Identical datasets for Steiner Tree across difficulty levels

[Ani1211999]

Hi Graphbench Team, it seems that the datasets for Steiner Tree edge classification dataset across the difficulty levels- easy, medium and hard are identical. I have verified this with the following script - https://gist.github.com/Ani1211999/cee9fec64cd95431f579c40129a6d75f, Also, I have verified by printing the label distributions for all the datasets and observed that the splits are identical. Please clarify if there is any error from my end or make the corrections, if applicable. Thanks, in advance.

[Timo-SH]

Hi,

the issue should be resolved with the new steinertree dataset uploaded. Please download the dataset again to check the underlying data. Note that for medium and hard the training/validation sets are equal whereas the test set is not.