Reinforcement learning (RL) algorithms involve the deep nesting of highly irregular computation patterns, each of which typically exhibits opportunities for distributed …
Reinforcement learning (RL) algorithms involve the deep nesting of distinct components, where each component typically exhibits opportunities for distributed computation. Current …
Deep reinforcement learning (RL) has led to many recent and groundbreaking advances. However, these advances have often come at the cost of both increased scale in the …
Training machine learning models in parallel is an increasingly important workload. We accelerate distributed parallel training by designing a communication primitive that uses a …
J Fang, G Zhao, H Xu, C Wu… - IEEE/ACM Transactions on …, 2023 - ieeexplore.ieee.org
As the scale of distributed training increases, it brings huge communication overhead in clusters. Some works try to reduce the communication cost through gradient compression or …
AV Clemente, HN Castejón, A Chandra - arXiv preprint arXiv:1705.04862, 2017 - arxiv.org
We propose a novel framework for efficient parallelization of deep reinforcement learning algorithms, enabling these algorithms to learn from multiple actors on a single machine. The …
Abstract Most Deep Reinforcement Learning (Deep RL) algorithms require a prohibitively large number of training samples for learning complex tasks. Many recent works on …
There has been significant progress in developing reinforcement learning (RL) training systems. Past works such as IMPALA, Apex, Seed RL, Sample Factory, and others, aim to …
We present a modern scalable reinforcement learning agent called SEED (Scalable, Efficient Deep-RL). By effectively utilizing modern accelerators, we show that it is not only …