Awesome

Target

This repository is for our conference paper, now it is anonymous review.

Environment

pip install -r requirements.txt or conda env create -f environment.yml

Usage

python main.py

For version use attention memory module in our model, just verify use_mem_network=True in main.py. For the version with ability to class-incremental relation extraction, just verify fix_labels=False in main.py.

Parameters

We provide default parameters with config-tacred.ini corresponding to TACRED, config-fewrel.ini corresponding to FewRel, change filename to config.ini and add your BERT path and run python main.py to reproduce results.# CRECL