Multi-domain Multilingual Question Answering
Sebastian Ruder
DECEMBER 6, 2021
Instead, strategies from continual learning such as L2 regularization ( Xu et al., 2008 ; Lin & Kuo, 2010 ; Shima & Mitamura, 2010 ). However, naive fine-tuning generally leads to deterioration of performance in the source domain, which may be undesirable.
52 
Let's personalize your content