Model ​
The Model resource allows you to create and manage AWS SageMaker Models for deploying machine learning algorithms and workflows.
Minimal Example ​
Create a basic SageMaker model with required properties and one optional property.
ts
import AWS from "alchemy/aws/control";
const simpleModel = await AWS.SageMaker.Model("simpleModel", {
ModelName: "simple-model",
ExecutionRoleArn: "arn:aws:iam::123456789012:role/SageMakerExecutionRole",
PrimaryContainer: {
Image: "123456789012.dkr.ecr.us-west-2.amazonaws.com/my-image:latest",
ModelDataUrl: "s3://my-bucket/model.tar.gz"
}
});
Advanced Configuration ​
Configure a SageMaker model with VPC settings and tags for better resource management.
ts
const advancedModel = await AWS.SageMaker.Model("advancedModel", {
ModelName: "advanced-model",
ExecutionRoleArn: "arn:aws:iam::123456789012:role/SageMakerExecutionRole",
PrimaryContainer: {
Image: "123456789012.dkr.ecr.us-west-2.amazonaws.com/my-advanced-image:latest",
ModelDataUrl: "s3://my-bucket/advanced-model.tar.gz"
},
VpcConfig: {
SecurityGroupIds: ["sg-0123456789abcdef0"],
Subnets: ["subnet-0123456789abcdef0"]
},
Tags: [
{
Key: "Environment",
Value: "Production"
},
{
Key: "Project",
Value: "MachineLearning"
}
]
});
Network Isolation ​
Create a SageMaker model with network isolation enabled for enhanced security during inference.
ts
const isolatedModel = await AWS.SageMaker.Model("isolatedModel", {
ModelName: "isolated-model",
ExecutionRoleArn: "arn:aws:iam::123456789012:role/SageMakerExecutionRole",
PrimaryContainer: {
Image: "123456789012.dkr.ecr.us-west-2.amazonaws.com/my-isolated-image:latest",
ModelDataUrl: "s3://my-bucket/isolated-model.tar.gz"
},
EnableNetworkIsolation: true
});
Custom Inference Execution Configuration ​
Define a custom inference execution configuration for a SageMaker model to specify execution settings.
ts
const customInferenceModel = await AWS.SageMaker.Model("customInferenceModel", {
ModelName: "custom-inference-model",
ExecutionRoleArn: "arn:aws:iam::123456789012:role/SageMakerExecutionRole",
PrimaryContainer: {
Image: "123456789012.dkr.ecr.us-west-2.amazonaws.com/my-custom-image:latest",
ModelDataUrl: "s3://my-bucket/custom-inference-model.tar.gz"
},
InferenceExecutionConfig: {
Mode: "SingleModel" // Options: "SingleModel", "MultiModel"
}
});