Chain-of-Thought Prompting Elicits Reasoning in Large Language Models

Overview

Chain-of-thought prompting is a technique that elicits step-by-step reasoning from large language models by providing few-shot examples that include intermediate reasoning steps. It improves performance on tasks requiring multi-step reasoning, such as math word problems and logical puzzles.

Best for

Best for
Developers needing improved reasoning capabilities from LLMs for arithmetic or logical problem-solving tasks

Use cases

• Solving multi-step math word problems
• Answering complex commonsense reasoning questions
• Generating explanations for model outputs

Notes

Chain-of-thought prompting is a technique that elicits step-by-step reasoning from large language models by providing few-shot examples that include intermediate reasoning steps. It improves performance on tasks requiring multi-step reasoning, such as math word problems and logical puzzles.

Use cases

• Solving multi-step math word problems
• Answering complex commonsense reasoning questions
• Generating explanations for model outputs

Pros

• Improves accuracy on reasoning tasks without fine-tuning
• Simple to implement with carefully crafted few-shot examples
• Provides interpretable reasoning traces

Cons

• Requires manual construction of few-shot examples
• May not generalize to out-of-distribution tasks
• Increases token output and latency

Indexed from awesome-llm and enriched against its public facts.