Elamur Research⌕
☰
home / papers / 2026 / Format-Adapter: Improving Reasoning Capability of LLMs by Adapting Suitable Format

Format-Adapter: Improving Reasoning Capability of LLMs by Adapting Suitable Format

Paper · 2026 · Dingzirui Wang, Xuanliang Zhang, Rongyu Cao et al.

Summary

Parallel test-time scaling (TTS) usually samples many answers using a single reasoning format chosen by a human. The authors argue this choice limits how much extra inference compute can help. They decompose ensemble error into two parts: the average error of individual predictors minus the divergence among them. Under this view, repeated sampling within one format mainly cancels random perturbations, which improves robustness. However, it cannot remove systematic errors that come from a poor fit between the model and the format. Mixing several formats adds diversity, which can lower those errors.

Building on this, they introduce FORMAT-ADAPTER, a training-free procedure with three steps:

  1. An LLM proposes candidate formats grouped into categories, guided by a task definition and an example.
  2. The model generates answers under each format.
  3. The method scores the candidates and selects formats within a fixed budget by minimizing the error measure.

On math and general reasoning benchmarks, it beats prior baselines at equal compute. The paper presents this as a new state of the art for training-free parallel-scaling TTS. The main lesson is that the choice of format matters for how well TTS works.

Key points

  • The analysis builds on the ensemble error decomposition of Wood et al. (2024). In it, the divergence among predictors produced by different formats offsets the average error.
  • Illustrative categories, each with the formats it contains:
    • English or Chinese, under Natural Language
    • LaTeX or Markdown, under Math Notation
    • Decimal or Character, under Number Format
  • Under the same TTS budget, the method achieves an average relative gain of 2.2% over existing methods.
  • The method's average Oracle score is 89.4%. The best single format trails this by 15.1%, so no one format works best everywhere.
  • The method selects more than 70% of formats in every category. The authors read this as a sign that LLM-based scoring rates candidates too highly. Applying the idea to post-training is left for future work.

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

🎓Harbin Institute of TechnologyHarbin · China158 works · 28,841 citations

 Fields, topics and concepts

📚 Shelves 5The 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 works

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

Cite

Cite as Dingzirui Wang, Xuanliang Zhang, Rongyu Cao, Longxu Dou, Xianzhen Luo, Yingwei MA, Qingfu Zhu, Binhua Li, Fei Huang and Yongbin Li. 2026. Format-Adapter: Improving Reasoning Capability of LLMs by Adapting Suitable Format. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1124

@inproceedings{wang2026format,
  title = {{Format-Adapter: Improving Reasoning Capability of LLMs by Adapting Suitable Format}},
  author = {Dingzirui Wang, Xuanliang Zhang, Rongyu Cao, Longxu Dou, Xianzhen Luo, Yingwei MA, Qingfu Zhu, Binhua Li, Fei Huang and Yongbin Li},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1124},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1124}
}

References

51 references, 25 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Aradhye Agarwal, Ayan Sengupta, and Tanmoy Chakraborty. 2025. The art of scaling test-time compute for large language models. Preprint, arXiv:2512.02008. — [arXiv:2512.02008](https://arxiv.org/abs/2512.02008) find
  2. Amanda Bertsch, Alex Xie, Graham Neubig, and Matthew Gormley. 2023. It’s MBR all the way down: Modern generation techniques through the lens of minimum Bayes risk. In Proceedings of the Big Picture Workshop, pages 108–122, Singapore. Association for Computational Linguistics. find
  3. Maciej Besta, Nils Blach, Ales Kubicek, Robert Gerstenberger, Michal Podstawski, Lukas Gianinazzi, Joanna Gajda, Tomasz Lehmann, Hubert Niewiadomski, Piotr Nyczyk, and Torsten Hoefler. 2024. Graph of thoughts: Solving elaborate problems with large language models. Proceedings of the AAAI Conference… find
  4. Zhenni Bi, Kai Han, Chuanjian Liu, Yehui Tang, and Yunhe Wang. 2025. Forest-of-thought: Scaling test-time compute for enhancing LLM reasoning. In Forty-second International Conference on Machine Learning.
  5. Bradley Brown, Jordan Juravsky, Ryan Ehrlich, Ronald Clark, Quoc V. Le, Christopher Ré, and Azalia Mirhoseini. 2024. Large language monkeys: Scaling inference compute with repeated sampling. Preprint, arXiv:2407.21787.
  6. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. Training verifiers to solve math word problems. Preprint, arXiv:2110.14168.
  7. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao, Aston Zhang, and 516 others. 202… — The Llama 3 Herd of Models
  8. Jonathan Geuter and Gregor Kornhardt. 2025. Robon: Routed online best-of-n for test-time scaling with multiple llms. Preprint, arXiv:2512.05542. — [arXiv:2512.05542](https://arxiv.org/abs/2512.05542) find
  9. Behrooz Ghorbani, Orhan Firat, Markus Freitag, Ankur Bapna, Maxim Krikun, Xavier Garcia, Ciprian Chelba, and Colin Cherry. 2022. Scaling laws for neural machine translation. In International Conference on Learning Representations. find
  10. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Peiyi Wang, Qihao Zhu, Runxin Xu, Ruisong Zhang, Shirong Ma, and 1 others. 2025. Deepseek-r1 incentivizes reasoning in llms through reinforcement learning. Nature, 645(8081):633–638. find
  11. David Heineman, Yao Dou, and Wei Xu. 2024. Improving minimum Bayes risk decoding with multi-prompt. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 22525–22545, Miami, Florida, USA. Association for Computational Linguistics. find
  12. Joel Hestness, Sharan Narang, Newsha Ardalani, Gregory Diamos, Heewoo Jun, Hassan Kianinejad, Md. Mostofa Ali Patwary, Yang Yang, and Yanqi Zhou. 2017. Deep learning scaling is predictable, empirically. Preprint, arXiv:1712.00409.
  13. Jordan Hoffmann, Sebastian Borgeaud, Arthur Mensch, Elena Buchatskaya, Trevor Cai, Eliza Rutherford, Diego de las Casas, Lisa Anne Hendricks, Johannes Welbl, Aidan Clark, Tom Hennigan, Eric Noland, Katherine Millican, George van den Driessche, Bogdan Damoc, Aurelia Guy, Simon Osindero, Karen Simonya… find
  14. Hidetaka Kamigaito, Hiroyuki Deguchi, Yusuke Sakai, Katsuhiko Hayashi, and Taro Watanabe. 2025. Diversity explains inference scaling laws: Through a case study of minimum Bayes risk decoding. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long P… find
  15. Zhewei Kang, Xuandong Zhao, and Dawn Song. 2025. Scalable best-of-n selection for large language models via self-certainty. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. find
  16. Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles.
  17. Dong Li, Xujiang Zhao, Linlin Yu, Yanchi Liu, Wei Cheng, Zhengzhang Chen, Zhong Chen, Feng Chen, Chen Zhao, and Haifeng Chen. 2025. SolverLLM: Leveraging test-time scaling for optimization problem via LLM-guided search. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. 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. Weizhe Lin, Xing Li, Zhiyuan Yang, Xiaojin Fu, Hui-Ling Zhen, Yaoyuan Wang, Xianzhi Yu, Wulong Liu, Xiaosong Li, and Mingxuan Yuan. 2025. Trimr: Verifier-based training-free thinking compression for efficient test-time scaling. Preprint, arXiv:2505.17155. — [arXiv:2505.17155](https://arxiv.org/abs/2505.17155) find
  20. Yexiang Liu, Zekun Li, Zhi Fang, Nan Xu, Ran He, and Tieniu Tan. 2025. Rethinking the role of prompting strategies in LLM test-time scaling: A perspective of probability theory. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages… find
  21. Xianzhen Luo, Qingfu Zhu, Zhiming Zhang, Libo Qin, Xuanyu Zhang, Qing Yang, Dongliang Xu, and Wanxiang Che. 2024. Python is not always the best choice: Embracing multilingual program of thoughts. Preprint, arXiv:2402.10691. find
  22. Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zettlemoyer, Percy Liang, Emmanuel Candes, and Tatsunori Hashimoto. 2025. s1: Simple test-time scaling. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 2…
  23. Tobias Nipkow, Lawrence C. Paulson, and Markus Wenzel. 2002. Isabelle/HOL — A Proof Assistant for Higher-Order Logic, volume 2283 of LNCS. Springer. find
  24. 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, and 244 others. 2024a. Openai o1 s…
  25. OpenAI, Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, Red Avila, Igor Babuschkin, Suchir Balaji, Valerie Balcom, Paul Baltescu, Haiming Bao, Mohammad Bavarian, Jeff Belgum, and 262 others.… — GPT-4 Technical Report
  26. Panupong Pasupat and Percy Liang. 2015. Compositional semantic parsing on semi-structured tables. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 1470–…
  27. Libo Qin, Qiguang Chen, Fuxuan Wei, Shijue Huang, and Wanxiang Che. 2023. Cross-lingual prompting: Improving zero-shot chain-of-thought reasoning across languages. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 2695–2709, Singapore. Association for C…
  28. 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.
  29. Bernardino Romera-Paredes, Mohammadamin Barekatain, Alexander Novikov, Matej Balog, M Pawan Kumar, Emilien Dupont, Francisco Zhernov, Heiko Ruiz, and 1 others. 2024. Mathematical discoveries from program search with large language models. Nature, 625(7995):468–477.
  30. Omer Sagi and Lior Rokach. 2018. Ensemble learning: A survey. WIREs Data Mining and Knowledge Discovery, 8(4):e1249. find
  31. David Saxton, Edward Grefenstette, Felix Hill, and Pushmeet Kohli. 2019. Analysing mathematical reasoning abilities of neural models. In International Conference on Learning Representations.
  32. Charlie Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. 2024. Scaling llm test-time compute optimally can be more effective than scaling model parameters. Preprint, arXiv:2408.03314.
  33. 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
  34. Zhende Song, Shengji Tang, Peng Ye, Jiayuan Fan, Lei Bai, Tao Chen, and Wanli Ouyang. 2025. Ctts: Collective test-time scaling. Preprint, arXiv:2508.03333. — [arXiv:2508.03333](https://arxiv.org/abs/2508.03333) find
  35. Xuezhi Wang, Haohan Wang, and Diyi Yang. 2022. Measure and improve robustness in NLP models: A survey. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4569–4586, Seattle, United States. Associati…
  36. 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.
  37. 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.
  38. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, and 3 others. 2020. Transf…
  39. Danny Wood, Tingting Mu, Andrew M. Webb, Henry W. J. Reeve, Mikel Luján, and Gavin Brown. 2024. A unified theory of diversity in ensemble learning. J. Mach. Learn. Res., 24(1). find
  40. Yangzhen Wu, Zhiqing Sun, Shanda Li, Sean Welleck, and Yiming Yang. 2025. Inference scaling laws: An empirical analysis of compute-optimal inference for LLM problem-solving. In The Thirteenth International Conference on Learning Representations. find
  41. Vikas Yadav, Steven Bethard, and Mihai Surdeanu. 2019. Quick and (not so) dirty: Unsupervised selection of justification sentences for multi-hop question answering. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on… find
  42. 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. Preprint, arXiv:2505.09388.
  43. Zhen Yang, Mingyang Zhang, Feng Chen, Ganggui Ding, Liang Hou, Xin Tao, Pengfei Wan, and Ying-Cong Chen. 2025b. Less is more: Improving llm reasoning with minimal test-time intervention. Preprint, arXiv:2510.13940.
  44. 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
  45. Zhiyuan Zeng, Qinyuan Cheng, Zhangyue Yin, Yunhua Zhou, and Xipeng Qiu. 2025. Revisiting the test-time scaling of o1-like models: Do they truly possess test-time scaling capabilities? In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers),…
  46. Xuanliang Zhang, Dingzirui Wang, Longxu Dou, Baoxin Wang, Dayong Wu, Qingfu Zhu, and Wanxiang Che. 2024. Flextaf: Enhancing table reasoning with flexible tabular formats. Preprint, arXiv:2408.08841. — [arXiv:2408.08841](https://arxiv.org/abs/2408.08841) find
  47. Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric Xing, Hao Zhang, Joseph E. Gonzalez, and Ion Stoica. 2023. Judging LLM-as-a-judge with MT-bench and chatbot arena. In Thirty-seventh Conference on Neural Information Processing…
  48. Tianyang Zhong, Zhengliang Liu, Yi Pan, Yutong Zhang, Yifan Zhou, Shizhe Liang, Zihao Wu, Yanjun Lyu, Peng Shu, Xiaowei Yu, Chao Cao, Hanqi Jiang, Hanxu Chen, Yiwei Li, Junhao Chen, Huawen Hu, Yihen Liu, Huaqin Zhao, Shaochen Xu, and 59 others. 2024. Evaluation of openai o1: Opportunities and challe…
  49. Jin Peng Zhou, Charles E Staats, Wenda Li, Christian Szegedy, Kilian Q Weinberger, and Yuhuai Wu. 2024. Don’t trust: Verify – grounding LLM quantitative reasoning with autoformalization. In The Twelfth International Conference on Learning Representations. find
  50. Yilun Zhou, Austin Xu, PeiFeng Wang, Caiming Xiong, and Shafiq Joty. 2025. Evaluating judges as evaluators: The JETTS benchmark of LLM-as-judges as test-time scaling evaluators. In Forty-second International Conference on Machine Learning. find
  51. King Zhu, Hanhao Li, Siwei Wu, Tianshun Xing, Dehua Ma, Xiangru Tang, Minghao Liu, Jian Yang, Jiaheng Liu, Yuchen Eleanor Jiang, Changwang Zhang, Chenghua Lin, Jun Wang, Ge Zhang, and Wangchunshu Zhou. 2025. Scaling test-time compute for llm agents. Preprint, arXiv:2506.12928. — [arXiv:2506.12928](https://arxiv.org/abs/2506.12928) find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wang-2026-format-adapter-improving-reasoning-capability/