Train a ML-agent with # RND

Hey, does anyone know RND? Is it included in the latest version of Unity ML-Agents?

I find the introduction here, but no more information about how to achieve it.

You can enable it in the training conf files