Calibrating Factual Knowledge in Pretrained Language Models
Open accessAbstract
Previous literature has proved that Pretrained Language Models (PLMs) can store factual knowledge.However, we find that facts stored in the PLMs are not always correct.It motivates us to explore a fundamental question: How do we calibrate factual knowledge in PLMs without re-training from scratch?In this work, we propose a simple and lightweight method CA-LINET to achieve this goal.To be specific, we first detect whether PLMs can learn the right facts via a contrastive score between right and fake facts.If not, we then use a lightweight method to add and adapt new parameters to specific factual texts.Experiments on the knowledge probing task show the calibration effectiveness and efficiency.In addition, through closed-book question answering, we find that the calibrated PLM possesses knowledge generalization ability after fine-tuning.Beyond the calibration performance, we further investigate and visualize the knowledge calibration mechanism.The code and data are available at https://github.com/dqxiu/CaliNet.
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?
Memory-based approaches require training to update memory representations or routing, and memory consumption grows linearly with the number of edits, limiting practicality in lifelong model editing.
“However, these methods still require training to update memory representations or routing, and inference must be rerouted through the memory. In addition, they maintain one entry per edit, causing memory consumption to grow linearly with the number of edits, which limits their practicality in lifelong model editing with high-frequency or large-scale updates.”
From UltraEdit: Training-, Subject-, and Memory-Free Lifelong Editing in Language Models (2025) · cites this paper as Dong et al., 2022
Section 3.3 (page not provided) says CALINET is trained by updating calibration memory slots, supporting the training portion of the claim. Section 4.2.5 (page not provided) reports that 256 slots are nearly sufficient for 1,000 facts, rather than one entry per edit. Our copy does not establish inference rerouting, linear memory growth, or a resulting limit on lifelong editing. (LLM, conf=0.84)
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
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929 · see citing works
Don’t Stop Pretraining: Adapt Language Models to Domains and Tasks
Gururangan, S., Marasović, A., Swayamdipta, S., Lo, K., Beltagy, I., Downey, D., & Smith, N. A. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.740 · Cited by 2,103 · see citing works
TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension
Joshi, M., Choi, E., Weld, D., & Zettlemoyer, L. (2017) · Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/p17-1147 · Cited by 1,690 · 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
Editing Factual Knowledge in Language Models
De Cao, N., Aziz, W., & Titov, I. (2021) · Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2021.emnlp-main.522 · Cited by 180 · see citing works
Flagged first, then by citations
Research works printed but not linked: 22
- Jonathan Berant, Andrew Chou, Roy Frostig, and Percy Liang. 2013. Semantic parsing on Freebase from question-answer pairs. In Proceedings of EMNLP.
- Zied Bouraoui, José Camacho-Collados, and Steven Schockaert. 2020. Inducing relational knowledge from bert. In Proceedings of AAAI.
- Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. Language models are few-shot learners. In Proceedings of NeurIPS.
- Boxi Cao, Hongyu Lin, Xianpei Han, Le Sun, Lingy- ong Yan, Meng Liao, Tong Xue, and Jin Xu. 2021a. 5974 Knowledgeable or educated guess? revisiting lan- guage models as knowledge bases. In Proceedings of ACL.
- Damai Dai, Li Dong, Yaru Hao, Zhifang Sui, Baobao Chang, and Furu Wei. 2022. Knowledge neurons in pretrained transformers. In Proceedings of ACL.
- Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understand- ing. In Proceedings of NAACL.
- Hady ElSahar, Pavlos Vougiouklis, Arslen Remaci, Christophe Gravier, Jonathon S. Hare, Frédérique Laforest, and Elena Simperl. 2018. T-rex: A large scale alignment of natural language with knowledge base triples. In Proceedings of LREC.
- Mor Geva, Avi Caciularu, Kevin Ro Wang, and Yoav Goldberg. 2022. Transformer feed-forward layers build predictions by promoting concepts in the vo- cabulary space. CoRR, abs/2203.14680. · arXiv
- Mor Geva, Roei Schuster, Jonathan Berant, and Omer Levy. 2021. Transformer feed-forward layers are key-value memories. In Proceedings of EMNLP.
- Zhengbao Jiang, Frank F. Xu, Jun Araki, and Graham Neubig. 2020. How can we know what language models know? TACL.
- Nayeon Lee, Belinda Z. Li, Sinong Wang, Wen-tau Yih, Hao Ma, and Madian Khabsa. 2020. Language models as fact checkers? In Proceedings of FEVER.
- Kevin Meng, David Bau, Alex Andonian, and Yonatan Belinkov. 2022. Locating and editing factual knowl- edge in GPT. CoRR, abs/2202.05262. · arXiv
- Eric Mitchell, Charles Lin, Antoine Bosselut, Chelsea Finn, and Christopher D Manning. 2022. Fast model editing at scale. In Proceedings of ICLR.
- Matthew E. Peters, Mark Neumann, Robert Logan, Roy Schwartz, Vidur Joshi, Sameer Singh, and Noah A. Smith. 2019. Knowledge enhanced contextual word representations. In Proceedings of EMNLP-IJCNLP.
- Fabio Petroni, Tim Rocktäschel, Sebastian Riedel, Patrick Lewis, Anton Bakhtin, Yuxiang Wu, and Alexander Miller. 2019. Language models as knowl- edge bases? In Proceedings of EMNLP-IJCNLP.
- Nina Pörner, Ulli Waltinger, and Hinrich Schütze. 2019. Bert is not a knowledge base (yet): Factual knowl- edge vs. name-based reasoning in unsupervised qa. CoRR, abs/1911.03681. · arXiv
- Adam Roberts, Colin Raffel, and Noam Shazeer. 2020a. How much knowledge can you pack into the parame- ters of a language model? In Proceedings of EMNLP.
- Adam Roberts, Colin Raffel, and Noam Shazeer. 2020b. How much knowledge can you pack into the parame- ters of a language model? In Proceedings of EMNLP.
- Ruize Wang, Duyu Tang, Nan Duan, Zhongyu Wei, Xuanjing Huang, Jianshu Ji, Guihong Cao, Daxin Jiang, and Ming Zhou. 2021. K-adapter: Infusing knowledge into pre-trained models with adapters. In Findings of ACL-IJCNLP, pages 1405-1418.
- Yunzhi Yao, Shaohan Huang, Ningyu Zhang, Li Dong, Furu Wei, and Huajun Chen. 2022. Kformer: Knowl- edge injection in transformer feed-forward layers. CoRR, abs/2201.05742. · arXiv
- Zhengyan Zhang, Xu Han, Zhiyuan Liu, Xin Jiang, Maosong Sun, and Qun Liu. 2019. Ernie: Enhanced language representation with informative entities. In Proceedings of ACL.
- Chen Zhu, Ankit Singh Rawat, Manzil Zaheer, Srinadh Bhojanapalli, Daliang Li, Felix X. Yu, and Sanjiv Kumar. 2020. Modifying memories in transformer models. CoRR, abs/2012.00363. · arXiv
Cited by
Sorted by citations
Kabir, S., Udo-Imeh, D. N., Kou, B., & Zhang, T. (2024) · Proceedings of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642596 · Cited by 160 · see citing works
The Life Cycle of Knowledge in Big Language Models: A Survey
Cao, B., Lin, H., Han, X., & Sun, L. (2023) · arXiv · doi:10.48550/arxiv.2303.07616 · Cited by 18
UltraEdit: Training-, Subject-, and Memory-Free Lifelong Editing in Language Models
Gu, X., Huang, Z., Gu, J.-C., & Zhang, K. (2025) · arXiv · doi:10.48550/arxiv.2505.14679
The AI Hippocampus: How Far are We From Human Memory?
Jia, Z., Li, J., Kang, Y., Wang, Y., Wu, T., Wang, Q., Wang, X., Zhang, S., Shen, J., Li, Q., et al. (2026) · arXiv · doi:10.48550/arxiv.2601.09113
- Authors
- Dong, Q., Dai, D., Song, Y., Xu, J., Sui, Z., & Li, L.
- Affiliations
- Peking University (CN)
- Shanghai Artificial Intelligence Laboratory (CN)
- University of California, Santa Barbara (US)
- Venue
- Findings of the Association for Computational Linguistics: EMNLP 2022
- Published
- 2022
- Pages
- 5937-5947
- OpenAlex
- W4385574113
- Cited by
- 39
- References
- 5
- Citations by year
- 2026: 8 · 2025: 12 · 2024: 8 · 2023: 11
- FWCI
- 4.29
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) (publishedVersion, cc-by)
- Funders
- National Natural Science Foundation of China (U19A2065)
- National Key Research and Development Program of China (2020AAA0106700)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- pre-trained language models, knowledge probing, contrastive learning, knowledge generalization, closed-book question answering, parameter adaptation
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
