RAG Application Demo скачать в хорошем качестве

RAG Application Demo

This project is a high-performance, containerized RAG application designed for real-time document intelligence. It features a decoupled architecture where a FastAPI producer queues ingestion tasks in Redis, allowing a background Worker to handle heavy PDF parsing and vectorization in ChromaDB without lagging the UI. Powered by Gemma 3:4b and Ollama, the system delivers token-by-token streaming responses with source citations directly to a Chainlit interface. It is optimized for local deployment, requiring approximately 4.0 GiB of memory to provide private, low-latency AI assistance.