Last week:
did some experiments of merging the two methods "regularization" and "pre-training" together in the RNN traning with knowledge transfer;
This week:
accomplish it.