You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi,
Recently, i read the articles about wide residual networks. I found that the latest version top1err is slightly better than the first version. 4.00% vs 4.17%, 4.27% vs 4.81%,4.53% vs 4.97%. I can't find the reason in the articles.Could you tell me the why? @szagoruyko
The text was updated successfully, but these errors were encountered:
Actually,I reproduced the wide residual networks, but i got slightly worse results. eg. wrn28-10 with top1err 4.20% . And i do exactly the same augmentation and preprocess. What should i do?
Thanks reply. @szagoruyko
Hi,
Recently, i read the articles about wide residual networks. I found that the latest version top1err is slightly better than the first version. 4.00% vs 4.17%, 4.27% vs 4.81%,4.53% vs 4.97%. I can't find the reason in the articles.Could you tell me the why?
@szagoruyko
The text was updated successfully, but these errors were encountered: