About training data for question generation?
linmou opened this issue · 0 comments
linmou commented
As it mentioned in paper, there was a corpra for trianing MT model , where S contains sentences extracted from Wikipedia, and Q contains user-generated questions from a QA site. While I didn't find that in this project.
I believe it is hard for personal researcher to rebuild up such corpra by themselves, not to mention to reproduce the entire model. Could you do us a favor to provide these data for further research? Or you have already provided it but I didn't find ?
Best wishes.