DAMO-ConvAI/pcll

PCLL

Introduction

PCLL (Prompt-Conditioned Lifelong Learning) is build by Conversational AI Team, Alibaba DAMO Academy.

The corresponding paper has been published at EMNLP 2022 main conference: "Prompt Conditioned VAE: Enhancing Generative Replay for Lifelong Learning in Task-Oriented Dialogue".

PCLL Implementation

Requirements

pip install -r requirements.txt

Dataset

Download the datasets from the Google Drive. Put the required datasets for intent detection and slot filling tasks under the folder DATA. The datasets process scripts are also contained DATA. You can read the files for more details.

Model Training and Evaluation

sh scripts/intent_all_train.sh # for lifelong intent detection task
sh scripts/slot_all_train.sh # for lifelong slot filling task

The files required for training are lltrain.py, mycvae/model.py, mycvae/trainer.py

Citation

If you use our code or find PCLL useful for your work, please cite our paper:

@inproceedings{zhao2022cvae,
  title={Prompt Conditioned VAE: Enhancing Generative Replay for Lifelong Learning in Task-Oriented Dialogue},
  author={Zhao, Yingxiu and Zheng, Yinhe and Tian, Zhiliang and Gao, Chang and Yu, Bowen and Yu, Haiyang and Li, Yongbin and Sun, Jian and Zhang, Nevin L.},
  booktitle={Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing},
  year={2022},
}