GitLab

Summary:
Addressed mthrok's comments in https://github.com/pytorch/audio/pull/2833:
- Moved model type from `_params` directly into the bundle definition. For now I defined model type as "WavLM" for WavLM bundles and "Wav2Vec2" for everything else. We can also distinguish between different Wav2Vec2 falvours - Hubert, VoxPopuli etc, but at the moment this won't imply any functional differences, so I didn't do it
- Expanded the title underline to match the title length

Pull Request resolved: https://github.com/pytorch/audio/pull/2895

Reviewed By: nateanl, mthrok

Differential Revision: D41799875

Pulled By: sgrigory

fbshipit-source-id: 0730d4f91ed60e900643bb74d6cccdd7aa5d7b39