QA-engine

QA-engine is a question-answering service that wraps multiple QA models.

Quick start

Requirements

Python 3
Python 2.7 (qa-bert dependency)

Cloning the project

Clone the repository with

git clone https://github.com/TrentoCrowdAI/qa-engine/

Create python virtual environment

virtualenv -p python3 venv3.7

Configure the project

First of all create your config.json file. Reference: TODO:link Copy and paste the config.example.json in the root of the project.

cp config.example.json config.json

You can choose to configure the project automatically or manually, as you want.

Automatically

Start the script "service_start.sh" with:

bash service_start.sh

All the requirements will be downloaded and installed. If all goes well, the service API service is started on port 80.

Manually

If you want to have full control on the preparation of the environment you have to follow these steps:

1. Enter in the python virtual environment source venv3.7/bin/activate
2. Download python dependencies pip install -r requirements_python2.txt pip3 install -r requirements_python3.txt
3. Download QA-models resources bash prepare_environment.sh

Start the QA-engine

For testing purposes you can launch: python3 app.py The API service will listen on port 5000.

Otherwise, in production environment you can run the service in background with: bash service_start.sh The API service will listen on port 80.

Read the WIKI

QA-Engine Wiki