公开(公告)号：US20230237352A1

公开(公告)日：2023-07-27

申请号：US17581782

申请日：2022-01-21

Applicant: salesforce.com, inc.

Inventor： Tian Lan ,  Stephan Tao Zheng ,  Sunil Srinivasa

IPC: G06N5/04 ,  G06F9/54 ,  G06N20/00 ,  G06F9/50

CPC classification number: G06N5/043 ,  G06F9/545 ,  G06N20/00 ,  G06F9/5072

Abstract: Embodiments provide a fast multi-agent reinforcement learning (RL) pipeline that runs the full RL workflow end-to-end on a single GPU, using a single store of data for simulation roll-outs, inference, and training. Specifically, simulations and agents in each simulation are run in tandem, taking advantage of the parallel capabilities of the GPU. This way, the costly GPU-CPU communication and copying is significantly reduced, and simulation sampling and learning rates are in turn improved. In this way, a large number of simulations may be concurrently run on the GPU, thus largely improving efficiency of the RL training.