Attention Mechanism

A key architecture component in modern deep learning (specifically Transformers) that allows models to dynamically focus on specific parts of an input sequence when generating output, enabling long-range contextual understanding.

Part of the Data & AI Terms glossary.