Large Language Models are Zero-Shot Reasoners

By Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, Yusuke Iwasawa. 2022

7,591 citations recorded by the source corpus.

Pretrained large language models (LLMs) are widely used in many sub-fields of natural language processing (NLP) and generally known as excellent few-shot learners with task-specific exemplars. Notably, chain of thought (CoT) prompting, a recent technique for eliciting complex multi-step reasoning through step-by-step answer examples, achieved the state-of-the-art performances in arithmetics and symbolic reasoning, difficult system-2 tasks that do not follow the standard scaling laws for LLMs. While these successes are often attributed to LLMs' ability for few-shot learning, we show that LLMs are decent zero-shot reasoners by simply adding"Let's think step by step"before each answer. Experimental results demonstrate that our Zero-shot-CoT, using the same single prompt template, significantly outperforms zero-shot LLM performances on diverse benchmark reasoning tasks including arithmetics (MultiArith, GSM8K, AQUA-RAT, SVAMP), symbolic reasoning (Last Letter, Coin Flip), and other logical reasoning tasks (Date Understanding, Tracking Shuffled Objects), without any hand-crafted few-shot examples, e.g. increasing the accuracy on MultiArith from 17.7% to 78.7% and GSM8K from 10.4% to 40.7% with large InstructGPT model (text-davinci-002), as well as similar magnitudes of improvements with another off-the-shelf large model, 540B parameter PaLM. The versatility of this single prompt across very diverse reasoning tasks hints at untapped and understudied fundamental zero-shot capabilities of LLMs, suggesting high-level, multi-task broad cognitive capabilities may be extracted by simple prompting. We hope our work not only serves as the minimal strongest zero-shot baseline for the challenging reasoning benchmarks, but also highlights the importance of carefully exploring and analyzing the enormous zero-shot knowledge hidden inside LLMs before crafting finetuning datasets or few-shot exemplars.

Subfields: Artificial Intelligence, Information Systems, Safety Research, Large Language Models, Prompting & Reasoning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

  1. Mohit Bansal
  2. Jianfeng Gao
  3. Percy Liang
  4. Luke S. Zettlemoyer
  5. Quoc V. Le
  6. Yoshua Bengio
  7. Sergey Levine
  8. Yejin Choi

In Information Systems

  1. Percy Liang
  2. Jimmy J. Lin
  3. Jianfeng Gao
  4. Tushar Khot
  5. Peter Clark
  6. Wayne Xin Zhao
  7. Ashish Sabharwal
  8. Ji-rong Wen

In Safety Research

  1. Denny Zhou
  2. Jason Wei
  3. Alessandro Stolfo
  4. Mrinmaya Sachan
  5. Kumar Shridhar
  6. Jie Huang
  7. Wenhu Chen
  8. Quoc V. Le

In Large Language Models

  1. Luke S. Zettlemoyer
  2. Yejin Choi
  3. Noah A. Smith
  4. Jianfeng Gao
  5. Hinrich Schütze
  6. Mohit Bansal
  7. Samuel R. Bowman
  8. Yi Tay

In Prompting & Reasoning

  1. Denny Zhou
  2. Luke S. Zettlemoyer
  3. Jie Huang
  4. Kevin Chen-Chuan Chang
  5. Swaroop Mishra
  6. Wenhu Chen
  7. Tushar Khot
  8. Peter Clark

Citations in the corpus