Elamur Research
☰
home / papers / 2022 / Large language models are zero-shot reasoners

Large language models are zero-shot reasoners

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/2205.11916

Summary

The paper asks whether large language models need hand-written, task-specific worked examples to solve multi-step reasoning problems. Earlier chain of thought (CoT) work relied on few-shot demonstrations. The authors propose Zero-shot-CoT, which uses no examples at all. The model receives the question followed by one fixed trigger, "Let's think step by step," and writes out its reasoning. A second prompt then pulls a correctly formatted answer out of that generated text. The same template was applied unchanged across arithmetic benchmarks (MultiArith, GSM8K, AQUA-RAT, SVAMP), symbolic tasks (Last Letter, Coin Flip), commonsense sets (CommonSenseQA, StrategyQA), and BIG-bench logic tasks (Date Understanding, Tracking Shuffled Objects). The authors tested InstructGPT, original GPT-3 and PaLM. Zero-shot-CoT far outperforms plain zero-shot prompting, although it still trails carefully engineered Few-shot-CoT. Its gains also increase with model size. The work matters because it sets a minimal, strong zero-shot baseline for reasoning tasks. It also suggests that LLMs hold general reasoning abilities that a single generic prompt can bring out.

Key points

  • With InstructGPT (text-davinci-002), compared with standard zero-shot prompting, MultiArith accuracy climbs to 78.7% (up from 17.7%) and GSM8K accuracy climbs to 40.7% (up from 10.4%).
  • PaLM (540B) shows improvements of a similar size on MultiArith and GSM8K. Gains also appear for vanilla GPT-3, so the method works independently of instruction tuning.
  • The pipeline has two stages: a reasoning-extraction prompt, then an answer-extraction prompt (e.g., asking for the answer in arabic numerals).
  • Few-shot-CoT stays stronger, but its accuracy falls when the example question types do not match the target task. Zero-shot-CoT needs no per-task engineering.
  • The authors name two limitations: little public information about the LLMs' training data, and biases the models inherit from web corpora. They argue the consistent gains across models point to general reasoning rather than memorization.

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: Mathematics & statistics

Cited by

2223 papers of our library cite it (of the 140673 whose bibliographies we read); 113 of them have a page here, the first 100 below, 2110 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. arXiv 2205.11916

@misc{kojima2022large,
  title = {{Large language models are zero-shot reasoners}},
  author = {Kojima, Takeshi and Gu, Shixiang Shane and Reid, Machel and Matsuo, Yutaka and Iwasawa, Yusuke},
  year = {2022},
  eprint = {2205.11916},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2205.11916}
}

References

49 references, 25 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Michael Ahn, Anthony Brohan, Noah Brown, Yevgen Chebotar, Omar Cortes, Byron David, Chelsea Finn, Keerthana Gopalakrishnan, Karol Hausman, Alex Herzog, Daniel Ho, Jasmine Hsu, Julian Ibarz, Brian Ichter, Alex Irpan, Eric Jang, Rosario Jauregui Ruano, Kyle Jeffrey, Sally Jesmonth, Nikhil J Joshi, Rya… find our card
  2. Sid Black, Leo Gao, Phil Wang, Connor Leahy, and Stella Biderman. GPT-Neo: Large Scale Autoregressive Language Modeling with Mesh-Tensorflow, March 2021. URL https://doi. org/10.5281/zenodo.5297715. — [doi:10.5281/zenodo.5297715](https://doi.org/10.5281/zenodo.5297715) find
  3. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D 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 Ziegler, Jeffrey Wu, Clemens Winter, Chri… find
  4. Curran Associates, Inc., 2020. URL https://proceedings.neurips.cc/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf. ↗
  5. François Chollet. On the measure of intelligence. arXiv preprint arXiv:1911.01547, 2019. URL https://arxiv.org/abs/1911.01547. ↗
  6. Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, Parker Schuh, Kensen Shi, Sasha Tsvyashchenko, Joshua Maynez, Abhishek Rao, Parker Barnes, Yi Tay, Noam Shazeer, Vinodkumar Prabhakaran, Emi… find
  7. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. Training verifiers to solve math word problems, 2021. URL https://arxiv.org/abs/2110.14168. ↗ our card
  8. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of NAACL, pages 4171–4186, 2019. URL https://aclanthology.org/N19-1423. ↗
  9. Leo Gao, Stella Biderman, Sid Black, Laurence Golding, Travis Hoppe, Charles Foster, Jason Phang, Horace He, Anish Thite, Noa Nabeshima, Shawn Presser, and Connor Leahy. The pile: An 800gb dataset of diverse text for language modeling. arXiv preprint arXiv: Arxiv-2101.00027, 2020. find
  10. Tianyu Gao, Adam Fisch, and Danqi Chen. Making pre-trained language models better few-shot learners. In Proceedings of ACL-IJCNLP, pages 3816–3830, 2021. URL https://aclanthology. org/2021.acl-long.295. find our card
  11. Mor Geva, Daniel Khashabi, Elad Segal, Tushar Khot, Dan Roth, and Jonathan Berant. Did aristotle use a laptop? a question answering benchmark with implicit reasoning strategies. TACL, 9:346–361, 2021. URL https://aclanthology.org/2021.tacl-1.21/. ↗
  12. Mohammad Javad Hosseini, Hannaneh Hajishirzi, Oren Etzioni, and Nate Kushman. Learning to solve arithmetic word problems with verb categorization. In EMNLP, volume 523533. Citeseer, 2014. URL https://aclanthology.org/D14-1058/. ↗ our card
  13. Wendy Johnson and Thomas J Bouchard Jr. The structure of human intelligence: It is verbal, perceptual, and image rotation (vpr), not fluid and crystallized. Intelligence, 33(4):393–416, 2005. find
  14. Rik Koncel-Kedziorski, Hannaneh Hajishirzi, Ashish Sabharwal, Oren Etzioni, and Siena Dumas Ang. Parsing algebraic word problems into equations. TACL, 3:585–597, 2015. URL https: //aclanthology.org/Q15-1042. find
  15. Rik Koncel-Kedziorski, Subhro Roy, Aida Amini, Nate Kushman, and Hannaneh Hajishirzi. MAWPS: A math word problem repository. In Proceedings of NAACL, pages 1152–1157, 2016. URL https://aclanthology.org/N16-1136. ↗
  16. Wang Ling, Dani Yogatama, Chris Dyer, and Phil Blunsom. Program induction by rationale generation: Learning to solve and explain algebraic word problems. In Proceedings of ACL, pages 158–167, 2017. URL https://aclanthology.org/P17-1015. ↗
  17. Jiachang Liu, Dinghan Shen, Yizhe Zhang, Bill Dolan, Lawrence Carin, and Weizhu Chen. What makes good in-context examples for gpt-3? arXiv preprint arXiv:2101.06804, 2021a. URL https://arxiv.org/abs/2101.06804. ↗ our card
  18. Pengfei Liu, Weizhe Yuan, Jinlan Fu, Zhengbao Jiang, Hiroaki Hayashi, and Graham Neubig. Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. arXiv preprint arXiv:2107.13586, 2021b. URL https://arxiv.org/abs/2107.13586. ↗ our card
  19. 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. In Proceedings of ACL, pages 8086–8098, 2022. URL https://aclanthology.org/2022.acl-long.556. ↗ our card
  20. Kevin S McGrew. The cattell-horn-carroll theory of cognitive abilities: Past, present, and future. 2005. find
  21. Stephen Merity, Caiming Xiong, James Bradbury, and Richard Socher. Pointer sentinel mixture models. arXiv preprint arXiv: Arxiv-1609.07843, 2016. URL https://arxiv.org/abs/1609.07843. ↗ our card
  22. Sewon Min, Xinxi Lyu, Ari Holtzman, Mikel Artetxe, Mike Lewis, Hannaneh Hajishirzi, and Luke Zettlemoyer. Rethinking the role of demonstrations: What makes in-context learning work? arXiv preprint arXiv:2202.12837, 2022. URL https://arxiv.org/pdf/2202.12837.pdf. ↗ our card
  23. Maxwell Nye, Anders Johan Andreassen, Guy Gur-Ari, Henryk Michalewski, Jacob Austin, David Bieber, David Dohan, Aitor Lewkowycz, Maarten Bosma, David Luan, Charles Sutton, and Augustus Odena. Show your work: Scratchpads for intermediate computation with language models. In Deep Learning for Code Wor… find our card
  24. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, and Ryan Lowe. Training la… find
  25. Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library. Advances in NeurIPS, 32:8026–8037, 2019. URL https://papers.nips.cc/paper/201… ↗ our card
  26. Arkil Patel, Satwik Bhattamishra, and Navin Goyal. Are NLP models really able to solve simple math word problems? In Proceedings of NAACL, pages 2080–2094, 2021. URL https://aclanthology.org/2021.naacl-main.168. ↗ our card
  27. Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. Language models are unsupervised multitask learners. OpenAI blog, page 9, 2019. URL http://www. persagen.com/files/misc/radford2019language.pdf. find
  28. Jack W. Rae, Sebastian Borgeaud, Trevor Cai, Katie Millican, Jordan Hoffmann, Francis Song, John Aslanides, Sarah Henderson, Roman Ring, Susannah Young, Eliza Rutherford, Tom Hennigan, Jacob Menick, Albin Cassirer, Richard Powell, George van den Driessche, Lisa Anne Hendricks, Maribeth Rauh, Po-Sen… find
  29. Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. Exploring the limits of transfer learning with a unified text-to-text transformer. JMLR, 21(140):1–67, 2020. URL http://jmlr.org/papers/v21/20-074.html. ↗
  30. Nazneen Fatema Rajani, Bryan McCann, Caiming Xiong, and Richard Socher. Explain yourself! leveraging language models for commonsense reasoning. In Proceedings of ACL, pages 4932–4942, 2019. URL https://aclanthology.org/P19-1487. ↗ our card
  31. 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, pages 1–7, 2021. URL https://arxiv.org/pdf/2102.07350.pdf. ↗ our card
  32. Subhro Roy and Dan Roth. Solving general arithmetic word problems. In Proceedings of EMNLP, pages 1743–1752, 2015. URL https://aclanthology.org/D15-1202. ↗ our card
  33. Victor Sanh, Albert Webson, Colin Raffel, Stephen Bach, Lintang Sutawika, Zaid Alyafeai, Antoine Chaffin, Arnaud Stiegler, Arun Raja, Manan Dey, M Saiful Bari, Canwen Xu, Urmish Thakker, Shanya Sharma Sharma, Eliza Szczechla, Taewoon Kim, Gunjan Chhablani, Nihal Nayak, Debajyoti Datta, Jonathan Chan… find our card
  34. Timo Schick and Hinrich Schütze. It’s not just size that matters: Small language models are also few-shot learners. In Proceedings of NAACL, pages 2339–2352, 2021. URL https://aclanthology. org/2021.naacl-main.185. find our card
  35. Taylor Shin, Yasaman Razeghi, Robert L. Logan IV, Eric Wallace, and Sameer Singh. Auto-Prompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts. In Proceedings of EMNLP, pages 4222–4235, 2020. URL https://aclanthology.org/2020. emnlp-main.346. ↗
  36. Vered Shwartz, Peter West, Ronan Le Bras, Chandra Bhagavatula, and Yejin Choi. Unsupervised commonsense question answering with self-talk. In Proceedings of EMNLP, pages 4615–4629, 2020. URL https://aclanthology.org/2020.emnlp-main.373. ↗
  37. Shaden Smith, Mostofa Patwary, Brandon Norick, Patrick LeGresley, Samyam Rajbhandari, Jared Casper, Zhun Liu, Shrimai Prabhumoye, George Zerveas, Vijay Korthikanti, Elton Zhang, Rewon Child, Reza Yazdani Aminabadi, Julie Bernauer, Xia Song, Mohammad Shoeybi, Yuxiong He, Michael Houston, Saurabh Tiwa… find our card
  38. 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
  39. Keith E Stanovich and Richard F West. Individual differences in reasoning: Implications for the rationality debate? Behavioral and brain sciences, 23(5):645–665, 2000. find
  40. Alon Talmor, Jonathan Herzig, Nicholas Lourie, and Jonathan Berant. Commonsenseqa: A question answering challenge targeting commonsense knowledge. In Proceedings of NAACL-HLT, pages 4149–4158, 2019. URL https://aclanthology.org/N19-1421/. ↗
  41. Romal Thoppilan, Daniel De Freitas, Jamie Hall, Noam Shazeer, Apoorv Kulshreshtha, Heng-Tze Cheng, Alicia Jin, Taylor Bos, Leslie Baker, Yu Du, YaGuang Li, Hongrae Lee, Huaixiu Steven Zheng, Amin Ghafouri, Marcelo Menegali, Yanping Huang, Maxim Krikun, Dmitry Lepikhin, James Qin, Dehao Chen, Yuanzho… find our card
  42. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. Attention is all you need. In I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, editors, Advances in NeurIPS, 2017. URL https://pro… find
  43. Ben Wang and Aran Komatsuzaki. GPT-J-6B: A 6 Billion Parameter Autoregressive Language Model. https://github.com/kingoflolz/mesh-transformer-jax, May 2021. ↗
  44. Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models. arXiv preprint arXiv:2203.11171, 2022. URL https://arxiv.org/abs/2203.11171. ↗ our card
  45. Albert Webson and Ellie Pavlick. Do prompt-based models really understand the meaning of their prompts? In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 2300–2344. Association for Computational Li… find our card
  46. 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, 2022. URL https: //arxiv.org/abs/2201.11903. find our card
  47. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lhoe… find our card
  48. Eric Zelikman, Yuhuai Wu, and Noah D. Goodman. Star: Bootstrapping reasoning with reasoning, 2022. URL https://arxiv.org/abs/2203.14465. ↗ our card
  49. 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. URL https://arxiv.org/abs/2205.01068. ↗