This repository offers basic Flask, Streamlit, and Docker examples for the LlamaIndex package, targeting developers needing to quickly build Proofs-of-Concept (POCs) for LLM-powered applications. It simplifies the initial setup and demonstration of LlamaIndex's capabilities.

How It Works

The examples showcase LlamaIndex's core functionality for indexing and querying text data. The Flask example uses a React frontend to interact with a Python API, demonstrating document upload and querying via dedicated endpoints. The Streamlit examples provide user-friendly interfaces for querying indexed data, including a SQL sandbox for Text2SQL capabilities and a term/definition extraction tool. The architecture separates the index management from the API for thread safety in the Flask example.

Quick Start & Requirements

• Install: conda create --name llama_index python=3.11 followed by pip install -r requirements.txt.
• Prerequisites: Python 3.11.
• Docker: Each example includes a Dockerfile for building images (approx. 600MB-900MB).
• Links: Paul Graham Essay (example data).

Highlighted Details

• Flask example includes separate API and React frontend services.
• Streamlit examples demonstrate SQL querying and agent-based chat history.
• Supports text extraction from documents and image uploads for term/definition building.
• Dockerfiles are provided for containerized deployment.

Maintenance & Community

• Contributions and suggestions are welcomed via Pull Requests.

Licensing & Compatibility

• The repository does not explicitly state a license in the README.

Limitations & Caveats

The Flask example's index management uses locks for multithreading, which might require careful handling in production environments. The provided examples are basic and intended for POCs, not production-ready applications.