It didn't perform any better than the regular model Removing batch normalisation significantly harmed training performance
7.5 KiB
7.5 KiB
It didn't perform any better than the regular model Removing batch normalisation significantly harmed training performance