I am working with newest dataset provided on phase1 , I tried with test_unseen.jsonl and test_seen.jsonl , what am I doing wrong ?
Have you checked the order of the IDs?
If I am not mistaken for phase 1 you should use the _seen version
Yes, I checked and I rearranged as shown in test_seen.jsonl . It fixed the issue.