End-to-End Document-Grounded Conversation with Encoder-Decoder Pre-Trained Language Model

Our codes are developed based on the DSTC9 Track 1 repository.

Please visit the track repository or the track overview paper for more information about the challenge.

Dataset

Copy the data directory from the DSTC9 Track 1 repository.

Environment

Requires Python version 3.6.

pip install -r requirements.txt

Snippet Filtering

Required before performing the training and evaluation.

./drqa.sh
python snippet_filtering.py

Training

./train.sh

Evaluation

./evaluate.sh