The train.py script should have the feature to continue training from a particular checkpoint
The train.py script should have the feature to continue training from a particular checkpoint