Evaluation accuracy - W2

In the Python script,(C1W2_Ungraded_Lab_Birds_Cats_Dogs.ipynb) we are using the evaluation accuracy, evaluation loss.

What exactly these terms mean? and what the distribution of evaluation accuracy, evaluation loss with the Accuracy and Loss show ?

hi @grvsinha , welcome to the course!

A good machine learning model is one that works well on unseen data. So to check that we still have a good model in training, we separate our data into two sets here: a training set and a validation set (or evaluation set). The term evaluation accuracy and evaluation loss is the accuracy and loss we got when we feed the model with evaluation data (not seen in training). This give us a sense of how good our model is. Usually evaluation accuracy is lower than (training) accuracy, and the evaluation loss is bigger than (training) loss.

Doest that answer your question?