Recent work has shown that fine-tuning large pre-trained language models...
Techniques such as ensembling and distillation promise model quality
imp...
We systematically explore regularizing neural networks by penalizing low...
Recurrent Neural Networks (RNNs) are powerful models for sequential data...
share