Glossary term

One-shot learning

Giving a model a single example in the prompt before asking it to do a task.

How it works

One-shot prompting gives exactly one example. It is the middle step between zero-shot (no examples) and few-shot (several).

In classic machine learning the term also refers to learning to recognize a new class from a single example.

Zero, one, few

A one-shot prompt

Convert to JSON.

"Ana, 31" -> {"name": "Ana", "age": 31}

"Luis, 45" ->

A single example is often enough to fix the output format.