Sequence Reducible Holdout Loss for Language Model Pretraining.

Raghuveer Thirukovalluru, Nicholas Monath, Bhuwan Dhingra, Sam Wiseman

Browse the full COLING paper archive.