Evaluation Metrics
Metrics will be calculated and displayed for each round of training
When the session is complete, a user will be able to see a set of metrics for all rounds of training, as well as metrics for the final model.
Metrics Displayed
Global Model Federated Loss is a model level metric reported for each round of training. Global Model Federated Loss is a weighted average loss across different silo users, weighted by the number of examples/samples from each silo.
The other metrics seen in the example tables above are all specific to each data silo. In the table, Client IDs represent each data silo contributing to the global model. The metrics displayed on the table are dynamic, and are based on which machine learning task the model is using. See the metrics by machine learning task in the following table:
Last updated
