patrickvonplaten

/

wav2vec2-conformer-rel-pos-large-960h-ft-4-gram

like 0

Hi, as the intro said "Wav2Vec2-Conformer with relative position embeddings, pre-trained and fine-tuned on 960 hours of Librispeech on 16kHz sampled speech audio", but in the Github readme, the pre-train dataset is 60k Libri-Light.

Which one is correct?

Thanks a lot!