arXiv (Cornell University) · 2017 · 331 citations · 0 references
ComputingComputer ScienceArtificial IntelligenceMachine LearningMarkov Decision ProcessesDistributed AlgorithmsDistributed ComputingPolicy ImprovementPolicy LearningReinforcement LearningReinforcement Learning FoundationsReinforcement Learning Theory
Reinforcement learning algorithms involve deeply nested irregular computation patterns that offer opportunities for distributed computation, and RLlib is available at https://rllib.io/. The authors propose distributing RL components in a composable way by adapting algorithms for top‑down hierarchical control. This approach encapsulates parallelism and resource requirements within short‑running compute tasks. RLlib demonstrates that scalable software primitives for RL enable a broad range of algorithms to be implemented with high performance, scalability, and substantial code reuse.
Reinforcement learning (RL) algorithms involve the deep nesting of highly irregular computation patterns, each of which typically exhibits opportunities for distributed computation. We argue for distributing RL components in a composable way by adapting algorithms for top-down hierarchical control, thereby encapsulating parallelism and resource requirements within short-running compute tasks. We demonstrate the benefits of this principle through RLlib: a library that provides scalable software primitives for RL. These primitives enable a broad range of algorithms to be implemented with high performance, scalability, and substantial code reuse. RLlib is available at https://rllib.io/.