GLOSSARY TERM

What is Self-Supervised Learning?

A learning technique where the model derives its own training labels from the input data.
Self-supervised paradigms mask or corrupt parts of an input and challenge the model to reconstruct it. It unlocks the ability to pre-train massive architectures on vast, completely unlabeled datasets, learning deep universal representations.

Unlock Unlabeled Data

Train foundational representation models without manual labeling via M1.