Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

✨ Add speech recognition training & other improvements #147

Merged
merged 36 commits into from
Feb 7, 2024
Merged

✨ Add speech recognition training & other improvements #147

merged 36 commits into from
Feb 7, 2024

Conversation

arxyzan
Copy link
Member

@arxyzan arxyzan commented Feb 7, 2024

Pull Request

Description

This PR mainly adds support for ASR regarding dataset, training, collator, etc. But in the process some other improvments and bug fixes were also introduced.

Changes

  • Add SpeechRecognitionDataset and SpeechRecognitionDataCollator
  • Add SpeechRecognitionMetricsHandler
  • Add training script for ASR in examples
  • Fix bugs in Tokenizer
  • Fix bugs in AudioFeatureExtractor
  • Fix bugs in config
  • Add dataset load script to templates for ASR
  • Improve tests

Related Issues

Mainly focuses on #72

Checklist

  • I have read and followed the project's contributing guidelines.
  • My code follows the project's coding style.
  • I have tested my changes thoroughly.
  • I have updated the documentation if necessary.
  • All existing tests pass.
  • I have added new tests to cover my changes.
  • My changes do not introduce any new warnings or errors.

Additional Comments

Reviewer Instructions

Author's Note

@arxyzan
✨ Implement metrics handler for speech recognition
f5b6ab4
@arxyzan
🍱 Add train_speech_recognition.py
fe7ff00
@arxyzan
Merge branch 'main' into asr-training
63af54b
@arxyzan
🐛 Fix resolve_inputs_length_for_padding bug
7a9a7b6
@arxyzan
✏️ Minor
e621612
@arxyzan
✏️ Minor
fa3c006
@arxyzan
🐛 Fix fields bug in Config
4672331
@arxyzan
🐛 Fix bug in AudioFeatureExtractor.pad()
f507b11
@arxyzan
✨ Add SpeechRecognitionDataCollator
56632f9
@arxyzan
🐛 Fix wrong attributes in WhisperBPEConfig
7068c2f
@arxyzan
🐛 Handle bugs in WhisperSpeechRecognition
bbbf902
@arxyzan
✨ Add dataset loading script for ASR
35a16a4
@arxyzan
✨ Add SpeechRecognitionDataset
d505b09
@arxyzan
🐛 Fix tokenizer max_length bug
6b1a7c7 
Setting `max_length` in tokenizer call had unexpected behavior
@arxyzan
✏️ Improve logging robustness in Trainer
f3b6988
@arxyzan
✏️ Update train_speech_recognition.py
bc30588
@arxyzan
✏️ Rename padding -> padding_type in `data_utils.resolve_inputs_l…
c4dd743 
…ength_for_padding()`
@arxyzan
🧪 Add speech_recognition to tests for datasets and trainer
c95c4f5
@arxyzan
🧪 Improve flexibility of tests in test_datasets.py
19574b6
@arxyzan
🧪 Ignore errors for rmtree in test_trainer.py
3f41b11
@arxyzan
✏️ Fix minor issues for asr
3eb23d5
@arxyzan
🧪 Limit max input lengths to prevent crash in CI
fae959d
@arxyzan
✨ Add clean_cache function to utils
f3c28d1
@arxyzan
🧪 Clean cache after every train process
e8eaa29
@arxyzan
🧪 Add CI_MODE to tests.yml
b37cd26
@arxyzan
🐛 Fix minor bug in speech_recognition_dataset.py
cdea95d
@arxyzan
🧪 Minor change
d9a298e
@arxyzan
🧪 Clean cache in test_datasets.py
045b820
@arxyzan
🧪 Minor change
d00c30a
@arxyzan
🧪 Clean cache after every test
ac6a8db
@arxyzan arxyzan merged commit d7dadd6 into main Feb 7, 2024
1 check passed
@arxyzan arxyzan mentioned this pull request Feb 7, 2024
Closed
@arxyzan arxyzan deleted the asr-training branch February 10, 2024 12:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@arxyzan