Hi, I noticed that the random seed does not affect the splits of the datasets using the scaffold splitting method. Thanks!