Glossary
What is MIRAS? Definition and meaning
A research framework that describes memory modules by their attentional bias, retention gate and update rule.
Definition
MIRAS: A research framework that describes memory modules by their attentional bias, retention gate and update rule.
In short: A framework for memory modules in sequence models.
When it is used
Designing memory-augmented models for continual learning.
Learn more
The full guide to MIRAS explains how it works, when to use it and how it compares with the alternatives, with examples. The glossary lists every term on the site from A to Z.
Previous term: Milvus · Next term: MirrorMaker 2