# Renumics RAG

A demo retrieval-augmented-generation assistant built with LangChain and Streamlit, intended to explore and visualize the data behind a RAG system rather than serve as a production chatbot.

Renumics RAG is a reference implementation of a retrieval-augmented-generation assistant, combining LangChain for the retrieval and generation pipeline with a Streamlit interface. It installs as an editable Python package with GPU or CPU-specific extras for its embedding and machine-learning dependencies, such as sentence-transformers and PyTorch, and supports pip or uv for setting up its environment.

It can be configured to use OpenAI models directly or through Azure OpenAI, or Hugging Face models, by setting the relevant API keys and endpoint values in an environment file, and is intended to be cloned and edited directly for anyone wanting to modify rather than just run the demo.

The project's emphasis, per its name, is on exploring and visualizing the data a RAG pipeline retrieves and generates from, rather than only answering questions.

Website: https://github.com/Renumics/renumics-rag