Chain-of-Thought Prompting
Chain-of-thought prompting instructs a language model to produce intermediate reasoning steps before its final answer, typically by adding cues such as 'think step by step' or showing worked examples. Surfacing intermediate steps tends to improve accuracy on multi-step arithmetic, logical, and planning tasks, at the cost of longer outputs, higher latency, and more tokens consumed per response.
Definition
Chain-of-thought prompting instructs a language model to produce intermediate reasoning steps before its final answer, typically by adding cues such as 'think step by step' or showing worked examples. Surfacing intermediate steps tends to improve accuracy on multi-step arithmetic, logical, and planning tasks, at the cost of longer outputs, higher latency, and more tokens consumed per response.
Examples
- Appending 'Let's reason step by step.' to a multi-step word problem.
- Showing a worked example with numbered reasoning lines before the question.
- Asking an agent to list sub-goals before selecting a tool to call.
Related terms
No credit card required
Frequently asked
- What is Chain-of-Thought Prompting?
- Chain-of-thought prompting instructs a language model to produce intermediate reasoning steps before its final answer, typically by adding cues such as 'think step by step' or showing worked examples. Surfacing intermediate steps tends to improve accuracy on multi-step arithmetic, logical, and planning tasks, at the cost of longer outputs, higher latency, and more tokens consumed per response.
- How is Chain-of-Thought Prompting used in practice?
- Chain-of-Thought Prompting is commonly used in scenarios such as: Appending 'Let's reason step by step.' to a multi-step word problem.; Showing a worked example with numbered reasoning lines before the question.; Asking an agent to list sub-goals before selecting a tool to call..
- How does yno.ai support Chain-of-Thought Prompting?
- yno.ai uses Chain-of-Thought Prompting to power its AI agents, which combine planning, tool use, and dynamic context to handle multi-step tasks.
- What other concepts are related to Chain-of-Thought Prompting?
- Chain-of-Thought Prompting relates to Prompt Engineering, Few-Shot Learning, Agentic AI, Inference. Each links to its own glossary page below.