[BugFix] Fix extract_features method for WavLM models

[nateanl]

resolve #3347

position_bias is ignored in extract_features method, this doesn't affect Wav2Vec2 or HuBERT models, but it changes the output of transformer layers (except the first layer) in WavLM model. This PR fixes it by adding position_bias to the method.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/audio/3350

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure, 3 Unrelated Failures

As of commit e8fa75a:

NEW FAILURE - The following job has failed:

• build (3.8) (gh)

BROKEN TRUNK - The following jobs failed but were present on the merge base f046f7e:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• tests (3.10, 11.7) / linux-job (gh)
• tests (3.8, 11.7) / linux-job (gh)
• tests (3.9, 11.7) / linux-job (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[sgrigory]

LGTM, thanks for fixing this! When verifying against HF results https://github.com/pytorch/audio/blob/f9b4f74fd5e8a3ec92d77b20102d187090a8ad15/test/torchaudio_unittest/models/wav2vec2/huggingface_intergration_test.py#LL310C9-L310C28 we never call model.extract_features or transformer.get_intermediate_outputs, that's probably why this went unnoticed

[facebook-github-bot]

@nateanl has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[facebook-github-bot]

@nateanl merged this pull request in 7d0f336.