Few-Shot Learning
Providing a small number of input-output examples in the prompt to teach the model a task pattern.
Full Definition
Few-shot learning places two to ten demonstrations of a task directly inside the prompt, allowing the model to infer the pattern and apply it to a new input — all without updating any weights. The quality of examples matters enormously: diverse, representative demonstrations outperform redundant ones. Few-shot prompting is powerful because it sidesteps the need for expensive fine-tuning while still adapting a general model to a niche format or domain. It sits between zero-shot (no examples) and fine-tuning (many examples used to update weights) on the effort-versus-performance spectrum.
Examples
Showing three (tweet, sentiment) pairs before asking the model to classify a fourth tweet as positive, negative, or neutral.
Providing two example SQL queries with their natural-language descriptions before asking the model to write a third query from a new description.
Apply this in your prompts
PromptITIN automatically uses techniques like Few-Shot Learning to build better prompts for you.
Related Terms
Zero-Shot Learning
Asking a model to perform a task it has never seen demonstrated, relying purely …
View →Instruction Prompting
Directly telling the model what to do using clear imperative commands.…
View →Prompt Template
A reusable prompt structure with placeholders that are filled in at runtime.…
View →