Roc analysis, tester, builder (training/discovery)

I’m curious about the builder section of ROC analysis. When we choose logistic regression in the ‘Select an algorithm’ option, we’re presented with the ‘LR Model (10-Fold CV)’ tab. Could you explain the concept of ‘Training/Discovery’ in evaluating the logistic regression model’s performance? How does it differ from 10-fold CV or 100 cross-validations?