Large Language Models are Zero-Shot Reasoners (May 2022)
improve zero-shot prompt performance of LLMs by adding “Let’s think step by step” before each answer
improve zero-shot prompt performance of LLMs by adding “Let’s think step by step” before each answer