Before reading any words, I guessed just by seeing how long the paragraph was.
Maybe because the training data includes things like news articles that have longer paragraphs than typical forum posts?
Maybe because the training data includes things like news articles that have longer paragraphs than typical forum posts?