Thanks for sharing the great work! I tried to reproduce the paper results for Visda17 datasets. The result is 63.98 obtained in the first epoch and then kind of decrease after the first epoch. The best result was obtained from the first epoch after 30 epoches training. May I ask what's the potential reason for that? Thanks!
Carina
Thanks for sharing the great work! I tried to reproduce the paper results for Visda17 datasets. The result is 63.98 obtained in the first epoch and then kind of decrease after the first epoch. The best result was obtained from the first epoch after 30 epoches training. May I ask what's the potential reason for that? Thanks!
Carina