fix: randomize all layer types #252
Open
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
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #252 +/- ##
==========================================
+ Coverage 92.11% 92.12% +0.01%
==========================================
Files 55 55
Lines 2345 2350 +5
==========================================
+ Hits 2160 2165 +5
Misses 185 185 ☔ View full report in Codecov by Sentry. |
dilyabareeva
requested changes
Dec 23, 2024
There was a problem hiding this comment.
Hi @aski02. Thanks for your work. I have two major comments:
- The tests are only checking if parameters have been changed after randomization. Could we add tests that do inference on a batch of random data and see if the model output NaNs? That was the main reason we limited randomization to Linear layers previously.
- As far as I remember, the current approach didn't work for BatchNorm layers. Could we add tests for them as well?
dilyabareeva
requested changes
Jan 14, 2025
| @@ -67,17 +69,25 @@ def test_randomization_metric( | |||
| tmp_path, | |||
| request, | |||
| ): | |||
There was a problem hiding this comment.
Can we add some additional (tiny) architectures to conftest.py, and add a couple of parameterizations (extra items in the parametrization list) with these architectures? Maybe the easiest would be to create those tiny architectures manually, and not train them.
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Outdated suggestions cannot be applied.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.
References #18.