tensorflow_multigpu_imagenet icon indicating copy to clipboard operation
tensorflow_multigpu_imagenet copied to clipboard

Published 20 hours ago • verified publisher icon arashno

VGG acc always keeps 0

Open GeKeShi opened this issue 5 years ago • 3 comments

I train vgg type D(default) from scratch, but the loss doesn't decline, and the acc always keeps 0, others model like resnet50 is fine. Is there anyone meet the same problem?

GeKeShi avatar Mar 20 '19 01:03 GeKeShi

Yes, sometimes it happens for the VGG. Have you tried lowering the learning rate?

arashno avatar Mar 23 '19 02:03 arashno

Have you solved this problem? Can you give me some suggestions for solving this problem? I tried to reduce the learning rate and still invalid

hiJulie avatar Apr 18 '19 02:04 hiJulie

Have you solved this problem? Can you give me some suggestions for solving this problem? I tried to reduce the learning rate and still invalid

no, I haven't, I replace it with alexnet and resnet, but the training speed is too slow to weeks

GeKeShi avatar Apr 18 '19 06:04 GeKeShi