Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Mitigating Knowledge Conflicts in Language Model-Driven Question Answering

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Training small bottleneck adapters or prefix-tuning adapters on entity-substituted question-answer triples makes a QA model override its memorized parametric answers and follow the supplied context instead, reducing a common form of…

desk verdict A plausible adapter-based idea for overriding memorized QA answers that is undermined by an evaluation unable to distinguish context-copying from knowledge override. read the letter →

arxiv 2411.11344 v3 pith:BI7LS762 submitted 2024-11-18 cs.CL cs.AI

Han Cao , Zhaoyang Zhang , Xiangtian Li , Chufan Wu , Hansong Zhang , Wenqing Zhang This is my paper · ORCID
classification cs.CLcs.AI
keywords knowledgeconflictsquestionansweringhallucinationmitigationadaptertuningprefixparametricanti-factualsubstitutionfrozenbackbone
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

The paper targets a failure mode in open-book question answering: a reader model that has memorized an answer can ignore retrieved context and repeat the memorized entity, producing a hallucination when the context says otherwise. The authors construct anti-factual triples by swapping the gold answer for another entity of the same type, then train either a bottleneck adapter or a prefix-tuning adapter on these triples while keeping the base GPT-2 frozen. They report that both adapters make the model switch to the substituted answer on the memorized subset, reaching about 92 percent exact-match accuracy on KMIR and about 64 percent on Natural Questions. The paper's intended conclusion is that parameter-efficient adapter training can steer a model toward context-grounded output without full fine-tuning.

What carries the argument

The load-bearing mechanism is the combination of an anti-factual dataset and a frozen-backbone adapter: a fine-tuned GPT-2 victim model is scored by a memorization rate that counts exact matches of original answers, a subset of memorized questions is rebuilt with same-type entity substitutions, and then cross-entropy training is run on the substituted triples while only the adapter parameters (bottleneck adapter or prefix-tuning vectors) are updated. The evaluation metric $p_s$ measures how often the model emits the substituted answer, which the paper takes as evidence that the context rather than parametric memory is driving generation.

What would settle it

Run the adapter-trained model on naturally occurring conflicting passages, such as an NQ passage stating a fact newer than or different from the answer the original model memorized, without any engineered entity substitution, and measure exact-match faithfulness; if $p_s$ stays high on engineered swaps while natural conflicts still receive the memorized answer, the central claim fails.

Watch

Extended reading notes

Core claim

The central discovery is that knowledge conflicts can be mitigated at the adapter level: with the base language model frozen, a small trainable module trained on substituted entity contexts is enough to override the model's memorized answer. The method builds an anti-factual dataset of triples $(q, c', x')$ where $x'$ is a same-type substitution of the original answer $x$ and $c'$ is the context with $x$ replaced by $x'$, then maximizes the language-modeling likelihood of $x'$ conditioned on $(q, c')$ through the adapter parameters only. On examples the victim model previously answered correctly, the fraction of substituted answers generated after adapter training reaches 92.9 percent on KMIR and 64.1 percent on Natural Questions with the bottleneck adapter, and 91.9 percent and 64.8 percent with prefix tuning. The authors conclude that both methods can effectively override memorized parameterized knowledge.

Load-bearing premise

The evaluation rests on the assumption that the anti-factual, same-type substituted dataset represents real-world knowledge conflicts and that raising substituted-answer accuracy on a pre-selected memorized subset reduces hallucination in general question answering.

Editorial extensions

If this is right

  • If the result holds, a small adapter trained on anti-factual examples can replace full fine-tuning for reducing entity-level hallucination, preserving the base weights and lowering compute.
  • The high substituted-answer accuracy indicates that models can be steered to trust retrieved context even when it contradicts a memorized fact, which is directly relevant to retriever-reader question-answering pipelines.
  • Because both bottleneck and prefix adapters achieve similar accuracy, the effect is not tied to one adapter architecture, suggesting a general property of small trainable modules.
  • The method targets cases where the model has a crisp memorized answer, so improving these cases should reduce clear-cut factual hallucinations of the exact-match type.

Reading between the lines

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

  • The paper does not establish that flip rate on engineered anti-factual substitutions transfers to natural hallucination; a testable extension would measure faithfulness on unmodified or naturally conflicting contexts, not only swapped entities.
  • Because adapters are small and additive, one could train separate context-trusting adapters for different knowledge domains and switch them at inference time, an application the paper does not explore.
  • The accuracy gap between KMIR (about 92 percent) and NQ (about 64 percent) suggests context complexity matters; a natural extension is to test whether longer or more adversarial contexts require larger adapters or additional training signal.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes to mitigate entity-based knowledge conflicts in QA by freezing a GPT-2 victim model and training bottleneck adapters or prefix-tuning adapters on 'anti-factual' triples, where the gold answer is replaced by a same-type entity and the context is rewritten correspondingly. It defines a memorization rate R_mem and a substituted-answer accuracy p_s, reports high p_s numbers on KMIR and NQ memorized subsets, and concludes in Section VI that these methods 'effectively override memorized parameterized knowledge.' The paper does not provide code, baselines, hyperparameters, or error bars.

Significance. If the central claim were established, the idea of using parameter-efficient adapters to steer a frozen model away from parametric memory and toward presented context would be a useful contribution to hallucination mitigation. The paper is clear about the problem formulation and the construction of anti-factual data, and it gives a concrete, falsifiable metric p_s. However, as it stands the evidence is not sufficient: the evaluation is confined to a specially selected memorized subset, the metric does not distinguish context-following from context-copying, and the promised 'minimum performance sacrifice' is never measured. The strengths are the clean experimental template and the parameter-efficient approach; the weaknesses are the missing controls and incomplete method specification.

major comments (4)
  1. [Section V.B, Eq. (p_s)] The evaluation metric p_s only checks whether the model's output equals the substituted entity that is already present in the context. Because the adapter is trained to maximize exactly this conditional probability on anti-factual triples, a high p_s is equally consistent with a shallow 'copy the entity in the context' behavior as with genuinely overriding parametric knowledge. To support the Section VI claim, the paper must report at least: (i) accuracy on the original, unsubstituted questions after adapter training (to measure the 'minimum performance sacrifice' promised in Section I), (ii) p_s on non-memorized or held-out conflict examples, and (iii) a comparison against a context-copying baseline and standard full fine-tuning on the same anti-factual data. Without these, the central claim is not established.
  2. [Section III.C] The 'adversarial way' of training is never specified: there is no adversarial loss formula, no adversary architecture, no training schedule, and no hyperparameters. This is a load-bearing part of the method because it is what allegedly distinguishes the proposal from ordinary supervised fine-tuning, yet it is not reproducible. In addition, Section III explicitly says the paper reports 'our tentative schedule for the project,' which suggests the method section describes a proposal rather than a completed and validated approach.
  3. [Section V.A, Table I] The qualitative example in Table I is inconsistent with the quantitative metric. For the Heinz Hohner question, the context contains 'French Navy,' the bottleneck adapter output is 'France,' and the text says the model 'successfully output the shuffled answer.' Under the exact-match definition of p_s in Section V.B, 'France' would not count as the substituted answer. The paper needs to reconcile this discrepancy: either the metric underreports the method's behavior (e.g., fine-grained or semantic matching should be used) or the qualitative claim is inaccurate.
  4. [Section IV.B and Section V.B] The test set is constructed by taking the 593/637 memorized samples from the training split and substituting the answer, so p_s is evaluated on the same distribution used to create the adapter training data. No evidence is given that improvements on this memorized subset transfer to broader QA faithfulness or to naturally occurring knowledge conflicts, which is the stated goal in the abstract. The paper needs an independent prediction—for example, behavior on non-memorized questions or on conflicts sampled from a realistic retrieval distribution—before the hallucination-reduction claim can be taken as supported.
minor comments (5)
  1. [Tables III and IV] The column headers 'accuracy on training set' and 'accuracy on test set' are misleading; these are p_s values on the anti-factual memorized subset, not general QA accuracy. Please rename them to make the scope explicit.
  2. [Throughout] There are numerous typos and formatting issues, including 'Datset' in Section IV, 'extend' for 'extent' in Section V.B, 'bottle neck adapter' in Section III.C, and inconsistent spelling of 'parameterized/parameterised.'
  3. [References] Several references appear unrelated to the claims they are attached to, for example [13] is cited to support a statement about hallucination risks but the reference is an alpha-mining model paper, and [17],[26],[33] are about resource allocation, transformer heart-disease prediction, and graph storage. The citation list should be cleaned so that each reference supports the specific statement.
  4. [Figure 1] The text refers to Figure 1 and contains a caption-like block 'FIG 1 THREE STAGES...', but no actual figure image appears in the manuscript. Either include the figure or remove the reference.
  5. [Section III.B and Section V.B] The notation is inconsistent: R_mem uses x'' for the predicted answer while p_s uses x-hat, and the indicator function notation is described loosely. Please unify the notation and define all variables at first use.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: held-out evaluation on anti-factual data is standard supervised testing, and self-citations are not load-bearing.

full rationale

The paper's claimed derivation chain is self-contained against external benchmarks and does not reduce any prediction to its fitting inputs. Stage 3 trains bottleneck/prefix adapters to maximize log P(x'|q,c') on anti-factual examples, while Section V.B evaluates p_s as the fraction of held-out anti-factual examples where the adapted model outputs x'. This is a supervised accuracy metric matching the training objective, but the test set is reported separately from the training set (Tables III and IV report train and test accuracy), and the anti-factual examples are built from externally sourced KMIR and NQ data; hence the held-out p_s is an empirical generalization result rather than an identity. The selection of memorized examples via Rmem uses the victim model's behavior on anti-factual contexts, but that selection targets the phenomenon the method is designed to address; it does not define the post-training outcome. The self-citations (e.g., [32] for adversarial-loss inspiration and [21] for related multimodal work) are tangential and are not used to justify the central override claim. The paper's evidentiary limitations, such as not reporting accuracy on original unsubstituted questions and not comparing against a simple context-copy baseline, are construct-validity concerns rather than circularity per the definitions used here.

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

The paper does not derive any analytic result or introduce free parameters; its contribution is an empirical training recipe. The main assumptions are domain-proxy assumptions about the anti-factual data and exact-match metrics, plus the unexamined choice of GPT-2.

assumptions (3)
  • domain assumption Entity-type-matched substitution produces anti-factual examples that are a valid model of real knowledge conflicts.
    The whole training and evaluation pipeline in Sections IV.B and V.B depends on this proxy; no evidence is given that the substituted-context distribution matches natural contexts where hallucination occurs.
  • domain assumption Exact string match is a sufficient measure of memorization and of context following.
    Section IV.C defines memorization by exact match, and Section V.B defines accuracy by exact match, while acknowledging that same-entity paraphrases are ignored, which could misclassify model behavior.
  • domain assumption GPT-2 fine-tuned for two to three epochs is representative enough for conclusions about parametric knowledge conflicts.
    All experiments use GPT-2; the paper generalizes conclusions to large language models without justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mitigating Knowledge Conflicts in Language Model-Driven Question Answering." pith.science (2026). https://pith.science/paper/BI7LS762

@misc{pith2026241111344,
  author       = {Pith},
  title        = {Pith review of: Mitigating Knowledge Conflicts in Language Model-Driven Question Answering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BI7LS762}},
  note         = {Machine review of arXiv:2411.11344}
}
read the original abstract

In the context of knowledge-driven seq-to-seq generation tasks, such as document-based question answering and document summarization systems, two fundamental knowledge sources play crucial roles: the inherent knowledge embedded within model parameters and the external knowledge obtained through context. Recent studies revealed a significant challenge: when there exists a misalignment between the model's inherent knowledge and the ground truth answers in training data, the system may exhibit problematic behaviors during inference, such as ignoring input context, or generating unfaithful content. Our investigation proposes a strategy to minimize hallucination by building explicit connection between source inputs and generated outputs. We specifically target a common hallucination pattern in question answering, examining how the correspondence between entities and their contexts during model training influences the system's performance at inference time.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Enhanced Recommendation Combining Collaborative Filtering and Large Language Models

    cs.AI 2024-12 reject novelty 1.0 of 10

    A simple weighted sum of collaborative filtering scores and LLM text embeddings is claimed to improve recommendation accuracy, but the reported experiments are not reproducible.

Reference graph

Works this paper leans on

42 extracted references · 22 canonical work pages · cited by 1 Pith paper

  1. [1]

    Common sense or world knowledge? investigating adapter - based knowledge injection into pretrained transformers,

    A. Lauscher, O. Majewska, L. F. Ribeiro, I. Gurevych, N. Rozanov, and G. Glavas, “Common sense or world knowledge? investigating adapter - based knowledge injection into pretrained transformers,” in Proceedings of Deep Learning Inside Out (DeeLIO): The First Workshop on Knowledge Extraction and Integration for Deep Learning Architectures, 2020, pp. 43–49

    work page 2020

  2. [2]

    Totto: A controlled table-to-text generation dataset,

    A. P. Parikh, X. Wang, S. Gehrmann, M. Faruqui, B. Dhingra, D. Yang, and D. Das, “Totto: A controlled table-to-text generation dataset,” arXiv preprint arXiv:2004.14373, 2020

    arXiv 2004

  3. [3]

    Learning to compose neural networks for question answering,

    J. Andreas, M. Rohrbach, T. Darrell, and D. Klein, “Learning to compose neural networks for question answering,”CoRR,vol. abs/1601.01705,2016. [Online]. Available: http://arxiv.org/abs/1601.01705

    arXiv 2016

  4. [4]

    Llmeasyquant –an easy to use toolkit for llm quantization,

    D. Liu, M. Jiang, and K. Pister, “Llmeasyquant –an easy to use toolkit for llm quantization,” arXiv preprint arXiv:2406.19657, 2024

    arXiv 2024

  5. [5]

    Opti- mization and application of cloud -based deep learning architecture for multi-source data prediction,

    Y. Zhang, F. Wang, X. Huang, X. Li, S. Liu, and H. Zhang, “Opti- mization and application of cloud -based deep learning architecture for multi-source data prediction,” arXiv preprint arXiv:2410.12642, 2024

    arXiv 2024

  6. [6]

    The curious case of hallucinations in neural machine translation,

    V. Raunak, A. Menezes, and M. Junczys -Dowmunt, “The curious case of hallucinations in neural machine translation,” arXiv preprint arXiv:2104.06683, 2021

    arXiv 2021

  7. [7]

    Extracting training data from large language models,

    N. Carlini, F. Tramer, E. Wallace, M. Jagielski, A. Herbert-Voss, K. Lee, A. Roberts, T. Brown, D. Song, U. Erlingsson et al., “Extracting training data from large language models,” in 30th USENIX Security Symposium (USENIX Security 21), 2021, pp. 2633–2650

    work page 2021

  8. [8]

    Neural text generation with unlikelihood training,

    S. Welleck, I. Kulikov, S. Roller, E. Dinan, K. Cho, and J. Weston, “Neural text generation with unlikelihood training,” arXiv preprint arXiv:1908.04319, 2019

    arXiv 1908

Show all 42 references
  1. [9]

    Sdrtv-to- hdrtv via hierarchical dynamic context feature mapping,

    G. He, K. Xu, L. Xu, C. Wu, M. Sun, X. Wen, and Y.-W. Tai, “Sdrtv-to- hdrtv via hierarchical dynamic context feature mapping,” in Proceedings of the 30th ACM International Conference on Multimedia, 2022, pp. 2890–2898

    2022

  2. [10]

    Unsupervised enrichment of persona-grounded dialog with background stories,

    B. P. Majumder, T. Berg -Kirkpatrick, J. McAuley, and H. Jhamtani, “Unsupervised enrichment of persona-grounded dialog with background stories,” arXiv preprint arXiv:2106.08364, 2021

    2021 arXiv

  3. [11]

    Research on improved u-net based remote sensing image segmentation algorithm,

    Q. Yang, Z. Wang, S. Liu, and Z. Li, “Research on improved u-net based remote sensing image segmentation algorithm,” arXiv preprint arXiv:2408.12672, 2024

    2024 arXiv

  4. [12]

    Efficient fine-tuning of large language models for automated medical documentation,

    H. Y. Leong, Y. F. Gao, J. Shuai, Y. Zhang, and U. Pamuksuz, “Efficient fine-tuning of large language models for automated medical documentation,” arXiv preprint arXiv:2409.09324, 2024

    2024

  5. [13]

    Alphanetv4: Alpha mining model,

    W. Wu, “Alphanetv4: Alpha mining model,” arXiv preprint arXiv:2411.04409, 2024

    2024 arXiv

  6. [14]

    CTRL: A conditional transformer language model for controllable generation,

    N. S. Keskar, B. McCann, L. R. Varshney, C. Xiong, and R. Socher, “CTRL: A conditional transformer language model for controllable generation,” CoRR, vol. abs/1909.05858, 2019. [Online]. Available: http://arxiv.org/abs/1909.05858

    1909 arXiv

  7. [15]

    Controlling linguistic style aspects in neural language generation,

    J. Ficler and Y. Goldberg, “Controlling linguistic style aspects in neural language generation,” CoRR, vol. abs/1707.02633, 2017. [Online]. Available: http://arxiv.org/abs/1707.02633

    2017 arXiv

  8. [16]

    A novel bayesian pay-as-you-drive insurance model with risk prediction and causal mapping,

    B. Wang, Y. Chen, and Z. Li, “A novel bayesian pay-as-you-drive insurance model with risk prediction and causal mapping,” Decision Analytics Journal, p. 100522, 2024

    2024

  9. [17]

    Recursive balanced k-subset sum partition for rule-constrained resource allocation,

    Z. Li, J. Cao, Z. Yao, W. Li, Y. Yang, and J. Wang, “Recursive balanced k-subset sum partition for rule-constrained resource allocation,” in Proceedings of the 29th ACM International Conference on Information & Knowledge Management, ser. CIKM ’20. New York, NY, USA: Associatio...

    2020 doi

  10. [18]

    Selecting workers wisely for crowdsourcing when copiers and domain experts co-exist,

    X. Fang, S. Si, G. Sun, Q. Z. Sheng, W. Wu, K. Wang, and H. Lv, “Selecting workers wisely for crowdsourcing when copiers and domain experts co-exist,” Future Internet, vol. 14, no. 2, p. 37, 2022

    2022

  11. [19]

    Machine learning -based research on the adaptability of adolescents to online education,

    M. Wang and S. Liu, “Machine learning -based research on the adaptability of adolescents to online education,” arXiv preprint arXiv:2408.16849, 2024

    2024 arXiv

  12. [21]

    Integration of mamba and transformer –mat for long -short range time series forecasting with application to weather dynamics,

    W. Zhang, J. Huang, R. Wang, C. Wei, W. Huang, and Y. Qiao, “Integration of mamba and transformer –mat for long -short range time series forecasting with application to weather dynamics,” arXiv preprint arXiv:2409.08530, 2024

    2024 arXiv

  13. [22]

    Enhanced self - checkout system for retail based on improved yolov10,

    L. Tan, S. Liu, J. Gao, X. Liu, L. Chu, and H. Jiang, “Enhanced self - checkout system for retail based on improved yolov10,” Journal of Imaging, vol. 10, no. 10, p. 248, 2024

    2024

  14. [23]

    Multimodalqa: complex question answering over text, tables and images,

    A. Talmor, O. Yoran, A. Catav, D. Lahav, Y. Wang, A. Asai, G. Ilharco, H. Hajishirzi, and J. Berant, “Multimodalqa: complex question answering over text, tables and images,” in International Conference on Learning Representations

  15. [24]

    Measuring digitalization capabilities using machine learning,

    J. Yang, J. Liu, Z. Yao, and C. Ma, “Measuring digitalization capabilities using machine learning,” Research in International Business and Finance, vol. 70, p. 102380, 2024.[Online]. Available: https://www.sciencedirect.com/science/article/pii/S0275531924001739

    2024

  16. [25]

    Reading wikipedia to answer open -domain questions,

    D. Chen, A. Fisch, J. Weston, and A. Bordes, “Reading wikipedia to answer open -domain questions,” CoRR, vol. abs/1704.00051, 2017.[Online]. Available: http://arxiv.org/abs/1704.00051

    2017 arXiv

  17. [26]

    Optimization of transformer heart disease prediction model based on particle swarm optimization algorithm,

    P. Yu, J. Yi, T. Huang, Z. Xu, and X. Xu, “Optimization of transformer heart disease prediction model based on particle swarm optimization algorithm,” 2024. [Online]. Available: https://arxiv.org/abs/2412.02801

    2024 arXiv

  18. [27]

    Long short-term memory based recurrent neural network architectures for large vocabulary speech recognition,

    H. Sak, A. W. Senior, and F. Beaufays, “Long short-term memory based recurrent neural network architectures for large vocabulary speech recognition,” CoRR, vol. abs/1402.1128, 2014. [Online]. Available: http://arxiv.org/abs/1402.1128

    2014 arXiv

  19. [28]

    Real -time idling vehicles detection using combined audio -visual deep learning,

    X. Li, T. Mangin, S. Saha, R. Mohammed, E. Blanchard, D. Tang, H. Poppe, O. Choi, K. Kelly, and R. Whitaker, “Real -time idling vehicles detection using combined audio -visual deep learning,” in Emerging Cutting-Edge Developments in Intelligent Traffic and Transportation Syste...

    2024

  20. [29]

    Handling divergent reference texts when evaluating table-to- text generation,

    B. Dhingra, M. Faruqui, A. Parikh, M. -W. Chang, D. Das, and W. Cohen, “Handling divergent reference texts when evaluating table-to- text generation,” in Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. Florence, Italy: Association for C...

    2019

  21. [30]

    K -adapter: Infusing knowledge into pre -trained models with adapters,

    R. Wang, D. Tang, N. Duan, Z. Wei, X. Huang, J. Ji, G. Cao, D. Jiang, and M. Zhou, “K -adapter: Infusing knowledge into pre -trained models with adapters,” in FINDINGS, 2021

    2021

  22. [31]

    Efficient transfer learning for quality estimation with bottleneck adapter layer,

    H. Yang, M. Wang, N. Xie, Y. Qin, and Y. Deng, “Efficient transfer learning for quality estimation with bottleneck adapter layer,” in EAMT, 2020

    2020

  23. [32]

    Improving text -image matching with adversarial learning and circle loss for multi -modal steganography,

    Y. Hu, H. Cao, Z. Yang, and Y. Huang, “Improving text -image matching with adversarial learning and circle loss for multi -modal steganography,” in International Workshop on Digital Watermarking. Springer, 2020, pp. 41–52

    2020

  24. [33]

    Graphsnapshot: Graph machine learning acceleration with fast storage and retrieval,

    D. Liu, R. Waleffe, M. Jiang, and S. Venkataraman, “Graphsnapshot: Graph machine learning acceleration with fast storage and retrieval,” arXiv preprint arXiv:2406.17918, 2024

    2024 arXiv

  25. [34]

    Ratt: Athought structure for coherent and correct llmreasoning,

    J. Zhang, X. Wang, W. Ren, L. Jiang, D. Wang, and K. Liu, “Ratt: Athought structure for coherent and correct llmreasoning,” arXiv preprint arXiv:2406.02746, 2024

    2024 arXiv

  26. [35]

    Irda: Incremental rein - forcement learning for dynamic resource allocation,

    J. Wang, J. Cao, S. Wang, Z. Yao, and W. Li, “Irda: Incremental rein - forcement learning for dynamic resource allocation,” IEEE Transactions on Big Data, vol. 8, no. 3, pp. 770–783, 2022

    2022

  27. [36]

    Hurdles to progress in long -form question answering,

    K. Krishna, A. Roy, and M. Iyyer, “Hurdles to progress in long -form question answering,” arXiv preprint arXiv:2103.06332, 2021

    2021 arXiv

  28. [37]

    KMIR: A benchmark for evaluating knowledge memorization, identification and reasoning abilities of language models,

    D. Gao, Y. Jia, L. Li, C. Fu, Z. Dou, H. Jiang, X. Zhang, L. Chen, and Z. Cao, “KMIR: A benchmark for evaluating knowledge memorization, identification and reasoning abilities of language models,” CoRR, vol. abs/2202.13529, 2022. [Online]. Available: https://arxiv.org/abs/2202.13529

    2022 arXiv

  29. [38]

    Entity-based knowledge conflicts in question answering,

    S. Longpre, K. Perisetla, A. Chen, N. Ramesh, C. DuBois, and S. Singh, “Entity-based knowledge conflicts in question answering,” in Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Online and Punta Cana, Dominican Republic: Association fo...

    2021

  30. [39]

    Natural questions: a benchmark for question answering research,

    T. Kwiatkowski, J. Palomaki, O. Redfield, M. Collins, A. Parikh, C. Alberti, D. Epstein, I. Polosukhin, M. Kelcey, J. Devlin, K. Lee, K. N. Toutanova, L. Jones, M. -W. Chang, A. Dai, J. Uszkoreit, Q. Le, and S. Petrov, “Natural questions: a benchmark for question answering res...

    2019

  31. [40]

    NewsQA: A machine comprehension dataset,

    A. Trischler, T. Wang, X. Yuan, J. Harris, A. Sordoni, P. Bachman, and K. Suleman, “NewsQA: A machine comprehension dataset,” in Proceedings of the 2nd Workshop on Representation Learning for NLP. Vancouver, Canada: Association for Computational Linguistics, Aug. 2017, pp. 191...

    2017

  32. [41]

    Prototypical reward network for data-efficient model alignment,

    J. Zhang, X. Wang, Y. Jin, C. Chen, X. Zhang, and K. Liu, “Prototypical reward network for data-efficient model alignment,” in Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2024, pp. 13 871–13 884

    2024

  33. [42]

    A consolidated volatility prediction with back propagation neural network and genetic algorithm,

    Z. Ke, J. Xu, Z. Zhang, Y. Cheng, and W. Wu, “A consolidated volatility prediction with back propagation neural network and genetic algorithm,” arXiv preprint arXiv:2412.07223, 2024

    2024 arXiv

  34. [2015]

    Available: http://arxiv.org/abs/1505.00468

    [Online]. Available: http://arxiv.org/abs/1505.00468

    arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.