Since the model has already been pre-trained, the model will easily be able to pick all the necessary features required to do the classification successfully. Also this pre-trained model allows us to partially re-train the classifier with out training data without the need of a large sample of data. Allowing us to train the classifier to our requirements.