{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,10]],"date-time":"2026-11-10T22:46:04Z","timestamp":1794350764823,"version":"4.9.0"},"reference-count":55,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,2,11]],"date-time":"2026-02-11T00:00:00Z","timestamp":1770768000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0"},{"start":{"date-parts":[[2026,3,6]],"date-time":"2026-03-06T00:00:00Z","timestamp":1772755200000},"content-version":"vor","delay-in-days":23,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Sci Rep"],"DOI":"10.1038\/s41598-026-39296-8","type":"journal-article","created":{"date-parts":[[2026,2,11]],"date-time":"2026-02-11T19:56:48Z","timestamp":1770839808000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["Use large language model to enhance reasoning of another large language model through reward updated GRPO"],"prefix":"10.1038","volume":"16","author":[{"given":"Yiqiao","family":"Yin","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,2,11]]},"reference":[{"key":"39296_CR1","doi-asserted-by":"crossref","unstructured":"Wissal Farsal, S. A. & Mohammed, R. Deep learning: An overview. In Proceedings of the 12th International Conference on Intelligent Systems: Theories and Applications, p 1\u20136, (2018).","DOI":"10.1145\/3289402.3289538"},{"key":"39296_CR2","doi-asserted-by":"crossref","unstructured":"Pramila P. S. & Seema, S. A review of machine learning and deep learning applications. In 2018 Fourth international conference on computing communication control and automation (ICCUBEA), p 1\u20136. (IEEE, 2018).","DOI":"10.1109\/ICCUBEA.2018.8697857"},{"key":"39296_CR3","unstructured":"Li, D. & Yang, L. Deep learning in natural language processing. (Springer, 2018)."},{"key":"39296_CR4","unstructured":"Uday, K., John, L. & James, W. Deep learning for NLP and speech recognition, vol.\u00a084. (Springer, 2019)."},{"key":"39296_CR5","doi-asserted-by":"crossref","unstructured":"Daniel W. O., Julian R. M. & Jugal K. K. A survey of the usages of deep learning for natural language processing. IEEE Trans. Neural Netw. Learn. Syst. 32(2), 604\u2013624 (2020).","DOI":"10.1109\/TNNLS.2020.2979670"},{"key":"39296_CR6","unstructured":"Amirsina, T., Rouzbeh A. S., Yaser, K., Nader, T., & Edward A. F. Natural language processing advancements by deep learning: A survey. arXiv:2003.01200 (2020)."},{"key":"39296_CR7","unstructured":"Jiuhai, C. & Jonas, M. Automated data curation for robust language model fine-tuning. arXiv:2403.12776 (2024)."},{"key":"39296_CR8","unstructured":"Xiaoqun, L., Jiacheng, L., Muchao, Y. & Zhaohan, X. Robustifying safety-aligned large language models through clean data curation. arXiv:2405.19358 (2024)."},{"key":"39296_CR9","doi-asserted-by":"crossref","unstructured":"Long, O. et al. Training language models to follow instructions with human feedback. Adv. Neural Inf. Process. Syst. 35, 27730\u201327744 (2022).","DOI":"10.52202\/068431-2011"},{"key":"39296_CR10","unstructured":"Yuntao, B. et al. Constitutional ai: Harmlessness from ai feedback. (2022)."},{"key":"39296_CR11","unstructured":"Yizhong, W. et al. Self-instruct: Aligning language models with self-generated instructions. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL) (2023)."},{"key":"39296_CR12","unstructured":"Zhihong, S. et al. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv:2402.03300 (2024)."},{"key":"39296_CR13","unstructured":"Raul, P., Ryan, S., Mostofa, P., Mohammad, S. & Bryan, C. Training question answering models from synthetic data. arXiv:2002.09599 (2020)."},{"key":"39296_CR14","unstructured":"Yash, G., Tejas, K., Douglas, S.-S., Dhruv, B., & Devi, P. Making the v in vqa matter: Elevating the role of image understanding in visual question answering. In Proceedings of the IEEE conference on computer vision and pattern recognition, p 6904\u20136913, (2017)."},{"key":"39296_CR15","doi-asserted-by":"crossref","unstructured":"Jason, W. et al. Chain-of-thought prompting elicits reasoning in large language models. Adv. Neural Inf. Process. Syst. 35, 24824\u201324837 (2022).","DOI":"10.52202\/068431-1800"},{"key":"39296_CR16","unstructured":"Tom, B. et al. Language models are few-shot learners. Adv. Neural Inf. Process. Syst. 33, 1877\u20131901 (2020)."},{"issue":"1","key":"39296_CR17","doi-asserted-by":"publisher","first-page":"12","DOI":"10.1038\/s41588-018-0295-5","volume":"51","author":"J Zou","year":"2019","unstructured":"Zou, J. et al. A primer on deep learning in genomics. Nat. Genet. 51(1), 12\u201318 (2019).","journal-title":"Nat. Genet."},{"key":"39296_CR18","doi-asserted-by":"publisher","first-page":"443","DOI":"10.1016\/j.neucom.2021.05.103","volume":"470","author":"I Lauriola","year":"2022","unstructured":"Lauriola, I., Lavelli, A. & Aiolli, F. An introduction to deep learning in natural language processing: Models, techniques, and tools. Neurocomputing 470, 443\u2013456 (2022).","journal-title":"Neurocomputing"},{"key":"39296_CR19","unstructured":"Dorottya, D., Kelvin, G., & Percy, L. Transforming question answering datasets into natural language inference datasets. arXiv:1809.02922, (2018)."},{"key":"39296_CR20","unstructured":"Fabio, P. et al. Language models as knowledge bases? arXiv:1909.01066, (2019)."},{"key":"39296_CR21","unstructured":"Kelvin, G., Kenton, L., Zora, T., Panupong, P. & Mingwei, C. Retrieval augmented language model pre-training. In International conference on machine learning, p 3929\u20133938. (PMLR, 2020)."},{"key":"39296_CR22","doi-asserted-by":"publisher","first-page":"453","DOI":"10.1162\/tacl_a_00276","volume":"7","author":"T Kwiatkowski","year":"2019","unstructured":"Kwiatkowski, T. et al. Natural questions: a benchmark for question answering research. Trans. Assoc. Comput. Linguist. 7, 453\u2013466 (2019).","journal-title":"Trans. Assoc. Comput. Linguist."},{"key":"39296_CR23","unstructured":"Li, D. et al. Unified language model pre-training for natural language understanding and generation. Adv. Neural Inf. Process. Syst. 32, (2019)."},{"key":"39296_CR24","unstructured":"Fernandes, J. & Ricardo, C. Improving model performance: comparing complete fine-tuning with parameter efficient language model tuning on a small, portuguese, domain-specific, dataset. B.S. thesis, Universidade Tecnol\u00f3gica Federal do Paran\u00e1, (2022)."},{"key":"39296_CR25","unstructured":"Jeremy, H. & Sebastian, R. Universal language model fine-tuning for text classification. arXiv:1801.06146 (2018)."},{"key":"39296_CR26","doi-asserted-by":"crossref","unstructured":"Zoltan P. M. Sample size considerations for fine-tuning large language models for named entity recognition tasks: methodological study. Jmir ai 3, e52095 (2024).","DOI":"10.2196\/52095"},{"key":"39296_CR27","unstructured":"Xiangyu, Q. et al. Fine-tuning aligned language models compromises safety, even when users do not intend to! arXiv:2310.03693 (2023)."},{"key":"39296_CR28","doi-asserted-by":"crossref","unstructured":"Jianqi, G. et al. Promoting knowledge base question answering by directing llms to generate task-relevant logical forms. In Proceedings of the Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI-25), p 23914\u201323922. (AAAI Press, 2025).","DOI":"10.1609\/aaai.v39i22.34564"},{"key":"39296_CR29","doi-asserted-by":"crossref","unstructured":"Yuan, S. et al. Fidelis: Faithful reasoning in large language models for knowledge graph question answering. In Findings of the Association for Computational Linguistics: ACL 2025, p 8315\u20138330, (Association for Computational Linguistics, Vienna, Austria, 2025).","DOI":"10.18653\/v1\/2025.findings-acl.436"},{"key":"39296_CR30","doi-asserted-by":"publisher","DOI":"10.1016\/j.inffus.2025.103426","volume":"125","author":"Q Liu","year":"2026","unstructured":"Liu, Q. et al. Legal knowledge infusion for large language models: A survey. Inf. Fusion 125, 103426 (2026).","journal-title":"Inf. Fusion"},{"key":"39296_CR31","doi-asserted-by":"crossref","unstructured":"Joao, M. et al. Repliqa: A question-answering dataset for benchmarking llms on unseen reference content. Adv. Neural Inf. Process. Syst. 37, 24242\u201324276 (2025).","DOI":"10.52202\/079017-0764"},{"key":"39296_CR32","doi-asserted-by":"crossref","unstructured":"Siqing, H., Negar, A. & Charles, C. Retrieving supporting evidence for generative question answering. In Proceedings of the annual international ACM SIGIR conference on research and development in information retrieval in the Asia Pacific region, p 11\u201320, (2023).","DOI":"10.1145\/3624918.3625336"},{"key":"39296_CR33","doi-asserted-by":"crossref","unstructured":"Utkarsh, U., Sai, S., Harsha, S., Sayantan, M. & Tulika, S. \u201dreasoning before responding\u201d: Towards legal long-form question answering with interpretability. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management, p. 4922\u20134930, (2024).","DOI":"10.1145\/3627673.3680082"},{"key":"39296_CR34","unstructured":"Somnath, B. et al. Context matters: Pushing the boundaries of open-ended answer generation with graph-structured knowledge context. arXiv:2401.12671 (2024)."},{"key":"39296_CR35","unstructured":"Shivam, R., Abhiroop, T., Raghav, C., Nikhil, A., & Vinayak K, D. Beyond qa pairs: Assessing parameter-efficient fine-tuning for fact embedding in llms. In Workshop on Preparing Good Data for Generative AI: Challenges and Approaches (2025)."},{"key":"39296_CR36","unstructured":"Keheng, W. et al. Knowledge-driven cot: Exploring faithful reasoning in llms for knowledge-intensive question answering. arXiv:2308.13259 (2023)."},{"key":"39296_CR37","first-page":"66383","volume":"37","author":"X Wang","year":"2025","unstructured":"Wang, X & Zhou, D. Chain-of-thought reasoning without prompting. Adv. Neural Inf. Process. Syst. 37, 66383\u201366409 (2025).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"39296_CR38","doi-asserted-by":"crossref","unstructured":"Junyi, Y. et al. Promptcot: Align prompt distribution via adapted chain-of-thought. In Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, p 7027\u20137037, (2024).","DOI":"10.1109\/CVPR52733.2024.00671"},{"key":"39296_CR39","unstructured":"Dingjun, W., Jing, Z. & Xinmei, H. Chain of thought prompting elicits knowledge augmentation. arXiv:2307.01640 (2023)."},{"key":"39296_CR40","unstructured":"Mirac, S. et al. Challenging big-bench tasks and whether chain-of-thought can solve them. arXiv:2210.09261 (2022)."},{"key":"39296_CR41","first-page":"29106","volume":"37","author":"K Stechly","year":"2025","unstructured":"Stechly, K., Valmeekam, K. & Kambhampati, S. Chain of thoughtlessness? an analysis of cot in planning. Adv. Neural Inf. Process. Syst. 37, 29106\u201329141 (2025).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"39296_CR42","unstructured":"Qingyuan, T., Hanlun, Z., Lei, W., Yang, L. & Yunshi, L. R3 prompting: Review, rephrase and resolve for chain-of-thought reasoning in large language models under noisy context. arXiv:2310.16535 (2023)."},{"key":"39296_CR43","unstructured":"Xiang, Z. & Dujian, D. Supervised chain of thought. arXiv:2410.14198 (2024)."},{"key":"39296_CR44","first-page":"333","volume":"37","author":"X Zhang","year":"2025","unstructured":"Zhang, X. et al. Chain of preference optimization: Improving chain-of-thought reasoning in llms. Adv. Neural Inf. Process. Syst. 37, 333\u2013356 (2025).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"39296_CR45","volume-title":"Dan Ley, Chirag Agarwal, and Himabindu Lakkaraju","author":"Sree Harsha Tanneru","year":"2024","unstructured":"Sree Harsha Tanneru. Dan Ley, Chirag Agarwal, and Himabindu Lakkaraju (On the difficulty of faithful chain-of-thought reasoning in large language models, In Trustworthy Multi-modal Foundation Models and AI Agents (TiFA), 2024)."},{"key":"39296_CR46","unstructured":"Qing, L. et al. Faithful chain-of-thought reasoning. In The 13th International Joint Conference on Natural Language Processing and the 3rd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (IJCNLP-AACL 2023), (2023)."},{"key":"39296_CR47","first-page":"54872","volume":"37","author":"Q Chen","year":"2025","unstructured":"Chen, Q., Qin, L., Wang, J., Zhou, J. & Che, W. Unlocking the capabilities of thought: A reasoning boundary framework to quantify and optimize chain-of-thought. Adv. Neural Inf. Process. Syst. 37, 54872\u201354904 (2025).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"39296_CR48","doi-asserted-by":"crossref","unstructured":"Shao, H. et al. Visual cot: Advancing multi-modal language models with a comprehensive dataset and benchmark for chain-of-thought reasoning. Adv. Neural Inf. Process. Syst. 37, 8612\u20138642 (2025).","DOI":"10.52202\/079017-0275"},{"key":"39296_CR49","unstructured":"Karl, C. et al. Training verifiers to solve math word problems. arXiv:2110.14168 (2021)."},{"key":"39296_CR50","first-page":"3843","volume":"35","author":"A Lewkowycz","year":"2022","unstructured":"Lewkowycz, A. et al. Solving quantitative reasoning problems with language models. Adv. Neural Inf. Process. Syst. 35, 3843\u20133857 (2022).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"39296_CR51","doi-asserted-by":"crossref","unstructured":"Takeshi, K. et al. Large language models are zero-shot reasoners. Adv. Neural Inf. Process. Syst., 35, 22199\u201322213 (2022).","DOI":"10.52202\/068431-1613"},{"key":"39296_CR52","unstructured":"Hunter, L. et al. Let\u2019s verify step by step. arXiv:2305.20050 (2023)."},{"key":"39296_CR53","unstructured":"Devichand, B. et al. Hierarchical prompting taxonomy: A universal evaluation framework for large language models aligned with human cognitive principles. In Proceedings of the ACM Workshop on Prompt Optimization (Prompt Optimization KDD 2025). (ACM, 2025)."},{"key":"39296_CR54","unstructured":"Qihuang, Z. et al Achievingspsdoigtsps 97% on gsm8k: Deeply understanding the problems makes llms better solvers for math word problems. arXiv:2404.14963 (2024)."},{"key":"39296_CR55","unstructured":"An, Y. et al. Qwen2 technical report, arXiv: 2407.10671 (2024)."}],"container-title":["Scientific Reports"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-39296-8","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-39296-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-39296-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,6]],"date-time":"2026-03-06T12:42:13Z","timestamp":1772800933000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-39296-8"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,11]]},"references-count":55,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["39296"],"URL":"https:\/\/doi.org\/10.1038\/s41598-026-39296-8","relation":{},"ISSN":["2045-2322"],"issn-type":[{"value":"2045-2322","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,11]]},"assertion":[{"value":"17 May 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"4 February 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"11 February 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors declare no competing interests.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}},{"value":"Warren Buffett Letters Dataset - Copyright Compliance. The Warren Buffett Letters dataset used in this research is derived from shareholder letters published by Berkshire Hathaway Inc. and made freely available to the public on their official website (\n                      \n                      ). These letters are published annually for public dissemination without access restrictions. Our use of this material constitutes fair use under 17 U.S.C.\n                      \n                      107 for the following reasons: (1)\n                      Transformative Purpose\n                      : We do not republish the original letters but create derivative question-answer-reasoning triplets for academic research in artificial intelligence and natural language processing, representing a fundamentally different purpose than the original corporate communication. (2)\n                      Limited Extraction\n                      : We extract small portions at the paragraph level to generate training samples, not entire letters. (3)\n                      Non-Commercial Research\n                      : This work is conducted for academic research and educational purposes. (4)\n                      No Market Substitution\n                      : Our dataset does not serve as a substitute for reading the original comprehensive letters; rather, it may increase interest in Buffett\u2019s investment philosophy.\n                      Attribution and Licensing.\n                      The original shareholder letters are\n                      \n                      Berkshire Hathaway Inc., authored by Warren Buffett. Our derivative dataset includes proper attribution to the original author and source. The dataset is released for non-commercial academic research and educational purposes only under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). Users planning commercial applications should consult legal counsel regarding appropriate licensing. Full attribution and links to original source materials are provided in the dataset repository documentation.\n                      Responsible Use and Takedown Policy.\n                      We have established a clear takedown policy in our dataset documentation. Should Berkshire Hathaway Inc., Warren Buffett, or their authorized representatives express concerns about our use of this material, we commit to: (1) immediately removing the dataset from public repositories, (2) notifying users who have downloaded the dataset, and (3) updating this publication accordingly. Contact information for such requests is provided in the dataset repository.\n                      Precedent in Academic Research.\n                      Our approach follows established practices in computational finance and NLP research, where numerous academic datasets have been created from publicly available corporate documents, SEC filings, earnings calls, and annual reports for research purposes. Examples include the Financial PhraseBank dataset, ECTSum corpus of earnings call transcripts, and various SEC filing datasets used widely in the research community.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Copyright and Data Ethics Statement"}}],"article-number":"8360"}}