Hello team!
I would like to know if the private test set has been impacted by the task 1 public data leak. Removing the overlap between task 2 public test set and task 1 public train set changed the apriori target distribution, increasing the number of exact labels from 66% to 74%.