I am trying to download audio files using streaming mode. The steps go like this Download the dataset, processor feature extractor, prepare the dataset, seq2seqtrainer, compute metrics and then trainer.train().
When the training steps starts, after few downloads getting an error in tokenizer.utils.base as “You need to specify either ’ text’ or ’ text_target’”.
Have two questions here, one is not getting this error while downloading non streaming mode. Secondly after passing the parameter as " tex" / " text_target" still getting this error.
Any help would be appreciated!!