Deep Learning Dictionary - Lightweight Crash Course

Overfitting - Deep Learning Dictionary

Generally speaking, we say that a model is overfitting, or more specifically, overfitting the training set, when it is unable to perform well or generalize on data that wasn't included in the training set.

The goal of training is to create a model that will be able to be deployed and generalize well to new data based on what it learned during the training process. If it can only perform on the training set alone, then the model will not have a practical purpose.