Hi @<1826066729852211200:profile|DullSwallow71> , I would suggest looking at it from another perspective. Check machine availability and only then push a job into a queue. You can see all the related information to usage in the 'Workers' screen in the 'Orchestration' tab.

Then you can either push jobs manually according to usage or write your own service to sample usage and push jobs accordingly.

Thanks for the reply, but I guess nothing ever came from it, right?

That's a bummer. I guess I have to hack something together myself then.

Thanks @<1523701070390366208:profile|CostlyOstrich36> for your suggestion. However I would like to automate this as much as possible and let it run outside of my working, so checking it manually would be my least favorite option.
The idea behind this would be to use the "downtime" of our machines to check for better hyperparameters, but because this hasn't the highest priority we would not kill any other process just for this.

I think your idea is related to issue 71 on GitHub . As far as I undestand it, it is not very straightforward to do it currently.