Empowering privacy-preserving LLM training with
secure multi-party computation.
Leverages GPU Acceleration for a Fast Training
Users can easily configure parameters to optimize performance and security
Ensures correct aggregation in the event of dropouts
Enables the communications of models exceeding the 2GB limit set by gRPC
Employs secure multi-party computation techniques for secure aggregation
Multiple servers collectively aggregate as opposed to a central aggregator
In 2023, genetic testing giant 23andMe suffered a massive breach,...
May 27, 2025
Business leaders are wary. They urgently want to reap the...
May 27, 2025
Join us on our journey to make federated
approaches available to everyone.
