Closed de9uch1 closed 2 years ago
Note that JParaCrawl has a different format for v2 and v3, and the tsv columns that should be extracted have changed.
Thanks @de9uch1 for this PR!
I hope you don't mind me taking these changes into develop
branch first and releasing a new version along with a few other changes.
Fixed a bug in the extraction of JParaCrawl v3 used in WMT22 en-ja translation task. The minor version has been bumped to need to update the index cache.