Few-Shot Prompting: Definition and Examples

Few-shot prompting provides a few examples (2-5) in your prompt to show the AI the expected format, tone, or type of response, without fine-tuning.

Full definition

Few-shot is one of the most powerful prompt engineering techniques. Instead of describing what you want, you show the expected result through concrete examples.

The principle: include 2-5 input/output pairs before your actual query. The model detects the pattern and applies it.

Few-shot differs from zero-shot (no examples) and one-shot (single example). Beyond 5 examples, gains are often marginal.

Example quality is crucial: representative, consistent, error-free. A bad example teaches bad habits.

Etymology

Comes from machine learning (few-shot learning). Popularized by the GPT-3 paper (2020) demonstrating in-context learning.

Concrete examples

Sentiment classification with 3 examples

Classify the sentiment:

Review: "Fast delivery"
Sentiment: Positive

Review: "Damaged package"
Sentiment: Negative

Review: "Customer service was incredible"
Sentiment:

Structured data extraction

Extract the info:

Text: "Marie Dupont - Marketing Dir. - Acme SAS"
JSON: {"name":"Marie Dupont","position":"Marketing Dir."}

Text: "Jean Martin, CTO @ StartupXYZ"
JSON:

Technical rewriting into plain language

Practical usage

Use few-shot when a simple instruction isn't enough. Choose 2-3 varied examples. Place them just before your query.

Related concepts

Zero-shotOne-shotIn-context learningChain-of-thought

FAQ

How many examples in few-shot?

2-5 are enough. Simple tasks: 2. Complex tasks: 3-5. Beyond 5, marginal gains.

Few-shot or fine-tuning?

Few-shot first — fast, free, flexible. Fine-tuning if you have thousands of examples or a very specialized domain.

Does the order of examples matter?

Yes. The last example has more influence (recency bias). Place the most representative one last.

How to use this prompt

Copy the prompt with the button above.
Paste it into ChatGPT, Claude or your favorite AI assistant.
Replace the bracketed variables with your details, then refine the result.

About Prompt Guide

Prompt Guide is a free library of 2500+ ready-to-use prompts for ChatGPT, Claude and other AIs, with guides to learn prompting and tools to build and optimize your own prompts.

Prompt library Learn prompting Prompt builder Prompt optimizer

More definitions

Fine Tuning: Definition and Examples

Fine tuning is the process of adjusting a pre-trained AI model on a specific dataset to improve its performance for a particular task or

Flash Attention: Definition and Examples

Flash Attention is an optimized algorithm for computing the attention mechanism in Transformers, drastically reducing memory consumption and accelerating...

Frequency Penalty: Definition and Examples

The Frequency Penalty is a language model parameter that penalizes tokens proportionally to the number of times they appear in the generated text

Function Calling: Definition and Examples

Function Calling is a capability of language models (LLMs) that allows them to identify when to call an external function and generate the required arguments.

Function Grounding: Definition and Examples

Function Grounding is a technique that anchors an AI model's responses in executable functions or tools, allowing it to interact with systems

GDPR AI: Definition and Examples

GDPR AI refers to the application of the General Data Protection Regulation to artificial intelligence systems, governing the collection, processing, and use of personal data by AI algorithms and models.

Get new prompts every week

Join our newsletter.