ChatDB augments Large Language Models (LLMs) with SQL databases to serve as symbolic memory, enabling complex multi-hop reasoning. This framework is designed for researchers and developers working on LLM memory augmentation and complex reasoning tasks.

How It Works

ChatDB instantiates a symbolic memory framework using an LLM and SQL databases. The LLM generates SQL queries to interact with the databases, allowing for structured data manipulation and retrieval. This approach leverages the precision and reliability of database systems for complex reasoning, overcoming limitations of purely neural memory mechanisms.

Quick Start & Requirements

• Install dependencies: conda create -n chatdb python=3.9, conda activate chatdb, pip install -r requirements.txt.
• Add API keys: Copy .env.template to .env and set OPENAI_API_KEY and MYSQL_PASSWORD.
• Run demo: python chatdb.py.
• Requires Python 3.9, OpenAI API key, and MySQL credentials.

Highlighted Details

• Augments LLMs with databases as symbolic memory.
• Enables complex multi-hop reasoning.
• LLM generates SQL instructions to manipulate databases.

Maintenance & Community

This is the official repository for the ChatDB paper. The authors state they will continuously add new features. No community links or contributor information are provided in the README.

Licensing & Compatibility

The repository does not explicitly state a license. The code is provided for research purposes, and commercial use implications are not detailed.

Limitations & Caveats

The project is presented as a research artifact with ongoing feature development. The effectiveness is validated on a synthetic dataset, and performance on real-world, diverse datasets is not detailed.