Amazon SageMaker AI in Amazon Web Services in China
Amazon SageMaker AI is a fully managed machine learning service. With SageMaker AI, data scientists and developers can quickly build and train machine learning models and then deploy them into a production-ready hosted environment.
Region availability
Amazon SageMaker AI is available in the following Amazon Web Services Regions in China:
-
Beijing Region
-
Ningxia Region
How SageMaker AI differs
The following differences apply to Amazon SageMaker AI:
-
Amazon Augmented AI (Amazon A2I) is not available.
-
Amazon SageMaker Canvas is not available.
-
Amazon SageMaker Edge Manager is not available.
-
Amazon SageMaker Elastic Inference is not available.
-
Amazon SageMaker geospatial capabilities are not available.
-
Amazon SageMaker Ground Truth is not available.
-
Amazon SageMaker Ground Truth Plus is not available.
-
Amazon SageMaker Inference Recommender is not available.
-
Amazon SageMaker model registry and projects are not available.
-
Reinforcement learning is not available.
-
Amazon SageMaker Serverless Inference is not available.
-
Amazon SageMaker Studio, the new integrated development environment (IDE) for machine learning in SageMaker AI, is not available. However, the previous Amazon SageMaker Studio experience, now named Amazon SageMaker Studio Classic, is available.
-
Amazon SageMaker Studio Lab is not available.
-
Amazon SageMaker Training Compiler is not available.
The following SageMaker AI features are limited in the Beijing and Ningxia Regions:
-
Amazon SageMaker Domain: On-boarding with Amazon SageMaker Domain is limited to Amazon Identity and Access Management (IAM).
-
Amazon SageMaker Autopilot: Configuring the default parameters of a SageMaker Autopilot experiment (for administrators) is not available.
-
Amazon SageMaker JumpStart is only supported in Beijing Region.
-
The following features are not available for Amazon SageMaker Inference.
-
Multi-model endpoints on GPU instances
-
Graviton-based, Inferentia-based, p4d, and g5 instance types
-
Increased shared memory size to almost half of instance memory
-
Deployment guardrails
-
EndpointConfigparametersContainerStartupHealthCheckTimeoutInSeconds,ModelDataDownloadTimeoutInSeconds, andVolumeSizeInGB
-
-
Amazon Web Services Marketplace China: You can’t create and list your SageMaker AI artifacts in Amazon Web Services Marketplace China.