Got an unexpected keyword argument when restore stt_zh_citrinet_512.nemo.

[Minxiangliu]

I tried to use stt_zh_citrinet_512.nemo for fine-tuning, but an error occurred when updating the vocabulary.

Traceback (most recent call last):
  File "---.py", line 67, in <module>
    model.change_vocabulary(new_tokenizer_dir=TOKENIZER_DIR, new_tokenizer_type="bpe")
TypeError: change_vocabulary() got an unexpected keyword argument 'new_tokenizer_dir'

By the way, this error does not appear in the stt_en_citrinet_512.nemo.

Any comments here?

[titu1994]

This model is not a BPE model - it does not use a tokenizer. Please refer to the documentation of EncDecCTCModel.change_vocabulary()