Hi SmallGiraffe94 ! Dataset.squash doesn't set as parents the ids you specify in dataset_ids . Also, notice that the current behaviour of squash is pulling the files from all the datasetes from a temp folder and re-uploading them. How about creating a new dataset with id1, id2, id3 as parents Dataset.create(..., parent_datasets=[id1, id2, id3]) instead? Would this fit your usecase?