Attention mechanism

An attention mechanism is a neural network component that lets a model weigh how relevant each part of its input is to every other part when building its representation of the data.

Attention mechanism sits in the Foundations part of the Agentik {OS} glossary, which defines the words used to build and run AI agent systems.

Also called Attention.