Fix broken test for models with batchnorm #17841
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Related to #17427, cc @amyeroberts . |
|
The documentation is not available anymore as the PR was closed or merged. |
...give you the same weights each time. I am good at machine learning.
amyeroberts
approved these changes
Jun 23, 2022
|
🧠 |
|
Give @amyeroberts the brain emoji for that one - she identified the whole problem, I just fixed the test! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
One of the Keras tests assumed that fitting a model for one iteration with a learning rate of zero would not change any weights. This is not true for
BatchNorm, which updates its running means and variances regardless! As a result, the model after the iteration had slightly different outputs, which caused the test to be very flaky. We now reinitialize the model after the single training epoch to make sure this doesn't happen.