Update layers.py #13829
Update layers.py #13829
Conversation
I have added a check on beta with reference to the following issue raised. tensorflow#11673 Please verify and get back. Thank you.
|
Can one of the admins verify this patch? |
|
@tensorflow-jenkins test this please |
|
Please add the test as requested in the previous version of the CL, otherwise this looks great, thanks! |
The following PR contains the required unit test case regarding issue tensorflow#11673 and it also has a fix in another pull request tensorflow#13829
|
#13864 contains the updated unit test case. |
|
Can one of the admins verify this patch? |
|
|
||
|
|
||
|
|
||
| class zero_debias_moving_mean(test.TestCase): |
There was a problem hiding this comment.
In this case, you can just add the test to the existing BatchNormTest suite, no need to add a new ones.
In general though, you can check the style of other class names in this file; one should use CamelCase test names here if you were adding a new class.
|
|
||
| class zero_debias_moving_mean(test.TestCase): | ||
|
|
||
| def Error_in_tf.contrib.layers.batch_norm_when(self): |
There was a problem hiding this comment.
test names should start with 'test':
def testBatchNormCenterFalse
| class zero_debias_moving_mean(test.TestCase): | ||
|
|
||
| def Error_in_tf.contrib.layers.batch_norm_when(self): | ||
| import tensorflow as tf |
There was a problem hiding this comment.
In internal tests, you have to import the modules individually, you cannot import tensorflow here. So you should use something like:
a = array_ops.placeholder(dtypes.float32, ...)
Take a look at the existing batch norm tests and use that as a guide for how to write the test where center=False
|
It looks like you may be having trouble updating this PR to include all the changes -- perhaps it makes sense to resend a new fresh PR with all of the requested changes (layers.py and layers_test.py change in the same PR). Thanks! |
The Changes have been made as suggested in my previous pull request. tensorflow#13829 The unit test case has been updated along with the update in layers.py with reference to the issue tensorflow#11673. Please verify and get back.
I have added a check on beta with reference to the following issue raised.
#11673
Please verify and get back.
Thank you.
I will add update the unit test cases to verify this patch.