RAGxplorer is an open-source tool designed for building visualisations of Retrieval Augmented Generation (RAG) systems. It targets developers and researchers working with RAG pipelines, offering a way to explore and understand the retrieval process.

How It Works

RAGxplorer leverages embedding models to represent text data and facilitates querying these embeddings. The core approach involves loading documents (e.g., PDFs), generating embeddings, and then visualizing the relationships between queries and retrieved documents. This allows users to gain insights into the retrieval effectiveness of their RAG systems.

Quick Start & Requirements

• Install: pip install ragxplorer
• Prerequisites: Requires an embedding model (e.g., thenlper/gte-large).
• Resources: A Jupyter notebook tutorial is available at https://github.com/gabrielchua/RAGxplorer/blob/main/tutorials/quickstart.ipynb, and a Streamlit demo is hosted at https://ragxplorer.streamlit.app/.

Highlighted Details

• Enables visualization of RAG system performance.
• Supports loading documents like PDFs.
• Integrates with embedding models for semantic understanding.
• Provides a Streamlit-based demo for immediate exploration.

Maintenance & Community

The project acknowledges inspiration from DeepLearning.AI and Chroma. Contributions are welcomed, with contributing guidelines noted as "WIP" (Work In Progress).

Licensing & Compatibility

Licensed under the MIT license, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

Contributing guidelines are still under development. The project is primarily demonstrated via a Streamlit app and a Jupyter notebook, suggesting a focus on interactive exploration rather than a standalone production-ready application.