Elamur Research⌕
☰
home / papers / 2025 / NaturalReasoning: Reasoning in the Wild with 2.8M Challenging Questions

NaturalReasoning: Reasoning in the Wild with 2.8M Challenging Questions

Paper · 2025 · Weizhe Yuan, Jane Yu, Song Jiang et al.

Summary

Most data for training reasoning in LLMs comes from math and coding, where answers can be checked automatically. Other fields lack varied, difficult training questions. The authors address this with NaturalReasoning, a set of 2.8M questions built entirely with LLMs and no human annotation. An LLM rates documents from the DCLM-baseline and FineMath pretraining corpora for reasoning depth. The strongest documents then serve as grounding for newly written, self-contained questions. A reference answer is attached whenever the source text supports one. Near-duplicates are removed with word-level MinHash, and questions overlapping MATH, GPQA, MMLU-Pro and MMLU-Stem are filtered out. Compared with MetaMathQA, NuminaMath, OpenMathInstruct-2 and WebInstruct, the questions cover more domains, are rated higher in quality and prompt longer responses. Experiments show the data works for sample-efficient distillation from a strong teacher. It also supports unsupervised self-training, where self-rewarding performs on par with external reward models. The resource matters because it extends reasoning post-training beyond easily verified math into physics, economics, social sciences and other open-ended areas.

Key points

  • The dataset has 2.8M unique questions with an average length of 55 words, longer than the comparison datasets (34–48 words). It is released at huggingface.co/datasets/facebook/natural_reasoning.
  • Three LLM judges rated 93% of the questions as high quality, compared with 79% for the next-best dataset. In a 100-question human study, experts gave a mean score of 6.45 versus 5.92 for the runner-up.
  • Difficulty was estimated by the median length of Llama3.3-70B-Instruct responses. NaturalReasoning reached 434 words, the longest among all compared datasets.
  • Decontamination used the 13-gram method from llm-eval-harness and removed only 0.026% of items. Deduplication used a MinHash similarity threshold of 0.55.
  • Self-training was tested on Llama3.1-8B-Instruct with RFT and DPO, against a baseline averaging 40.81 on GPQA-Diamond and MMLU-Pro. Self-score-filtered DPO gave the best average (43.67) and 35.02 on GPQA-Diamond, matching or beating external reward models such as INF-ORM-Llama3.1-70B and Qwen2.5-Math-RM-72B.

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

🏢MetaMenlo Park · United States224 works · 241,960 citations🎓New York UniversityNew York · United States233 works · 104,003 citations🎓The University of Texas at ArlingtonArlington · United States8 works · 1,542 citations🏢GenentechSan Francisco · United States2 works · 264 citations🔬Tsinghua–Berkeley Shenzhen InstituteShenzhen · China3 works · 422 citations🔬Facility for Antiproton and Ion ResearchDarmstadt · Germany2 works · 90 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 19,503 worksArtificial Intelligence · 18,805 worksArts and Humanities · 17,959 worksLanguage and Linguistics · 17,789 works

🧭 Topics 1Research subjects and languages the work is about.Reasoning · 1,948 works

Cite

Cite as Weizhe Yuan, Jane Yu, Song Jiang, Karthik Padthe, Yang Li, Ilia Kulikov, Kyunghyun Cho, Dong Wang, Yuandong Tian, Jason E Weston and Xian Li. 2025. NaturalReasoning: Reasoning in the Wild with 2.8M Challenging Questions. arXiv:2502.13124.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2502.13124 · arXiv 2502.13124

@misc{yuan2025naturalreasoning,
  title = {{NaturalReasoning: Reasoning in the Wild with 2.8M Challenging Questions}},
  author = {Weizhe Yuan, Jane Yu, Song Jiang, Karthik Padthe, Yang Li, Ilia Kulikov, Kyunghyun Cho, Dong Wang, Yuandong Tian, Jason E Weston and Xian Li},
  year = {2025},
  doi = {10.48550/arxiv.2502.13124},
  eprint = {2502.13124},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2502.13124}
}

References

50 references, 39 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Loubna Ben Allal, Anton Lozhkov, Elie Bakouch, Gabriel Martín Blázquez, Guilherme Penedo, Lewis Tunstall, Andrés Marafioti, Hynek Kydlíček, Agustín Piqueres Lajarín, Vaibhav Srivastav, Joshua Lochner, Caleb Fahlgren, Xuan-Son Nguyen, Clémentine Fourrier, Ben Burtenshaw, Hugo Larcher, Haojun Zhao, C… — SmolLM2: When Smol Goes Big — Data-Centric Training of a Small Language Model
  2. Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, P… — Qwen Technical Report
  3. Can Balioglu. fairseq2, 2023. URL http://github.com/facebookresearch/fairseq2. ↗
  4. Zixiang Chen, Yihe Deng, Huizhuo Yuan, Kaixuan Ji, and Quanquan Gu. Self-play fine-tuning converts weak language models to strong language models. In Forty-first International Conference on Machine Learning, 2024. URL https://openreview.net/forum?id=O4cHTxW9BS. ↗
  5. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168, 2021.
  6. Aniket Didolkar, Anirudh Goyal, Nan Rosemary Ke, Siyuan Guo, Michal Valko, Timothy Lillicrap, Danilo Rezende, Yoshua Bengio, Michael Mozer, and Sanjeev Arora. Metacognitive capabilities of llms: An exploration in mathematical problem solving. arXiv preprint arXiv:2405.12205, 2024.
  7. Hanze Dong, Wei Xiong, Deepanshu Goyal, Yihan Zhang, Winnie Chow, Rui Pan, Shizhe Diao, Jipeng Zhang, KaShun Shum, and Tong Zhang. RAFT: Reward ranked finetuning for generative foundation model alignment. Transactions on Machine Learning Research, 2023. ISSN 2835-8856. URL https://openreview.net/for… ↗
  8. Leo Gao, Jonathan Tow, Baber Abbasi, Stella Biderman, Sid Black, Anthony DiPofi, Charles Foster, Laurence Golding, Jeffrey Hsu, Alain Le Noac’h, Haonan Li, Kyle McDonell, Niklas Muennighoff, Chris Ociepa, Jason Phang, Laria Reynolds, Hailey Schoelkopf, Aviya Skowron, Lintang Sutawika, Eric Tang, Ani… find
  9. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao,… — The Llama 3 Herd of Models
  10. Etash Guha, Ryan Marten, Sedrick Keh, Negin Raoof, Georgios Smyrnis, Hritik Bansal, Marianna Nezhurina, Jean Mercat, Trung Vu, Zayne Sprague, Ashima Suvarna, Benjamin Feuer, Liangyu Chen, Zaid Khan, Eric Frankel, Sachin Grover, Caroline Choi, Niklas Muennighoff, Shiye Su, Wanjia Zhao, John Yang, Shr… — OpenThoughts: Data Recipes for Reasoning Models
  11. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948, 2025.
  12. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. Measuring massive multitask language understanding. Proceedings of the International Conference on Learning Representations (ICLR), 2021a.
  13. Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. Measuring mathematical problem solving with the math dataset. NeurIPS, 2021b.
  14. Albert Q. Jiang, Alexandre Sablayrolles, Antoine Roux, Arthur Mensch, Blanche Savary, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Emma Bou Hanna, Florian Bressand, Gianna Lengyel, Guillaume Bour, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock,… find
  15. Mingyu Jin, Qinkai Yu, Dong Shu, Haiyan Zhao, Wenyue Hua, Yanda Meng, Yongfeng Zhang, and Mengnan Du. The impact of reasoning step length on large language models. arXiv preprint arXiv:2401.04925, 2024.
  16. Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph Gonzalez, Hao Zhang, and Ion Stoica. Efficient memory management for large language model serving with pagedattention. In Proceedings of the 29th Symposium on Operating Systems Principles, SOSP ’23, page 611–626.…
  17. Nathan Lambert, Jacob Morrison, Valentina Pyatkin, Shengyi Huang, Hamish Ivison, Faeze Brahman, Lester James V. Miranda, Alisa Liu, Nouha Dziri, Shane Lyu, Yuling Gu, Saumya Malik, Victoria Graf, Jena D. Hwang, Jiangjiang Yang, Ronan Le Bras, Oyvind Tafjord, Chris Wilhelm, Luca Soldaini, Noah A. Smi… — Tulu 3: Pushing Frontiers in Open Language Model Post-Training
  18. Chen Li, Weiqi Wang, Jingcheng Hu, Yixuan Wei, Nanning Zheng, Han Hu, Zheng Zhang, and Houwen Peng. Common 7b language models already possess strong math capabilities. arXiv preprint arXiv:2403.04706, 2024a.
  19. Jeffrey Li, Alex Fang, Georgios Smyrnis, Maor Ivgi, Matt Jordan, Samir Gadre, Hritik Bansal, Etash Guha, Sedrick Keh, Kushal Arora, et al. Datacomp-lm: In search of the next generation of training sets for language models. arXiv preprint arXiv:2406.11794, 2024b.
  20. Jia Li, Edward Beeching, Lewis Tunstall, Ben Lipkin, Roman Soletskyi, Shengyi Costa Huang, Kashif Rasul, Longhui Yu, Albert Jiang, Ziju Shen, Zihan Qin, Bin Dong, Li Zhou, Yann Fleureau, Guillaume Lample, and Stanislas Polu. Numinamath. [https://huggingface.co/AI-MO/NuminaMath-CoT](https://github.co… find
  21. Xian Li, Ping Yu, Chunting Zhou, Timo Schick, Omer Levy, Luke Zettlemoyer, Jason Weston, and Mike Lewis. Self-alignment with instruction backtranslation. arXiv preprint arXiv:2308.06259, 2023.
  22. Hunter Lightman, Vineet Kosaraju, Yura Burda, Harri Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. Let’s verify step by step. arXiv preprint arXiv:2305.20050, 2023.
  23. Haoxiong Liu, Yifan Zhang, Yifan Luo, and Andrew C Yao. Augmenting math word problems via iterative question composing. In ICLR 2024 Workshop on Navigating and Addressing Data Problems for Foundation Models.
  24. Xueguang Ma, Qian Liu, Dongfu Jiang, Ge Zhang, Zejun Ma, and Wenhu Chen. General-reasoner: Advancing llm reasoning across all domains. arXiv:2505.14652, 2025. URL https://arxiv.org/abs/2505.14652. ↗
  25. Leland McInnes, John Healy, and James Melville. Umap: Uniform manifold approximation and projection for dimension reduction. arXiv preprint arXiv:1802.03426, 2018.
  26. Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zettlemoyer, Percy Liang, Emmanuel Candès, and Tatsunori Hashimoto. s1: Simple test-time scaling, 2025. URL https://arxiv.org/abs/2501.19393. ↗
  27. OpenAI, :, Aaron Jaech, Adam Kalai, Adam Lerer, Adam Richardson, Ahmed El-Kishky, Aiden Low, Alec Helyar, Aleksander Madry, Alex Beutel, Alex Carney, Alex Iftimie, Alex Karpenko, Alex Tachard Passos, Alexander Neitz, Alexander Prokofiev, Alexander Wei, Allison Tam, Ally Bennett, Ananya Kumar, Andre… — OpenAI o1 System Card
  28. Richard Yuanzhe Pang, Weizhe Yuan, Kyunghyun Cho, He He, Sainbayar Sukhbaatar, and Jason Weston. Iterative reasoning preference optimization. arXiv preprint arXiv:2404.19733, 2024.
  29. Archiki Prasad, Weizhe Yuan, Richard Yuanzhe Pang, Jing Xu, Maryam Fazel-Zarandi, Mohit Bansal, Sainbayar Sukhbaatar, Jason Weston, and Jane Yu. Self-consistency preference optimization. arXiv preprint arXiv:2411.04109, 2024. find
  30. Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. Direct preference optimization: Your language model is secretly a reward model. In Thirty-seventh Conference on Neural Information Processing Systems, 2023. URL https://arxiv.org/abs/2305.18290. ↗
  31. David Rein, Betty Li Hou, Asa Cooper Stickland, Jackson Petty, Richard Yuanzhe Pang, Julien Dirani, Julian Michael, and Samuel R. Bowman. GPQA: A graduate-level google-proof q&a benchmark. In First Conference on Language Modeling, 2024. URL https://openreview.net/forum? id=Ti67584b98. ↗
  32. Corby Rosset, Ching-An Cheng, Arindam Mitra, Michael Santacroce, Ahmed Awadallah, and Tengyang Xie. Direct nash optimization: Teaching language models to self-improve with general preferences. arXiv preprint arXiv:2404.03715, 2024.
  33. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, Y. K. Li, Y. Wu, and Daya Guo. Deepseekmath: Pushing the limits of mathematical reasoning in open language models, 2024. URL https://arxiv.org/abs/2402.03300. ↗
  34. Shuaijie She, Shujian Huang, Wei Zou, Wenhao Zhu, Xiang Liu, Xiang Geng, and Jiajun Chen. MAPO: Advancing multilingual reasoning through multilingual alignment-as-preference optimization. arXiv preprint arXiv:2401.06838, 2024.
  35. Avi Singh, John D Co-Reyes, Rishabh Agarwal, Ankesh Anand, Piyush Patil, Xavier Garcia, Peter J Liu, James Harrison, Jaehoon Lee, Kelvin Xu, Aaron T Parisi, Abhishek Kumar, Alexander A Alemi, Alex Rizkowsky, Azade Nova, Ben Adlam, Bernd Bohnet, Gamaleldin Fathy Elsayed, Hanie Sedghi, Igor Mordatch,… — Beyond Human Data: Scaling Self-Training for Problem-Solving with Language Models
  36. M-A-P Team, Xinrun Du, Yifan Yao, Kaijing Ma, Bingli Wang, Tianyu Zheng, Kang Zhu, Minghao Liu, Yiming Liang, Xiaolong Jin, Zhenlin Wei, Chujie Zheng, Kaixin Deng, Shian Jia, Sichao Jiang, Yiyan Liao, Rui Li, Qinrui Li, Sirun Li, Yizhi Li, Yunwen Li, Dehua Ma, Yuansheng Ni, Haoran Que, Qiyao Wang, Z… — SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines
  37. Shubham Toshniwal, Wei Du, Ivan Moshkov, Branislav Kisacanin, Alexan Ayrapetyan, and Igor Gitman. Openmathinstruct-2: Accelerating ai for math with massive open-source instruction data. In The 4th Workshop on Mathematical Reasoning and AI at NeurIPS’24. Shubham Toshniwal, Wei Du, Ivan Moshkov, Brani…
  38. Hoang Tran, Chris Glaze, and Braden Hancock. Iterative DPO alignment. Technical report, Snorkel AI, 2023. find
  39. Jonathan Uesato, Nate Kushman, Ramana Kumar, Francis Song, Noah Siegel, Lisa Wang, Antonia Creswell, Geoffrey Irving, and Irina Higgins. Solving math word problems with process-and outcome-based feedback. arXiv preprint arXiv:2211.14275, 2022.
  40. Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A Smith, Daniel Khashabi, and Hannaneh Hajishirzi. Self-instruct: Aligning language models with self-generated instructions. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)… find
  41. Yubo Wang, Xueguang Ma, Ge Zhang, Yuansheng Ni, Abhranil Chandra, Shiguang Guo, Weiming Ren, Aaran Arulraj, Xuan He, Ziyan Jiang, et al. Mmlu-pro: A more robust and challenging multi-task language understanding benchmark. arXiv preprint arXiv:2406.01574, 2024.
  42. Junjie Wu. Advances in K-means clustering: a data mining thinking. Springer Science & Business Media, 2012. find
  43. An Yang, Beichen Zhang, Binyuan Hui, Bofei Gao, Bowen Yu, Chengpeng Li, Dayiheng Liu, Jianhong Tu, Jingren Zhou, Junyang Lin, Keming Lu, Mingfeng Xue, Runji Lin, Tianyu Liu, Xingzhang Ren, and Zhenru Zhang. Qwen2.5-math technical report: Toward mathematical expert model via self-improvement. arXiv p…
  44. Yixin Ye, Zhen Huang, Yang Xiao, Ethan Chern, Shijie Xia, and Pengfei Liu. Limo: Less is more for reasoning, 2025. URL https://arxiv.org/abs/2502.03387. ↗
  45. Longhui Yu, Weisen Jiang, Han Shi, YU Jincheng, Zhengying Liu, Yu Zhang, James Kwok, Zhenguo Li, Adrian Weller, and Weiyang Liu. Metamath: Bootstrap your own mathematical questions for large language models. In The Twelfth International Conference on Learning Representations. Longhui Yu, Weisen Jian…
  46. OpenReview.net, 2024. URL https://openreview.net/forum?id=N8N0hgNDRt. ↗
  47. Weizhe Yuan, Richard Yuanzhe Pang, Kyunghyun Cho, Xian Li, Sainbayar Sukhbaatar, Jing Xu, and Jason E Weston. Self-rewarding language models. In Forty-first International Conference on Machine Learning. Weizhe Yuan, Richard Yuanzhe Pang, Kyunghyun Cho, Xian Li, Sainbayar Sukhbaatar, Jing Xu, and Jas… find
  48. Xiang Yue, Tuney Zheng, Ge Zhang, and Wenhu Chen. Mammoth2: Scaling instructions from the web. Advances in Neural Information Processing Systems, 2024.
  49. Eric Zelikman, Yuhuai Wu, Jesse Mu, and Noah Goodman. Star: Bootstrapping reasoning with reasoning. In Advances in Neural Information Processing Systems, 2022.
  50. Lianmin Zheng, Liangsheng Yin, Zhiqiang Xie, Chuyue Sun, Jeff Huang, Cody Hao Yu, Shiyi Cao, Christos Kozyrakis, Ion Stoica, Joseph E. Gonzalez, Clark Barrett, and Ying Sheng. Sglang: Efficient execution of structured language model programs, 2023.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yuan-2025-naturalreasoning-reasoning-wild-2-8m/