Published August 1, 2021 | Version 1.0
Dataset Open

BERTGen: Multi-task Generation through BERT

Creators

  • 1. Imperial College London

Description

Supplements for the paper entitled "BERTGen: Multi-task Generation through BERT", accepted to appear at ACL-IJCNLP 2021 conference. Further instructions on how to use these resources are explained at https://github.com/ImperialNLP/BertGen

This entry contains:

  • The datasets used to train and test the BertGen model, packed into a particular format required by the codebase
  • The best model checkpoint used to generate the results in the paper

 

Files

Files (12.5 GB)

Name Size Download all
md5:982b3f33bb3ab33ec7dd7191cdb37645
10.3 GB Download
md5:fae1365f322b2bc8047f554c46e6c79a
2.2 GB Download

Additional details

Funding

MultiMT – Multi-modal Context Modelling for Machine Translation 678017
European Commission