Glossary term
Glossary term
Foundations
A plot of both training loss and validation loss as a function of the number of iterations.
A generalization curve can help you detect possible overfitting. For example, the following generalization curve suggests overfitting because validation loss ultimately becomes significantly higher than training loss.
See Generalization in Machine Learning Crash Course for more information.
Created for this library
An ML platform team requires a generalization curve in every release review to detect overfitting before promotion to production.
A computer vision team uses generalization curves to decide when to stop training and which checkpoint to deploy.
A speech recognition team tracks generalization curves across model sizes to find the smallest model that still generalizes well to test audio.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License