Dear Participants,
We discovered that NaN values were unintentionally excluded from metric calculations, which resulted in some unusually high scores. This issue has now been fixed, and all affected metrics have been updated accordingly.
Thank you for your understanding.
Best regards,
Posted by: AlperBahcekapili @ May 7, 2025, 8:06 a.m.I was wondering how server's test metric calculations were much better than my local validation calculations :)
Posted by: caglarmert @ May 7, 2025, 4:36 p.m.