lunny/keras
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix ValueError(ndim of gamma and beta) of batch normalization when using Theano (#3740)

Browse Source 
* Fix ndim mismatch error when using theano

* Change keras backend call
This commit is contained in:
Seonghyeon Nam 2016-09-16 10:09:02 +09:00 committed by François Chollet
parent 8b42fff90e
commit 56f3c85b87
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
keras/layers/normalization.py
View File

@ -140,7 +140,7 @@ class BatchNormalization(Layer):
self.updates = [K.moving_average_update(self.running_mean, mean, self.momentum),
K.moving_average_update(self.running_std, std, self.momentum)]
if sorted(reduction_axes) == range(K.ndim(x))[:-1]:
if K.backend() == 'tensorflow' and sorted(reduction_axes) == range(K.ndim(x))[:-1]:
x_normed_running = K.batch_normalization(
x, self.running_mean, self.running_std,
self.beta, self.gamma,