Already done:

1)the usefulness of yield in processing data in ml

2)Jupiter notebook transfer to & from python

3)importlib is to import lib dynamically. (you can designate lib name by string)

 

4)one day to understand yelp_prepare.py / yelp.py code in Hierarchical Attention.

5)one day to run yelp_prepare.py (16hours for now)

produce: word_freq.pickle / vocal.pickle / train.dataset / dev.dataset / test.dataset