Benchmarking Uncertainty Quantification Methods for Large Language Models with LM-Polygraph
Open accessThe DOI matches its Crossref record; 78 of the 78 references it cites were found in a registry or an index; 1 of the 20 sources its claims cite were checked for support.
Abstract
The rapid proliferation of large language models (LLMs) has stimulated researchers to seek effective and efficient approaches to deal with LLM hallucinations and low-quality outputs. Uncertainty quantification (UQ) is a key element of machine learning applications in dealing with such challenges. However, research to date on UQ for LLMs has been fragmented in terms of techniques and evaluation methodologies. In this work, we address this issue by introducing a novel benchmark that implements a collection of state-of-the-art UQ baselines and offers an environment for controllable and consistent evaluation of novel UQ techniques over various text generation tasks. Our benchmark also supports the assessment of confidence normalization methods in terms of their ability to provide interpretable scores. Using our benchmark, we conduct a large-scale empirical investigation of UQ and normalization techniques across eleven tasks, identifying the most effective approaches.
Overview
CoChat Summary
TL;DR
The paper introduces LM-Polygraph, a benchmark for consistent evaluation of uncertainty quantification (UQ) and confidence normalization methods for LLM text generation across selective QA/generation, claim-level fact-checking, and calibration. A large-scale empirical study finds sample-diversity methods strong overall, information-based methods best for short outputs, CCP best for claim-level fact-checking, and isotonic PCC useful for interpretable confidence without degrading ranking performance.
Key findings
- In selective QA/generation, sample-diversity methods generally perform well, especially for longer outputs; for short generations (<7 symbols), information-based methods such as MSP, Perplexity, Mean Token Entropy, CCP, and TokenSAR perform best. SAR is highlighted as robust across output lengths.
- For claim-level fact-checking, CCP achieves the best ROC-AUC/PR-AUC for English Mistral 7B (0.74/0.46), English Jais 13B (0.68/0.47), Arabic Jais 13B (0.73/0.30), and Russian Vikhr 7B (0.67/0.80); for Chinese Yi 6B, CCP and P(True) are comparable (both 0.61 ROC-AUC, 0.25 PR-AUC).
- MSP is a strong and robust baseline across tasks and should not be discarded; P(True) often performs poorly, especially with smaller LLMs.
- Binned and isotonic PCC confidently outperform linear and quantile normalization for calibration (MSE), and isotonic PCC preserves PRR relative to raw uncertainty scores (Table 13 shows mostly 0.000 difference).
Methods
- The benchmark is built on the LM-Polygraph framework and implements white-box and black-box UQ baselines, claim-level extensions, and normalization methods including linear scaling, quantile scaling, Binned PCC, and Isotonic PCC.
- Selective QA/generation uses CoQA, TriviaQA, MMLU, GSM8k, XSum, WMT14 Fr-En, and WMT19 De-En; claim-level fact-checking uses automatically generated biographies in English, Arabic, Russian, and Chinese, with GPT-4 for claim extraction, token mapping, and annotation.
- Models include Mistral 7B v0.2, Stable LM 2 12B, GPT-4o-mini, Jais 13B, Vikhr 7B-0.2, and Yi 7B/6B; evaluation metrics include PRR, ROC-AUC/PR-AUC, and MSE between normalized confidence and quality.
Limitations
- The paper does not include a dedicated limitations section; it states that about 5% of claims cannot be mapped to tokens and are excluded, and that no more than 8% of claims were classified as unknown and discarded.
- GPT-4 is used as an automatic annotator and judge; the paper validates it against manual annotation (e.g., English Mistral-7B-v0.1 accuracy 0.98, F1 0.93; Fleiss Kappa 0.90), but this remains a proxy for human judgment.
- Evident from the text: results are tied to the specific models, datasets, languages, and prompt formats tested, so generalization to other LLMs, tasks, or domains is not established.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
1 Valid · 1 Not valid · 16 Not checked · 78 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“We argue that such an approach introduces a big discrepancy between the generated text and what a proxy LLM actually wants to generate, which results in biased UQ performance.”
What the sources say
- Supportedcites Manakul et al., 2023 matched to “SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models” (2023) source
The Proxy LLMs section (p. 3) describes using another LLM to approximate the generating model’s token probabilities. Section 7.1 (p. 6) reports worse proxy-LLM detection performance and suggests that differing generation patterns can assign low probabilities to the generated text. The citing authors frame the resulting bias as their own argument, rather than attributing that conclusion directly to Manakul et al. (LLM, conf=0.88)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Despite being a simplistic baseline, MSP demonstrates strong performance.”
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, for MMLU, these methods fall significantly behind informationbased techniques.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“State-of-the-art methods based on sample diversity typically perform well on CoQA and TriviaQA.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The majority of density-based methods demonstrate poor results across all tasks.”
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
34 more claims
Create a free account to see all 96 claims and to verify the 19 references still open.
- It may state two separate claims as one. 31
- The quote only partly states this claim. 1
- Probably not a substantial claim. 36
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?
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“Notably, there is growing attention on assessing LLM bias [63-64] and robustness [323-324].”
From A Survey of Large Language Models (2026) · cites this paper as [323]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 5.4 Models · p. 9
“Figure 1: Mean PRR ↑ aggregated over all selective QA tasks for each white-box LLM separately (the lower the better).”
§ 5.1 Selective QA / Generation · p. 7
“A higher PRR indicates a better uncertainty score.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 6.2 Fact-Checking · p. 11
“In the fact-checking evaluation, we generated biographies for 100 individuals across five languages: English (using Mistral 7B-v0.1 and Jais 13B), Arabic (using Jais 13B), Russian (using Vikhr 7B-0.2), and Chinese (using Yi 6B).”
§ 5.2 Claim-Level Fact-Checking · p. 8
“The pipeline supports English, Mandarin Chinese, Arabic, and Russian.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 5.4 Models · p. 9
“Yi 7B (for Chinese).”
§ 6.2 Fact-Checking · p. 11
“Chinese (using Yi 6B).”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 6.1 Selective QA and Generation · p. 11
“For both models, on the text summarization task, the best results are achieved by sample-diversity techniques in terms of both metrics.”
§ 6.1 Selective QA and Generation · p. 11
“One exception is the performance of Conditional PMI in terms of ROUGE-L-based PRR, which is drastically better than the performance of all other methods.”
Citations
Cites
Advances in Neural Information Processing Systems 19
(2007) · The MIT Press eBooks · doi:10.52202/085713 · Cited by 15,539
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 · cited 2×
“Ren et al. (2023) showed that it might be useful to adjust the Mahalanobis distance score by subtracting from it the other Mahalanobis distance MD 0 ( x ) computed for some large general purpose dataset covering many domains like C4 (Raffel et al., 2020).”
Least Median of Squares Regression
Rousseeuw, P. J. (1984) · Journal of the American Statistical Association · doi:10.1080/01621459.1984.10477105 · Cited by 3,835 · cited 1×
“Additionally, computing of the covariance matrix Σ for each individual class is done by us-ing the Minimum Covariance Determinant estimation (Rousseeuw, 1984).”
A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks
Lee, K., Lee, K., Lee, H., & Shin, J. (2018) · arXiv · doi:10.48550/arxiv.1807.03888 · Cited by 2,450 · cited 2×
“Several variations of these methods have been proposed recently (Lee et al., 2018; Yoo et al., 2022; Kotelevskii et al., 2022; Ren et al., 2023).”
DEBERTA: DECODING-ENHANCED BERT WITH DISENTANGLED ATTENTION
He, P., Liu, X., Gao, J., & Chen, W. (2021) · International Conference on Learning Representations · Cited by 2,445
Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
Zheng, L., Chiang, W.-L., Sheng, Y., Zhuang, S., Wu, Z., Zhuang, Y., Lin, Z., Li, Z., Li, D., Xing, E. P., et al. (2023) · arXiv · doi:10.48550/arxiv.2306.05685 · Cited by 1,888 · cited 1×
“We provide a more detailed discussion on the specifics of the normalization methods in Appendix B.”
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 · cited 1×
“The first two datasets are used in many works on UQ, while the last two are popular datasets for evaluating the English-language generation quality of LLMs.”
TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension
Joshi, M., Choi, E., Weld, D. S., & Zettlemoyer, L. (2017) · arXiv · doi:10.48550/arxiv.1705.03551 · Cited by 1,650 · cited 1×
“The first two datasets are used in many works on UQ, while the last two are popular datasets for evaluating the English-language generation quality of LLMs.”
Weight Uncertainty in Neural Network
Blundell, C., Cornebise, J., Kavukcuoglu, K., & Wierstra, D. (2015) · International Conference on Machine Learning · Cited by 1,335
Narayan, S., Cohen, S. B., & Lapata, M. (2018) · Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d18-1206 · Cited by 1,207 · cited 1×
“For selective generation, we use two MT datasets: WMT-14 French to English (Bojar et al., 2014), WMT-19 German to English (Barrault et al., 2019), and one text summarization dataset – XSum (Narayan et al., 2018).”
CoQA: A Conversational Question Answering Challenge
Reddy, S., Chen, D., & Manning, C. D. (2019) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00266 · Cited by 975 · cited 1×
“Our benchmark features three sections: (1) evaluation of the UQ performance in selective classification/generation; (2) evaluation of UQ performance in fact-checking; and (3) evaluation of confidence calibration.”
Deep Bayesian Active Learning with Image Data
Gal, Y., Islam, R., & Ghahramani, Z. (2017) · arXiv · doi:10.48550/arxiv.1703.02910 · Cited by 900 · cited 1×
“There has also been a surge of research devoted to UQ specifically in the context of encoder-only language models (LMs) such as BERT (Zhang et al., 2019; He et al., 2020; Shelmanov et al., 2021; Xin et al., 2021a; Vazhentsev et al., 2022; Kotelevskii et al., 2022; Wang et al., 2022; Kuzmin et al.,…”
Kuhn, L., Gal, Y., & Farquhar, S. (2023) · arXiv · doi:10.48550/arxiv.2302.09664 · Cited by 700
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 · cited 1×
“Building an efficient external system to verify LLM output for every possible task is currently infeasible.”
COMET: A Neural Framework for MT Evaluation
Rei, R., Stewart, C., Farinha, A. C., & Lavie, A. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.213 · Cited by 617
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 · cited 1×
“However, this raises the problem that each generation is unique and needs to be reannotated, posing a significant challenge for human annotation.”
SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models
Manakul, P., Liusie, A., & Gales, M. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.557 · Cited by 557
Findings of the 2014 Workshop on Statistical Machine Translation
Bojar, O., Buck, C., Federmann, C., Haddow, B., Koehn, P., Leveling, J., Monz, C., Pecina, P., Post, M., Saint-Amand, H., et al. (2014) · Proceedings of the Ninth Workshop on Statistical Machine Translation · doi:10.3115/v1/w14-3302 · Cited by 529 · cited 1×
“We also reserve 1,000 instances from the training set for UQ techniques that require “pre-training”, such as density-based methods.”
Socratic guiding questions in synthetic arithmetic data: matched LoRA runs (revision v2)
O'Grady, J., Gürhan, A. I., Chee, F. T., & Ramlan, E. I. (2026) · Zenodo (CERN European Organization for Nuclear Research) · doi:10.5281/zenodo.23236017 · Cited by 462
Findings of the 2019 Conference on Machine Translation (WMT19)
Barrault, L., Bojar, O., Costa-jussà, M. R., Federmann, C., Fishel, M., Graham, Y., Haddow, B., Huck, M., Koehn, P., Malmasi, S., et al. (2019) · Proceedings of the Fourth Conference on Machine Translation (Volume 2: Shared Task Papers, Day 1) · doi:10.18653/v1/w19-5301 · Cited by 461 · cited 1×
“More detailed statistics about the datasets are given in Appendix D.”
(2021) · International Conference on Learning Representations · Cited by 435
Confident Adaptive Language ModelingPreprint
Schuster, T., Fisch, A., Gupta, J., et al. (2022) · Neural Information Processing Systems · doi:10.48550/arxiv.2207.07061 · Cited by 419
Sengupta, N., Sahu, S. K., Jia, B., Katipomu, S., Li, H., Koto, F., Marshall, W., Gosal, G., Liu, C., Chen, Z., et al. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2308.16149 · Cited by 378
Self-citationDeeBERT: Dynamic Early Exiting for Accelerating BERT Inference
Xin, J., Tang, R., Lee, J., Yu, Y., & Lin, J. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.204 · Cited by 329
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
Rafailov, R., Sharma, A., Mitchell, E., Ermon, S., Manning, C. D., & Finn, C. (2023) · arXiv · doi:10.48550/arxiv.2305.18290 · Cited by 293 · cited 1×
“Building an efficient external system to verify LLM output for every possible task is currently infeasible.”
Language Models (Mostly) Know What They Know
Kadavath, S., Conerly, T., Askell, A., Henighan, T., Drain, D., Perez, E., Schiefer, N., Hatfield-Dodds, Z., DasSarma, N., Tran-Johnson, E., et al. (2022) · arXiv · doi:10.48550/arxiv.2207.05221 · Cited by 187
Pitfalls of In-Domain Uncertainty Estimation and Ensembling in Deep Learning
Ashukha, A., Lyzhov, A., Molchanov, D., & Vetrov, D. (2020) · arXiv · doi:10.48550/arxiv.2002.06470 · Cited by 178 · cited 1×
“For example, one of the popular approaches to UQ is ensembling (Ashukha et al., 2019).”
Understanding Measures of Uncertainty for Adversarial Example Detection
Smith, L., & Gal, Y. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1803.08533 · Cited by 162 · cited 1×
“…has also been a surge of research devoted to UQ specifically in the context of encoder-only language models (LMs) such as BERT (Zhang et al., 2019; He et al., 2020; Shelmanov et al., 2021; Xin et al., 2021a; Vazhentsev et al., 2022; Kotelevskii et al., 2022; Wang et al., 2022; Kuzmin et al., 2023).”
Enhancing Uncertainty-Based Hallucination Detection with Stronger Focus
Zhang, T., Qiu, L., Guo, Q., Deng, C., Zhang, Y., Zhang, Z., Zhou, C., Wang, X., & Fu, L. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.58 · Cited by 158
Stable LM 2 1.6B Technical ReportPreprint
Bellagente, M., Tow, J., Mahan, D., Phung, D., Zhuravinskyi, M., Adithyan, R., Baicoianu, J., Brooks, B., Cooper, N., Datta, A., et al. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2402.17834 · Cited by 158
Out-of-Distribution Detection and Selective Generation for Conditional Language ModelsPreprint
Ren, J., Luo, J., Zhao, Y., Krishna, K. S., Saleh, M., Lakshminarayanan, B., & Liu, P. J. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2209.15558 · Cited by 148 · cited 4×
“In this work, we strive to bridge these disparate research efforts and to resolve some issues found in their evaluation protocols by developing a benchmark for UQ techniques in text generation tasks.”
InfluentialThe Right Tool for the Job: Matching Model and Instance Complexities
Schwartz, R., Stanovsky, G., Swayamdipta, S., Dodge, J., & Smith, N. A. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.593 · Cited by 128
LUQ: Long-text Uncertainty Quantification for LLMs
Zhang, C., Liu, F., Basaldella, M., & Collier, N. (2024) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2024.emnlp-main.299 · Cited by 124
Tian, K., Mitchell, E., Zhou, A., Sharma, A., Rafailov, R., Yao, H., Finn, C., & Manning, C. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.330 · Cited by 123
On the Origin of Hallucinations in Conversational Models: Is it the Datasets or the Models?
Dziri, N., Milton, S., Yu, M., Zaiane, O., & Reddy, S. (2022) · Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies · doi:10.18653/v1/2022.naacl-main.387 · Cited by 121
A framework for few-shot language model evaluation
Sutawika, L., Gao, L., Schoelkopf, H., Biderman, S., Tow, J., Abbasi, B., fattori, B., Lovering, C., farzanehnakhaee70, Phang, J., et al. (2021) · Zenodo (CERN European Organization for Nuclear Research) · doi:10.5281/zenodo.5371628 · Cited by 121
Revisiting Mahalanobis Distance for Transformer-Based Out-of-Domain Detection
Podolskiy, A., Lipin, D., Bout, A., Artemova, E., & Piontkovskaya, I. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i15.17612 · Cited by 113
Mutual Information Alleviates Hallucinations in Abstractive Summarization
van der Poel, L., Cotterell, R., & Meister, C. (2022) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2022.emnlp-main.399 · Cited by 105
Confident adaptive language modeling
Schuster, T., Fisch, A., Gupta, J., Dehghani, M., Bahri, D., Tran, V., Tay, Y., & Metzler, D. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1269 · Cited by 91
AlignScore: Evaluating factual consistency with a unified alignment function
Zha, Y., Yang, Y., Li, R., & Hu, Z. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.634 · Cited by 81
Active Learning for Sequence Tagging with Deep Pre-trained Models and Bayesian Uncertainty Estimates
Shelmanov, A., Puzyrev, D., Kupriyanova, L., Belyakov, D., Larionov, D., Khromov, N., Kozlova, O., Artemova, E., Dylov, D. V., & Panchenko, A. (2021) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2021.eacl-main.145 · Cited by 81
Self-citationOn hallucination and predictive uncertainty in conditional language generation
Xiao, Y., & Wang, W. Y. (2021) · Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume · doi:10.18653/v1/2021.eacl-main.236 · Cited by 79
Fast Inference from Transformers via Speculative Decoding
Leviathan, Y., Kalman, M., & Matias, Y. (2022) · arXiv · doi:10.48550/arxiv.2211.17192 · Cited by 71 · cited 1×
“The rapid proliferation of large language models (LLMs) has stimulated researchers to seek efficient and effective approaches to UQ in text generation tasks, in an attempt to make them safer to use in downstream applications.”
Uncertainty Estimation of Transformer Predictions for Misclassification Detection
Vazhentsev, A., Kuzmin, G., Shelmanov, A., Tsvigun, A., Tsymbalov, E., Fedyanin, K., Panov, M., Panchenko, A., Gusev, G., Burtsev, M., et al. (2022) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2022.acl-long.566 · Cited by 70
Self-citationNonparametric Uncertainty Quantification for Single Deterministic Neural Network
Kotelevskii, N., Artemenkov, A., Fedyanin, K., Noskov, F., Fishkov, A., Shelmanov, A., Vazhentsev, A., Petiushko, A., & Panov, M. (2022) · neural information processing systems · doi:10.52202/068431-2631 · Cited by 66 · cited 2×
“Finally, different methods might be restricted only to UQ of the whole text (sequence level), while others might be also applicable to text fragments such as atomic claims (claim level).”
Self-citationHow Certain is Your Transformer?
Shelmanov, A., Tsymbalov, E., Puzyrev, D., Fedyanin, K., Panchenko, A., & Panov, M. (2021) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2021.eacl-main.157 · Cited by 64 · cited 2×
“One of the main research questions that has been addressed in recent work related to classification and UQ in general is how to perform it efficiently, while keeping the performance of the scores reliably high (Shelmanov et al., 2021).”
Self-citationHybrid Uncertainty Quantification for Selective Text Classification in Ambiguous Tasks
Vazhentsev, A., Kuzmin, G., Tsvigun, A., Panchenko, A., Panov, M., Burtsev, M., & Shelmanov, A. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.652 · Cited by 58
Self-citationTowards More Accurate Uncertainty Estimation In Text Classification
He, J., Zhang, X., Lei, S., Chen, Z., Chen, F., Alhamadani, A., Xiao, B., & Lu, C. (2020) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.671 · Cited by 55 · cited 1×
“…has also been a surge of research devoted to UQ specifically in the context of encoder-only language models (LMs) such as BERT (Zhang et al., 2019; He et al., 2020; Shelmanov et al., 2021; Xin et al., 2021a; Vazhentsev et al., 2022; Kotelevskii et al., 2022; Wang et al., 2022; Kuzmin et al.,…”
Understanding and Detecting Hallucinations in Neural Machine Translation via Model Introspection
Xu, W., Agrawal, S., Briakou, E., Martindale, M. J., & Carpuat, M. (2023) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00563 · Cited by 53 · cited 1×
“For instance, fact-checkers target a very narrow sub-problem, and usually require ex-ternal knowledge sources such as knowledge bases that are generally incomplete.”
Mitigating Uncertainty in Document Classification
Zhang, X., Chen, F., Lu, C., & Ramakrishnan, N. (2019) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/n19-1316 · Cited by 50 · cited 1×
“As with any ML model, LLMs can make incorrect predictions, “hallucinate” by fabricating claims (Xiao and Wang, 2021; Dziri et al., 2022), or simply generate low-quality outputs.”
Uncertainty Estimation in Autoregressive Structured Prediction
Gales, M., & Malinin, A. (2021) · Apollo - University of Cambridge Repository · doi:10.17863/cam.63497 · Cited by 48 · cited 7×
“To facilitate this, we propose two methods of such calibration, which we call Performance-Calibrated Confidence (PCC).”
InfluentialUncertainty Estimation in Autoregressive Structured Prediction
Malinin, A., & Gales, M. (2020) · arXiv · doi:10.48550/arxiv.2002.07650 · Cited by 48
Centered Isotonic Regression: Point and Interval Estimation for Dose–Response Studies
Oron, A. P., & Flournoy, N. (2017) · Statistics in Biopharmaceutical Research · doi:10.1080/19466315.2017.1286256 · Cited by 46 · cited 3×
“It fits Centered Isotonic Regression (CIR) (Oron and Flournoy, 2017) on pairs of uncertainty and quality values from the calibration set.”
InfluentialThe Art of Abstention: Selective Prediction and Error Regularization for Natural Language Processing
Xin, J., Tang, R., Yu, Y., & Lin, J. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.84 · Cited by 45
ParaDetox: Detoxification with Parallel Data
Logacheva, V., Dementieva, D., Ustyantsev, S., Moskovskiy, D., Dale, D., Krotova, I., Semenov, N., & Panchenko, A. (2022) · Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2022.acl-long.469 · Cited by 44
Self-citationEfficient Out-of-Domain Detection for Sequence to Sequence Models
Vazhentsev, A., Tsvigun, A., Vashurin, R., Petrakov, S., Vasilev, D., Panov, M., Panchenko, A., & Shelmanov, A. (2023) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2023.findings-acl.93 · Cited by 44 · cited 3×
“There has also been a surge of research devoted to UQ specifically in the context of encoder-only language models (LMs) such as BERT (Zhang et al., 2019; He et al., 2020; Shelmanov et al., 2021; Xin et al., 2021a; Vazhentsev et al., 2022; Kotelevskii et al., 2022; Wang et al., 2022; Kuzmin et al.,…”
Self-citationAccelerating Large Language Model Decoding with Speculative Sampling
Chen, C., Borgeaud, S., Irving, G., Lespiau, J.-B., Sifre, L., & Jumper, J. (2023) · arXiv · doi:10.48550/arxiv.2302.01318 · Cited by 40 · cited 1×
“The rapid proliferation of large language models (LLMs) has stimulated researchers to seek efficient and effective approaches to UQ in text generation tasks, in an attempt to make them safer to use in downstream applications.”
Duan, J., Cheng, H., Wang, S., Zavalny, A., Wang, C., Xu, R., Kailkhura, B., & Xu, K. (2024) · Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2024.acl-long.276 · Cited by 40 · cited 3×
“…u i = U ( x i ) as well as the generation quality f ( x i ) : Among simple normalization approaches we consider linear scaling and quantile scaling , as they provide simple rules to normalize uncertainty scores in [0 , 1] based solely on uncertainty values u i ; see Appendix B for more detail.”
SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models
Manakul, P., Liusie, A., & Gales, M. J. F. (2023) · arXiv · doi:10.48550/arxiv.2303.08896 · Cited by 38
Generating with Confidence: Uncertainty Quantification for Black-box Large Language Models
Lin, Z., Trivedi, S., & Sun, J. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2305.19187 · Cited by 35
Yoo, K., Kim, J., Jang, J., & Kwak, N. (2022) · Findings of the Association for Computational Linguistics: ACL 2022 · doi:10.18653/v1/2022.findings-acl.289 · Cited by 33
Quantifying Aleatoric and Epistemic Uncertainty with Proper Scoring RulesPreprint
Hofman, P., Sale, Y., & Hüllermeier, E. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2404.12215 · Cited by 31 · cited 1×
“Despite recent efforts to establish a common definition of predictive uncertainty (Kotelevskii and Panov, 2024; Hofman et al., 2024), multiple approaches to its quantification exist based on probabilities, entropies, distances, risks, etc.”
Incorporating uncertainty into deep learning for spoken language assessment
Malinin, A., Ragni, A., Knill, K., & Gales, M. (2017) · Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) · doi:10.18653/v1/p17-2008 · Cited by 29 · cited 2×
“These pairs of uncertainty thresholds and associated average output quality give a prediction–rejection curve, and the area under this curve demonstrates the overall quality of a UQ method (Malinin et al., 2017).”
Fact-checking the output of large language models via token-level uncertainty quantification
Fadeeva, E., Rubashevskii, A., Shelmanov, A., Petrakov, S., Li, H., Mubarak, H., Tsymbalov, E., Kuzmin, G., Panchenko, A., Baldwin, T., et al. (2024) · Findings of the Association for Computational Linguistics ACL 2024 · doi:10.18653/v1/2024.findings-acl.558 · Cited by 29
Self-citationTowards Computationally Feasible Deep Active Learning
Tsvigun, A., Shelmanov, A., Kuzmin, G., Sanochkin, L., Larionov, D., Gusev, G., Avetisian, M., & Евгеньевич, Ж. Л. (2022) · Findings of the Association for Computational Linguistics: NAACL · doi:10.18653/v1/2022.findings-naacl.90 · Cited by 26
Self-citationRainProof: An Umbrella to Shield Text Generator from Out-Of-Distribution Data
Darrin, M., Piantanida, P., & Colombo, P. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.357 · Cited by 24
Uncertainty Estimation and Reduction of Pre-trained Models for Text Regression
Wang, Y., Beck, D., Baldwin, T., & Verspoor, K. (2022) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00483 · Cited by 23 · cited 1×
“These problems stem partially from the peculiarities of the LLM training objec-tive, and are partially inherited from the nature of ML models in general.”
Self-citationLM-Polygraph: Uncertainty estimation for language models
Fadeeva, E., Vashurin, R., Tsvigun, A., Vazhentsev, A., Petrakov, S., Fedyanin, K., Vasilev, D., Goncharova, E., Panchenko, A., Panov, M., et al. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations · doi:10.18653/v1/2023.emnlp-demo.41 · Cited by 20
Self-citationRelevant and Informative Response Generation using Pointwise Mutual Information
Takayama, J., & Arase, Y. (2019) · Workshop on NLP for Conversational AI (NLP4ConvAI) · doi:10.18653/v1/w19-4115 · Cited by 19 · cited 1×
“Finally, one can consider negative mean Point-wise Mutual Information (PMI; Takayama and Arase (2019)) which is given by This method was extended in (van der Poel et al., 2022) by considering only those marginal probabilities for which the entropy of the conditional distribution is above certain…”
On the Origin of Hallucinations in Conversational Models: Is it the Datasets or the Models?
Dziri, N., Milton, S., Yu, M., Zaiane, O., & Reddy, S. (2022) · arXiv · doi:10.48550/arxiv.2204.07931 · Cited by 13 · cited 1×
“These include filtering using stop-word lists, post-processing using classifiers (Xu et al., 2023), rewriting toxic outputs (Logacheva et al., 2022), and extended fine-tuning with reinforcement learning from human feedback (RLHF).”
From Risk to Uncertainty: Generating Predictive Uncertainty Measures via Bayesian EstimationPreprint
Kotelevskii, N., Kondratyev, V., Takáč, M., Moulines, É., & Panov, M. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2402.10727 · Cited by 12
Self-citationUnsupervised Quality Estimation for Neural Machine Translation
Fomicheva, M., Sun, S., Yankovskaya, L., Blain, F., Guzmán, F., Fishel, M., Aletras, N., Chaudhary, V., & Specia, L. (2020) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00330 · Cited by 10 · cited 3×
“In this paper, we consider several black-box approaches, including Lexical Similarity, Number of Semantic Sets, Sum of Eigen-values of the Graph Laplacian, Degree Matrix, and Eccentricity, all of which have been shown to perform well in other studies (Fomicheva et al., 2020; Kuhn et al., 2023).”
Nikolich, A., Korolev, K., Bratchikov, S., Kiselev, I., & Shelmanov, A. (2024) · Workshop on Multilingual Representation Learning (MRL) · doi:10.18653/v1/2024.mrl-1.15 · Cited by 10
Self-citationUncertainty Estimation for Debiased Models: Does Fairness Hurt Reliability?
Kuzmin, G., Vazhentsev, A., Shelmanov, A., Han, X., Šuster, S., Panov, M., Panchenko, A., & Baldwin, T. (2023) · International Joint Conference on Natural Language Processing and Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (IJCNLP-AACL) · doi:10.18653/v1/2023.ijcnlp-main.48 · Cited by 6 · cited 1×
“…has also been a surge of research devoted to UQ specifically in the context of encoder-only language models (LMs) such as BERT (Zhang et al., 2019; He et al., 2020; Shelmanov et al., 2021; Xin et al., 2021a; Vazhentsev et al., 2022; Kotelevskii et al., 2022; Wang et al., 2022; Kuzmin et al., 2023).”
Self-citationTian, K., Mitchell, E., Zhou, A., Sharma, A., Rafailov, R., Yao, H., Finn, C., & Manning, C. D. (2023) · arXiv · doi:10.48550/arxiv.2305.14975 · Cited by 5
Scalable Batch Acquisition for Deep Bayesian Active Learning
Rubashevskii, A., Kotova, D., & Panov, M. (2023) · Society for Industrial and Applied Mathematics eBooks · doi:10.1137/1.9781611977653.ch83 · Cited by 3
Self-citationFactcheck-Bench: Fine-Grained Evaluation Benchmark for Automatic Fact-checkers
Wang, Y., Reddy, R. G., Mujahid, Z. M., Arora, A., Rubashevskii, A., Geng, J., Afzal, O. M., Pan, L., Borenstein, N., Pillai, A., et al. (2023) · arXiv · doi:10.48550/arxiv.2311.09000 · Cited by 3 · cited 1×
“For instance, fact-checkers target a very narrow sub-problem, and usually require ex-ternal knowledge sources such as knowledge bases that are generally incomplete.”
Self-citation(2010) · doi:10.5555/1756006.1859904
Research works printed but not linked: 4
- Yarin Gal. 2016. Uncertainty in Deep Learning . Ph.D. thesis, University of Cambridge.
- 16-20, 2020 , pages 8362-8372. Association for Computational Linguistics.
- Pengcheng He, Xiaodong Liu, Jianfeng Gao, and Weizhu Chen. 2021. DeBERTa: Decodingenhanced bert with disentangled attention. In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 .
- Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de Las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, Lélio Renard Lavaud, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, and William El Sayed. 2023. Mistral 7b. CoRR , abs/2310.06825. · arXiv
Cited by
A Survey of Large Language Models
Zhao, W. X., Zhou, K., Li, J., Tang, T., Dong, Z., Hou, Y., Zhang, B., Min, Y., Zhang, J., Liu, P., et al. (2026) · Frontiers of Computer Science · doi:10.1007/s11704-026-60308-3 · Cited by 2,093
Shorinwa, O., Mei, Z., Lidard, J., Ren, A. Z., & Majumdar, A. (2026) · ACM Computing Surveys · doi:10.1145/3744238 · Cited by 61
“Lastly, we note that there has been some work that aims to standardize the tasks for evaluating the performance of LLMs by explicitly accounting for the uncertainty of LLMs in specific tasks, e.g., based on selective classification and generation [209] or conformal prediction [234].”
Uncertainty Quantification and Confidence Calibration in Large Language Models: A Survey
Liu, X., Chen, T., Da, L., Chen, C., Lin, Z., & Wei, H. (2025) · Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 · doi:10.1145/3711896.3736569 · Cited by 53
A Survey of Uncertainty Estimation Methods on Large Language Models
Xia, Z., Xu, J., Zhang, Y., & Liu, H. (2025) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2025.findings-acl.1101 · Cited by 120
A Survey on Recent Advances in Conversational Data Generation
Soudani, H., Petcu, R., Kanoulas, E., & Hasibi, F. (2026) · ACM Computing Surveys · doi:10.1145/3795686 · Cited by 39
Metadata
- Authors
- Vashurin, R., Fadeeva, E., Vazhentsev, A., Rvanova, L., Vasilev, D., Tsvigun, A., Petrakov, S., Xing, R., Sadallah, A., Grishchenkov, K., Panchenko, A., Baldwin, T., Nakov, P., Panov, M., & Shelmanov, A.
- Affiliations
- Mohamed bin Zayed University of Artificial Intelligence (AE)
- ETH Zurich (CH)
- Center for Artificial Intelligence Technology, Russia. vazhentsev@airi.net, artiomvazh99@gmail.com
- Laboratory for Analysis and Controllable Text Generation Technologies RAS, Russia. Milarv99@gmail.com
- National Research University Higher School of Economics (RU)
- Nebius. aktsvigun@gmail.com
- Center for Artificial Intelligence Technology, Russia. sergeypetrakof@gmail.com
- The University of Melbourne (AU)
- Independent Researcher. kirillgrish@gmail.com
- Center for Artificial Intelligence Technology, Russia. panchenkoalexander@gmail.com, panchenko@airi.net
- ORCID
- Venue
- Transactions of the Association for Computational Linguistics
- Publisher
- MIT Press
- Published
- 2025
- Volume
- 13
- Pages
- 220-248
- ISSN
- 2307-387X (ISSN-L)
- OpenAlex
- W4409218928
- Semantic Scholar
- cc0c6f4dbbfc163cfae15724da1d7e3042fa099c
- Cited by
- 21
- References
- 75
- Citations by year
- 2026: 17 · 2025: 4
- FWCI
- 8.42
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- Locations
- Transactions of the Association for Computational Linguistics (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion, cc-by)
- Minerva Access (University of Melbourne) (submittedVersion, cc-by)
- arXiv (Cornell University) (cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Keywords
- uncertainty quantification, large language models, LLM hallucination, benchmark, text generation
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.
A language model drew the claims from the paper on 2026-10-04, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-04 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.