>trainq1_trans = tfidf_vect.transform(df['question1'].values)

Thanks for the article. But what does it mean to train on some numbers; The tfid vectroization output is tokenid,documentid tuple against the tfidf score. What does it signify to train against those