For binary classification models, in addition to accuracy, it's standard practice to compute additional metrics: precision, recall and F1 score. After evaluating the trained network, the demo saves ...