Elamur Research⌕
☰
home / papers / 2024 / 大语言模型的幻觉问题研究综述

大语言模型的幻觉问题研究综述

Summary

This Chinese-language survey from Tsinghua University and the University of International Business and Economics reviews hallucination in large language models (LLMs), meaning fluent output that conflicts with the input or with real-world facts. Such errors erode trust and create legal and ethical risks, especially in court rulings, medical advice and translated drug leaflets. The authors define the phenomenon and organise it with three complementary taxonomies. They then trace its causes across three layers:

  • Data: heuristic collection, duplicated or missing data, and machine-assisted annotation.
  • Model: encoder/decoder flaws, reliance on memorised parametric knowledge, and catastrophic forgetting.
  • Application: domain specialisation, architectural homogenisation, multimodality, and prompting techniques such as instruction tuning and chain-of-thought.

Next, they group evaluation approaches into text-statistic metrics, model-based metrics and multi-task application benchmarks. They also classify mitigation methods and compare their trade-offs for each task. The survey ends with open challenges for both evaluation and mitigation. It is a useful map for teams building reliable generation systems or factuality benchmarks.

Key points

  • Taxonomies: intrinsic vs. extrinsic hallucination; closed-domain vs. open-domain (the framing OpenAI used for GPT-4); input-, context- and fact-conflicting hallucination.
  • Scale of the problem: about 30% of summaries from state-of-the-art models contain unfaithful content. In RotoWire, roughly 40% of reference summary content cannot be linked to any input table record.
  • Data-based metrics: PARENT and PARENT-T for table-to-text, Knowledge F1 for knowledge-grounded dialogue, and chrF and BVSS for machine translation. Wu et al. also built a RAG-based hallucination corpus with 18,000 responses.
  • Model-based evaluation:
    • Information-extraction methods, including NER-based F1.
    • NLI-based methods: SummaC Conv, FactCC, DocNLI, dependency-level NLI and the BEGIN benchmark.
    • Typical NLI backbones are BERT, RoBERTa, DeBERTa and XLNet, tuned on MNLI or ANLI.
  • Publication: Journal of Software, vol. 36(3), pp. 1152–1185. It is the 2025 issue, published online in December 2024, and has 202 references.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages · Natural language processing

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.13328/j.cnki.jos.007242

@article{2024,
  title = {{大语言模型的幻觉问题研究综述}},
  author = {刘泽垣 and 王鹏江 and 宋晓斌 and 张欣 and 江奔奔},
  journal = {软件学报 (Journal of Software)},
  year = {2024},
  doi = {10.13328/j.cnki.jos.007242},
  url = {https://doi.org/10.13328/j.cnki.jos.007242}
}

References

202 references, 50 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Yu TR, Jin R, Han XZ, Li JH, Yu T. Review of pre-training models for natural language processing. Computer Engineering and Applications, 2020, 56(23): 12–22 (in Chinese with English abstract). [doi: 10.3778/j.issn.1002-8331.2006-0040] — [doi:10.3778/j.issn.1002-8331.2006-0040](https://doi.org/10.3778/j.issn.1002-8331.2006-0040) find
  2. Bao SQ, He H, Wang F, Wu H, Wang HF. PLATO: Pre-trained dialogue generation model with discrete latent variable. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. ACL, 2020. 85–96. [doi: 10.18653/v1/2020.acl-main.9] find
  3. Chen WH, Su Y, Yan XF, Wang WY. KGPT: Knowledge-grounded pre-training for data-to-text generation. arXiv:2010.02307, 2020. find
  4. Liu YH, Gu JT, Goyal N, Li X, Edunov S, Ghazvininejad M, Lewis M, Zettlemoyer L. Multilingual denoising pre-training for neural machine translation. Trans. of the Association for Computational Linguistics, 2020, 8: 726–742. [doi: 10.1162/tacl_a_00343] find
  5. Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser Ł, Polosukhin I. Attention is all you need. In: Proc. of the 31st Int’l Conf. on Neural Information Processing Systems. Long Beach: Curran Associates Inc., 2017. 6000–6010. find
  6. Zhao WX, Zhou K, Li JY, Tang TY, Wang XL, Hou YP, Min YQ, Zhang BC, Zhang JJ, Dong ZC, Du YF, Yang C, Chen YS, Chen ZP, Jiang JH, Ren RY, Li YF, Tang XY, Liu ZK, Liu PY, Nie JY, Wen JR. A survey of large language models. arXiv:2303.18223, 2023. find our card
  7. Pagnoni A, Balachandran V, Tsvetkov Y. Understanding factuality in abstractive summarization with FRANK: A benchmark for factuality metrics. In: Proc. of the 2021 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Comput… find our card
  8. Liu PJ, Saleh M, Pot E, Goodrich B, Sepassi R, Kaiser L, Shazeer N. Generating Wikipedia by summarizing long sequences. In: Proc. of the 6th Int’l Conf. on Learning Representations. 2018. 1–18. find our card
  9. Wiseman S, Shieber S, Rush A. Challenges in data-to-document generation. In: Proc. of the 2017 Conf. on Empirical Methods in Natural Language Processing. Copenhagen: Association for Computational Linguistics, 2017. 2253–2263. [doi: 10.18653/v1/D17-1239] find our card
  10. Zhou CT, Neubig G, Gu JT, Diab M, Guzmán F, Zettlemoyer L, Ghazvininejad M. Detecting hallucinated content in conditional neural sequence generation. In: Proc. of the Association for Computational Linguistics: ACL-IJCNLP 2021. Association for Computational Linguistics, 2021. 1393–1404. [doi: 10.1865… find
  11. Zhang C, Lee G, D'Haro LF, Li HZ. D-Score: Holistic dialogue evaluation without reference. IEEE/ACM Trans. on Audio, Speech, and Language Processing, 2021, 29: 2502–2516. [doi: 10.1109/TASLP.2021.3074012] — [doi:10.1109/taslp.2021.3074012](https://doi.org/10.1109/taslp.2021.3074012) find
  12. Zhang XC, Ghorbani AA. An overview of online fake news: Characterization, detection, and discussion. Information Processing & Management, 2020, 57(2): 102025. [doi: 10.1016/j.ipm.2019.03.004] — [doi:10.1016/j.ipm.2019.03.004](https://doi.org/10.1016/j.ipm.2019.03.004) find
  13. Huang YC, Feng XC, Feng XC, Qin B. The factual inconsistency problem in abstractive text summarization: A survey. arXiv:2104. 14839, 2023. find
  14. Cao ZQ, Wei FR, Li WJ, Li SJ. Faithful to the original: Fact aware neural abstractive summarization. In: Proc. of the 32nd AAAI Conf. on Artificial Intelligence. Palo Alto: AAAI, 2018. 4784–4791. [doi: 10.1609/aaai.v32i1.11912] — [doi:10.1609/aaai.v32i1.11912](https://doi.org/10.1609/aaai.v32i1.11912) find
  15. Sobieszek A, Price T. Playing games with ais: The limits of GPT-3 and similar large language models. Minds & Machines, 2022, 32(2): 341–364. [doi: 10.1007/s11023-022-09602-0] — [doi:10.1007/s11023-022-09602-0](https://doi.org/10.1007/s11023-022-09602-0) find
  16. Ji ZW, Lee N, Frieske R, Yu TZ, Su D, Xu Y, Ishii E, Bang YJ, Madotto A, Fung P. Survey of hallucination in natural language generation. ACM Computing Surveys, 2023, 55(12): 248. [doi: 10.1145/3571730] find our card
  17. OpenAI: GPT-4 System Card. 2023. https://cdn.openai.com/papers/gpt-4-system-card.pdf ↗
  18. Evans O, Cotton-Barratt O, Finnveden L, Bales A, Balwit A, Wills P, Righetti L, Saunders W. Truthful AI: Developing and governing AI that does not lie. arXiv:2110.06674, 2021. — [arXiv:2110.06674](https://arxiv.org/abs/2110.06674) find
  19. Zhang Y, Li YF, Cui LY, Cai D, Liu LM, Fu TC, Huang XT, Zhao EB, Zhang Y, Chen YL, Wang LY, Luu AT, Bi W, Shi F, Shi SM. Siren’s song in the AI ocean: A survey on hallucination in large language models. arXiv:2309.01219, 2023. find our card
  20. Wang HM. Revisiting challenges in data-to-text generation with fact grounding. arXiv:2001.03830, 2020. find
  21. Lee K, Ippolito D, Nystrom A, Zhang CY, Eck D, Callison-Burch C, Carlini N. Deduplicating training data makes language models better. arXiv:2107.06499, 2022. find
  22. Jo ES, Gebru T. Lessons from archives: Strategies for collecting sociocultural data in machine learning. In: Proc. of the 2020 Conf. on Fairness, Accountability, and Transparency. Barcelona: ACM, 2020. 306–316. [doi: 10.1145/3351095.3372829] — [doi:10.1145/3351095.3372829](https://doi.org/10.1145/3351095.3372829) find
  23. Parikh A, Wang XZ, Gehrmann S, Faruqui M, Dhingra B, Yang DY, Das D. ToTTo: A controlled table-to-text generation dataset. In: Proc. of the 2020 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2020. 1173–1186. [doi: 10.18653/v1/2020.emnlp-main.89… find our card
  24. Tian R, Narayan S, Sellam S, Parikh AP. Sticking to the facts: Confident decoding for faithful data-to-text generation. arXiv:1910. 08684, 2020. find
  25. Longpre S, Perisetla K, Chen A, Ramesh N, DuBois C, Singh S. Entity-based knowledge conflicts in question answering. In: Proc. of the 2021 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2021. 7052–7063. [doi: 10.18653/v1/2021.emnlp-main.565] find our card
  26. Bommasani R, Hudson DA, Adeli E, et al. On the opportunities and risks of foundation models. arXiv:2108.07258, 2022. find our card
  27. Wei J, Bosma M, Zhao VY, Guu K, Yu AW, Lester B, Du N, Dai AM, Le QV. Finetuned language models are zero-shot learners. arXiv:2109.01652, 2022. find
  28. Wei J, Wang XZ, Schuurmans D, Bosma M, Ichter B, Xia F, Chi E, Le Q, Zhou D. Chain-of-thought prompting elicits reasoning in large language models. In: Proc. of the 36th Int’l Conf. on Neural Information Processing Systems. New Orleans: Curran Associates Inc., 2022. 24824–24837. find our card
  29. Liu FX, Lin K, Li LJ, Wang JF, Yacoob Y, Wang LJ. Mitigating hallucination in large multi-modal models via robust instruction tuning. arXiv:2306.14565, 2024. find our card
  30. Kojima T, Gu SS, Reid M, Matsuo Y, Iwasawa Y. Large language models are zero-shot reasoners. In: Proc. of the 36th Int’l Conf. on Neural Information Processing Systems. New Orleans: Curran Associates Inc., 2022. 22199–22213. find our card
  31. Durmus E, He H, Diab M. FEQA: A question answering evaluation framework for faithfulness assessment in abstractive summarization. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 5055–5070. [doi: 10.18653/v1/2020.… find our card
  32. Dhingra B, Faruqui M, Parikh A, Chang MW, Das D, Cohen W. Handling divergent reference texts when evaluating table-to-text generation. In: Proc. of the 57th Annual Meeting of the Association for Computational Linguistics. Florence: Association for Computational Linguistics, 2019. 4884–4895. [doi: 10… find our card
  33. Manakul P, Liusie A, Gales MJF. SelfCheckGPT: Zero-resource black-box hallucination detection for generative large language models. arXiv:2303.08896, 2023. find
  34. Niu C, Wu YH, Zhu J, Xu SL, Shum K, Zhong R, Song JT, Zhang T. RAGTruth: A hallucination corpus for developing trustworthy retrieval-augmented language models. arXiv:2401.00396, 2024. find
  35. Wang ZY, Wang XY, An B, Yu D, Chen CY. Towards faithful neural table-to-text generation with content-matching constraints. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 1072–1086. [doi: 10.18653/v1/2020.acl-mai… find
  36. Shuster K, Poff S, Chen MY, Kiela D, Weston J. Retrieval augmentation reduces hallucination in conversation. In: Proc. of the Association for Computational Linguistics: EMNLP. Punta Cana: Association for Computational Linguistics, 2021. 3784–3803. [doi: 10.18653/v1/2021.findings-emnlp.320] find our card
  37. Popović M. chrF: Character n-gram F-score for automatic MT evaluation. In: Proc. of the 10th Workshop on Statistical Machine Translation. Lisbon: Association for Computational Linguistics, 2015. 392–395. [doi: 10.18653/v1/W15-3049] find our card
  38. Martindale M, Carpuat M, Duh K, McNamee P. Identifying fluently inadequate output in neural and statistical machine translation. In: Proc. of Machine Translation Summit XVII: Research Track. Dublin: European Association for Machine Translation, 2019. 233–243. find
  39. Banko M, Cafarella MJ, Soderland S, Broadhead M, Etzioni O. Open information extraction from the Web. In: Proc. of the 20th Int’l Joint Conf. on Artifical Intelligence. Hyderabad: Morgan Kaufmann Publishers Inc., 2007. 2670–2676. find our card
  40. Goodrich B, Rao V, Liu PJ, Saleh M. Assessing the factual accuracy of generated text. In: Proc. of the 25th ACM SIGKDD Int’l Conf. on Knowledge Discovery & Data Mining. Anchorage: ACM, 2019. 166–175. [doi: 10.1145/3292500.3330955] — [doi:10.1145/3292500.3330955](https://doi.org/10.1145/3292500.3330955) find
  41. Nan F, Nallapati R, Wang ZG, dos Santos CN, Zhu HH, Zhang DJ, McKeown K, Xiang B. Entity-level factual consistency of abstractive text summarization. In: Proc. of the 16th Conf. of the European Chapter of the Association for Computational Linguistics: Main Volume. Association for Computational Lingu… find
  42. Lee N, Ping W, Xu P, Patwary M, Fung P, Shoeybi M, Catanzaro B. Factuality enhanced language models for open-ended text generation. In: Proc. of the 36th Int’l Conf. on Neural Information Processing Systems. New Orleans: Curran Associates Inc., 2022. 34586–34599. find
  43. Dušek O, Novikova J, Rieser V. Evaluating the state-of-the-art of end-to-end natural language generation: The E2E NLG challenge. Computer Speech & Language, 2020, 59: 123–156. [doi: 10.1016/j.csl.2019.06.009] — [doi:10.1016/j.csl.2019.06.009](https://doi.org/10.1016/j.csl.2019.06.009) find
  44. Dušek O, Kasner Z. Evaluating semantic accuracy of data-to-text generation with natural language inference. In: Proc. of the 13th Int’l Conf. on Natural Language Generation. Dublin: Association for Computational Linguistics, 2020. 131–137. [doi: 10.18653/v1/2020.inlg-1.19] find
  45. Williams A, Nangia N, Bowman S. A broad-coverage challenge corpus for sentence understanding through inference. In: Proc. of the 2018 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies , Vol. 1 (Long Papers). New Orleans: Association for… find
  46. Nie YX, Williams A, Dinan E, Bansal M, Weston J, Kiela D. Adversarial NLI: A new benchmark for natural language understanding. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 4885–4901. [doi: 10.18653/v1/2020.acl… find
  47. Laban P, Schnabel T, Bennett PN, Hearst MA. SummaC: Re-visiting NLI-based models for inconsistency detection in summarization. Trans. of the Association for Computational Linguistics, 2022, 10: 163–177. [doi: 10.1162/TACL_A_00453] find
  48. Kryściński W, McCann B, Xiong CM, Socher R. Evaluating the factual consistency of abstractive text summarization. In: Proc. of the 2020 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2020. 9332–9346. [doi: 10.18653/v1/2020.emnlp-main.750] find
  49. Yin WP, Radev D, Xiong CM. DocNLI: A large-scale dataset for document-level natural language inference. In: Proc. of the 2021 Association for Computational Linguistics. Association for Computational Linguistics, 2021. 4913–4922. [doi: 10.18653/v1/2021. findings-acl.435] find our card
  50. Devlin J, Chang MW, Lee K, Toutanova K. BERT: Pre-training of deep bidirectional transformers for language understanding. arXiv:1810.04805, 2019. find
  51. Liu YH, Ott M, Goyal N, Du JF, Joshi M, Chen DQ, Levy O, Lewis M, Zettlemoyer L, Stoyanov V. RoBERTa: A robustly optimized BERT pretraining approach. arXiv:1907.11692, 2019. find
  52. He PC, Li XD, Gao JF, Chen WZ. DeBERTa: Decoding-enhanced BERT with disentangled attention. arXiv:2006.03654, 2021. find our card
  53. Goyal T, Durrett G. Evaluating Factuality in generation with dependency-level entailment. In: Proc. of the Association for Computational Linguistics: EMNLP. Association for Computational Linguistics, 2020. 3592–3603. [doi: 10.18653/v1/2020.findings-emnlp.322] find our card
  54. Falke T, Ribeiro LFR, Utama PA, Dagan I, Gurevych I. Ranking generated summaries by correctness: An interesting but challenging application for natural language inference. In: Proc. of the 57th Annual Meeting of the Association for Computational Linguistics. Florence: Association for Computational L… find our card
  55. Honovich O, Choshen L, Aharoni R, Neeman E, Szpektor I, Abend O. Q 2 : Evaluating factual consistency in knowledge-grounded dialogues via question generation and question answering. In: Proc. of the 2021 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguis… find
  56. Fabbri AR, Wu CS, Liu WH, Xiong CM. QAFactEval: Improved QA-based factual consistency evaluation for summarization. In: Proc. of the 2022 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Seattle: Association for Computational Linguist… find our card
  57. Barrantes M, Herudek B, Wang R. Adversarial NLI for factual correctness in text summarisation models. arXiv:2005.11739, 2020. — [arXiv:2005.11739](https://arxiv.org/abs/2005.11739) find
  58. Yang ZL, Dai ZH, Yang YM, Carbonell J, Salakhutdinov R, Le QV. XLNet: Generalized autoregressive pretraining for language understanding. In: Proc. of the 33rd Int’l Conf. on Neural Information Processing Systems. Vancouver: Curran Associates Inc., 2019. 5735–5763. find our card
  59. Dziri N, Rashkin H, Linzen T, Reitter D. Evaluating groundedness in dialogue systems: The BEGIN benchmark. arXiv:2105.00071, 2022. — [arXiv:2105.00071](https://arxiv.org/abs/2105.00071) find
  60. Filippova K. Controlled Hallucinations: Learning to generate faithfully from noisy data. In: Proc. of the Association for Computational Linguistics: EMNLP 2020. Association for Computational Linguistics, 2020. 864–870. [doi: 10.18653/v1/2020.findings-emnlp.76] find
  61. Cao M, Dong Y, Cheung J. Hallucinated but factual! Inspecting the factuality of hallucinations in abstractive summarization. In: Proc. of the 60th Annual Meeting of the Association for Computational Linguistics (Vol. 1: Long Papers). Dublin: Association for Computational Linguistics, 2022. 3340–3354… find
  62. Yu JF, Wang XZ, Tu SQ, Cao SL, Zhang-Li D, Lv X, Peng H, Yao ZJ, Zhang XH, Li HM, Li CY, Zhang ZY, Bai YS, Liu YT, Xin A, Lin NY, Yun KF, Gong LL, Chen JH, Wu ZL, Qi YJ, Li WK, Guan Y, Zeng KS, Qi J, Jin HL, Liu JX, Gu Y, Yao Y, Ding N, Hou L, Liu ZY, Xu B, Tang J, Li JZ. KoLA: Carefully benchmarkin… — [arXiv:2306.09296](https://arxiv.org/abs/2306.09296) find
  63. Deng MK, Tan BW, Liu ZZ, Xing E, Hu ZT. Compression, transduction, and creation: A unified framework for evaluating natural language generation. In: Proc. of the 2021 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2021. 7580–7605. [doi: 10.18653… find
  64. Zha YH, Yang YC, Li RC, Hu ZT. AlignScore: Evaluating factual consistency with a unified alignment function. arXiv:2305.16739, 2023. find our card
  65. Yue X, Wang BS, Chen ZR, Zhang K, Su Y, Sun H. Automatic evaluation of attribution by large language models. arXiv:2305.06311, 2023. find
  66. Zhong M, Liu Y, Yin D, Mao YN, Jiao YZ, Liu PF, Zhu CG, Ji H, Han JW. Towards a unified multi-dimensional evaluator for text generation. In: Proc. of the 2022 Conf. on Empirical Methods in Natural Language Processing. Abu Dhabi: Association for Computational Linguistics, 2022. 2023–2038. [doi: 10.18… find
  67. Clark C, Lee K, Chang MW, Kwiatkowski T, Collins M, Toutanova K. BoolQ: Exploring the surprising difficulty of natural yes/no questions. In: Proc. of the 2019 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Vol. 1 (Long and Short Pap… find our card
  68. Zhang TY, Kishore V, Wu F, Weinberger KQ, Artzi Y. BERTScore: Evaluating text generation with BERT. arXiv:1904.09675, 2020. find
  69. Yuan WZ. Neubig G, Liu PF. BARTScore: Evaluating generated text as text generation. In: Proc. of the 35th Int’l Conf. on Neural Information Processing Systems. Curran Associates Inc., 2021. 27263–27277. find our card
  70. Mehri S, Eskenazi M. USR: An unsupervised and reference free evaluation metric for dialog generation. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 681–707. [doi: 10.18653/v1/2020.acl-main.64] find our card
  71. Wei JH, Yao YS, Ton JF, Guo HY, Estornell A, Liu Y. Measuring and reducing LLM hallucination without gold-standard answers. arXiv:2402.10412, 2024. — [arXiv:2402.10412](https://arxiv.org/abs/2402.10412) find
  72. Su WH, Wang CY, Ai QY, Hu YR, Wu ZJ, Zhou YJ, Liu YQ. Unsupervised real-time hallucination detection based on the internal states of large language models. arXiv:2403.06448, 2024. find
  73. Zhang SY, Li Y, Wu R, Huang XT, Chen YR, Xu WH, Qi GL. DEE: Dual-stage explainable evaluation method for text generation. arXiv:2403.11509, 2024. — [arXiv:2403.11509](https://arxiv.org/abs/2403.11509) find
  74. Wang A, Cho K, Lewis M. Asking and answering questions to evaluate the factual consistency of summaries. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 5008–5020. [doi: 10.18653/v1/2020.acl-main.450] find
  75. Nan F, dos Santos CN, Zhu HH, Ng P, McKeow K, Nallapati R, Zhang DJ, Wang ZG, Arnold AO, Xiang B. Improving factual consistency of abstractive summarization via question answering. In: Proc. of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th Int’l Joint Conf. on… find
  76. Shakeri S, dos Santos CN, Zhu HH, Ng P, Nan F, Wang ZG, Nallapati R, Xiang B. End-to-end synthetic data generation for domain adaptation of question answering systems. In: In: Proc. of the 2020 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2020… find
  77. Scialom T, Dray PA, Lamprier S, Piwowarski B, Staiano J, Wang A, Gallinari P. QuestEval: Summarization asks for fact-based evaluation. In: Proc. of the 2021 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2021. 6594–6604. [doi: 10.18653/v1/2021.e… find our card
  78. Rebuffel C, Scialom T, Soulier L, Piwowarski B, Lamprier S, Staiano J, Scoutheeten G, Gallinari P. Data-QuestEval: A referenceless metric for data-to-text semantic evaluation. In: Proc. of the 2021 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics,… find
  79. Yehuda Y, Malkiel I, Barkan O, Weill J, Ronen R, Koenigstein N. In search of truth: An interrogation approach to hallucination detection. arXiv:2403.02889, 2024. — [arXiv:2403.02889](https://arxiv.org/abs/2403.02889) find
  80. Vu T, Iyyer M, Wang XZ, Constant N, Wei J, Wei J, Tar C, Sung YH, Zhou D, Le Q, Luong T. FreshLLMs: Refreshing large language models with search engine augmentation. arXiv:2310.03214, 2023. find our card
  81. Zhang JX, Li ZH, Das K, Malin BA, Kumar S. SAC3: Reliable hallucination detection in black-box language models via semantic-aware cross-check consistency. arXiv:2311.01740, 2024. find
  82. Dinan E, Roller S, Shuster K, Fan A, Auli M, Weston J. Wizard of Wikipedia: Knowledge-powered conversational agents. arXiv:1811.01241, 2019. find
  83. Santhanam S, Hedayatnia B, Gella S, Padmakumar A, Kim S, Liu Y, Hakkani-Tur D. Rome was built in 1776: A case study on factual correctness in knowledge-grounded response generation. arXiv:2110.05456, 2022. — [arXiv:2110.05456](https://arxiv.org/abs/2110.05456) find
  84. Utama P, Bambrick J, Moosavi N, Gurevych I. Falsesum: Generating document-level NLI examples for recognizing factual inconsistency in summarization. In: Proc. of the 2022 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Seattle: Assoc… find
  85. Jiang C, Qi BQ, Hong XY, Fu DY, Cheng Y, Meng FD, Yu M, Zhou BW, Zhou J. On large language models’ hallucination with regard to known facts. arXiv:2403.20009, 2024. find
  86. Li JY, Cheng XX, Zhao WX, Nie JY, Wen JR. HaluEval: A large-scale hallucination evaluation benchmark for large language models. arXiv:2305.11747, 2023. find our card
  87. Lin S, Hilton J, Evans O. TruthfulQA: Measuring how models mimic human falsehoods. In: Proc. of the 60th Annual Meeting of the Association for Computational Linguistics (Vol. 1: Long Papers). Dublin: Association for Computational Linguistics, 2022. 3214–3252. [doi: 10.18653/v1/2022.acl-long.229] find our card
  88. Tang LY, Shalyminov I, Wong AWM, Burnsky J, Vincent JW, Yang YA, Singh S, Feng S, Song H, Su H, Sun LJ, Zhang Y, Mansour S, McKeown K. TofuEval: Evaluating hallucinations of LLMs on topic-focused dialogue summarization. arXiv:2402.13249, 2024. find
  89. Chen KD, Chen Q, Zhou J, He YS, He L. DiaHalu: A dialogue-level hallucination evaluation benchmark for large language models. arXiv:2403.00896, 2024. find
  90. Zhu ZY, Yang YM, Sun ZQ. HaluEval-Wild: Evaluating hallucinations of language models in the wild. arXiv:2403.04307, 2024. — [arXiv:2403.04307](https://arxiv.org/abs/2403.04307) find
  91. Chen SQ, Zhao YR, Zhang JH, Chern IC, Gao SY, Liu PF, He JX. FELM: Benchmarking factuality evaluation of large language models. In: Proc. of the 37th Int’l Conf. on Neural Information Processing Systems. New Orleans: Curran Associates Inc., 2023. 44502–44523. find
  92. Yang SP, Sun RL, Wan XJ. A new benchmark and reverse validation method for passage-level hallucination detection. arXiv:2310. 06498, 2023. find
  93. Lattimer BM, Chen P, Zhang XY, Yang Y. Fast and accurate factual inconsistency detection over long documents. arXiv:2310.13189, 2023. find
  94. Muhlgay D, Ram O, Magar I, Levine Y, Ratner N, Belinkov Y, Abend O, Leyton-Brown K, Shashua A, Shoham Y. Generating benchmarks for factuality evaluation of language models. arXiv:2307.06908, 2024. find
  95. Mündler N, He JX, Jenko S, Vechev M. Self-contradictory hallucinations of large language models: Evaluation, detection and mitigation. arXiv:2305.15852, 2024. find
  96. Kocmi T, Federmann C. Large language models are state-of-the-art evaluators of translation quality. arXiv:2302.14520, 2023. find
  97. Gao MQ, Ruan J, Sun RL, Yin XJ, Yang SP, Wan XJ. Human-like summarization evaluation with chatgpt. arXiv:2304.02554, 2023. find
  98. Liu Y, Iter D, Xu YC, Wang SH, Xu RC, Zhu CG. G-Eval: NLG evaluation using GPT-4 with better human alignment. arXiv:2303. 16634, 2023. find
  99. Min S, Krishna K, Lyu XX, Lewis M, Yih WT, Koh PW, Iyyer M, Zettlemoyer L, Hajishirzi H. FActScore: Fine-grained atomic evaluation of factual precision in long form text generation. arXiv:2305.14251, 2023. find our card
  100. Shafayat S, Kim E, Oh J, Oh A. Multi-FAct: Assessing multilingual LLMs’ multi-regional knowledge using FActScore. arXiv:2402. 18045, 2024. find
  101. Gekhman Z, Herzig J, Aharoni R, Elkind C, Szpektor I. TrueTeacher: Learning factual consistency evaluation with large language models. arXiv:2305.11171, 2023. find
  102. Gardent C, Shimorina A, Narayan S, Perez-Beltrachini L. Creating training corpora for NLG micro-planning. In: Proc. of the 55th Annual Meeting of the Association for Computational Linguistics (Vol. 1: Long Papers). Vancouver: Association for Computational Linguistics, 2017. 179–188. [doi: 10.18653/v… find our card
  103. Gabriel S, Celikyilmaz A, Jha R, Choi Y, Gao JF. GO FIGURE: A meta evaluation of factuality in summarization. In: Proc. of the Association for Computational Linguistics: ACL-IJCNLP 2021. Association for Computational Linguistics, 2021. 478–487. [doi: 10.18653/v1/2021.findings-acl.42] find
  104. Dziri N, Kamalloo E, Milton S, Zaiane O, Yu M, Ponti EM, Reddy S. FaithDial: A faithful benchmark for information-seeking dialogue. Trans. of the Association for Computational Linguistics, 2022, 10: 1473–1490. [doi: 10.1162/tacl_a_00529] find
  105. Cheng ZJ, Dong HY, Wang ZR, Jia R, Guo JQ, Gao Y, Han S, Lou JG, Zhang DM. HiTab: A hierarchical table dataset for question answering and natural language generation. arXiv:2108.06712, 2022. find
  106. Chen ZY, Chen WH, Zha HW, Zhou XY, Zhang YK, Sundaresan S, Wang WY. Logic2Text: High-fidelity natural language generation from logical forms. arXiv:2004.14579, 2020. find
  107. Xu L, Li AQ, Zhu L, Xue H, Zhu CT, Zhao KK, He HN, Zhang XW, Kang QY, Lan ZZ. SuperCLUE: A comprehensive Chinese large language model benchmark. arXiv:2307.15020, 2023. find
  108. Raunak V, Menezes A, Junczys-Dowmunt M. The curious case of hallucinations in neural machine translation. In: Proc. of the 2021 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologys. Association for Computational Linguistics, 2021. 1172–1183… find our card
  109. Nie F, Yao JG, Wang JP, Pan R, Lin CY. A simple recipe towards reducing hallucination in neural surface realisation. In: Proc. of the 57th Annual Meeting of the Association for Computational Linguistics. Florence: Association for Computational Linguistics, 2019. 2673–2679. [doi: 10.18653/v1/P19-1256… find
  110. Liu TY, Zheng X, Chang BB, Sui ZF. Towards faithfulness in open domain table-to-text generation from an entity-centric view. In: Proc. of the AAAI Conf. on Artificial Intelligence. Virtually: AAAI Press, 2021. 13415–13423. [doi: 10.1609/aaai.v35i15.17583] — [doi:10.1609/aaai.v35i15.17583](https://doi.org/10.1609/aaai.v35i15.17583) find
  111. Shen L, Zhan HL, Shen X, Chen HS, Zhao XF, Zhu XD. Identifying untrustworthy samples: Data filtering for open-domain dialogues with Bayesian optimization. In: Proc. of the 30th ACM Int’l Conf. on Information and Knowledge Management. Virtual Event: ACM, 2021. 1598–1608. [doi: 10.1145/3459637.3482352… — [doi:10.1145/3459637.3482352](https://doi.org/10.1145/3459637.3482352) find
  112. Rebuffel C, Roberti M, Soulier L, Scoutheeten G, Cancelliere R, Gallinari P. Controlling hallucinations at word level in data-to-text generation. Data Mining and Knowledge Discovery, 2022, 36(1): 318–354. [doi: 10.1007/s10618-021-00801-4] — [doi:10.1007/s10618-021-00801-4](https://doi.org/10.1007/s10618-021-00801-4) find
  113. Dušek O, Howcroft DM, Rieser V. Semantic noise matters for neural natural language generation. In: Proc. of the 12th Int’l Conf. on Natural Language Generation. Tokyo: Association for Computational Linguistics, 2019. 421–426. [doi: 10.18653/v1/W19-8652] find
  114. Honnibal M, Montani I. spaCy 2: Natural language understanding with bloom embeddings, convolutional neural networks and incremental parsing. Neural Machine Translation. Proc. of the Association for Computational Linguistics. 2017. 688–697. find
  115. Junczys-Dowmunt M. Dual conditional cross-entropy filtering of noisy parallel corpora. arXiv:1809.00197, 2019. find our card
  116. Zhang BL, Nagesh A, Knight K. Parallel corpus filtering via pre-trained language models. arXiv:2005.06166, 2020. find
  117. Nie F, Wang JP, Yao JG, Pan R, Lin CY. Operation-guided neural networks for high fidelity data-to-text generation. In: Proc. of the 2018 Conf. on Empirical Methods in Natural Language Processing. Brussels: Association for Computational Linguistics, 2018. 3879–3889. [doi: 10.18653/v1/D18-1422] find
  118. Wang TS, Ladhak F, Durmus E, He H. Improving faithfulness by augmenting negative summaries from fake documents. In: Proc. of the 2022 Conf. on Empirical Methods in Natural Language Processing. Abu Dhabi: Association for Computational Linguistics, 2022. 11913–11921. [doi: 10.18653/v1/2022.emnlp-main.… find
  119. Longpre S, Perisetla K, Chen A, Ramesh N, DuBois C, Singh S. Entity-based knowledge conflicts in question answering. arXiv:2109.05052, 2022. find our card
  120. Chen SH, Zhang F, Sone K, Roth D. Improving faithfulness in abstractive summarization with contrast candidate generation and selection. In: Proc. of the 2021 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computation… find
  121. Biten AF, Gómez L, Karatzas D. Let there be a clock on the beach: Reducing object hallucination in image captioning. In: Proc. of the 2022 IEEE/CVF Winter Conf. on Applications of Computer Vision. Waikoloa: IEEE, 2022. 2473–2482. [doi: 10.1109/WACV51458.2022.00253] — [doi:10.1109/wacv51458.2022.00253](https://doi.org/10.1109/wacv51458.2022.00253) find
  122. Bi B, Wu C, Yan M, Wang W, Xia JN, Li CL. Incorporating external knowledge into machine reading for generative question answering. In: Proc. of the 2019 Conf. on Empirical Methods in Natural Language Processing and the 9th Int’l Joint Conf. on Natural Language Processing. Hong Kong: Association for… find
  123. Lim J, Kang M, Hur Y, Jung S, Kim J, Jang Y, Lee D, Ji H, Shin D, Kim S, Lim H. You truly understand what I need: Intellectual and friendly dialogue agents grounding knowledge and persona. arXiv:2301.02401, 2023. — [arXiv:2301.02401](https://arxiv.org/abs/2301.02401) find
  124. Zhu CG, William H, Xu RC, Zeng QK, Zeng M, Huang XD, Jiang M. Enhancing factual consistency of abstractive summarization. In: Proc. of the 2021 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics… find
  125. Huang LY, Wu LF, Wang L. Knowledge graph-augmented abstractive summarization with semantic-driven cloze reward. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 5094–5107. [doi: 10.18653/v1/2020.acl-main.457] find
  126. Liu A, Sap M, Lu XM, Swayamdipta S, Bhagavatula C, Smith NA, Choi Y. DExperts: Decoding-time controlled text generation with experts and anti-experts. arXiv:2105.03023, 2021. find our card
  127. Xiao YJ, Wang WY. On hallucination and predictive uncertainty in conditional language generation. In: Proc. of the 16th Conf. of the European Chapter of the Association for Computational Linguistics: Main Volume. Association for Computational Linguistics, 2021. 2734–2744. [doi: 10.18653/v1/2021.eacl… find
  128. Song KQ, Lebanoff L, Guo QP, Qiu XP, Xue XY, Li C, Yu D, Liu F. Joint parsing and generation for abstractive summarization. In: Proc. of the 34th AAAI Conf. on Artificial Intelligence. New York: AAAI, 2020. 8894–8901. [doi: 10.1609/aaai.v34i05.6419] — [doi:10.1609/aaai.v34i05.6419](https://doi.org/10.1609/aaai.v34i05.6419) find
  129. Balakrishnan A, Rao JF, Upasani K, White M, Subba R. Constrained decoding for neural NLG from compositional representations in task-oriented dialogue. In: Proc. of the 57th Annual Meeting of the Association for Computational Linguistics. Florence: Association for Computational Linguistics, 2019. 831… find
  130. Lee K, Firat O, Agarwal A, Fannjiang C, Sussillo D. Hallucinations in neural machine translation. In: Proc. of the 2019 Int’l Conf. on Learning Representations. 2019. find
  131. Kang D, Hashimoto T. Improved natural language generation via loss truncation. arXiv:2004.14589, 2020. find
  132. Yoon S, Yoon E, Yoon HS, Kim J, Yoo CD. Information-theoretic text hallucination reduction for video-grounded dialogue. arXiv:2212. 05765, 2022. find
  133. Dai WL, Liu ZH, Ji ZW, Su D, Fung P. Plausible may not be faithful: Probing object hallucination in vision-language pre-training. arXiv:2210.07688, 2023. find
  134. Li T, Beirami A, Sanjabi M, Smith V. Tilted empirical risk minimization. In: Proc. of the 9th Int’l Conf. on Learning Representations. Virtual Event, 2021. find
  135. Wang CJ, Sennrich R. On exposure bias, hallucination and domain shift in neural machine translation. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 3544–3552. [doi: 10.18653/v1/2020.acl-main.326] find
  136. Ranzato MA, Chopra S, Auli M, Zaremba W. Sequence level training with recurrent neural networks. In: Proc. of the 4th Int’l Conf. on Learning Representations. San Juan, 2016. find our card
  137. Mesgar M, Simpson E, Gurevych I. Improving factual consistency between a response and persona facts. In: Proc. of the 16th Conf. of the European Chapter of the Association for Computational Linguistics: Main Volume. Association for Computational Linguistics, 2021. 549–562. [doi: 10.18653/v1/2021.eac… find
  138. Song HY, Zhang WN, Hu JW, Liu T. Generating persona consistent dialogues by exploiting natural language inference. In: Proc. of the 34th AAAI Conf. on Artificial Intelligence. New York: AAAI, 2020. 8878–8885. [doi: 10.1609/aaai.v34i05.6417] — [doi:10.1609/aaai.v34i05.6417](https://doi.org/10.1609/aaai.v34i05.6417) find
  139. Schulman J. Reinforcement learning from human feedback: Progress and challenges. 2023. https://eecs.berkeley.edu/research/colloquium/230419-2/ ↗
  140. Weng RX, Yu H, Wei XP, Liu WH. Towards enhancing faithfulness for neural machine translation. In: Proc. of the 2020 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2020. 2675–2684. [doi: 10.18653/v1/2020.emnlp-main.212] find
  141. Li CL, Bi B, Yan M, Wang W, Huang SF. Addressing semantic drift in generative question answering with auxiliary extraction. In: Proc. of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th Int’l Joint Conf. on Natural Language Processing. Association for Computation… find
  142. Cao M, Dong Y, Wu JP, Cheung JCK. Factual error correction for abstractive summarization models. In: Proc. of the 2020 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2020. 6251–6258. [doi: 10.18653/v1/2020.emnlp-main.506] find
  143. Dong Y, Wang SH, Gan Z, Cheng Y, Cheung JCK, Liu JJ. Multi-fact correction in abstractive text summarization. In: Proc. of the 2020 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2020. 9320–9331. [doi: 10.18653/v1/2020.emnlp-main.749] find
  144. Song HY, Wang Y, Zhang WN, Liu XJ, Liu T. Generate, delete and rewrite: A three-stage framework for improving persona consistency of dialogue generation. In: Proc. of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2020. 5821–5831.… find
  145. Dziri N, Madotto A, Zaïane O, Bose AJ. Neural path hunter: Reducing hallucination in dialogue systems via path grounding. In: Proc. of the 2021 Conf. on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2021. 2197–2214. [doi: 10.18653/v1/2021.emnlp-main.168… find
  146. Qiu YF, Ziser Y, Korhonen A, Ponti EM, Cohen SB. Detecting and mitigating hallucinations in multilingual summarisation. arXiv:2305.13632, 2023. find
  147. Ilharco G, Ribeiro MT, Wortsman M, Gururangan S, Schmidt L, Hajishirzi H, Farhadi A. Editing models with task arithmetic. arXiv:2212.04089, 2023. find our card
  148. Choubey PK, Fabbri AR, Vig J, Wu CS, Liu WH, Rajani NF. CaPE: Contrastive parameter ensembling for reducing hallucination in abstractive summarization. arXiv:2110.07166, 2022. find
  149. Lightman H, Kosaraju V, Burda Y, Edwards H, Baker B, Lee T, Leike J, Schulman J, Sutskever I, Cobbe K. Let’s verify step by step. arXiv:2305.20050, 2023. find our card
  150. Li YF, Du YF, Zhou K, Wang JP, Zhao WX, Wen JR. Evaluating object hallucination in large vision-language models. arXiv:2305. 10355, 2023. find our card
  151. Kumar K. Geotechnical Parrot Tales (GPT): Harnessing large language models in geotechnical engineering. arXiv:2304.02138, 2023. — [arXiv:2304.02138](https://arxiv.org/abs/2304.02138) find
  152. Zhang MR, Press O, Merrill W, Liu A, Smith NA. How language model hallucinations can snowball. arXiv:2305.13534, 2023. find our card
  153. Martino A, Iannelli M, Truong C. Knowledge injection to counter large language model (LLM) hallucination. In: Proc. of the Semantic Web: ESWC 2023 Satellite Events. Hersonissos: Springer, 2023. 182–185. [doi: 10.1007/978-3-031-43458-7_34] — [doi:10.1007/978-3-031-43458-7_34](https://doi.org/10.1007/978-3-031-43458-7_34) find
  154. Li CL, Xu WR, Li S, Gao S. Guiding generation for abstractive text summarization based on key information guide network. In: Proc. of the 2018 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Vol.2 (Short Papers). New Orleans: Associa… find
  155. Saito I, Nishida K, Nishida K, Tomita J. Abstractive summarization with combination of pre-trained sequence-to-sequence and saliency models. arXiv:2003.13028, 2020. — [arXiv:2003.13028](https://arxiv.org/abs/2003.13028) find
  156. Dong Y, Wieting J, Verga P. Faithful to the document or to the world? Mitigating hallucinations via entity-linked knowledge in abstractive summarization. arXiv:2204.13761, 2022. find
  157. Zhang S, Pan LM, Zhao JZ, Wang WY. The knowledge alignment problem: Bridging human and external knowledge for large language models. arXiv:2305.13669, 2024. find
  158. Bran AM, Cox S, Schilter O, Baldassari C, White AD, Schwaller P. ChemCrow: Augmenting large-language models with chemistry tools. arXiv:2304.05376, 2023. find
  159. Ullah N, Mohanta PP. Thinking hallucination for video captioning. In: Proc. of the 16th Asian Conf. on Computer Vision. Macao: Springer, 2022. 623–640. [doi: 10.1007/978-3-031-26316-3_37] — [doi:10.1007/978-3-031-26316-3_37](https://doi.org/10.1007/978-3-031-26316-3_37) find
  160. Li JY, Chen J, Ren RY, Cheng XX, Zhao WX, Nie JY, Wen JR. The dawn after the dark: An empirical study on factuality hallucination in large language models. arXiv:2401.03205, 2024. find
  161. Xu ZW, Jain S, Kankanhalli M. Hallucination is inevitable: An innate limitation of large language models. arXiv:2401.11817, 2024. find our card
  162. Bi BL, Liu SH, Wang YW, Mei LR, Cheng XQ. Is factuality decoding a free lunch for LLMs? Evaluation on knowledge editing benchmark. arXiv:2404.00216, 2024. — [arXiv:2404.00216](https://arxiv.org/abs/2404.00216) find
  163. Huang L, Yu WJ, Ma WT, Zhong WH, Feng ZY, Wang HT, Chen QL, Peng WH, Feng XC, Qin B, Liu T. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. arXiv:2311.05232, 2024. find our card
  164. Jiang XH, Tian YX, Hua FR, Xu CJ, Wang YZ, Guo J. A survey on large language model hallucination via a creativity perspective. arXiv:2402.06647, 2024. — [arXiv:2402.06647](https://arxiv.org/abs/2402.06647) find
  165. Lee M. A mathematical investigation of hallucination and creativity in gpt models. Mathematics, 2023, 11(10): 2320. [doi: 10.3390/math11102320] — [doi:10.3390/math11102320](https://doi.org/10.3390/math11102320) find
  166. Wang F. Lighthouse: A survey of AGI hallucination. arXiv:2401.06792, 2024. — [arXiv:2401.06792](https://arxiv.org/abs/2401.06792) find
  167. Rawte V, Sheth A, Das A. A survey of hallucination in large foundation models. arXiv:2309.05922, 2023. find our card
  168. Jablonka KM, Ai QX, Al-Feghali A, et al. 14 examples of how LLMs can transform materials science and chemistry: A reflection on a large language model hackathon. Digital Discovery, 2023, 2(5): 1233–1250. [doi: 10.1039/D3DD00113J] — [doi:10.1039/d3dd00113j](https://doi.org/10.1039/d3dd00113j) find
  169. Gupta AK, Raghavachari K. Three-dimensional convolutional neural networks utilizing molecular topological features for accurate atomization energy predictions. Journal of Chemical Theory and Computation, 2022, 18(4): 2132–2143. [doi: 10.1021/acs.jctc.1c00504] — [doi:10.1021/acs.jctc.1c00504](https://doi.org/10.1021/acs.jctc.1c00504) find
  170. Völker C, Rug T, Jablonka KM, Kruschwitz S. LLMs can design sustainable concrete—A systematic benchmark. 2024. [doi: 10.21203/rs.3.rs-3913272/v1] — [doi:10.21203/rs.3.rs-3913272/v1](https://doi.org/10.21203/rs.3.rs-3913272/v1) find
  171. Hong. tuhz/Molecule-discovery-by-context: LLM hackthon release (0.1). Zenodo. 2023. [doi: 10.5281/zenodo.8122087] — [doi:10.5281/zenodo.8122087](https://doi.org/10.5281/zenodo.8122087) find
  172. Abramson J, Adler J, Dunger J, et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature, 2024, 630(8016): 493–500. [doi: 10.1038/s41586-024-07487-w] find
  173. Bleumink AG, Shikhule A. Keeping AI honest in education: Identifying GPT-generated text. Edukado AI Research, 2023: 1–5. find
  174. Liu R, Sumers TR, Dasgupta I, et al. How do large language models navigate conflicts between honesty and helpfulness? arXiv:2402. 07282, 2024. find
  175. Rashkin H, Reitter D, Tomar GS, et al. Increasing faithfulness in knowledge-grounded dialogue with controllable features. arXiv:2107.06963, 2021. find
  176. Wu ZQ, Galley M, Brockett C, Zhang YZ, Gao X, Quirk C, Koncel-Kedziorski R, Gao JF, Hajishirzi H, Ostendorf M, Dolan B. A controllable model of grounded response generation. In: Proc. of the 35th AAAI Conf. on Artificial Intelligence. Virtually: AAAI, 2021. 14085–14093. [doi: 10.1609/aaai.v35i16.176… — [doi:10.1609/aaai.v35i16.17658](https://doi.org/10.1609/aaai.v35i16.17658) find
  177. Yu L, Cao M, Cheung JCK, Dong Y. Mechanistic understanding and mitigation of language model non-factual hallucinations. arXiv:2403.18167, 2024. find
  178. Yang YQ, Chern E, Qiu XP, Neubig G, Liu PF. Alignment for honesty. arXiv:2312.07000, 2023. find
  179. Tonmoy SMTI, Zaman SMM, Jain V, Rani A, Rawte V, Chadha A, Das A. A comprehensive survey of hallucination mitigation techniques in large language models. arXiv:2401.01313, 2024. find our card
  180. Thorne J, Vlachos A, Christodoulopoulos C, Mittal A. Evaluating adversarial attacks against multiple fact verification systems. In: Proc. of the 2019 Conf. on Empirical Methods in Natural Language Processing and the 9th Int’l Joint Conf. on Natural Language Processing. Hong Kong: Association for Com… find
  181. Zhang QS, Wu YN, Zhu SC. Interpretable convolutional neural networks. In: Proc. of the 2018 IEEE/CVF Conf. on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018. 8827–8836. [doi: 10.1109/CVPR.2018.00920] — [doi:10.1109/cvpr.2018.00920](https://doi.org/10.1109/cvpr.2018.00920) find
  182. Chen CF, Li O, Tao DF, Barnett AJ, Su J, Rudin C. This looks like that: Deep learning for interpretable image recognition. In: Proc. of the 33rd Int’l Conf. on Neural Information Processing Systems. Vancouver: Curran Associates Inc., 2019. 8930–8941. find
  183. Wang DL, Yang S, Ouyang WL, Li BP, Zhou Y. Explainability of artificial intelligence: Development and application. Computer Science, 2023, 50(6A): 220600212 (in Chinese with English abstract). [doi: 10.11896/jsjkx.220600212] — [doi:10.11896/jsjkx.220600212](https://doi.org/10.11896/jsjkx.220600212) find
  184. Ribeiro M, Singh S, Guestrin C. “Why should I trust you?”: Explaining the predictions of any classifier. In: Proc. of the 22nd ACM SIGKDD Int’l Conf. on Knowledge Discovery and Data Mining. San Francisco: ACM, 2016. 1135–1144. [doi: 10.1145/2939672.2939778] find our card
  185. Hinton G, Vinyals O, Dean J. Distilling the knowledge in a neural network. arXiv:1503.02531, 2015. find our card
  186. Zhao L, Peng X, Chen YX, Kapadia M, Metaxas DN. Knowledge as priors: Cross-modal knowledge generalization for datasets without superior knowledge. In: Proc. of the 2020 IEEE/CVF Conf. on Computer Vision and Pattern Recognition. Seattle: IEEE, 2020. 6527–6536. [doi: 10.1109/CVPR42600.2020.00656] — [doi:10.1109/cvpr42600.2020.00656](https://doi.org/10.1109/cvpr42600.2020.00656) find
  187. Sundararajan M, Taly A, Yan QQ. Axiomatic attribution for deep networks. In: Proc. of the 34th Int’l Conf. on Machine Learning. Sydney: JMLR.org, 2017. 3319–3328. find
  188. Smilkov D, Thorat N, Kim B, Viégas F, Wattenberg M. SmoothGrad: Removing noise by adding noise. arXiv:1706.03825, 2017. — [arXiv:1706.03825](https://arxiv.org/abs/1706.03825) find
  189. Lin ZH, Feng MW, dos Santos CN, Yu M, Xiang B, Zhou BW, Bengio Y. A structured self-attentive sentence embedding. In: Proc. of the 5th Int’l Conf. on Learning Representations. 2017. find our card
  190. Ali S, Abuhmed T, El-Sappagh S, Muhammad K, Alonso-Moral JM, Confalonieri R, Guidotti R, Del Ser J, Díaz-Rodríguez N, Herrera F. Explainable artificial intelligence (XAI): What we know and what is left to attain Trustworthy Artificial Intelligence. Information Fusion, 2023, 99: 101805. [doi: 10.1016… — [doi:10.1016/j.inffus.2023.101805](https://doi.org/10.1016/j.inffus.2023.101805) find
  191. Michaud EJ, Liu ZM, Girit U, Tegmark M. The quantization model of neural scaling. arXiv:2303.13506, 2024. find
  192. Villalobos P, Ho A, Sevilla J, Besiroglu T, Heim L, Hobbhahn M, Ho A. Will we run out of data? Limits of LLM scaling based on human-generated data. arXiv:2211.04325, 2024. find
  193. Li BH, Hou YT, Che WX. Data augmentation approaches in natural language processing: A survey. arXiv:2110.01852, 2022. — [arXiv:2110.01852](https://arxiv.org/abs/2110.01852) find
  194. Li W, Gao C, Niu GC, Xiao XY, Liu H, Liu JC, Wu H, Wang HF. UNIMO: Towards unified-modal understanding and generation via cross-modal contrastive learning. In: Proc. of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th Int’l Joint Conf. on Natural Language Process… find
  195. Thawani A, Pujara J, Ilievski F, Szekely P. Representing numbers in NLP: A survey and a vision. In: Proc. of the 2021 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics, 2021. 644–656. [doi: 10.… find
  196. Kıcıman E, Ness R, Sharma A, Tan C. Causal reasoning and large language models: Opening a new frontier for causality. arXiv:2305. 00050, 2023. find
  197. Guidotti R, Monreale A, Ruggieri S, Turini F, Giannotti F, Pedreschi D. A survey of methods for explaining black box models. ACM Computing Surveys, 2018, 51(5): 93. [doi: 10.1145/3236009] — [doi:10.1145/3236009](https://doi.org/10.1145/3236009) find
  198. Liang Z, Wang HZ, Dai JJ, Shao XY, Ding XO, Mu TY. Interpretability of entity matching based on pre-trained language model. Ruan Jian Xue Bao/Journal of Software, 2023, 34(3): 1087–1108 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/6794. htm [doi: 10.13328/j.cnki.jos.006794] — [doi:10.13328/j.cnki.jos.006794](https://doi.org/10.13328/j.cnki.jos.006794) ↗
  199. Amini A, Gabriel S, Lin SC, Koncel-Kedziorski R, Choi Y, Hajishirzi H. MathQA: Towards interpretable math word problem solving with operation-based formalisms. In: Proc. of the 2019 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Vol… find our card
  200. Zhou MT, Huang ML, Zhu XY. An interpretable reasoning network for multi-relation question answering. In: Proc. of the 27th Int’l Conf. on Computational Linguistics. Santa Fe: Association for Computational Linguistics, 2018. 2010–2022. find
  201. Li W, Wu WH, Chen MY, Liu JC, Xiao XY, Wu H. Faithfulness in natural language generation: A systematic survey of analysis, evaluation and optimization methods. arXiv:2203.05227, 2022. — [arXiv:2203.05227](https://arxiv.org/abs/2203.05227) find
  202. Li G, Peng X, Wang QX, Xie T, Jin Z, Wang J, Ma XX, Li XD. Challenges from LLMs as a natural language based human-machine collaborative tool for software development and evolution. Ruan Jian Xue Bao/Journal of Software, 2023, 34(10): 4601–4606 (in Chinese with English abstract). http://www.jos.org.c… — [doi:10.13328/j.cnki.jos.007008](https://doi.org/10.13328/j.cnki.jos.007008) ↗

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/liu-2024-hallucinations-large-language/