Zero-Shot CoT

Appears in 1 paper

A variant of chain-of-thought prompting where no examples are provided.

As used in Paper 14 — Chain-of-Thought Prompting Elicits Reasoning in Large Language Models →

A variant of chain-of-thought prompting where no examples are provided. Instead, the prompt simply includes "Let's think step by step" before the question. First demonstrated by Kojima et al. (2022), zero-shot CoT achieves strong performance on large models without requiring human-authored reasoning examples. On GSM8K, GPT-3 achieved 41% accuracy with zero-shot CoT (vs. 17% standard prompting).