rAlexandre

Cross-Silo Horizontal Loading all datasets to all clientes

6

From what I can see, on every example for cross-silo FL, the datasets of all clients are loaded into memory. For example [here](https://github.com/FedML-AI/FedML/blob/master/python/examples/cross_silo/grpc_fedavg_mnist_lr_example/custom_data_and_model/torch_client.py). It would be nice to add a...

Closes #611

[FedML](https://github.com/FedML-AI/FedML)/[python](https://github.com/FedML-AI/FedML/tree/master/python)/[fedml](https://github.com/FedML-AI/FedML/tree/master/python/fedml)/[__init__.py](https://github.com/FedML-AI/FedML/blob/956639046be40ba45e6a117273a3e51f117961ce/python/fedml/__init__.py) The server (rank 0) should be getting his client list derivated from client_num_in_totals and not only the clients_per_round. This makes the clients chosen for each round always the same...