Running Pytorch on CHTC

Sign up for our training!

Curious about running Pytorch workflows on CHTC systems? Join our training!

Topics covered:

• GPU machines: open capacity, researcher-owned machines, and backfill capacity
• Build your Pytorch software stack in a container
• Submit training and inference jobs
• Best practices for handling data
• Strategies for getting more throughput and common issues

Prerequisites

• Familiarity with submitting jobs with HTCondor
• Basic machine learning concepts
• A CHTC account

Note

This training does not teach in-depth Pytorch concepts! This training assumes that the user has some experience running Pytorch workflows on a local workstation or other cluster computing system and will be focused mainly on how best to run the workflow on CHTC's HTC system.