Glossary

Overfitting — Glossary

When a model fits training data quirks and fails to generalize to new data.

Reviewed:

Full definition

Classic signs include a large gap between training and validation metrics. Mitigations include more data, regularization, early stopping, and simpler models. Distribution shift is a related but distinct failure mode.

Related encyclopedia topic

Overfitting — Overfitting happens when a model memorizes training data quirks and fails to generalize to new data, producing optimistic offline scores and disappointing…

← Back to glossary · Learning hub