RST Discourse Parsing with Second-Stage EDU-Level Pre-training

Implementation of the paper RST Discourse Parsing with Second-Stage EDU-Level Pre-training, ACL'22

Requirements

PyTorch 1.7.0

Python 3.6

Transformers 3.5.0

Data

RST-DT

GUM

PLM

XLNET+NEP+DMP

Training

python -u driver/TrainTest.py --config_file config.rst.xlnet.nep.dmp

Testing

We provide our trained model for testing.

python -u driver/Test.py --config_file rst_model/config.cfg --test_file experiments/rst/sample.txt