/Experiments-on-DF-Net-and-GPT-3

Code for "DAISY: An Implementation of Five Core Principles for Transparent and Accountable Conversational AI" 2022. Fork from ACL 2020 Paper "Dynamic Fusion Network for Multi-Domain End-to-end Task-Oriented Dialog"

Primary LanguagePython

Experiments on deep neural networks for task-oriented conversational agents

This fork is based on the work Dynamic Fusion Network for Multi-Domain End-to-end Task-Oriented Dialog by Qin et al. 2020. Please refer to the original repository for information on that paper. I used this codebase to carry out further experiments for our paper:

@article{WahdeVirgolin2022DAISY,
  author = {Mattias Wahde and Marco Virgolin},
  title = {DAISY: An Implementation of Five Core Principles for Transparent and Accountable Conversational AI},
  journal = {International Journal of Human–Computer Interaction},
  volume = {0},
  number = {0},
  pages = {1-18},
  year  = {2022},
  publisher = {Taylor & Francis},
  doi = {10.1080/10447318.2022.2081762},
  URL = {https://doi.org/10.1080/10447318.2022.2081762},
  eprint = {https://doi.org/10.1080/10447318.2022.2081762}
}

Changes included here

Changes include:

  • Modifications to generate and use a smaller version of the data sets considered in the paper (SMD aka KVR, and MultiWOZ 2.1), see generate_our_test.py and files referenced there.
  • Minor changes to the original code base to test DF-Net on such data sets.
  • Code to query GPT-3 on those data sets, using OpenAI's API.