EU AI Law Chatbot 🇪🇺🤖🧑‍⚖️

Intro

This repo implements a RAG pipeline using AWS bedrock llm models, gradio for the frontend, and AWS lambda for serveless hosting. The vector database indexes the recently approved EU AI law in the English and Spanish version using the Titan Text Embeddings V2 model, whereas the chat model is Antrophic's Claude 3 Haiku.

Start Locally

• To run it locally, create a virtual environment, activate and, navigate to rag_app to run pip install -e ..
• Create the following directories rag_app/data/input_data and rag_app/data/vector_db.
• Put the PDFs in rag_app/data/input_data and then run python rag_app/populate_database.py which will put the vector embeddings in rag_app/data/vector_db.
• To create the docker image, from rag_app run docker build -t rag_exp:latest . , then docker build -t rag_exp:latest . to test it.

Deploy to AWS Lambda

WIP