NaN's with multiple pooling layers with MNIST clouds :(

Browse Source

Tried building a model with these ops, but with multiple layers of pooling I get NaN value after a while. I think there is a problem with an unstable gradient in the this layer.

...

This commit is contained in:

Mike Clark

2018-10-16 08:14:51 +08:00

committed by GitHub

parent e2fdad61d9

commit aae13f0126

1 changed files with 1009 additions and 0 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

mnist_clouds.ipynb

+1009

File diff suppressed because it is too large Load Diff