bc_learning_sound icon indicating copy to clipboard operation
bc_learning_sound copied to clipboard

Published 20 hours ago • verified publisher icon mil-tokyo

Loss NAN from epoch 1

Open mayankb2103 opened this issue 6 years ago • 1 comments

I was trying to replicate your code. System details: Chainer 1.24, Cuda8, cudnn5.1 I have also generated dataset as given. I am using envnet, esc50, no BC Learning, no strong augment. However I am getting Loss nan, top1 98.00 for a train and val both

mayankb2103 avatar Dec 28 '18 07:12 mayankb2103

Hello, have you solved this problem? I also have this problem, can you help me?

kangshuning avatar Sep 17 '19 07:09 kangshuning