RAGmap 🗺️🔍

Overview

RAGmap is a simple RAG visualization tool for exploring document chunks and queries in embedding space.

Inspired by DeepLearning.ai's short course on Advanced Retrieval for AI with Chroma and Gabriel Chua's award-winning RAGxplorer.

🌐 NEW: Try the live demo hosted on Streamlit Community Cloud.

📢 NEW: RAGmap now supports Hugging Face 🤗 models!

What's inside?

RAGmap supports the following features:

• Multiple document formats 📄
  • PDF
  • DOCX
  • PPTX
• Multiple embedding models
  • Hugging Face 🤗
  • Amazon Bedrock ⛰️
    • Titan Text Embeddings
    • Titan Multimodal Embeddings
    • Cohere Embed English
    • Cohere Embed Multilingual
• Dimensionality reduction (2D and 3D)
  • UMAP
  • t-SNE
  • PCA
• Natural language queries
• Advanced query augmentation
  • Generated Answers (HyDE)
  • Multiple Queries
• ... and more!

☝️⚠️ Important notice: As of January 2024, chromadb's AmazonBedrockEmbeddingFunction only works with Titan models. Feel free to upvote this PR to add support for Cohere Embed models.

How to use

Prerequisites

Enable access to the embedding (Titan Embeddings, Cohere Embed) and text (Anthropic Claude) models via Amazon Bedrock.

For more information on how to request model access, please refer to the Amazon Bedrock User Guide (Set up > Model access)

Option 1 💻

1. Install dependencies

   pip install -r requirements.txt
   

2. Run the application

   streamlit run app.py
   

3. Point your browser to http://localhost:8501

Option 2 🐳

1. Run the following command to start the application

   docker-compose up
   

2. Once the service is up and running, head over to http://localhost:8501

Example: Amazon shareholder letters

References

• (AWS) What is Retrieval-Augmented Generation?
• (DeepLearning.ai) Advanced Retrieval for AI with Chroma