The purpose of this demo is to combine the processing of PDF files, embedding generation, multiple retrieval metrics and a user interface with streamlit that also consider agent memory.
pip install -r requirements.txt
Create an account and a Vector DB at (astra.datastax.com).
Define the AstraDB credentials and Open AI API Key in the .env file.
Copy .env.sample to .env
streamlit run app.py
I uploaded and converted PDF using the notebook "Explicando Retrieval Augmented Generation.ipynb".