Elamur Research⌕
☰
home / papers / 2023 / Benchmarking Foundation Models with Language-Model-as-an-Examiner

Benchmarking Foundation Models with Language-Model-as-an-Examiner

Paper · 2023 · Yushi Bai, Jiahao Ying, Yixin Cao et al.

Summary

Open-ended question answering benchmarks face two obstacles. Test items may already appear in pre-training data, which inflates scores. Free-form answers are also hard to grade automatically because no single reference covers every acceptable response. Bai et al. propose Language-Model-as-an-Examiner, in which a strong LM both writes the questions and grades other models' answers without references. Since the examiner creates questions from its own knowledge, it is better equipped to judge them, and fresh question sets can be regenerated if leakage becomes a concern. With GPT-4 as examiner, the authors build LMExamQA. It contains questions drawn from many Google Trends domains, spread over three cognitive levels, plus multi-turn follow-ups that probe depth. Grading combines Likert-scale scores with response rankings, which the authors report match human judgments much better than earlier metrics. To reduce the bias of a single examiner, a decentralized peer-examination scheme lets several models examine one another. Eight popular foundation models are benchmarked, offering a scalable and renewable alternative to static, human-curated QA sets.

Key points

  • Motivating pilot study: GPT-4 acting only as a judge of LLaMA answers on Natural Questions made wrong calls in 18 of 100 cases. The authors attribute this mainly to gaps in the judge's own knowledge.
  • LMExamQA uses the Google Trends Categories as its domain taxonomy. Questions are labeled by Bloom's taxonomy levels as memorization (22%), comprehension (43%) and analysis (35%). This mix is more balanced and harder than MS MARCO, SQuAD2.0, Natural Questions or ELI5.
  • Leakage check: questions were compared with Google search results (100 sampled questions) and with the Flan collection (1,836 tasks, 15M instances) via BM25 retrieval. Average ROUGE-L to the closest match was only 0.293 for web data and 0.046 for Flan.
  • Peer-examination produces more diverse questions and evens out individual examiners' preferences. However, only four accessible models were judged capable enough to act as examiners.
  • Published at NeurIPS 2023 (Datasets and Benchmarks track). Data and results are available at lmexam.xlore.cn.

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, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksSocial Sciences · 1,954 worksEducation · 212 works

Cite

Cite as Yushi Bai, Jiahao Ying, Yixin Cao, Xin Lv, Yuze He, Xiaozhi Wang, Jifan Yu, Kaisheng Zeng, Yijia Xiao, Haozhe Lyu, Jiayin Zhang, Juanzi Li and Lei Hou. 2023. Benchmarking Foundation Models with Language-Model-as-an-Examiner. arXiv:2306.04181.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2306.04181 · arXiv 2306.04181

@misc{bai2023benchmarking,
  title = {{Benchmarking Foundation Models with Language-Model-as-an-Examiner}},
  author = {Yushi Bai, Jiahao Ying, Yixin Cao, Xin Lv, Yuze He, Xiaozhi Wang, Jifan Yu, Kaisheng Zeng, Yijia Xiao, Haozhe Lyu, Jiayin Zhang, Juanzi Li and Lei Hou},
  year = {2023},
  doi = {10.48550/arxiv.2306.04181},
  eprint = {2306.04181},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2306.04181}
}

References

64 references, 35 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. R. Bommasani, D. A. Hudson, E. Adeli, R. Altman, S. Arora, S. von Arx, M. S. Bernstein, J. Bohg, A. Bosselut, E. Brunskill, et al., “On the opportunities and risks of foundation models,” arXiv preprint arXiv:2108.07258, 2021.
  2. OpenAI, “Introducing chatgpt,” 2022. find
  3. H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozière, N. Goyal, E. Hambro, F. Azhar, et al., “Llama: Open and efficient foundation language models,” arXiv preprint arXiv:2302.13971, 2023.
  4. A. Chowdhery, S. Narang, J. Devlin, M. Bosma, G. Mishra, A. Roberts, P. Barham, H. W. Chung, C. Sutton, S. Gehrmann, et al., “Palm: Scaling language modeling with pathways,” arXiv preprint arXiv:2204.02311, 2022.
  5. Z. Ji, N. Lee, R. Frieske, T. Yu, D. Su, Y. Xu, E. Ishii, Y. J. Bang, A. Madotto, and P. Fung, “Survey of hallucination in natural language generation,” ACM Computing Surveys, vol. 55, no. 12, pp. 1–38, 2023.
  6. A. Wang, Y. Pruksachatkun, N. Nangia, A. Singh, J. Michael, F. Hill, O. Levy, and S. Bowman, “Superglue: A stickier benchmark for general-purpose language understanding systems,” Advances in neural information processing systems, vol. 32, 2019.
  7. D. Hendrycks, C. Burns, S. Basart, A. Zou, M. Mazeika, D. Song, and J. Steinhardt, “Measuring massive multitask language understanding,” in International Conference on Learning Representations, 2021.
  8. P. Liang, R. Bommasani, T. Lee, D. Tsipras, D. Soylu, M. Yasunaga, Y. Zhang, D. Narayanan, Y. Wu, A. Kumar, et al., “Holistic evaluation of language models,” arXiv preprint arXiv:2211.09110, 2022. find
  9. W. Zhong, R. Cui, Y. Guo, Y. Liang, S. Lu, Y. Wang, A. Saied, W. Chen, and N. Duan, “Agieval: A human-centric benchmark for evaluating foundation models,” arXiv preprint arXiv:2304.06364, 2023.
  10. J. Yu, X. Wang, S. Tu, S. Cao, D. Zhang-Li, X. Lv, H. Peng, Z. Yao, X. Zhang, H. Li, et al., “Kola: Carefully benchmarking world knowledge of large language models,” arXiv preprint arXiv:2306.09296, 2023. — [arXiv:2306.09296](https://arxiv.org/abs/2306.09296) find
  11. Y. Bai, X. Lv, J. Zhang, H. Lyu, J. Tang, Z. Huang, Z. Du, X. Liu, A. Zeng, L. Hou, Y. Dong, J. Tang, and J. Li, “Longbench: A bilingual, multitask benchmark for long context understanding,” arXiv preprint arXiv:2308.14508, 2023.
  12. A. Rogers, M. Gardner, and I. Augenstein, “Qa dataset explosion: A taxonomy of nlp resources for question answering and reading comprehension,” ACM Computing Surveys, vol. 55, no. 10, pp. 1–45, 2023. find
  13. C.-Y. Lin, “Rouge: A package for automatic evaluation of summaries,” in Text summarization branches out, pp. 74–81, 2004.
  14. T. Zhang, V. Kishore, F. Wu, K. Q. Weinberger, and Y. Artzi, “Bertscore: Evaluating text generation with bert,” in International Conference on Learning Representations, 2020. find
  15. A. Chen, G. Stanovsky, S. Singh, and M. Gardner, “Evaluating question answering evaluation,” in Proceedings of the 2nd workshop on machine reading for question answering, pp. 119–124, 2019. find
  16. K. Krishna, A. Roy, and M. Iyyer, “Hurdles to progress in long-form question answering,” in Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2021.
  17. J. Fu, S.-K. Ng, Z. Jiang, and P. Liu, “Gptscore: Evaluate as you desire,” arXiv preprint arXiv:2302.04166, 2023. find
  18. Y. Liu, D. Iter, Y. Xu, S. Wang, R. Xu, and C. Zhu, “Gpteval: Nlg evaluation using gpt-4 with better human alignment,” arXiv preprint arXiv:2303.16634, 2023.
  19. J. Wang, Y. Liang, F. Meng, H. Shi, Z. Li, J. Xu, J. Qu, and J. Zhou, “Is chatgpt a good nlg evaluator? a preliminary study,” arXiv preprint arXiv:2303.04048, 2023.
  20. OpenAI, “Openai: Gpt-4,” 2023. find
  21. T. Kwiatkowski, J. Palomaki, O. Redfield, M. Collins, A. Parikh, C. Alberti, D. Epstein, I. Polosukhin, J. Devlin, K. Lee, et al., “Natural questions: a benchmark for question answering research,” Transactions of the Association for Computational Linguistics, vol. 7, pp. 453–466, 2019.
  22. T. Nguyen, M. Rosenberg, X. Song, J. Gao, S. Tiwary, R. Majumder, and L. Deng, “Ms marco: A human generated machine reading comprehension dataset,” choice, vol. 2640, p. 660, 2016.
  23. P. Rajpurkar, J. Zhang, K. Lopyrev, and P. Liang, “Squad: 100,000+ questions for machine comprehension of text,” in Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pp. 2383–2392, 2016.
  24. P. Rajpurkar, R. Jia, and P. Liang, “Know what you don’t know: Unanswerable questions for squad,” in Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pp. 784–789, 2018. — Know What You Don't Know: Unanswerable Questions for SQuAD
  25. J. Berant, A. Chou, R. Frostig, and P. Liang, “Semantic parsing on freebase from question-answer pairs,” in Proceedings of the 2013 conference on empirical methods in natural language processing, pp. 1533–1544, 2013.
  26. T. Mihaylov, P. Clark, T. Khot, and A. Sabharwal, “Can a suit of armor conduct electricity? a new dataset for open book question answering,” in Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pp. 2381–2391, 2018. find
  27. A. Fan, Y. Jernite, E. Perez, D. Grangier, J. Weston, and M. Auli, “Eli5: Long form question answering,” in Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp. 3558–3567, 2019.
  28. K. Papineni, S. Roukos, T. Ward, and W.-J. Zhu, “Bleu: a method for automatic evaluation of machine translation,” in Proceedings of the 40th annual meeting of the Association for Computational Linguistics, pp. 311–318, 2002.
  29. S. Banerjee and A. Lavie, “Meteor: An automatic metric for mt evaluation with improved correlation with human judgments,” in Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summarization, pp. 65–72, 2005.
  30. A. Chen, G. Stanovsky, S. Singh, and M. Gardner, “Mocha: A dataset for training and evaluating generative reading comprehension metrics,” in Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 6521–6532, 2020. find
  31. C. Si, C. Zhao, and J. Boyd-Graber, “What’s in a name? answer equivalence for open-domain question answering,” in Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pp. 9623–9629, 2021.
  32. T. Sellam, D. Das, and A. Parikh, “Bleurt: Learning robust metrics for text generation,” in Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 7881–7892, 2020.
  33. R. Rei, C. Stewart, A. C. Farinha, and A. Lavie, “Comet: A neural framework for mt evaluation,” in Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 2685–2702, 2020.
  34. W. Yuan, G. Neubig, and P. Liu, “Bartscore: Evaluating generated text as text generation,” Advances in Neural Information Processing Systems, vol. 34, pp. 27263–27277, 2021.
  35. A. B. Sai, A. K. Mohankumar, and M. M. Khapra, “A survey of evaluation metrics used for nlg systems,” ACM Computing Surveys (CSUR), vol. 55, no. 2, pp. 1–39, 2022. find
  36. W. Zhao, M. Peyrard, F. Liu, Y. Gao, C. M. Meyer, and S. Eger, “Moverscore: Text generation evaluating with contextualized embeddings and earth mover distance,” in Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2019.
  37. S. Gehrmann, E. Clark, and T. Sellam, “Repairing the cracked foundation: A survey of obstacles in evaluation practices for generated text,” arXiv preprint arXiv:2202.06935, 2022. find
  38. M. Zhong, Y. Liu, D. Yin, Y. Mao, Y. Jiao, P. Liu, C. Zhu, H. Ji, and J. Han, “Towards a unified multi-dimensional evaluator for text generation,” in Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pp. 2023–2038, 2022.
  39. W.-L. Chiang, Z. Li, Z. Lin, Y. Sheng, Z. Wu, H. Zhang, L. Zheng, S. Zhuang, Y. Zhuang, J. E. Gonzalez, I. Stoica, and E. P. Xing, “Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality,” March 2023. find
  40. Y. Chen, R. Wang, H. Jiang, S. Shi, and R. Xu, “Exploring the use of large language models for reference-free text quality evaluation: A preliminary empirical study,” arXiv preprint arXiv:2304.00723, 2023. — [arXiv:2304.00723](https://arxiv.org/abs/2304.00723) find
  41. W. Yidong, Y. Zhuohao, Z. Zhengran, Y. Linyi, H. Qiang, W. Cunxiang, C. Hao, J. Chaoya, X. Rui, W. Jindong, X. Xing, Y. Wei, Z. Shikun, and Z. Yue, “Pandalm: Reproducible and automated language model assessment.” https://github.com/WeOpenML/PandaLM, 2023. ↗
  42. M. Gao, J. Ruan, R. Sun, X. Yin, S. Yang, and X. Wan, “Human-like summarization evaluation with chatgpt,” arXiv preprint arXiv:2304.02554, 2023.
  43. A. R. Fabbri, W. Kryściński, B. McCann, C. Xiong, R. Socher, and D. Radev, “Summeval: Re-evaluating summarization evaluation,” Transactions of the Association for Computational Linguistics, vol. 9, pp. 391–409, 2021. find
  44. H. Chen, D. M. Vo, H. Takamura, Y. Miyao, and H. Nakayama, “Storyer: Automatic story evaluation via ranking, rating and reasoning,” Journal of Natural Language Processing, vol. 30, no. 1, pp. 243–249, 2023. find
  45. B. S. Bloom and D. R. Krathwohl, Taxonomy of educational objectives: The classification of educational goals. Book 1, Cognitive domain. longman, 2020. find
  46. E. Choi, H. He, M. Iyyer, M. Yatskar, W.-t. Yih, Y. Choi, P. Liang, and L. Zettlemoyer, “Quac: Question answering in context,” in Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pp. 2174–2184, 2018.
  47. S. Reddy, D. Chen, and C. D. Manning, “Coqa: A conversational question answering challenge,” Transactions of the Association for Computational Linguistics, vol. 7, pp. 249–266, 2019.
  48. J. A. Campos, A. Otegi, A. Soroa, J. M. Deriu, M. Cieliebak, and E. Agirre, “Doqa-accessing domain-specific faqs via conversational qa,” in Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 7302–7314, 2020. find
  49. H. W. Chung, L. Hou, S. Longpre, B. Zoph, Y. Tay, W. Fedus, E. Li, X. Wang, M. Dehghani, S. Brahma, et al., “Scaling instruction-finetuned language models,” arXiv preprint arXiv:2210.11416, 2022. find
  50. OpenAI, “Gpt-4 technical report,” arXiv preprint arXiv:2303.08774, 2023.
  51. S. Bubeck, V. Chandrasekaran, R. Eldan, J. Gehrke, E. Horvitz, E. Kamar, P. Lee, Y. T. Lee, Y. Li, S. Lundberg, et al., “Sparks of artificial general intelligence: Early experiments with gpt-4,” arXiv preprint arXiv:2303.12712, 2023.
  52. Anthropic, “Anthropic: Claude,” 2023. find
  53. Google, “Google: Bard,” 2023. find
  54. N. Muennighoff, T. Wang, L. Sutawika, A. Roberts, S. Biderman, T. L. Scao, M. S. Bari, S. Shen, Z.-X. Yong, H. Schoelkopf, et al., “Crosslingual generalization through multitask finetuning,” arXiv preprint arXiv:2211.01786, 2022.
  55. H. W. Chung, L. Hou, S. Longpre, B. Zoph, Y. Tay, W. Fedus, E. Li, X. Wang, M. Dehghani, S. Brahma, et al., “Scaling instruction-finetuned language models,” arXiv preprint arXiv:2210.11416, 2022. find
  56. A. Zeng, X. Liu, Z. Du, Z. Wang, H. Lai, M. Ding, Z. Yang, Y. Xu, W. Zheng, X. Xia, W. L. Tam, Z. Ma, Y. Xue, J. Zhai, W. Chen, Z. Liu, P. Zhang, Y. Dong, and J. Tang, “GLM-130b: An open bilingual pre-trained model,” in The Eleventh International Conference on Learning Representations, 2023.
  57. J. Kaplan, S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Radford, J. Wu, and D. Amodei, “Scaling laws for neural language models,” arXiv preprint arXiv:2001.08361, 2020.
  58. LMSYS, “Lmsys org: Chatbot arena,” 2023. find
  59. OpenAI, “Openai: New and improved embedding model,” 2022. find
  60. A. Radford, J. Wu, R. Child, D. Luan, D. Amodei, I. Sutskever, et al., “Language models are unsupervised multitask learners,” find
  61. Y. Wang, Y. Kordi, S. Mishra, A. Liu, N. A. Smith, D. Khashabi, and H. Hajishirzi, “Self-instruct: Aligning language model with self generated instructions,” arXiv preprint arXiv:2212.10560, 2022. find
  62. R. Taori, I. Gulrajani, T. Zhang, Y. Dubois, X. Li, C. Guestrin, P. Liang, and T. B. Hashimoto, “Stanford alpaca: An instruction-following llama model.” https://github.com/tatsu-lab/stanford_alpaca, 2023. ↗
  63. S. Longpre, L. Hou, T. Vu, A. Webson, H. W. Chung, Y. Tay, D. Zhou, Q. V. Le, B. Zoph, J. Wei, et al., “The flan collection: Designing data and methods for effective instruction tuning,” arXiv preprint arXiv:2301.13688, 2023.
  64. Y. Xiao, Y. Jin, Y. Bai, Y. Wu, X. Yang, X. Luo, W. Yu, X. Zhao, Y. Liu, H. Chen, et al., “Large language models can be good privacy protection learners,” arXiv preprint arXiv:2310.02469, 2023. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/bai-2023-benchmarking-foundation-models-language-model/