> Leaderboard

Hi all,

We are having a small set of technical difficulties with the site that is supposed to be able to automate submissions. We are still running your submissions and have generated a leaderboard here: https://medici-codalab-main.eastus.cloudapp.azure.com/competitions/24#results

Stay tuned for updates, but so far initial submissions are running well!

Posted by: challenge-organizer @ June 22, 2022, 5:45 p.m.

FYI, we are only evaluating "best_FL_global_model.pt". This model is selected based on the best validation scores of each client using IntimeModelSelectionHandler configured here https://github.com/Project-MONAI/tutorials/blob/3848027d8dee0b0688aa0e61e01fdd434d57e06c/federated_learning/breast_density_challenge/code/configs/mammo_fedavg/config/config_fed_server.json#L41

For the current selector to work, you need to run at least 2 rounds of FL (set num_round=2 in config_fed_server.json). Of course, feel free to implement your own model selection but the file "best_FL_global_model.pt" needs to be generated on the server in order for it to be evaluated during testing.

Posted by: hroth @ June 22, 2022, 10:43 p.m.
Post in this thread