente, Chiara von Gerlach, Lucas Irwin, Yihan Jiang, Arda Kaz, Windsor Nguyen, Sewoong Oh, and 1 others. 2025. Open deep search: Democ- ratizing search with open-source reasoning agents. arXiv preprint arXiv:2503.20201. · arXiv
  2. Mingyang Chen, Tianpeng Li, Haoze Sun, Yijie Zhou, Chenzheng Zhu, Haofen Wang, Jeff Z. Pan, Wen Zhang, Huajun Chen, Fan Yang, Zenan Zhou, and Weipeng Chen. 2025. Research: Learning to rea- son with search for llms via reinforcement learning. Preprint, arXiv:2503.19470. · arXiv
  3. Wenfeng Feng, Chuzhan Hao, Yuewei Zhang, Jingyi Song, and Hao Wang. 2025. Airrag: Activat- ing intrinsic reasoning for retrieval augmented gen- eration via tree-based search. arXiv preprint arXiv:2501.10053. · arXiv
  4. Dayuan Fu, Keqing He, Yejie Wang, Wentao Hong, Zhuoma Gongque, Weihao Zeng, Wei Wang, Jin- gang Wang, Xunliang Cai, and Weiran Xu. 2025. Agentrefine: Enhancing agent generalization through refinement tuning. arXiv preprint arXiv:2501.01702. · arXiv
  5. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shi- rong Ma, Peiyi Wang, Xiao Bi, and 1 others. 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948. · arXiv
  6. Xanh Ho, Anh-Khoa Duong Nguyen, Saku Sugawara, and Akiko Aizawa. 2020. Constructing a multi- hop QA dataset for comprehensive evaluation of reasoning steps. In Proceedings of the 28th Inter- national Conference on Computational Linguistics, pages 6609-6625, Barcelona, Spain (Online). Inter- national Committee on Computational Linguistics.
  7. Sirui Hong, Mingchen Zhuge, Jonathan Chen, Xiawu Zheng, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, Chenyu Ran, Lingfeng Xiao, Chenglin Wu, and Jürgen Schmidhuber. 2024. MetaGPT: Meta pro- gramming for a multi-agent collaborative framework. In The Twelfth International Conference on Learning Representations.
  8. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, and 1 others. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276. · arXiv
  9. Mandar Joshi, Eunsol Choi, Daniel Weld, and Luke Zettlemoyer. 2017. TriviaQA: A large scale distantly supervised challenge dataset for reading comprehen- sion. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Vol- ume 1: Long Papers), pages 1601-1611, Vancouver, Canada. Association for Computational Linguistics. · DOI · Paper page
  10. Tom Kwiatkowski, Jennimaria Palomaki, Olivia Red- field, Michael Collins, Ankur Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Ken- ton Lee, Kristina Toutanova, Llion Jones, Matthew Kelcey, Ming-Wei Chang, Andrew M. Dai, Jakob Uszkoreit, Quoc Le, and Slav Petrov. 2019. Natu- ral questions: A benchmark for question answering research. Transactions of the Association for Compu- tational Linguistics, 7:452-466. · DOI · Paper page
  11. Xiaoxi Li, Guanting Dong, Jiajie Jin, Yuyao Zhang, Yujia Zhou, Yutao Zhu, Peitian Zhang, and Zhicheng Dou. 2025a. Search-o1: Agentic search- enhanced large reasoning models. arXiv preprint arXiv:2501.05366. · arXiv
  12. Xuefeng Li, Haoyang Zou, and Pengfei Liu. 2025b. Limr: Less is more for rl scaling. Preprint, arXiv:2502.11886. · arXiv
  13. Xuefeng Li, Haoyang Zou, and Pengfei Liu. 2025c. Torl: Scaling tool-integrated rl. Preprint, arXiv:2503.23383. · arXiv
  14. Alex Mallen, Akari Asai, Victor Zhong, Rajarshi Das, Hannaneh Hajishirzi, and Daniel Khashabi. 2022. When not to trust language models: Investigating effectiveness and limitations of parametric and non- parametric memories. arXiv preprint.
  15. Ofir Press, Muru Zhang, Sewon Min, Ludwig Schmidt, Noah A Smith, and Mike Lewis. 2022. Measuring and narrowing the compositionality gap in language models. arXiv preprint arXiv:2210.03350. · arXiv
  16. Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, and 1 others. 2023. Toolllm: Facilitating large language models to master 16000+ real-world apis. arXiv preprint arXiv:2307.16789. · arXiv
  17. 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 oth- ers. 2025. Qwen2.5 technical report. Preprint, arXiv:2412.15115. · arXiv
  18. Timo Schick, Jane Dwivedi-Yu, Roberto Dessì, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettle- moyer, Nicola Cancedda, and Thomas Scialom. 2023. Toolformer: Language models can teach themselves to use tools. Advances in Neural Information Pro- cessing Systems, 36:68539-68551.
  19. 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. · arXiv
  20. Kimi Team, Angang Du, Bofei Gao, Bowei Xing, Changjiu Jiang, Cheng Chen, Cheng Li, Chenjun Xiao, Chenzhuang Du, Chonghua Liao, and 1 others. 2025. Kimi k1.5: Scaling reinforcement learning with llms. arXiv preprint arXiv:2501.12599. · arXiv
  21. Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022. MuSiQue: Multi- hop questions via single-hop question composition. Transactions of the Association for Computational Linguistics.
  22. Prakhar Verma, Sukruta Prakash Midigeshi, Gaurav Sinha, Arno Solin, Nagarajan Natarajan, and Amit Sharma. 2025. Plan*rag: Efficient test-time plan- ning for retrieval augmented generation. Preprint, arXiv:2410.20753. · arXiv
  23. Xiaohua Wang, Zhenghua Wang, Xuan Gao, Feiran Zhang, Yixin Wu, Zhibo Xu, Tianyuan Shi, Zhengyuan Wang, Shizheng Li, Qi Qian, and 1 oth- ers. 2024a. Searching for best practices in retrieval- augmented generation. In Proceedings of the 2024 Conference on Empirical Methods in Natural Lan- guage Processing, pages 17716-17736.
  24. Ziting Wang, Haitao Yuan, Wei Dong, Gao Cong, and Feifei Li. 2024b. Corag: A cost-constrained retrieval optimization system for retrieval-augmented genera- tion. arXiv preprint arXiv:2411.00744. xAI. 2025. Grok 3. · arXiv
  25. Yuqing Yang, Ethan Chern, Xipeng Qiu, Graham Neu- big, and Pengfei Liu. 2024. Alignment for honesty. Advances in Neural Information Processing Systems, 37:63565-63598.
  26. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Ben- gio, William W. Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. HotpotQA: A dataset for diverse, explainable multi-hop question answer- ing. In Conference on Empirical Methods in Natural Language Processing (EMNLP).
  27. Tian Yu, Shaolei Zhang, and Yang Feng. 2024. Auto-rag: Autonomous retrieval-augmented gener- ation for large language models. arXiv preprint arXiv:2411.19443. · arXiv
  28. Murong Yue, Wenlin Yao, Haitao Mi, Dian Yu, Ziyu Yao, and Dong Yu. 2024a. Dots: Learning to reason dynamically in llms via optimal reasoning trajectories search. arXiv preprint arXiv:2410.03864. · arXiv
  29. Zhenrui Yue, Honglei Zhuang, Aijun Bai, Kai Hui, Rolf Jagerman, Hansi Zeng, Zhen Qin, Dong Wang, Xuan- hui Wang, and Michael Bendersky. 2024b. Inference scaling for long-context retrieval augmented genera- tion. arXiv preprint arXiv:2410.04343. · arXiv
  30. Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric P. Xing, Hao Zhang, Joseph E. Gonzalez, and Ion Stoica. 2023. Judging llm-as-a-judge with mt-bench and chatbot arena. In Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Pro- cessing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 -16, 2023.
  31. Yuxiang Zheng, Shichao Sun, Lin Qiu, Dongyu Ru, Cheng Jiayang, Xuefeng Li, Jifan Lin, Binjie Wang, Yun Luo, Renjie Pan, Yang Xu, Qingkai Min, Zizhao Zhang, Yiwen Wang, Wenjie Li, and Pengfei Liu. 2024. OpenResearcher: Unleashing AI for accel- erated scientific research. In Proceedings of the 2024 Conference on Empirical Methods in Natu- ral Language Processing: System Demonstrations, pages 209-218, Miami, Florida, USA. Association for Computational Linguistics. · DOI · Paper page

Cited by

See all 35

Metadata

Authors
Zheng, Y., Fu, D., Hu, X., Cai, X., Ye, L., Lu, P., & Liu, P.
Venue
Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
Publisher
Association for Computational Linguistics
Published
2025
Pages
414-431
DOI
10.18653/v1/2025.emnlp-main.22
Cited by
35
References
2

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.