It is simply a continuation of training with a extended dataset.
With many diffent styles, eyes getting a bit unprecise, testing it only for training check. I can fix that, but i spend that time better on further training and dataset working, instead staying and spending time on a between Version.