How is this different from argparser btw?

Not different, just a dedicated section 🙂 Maybe we should do that automatically, the only "downside" is you will have to name the Dataset when getting it (so it will have an entry name in the Dataset section), wdyt ?