FATE-Serving is a high-performance, industrialized serving system for federated learning models, designed for production environments.
- High performance online Federated Learning algorithms.
- Real-time inference using federated learning models.
- Support parallel inference between guest and host.
- Support parallel computing in a inference request.
- Provide service managerment for grpc interface by using zookeeper as registry.
- Visualization tools are provided for cluster management and model management.