***** March 30th, 2020 *****

It is the original PyTorch implementation to this paper: Gate-Enhanced Multi-Domain Dialog State Tracking for Task-Oriented Dialogue Systems

The code is written and tested with PyTorch == 1.1.0.

Our experiments mainly conduct on MultiWOZ 2.1 corpus, and the 2.0 verison is also supported.

• python 3.6

• pytorch >= 1.1.0

• Install python packages:

  • pip install transformers

    (The pretrained BERT Models need to be downloaded here in advance.)

python3 create_data.py 

This Python script inherits from the same-name script in TRADE with minor modifications.

python3 make_dataset.py

python3 run.py

• -dataset: which corpus to train
• -path: model saved path
• -train_m: training gate mode

python3 run_test.py -path=${save_path}

• -dataset: which corpus to train
• -path: model saved path
• -test_m: testing gate mode

Contact: Changhong Yu (CharpYu@bupt.edu.cn)