search

wjko2 / Domain-Agnostic-Sentence-Specificity-Prediction

13 stars 10 forks source link

Training new unlabeled data #7

Open radiantone opened 5 years ago

radiantone commented 5 years ago

Hi

The previous instructions for training new unseen data and making accurate predictions on it said to change paths in train.py and data2.py. These files specifically refer to a finite set of known datasets such as Yelp and twitter.

Its unclear exactly how to train on new domain sentences given the current code.

wjko2 commented 5 years ago

Please type there the file paths of your own file containing data from the new domain.