On the importance of pre-training data volume for compact language models.

Vincent Micheli, Martin d'Hoffschmidt, Franois Fleuret

Browse the full EMNLP paper archive.