Elamur Research⌕
☰
home / papers / 2023 / Mind the instructions: a holistic evaluation of consistency and interactions in prompt-based learning

Mind the instructions: a holistic evaluation of consistency and interactions in prompt-based learning

Paper · 2023 · Lucas Weber, Elia Bruni and Dieuwke Hupkes

Summary

In-context learning (ICL) lets large language models perform tasks without parameter updates, but its results are notoriously unstable. It is also unclear which prompt design decisions cause this instability and how those decisions affect one another. The authors address this in two steps. First, they compare task-tuned RoBERTa models with vanilla LLaMA and instruction-tuned Alpaca models on three task types: NLI, paraphrase identification and extractive QA. Each model is adapted on a base dataset that contains known spurious correlations and then tested on adversarial counterparts. Unlike task tuning, ICL turns out to be barely affected by such shortcuts. Second, they run a full grid over combinations of prompt factors and use ANOVA to estimate main effects and interactions. The wording of the instruction template used to query the target matters far more than any other factor, and its effect also varies most across settings. Details of the in-context examples have small but dependable effects. These results show practitioners where prompt engineering effort pays off and where caution is needed.

Key points

  • Models: RoBERTa-base and RoBERTa-large for task tuning; LLaMA (7B, 13B, 30B/33B, 65B) and the matching Alpaca models, which are LLaMA fine-tuned with LoRA on self-instruct data.
  • Base and adversarial datasets: MNLI paired with HANS and ANLI; QQP paired with PAWS; SQuAD paired with SQuAD adv., adv. QA and SQuAD shifts. The analyses are classified using the GenBench taxonomy.
  • Depending on the prompt setup, ICL accuracy differed by up to 40%. Instruction quality was both the most influential factor and the one with the most interactions; calibration also interacted strongly with other factors.
  • Three factors showed almost no interactions, so their effects carry over between setups:
    • adding more in-context examples gives small, reliable gains;
    • balancing the labels gives small, reliable gains;
    • showing only one label lowers accuracy for large models.
  • Larger models and instruction tuning consistently made predictions more stable. The authors recommend using tuned models and evaluating with diverse setups and proper statistical tests.

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

 Institutions

🎓Universitat Pompeu FabraBarcelona · Spain47 works · 11,723 citations🎓Osnabrück UniversityOsnabrück · Germany7 works · 250 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

Cite

Cite as Lucas Weber, Elia Bruni and Dieuwke Hupkes. 2023. Mind the instructions: a holistic evaluation of consistency and interactions in prompt-based learning. Proceedings of the 27th Conference on Computational Natural Language Learning (CoNLL).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2023.conll-1.20

@inproceedings{weber2023mind,
  title = {{Mind the instructions: a holistic evaluation of consistency and interactions in prompt-based learning}},
  author = {Lucas Weber, Elia Bruni and Dieuwke Hupkes},
  booktitle = {Proceedings of the 27th Conference on Computational Natural Language Learning (CoNLL)},
  year = {2023},
  doi = {10.18653/v1/2023.conll-1.20},
  url = {https://doi.org/10.18653/v1/2023.conll-1.20}
}

References

51 references, 27 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Ekin Akyürek, Dale Schuurmans, Jacob Andreas, Tengyu Ma, and Denny Zhou. 2022. What learning algorithm is in-context learning? investigations with linear models. ArXiv preprint, abs/2211.15661.
  2. Anas Awadalla, Mitchell Wortsman, Gabriel Ilharco, Sewon Min, Ian Magnusson, Hannaneh Hajishirzi, and Ludwig Schmidt. 2022. Exploring the landscape of distributional robustness for question answering models. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 5971–5987, A… find
  3. Stephen Bach, Victor Sanh, Zheng Xin Yong, Albert Webson, Colin Raffel, Nihal V. Nayak, Abheesht Sharma, Taewoon Kim, M Saiful Bari, Thibault Fevry, Zaid Alyafeai, Manan Dey, Andrea Santilli, Zhiqing Sun, Srulik Ben-david, Canwen Xu, Gunjan Chhablani, Han Wang, Jason Fries, Maged Al-shaibani, Shanya… find
  4. Yejin Bang, Samuel Cahyawijaya, Nayeon Lee, Wenliang Dai, Dan Su, Bryan Wilie, Holy Lovenia, Ziwei Ji, Tiezheng Yu, Willy Chung, et al. 2023. A multi-task, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity. ArXiv preprint, abs/2302.04023.
  5. Max Bartolo, Alastair Roberts, Johannes Welbl, Sebastian Riedel, and Pontus Stenetorp. 2020. Beat the AI: Investigating adversarial human annotation for reading comprehension. Transactions of the Association for Computational Linguistics, 8:662–678. find
  6. Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter,… — Language Models are Few-Shot Learners
  7. Jacob Cohen. 1960. A coefficient of agreement for nominal scales. Educational and psychological measurement, 20(1):37–46. find
  8. Tim Dettmers, Mike Lewis, Younes Belkada, and Luke Zettlemoyer. 2022. Llm. int8 (): 8-bit matrix multiplication for transformers at scale. ArXiv preprint, abs/2208.07339. find
  9. Shivam Garg, Dimitris Tsipras, Percy S Liang, and Gregory Valiant. 2022. What can transformers learn in-context? a case study of simple function classes. Advances in Neural Information Processing Systems, 35:30583–30598. find
  10. Robert Geirhos, Jörn-Henrik Jacobsen, Claudio Michaelis, Richard Zemel, Wieland Brendel, Matthias Bethge, and Felix A Wichmann. 2020. Shortcut learning in deep neural networks. Nature Machine Intelligence, 2(11):665–673. find
  11. Mor Geva, Yoav Goldberg, and Jonathan Berant. 2019. Are we modeling the task or the annotator? an investigation of annotator bias in natural language understanding datasets. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conf… find
  12. Suchin Gururangan, Swabha Swayamdipta, Omer Levy, Roy Schwartz, Samuel Bowman, and Noah A. Smith. 2018. Annotation artifacts in natural language inference data. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Techno…
  13. Dan Hendrycks, Kimin Lee, and Mantas Mazeika. 2019. Using pre-training can improve model robustness and uncertainty. In Proceedings of the 36th International Conference on Machine Learning, ICML 2019, 9-15 June 2019, Long Beach, California, USA, volume 97 of Proceedings of Machine Learning Research,… find
  14. Dan Hendrycks, Xiaoyuan Liu, Eric Wallace, Adam Dziedzic, Rishabh Krishnan, and Dawn Song. 2020. Pretrained transformers improve out-of-distribution robustness. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2744–2751, Online. Association for Comput…
  15. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. Lora: Low-rank adaptation of large language models. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022. OpenReview.net.
  16. D. Hupkes, M. Giulianelli, V. Dankers, et al. 2023. A taxonomy and review of generalization research in nlp. Nature Machine Intelligence, 5:1161–1174. find
  17. John PA Ioannidis. 2005. Why most published research findings are false. PLoS medicine, 2(8):e124. find
  18. Robin Jia and Percy Liang. 2017. Adversarial examples for evaluating reading comprehension systems. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2021–2031, Copenhagen, Denmark. Association for Computational Linguistics.
  19. Divyansh Kaushik, Eduard H. Hovy, and Zachary Chase Lipton. 2020. Learning the difference that makes A difference with counterfactually-augmented data. In 8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 2020. OpenReview.net. find
  20. Pride Kavumba, Naoya Inoue, Benjamin Heinzerling, Keshav Singh, Paul Reisert, and Kentaro Inui. 2019. When choosing plausible alternatives, clever hans can be clever. In Proceedings of the First Workshop on Commonsense Inference in Natural Language Processing, pages 33–42, Hong Kong, China. Associat… find
  21. Pride Kavumba, Ryo Takahashi, and Yusuke Oda. 2022. Are prompt-based models clueless? In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2333–2352, Dublin, Ireland. Association for Computational Linguistics. find
  22. Daniel Khashabi, Xinxi Lyu, Sewon Min, Lianhui Qin, Kyle Richardson, Sean Welleck, Hannaneh Hajishirzi, Tushar Khot, Ashish Sabharwal, Sameer Singh, et al. 2022. Prompt waywardness: The curious case of discretized interpretation of continuous prompts. In Proceedings of the 2022 Conference of the Nor… find
  23. Percy Liang, Rishi Bommasani, Tony Lee, Dimitris Tsipras, Dilara Soylu, Michihiro Yasunaga, Yian Zhang, Deepak Narayanan, Yuhuai Wu, Ananya Kumar, et al. 2022. Holistic evaluation of language models. ArXiv preprint, abs/2211.09110. find
  24. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining approach. ArXiv preprint, abs/1907.11692.
  25. Yao Lu, Max Bartolo, Alastair Moore, Sebastian Riedel, and Pontus Stenetorp. 2022. Fantastically ordered prompts and where to find them: Overcoming few-shot prompt order sensitivity. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), p…
  26. Tom McCoy, Ellie Pavlick, and Tal Linzen. 2019. Right for the wrong reasons: Diagnosing syntactic heuristics in natural language inference. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3428–3448, Florence, Italy. Association for Computational Ling…
  27. John Miller, Karl Krauth, Benjamin Recht, and Ludwig Schmidt. 2020. The effect of natural distribution shift on question answering models. In Proceedings of the 37th International Conference on Machine Learning, ICML 2020, 13-18 July 2020, Virtual Event, volume 119 of Proceedings of Machine Learning… find
  28. Sewon Min, Xinxi Lyu, Ari Holtzman, Mikel Artetxe, Mike Lewis, Hannaneh Hajishirzi, and Luke Zettlemoyer. 2022. Rethinking the role of demonstrations: What makes in-context learning work? In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 11048–11064, Ab…
  29. Swaroop Mishra, Daniel Khashabi, Chitta Baral, Yejin Choi, and Hannaneh Hajishirzi. 2022. Reframing instructional prompts to GPTk’s language. In Findings of the Association for Computational Linguistics: ACL 2022, pages 589–612, Dublin, Ireland. Association for Computational Linguistics.
  30. Marius Mosbach, Tiago Pimentel, Shauli Ravfogel, Dietrich Klakow, and Yanai Elazar. 2023. Few-shot fine-tuning vs. in-context learning: A fair comparison and evaluation. ArXiv preprint, abs/2305.16938.
  31. Yixin Nie, Adina Williams, Emily Dinan, Mohit Bansal, Jason Weston, and Douwe Kiela. 2020. 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 Computa…
  32. Timothy Niven and Hung-Yu Kao. 2019. Probing neural network comprehension of natural language arguments. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4658–4664, Florence, Italy. Association for Computational Linguistics. find
  33. Xenia Ohmer, Elia Bruni, and Dieuwke Hupkes. 2023. Evaluating task understanding through multilingual consistency: A chatgpt case study. ArXiv preprint, abs/2305.11662. — [arXiv:2305.11662](https://arxiv.org/abs/2305.11662) find
  34. Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35:27730–27744.
  35. Adam Poliak, Jason Naradowsky, Aparajita Haldar, Rachel Rudinger, and Benjamin Van Durme. 2018. Hypothesis only baselines in natural language inference. In Proceedings of the Seventh Joint Conference on Lexical and Computational Semantics, pages 180–191, New Orleans, Louisiana. Association for Compu…
  36. Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. 2016. SQuAD: 100,000+ questions for machine comprehension of text. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2383–2392, Austin, Texas. Association for Computational Linguistics.
  37. Bernhard Schölkopf, Dominik Janzing, Jonas Peters, Eleni Sgouritsa, Kun Zhang, and Joris M. Mooij. 2012. On causal and anticausal learning. In Proceedings of the 29th International Conference on Machine Learning, ICML 2012, Edinburgh, Scotland, UK, June 26 - July 1, 2012. icml.cc / Omnipress. find
  38. Chenglei Si, Zhe Gan, Zhengyuan Yang, Shuohang Wang, Jianfeng Wang, Jordan Boyd-Graber, and Lijuan Wang. 2023. Prompting gpt-3 to be reliable.
  39. Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B. Hashimoto. 2023. Stanford alpaca: An instruction-following llama model. https://github.com/tatsu-lab/stanford_alpaca. ↗
  40. Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. ArXiv preprint, abs/2302.13971.
  41. Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A Smith, Daniel Khashabi, and Hannaneh Hajishirzi. 2022. Self-instruct: Aligning language model with self generated instructions. ArXiv preprint, abs/2212.10560. find
  42. Zhiguo Wang, Wael Hamza, and Radu Florian. 2017. Bilateral multi-perspective matching for natural language sentences. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI 2017, Melbourne, Australia, August 19-25, 2017, pages 4144–4150. ijcai.org.
  43. Jason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai, and Quoc V. Le. 2022. Finetuned language models are zero-shot learners. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022. OpenReview…
  44. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, et al. 2019. Huggingface’s transformers: State-ofthe-art natural language processing. ArXiv preprint, abs/1910.03771.
  45. Kang Min Yoo, Junyeob Kim, Hyuhng Joon Kim, Hyunsoo Cho, Hwiyeol Jo, Sang-Woo Lee, Sang-goo Lee, and Taeuk Kim. 2022. Ground-truth labels matter: A deeper look into input-label demonstrations. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 2422– 2437…
  46. Yuan Zhang, Jason Baldridge, and Luheng He. 2019. PAWS: Paraphrase adversaries from word scrambling. 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 1298–1308, M…
  47. Zihao Zhao, Eric Wallace, Shi Feng, Dan Klein, and Sameer Singh. 2021. Calibrate before use: Improving few-shot performance of language models. In Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtual Event, volume 139 of Proceedings of Machine Lea…
  48. Ruiqi Zhong, Kristy Lee, Zheng Zhang, and Dan Klein. 2021. Adapting language models for zero-shot learning by meta-tuning on dataset and prompt collections. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 2856–2878, Punta Cana, Dominican Republic. Association for Comp… find
  49. Chunting Zhou, Pengfei Liu, Puxin Xu, Srini Iyer, Jiao Sun, Yuning Mao, Xuezhe Ma, Avia Efrat, Ping Yu, Lili Yu, Susan Zhang, Gargi Ghosh, Mike Lewis, Luke Zettlemoyer, and Omer Levy. 2023. Lima: Less is more for alignment. find
  50. Jerry Wei, Jason Wei, Yi Tay, Dustin Tran, Albert Webson, Yifeng Lu, Xinyun Chen, Hanxiao Liu, Da Huang, Denny Zhou, et al. 2023. Larger language models do in-context learning differently. ArXiv preprint, abs/2303.03846.
  51. Adina Williams, Nikita Nangia, and Samuel Bowman. 2018. A broad-coverage challenge corpus for sentence understanding through inference. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Pa…

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/weber-2023-mind-instructions-holistic-evaluation-consistency/