Hanscal / DCKS

The official implementation for the paper Improving Empathetic Dialogue Generation by Dynamically Infusing Commonsense Knowledge.

DCKS

The official implementation for the paper Improving Empathetic Dialogue Generation by Dynamically Infusing Commonsense Knowledge.

Usage

Dependencies

Install the required libraries (Python 3.8.5 | CUDA 10.2)

pip install -r requirements.txt 

Dataset

The preprocessed dataset is already provided, you can download the data from Baidu Netdisk (passcode: 4b8l) , and then put it in the folder /data/ED/dataset.
If you want to create the dataset yourself, delete this file, download the COMET checkpoint and place it in /data. The preprocessed dataset would be generated after the training script.

Training

bash run_train.sh

where data_ratio can be used for low resource ablation studies.

Evaluation

bash run_evaluate.sh

Citation

If you find our work useful for your research, please kindly cite our paper as follows:

@article{DCKS2023,
      title={Improving Empathetic Dialogue Generation by Dynamically Infusing Commonsense Knowledge}, 
      author={Hua Cai, Xuli Shen, Qing Xu, Weilin Shen, Xiaomei Wang, Weifeng Ge, Xiaoqing Zheng, Xiangyang Xue},
      year={2023},
}