[Wav2Vec2] Fix mask_feature_prob #13921
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
4 tasks
|
@anton-l - feel free to merge if the change is OK for you :-) |
patrickvonplaten
commented
Oct 7, 2021
| @@ -433,6 +434,52 @@ def _mock_init_weights(self, module): | |||
| if hasattr(module, "masked_spec_embed") and module.masked_spec_embed is not None: | |||
| module.masked_spec_embed.data.fill_(3) | |||
|
|
|||
| def test_mask_feature_prob_ctc(self): | |||
There was a problem hiding this comment.
this test was failing before
|
Makes sense, thanks for adding the tests too! 👍 |
Albertobegue
pushed a commit
to Albertobegue/transformers
that referenced
this pull request
Jan 27, 2022
* up * overwrite hubert
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Outdated suggestions cannot be applied.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.
What does this PR do?
When using
mask_feature_prob > 0it's important that theattention_maskis not passed as the attention_mask's dimension is[batch_size, sequence_length]. Since attention_mask = 1 vectors are padded anyways it doesn't matter whether some of those vectors are zero-ed out or not. Thus we should just never passattention_maskformask_feature_prob > 0.We however need it for
mask_time_probsincemask_time_probis used forWav2Vec2ForPretrainingwhere the created outputs will be used as labels.