This only could affect claim-1 results and none of those results are messed up in this way bc we didnt use these values, but we should make the following printouts and values also use min:
analyze first_train_cost a bit, it's weird shouldnt it min within a task like the other?
note that current compression_ratio stuff is also not task aware
original_cost also uses the non-min cost
likewise with the integration_tests version of origimal_cost
This only could affect claim-1 results and none of those results are messed up in this way bc we didnt use these values, but we should make the following printouts and values also use min:
first_train_costa bit, it's weird shouldnt it min within a task like the other?original_costalso uses the non-min cost