In-Context Learning

A large language model's ability to temporarily learn from prompts, without any weight updates — an emergent property of scale.

Specifications & Metrics

type

Prompting & Reasoning

Core Principles & Analysis

• A model's ability to temporarily learn from prompts is known as in-context learning.
• It is an emergent ability of large language models.
• It is an emergent property of model scale: breaks in scaling laws occur, so its efficacy increases at a different rate in larger models than in smaller ones.

Sources

• Wikipedia: In-context learning