Wav2 vec2 phoneme ctc tokenizer optimisation #16817
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
Markdown references in headings such as '####' don't render well. Replaced it with <h4>...<a></a></h> banners.
The backend should only be initialized once, otherwise it is reloaded. Added `init_backend` function, intializes a backend attribute. Phonemize re-uses self.backend. Should give ~10 times faster phonemization.
|
The documentation is not available anymore as the PR was closed or merged. |
sgugger
approved these changes
Apr 18, 2022
There was a problem hiding this comment.
Congrats on your first PR!
LGTM with a few nits, but let's wait for @patrickvonplaten approval before merging.
src/transformers/models/wav2vec2_phoneme/tokenization_wav2vec2_phoneme.py
Outdated
Show resolved
Hide resolved
src/transformers/models/wav2vec2_phoneme/tokenization_wav2vec2_phoneme.py
Outdated
Show resolved
Hide resolved
patrickvonplaten
approved these changes
Apr 18, 2022
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
elusenji
pushed a commit
to elusenji/transformers
that referenced
this pull request
Jun 12, 2022
* Solved href rendering issue in heading Markdown references in headings such as '####' don't render well. Replaced it with <h4>...<a></a></h> banners. * PhonemeTokenizer optimization using phonemizer lib The backend should only be initialized once, otherwise it is reloaded. Added `init_backend` function, intializes a backend attribute. Phonemize re-uses self.backend. Should give ~10 times faster phonemization. * formatted file with make style * Documentation suggestion Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com> * Update /tokenization_wav2vec2_phoneme.py based on PR suggestion Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com> * Update CONTRIBUTING.md Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
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?
This is my FIRST PR!
The Wav2Vec2PhonemCTCTokenizer is slow when its argument
do_phonemizeis set to True. It re-initialises the backend at each forward pass. This is adressed using a class argument.There was also an H4 title in the documentation which had a link which did not render(
<h4></h4>used to replace####)Tests were passed, no additional ones were created. Runtime experiments to phonemize the entire 'tr' (turkish) subset of the common voice dataset gives a x10 boost in performances.
Models:
Documentation: @sgugger