smj0 / GOT

Source code for 《Energy-based Unknown Intent Detection with Data Manipulation》, which is accepted by Findings of ACL, 2021.

GOT

This repository (under construction) is the implementation of 《Energy-based Unknown Intent Detection with Data Manipulation》, which is accepted by Findings of ACL, 2021.

Requirements

Install requirements:

pip install -r requirements.txt

Other preparations:

• Download bert-base-uncased, gpt2-base and put them to the pretrained folder. You can also customize their location by modifying the *.yaml files in the configs folder.

Usage

Unknown intent detection:

bash train.sh

Generate OOD utteranes:

bash locating.sh
bash generating.sh
bash weighting.sh

Citation

@inproceedings{ouyang-etal-2021-energy,
    title = "Energy-based Unknown Intent Detection with Data Manipulation",
    author = "Ouyang, Yawen  and
      Ye, Jiasheng  and
      Chen, Yu  and
      Dai, Xinyu  and
      Huang, Shujian  and
      Chen, Jiajun",
    booktitle = "Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021",
    month = aug,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.findings-acl.252",
    doi = "10.18653/v1/2021.findings-acl.252",
    pages = "2852--2861",
}