[Adapt] [Seminar] Commonsense Resources and Application

[任思宇]

Hi Adapters,


Despite the unprecedentedly rocketing model performance on a wide spectrum of NLP tasks, we still do not understand a lot about the inner workings of these models. Moreover, recent works have shown that these deep neural models(e.g., BERT, RoBERTa, etc) are vulnerable to adversarial attacks and exhibit a deficiency of basic commonsense knowledge and reasoning skills to solve tasks that are easy to human. In this talk, I will introduce some widely used commonsense resources and examples of how researchers typically integrate it into neural models.


Time: Wed 4:00pm
Venue: SEIEE 3-414

See you there!
Best Regards,
Roy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20201013/eabc0f9e/attachment.html>