You’ve built a machine learning model that performs perfectly on training data but fails on new examples.