Рет қаралды 75
Abstract: This tutorial aims to provide researchers with a comprehensive introduction to the latest reproducible Machine Learning (ML) workflows and tooling provided by the NSF-funded Tapis v3 Application Programming Interface (API) and User Interface (UI). Through hands-on exercises, participants will gain experience in querying pre-trained ML models from public ML infrastructures such as Hugging Face or the ICICLE AI Institute, deploying various ML work-environments, and developing ML workflows on national-scale supercomputing resources. Throughout this tutorial, we will focus on utilizing the Tapis Workflows API, Tapis Pods API, Tapis ML Hub API, and TapisUI. These production-grade services are designed to simplify the facilitation and creation of trustworthy, reproducible, scientific workflows. By abstracting the complexities of underlying technologies behind user-friendly APIs, the Tapis services enable seamless integration with high performance computing (HPC) resources available at institutions with a Tapis deployment. Ultimately the tutorial aims to empower researchers to efficiently develop, deploy, and maintain their own ML workflows.
Authors: Joe Stubbs, Nathan Freeman, Anagha Jamthe, Christian Garcia, and Dhanny Indrakusuma
Intro/outro music by The Known Sea, "Emperor".