Hello,
Thanks for this amazing project that aims at solving the bottleneck of environnement interactions in DRL.
I have read the paper several times and I am starting to make the code work on a GPU cloud plateform.
But unfortunately, in the process of connecting the concepts of the paper and the code of the repo, I cannot find, in the latter, these differents pieces:
- how and where Kubernetes is used ?
- where is the Orchestrator code ?
- where is the Leaderboard ?
The only elements I could find so far are the Pod elements; the worker, learner and evaluator.
Thanks for your attention and help.
Aymeric