No default predictor for model type bert_crf_tagger.

[dpappas]

Hi everyone!

I tried the following command to use my fine-tuned model for NER.

python -m allennlp.run predict --output-file=out.txt --include-package=scibert ./ner_model_bc5cdr/model.tar.gz ./data/ner/bc5cdr/test.txt

I have used the same commands you mention in your readme file and a virtual environment so the versions I use are exactly the same as yours.

I get the following error:
'No default predictor for model type bert_crf_tagger.\nPlease specify a predictor explicitly.'

How can I overcome this problem? Also, how do I specify the predictor?
I tried
--predictor=ner_model_bc5cdr
--predictor=./scibert/models/bert_crf_tagger
--predictor=./scibert/models/bert_crf_tagger.py
but nothing worked.

Can you help me?

[aadarshsingh191198]

--predictor sentence-tagger is probably what you are looking for