Today, Amazon Web Services announces that SageMaker Hyperpod Flexible Training Plans is available in six new regions: US West (N. California), Asia Pacific (Sydney, Mumbai), Europe (Stockholm, London), and South America (São Paulo). SageMaker Training Plans allows you to gain predictable model training timelines and run training workloads within your budget requirements. You can use SageMaker Training Plans to reserve highly sought-after GPU instances (P4d, P5, P5e, P5en) for Hyperpod for a future date.
SageMaker Training Plans enables you to reserve GPU capacity up to eight weeks in advance for durations up to six months in cluster sizes of 1 to 256 instances, giving you the flexibility to run a broad range of ML workloads. It is ideal for short-duration pre-training and fine-tuning workloads and rapid prototyping. SageMaker automatically provisions the infrastructure and runs the training workloads on these compute resources without requiring any manual intervention.
With this expansion, SageMaker Training Plans is available for the following instance types and AWS Regions: P5 instances in US West (Oregon, N. California), Asia Pacific (Sydney, Melbourne, Mumbai), Europe (London, Stockholm), and South America (São Paulo); P5e instances in US East (N. Virginia), US West (Oregon, N. California), Asia Pacific (Sydney, Mumbai), Europe (London, Stockholm), and South America (São Paulo); P5en instances in US East (N. Virginia), US West (Oregon, N. California), Asia Pacific (Tokyo, Mumbai), and Europe (Stockholm); Trn1 in US East (Ohio, N. Virginia) and US West (Oregon); and P4d in US East (N. Virginia).
To learn more, visit: SageMaker HyperPod , documentation , and the announcement blog .