Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
97 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
5 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Explanation Regeneration via Information Bottleneck (2212.09603v2)

Published 19 Dec 2022 in cs.CL

Abstract: Explaining the black-box predictions of NLP models naturally and accurately is an important open problem in natural language generation. These free-text explanations are expected to contain sufficient and carefully-selected evidence to form supportive arguments for predictions. Due to the superior generative capacity of large pretrained LLMs, recent work built on prompt engineering enables explanation generation without specific training. However, explanation generated through single-pass prompting often lacks sufficiency and conciseness. To address this problem, we develop an information bottleneck method EIB to produce refined explanations that are sufficient and concise. Our approach regenerates the free-text explanation by polishing the single-pass output from the pretrained LLM but retaining the information that supports the contents being explained. Experiments on two out-of-domain tasks verify the effectiveness of EIB through automatic evaluation and thoroughly-conducted human evaluation.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (55)
  1. Explanations for commonsenseqa: New dataset and models. In Association for Computational Linguistics and International Joint Conference on Natural Language Processing (ACL-IJCNLP), pages 3050–3065.
  2. Deep variational information bottleneck. In International Conference on Learning Representations (ICLR).
  3. Where is your evidence: Improving fact-checking by justification modeling. In First Workshop on Fact Extraction and VERification (FEVER), pages 85–90.
  4. Learning to rationalize for nonmonotonic reasoning with distant supervision. In Association for the Advancement of Artificial Intelligence (AAAI), pages 12592–12601.
  5. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901.
  6. e-snli: Natural language inference with natural language explanations. Advances in Neural Information Processing Systems (NeurIPS), 31.
  7. Make up your mind! adversarial generation of inconsistent natural language explanations. In Association for Computational Linguistics (ACL), pages 4157–4165.
  8. Palm: Scaling language modeling with pathways. arXiv preprint arXiv:2204.02311.
  9. GLM: general language model pretraining with autoregressive blank infilling. In Association for Computational Linguistics (ACL), pages 320–335.
  10. Understanding dataset difficulty with V-usable information. In International Conference on Machine Learning (ICML), volume 162, pages 5988–6008.
  11. Learning to scaffold: Optimizing model explanations for teaching. Advances in Neural Information Processing Systems (NeurIPS), 35:36108–36122.
  12. News summarization and evaluation in the era of gpt-3. arXiv preprint arXiv:2209.12356.
  13. Jian Guan and Minlie Huang. 2020. Union: An unreferenced metric for evaluating open-ended story generation. In Empirical Methods in Natural Language Processing (EMNLP), pages 9157–9166.
  14. Summarize-then-answer: Generating concise explanations for multi-hop reading comprehension. In Empirical Methods in Natural Language Processing (EMNLP), pages 6064–6080.
  15. Towards unsupervised dense information retrieval with contrastive learning. arXiv preprint arXiv:2112.09118.
  16. What’s in an explanation? characterizing knowledge and inference requirements for elementary science exams. In International Conference on Computational Linguistics (COLING), pages 2956–2965.
  17. Leveraging information bottleneck for scientific document summarization. In Findings of the Association for Computational Linguistics (Findings of EMNLP), pages 4091–4098.
  18. Maieutic prompting: Logically consistent reasoning with recursive explanations. arXiv preprint arXiv:2205.11822.
  19. Nora Kassner and Hinrich Schütze. 2020. Negated and misprimed probes for pretrained language models: Birds can talk, but cannot fly. In Association for Computational Linguistics (ACL), pages 7811–7818.
  20. Are prompt-based models clueless? In Association for Computational Linguistics (ACL), pages 2333–2352.
  21. Diederik P. Kingma and Jimmy Ba. 2015. Adam: A method for stochastic optimization. In International Conference on Learning Representations (ICLR).
  22. Neema Kotonya and Francesca Toni. 2020. Explainable automated fact-checking for public health claims. In Empirical Methods in Natural Language Processing (EMNLP), pages 7740–7754.
  23. Tell me why! explanations support learning relational and causal structure. In International Conference on Machine Learning (ICML), volume 162, pages 11868–11890.
  24. Veronica Latcinnik and Jonathan Berant. 2020. Explaining question answering models through text generation. arXiv preprint arXiv:2004.05569.
  25. A diversity-promoting objective function for neural conversation models. In North American Chapter of the Association for Computational Linguistics (NAACL), pages 110–119.
  26. Event transition planning for open-ended text generation. In Findings of the Association for Computational Linguistics (Findings of ACL), pages 3412–3426.
  27. Xiang Lisa Li and Percy Liang. 2021. Prefix-tuning: Optimizing continuous prompts for generation. In Association for Computational Linguistics and International Joint Conference on Natural Language Processing (ACL-IJCNLP), pages 4582–4597.
  28. Holistic evaluation of language models. CoRR, abs/2211.09110.
  29. Chin-Yew Lin and Eduard Hovy. 2002. Manual and automatic evaluation of summaries. In Proceedings of the ACL-02 Workshop on Automatic Summarization, pages 45–51.
  30. Teaching language models to support answers with verified quotes. arXiv preprint arXiv:2203.11147.
  31. Tim Miller. 2019. Explanation in artificial intelligence: Insights from the social sciences. Artificial intelligence, 267:1–38.
  32. Representation learning with contrastive predictive coding. arXiv preprint arXiv:1807.03748.
  33. Bleu: a method for automatic evaluation of machine translation. In Association for Computational Linguistics (ACL), pages 311–318.
  34. An information bottleneck approach for controlling conciseness in rationale extraction. In Empirical Methods in Natural Language Processing (EMNLP), pages 1938–1952.
  35. Language models are unsupervised multitask learners. OpenAI blog, 1(8):9.
  36. Explain yourself! leveraging language models for commonsense reasoning. In Association for Computational Linguistics (ACL), pages 4932–4942.
  37. Referee: Reference-free sentence summarization with sharper controllability through symbolic knowledge distillation. arXiv preprint arXiv:2210.13800.
  38. Chenhao Tan. 2021. On the diversity and limits of human explanations. arXiv preprint arXiv:2106.11988.
  39. The information bottleneck method. arXiv preprint physics/0004057.
  40. Naftali Tishby and Noga Zaslavsky. 2015. Deep learning and the information bottleneck principle. In 2015 IEEE Information Theory Workshop (ITW).
  41. Cider: Consensus-based image description evaluation. In Computer Vision and Pattern Recognition (CVPR), pages 4566–4575.
  42. Does it make sense? and why? a pilot study for sense making and explanation. In Association for Computational Linguistics (ACL), pages 4020–4026.
  43. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems (NeurIPS).
  44. Generating sequences by learning to self-correct. arXiv preprint arXiv:2211.00053.
  45. Bottlesum: Unsupervised and self-supervised sentence summarization using the information bottleneck principle. In Empirical Methods in Natural Language Processing and International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 3750–3759.
  46. Reframing human-ai collaboration for generating free-text explanations. In North American Chapter of the Association for Computational Linguistics (NAACL), pages 632–658.
  47. Sarah Wiegreffe and Ana Marasovic. 2021. Teach me to explain: A review of datasets for explainable natural language processing. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 1).
  48. Huggingface’s transformers: State-of-the-art natural language processing. arXiv preprint arXiv:1910.03771.
  49. A theory of usable information under computational constraints. In International Conference on Learning Representations (ICLR).
  50. The unreliability of explanations in few-shot prompting for textual reasoning. In Advances in Neural Information Processing Systems ((NeurIPS)).
  51. Rethinking cooperative rationalization: Introspective extraction and complement control. In Empirical Methods in Natural Language Processing and International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 4094–4103.
  52. Star: Bootstrapping reasoning with reasoning. Advances in Neural Information Processing Systems (NeurIPS), 35:15476–15488.
  53. Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068.
  54. Bertscore: Evaluating text generation with BERT. In International Conference on Learning Representations (ICLR).
  55. Probing commonsense explanation in dialogue response generation. In Findings of Empirical Methods in Natural Language Processing (Findings of EMNLP), pages 4132–4146.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (4)
  1. Qintong Li (17 papers)
  2. Zhiyong Wu (171 papers)
  3. Lingpeng Kong (134 papers)
  4. Wei Bi (62 papers)
Citations (3)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now