Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
41 tokens/sec
GPT-4o
59 tokens/sec
Gemini 2.5 Pro Pro
41 tokens/sec
o3 Pro
7 tokens/sec
GPT-4.1 Pro
50 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

The Critique of Critique (2401.04518v2)

Published 9 Jan 2024 in cs.CL and cs.AI

Abstract: Critique, as a natural language description for assessing the quality of model-generated content, has played a vital role in the training, evaluation, and refinement of LLMs. However, a systematic method to evaluate the quality of critique is lacking. In this paper, we pioneer the critique of critique, termed MetaCritique, which builds specific quantification criteria. To achieve a reliable evaluation outcome, we propose Atomic Information Units (AIUs), which describe the critique in a more fine-grained manner. MetaCritique aggregates each AIU's judgment for the overall score. Moreover, MetaCritique delivers a natural language rationale for the intricate reasoning within each judgment. Lastly, we construct a meta-evaluation dataset covering 4 tasks across 16 public datasets involving human-written and LLM-generated critiques. Experiments demonstrate that MetaCritique can achieve near-human performance. Our study can facilitate future research in LLM critiques based on our following observations and released resources: (1) superior critiques judged by MetaCritique can lead to better refinements, indicating that it can potentially enhance the alignment of existing LLMs; (2) the leaderboard of critique models reveals that open-source critique models commonly suffer from factuality issues; (3) relevant code and data are publicly available at https://github.com/GAIR-NLP/MetaCritique to support deeper exploration; (4) an API at PyPI with the usage documentation in Appendix C allows users to assess the critique conveniently.

PDF HTML Abstract
Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize File Document Download Save Streamline Icon: https://streamlinehq.com PDF File Document Download Save Streamline Icon: https://streamlinehq.com Markdown Bookmark Chat Bubble Oval Streamline Icon: https://streamlinehq.com Chat (Pro)
Definition Search Book Streamline Icon: https://streamlinehq.com
References (72)
  1. Explanations for CommonsenseQA: New Dataset and Models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 3050–3065, Online. Association for Computational Linguistics.
  2. RL4F: Generating natural language feedback with reinforcement learning for repairing model outputs. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7716–7733, Toronto, Canada. Association for Computational Linguistics.
  3. MultiFC: A real-world multi-domain dataset for evidence-based fact checking of claims. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 4685–4697, Hong Kong, China. Association for Computational Linguistics.
  4. Constitutional ai: Harmlessness from ai feedback. arXiv preprint arXiv:2212.08073.
  5. Piqa: Reasoning about physical commonsense in natural language. In Proceedings of the AAAI conference on artificial intelligence, pages 7432–7439.
  6. Re-evaluating the role of Bleu in machine translation research. In 11th Conference of the European Chapter of the Association for Computational Linguistics, pages 249–256, Trento, Italy. Association for Computational Linguistics.
  7. e-snli: Natural language inference with natural language explanations. Advances in Neural Information Processing Systems, 31.
  8. Chateval: Towards better llm-based evaluators through multi-agent debate. arXiv preprint arXiv:2308.07201.
  9. Factool: Factuality detection in generative ai–a tool augmented framework for multi-task and multi-domain scenarios. arXiv preprint arXiv:2307.13528.
  10. Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality.
  11. BoolQ: Exploring the surprising difficulty of natural yes/no questions. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 2924–2936, Minneapolis, Minnesota. Association for Computational Linguistics.
  12. Think you have solved question answering? try arc, the ai2 reasoning challenge. arXiv preprint arXiv:1803.05457.
  13. Ultrafeedback: Boosting language models with high-quality feedback. arXiv preprint arXiv:2310.01377.
  14. Explaining answers with entailment trees. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 7358–7370, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
  15. Gptscore: Evaluate as you desire. arXiv preprint arXiv:2302.04166.
  16. Rarr: Researching and revising what language models say, using language models. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 16477–16508.
  17. Critic: Large language models can self-correct with tool-interactive critiquing. arXiv preprint arXiv:2305.11738.
  18. News summarization and evaluation in the era of gpt-3. arXiv preprint arXiv:2209.12356.
  19. A survey on automated fact-checking. Transactions of the Association for Computational Linguistics, 10:178–206.
  20. Micah Hodosh and Julia Hockenmaier. 2016. Focused evaluation for image description with binary forced-choice tasks. In Proceedings of the 5th Workshop on Vision and Language, pages 19–28, Berlin, Germany. Association for Computational Linguistics.
  21. Cosmos QA: Machine reading comprehension with contextual commonsense reasoning. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 2391–2401, Hong Kong, China. Association for Computational Linguistics.
  22. TriviaQA: A large scale distantly supervised challenge dataset for reading comprehension. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1601–1611, Vancouver, Canada. Association for Computational Linguistics.
  23. Wice: Real-world entailment for claims in wikipedia. arXiv preprint arXiv:2303.01432.
  24. Critiquellm: Scaling llm-as-critic for effective and explainable evaluation of large language model generation. arXiv preprint arXiv:2311.18702.
  25. Maurice G Kendall. 1938. A new measure of rank correlation. Biometrika, 30(1/2):81–93.
  26. Re-evaluating automatic metrics for image captioning. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pages 199–209, Valencia, Spain. Association for Computational Linguistics.
  27. Prometheus: Inducing fine-grained evaluation capability in language models. arXiv preprint arXiv:2310.08491.
  28. Philipp Koehn. 2004. Statistical significance tests for machine translation evaluation. In Proceedings of the 2004 Conference on Empirical Methods in Natural Language Processing, pages 388–395, Barcelona, Spain. Association for Computational Linguistics.
  29. ProoFVer: Natural logic theorem proving for fact verification. Transactions of the Association for Computational Linguistics, 10:1013–1030.
  30. Evaluating the factual consistency of abstractive text summarization. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 9332–9346, Online. Association for Computational Linguistics.
  31. Natural questions: A benchmark for question answering research. Transactions of the Association for Computational Linguistics, 7:452–466.
  32. RACE: Large-scale ReAding comprehension dataset from examinations. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 785–794, Copenhagen, Denmark. Association for Computational Linguistics.
  33. Joseph Lee Rodgers and W Alan Nicewander. 1988. Thirteen ways to look at the correlation coefficient. The American Statistician, 42(1):59–66.
  34. Adversarial learning for neural dialogue generation. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2157–2169, Copenhagen, Denmark. Association for Computational Linguistics.
  35. Generative judge for evaluating alignment. arXiv preprint arXiv:2310.05470.
  36. Chin-Yew Lin. 2004. ROUGE: A package for automatic evaluation of summaries. In Text Summarization Branches Out, pages 74–81, Barcelona, Spain. Association for Computational Linguistics.
  37. On improving summarization factual consistency from natural language feedback. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 15144–15161, Toronto, Canada. Association for Computational Linguistics.
  38. On learning to summarize with large language models as references. arXiv preprint arXiv:2305.14239.
  39. Self-refine: Iterative refinement with self-feedback. arXiv preprint arXiv:2303.17651.
  40. Mavuto M Mukaka. 2012. A guide to appropriate use of correlation coefficient in medical research. Malawi medical journal, 24(3):69–71.
  41. Adversarial NLI: A new benchmark for natural language understanding. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 4885–4901, Online. Association for Computational Linguistics.
  42. OpenAI. 2023. Gpt-4 technical report. ArXiv, abs/2303.08774.
  43. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pages 311–318, Philadelphia, Pennsylvania, USA. Association for Computational Linguistics.
  44. Instruction tuning with gpt-4. arXiv preprint arXiv:2304.03277.
  45. Marc Uri Porat. 1977. The Information Economy: Definition and Measurement. ERIC.
  46. Re-evaluating adem: A deeper look at scoring dialogue responses. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01):6220–6227.
  47. A survey of evaluation metrics used for nlg systems. ACM Computing Surveys (CSUR), 55(2):1–39.
  48. Winogrande: An adversarial winograd schema challenge at scale. Communications of the ACM, 64(9):99–106.
  49. Social IQa: Commonsense reasoning about social interactions. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 4463–4473, Hong Kong, China. Association for Computational Linguistics.
  50. Self-critiquing models for assisting human evaluators. arXiv preprint arXiv:2206.05802.
  51. Training language models with language feedback at scale. arXiv preprint arXiv:2303.16755.
  52. Ehsan S Soofi. 1994. Capturing the intangible concept of information. Journal of the American Statistical Association, 89(428):1243–1254.
  53. FEVER: a large-scale dataset for fact extraction and VERification. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), pages 809–819, New Orleans, Louisiana. Association for Computational Linguistics.
  54. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288.
  55. Fact or fiction: Verifying scientific claims. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 7534–7550, Online. Association for Computational Linguistics.
  56. Asking and answering questions to evaluate the factual consistency of summaries. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5008–5020, Online. Association for Computational Linguistics.
  57. Shepherd: A critic for language model generation. arXiv preprint arXiv:2308.04592.
  58. William Yang Wang. 2017. “liar, liar pants on fire”: A new benchmark dataset for fake news detection. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 422–426, Vancouver, Canada. Association for Computational Linguistics.
  59. Pandalm: An automatic evaluation benchmark for llm instruction tuning optimization. arXiv preprint arXiv:2306.05087.
  60. Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35:24824–24837.
  61. Generating sequences by learning to self-correct. In The Eleventh International Conference on Learning Representations.
  62. Evan J. Williams. 1959. Regression analysis. Wiley, New York, 14.
  63. Fine-grained human feedback gives better rewards for language model training. arXiv preprint arXiv:2306.01693.
  64. Wizardlm: Empowering large language models to follow complex instructions. arXiv preprint arXiv:2304.12244.
  65. Selfee: Iterative self-revising llm empowered by self-feedback generation. Blog post.
  66. Bartscore: Evaluating generated text as text generation. Advances in Neural Information Processing Systems, 34:27263–27277.
  67. Jerrold H Zar. 2005. Spearman rank correlation. Encyclopedia of biostatistics, 7.
  68. HellaSwag: Can a machine really finish your sentence? In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4791–4800, Florence, Italy. Association for Computational Linguistics.
  69. Bertscore: Evaluating text generation with bert. In International Conference on Learning Representations.
  70. Judging llm-as-a-judge with mt-bench and chatbot arena. arXiv preprint arXiv:2306.05685.
  71. Reasoning over semantic-level graph for fact checking. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6170–6180, Online. Association for Computational Linguistics.
  72. Least-to-most prompting enables complex reasoning in large language models. In The Eleventh International Conference on Learning Representations.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (6)
  1. Shichao Sun (15 papers)
  2. Junlong Li (22 papers)
  3. Weizhe Yuan (25 papers)
  4. Ruifeng Yuan (19 papers)
  5. Wenjie Li (183 papers)
  6. Pengfei Liu (191 papers)
Github Logo Streamline Icon: https://streamlinehq.com

GitHub

  1. GitHub - GAIR-NLP/MetaCritique: Evaluate the Quality of Critique (35 stars)
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets