Nuggt is an open-source autonomous LLM agent designed to democratize powerful AI task automation using smaller, quantized models. It targets developers and researchers looking to build controlled, efficient bots capable of integrating with various tools and services, benefiting from a community-driven development approach.

How It Works

Nuggt leverages a 4-bit quantized version of the Wizcoder-15B model as its core LLM. This approach allows for efficient execution on more accessible hardware while maintaining significant capabilities. The agent is designed as an all-in-one platform, enabling users to define complex functionalities through single instruction sets and integrate seamlessly with tools like Python, document processors, video analysis, Google Search, and web browsing.

Quick Start & Requirements

• Install: pip install -r requirements.txt
• Run: python3 server.py and streamlit run nuggt.py
• Prerequisites: Requires Wizcoder-15B 4-bit Quantised model (download from Hugging Face). API keys need to be configured.
• Documentation: How to use Nuggt WizardCoder Version

Highlighted Details

• Powered by Wizcoder-15B (4-bit Quantised).
• Supports integration with Python, document processing, video analysis, Google Search, and web browsing.
• Community-driven development with active feedback channels.
• Demo available showcasing RandomForest implementation on the Titanic dataset.

Maintenance & Community

• Active community engagement via Discord Server and Twitter.
• Contribution guidelines are provided for community involvement.

Licensing & Compatibility

• The README does not explicitly state a license. This requires clarification for commercial use or closed-source linking.

Limitations & Caveats

As of June 9th, 2023, Nuggt is an experimental project and not fully developed. Thorough testing is strongly recommended before deployment due to potential security and functionality concerns. The project disclaims all warranties and is not suitable for production or critical systems without extensive evaluation.