anton-l

/

xtreme_s_xlsr_300m_voxpopuli_en

like 0

Hello

I'm trying to use xtreme_s_xlsr_300m_voxpopuli_en to perform ASR (without fine-tuning), but I have the following error while launching the processor:

pretrained_model="anton-l/xtreme_s_xlsr_300m_voxpopuli_en"
model = Wav2Vec2ForCTC.from_pretrained(pretrained_model).to("cuda")
processor = Wav2Vec2Processor.from_pretrained(pretrained_model)
Traceback (most recent call last):
File "/home/brun/anaconda3/lib/python3.9/site-packages/transformers/models/wav2vec2/processing_wav2vec2.py", line 50, in from_pretrained
return super().from_pretrained(pretrained_model_name_or_path, **kwargs)
File "/home/brun/anaconda3/lib/python3.9/site-packages/transformers/processing_utils.py", line 187, in from_pretrained
args = cls._get_arguments_from_pretrained(pretrained_model_name_or_path, **kwargs)
File "/home/brun/anaconda3/lib/python3.9/site-packages/transformers/processing_utils.py", line 231, in _get_arguments_from_pretrained
args.append(attribute_class.from_pretrained(pretrained_model_name_or_path, **kwargs))
File "/home/brun/anaconda3/lib/python3.9/site-packages/transformers/models/auto/tokenization_auto.py", line 549...

Do you have an idea how to solve this problem ?

Many thanks!

Regards