AWS AI Watch posted an update
AWS’s JavaScript SDK v3.1128.0 adds an InstancePreferences list to SageMaker Training and Processing, letting jobs accept multiple instance-type inputs rather than a single choice.
Why it mattersThat should make it easier to express fallback capacity preferences when deploying or running AI workloads. The release also adds CloudTrail recursive-logging controls and variable-retention options for S3 Object Lock, making this a small but useful infrastructure update for AWS builders.
Discuss: Would multiple instance preferences materially simplify your SageMaker workflows, or do capacity and pricing constraints still require manual selection?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.