Within Federated Learning, what remains on the user's device during model training?

Within Federated Learning, the raw training data remains on the user's device during model training. Federated Learning is a machine learning technique that allows a global model to be trained across multiple decentralized devices or servers holding local data samples, without exchanging them. Instead of sending the raw data to a central server, each device trains a local model on its own data and then sends only the updated model parameters (e.g., weights and biases) to the central server. The central server aggregates these updates to create a better global model. This protects user privacy because the sensitive raw data never leaves the device.