Is it correct to use a BERT (or any other Transformers-based model) finetuned with a dataset like IMDb if I want to do sentiment analysis in a generic text (unrelated with movie
