Is there a way or method to exclude a word from corpus?

[Pooja-Vishnoi]

Like if word is incorrect for us and it considers word as correct as it might be in its built in corpus.
Ex clf.ignore['abc'. 'def']
before or after training model.