Datasets:
wsc.fixed span2_index off by one
#12
by
ianporada
- opened
For validation example index 42, span2_index seems off by one. Right now it's 30, but `text.split(" ")[30] is "use" whereas the span2_text is "it" (at index 31).
Thanks for reporting, @ianporada .
Yes, that is an issue with the original data. Maybe, it would be worth informing the authors of the WSC subset...
I see. You are right. This is a typo in the original dataset. I'll close the issue.
ianporada
changed discussion status to
closed