Training Hybrid Language Models by Marginalizing over Segmentations.

Edouard Grave, Sainbayar Sukhbaatar, Piotr Bojanowski, Armand Joulin

Browse the full ACL paper archive.