FLock.io

FLock.io supports decentralized training and fine-tuning of AI models through federated learning without making the original data publicly available. Participants stake tokens to train and validate models in AI Arena, where consensus is formed from weighted evaluations.

In FL Alliance, models are fine-tuned locally on user data and their updates are aggregated with zero-knowledge proofs without exposing the underlying information. The project has open-source code on GitHub and supports fine-tuning for widely used large language models.

It is a core AI protocol focused on secure, decentralized model training for enterprises and communities.