Due to people’s emerging concern about data privacy, federated learning(FL) is currently being widely used. Conventional federated learning uses a highly centralized architecture, but in a real federated learning scenario, due to the highly distributed of data nodes and the existence of malicious data nodes, It is of great challenges for conventional federated learning to improve the utilization of network bandwidth and maintained the security and robustness of federated learning under malicious node attacks. As a result, we propose a practical model training paradigm based on Blockchain, named GalaxyLearning, which aims to train a model with distributed data and to reserve the data ownership for their owners and the interest of trained model.