English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep reinforcement learning (DRL) has revolutionized learning and actuationin applications such as game playing and robotic control. The cost of datacollection, i.e., generating transitions from agent-environment interactions,remains a major challenge for wider DRL adoption in complex real-worldproblems. Following a cloud-native paradigm to train DRL agents on a GPU cloudplatform is a promising solution. In this paper, we present a scalable andelastic library ElegantRL-podracer for cloud-native deep reinforcementlearning, which efficiently supports millions of GPU cores to carry outmassively parallel training at multiple levels. At a high-level,ElegantRL-podracer employs a tournament-based ensemble scheme to orchestratethe training process on hundreds or even thousands of GPUs, scheduling theinteractions between a leaderboard and a training pool with hundreds of pods.At a low-level, each pod simulates agent-environment interactions in parallelby fully utilizing nearly 7,000 GPU CUDA cores in a single GPU. OurElegantRL-podracer library features high scalability, elasticity andaccessibility by following the development principles of containerization,microservices and MLOps. Using an NVIDIA DGX SuperPOD cloud, we conductextensive experiments on various tasks in locomotion and stock trading and showthat ElegantRL-podracer substantially outperforms RLlib. Our codes areavailable on GitHub.

ElegantRL-Podracer: Scalable and Elastic Library for Cloud-Native Deep Reinforcement Learning