Self-Training Meets Consistency: Improving LLMs’ Reasoning with Consistency-Driven Rationale Evaluation
Open accessAbstract
Jaehyeok Lee, Keisuke Sakaguchi, JinYeong Bak. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
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.
Integrity signals
Registry
Cites
Statistics
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.
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?
CoT explanations often diverge from the model's latent computation path, and coherent rationales can be post-hoc fabrications rather than faithful reasoning paths.
“These findings align with evidence that CoT explanations often diverge from the model's latent computation path (Barez et al., 2025;Turpin et al., 2023), and that coherent rationales can be post-hoc fabrications rather than faithful reasoning paths (Lee et al., 2025;Chua et al., 2024;Lanham et al., 2023).”
From Good Arguments Against the People Pleasers: How Reasoning Mitigates (Yet Masks) LLM Sycophancy (2026) · cites this paper as Lee et al., 2025
The Introduction (page unavailable) says an LLM can reach a correct answer through inappropriate reasoning steps, and its example includes an incomplete rationale that would be treated as appropriate because it leads to the correct answer. This supports the warning that an apparently acceptable rationale need not reflect sound reasoning. The paper does not establish that such rationales are post-hoc fabrications or compare them with the model’s latent computation path; the citation to Lee et al. is attached to the second clause, not the first. (LLM, conf=0.83)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Decoupled Weight Decay Regularization
Loshchilov, I., & Hutter, F. (2017) · arXiv · doi:10.48550/arxiv.1711.05101 · Cited by 12,522 · see citing works
Proximal Policy Optimization Algorithms
Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017) · arXiv · doi:10.48550/arxiv.1707.06347 · Cited by 10,625 · see citing works
Transformers: State-of-the-Art Natural Language Processing
Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations · doi:10.18653/v1/2020.emnlp-demos.6 · Cited by 8,741 · see citing works
Chain-Of-Thought Prompting Elicits Reasoning in Large Language Models
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E., Le, Q. V., & Zhou, D. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1800 · Cited by 4,505 · see citing works
CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge
Talmor, A., Herzig, J., Lourie, N., & Berant, J. (2019) · North American Chapter of the Association for Computational Linguistics · doi:10.18653/v1/n19-1421 · Cited by 2,805 · see citing works
ConceptNet 5.5: An Open Multilingual Graph of General Knowledge
Speer, R., Chin, J., & Havasi, C. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.11164 · Cited by 2,312 · see citing works
Large language models are zero-shot reasoners
Kojima, T., Gu, S. (., Reid, M., Matsuo, Y., & Iwasawa, Y. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1613 · Cited by 2,013 · see citing works
Efficient Memory Management for Large Language Model Serving with PagedAttention
Kwon, W., Li, Z., Zhuang, S., Sheng, Y., Zheng, L., Yu, C. H., Gonzalez, J., Zhang, H., & Stoica, I. (2023) · Proceedings of the 29th Symposium on Operating Systems Principles · doi:10.1145/3600006.3613165 · Cited by 1,817 · see citing works
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
Rafailov, R., Sharma, A., Mitchell, E., Manning, C. D., Ermon, S., & Finn, C. (2023) · Advances in Neural Information Processing Systems 36 · doi:10.52202/075280-2338 · Cited by 686 · see citing works
Training Verifiers to Solve Math Word Problems
Cobbe, K., Kosaraju, V., Bavarian, M., Chen, M., Jun, H., Kaiser, L., Plappert, M., Tworek, J., Hilton, J., Nakano, R., et al. (2021) · arXiv · doi:10.48550/arxiv.2110.14168 · Cited by 575 · see citing works
Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
Clark, P., Cowhey, I., Etzioni, O., Khot, T., Sabharwal, A., Schoenick, C., & Tafjord, O. (2018) · arXiv · doi:10.48550/arxiv.1803.05457 · Cited by 462 · see citing works
Hsieh, C.-Y., Li, C.-L., Yeh, C.-K., Nakhost, H., Fujii, Y., Ratner, A., Krishna, R., Lee, C.-Y., & Pfister, T. (2023) · Findings of the Association for Computational Linguistics: ACL 2023 · doi:10.18653/v1/2023.findings-acl.507 · Cited by 316 · see citing works
Gemma: Open Models Based on Gemini Research and Technology
Gemma Team, Mesnard, T., Hardin, C., Dadashi, R., Bhupatiraju, S., Pathak, S., Sifre, L., Rivière, M., Kale, M. S., Love, J., et al. (2024) · arXiv · doi:10.48550/arxiv.2403.08295 · Cited by 284 · see citing works
Large Language Models Can Self-Improve
Huang, J., Gu, S., Hou, L., Wu, Y., Wang, X., Yu, H., & Han, J. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.67 · Cited by 230 · see citing works
Measuring and improving consistency in pretrained language models
Elazar, Y., Kassner, N., Ravfogel, S., Ravichander, A., Hovy, E., Schütze, H., & Goldberg, Y. (2021) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00410 · Cited by 220 · see citing works
Synthetic QA Corpora Generation with Roundtrip Consistency
Alberti, C., Andor, D., Pitler, E., Devlin, J., & Collins, M. (2019) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/p19-1620 · Cited by 208 · see citing works
(2022) · doi:10.52202/068431-1126 · Cited by 182 · see citing works
Are Red Roses Red? Evaluating Consistency of Question-Answering Models
Ribeiro, M. T., Guestrin, C., & Singh, S. (2019) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/p19-1621 · Cited by 93 · see citing works
Distilling Reasoning Capabilities into Smaller Language Models
Shridhar, K., Stolfo, A., & Sachan, M. (2023) · Findings of the Association for Computational Linguistics: ACL 2023 · doi:10.18653/v1/2023.findings-acl.441 · Cited by 79 · see citing works
Logic-Guided Data Augmentation and Regularization for Consistent Question Answering
Asai, A., & Hajishirzi, H. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.499 · Cited by 66 · see citing works
Iterative Reasoning Preference Optimization
Pang, R. Y., Yuan, W., Cho, K., He, H., Sukhbaatar, S., & Weston, J. (2024) · neural information processing systems · doi:10.52202/079017-3702 · Cited by 57 · see citing works
BeliefBank: Adding Memory to a Pre-Trained Language Model for a Systematic Notion of Belief
Kassner, N., Tafjord, O., Schütze, H., & Clark, P. E. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.697 · Cited by 31 · see citing works
LogiCoT: Logical Chain-of-Thought Instruction Tuning
Liu, H., Teng, Z., Cui, L., Zhang, C., Zhou, Q., & Zhang, Y. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.191 · Cited by 21 · see citing works
LLMs May Perform MCQA by Selecting the Least Incorrect OptionPreprint
Wang, H., Zhao, S., Qiang, Z., Xi, N., Qin, B., & Liu, T. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2402.01349 · Cited by 17 · see citing works
How Likely Do LLMs with CoT Mimic Human Reasoning?Preprint
Bao, G., Zhang, H., Wang, C., Yang, L., & Zhang, Y. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2402.16048 · Cited by 16 · see citing works
Iterative Reasoning Preference Optimization
Pang, R. Y., Yuan, W., Cho, K., He, H., Sukhbaatar, S., & Weston, J. (2024) · arXiv · doi:10.48550/arxiv.2404.19733 · Cited by 15 · see citing works
Evaluating Superhuman Models with Consistency Checks
Fluri, L., Paleka, D., & Tramèr, F. (2024) · IEEE Conference on Secure and Trustworthy Machine Learning (SaTML) · doi:10.1109/satml59370.2024.00017 · Cited by 14 · see citing works
Scaling Relationship on Learning Mathematical Reasoning with Large Language Models
Yuan, Z., Yuan, H., Li, C., Dong, G., Lu, K., Tan, C., Zhou, C., & Zhou, J. (2023) · arXiv · doi:10.48550/arxiv.2308.01825 · Cited by 10 · see citing works
Language Models with Rationality
Kassner, N., Tafjord, O., Sabharwal, A., Richardson, K., Schuetze, H., & Clark, P. A. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.877 · Cited by 10 · see citing works
Abstract Meaning Representation-Based Logic-Driven Data Augmentation for Logical Reasoning
Bao, Q., Peng, A. Y., Deng, Z., Zhong, W., Gendron, G., Pistotti, T., Tan, N. Ö., Young, N., Chen, Y., Zhu, Y., et al. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2305.12599 · Cited by 6 · see citing works
Towards Consistent Natural-Language Explanations via Explanation-Consistency FinetuningPreprint
Chen, Y., Singh, C., Liu, X., Zuo, S., Yu, B., He, H., & Gao, J. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2401.13986 · Cited by 6 · see citing works
Evaluating the Rationale Understanding of Critical Reasoning in Logical Reading Comprehension
Kawabata, A., & Sugawara, S. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.9 · Cited by 2 · see citing works
Improving Language Model Reasoning with Self-motivated Learning
Feng, Y., Xu, Y., Qin, L., Wang, Y., & Che, W. (2024) · Language Resources and Evaluation Conference (LREC) · doi:10.63317/2whnfttyog7w · Cited by 1 · see citing works
Self-Explore: Enhancing Mathematical Reasoning in Language Models with Fine-grained Rewards
Hwang, H., Kim, D., Kim, S., Ye, S., & Seo, M. (2024) · arXiv · doi:10.48550/arxiv.2404.10346
Flagged first, then by citations
Research works printed but not linked: 12
- Angelica Chen, Jason Phang, Alicia Parrish, Vishakh Padmakumar, Chen Zhao, Samuel R. Bowman, and Kyunghyun Cho. 2024a. Two failures of self- consistency in the multi-step reasoning of LLMs. Transactions on Machine Learning Research.
- Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Al- bert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdh- ery, Alex Castro-Ros, Marie Pellat, Kevin Robinson, Dasha Valter, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Ja- cob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei. 2024. Scaling instruction-finetuned language models. Journal of Machine Learning Re- search, 25(70):1-53.
- Matthew Douglas Hoffman, Du Phan, david dohan, Sholto Douglas, Tuan Anh Le, Aaron T Parisi, Pavel Sountsov, Charles Sutton, Sharad Vikram, and Rif A. Saurous. 2023. Training chain-of-thought via latent- variable inference. In Thirty-seventh Conference on Neural Information Processing Systems.
- Arian Hosseini, Xingdi Yuan, Nikolay Malkin, Aaron Courville, Alessandro Sordoni, and Rishabh Agarwal. 2024. V-STar: Training verifiers for self-taught rea- soners. In First Conference on Language Modeling.
- Edward J Hu, yelong shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. LoRA: Low-rank adaptation of large language models. In International Conference on Learning Representations.
- Myeongjun Jang, Deuk Sin Kwon, and Thomas Lukasiewicz. 2022. BECEL: Benchmark for con- sistency evaluation of language models. In Proceed- ings of the 29th International Conference on Com- putational Linguistics, pages 3680-3696, Gyeongju, Republic of Korea. International Committee on Com- putational Linguistics.
- Chen Ling, Xuchao Zhang, Xujiang Zhao, Yanchi Liu, Wei Cheng, Mika Oishi, Takao Osaki, Katsushi Mat- suda, Haifeng Chen, and Liang Zhao. 2023. Open- ended commonsense reasoning with unrestricted an- swer candidates. In The 2023 Conference on Empiri- cal Methods in Natural Language Processing.
- 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 Mor- datch, Isabelle Simpson, Izzeddin Gur, Jasper Snoek, Jeffrey Pennington, Jiri Hron, Kathleen Kenealy, Kevin Swersky, Kshiteej Mahajan, Laura A Culp, Lechao Xiao, Maxwell Bileschi, Noah Constant, Ro- man Novak, Rosanne Liu, Tris Warkentin, Yamini Bansal, Ethan Dyer, Behnam Neyshabur, Jascha Sohl- Dickstein, and Noah Fiedel. 2024. Beyond human data: Scaling self-training for problem-solving with language models. Transactions on Machine Learning Research. Expert Certification.
- Yeo Wei Jie, Teddy Ferdinan, Przemyslaw Kazienko, Ranjan Satapathy, and Erik Cambria. 2024. Self- training large language models through knowledge detection. In Findings of the Association for Com- putational Linguistics: EMNLP 2024, pages 15033- 15045, Miami, Florida, USA. Association for Com- putational Linguistics. · DOI
- Zihang Xu, Ziqing Yang, Yiming Cui, and Shijin Wang. 2023. IDOL: Indicator-oriented logic pre-training for logical reasoning. In Findings of the Associa- tion for Computational Linguistics: ACL 2023, pages 8099-8111, Toronto, Canada. Association for Com- putational Linguistics. · DOI
- Seonghyeon Ye, Doyoung Kim, Sungdong Kim, Hyeon- bin Hwang, Seungone Kim, Yongrae Jo, James Thorne, Juho Kim, and Minjoon Seo. 2024. FLASK: Fine-grained language model evaluation based on alignment skill sets. In The Twelfth International Conference on Learning Representations.
- Weihao Yu, Zihang Jiang, Yanfei Dong, and Jiashi Feng. 2020. Reclor: A reading comprehension dataset re- quiring logical reasoning. In International Confer- ence on Learning Representations (ICLR).
Cited by
Sorted by citations
Good Arguments Against the People Pleasers: How Reasoning Mitigates (Yet Masks) LLM Sycophancy
Feng, Z., Chen, Z., Ma, J., Po, Y. T., Chersoni, E., & Li, B. (2026) · Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2026.acl-long.1126 · Cited by 2 · see citing works
Shin, J., Hur, C., Cho, S., Choi, J., & Park, H. (2026) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61042.2026.00738
Kajare, P. V., Priya, P., Santra, B., & Ekbal, A. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.2113
- Authors
- Lee, J., Sakaguchi, K., & Bak, J.
- Affiliations
- Sungkyunkwan University (KR)
- Tohoku University (JP)
- Venue
- Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
- Published
- 2025
- Pages
- 10519-10539
- OpenAlex
- W4411119276
- Cited by
- 2
- References
- 8
- Citations by year
- 2026: 2
- FWCI
- 1.40
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion, cc-by-sa)
- arXiv (Cornell University)
- Funders
- Sungkyunkwan University (RS[2019-II19(0421)])
- Ministry of Science and ICT, South Korea (RS-2019-II190421)
- Institute for Information and Communications Technology Promotion (RS-2019-II190421, 2022-, 2022-0-00680)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Keywords
- self-training, consistency, large language models, reasoning, self-supervised learning
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
