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.

RAGmap supports the following features:

• Multiple vector stores
  • ChromaDB
  • LanceDB (NEW!)
• Multiple document formats 📄
  • PDF
  • DOCX
  • PPTX
• Multiple embedding models
  • Amazon Bedrock ⛰️
    • Titan Text Embeddings
    • Titan Multimodal Embeddings
    • Cohere Embed English
    • Cohere Embed Multilingual
  • Hugging Face 🤗
  • OpenAI ֎ (NEW!)
• 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.

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)

1. Install dependencies

   pip install -r requirements.txt

2. Run the application

   # ChromaDB
   streamlit run app.py
   
   # LanceDB (NEW!) 🧪
   streamlit run app_lancedb.py

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

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

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