Elamur Research⌕
☰
home / papers / 2025 / FineReason: Evaluating and Improving LLMs' Deliberate Reasoning through Reflective Puzzle Solving

FineReason: Evaluating and Improving LLMs' Deliberate Reasoning through Reflective Puzzle Solving

Paper · 2025 · Guizhen Chen, Weiwen Xu, Hao Zhang et al.

Summary

Most reasoning benchmarks for LLMs score only the final answer. They do not show whether a model can examine its intermediate steps, notice a dead end, and recover from it. FineReason fills this gap with logic puzzles that break down into small verifiable steps. It covers four puzzle types: Sudoku, Graph Coloring, Game of 24, and logic grid puzzles. For each puzzle, depth-first search over single-step "minimal moves" builds a full solution tree, and every state is checked automatically by code that encodes the puzzle rules. Two tasks are drawn from these trees. State checking asks whether a given state can still reach a solution. State transition asks for the next valid move, which may be a step forward or a backtrack. Under this evaluation, general-purpose models struggle, and even strong reasoning models differ substantially. The authors also release a puzzle training set. Adding it to standard reasoning data improves math and general reasoning, which suggests that skills like reflection and backtracking learned on puzzles carry over to other domains.

Key points

  • Data sources:
    • Sudoku puzzles come from a Kaggle dataset.
    • Game of 24 problems come from Yao et al. (2023).
    • Grid puzzles come from Tyagi et al. (2024). Their textual clues were turned into checkable code using GPT-4o one-shot prompting, followed by manual verification.
    • Graph Coloring instances are randomly generated graphs.
  • Gaps between reasoning models: OpenAI-o1 outperforms Gemini-2.0-Flash-Thinking by 19.7%, a difference that saturated math and coding benchmarks do not reveal.
  • General-purpose models: GPT-4o performs close to random on state checking and poorly on state transition.
  • Math transfer: Adding puzzle data to math-only fine-tuning of DeepSeek-R1-Distill-Qwen-7B raises GSM8K accuracy from 82.3% to 87.4%.
  • MMLU transfer: Compared with training on MMLU-train alone, mixed training raises MMLU-test scores from 41.1 to 44.9 (1.5B model) and from 61.2 to 64.2 (7B model). The largest gains are in STEM, at about +10%.

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

🏢Alibaba GroupHangzhou · China168 works · 15,386 citations🎓Nanyang Technological UniversitySingapore · Singapore156 works · 25,644 citations🏢Alibaba DAMO AcademyHangzhou · China39 works · 6,378 citations

 Fields, topics and concepts

📚 Shelves 4The 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 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,320 worksReasoning · 1,426 works

Cite

Cite as Guizhen Chen, Weiwen Xu, Hao Zhang, Hou Pong Chan, Chaoqun Liu, Lidong Bing, Deli Zhao, Anh Tuan Luu and Yu Rong. 2025. FineReason: Evaluating and Improving LLMs' Deliberate Reasoning through Reflective Puzzle Solving. Proceedings of the 63rd 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/v1/2025.acl-long.333

@inproceedings{chen2025finereason,
  title = {{FineReason: Evaluating and Improving LLMs' Deliberate Reasoning through Reflective Puzzle Solving}},
  author = {Guizhen Chen, Weiwen Xu, Hao Zhang, Hou Pong Chan, Chaoqun Liu, Lidong Bing, Deli Zhao, Anh Tuan Luu and Yu Rong},
  booktitle = {Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2025},
  doi = {10.18653/v1/2025.acl-long.333},
  url = {https://doi.org/10.18653/v1/2025.acl-long.333}
}

References

58 references, 26 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Arpit Agarwal, Katharina Muelling, and Katerina Fragkiadaki. 2019. Model learning for look-ahead exploration in continuous control. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 3151–3158. find
  2. Jacob Austin, Augustus Odena, Maxwell Nye, Maarten Bosma, Henryk Michalewski, David Dohan, Ellen Jiang, Carrie Cai, Michael Terry, Quoc Le, et al. 2021. Program synthesis with large language models. arXiv preprint arXiv:2108.07732.
  3. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde De Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, et al. 2021. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374.
  4. Wenhu Chen, Ming Yin, Max Ku, Pan Lu, Yixin Wan, Xueguang Ma, Jianyu Xu, Xinyi Wang, and Tony Xia. 2023. TheoremQA: A theorem-driven question answering dataset. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 7889–7901, Singapore. Association for Comp…
  5. Yew Ken Chia, Guizhen Chen, Weiwen Xu, Anh Tuan Luu, Soujanya Poria, and Lidong Bing. 2024a. Reasoning paths optimization: Learning to reason and explore from diverse paths. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 16763–16780, Miami, Florida, USA. Association… find
  6. Yew Ken Chia, Vernon Toh, Deepanway Ghosal, Lidong Bing, and Soujanya Poria. 2024b. PuzzleVQA: Diagnosing multimodal reasoning challenges of language models with abstract visual patterns. In Findings of the Association for Computational Linguistics: ACL 2024, pages 16259–16273, Bangkok, Thailand. As… find
  7. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168.
  8. Antonia Creswell, Murray Shanahan, and Irina Higgins. 2023. Selection-inference: Exploiting large language models for interpretable logical reasoning. In The Eleventh International Conference on Learning Representations. find
  9. DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, et al. 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. Preprint, arXiv:2501.12948.
  10. Ruomeng Ding, Chaoyun Zhang, Lu Wang, Yong Xu, Minghua Ma, Wei Zhang, Si Qin, Saravan Rajmohan, Qingwei Lin, and Dongmei Zhang. 2024. Everything of thoughts: Defying the law of penrose triangle for thought generation. In Findings of the Association for Computational Linguistics: ACL 2024, pages 1638… find
  11. Nouha Dziri, Ximing Lu, Melanie Sclar, Xiang Lorraine Li, Liwei Jiang, Bill Yuchen Lin, Sean Welleck, Peter West, Chandra Bhagavatula, Ronan Le Bras, Jena D. Hwang, Soumya Sanyal, Xiang Ren, Allyson Ettinger, Zaid Harchaoui, and Yejin Choi. 2023. Faith and fate: Limits of transformers on composition…
  12. Daniel Enström, Viktor Kjellberg, and Moa Johansson. 2024. Reasoning in transformers - mitigating spurious correlations and reasoning shortcuts. CoRR, abs/2403.11314. — [arXiv:2403.11314](https://arxiv.org/abs/2403.11314) find
  13. Xidong Feng, Yicheng Luo, Ziyan Wang, Hongrui Tang, Mengyue Yang, Kun Shao, David Henry Mguni, Yali Du, and Jun Wang. 2023. ChessGPT: Bridging policy learning and language modeling. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track. find
  14. Panagiotis Giadikiaroglou, Maria Lymperaiou, Giorgos Filandrianos, and Giorgos Stamou. 2024. Puzzle solving using reasoning of large language models: A survey. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 11574–11591, Miami, Florida, USA. Associati…
  15. Google. 2024. Gemini-2.0-flash family. find
  16. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2021a. Measuring massive multitask language understanding. In International Conference on Learning Representations.
  17. American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 59–81, Mexico City, Mexico. Association for Computational Linguistics. find
  18. Hunter Lightman, Vineet Kosaraju, Yuri Burda, Harrison Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. 2024. Let’s verify step by step. In The Twelfth International Conference on Learning Representations.
  19. Jieyi Long. 2023. Large language model guided tree-of-thought. find
  20. Jingkun Ma, Runzhe Zhan, Derek F. Wong, Yang Li, Di Sun, Hou Pong Chan, and Lidia S. Chao. 2024. Visaidmath: Benchmarking visual-aided mathematical reasoning. CoRR, abs/2410.22995.
  21. Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. 2021b. Measuring mathematical problem solving with the MATH dataset. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2).
  22. Todor Mihaylov, Peter Clark, Tushar Khot, and Ashish Sabharwal. 2018. Can a suit of armor conduct electricity? a new dataset for open book question answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 2381–2391, Brussels, Belgium. Association fo… find
  23. Kung-Hsiang Huang, Hou Pong Chan, May Fung, Haoyi Qiu, Mingyang Zhou, Shafiq Joty, Shih-Fu Chang, and Heng Ji. 2025. From pixels to insights: A survey on automatic chart understanding in the era of large foundation models. IEEE Trans. Knowl. Data Eng., 37(5):2550–2568. find
  24. Chinmay Mittal, Krishna Kartik, Parag Singla, et al. 2024. Puzzlebench: Can llms solve challenging first-order combinatorial reasoning problems? arXiv preprint arXiv:2402.02611. — [arXiv:2402.02611](https://arxiv.org/abs/2402.02611) find
  25. Kung-Hsiang Huang, Mingyang Zhou, Hou Pong Chan, Yi Fung, Zhenhailong Wang, Lingyu Zhang, Shih-Fu Chang, and Heng Ji. 2024. Do lvlms understand charts? analyzing and correcting factual errors in chart captioning. In Findings of the Association for Computational Linguistics, ACL 2024, Bangkok, Thaila… find
  26. Adam Ishay, Zhun Yang, and Joohyung Lee. 2023. Leveraging large language models to generate answer set programs. arXiv preprint arXiv:2307.07699. — [arXiv:2307.07699](https://arxiv.org/abs/2307.07699) find
  27. Daniel Kahneman. 2011. Thinking, fast and slow. Farrar, Straus and Giroux. find
  28. Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwasawa. 2022. Large language models are zero-shot reasoners. In Advances in Neural Information Processing Systems.
  29. Xingxuan Li, Weiwen Xu, Ruochen Zhao, Fangkai Jiao, Shafiq Joty, and Lidong Bing. 2024a. Can we further elicit reasoning in llms? critic-guided planning with retrieval-augmentation for solving challenging tasks. arXiv preprint arXiv:2410.01428.
  30. Yinghao Li, Haorui Wang, and Chao Zhang. 2024b. Assessing logical puzzle solving in large language models: Insights from a minesweeper case study. In Proceedings of the 2024 Conference of the North Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zett…
  31. David A. Noever and Ryerson Burdick. 2021. Puzzle solving without search or human knowledge: An unnatural language approach. ArXiv, abs/2109.02797. — [arXiv:2109.02797](https://arxiv.org/abs/2109.02797) find
  32. OpenAI. 2024. Learning to reason with llms. find
  33. OpenAI, Aaron Hurst, Adam Lerer, Adam P. Goucher, et al. 2024. GPT-4o system card. ArXiv, abs/2410.21276.
  34. Yiwei Qin, Xuefeng Li, Haoyang Zou, Yixiu Liu, Shijie Xia, Zhen Huang, Yixin Ye, Weizhe Yuan, Hector Liu, Yuanzhi Li, et al. 2024. O1 replication journey: A strategic progress report–part 1. arXiv preprint arXiv:2410.18982. find
  35. An Yang Qwen, 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, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Mei Li, Mingfeng Xue… find
  36. David Rein, Betty Li Hou, Asa Cooper Stickland, Jackson Petty, Richard Yuanzhe Pang, Julien Dirani, Julian Michael, and Samuel R. Bowman. 2024. GPQA: A graduate-level google-proof q&a benchmark. In First Conference on Language Modeling.
  37. Rebecca Roelofs, Vaishaal Shankar, Benjamin Recht, Sara Fridovich-Keil, Moritz Hardt, John Miller, and Ludwig Schmidt. 2019. A meta-analysis of overfitting in machine learning. Advances in Neural Information Processing Systems, 32. find
  38. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, Y. K. Li, Y. Wu, and Daya Guo. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. Preprint, arXiv:2402.03300.
  39. Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik R Narasimhan, and Shunyu Yao. 2023. Reflexion: language agents with verbal reinforcement learning. In Thirty-seventh Conference on Neural Information Processing Systems.
  40. Joykirat Singh, Akshay Nambi, and Vibhav Vineet. 2024. Exposing the achilles’ heel: Evaluating llms ability to handle mistakes in mathematical reasoning. arXiv preprint arXiv:2406.10834.
  41. Charlie Victor Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. 2025. Scaling LLM test-time compute optimally can be more effective than scaling parameters for reasoning. In The Thirteenth International Conference on Learning Representations. find
  42. Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao, Abu Awal Md Shoeb, Abubakar Abid, Adam Fisch, Adam R Brown, Adam Santoro, Aditya Gupta, Adrià Garriga-Alonso, et al. 2023. Beyond the imitation game: Quantifying and extrapolating the capabilities of language models. Transactions on Machine Learning…
  43. Kimi Team. 2025. Kimi k1.5: Scaling reinforcement learning with llms. ArXiv, abs/2501.12599.
  44. Nemika Tyagi, Mihir Parmar, Mohith Kulkarni, Aswin Rrv, Nisarg Patel, Mutsumi Nakamura, Arindam Mitra, and Chitta Baral. 2024. Step-by-step reasoning to solve grid puzzles: Where do LLMs falter? In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 19898–19… find
  45. Peter van Beek. 2006. Chapter 4 - backtracking search algorithms. In Francesca Rossi, Peter van Beek, and Toby Walsh, editors, Handbook of Constraint Programming, volume 2 of Foundations of Artificial Intelligence, pages 85–134. Elsevier. find
  46. Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc V Le, Ed H. Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2023. Self-consistency improves chain of thought reasoning in language models. In The Eleventh International Conference on Learning Representations.
  47. Zihan Wang, Xiangyang Li, Jiahao Yang, Yeqi Liu, Junjie Hu, Ming Jiang, and Shuqiang Jiang. 2024. Lookahead exploration with neural radiance representation for continuous vision-language navigation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13753– 13… find
  48. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, brian ichter, Fei Xia, Ed H. Chi, Quoc V Le, and Denny Zhou. 2022. Chain of thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems.
  49. Larry Wos, Ross Overbeek, Ewing Lusk, and Jim Boyle. 1992. Automated reasoning introduction and applications. McGraw-Hill, Inc. find
  50. Weiwen Xu, Deng Cai, Zhisong Zhang, Wai Lam, and Shuming Shi. 2024. Reasons to reject? aligning language models with judgments. In Findings of the Association for Computational Linguistics: ACL 2024, pages 12288–12304, Bangkok, Thailand. Association for Computational Linguistics. find
  51. Weiwen Xu, Yang Deng, Huihui Zhang, Deng Cai, and Wai Lam. 2021a. Exploiting reasoning chains for multi-hop science question answering. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 1143–1156, Punta Cana, Dominican Republic. Association for Computational Linguistics… find
  52. Weiwen Xu, Xin Li, Yang Deng, Wai Lam, and Lidong Bing. 2023. PeerDA: Data augmentation via modeling peer relation for span identification tasks. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8681–8699, Toronto, Canada. Asso… find
  53. Weiwen Xu, Huihui Zhang, Deng Cai, and Wai Lam. 2021b. Dynamic semantic graph construction and reasoning for explainable multi-hop science question answering. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 1044–1056, Online. Association for Computational Linguis… find
  54. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. HotpotQA: A dataset for diverse, explainable multi-hop question answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 2369…
  55. Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Thomas L. Griffiths, Yuan Cao, and Karthik R Narasimhan. 2023. Tree of thoughts: Deliberate problem solving with large language models. In Thirty-seventh Conference on Neural Information Processing Systems. find
  56. Longhui Yu, Weisen Jiang, Han Shi, Jincheng YU, Zhengying Liu, Yu Zhang, James Kwok, Zhenguo Li, Adrian Weller, and Weiyang Liu. 2024. Metamath: Bootstrap your own mathematical questions for large language models. In The Twelfth International Conference on Learning Representations.
  57. Eric Zelikman, Yuhuai Wu, Jesse Mu, and Noah Goodman. 2022. STar: Bootstrapping reasoning with reasoning. In Advances in Neural Information Processing Systems.
  58. Zhongshen Zeng, Yinhong Liu, Yingjia Wan, Jingyao Li, Pengguang Chen, Jianbo Dai, Yuxuan Yao, Rongwu Xu, Zehan Qi, Wanru Zhao, Linling Shen, Jianqiao Lu, Haochen Tan, Yukang Chen, Hao Zhang, Zhan Shi, Bailin Wang, Zhijiang Guo, and Jiaya Jia. 2024. MR-ben: A meta-reasoning benchmark for evaluating s… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/chen-2025-finereason-evaluating-improving-llms-deliberate/