Gradient Vaccine: Investigating and Improving Multi-task Optimization in Massively Multilingual Models.

Zirui Wang, Yulia Tsvetkov, Orhan Firat, Yuan Cao

Browse the full ICLR paper archive.