lizard24

I have experienced similar problems. E.g., the training stopped after the first completed epoch without exiting - that was after executing train.py in a jobscript on a server. Other times...