Closed krfricke closed 3 years ago
Closes #12, concerns #81
Currently we don't check if we assign too many actors to too few data shards. This PR introduces a check that will raise an error if we assign too many workers to too few data sources, which would leave some actors without any data to train on.
Closes #12, concerns #81
Currently we don't check if we assign too many actors to too few data shards. This PR introduces a check that will raise an error if we assign too many workers to too few data sources, which would leave some actors without any data to train on.