fjolsvin/chainlit-rag-poc

🚀 Chainlit RAG POC

📖 Description

This project is a Proof of Concept (POC) for a Retrieval-Augmented Generation (RAG) system using Chainlit. It demonstrates how to build a conversational AI application that can search and retrieve information from a vector database to enhance its responses.

✨ Features

• —🤖 Conversational AI interface powered by Chainlit
• —🔍 Retrieval-Augmented Generation (RAG) for enhanced responses
• —🗃️ Flexible database backend (SQLite or Turso)
• —📊 Configurable vector database (in-memory or Qdrant)
• —📄 PDF and text file processing
• —🔐 Secure authentication system

🔧 Configuration

Create a .env file in the root directory and set the following environment variables:

🔑 Required Environment Variables

• —OPENAI_API_KEY: Your OpenAI API key

📊 Vector Database Configuration

• —VECTOR_STORE_TYPE: Type of vector store to use (default: "in_memory")
• —Options: "in_memory", "qdrant"

Qdrant-specific Configuration (if using Qdrant)

• —QDRANT_URL: URL of your Qdrant instance
• —QDRANT_API_KEY: API key for Qdrant
• —QDRANT_COLLECTION_NAME: Name of the Qdrant collection (default: "chainlit-rag-poc")

🗃️ Database Configuration

• —TURSO_DATABASE_URL: URL for Turso database (optional)
• —TURSO_AUTH_TOKEN: Authentication token for Turso (optional)
• —SQLITE_PATH: Path for SQLite database (default: "./local.db")

🔊 Logging Configuration

• —LOG_LEVEL: Logging level (default: "INFO")
• —LOG_TO_FILE: Whether to log to a file (default: "False")
• —LOG_TO_CONSOLE: Whether to log to console (default: "True")
• —LOG_FILE: Path to log file (default: "app.log")
• —LOG_COLORFUL: Whether to use colorful logs (default: "True")

💾 Database and Vector Database Operations

The application's behavior changes based on the environment variables:

Database Operations

• —If both TURSO_DATABASE_URL and TURSO_AUTH_TOKEN are set, the application will use Turso as the database backend.
• —If either of these is not set, it will fall back to using a local SQLite database at the path specified by SQLITE_PATH.

Vector Database Operations

• —If VECTOR_STORE_TYPE is set to "qdrant" and both QDRANT_URL and QDRANT_API_KEY are provided, the application will use Qdrant as the vector store.
• —If VECTOR_STORE_TYPE is "in_memory" or the Qdrant credentials are not provided, it will use an in-memory vector store.

🚀 Running the Application

To run the application, use the following command:

poetry run python -m app

This will start the Chainlit server, and you can access the application through your web browser.

📘 Usage Guide

1. 1.Start the Application: Run the command mentioned above to start the Chainlit server.

1. 1.Access the Web Interface: Open your web browser and navigate to the URL provided in the console (typically http://localhost:8000).

1. 1.Upload a Document: Use the file upload feature to submit a PDF or text document. The system will process and index the document's content.

1. 1.Ask Questions: Once the document is processed, you can start asking questions related to the content of the uploaded document.

1. 1.Interact with the AI: The AI will use the RAG system to provide informed responses based on the uploaded document and its general knowledge.

1. 1.Review Sources: The AI may provide sources or references from the uploaded document to support its answers.

1. 1.Continue the Conversation: You can ask follow-up questions or upload additional documents as needed.

🧩 Project Structure

• —app/: Main application directory
• —__main__.py: Entry point of the application
• —auth/: Authentication-related modules
• —chainlit/: Chainlit-specific modules
• —chat/: Chat-related modules
• —database/: Database-related modules
• —embedding/: Embedding-related modules
• —models/: Database models
• —prompt/: Prompt-related modules
• —rag/: RAG-specific modules
• —text/: Text processing modules
• —vectordb/: Vector database modules

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.