Few-Shot Learning
Few-shot learning is a prompting technique where a model is given several input-output examples directly in the prompt to demonstrate the desired task before being asked to handle a new instance. The examples act as an in-context training signal, guiding format, reasoning style, and label conventions without modifying the model's underlying weights through fine-tuning.
Definition
Few-shot learning is a prompting technique where a model is given several input-output examples directly in the prompt to demonstrate the desired task before being asked to handle a new instance. The examples act as an in-context training signal, guiding format, reasoning style, and label conventions without modifying the model's underlying weights through fine-tuning.
Examples
- Showing three labeled review-to-sentiment pairs before classifying a new review.
- Providing two question-and-SQL pairs to teach a target query dialect.
- Including sample translations to lock vocabulary choices for a domain.
Related terms
No credit card required
Frequently asked
- What is Few-Shot Learning?
- Few-shot learning is a prompting technique where a model is given several input-output examples directly in the prompt to demonstrate the desired task before being asked to handle a new instance. The examples act as an in-context training signal, guiding format, reasoning style, and label conventions without modifying the model's underlying weights through fine-tuning.
- How is Few-Shot Learning used in practice?
- Few-Shot Learning is commonly used in scenarios such as: Showing three labeled review-to-sentiment pairs before classifying a new review.; Providing two question-and-SQL pairs to teach a target query dialect.; Including sample translations to lock vocabulary choices for a domain..
- How does yno.ai support Few-Shot Learning?
- yno.ai supports Few-Shot Learning as part of its agentic AI platform — combining multiple frontier models with extensible tool use.
- What other concepts are related to Few-Shot Learning?
- Few-Shot Learning relates to Zero-Shot Learning, Prompt Engineering, Training Examples, Context Window. Each links to its own glossary page below.