Background: Training already has a stopping condition..., "Auto-stop after Epochs without an error rate low" controlled by a checkbox. This request is to Auto-stop on the next error rate low, also controlled by a checkbox. The intent is to stop training on a "local best" minimum error. When the box is checked, training should stop the next time "Epochs since Error rate low" equals zero.

Besides the checkbox and label, a single well-placed "if" statement might do it:

CODE:
Please log in to see this code.

My current workaround is to Clear Graphs, then Stop/Continue repetitively until the error is low in the graph. But that's a pain.

My suggested screen location is on the attached screen shot.

That sounds easy enough. Please create a ticket and I'll try to work up a test version for you.