Yey! BTW: what the setup you are running it with ? does it include "manual" tasks? Do you also report on completed experiments (not just failed ones)? Do you filter by iteration numbers?