Duplicates in english split
#2
by
TomData
- opened
Hello,
I found some duplicates in the dataset for the english split. And for one case I also with different labels.
Is this intended, or something that has been overlooked?
Thank you very much for your reply!
Hey! I noticed that for some languages (Indonesian, Javanese, Xhosa, Zulu) we have dev and test but no train. How come? Would it be possible to have train and dev but no test instead?
Hi @MauroExtrac ,
Please refer to our dataset paper https://aclanthology.org/2025.acl-long.436/ (see Table 1). These were used for testing only (in crosslingual settings). You are welcome to change the splits though.