Description: We present an overview and tutorial for utilizing DoD HPC systems for training deep learning and deep reinforcement learning models across multiple GPU or CPU nodes. We demonstrate how these goals can be achieved using the PyTorch Distributed Data Parallel (DDP) Module and the Ray Framework for building and running distributed applications.
Presenter(s): Jamison Moody and Tanner Norton (Brigham Young University)
Location: Webcast
Date & Time: August 7, 2020, 12:00p ET
Additional Notes: Jamison Moody and Tanner Norton are working on Computer Science Master’s degrees at Brigham Young University and are participating as Wright State University contractors in the AFRL Autonomous Technology Research Center (ATRC) Summer Internship program under the direction of Dr. Oliver Nina from AFRL/RYAT.
Distribution Statement C. Distribution authorized to U.S. Government Agencies and their contractors, administrative or operational use, 7 August 2020. Other requests for this document shall be referred to AFRL/RYA, 2241 Avionics Circle, Wright-Patterson AFB, OH 45433.