Question about whether the RNN model is trained in run_reweight.py

[LindaLydia]

A question about run_reweight.py.
It seems that, in the current version code you provided on branch "main", the RNN model is never updated. As in function train_to_converge(), a model_copy is copied from the constructed RNN model using deepcopy and trained. But no back copy or assignment is found.
Did I miss anything or did I made any misunderstanding? Looking forward to your rely! Thanks a lot!