Closed chiragjn closed 2 years ago
@chiragjn LGTM, should we merge this?
unfortunately, this still does not solve the problem, without going component by component it is hard to tell what is still causing the variance :(
Closing this for now, this still remains a issue :upside_down_face:
Attempts to fix #45
Taken directly from https://github.com/zihangdai/xlnet/pull/151/commits/d1d0ff4d0d2e536e897adabcdd04683e62dfa7a4
Todo: