A virtual assistant based on the RAG architecture. This project implements the question-answering part of the architecture, utilizing a prepopulated DB. The DB used is Qdrant. And the Azure OpenAI api is used to query an LLM.
The project is written in Python version 3.10. To install the requirements please run:
pip install -r requirements.txt
The project uses a Qdrant DB to store the embeddings. The DB is set up in a docker container and port 6334 is open to access it. The docker-compose.yml is provided. To start the DB run in your terminal:
docker-compose up -d
This DB is initially empty. So, if this is the first time we need to add data.
To fill it with data, please run the python script store_pdfs.py. It reads pdf files from a user-provided filepath.
The necessary parametrization can be provided using a conf.yml file in the same directory as your working environment. The conf-example.yml file is provided as an example.
To start the virtual assistant run the app.py script.
As previously, the necessary parametrization can be provided using a conf.yml file in the same directory as your working environment. The conf-example.yml file is provided as an example.
The app runs in localhost:5000. And the UI looks like the following image. You can start interacting with the virtual assistant! The history is also printed in the screen.
This project is licensed under the MIT License - see the LICENSE file for details.