The Adapter-Bot: All-In-One Controllable Conversational Model

This is the implementation of the paper: The Adapter-Bot: All-In-One Controllable Conversational Model. Zhaojiang Lin, Andrea Madotto, Yejin Bang, Pascale Fung AAAI-DEMO [PDF]

Citation:

If you find this paper and code useful, please cite our paper:

@article{madotto2020adapter,
  title={The Adapter-Bot: All-In-One Controllable Conversational Model},
  author={Madotto, Andrea and Lin, Zhaojiang and Bang, Yejin and Fung, Pascale},
  journal={arXiv preprint arXiv:2008.12579},
  year={2020}
}

Basic Installation

In this repository, we release the trained model, the knowledge retriever, and the interactive script (both via termial and the UI) of the adapter-bot.

Download models

To download the pretrained model run the following commands:

## pip install gdown
import gdown
import zipfile
import os

url = 'https://drive.google.com/uc?id=1JQZex-AD-sa5WUT5U4lIn1K2sPW-us8a/'
output = 'models.zip'
gdown.download(url, output, quiet=False)
with zipfile.ZipFile(output, 'r') as zip_ref:
    zip_ref.extractall()
os.remove(output)

Download and install knowledge retriever (KG and Wiki)

To download and install the knowledge retrievers you can have to follow the step in the retriever folder. Specifically, for the knowledge graph follow the read me at:

https://github.com/HLTCHKUST/adapterbot/tree/main/retriever/graphdb#installing-neo4j

which provides instructions to install neo4j and load opendialoKG. For the wikipedia knowledge, we use DrQA. Also in this case follow the read me at:

https://github.com/HLTCHKUST/adapterbot/tree/main/retriever/doc_ret

which provides a simple script for download the wikidump and train the tf-idf retriever.

Run the interactive script

To interact with the model via command line use the following script:

>>> python interact_adapter.py --interact

HLTCHKUST / adapterbot