Overfitting

A model overfits when it memorises its training examples, noise and all, instead of learning patterns that carry over. The tell-tale sign: training loss keeps falling while the loss on held-out test data starts rising. Remedies include more and more varied data, a smaller model, stopping training early, and regularisation.

Why it matters: It's why models are judged on data they haven't seen, and why more data usually helps.

Builds on: Training versus using a model, Loss functions

Learn it, then prove it

Sign in (free) to take this step's mastery check and save your journey.