Pith. sign in

REVIEW 3 major objections 6 minor 42 references

A Head to Predict and a Head to Question: Pre-trained Uncertainty Quantification Heads for Hallucination Detection in LLM Outputs

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A pre-trained Transformer head that reads a frozen LLM's attention maps and top-k token probabilities achieves state-of-the-art claim-level hallucination detection, including out-of-domain and cross-lingual settings, with about 5% overhead.

desk verdict A useful plug-and-play hallucination detector with broad evaluation, but the GPT-4o label oracle and missing variance reporting temper the SOTA claim. read the letter →

arxiv 2505.08200 v1 pith:5CTSE3WS submitted 2025-05-13 cs.CL cs.AI

Artem Shelmanov , Ekaterina Fadeeva , Akim Tsvigun , Ivan Tsvigun , Zhuohan Xie , Igor Kiselev , Nico Daheim , Caiqi Zhang
show 4 more authors
Artem Vazhentsev Mrinmaya Sachan Preslav Nakov Timothy Baldwin
This is my paper · ORCID
classification cs.CLcs.AI
keywords hallucinationdetectionuncertaintyquantificationattentionmapsclaim-levelpre-trainedheadlargelanguagemodelscross-lingualgeneralization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a small supervised module called an uncertainty quantification head (UHead) detects hallucinated atomic claims — indivisible factual statements in an LLM's generated text — more accurately than existing unsupervised uncertainty scores and supervised probes. UHead reads the frozen LLM's raw attention weights to the handful of preceding tokens and the log-probabilities of the top candidate tokens, then passes them through a one- or two-layer Transformer encoder that marks which tokens belong to the claim being scored. Trained on native LLM outputs labeled by GPT-4o, UHead reports the best precision-recall area under the curve for claim-level supported/unsupported classification, both in-domain and out-of-domain, and on Russian, Chinese, and German prompts when trained only on English biographies. The design is practical because it adds roughly 5 percent inference overhead and about 40 MB of parameters to a multi-billion-parameter LLM.

What carries the argument

The central object is the UHead architecture: a Transformer encoder mounted on a frozen LLM. Its input is a token-level feature vector that concatenates, for each attention head and layer, the raw attention weights from the current token to the previous $k$ tokens (zero-padded where positions do not exist) with the logarithms of the top-$m$ next-token probabilities. A learned embedding is added to each token to indicate whether it belongs to the claim being scored, so the encoder can attend selectively to claim-relevant context. The encoder outputs are mean-pooled over the claim's tokens and fed to a two-layer classifier trained with binary cross-entropy. The load-bearing mechanism is the raw attention map itself: unlike the aggregated lookback ratios of prior work, or hidden-state features that the authors find overfit after 1-3 epochs, flattened attention weights to a short window of preceding tokens continue to generalize, and the per-head analysis suggests a minority of attention heads carry the hallucination signal.

What would settle it

Re-annotate a sample of the test claims with human fact-checkers using verifiable sources, then recompute PR-AUC for UHead and the strongest baselines on those human labels; if UHead's advantage shrinks or reverses, the reported state of the art is an artifact of the GPT-4o labeling protocol.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the signal for whether an atomic claim is unsupported lives primarily in the LLM's attention maps, not in its hidden states, and that a contextualized Transformer head can extract that signal. The authors formalize claim-level hallucination detection as a function $U(c_i\mid x,y)$ assigning each claim a probability of being unsupported. The head consumes flattened attention weights from every attention head and layer to the previous $k$ tokens (empirically $k=2$ to $5$ suffices) concatenated with the logarithms of the top-$m$ token probabilities, plus a learned embedding that indicates claim membership. In their experiments on eight English test domains, UHead achieves the highest PR-AUC among all baselines; for Mistral 7B it beats the best unsupervised method (CCP) by 16 percentage points in-domain and by up to 23 points out-of-domain, and it surpasses the closest supervised competitor (Factoscope) by 5 points in-domain. For Gemma 2, a head trained only on English biographies improves over the best unsupervised method by 9-13 percentage points on Russian, Chinese, and German test sets. The authors read these results as showing that attention-based features generalize across domains and languages while hidden-state features overfit.

Load-bearing premise

The training and evaluation labels come from GPT-4o's two-stage supported/unsupported judgments about each claim; if those judgments do not track real hallucination, the reported scores measure agreement with GPT-4o rather than true detection quality.

Editorial extensions

If this is right

  • Claim-level hallucination detection can be added to a frozen LLM without retraining it or altering its generated text, so deployment can be a drop-in adapter.
  • Because attention features generalize, a detector trained on one domain (biographies) transfers to movies, books, inventions, events, landmarks, cities, and artworks without retraining.
  • English-only training transfers to Russian, Chinese, and German prompts, at least for Gemma 2 9B, suggesting the uncertainty signal is at least partly language-independent.
  • The low overhead (about 5% extra compute and 40 MB) means per-token uncertainty scoring can run in the same generation pass in practical applications.
  • Adding more diverse training domains yields small further gains, so scaling the annotated data is a direct route to better detectors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the evaluation labels are swapped from GPT-4o to human fact-checkers or another oracle, the ranking of methods could shift; the paper's headline numbers should be read as measuring agreement with GPT-4o's supportedness judgments.
  • The finding that only 2-5 preceding tokens matter suggests hallucination tendency is a local attention phenomenon; a targeted experiment could test whether removing or reweighting the high-correlation heads identified in the paper changes generation-time confidence without hurting fluency.
  • The same head design could transfer to other claim-bearing tasks such as factual consistency of summaries, retrieval-augmented answers, or machine-translation confidence, since its features are not domain-specific.
  • Because the head relies on attention and token probabilities rather than hidden states, it points toward a lightweight direction for uncertainty scoring in settings where only attention maps are accessible.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper introduces UHead, a supervised auxiliary module for claim-level hallucination detection in LLM outputs. The module consists of a small Transformer encoder that processes token-level features, namely raw attention weights to the preceding k tokens and log-probabilities of the top-m token predictions, all extracted from a frozen LLM; a learned claim-token embedding marks the span of the claim being scored. Training data are produced by prompting the target LLM (Mistral 7B Instruct v0.2 or Gemma 2 9B Instruct) with biography questions, extracting atomic claims with GPT-4o, and labeling each claim as supported, unsupported, or unknown via a two-stage chain-of-thought prompt. Experiments compare UHead with unsupervised and supervised baselines on eight English domains (one in-domain, seven out-of-domain) and on Russian, Chinese, and German biography prompts, reporting PR-AUC as the main metric. The paper also reports about 5% computational overhead and a 40 MB memory footprint, and it releases code and pre-trained heads for several LLM families.

Significance. The empirical design is systematic in several respects: supervised baselines are retrained and tuned on the same biography data, hyperparameters are selected on a validation set, test domains and languages are disjoint from training, and the comparison includes both unsupervised scores and three supervised methods. The released code and pre-trained heads are a concrete practical contribution, and the analysis of which attention layers and token windows carry hallucination-related signal is a useful starting point for future work. If the results survive an independent label oracle, UHead would be a strong and lightweight off-the-shelf hallucination detector. The main risk is that GPT-4o serves as both the training and the test label oracle, so the absolute PR-AUC numbers may quantify agreement with GPT-4o's supportedness judgments rather than veridical hallucination detection; in addition, the absence of variance or significance reporting leaves the size of the reported gains uncertain. These concerns are central to the paper's claims and are addressable within the manuscript's scope.

major comments (3)
  1. [Section 4 and Appendix B] Because GPT-4o produces both the training and the test labels (Section 4, Table 6), the reported PR-AUC values in Tables 1, 2, and 4 measure agreement with GPT-4o's supportedness judgments rather than veridical hallucination detection. The paper's own Ethical Considerations state that this 'may introduce cultural, linguistic, or other biases into the data,' yet no independent validation of the label oracle is provided. I request a human-annotated sample (or an alternative oracle) for at least the out-of-domain and cross-lingual test sets, together with agreement scores, to establish that UHead detects actual hallucinations and not GPT-4o's labeling heuristics.
  2. [Tables 1, 2, and 4] The headline tables report single-run PR-AUC values with no variance, confidence intervals, or significance tests. Some comparative margins are small—for example, UHead trails Factoscope by 1 pp on landmarks in Table 1, and the differences in Table 2 are at most 0.02 PR-AUC—so the claimed state-of-the-art performance may be within run-to-run noise. Please report results over multiple seeds, or at least bootstrap confidence intervals, for the key comparisons that support the main claims.
  3. [Table 4] The cross-lingual generalization claim is especially sensitive to the label-oracle issue: GPT-4o's labeling behavior may differ across Russian, Chinese, and German, so the observed 'generalization' could reflect transfer of a proxy for GPT-4o's cross-lingual biases rather than transfer of a general hallucination-detection capability. I request per-language human validation or a comparison with an independent oracle on a sample of each language before the cross-lingual result can be interpreted as evidence for the stated claim.
minor comments (6)
  1. [Equation (1)] Equation (1) appears to contain a typo: the concatenation lists h1(t) twice instead of showing a sequence of layer indices; it should likely be h^l(t) for l = 1, ..., L.
  2. [Equation (6)] Equation (6) uses n in the superscript of the feature set definition, but the attention window over generated tokens should range over the generated sequence length T, not the prompt length n; please clarify the indexing.
  3. [Table 6] Table 6 includes an 'Accuracy' column with values such as 72.9% and 87.4%, but neither the caption nor the text defines what this accuracy measures or on which split it is computed; please add an explanation.
  4. [Section 4 vs. Appendix B] The annotation budget statement in Section 4 ('around $100' for 3,300 prompts) is difficult to reconcile with the 'approximately $4000' total API expenses in Appendix B; please clarify the accounting.
  5. [Table 7] In Table 7, the row for UHead (Factoscope) is missing a warmup value and the architecture cell appears to contain '0.05' followed by a dash, which is ambiguous; please align and label all hyperparameter columns consistently.
  6. [Table 8] Table 8 reports memory footprints for UHead (only hidden states) and UHead (att. + prob. + hs.) but does not provide the corresponding computational overhead percentages that are given for the other rows; please complete these entries.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: UHead is a supervised detector trained on external GPT-4o labels and evaluated on held-out domains and languages; its claims do not reduce to its inputs.

full rationale

The paper's derivation chain is an empirical supervised-learning pipeline, not a formal derivation that reduces to its inputs. The UHead features are raw attention maps and top-k token log-probabilities (Eqs. 6-7), and the target labels are produced by GPT-4o through claim extraction and two-stage chain-of-thought labeling (Section 4). The head is trained with binary cross-entropy and then evaluated on held-out English domains and on Russian, Chinese, and German prompts (Tables 1, 2, and 4); those evaluations are predictions on data not used for tuning. No equation in the paper defines the target in terms of the model's own output, and no fitted parameter is renamed as a prediction. Self-citations (LM-Polygraph, Fadeeva et al. 2024, Vashurin et al. 2024) supply evaluation infrastructure and prompt templates, not the UHead result, so they are not load-bearing for the central claim. The manuscript itself notes in Ethical Considerations that GPT-4o annotation 'may introduce cultural, linguistic, or other biases into the data used to train the uncertainty heads.' This is a legitimate external-validity concern about label quality that would affect any supervised detector trained and tested on the same oracle, but it is not a circularity of the kind where an equation or parameter reduces to its input. The derivation is therefore self-contained with respect to circularity.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The headline results depend on several tuned hyperparameters and on two domain assumptions: GPT-4o labels are treated as ground truth, and the frozen LLM's attention and probability features contain hallucination-relevant signal. There are no invented physical or conceptual entities beyond the trained UQ head module itself.

free parameters (5)
  • attention window size k = 2 (chosen for both Mistral and Gemma 2; grid included values 1,2,3,4,5,10)
    Number of preceding tokens whose attention weights are included in the feature vector F_att; tuned on the biographies validation set (Section 5.1, Table 7, Figure 5).
  • learning rate = 2e-4
    Selected for UHead via Bayesian optimization on validation PR-AUC (Section 5.1, Table 7).
  • number of training epochs = 7 for Mistral 7B, 6 for Gemma 2 9B
    Tuned on validation PR-AUC (Section 5.1, Table 7).
  • positive-class weight in cross-entropy loss = not reported numerically
    Explicitly tuned as a hyperparameter (Section 5.1) but the selected values are omitted from Table 7, so this part of the training setup is underspecified.
  • Transformer encoder size for the UQ head = 2 layers / 768 width / 4 heads (Mistral); 1 layer / 768 width / 16 heads (Gemma 2)
    Architecture capacity tuned on the validation set (Table 7).
assumptions (3)
  • domain assumption GPT-4o two-stage claim extraction and labeling (supported/unsupported/unknown) produces sufficiently accurate ground truth for training and evaluation.
    Section 4 and Appendix B describe the pipeline; all PR-AUC numbers are computed against these labels, so systematic annotation errors would directly distort the headline comparisons.
  • domain assumption LLM attention maps and top-k token log-probabilities encode useful signal about claim-level hallucination.
    This is the core informativeness assumption behind F_att and F_prob (Section 3.2); the paper's feature analysis supports it empirically, but the claim is assumed before training.
  • domain assumption Supervised training on English biographies transfers to other domains and languages.
    The entire OOD and cross-lingual evaluation rests on this transferability, which is tested rather than proven; the paper reports positive transfer.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Head to Predict and a Head to Question: Pre-trained Uncertainty Quantification Heads for Hallucination Detection in LLM Outputs." pith.science (2026). https://pith.science/paper/5CTSE3WS

@misc{pith2026250508200,
  author       = {Pith},
  title        = {Pith review of: A Head to Predict and a Head to Question: Pre-trained Uncertainty Quantification Heads for Hallucination Detection in LLM Outputs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5CTSE3WS}},
  note         = {Machine review of arXiv:2505.08200}
}
read the original abstract

Large Language Models (LLMs) have the tendency to hallucinate, i.e., to sporadically generate false or fabricated information. This presents a major challenge, as hallucinations often appear highly convincing and users generally lack the tools to detect them. Uncertainty quantification (UQ) provides a framework for assessing the reliability of model outputs, aiding in the identification of potential hallucinations. In this work, we introduce pre-trained UQ heads: supervised auxiliary modules for LLMs that substantially enhance their ability to capture uncertainty compared to unsupervised UQ methods. Their strong performance stems from the powerful Transformer architecture in their design and informative features derived from LLM attention maps. Experimental evaluation shows that these heads are highly robust and achieve state-of-the-art performance in claim-level hallucination detection across both in-domain and out-of-domain prompts. Moreover, these modules demonstrate strong generalization to languages they were not explicitly trained on. We pre-train a collection of UQ heads for popular LLM series, including Mistral, Llama, and Gemma 2. We publicly release both the code and the pre-trained heads.

Figures

Figures reproduced from arXiv: 2505.08200 by the authors.

Figure 1
Figure 1. The architecture of uncertainty quantification heads. The example represents a text generated using an [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. (a) The distribution of correlations between [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Code example for using uncertainty heads. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: The training data generation pipeline. B Dataset Details B.1 Dataset Construction We used few-shot learning to better guide the LLM to generate the items for the desired domain. The structure of the prompts looks as follows: Continue the list of 100 most famous { domai…
Figure 5
Figure 5. Figure 5: PR-AUC for different attention window sizes using UHead for Mistral 7B Instruct v0.2 model. [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: PR-AUC as a function of layer number used for attention features in UHead for Mistral 7B Instruct v0.2 [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 3 canonical work pages

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Amos Azaria and Tom Mitchell. 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.68 The internal state of an LLM knows when it ' s lying . In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 967--976, Singapore. Association for Computational Linguistics

    doi:10.18653/v1/2023.findings-emnlp.68 2023

  4. [4]

    Joris Baan, Nico Daheim, Evgenia Ilia, Dennis Ulmer, Haau-Sing Li, Raquel Fern \'a ndez, Barbara Plank, Rico Sennrich, Chrysoula Zerva, and Wilker Aziz. 2023. Uncertainty in natural language generation: From theory to applications. arXiv preprint arXiv:2307.15703

    arXiv 2023

  5. [5]

    Sky CH-Wang, Benjamin Van Durme, Jason Eisner, and Chris Kedzie. 2024. https://doi.org/10.18653/v1/2024.findings-acl.260 Do androids know they ' re only dreaming of electric sheep? In Findings of the Association for Computational Linguistics: ACL 2024, pages 4401--4420, Bangkok, Thailand. Association for Computational Linguistics

    doi:10.18653/v1/2024.findings-acl.260 2024

  6. [6]

    Yuyan Chen, Qiang Fu, Yichen Yuan, Zhihao Wen, Ge Fan, Dayiheng Liu, Dongmei Zhang, Zhixu Li, and Yanghua Xiao. 2023. Hallucination detection: Robustly discerning reliable answers in large language models. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, pages 245--255

    2023

  7. [7]

    Yung-Sung Chuang, Linlu Qiu, Cheng-Yu Hsieh, Ranjay Krishna, Yoon Kim, and James Glass. 2024. https://arxiv.org/abs/2407.07071 Lookback lens: Detecting and mitigating contextual hallucinations in large language models using only attention maps . Preprint, arXiv:2407.07071

    arXiv 2024

  8. [8]

    Jinhao Duan, Hao Cheng, Shiqi Wang, Alex Zavalny, Chenan Wang, Renjing Xu, Bhavya Kailkhura, and Kaidi Xu. 2024. https://doi.org/10.18653/v1/2024.acl-long.276 Shifting attention to relevance: Towards the predictive uncertainty quantification of free-form large language models . In Proceedings of the 62nd Annual Meeting of the Association for Computational...

    doi:10.18653/v1/2024.acl-long.276 2024

Show all 42 references
  1. [9]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783

    2024 arXiv

  2. [10]

    Ekaterina Fadeeva, Aleksandr Rubashevskii, Artem Shelmanov, Sergey Petrakov, Haonan Li, Hamdy Mubarak, Evgenii Tsymbalov, Gleb Kuzmin, Alexander Panchenko, Timothy Baldwin, Preslav Nakov, and Maxim Panov. 2024. https://doi.org/10.18653/v1/2024.findings-acl.558 Fact-checking th...

    2024 doi

  3. [11]

    Ekaterina Fadeeva, Roman Vashurin, Akim Tsvigun, Artem Vazhentsev, Sergey Petrakov, Kirill Fedyanin, Daniil Vasilev, Elizaveta Goncharova, Alexander Panchenko, Maxim Panov, Timothy Baldwin, and Artem Shelmanov. 2023. https://doi.org/10.48550/arXiv.2311.07383 LM-Polygraph : Unc...

    2023 doi arXiv

  4. [12]

    Sebastian Farquhar, Jannik Kossen, Lorenz Kuhn, and Yarin Gal. 2024. Detecting hallucinations in large language models using semantic entropy. Nature, 630(8017):625--630

    2024

  5. [13]

    Yarin Gal and Zoubin Ghahramani. 2016. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. In international conference on machine learning, pages 1050--1059. PMLR

    2016

  6. [14]

    Jiahui Geng, Fengyu Cai, Yuxia Wang, Heinz Koeppl, Preslav Nakov, and Iryna Gurevych. 2024. https://doi.org/10.18653/v1/2024.naacl-long.366 A survey of confidence estimation and calibration in large language models . In Proceedings of the 2024 Conference of the North American ...

    2024 doi

  7. [15]

    Jianfeng He, Linlin Yu, Shuo Lei, Chang-Tien Lu, and Feng Chen. 2024 a . https://doi.org/10.18653/v1/2024.findings-naacl.180 Uncertainty estimation on sequential labeling via uncertainty transmission . In Findings of the Association for Computational Linguistics: NAACL 2024, p...

    2024 doi

  8. [16]

    Jianfeng He, Xuchao Zhang, Shuo Lei, Zhiqian Chen, Fanglan Chen, Abdulaziz Alhamadani, Bei Xiao, and ChangTien Lu. 2020. Towards more accurate uncertainty estimation in text classification. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processi...

    2020

  9. [17]

    Jinwen He, Yujia Gong, Zijin Lin, Yue Zhao, Kai Chen, et al. 2024 b . Llm factoscope: Uncovering llms’ factual discernment through measuring inner states. In Findings of the Association for Computational Linguistics ACL 2024, pages 10218--10230

    2024

  10. [18]

    Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, et al. 2025. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. ACM Transactions on Info...

    2025

  11. [19]

    Yukun Huang, Yixin Liu, Raghuveer Thirukovalluru, Arman Cohan, and Bhuwan Dhingra. 2024. https://doi.org/10.18653/v1/2024.findings-emnlp.785 Calibrating long-form generations from large language models . In Findings of the Association for Computational Linguistics: EMNLP 2024,...

    2024 doi

  12. [20]

    Albert Q Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, et al. 2023 a . Mistral 7b. arXiv preprint arXiv:2310.06825

    2023 arXiv

  13. [21]

    Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de Las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, L \' e lio Renard Lavaud, Marie - Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, ...

    2023 doi arXiv

  14. [22]

    Lorenz Kuhn, Yarin Gal, and Sebastian Farquhar. 2023. https://openreview.net/pdf?id=VD-AYtP0dve Semantic uncertainty: Linguistic invariances for uncertainty estimation in natural language generation . In The Eleventh International Conference on Learning Representations, ICLR 2...

    2023

  15. [23]

    Qing Li, Jiahui Geng, Chenyang Lyu, Derui Zhu, Maxim Panov, and Fakhri Karray. 2024. https://doi.org/10.18653/v1/2024.findings-emnlp.262 Reference-free hallucination detection for large vision-language models . In Findings of the Association for Computational Linguistics: EMNL...

    2024 doi

  16. [24]

    Zhen Lin, Shubhendu Trivedi, and Jimeng Sun. 2023. https://doi.org/10.48550/arXiv.2305.19187 Generating with confidence: Uncertainty quantification for black-box large language models . CoRR, abs/2305.19187

    2023 doi arXiv

  17. [25]

    Andrey Malinin and Mark J. F. Gales. 2021. https://openreview.net/forum?id=jN5y-zb5Q7m Uncertainty estimation in autoregressive structured prediction . In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . OpenReview.net

    2021

  18. [26]

    Potsawee Manakul, Adian Liusie, and Mark Gales. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.557 S elf C heck GPT : Zero-resource black-box hallucination detection for generative large language models . In Proceedings of the 2023 Conference on Empirical Methods in Natural...

    2023 doi

  19. [27]

    Sewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis, Wen-tau Yih, Pang Koh, Mohit Iyyer, Luke Zettlemoyer, and Hannaneh Hajishirzi. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.741 FA ct S core: Fine-grained atomic evaluation of factual precision in long form text generatio...

    2023 doi

  20. [28]

    Xin Qiu and Risto Miikkulainen. 2024. https://proceedings.neurips.cc/paper_files/paper/2024/file/f26d4fbaf7dfa115f1d4b3f104e26bce-Paper-Conference.pdf Semantic density: Uncertainty quantification for large language models through confidence measurement in semantic space . In A...

    2024

  21. [29]

    Aviv Slobodkin, Omer Goldman, Avi Caciularu, Ido Dagan, and Shauli Ravfogel. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.220 The curious case of hallucinatory (un)answerability: Finding truths in the hidden states of over-confident large language models . In Proceedings ...

    2023 doi

  22. [30]

    Weihang Su, Changyue Wang, Qingyao Ai, Yiran Hu, Zhijing Wu, Yujia Zhou, and Yiqun Liu. 2024. https://doi.org/10.18653/v1/2024.findings-acl.854 Unsupervised real-time hallucination detection based on the internal states of large language models . In Findings of the Association...

    2024 doi

  23. [31]

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Yonghui Wu, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805

    2023 arXiv

  24. [32]

    Katherine Tian, Eric Mitchell, Allan Zhou, Archit Sharma, Rafael Rafailov, Huaxiu Yao, Chelsea Finn, and Christopher Manning. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.330 Just ask for calibration: Strategies for eliciting calibrated confidence scores from language mod...

    2023 doi

  25. [33]

    Roman Vashurin, Ekaterina Fadeeva, Artem Vazhentsev, Akim Tsvigun, Daniil Vasilev, Rui Xing, Abdelrahman Boda Sadallah, Lyudmila Rvanova, Sergey Petrakov, Alexander Panchenko, et al. 2024. Benchmarking uncertainty quantification methods for large language models with lm-polygr...

    2024 arXiv

  26. [34]

    Artem Vazhentsev, Gleb Kuzmin, Artem Shelmanov, Akim Tsvigun, Evgenii Tsymbalov, Kirill Fedyanin, Maxim Panov, Alexander Panchenko, Gleb Gusev, Mikhail Burtsev, Manvel Avetisian, and Leonid Zhukov. 2022. https://doi.org/10.18653/v1/2022.acl-long.566 Uncertainty estimation of t...

    2022 doi

  27. [35]

    Artem Vazhentsev, Gleb Kuzmin, Akim Tsvigun, Alexander Panchenko, Maxim Panov, Mikhail Burtsev, and Artem Shelmanov. 2023. https://aclanthology.org/2023.acl-long.652 Hybrid uncertainty quantification for selective text classification in ambiguous tasks . In Proceedings of the ...

    2023

  28. [36]

    Yuxia Wang, Daniel Beck, Timothy Baldwin, and Karin Verspoor. 2022. https://doi.org/10.1162/tacl_a_00483 Uncertainty estimation and reduction of pre-trained models for text regression . Transactions of the Association for Computational Linguistics, 10:680--696

    2022 doi

  29. [37]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35:24824--24837

    2022

  30. [38]

    Ji Xin, Raphael Tang, Yaoliang Yu, and Jimmy Lin. 2021. https://doi.org/10.18653/v1/2021.acl-long.84 The art of abstention: Selective prediction and error regularization for natural language processing . In Proceedings of the 59th Annual Meeting of the Association for Computat...

    2021 doi

  31. [39]

    Caiqi Zhang, Fangyu Liu, Marco Basaldella, and Nigel Collier. 2024 a . https://doi.org/10.18653/v1/2024.emnlp-main.299 LUQ : Long-text uncertainty quantification for LLM s . In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 5244--...

    2024 doi

  32. [40]

    Caiqi Zhang, Ruihan Yang, Zhisong Zhang, Xinting Huang, Sen Yang, Dong Yu, and Nigel Collier. 2024 b . https://arxiv.org/abs/2410.13246 Atomic calibration of llms in long-form generations . Preprint, arXiv:2410.13246

    2024

  33. [41]

    Tianhang Zhang, Lin Qiu, Qipeng Guo, Cheng Deng, Yue Zhang, Zheng Zhang, Chenghu Zhou, Xinbing Wang, and Luoyi Fu. 2023. https://doi.org/10.48550/arXiv.2311.13230 Enhancing uncertainty-based hallucination detection with stronger focus . In Proceedings of the 2023 Conference on...

    2023 doi arXiv

  34. [42]

    Xuchao Zhang, Fanglan Chen, Chang-Tien Lu, and Naren Ramakrishnan. 2019. https://doi.org/10.18653/v1/N19-1316 Mitigating uncertainty in document classification . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Linguisti...

    2019 doi

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.