Support public MMS demo #1187

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

Merged

cbalioglu merged 18 commits into v0.4.6 from mattsetzler/mms_demo

May 29, 2025

Contributor

mattsetz commented May 27, 2025 •

edited

Loading

What does this PR do? Please describe:

exposes load_from_path in model hub
implements HuggingFaceAsrDataset

Does your PR introduce any breaking changes? If yes, please list them:
No.

Testing
Tested new functionality in jupyter notebook.

mattsetzler and others added 4 commits

May 25, 2025 08:03


          Add support for mms ASR finetuning of 300M param model.

3e1351e


          Expose load_from_path in ModelHub.

d4724ba


          Fixed bug in model hub load_from_path.

d0ce8a6


          Add HuggingFaceDataset implementation for speech ASR.

e376e5b

mattsetz requested a review from cbalioglu as a code owner

May 27, 2025 15:27

facebook-github-bot added the CLA Signed label

mattsetz commented

View reviewed changes

src/fairseq2/assets/cards/datasets/bible.yaml Outdated Show resolved Hide resolved

mattsetz commented

View reviewed changes

src/fairseq2/assets/cards/models/wav2vec2.yaml Outdated Show resolved Hide resolved

mattsetz commented

View reviewed changes

src/fairseq2/datasets/asr.py Outdated Show resolved Hide resolved

mattsetz commented

View reviewed changes

src/fairseq2/datasets/asr.py Outdated Show resolved Hide resolved

mattsetz commented

View reviewed changes

src/fairseq2/recipes/wav2vec2/asr/_train.py Outdated Show resolved Hide resolved

mattsetzler and others added 7 commits

May 27, 2025 11:18


          Support loading of HuggingFaceAsrDataset in load_wav2vec2_asr_trainer.

dd50d42


          Fixes for HuggingFaceAsrDataset.

35db4e7


          Fix linter issues and clean up naming.

dfe7154


          Updated naming.

e5de723


          Add 1B arch config for w2v2_asr model.

a355779


          Remove MMS-specific data and model cards.

9d0f2df


          Revert unnecessary formatting in imports.

b83909f

mattsetz commented

View reviewed changes

src/fairseq2/datasets/asr.py Outdated Show resolved Hide resolved

src/fairseq2/recipes/wav2vec2/asr/_train.py Show resolved Hide resolved

mattsetz changed the title ~~[WIP] Support public MMS demo~~ Support public MMS demo

cbalioglu reviewed

View reviewed changes

Contributor

cbalioglu left a comment

Mostly looks good to me. Once you fix all linter errors and make datasets dependency optional, it is good to go.

src/fairseq2/models/wav2vec2/_handler.py Outdated Show resolved Hide resolved

src/fairseq2/datasets/asr.py Outdated Show resolved Hide resolved

src/fairseq2/datasets/asr.py Outdated Show resolved Hide resolved

mattsetzler user and others added 7 commits

May 28, 2025 19:30


          Address nits and make datasets library optional.

72cee3e


          Fix linter errors and add transcript_key arg to HuggingFaceAsrDataset.

b039932


          Silence linter for file unmodified by this PR.

a795bb4


          Formatting changes after running isort . && black .

cab1a23


          Add all datasets imports to try/catch block.

b0ef189


          Update version to v0.4.6

dd7c275


          Revert "Update version to v0.4.6"

58272be

This reverts commit dd7c275.

cbalioglu changed the base branch from v0.4.5 to v0.4.6

May 29, 2025 11:32

cbalioglu merged commit 249392a into v0.4.6

15 checks passed

cbalioglu deleted the mattsetzler/mms_demo branch

May 29, 2025 11:39

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels