Techniques

Zero-Shot Learning

When an AI model completes a task without being given any specific examples in the prompt.

Detailed Explanation

Zero-Shot Learning refers to a model's ability to correctly complete a task it has never explicitly been trained to do, simply based on its general understanding of the prompt. For example, asking ChatGPT to translate English to French without providing any examples of French translations in the prompt is a zero-shot task. Models get better at zero-shot tasks as they scale in size.

Related Terms

Few-Shot Learning

A technique where the AI is provided with a few examples in the prompt to guide its behavior.

Prompt Engineering

The practice of designing and refining inputs (prompts) to get optimal results from AI models.

Back to Glossary