WIT corpus, SETimes corpus, newsdev2016, newstest2016, and newstest2017

The dataset used in the paper is the WIT corpus, SETimes corpus, newsdev2016, newstest2016, and newstest2017.

BibTex: