How to evaluate the BLEU score and train on other datasets?

[juntang-zhuang]

Hi, thanks for the nice repo. I have a quite naive question, is there any script to evaluate the BLEU score during the training, or test the trained model? Also if we want to experiment with other datasets, say WMT, could I just replace the dataset by torchnlp.datasets.wmt_dataset? Thanks a lot.