Ray RLlib is a production-grade reinforcement learning library. Its main advantages are robust code base and massive scalability. On the other hand, it's less research-oriented than OpenAI baselines or Stable baselines.
This 1st step will bring following features:
demonstration of Ray usage with local training
update code base to support the change
split requirements.txt (distinct dependencies for baselines and ray)
Ray RLlib is a production-grade reinforcement learning library. Its main advantages are robust code base and massive scalability. On the other hand, it's less research-oriented than OpenAI baselines or Stable baselines.
This 1st step will bring following features:
Covered later: