Revert trainNetworks(...) CNN to an older network.

In matlab i am training a shallow convolutional neural network with the trainNetwork(...) function. After i stop the training early from not improving for x iterations, how do i revert the model back to the iteration with best validation accuracy? The function as it is now will save the name with an auto generated name - this means that i have to do it manually.

This early stop method is implemented in python in this paper: https://onlinelibrary.wiley.com/doi/abs/10.1002/hbm.23730

and their git: https://github.com/robintibor/braindecode