Stratify cross fitting?

[quinn-ca]

I'm interested in using DoubleML for my project, where I have a binary outcome with a proportionally rare response (85% 0, 15% 1). Is there a way I can stratify the cross fitting, so I ensure the rare outcome is represented in each of the folds? I'm using the R implementation. I've read through the User Guide and didn't see this mentioned, apologies if I missed it. I appreciate this awesome package and the thorough documentation.