Object Storage Driven Machine Language Workloads - Daniel Valdivia, MinIO
Автор: The Linux Foundation
Загружено: 2023-05-25
Просмотров: 165
Object Storage Driven Machine Language Workloads - Daniel Valdivia, MinIO
As a result of the rapid implementation of Machine Learning (ML) technologies, the industry is undergoing a major shift towards AI-first architectures that leverage object storage. Current best practice is for ML frameworks such as Tensorflow and Kubeflow to rely on the de facto standard of the S3 API to interact with models, data and artifacts stored within object storage. By doing so they unlock the massive scalability and high performance of object storage coupled with immutability and sophisticated object lifecycle features. Models and data can be tiered across Kubernetes-native object storage classes and replicated between datacenters, the edge and clouds so everyone is working with the most up-to-date models and data. In this talk, I will provide an overview of object storage for ML, and then demonstrate deploying open source MinIO and Tensorflow on Kubernetes. You will learn to build and operate a secure data pipeline and automate ML testing and operations.
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
