Fewer Truncations Improve Language Modeling.

Hantian Ding, Zijian Wang, Giovanni Paolini, Varun Kumar, Anoop Deoras, Dan Roth, Stefano Soatto

Browse the full ICML paper archive.