Amazon Sagemaker Edge Manager

2020/12/08 - Amazon Sagemaker Edge Manager - 2 new api methods

Changes  This feature helps you monitor model performance characteristics such as accuracy, identify undesired bias in your ML models, and explain model decisions better with explainability drift detection.

SendHeartbeat (new) Link ¶

Use to get the current status of devices registered on SageMaker Edge Manager.

See also: AWS API Documentation

Request Syntax

client.send_heartbeat(
    AgentMetrics=[
        {
            'Dimension': 'string',
            'MetricName': 'string',
            'Value': 123.0,
            'Timestamp': datetime(2015, 1, 1)
        },
    ],
    Models=[
        {
            'ModelName': 'string',
            'ModelVersion': 'string',
            'LatestSampleTime': datetime(2015, 1, 1),
            'LatestInference': datetime(2015, 1, 1),
            'ModelMetrics': [
                {
                    'Dimension': 'string',
                    'MetricName': 'string',
                    'Value': 123.0,
                    'Timestamp': datetime(2015, 1, 1)
                },
            ]
        },
    ],
    AgentVersion='string',
    DeviceName='string',
    DeviceFleetName='string'
)
type AgentMetrics

list

param AgentMetrics

For internal use. Returns a list of SageMaker Edge Manager agent operating metrics.

  • (dict) --

    Information required for edge device metrics.

    • Dimension (string) --

      The dimension of metrics published.

    • MetricName (string) --

      Returns the name of the metric.

    • Value (float) --

      Returns the value of the metric.

    • Timestamp (datetime) --

      Timestamp of when the metric was requested.

type Models

list

param Models

Returns a list of models deployed on the the device.

  • (dict) --

    Information about a model deployed on an edge device that is registered with SageMaker Edge Manager.

    • ModelName (string) --

      The name of the model.

    • ModelVersion (string) --

      The version of the model.

    • LatestSampleTime (datetime) --

      The timestamp of the last data sample taken.

    • LatestInference (datetime) --

      The timestamp of the last inference that was made.

    • ModelMetrics (list) --

      Information required for model metrics.

      • (dict) --

        Information required for edge device metrics.

        • Dimension (string) --

          The dimension of metrics published.

        • MetricName (string) --

          Returns the name of the metric.

        • Value (float) --

          Returns the value of the metric.

        • Timestamp (datetime) --

          Timestamp of when the metric was requested.

type AgentVersion

string

param AgentVersion

[REQUIRED]

Returns the version of the agent.

type DeviceName

string

param DeviceName

[REQUIRED]

The unique name of the device.

type DeviceFleetName

string

param DeviceFleetName

[REQUIRED]

The name of the fleet that the device belongs to.

returns

None

GetDeviceRegistration (new) Link ¶

Use to check if a device is registered with SageMaker Edge Manager.

See also: AWS API Documentation

Request Syntax

client.get_device_registration(
    DeviceName='string',
    DeviceFleetName='string'
)
type DeviceName

string

param DeviceName

[REQUIRED]

The unique name of the device you want to get the registration status from.

type DeviceFleetName

string

param DeviceFleetName

[REQUIRED]

The name of the fleet that the device belongs to.

rtype

dict

returns

Response Syntax

{
    'DeviceRegistration': 'string',
    'CacheTTL': 'string'
}

Response Structure

  • (dict) --

    • DeviceRegistration (string) --

      Describes if the device is currently registered with SageMaker Edge Manager.

    • CacheTTL (string) --

      The amount of time, in seconds, that the registration status is stored on the device’s cache before it is refreshed.