This system contains 2 components:

• Retriever: retrieve the relevant document from the input question.
• Reader: extract the answer from the pair of question and retrieved document.

conda create -n openqa python=3.8
conda activate openqa
pip install -r requirements.txt

Note: please check file requirements.txt to install pyserini library in right way. Or check the official homepage of this package.

# these commands is ran from root folder of this project.

# if you want to train Reader locally
git submodule init 
git submodule update 

# run server
python src/server.py

Note: when you run this command, it will download a large corpus (e.g. Wikipedia indexed corpus) size 40 - 50 GB.

Access to your localhost web-server https://localhost:7009/ to enjoy the result.

MIT-licensed.