Is there a recommended infrastructure pattern for RLLib on Unity?


I have questions about training agents on Unity environments using RLLib. I’m wondering: what is the recommended architecture for infra? Are there documents and best practices to share? In particular:

should agent training and Unit data gen run on the same machine? container? Is it more accepted to use separate instances for Unity and agent training?

Curious about documentation on this

@sven1977 could you take a look?