Expose vocab_info in wav2vec2 ASR model

facebookresearch / fairseq2

FAIR Sequence Modeling Toolkit 2

https://facebookresearch.github.io/fairseq2/

MIT License

613 stars 59 forks source link

Closed cbalioglu closed 3 days ago

cbalioglu commented 3 days ago

This is a small PR that slightly updates Wav2Vec2AsrModel to follow the vocab_info accessor convention of other models in fairseq2.