/DMBERT

A temporary repo to share the DMBERT code for Event Detection

Primary LanguagePythonMIT LicenseMIT

DMBERT

A temporary repo to share the DMBERT code for Event Detection

This repo includes a recent implementation of the DMBERT model in the NAACL work with the support of the newest transformers library.

Just a temporary repo, not well-documented, will be merged into some to-be-announced works in the future.

Most of the codes are changed from the examples in transformers, you can refer to that repo for most of your questions.

Requirments

transformers

torch >= 1.2.0

Usage

To run this code, you need to:

  • preprocess your ACE2005 dataset as the json format specified in the HMEAE repo
  • change the parameters in ```run.sh`` as your environment needs and run it

Cite

If the codes help you, please cite our paper:

Adversarial Training for Weakly Supervised Event Detection. Xiaozhi Wang, Xu Han, Zhiyuan Liu, Maosong Sun, Peng Li. NAACL-HLT 2019.