Elamur Research
☰
home / papers / 2022 / Large Language Models Are Human-Level Prompt Engineers

Large Language Models Are Human-Level Prompt Engineers

Written for the Research library (claude-opus-5-5) from the paper text, in our own words: the license "arXiv — non-exclusive licence to distribute" does not allow redistributing the text itself. Original — arxiv.org/pdf/2211.01910

Summary

How well a large language model does a task depends heavily on the wording of its instruction, and good prompts are usually written by hand through trial and error. The authors treat instruction writing as natural language program synthesis. In this framing, the instruction is a "program" for a black-box model, chosen to maximise a score on a small set of input–output demonstrations. Their method, Automatic Prompt Engineer (APE), uses an LLM three ways. First, it infers candidate instructions from the demonstrations. Second, the target model scores each candidate, either by execution accuracy or by the log-likelihood of the correct outputs. Third, an optional iterative Monte Carlo step asks the LLM to rewrite the best candidates as close variants. The selected instruction is then tested zero-shot on another model. Across GPT-3 and InstructGPT models of several sizes, APE instructions clearly beat a greedy LLM baseline and match or exceed human-written instructions on most benchmark tasks. This makes prompt optimisation possible through an API alone, without gradients or hand-made templates.

Key points

  • On zero-shot evaluation, APE matches or beats human-written instructions on 24/24 Instruction Induction tasks (Honovich et al., 2022) and on 17/21 curated BIG-Bench tasks.
  • Measured by the interquartile mean over the 24 tasks, APE with InstructGPT (175B) scores higher than the human prompt engineer baseline. Greedy generation with no search scores much lower.
  • Eight models were tested: GPT-3 and InstructGPT at 350M, 1.3B, 6.7B and 175B parameters. A single LLM handles candidate generation, scoring and resampling, with no task-specific training.
  • APE has other uses too. Adding its instructions in front of few-shot prompts improves in-context learning. It also finds stronger zero-shot chain-of-thought trigger phrases and can push models toward truthful or informative answers on TruthfulQA.
  • The code is released at github.com/keirp/automatic_prompt_engineer.

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: Other

Cited by

395 papers of our library cite it (of the 140673 whose bibliographies we read); 17 of them have a page here, 378 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2211.01910 · arXiv 2211.01910

@misc{zhou2022large,
  title = {{Large Language Models Are Human-Level Prompt Engineers}},
  author = {Zhou, Yongchao and Muresanu, Andrei Ioan and Han, Ziwen and Paster, Keiran and Pitis, Silviu and Chan, Harris and Ba, Jimmy},
  year = {2022},
  doi = {10.48550/arxiv.2211.01910},
  eprint = {2211.01910},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2211.01910}
}

References

58 references, 30 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Rishabh Agarwal, Max Schwarzer, Pablo Samuel Castro, Aaron Courville, and Marc G Bellemare. Deep reinforcement learning at the edge of the statistical precipice. Advances in Neural Information Processing Systems, 2021. find
  2. Michael Ahn, Anthony Brohan, Noah Brown, Yevgen Chebotar, Omar Cortes, Byron David, Chelsea Finn, Keerthana Gopalakrishnan, Karol Hausman, Alex Herzog, et al. Do as i can, not as i say: Grounding language in robotic affordances. arXiv preprint arXiv:2204.01691, 2022. find our card
  3. Amanda Askell, Yuntao Bai, Anna Chen, Dawn Drain, Deep Ganguli, Tom Henighan, Andy Jones, Nicholas Joseph, Ben Mann, Nova DasSarma, et al. A general language assistant as a laboratory for alignment. arXiv preprint arXiv:2112.00861, 2021. find our card
  4. Jacob Austin, Augustus Odena, Maxwell Nye, Maarten Bosma, Henryk Michalewski, David Dohan, Ellen Jiang, Carrie Cai, Michael Terry, Quoc Le, et al. Program synthesis with large language models. arXiv preprint arXiv:2108.07732, 2021. find our card
  5. Mohammad Bavarian, Heewoo Jun, Nikolas Tezak, John Schulman, Christine McLeavey, Jerry Tworek, and Mark Chen. Efficient training of language models to fill in the middle. arXiv preprint arXiv:2207.14255, 2022. find our card
  6. Eyal Ben-David, Nadav Oved, and Roi Reichart. Pada: A prompt-based autoregressive approach for adaptation to unseen domains. arXiv preprint arXiv:2102.12206, 2021. — [arXiv:2102.12206](https://arxiv.org/abs/2102.12206) find
  7. Gregor Betz, Kyle Richardson, and Christian Voigt. Thinking aloud: Dynamic context generation improves zero-shot reasoning performance of gpt-2. arXiv preprint arXiv:2103.13033, 2021. — [arXiv:2103.13033](https://arxiv.org/abs/2103.13033) find
  8. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901, 2020. find
  9. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, et al. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374, 2021. find our card
  10. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168, 2021. find our card
  11. Joe Davison, Joshua Feldman, and Alexander M Rush. Commonsense knowledge mining from pretrained models. 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), pp. 1173–1178, 2… find
  12. Jacob Devlin, Rudy R Bunel, Rishabh Singh, Matthew Hausknecht, and Pushmeet Kohli. Neural program meta-induction. Advances in Neural Information Processing Systems, 30, 2017. find
  13. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805, 2018. find
  14. Zhengxiao Du, Yujie Qian, Xiao Liu, Ming Ding, Jiezhong Qiu, Zhilin Yang, and Jie Tang. GLM: General language model pretraining with autoregressive blank infilling. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 320–335, Dublin… find our card
  15. Kevin Ellis, Lucas Morales, Mathias Sablé-Meyer, Armando Solar-Lezama, and Josh Tenenbaum. Learning libraries of subroutines for neurally–guided bayesian program induction. In S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett (eds.), Advances in Neural Information Pro… find
  16. Kevin Ellis, Catherine Wong, Maxwell Nye, Mathias Sablé-Meyer, Lucas Morales, Luke Hewitt, Luc Cary, Armando Solar-Lezama, and Joshua B Tenenbaum. Dreamcoder: Bootstrapping inductive program synthesis with wake-sleep library learning. In Proceedings of the 42nd acm sigplan international conference o… find
  17. Tianyu Gao. Prompting: Better ways of using language models for nlp tasks. The Gradient, 2021. find
  18. Tianyu Gao, Adam Fisch, and Danqi Chen. Making pre-trained language models better few-shot learners. 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), pp. 3816… find our card
  19. Sumit Gulwani, Oleksandr Polozov, Rishabh Singh, et al. Program synthesis. Foundations and Trends® in Programming Languages, 4(1-2):1–119, 2017. find
  20. Or Honovich, Uri Shaham, Samuel R Bowman, and Omer Levy. Instruction induction: From few examples to natural language task descriptions. arXiv preprint arXiv:2205.10782, 2022. find
  21. Naman Jain, Skanda Vaidyanath, Arun Iyer, Nagarajan Natarajan, Suresh Parthasarathy, Sriram Rajamani, and Rahul Sharma. Jigsaw: Large language models meet program synthesis. In Proceedings of the 44th International Conference on Software Engineering, pp. 1219–1231, 2022. find
  22. Zhengbao Jiang, Frank F Xu, Jun Araki, and Graham Neubig. How can we know what language models know? Transactions of the Association for Computational Linguistics, 8:423–438, 2020. find our card
  23. Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361, 2020. find
  24. Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwasawa. Large language models are zero-shot reasoners. arXiv preprint arXiv:2205.11916, 2022. find our card
  25. Woosuk Lee, Kihong Heo, Rajeev Alur, and Mayur Naik. Accelerating search-based program synthesis using learned probabilistic models. ACM SIGPLAN Notices, 53(4):436–449, 2018. find
  26. Brian Lester, Rami Al-Rfou, and Noah Constant. The power of scale for parameter-efficient prompt tuning. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pp. 3045–3059, 2021. find
  27. Yujia Li, David Choi, Junyoung Chung, Nate Kushman, Julian Schrittwieser, Rémi Leblond, Tom Eccles, James Keeling, Felix Gimeno, Agustin Dal Lago, et al. Competition-level code generation with alphacode. arXiv preprint arXiv:2203.07814, 2022. find our card
  28. Percy Liang, Michael I. Jordan, and Dan Klein. Learning programs: A hierarchical bayesian approach. In Johannes Fürnkranz and Thorsten Joachims (eds.), Proceedings of the 27th International Conference on Machine Learning (ICML-10), June 21-24, 2010, Haifa, Israel, pp. 639–646. find
  29. Omnipress, 2010. URL https://icml.cc/Conferences/2010/papers/568.pdf. ↗
  30. Stephanie Lin, Jacob Hilton, and Owain Evans. TruthfulQA: Measuring how models mimic human falsehoods. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 3214–3252, Dublin, Ireland, May 2022. Association for Computational Linguisti… find our card
  31. Xiao Liu, Yanan Zheng, Zhengxiao Du, Ming Ding, Yujie Qian, Zhilin Yang, and Jie Tang. Gpt understands, too. arXiv preprint arXiv:2103.10385, 2021. find our card
  32. Yao Lu, Max Bartolo, Alastair Moore, Sebastian Riedel, and Pontus Stenetorp. Fantastically ordered prompts and where to find them: Overcoming few-shot prompt order sensitivity. arXiv preprint arXiv:2104.08786, 2021. find our card
  33. Dougal Maclaurin and Ryan Prescott Adams. Firefly monte carlo: Exact mcmc with subsets of data. In Twenty-Fourth International Joint Conference on Artificial Intelligence, 2015. find
  34. Aditya Menon, Omer Tamuz, Sumit Gulwani, Butler Lampson, and Adam Kalai. A machine learning framework for programming by example. In International Conference on Machine Learning, pp. 187–195. PMLR, 2013. find
  35. Maxwell Nye, Anders Johan Andreassen, Guy Gur-Ari, Henryk Michalewski, Jacob Austin, David Bieber, David Dohan, Aitor Lewkowycz, Maarten Bosma, David Luan, et al. Show your work: Scratchpads for intermediate computation with language models. arXiv preprint arXiv:2112.00114, 2021. find our card
  36. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. arXiv preprint arXiv:2203.02155, 2022. find
  37. Ethan Perez, Douwe Kiela, and Kyunghyun Cho. True few-shot learning with language models. Advances in Neural Information Processing Systems, 34:11054–11070, 2021. find our card
  38. Guanghui Qin and Jason Eisner. Learning how to ask: Querying lms with mixtures of soft prompts. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 5203–5212, 2021. find our card
  39. Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, Peter J Liu, et al. Exploring the limits of transfer learning with a unified text-to-text transformer. J. Mach. Learn. Res., 21(140):1–67, 2020. find
  40. Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125, 2022. find our card
  41. Laria Reynolds and Kyle McDonell. Prompt programming for large language models: Beyond the few-shot paradigm. In Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems, pp. 1–7, 2021. find our card
  42. Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10684–10695, 2022. find
  43. Subhro Roy and Dan Roth. arXiv:1608.01413, 2016. find our card
  44. Solving general arithmetic word problems. Victor Sanh, Albert Webson, Colin Raffel, Stephen Bach, Lintang Sutawika, Zaid Alyafeai, Antoine Chaffin, Arnaud Stiegler, Teven Le Scao, Arun Raja, et al. Multitask prompted training enables zero-shot task generalization. In The Tenth International Conferen… find our card
  45. Timo Schick and Hinrich Schütze. Exploiting cloze-questions for few-shot text classification and natural language inference. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pp. 255–269, 2021. arXiv preprint Taylor Shin, Yas… find our card
  46. Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao, Abu Awal Md Shoeb, Abubakar Abid, Adam Fisch, Adam R Brown, Adam Santoro, Aditya Gupta, Adrià Garriga-Alonso, et al. Beyond the imitation game: Quantifying and extrapolating the capabilities of language models. arXiv preprint arXiv:2206.04615, 2022. find our card
  47. Mirac Suzgun, Nathan Scales, Nathanael Schärli, Sebastian Gehrmann, Yi Tay, Hyung Won Chung, Aakanksha Chowdhery, Quoc V Le, Ed H Chi, Denny Zhou, et al. Challenging big-bench tasks and whether chain-of-thought can solve them. arXiv preprint arXiv:2210.09261, 2022. find our card
  48. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017. find
  49. Albert Webson and Ellie Pavlick. Do prompt-based models really understand the meaning of their prompts? arXiv preprint arXiv:2109.01247, 2021. find our card
  50. Jason Wei, Maarten Bosma, Vincent Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M Dai, and Quoc V Le. Finetuned language models are zero-shot learners. In International Conference on Learning Representations, 2021. find
  51. Jason Wei, Yi Tay, Rishi Bommasani, Colin Raffel, Barret Zoph, Sebastian Borgeaud, Dani Yogatama, Maarten Bosma, Denny Zhou, Donald Metzler, et al. Emergent abilities of large language models. arXiv preprint arXiv:2206.07682, 2022a. find our card
  52. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou. Chain of thought prompting elicits reasoning in large language models. arXiv preprint arXiv:2201.11903, 2022b. find our card
  53. Catherine Wong, Kevin M Ellis, Joshua Tenenbaum, and Jacob Andreas. Leveraging language to learn program abstractions and search heuristics. In International Conference on Machine Learning, pp. 11193–11204. PMLR, 2021. find
  54. Weizhe Yuan, Graham Neubig, and Pengfei Liu. Bartscore: Evaluating generated text as text generation. Advances in Neural Information Processing Systems, 34:27263–27277, 2021. find our card
  55. Eric Zelikman, Yuhuai Wu, and Noah D Goodman. Star: Bootstrapping reasoning with reasoning. arXiv preprint arXiv:2203.14465, 2022. find our card
  56. Aohan Zeng, Xiao Liu, Zhengxiao Du, Zihan Wang, Hanyu Lai, Ming Ding, Zhuoyi Yang, Yifan Xu, Wendi Zheng, Xiao Xia, et al. Glm-130b: An open bilingual pre-trained model. arXiv preprint arXiv:2210.02414, 2022. find our card
  57. Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen, Christopher Dewan, Mona Diab, Xian Li, Xi Victoria Lin, et al. Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068, 2022. find
  58. Daniel M Ziegler, Nisan Stiennon, Jeffrey Wu, Tom B Brown, Alec Radford, Dario Amodei, Paul Christiano, and Geoffrey Irving. Fine-tuning language models from human preferences. arXiv preprint arXiv:1909.08593, 2019. find