Closed ahmednasserswe closed 4 years ago
Hey Ahmed, I just uploaded the train and test splits as separate files, the split is 50-50 and is based on articles, paragraphs from the same article only appear in one of the splits. If you have the computational resources you could do two-fold CV (switch roles of train and test, or do a different split) for a better validation of your model.
Hello,
I can not find the ARCD test and train sets, I can only find one file under
/data
that is calledarcd.json
. How can I find the test and train split as indicated in the paper? I'd like to compare my results to yours that's why I am interested in finding the exact datasets used.شكرا!