AWS SageMaker支持按实例度量自动扩展功能。要使用此功能，您需要执行以下步骤：

1. 创建SageMaker Notebook实例：

import sagemaker

sagemaker_session = sagemaker.Session()
role = sagemaker.get_execution_role()

sagemaker_session.create_notebook_instance(
    notebook_instance_name='my-notebook-instance',
    instance_type='ml.t2.medium',
    role_arn=role
)

2. 创建SageMaker自动扩展策略：

import boto3

autoscaling_client = boto3.client('application-autoscaling')

autoscaling_client.register_scalable_target(
    ServiceNamespace='sagemaker',
    ResourceId='endpoint/your-endpoint-name/variant/your-variant-name',
    ScalableDimension='sagemaker:variant:DesiredInstanceCount',
    MinCapacity=1,
    MaxCapacity=10
)

autoscaling_client.put_scaling_policy(
    PolicyName='your-scaling-policy-name',
    ServiceNamespace='sagemaker',
    ResourceId='endpoint/your-endpoint-name/variant/your-variant-name',
    ScalableDimension='sagemaker:variant:DesiredInstanceCount',
    PolicyType='TargetTrackingScaling',
    TargetTrackingScalingPolicyConfiguration={
        'TargetValue': 1000,
        'PredefinedMetricSpecification': {
            'PredefinedMetricType': 'SageMakerVariantInvocationsPerInstance'
        },
        'ScaleInCooldown': 60,
        'ScaleOutCooldown': 60,
        'DisableScaleIn': False
    }
)

3. 启动SageMaker实例并监控自动扩展：

import time

sagemaker_client = boto3.client('sagemaker')

sagemaker_client.update_endpoint_weights_and_capacities(
    EndpointName='your-endpoint-name',
    DesiredWeightsAndCapacities=[
        {
            'VariantName': 'your-variant-name',
            'DesiredWeight': 1,
            'DesiredInstanceCount': 1
        }
    ]
)

# 等待自动扩展生效
time.sleep(300)

# 获取SageMaker实例状态
response = sagemaker_client.describe_endpoint(
    EndpointName='your-endpoint-name'
)

# 打印实例状态
print(response['EndpointStatus'])

以上代码示例了如何使用AWS SDK for Python（Boto3）在SageMaker中创建自动扩展策略。您需要根据自己的实际情况进行适当的更改，如替换your-endpoint-name和your-variant-name等参数。请确保您已正确配置AWS访问凭证，并具有执行这些操作所需的权限。