I have a huge data set of tweet texts (almost 3 billion tweets). I finetuned a BERT model with an annotated data set for a classification task.
Is there a way to make pre