richardbaihe / paperreading

NLP papers
MIT License
2 stars 0 forks source link

Arixv 2021 | Pitfalls of Static Language Modelling #60

Closed richardbaihe closed 3 years ago

richardbaihe commented 3 years ago

This paper proposes to test LM's ability for future text modeling.

For example, training with 2016's news data and testing on 2017' news data.

The main contribution I think is the problem they investigate and how to address such a problem.