How to use kul-speech-lab/wav2vec2-xls-r-300m-lay21 with Transformers:
# Load model directly from transformers import AutoProcessor, AutoModelForPreTraining processor = AutoProcessor.from_pretrained("kul-speech-lab/wav2vec2-xls-r-300m-lay21") model = AutoModelForPreTraining.from_pretrained("kul-speech-lab/wav2vec2-xls-r-300m-lay21")
This repository contains the wav2vec2 XLS-R 300M model truncated to 21 layers.
See also: https://github.com/lcn-kul/xls-r-analysis-sqa
