Skip to content
GitHubXDiscordRSS

ModelPackage

Learn how to create, update, and manage AWS SageMaker ModelPackages using Alchemy Cloud Control.

The ModelPackage resource lets you create and manage AWS SageMaker ModelPackages which are used to package machine learning models for deployment and sharing.

Create a basic ModelPackage with required properties and a common optional property.

import AWS from "alchemy/aws/control";
const modelPackage = await AWS.SageMaker.ModelPackage("basicModelPackage", {
ModelPackageName: "MyFirstModelPackage",
ModelPackageGroupName: "MyModelPackageGroup",
ModelApprovalStatus: "PendingManualApproval",
ModelPackageDescription: "This is my first model package created with Alchemy."
});

Configure a ModelPackage with additional properties for enhanced functionality.

const advancedModelPackage = await AWS.SageMaker.ModelPackage("advancedModelPackage", {
ModelPackageName: "AdvancedModelPackage",
ModelPackageGroupName: "AdvancedModelGroup",
ModelApprovalStatus: "Approved",
ModelPackageDescription: "This model package has advanced configurations.",
InferenceSpecification: {
Containers: [{
Image: "123456789012.dkr.ecr.us-west-2.amazonaws.com/my-custom-image:latest",
ModelDataUrl: "s3://my-bucket/my-model.tar.gz"
}],
SupportedContentTypes: ["application/json"],
SupportedResponseMIMETypes: ["application/json"],
SupportedProtocols: ["HTTP"]
},
DriftCheckBaselines: {
Baselines: [{
ModelDataUrl: "s3://my-bucket/baseline-data.json",
DriftCheckBaselines: {
BaselineData: "my-baseline-data",
BaselineMetrics: "my-baseline-metrics"
}
}]
},
Tags: [{
Key: "Environment",
Value: "Production"
}]
});

Define a ModelPackage with a custom inference specification.

const customInferenceModelPackage = await AWS.SageMaker.ModelPackage("customInferenceModelPackage", {
ModelPackageName: "CustomInferenceModelPackage",
ModelPackageGroupName: "CustomInferenceGroup",
ModelApprovalStatus: "PendingManualApproval",
InferenceSpecification: {
Containers: [{
Image: "123456789012.dkr.ecr.us-west-2.amazonaws.com/my-inference-image:latest",
ModelDataUrl: "s3://my-bucket/custom-inference-model.tar.gz"
}],
SupportedContentTypes: ["text/csv"],
SupportedProtocols: ["HTTP"],
SupportedResponseMIMETypes: ["application/json"]
},
SamplePayloadUrl: "s3://my-bucket/sample-payload.json",
ModelMetrics: {
ModelQuality: {
Statistics: {
S3Uri: "s3://my-bucket/model-quality-statistics.json"
}
}
}
});