Closed raja1196 closed 3 years ago
Changed the wav2vec.Dockerfile to make it work with the current wave2vec 2.0 and the updated fairseq repository. Added the base_architecture definition locally to recognize.py as it was no longer available in fairseq Wav2vec models (https://github.com/pytorch/fairseq/blob/master/fairseq/models/wav2vec/wav2vec2_asr.py).
Thank you
loretoparisi
commented
3 years ago loretoparisi
commented
3 years ago
Changed the wav2vec.Dockerfile to make it work with the current wave2vec 2.0 and the updated fairseq repository. Added the base_architecture definition locally to recognize.py as it was no longer available in fairseq Wav2vec models (https://github.com/pytorch/fairseq/blob/master/fairseq/models/wav2vec/wav2vec2_asr.py).