Backprop Test for Freeze FlaxWav2Vec2 Feature Encoder #15938
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
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
sanchit-gandhi
force-pushed
the
flax-wav2vec2
branch
from
85ad7e8
to
bf29d36
Compare
patrickvonplaten
approved these changes
Mar 5, 2022
|
If @patil-suraj is happy with this test I'll merge! |
patil-suraj
approved these changes
Mar 7, 2022
Comment on lines
276
to
283
| # ensure that the gradients of the frozen layers differ, i.e. that the feature encoder is properly frozen | ||
| feature_extractor_grads = tuple(grads[k] for k in grads if "feature_extractor" in k) | ||
| feature_extractor_grads_frozen = tuple(grads_frozen[k] for k in grads_frozen if "feature_extractor" in k) | ||
|
|
||
| for feature_extractor_grad, feature_extractor_grad_frozen in zip( | ||
| feature_extractor_grads, feature_extractor_grads_frozen | ||
| ): | ||
| self.assert_difference(feature_extractor_grad, feature_extractor_grad_frozen, 1e-7) |
There was a problem hiding this comment.
could we also add one more check to see if the grads of frozen module are all precisely zero, since that's what jax.lax.stop_gradient is supposed to do.
There was a problem hiding this comment.
Thank you for the feedback! Sure thing, I'll look into that now!
There was a problem hiding this comment.
The most recent commit (ca918e9) adds an assertion that verifies that the gradients of the frozen feature encoder layers are precisely zero!
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Outdated suggestions cannot be applied.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.
This PR correctly implements a back propagation test to verify the functionality of the
freeze_feature_encoderargument added to the FlaxWav2Vec2 Model in #15873. It tests: