Show HN:Basic Distributed AI Train Tool

I built a Python-based distributed AI training system that makes collaborative model training accessible to everyone. It automatically splits datasets and models among participants based on their performance scores, allowing stronger GPUs to handle more data while lighter machines still contribute. Communication is done via simple socket connections, supporting both LAN and wider networks. Models and vocabularies are merged automatically to improve continuously. Open source and easy to integrate, it supports both P2P and centralized server setups. In short, it’s a practical framework for fair, dynamic, and scalable distributed AI training. Check out the details and code on GitHub.