Skip to content
AWS for Query.Farm

aws.sagemaker.search

Example SQL Queries

SELECT * FROM
aws.sagemaker.search
WHERE
"resource" = 'VALUE';

Description

Finds SageMaker resources that match a search query. Matching resources are returned as a list of SearchRecord objects in the response. You can sort the search results by any resource property in a ascending or descending order.

You can query against the following value types: numeric, text, Boolean, and timestamp.

The Search API may provide access to otherwise restricted data. See Amazon SageMaker API Permissions: Actions, Permissions, and Resources Reference for more information.

Table Definition

Column NameColumn Data Type
resource Required Input Column

The name of the SageMaker resource to search for.

VARCHAR
cross_account_filter_option Input Column

A cross account filter option. When the value is "CrossAccount" the search results will only include resources made discoverable to you from other accounts. When the value is "SameAccount" or null the search results will only include resources from your account. Default is null. For more information on searching for resources made discoverable to your account, see Search discoverable resources in the SageMaker Developer Guide. The maximum number of ResourceCatalogs viewable is 1000. 

VARCHAR
search_expression Input Column

A Boolean conditional statement. Resources must satisfy this condition to be included in search results. You must provide at least one subexpression, filter, or nested filter. The maximum number of recursive SubExpressions, NestedFilters, and Filters that can be included in a SearchExpression object is 50.

STRUCT(
    "filters" STRUCT(
        "name" VARCHAR,
        "operator" VARCHAR,
        "value" VARCHAR
    )[],
    "nested_filters" STRUCT(
        "nested_property_name" VARCHAR,
        "filters" STRUCT(
            "name" VARCHAR,
            "operator" VARCHAR,
            "value" VARCHAR
        )[]
    )[],
    "operator" VARCHAR
)
Show child fields
search_expression.filters[]
Show child fields
search_expression.filters[].name

A resource property name. For example, TrainingJobName. For valid property names, see SearchRecord. You must specify a valid property for the resource.

search_expression.filters[].operator

A Boolean binary operator that is used to evaluate the filter. The operator field contains one of the following values:

Equals

The value of Name equals Value.

NotEquals

The value of Name doesn't equal Value.

Exists

The Name property exists.

NotExists

The Name property does not exist.

GreaterThan

The value of Name is greater than Value. Not supported for text properties.

GreaterThanOrEqualTo

The value of Name is greater than or equal to Value. Not supported for text properties.

LessThan

The value of Name is less than Value. Not supported for text properties.

LessThanOrEqualTo

The value of Name is less than or equal to Value. Not supported for text properties.

In

The value of Name is one of the comma delimited strings in Value. Only supported for text properties.

Contains

The value of Name contains the string Value. Only supported for text properties.

A SearchExpression can include the Contains operator multiple times when the value of Name is one of the following:

  • Experiment.DisplayName

  • Experiment.ExperimentName

  • Experiment.Tags

  • Trial.DisplayName

  • Trial.TrialName

  • Trial.Tags

  • TrialComponent.DisplayName

  • TrialComponent.TrialComponentName

  • TrialComponent.Tags

  • TrialComponent.InputArtifacts

  • TrialComponent.OutputArtifacts

A SearchExpression can include only one Contains operator for all other values of Name. In these cases, if you include multiple Contains operators in the SearchExpression, the result is the following error message: "'CONTAINS' operator usage limit of 1 exceeded."

search_expression.filters[].value

A value used with Name and Operator to determine which resources satisfy the filter's condition. For numerical properties, Value must be an integer or floating-point decimal. For timestamp properties, Value must be an ISO 8601 date-time string of the following format: YYYY-mm-dd'T'HH:MM:SS.

search_expression.nested_filters[]
Show child fields
search_expression.nested_filters[].filters[]
Show child fields
search_expression.nested_filters[].filters[].name

A resource property name. For example, TrainingJobName. For valid property names, see SearchRecord. You must specify a valid property for the resource.

search_expression.nested_filters[].filters[].operator

A Boolean binary operator that is used to evaluate the filter. The operator field contains one of the following values:

Equals

The value of Name equals Value.

NotEquals

The value of Name doesn't equal Value.

Exists

The Name property exists.

NotExists

The Name property does not exist.

GreaterThan

The value of Name is greater than Value. Not supported for text properties.

GreaterThanOrEqualTo

The value of Name is greater than or equal to Value. Not supported for text properties.

LessThan

The value of Name is less than Value. Not supported for text properties.

LessThanOrEqualTo

The value of Name is less than or equal to Value. Not supported for text properties.

In

The value of Name is one of the comma delimited strings in Value. Only supported for text properties.

Contains

The value of Name contains the string Value. Only supported for text properties.

A SearchExpression can include the Contains operator multiple times when the value of Name is one of the following:

  • Experiment.DisplayName

  • Experiment.ExperimentName

  • Experiment.Tags

  • Trial.DisplayName

  • Trial.TrialName

  • Trial.Tags

  • TrialComponent.DisplayName

  • TrialComponent.TrialComponentName

  • TrialComponent.Tags

  • TrialComponent.InputArtifacts

  • TrialComponent.OutputArtifacts

A SearchExpression can include only one Contains operator for all other values of Name. In these cases, if you include multiple Contains operators in the SearchExpression, the result is the following error message: "'CONTAINS' operator usage limit of 1 exceeded."

search_expression.nested_filters[].filters[].value

A value used with Name and Operator to determine which resources satisfy the filter's condition. For numerical properties, Value must be an integer or floating-point decimal. For timestamp properties, Value must be an ISO 8601 date-time string of the following format: YYYY-mm-dd'T'HH:MM:SS.

search_expression.nested_filters[].nested_property_name

The name of the property to use in the nested filters. The value must match a listed property name, such as InputDataConfig.

search_expression.operator

A Boolean operator used to evaluate the search expression. If you want every conditional statement in all lists to be satisfied for the entire search expression to be true, specify And. If only a single conditional statement needs to be true for the entire search expression to be true, specify Or. The default value is And.

sort_by Input Column

The name of the resource property used to sort the SearchResults. The default is LastModifiedTime.

VARCHAR
sort_order Input Column

How SearchResults are ordered. Valid values are Ascending or Descending. The default is Descending.

VARCHAR
visibility_conditions Input Column

Limits the results of your search request to the resources that you can access. 

STRUCT(
    "key" VARCHAR,
    "value" VARCHAR
)[]
Show child fields
visibility_conditions[]
Show child fields
visibility_conditions[].key

The key that specifies the tag that you're using to filter the search results. It must be in the following format: Tags.<key>.

visibility_conditions[].value

The value for the tag that you're using to filter the search results.

_aws_profile Input Column

The AWS profile defines the AWS identity used. It can be defined via credentials or by assuming a IAM role.

STRUCT(
    "type" VARCHAR,
    "name" VARCHAR,
    "account_id" VARCHAR,
    "via_profile_name" VARCHAR,
    "assumed_role_arn" VARCHAR,
    "organization" STRUCT(
        "account_name" VARCHAR,
        "id" VARCHAR,
        "tags" STRUCT(
            "key" VARCHAR,
            "value" VARCHAR
        )[],
        "master_account" STRUCT(
            "id" VARCHAR,
            "email" VARCHAR
        ),
        "parents" STRUCT(
            "type" VARCHAR,
            "id" VARCHAR,
            "name" VARCHAR,
            "tags" STRUCT(
                "key" VARCHAR,
                "value" VARCHAR
            )[]
        )[]
    )
)
Show child fields
_aws_profile.account_id

The AWS account id

_aws_profile.assumed_role_arn

The ARN of the assumed role

_aws_profile.name

The unique name of the profile.

_aws_profile.organization

Information about this profile's membership in the AWS organization.

Show child fields
_aws_profile.organization.account_name

The name of account speciifed by the organization

_aws_profile.organization.id

The organization id

_aws_profile.organization.master_account
Show child fields
_aws_profile.organization.master_account.email

The organization master account email address

_aws_profile.organization.master_account.id

The organization master account id

_aws_profile.organization.parents[]
Show child fields
_aws_profile.organization.parents[].id

The id of the parent

_aws_profile.organization.parents[].name

The name of the parent

_aws_profile.organization.parents[].tags[]
Show child fields
_aws_profile.organization.parents[].tags[].key
_aws_profile.organization.parents[].tags[].value
_aws_profile.organization.parents[].type

The type of parent can be an organization unit or a root

_aws_profile.organization.tags[]
Show child fields
_aws_profile.organization.tags[].key
_aws_profile.organization.tags[].value
_aws_profile.type

The type of profile, either 'credentials' or 'assumed_role'

_aws_profile.via_profile_name

This IAM role for this profile is assumed by first utilizing another profile with this name to obtain credentials.

endpoint

A hosted endpoint for real-time inference.

STRUCT(
    "endpoint_name" VARCHAR,
    "endpoint_arn" VARCHAR,
    "endpoint_config_name" VARCHAR,
    "production_variants" STRUCT(
        "variant_name" VARCHAR,
        "deployed_images" STRUCT(
            "specified_image" VARCHAR,
            "resolved_image" VARCHAR,
            "resolution_time" TIMESTAMP_S
        )[],
        "current_weight" DOUBLE,
        "desired_weight" DOUBLE,
        "current_instance_count" BIGINT,
        "desired_instance_count" BIGINT,
        "variant_status" STRUCT(
            "status" VARCHAR,
            "status_message" VARCHAR,
            "start_time" TIMESTAMP_S
        )[],
        "current_serverless_config" STRUCT(
            "memory_size_in_mb" BIGINT,
            "max_concurrency" BIGINT,
            "provisioned_concurrency" BIGINT
        ),
        "desired_serverless_config" STRUCT(
            "memory_size_in_mb" BIGINT,
            "max_concurrency" BIGINT,
            "provisioned_concurrency" BIGINT
        ),
        "managed_instance_scaling" STRUCT(
            "status" VARCHAR,
            "min_instance_count" BIGINT,
            "max_instance_count" BIGINT
        ),
        "routing_config" STRUCT(
            "routing_strategy" VARCHAR
        )
    )[],
    "data_capture_config" STRUCT(
        "enable_capture" BOOLEAN,
        "capture_status" VARCHAR,
        "current_sampling_percentage" BIGINT,
        "destination_s3_uri" VARCHAR,
        "kms_key_id" VARCHAR
    ),
    "endpoint_status" VARCHAR,
    "failure_reason" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "last_modified_time" TIMESTAMP_S,
    "monitoring_schedules" STRUCT(
        "monitoring_schedule_arn" VARCHAR,
        "monitoring_schedule_name" VARCHAR,
        "monitoring_schedule_status" VARCHAR,
        "monitoring_type" VARCHAR,
        "failure_reason" VARCHAR,
        "creation_time" TIMESTAMP_S,
        "last_modified_time" TIMESTAMP_S,
        "monitoring_schedule_config" STRUCT(
            "schedule_config" STRUCT(
                "schedule_expression" VARCHAR,
                "data_analysis_start_time" VARCHAR,
                "data_analysis_end_time" VARCHAR
            ),
            "monitoring_job_definition" STRUCT(
                "baseline_config" STRUCT(
                    "baselining_job_name" VARCHAR,
                    "constraints_resource" STRUCT(
                        "s3_uri" VARCHAR
                    ),
                    "statistics_resource" STRUCT(
                        "s3_uri" VARCHAR
                    )
                ),
                "monitoring_inputs" STRUCT(
                    "endpoint_input" STRUCT(
                        "endpoint_name" VARCHAR,
                        "local_path" VARCHAR,
                        "s3_input_mode" VARCHAR,
                        "s3_data_distribution_type" VARCHAR,
                        "features_attribute" VARCHAR,
                        "inference_attribute" VARCHAR,
                        "probability_attribute" VARCHAR,
                        "probability_threshold_attribute" DOUBLE,
                        "start_time_offset" VARCHAR,
                        "end_time_offset" VARCHAR,
                        "exclude_features_attribute" VARCHAR
                    ),
                    "batch_transform_input" STRUCT(
                        "data_captured_destination_s3_uri" VARCHAR,
                        "dataset_format" STRUCT(
                            "csv" STRUCT(
                                "header" BOOLEAN
                            ),
                            "json" STRUCT(
                                "line" BOOLEAN
                            ),
                            "parquet" BOOLEAN
                        ),
                        "local_path" VARCHAR,
                        "s3_input_mode" VARCHAR,
                        "s3_data_distribution_type" VARCHAR,
                        "features_attribute" VARCHAR,
                        "inference_attribute" VARCHAR,
                        "probability_attribute" VARCHAR,
                        "probability_threshold_attribute" DOUBLE,
                        "start_time_offset" VARCHAR,
                        "end_time_offset" VARCHAR,
                        "exclude_features_attribute" VARCHAR
                    )
                )[],
                "monitoring_output_config" STRUCT(
                    "monitoring_outputs" STRUCT(
                        "s3_output" STRUCT(
                            "s3_uri" VARCHAR,
                            "local_path" VARCHAR,
                            "s3_upload_mode" VARCHAR
                        )
                    )[],
                    "kms_key_id" VARCHAR
                ),
                "monitoring_resources" STRUCT(
                    "cluster_config" STRUCT(
                        "instance_count" BIGINT,
                        "instance_type" VARCHAR,
                        "volume_size_in_gb" BIGINT,
                        "volume_kms_key_id" VARCHAR
                    )
                ),
                "monitoring_app_specification" STRUCT(
                    "image_uri" VARCHAR,
                    "container_entrypoint" VARCHAR[],
                    "container_arguments" VARCHAR[],
                    "record_preprocessor_source_uri" VARCHAR,
                    "post_analytics_processor_source_uri" VARCHAR
                ),
                "stopping_condition" STRUCT(
                    "max_runtime_in_seconds" BIGINT
                ),
                "environment" MAP(VARCHAR, VARCHAR),
                "network_config" STRUCT(
                    "enable_inter_container_traffic_encryption" BOOLEAN,
                    "enable_network_isolation" BOOLEAN,
                    "vpc_config" STRUCT(
                        "security_group_ids" VARCHAR[],
                        "subnets" VARCHAR[]
                    )
                ),
                "role_arn" VARCHAR
            ),
            "monitoring_job_definition_name" VARCHAR,
            "monitoring_type" VARCHAR
        ),
        "endpoint_name" VARCHAR,
        "last_monitoring_execution_summary" STRUCT(
            "monitoring_schedule_name" VARCHAR,
            "scheduled_time" TIMESTAMP_S,
            "creation_time" TIMESTAMP_S,
            "last_modified_time" TIMESTAMP_S,
            "monitoring_execution_status" VARCHAR,
            "processing_job_arn" VARCHAR,
            "endpoint_name" VARCHAR,
            "failure_reason" VARCHAR,
            "monitoring_job_definition_name" VARCHAR,
            "monitoring_type" VARCHAR
        ),
        "tags" STRUCT(
            "key" VARCHAR,
            "value" VARCHAR
        )[]
    )[],
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[],
    "shadow_production_variants" STRUCT(
        "variant_name" VARCHAR,
        "deployed_images" STRUCT(
            "specified_image" VARCHAR,
            "resolved_image" VARCHAR,
            "resolution_time" TIMESTAMP_S
        )[],
        "current_weight" DOUBLE,
        "desired_weight" DOUBLE,
        "current_instance_count" BIGINT,
        "desired_instance_count" BIGINT,
        "variant_status" STRUCT(
            "status" VARCHAR,
            "status_message" VARCHAR,
            "start_time" TIMESTAMP_S
        )[],
        "current_serverless_config" STRUCT(
            "memory_size_in_mb" BIGINT,
            "max_concurrency" BIGINT,
            "provisioned_concurrency" BIGINT
        ),
        "desired_serverless_config" STRUCT(
            "memory_size_in_mb" BIGINT,
            "max_concurrency" BIGINT,
            "provisioned_concurrency" BIGINT
        ),
        "managed_instance_scaling" STRUCT(
            "status" VARCHAR,
            "min_instance_count" BIGINT,
            "max_instance_count" BIGINT
        ),
        "routing_config" STRUCT(
            "routing_strategy" VARCHAR
        )
    )[]
)
Show child fields
endpoint.creation_time

The time that the endpoint was created.

endpoint.data_capture_config

The currently active data capture configuration used by your Endpoint.

Show child fields
endpoint.data_capture_config.capture_status

Whether data capture is currently functional.

endpoint.data_capture_config.current_sampling_percentage

The percentage of requests being captured by your Endpoint.

endpoint.data_capture_config.destination_s3_uri

The Amazon S3 location being used to capture the data.

endpoint.data_capture_config.enable_capture

Whether data capture is enabled or disabled.

endpoint.data_capture_config.kms_key_id

The KMS key being used to encrypt the data in Amazon S3.

endpoint.endpoint_arn

The Amazon Resource Name (ARN) of the endpoint.

endpoint.endpoint_config_name

The endpoint configuration associated with the endpoint.

endpoint.endpoint_name

The name of the endpoint.

endpoint.endpoint_status

The status of the endpoint.

endpoint.failure_reason

If the endpoint failed, the reason it failed.

endpoint.last_modified_time

The last time the endpoint was modified.

endpoint.monitoring_schedules[]
Show child fields
endpoint.monitoring_schedules[].creation_time

The time that the monitoring schedule was created.

endpoint.monitoring_schedules[].endpoint_name

The endpoint that hosts the model being monitored.

endpoint.monitoring_schedules[].failure_reason

If the monitoring schedule failed, the reason it failed.

endpoint.monitoring_schedules[].last_modified_time

The last time the monitoring schedule was changed.

endpoint.monitoring_schedules[].last_monitoring_execution_summary

Summary of information about the last monitoring job to run.

Show child fields
endpoint.monitoring_schedules[].last_monitoring_execution_summary.creation_time

The time at which the monitoring job was created.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.endpoint_name

The name of the endpoint used to run the monitoring job.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.failure_reason

Contains the reason a monitoring job failed, if it failed.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.last_modified_time

A timestamp that indicates the last time the monitoring job was modified.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.monitoring_execution_status

The status of the monitoring job.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.monitoring_job_definition_name

The name of the monitoring job.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.monitoring_schedule_name

The name of the monitoring schedule.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.monitoring_type

The type of the monitoring job.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.processing_job_arn

The Amazon Resource Name (ARN) of the monitoring job.

endpoint.monitoring_schedules[].last_monitoring_execution_summary.scheduled_time

The time the monitoring job was scheduled.

endpoint.monitoring_schedules[].monitoring_schedule_arn

The Amazon Resource Name (ARN) of the monitoring schedule.

endpoint.monitoring_schedules[].monitoring_schedule_config

Configures the monitoring schedule and defines the monitoring job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition

Defines the monitoring job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config

Baseline configuration used to validate that the data conforms to the specified constraints and statistics

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.baselining_job_name

The name of the job that performs baselining for the monitoring job.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.constraints_resource

The baseline constraint file in Amazon S3 that the current monitoring job should validated against.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.constraints_resource.s3_uri

The Amazon S3 URI for the constraints resource.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.statistics_resource

The baseline statistics file in Amazon S3 that the current monitoring job should be validated against.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.statistics_resource.s3_uri

The Amazon S3 URI for the statistics resource.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.environment

Sets the environment variables in the Docker container.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification

Configures the monitoring job to run a specified Docker container image.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.container_arguments[]
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.container_entrypoint[]
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.image_uri

The container image to be run by the monitoring job.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.post_analytics_processor_source_uri

An Amazon S3 URI to a script that is called after analysis has been performed. Applicable only for the built-in (first party) containers.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.record_preprocessor_source_uri

An Amazon S3 URI to a script that is called per row prior to running analysis. It can base64 decode the payload and convert it into a flattened JSON so that the built-in container can use the converted data. Applicable only for the built-in (first party) containers.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[]
Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input

Input object for the batch transform job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.data_captured_destination_s3_uri

The Amazon S3 location being used to capture the data.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format

The dataset format for your batch transform job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.csv

The CSV dataset used in the monitoring job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.csv.header

Indicates if the CSV data has a header.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.json

The JSON dataset used in the monitoring job

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.json.line

Indicates if the file should be read as a JSON object per line.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.parquet

The Parquet dataset used in the monitoring job

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.end_time_offset

If specified, monitoring jobs subtract this time from the end time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.exclude_features_attribute

The attributes of the input data to exclude from the analysis.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.features_attribute

The attributes of the input data that are the input features.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.inference_attribute

The attribute of the input data that represents the ground truth label.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.local_path

Path to the filesystem where the batch transform data is available to the container.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.probability_attribute

In a classification problem, the attribute that represents the class probability.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.probability_threshold_attribute

The threshold for the class probability to be evaluated as a positive result.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.s3_data_distribution_type

Whether input data distributed in Amazon S3 is fully replicated or sharded by an S3 key. Defaults to FullyReplicated

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.s3_input_mode

Whether the Pipe or File is used as the input mode for transferring data for the monitoring job. Pipe mode is recommended for large datasets. File mode is useful for small files that fit in memory. Defaults to File.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.start_time_offset

If specified, monitoring jobs substract this time from the start time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input

The endpoint for a monitoring job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.end_time_offset

If specified, monitoring jobs substract this time from the end time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.endpoint_name

An endpoint in customer's account which has enabled DataCaptureConfig enabled.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.exclude_features_attribute

The attributes of the input data to exclude from the analysis.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.features_attribute

The attributes of the input data that are the input features.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.inference_attribute

The attribute of the input data that represents the ground truth label.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.local_path

Path to the filesystem where the endpoint data is available to the container.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.probability_attribute

In a classification problem, the attribute that represents the class probability.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.probability_threshold_attribute

The threshold for the class probability to be evaluated as a positive result.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.s3_data_distribution_type

Whether input data distributed in Amazon S3 is fully replicated or sharded by an Amazon S3 key. Defaults to FullyReplicated

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.s3_input_mode

Whether the Pipe or File is used as the input mode for transferring data for the monitoring job. Pipe mode is recommended for large datasets. File mode is useful for small files that fit in memory. Defaults to File.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.start_time_offset

If specified, monitoring jobs substract this time from the start time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config

The array of outputs from the monitoring job to be uploaded to Amazon S3.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.kms_key_id

The Key Management Service (KMS) key that Amazon SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[]
Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output

The Amazon S3 storage location where the results of a monitoring job are saved.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output.local_path

The local path to the Amazon S3 storage location where Amazon SageMaker saves the results of a monitoring job. LocalPath is an absolute path for the output data.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output.s3_upload_mode

Whether to upload the results of the monitoring job continuously or after the job completes.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output.s3_uri

A URI that identifies the Amazon S3 storage location where Amazon SageMaker saves the results of a monitoring job.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources

Identifies the resources, ML compute instances, and ML storage volumes to deploy for a monitoring job. In distributed processing, you specify more than one instance.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config

The configuration for the cluster resources used to run the processing job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.instance_count

The number of ML compute instances to use in the model monitoring job. For distributed processing jobs, specify a value greater than 1. The default value is 1.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.instance_type

The ML compute instance type for the processing job.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.volume_kms_key_id

The Key Management Service (KMS) key that Amazon SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) that run the model monitoring job.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.volume_size_in_gb

The size of the ML storage volume, in gigabytes, that you want to provision. You must specify sufficient ML storage for your scenario.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config

Specifies networking options for an monitoring job.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.enable_inter_container_traffic_encryption

Whether to encrypt all communications between distributed processing jobs. Choose True to encrypt communications. Encryption provides greater security for distributed processing jobs, but the processing might take longer.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.enable_network_isolation

Whether to allow inbound and outbound network calls to and from the containers used for the processing job.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.vpc_config

Specifies an Amazon Virtual Private Cloud (VPC) that your SageMaker jobs, hosted models, and compute resources have access to. You can control access to and from your resources by configuring a VPC. For more information, see Give SageMaker Access to Resources in your Amazon VPC.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.vpc_config.security_group_ids[]
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.vpc_config.subnets[]
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.role_arn

The Amazon Resource Name (ARN) of an IAM role that Amazon SageMaker can assume to perform tasks on your behalf.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.stopping_condition

Specifies a time limit for how long the monitoring job is allowed to run.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.stopping_condition.max_runtime_in_seconds

The maximum runtime allowed in seconds.

The MaxRuntimeInSeconds cannot exceed the frequency of the job. For data quality and model explainability, this can be up to 3600 seconds for an hourly schedule. For model bias and model quality hourly schedules, this can be up to 1800 seconds.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition_name

The name of the monitoring job definition to schedule.

endpoint.monitoring_schedules[].monitoring_schedule_config.monitoring_type

The type of the monitoring job definition to schedule.

endpoint.monitoring_schedules[].monitoring_schedule_config.schedule_config

Configures the monitoring schedule.

Show child fields
endpoint.monitoring_schedules[].monitoring_schedule_config.schedule_config.data_analysis_end_time

Sets the end time for a monitoring job window. Express this time as an offset to the times that you schedule your monitoring jobs to run. You schedule monitoring jobs with the ScheduleExpression parameter. Specify this offset in ISO 8601 duration format. For example, if you want to end the window one hour before the start of each monitoring job, you would specify: "-PT1H".

The end time that you specify must not follow the start time that you specify by more than 24 hours. You specify the start time with the DataAnalysisStartTime parameter.

If you set ScheduleExpression to NOW, this parameter is required.

endpoint.monitoring_schedules[].monitoring_schedule_config.schedule_config.data_analysis_start_time

Sets the start time for a monitoring job window. Express this time as an offset to the times that you schedule your monitoring jobs to run. You schedule monitoring jobs with the ScheduleExpression parameter. Specify this offset in ISO 8601 duration format. For example, if you want to monitor the five hours of data in your dataset that precede the start of each monitoring job, you would specify: "-PT5H".

The start time that you specify must not precede the end time that you specify by more than 24 hours. You specify the end time with the DataAnalysisEndTime parameter.

If you set ScheduleExpression to NOW, this parameter is required.

endpoint.monitoring_schedules[].monitoring_schedule_config.schedule_config.schedule_expression

A cron expression that describes details about the monitoring schedule.

The supported cron expressions are:

  • If you want to set the job to start every hour, use the following:

    Hourly: cron(0 * ? * * *)

  • If you want to start the job daily:

    cron(0 [00-23] ? * * *)

  • If you want to run the job one time, immediately, use the following keyword:

    NOW

For example, the following are valid cron expressions:

  • Daily at noon UTC: cron(0 12 ? * * *)

  • Daily at midnight UTC: cron(0 0 ? * * *)

To support running every 6, 12 hours, the following are also supported:

cron(0 [00-23]/[01-24] ? * * *)

For example, the following are valid cron expressions:

  • Every 12 hours, starting at 5pm UTC: cron(0 17/12 ? * * *)

  • Every two hours starting at midnight: cron(0 0/2 ? * * *)

  • Even though the cron expression is set to start at 5PM UTC, note that there could be a delay of 0-20 minutes from the actual requested time to run the execution.

  • We recommend that if you would like a daily schedule, you do not provide this parameter. Amazon SageMaker will pick a time for running every day.

You can also specify the keyword NOW to run the monitoring job immediately, one time, without recurring.

endpoint.monitoring_schedules[].monitoring_schedule_name

The name of the monitoring schedule.

endpoint.monitoring_schedules[].monitoring_schedule_status

The status of the monitoring schedule. This can be one of the following values.

  • PENDING - The schedule is pending being created.

  • FAILED - The schedule failed.

  • SCHEDULED - The schedule was successfully created.

  • STOPPED - The schedule was stopped.

endpoint.monitoring_schedules[].monitoring_type

The type of the monitoring job definition to schedule.

endpoint.monitoring_schedules[].tags[]
Show child fields
endpoint.monitoring_schedules[].tags[].key

The tag key. Tag keys must be unique per resource.

endpoint.monitoring_schedules[].tags[].value

The tag value.

endpoint.production_variants[]
Show child fields
endpoint.production_variants[].current_instance_count

The number of instances associated with the variant.

endpoint.production_variants[].current_serverless_config

The serverless configuration for the endpoint.

Show child fields
endpoint.production_variants[].current_serverless_config.max_concurrency

The maximum number of concurrent invocations your serverless endpoint can process.

endpoint.production_variants[].current_serverless_config.memory_size_in_mb

The memory size of your serverless endpoint. Valid values are in 1 GB increments: 1024 MB, 2048 MB, 3072 MB, 4096 MB, 5120 MB, or 6144 MB.

endpoint.production_variants[].current_serverless_config.provisioned_concurrency

The amount of provisioned concurrency to allocate for the serverless endpoint. Should be less than or equal to MaxConcurrency.

This field is not supported for serverless endpoint recommendations for Inference Recommender jobs. For more information about creating an Inference Recommender job, see CreateInferenceRecommendationsJobs.

endpoint.production_variants[].current_weight

The weight associated with the variant.

endpoint.production_variants[].deployed_images[]
Show child fields
endpoint.production_variants[].deployed_images[].resolution_time

The date and time when the image path for the model resolved to the ResolvedImage

endpoint.production_variants[].deployed_images[].resolved_image

The specific digest path of the image hosted in this ProductionVariant.

endpoint.production_variants[].deployed_images[].specified_image

The image path you specified when you created the model.

endpoint.production_variants[].desired_instance_count

The number of instances requested in the UpdateEndpointWeightsAndCapacities request.

endpoint.production_variants[].desired_serverless_config

The serverless configuration requested for the endpoint update.

Show child fields
endpoint.production_variants[].desired_serverless_config.max_concurrency

The maximum number of concurrent invocations your serverless endpoint can process.

endpoint.production_variants[].desired_serverless_config.memory_size_in_mb

The memory size of your serverless endpoint. Valid values are in 1 GB increments: 1024 MB, 2048 MB, 3072 MB, 4096 MB, 5120 MB, or 6144 MB.

endpoint.production_variants[].desired_serverless_config.provisioned_concurrency

The amount of provisioned concurrency to allocate for the serverless endpoint. Should be less than or equal to MaxConcurrency.

This field is not supported for serverless endpoint recommendations for Inference Recommender jobs. For more information about creating an Inference Recommender job, see CreateInferenceRecommendationsJobs.

endpoint.production_variants[].desired_weight

The requested weight, as specified in the UpdateEndpointWeightsAndCapacities request.

endpoint.production_variants[].managed_instance_scaling

Settings that control the range in the number of instances that the endpoint provisions as it scales up or down to accommodate traffic.

Show child fields
endpoint.production_variants[].managed_instance_scaling.max_instance_count

The maximum number of instances that the endpoint can provision when it scales up to accommodate an increase in traffic.

endpoint.production_variants[].managed_instance_scaling.min_instance_count

The minimum number of instances that the endpoint must retain when it scales down to accommodate a decrease in traffic.

endpoint.production_variants[].managed_instance_scaling.status

Indicates whether managed instance scaling is enabled.

endpoint.production_variants[].routing_config

Settings that control how the endpoint routes incoming traffic to the instances that the endpoint hosts.

Show child fields
endpoint.production_variants[].routing_config.routing_strategy

Sets how the endpoint routes incoming traffic:

  • LEAST_OUTSTANDING_REQUESTS: The endpoint routes requests to the specific instances that have more capacity to process them.

  • RANDOM: The endpoint routes each request to a randomly chosen instance.

endpoint.production_variants[].variant_name

The name of the variant.

endpoint.production_variants[].variant_status[]
Show child fields
endpoint.production_variants[].variant_status[].start_time

The start time of the current status change.

endpoint.production_variants[].variant_status[].status

The endpoint variant status which describes the current deployment stage status or operational status.

  • Creating: Creating inference resources for the production variant.

  • Deleting: Terminating inference resources for the production variant.

  • Updating: Updating capacity for the production variant.

  • ActivatingTraffic: Turning on traffic for the production variant.

  • Baking: Waiting period to monitor the CloudWatch alarms in the automatic rollback configuration.

endpoint.production_variants[].variant_status[].status_message

A message that describes the status of the production variant.

endpoint.shadow_production_variants[]
Show child fields
endpoint.shadow_production_variants[].current_instance_count

The number of instances associated with the variant.

endpoint.shadow_production_variants[].current_serverless_config

The serverless configuration for the endpoint.

Show child fields
endpoint.shadow_production_variants[].current_serverless_config.max_concurrency

The maximum number of concurrent invocations your serverless endpoint can process.

endpoint.shadow_production_variants[].current_serverless_config.memory_size_in_mb

The memory size of your serverless endpoint. Valid values are in 1 GB increments: 1024 MB, 2048 MB, 3072 MB, 4096 MB, 5120 MB, or 6144 MB.

endpoint.shadow_production_variants[].current_serverless_config.provisioned_concurrency

The amount of provisioned concurrency to allocate for the serverless endpoint. Should be less than or equal to MaxConcurrency.

This field is not supported for serverless endpoint recommendations for Inference Recommender jobs. For more information about creating an Inference Recommender job, see CreateInferenceRecommendationsJobs.

endpoint.shadow_production_variants[].current_weight

The weight associated with the variant.

endpoint.shadow_production_variants[].deployed_images[]
Show child fields
endpoint.shadow_production_variants[].deployed_images[].resolution_time

The date and time when the image path for the model resolved to the ResolvedImage

endpoint.shadow_production_variants[].deployed_images[].resolved_image

The specific digest path of the image hosted in this ProductionVariant.

endpoint.shadow_production_variants[].deployed_images[].specified_image

The image path you specified when you created the model.

endpoint.shadow_production_variants[].desired_instance_count

The number of instances requested in the UpdateEndpointWeightsAndCapacities request.

endpoint.shadow_production_variants[].desired_serverless_config

The serverless configuration requested for the endpoint update.

Show child fields
endpoint.shadow_production_variants[].desired_serverless_config.max_concurrency

The maximum number of concurrent invocations your serverless endpoint can process.

endpoint.shadow_production_variants[].desired_serverless_config.memory_size_in_mb

The memory size of your serverless endpoint. Valid values are in 1 GB increments: 1024 MB, 2048 MB, 3072 MB, 4096 MB, 5120 MB, or 6144 MB.

endpoint.shadow_production_variants[].desired_serverless_config.provisioned_concurrency

The amount of provisioned concurrency to allocate for the serverless endpoint. Should be less than or equal to MaxConcurrency.

This field is not supported for serverless endpoint recommendations for Inference Recommender jobs. For more information about creating an Inference Recommender job, see CreateInferenceRecommendationsJobs.

endpoint.shadow_production_variants[].desired_weight

The requested weight, as specified in the UpdateEndpointWeightsAndCapacities request.

endpoint.shadow_production_variants[].managed_instance_scaling

Settings that control the range in the number of instances that the endpoint provisions as it scales up or down to accommodate traffic.

Show child fields
endpoint.shadow_production_variants[].managed_instance_scaling.max_instance_count

The maximum number of instances that the endpoint can provision when it scales up to accommodate an increase in traffic.

endpoint.shadow_production_variants[].managed_instance_scaling.min_instance_count

The minimum number of instances that the endpoint must retain when it scales down to accommodate a decrease in traffic.

endpoint.shadow_production_variants[].managed_instance_scaling.status

Indicates whether managed instance scaling is enabled.

endpoint.shadow_production_variants[].routing_config

Settings that control how the endpoint routes incoming traffic to the instances that the endpoint hosts.

Show child fields
endpoint.shadow_production_variants[].routing_config.routing_strategy

Sets how the endpoint routes incoming traffic:

  • LEAST_OUTSTANDING_REQUESTS: The endpoint routes requests to the specific instances that have more capacity to process them.

  • RANDOM: The endpoint routes each request to a randomly chosen instance.

endpoint.shadow_production_variants[].variant_name

The name of the variant.

endpoint.shadow_production_variants[].variant_status[]
Show child fields
endpoint.shadow_production_variants[].variant_status[].start_time

The start time of the current status change.

endpoint.shadow_production_variants[].variant_status[].status

The endpoint variant status which describes the current deployment stage status or operational status.

  • Creating: Creating inference resources for the production variant.

  • Deleting: Terminating inference resources for the production variant.

  • Updating: Updating capacity for the production variant.

  • ActivatingTraffic: Turning on traffic for the production variant.

  • Baking: Waiting period to monitor the CloudWatch alarms in the automatic rollback configuration.

endpoint.shadow_production_variants[].variant_status[].status_message

A message that describes the status of the production variant.

endpoint.tags[]
Show child fields
endpoint.tags[].key

The tag key. Tag keys must be unique per resource.

endpoint.tags[].value

The tag value.

experiment

The properties of an experiment.

STRUCT(
    "experiment_name" VARCHAR,
    "experiment_arn" VARCHAR,
    "display_name" VARCHAR,
    "source" STRUCT(
        "source_arn" VARCHAR,
        "source_type" VARCHAR
    ),
    "description" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "last_modified_time" TIMESTAMP_S,
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
experiment.created_by

Who created the experiment.

Show child fields
experiment.created_by.domain_id

The domain associated with the user.

experiment.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
experiment.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

experiment.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

experiment.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

experiment.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

experiment.created_by.user_profile_name

The name of the user's profile.

experiment.creation_time

When the experiment was created.

experiment.description

The description of the experiment.

experiment.display_name

The name of the experiment as displayed. If DisplayName isn't specified, ExperimentName is displayed.

experiment.experiment_arn

The Amazon Resource Name (ARN) of the experiment.

experiment.experiment_name

The name of the experiment.

experiment.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
experiment.last_modified_by.domain_id

The domain associated with the user.

experiment.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
experiment.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

experiment.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

experiment.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

experiment.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

experiment.last_modified_by.user_profile_name

The name of the user's profile.

experiment.last_modified_time

When the experiment was last modified.

experiment.source

The source of the experiment.

Show child fields
experiment.source.source_arn

The Amazon Resource Name (ARN) of the source.

experiment.source.source_type

The source type.

experiment.tags[]
Show child fields
experiment.tags[].key

The tag key. Tag keys must be unique per resource.

experiment.tags[].value

The tag value.

feature_group

Amazon SageMaker Feature Store stores features in a collection called Feature Group. A Feature Group can be visualized as a table which has rows, with a unique identifier for each row where each column in the table is a feature. In principle, a Feature Group is composed of features and values per features.

STRUCT(
    "feature_group_arn" VARCHAR,
    "feature_group_name" VARCHAR,
    "record_identifier_feature_name" VARCHAR,
    "event_time_feature_name" VARCHAR,
    "feature_definitions" STRUCT(
        "feature_name" VARCHAR,
        "feature_type" VARCHAR,
        "collection_type" VARCHAR,
        "collection_config" STRUCT(
            "vector_config" STRUCT(
                "dimension" BIGINT
            )
        )
    )[],
    "creation_time" TIMESTAMP_S,
    "last_modified_time" TIMESTAMP_S,
    "online_store_config" STRUCT(
        "security_config" STRUCT(
            "kms_key_id" VARCHAR
        ),
        "enable_online_store" BOOLEAN,
        "ttl_duration" STRUCT(
            "unit" VARCHAR,
            "value" BIGINT
        ),
        "storage_type" VARCHAR
    ),
    "offline_store_config" STRUCT(
        "s3_storage_config" STRUCT(
            "s3_uri" VARCHAR,
            "kms_key_id" VARCHAR,
            "resolved_output_s3_uri" VARCHAR
        ),
        "disable_glue_table_creation" BOOLEAN,
        "data_catalog_config" STRUCT(
            "table_name" VARCHAR,
            "catalog" VARCHAR,
            "database" VARCHAR
        ),
        "table_format" VARCHAR
    ),
    "role_arn" VARCHAR,
    "feature_group_status" VARCHAR,
    "offline_store_status" STRUCT(
        "status" VARCHAR,
        "blocked_reason" VARCHAR
    ),
    "last_update_status" STRUCT(
        "status" VARCHAR,
        "failure_reason" VARCHAR
    ),
    "failure_reason" VARCHAR,
    "description" VARCHAR,
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
feature_group.creation_time

The time a FeatureGroup was created.

feature_group.description

A free form description of a FeatureGroup.

feature_group.event_time_feature_name

The name of the feature that stores the EventTime of a Record in a FeatureGroup.

A EventTime is point in time when a new event occurs that corresponds to the creation or update of a Record in FeatureGroup. All Records in the FeatureGroup must have a corresponding EventTime.

feature_group.failure_reason

The reason that the FeatureGroup failed to be replicated in the OfflineStore. This is failure may be due to a failure to create a FeatureGroup in or delete a FeatureGroup from the OfflineStore.

feature_group.feature_definitions[]
Show child fields
feature_group.feature_definitions[].collection_config

Configuration for your collection.

Show child fields
feature_group.feature_definitions[].collection_config.vector_config

Configuration for your vector collection type.

  • Dimension: The number of elements in your vector.

Show child fields
feature_group.feature_definitions[].collection_config.vector_config.dimension

The number of elements in your vector.

feature_group.feature_definitions[].collection_type

A grouping of elements where each element within the collection must have the same feature type (String, Integral, or Fractional).

  • List: An ordered collection of elements.

  • Set: An unordered collection of unique elements.

  • Vector: A specialized list that represents a fixed-size array of elements. The vector dimension is determined by you. Must have elements with fractional feature types.

feature_group.feature_definitions[].feature_name

The name of a feature. The type must be a string. FeatureName cannot be any of the following: is_deleted, write_time, api_invocation_time.

The name:

  • Must start with an alphanumeric character.

  • Can only include alphanumeric characters, underscores, and hyphens. Spaces are not allowed.

feature_group.feature_definitions[].feature_type

The value type of a feature. Valid values are Integral, Fractional, or String.

feature_group.feature_group_arn

The Amazon Resource Name (ARN) of a FeatureGroup.

feature_group.feature_group_name

The name of the FeatureGroup.

feature_group.feature_group_status

A FeatureGroup status.

feature_group.last_modified_time

A timestamp indicating the last time you updated the feature group.

feature_group.last_update_status

A value that indicates whether the feature group was updated successfully.

Show child fields
feature_group.last_update_status.failure_reason

If the update wasn't successful, indicates the reason why it failed.

feature_group.last_update_status.status

A value that indicates whether the update was made successful.

feature_group.offline_store_config

The configuration of an OfflineStore.

Provide an OfflineStoreConfig in a request to CreateFeatureGroup to create an OfflineStore.

To encrypt an OfflineStore using at rest data encryption, specify Amazon Web Services Key Management Service (KMS) key ID, or KMSKeyId, in S3StorageConfig.

Show child fields
feature_group.offline_store_config.data_catalog_config

The meta data of the Glue table that is autogenerated when an OfflineStore is created.

Show child fields
feature_group.offline_store_config.data_catalog_config.catalog

The name of the Glue table catalog.

feature_group.offline_store_config.data_catalog_config.database

The name of the Glue table database.

feature_group.offline_store_config.data_catalog_config.table_name

The name of the Glue table.

feature_group.offline_store_config.disable_glue_table_creation

Set to True to disable the automatic creation of an Amazon Web Services Glue table when configuring an OfflineStore. If set to False, Feature Store will name the OfflineStore Glue table following Athena's naming recommendations.

The default value is False.

feature_group.offline_store_config.s3_storage_config

The Amazon Simple Storage (Amazon S3) location of OfflineStore.

Show child fields
feature_group.offline_store_config.s3_storage_config.kms_key_id

The Amazon Web Services Key Management Service (KMS) key ARN of the key used to encrypt any objects written into the OfflineStore S3 location.

The IAM roleARN that is passed as a parameter to CreateFeatureGroup must have below permissions to the KmsKeyId:

  • "kms:GenerateDataKey"

feature_group.offline_store_config.s3_storage_config.resolved_output_s3_uri

The S3 path where offline records are written.

feature_group.offline_store_config.s3_storage_config.s3_uri

The S3 URI, or location in Amazon S3, of OfflineStore.

S3 URIs have a format similar to the following: s3://example-bucket/prefix/.

feature_group.offline_store_config.table_format

Format for the offline store table. Supported formats are Glue (Default) and Apache Iceberg.

feature_group.offline_store_status

The status of OfflineStore.

Show child fields
feature_group.offline_store_status.blocked_reason

The justification for why the OfflineStoreStatus is Blocked (if applicable).

feature_group.offline_store_status.status

An OfflineStore status.

feature_group.online_store_config

Use this to specify the Amazon Web Services Key Management Service (KMS) Key ID, or KMSKeyId, for at rest data encryption. You can turn OnlineStore on or off by specifying the EnableOnlineStore flag at General Assembly.

The default value is False.

Show child fields
feature_group.online_store_config.enable_online_store

Turn OnlineStore off by specifying False for the EnableOnlineStore flag. Turn OnlineStore on by specifying True for the EnableOnlineStore flag.

The default value is False.

feature_group.online_store_config.security_config

Use to specify KMS Key ID (KMSKeyId) for at-rest encryption of your OnlineStore.

Show child fields
feature_group.online_store_config.security_config.kms_key_id

The Amazon Web Services Key Management Service (KMS) key ARN that SageMaker Feature Store uses to encrypt the Amazon S3 objects at rest using Amazon S3 server-side encryption.

The caller (either user or IAM role) of CreateFeatureGroup must have below permissions to the OnlineStore KmsKeyId:

  • "kms:Encrypt"

  • "kms:Decrypt"

  • "kms:DescribeKey"

  • "kms:CreateGrant"

  • "kms:RetireGrant"

  • "kms:ReEncryptFrom"

  • "kms:ReEncryptTo"

  • "kms:GenerateDataKey"

  • "kms:ListAliases"

  • "kms:ListGrants"

  • "kms:RevokeGrant"

The caller (either user or IAM role) to all DataPlane operations (PutRecord, GetRecord, DeleteRecord) must have the following permissions to the KmsKeyId:

  • "kms:Decrypt"

feature_group.online_store_config.storage_type

Option for different tiers of low latency storage for real-time data retrieval.

  • Standard: A managed low latency data store for feature groups.

  • InMemory: A managed data store for feature groups that supports very low latency retrieval.

feature_group.online_store_config.ttl_duration

Time to live duration, where the record is hard deleted after the expiration time is reached; ExpiresAt = EventTime + TtlDuration. For information on HardDelete, see the DeleteRecord API in the Amazon SageMaker API Reference guide.

Show child fields
feature_group.online_store_config.ttl_duration.unit

TtlDuration time unit.

feature_group.online_store_config.ttl_duration.value

TtlDuration time value.

feature_group.record_identifier_feature_name

The name of the Feature whose value uniquely identifies a Record defined in the FeatureGroup FeatureDefinitions.

feature_group.role_arn

The Amazon Resource Name (ARN) of the IAM execution role used to create the feature group.

feature_group.tags[]
Show child fields
feature_group.tags[].key

The tag key. Tag keys must be unique per resource.

feature_group.tags[].value

The tag value.

feature_metadata

The feature metadata used to search through the features.

STRUCT(
    "feature_group_arn" VARCHAR,
    "feature_group_name" VARCHAR,
    "feature_name" VARCHAR,
    "feature_type" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "last_modified_time" TIMESTAMP_S,
    "description" VARCHAR,
    "parameters" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
feature_metadata.creation_time

A timestamp indicating when the feature was created.

feature_metadata.description

An optional description that you specify to better describe the feature.

feature_metadata.feature_group_arn

The Amazon Resource Number (ARN) of the feature group.

feature_metadata.feature_group_name

The name of the feature group containing the feature.

feature_metadata.feature_name

The name of feature.

feature_metadata.feature_type

The data type of the feature.

feature_metadata.last_modified_time

A timestamp indicating when the feature was last modified.

feature_metadata.parameters[]
Show child fields
feature_metadata.parameters[].key

A key that must contain a value to describe the feature.

feature_metadata.parameters[].value

The value that belongs to a key.

hyper_parameter_tuning_job

The properties of a hyperparameter tuning job.

STRUCT(
    "hyper_parameter_tuning_job_name" VARCHAR,
    "hyper_parameter_tuning_job_arn" VARCHAR,
    "hyper_parameter_tuning_job_config" STRUCT(
        "strategy" VARCHAR,
        "strategy_config" STRUCT(
            "hyperband_strategy_config" STRUCT(
                "min_resource" BIGINT,
                "max_resource" BIGINT
            )
        ),
        "hyper_parameter_tuning_job_objective" STRUCT(
            "type" VARCHAR,
            "metric_name" VARCHAR
        ),
        "resource_limits" STRUCT(
            "max_number_of_training_jobs" BIGINT,
            "max_parallel_training_jobs" BIGINT,
            "max_runtime_in_seconds" BIGINT
        ),
        "parameter_ranges" STRUCT(
            "integer_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "min_value" VARCHAR,
                "max_value" VARCHAR,
                "scaling_type" VARCHAR
            )[],
            "continuous_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "min_value" VARCHAR,
                "max_value" VARCHAR,
                "scaling_type" VARCHAR
            )[],
            "categorical_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "values" VARCHAR[]
            )[],
            "auto_parameters" STRUCT(
                "name" VARCHAR,
                "value_hint" VARCHAR
            )[]
        ),
        "training_job_early_stopping_type" VARCHAR,
        "tuning_job_completion_criteria" STRUCT(
            "target_objective_metric_value" DOUBLE,
            "best_objective_not_improving" STRUCT(
                "max_number_of_training_jobs_not_improving" BIGINT
            ),
            "convergence_detected" STRUCT(
                "complete_on_convergence" VARCHAR
            )
        ),
        "random_seed" BIGINT
    ),
    "training_job_definition" STRUCT(
        "definition_name" VARCHAR,
        "tuning_objective" STRUCT(
            "type" VARCHAR,
            "metric_name" VARCHAR
        ),
        "hyper_parameter_ranges" STRUCT(
            "integer_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "min_value" VARCHAR,
                "max_value" VARCHAR,
                "scaling_type" VARCHAR
            )[],
            "continuous_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "min_value" VARCHAR,
                "max_value" VARCHAR,
                "scaling_type" VARCHAR
            )[],
            "categorical_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "values" VARCHAR[]
            )[],
            "auto_parameters" STRUCT(
                "name" VARCHAR,
                "value_hint" VARCHAR
            )[]
        ),
        "static_hyper_parameters" MAP(VARCHAR, VARCHAR),
        "algorithm_specification" STRUCT(
            "training_image" VARCHAR,
            "training_input_mode" VARCHAR,
            "algorithm_name" VARCHAR,
            "metric_definitions" STRUCT(
                "name" VARCHAR,
                "regex" VARCHAR
            )[]
        ),
        "role_arn" VARCHAR,
        "input_data_config" STRUCT(
            "channel_name" VARCHAR,
            "data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_data_type" VARCHAR,
                    "s3_uri" VARCHAR,
                    "s3_data_distribution_type" VARCHAR,
                    "attribute_names" VARCHAR[],
                    "instance_group_names" VARCHAR[]
                ),
                "file_system_data_source" STRUCT(
                    "file_system_id" VARCHAR,
                    "file_system_access_mode" VARCHAR,
                    "file_system_type" VARCHAR,
                    "directory_path" VARCHAR
                )
            ),
            "content_type" VARCHAR,
            "compression_type" VARCHAR,
            "record_wrapper_type" VARCHAR,
            "input_mode" VARCHAR,
            "shuffle_config" STRUCT(
                "seed" BIGINT
            )
        )[],
        "vpc_config" STRUCT(
            "security_group_ids" VARCHAR[],
            "subnets" VARCHAR[]
        ),
        "output_data_config" STRUCT(
            "kms_key_id" VARCHAR,
            "s3_output_path" VARCHAR,
            "compression_type" VARCHAR
        ),
        "resource_config" STRUCT(
            "instance_type" VARCHAR,
            "instance_count" BIGINT,
            "volume_size_in_gb" BIGINT,
            "volume_kms_key_id" VARCHAR,
            "keep_alive_period_in_seconds" BIGINT,
            "instance_groups" STRUCT(
                "instance_type" VARCHAR,
                "instance_count" BIGINT,
                "instance_group_name" VARCHAR
            )[]
        ),
        "hyper_parameter_tuning_resource_config" STRUCT(
            "instance_type" VARCHAR,
            "instance_count" BIGINT,
            "volume_size_in_gb" BIGINT,
            "volume_kms_key_id" VARCHAR,
            "allocation_strategy" VARCHAR,
            "instance_configs" STRUCT(
                "instance_type" VARCHAR,
                "instance_count" BIGINT,
                "volume_size_in_gb" BIGINT
            )[]
        ),
        "stopping_condition" STRUCT(
            "max_runtime_in_seconds" BIGINT,
            "max_wait_time_in_seconds" BIGINT,
            "max_pending_time_in_seconds" BIGINT
        ),
        "enable_network_isolation" BOOLEAN,
        "enable_inter_container_traffic_encryption" BOOLEAN,
        "enable_managed_spot_training" BOOLEAN,
        "checkpoint_config" STRUCT(
            "s3_uri" VARCHAR,
            "local_path" VARCHAR
        ),
        "retry_strategy" STRUCT(
            "maximum_retry_attempts" BIGINT
        ),
        "environment" MAP(VARCHAR, VARCHAR)
    ),
    "training_job_definitions" STRUCT(
        "definition_name" VARCHAR,
        "tuning_objective" STRUCT(
            "type" VARCHAR,
            "metric_name" VARCHAR
        ),
        "hyper_parameter_ranges" STRUCT(
            "integer_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "min_value" VARCHAR,
                "max_value" VARCHAR,
                "scaling_type" VARCHAR
            )[],
            "continuous_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "min_value" VARCHAR,
                "max_value" VARCHAR,
                "scaling_type" VARCHAR
            )[],
            "categorical_parameter_ranges" STRUCT(
                "name" VARCHAR,
                "values" VARCHAR[]
            )[],
            "auto_parameters" STRUCT(
                "name" VARCHAR,
                "value_hint" VARCHAR
            )[]
        ),
        "static_hyper_parameters" MAP(VARCHAR, VARCHAR),
        "algorithm_specification" STRUCT(
            "training_image" VARCHAR,
            "training_input_mode" VARCHAR,
            "algorithm_name" VARCHAR,
            "metric_definitions" STRUCT(
                "name" VARCHAR,
                "regex" VARCHAR
            )[]
        ),
        "role_arn" VARCHAR,
        "input_data_config" STRUCT(
            "channel_name" VARCHAR,
            "data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_data_type" VARCHAR,
                    "s3_uri" VARCHAR,
                    "s3_data_distribution_type" VARCHAR,
                    "attribute_names" VARCHAR[],
                    "instance_group_names" VARCHAR[]
                ),
                "file_system_data_source" STRUCT(
                    "file_system_id" VARCHAR,
                    "file_system_access_mode" VARCHAR,
                    "file_system_type" VARCHAR,
                    "directory_path" VARCHAR
                )
            ),
            "content_type" VARCHAR,
            "compression_type" VARCHAR,
            "record_wrapper_type" VARCHAR,
            "input_mode" VARCHAR,
            "shuffle_config" STRUCT(
                "seed" BIGINT
            )
        )[],
        "vpc_config" STRUCT(
            "security_group_ids" VARCHAR[],
            "subnets" VARCHAR[]
        ),
        "output_data_config" STRUCT(
            "kms_key_id" VARCHAR,
            "s3_output_path" VARCHAR,
            "compression_type" VARCHAR
        ),
        "resource_config" STRUCT(
            "instance_type" VARCHAR,
            "instance_count" BIGINT,
            "volume_size_in_gb" BIGINT,
            "volume_kms_key_id" VARCHAR,
            "keep_alive_period_in_seconds" BIGINT,
            "instance_groups" STRUCT(
                "instance_type" VARCHAR,
                "instance_count" BIGINT,
                "instance_group_name" VARCHAR
            )[]
        ),
        "hyper_parameter_tuning_resource_config" STRUCT(
            "instance_type" VARCHAR,
            "instance_count" BIGINT,
            "volume_size_in_gb" BIGINT,
            "volume_kms_key_id" VARCHAR,
            "allocation_strategy" VARCHAR,
            "instance_configs" STRUCT(
                "instance_type" VARCHAR,
                "instance_count" BIGINT,
                "volume_size_in_gb" BIGINT
            )[]
        ),
        "stopping_condition" STRUCT(
            "max_runtime_in_seconds" BIGINT,
            "max_wait_time_in_seconds" BIGINT,
            "max_pending_time_in_seconds" BIGINT
        ),
        "enable_network_isolation" BOOLEAN,
        "enable_inter_container_traffic_encryption" BOOLEAN,
        "enable_managed_spot_training" BOOLEAN,
        "checkpoint_config" STRUCT(
            "s3_uri" VARCHAR,
            "local_path" VARCHAR
        ),
        "retry_strategy" STRUCT(
            "maximum_retry_attempts" BIGINT
        ),
        "environment" MAP(VARCHAR, VARCHAR)
    )[],
    "hyper_parameter_tuning_job_status" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "hyper_parameter_tuning_end_time" TIMESTAMP_S,
    "last_modified_time" TIMESTAMP_S,
    "training_job_status_counters" STRUCT(
        "completed" BIGINT,
        "in_progress" BIGINT,
        "retryable_error" BIGINT,
        "non_retryable_error" BIGINT,
        "stopped" BIGINT
    ),
    "objective_status_counters" STRUCT(
        "succeeded" BIGINT,
        "pending" BIGINT,
        "failed" BIGINT
    ),
    "best_training_job" STRUCT(
        "training_job_definition_name" VARCHAR,
        "training_job_name" VARCHAR,
        "training_job_arn" VARCHAR,
        "tuning_job_name" VARCHAR,
        "creation_time" TIMESTAMP_S,
        "training_start_time" TIMESTAMP_S,
        "training_end_time" TIMESTAMP_S,
        "training_job_status" VARCHAR,
        "tuned_hyper_parameters" MAP(VARCHAR, VARCHAR),
        "failure_reason" VARCHAR,
        "final_hyper_parameter_tuning_job_objective_metric" STRUCT(
            "type" VARCHAR,
            "metric_name" VARCHAR,
            "value" DOUBLE
        ),
        "objective_status" VARCHAR
    ),
    "overall_best_training_job" STRUCT(
        "training_job_definition_name" VARCHAR,
        "training_job_name" VARCHAR,
        "training_job_arn" VARCHAR,
        "tuning_job_name" VARCHAR,
        "creation_time" TIMESTAMP_S,
        "training_start_time" TIMESTAMP_S,
        "training_end_time" TIMESTAMP_S,
        "training_job_status" VARCHAR,
        "tuned_hyper_parameters" MAP(VARCHAR, VARCHAR),
        "failure_reason" VARCHAR,
        "final_hyper_parameter_tuning_job_objective_metric" STRUCT(
            "type" VARCHAR,
            "metric_name" VARCHAR,
            "value" DOUBLE
        ),
        "objective_status" VARCHAR
    ),
    "warm_start_config" STRUCT(
        "parent_hyper_parameter_tuning_jobs" STRUCT(
            "hyper_parameter_tuning_job_name" VARCHAR
        )[],
        "warm_start_type" VARCHAR
    ),
    "failure_reason" VARCHAR,
    "tuning_job_completion_details" STRUCT(
        "number_of_training_jobs_objective_not_improving" BIGINT,
        "convergence_detected_time" TIMESTAMP_S
    ),
    "consumed_resources" STRUCT(
        "runtime_in_seconds" BIGINT
    ),
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
hyper_parameter_tuning_job.best_training_job

The container for the summary information about a training job.

Show child fields
hyper_parameter_tuning_job.best_training_job.creation_time

The date and time that the training job was created.

hyper_parameter_tuning_job.best_training_job.failure_reason

The reason that the training job failed.

hyper_parameter_tuning_job.best_training_job.final_hyper_parameter_tuning_job_objective_metric

The FinalHyperParameterTuningJobObjectiveMetric object that specifies the value of the objective metric of the tuning job that launched this training job.

Show child fields
hyper_parameter_tuning_job.best_training_job.final_hyper_parameter_tuning_job_objective_metric.metric_name

The name of the objective metric. For SageMaker built-in algorithms, metrics are defined per algorithm. See the metrics for XGBoost as an example. You can also use a custom algorithm for training and define your own metrics. For more information, see Define metrics and environment variables.

hyper_parameter_tuning_job.best_training_job.final_hyper_parameter_tuning_job_objective_metric.type

Select if you want to minimize or maximize the objective metric during hyperparameter tuning.

hyper_parameter_tuning_job.best_training_job.final_hyper_parameter_tuning_job_objective_metric.value

The value of the objective metric.

hyper_parameter_tuning_job.best_training_job.objective_status

The status of the objective metric for the training job:

  • Succeeded: The final objective metric for the training job was evaluated by the hyperparameter tuning job and used in the hyperparameter tuning process.

  • Pending: The training job is in progress and evaluation of its final objective metric is pending.

  • Failed: The final objective metric for the training job was not evaluated, and was not used in the hyperparameter tuning process. This typically occurs when the training job failed or did not emit an objective metric.

hyper_parameter_tuning_job.best_training_job.training_end_time

Specifies the time when the training job ends on training instances. You are billed for the time interval between the value of TrainingStartTime and this time. For successful jobs and stopped jobs, this is the time after model artifacts are uploaded. For failed jobs, this is the time when SageMaker detects a job failure.

hyper_parameter_tuning_job.best_training_job.training_job_arn

The Amazon Resource Name (ARN) of the training job.

hyper_parameter_tuning_job.best_training_job.training_job_definition_name

The training job definition name.

hyper_parameter_tuning_job.best_training_job.training_job_name

The name of the training job.

hyper_parameter_tuning_job.best_training_job.training_job_status

The status of the training job.

hyper_parameter_tuning_job.best_training_job.training_start_time

The date and time that the training job started.

hyper_parameter_tuning_job.best_training_job.tuned_hyper_parameters

A list of the hyperparameters for which you specified ranges to search.

hyper_parameter_tuning_job.best_training_job.tuning_job_name

The HyperParameter tuning job that launched the training job.

hyper_parameter_tuning_job.consumed_resources

The total amount of resources consumed by a hyperparameter tuning job.

Show child fields
hyper_parameter_tuning_job.consumed_resources.runtime_in_seconds

The wall clock runtime in seconds used by your hyperparameter tuning job.

hyper_parameter_tuning_job.creation_time

The time that a hyperparameter tuning job was created.

hyper_parameter_tuning_job.failure_reason

The error that was created when a hyperparameter tuning job failed.

hyper_parameter_tuning_job.hyper_parameter_tuning_end_time

The time that a hyperparameter tuning job ended.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_arn

The Amazon Resource Name (ARN) of a hyperparameter tuning job.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config

Configures a hyperparameter tuning job.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.hyper_parameter_tuning_job_objective

The HyperParameterTuningJobObjective specifies the objective metric used to evaluate the performance of training jobs launched by this tuning job.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.hyper_parameter_tuning_job_objective.metric_name

The name of the metric to use for the objective metric.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.hyper_parameter_tuning_job_objective.type

Whether to minimize or maximize the objective metric.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges

The ParameterRanges object that specifies the ranges of hyperparameters that this tuning job searches over to find the optimal configuration for the highest model performance against your chosen objective metric.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.auto_parameters[]
Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.auto_parameters[].name

The name of the hyperparameter to optimize using Autotune.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.auto_parameters[].value_hint

An example value of the hyperparameter to optimize using Autotune.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.categorical_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.categorical_parameter_ranges[].name

The name of the categorical hyperparameter to tune.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.categorical_parameter_ranges[].values[]
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.continuous_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.continuous_parameter_ranges[].max_value

The maximum value for the hyperparameter. The tuning job uses floating-point values between MinValue value and this value for tuning.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.continuous_parameter_ranges[].min_value

The minimum value for the hyperparameter. The tuning job uses floating-point values between this value and MaxValuefor tuning.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.continuous_parameter_ranges[].name

The name of the continuous hyperparameter to tune.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.continuous_parameter_ranges[].scaling_type

The scale that hyperparameter tuning uses to search the hyperparameter range. For information about choosing a hyperparameter scale, see Hyperparameter Scaling. One of the following values:

Auto

SageMaker hyperparameter tuning chooses the best scale for the hyperparameter.

Linear

Hyperparameter tuning searches the values in the hyperparameter range by using a linear scale.

Logarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a logarithmic scale.

Logarithmic scaling works only for ranges that have only values greater than 0.

ReverseLogarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a reverse logarithmic scale.

Reverse logarithmic scaling works only for ranges that are entirely within the range 0<=x<1.0.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.integer_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.integer_parameter_ranges[].max_value

The maximum value of the hyperparameter to search.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.integer_parameter_ranges[].min_value

The minimum value of the hyperparameter to search.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.integer_parameter_ranges[].name

The name of the hyperparameter to search.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.parameter_ranges.integer_parameter_ranges[].scaling_type

The scale that hyperparameter tuning uses to search the hyperparameter range. For information about choosing a hyperparameter scale, see Hyperparameter Scaling. One of the following values:

Auto

SageMaker hyperparameter tuning chooses the best scale for the hyperparameter.

Linear

Hyperparameter tuning searches the values in the hyperparameter range by using a linear scale.

Logarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a logarithmic scale.

Logarithmic scaling works only for ranges that have only values greater than 0.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.random_seed

A value used to initialize a pseudo-random number generator. Setting a random seed and using the same seed later for the same tuning job will allow hyperparameter optimization to find more a consistent hyperparameter configuration between the two runs.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.resource_limits

The ResourceLimits object that specifies the maximum number of training and parallel training jobs that can be used for this hyperparameter tuning job.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.resource_limits.max_number_of_training_jobs

The maximum number of training jobs that a hyperparameter tuning job can launch.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.resource_limits.max_parallel_training_jobs

The maximum number of concurrent training jobs that a hyperparameter tuning job can launch.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.resource_limits.max_runtime_in_seconds

The maximum time in seconds that a hyperparameter tuning job can run.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.strategy

Specifies how hyperparameter tuning chooses the combinations of hyperparameter values to use for the training job it launches. For information about search strategies, see How Hyperparameter Tuning Works.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.strategy_config

The configuration for the Hyperband optimization strategy. This parameter should be provided only if Hyperband is selected as the strategy for HyperParameterTuningJobConfig.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.strategy_config.hyperband_strategy_config

The configuration for the object that specifies the Hyperband strategy. This parameter is only supported for the Hyperband selection for Strategy within the HyperParameterTuningJobConfig API.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.strategy_config.hyperband_strategy_config.max_resource

The maximum number of resources (such as epochs) that can be used by a training job launched by a hyperparameter tuning job. Once a job reaches the MaxResource value, it is stopped. If a value for MaxResource is not provided, and Hyperband is selected as the hyperparameter tuning strategy, HyperbandTraining attempts to infer MaxResource from the following keys (if present) in StaticsHyperParameters:

  • epochs

  • numepochs

  • n-epochs

  • n_epochs

  • num_epochs

If HyperbandStrategyConfig is unable to infer a value for MaxResource, it generates a validation error. The maximum value is 20,000 epochs. All metrics that correspond to an objective metric are used to derive early stopping decisions. For distributed training jobs, ensure that duplicate metrics are not printed in the logs across the individual nodes in a training job. If multiple nodes are publishing duplicate or incorrect metrics, training jobs may make an incorrect stopping decision and stop the job prematurely.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.strategy_config.hyperband_strategy_config.min_resource

The minimum number of resources (such as epochs) that can be used by a training job launched by a hyperparameter tuning job. If the value for MinResource has not been reached, the training job is not stopped by Hyperband.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.training_job_early_stopping_type

Specifies whether to use early stopping for training jobs launched by the hyperparameter tuning job. Because the Hyperband strategy has its own advanced internal early stopping mechanism, TrainingJobEarlyStoppingType must be OFF to use Hyperband. This parameter can take on one of the following values (the default value is OFF):

OFF

Training jobs launched by the hyperparameter tuning job do not use early stopping.

AUTO

SageMaker stops training jobs launched by the hyperparameter tuning job when they are unlikely to perform better than previously completed training jobs. For more information, see Stop Training Jobs Early.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.tuning_job_completion_criteria

The tuning job's completion criteria.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.tuning_job_completion_criteria.best_objective_not_improving

A flag to stop your hyperparameter tuning job if model performance fails to improve as evaluated against an objective function.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.tuning_job_completion_criteria.best_objective_not_improving.max_number_of_training_jobs_not_improving

The number of training jobs that have failed to improve model performance by 1% or greater over prior training jobs as evaluated against an objective function.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.tuning_job_completion_criteria.convergence_detected

A flag to top your hyperparameter tuning job if automatic model tuning (AMT) has detected that your model has converged as evaluated against your objective function.

Show child fields
hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.tuning_job_completion_criteria.convergence_detected.complete_on_convergence

A flag to stop a tuning job once AMT has detected that the job has converged.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_config.tuning_job_completion_criteria.target_objective_metric_value

The value of the objective metric.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_name

The name of a hyperparameter tuning job.

hyper_parameter_tuning_job.hyper_parameter_tuning_job_status

The status of a hyperparameter tuning job.

hyper_parameter_tuning_job.last_modified_time

The time that a hyperparameter tuning job was last modified.

hyper_parameter_tuning_job.objective_status_counters

Specifies the number of training jobs that this hyperparameter tuning job launched, categorized by the status of their objective metric. The objective metric status shows whether the final objective metric for the training job has been evaluated by the tuning job and used in the hyperparameter tuning process.

Show child fields
hyper_parameter_tuning_job.objective_status_counters.failed

The number of training jobs whose final objective metric was not evaluated and used in the hyperparameter tuning process. This typically occurs when the training job failed or did not emit an objective metric.

hyper_parameter_tuning_job.objective_status_counters.pending

The number of training jobs that are in progress and pending evaluation of their final objective metric.

hyper_parameter_tuning_job.objective_status_counters.succeeded

The number of training jobs whose final objective metric was evaluated by the hyperparameter tuning job and used in the hyperparameter tuning process.

hyper_parameter_tuning_job.overall_best_training_job

The container for the summary information about a training job.

Show child fields
hyper_parameter_tuning_job.overall_best_training_job.creation_time

The date and time that the training job was created.

hyper_parameter_tuning_job.overall_best_training_job.failure_reason

The reason that the training job failed.

hyper_parameter_tuning_job.overall_best_training_job.final_hyper_parameter_tuning_job_objective_metric

The FinalHyperParameterTuningJobObjectiveMetric object that specifies the value of the objective metric of the tuning job that launched this training job.

Show child fields
hyper_parameter_tuning_job.overall_best_training_job.final_hyper_parameter_tuning_job_objective_metric.metric_name

The name of the objective metric. For SageMaker built-in algorithms, metrics are defined per algorithm. See the metrics for XGBoost as an example. You can also use a custom algorithm for training and define your own metrics. For more information, see Define metrics and environment variables.

hyper_parameter_tuning_job.overall_best_training_job.final_hyper_parameter_tuning_job_objective_metric.type

Select if you want to minimize or maximize the objective metric during hyperparameter tuning.

hyper_parameter_tuning_job.overall_best_training_job.final_hyper_parameter_tuning_job_objective_metric.value

The value of the objective metric.

hyper_parameter_tuning_job.overall_best_training_job.objective_status

The status of the objective metric for the training job:

  • Succeeded: The final objective metric for the training job was evaluated by the hyperparameter tuning job and used in the hyperparameter tuning process.

  • Pending: The training job is in progress and evaluation of its final objective metric is pending.

  • Failed: The final objective metric for the training job was not evaluated, and was not used in the hyperparameter tuning process. This typically occurs when the training job failed or did not emit an objective metric.

hyper_parameter_tuning_job.overall_best_training_job.training_end_time

Specifies the time when the training job ends on training instances. You are billed for the time interval between the value of TrainingStartTime and this time. For successful jobs and stopped jobs, this is the time after model artifacts are uploaded. For failed jobs, this is the time when SageMaker detects a job failure.

hyper_parameter_tuning_job.overall_best_training_job.training_job_arn

The Amazon Resource Name (ARN) of the training job.

hyper_parameter_tuning_job.overall_best_training_job.training_job_definition_name

The training job definition name.

hyper_parameter_tuning_job.overall_best_training_job.training_job_name

The name of the training job.

hyper_parameter_tuning_job.overall_best_training_job.training_job_status

The status of the training job.

hyper_parameter_tuning_job.overall_best_training_job.training_start_time

The date and time that the training job started.

hyper_parameter_tuning_job.overall_best_training_job.tuned_hyper_parameters

A list of the hyperparameters for which you specified ranges to search.

hyper_parameter_tuning_job.overall_best_training_job.tuning_job_name

The HyperParameter tuning job that launched the training job.

hyper_parameter_tuning_job.tags[]
Show child fields
hyper_parameter_tuning_job.tags[].key

The tag key. Tag keys must be unique per resource.

hyper_parameter_tuning_job.tags[].value

The tag value.

hyper_parameter_tuning_job.training_job_definition

Defines the training jobs launched by a hyperparameter tuning job.

Show child fields
hyper_parameter_tuning_job.training_job_definition.algorithm_specification

The HyperParameterAlgorithmSpecification object that specifies the resource algorithm to use for the training jobs that the tuning job launches.

Show child fields
hyper_parameter_tuning_job.training_job_definition.algorithm_specification.algorithm_name

The name of the resource algorithm to use for the hyperparameter tuning job. If you specify a value for this parameter, do not specify a value for TrainingImage.

hyper_parameter_tuning_job.training_job_definition.algorithm_specification.metric_definitions[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.algorithm_specification.metric_definitions[].name

The name of the metric.

hyper_parameter_tuning_job.training_job_definition.algorithm_specification.metric_definitions[].regex

A regular expression that searches the output of a training job and gets the value of the metric. For more information about using regular expressions to define metrics, see Defining metrics and environment variables.

hyper_parameter_tuning_job.training_job_definition.algorithm_specification.training_image

The registry path of the Docker image that contains the training algorithm. For information about Docker registry paths for built-in algorithms, see Algorithms Provided by Amazon SageMaker: Common Parameters. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information, see Using Your Own Algorithms with Amazon SageMaker.

hyper_parameter_tuning_job.training_job_definition.algorithm_specification.training_input_mode

The training input mode that the algorithm supports. For more information about input modes, see Algorithms.

Pipe mode

If an algorithm supports Pipe mode, Amazon SageMaker streams data directly from Amazon S3 to the container.

File mode

If an algorithm supports File mode, SageMaker downloads the training data from S3 to the provisioned ML storage volume, and mounts the directory to the Docker volume for the training container.

You must provision the ML storage volume with sufficient capacity to accommodate the data downloaded from S3. In addition to the training data, the ML storage volume also stores the output model. The algorithm container uses the ML storage volume to also store intermediate information, if any.

For distributed algorithms, training data is distributed uniformly. Your training duration is predictable if the input data objects sizes are approximately the same. SageMaker does not split the files any further for model training. If the object sizes are skewed, training won't be optimal as the data distribution is also skewed when one host in a training cluster is overloaded, thus becoming a bottleneck in training.

FastFile mode

If an algorithm supports FastFile mode, SageMaker streams data directly from S3 to the container with no code changes, and provides file system access to the data. Users can author their training script to interact with these files as if they were stored on disk.

FastFile mode works best when the data is read sequentially. Augmented manifest files aren't supported. The startup time is lower when there are fewer files in the S3 bucket provided.

hyper_parameter_tuning_job.training_job_definition.checkpoint_config

Contains information about the output location for managed spot training checkpoint data.

Show child fields
hyper_parameter_tuning_job.training_job_definition.checkpoint_config.local_path

(Optional) The local directory where checkpoints are written. The default directory is /opt/ml/checkpoints/.

hyper_parameter_tuning_job.training_job_definition.checkpoint_config.s3_uri

Identifies the S3 path where you want SageMaker to store checkpoints. For example, s3://bucket-name/key-name-prefix.

hyper_parameter_tuning_job.training_job_definition.definition_name

The job definition name.

hyper_parameter_tuning_job.training_job_definition.enable_inter_container_traffic_encryption

To encrypt all communications between ML compute instances in distributed training, choose True. Encryption provides greater security for distributed training, but training might take longer. How long it takes depends on the amount of communication between compute instances, especially if you use a deep learning algorithm in distributed training.

hyper_parameter_tuning_job.training_job_definition.enable_managed_spot_training

A Boolean indicating whether managed spot training is enabled (True) or not (False).

hyper_parameter_tuning_job.training_job_definition.enable_network_isolation

Isolates the training container. No inbound or outbound network calls can be made, except for calls between peers within a training cluster for distributed training. If network isolation is used for training jobs that are configured to use a VPC, SageMaker downloads and uploads customer data and model artifacts through the specified VPC, but the training container does not have network access.

hyper_parameter_tuning_job.training_job_definition.environment

An environment variable that you can pass into the SageMaker CreateTrainingJob API. You can use an existing environment variable from the training container or use your own. See Define metrics and variables for more information.

The maximum number of items specified for Map Entries refers to the maximum number of environment variables for each TrainingJobDefinition and also the maximum for the hyperparameter tuning job itself. That is, the sum of the number of environment variables for all the training job definitions can't exceed the maximum number specified.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges

Specifies ranges of integer, continuous, and categorical hyperparameters that a hyperparameter tuning job searches. The hyperparameter tuning job launches training jobs with hyperparameter values within these ranges to find the combination of values that result in the training job with the best performance as measured by the objective metric of the hyperparameter tuning job.

The maximum number of items specified for Array Members refers to the maximum number of hyperparameters for each range and also the maximum for the hyperparameter tuning job itself. That is, the sum of the number of hyperparameters for all the ranges can't exceed the maximum number specified.

Show child fields
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.auto_parameters[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.auto_parameters[].name

The name of the hyperparameter to optimize using Autotune.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.auto_parameters[].value_hint

An example value of the hyperparameter to optimize using Autotune.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.categorical_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.categorical_parameter_ranges[].name

The name of the categorical hyperparameter to tune.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.categorical_parameter_ranges[].values[]
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.continuous_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.continuous_parameter_ranges[].max_value

The maximum value for the hyperparameter. The tuning job uses floating-point values between MinValue value and this value for tuning.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.continuous_parameter_ranges[].min_value

The minimum value for the hyperparameter. The tuning job uses floating-point values between this value and MaxValuefor tuning.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.continuous_parameter_ranges[].name

The name of the continuous hyperparameter to tune.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.continuous_parameter_ranges[].scaling_type

The scale that hyperparameter tuning uses to search the hyperparameter range. For information about choosing a hyperparameter scale, see Hyperparameter Scaling. One of the following values:

Auto

SageMaker hyperparameter tuning chooses the best scale for the hyperparameter.

Linear

Hyperparameter tuning searches the values in the hyperparameter range by using a linear scale.

Logarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a logarithmic scale.

Logarithmic scaling works only for ranges that have only values greater than 0.

ReverseLogarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a reverse logarithmic scale.

Reverse logarithmic scaling works only for ranges that are entirely within the range 0<=x<1.0.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.integer_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.integer_parameter_ranges[].max_value

The maximum value of the hyperparameter to search.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.integer_parameter_ranges[].min_value

The minimum value of the hyperparameter to search.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.integer_parameter_ranges[].name

The name of the hyperparameter to search.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_ranges.integer_parameter_ranges[].scaling_type

The scale that hyperparameter tuning uses to search the hyperparameter range. For information about choosing a hyperparameter scale, see Hyperparameter Scaling. One of the following values:

Auto

SageMaker hyperparameter tuning chooses the best scale for the hyperparameter.

Linear

Hyperparameter tuning searches the values in the hyperparameter range by using a linear scale.

Logarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a logarithmic scale.

Logarithmic scaling works only for ranges that have only values greater than 0.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config

The configuration for the hyperparameter tuning resources, including the compute instances and storage volumes, used for training jobs launched by the tuning job. By default, storage volumes hold model artifacts and incremental states. Choose File for TrainingInputMode in the AlgorithmSpecification parameter to additionally store training data in the storage volume (optional).

Show child fields
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.allocation_strategy

The strategy that determines the order of preference for resources specified in InstanceConfigs used in hyperparameter optimization.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.instance_configs[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.instance_configs[].instance_count

The number of instances of the type specified by InstanceType. Choose an instance count larger than 1 for distributed training algorithms. See Step 2: Launch a SageMaker Distributed Training Job Using the SageMaker Python SDK for more information.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.instance_configs[].instance_type

The instance type used for processing of hyperparameter optimization jobs. Choose from general purpose (no GPUs) instance types: ml.m5.xlarge, ml.m5.2xlarge, and ml.m5.4xlarge or compute optimized (no GPUs) instance types: ml.c5.xlarge and ml.c5.2xlarge. For more information about instance types, see instance type descriptions.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.instance_configs[].volume_size_in_gb

The volume size in GB of the data to be processed for hyperparameter optimization (optional).

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.instance_count

The number of compute instances of type InstanceType to use. For distributed training, select a value greater than 1.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.instance_type

The instance type used to run hyperparameter optimization tuning jobs. See descriptions of instance types for more information.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.volume_kms_key_id

A key used by Amazon Web Services Key Management Service to encrypt data on the storage volume attached to the compute instances used to run the training job. You can use either of the following formats to specify a key.

KMS Key ID:

"1234abcd-12ab-34cd-56ef-1234567890ab"

Amazon Resource Name (ARN) of a KMS key:

"arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

Some instances use local storage, which use a hardware module to encrypt storage volumes. If you choose one of these instance types, you cannot request a VolumeKmsKeyId. For a list of instance types that use local storage, see instance store volumes. For more information about Amazon Web Services Key Management Service, see KMS encryption for more information.

hyper_parameter_tuning_job.training_job_definition.hyper_parameter_tuning_resource_config.volume_size_in_gb

The volume size in GB for the storage volume to be used in processing hyperparameter optimization jobs (optional). These volumes store model artifacts, incremental states and optionally, scratch space for training algorithms. Do not provide a value for this parameter if a value for InstanceConfigs is also specified.

Some instance types have a fixed total local storage size. If you select one of these instances for training, VolumeSizeInGB cannot be greater than this total size. For a list of instance types with local instance storage and their sizes, see instance store volumes.

SageMaker supports only the General Purpose SSD (gp2) storage volume type.

hyper_parameter_tuning_job.training_job_definition.input_data_config[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.input_data_config[].channel_name

The name of the channel.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].compression_type

If training data is compressed, the compression type. The default value is None. CompressionType is used only in Pipe input mode. In File mode, leave this field unset or set it to None.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].content_type

The MIME type of the data.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source

The location of the channel data.

Show child fields
hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.file_system_data_source

The file system that is associated with a channel.

Show child fields
hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.file_system_data_source.directory_path

The full path to the directory to associate with the channel.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.file_system_data_source.file_system_access_mode

The access mode of the mount of the directory associated with the channel. A directory can be mounted either in ro (read-only) or rw (read-write) mode.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.file_system_data_source.file_system_id

The file system id.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.file_system_data_source.file_system_type

The file system type.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.s3_data_source

The S3 location of the data source that is associated with a channel.

Show child fields
hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.s3_data_source.attribute_names[]
hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.s3_data_source.instance_group_names[]
hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.s3_data_source.s3_data_distribution_type

If you want SageMaker to replicate the entire dataset on each ML compute instance that is launched for model training, specify FullyReplicated.

If you want SageMaker to replicate a subset of data on each ML compute instance that is launched for model training, specify ShardedByS3Key. If there are n ML compute instances launched for a training job, each instance gets approximately 1/n of the number of S3 objects. In this case, model training on each machine uses only the subset of training data.

Don't choose more ML compute instances for training than available S3 objects. If you do, some nodes won't get any data and you will pay for nodes that aren't getting any training data. This applies in both File and Pipe modes. Keep this in mind when developing algorithms.

In distributed training, where you use multiple ML compute EC2 instances, you might choose ShardedByS3Key. If the algorithm requires copying training data to the ML storage volume (when TrainingInputMode is set to File), this copies 1/n of the number of objects.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.s3_data_source.s3_data_type

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix for model training.

If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want SageMaker to use for model training.

If you choose AugmentedManifestFile, S3Uri identifies an object that is an augmented manifest file in JSON lines format. This file contains the data you want to use for model training. AugmentedManifestFile can only be used if the Channel's input mode is Pipe.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].data_source.s3_data_source.s3_uri

Depending on the value specified for the S3DataType, identifies either a key name prefix or a manifest. For example:

  • A key name prefix might look like this: s3://bucketname/exampleprefix/

  • A manifest might look like this: s3://bucketname/example.manifest

    A manifest is an S3 object which is a JSON file consisting of an array of elements. The first element is a prefix which is followed by one or more suffixes. SageMaker appends the suffix elements to the prefix to get a full set of S3Uri. Note that the prefix must be a valid non-empty S3Uri that precludes users from specifying a manifest whose individual S3Uri is sourced from different S3 buckets.

    The following code example shows a valid manifest format:

    [ {"prefix": "s3://customer_bucket/some/prefix/"},

    "relative/path/to/custdata-1",

    "relative/path/custdata-2",

    ...

    "relative/path/custdata-N"

    ]

    This JSON is equivalent to the following S3Uri list:

    s3://customer_bucket/some/prefix/relative/path/to/custdata-1

    s3://customer_bucket/some/prefix/relative/path/custdata-2

    ...

    s3://customer_bucket/some/prefix/relative/path/custdata-N

    The complete set of S3Uri in this manifest is the input data for the channel for this data source. The object that each S3Uri points to must be readable by the IAM role that SageMaker uses to perform tasks on your behalf.

Your input bucket must be located in same Amazon Web Services region as your training job.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].input_mode

(Optional) The input mode to use for the data channel in a training job. If you don't set a value for InputMode, SageMaker uses the value set for TrainingInputMode. Use this parameter to override the TrainingInputMode setting in a AlgorithmSpecification request when you have a channel that needs a different input mode from the training job's general setting. To download the data from Amazon Simple Storage Service (Amazon S3) to the provisioned ML storage volume, and mount the directory to a Docker volume, use File input mode. To stream data directly from Amazon S3 to the container, choose Pipe input mode.

To use a model for incremental training, choose File input model.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].record_wrapper_type

Specify RecordIO as the value when input data is in raw format but the training algorithm requires the RecordIO format. In this case, SageMaker wraps each individual S3 object in a RecordIO record. If the input data is already in RecordIO format, you don't need to set this attribute. For more information, see Create a Dataset Using RecordIO.

In File mode, leave this field unset or set it to None.

hyper_parameter_tuning_job.training_job_definition.input_data_config[].shuffle_config

A configuration for a shuffle option for input data in a channel. If you use S3Prefix for S3DataType, this shuffles the results of the S3 key prefix matches. If you use ManifestFile, the order of the S3 object references in the ManifestFile is shuffled. If you use AugmentedManifestFile, the order of the JSON lines in the AugmentedManifestFile is shuffled. The shuffling order is determined using the Seed value.

For Pipe input mode, shuffling is done at the start of every epoch. With large datasets this ensures that the order of the training data is different for each epoch, it helps reduce bias and possible overfitting. In a multi-node training job when ShuffleConfig is combined with S3DataDistributionType of ShardedByS3Key, the data is shuffled across nodes so that the content sent to a particular node on the first epoch might be sent to a different node on the second epoch.

Show child fields
hyper_parameter_tuning_job.training_job_definition.input_data_config[].shuffle_config.seed

Determines the shuffling order in ShuffleConfig value.

hyper_parameter_tuning_job.training_job_definition.output_data_config

Specifies the path to the Amazon S3 bucket where you store model artifacts from the training jobs that the tuning job launches.

Show child fields
hyper_parameter_tuning_job.training_job_definition.output_data_config.compression_type

The model output compression type. Select None to output an uncompressed model, recommended for large model outputs. Defaults to gzip.

hyper_parameter_tuning_job.training_job_definition.output_data_config.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption. The KmsKeyId can be any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

  • // KMS Key Alias

    "alias/ExampleAlias"

  • // Amazon Resource Name (ARN) of a KMS Key Alias

    "arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias"

If you use a KMS key ID or an alias of your KMS key, the SageMaker execution role must include permissions to call kms:Encrypt. If you don't provide a KMS key ID, SageMaker uses the default KMS key for Amazon S3 for your role's account. For more information, see KMS-Managed Encryption Keys in the Amazon Simple Storage Service Developer Guide. If the output data is stored in Amazon S3 Express One Zone, it is encrypted with server-side encryption with Amazon S3 managed keys (SSE-S3). KMS key is not supported for Amazon S3 Express One Zone

The KMS key policy must grant permission to the IAM role that you specify in your CreateTrainingJob, CreateTransformJob, or CreateHyperParameterTuningJob requests. For more information, see Using Key Policies in Amazon Web Services KMS in the Amazon Web Services Key Management Service Developer Guide.

hyper_parameter_tuning_job.training_job_definition.output_data_config.s3_output_path

Identifies the S3 path where you want SageMaker to store the model artifacts. For example, s3://bucket-name/key-name-prefix.

hyper_parameter_tuning_job.training_job_definition.resource_config

The resources, including the compute instances and storage volumes, to use for the training jobs that the tuning job launches.

Storage volumes store model artifacts and incremental states. Training algorithms might also use storage volumes for scratch space. If you want SageMaker to use the storage volume to store the training data, choose File as the TrainingInputMode in the algorithm specification. For distributed training algorithms, specify an instance count greater than 1.

If you want to use hyperparameter optimization with instance type flexibility, use HyperParameterTuningResourceConfig instead.

Show child fields
hyper_parameter_tuning_job.training_job_definition.resource_config.instance_count

The number of ML compute instances to use. For distributed training, provide a value greater than 1.

hyper_parameter_tuning_job.training_job_definition.resource_config.instance_groups[]
Show child fields
hyper_parameter_tuning_job.training_job_definition.resource_config.instance_groups[].instance_count

Specifies the number of instances of the instance group.

hyper_parameter_tuning_job.training_job_definition.resource_config.instance_groups[].instance_group_name

Specifies the name of the instance group.

hyper_parameter_tuning_job.training_job_definition.resource_config.instance_groups[].instance_type

Specifies the instance type of the instance group.

hyper_parameter_tuning_job.training_job_definition.resource_config.instance_type

The ML compute instance type.

SageMaker Training on Amazon Elastic Compute Cloud (EC2) P4de instances is in preview release starting December 9th, 2022.

Amazon EC2 P4de instances (currently in preview) are powered by 8 NVIDIA A100 GPUs with 80GB high-performance HBM2e GPU memory, which accelerate the speed of training ML models that need to be trained on large datasets of high-resolution data. In this preview release, Amazon SageMaker supports ML training jobs on P4de instances (ml.p4de.24xlarge) to reduce model training time. The ml.p4de.24xlarge instances are available in the following Amazon Web Services Regions.

  • US East (N. Virginia) (us-east-1)

  • US West (Oregon) (us-west-2)

To request quota limit increase and start using P4de instances, contact the SageMaker Training service team through your account team.

hyper_parameter_tuning_job.training_job_definition.resource_config.keep_alive_period_in_seconds

The duration of time in seconds to retain configured resources in a warm pool for subsequent training jobs.

hyper_parameter_tuning_job.training_job_definition.resource_config.volume_kms_key_id

The Amazon Web Services KMS key that SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) that run the training job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

The VolumeKmsKeyId can be in any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

hyper_parameter_tuning_job.training_job_definition.resource_config.volume_size_in_gb

The size of the ML storage volume that you want to provision.

ML storage volumes store model artifacts and incremental states. Training algorithms might also use the ML storage volume for scratch space. If you want to store the training data in the ML storage volume, choose File as the TrainingInputMode in the algorithm specification.

When using an ML instance with NVMe SSD volumes, SageMaker doesn't provision Amazon EBS General Purpose SSD (gp2) storage. Available storage is fixed to the NVMe-type instance's storage capacity. SageMaker configures storage paths for training datasets, checkpoints, model artifacts, and outputs to use the entire capacity of the instance storage. For example, ML instance families with the NVMe-type instance storage include ml.p4d, ml.g4dn, and ml.g5.

When using an ML instance with the EBS-only storage option and without instance storage, you must define the size of EBS volume through VolumeSizeInGB in the ResourceConfig API. For example, ML instance families that use EBS volumes include ml.c5 and ml.p2.

To look up instance types and their instance storage types and volumes, see Amazon EC2 Instance Types.

To find the default local paths defined by the SageMaker training platform, see Amazon SageMaker Training Storage Folders for Training Datasets, Checkpoints, Model Artifacts, and Outputs.

hyper_parameter_tuning_job.training_job_definition.retry_strategy

The number of times to retry the job when the job fails due to an InternalServerError.

Show child fields
hyper_parameter_tuning_job.training_job_definition.retry_strategy.maximum_retry_attempts

The number of times to retry the job. When the job is retried, it's SecondaryStatus is changed to STARTING.

hyper_parameter_tuning_job.training_job_definition.role_arn

The Amazon Resource Name (ARN) of the IAM role associated with the training jobs that the tuning job launches.

hyper_parameter_tuning_job.training_job_definition.static_hyper_parameters

Specifies the values of hyperparameters that do not change for the tuning job.

hyper_parameter_tuning_job.training_job_definition.stopping_condition

Specifies a limit to how long a model hyperparameter training job can run. It also specifies how long a managed spot training job has to complete. When the job reaches the time limit, SageMaker ends the training job. Use this API to cap model training costs.

Show child fields
hyper_parameter_tuning_job.training_job_definition.stopping_condition.max_pending_time_in_seconds

The maximum length of time, in seconds, that a training or compilation job can be pending before it is stopped.

hyper_parameter_tuning_job.training_job_definition.stopping_condition.max_runtime_in_seconds

The maximum length of time, in seconds, that a training or compilation job can run before it is stopped.

For compilation jobs, if the job does not complete during this time, a TimeOut error is generated. We recommend starting with 900 seconds and increasing as necessary based on your model.

For all other jobs, if the job does not complete during this time, SageMaker ends the job. When RetryStrategy is specified in the job request, MaxRuntimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt. The default value is 1 day. The maximum value is 28 days.

The maximum time that a TrainingJob can run in total, including any time spent publishing metrics or archiving and uploading models after it has been stopped, is 30 days.

hyper_parameter_tuning_job.training_job_definition.stopping_condition.max_wait_time_in_seconds

The maximum length of time, in seconds, that a managed Spot training job has to complete. It is the amount of time spent waiting for Spot capacity plus the amount of time the job can run. It must be equal to or greater than MaxRuntimeInSeconds. If the job does not complete during this time, SageMaker ends the job.

When RetryStrategy is specified in the job request, MaxWaitTimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt.

hyper_parameter_tuning_job.training_job_definition.tuning_objective

Defines the objective metric for a hyperparameter tuning job. Hyperparameter tuning uses the value of this metric to evaluate the training jobs it launches, and returns the training job that results in either the highest or lowest value for this metric, depending on the value you specify for the Type parameter. If you want to define a custom objective metric, see Define metrics and environment variables.

Show child fields
hyper_parameter_tuning_job.training_job_definition.tuning_objective.metric_name

The name of the metric to use for the objective metric.

hyper_parameter_tuning_job.training_job_definition.tuning_objective.type

Whether to minimize or maximize the objective metric.

hyper_parameter_tuning_job.training_job_definition.vpc_config

The VpcConfig object that specifies the VPC that you want the training jobs that this hyperparameter tuning job launches to connect to. Control access to and from your training container by configuring the VPC. For more information, see Protect Training Jobs by Using an Amazon Virtual Private Cloud.

Show child fields
hyper_parameter_tuning_job.training_job_definition.vpc_config.security_group_ids[]
hyper_parameter_tuning_job.training_job_definition.vpc_config.subnets[]
hyper_parameter_tuning_job.training_job_definitions[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].algorithm_specification

The HyperParameterAlgorithmSpecification object that specifies the resource algorithm to use for the training jobs that the tuning job launches.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].algorithm_specification.algorithm_name

The name of the resource algorithm to use for the hyperparameter tuning job. If you specify a value for this parameter, do not specify a value for TrainingImage.

hyper_parameter_tuning_job.training_job_definitions[].algorithm_specification.metric_definitions[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].algorithm_specification.metric_definitions[].name

The name of the metric.

hyper_parameter_tuning_job.training_job_definitions[].algorithm_specification.metric_definitions[].regex

A regular expression that searches the output of a training job and gets the value of the metric. For more information about using regular expressions to define metrics, see Defining metrics and environment variables.

hyper_parameter_tuning_job.training_job_definitions[].algorithm_specification.training_image

The registry path of the Docker image that contains the training algorithm. For information about Docker registry paths for built-in algorithms, see Algorithms Provided by Amazon SageMaker: Common Parameters. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information, see Using Your Own Algorithms with Amazon SageMaker.

hyper_parameter_tuning_job.training_job_definitions[].algorithm_specification.training_input_mode

The training input mode that the algorithm supports. For more information about input modes, see Algorithms.

Pipe mode

If an algorithm supports Pipe mode, Amazon SageMaker streams data directly from Amazon S3 to the container.

File mode

If an algorithm supports File mode, SageMaker downloads the training data from S3 to the provisioned ML storage volume, and mounts the directory to the Docker volume for the training container.

You must provision the ML storage volume with sufficient capacity to accommodate the data downloaded from S3. In addition to the training data, the ML storage volume also stores the output model. The algorithm container uses the ML storage volume to also store intermediate information, if any.

For distributed algorithms, training data is distributed uniformly. Your training duration is predictable if the input data objects sizes are approximately the same. SageMaker does not split the files any further for model training. If the object sizes are skewed, training won't be optimal as the data distribution is also skewed when one host in a training cluster is overloaded, thus becoming a bottleneck in training.

FastFile mode

If an algorithm supports FastFile mode, SageMaker streams data directly from S3 to the container with no code changes, and provides file system access to the data. Users can author their training script to interact with these files as if they were stored on disk.

FastFile mode works best when the data is read sequentially. Augmented manifest files aren't supported. The startup time is lower when there are fewer files in the S3 bucket provided.

hyper_parameter_tuning_job.training_job_definitions[].checkpoint_config

Contains information about the output location for managed spot training checkpoint data.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].checkpoint_config.local_path

(Optional) The local directory where checkpoints are written. The default directory is /opt/ml/checkpoints/.

hyper_parameter_tuning_job.training_job_definitions[].checkpoint_config.s3_uri

Identifies the S3 path where you want SageMaker to store checkpoints. For example, s3://bucket-name/key-name-prefix.

hyper_parameter_tuning_job.training_job_definitions[].definition_name

The job definition name.

hyper_parameter_tuning_job.training_job_definitions[].enable_inter_container_traffic_encryption

To encrypt all communications between ML compute instances in distributed training, choose True. Encryption provides greater security for distributed training, but training might take longer. How long it takes depends on the amount of communication between compute instances, especially if you use a deep learning algorithm in distributed training.

hyper_parameter_tuning_job.training_job_definitions[].enable_managed_spot_training

A Boolean indicating whether managed spot training is enabled (True) or not (False).

hyper_parameter_tuning_job.training_job_definitions[].enable_network_isolation

Isolates the training container. No inbound or outbound network calls can be made, except for calls between peers within a training cluster for distributed training. If network isolation is used for training jobs that are configured to use a VPC, SageMaker downloads and uploads customer data and model artifacts through the specified VPC, but the training container does not have network access.

hyper_parameter_tuning_job.training_job_definitions[].environment

An environment variable that you can pass into the SageMaker CreateTrainingJob API. You can use an existing environment variable from the training container or use your own. See Define metrics and variables for more information.

The maximum number of items specified for Map Entries refers to the maximum number of environment variables for each TrainingJobDefinition and also the maximum for the hyperparameter tuning job itself. That is, the sum of the number of environment variables for all the training job definitions can't exceed the maximum number specified.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges

Specifies ranges of integer, continuous, and categorical hyperparameters that a hyperparameter tuning job searches. The hyperparameter tuning job launches training jobs with hyperparameter values within these ranges to find the combination of values that result in the training job with the best performance as measured by the objective metric of the hyperparameter tuning job.

The maximum number of items specified for Array Members refers to the maximum number of hyperparameters for each range and also the maximum for the hyperparameter tuning job itself. That is, the sum of the number of hyperparameters for all the ranges can't exceed the maximum number specified.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.auto_parameters[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.auto_parameters[].name

The name of the hyperparameter to optimize using Autotune.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.auto_parameters[].value_hint

An example value of the hyperparameter to optimize using Autotune.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.categorical_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.categorical_parameter_ranges[].name

The name of the categorical hyperparameter to tune.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.categorical_parameter_ranges[].values[]
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.continuous_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.continuous_parameter_ranges[].max_value

The maximum value for the hyperparameter. The tuning job uses floating-point values between MinValue value and this value for tuning.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.continuous_parameter_ranges[].min_value

The minimum value for the hyperparameter. The tuning job uses floating-point values between this value and MaxValuefor tuning.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.continuous_parameter_ranges[].name

The name of the continuous hyperparameter to tune.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.continuous_parameter_ranges[].scaling_type

The scale that hyperparameter tuning uses to search the hyperparameter range. For information about choosing a hyperparameter scale, see Hyperparameter Scaling. One of the following values:

Auto

SageMaker hyperparameter tuning chooses the best scale for the hyperparameter.

Linear

Hyperparameter tuning searches the values in the hyperparameter range by using a linear scale.

Logarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a logarithmic scale.

Logarithmic scaling works only for ranges that have only values greater than 0.

ReverseLogarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a reverse logarithmic scale.

Reverse logarithmic scaling works only for ranges that are entirely within the range 0<=x<1.0.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.integer_parameter_ranges[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.integer_parameter_ranges[].max_value

The maximum value of the hyperparameter to search.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.integer_parameter_ranges[].min_value

The minimum value of the hyperparameter to search.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.integer_parameter_ranges[].name

The name of the hyperparameter to search.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_ranges.integer_parameter_ranges[].scaling_type

The scale that hyperparameter tuning uses to search the hyperparameter range. For information about choosing a hyperparameter scale, see Hyperparameter Scaling. One of the following values:

Auto

SageMaker hyperparameter tuning chooses the best scale for the hyperparameter.

Linear

Hyperparameter tuning searches the values in the hyperparameter range by using a linear scale.

Logarithmic

Hyperparameter tuning searches the values in the hyperparameter range by using a logarithmic scale.

Logarithmic scaling works only for ranges that have only values greater than 0.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config

The configuration for the hyperparameter tuning resources, including the compute instances and storage volumes, used for training jobs launched by the tuning job. By default, storage volumes hold model artifacts and incremental states. Choose File for TrainingInputMode in the AlgorithmSpecification parameter to additionally store training data in the storage volume (optional).

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.allocation_strategy

The strategy that determines the order of preference for resources specified in InstanceConfigs used in hyperparameter optimization.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.instance_configs[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.instance_configs[].instance_count

The number of instances of the type specified by InstanceType. Choose an instance count larger than 1 for distributed training algorithms. See Step 2: Launch a SageMaker Distributed Training Job Using the SageMaker Python SDK for more information.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.instance_configs[].instance_type

The instance type used for processing of hyperparameter optimization jobs. Choose from general purpose (no GPUs) instance types: ml.m5.xlarge, ml.m5.2xlarge, and ml.m5.4xlarge or compute optimized (no GPUs) instance types: ml.c5.xlarge and ml.c5.2xlarge. For more information about instance types, see instance type descriptions.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.instance_configs[].volume_size_in_gb

The volume size in GB of the data to be processed for hyperparameter optimization (optional).

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.instance_count

The number of compute instances of type InstanceType to use. For distributed training, select a value greater than 1.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.instance_type

The instance type used to run hyperparameter optimization tuning jobs. See descriptions of instance types for more information.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.volume_kms_key_id

A key used by Amazon Web Services Key Management Service to encrypt data on the storage volume attached to the compute instances used to run the training job. You can use either of the following formats to specify a key.

KMS Key ID:

"1234abcd-12ab-34cd-56ef-1234567890ab"

Amazon Resource Name (ARN) of a KMS key:

"arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

Some instances use local storage, which use a hardware module to encrypt storage volumes. If you choose one of these instance types, you cannot request a VolumeKmsKeyId. For a list of instance types that use local storage, see instance store volumes. For more information about Amazon Web Services Key Management Service, see KMS encryption for more information.

hyper_parameter_tuning_job.training_job_definitions[].hyper_parameter_tuning_resource_config.volume_size_in_gb

The volume size in GB for the storage volume to be used in processing hyperparameter optimization jobs (optional). These volumes store model artifacts, incremental states and optionally, scratch space for training algorithms. Do not provide a value for this parameter if a value for InstanceConfigs is also specified.

Some instance types have a fixed total local storage size. If you select one of these instances for training, VolumeSizeInGB cannot be greater than this total size. For a list of instance types with local instance storage and their sizes, see instance store volumes.

SageMaker supports only the General Purpose SSD (gp2) storage volume type.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].channel_name

The name of the channel.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].compression_type

If training data is compressed, the compression type. The default value is None. CompressionType is used only in Pipe input mode. In File mode, leave this field unset or set it to None.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].content_type

The MIME type of the data.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source

The location of the channel data.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.file_system_data_source

The file system that is associated with a channel.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.file_system_data_source.directory_path

The full path to the directory to associate with the channel.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.file_system_data_source.file_system_access_mode

The access mode of the mount of the directory associated with the channel. A directory can be mounted either in ro (read-only) or rw (read-write) mode.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.file_system_data_source.file_system_id

The file system id.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.file_system_data_source.file_system_type

The file system type.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.s3_data_source

The S3 location of the data source that is associated with a channel.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.s3_data_source.attribute_names[]
hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.s3_data_source.instance_group_names[]
hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.s3_data_source.s3_data_distribution_type

If you want SageMaker to replicate the entire dataset on each ML compute instance that is launched for model training, specify FullyReplicated.

If you want SageMaker to replicate a subset of data on each ML compute instance that is launched for model training, specify ShardedByS3Key. If there are n ML compute instances launched for a training job, each instance gets approximately 1/n of the number of S3 objects. In this case, model training on each machine uses only the subset of training data.

Don't choose more ML compute instances for training than available S3 objects. If you do, some nodes won't get any data and you will pay for nodes that aren't getting any training data. This applies in both File and Pipe modes. Keep this in mind when developing algorithms.

In distributed training, where you use multiple ML compute EC2 instances, you might choose ShardedByS3Key. If the algorithm requires copying training data to the ML storage volume (when TrainingInputMode is set to File), this copies 1/n of the number of objects.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.s3_data_source.s3_data_type

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix for model training.

If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want SageMaker to use for model training.

If you choose AugmentedManifestFile, S3Uri identifies an object that is an augmented manifest file in JSON lines format. This file contains the data you want to use for model training. AugmentedManifestFile can only be used if the Channel's input mode is Pipe.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].data_source.s3_data_source.s3_uri

Depending on the value specified for the S3DataType, identifies either a key name prefix or a manifest. For example:

  • A key name prefix might look like this: s3://bucketname/exampleprefix/

  • A manifest might look like this: s3://bucketname/example.manifest

    A manifest is an S3 object which is a JSON file consisting of an array of elements. The first element is a prefix which is followed by one or more suffixes. SageMaker appends the suffix elements to the prefix to get a full set of S3Uri. Note that the prefix must be a valid non-empty S3Uri that precludes users from specifying a manifest whose individual S3Uri is sourced from different S3 buckets.

    The following code example shows a valid manifest format:

    [ {"prefix": "s3://customer_bucket/some/prefix/"},

    "relative/path/to/custdata-1",

    "relative/path/custdata-2",

    ...

    "relative/path/custdata-N"

    ]

    This JSON is equivalent to the following S3Uri list:

    s3://customer_bucket/some/prefix/relative/path/to/custdata-1

    s3://customer_bucket/some/prefix/relative/path/custdata-2

    ...

    s3://customer_bucket/some/prefix/relative/path/custdata-N

    The complete set of S3Uri in this manifest is the input data for the channel for this data source. The object that each S3Uri points to must be readable by the IAM role that SageMaker uses to perform tasks on your behalf.

Your input bucket must be located in same Amazon Web Services region as your training job.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].input_mode

(Optional) The input mode to use for the data channel in a training job. If you don't set a value for InputMode, SageMaker uses the value set for TrainingInputMode. Use this parameter to override the TrainingInputMode setting in a AlgorithmSpecification request when you have a channel that needs a different input mode from the training job's general setting. To download the data from Amazon Simple Storage Service (Amazon S3) to the provisioned ML storage volume, and mount the directory to a Docker volume, use File input mode. To stream data directly from Amazon S3 to the container, choose Pipe input mode.

To use a model for incremental training, choose File input model.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].record_wrapper_type

Specify RecordIO as the value when input data is in raw format but the training algorithm requires the RecordIO format. In this case, SageMaker wraps each individual S3 object in a RecordIO record. If the input data is already in RecordIO format, you don't need to set this attribute. For more information, see Create a Dataset Using RecordIO.

In File mode, leave this field unset or set it to None.

hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].shuffle_config

A configuration for a shuffle option for input data in a channel. If you use S3Prefix for S3DataType, this shuffles the results of the S3 key prefix matches. If you use ManifestFile, the order of the S3 object references in the ManifestFile is shuffled. If you use AugmentedManifestFile, the order of the JSON lines in the AugmentedManifestFile is shuffled. The shuffling order is determined using the Seed value.

For Pipe input mode, shuffling is done at the start of every epoch. With large datasets this ensures that the order of the training data is different for each epoch, it helps reduce bias and possible overfitting. In a multi-node training job when ShuffleConfig is combined with S3DataDistributionType of ShardedByS3Key, the data is shuffled across nodes so that the content sent to a particular node on the first epoch might be sent to a different node on the second epoch.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].input_data_config[].shuffle_config.seed

Determines the shuffling order in ShuffleConfig value.

hyper_parameter_tuning_job.training_job_definitions[].output_data_config

Specifies the path to the Amazon S3 bucket where you store model artifacts from the training jobs that the tuning job launches.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].output_data_config.compression_type

The model output compression type. Select None to output an uncompressed model, recommended for large model outputs. Defaults to gzip.

hyper_parameter_tuning_job.training_job_definitions[].output_data_config.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption. The KmsKeyId can be any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

  • // KMS Key Alias

    "alias/ExampleAlias"

  • // Amazon Resource Name (ARN) of a KMS Key Alias

    "arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias"

If you use a KMS key ID or an alias of your KMS key, the SageMaker execution role must include permissions to call kms:Encrypt. If you don't provide a KMS key ID, SageMaker uses the default KMS key for Amazon S3 for your role's account. For more information, see KMS-Managed Encryption Keys in the Amazon Simple Storage Service Developer Guide. If the output data is stored in Amazon S3 Express One Zone, it is encrypted with server-side encryption with Amazon S3 managed keys (SSE-S3). KMS key is not supported for Amazon S3 Express One Zone

The KMS key policy must grant permission to the IAM role that you specify in your CreateTrainingJob, CreateTransformJob, or CreateHyperParameterTuningJob requests. For more information, see Using Key Policies in Amazon Web Services KMS in the Amazon Web Services Key Management Service Developer Guide.

hyper_parameter_tuning_job.training_job_definitions[].output_data_config.s3_output_path

Identifies the S3 path where you want SageMaker to store the model artifacts. For example, s3://bucket-name/key-name-prefix.

hyper_parameter_tuning_job.training_job_definitions[].resource_config

The resources, including the compute instances and storage volumes, to use for the training jobs that the tuning job launches.

Storage volumes store model artifacts and incremental states. Training algorithms might also use storage volumes for scratch space. If you want SageMaker to use the storage volume to store the training data, choose File as the TrainingInputMode in the algorithm specification. For distributed training algorithms, specify an instance count greater than 1.

If you want to use hyperparameter optimization with instance type flexibility, use HyperParameterTuningResourceConfig instead.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].resource_config.instance_count

The number of ML compute instances to use. For distributed training, provide a value greater than 1.

hyper_parameter_tuning_job.training_job_definitions[].resource_config.instance_groups[]
Show child fields
hyper_parameter_tuning_job.training_job_definitions[].resource_config.instance_groups[].instance_count

Specifies the number of instances of the instance group.

hyper_parameter_tuning_job.training_job_definitions[].resource_config.instance_groups[].instance_group_name

Specifies the name of the instance group.

hyper_parameter_tuning_job.training_job_definitions[].resource_config.instance_groups[].instance_type

Specifies the instance type of the instance group.

hyper_parameter_tuning_job.training_job_definitions[].resource_config.instance_type

The ML compute instance type.

SageMaker Training on Amazon Elastic Compute Cloud (EC2) P4de instances is in preview release starting December 9th, 2022.

Amazon EC2 P4de instances (currently in preview) are powered by 8 NVIDIA A100 GPUs with 80GB high-performance HBM2e GPU memory, which accelerate the speed of training ML models that need to be trained on large datasets of high-resolution data. In this preview release, Amazon SageMaker supports ML training jobs on P4de instances (ml.p4de.24xlarge) to reduce model training time. The ml.p4de.24xlarge instances are available in the following Amazon Web Services Regions.

  • US East (N. Virginia) (us-east-1)

  • US West (Oregon) (us-west-2)

To request quota limit increase and start using P4de instances, contact the SageMaker Training service team through your account team.

hyper_parameter_tuning_job.training_job_definitions[].resource_config.keep_alive_period_in_seconds

The duration of time in seconds to retain configured resources in a warm pool for subsequent training jobs.

hyper_parameter_tuning_job.training_job_definitions[].resource_config.volume_kms_key_id

The Amazon Web Services KMS key that SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) that run the training job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

The VolumeKmsKeyId can be in any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

hyper_parameter_tuning_job.training_job_definitions[].resource_config.volume_size_in_gb

The size of the ML storage volume that you want to provision.

ML storage volumes store model artifacts and incremental states. Training algorithms might also use the ML storage volume for scratch space. If you want to store the training data in the ML storage volume, choose File as the TrainingInputMode in the algorithm specification.

When using an ML instance with NVMe SSD volumes, SageMaker doesn't provision Amazon EBS General Purpose SSD (gp2) storage. Available storage is fixed to the NVMe-type instance's storage capacity. SageMaker configures storage paths for training datasets, checkpoints, model artifacts, and outputs to use the entire capacity of the instance storage. For example, ML instance families with the NVMe-type instance storage include ml.p4d, ml.g4dn, and ml.g5.

When using an ML instance with the EBS-only storage option and without instance storage, you must define the size of EBS volume through VolumeSizeInGB in the ResourceConfig API. For example, ML instance families that use EBS volumes include ml.c5 and ml.p2.

To look up instance types and their instance storage types and volumes, see Amazon EC2 Instance Types.

To find the default local paths defined by the SageMaker training platform, see Amazon SageMaker Training Storage Folders for Training Datasets, Checkpoints, Model Artifacts, and Outputs.

hyper_parameter_tuning_job.training_job_definitions[].retry_strategy

The number of times to retry the job when the job fails due to an InternalServerError.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].retry_strategy.maximum_retry_attempts

The number of times to retry the job. When the job is retried, it's SecondaryStatus is changed to STARTING.

hyper_parameter_tuning_job.training_job_definitions[].role_arn

The Amazon Resource Name (ARN) of the IAM role associated with the training jobs that the tuning job launches.

hyper_parameter_tuning_job.training_job_definitions[].static_hyper_parameters

Specifies the values of hyperparameters that do not change for the tuning job.

hyper_parameter_tuning_job.training_job_definitions[].stopping_condition

Specifies a limit to how long a model hyperparameter training job can run. It also specifies how long a managed spot training job has to complete. When the job reaches the time limit, SageMaker ends the training job. Use this API to cap model training costs.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].stopping_condition.max_pending_time_in_seconds

The maximum length of time, in seconds, that a training or compilation job can be pending before it is stopped.

hyper_parameter_tuning_job.training_job_definitions[].stopping_condition.max_runtime_in_seconds

The maximum length of time, in seconds, that a training or compilation job can run before it is stopped.

For compilation jobs, if the job does not complete during this time, a TimeOut error is generated. We recommend starting with 900 seconds and increasing as necessary based on your model.

For all other jobs, if the job does not complete during this time, SageMaker ends the job. When RetryStrategy is specified in the job request, MaxRuntimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt. The default value is 1 day. The maximum value is 28 days.

The maximum time that a TrainingJob can run in total, including any time spent publishing metrics or archiving and uploading models after it has been stopped, is 30 days.

hyper_parameter_tuning_job.training_job_definitions[].stopping_condition.max_wait_time_in_seconds

The maximum length of time, in seconds, that a managed Spot training job has to complete. It is the amount of time spent waiting for Spot capacity plus the amount of time the job can run. It must be equal to or greater than MaxRuntimeInSeconds. If the job does not complete during this time, SageMaker ends the job.

When RetryStrategy is specified in the job request, MaxWaitTimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt.

hyper_parameter_tuning_job.training_job_definitions[].tuning_objective

Defines the objective metric for a hyperparameter tuning job. Hyperparameter tuning uses the value of this metric to evaluate the training jobs it launches, and returns the training job that results in either the highest or lowest value for this metric, depending on the value you specify for the Type parameter. If you want to define a custom objective metric, see Define metrics and environment variables.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].tuning_objective.metric_name

The name of the metric to use for the objective metric.

hyper_parameter_tuning_job.training_job_definitions[].tuning_objective.type

Whether to minimize or maximize the objective metric.

hyper_parameter_tuning_job.training_job_definitions[].vpc_config

The VpcConfig object that specifies the VPC that you want the training jobs that this hyperparameter tuning job launches to connect to. Control access to and from your training container by configuring the VPC. For more information, see Protect Training Jobs by Using an Amazon Virtual Private Cloud.

Show child fields
hyper_parameter_tuning_job.training_job_definitions[].vpc_config.security_group_ids[]
hyper_parameter_tuning_job.training_job_definitions[].vpc_config.subnets[]
hyper_parameter_tuning_job.training_job_status_counters

The numbers of training jobs launched by a hyperparameter tuning job, categorized by status.

Show child fields
hyper_parameter_tuning_job.training_job_status_counters.completed

The number of completed training jobs launched by the hyperparameter tuning job.

hyper_parameter_tuning_job.training_job_status_counters.in_progress

The number of in-progress training jobs launched by a hyperparameter tuning job.

hyper_parameter_tuning_job.training_job_status_counters.non_retryable_error

The number of training jobs that failed and can't be retried. A failed training job can't be retried if it failed because a client error occurred.

hyper_parameter_tuning_job.training_job_status_counters.retryable_error

The number of training jobs that failed, but can be retried. A failed training job can be retried only if it failed because an internal service error occurred.

hyper_parameter_tuning_job.training_job_status_counters.stopped

The number of training jobs launched by a hyperparameter tuning job that were manually stopped.

hyper_parameter_tuning_job.tuning_job_completion_details

Information about either a current or completed hyperparameter tuning job.

Show child fields
hyper_parameter_tuning_job.tuning_job_completion_details.convergence_detected_time

The time in timestamp format that AMT detected model convergence, as defined by a lack of significant improvement over time based on criteria developed over a wide range of diverse benchmarking tests.

hyper_parameter_tuning_job.tuning_job_completion_details.number_of_training_jobs_objective_not_improving

The number of training jobs launched by a tuning job that are not improving (1% or less) as measured by model performance evaluated against an objective function.

hyper_parameter_tuning_job.warm_start_config

Specifies the configuration for a hyperparameter tuning job that uses one or more previous hyperparameter tuning jobs as a starting point. The results of previous tuning jobs are used to inform which combinations of hyperparameters to search over in the new tuning job.

All training jobs launched by the new hyperparameter tuning job are evaluated by using the objective metric, and the training job that performs the best is compared to the best training jobs from the parent tuning jobs. From these, the training job that performs the best as measured by the objective metric is returned as the overall best training job.

All training jobs launched by parent hyperparameter tuning jobs and the new hyperparameter tuning jobs count against the limit of training jobs for the tuning job.

Show child fields
hyper_parameter_tuning_job.warm_start_config.parent_hyper_parameter_tuning_jobs[]
Show child fields
hyper_parameter_tuning_job.warm_start_config.parent_hyper_parameter_tuning_jobs[].hyper_parameter_tuning_job_name

The name of the hyperparameter tuning job to be used as a starting point for a new hyperparameter tuning job.

hyper_parameter_tuning_job.warm_start_config.warm_start_type

Specifies one of the following:

IDENTICAL_DATA_AND_ALGORITHM

The new hyperparameter tuning job uses the same input data and training image as the parent tuning jobs. You can change the hyperparameter ranges to search and the maximum number of training jobs that the hyperparameter tuning job launches. You cannot use a new version of the training algorithm, unless the changes in the new version do not affect the algorithm itself. For example, changes that improve logging or adding support for a different data format are allowed. You can also change hyperparameters from tunable to static, and from static to tunable, but the total number of static plus tunable hyperparameters must remain the same as it is in all parent jobs. The objective metric for the new tuning job must be the same as for all parent jobs.

TRANSFER_LEARNING

The new hyperparameter tuning job can include input data, hyperparameter ranges, maximum number of concurrent training jobs, and maximum number of training jobs that are different than those of its parent hyperparameter tuning jobs. The training image can also be a different version from the version used in the parent hyperparameter tuning job. You can also change hyperparameters from tunable to static, and from static to tunable, but the total number of static plus tunable hyperparameters must remain the same as it is in all parent jobs. The objective metric for the new tuning job must be the same as for all parent jobs.

model

A model displayed in the Amazon SageMaker Model Dashboard.

STRUCT(
    "model" STRUCT(
        "model_name" VARCHAR,
        "primary_container" STRUCT(
            "container_hostname" VARCHAR,
            "image" VARCHAR,
            "image_config" STRUCT(
                "repository_access_mode" VARCHAR,
                "repository_auth_config" STRUCT(
                    "repository_credentials_provider_arn" VARCHAR
                )
            ),
            "mode" VARCHAR,
            "model_data_url" VARCHAR,
            "model_data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_uri" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "model_access_config" STRUCT(
                        "accept_eula" BOOLEAN
                    ),
                    "hub_access_config" STRUCT(
                        "hub_content_arn" VARCHAR
                    )
                )
            ),
            "additional_model_data_sources" STRUCT(
                "channel_name" VARCHAR,
                "s3_data_source" STRUCT(
                    "s3_uri" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "model_access_config" STRUCT(
                        "accept_eula" BOOLEAN
                    ),
                    "hub_access_config" STRUCT(
                        "hub_content_arn" VARCHAR
                    )
                )
            )[],
            "environment" MAP(VARCHAR, VARCHAR),
            "model_package_name" VARCHAR,
            "inference_specification_name" VARCHAR,
            "multi_model_config" STRUCT(
                "model_cache_setting" VARCHAR
            )
        ),
        "containers" STRUCT(
            "container_hostname" VARCHAR,
            "image" VARCHAR,
            "image_config" STRUCT(
                "repository_access_mode" VARCHAR,
                "repository_auth_config" STRUCT(
                    "repository_credentials_provider_arn" VARCHAR
                )
            ),
            "mode" VARCHAR,
            "model_data_url" VARCHAR,
            "model_data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_uri" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "model_access_config" STRUCT(
                        "accept_eula" BOOLEAN
                    ),
                    "hub_access_config" STRUCT(
                        "hub_content_arn" VARCHAR
                    )
                )
            ),
            "additional_model_data_sources" STRUCT(
                "channel_name" VARCHAR,
                "s3_data_source" STRUCT(
                    "s3_uri" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "model_access_config" STRUCT(
                        "accept_eula" BOOLEAN
                    ),
                    "hub_access_config" STRUCT(
                        "hub_content_arn" VARCHAR
                    )
                )
            )[],
            "environment" MAP(VARCHAR, VARCHAR),
            "model_package_name" VARCHAR,
            "inference_specification_name" VARCHAR,
            "multi_model_config" STRUCT(
                "model_cache_setting" VARCHAR
            )
        )[],
        "inference_execution_config" STRUCT(
            "mode" VARCHAR
        ),
        "execution_role_arn" VARCHAR,
        "vpc_config" STRUCT(
            "security_group_ids" VARCHAR[],
            "subnets" VARCHAR[]
        ),
        "creation_time" TIMESTAMP_S,
        "model_arn" VARCHAR,
        "enable_network_isolation" BOOLEAN,
        "tags" STRUCT(
            "key" VARCHAR,
            "value" VARCHAR
        )[],
        "deployment_recommendation" STRUCT(
            "recommendation_status" VARCHAR,
            "real_time_inference_recommendations" STRUCT(
                "recommendation_id" VARCHAR,
                "instance_type" VARCHAR,
                "environment" MAP(VARCHAR, VARCHAR)
            )[]
        )
    ),
    "endpoints" STRUCT(
        "endpoint_name" VARCHAR,
        "endpoint_arn" VARCHAR,
        "creation_time" TIMESTAMP_S,
        "last_modified_time" TIMESTAMP_S,
        "endpoint_status" VARCHAR
    )[],
    "last_batch_transform_job" STRUCT(
        "transform_job_name" VARCHAR,
        "transform_job_arn" VARCHAR,
        "transform_job_status" VARCHAR,
        "failure_reason" VARCHAR,
        "model_name" VARCHAR,
        "max_concurrent_transforms" BIGINT,
        "model_client_config" STRUCT(
            "invocations_timeout_in_seconds" BIGINT,
            "invocations_max_retries" BIGINT
        ),
        "max_payload_in_mb" BIGINT,
        "batch_strategy" VARCHAR,
        "environment" MAP(VARCHAR, VARCHAR),
        "transform_input" STRUCT(
            "data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_data_type" VARCHAR,
                    "s3_uri" VARCHAR
                )
            ),
            "content_type" VARCHAR,
            "compression_type" VARCHAR,
            "split_type" VARCHAR
        ),
        "transform_output" STRUCT(
            "s3_output_path" VARCHAR,
            "accept" VARCHAR,
            "assemble_with" VARCHAR,
            "kms_key_id" VARCHAR
        ),
        "data_capture_config" STRUCT(
            "destination_s3_uri" VARCHAR,
            "kms_key_id" VARCHAR,
            "generate_inference_id" BOOLEAN
        ),
        "transform_resources" STRUCT(
            "instance_type" VARCHAR,
            "instance_count" BIGINT,
            "volume_kms_key_id" VARCHAR
        ),
        "creation_time" TIMESTAMP_S,
        "transform_start_time" TIMESTAMP_S,
        "transform_end_time" TIMESTAMP_S,
        "labeling_job_arn" VARCHAR,
        "auto_ml_job_arn" VARCHAR,
        "data_processing" STRUCT(
            "input_filter" VARCHAR,
            "output_filter" VARCHAR,
            "join_source" VARCHAR
        ),
        "experiment_config" STRUCT(
            "experiment_name" VARCHAR,
            "trial_name" VARCHAR,
            "trial_component_display_name" VARCHAR,
            "run_name" VARCHAR
        ),
        "tags" STRUCT(
            "key" VARCHAR,
            "value" VARCHAR
        )[]
    ),
    "monitoring_schedules" STRUCT(
        "monitoring_schedule_arn" VARCHAR,
        "monitoring_schedule_name" VARCHAR,
        "monitoring_schedule_status" VARCHAR,
        "monitoring_type" VARCHAR,
        "failure_reason" VARCHAR,
        "creation_time" TIMESTAMP_S,
        "last_modified_time" TIMESTAMP_S,
        "monitoring_schedule_config" STRUCT(
            "schedule_config" STRUCT(
                "schedule_expression" VARCHAR,
                "data_analysis_start_time" VARCHAR,
                "data_analysis_end_time" VARCHAR
            ),
            "monitoring_job_definition" STRUCT(
                "baseline_config" STRUCT(
                    "baselining_job_name" VARCHAR,
                    "constraints_resource" STRUCT(
                        "s3_uri" VARCHAR
                    ),
                    "statistics_resource" STRUCT(
                        "s3_uri" VARCHAR
                    )
                ),
                "monitoring_inputs" STRUCT(
                    "endpoint_input" STRUCT(
                        "endpoint_name" VARCHAR,
                        "local_path" VARCHAR,
                        "s3_input_mode" VARCHAR,
                        "s3_data_distribution_type" VARCHAR,
                        "features_attribute" VARCHAR,
                        "inference_attribute" VARCHAR,
                        "probability_attribute" VARCHAR,
                        "probability_threshold_attribute" DOUBLE,
                        "start_time_offset" VARCHAR,
                        "end_time_offset" VARCHAR,
                        "exclude_features_attribute" VARCHAR
                    ),
                    "batch_transform_input" STRUCT(
                        "data_captured_destination_s3_uri" VARCHAR,
                        "dataset_format" STRUCT(
                            "csv" STRUCT(
                                "header" BOOLEAN
                            ),
                            "json" STRUCT(
                                "line" BOOLEAN
                            ),
                            "parquet" BOOLEAN
                        ),
                        "local_path" VARCHAR,
                        "s3_input_mode" VARCHAR,
                        "s3_data_distribution_type" VARCHAR,
                        "features_attribute" VARCHAR,
                        "inference_attribute" VARCHAR,
                        "probability_attribute" VARCHAR,
                        "probability_threshold_attribute" DOUBLE,
                        "start_time_offset" VARCHAR,
                        "end_time_offset" VARCHAR,
                        "exclude_features_attribute" VARCHAR
                    )
                )[],
                "monitoring_output_config" STRUCT(
                    "monitoring_outputs" STRUCT(
                        "s3_output" STRUCT(
                            "s3_uri" VARCHAR,
                            "local_path" VARCHAR,
                            "s3_upload_mode" VARCHAR
                        )
                    )[],
                    "kms_key_id" VARCHAR
                ),
                "monitoring_resources" STRUCT(
                    "cluster_config" STRUCT(
                        "instance_count" BIGINT,
                        "instance_type" VARCHAR,
                        "volume_size_in_gb" BIGINT,
                        "volume_kms_key_id" VARCHAR
                    )
                ),
                "monitoring_app_specification" STRUCT(
                    "image_uri" VARCHAR,
                    "container_entrypoint" VARCHAR[],
                    "container_arguments" VARCHAR[],
                    "record_preprocessor_source_uri" VARCHAR,
                    "post_analytics_processor_source_uri" VARCHAR
                ),
                "stopping_condition" STRUCT(
                    "max_runtime_in_seconds" BIGINT
                ),
                "environment" MAP(VARCHAR, VARCHAR),
                "network_config" STRUCT(
                    "enable_inter_container_traffic_encryption" BOOLEAN,
                    "enable_network_isolation" BOOLEAN,
                    "vpc_config" STRUCT(
                        "security_group_ids" VARCHAR[],
                        "subnets" VARCHAR[]
                    )
                ),
                "role_arn" VARCHAR
            ),
            "monitoring_job_definition_name" VARCHAR,
            "monitoring_type" VARCHAR
        ),
        "endpoint_name" VARCHAR,
        "monitoring_alert_summaries" STRUCT(
            "monitoring_alert_name" VARCHAR,
            "creation_time" TIMESTAMP_S,
            "last_modified_time" TIMESTAMP_S,
            "alert_status" VARCHAR,
            "datapoints_to_alert" BIGINT,
            "evaluation_period" BIGINT,
            "actions" STRUCT(
                "model_dashboard_indicator" STRUCT(
                    "enabled" BOOLEAN
                )
            )
        )[],
        "last_monitoring_execution_summary" STRUCT(
            "monitoring_schedule_name" VARCHAR,
            "scheduled_time" TIMESTAMP_S,
            "creation_time" TIMESTAMP_S,
            "last_modified_time" TIMESTAMP_S,
            "monitoring_execution_status" VARCHAR,
            "processing_job_arn" VARCHAR,
            "endpoint_name" VARCHAR,
            "failure_reason" VARCHAR,
            "monitoring_job_definition_name" VARCHAR,
            "monitoring_type" VARCHAR
        ),
        "batch_transform_input" STRUCT(
            "data_captured_destination_s3_uri" VARCHAR,
            "dataset_format" STRUCT(
                "csv" STRUCT(
                    "header" BOOLEAN
                ),
                "json" STRUCT(
                    "line" BOOLEAN
                ),
                "parquet" BOOLEAN
            ),
            "local_path" VARCHAR,
            "s3_input_mode" VARCHAR,
            "s3_data_distribution_type" VARCHAR,
            "features_attribute" VARCHAR,
            "inference_attribute" VARCHAR,
            "probability_attribute" VARCHAR,
            "probability_threshold_attribute" DOUBLE,
            "start_time_offset" VARCHAR,
            "end_time_offset" VARCHAR,
            "exclude_features_attribute" VARCHAR
        )
    )[],
    "model_card" STRUCT(
        "model_card_arn" VARCHAR,
        "model_card_name" VARCHAR,
        "model_card_version" BIGINT,
        "model_card_status" VARCHAR,
        "security_config" STRUCT(
            "kms_key_id" VARCHAR
        ),
        "creation_time" TIMESTAMP_S,
        "created_by" STRUCT(
            "user_profile_arn" VARCHAR,
            "user_profile_name" VARCHAR,
            "domain_id" VARCHAR,
            "iam_identity" STRUCT(
                "arn" VARCHAR,
                "principal_id" VARCHAR,
                "source_identity" VARCHAR
            )
        ),
        "last_modified_time" TIMESTAMP_S,
        "last_modified_by" STRUCT(
            "user_profile_arn" VARCHAR,
            "user_profile_name" VARCHAR,
            "domain_id" VARCHAR,
            "iam_identity" STRUCT(
                "arn" VARCHAR,
                "principal_id" VARCHAR,
                "source_identity" VARCHAR
            )
        ),
        "tags" STRUCT(
            "key" VARCHAR,
            "value" VARCHAR
        )[],
        "model_id" VARCHAR,
        "risk_rating" VARCHAR
    )
)
Show child fields
model.endpoints[]
Show child fields
model.endpoints[].creation_time

A timestamp that indicates when the endpoint was created.

model.endpoints[].endpoint_arn

The Amazon Resource Name (ARN) of the endpoint.

model.endpoints[].endpoint_name

The endpoint name.

model.endpoints[].endpoint_status

The endpoint status.

model.endpoints[].last_modified_time

The last time the endpoint was modified.

model.last_batch_transform_job

A batch transform job. For information about SageMaker batch transform, see Use Batch Transform.

Show child fields
model.last_batch_transform_job.auto_ml_job_arn

The Amazon Resource Name (ARN) of the AutoML job that created the transform job.

model.last_batch_transform_job.batch_strategy

Specifies the number of records to include in a mini-batch for an HTTP inference request. A record is a single unit of input data that inference can be made on. For example, a single line in a CSV file is a record.

model.last_batch_transform_job.creation_time

A timestamp that shows when the transform Job was created.

model.last_batch_transform_job.data_capture_config

Configuration to control how SageMaker captures inference data for batch transform jobs.

Show child fields
model.last_batch_transform_job.data_capture_config.destination_s3_uri

The Amazon S3 location being used to capture the data.

model.last_batch_transform_job.data_capture_config.generate_inference_id

Flag that indicates whether to append inference id to the output.

model.last_batch_transform_job.data_capture_config.kms_key_id

The Amazon Resource Name (ARN) of a Amazon Web Services Key Management Service key that SageMaker uses to encrypt data on the storage volume attached to the ML compute instance that hosts the batch transform job.

The KmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

model.last_batch_transform_job.data_processing

The data structure used to specify the data to be used for inference in a batch transform job and to associate the data that is relevant to the prediction results in the output. The input filter provided allows you to exclude input data that is not needed for inference in a batch transform job. The output filter provided allows you to include input data relevant to interpreting the predictions in the output from the job. For more information, see Associate Prediction Results with their Corresponding Input Records.

Show child fields
model.last_batch_transform_job.data_processing.input_filter

A JSONPath expression used to select a portion of the input data to pass to the algorithm. Use the InputFilter parameter to exclude fields, such as an ID column, from the input. If you want SageMaker to pass the entire input dataset to the algorithm, accept the default value $.

Examples: "$", "$[1:]", "$.features"

model.last_batch_transform_job.data_processing.join_source

Specifies the source of the data to join with the transformed data. The valid values are None and Input. The default value is None, which specifies not to join the input with the transformed data. If you want the batch transform job to join the original input data with the transformed data, set JoinSource to Input. You can specify OutputFilter as an additional filter to select a portion of the joined dataset and store it in the output file.

For JSON or JSONLines objects, such as a JSON array, SageMaker adds the transformed data to the input JSON object in an attribute called SageMakerOutput. The joined result for JSON must be a key-value pair object. If the input is not a key-value pair object, SageMaker creates a new JSON file. In the new JSON file, and the input data is stored under the SageMakerInput key and the results are stored in SageMakerOutput.

For CSV data, SageMaker takes each row as a JSON array and joins the transformed data with the input by appending each transformed row to the end of the input. The joined data has the original input data followed by the transformed data and the output is a CSV file.

For information on how joining in applied, see Workflow for Associating Inferences with Input Records.

model.last_batch_transform_job.data_processing.output_filter

A JSONPath expression used to select a portion of the joined dataset to save in the output file for a batch transform job. If you want SageMaker to store the entire input dataset in the output file, leave the default value, $. If you specify indexes that aren't within the dimension size of the joined dataset, you get an error.

Examples: "$", "$[0,5:]", "$['id','SageMakerOutput']"

model.last_batch_transform_job.environment

The environment variables to set in the Docker container. We support up to 16 key and values entries in the map.

model.last_batch_transform_job.experiment_config

Associates a SageMaker job as a trial component with an experiment and trial. Specified when you call the following APIs:

Show child fields
model.last_batch_transform_job.experiment_config.experiment_name

The name of an existing experiment to associate with the trial component.

model.last_batch_transform_job.experiment_config.run_name

The name of the experiment run to associate with the trial component.

model.last_batch_transform_job.experiment_config.trial_component_display_name

The display name for the trial component. If this key isn't specified, the display name is the trial component name.

model.last_batch_transform_job.experiment_config.trial_name

The name of an existing trial to associate the trial component with. If not specified, a new trial is created.

model.last_batch_transform_job.failure_reason

If the transform job failed, the reason it failed.

model.last_batch_transform_job.labeling_job_arn

The Amazon Resource Name (ARN) of the labeling job that created the transform job.

model.last_batch_transform_job.max_concurrent_transforms

The maximum number of parallel requests that can be sent to each instance in a transform job. If MaxConcurrentTransforms is set to 0 or left unset, SageMaker checks the optional execution-parameters to determine the settings for your chosen algorithm. If the execution-parameters endpoint is not enabled, the default value is 1. For built-in algorithms, you don't need to set a value for MaxConcurrentTransforms.

model.last_batch_transform_job.max_payload_in_mb

The maximum allowed size of the payload, in MB. A payload is the data portion of a record (without metadata). The value in MaxPayloadInMB must be greater than, or equal to, the size of a single record. To estimate the size of a record in MB, divide the size of your dataset by the number of records. To ensure that the records fit within the maximum payload size, we recommend using a slightly larger value. The default value is 6 MB. For cases where the payload might be arbitrarily large and is transmitted using HTTP chunked encoding, set the value to 0. This feature works only in supported algorithms. Currently, SageMaker built-in algorithms do not support HTTP chunked encoding.

model.last_batch_transform_job.model_client_config

Configures the timeout and maximum number of retries for processing a transform job invocation.

Show child fields
model.last_batch_transform_job.model_client_config.invocations_max_retries

The maximum number of retries when invocation requests are failing. The default value is 3.

model.last_batch_transform_job.model_client_config.invocations_timeout_in_seconds

The timeout value in seconds for an invocation request. The default value is 600.

model.last_batch_transform_job.model_name

The name of the model associated with the transform job.

model.last_batch_transform_job.tags[]
Show child fields
model.last_batch_transform_job.tags[].key

The tag key. Tag keys must be unique per resource.

model.last_batch_transform_job.tags[].value

The tag value.

model.last_batch_transform_job.transform_end_time

Indicates when the transform job has been completed, or has stopped or failed. You are billed for the time interval between this time and the value of TransformStartTime.

model.last_batch_transform_job.transform_input

Describes the input source of a transform job and the way the transform job consumes it.

Show child fields
model.last_batch_transform_job.transform_input.compression_type

If your transform data is compressed, specify the compression type. Amazon SageMaker automatically decompresses the data for the transform job accordingly. The default value is None.

model.last_batch_transform_job.transform_input.content_type

The multipurpose internet mail extension (MIME) type of the data. Amazon SageMaker uses the MIME type with each http call to transfer data to the transform job.

model.last_batch_transform_job.transform_input.data_source

Describes the location of the channel data, which is, the S3 location of the input data that the model can consume.

Show child fields
model.last_batch_transform_job.transform_input.data_source.s3_data_source

The S3 location of the data source that is associated with a channel.

Show child fields
model.last_batch_transform_job.transform_input.data_source.s3_data_source.s3_data_type

If you choose S3Prefix, S3Uri identifies a key name prefix. Amazon SageMaker uses all objects with the specified key name prefix for batch transform.

If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want Amazon SageMaker to use for batch transform.

The following values are compatible: ManifestFile, S3Prefix

The following value is not compatible: AugmentedManifestFile

model.last_batch_transform_job.transform_input.data_source.s3_data_source.s3_uri

Depending on the value specified for the S3DataType, identifies either a key name prefix or a manifest. For example:

  • A key name prefix might look like this: s3://bucketname/exampleprefix/.

  • A manifest might look like this: s3://bucketname/example.manifest

    The manifest is an S3 object which is a JSON file with the following format:

    [ {"prefix": "s3://customer_bucket/some/prefix/"},

    "relative/path/to/custdata-1",

    "relative/path/custdata-2",

    ...

    "relative/path/custdata-N"

    ]

    The preceding JSON matches the following S3Uris:

    s3://customer_bucket/some/prefix/relative/path/to/custdata-1

    s3://customer_bucket/some/prefix/relative/path/custdata-2

    ...

    s3://customer_bucket/some/prefix/relative/path/custdata-N

    The complete set of S3Uris in this manifest constitutes the input data for the channel for this datasource. The object that each S3Uris points to must be readable by the IAM role that Amazon SageMaker uses to perform tasks on your behalf.

model.last_batch_transform_job.transform_input.split_type

The method to use to split the transform job's data files into smaller batches. Splitting is necessary when the total size of each object is too large to fit in a single request. You can also use data splitting to improve performance by processing multiple concurrent mini-batches. The default value for SplitType is None, which indicates that input data files are not split, and request payloads contain the entire contents of an input object. Set the value of this parameter to Line to split records on a newline character boundary. SplitType also supports a number of record-oriented binary data formats. Currently, the supported record formats are:

  • RecordIO

  • TFRecord

When splitting is enabled, the size of a mini-batch depends on the values of the BatchStrategy and MaxPayloadInMB parameters. When the value of BatchStrategy is MultiRecord, Amazon SageMaker sends the maximum number of records in each request, up to the MaxPayloadInMB limit. If the value of BatchStrategy is SingleRecord, Amazon SageMaker sends individual records in each request.

Some data formats represent a record as a binary payload wrapped with extra padding bytes. When splitting is applied to a binary data format, padding is removed if the value of BatchStrategy is set to SingleRecord. Padding is not removed if the value of BatchStrategy is set to MultiRecord.

For more information about RecordIO, see Create a Dataset Using RecordIO in the MXNet documentation. For more information about TFRecord, see Consuming TFRecord data in the TensorFlow documentation.

model.last_batch_transform_job.transform_job_arn

The Amazon Resource Name (ARN) of the transform job.

model.last_batch_transform_job.transform_job_name

The name of the transform job.

model.last_batch_transform_job.transform_job_status

The status of the transform job.

Transform job statuses are:

  • InProgress - The job is in progress.

  • Completed - The job has completed.

  • Failed - The transform job has failed. To see the reason for the failure, see the FailureReason field in the response to a DescribeTransformJob call.

  • Stopping - The transform job is stopping.

  • Stopped - The transform job has stopped.

model.last_batch_transform_job.transform_output

Describes the results of a transform job.

Show child fields
model.last_batch_transform_job.transform_output.accept

The MIME type used to specify the output data. Amazon SageMaker uses the MIME type with each http call to transfer data from the transform job.

model.last_batch_transform_job.transform_output.assemble_with

Defines how to assemble the results of the transform job as a single S3 object. Choose a format that is most convenient to you. To concatenate the results in binary format, specify None. To add a newline character at the end of every transformed record, specify Line.

model.last_batch_transform_job.transform_output.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption. The KmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

If you don't provide a KMS key ID, Amazon SageMaker uses the default KMS key for Amazon S3 for your role's account. For more information, see KMS-Managed Encryption Keys in the Amazon Simple Storage Service Developer Guide.

The KMS key policy must grant permission to the IAM role that you specify in your CreateModel request. For more information, see Using Key Policies in Amazon Web Services KMS in the Amazon Web Services Key Management Service Developer Guide.

model.last_batch_transform_job.transform_output.s3_output_path

The Amazon S3 path where you want Amazon SageMaker to store the results of the transform job. For example, s3://bucket-name/key-name-prefix.

For every S3 object used as input for the transform job, batch transform stores the transformed data with an .out suffix in a corresponding subfolder in the location in the output prefix. For example, for the input data stored at s3://bucket-name/input-name-prefix/dataset01/data.csv, batch transform stores the transformed data at s3://bucket-name/output-name-prefix/input-name-prefix/data.csv.out. Batch transform doesn't upload partially processed objects. For an input S3 object that contains multiple records, it creates an .out file only if the transform job succeeds on the entire file. When the input contains multiple S3 objects, the batch transform job processes the listed S3 objects and uploads only the output for successfully processed objects. If any object fails in the transform job batch transform marks the job as failed to prompt investigation.

model.last_batch_transform_job.transform_resources

Describes the resources, including ML instance types and ML instance count, to use for transform job.

Show child fields
model.last_batch_transform_job.transform_resources.instance_count

The number of ML compute instances to use in the transform job. The default value is 1, and the maximum is 100. For distributed transform jobs, specify a value greater than 1.

model.last_batch_transform_job.transform_resources.instance_type

The ML compute instance type for the transform job. If you are using built-in algorithms to transform moderately sized datasets, we recommend using ml.m4.xlarge or ml.m5.largeinstance types.

model.last_batch_transform_job.transform_resources.volume_kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt model data on the storage volume attached to the ML compute instance(s) that run the batch transform job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

The VolumeKmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

model.last_batch_transform_job.transform_start_time

Indicates when the transform job starts on ML instances. You are billed for the time interval between this time and the value of TransformEndTime.

model.model

A model displayed in the Model Dashboard.

Show child fields
model.model.containers[]
Show child fields
model.model.containers[].additional_model_data_sources[]
Show child fields
model.model.containers[].additional_model_data_sources[].channel_name

A custom name for this AdditionalModelDataSource object.

model.model.containers[].additional_model_data_sources[].s3_data_source

Specifies the S3 location of ML model data to deploy.

Show child fields
model.model.containers[].additional_model_data_sources[].s3_data_source.compression_type

Specifies how the ML model data is prepared.

If you choose Gzip and choose S3Object as the value of S3DataType, S3Uri identifies an object that is a gzip-compressed TAR archive. SageMaker will attempt to decompress and untar the object during model deployment.

If you choose None and chooose S3Object as the value of S3DataType, S3Uri identifies an object that represents an uncompressed ML model to deploy.

If you choose None and choose S3Prefix as the value of S3DataType, S3Uri identifies a key name prefix, under which all objects represents the uncompressed ML model to deploy.

If you choose None, then SageMaker will follow rules below when creating model data files under /opt/ml/model directory for use by your inference code:

  • If you choose S3Object as the value of S3DataType, then SageMaker will split the key of the S3 object referenced by S3Uri by slash (/), and use the last part as the filename of the file holding the content of the S3 object.

  • If you choose S3Prefix as the value of S3DataType, then for each S3 object under the key name pefix referenced by S3Uri, SageMaker will trim its key by the prefix, and use the remainder as the path (relative to /opt/ml/model) of the file holding the content of the S3 object. SageMaker will split the remainder by slash (/), using intermediate parts as directory names and the last part as filename of the file holding the content of the S3 object.

  • Do not use any of the following as file names or directory names:

    • An empty or blank string

    • A string which contains null bytes

    • A string longer than 255 bytes

    • A single dot (.)

    • A double dot (..)

  • Ambiguous file names will result in model deployment failure. For example, if your uncompressed ML model consists of two S3 objects s3://mybucket/model/weights and s3://mybucket/model/weights/part1 and you specify s3://mybucket/model/ as the value of S3Uri and S3Prefix as the value of S3DataType, then it will result in name clash between /opt/ml/model/weights (a regular file) and /opt/ml/model/weights/ (a directory).

  • Do not organize the model artifacts in S3 console using folders. When you create a folder in S3 console, S3 creates a 0-byte object with a key set to the folder name you provide. They key of the 0-byte object ends with a slash (/) which violates SageMaker restrictions on model artifact file names, leading to model deployment failure.

model.model.containers[].additional_model_data_sources[].s3_data_source.hub_access_config

Configuration information for hub access.

Show child fields
model.model.containers[].additional_model_data_sources[].s3_data_source.hub_access_config.hub_content_arn

The ARN of the hub content for which deployment access is allowed.

model.model.containers[].additional_model_data_sources[].s3_data_source.model_access_config

Specifies the access configuration file for the ML model. You can explicitly accept the model end-user license agreement (EULA) within the ModelAccessConfig. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

Show child fields
model.model.containers[].additional_model_data_sources[].s3_data_source.model_access_config.accept_eula

Specifies agreement to the model end-user license agreement (EULA). The AcceptEula value must be explicitly defined as True in order to accept the EULA that this model requires. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

model.model.containers[].additional_model_data_sources[].s3_data_source.s3_data_type

Specifies the type of ML model data to deploy.

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix as part of the ML model data to deploy. A valid key name prefix identified by S3Uri always ends with a forward slash (/).

If you choose S3Object, S3Uri identifies an object that is the ML model data to deploy.

model.model.containers[].additional_model_data_sources[].s3_data_source.s3_uri

Specifies the S3 path of ML model data to deploy.

model.model.containers[].container_hostname

This parameter is ignored for models that contain only a PrimaryContainer.

When a ContainerDefinition is part of an inference pipeline, the value of the parameter uniquely identifies the container for the purposes of logging and metrics. For information, see Use Logs and Metrics to Monitor an Inference Pipeline. If you don't specify a value for this parameter for a ContainerDefinition that is part of an inference pipeline, a unique name is automatically assigned based on the position of the ContainerDefinition in the pipeline. If you specify a value for the ContainerHostName for any ContainerDefinition that is part of an inference pipeline, you must specify a value for the ContainerHostName parameter of every ContainerDefinition in that pipeline.

model.model.containers[].environment

The environment variables to set in the Docker container.

The maximum length of each key and value in the Environment map is 1024 bytes. The maximum length of all keys and values in the map, combined, is 32 KB. If you pass multiple containers to a CreateModel request, then the maximum length of all of their maps, combined, is also 32 KB.

model.model.containers[].image

The path where inference code is stored. This can be either in Amazon EC2 Container Registry or in a Docker registry that is accessible from the same VPC that you configure for your endpoint. If you are using your own custom algorithm instead of an algorithm provided by SageMaker, the inference code must meet SageMaker requirements. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information, see Using Your Own Algorithms with Amazon SageMaker.

The model artifacts in an Amazon S3 bucket and the Docker image for inference container in Amazon EC2 Container Registry must be in the same region as the model or endpoint you are creating.

model.model.containers[].image_config

Specifies whether the model container is in Amazon ECR or a private Docker registry accessible from your Amazon Virtual Private Cloud (VPC). For information about storing containers in a private Docker registry, see Use a Private Docker Registry for Real-Time Inference Containers.

The model artifacts in an Amazon S3 bucket and the Docker image for inference container in Amazon EC2 Container Registry must be in the same region as the model or endpoint you are creating.

Show child fields
model.model.containers[].image_config.repository_access_mode

Set this to one of the following values:

  • Platform - The model image is hosted in Amazon ECR.

  • Vpc - The model image is hosted in a private Docker registry in your VPC.

model.model.containers[].image_config.repository_auth_config

(Optional) Specifies an authentication configuration for the private docker registry where your model image is hosted. Specify a value for this property only if you specified Vpc as the value for the RepositoryAccessMode field, and the private Docker registry where the model image is hosted requires authentication.

Show child fields
model.model.containers[].image_config.repository_auth_config.repository_credentials_provider_arn

The Amazon Resource Name (ARN) of an Amazon Web Services Lambda function that provides credentials to authenticate to the private Docker registry where your model image is hosted. For information about how to create an Amazon Web Services Lambda function, see Create a Lambda function with the console in the Amazon Web Services Lambda Developer Guide.

model.model.containers[].inference_specification_name

The inference specification name in the model package version.

model.model.containers[].mode

Whether the container hosts a single model or multiple models.

model.model.containers[].model_data_source

Specifies the location of ML model data to deploy.

Currently you cannot use ModelDataSource in conjunction with SageMaker batch transform, SageMaker serverless endpoints, SageMaker multi-model endpoints, and SageMaker Marketplace.

Show child fields
model.model.containers[].model_data_source.s3_data_source

Specifies the S3 location of ML model data to deploy.

Show child fields
model.model.containers[].model_data_source.s3_data_source.compression_type

Specifies how the ML model data is prepared.

If you choose Gzip and choose S3Object as the value of S3DataType, S3Uri identifies an object that is a gzip-compressed TAR archive. SageMaker will attempt to decompress and untar the object during model deployment.

If you choose None and chooose S3Object as the value of S3DataType, S3Uri identifies an object that represents an uncompressed ML model to deploy.

If you choose None and choose S3Prefix as the value of S3DataType, S3Uri identifies a key name prefix, under which all objects represents the uncompressed ML model to deploy.

If you choose None, then SageMaker will follow rules below when creating model data files under /opt/ml/model directory for use by your inference code:

  • If you choose S3Object as the value of S3DataType, then SageMaker will split the key of the S3 object referenced by S3Uri by slash (/), and use the last part as the filename of the file holding the content of the S3 object.

  • If you choose S3Prefix as the value of S3DataType, then for each S3 object under the key name pefix referenced by S3Uri, SageMaker will trim its key by the prefix, and use the remainder as the path (relative to /opt/ml/model) of the file holding the content of the S3 object. SageMaker will split the remainder by slash (/), using intermediate parts as directory names and the last part as filename of the file holding the content of the S3 object.

  • Do not use any of the following as file names or directory names:

    • An empty or blank string

    • A string which contains null bytes

    • A string longer than 255 bytes

    • A single dot (.)

    • A double dot (..)

  • Ambiguous file names will result in model deployment failure. For example, if your uncompressed ML model consists of two S3 objects s3://mybucket/model/weights and s3://mybucket/model/weights/part1 and you specify s3://mybucket/model/ as the value of S3Uri and S3Prefix as the value of S3DataType, then it will result in name clash between /opt/ml/model/weights (a regular file) and /opt/ml/model/weights/ (a directory).

  • Do not organize the model artifacts in S3 console using folders. When you create a folder in S3 console, S3 creates a 0-byte object with a key set to the folder name you provide. They key of the 0-byte object ends with a slash (/) which violates SageMaker restrictions on model artifact file names, leading to model deployment failure.

model.model.containers[].model_data_source.s3_data_source.hub_access_config

Configuration information for hub access.

Show child fields
model.model.containers[].model_data_source.s3_data_source.hub_access_config.hub_content_arn

The ARN of the hub content for which deployment access is allowed.

model.model.containers[].model_data_source.s3_data_source.model_access_config

Specifies the access configuration file for the ML model. You can explicitly accept the model end-user license agreement (EULA) within the ModelAccessConfig. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

Show child fields
model.model.containers[].model_data_source.s3_data_source.model_access_config.accept_eula

Specifies agreement to the model end-user license agreement (EULA). The AcceptEula value must be explicitly defined as True in order to accept the EULA that this model requires. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

model.model.containers[].model_data_source.s3_data_source.s3_data_type

Specifies the type of ML model data to deploy.

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix as part of the ML model data to deploy. A valid key name prefix identified by S3Uri always ends with a forward slash (/).

If you choose S3Object, S3Uri identifies an object that is the ML model data to deploy.

model.model.containers[].model_data_source.s3_data_source.s3_uri

Specifies the S3 path of ML model data to deploy.

model.model.containers[].model_data_url

The S3 path where the model artifacts, which result from model training, are stored. This path must point to a single gzip compressed tar archive (.tar.gz suffix). The S3 path is required for SageMaker built-in algorithms, but not if you use your own algorithms. For more information on built-in algorithms, see Common Parameters.

The model artifacts must be in an S3 bucket that is in the same region as the model or endpoint you are creating.

If you provide a value for this parameter, SageMaker uses Amazon Web Services Security Token Service to download model artifacts from the S3 path you provide. Amazon Web Services STS is activated in your Amazon Web Services account by default. If you previously deactivated Amazon Web Services STS for a region, you need to reactivate Amazon Web Services STS for that region. For more information, see Activating and Deactivating Amazon Web Services STS in an Amazon Web Services Region in the Amazon Web Services Identity and Access Management User Guide.

If you use a built-in algorithm to create a model, SageMaker requires that you provide a S3 path to the model artifacts in ModelDataUrl.

model.model.containers[].model_package_name

The name or Amazon Resource Name (ARN) of the model package to use to create the model.

model.model.containers[].multi_model_config

Specifies additional configuration for multi-model endpoints.

Show child fields
model.model.containers[].multi_model_config.model_cache_setting

Whether to cache models for a multi-model endpoint. By default, multi-model endpoints cache models so that a model does not have to be loaded into memory each time it is invoked. Some use cases do not benefit from model caching. For example, if an endpoint hosts a large number of models that are each invoked infrequently, the endpoint might perform better if you disable model caching. To disable model caching, set the value of this parameter to Disabled.

model.model.creation_time

A timestamp that indicates when the model was created.

model.model.deployment_recommendation

A set of recommended deployment configurations for the model.

Show child fields
model.model.deployment_recommendation.real_time_inference_recommendations[]
Show child fields
model.model.deployment_recommendation.real_time_inference_recommendations[].environment

The recommended environment variables to set in the model container for Real-Time Inference.

model.model.deployment_recommendation.real_time_inference_recommendations[].instance_type

The recommended instance type for Real-Time Inference.

model.model.deployment_recommendation.real_time_inference_recommendations[].recommendation_id

The recommendation ID which uniquely identifies each recommendation.

model.model.deployment_recommendation.recommendation_status

Status of the deployment recommendation. The status NOT_APPLICABLE means that SageMaker is unable to provide a default recommendation for the model using the information provided. If the deployment status is IN_PROGRESS, retry your API call after a few seconds to get a COMPLETED deployment recommendation.

model.model.enable_network_isolation

Isolates the model container. No inbound or outbound network calls can be made to or from the model container.

model.model.execution_role_arn

The Amazon Resource Name (ARN) of the IAM role that you specified for the model.

model.model.inference_execution_config

Specifies details about how containers in a multi-container endpoint are run.

Show child fields
model.model.inference_execution_config.mode

How containers in a multi-container are run. The following values are valid.

  • SERIAL - Containers run as a serial pipeline.

  • DIRECT - Only the individual container that you specify is run.

model.model.model_arn

The Amazon Resource Name (ARN) of the model.

model.model.model_name

The name of the model.

model.model.primary_container

Describes the container, as part of model definition.

Show child fields
model.model.primary_container.additional_model_data_sources[]
Show child fields
model.model.primary_container.additional_model_data_sources[].channel_name

A custom name for this AdditionalModelDataSource object.

model.model.primary_container.additional_model_data_sources[].s3_data_source

Specifies the S3 location of ML model data to deploy.

Show child fields
model.model.primary_container.additional_model_data_sources[].s3_data_source.compression_type

Specifies how the ML model data is prepared.

If you choose Gzip and choose S3Object as the value of S3DataType, S3Uri identifies an object that is a gzip-compressed TAR archive. SageMaker will attempt to decompress and untar the object during model deployment.

If you choose None and chooose S3Object as the value of S3DataType, S3Uri identifies an object that represents an uncompressed ML model to deploy.

If you choose None and choose S3Prefix as the value of S3DataType, S3Uri identifies a key name prefix, under which all objects represents the uncompressed ML model to deploy.

If you choose None, then SageMaker will follow rules below when creating model data files under /opt/ml/model directory for use by your inference code:

  • If you choose S3Object as the value of S3DataType, then SageMaker will split the key of the S3 object referenced by S3Uri by slash (/), and use the last part as the filename of the file holding the content of the S3 object.

  • If you choose S3Prefix as the value of S3DataType, then for each S3 object under the key name pefix referenced by S3Uri, SageMaker will trim its key by the prefix, and use the remainder as the path (relative to /opt/ml/model) of the file holding the content of the S3 object. SageMaker will split the remainder by slash (/), using intermediate parts as directory names and the last part as filename of the file holding the content of the S3 object.

  • Do not use any of the following as file names or directory names:

    • An empty or blank string

    • A string which contains null bytes

    • A string longer than 255 bytes

    • A single dot (.)

    • A double dot (..)

  • Ambiguous file names will result in model deployment failure. For example, if your uncompressed ML model consists of two S3 objects s3://mybucket/model/weights and s3://mybucket/model/weights/part1 and you specify s3://mybucket/model/ as the value of S3Uri and S3Prefix as the value of S3DataType, then it will result in name clash between /opt/ml/model/weights (a regular file) and /opt/ml/model/weights/ (a directory).

  • Do not organize the model artifacts in S3 console using folders. When you create a folder in S3 console, S3 creates a 0-byte object with a key set to the folder name you provide. They key of the 0-byte object ends with a slash (/) which violates SageMaker restrictions on model artifact file names, leading to model deployment failure.

model.model.primary_container.additional_model_data_sources[].s3_data_source.hub_access_config

Configuration information for hub access.

Show child fields
model.model.primary_container.additional_model_data_sources[].s3_data_source.hub_access_config.hub_content_arn

The ARN of the hub content for which deployment access is allowed.

model.model.primary_container.additional_model_data_sources[].s3_data_source.model_access_config

Specifies the access configuration file for the ML model. You can explicitly accept the model end-user license agreement (EULA) within the ModelAccessConfig. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

Show child fields
model.model.primary_container.additional_model_data_sources[].s3_data_source.model_access_config.accept_eula

Specifies agreement to the model end-user license agreement (EULA). The AcceptEula value must be explicitly defined as True in order to accept the EULA that this model requires. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

model.model.primary_container.additional_model_data_sources[].s3_data_source.s3_data_type

Specifies the type of ML model data to deploy.

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix as part of the ML model data to deploy. A valid key name prefix identified by S3Uri always ends with a forward slash (/).

If you choose S3Object, S3Uri identifies an object that is the ML model data to deploy.

model.model.primary_container.additional_model_data_sources[].s3_data_source.s3_uri

Specifies the S3 path of ML model data to deploy.

model.model.primary_container.container_hostname

This parameter is ignored for models that contain only a PrimaryContainer.

When a ContainerDefinition is part of an inference pipeline, the value of the parameter uniquely identifies the container for the purposes of logging and metrics. For information, see Use Logs and Metrics to Monitor an Inference Pipeline. If you don't specify a value for this parameter for a ContainerDefinition that is part of an inference pipeline, a unique name is automatically assigned based on the position of the ContainerDefinition in the pipeline. If you specify a value for the ContainerHostName for any ContainerDefinition that is part of an inference pipeline, you must specify a value for the ContainerHostName parameter of every ContainerDefinition in that pipeline.

model.model.primary_container.environment

The environment variables to set in the Docker container.

The maximum length of each key and value in the Environment map is 1024 bytes. The maximum length of all keys and values in the map, combined, is 32 KB. If you pass multiple containers to a CreateModel request, then the maximum length of all of their maps, combined, is also 32 KB.

model.model.primary_container.image

The path where inference code is stored. This can be either in Amazon EC2 Container Registry or in a Docker registry that is accessible from the same VPC that you configure for your endpoint. If you are using your own custom algorithm instead of an algorithm provided by SageMaker, the inference code must meet SageMaker requirements. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information, see Using Your Own Algorithms with Amazon SageMaker.

The model artifacts in an Amazon S3 bucket and the Docker image for inference container in Amazon EC2 Container Registry must be in the same region as the model or endpoint you are creating.

model.model.primary_container.image_config

Specifies whether the model container is in Amazon ECR or a private Docker registry accessible from your Amazon Virtual Private Cloud (VPC). For information about storing containers in a private Docker registry, see Use a Private Docker Registry for Real-Time Inference Containers.

The model artifacts in an Amazon S3 bucket and the Docker image for inference container in Amazon EC2 Container Registry must be in the same region as the model or endpoint you are creating.

Show child fields
model.model.primary_container.image_config.repository_access_mode

Set this to one of the following values:

  • Platform - The model image is hosted in Amazon ECR.

  • Vpc - The model image is hosted in a private Docker registry in your VPC.

model.model.primary_container.image_config.repository_auth_config

(Optional) Specifies an authentication configuration for the private docker registry where your model image is hosted. Specify a value for this property only if you specified Vpc as the value for the RepositoryAccessMode field, and the private Docker registry where the model image is hosted requires authentication.

Show child fields
model.model.primary_container.image_config.repository_auth_config.repository_credentials_provider_arn

The Amazon Resource Name (ARN) of an Amazon Web Services Lambda function that provides credentials to authenticate to the private Docker registry where your model image is hosted. For information about how to create an Amazon Web Services Lambda function, see Create a Lambda function with the console in the Amazon Web Services Lambda Developer Guide.

model.model.primary_container.inference_specification_name

The inference specification name in the model package version.

model.model.primary_container.mode

Whether the container hosts a single model or multiple models.

model.model.primary_container.model_data_source

Specifies the location of ML model data to deploy.

Currently you cannot use ModelDataSource in conjunction with SageMaker batch transform, SageMaker serverless endpoints, SageMaker multi-model endpoints, and SageMaker Marketplace.

Show child fields
model.model.primary_container.model_data_source.s3_data_source

Specifies the S3 location of ML model data to deploy.

Show child fields
model.model.primary_container.model_data_source.s3_data_source.compression_type

Specifies how the ML model data is prepared.

If you choose Gzip and choose S3Object as the value of S3DataType, S3Uri identifies an object that is a gzip-compressed TAR archive. SageMaker will attempt to decompress and untar the object during model deployment.

If you choose None and chooose S3Object as the value of S3DataType, S3Uri identifies an object that represents an uncompressed ML model to deploy.

If you choose None and choose S3Prefix as the value of S3DataType, S3Uri identifies a key name prefix, under which all objects represents the uncompressed ML model to deploy.

If you choose None, then SageMaker will follow rules below when creating model data files under /opt/ml/model directory for use by your inference code:

  • If you choose S3Object as the value of S3DataType, then SageMaker will split the key of the S3 object referenced by S3Uri by slash (/), and use the last part as the filename of the file holding the content of the S3 object.

  • If you choose S3Prefix as the value of S3DataType, then for each S3 object under the key name pefix referenced by S3Uri, SageMaker will trim its key by the prefix, and use the remainder as the path (relative to /opt/ml/model) of the file holding the content of the S3 object. SageMaker will split the remainder by slash (/), using intermediate parts as directory names and the last part as filename of the file holding the content of the S3 object.

  • Do not use any of the following as file names or directory names:

    • An empty or blank string

    • A string which contains null bytes

    • A string longer than 255 bytes

    • A single dot (.)

    • A double dot (..)

  • Ambiguous file names will result in model deployment failure. For example, if your uncompressed ML model consists of two S3 objects s3://mybucket/model/weights and s3://mybucket/model/weights/part1 and you specify s3://mybucket/model/ as the value of S3Uri and S3Prefix as the value of S3DataType, then it will result in name clash between /opt/ml/model/weights (a regular file) and /opt/ml/model/weights/ (a directory).

  • Do not organize the model artifacts in S3 console using folders. When you create a folder in S3 console, S3 creates a 0-byte object with a key set to the folder name you provide. They key of the 0-byte object ends with a slash (/) which violates SageMaker restrictions on model artifact file names, leading to model deployment failure.

model.model.primary_container.model_data_source.s3_data_source.hub_access_config

Configuration information for hub access.

Show child fields
model.model.primary_container.model_data_source.s3_data_source.hub_access_config.hub_content_arn

The ARN of the hub content for which deployment access is allowed.

model.model.primary_container.model_data_source.s3_data_source.model_access_config

Specifies the access configuration file for the ML model. You can explicitly accept the model end-user license agreement (EULA) within the ModelAccessConfig. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

Show child fields
model.model.primary_container.model_data_source.s3_data_source.model_access_config.accept_eula

Specifies agreement to the model end-user license agreement (EULA). The AcceptEula value must be explicitly defined as True in order to accept the EULA that this model requires. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

model.model.primary_container.model_data_source.s3_data_source.s3_data_type

Specifies the type of ML model data to deploy.

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix as part of the ML model data to deploy. A valid key name prefix identified by S3Uri always ends with a forward slash (/).

If you choose S3Object, S3Uri identifies an object that is the ML model data to deploy.

model.model.primary_container.model_data_source.s3_data_source.s3_uri

Specifies the S3 path of ML model data to deploy.

model.model.primary_container.model_data_url

The S3 path where the model artifacts, which result from model training, are stored. This path must point to a single gzip compressed tar archive (.tar.gz suffix). The S3 path is required for SageMaker built-in algorithms, but not if you use your own algorithms. For more information on built-in algorithms, see Common Parameters.

The model artifacts must be in an S3 bucket that is in the same region as the model or endpoint you are creating.

If you provide a value for this parameter, SageMaker uses Amazon Web Services Security Token Service to download model artifacts from the S3 path you provide. Amazon Web Services STS is activated in your Amazon Web Services account by default. If you previously deactivated Amazon Web Services STS for a region, you need to reactivate Amazon Web Services STS for that region. For more information, see Activating and Deactivating Amazon Web Services STS in an Amazon Web Services Region in the Amazon Web Services Identity and Access Management User Guide.

If you use a built-in algorithm to create a model, SageMaker requires that you provide a S3 path to the model artifacts in ModelDataUrl.

model.model.primary_container.model_package_name

The name or Amazon Resource Name (ARN) of the model package to use to create the model.

model.model.primary_container.multi_model_config

Specifies additional configuration for multi-model endpoints.

Show child fields
model.model.primary_container.multi_model_config.model_cache_setting

Whether to cache models for a multi-model endpoint. By default, multi-model endpoints cache models so that a model does not have to be loaded into memory each time it is invoked. Some use cases do not benefit from model caching. For example, if an endpoint hosts a large number of models that are each invoked infrequently, the endpoint might perform better if you disable model caching. To disable model caching, set the value of this parameter to Disabled.

model.model.tags[]
Show child fields
model.model.tags[].key

The tag key. Tag keys must be unique per resource.

model.model.tags[].value

The tag value.

model.model.vpc_config

Specifies an Amazon Virtual Private Cloud (VPC) that your SageMaker jobs, hosted models, and compute resources have access to. You can control access to and from your resources by configuring a VPC. For more information, see Give SageMaker Access to Resources in your Amazon VPC.

Show child fields
model.model.vpc_config.security_group_ids[]
model.model.vpc_config.subnets[]
model.model_card

The model card for a model.

Show child fields
model.model_card.created_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
model.model_card.created_by.domain_id

The domain associated with the user.

model.model_card.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
model.model_card.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

model.model_card.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

model.model_card.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

model.model_card.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

model.model_card.created_by.user_profile_name

The name of the user's profile.

model.model_card.creation_time

A timestamp that indicates when the model card was created.

model.model_card.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
model.model_card.last_modified_by.domain_id

The domain associated with the user.

model.model_card.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
model.model_card.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

model.model_card.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

model.model_card.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

model.model_card.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

model.model_card.last_modified_by.user_profile_name

The name of the user's profile.

model.model_card.last_modified_time

A timestamp that indicates when the model card was last updated.

model.model_card.model_card_arn

The Amazon Resource Name (ARN) for a model card.

model.model_card.model_card_name

The name of a model card.

model.model_card.model_card_status

The model card status.

model.model_card.model_card_version

The model card version.

model.model_card.model_id

For models created in SageMaker, this is the model ARN. For models created outside of SageMaker, this is a user-customized string.

model.model_card.risk_rating

A model card's risk rating. Can be low, medium, or high.

model.model_card.security_config

The KMS Key ID (KMSKeyId) for encryption of model card information.

Show child fields
model.model_card.security_config.kms_key_id

A Key Management Service key ID to use for encrypting a model card.

model.model_card.tags[]
Show child fields
model.model_card.tags[].key

The tag key. Tag keys must be unique per resource.

model.model_card.tags[].value

The tag value.

model.monitoring_schedules[]
Show child fields
model.monitoring_schedules[].batch_transform_input

Input object for the batch transform job.

Show child fields
model.monitoring_schedules[].batch_transform_input.data_captured_destination_s3_uri

The Amazon S3 location being used to capture the data.

model.monitoring_schedules[].batch_transform_input.dataset_format

The dataset format for your batch transform job.

Show child fields
model.monitoring_schedules[].batch_transform_input.dataset_format.csv

The CSV dataset used in the monitoring job.

Show child fields
model.monitoring_schedules[].batch_transform_input.dataset_format.csv.header

Indicates if the CSV data has a header.

model.monitoring_schedules[].batch_transform_input.dataset_format.json

The JSON dataset used in the monitoring job

Show child fields
model.monitoring_schedules[].batch_transform_input.dataset_format.json.line

Indicates if the file should be read as a JSON object per line.

model.monitoring_schedules[].batch_transform_input.dataset_format.parquet

The Parquet dataset used in the monitoring job

model.monitoring_schedules[].batch_transform_input.end_time_offset

If specified, monitoring jobs subtract this time from the end time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

model.monitoring_schedules[].batch_transform_input.exclude_features_attribute

The attributes of the input data to exclude from the analysis.

model.monitoring_schedules[].batch_transform_input.features_attribute

The attributes of the input data that are the input features.

model.monitoring_schedules[].batch_transform_input.inference_attribute

The attribute of the input data that represents the ground truth label.

model.monitoring_schedules[].batch_transform_input.local_path

Path to the filesystem where the batch transform data is available to the container.

model.monitoring_schedules[].batch_transform_input.probability_attribute

In a classification problem, the attribute that represents the class probability.

model.monitoring_schedules[].batch_transform_input.probability_threshold_attribute

The threshold for the class probability to be evaluated as a positive result.

model.monitoring_schedules[].batch_transform_input.s3_data_distribution_type

Whether input data distributed in Amazon S3 is fully replicated or sharded by an S3 key. Defaults to FullyReplicated

model.monitoring_schedules[].batch_transform_input.s3_input_mode

Whether the Pipe or File is used as the input mode for transferring data for the monitoring job. Pipe mode is recommended for large datasets. File mode is useful for small files that fit in memory. Defaults to File.

model.monitoring_schedules[].batch_transform_input.start_time_offset

If specified, monitoring jobs substract this time from the start time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

model.monitoring_schedules[].creation_time

A timestamp that indicates when the monitoring schedule was created.

model.monitoring_schedules[].endpoint_name

The endpoint which is monitored.

model.monitoring_schedules[].failure_reason

If a monitoring job failed, provides the reason.

model.monitoring_schedules[].last_modified_time

A timestamp that indicates when the monitoring schedule was last updated.

model.monitoring_schedules[].last_monitoring_execution_summary

Summary of information about the last monitoring job to run.

Show child fields
model.monitoring_schedules[].last_monitoring_execution_summary.creation_time

The time at which the monitoring job was created.

model.monitoring_schedules[].last_monitoring_execution_summary.endpoint_name

The name of the endpoint used to run the monitoring job.

model.monitoring_schedules[].last_monitoring_execution_summary.failure_reason

Contains the reason a monitoring job failed, if it failed.

model.monitoring_schedules[].last_monitoring_execution_summary.last_modified_time

A timestamp that indicates the last time the monitoring job was modified.

model.monitoring_schedules[].last_monitoring_execution_summary.monitoring_execution_status

The status of the monitoring job.

model.monitoring_schedules[].last_monitoring_execution_summary.monitoring_job_definition_name

The name of the monitoring job.

model.monitoring_schedules[].last_monitoring_execution_summary.monitoring_schedule_name

The name of the monitoring schedule.

model.monitoring_schedules[].last_monitoring_execution_summary.monitoring_type

The type of the monitoring job.

model.monitoring_schedules[].last_monitoring_execution_summary.processing_job_arn

The Amazon Resource Name (ARN) of the monitoring job.

model.monitoring_schedules[].last_monitoring_execution_summary.scheduled_time

The time the monitoring job was scheduled.

model.monitoring_schedules[].monitoring_alert_summaries[]
Show child fields
model.monitoring_schedules[].monitoring_alert_summaries[].actions

A list of alert actions taken in response to an alert going into InAlert status.

Show child fields
model.monitoring_schedules[].monitoring_alert_summaries[].actions.model_dashboard_indicator

An alert action taken to light up an icon on the Model Dashboard when an alert goes into InAlert status.

Show child fields
model.monitoring_schedules[].monitoring_alert_summaries[].actions.model_dashboard_indicator.enabled

Indicates whether the alert action is turned on.

model.monitoring_schedules[].monitoring_alert_summaries[].alert_status

The current status of an alert.

model.monitoring_schedules[].monitoring_alert_summaries[].creation_time

A timestamp that indicates when a monitor alert was created.

model.monitoring_schedules[].monitoring_alert_summaries[].datapoints_to_alert

Within EvaluationPeriod, how many execution failures will raise an alert.

model.monitoring_schedules[].monitoring_alert_summaries[].evaluation_period

The number of most recent monitoring executions to consider when evaluating alert status.

model.monitoring_schedules[].monitoring_alert_summaries[].last_modified_time

A timestamp that indicates when a monitor alert was last updated.

model.monitoring_schedules[].monitoring_alert_summaries[].monitoring_alert_name

The name of a monitoring alert.

model.monitoring_schedules[].monitoring_schedule_arn

The Amazon Resource Name (ARN) of a monitoring schedule.

model.monitoring_schedules[].monitoring_schedule_config

Configures the monitoring schedule and defines the monitoring job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition

Defines the monitoring job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config

Baseline configuration used to validate that the data conforms to the specified constraints and statistics

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.baselining_job_name

The name of the job that performs baselining for the monitoring job.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.constraints_resource

The baseline constraint file in Amazon S3 that the current monitoring job should validated against.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.constraints_resource.s3_uri

The Amazon S3 URI for the constraints resource.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.statistics_resource

The baseline statistics file in Amazon S3 that the current monitoring job should be validated against.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.baseline_config.statistics_resource.s3_uri

The Amazon S3 URI for the statistics resource.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.environment

Sets the environment variables in the Docker container.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification

Configures the monitoring job to run a specified Docker container image.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.container_arguments[]
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.container_entrypoint[]
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.image_uri

The container image to be run by the monitoring job.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.post_analytics_processor_source_uri

An Amazon S3 URI to a script that is called after analysis has been performed. Applicable only for the built-in (first party) containers.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_app_specification.record_preprocessor_source_uri

An Amazon S3 URI to a script that is called per row prior to running analysis. It can base64 decode the payload and convert it into a flattened JSON so that the built-in container can use the converted data. Applicable only for the built-in (first party) containers.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[]
Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input

Input object for the batch transform job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.data_captured_destination_s3_uri

The Amazon S3 location being used to capture the data.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format

The dataset format for your batch transform job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.csv

The CSV dataset used in the monitoring job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.csv.header

Indicates if the CSV data has a header.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.json

The JSON dataset used in the monitoring job

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.json.line

Indicates if the file should be read as a JSON object per line.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.dataset_format.parquet

The Parquet dataset used in the monitoring job

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.end_time_offset

If specified, monitoring jobs subtract this time from the end time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.exclude_features_attribute

The attributes of the input data to exclude from the analysis.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.features_attribute

The attributes of the input data that are the input features.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.inference_attribute

The attribute of the input data that represents the ground truth label.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.local_path

Path to the filesystem where the batch transform data is available to the container.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.probability_attribute

In a classification problem, the attribute that represents the class probability.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.probability_threshold_attribute

The threshold for the class probability to be evaluated as a positive result.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.s3_data_distribution_type

Whether input data distributed in Amazon S3 is fully replicated or sharded by an S3 key. Defaults to FullyReplicated

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.s3_input_mode

Whether the Pipe or File is used as the input mode for transferring data for the monitoring job. Pipe mode is recommended for large datasets. File mode is useful for small files that fit in memory. Defaults to File.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].batch_transform_input.start_time_offset

If specified, monitoring jobs substract this time from the start time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input

The endpoint for a monitoring job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.end_time_offset

If specified, monitoring jobs substract this time from the end time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.endpoint_name

An endpoint in customer's account which has enabled DataCaptureConfig enabled.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.exclude_features_attribute

The attributes of the input data to exclude from the analysis.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.features_attribute

The attributes of the input data that are the input features.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.inference_attribute

The attribute of the input data that represents the ground truth label.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.local_path

Path to the filesystem where the endpoint data is available to the container.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.probability_attribute

In a classification problem, the attribute that represents the class probability.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.probability_threshold_attribute

The threshold for the class probability to be evaluated as a positive result.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.s3_data_distribution_type

Whether input data distributed in Amazon S3 is fully replicated or sharded by an Amazon S3 key. Defaults to FullyReplicated

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.s3_input_mode

Whether the Pipe or File is used as the input mode for transferring data for the monitoring job. Pipe mode is recommended for large datasets. File mode is useful for small files that fit in memory. Defaults to File.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_inputs[].endpoint_input.start_time_offset

If specified, monitoring jobs substract this time from the start time. For information about using offsets for scheduling monitoring jobs, see Schedule Model Quality Monitoring Jobs.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config

The array of outputs from the monitoring job to be uploaded to Amazon S3.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.kms_key_id

The Key Management Service (KMS) key that Amazon SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[]
Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output

The Amazon S3 storage location where the results of a monitoring job are saved.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output.local_path

The local path to the Amazon S3 storage location where Amazon SageMaker saves the results of a monitoring job. LocalPath is an absolute path for the output data.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output.s3_upload_mode

Whether to upload the results of the monitoring job continuously or after the job completes.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_output_config.monitoring_outputs[].s3_output.s3_uri

A URI that identifies the Amazon S3 storage location where Amazon SageMaker saves the results of a monitoring job.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources

Identifies the resources, ML compute instances, and ML storage volumes to deploy for a monitoring job. In distributed processing, you specify more than one instance.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config

The configuration for the cluster resources used to run the processing job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.instance_count

The number of ML compute instances to use in the model monitoring job. For distributed processing jobs, specify a value greater than 1. The default value is 1.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.instance_type

The ML compute instance type for the processing job.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.volume_kms_key_id

The Key Management Service (KMS) key that Amazon SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) that run the model monitoring job.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.monitoring_resources.cluster_config.volume_size_in_gb

The size of the ML storage volume, in gigabytes, that you want to provision. You must specify sufficient ML storage for your scenario.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config

Specifies networking options for an monitoring job.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.enable_inter_container_traffic_encryption

Whether to encrypt all communications between distributed processing jobs. Choose True to encrypt communications. Encryption provides greater security for distributed processing jobs, but the processing might take longer.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.enable_network_isolation

Whether to allow inbound and outbound network calls to and from the containers used for the processing job.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.vpc_config

Specifies an Amazon Virtual Private Cloud (VPC) that your SageMaker jobs, hosted models, and compute resources have access to. You can control access to and from your resources by configuring a VPC. For more information, see Give SageMaker Access to Resources in your Amazon VPC.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.vpc_config.security_group_ids[]
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.network_config.vpc_config.subnets[]
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.role_arn

The Amazon Resource Name (ARN) of an IAM role that Amazon SageMaker can assume to perform tasks on your behalf.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.stopping_condition

Specifies a time limit for how long the monitoring job is allowed to run.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition.stopping_condition.max_runtime_in_seconds

The maximum runtime allowed in seconds.

The MaxRuntimeInSeconds cannot exceed the frequency of the job. For data quality and model explainability, this can be up to 3600 seconds for an hourly schedule. For model bias and model quality hourly schedules, this can be up to 1800 seconds.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_job_definition_name

The name of the monitoring job definition to schedule.

model.monitoring_schedules[].monitoring_schedule_config.monitoring_type

The type of the monitoring job definition to schedule.

model.monitoring_schedules[].monitoring_schedule_config.schedule_config

Configures the monitoring schedule.

Show child fields
model.monitoring_schedules[].monitoring_schedule_config.schedule_config.data_analysis_end_time

Sets the end time for a monitoring job window. Express this time as an offset to the times that you schedule your monitoring jobs to run. You schedule monitoring jobs with the ScheduleExpression parameter. Specify this offset in ISO 8601 duration format. For example, if you want to end the window one hour before the start of each monitoring job, you would specify: "-PT1H".

The end time that you specify must not follow the start time that you specify by more than 24 hours. You specify the start time with the DataAnalysisStartTime parameter.

If you set ScheduleExpression to NOW, this parameter is required.

model.monitoring_schedules[].monitoring_schedule_config.schedule_config.data_analysis_start_time

Sets the start time for a monitoring job window. Express this time as an offset to the times that you schedule your monitoring jobs to run. You schedule monitoring jobs with the ScheduleExpression parameter. Specify this offset in ISO 8601 duration format. For example, if you want to monitor the five hours of data in your dataset that precede the start of each monitoring job, you would specify: "-PT5H".

The start time that you specify must not precede the end time that you specify by more than 24 hours. You specify the end time with the DataAnalysisEndTime parameter.

If you set ScheduleExpression to NOW, this parameter is required.

model.monitoring_schedules[].monitoring_schedule_config.schedule_config.schedule_expression

A cron expression that describes details about the monitoring schedule.

The supported cron expressions are:

  • If you want to set the job to start every hour, use the following:

    Hourly: cron(0 * ? * * *)

  • If you want to start the job daily:

    cron(0 [00-23] ? * * *)

  • If you want to run the job one time, immediately, use the following keyword:

    NOW

For example, the following are valid cron expressions:

  • Daily at noon UTC: cron(0 12 ? * * *)

  • Daily at midnight UTC: cron(0 0 ? * * *)

To support running every 6, 12 hours, the following are also supported:

cron(0 [00-23]/[01-24] ? * * *)

For example, the following are valid cron expressions:

  • Every 12 hours, starting at 5pm UTC: cron(0 17/12 ? * * *)

  • Every two hours starting at midnight: cron(0 0/2 ? * * *)

  • Even though the cron expression is set to start at 5PM UTC, note that there could be a delay of 0-20 minutes from the actual requested time to run the execution.

  • We recommend that if you would like a daily schedule, you do not provide this parameter. Amazon SageMaker will pick a time for running every day.

You can also specify the keyword NOW to run the monitoring job immediately, one time, without recurring.

model.monitoring_schedules[].monitoring_schedule_name

The name of a monitoring schedule.

model.monitoring_schedules[].monitoring_schedule_status

The status of the monitoring schedule.

model.monitoring_schedules[].monitoring_type

The monitor type of a model monitor.

model_card

An Amazon SageMaker Model Card that documents details about a machine learning model.

STRUCT(
    "model_card_arn" VARCHAR,
    "model_card_name" VARCHAR,
    "model_card_version" BIGINT,
    "content" VARCHAR,
    "model_card_status" VARCHAR,
    "security_config" STRUCT(
        "kms_key_id" VARCHAR
    ),
    "creation_time" TIMESTAMP_S,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "last_modified_time" TIMESTAMP_S,
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[],
    "model_id" VARCHAR,
    "risk_rating" VARCHAR,
    "model_package_group_name" VARCHAR
)
Show child fields
model_card.content

The content of the model card. Content uses the model card JSON schema and provided as a string.

model_card.created_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
model_card.created_by.domain_id

The domain associated with the user.

model_card.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
model_card.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

model_card.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

model_card.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

model_card.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

model_card.created_by.user_profile_name

The name of the user's profile.

model_card.creation_time

The date and time that the model card was created.

model_card.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
model_card.last_modified_by.domain_id

The domain associated with the user.

model_card.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
model_card.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

model_card.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

model_card.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

model_card.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

model_card.last_modified_by.user_profile_name

The name of the user's profile.

model_card.last_modified_time

The date and time that the model card was last modified.

model_card.model_card_arn

The Amazon Resource Name (ARN) of the model card.

model_card.model_card_name

The unique name of the model card.

model_card.model_card_status

The approval status of the model card within your organization. Different organizations might have different criteria for model card review and approval.

  • Draft: The model card is a work in progress.

  • PendingReview: The model card is pending review.

  • Approved: The model card is approved.

  • Archived: The model card is archived. No more updates should be made to the model card, but it can still be exported.

model_card.model_card_version

The version of the model card.

model_card.model_id

The unique name (ID) of the model.

model_card.model_package_group_name

The model package group that contains the model package. Only relevant for model cards created for model packages in the Amazon SageMaker Model Registry.

model_card.risk_rating

The risk rating of the model. Different organizations might have different criteria for model card risk ratings. For more information, see Risk ratings.

model_card.security_config

The security configuration used to protect model card data.

Show child fields
model_card.security_config.kms_key_id

A Key Management Service key ID to use for encrypting a model card.

model_card.tags[]
Show child fields
model_card.tags[].key

The tag key. Tag keys must be unique per resource.

model_card.tags[].value

The tag value.

model_package

A versioned model that can be deployed for SageMaker inference.

STRUCT(
    "model_package_name" VARCHAR,
    "model_package_group_name" VARCHAR,
    "model_package_version" BIGINT,
    "model_package_arn" VARCHAR,
    "model_package_description" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "inference_specification" STRUCT(
        "containers" STRUCT(
            "container_hostname" VARCHAR,
            "image" VARCHAR,
            "image_digest" VARCHAR,
            "model_data_url" VARCHAR,
            "model_data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_uri" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "model_access_config" STRUCT(
                        "accept_eula" BOOLEAN
                    ),
                    "hub_access_config" STRUCT(
                        "hub_content_arn" VARCHAR
                    )
                )
            ),
            "product_id" VARCHAR,
            "environment" MAP(VARCHAR, VARCHAR),
            "model_input" STRUCT(
                "data_input_config" VARCHAR
            ),
            "framework" VARCHAR,
            "framework_version" VARCHAR,
            "nearest_model_name" VARCHAR,
            "additional_s3_data_source" STRUCT(
                "s3_data_type" VARCHAR,
                "s3_uri" VARCHAR,
                "compression_type" VARCHAR
            )
        )[],
        "supported_transform_instance_types" VARCHAR[],
        "supported_realtime_inference_instance_types" VARCHAR[],
        "supported_content_types" VARCHAR[],
        "supported_response_mime_types" VARCHAR[]
    ),
    "source_algorithm_specification" STRUCT(
        "source_algorithms" STRUCT(
            "model_data_url" VARCHAR,
            "model_data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_uri" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "model_access_config" STRUCT(
                        "accept_eula" BOOLEAN
                    ),
                    "hub_access_config" STRUCT(
                        "hub_content_arn" VARCHAR
                    )
                )
            ),
            "algorithm_name" VARCHAR
        )[]
    ),
    "validation_specification" STRUCT(
        "validation_role" VARCHAR,
        "validation_profiles" STRUCT(
            "profile_name" VARCHAR,
            "transform_job_definition" STRUCT(
                "max_concurrent_transforms" BIGINT,
                "max_payload_in_mb" BIGINT,
                "batch_strategy" VARCHAR,
                "environment" MAP(VARCHAR, VARCHAR),
                "transform_input" STRUCT(
                    "data_source" STRUCT(
                        "s3_data_source" STRUCT(
                            "s3_data_type" VARCHAR,
                            "s3_uri" VARCHAR
                        )
                    ),
                    "content_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "split_type" VARCHAR
                ),
                "transform_output" STRUCT(
                    "s3_output_path" VARCHAR,
                    "accept" VARCHAR,
                    "assemble_with" VARCHAR,
                    "kms_key_id" VARCHAR
                ),
                "transform_resources" STRUCT(
                    "instance_type" VARCHAR,
                    "instance_count" BIGINT,
                    "volume_kms_key_id" VARCHAR
                )
            )
        )[]
    ),
    "model_package_status" VARCHAR,
    "model_package_status_details" STRUCT(
        "validation_statuses" STRUCT(
            "name" VARCHAR,
            "status" VARCHAR,
            "failure_reason" VARCHAR
        )[],
        "image_scan_statuses" STRUCT(
            "name" VARCHAR,
            "status" VARCHAR,
            "failure_reason" VARCHAR
        )[]
    ),
    "certify_for_marketplace" BOOLEAN,
    "model_approval_status" VARCHAR,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "metadata_properties" STRUCT(
        "commit_id" VARCHAR,
        "repository" VARCHAR,
        "generated_by" VARCHAR,
        "project_id" VARCHAR
    ),
    "model_metrics" STRUCT(
        "model_quality" STRUCT(
            "statistics" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "constraints" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        ),
        "model_data_quality" STRUCT(
            "statistics" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "constraints" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        ),
        "bias" STRUCT(
            "report" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "pre_training_report" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "post_training_report" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        ),
        "explainability" STRUCT(
            "report" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        )
    ),
    "last_modified_time" TIMESTAMP_S,
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "approval_description" VARCHAR,
    "domain" VARCHAR,
    "task" VARCHAR,
    "sample_payload_url" VARCHAR,
    "additional_inference_specifications" STRUCT(
        "name" VARCHAR,
        "description" VARCHAR,
        "containers" STRUCT(
            "container_hostname" VARCHAR,
            "image" VARCHAR,
            "image_digest" VARCHAR,
            "model_data_url" VARCHAR,
            "model_data_source" STRUCT(
                "s3_data_source" STRUCT(
                    "s3_uri" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "compression_type" VARCHAR,
                    "model_access_config" STRUCT(
                        "accept_eula" BOOLEAN
                    ),
                    "hub_access_config" STRUCT(
                        "hub_content_arn" VARCHAR
                    )
                )
            ),
            "product_id" VARCHAR,
            "environment" MAP(VARCHAR, VARCHAR),
            "model_input" STRUCT(
                "data_input_config" VARCHAR
            ),
            "framework" VARCHAR,
            "framework_version" VARCHAR,
            "nearest_model_name" VARCHAR,
            "additional_s3_data_source" STRUCT(
                "s3_data_type" VARCHAR,
                "s3_uri" VARCHAR,
                "compression_type" VARCHAR
            )
        )[],
        "supported_transform_instance_types" VARCHAR[],
        "supported_realtime_inference_instance_types" VARCHAR[],
        "supported_content_types" VARCHAR[],
        "supported_response_mime_types" VARCHAR[]
    )[],
    "source_uri" VARCHAR,
    "security_config" STRUCT(
        "kms_key_id" VARCHAR
    ),
    "model_card" STRUCT(
        "model_card_content" VARCHAR,
        "model_card_status" VARCHAR
    ),
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[],
    "customer_metadata_properties" MAP(VARCHAR, VARCHAR),
    "drift_check_baselines" STRUCT(
        "bias" STRUCT(
            "config_file" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "pre_training_constraints" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "post_training_constraints" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        ),
        "explainability" STRUCT(
            "constraints" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "config_file" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        ),
        "model_quality" STRUCT(
            "statistics" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "constraints" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        ),
        "model_data_quality" STRUCT(
            "statistics" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            ),
            "constraints" STRUCT(
                "content_type" VARCHAR,
                "content_digest" VARCHAR,
                "s3_uri" VARCHAR
            )
        )
    ),
    "skip_model_validation" VARCHAR
)
Show child fields
model_package.additional_inference_specifications[]
Show child fields
model_package.additional_inference_specifications[].containers[]
Show child fields
model_package.additional_inference_specifications[].containers[].additional_s3_data_source

The additional data source that is used during inference in the Docker container for your model package.

Show child fields
model_package.additional_inference_specifications[].containers[].additional_s3_data_source.compression_type

The type of compression used for an additional data source used in inference or training. Specify None if your additional data source is not compressed.

model_package.additional_inference_specifications[].containers[].additional_s3_data_source.s3_data_type

The data type of the additional data source that you specify for use in inference or training.

model_package.additional_inference_specifications[].containers[].additional_s3_data_source.s3_uri

The uniform resource identifier (URI) used to identify an additional data source used in inference or training.

model_package.additional_inference_specifications[].containers[].container_hostname

The DNS host name for the Docker container.

model_package.additional_inference_specifications[].containers[].environment

The environment variables to set in the Docker container. Each key and value in the Environment string to string map can have length of up to 1024. We support up to 16 entries in the map.

model_package.additional_inference_specifications[].containers[].framework

The machine learning framework of the model package container image.

model_package.additional_inference_specifications[].containers[].framework_version

The framework version of the Model Package Container Image.

model_package.additional_inference_specifications[].containers[].image

The Amazon EC2 Container Registry (Amazon ECR) path where inference code is stored.

If you are using your own custom algorithm instead of an algorithm provided by SageMaker, the inference code must meet SageMaker requirements. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information, see Using Your Own Algorithms with Amazon SageMaker.

model_package.additional_inference_specifications[].containers[].image_digest

An MD5 hash of the training algorithm that identifies the Docker image used for training.

model_package.additional_inference_specifications[].containers[].model_data_source

Specifies the location of ML model data to deploy during endpoint creation.

Show child fields
model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source

Specifies the S3 location of ML model data to deploy.

Show child fields
model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source.compression_type

Specifies how the ML model data is prepared.

If you choose Gzip and choose S3Object as the value of S3DataType, S3Uri identifies an object that is a gzip-compressed TAR archive. SageMaker will attempt to decompress and untar the object during model deployment.

If you choose None and chooose S3Object as the value of S3DataType, S3Uri identifies an object that represents an uncompressed ML model to deploy.

If you choose None and choose S3Prefix as the value of S3DataType, S3Uri identifies a key name prefix, under which all objects represents the uncompressed ML model to deploy.

If you choose None, then SageMaker will follow rules below when creating model data files under /opt/ml/model directory for use by your inference code:

  • If you choose S3Object as the value of S3DataType, then SageMaker will split the key of the S3 object referenced by S3Uri by slash (/), and use the last part as the filename of the file holding the content of the S3 object.

  • If you choose S3Prefix as the value of S3DataType, then for each S3 object under the key name pefix referenced by S3Uri, SageMaker will trim its key by the prefix, and use the remainder as the path (relative to /opt/ml/model) of the file holding the content of the S3 object. SageMaker will split the remainder by slash (/), using intermediate parts as directory names and the last part as filename of the file holding the content of the S3 object.

  • Do not use any of the following as file names or directory names:

    • An empty or blank string

    • A string which contains null bytes

    • A string longer than 255 bytes

    • A single dot (.)

    • A double dot (..)

  • Ambiguous file names will result in model deployment failure. For example, if your uncompressed ML model consists of two S3 objects s3://mybucket/model/weights and s3://mybucket/model/weights/part1 and you specify s3://mybucket/model/ as the value of S3Uri and S3Prefix as the value of S3DataType, then it will result in name clash between /opt/ml/model/weights (a regular file) and /opt/ml/model/weights/ (a directory).

  • Do not organize the model artifacts in S3 console using folders. When you create a folder in S3 console, S3 creates a 0-byte object with a key set to the folder name you provide. They key of the 0-byte object ends with a slash (/) which violates SageMaker restrictions on model artifact file names, leading to model deployment failure.

model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source.hub_access_config

Configuration information for hub access.

Show child fields
model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source.hub_access_config.hub_content_arn

The ARN of the hub content for which deployment access is allowed.

model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source.model_access_config

Specifies the access configuration file for the ML model. You can explicitly accept the model end-user license agreement (EULA) within the ModelAccessConfig. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

Show child fields
model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source.model_access_config.accept_eula

Specifies agreement to the model end-user license agreement (EULA). The AcceptEula value must be explicitly defined as True in order to accept the EULA that this model requires. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source.s3_data_type

Specifies the type of ML model data to deploy.

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix as part of the ML model data to deploy. A valid key name prefix identified by S3Uri always ends with a forward slash (/).

If you choose S3Object, S3Uri identifies an object that is the ML model data to deploy.

model_package.additional_inference_specifications[].containers[].model_data_source.s3_data_source.s3_uri

Specifies the S3 path of ML model data to deploy.

model_package.additional_inference_specifications[].containers[].model_data_url

The Amazon S3 path where the model artifacts, which result from model training, are stored. This path must point to a single gzip compressed tar archive (.tar.gz suffix).

The model artifacts must be in an S3 bucket that is in the same region as the model package.

model_package.additional_inference_specifications[].containers[].model_input

A structure with Model Input details.

Show child fields
model_package.additional_inference_specifications[].containers[].model_input.data_input_config

The input configuration object for the model.

model_package.additional_inference_specifications[].containers[].nearest_model_name

The name of a pre-trained machine learning benchmarked by Amazon SageMaker Inference Recommender model that matches your model. You can find a list of benchmarked models by calling ListModelMetadata.

model_package.additional_inference_specifications[].containers[].product_id

The Amazon Web Services Marketplace product ID of the model package.

model_package.additional_inference_specifications[].description

A description of the additional Inference specification

model_package.additional_inference_specifications[].name

A unique name to identify the additional inference specification. The name must be unique within the list of your additional inference specifications for a particular model package.

model_package.additional_inference_specifications[].supported_content_types[]
model_package.additional_inference_specifications[].supported_realtime_inference_instance_types[]
model_package.additional_inference_specifications[].supported_response_mime_types[]
model_package.additional_inference_specifications[].supported_transform_instance_types[]
model_package.approval_description

A description provided when the model approval is set.

model_package.certify_for_marketplace

Whether the model package is to be certified to be listed on Amazon Web Services Marketplace. For information about listing model packages on Amazon Web Services Marketplace, see List Your Algorithm or Model Package on Amazon Web Services Marketplace.

model_package.created_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, or project.

Show child fields
model_package.created_by.domain_id

The domain associated with the user.

model_package.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
model_package.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

model_package.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

model_package.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

model_package.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

model_package.created_by.user_profile_name

The name of the user's profile.

model_package.creation_time

The time that the model package was created.

model_package.customer_metadata_properties

The metadata properties for the model package.

model_package.domain

The machine learning domain of your model package and its components. Common machine learning domains include computer vision and natural language processing.

model_package.drift_check_baselines

Represents the drift check baselines that can be used when the model monitor is set using the model package.

Show child fields
model_package.drift_check_baselines.bias

Represents the drift check bias baselines that can be used when the model monitor is set using the model package.

Show child fields
model_package.drift_check_baselines.bias.config_file

The bias config file for a model.

Show child fields
model_package.drift_check_baselines.bias.config_file.content_digest

The digest of the file source.

model_package.drift_check_baselines.bias.config_file.content_type

The type of content stored in the file source.

model_package.drift_check_baselines.bias.config_file.s3_uri

The Amazon S3 URI for the file source.

model_package.drift_check_baselines.bias.post_training_constraints

The post-training constraints.

Show child fields
model_package.drift_check_baselines.bias.post_training_constraints.content_digest

The hash key used for the metrics source.

model_package.drift_check_baselines.bias.post_training_constraints.content_type

The metric source content type.

model_package.drift_check_baselines.bias.post_training_constraints.s3_uri

The S3 URI for the metrics source.

model_package.drift_check_baselines.bias.pre_training_constraints

The pre-training constraints.

Show child fields
model_package.drift_check_baselines.bias.pre_training_constraints.content_digest

The hash key used for the metrics source.

model_package.drift_check_baselines.bias.pre_training_constraints.content_type

The metric source content type.

model_package.drift_check_baselines.bias.pre_training_constraints.s3_uri

The S3 URI for the metrics source.

model_package.drift_check_baselines.explainability

Represents the drift check explainability baselines that can be used when the model monitor is set using the model package.

Show child fields
model_package.drift_check_baselines.explainability.config_file

The explainability config file for the model.

Show child fields
model_package.drift_check_baselines.explainability.config_file.content_digest

The digest of the file source.

model_package.drift_check_baselines.explainability.config_file.content_type

The type of content stored in the file source.

model_package.drift_check_baselines.explainability.config_file.s3_uri

The Amazon S3 URI for the file source.

model_package.drift_check_baselines.explainability.constraints

The drift check explainability constraints.

Show child fields
model_package.drift_check_baselines.explainability.constraints.content_digest

The hash key used for the metrics source.

model_package.drift_check_baselines.explainability.constraints.content_type

The metric source content type.

model_package.drift_check_baselines.explainability.constraints.s3_uri

The S3 URI for the metrics source.

model_package.drift_check_baselines.model_data_quality

Represents the drift check model data quality baselines that can be used when the model monitor is set using the model package.

Show child fields
model_package.drift_check_baselines.model_data_quality.constraints

The drift check model data quality constraints.

Show child fields
model_package.drift_check_baselines.model_data_quality.constraints.content_digest

The hash key used for the metrics source.

model_package.drift_check_baselines.model_data_quality.constraints.content_type

The metric source content type.

model_package.drift_check_baselines.model_data_quality.constraints.s3_uri

The S3 URI for the metrics source.

model_package.drift_check_baselines.model_data_quality.statistics

The drift check model data quality statistics.

Show child fields
model_package.drift_check_baselines.model_data_quality.statistics.content_digest

The hash key used for the metrics source.

model_package.drift_check_baselines.model_data_quality.statistics.content_type

The metric source content type.

model_package.drift_check_baselines.model_data_quality.statistics.s3_uri

The S3 URI for the metrics source.

model_package.drift_check_baselines.model_quality

Represents the drift check model quality baselines that can be used when the model monitor is set using the model package.

Show child fields
model_package.drift_check_baselines.model_quality.constraints

The drift check model quality constraints.

Show child fields
model_package.drift_check_baselines.model_quality.constraints.content_digest

The hash key used for the metrics source.

model_package.drift_check_baselines.model_quality.constraints.content_type

The metric source content type.

model_package.drift_check_baselines.model_quality.constraints.s3_uri

The S3 URI for the metrics source.

model_package.drift_check_baselines.model_quality.statistics

The drift check model quality statistics.

Show child fields
model_package.drift_check_baselines.model_quality.statistics.content_digest

The hash key used for the metrics source.

model_package.drift_check_baselines.model_quality.statistics.content_type

The metric source content type.

model_package.drift_check_baselines.model_quality.statistics.s3_uri

The S3 URI for the metrics source.

model_package.inference_specification

Defines how to perform inference generation after a training job is run.

Show child fields
model_package.inference_specification.containers[]
Show child fields
model_package.inference_specification.containers[].additional_s3_data_source

The additional data source that is used during inference in the Docker container for your model package.

Show child fields
model_package.inference_specification.containers[].additional_s3_data_source.compression_type

The type of compression used for an additional data source used in inference or training. Specify None if your additional data source is not compressed.

model_package.inference_specification.containers[].additional_s3_data_source.s3_data_type

The data type of the additional data source that you specify for use in inference or training.

model_package.inference_specification.containers[].additional_s3_data_source.s3_uri

The uniform resource identifier (URI) used to identify an additional data source used in inference or training.

model_package.inference_specification.containers[].container_hostname

The DNS host name for the Docker container.

model_package.inference_specification.containers[].environment

The environment variables to set in the Docker container. Each key and value in the Environment string to string map can have length of up to 1024. We support up to 16 entries in the map.

model_package.inference_specification.containers[].framework

The machine learning framework of the model package container image.

model_package.inference_specification.containers[].framework_version

The framework version of the Model Package Container Image.

model_package.inference_specification.containers[].image

The Amazon EC2 Container Registry (Amazon ECR) path where inference code is stored.

If you are using your own custom algorithm instead of an algorithm provided by SageMaker, the inference code must meet SageMaker requirements. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information, see Using Your Own Algorithms with Amazon SageMaker.

model_package.inference_specification.containers[].image_digest

An MD5 hash of the training algorithm that identifies the Docker image used for training.

model_package.inference_specification.containers[].model_data_source

Specifies the location of ML model data to deploy during endpoint creation.

Show child fields
model_package.inference_specification.containers[].model_data_source.s3_data_source

Specifies the S3 location of ML model data to deploy.

Show child fields
model_package.inference_specification.containers[].model_data_source.s3_data_source.compression_type

Specifies how the ML model data is prepared.

If you choose Gzip and choose S3Object as the value of S3DataType, S3Uri identifies an object that is a gzip-compressed TAR archive. SageMaker will attempt to decompress and untar the object during model deployment.

If you choose None and chooose S3Object as the value of S3DataType, S3Uri identifies an object that represents an uncompressed ML model to deploy.

If you choose None and choose S3Prefix as the value of S3DataType, S3Uri identifies a key name prefix, under which all objects represents the uncompressed ML model to deploy.

If you choose None, then SageMaker will follow rules below when creating model data files under /opt/ml/model directory for use by your inference code:

  • If you choose S3Object as the value of S3DataType, then SageMaker will split the key of the S3 object referenced by S3Uri by slash (/), and use the last part as the filename of the file holding the content of the S3 object.

  • If you choose S3Prefix as the value of S3DataType, then for each S3 object under the key name pefix referenced by S3Uri, SageMaker will trim its key by the prefix, and use the remainder as the path (relative to /opt/ml/model) of the file holding the content of the S3 object. SageMaker will split the remainder by slash (/), using intermediate parts as directory names and the last part as filename of the file holding the content of the S3 object.

  • Do not use any of the following as file names or directory names:

    • An empty or blank string

    • A string which contains null bytes

    • A string longer than 255 bytes

    • A single dot (.)

    • A double dot (..)

  • Ambiguous file names will result in model deployment failure. For example, if your uncompressed ML model consists of two S3 objects s3://mybucket/model/weights and s3://mybucket/model/weights/part1 and you specify s3://mybucket/model/ as the value of S3Uri and S3Prefix as the value of S3DataType, then it will result in name clash between /opt/ml/model/weights (a regular file) and /opt/ml/model/weights/ (a directory).

  • Do not organize the model artifacts in S3 console using folders. When you create a folder in S3 console, S3 creates a 0-byte object with a key set to the folder name you provide. They key of the 0-byte object ends with a slash (/) which violates SageMaker restrictions on model artifact file names, leading to model deployment failure.

model_package.inference_specification.containers[].model_data_source.s3_data_source.hub_access_config

Configuration information for hub access.

Show child fields
model_package.inference_specification.containers[].model_data_source.s3_data_source.hub_access_config.hub_content_arn

The ARN of the hub content for which deployment access is allowed.

model_package.inference_specification.containers[].model_data_source.s3_data_source.model_access_config

Specifies the access configuration file for the ML model. You can explicitly accept the model end-user license agreement (EULA) within the ModelAccessConfig. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

Show child fields
model_package.inference_specification.containers[].model_data_source.s3_data_source.model_access_config.accept_eula

Specifies agreement to the model end-user license agreement (EULA). The AcceptEula value must be explicitly defined as True in order to accept the EULA that this model requires. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

model_package.inference_specification.containers[].model_data_source.s3_data_source.s3_data_type

Specifies the type of ML model data to deploy.

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix as part of the ML model data to deploy. A valid key name prefix identified by S3Uri always ends with a forward slash (/).

If you choose S3Object, S3Uri identifies an object that is the ML model data to deploy.

model_package.inference_specification.containers[].model_data_source.s3_data_source.s3_uri

Specifies the S3 path of ML model data to deploy.

model_package.inference_specification.containers[].model_data_url

The Amazon S3 path where the model artifacts, which result from model training, are stored. This path must point to a single gzip compressed tar archive (.tar.gz suffix).

The model artifacts must be in an S3 bucket that is in the same region as the model package.

model_package.inference_specification.containers[].model_input

A structure with Model Input details.

Show child fields
model_package.inference_specification.containers[].model_input.data_input_config

The input configuration object for the model.

model_package.inference_specification.containers[].nearest_model_name

The name of a pre-trained machine learning benchmarked by Amazon SageMaker Inference Recommender model that matches your model. You can find a list of benchmarked models by calling ListModelMetadata.

model_package.inference_specification.containers[].product_id

The Amazon Web Services Marketplace product ID of the model package.

model_package.inference_specification.supported_content_types[]
model_package.inference_specification.supported_realtime_inference_instance_types[]
model_package.inference_specification.supported_response_mime_types[]
model_package.inference_specification.supported_transform_instance_types[]
model_package.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, or project.

Show child fields
model_package.last_modified_by.domain_id

The domain associated with the user.

model_package.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
model_package.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

model_package.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

model_package.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

model_package.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

model_package.last_modified_by.user_profile_name

The name of the user's profile.

model_package.last_modified_time

The last time the model package was modified.

model_package.metadata_properties

Metadata properties of the tracking entity, trial, or trial component.

Show child fields
model_package.metadata_properties.commit_id

The commit ID.

model_package.metadata_properties.generated_by

The entity this entity was generated by.

model_package.metadata_properties.project_id

The project ID.

model_package.metadata_properties.repository

The repository.

model_package.model_approval_status

The approval status of the model. This can be one of the following values.

  • APPROVED - The model is approved

  • REJECTED - The model is rejected.

  • PENDING_MANUAL_APPROVAL - The model is waiting for manual approval.

model_package.model_card

The model card associated with the model package. Since ModelPackageModelCard is tied to a model package, it is a specific usage of a model card and its schema is simplified compared to the schema of ModelCard. The ModelPackageModelCard schema does not include model_package_details, and model_overview is composed of the model_creator and model_artifact properties. For more information about the model package model card schema, see Model package model card schema. For more information about the model card associated with the model package, see View the Details of a Model Version.

Show child fields
model_package.model_card.model_card_content

The content of the model card. The content must follow the schema described in Model Package Model Card Schema.

model_package.model_card.model_card_status

The approval status of the model card within your organization. Different organizations might have different criteria for model card review and approval.

  • Draft: The model card is a work in progress.

  • PendingReview: The model card is pending review.

  • Approved: The model card is approved.

  • Archived: The model card is archived. No more updates can be made to the model card content. If you try to update the model card content, you will receive the message Model Card is in Archived state.

model_package.model_metrics

Metrics for the model.

Show child fields
model_package.model_metrics.bias

Metrics that measure bias in a model.

Show child fields
model_package.model_metrics.bias.post_training_report

The post-training bias report for a model.

Show child fields
model_package.model_metrics.bias.post_training_report.content_digest

The hash key used for the metrics source.

model_package.model_metrics.bias.post_training_report.content_type

The metric source content type.

model_package.model_metrics.bias.post_training_report.s3_uri

The S3 URI for the metrics source.

model_package.model_metrics.bias.pre_training_report

The pre-training bias report for a model.

Show child fields
model_package.model_metrics.bias.pre_training_report.content_digest

The hash key used for the metrics source.

model_package.model_metrics.bias.pre_training_report.content_type

The metric source content type.

model_package.model_metrics.bias.pre_training_report.s3_uri

The S3 URI for the metrics source.

model_package.model_metrics.bias.report

The bias report for a model

Show child fields
model_package.model_metrics.bias.report.content_digest

The hash key used for the metrics source.

model_package.model_metrics.bias.report.content_type

The metric source content type.

model_package.model_metrics.bias.report.s3_uri

The S3 URI for the metrics source.

model_package.model_metrics.explainability

Metrics that help explain a model.

Show child fields
model_package.model_metrics.explainability.report

The explainability report for a model.

Show child fields
model_package.model_metrics.explainability.report.content_digest

The hash key used for the metrics source.

model_package.model_metrics.explainability.report.content_type

The metric source content type.

model_package.model_metrics.explainability.report.s3_uri

The S3 URI for the metrics source.

model_package.model_metrics.model_data_quality

Metrics that measure the quality of the input data for a model.

Show child fields
model_package.model_metrics.model_data_quality.constraints

Data quality constraints for a model.

Show child fields
model_package.model_metrics.model_data_quality.constraints.content_digest

The hash key used for the metrics source.

model_package.model_metrics.model_data_quality.constraints.content_type

The metric source content type.

model_package.model_metrics.model_data_quality.constraints.s3_uri

The S3 URI for the metrics source.

model_package.model_metrics.model_data_quality.statistics

Data quality statistics for a model.

Show child fields
model_package.model_metrics.model_data_quality.statistics.content_digest

The hash key used for the metrics source.

model_package.model_metrics.model_data_quality.statistics.content_type

The metric source content type.

model_package.model_metrics.model_data_quality.statistics.s3_uri

The S3 URI for the metrics source.

model_package.model_metrics.model_quality

Metrics that measure the quality of a model.

Show child fields
model_package.model_metrics.model_quality.constraints

Model quality constraints.

Show child fields
model_package.model_metrics.model_quality.constraints.content_digest

The hash key used for the metrics source.

model_package.model_metrics.model_quality.constraints.content_type

The metric source content type.

model_package.model_metrics.model_quality.constraints.s3_uri

The S3 URI for the metrics source.

model_package.model_metrics.model_quality.statistics

Model quality statistics.

Show child fields
model_package.model_metrics.model_quality.statistics.content_digest

The hash key used for the metrics source.

model_package.model_metrics.model_quality.statistics.content_type

The metric source content type.

model_package.model_metrics.model_quality.statistics.s3_uri

The S3 URI for the metrics source.

model_package.model_package_arn

The Amazon Resource Name (ARN) of the model package.

model_package.model_package_description

The description of the model package.

model_package.model_package_group_name

The model group to which the model belongs.

model_package.model_package_name

The name of the model.

model_package.model_package_status

The status of the model package. This can be one of the following values.

  • PENDING - The model package is pending being created.

  • IN_PROGRESS - The model package is in the process of being created.

  • COMPLETED - The model package was successfully created.

  • FAILED - The model package failed.

  • DELETING - The model package is in the process of being deleted.

model_package.model_package_status_details

Specifies the validation and image scan statuses of the model package.

Show child fields
model_package.model_package_status_details.image_scan_statuses[]
Show child fields
model_package.model_package_status_details.image_scan_statuses[].failure_reason

if the overall status is Failed, the reason for the failure.

model_package.model_package_status_details.image_scan_statuses[].name

The name of the model package for which the overall status is being reported.

model_package.model_package_status_details.image_scan_statuses[].status

The current status.

model_package.model_package_status_details.validation_statuses[]
Show child fields
model_package.model_package_status_details.validation_statuses[].failure_reason

if the overall status is Failed, the reason for the failure.

model_package.model_package_status_details.validation_statuses[].name

The name of the model package for which the overall status is being reported.

model_package.model_package_status_details.validation_statuses[].status

The current status.

model_package.model_package_version

The version number of a versioned model.

model_package.sample_payload_url

The Amazon Simple Storage Service path where the sample payload are stored. This path must point to a single gzip compressed tar archive (.tar.gz suffix).

model_package.security_config

An optional Key Management Service key to encrypt, decrypt, and re-encrypt model package information for regulated workloads with highly sensitive data.

Show child fields
model_package.security_config.kms_key_id

The KMS Key ID (KMSKeyId) used for encryption of model package information.

model_package.skip_model_validation

Indicates if you want to skip model validation.

model_package.source_algorithm_specification

A list of algorithms that were used to create a model package.

Show child fields
model_package.source_algorithm_specification.source_algorithms[]
Show child fields
model_package.source_algorithm_specification.source_algorithms[].algorithm_name

The name of an algorithm that was used to create the model package. The algorithm must be either an algorithm resource in your SageMaker account or an algorithm in Amazon Web Services Marketplace that you are subscribed to.

model_package.source_algorithm_specification.source_algorithms[].model_data_source

Specifies the location of ML model data to deploy during endpoint creation.

Show child fields
model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source

Specifies the S3 location of ML model data to deploy.

Show child fields
model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source.compression_type

Specifies how the ML model data is prepared.

If you choose Gzip and choose S3Object as the value of S3DataType, S3Uri identifies an object that is a gzip-compressed TAR archive. SageMaker will attempt to decompress and untar the object during model deployment.

If you choose None and chooose S3Object as the value of S3DataType, S3Uri identifies an object that represents an uncompressed ML model to deploy.

If you choose None and choose S3Prefix as the value of S3DataType, S3Uri identifies a key name prefix, under which all objects represents the uncompressed ML model to deploy.

If you choose None, then SageMaker will follow rules below when creating model data files under /opt/ml/model directory for use by your inference code:

  • If you choose S3Object as the value of S3DataType, then SageMaker will split the key of the S3 object referenced by S3Uri by slash (/), and use the last part as the filename of the file holding the content of the S3 object.

  • If you choose S3Prefix as the value of S3DataType, then for each S3 object under the key name pefix referenced by S3Uri, SageMaker will trim its key by the prefix, and use the remainder as the path (relative to /opt/ml/model) of the file holding the content of the S3 object. SageMaker will split the remainder by slash (/), using intermediate parts as directory names and the last part as filename of the file holding the content of the S3 object.

  • Do not use any of the following as file names or directory names:

    • An empty or blank string

    • A string which contains null bytes

    • A string longer than 255 bytes

    • A single dot (.)

    • A double dot (..)

  • Ambiguous file names will result in model deployment failure. For example, if your uncompressed ML model consists of two S3 objects s3://mybucket/model/weights and s3://mybucket/model/weights/part1 and you specify s3://mybucket/model/ as the value of S3Uri and S3Prefix as the value of S3DataType, then it will result in name clash between /opt/ml/model/weights (a regular file) and /opt/ml/model/weights/ (a directory).

  • Do not organize the model artifacts in S3 console using folders. When you create a folder in S3 console, S3 creates a 0-byte object with a key set to the folder name you provide. They key of the 0-byte object ends with a slash (/) which violates SageMaker restrictions on model artifact file names, leading to model deployment failure.

model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source.hub_access_config

Configuration information for hub access.

Show child fields
model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source.hub_access_config.hub_content_arn

The ARN of the hub content for which deployment access is allowed.

model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source.model_access_config

Specifies the access configuration file for the ML model. You can explicitly accept the model end-user license agreement (EULA) within the ModelAccessConfig. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

Show child fields
model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source.model_access_config.accept_eula

Specifies agreement to the model end-user license agreement (EULA). The AcceptEula value must be explicitly defined as True in order to accept the EULA that this model requires. You are responsible for reviewing and complying with any applicable license terms and making sure they are acceptable for your use case before downloading or using a model.

model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source.s3_data_type

Specifies the type of ML model data to deploy.

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix as part of the ML model data to deploy. A valid key name prefix identified by S3Uri always ends with a forward slash (/).

If you choose S3Object, S3Uri identifies an object that is the ML model data to deploy.

model_package.source_algorithm_specification.source_algorithms[].model_data_source.s3_data_source.s3_uri

Specifies the S3 path of ML model data to deploy.

model_package.source_algorithm_specification.source_algorithms[].model_data_url

The Amazon S3 path where the model artifacts, which result from model training, are stored. This path must point to a single gzip compressed tar archive (.tar.gz suffix).

The model artifacts must be in an S3 bucket that is in the same Amazon Web Services region as the algorithm.

model_package.source_uri

The URI of the source for the model package.

model_package.tags[]
Show child fields
model_package.tags[].key

The tag key. Tag keys must be unique per resource.

model_package.tags[].value

The tag value.

model_package.task

The machine learning task your model package accomplishes. Common machine learning tasks include object detection and image classification.

model_package.validation_specification

Specifies batch transform jobs that SageMaker runs to validate your model package.

Show child fields
model_package.validation_specification.validation_profiles[]
Show child fields
model_package.validation_specification.validation_profiles[].profile_name

The name of the profile for the model package.

model_package.validation_specification.validation_profiles[].transform_job_definition

The TransformJobDefinition object that describes the transform job used for the validation of the model package.

Show child fields
model_package.validation_specification.validation_profiles[].transform_job_definition.batch_strategy

A string that determines the number of records included in a single mini-batch.

SingleRecord means only one record is used per mini-batch. MultiRecord means a mini-batch is set to contain as many records that can fit within the MaxPayloadInMB limit.

model_package.validation_specification.validation_profiles[].transform_job_definition.environment

The environment variables to set in the Docker container. We support up to 16 key and values entries in the map.

model_package.validation_specification.validation_profiles[].transform_job_definition.max_concurrent_transforms

The maximum number of parallel requests that can be sent to each instance in a transform job. The default value is 1.

model_package.validation_specification.validation_profiles[].transform_job_definition.max_payload_in_mb

The maximum payload size allowed, in MB. A payload is the data portion of a record (without metadata).

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input

A description of the input source and the way the transform job consumes it.

Show child fields
model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input.compression_type

If your transform data is compressed, specify the compression type. Amazon SageMaker automatically decompresses the data for the transform job accordingly. The default value is None.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input.content_type

The multipurpose internet mail extension (MIME) type of the data. Amazon SageMaker uses the MIME type with each http call to transfer data to the transform job.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input.data_source

Describes the location of the channel data, which is, the S3 location of the input data that the model can consume.

Show child fields
model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input.data_source.s3_data_source

The S3 location of the data source that is associated with a channel.

Show child fields
model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input.data_source.s3_data_source.s3_data_type

If you choose S3Prefix, S3Uri identifies a key name prefix. Amazon SageMaker uses all objects with the specified key name prefix for batch transform.

If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want Amazon SageMaker to use for batch transform.

The following values are compatible: ManifestFile, S3Prefix

The following value is not compatible: AugmentedManifestFile

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input.data_source.s3_data_source.s3_uri

Depending on the value specified for the S3DataType, identifies either a key name prefix or a manifest. For example:

  • A key name prefix might look like this: s3://bucketname/exampleprefix/.

  • A manifest might look like this: s3://bucketname/example.manifest

    The manifest is an S3 object which is a JSON file with the following format:

    [ {"prefix": "s3://customer_bucket/some/prefix/"},

    "relative/path/to/custdata-1",

    "relative/path/custdata-2",

    ...

    "relative/path/custdata-N"

    ]

    The preceding JSON matches the following S3Uris:

    s3://customer_bucket/some/prefix/relative/path/to/custdata-1

    s3://customer_bucket/some/prefix/relative/path/custdata-2

    ...

    s3://customer_bucket/some/prefix/relative/path/custdata-N

    The complete set of S3Uris in this manifest constitutes the input data for the channel for this datasource. The object that each S3Uris points to must be readable by the IAM role that Amazon SageMaker uses to perform tasks on your behalf.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_input.split_type

The method to use to split the transform job's data files into smaller batches. Splitting is necessary when the total size of each object is too large to fit in a single request. You can also use data splitting to improve performance by processing multiple concurrent mini-batches. The default value for SplitType is None, which indicates that input data files are not split, and request payloads contain the entire contents of an input object. Set the value of this parameter to Line to split records on a newline character boundary. SplitType also supports a number of record-oriented binary data formats. Currently, the supported record formats are:

  • RecordIO

  • TFRecord

When splitting is enabled, the size of a mini-batch depends on the values of the BatchStrategy and MaxPayloadInMB parameters. When the value of BatchStrategy is MultiRecord, Amazon SageMaker sends the maximum number of records in each request, up to the MaxPayloadInMB limit. If the value of BatchStrategy is SingleRecord, Amazon SageMaker sends individual records in each request.

Some data formats represent a record as a binary payload wrapped with extra padding bytes. When splitting is applied to a binary data format, padding is removed if the value of BatchStrategy is set to SingleRecord. Padding is not removed if the value of BatchStrategy is set to MultiRecord.

For more information about RecordIO, see Create a Dataset Using RecordIO in the MXNet documentation. For more information about TFRecord, see Consuming TFRecord data in the TensorFlow documentation.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_output

Identifies the Amazon S3 location where you want Amazon SageMaker to save the results from the transform job.

Show child fields
model_package.validation_specification.validation_profiles[].transform_job_definition.transform_output.accept

The MIME type used to specify the output data. Amazon SageMaker uses the MIME type with each http call to transfer data from the transform job.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_output.assemble_with

Defines how to assemble the results of the transform job as a single S3 object. Choose a format that is most convenient to you. To concatenate the results in binary format, specify None. To add a newline character at the end of every transformed record, specify Line.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_output.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption. The KmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

If you don't provide a KMS key ID, Amazon SageMaker uses the default KMS key for Amazon S3 for your role's account. For more information, see KMS-Managed Encryption Keys in the Amazon Simple Storage Service Developer Guide.

The KMS key policy must grant permission to the IAM role that you specify in your CreateModel request. For more information, see Using Key Policies in Amazon Web Services KMS in the Amazon Web Services Key Management Service Developer Guide.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_output.s3_output_path

The Amazon S3 path where you want Amazon SageMaker to store the results of the transform job. For example, s3://bucket-name/key-name-prefix.

For every S3 object used as input for the transform job, batch transform stores the transformed data with an .out suffix in a corresponding subfolder in the location in the output prefix. For example, for the input data stored at s3://bucket-name/input-name-prefix/dataset01/data.csv, batch transform stores the transformed data at s3://bucket-name/output-name-prefix/input-name-prefix/data.csv.out. Batch transform doesn't upload partially processed objects. For an input S3 object that contains multiple records, it creates an .out file only if the transform job succeeds on the entire file. When the input contains multiple S3 objects, the batch transform job processes the listed S3 objects and uploads only the output for successfully processed objects. If any object fails in the transform job batch transform marks the job as failed to prompt investigation.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_resources

Identifies the ML compute instances for the transform job.

Show child fields
model_package.validation_specification.validation_profiles[].transform_job_definition.transform_resources.instance_count

The number of ML compute instances to use in the transform job. The default value is 1, and the maximum is 100. For distributed transform jobs, specify a value greater than 1.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_resources.instance_type

The ML compute instance type for the transform job. If you are using built-in algorithms to transform moderately sized datasets, we recommend using ml.m4.xlarge or ml.m5.largeinstance types.

model_package.validation_specification.validation_profiles[].transform_job_definition.transform_resources.volume_kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt model data on the storage volume attached to the ML compute instance(s) that run the batch transform job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

The VolumeKmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

model_package.validation_specification.validation_role

The IAM roles to be used for the validation of the model package.

model_package_group

A group of versioned models in the model registry.

STRUCT(
    "model_package_group_name" VARCHAR,
    "model_package_group_arn" VARCHAR,
    "model_package_group_description" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "model_package_group_status" VARCHAR,
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
model_package_group.created_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
model_package_group.created_by.domain_id

The domain associated with the user.

model_package_group.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
model_package_group.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

model_package_group.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

model_package_group.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

model_package_group.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

model_package_group.created_by.user_profile_name

The name of the user's profile.

model_package_group.creation_time

The time that the model group was created.

model_package_group.model_package_group_arn

The Amazon Resource Name (ARN) of the model group.

model_package_group.model_package_group_description

The description for the model group.

model_package_group.model_package_group_name

The name of the model group.

model_package_group.model_package_group_status

The status of the model group. This can be one of the following values.

  • PENDING - The model group is pending being created.

  • IN_PROGRESS - The model group is in the process of being created.

  • COMPLETED - The model group was successfully created.

  • FAILED - The model group failed.

  • DELETING - The model group is in the process of being deleted.

  • DELETE_FAILED - SageMaker failed to delete the model group.

model_package_group.tags[]
Show child fields
model_package_group.tags[].key

The tag key. Tag keys must be unique per resource.

model_package_group.tags[].value

The tag value.

pipeline

A SageMaker Model Building Pipeline instance.

STRUCT(
    "pipeline_arn" VARCHAR,
    "pipeline_name" VARCHAR,
    "pipeline_display_name" VARCHAR,
    "pipeline_description" VARCHAR,
    "role_arn" VARCHAR,
    "pipeline_status" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "last_modified_time" TIMESTAMP_S,
    "last_run_time" TIMESTAMP_S,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "parallelism_configuration" STRUCT(
        "max_parallel_execution_steps" BIGINT
    ),
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
pipeline.created_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
pipeline.created_by.domain_id

The domain associated with the user.

pipeline.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
pipeline.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

pipeline.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

pipeline.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

pipeline.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

pipeline.created_by.user_profile_name

The name of the user's profile.

pipeline.creation_time

The creation time of the pipeline.

pipeline.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
pipeline.last_modified_by.domain_id

The domain associated with the user.

pipeline.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
pipeline.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

pipeline.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

pipeline.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

pipeline.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

pipeline.last_modified_by.user_profile_name

The name of the user's profile.

pipeline.last_modified_time

The time that the pipeline was last modified.

pipeline.last_run_time

The time when the pipeline was last run.

pipeline.parallelism_configuration

The parallelism configuration applied to the pipeline.

Show child fields
pipeline.parallelism_configuration.max_parallel_execution_steps

The max number of steps that can be executed in parallel.

pipeline.pipeline_arn

The Amazon Resource Name (ARN) of the pipeline.

pipeline.pipeline_description

The description of the pipeline.

pipeline.pipeline_display_name

The display name of the pipeline.

pipeline.pipeline_name

The name of the pipeline.

pipeline.pipeline_status

The status of the pipeline.

pipeline.role_arn

The Amazon Resource Name (ARN) of the role that created the pipeline.

pipeline.tags[]
Show child fields
pipeline.tags[].key

The tag key. Tag keys must be unique per resource.

pipeline.tags[].value

The tag value.

pipeline_execution

An execution of a pipeline.

STRUCT(
    "pipeline_arn" VARCHAR,
    "pipeline_execution_arn" VARCHAR,
    "pipeline_execution_display_name" VARCHAR,
    "pipeline_execution_status" VARCHAR,
    "pipeline_execution_description" VARCHAR,
    "pipeline_experiment_config" STRUCT(
        "experiment_name" VARCHAR,
        "trial_name" VARCHAR
    ),
    "failure_reason" VARCHAR,
    "creation_time" TIMESTAMP_S,
    "last_modified_time" TIMESTAMP_S,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "parallelism_configuration" STRUCT(
        "max_parallel_execution_steps" BIGINT
    ),
    "selective_execution_config" STRUCT(
        "source_pipeline_execution_arn" VARCHAR,
        "selected_steps" STRUCT(
            "step_name" VARCHAR
        )[]
    ),
    "pipeline_parameters" STRUCT(
        "name" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
pipeline_execution.created_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
pipeline_execution.created_by.domain_id

The domain associated with the user.

pipeline_execution.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
pipeline_execution.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

pipeline_execution.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

pipeline_execution.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

pipeline_execution.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

pipeline_execution.created_by.user_profile_name

The name of the user's profile.

pipeline_execution.creation_time

The creation time of the pipeline execution.

pipeline_execution.failure_reason

If the execution failed, a message describing why.

pipeline_execution.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
pipeline_execution.last_modified_by.domain_id

The domain associated with the user.

pipeline_execution.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
pipeline_execution.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

pipeline_execution.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

pipeline_execution.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

pipeline_execution.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

pipeline_execution.last_modified_by.user_profile_name

The name of the user's profile.

pipeline_execution.last_modified_time

The time that the pipeline execution was last modified.

pipeline_execution.parallelism_configuration

The parallelism configuration applied to the pipeline execution.

Show child fields
pipeline_execution.parallelism_configuration.max_parallel_execution_steps

The max number of steps that can be executed in parallel.

pipeline_execution.pipeline_arn

The Amazon Resource Name (ARN) of the pipeline that was executed.

pipeline_execution.pipeline_execution_arn

The Amazon Resource Name (ARN) of the pipeline execution.

pipeline_execution.pipeline_execution_description

The description of the pipeline execution.

pipeline_execution.pipeline_execution_display_name

The display name of the pipeline execution.

pipeline_execution.pipeline_execution_status

The status of the pipeline status.

pipeline_execution.pipeline_experiment_config

Specifies the names of the experiment and trial created by a pipeline.

Show child fields
pipeline_execution.pipeline_experiment_config.experiment_name

The name of the experiment.

pipeline_execution.pipeline_experiment_config.trial_name

The name of the trial.

pipeline_execution.pipeline_parameters[]
Show child fields
pipeline_execution.pipeline_parameters[].name

The name of the parameter to assign a value to. This parameter name must match a named parameter in the pipeline definition.

pipeline_execution.pipeline_parameters[].value

The literal value for the parameter.

pipeline_execution.selective_execution_config

The selective execution configuration applied to the pipeline run.

Show child fields
pipeline_execution.selective_execution_config.selected_steps[]
Show child fields
pipeline_execution.selective_execution_config.selected_steps[].step_name

The name of the pipeline step.

pipeline_execution.selective_execution_config.source_pipeline_execution_arn

The ARN from a reference execution of the current pipeline. Used to copy input collaterals needed for the selected steps to run. The execution status of the pipeline can be either Failed or Success.

This field is required if the steps you specify for SelectedSteps depend on output collaterals from any non-specified pipeline steps. For more information, see Selective Execution for Pipeline Steps.

project

The properties of a project.

STRUCT(
    "project_arn" VARCHAR,
    "project_name" VARCHAR,
    "project_id" VARCHAR,
    "project_description" VARCHAR,
    "service_catalog_provisioning_details" STRUCT(
        "product_id" VARCHAR,
        "provisioning_artifact_id" VARCHAR,
        "path_id" VARCHAR,
        "provisioning_parameters" STRUCT(
            "key" VARCHAR,
            "value" VARCHAR
        )[]
    ),
    "service_catalog_provisioned_product_details" STRUCT(
        "provisioned_product_id" VARCHAR,
        "provisioned_product_status_message" VARCHAR
    ),
    "project_status" VARCHAR,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "creation_time" TIMESTAMP_S,
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[],
    "last_modified_time" TIMESTAMP_S,
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    )
)
Show child fields
project.created_by

Who created the project.

Show child fields
project.created_by.domain_id

The domain associated with the user.

project.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
project.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

project.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

project.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

project.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

project.created_by.user_profile_name

The name of the user's profile.

project.creation_time

A timestamp specifying when the project was created.

project.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
project.last_modified_by.domain_id

The domain associated with the user.

project.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
project.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

project.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

project.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

project.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

project.last_modified_by.user_profile_name

The name of the user's profile.

project.last_modified_time

A timestamp container for when the project was last modified.

project.project_arn

The Amazon Resource Name (ARN) of the project.

project.project_description

The description of the project.

project.project_id

The ID of the project.

project.project_name

The name of the project.

project.project_status

The status of the project.

project.service_catalog_provisioned_product_details

Details of a provisioned service catalog product. For information about service catalog, see What is Amazon Web Services Service Catalog.

Show child fields
project.service_catalog_provisioned_product_details.provisioned_product_id

The ID of the provisioned product.

project.service_catalog_provisioned_product_details.provisioned_product_status_message

The current status of the product.

  • AVAILABLE - Stable state, ready to perform any operation. The most recent operation succeeded and completed.

  • UNDER_CHANGE - Transitive state. Operations performed might not have valid results. Wait for an AVAILABLE status before performing operations.

  • TAINTED - Stable state, ready to perform any operation. The stack has completed the requested operation but is not exactly what was requested. For example, a request to update to a new version failed and the stack rolled back to the current version.

  • ERROR - An unexpected error occurred. The provisioned product exists but the stack is not running. For example, CloudFormation received a parameter value that was not valid and could not launch the stack.

  • PLAN_IN_PROGRESS - Transitive state. The plan operations were performed to provision a new product, but resources have not yet been created. After reviewing the list of resources to be created, execute the plan. Wait for an AVAILABLE status before performing operations.

project.service_catalog_provisioning_details

Details that you specify to provision a service catalog product. For information about service catalog, see What is Amazon Web Services Service Catalog.

Show child fields
project.service_catalog_provisioning_details.path_id

The path identifier of the product. This value is optional if the product has a default path, and required if the product has more than one path.

project.service_catalog_provisioning_details.product_id

The ID of the product to provision.

project.service_catalog_provisioning_details.provisioning_artifact_id

The ID of the provisioning artifact.

project.service_catalog_provisioning_details.provisioning_parameters[]
Show child fields
project.service_catalog_provisioning_details.provisioning_parameters[].key

The key that identifies a provisioning parameter.

project.service_catalog_provisioning_details.provisioning_parameters[].value

The value of the provisioning parameter.

project.tags[]
Show child fields
project.tags[].key

The tag key. Tag keys must be unique per resource.

project.tags[].value

The tag value.

training_job

The properties of a training job.

STRUCT(
    "training_job_name" VARCHAR,
    "training_job_arn" VARCHAR,
    "tuning_job_arn" VARCHAR,
    "labeling_job_arn" VARCHAR,
    "auto_ml_job_arn" VARCHAR,
    "model_artifacts" STRUCT(
        "s3_model_artifacts" VARCHAR
    ),
    "training_job_status" VARCHAR,
    "secondary_status" VARCHAR,
    "failure_reason" VARCHAR,
    "hyper_parameters" MAP(VARCHAR, VARCHAR),
    "algorithm_specification" STRUCT(
        "training_image" VARCHAR,
        "algorithm_name" VARCHAR,
        "training_input_mode" VARCHAR,
        "metric_definitions" STRUCT(
            "name" VARCHAR,
            "regex" VARCHAR
        )[],
        "enable_sage_maker_metrics_time_series" BOOLEAN,
        "container_entrypoint" VARCHAR[],
        "container_arguments" VARCHAR[],
        "training_image_config" STRUCT(
            "training_repository_access_mode" VARCHAR,
            "training_repository_auth_config" STRUCT(
                "training_repository_credentials_provider_arn" VARCHAR
            )
        )
    ),
    "role_arn" VARCHAR,
    "input_data_config" STRUCT(
        "channel_name" VARCHAR,
        "data_source" STRUCT(
            "s3_data_source" STRUCT(
                "s3_data_type" VARCHAR,
                "s3_uri" VARCHAR,
                "s3_data_distribution_type" VARCHAR,
                "attribute_names" VARCHAR[],
                "instance_group_names" VARCHAR[]
            ),
            "file_system_data_source" STRUCT(
                "file_system_id" VARCHAR,
                "file_system_access_mode" VARCHAR,
                "file_system_type" VARCHAR,
                "directory_path" VARCHAR
            )
        ),
        "content_type" VARCHAR,
        "compression_type" VARCHAR,
        "record_wrapper_type" VARCHAR,
        "input_mode" VARCHAR,
        "shuffle_config" STRUCT(
            "seed" BIGINT
        )
    )[],
    "output_data_config" STRUCT(
        "kms_key_id" VARCHAR,
        "s3_output_path" VARCHAR,
        "compression_type" VARCHAR
    ),
    "resource_config" STRUCT(
        "instance_type" VARCHAR,
        "instance_count" BIGINT,
        "volume_size_in_gb" BIGINT,
        "volume_kms_key_id" VARCHAR,
        "keep_alive_period_in_seconds" BIGINT,
        "instance_groups" STRUCT(
            "instance_type" VARCHAR,
            "instance_count" BIGINT,
            "instance_group_name" VARCHAR
        )[]
    ),
    "vpc_config" STRUCT(
        "security_group_ids" VARCHAR[],
        "subnets" VARCHAR[]
    ),
    "stopping_condition" STRUCT(
        "max_runtime_in_seconds" BIGINT,
        "max_wait_time_in_seconds" BIGINT,
        "max_pending_time_in_seconds" BIGINT
    ),
    "creation_time" TIMESTAMP_S,
    "training_start_time" TIMESTAMP_S,
    "training_end_time" TIMESTAMP_S,
    "last_modified_time" TIMESTAMP_S,
    "secondary_status_transitions" STRUCT(
        "status" VARCHAR,
        "start_time" TIMESTAMP_S,
        "end_time" TIMESTAMP_S,
        "status_message" VARCHAR
    )[],
    "final_metric_data_list" STRUCT(
        "metric_name" VARCHAR,
        "value" DOUBLE,
        "timestamp" TIMESTAMP_S
    )[],
    "enable_network_isolation" BOOLEAN,
    "enable_inter_container_traffic_encryption" BOOLEAN,
    "enable_managed_spot_training" BOOLEAN,
    "checkpoint_config" STRUCT(
        "s3_uri" VARCHAR,
        "local_path" VARCHAR
    ),
    "training_time_in_seconds" BIGINT,
    "billable_time_in_seconds" BIGINT,
    "debug_hook_config" STRUCT(
        "local_path" VARCHAR,
        "s3_output_path" VARCHAR,
        "hook_parameters" MAP(VARCHAR, VARCHAR),
        "collection_configurations" STRUCT(
            "collection_name" VARCHAR,
            "collection_parameters" MAP(VARCHAR, VARCHAR)
        )[]
    ),
    "experiment_config" STRUCT(
        "experiment_name" VARCHAR,
        "trial_name" VARCHAR,
        "trial_component_display_name" VARCHAR,
        "run_name" VARCHAR
    ),
    "debug_rule_configurations" STRUCT(
        "rule_configuration_name" VARCHAR,
        "local_path" VARCHAR,
        "s3_output_path" VARCHAR,
        "rule_evaluator_image" VARCHAR,
        "instance_type" VARCHAR,
        "volume_size_in_gb" BIGINT,
        "rule_parameters" MAP(VARCHAR, VARCHAR)
    )[],
    "tensor_board_output_config" STRUCT(
        "local_path" VARCHAR,
        "s3_output_path" VARCHAR
    ),
    "debug_rule_evaluation_statuses" STRUCT(
        "rule_configuration_name" VARCHAR,
        "rule_evaluation_job_arn" VARCHAR,
        "rule_evaluation_status" VARCHAR,
        "status_details" VARCHAR,
        "last_modified_time" TIMESTAMP_S
    )[],
    "profiler_config" STRUCT(
        "s3_output_path" VARCHAR,
        "profiling_interval_in_milliseconds" BIGINT,
        "profiling_parameters" MAP(VARCHAR, VARCHAR),
        "disable_profiler" BOOLEAN
    ),
    "environment" MAP(VARCHAR, VARCHAR),
    "retry_strategy" STRUCT(
        "maximum_retry_attempts" BIGINT
    ),
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[]
)
Show child fields
training_job.algorithm_specification

Information about the algorithm used for training, and algorithm metadata.

Show child fields
training_job.algorithm_specification.algorithm_name

The name of the algorithm resource to use for the training job. This must be an algorithm resource that you created or subscribe to on Amazon Web Services Marketplace.

You must specify either the algorithm name to the AlgorithmName parameter or the image URI of the algorithm container to the TrainingImage parameter.

Note that the AlgorithmName parameter is mutually exclusive with the TrainingImage parameter. If you specify a value for the AlgorithmName parameter, you can't specify a value for TrainingImage, and vice versa.

If you specify values for both parameters, the training job might break; if you don't specify any value for both parameters, the training job might raise a null error.

training_job.algorithm_specification.container_arguments[]
training_job.algorithm_specification.container_entrypoint[]
training_job.algorithm_specification.enable_sage_maker_metrics_time_series

To generate and save time-series metrics during training, set to true. The default is false and time-series metrics aren't generated except in the following cases:

training_job.algorithm_specification.metric_definitions[]
Show child fields
training_job.algorithm_specification.metric_definitions[].name

The name of the metric.

training_job.algorithm_specification.metric_definitions[].regex

A regular expression that searches the output of a training job and gets the value of the metric. For more information about using regular expressions to define metrics, see Defining metrics and environment variables.

training_job.algorithm_specification.training_image

The registry path of the Docker image that contains the training algorithm. For information about docker registry paths for SageMaker built-in algorithms, see Docker Registry Paths and Example Code in the Amazon SageMaker developer guide. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information about using your custom training container, see Using Your Own Algorithms with Amazon SageMaker.

You must specify either the algorithm name to the AlgorithmName parameter or the image URI of the algorithm container to the TrainingImage parameter.

For more information, see the note in the AlgorithmName parameter description.

training_job.algorithm_specification.training_image_config

The configuration to use an image from a private Docker registry for a training job.

Show child fields
training_job.algorithm_specification.training_image_config.training_repository_access_mode

The method that your training job will use to gain access to the images in your private Docker registry. For access to an image in a private Docker registry, set to Vpc.

training_job.algorithm_specification.training_image_config.training_repository_auth_config

An object containing authentication information for a private Docker registry containing your training images.

Show child fields
training_job.algorithm_specification.training_image_config.training_repository_auth_config.training_repository_credentials_provider_arn

The Amazon Resource Name (ARN) of an Amazon Web Services Lambda function used to give SageMaker access credentials to your private Docker registry.

training_job.algorithm_specification.training_input_mode

The training input mode that the algorithm supports. For more information about input modes, see Algorithms.

Pipe mode

If an algorithm supports Pipe mode, Amazon SageMaker streams data directly from Amazon S3 to the container.

File mode

If an algorithm supports File mode, SageMaker downloads the training data from S3 to the provisioned ML storage volume, and mounts the directory to the Docker volume for the training container.

You must provision the ML storage volume with sufficient capacity to accommodate the data downloaded from S3. In addition to the training data, the ML storage volume also stores the output model. The algorithm container uses the ML storage volume to also store intermediate information, if any.

For distributed algorithms, training data is distributed uniformly. Your training duration is predictable if the input data objects sizes are approximately the same. SageMaker does not split the files any further for model training. If the object sizes are skewed, training won't be optimal as the data distribution is also skewed when one host in a training cluster is overloaded, thus becoming a bottleneck in training.

FastFile mode

If an algorithm supports FastFile mode, SageMaker streams data directly from S3 to the container with no code changes, and provides file system access to the data. Users can author their training script to interact with these files as if they were stored on disk.

FastFile mode works best when the data is read sequentially. Augmented manifest files aren't supported. The startup time is lower when there are fewer files in the S3 bucket provided.

training_job.auto_ml_job_arn

The Amazon Resource Name (ARN) of the job.

training_job.billable_time_in_seconds

The billable time in seconds.

training_job.checkpoint_config

Contains information about the output location for managed spot training checkpoint data.

Show child fields
training_job.checkpoint_config.local_path

(Optional) The local directory where checkpoints are written. The default directory is /opt/ml/checkpoints/.

training_job.checkpoint_config.s3_uri

Identifies the S3 path where you want SageMaker to store checkpoints. For example, s3://bucket-name/key-name-prefix.

training_job.creation_time

A timestamp that indicates when the training job was created.

training_job.debug_hook_config

Configuration information for the Amazon SageMaker Debugger hook parameters, metric and tensor collections, and storage paths. To learn more about how to configure the DebugHookConfig parameter, see Use the SageMaker and Debugger Configuration API Operations to Create, Update, and Debug Your Training Job.

Show child fields
training_job.debug_hook_config.collection_configurations[]
Show child fields
training_job.debug_hook_config.collection_configurations[].collection_name

The name of the tensor collection. The name must be unique relative to other rule configuration names.

training_job.debug_hook_config.collection_configurations[].collection_parameters

Parameter values for the tensor collection. The allowed parameters are "name", "include_regex", "reduction_config", "save_config", "tensor_names", and "save_histogram".

training_job.debug_hook_config.hook_parameters

Configuration information for the Amazon SageMaker Debugger hook parameters.

training_job.debug_hook_config.local_path

Path to local storage location for metrics and tensors. Defaults to /opt/ml/output/tensors/.

training_job.debug_hook_config.s3_output_path

Path to Amazon S3 storage location for metrics and tensors.

training_job.debug_rule_configurations[]
Show child fields
training_job.debug_rule_configurations[].instance_type

The instance type to deploy a custom rule for debugging a training job.

training_job.debug_rule_configurations[].local_path

Path to local storage location for output of rules. Defaults to /opt/ml/processing/output/rule/.

training_job.debug_rule_configurations[].rule_configuration_name

The name of the rule configuration. It must be unique relative to other rule configuration names.

training_job.debug_rule_configurations[].rule_evaluator_image

The Amazon Elastic Container (ECR) Image for the managed rule evaluation.

training_job.debug_rule_configurations[].rule_parameters

Runtime configuration for rule container.

training_job.debug_rule_configurations[].s3_output_path

Path to Amazon S3 storage location for rules.

training_job.debug_rule_configurations[].volume_size_in_gb

The size, in GB, of the ML storage volume attached to the processing instance.

training_job.debug_rule_evaluation_statuses[]
Show child fields
training_job.debug_rule_evaluation_statuses[].last_modified_time

Timestamp when the rule evaluation status was last modified.

training_job.debug_rule_evaluation_statuses[].rule_configuration_name

The name of the rule configuration.

training_job.debug_rule_evaluation_statuses[].rule_evaluation_job_arn

The Amazon Resource Name (ARN) of the rule evaluation job.

training_job.debug_rule_evaluation_statuses[].rule_evaluation_status

Status of the rule evaluation.

training_job.debug_rule_evaluation_statuses[].status_details

Details from the rule evaluation.

training_job.enable_inter_container_traffic_encryption

To encrypt all communications between ML compute instances in distributed training, choose True. Encryption provides greater security for distributed training, but training might take longer. How long it takes depends on the amount of communication between compute instances, especially if you use a deep learning algorithm in distributed training.

training_job.enable_managed_spot_training

When true, enables managed spot training using Amazon EC2 Spot instances to run training jobs instead of on-demand instances. For more information, see Managed Spot Training.

training_job.enable_network_isolation

If the TrainingJob was created with network isolation, the value is set to true. If network isolation is enabled, nodes can't communicate beyond the VPC they run in.

training_job.environment

The environment variables to set in the Docker container.

training_job.experiment_config

Associates a SageMaker job as a trial component with an experiment and trial. Specified when you call the following APIs:

Show child fields
training_job.experiment_config.experiment_name

The name of an existing experiment to associate with the trial component.

training_job.experiment_config.run_name

The name of the experiment run to associate with the trial component.

training_job.experiment_config.trial_component_display_name

The display name for the trial component. If this key isn't specified, the display name is the trial component name.

training_job.experiment_config.trial_name

The name of an existing trial to associate the trial component with. If not specified, a new trial is created.

training_job.failure_reason

If the training job failed, the reason it failed.

training_job.final_metric_data_list[]
Show child fields
training_job.final_metric_data_list[].metric_name

The name of the metric.

training_job.final_metric_data_list[].timestamp

The date and time that the algorithm emitted the metric.

training_job.final_metric_data_list[].value

The value of the metric.

training_job.hyper_parameters

Algorithm-specific parameters.

training_job.input_data_config[]
Show child fields
training_job.input_data_config[].channel_name

The name of the channel.

training_job.input_data_config[].compression_type

If training data is compressed, the compression type. The default value is None. CompressionType is used only in Pipe input mode. In File mode, leave this field unset or set it to None.

training_job.input_data_config[].content_type

The MIME type of the data.

training_job.input_data_config[].data_source

The location of the channel data.

Show child fields
training_job.input_data_config[].data_source.file_system_data_source

The file system that is associated with a channel.

Show child fields
training_job.input_data_config[].data_source.file_system_data_source.directory_path

The full path to the directory to associate with the channel.

training_job.input_data_config[].data_source.file_system_data_source.file_system_access_mode

The access mode of the mount of the directory associated with the channel. A directory can be mounted either in ro (read-only) or rw (read-write) mode.

training_job.input_data_config[].data_source.file_system_data_source.file_system_id

The file system id.

training_job.input_data_config[].data_source.file_system_data_source.file_system_type

The file system type.

training_job.input_data_config[].data_source.s3_data_source

The S3 location of the data source that is associated with a channel.

Show child fields
training_job.input_data_config[].data_source.s3_data_source.attribute_names[]
training_job.input_data_config[].data_source.s3_data_source.instance_group_names[]
training_job.input_data_config[].data_source.s3_data_source.s3_data_distribution_type

If you want SageMaker to replicate the entire dataset on each ML compute instance that is launched for model training, specify FullyReplicated.

If you want SageMaker to replicate a subset of data on each ML compute instance that is launched for model training, specify ShardedByS3Key. If there are n ML compute instances launched for a training job, each instance gets approximately 1/n of the number of S3 objects. In this case, model training on each machine uses only the subset of training data.

Don't choose more ML compute instances for training than available S3 objects. If you do, some nodes won't get any data and you will pay for nodes that aren't getting any training data. This applies in both File and Pipe modes. Keep this in mind when developing algorithms.

In distributed training, where you use multiple ML compute EC2 instances, you might choose ShardedByS3Key. If the algorithm requires copying training data to the ML storage volume (when TrainingInputMode is set to File), this copies 1/n of the number of objects.

training_job.input_data_config[].data_source.s3_data_source.s3_data_type

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix for model training.

If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want SageMaker to use for model training.

If you choose AugmentedManifestFile, S3Uri identifies an object that is an augmented manifest file in JSON lines format. This file contains the data you want to use for model training. AugmentedManifestFile can only be used if the Channel's input mode is Pipe.

training_job.input_data_config[].data_source.s3_data_source.s3_uri

Depending on the value specified for the S3DataType, identifies either a key name prefix or a manifest. For example:

  • A key name prefix might look like this: s3://bucketname/exampleprefix/

  • A manifest might look like this: s3://bucketname/example.manifest

    A manifest is an S3 object which is a JSON file consisting of an array of elements. The first element is a prefix which is followed by one or more suffixes. SageMaker appends the suffix elements to the prefix to get a full set of S3Uri. Note that the prefix must be a valid non-empty S3Uri that precludes users from specifying a manifest whose individual S3Uri is sourced from different S3 buckets.

    The following code example shows a valid manifest format:

    [ {"prefix": "s3://customer_bucket/some/prefix/"},

    "relative/path/to/custdata-1",

    "relative/path/custdata-2",

    ...

    "relative/path/custdata-N"

    ]

    This JSON is equivalent to the following S3Uri list:

    s3://customer_bucket/some/prefix/relative/path/to/custdata-1

    s3://customer_bucket/some/prefix/relative/path/custdata-2

    ...

    s3://customer_bucket/some/prefix/relative/path/custdata-N

    The complete set of S3Uri in this manifest is the input data for the channel for this data source. The object that each S3Uri points to must be readable by the IAM role that SageMaker uses to perform tasks on your behalf.

Your input bucket must be located in same Amazon Web Services region as your training job.

training_job.input_data_config[].input_mode

(Optional) The input mode to use for the data channel in a training job. If you don't set a value for InputMode, SageMaker uses the value set for TrainingInputMode. Use this parameter to override the TrainingInputMode setting in a AlgorithmSpecification request when you have a channel that needs a different input mode from the training job's general setting. To download the data from Amazon Simple Storage Service (Amazon S3) to the provisioned ML storage volume, and mount the directory to a Docker volume, use File input mode. To stream data directly from Amazon S3 to the container, choose Pipe input mode.

To use a model for incremental training, choose File input model.

training_job.input_data_config[].record_wrapper_type

Specify RecordIO as the value when input data is in raw format but the training algorithm requires the RecordIO format. In this case, SageMaker wraps each individual S3 object in a RecordIO record. If the input data is already in RecordIO format, you don't need to set this attribute. For more information, see Create a Dataset Using RecordIO.

In File mode, leave this field unset or set it to None.

training_job.input_data_config[].shuffle_config

A configuration for a shuffle option for input data in a channel. If you use S3Prefix for S3DataType, this shuffles the results of the S3 key prefix matches. If you use ManifestFile, the order of the S3 object references in the ManifestFile is shuffled. If you use AugmentedManifestFile, the order of the JSON lines in the AugmentedManifestFile is shuffled. The shuffling order is determined using the Seed value.

For Pipe input mode, shuffling is done at the start of every epoch. With large datasets this ensures that the order of the training data is different for each epoch, it helps reduce bias and possible overfitting. In a multi-node training job when ShuffleConfig is combined with S3DataDistributionType of ShardedByS3Key, the data is shuffled across nodes so that the content sent to a particular node on the first epoch might be sent to a different node on the second epoch.

Show child fields
training_job.input_data_config[].shuffle_config.seed

Determines the shuffling order in ShuffleConfig value.

training_job.labeling_job_arn

The Amazon Resource Name (ARN) of the labeling job.

training_job.last_modified_time

A timestamp that indicates when the status of the training job was last modified.

training_job.model_artifacts

Information about the Amazon S3 location that is configured for storing model artifacts.

Show child fields
training_job.model_artifacts.s3_model_artifacts

The path of the S3 object that contains the model artifacts. For example, s3://bucket-name/keynameprefix/model.tar.gz.

training_job.output_data_config

The S3 path where model artifacts that you configured when creating the job are stored. SageMaker creates subfolders for model artifacts.

Show child fields
training_job.output_data_config.compression_type

The model output compression type. Select None to output an uncompressed model, recommended for large model outputs. Defaults to gzip.

training_job.output_data_config.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption. The KmsKeyId can be any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

  • // KMS Key Alias

    "alias/ExampleAlias"

  • // Amazon Resource Name (ARN) of a KMS Key Alias

    "arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias"

If you use a KMS key ID or an alias of your KMS key, the SageMaker execution role must include permissions to call kms:Encrypt. If you don't provide a KMS key ID, SageMaker uses the default KMS key for Amazon S3 for your role's account. For more information, see KMS-Managed Encryption Keys in the Amazon Simple Storage Service Developer Guide. If the output data is stored in Amazon S3 Express One Zone, it is encrypted with server-side encryption with Amazon S3 managed keys (SSE-S3). KMS key is not supported for Amazon S3 Express One Zone

The KMS key policy must grant permission to the IAM role that you specify in your CreateTrainingJob, CreateTransformJob, or CreateHyperParameterTuningJob requests. For more information, see Using Key Policies in Amazon Web Services KMS in the Amazon Web Services Key Management Service Developer Guide.

training_job.output_data_config.s3_output_path

Identifies the S3 path where you want SageMaker to store the model artifacts. For example, s3://bucket-name/key-name-prefix.

training_job.profiler_config

Configuration information for Amazon SageMaker Debugger system monitoring, framework profiling, and storage paths.

Show child fields
training_job.profiler_config.disable_profiler

Configuration to turn off Amazon SageMaker Debugger's system monitoring and profiling functionality. To turn it off, set to True.

training_job.profiler_config.profiling_interval_in_milliseconds

A time interval for capturing system metrics in milliseconds. Available values are 100, 200, 500, 1000 (1 second), 5000 (5 seconds), and 60000 (1 minute) milliseconds. The default value is 500 milliseconds.

training_job.profiler_config.profiling_parameters

Configuration information for capturing framework metrics. Available key strings for different profiling options are DetailedProfilingConfig, PythonProfilingConfig, and DataLoaderProfilingConfig. The following codes are configuration structures for the ProfilingParameters parameter. To learn more about how to configure the ProfilingParameters parameter, see Use the SageMaker and Debugger Configuration API Operations to Create, Update, and Debug Your Training Job.

training_job.profiler_config.s3_output_path

Path to Amazon S3 storage location for system and framework metrics.

training_job.resource_config

Resources, including ML compute instances and ML storage volumes, that are configured for model training.

Show child fields
training_job.resource_config.instance_count

The number of ML compute instances to use. For distributed training, provide a value greater than 1.

training_job.resource_config.instance_groups[]
Show child fields
training_job.resource_config.instance_groups[].instance_count

Specifies the number of instances of the instance group.

training_job.resource_config.instance_groups[].instance_group_name

Specifies the name of the instance group.

training_job.resource_config.instance_groups[].instance_type

Specifies the instance type of the instance group.

training_job.resource_config.instance_type

The ML compute instance type.

SageMaker Training on Amazon Elastic Compute Cloud (EC2) P4de instances is in preview release starting December 9th, 2022.

Amazon EC2 P4de instances (currently in preview) are powered by 8 NVIDIA A100 GPUs with 80GB high-performance HBM2e GPU memory, which accelerate the speed of training ML models that need to be trained on large datasets of high-resolution data. In this preview release, Amazon SageMaker supports ML training jobs on P4de instances (ml.p4de.24xlarge) to reduce model training time. The ml.p4de.24xlarge instances are available in the following Amazon Web Services Regions.

  • US East (N. Virginia) (us-east-1)

  • US West (Oregon) (us-west-2)

To request quota limit increase and start using P4de instances, contact the SageMaker Training service team through your account team.

training_job.resource_config.keep_alive_period_in_seconds

The duration of time in seconds to retain configured resources in a warm pool for subsequent training jobs.

training_job.resource_config.volume_kms_key_id

The Amazon Web Services KMS key that SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) that run the training job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

The VolumeKmsKeyId can be in any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

training_job.resource_config.volume_size_in_gb

The size of the ML storage volume that you want to provision.

ML storage volumes store model artifacts and incremental states. Training algorithms might also use the ML storage volume for scratch space. If you want to store the training data in the ML storage volume, choose File as the TrainingInputMode in the algorithm specification.

When using an ML instance with NVMe SSD volumes, SageMaker doesn't provision Amazon EBS General Purpose SSD (gp2) storage. Available storage is fixed to the NVMe-type instance's storage capacity. SageMaker configures storage paths for training datasets, checkpoints, model artifacts, and outputs to use the entire capacity of the instance storage. For example, ML instance families with the NVMe-type instance storage include ml.p4d, ml.g4dn, and ml.g5.

When using an ML instance with the EBS-only storage option and without instance storage, you must define the size of EBS volume through VolumeSizeInGB in the ResourceConfig API. For example, ML instance families that use EBS volumes include ml.c5 and ml.p2.

To look up instance types and their instance storage types and volumes, see Amazon EC2 Instance Types.

To find the default local paths defined by the SageMaker training platform, see Amazon SageMaker Training Storage Folders for Training Datasets, Checkpoints, Model Artifacts, and Outputs.

training_job.retry_strategy

The number of times to retry the job when the job fails due to an InternalServerError.

Show child fields
training_job.retry_strategy.maximum_retry_attempts

The number of times to retry the job. When the job is retried, it's SecondaryStatus is changed to STARTING.

training_job.role_arn

The Amazon Web Services Identity and Access Management (IAM) role configured for the training job.

training_job.secondary_status

Provides detailed information about the state of the training job. For detailed information about the secondary status of the training job, see StatusMessage under SecondaryStatusTransition.

SageMaker provides primary statuses and secondary statuses that apply to each of them:

InProgress

  • Starting - Starting the training job.

  • Downloading - An optional stage for algorithms that support File training input mode. It indicates that data is being downloaded to the ML storage volumes.

  • Training - Training is in progress.

  • Uploading - Training is complete and the model artifacts are being uploaded to the S3 location.

Completed

  • Completed - The training job has completed.

Failed

  • Failed - The training job has failed. The reason for the failure is returned in the FailureReason field of DescribeTrainingJobResponse.

Stopped

  • MaxRuntimeExceeded - The job stopped because it exceeded the maximum allowed runtime.

  • Stopped - The training job has stopped.

Stopping

  • Stopping - Stopping the training job.

Valid values for SecondaryStatus are subject to change.

We no longer support the following secondary statuses:

  • LaunchingMLInstances

  • PreparingTrainingStack

  • DownloadingTrainingImage

training_job.secondary_status_transitions[]
Show child fields
training_job.secondary_status_transitions[].end_time

A timestamp that shows when the training job transitioned out of this secondary status state into another secondary status state or when the training job has ended.

training_job.secondary_status_transitions[].start_time

A timestamp that shows when the training job transitioned to the current secondary status state.

training_job.secondary_status_transitions[].status

Contains a secondary status information from a training job.

Status might be one of the following secondary statuses:

InProgress

  • Starting - Starting the training job.

  • Downloading - An optional stage for algorithms that support File training input mode. It indicates that data is being downloaded to the ML storage volumes.

  • Training - Training is in progress.

  • Uploading - Training is complete and the model artifacts are being uploaded to the S3 location.

Completed

  • Completed - The training job has completed.

Failed

  • Failed - The training job has failed. The reason for the failure is returned in the FailureReason field of DescribeTrainingJobResponse.

Stopped

  • MaxRuntimeExceeded - The job stopped because it exceeded the maximum allowed runtime.

  • Stopped - The training job has stopped.

Stopping

  • Stopping - Stopping the training job.

We no longer support the following secondary statuses:

  • LaunchingMLInstances

  • PreparingTrainingStack

  • DownloadingTrainingImage

training_job.secondary_status_transitions[].status_message

A detailed description of the progress within a secondary status.

SageMaker provides secondary statuses and status messages that apply to each of them:

Starting

  • Starting the training job.

  • Launching requested ML instances.

  • Insufficient capacity error from EC2 while launching instances, retrying!

  • Launched instance was unhealthy, replacing it!

  • Preparing the instances for training.

Training

  • Training image download completed. Training in progress.

Status messages are subject to change. Therefore, we recommend not including them in code that programmatically initiates actions. For examples, don't use status messages in if statements.

To have an overview of your training job's progress, view TrainingJobStatus and SecondaryStatus in DescribeTrainingJob, and StatusMessage together. For example, at the start of a training job, you might see the following:

  • TrainingJobStatus - InProgress

  • SecondaryStatus - Training

  • StatusMessage - Downloading the training image

training_job.stopping_condition

Specifies a limit to how long a model training job can run. It also specifies how long a managed Spot training job has to complete. When the job reaches the time limit, SageMaker ends the training job. Use this API to cap model training costs.

To stop a job, SageMaker sends the algorithm the SIGTERM signal, which delays job termination for 120 seconds. Algorithms can use this 120-second window to save the model artifacts, so the results of training are not lost.

Show child fields
training_job.stopping_condition.max_pending_time_in_seconds

The maximum length of time, in seconds, that a training or compilation job can be pending before it is stopped.

training_job.stopping_condition.max_runtime_in_seconds

The maximum length of time, in seconds, that a training or compilation job can run before it is stopped.

For compilation jobs, if the job does not complete during this time, a TimeOut error is generated. We recommend starting with 900 seconds and increasing as necessary based on your model.

For all other jobs, if the job does not complete during this time, SageMaker ends the job. When RetryStrategy is specified in the job request, MaxRuntimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt. The default value is 1 day. The maximum value is 28 days.

The maximum time that a TrainingJob can run in total, including any time spent publishing metrics or archiving and uploading models after it has been stopped, is 30 days.

training_job.stopping_condition.max_wait_time_in_seconds

The maximum length of time, in seconds, that a managed Spot training job has to complete. It is the amount of time spent waiting for Spot capacity plus the amount of time the job can run. It must be equal to or greater than MaxRuntimeInSeconds. If the job does not complete during this time, SageMaker ends the job.

When RetryStrategy is specified in the job request, MaxWaitTimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt.

training_job.tags[]
Show child fields
training_job.tags[].key

The tag key. Tag keys must be unique per resource.

training_job.tags[].value

The tag value.

training_job.tensor_board_output_config

Configuration of storage locations for the Amazon SageMaker Debugger TensorBoard output data.

Show child fields
training_job.tensor_board_output_config.local_path

Path to local storage location for tensorBoard output. Defaults to /opt/ml/output/tensorboard.

training_job.tensor_board_output_config.s3_output_path

Path to Amazon S3 storage location for TensorBoard output.

training_job.training_end_time

Indicates the time when the training job ends on training instances. You are billed for the time interval between the value of TrainingStartTime and this time. For successful jobs and stopped jobs, this is the time after model artifacts are uploaded. For failed jobs, this is the time when SageMaker detects a job failure.

training_job.training_job_arn

The Amazon Resource Name (ARN) of the training job.

training_job.training_job_name

The name of the training job.

training_job.training_job_status

The status of the training job.

Training job statuses are:

  • InProgress - The training is in progress.

  • Completed - The training job has completed.

  • Failed - The training job has failed. To see the reason for the failure, see the FailureReason field in the response to a DescribeTrainingJobResponse call.

  • Stopping - The training job is stopping.

  • Stopped - The training job has stopped.

For more detailed information, see SecondaryStatus.

training_job.training_start_time

Indicates the time when the training job starts on training instances. You are billed for the time interval between this time and the value of TrainingEndTime. The start time in CloudWatch Logs might be later than this time. The difference is due to the time it takes to download the training data and to the size of the training container.

training_job.training_time_in_seconds

The training time in seconds.

training_job.tuning_job_arn

The Amazon Resource Name (ARN) of the associated hyperparameter tuning job if the training job was launched by a hyperparameter tuning job.

training_job.vpc_config

A VpcConfig object that specifies the VPC that this training job has access to. For more information, see Protect Training Jobs by Using an Amazon Virtual Private Cloud.

Show child fields
training_job.vpc_config.security_group_ids[]
training_job.vpc_config.subnets[]
trial

The properties of a trial.

STRUCT(
    "trial_name" VARCHAR,
    "trial_arn" VARCHAR,
    "display_name" VARCHAR,
    "experiment_name" VARCHAR,
    "source" STRUCT(
        "source_arn" VARCHAR,
        "source_type" VARCHAR
    ),
    "creation_time" TIMESTAMP_S,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "last_modified_time" TIMESTAMP_S,
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "metadata_properties" STRUCT(
        "commit_id" VARCHAR,
        "repository" VARCHAR,
        "generated_by" VARCHAR,
        "project_id" VARCHAR
    ),
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[],
    "trial_component_summaries" STRUCT(
        "trial_component_name" VARCHAR,
        "trial_component_arn" VARCHAR,
        "trial_component_source" STRUCT(
            "source_arn" VARCHAR,
            "source_type" VARCHAR
        ),
        "creation_time" TIMESTAMP_S,
        "created_by" STRUCT(
            "user_profile_arn" VARCHAR,
            "user_profile_name" VARCHAR,
            "domain_id" VARCHAR,
            "iam_identity" STRUCT(
                "arn" VARCHAR,
                "principal_id" VARCHAR,
                "source_identity" VARCHAR
            )
        )
    )[]
)
Show child fields
trial.created_by

Who created the trial.

Show child fields
trial.created_by.domain_id

The domain associated with the user.

trial.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
trial.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

trial.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

trial.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

trial.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

trial.created_by.user_profile_name

The name of the user's profile.

trial.creation_time

When the trial was created.

trial.display_name

The name of the trial as displayed. If DisplayName isn't specified, TrialName is displayed.

trial.experiment_name

The name of the experiment the trial is part of.

trial.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
trial.last_modified_by.domain_id

The domain associated with the user.

trial.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
trial.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

trial.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

trial.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

trial.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

trial.last_modified_by.user_profile_name

The name of the user's profile.

trial.last_modified_time

Who last modified the trial.

trial.metadata_properties

Metadata properties of the tracking entity, trial, or trial component.

Show child fields
trial.metadata_properties.commit_id

The commit ID.

trial.metadata_properties.generated_by

The entity this entity was generated by.

trial.metadata_properties.project_id

The project ID.

trial.metadata_properties.repository

The repository.

trial.source

The source of the trial.

Show child fields
trial.source.source_arn

The Amazon Resource Name (ARN) of the source.

trial.source.source_type

The source job type.

trial.tags[]
Show child fields
trial.tags[].key

The tag key. Tag keys must be unique per resource.

trial.tags[].value

The tag value.

trial.trial_arn

The Amazon Resource Name (ARN) of the trial.

trial.trial_component_summaries[]
Show child fields
trial.trial_component_summaries[].created_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
trial.trial_component_summaries[].created_by.domain_id

The domain associated with the user.

trial.trial_component_summaries[].created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
trial.trial_component_summaries[].created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

trial.trial_component_summaries[].created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

trial.trial_component_summaries[].created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

trial.trial_component_summaries[].created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

trial.trial_component_summaries[].created_by.user_profile_name

The name of the user's profile.

trial.trial_component_summaries[].creation_time

When the component was created.

trial.trial_component_summaries[].trial_component_arn

The Amazon Resource Name (ARN) of the trial component.

trial.trial_component_summaries[].trial_component_name

The name of the trial component.

trial.trial_component_summaries[].trial_component_source

The Amazon Resource Name (ARN) and job type of the source of a trial component.

Show child fields
trial.trial_component_summaries[].trial_component_source.source_arn

The source Amazon Resource Name (ARN).

trial.trial_component_summaries[].trial_component_source.source_type

The source job type.

trial.trial_name

The name of the trial.

trial_component

The properties of a trial component.

STRUCT(
    "trial_component_name" VARCHAR,
    "display_name" VARCHAR,
    "trial_component_arn" VARCHAR,
    "source" STRUCT(
        "source_arn" VARCHAR,
        "source_type" VARCHAR
    ),
    "status" STRUCT(
        "primary_status" VARCHAR,
        "message" VARCHAR
    ),
    "start_time" TIMESTAMP_S,
    "end_time" TIMESTAMP_S,
    "creation_time" TIMESTAMP_S,
    "created_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "last_modified_time" TIMESTAMP_S,
    "last_modified_by" STRUCT(
        "user_profile_arn" VARCHAR,
        "user_profile_name" VARCHAR,
        "domain_id" VARCHAR,
        "iam_identity" STRUCT(
            "arn" VARCHAR,
            "principal_id" VARCHAR,
            "source_identity" VARCHAR
        )
    ),
    "parameters" MAP(VARCHAR, STRUCT(
            "string_value" VARCHAR,
            "number_value" DOUBLE
        )),
    "input_artifacts" MAP(VARCHAR, STRUCT(
            "media_type" VARCHAR,
            "value" VARCHAR
        )),
    "output_artifacts" MAP(VARCHAR, STRUCT(
            "media_type" VARCHAR,
            "value" VARCHAR
        )),
    "metrics" STRUCT(
        "metric_name" VARCHAR,
        "source_arn" VARCHAR,
        "time_stamp" TIMESTAMP_S,
        "max" DOUBLE,
        "min" DOUBLE,
        "last" DOUBLE,
        "count" BIGINT,
        "avg" DOUBLE,
        "std_dev" DOUBLE
    )[],
    "metadata_properties" STRUCT(
        "commit_id" VARCHAR,
        "repository" VARCHAR,
        "generated_by" VARCHAR,
        "project_id" VARCHAR
    ),
    "source_detail" STRUCT(
        "source_arn" VARCHAR,
        "training_job" STRUCT(
            "training_job_name" VARCHAR,
            "training_job_arn" VARCHAR,
            "tuning_job_arn" VARCHAR,
            "labeling_job_arn" VARCHAR,
            "auto_ml_job_arn" VARCHAR,
            "model_artifacts" STRUCT(
                "s3_model_artifacts" VARCHAR
            ),
            "training_job_status" VARCHAR,
            "secondary_status" VARCHAR,
            "failure_reason" VARCHAR,
            "hyper_parameters" MAP(VARCHAR, VARCHAR),
            "algorithm_specification" STRUCT(
                "training_image" VARCHAR,
                "algorithm_name" VARCHAR,
                "training_input_mode" VARCHAR,
                "metric_definitions" STRUCT(
                    "name" VARCHAR,
                    "regex" VARCHAR
                )[],
                "enable_sage_maker_metrics_time_series" BOOLEAN,
                "container_entrypoint" VARCHAR[],
                "container_arguments" VARCHAR[],
                "training_image_config" STRUCT(
                    "training_repository_access_mode" VARCHAR,
                    "training_repository_auth_config" STRUCT(
                        "training_repository_credentials_provider_arn" VARCHAR
                    )
                )
            ),
            "role_arn" VARCHAR,
            "input_data_config" STRUCT(
                "channel_name" VARCHAR,
                "data_source" STRUCT(
                    "s3_data_source" STRUCT(
                        "s3_data_type" VARCHAR,
                        "s3_uri" VARCHAR,
                        "s3_data_distribution_type" VARCHAR,
                        "attribute_names" VARCHAR[],
                        "instance_group_names" VARCHAR[]
                    ),
                    "file_system_data_source" STRUCT(
                        "file_system_id" VARCHAR,
                        "file_system_access_mode" VARCHAR,
                        "file_system_type" VARCHAR,
                        "directory_path" VARCHAR
                    )
                ),
                "content_type" VARCHAR,
                "compression_type" VARCHAR,
                "record_wrapper_type" VARCHAR,
                "input_mode" VARCHAR,
                "shuffle_config" STRUCT(
                    "seed" BIGINT
                )
            )[],
            "output_data_config" STRUCT(
                "kms_key_id" VARCHAR,
                "s3_output_path" VARCHAR,
                "compression_type" VARCHAR
            ),
            "resource_config" STRUCT(
                "instance_type" VARCHAR,
                "instance_count" BIGINT,
                "volume_size_in_gb" BIGINT,
                "volume_kms_key_id" VARCHAR,
                "keep_alive_period_in_seconds" BIGINT,
                "instance_groups" STRUCT(
                    "instance_type" VARCHAR,
                    "instance_count" BIGINT,
                    "instance_group_name" VARCHAR
                )[]
            ),
            "vpc_config" STRUCT(
                "security_group_ids" VARCHAR[],
                "subnets" VARCHAR[]
            ),
            "stopping_condition" STRUCT(
                "max_runtime_in_seconds" BIGINT,
                "max_wait_time_in_seconds" BIGINT,
                "max_pending_time_in_seconds" BIGINT
            ),
            "creation_time" TIMESTAMP_S,
            "training_start_time" TIMESTAMP_S,
            "training_end_time" TIMESTAMP_S,
            "last_modified_time" TIMESTAMP_S,
            "secondary_status_transitions" STRUCT(
                "status" VARCHAR,
                "start_time" TIMESTAMP_S,
                "end_time" TIMESTAMP_S,
                "status_message" VARCHAR
            )[],
            "final_metric_data_list" STRUCT(
                "metric_name" VARCHAR,
                "value" DOUBLE,
                "timestamp" TIMESTAMP_S
            )[],
            "enable_network_isolation" BOOLEAN,
            "enable_inter_container_traffic_encryption" BOOLEAN,
            "enable_managed_spot_training" BOOLEAN,
            "checkpoint_config" STRUCT(
                "s3_uri" VARCHAR,
                "local_path" VARCHAR
            ),
            "training_time_in_seconds" BIGINT,
            "billable_time_in_seconds" BIGINT,
            "debug_hook_config" STRUCT(
                "local_path" VARCHAR,
                "s3_output_path" VARCHAR,
                "hook_parameters" MAP(VARCHAR, VARCHAR),
                "collection_configurations" STRUCT(
                    "collection_name" VARCHAR,
                    "collection_parameters" MAP(VARCHAR, VARCHAR)
                )[]
            ),
            "experiment_config" STRUCT(
                "experiment_name" VARCHAR,
                "trial_name" VARCHAR,
                "trial_component_display_name" VARCHAR,
                "run_name" VARCHAR
            ),
            "debug_rule_configurations" STRUCT(
                "rule_configuration_name" VARCHAR,
                "local_path" VARCHAR,
                "s3_output_path" VARCHAR,
                "rule_evaluator_image" VARCHAR,
                "instance_type" VARCHAR,
                "volume_size_in_gb" BIGINT,
                "rule_parameters" MAP(VARCHAR, VARCHAR)
            )[],
            "tensor_board_output_config" STRUCT(
                "local_path" VARCHAR,
                "s3_output_path" VARCHAR
            ),
            "debug_rule_evaluation_statuses" STRUCT(
                "rule_configuration_name" VARCHAR,
                "rule_evaluation_job_arn" VARCHAR,
                "rule_evaluation_status" VARCHAR,
                "status_details" VARCHAR,
                "last_modified_time" TIMESTAMP_S
            )[],
            "profiler_config" STRUCT(
                "s3_output_path" VARCHAR,
                "profiling_interval_in_milliseconds" BIGINT,
                "profiling_parameters" MAP(VARCHAR, VARCHAR),
                "disable_profiler" BOOLEAN
            ),
            "environment" MAP(VARCHAR, VARCHAR),
            "retry_strategy" STRUCT(
                "maximum_retry_attempts" BIGINT
            ),
            "tags" STRUCT(
                "key" VARCHAR,
                "value" VARCHAR
            )[]
        ),
        "processing_job" STRUCT(
            "processing_inputs" STRUCT(
                "input_name" VARCHAR,
                "app_managed" BOOLEAN,
                "s3_input" STRUCT(
                    "s3_uri" VARCHAR,
                    "local_path" VARCHAR,
                    "s3_data_type" VARCHAR,
                    "s3_input_mode" VARCHAR,
                    "s3_data_distribution_type" VARCHAR,
                    "s3_compression_type" VARCHAR
                ),
                "dataset_definition" STRUCT(
                    "athena_dataset_definition" STRUCT(
                        "catalog" VARCHAR,
                        "database" VARCHAR,
                        "query_string" VARCHAR,
                        "work_group" VARCHAR,
                        "output_s3_uri" VARCHAR,
                        "kms_key_id" VARCHAR,
                        "output_format" VARCHAR,
                        "output_compression" VARCHAR
                    ),
                    "redshift_dataset_definition" STRUCT(
                        "cluster_id" VARCHAR,
                        "database" VARCHAR,
                        "db_user" VARCHAR,
                        "query_string" VARCHAR,
                        "cluster_role_arn" VARCHAR,
                        "output_s3_uri" VARCHAR,
                        "kms_key_id" VARCHAR,
                        "output_format" VARCHAR,
                        "output_compression" VARCHAR
                    ),
                    "local_path" VARCHAR,
                    "data_distribution_type" VARCHAR,
                    "input_mode" VARCHAR
                )
            )[],
            "processing_output_config" STRUCT(
                "outputs" STRUCT(
                    "output_name" VARCHAR,
                    "s3_output" STRUCT(
                        "s3_uri" VARCHAR,
                        "local_path" VARCHAR,
                        "s3_upload_mode" VARCHAR
                    ),
                    "feature_store_output" STRUCT(
                        "feature_group_name" VARCHAR
                    ),
                    "app_managed" BOOLEAN
                )[],
                "kms_key_id" VARCHAR
            ),
            "processing_job_name" VARCHAR,
            "processing_resources" STRUCT(
                "cluster_config" STRUCT(
                    "instance_count" BIGINT,
                    "instance_type" VARCHAR,
                    "volume_size_in_gb" BIGINT,
                    "volume_kms_key_id" VARCHAR
                )
            ),
            "stopping_condition" STRUCT(
                "max_runtime_in_seconds" BIGINT
            ),
            "app_specification" STRUCT(
                "image_uri" VARCHAR,
                "container_entrypoint" VARCHAR[],
                "container_arguments" VARCHAR[]
            ),
            "environment" MAP(VARCHAR, VARCHAR),
            "network_config" STRUCT(
                "enable_inter_container_traffic_encryption" BOOLEAN,
                "enable_network_isolation" BOOLEAN,
                "vpc_config" STRUCT(
                    "security_group_ids" VARCHAR[],
                    "subnets" VARCHAR[]
                )
            ),
            "role_arn" VARCHAR,
            "experiment_config" STRUCT(
                "experiment_name" VARCHAR,
                "trial_name" VARCHAR,
                "trial_component_display_name" VARCHAR,
                "run_name" VARCHAR
            ),
            "processing_job_arn" VARCHAR,
            "processing_job_status" VARCHAR,
            "exit_message" VARCHAR,
            "failure_reason" VARCHAR,
            "processing_end_time" TIMESTAMP_S,
            "processing_start_time" TIMESTAMP_S,
            "last_modified_time" TIMESTAMP_S,
            "creation_time" TIMESTAMP_S,
            "monitoring_schedule_arn" VARCHAR,
            "auto_ml_job_arn" VARCHAR,
            "training_job_arn" VARCHAR,
            "tags" STRUCT(
                "key" VARCHAR,
                "value" VARCHAR
            )[]
        ),
        "transform_job" STRUCT(
            "transform_job_name" VARCHAR,
            "transform_job_arn" VARCHAR,
            "transform_job_status" VARCHAR,
            "failure_reason" VARCHAR,
            "model_name" VARCHAR,
            "max_concurrent_transforms" BIGINT,
            "model_client_config" STRUCT(
                "invocations_timeout_in_seconds" BIGINT,
                "invocations_max_retries" BIGINT
            ),
            "max_payload_in_mb" BIGINT,
            "batch_strategy" VARCHAR,
            "environment" MAP(VARCHAR, VARCHAR),
            "transform_input" STRUCT(
                "data_source" STRUCT(
                    "s3_data_source" STRUCT(
                        "s3_data_type" VARCHAR,
                        "s3_uri" VARCHAR
                    )
                ),
                "content_type" VARCHAR,
                "compression_type" VARCHAR,
                "split_type" VARCHAR
            ),
            "transform_output" STRUCT(
                "s3_output_path" VARCHAR,
                "accept" VARCHAR,
                "assemble_with" VARCHAR,
                "kms_key_id" VARCHAR
            ),
            "data_capture_config" STRUCT(
                "destination_s3_uri" VARCHAR,
                "kms_key_id" VARCHAR,
                "generate_inference_id" BOOLEAN
            ),
            "transform_resources" STRUCT(
                "instance_type" VARCHAR,
                "instance_count" BIGINT,
                "volume_kms_key_id" VARCHAR
            ),
            "creation_time" TIMESTAMP_S,
            "transform_start_time" TIMESTAMP_S,
            "transform_end_time" TIMESTAMP_S,
            "labeling_job_arn" VARCHAR,
            "auto_ml_job_arn" VARCHAR,
            "data_processing" STRUCT(
                "input_filter" VARCHAR,
                "output_filter" VARCHAR,
                "join_source" VARCHAR
            ),
            "experiment_config" STRUCT(
                "experiment_name" VARCHAR,
                "trial_name" VARCHAR,
                "trial_component_display_name" VARCHAR,
                "run_name" VARCHAR
            ),
            "tags" STRUCT(
                "key" VARCHAR,
                "value" VARCHAR
            )[]
        )
    ),
    "lineage_group_arn" VARCHAR,
    "tags" STRUCT(
        "key" VARCHAR,
        "value" VARCHAR
    )[],
    "parents" STRUCT(
        "trial_name" VARCHAR,
        "experiment_name" VARCHAR
    )[],
    "run_name" VARCHAR
)
Show child fields
trial_component.created_by

Who created the trial component.

Show child fields
trial_component.created_by.domain_id

The domain associated with the user.

trial_component.created_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
trial_component.created_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

trial_component.created_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

trial_component.created_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

trial_component.created_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

trial_component.created_by.user_profile_name

The name of the user's profile.

trial_component.creation_time

When the component was created.

trial_component.display_name

The name of the component as displayed. If DisplayName isn't specified, TrialComponentName is displayed.

trial_component.end_time

When the component ended.

trial_component.input_artifacts

The input artifacts of the component.

trial_component.last_modified_by

Information about the user who created or modified an experiment, trial, trial component, lineage group, project, or model card.

Show child fields
trial_component.last_modified_by.domain_id

The domain associated with the user.

trial_component.last_modified_by.iam_identity

The IAM Identity details associated with the user. These details are associated with model package groups, model packages, and project entities only.

Show child fields
trial_component.last_modified_by.iam_identity.arn

The Amazon Resource Name (ARN) of the IAM identity.

trial_component.last_modified_by.iam_identity.principal_id

The ID of the principal that assumes the IAM identity.

trial_component.last_modified_by.iam_identity.source_identity

The person or application which assumes the IAM identity.

trial_component.last_modified_by.user_profile_arn

The Amazon Resource Name (ARN) of the user's profile.

trial_component.last_modified_by.user_profile_name

The name of the user's profile.

trial_component.last_modified_time

When the component was last modified.

trial_component.lineage_group_arn

The Amazon Resource Name (ARN) of the lineage group resource.

trial_component.metadata_properties

Metadata properties of the tracking entity, trial, or trial component.

Show child fields
trial_component.metadata_properties.commit_id

The commit ID.

trial_component.metadata_properties.generated_by

The entity this entity was generated by.

trial_component.metadata_properties.project_id

The project ID.

trial_component.metadata_properties.repository

The repository.

trial_component.metrics[]
Show child fields
trial_component.metrics[].avg

The average value of the metric.

trial_component.metrics[].count

The number of samples used to generate the metric.

trial_component.metrics[].last

The most recent value of the metric.

trial_component.metrics[].max

The maximum value of the metric.

trial_component.metrics[].metric_name

The name of the metric.

trial_component.metrics[].min

The minimum value of the metric.

trial_component.metrics[].source_arn

The Amazon Resource Name (ARN) of the source.

trial_component.metrics[].std_dev

The standard deviation of the metric.

trial_component.metrics[].time_stamp

When the metric was last updated.

trial_component.output_artifacts

The output artifacts of the component.

trial_component.parameters

The hyperparameters of the component.

trial_component.parents[]
Show child fields
trial_component.parents[].experiment_name

The name of the experiment.

trial_component.parents[].trial_name

The name of the trial.

trial_component.run_name

The name of the experiment run.

trial_component.source

The Amazon Resource Name (ARN) and job type of the source of the component.

Show child fields
trial_component.source.source_arn

The source Amazon Resource Name (ARN).

trial_component.source.source_type

The source job type.

trial_component.source_detail

Details of the source of the component.

Show child fields
trial_component.source_detail.processing_job

Information about a processing job that's the source of a trial component.

Show child fields
trial_component.source_detail.processing_job.app_specification

Configuration to run a processing job in a specified container image.

Show child fields
trial_component.source_detail.processing_job.app_specification.container_arguments[]
trial_component.source_detail.processing_job.app_specification.container_entrypoint[]
trial_component.source_detail.processing_job.app_specification.image_uri

The container image to be run by the processing job.

trial_component.source_detail.processing_job.auto_ml_job_arn

The Amazon Resource Name (ARN) of the AutoML job associated with this processing job.

trial_component.source_detail.processing_job.creation_time

The time the processing job was created.

trial_component.source_detail.processing_job.environment

Sets the environment variables in the Docker container.

trial_component.source_detail.processing_job.exit_message

A string, up to one KB in size, that contains metadata from the processing container when the processing job exits.

trial_component.source_detail.processing_job.experiment_config

Associates a SageMaker job as a trial component with an experiment and trial. Specified when you call the following APIs:

Show child fields
trial_component.source_detail.processing_job.experiment_config.experiment_name

The name of an existing experiment to associate with the trial component.

trial_component.source_detail.processing_job.experiment_config.run_name

The name of the experiment run to associate with the trial component.

trial_component.source_detail.processing_job.experiment_config.trial_component_display_name

The display name for the trial component. If this key isn't specified, the display name is the trial component name.

trial_component.source_detail.processing_job.experiment_config.trial_name

The name of an existing trial to associate the trial component with. If not specified, a new trial is created.

trial_component.source_detail.processing_job.failure_reason

A string, up to one KB in size, that contains the reason a processing job failed, if it failed.

trial_component.source_detail.processing_job.last_modified_time

The time the processing job was last modified.

trial_component.source_detail.processing_job.monitoring_schedule_arn

The ARN of a monitoring schedule for an endpoint associated with this processing job.

trial_component.source_detail.processing_job.network_config

Networking options for a job, such as network traffic encryption between containers, whether to allow inbound and outbound network calls to and from containers, and the VPC subnets and security groups to use for VPC-enabled jobs.

Show child fields
trial_component.source_detail.processing_job.network_config.enable_inter_container_traffic_encryption

Whether to encrypt all communications between distributed processing jobs. Choose True to encrypt communications. Encryption provides greater security for distributed processing jobs, but the processing might take longer.

trial_component.source_detail.processing_job.network_config.enable_network_isolation

Whether to allow inbound and outbound network calls to and from the containers used for the processing job.

trial_component.source_detail.processing_job.network_config.vpc_config

Specifies an Amazon Virtual Private Cloud (VPC) that your SageMaker jobs, hosted models, and compute resources have access to. You can control access to and from your resources by configuring a VPC. For more information, see Give SageMaker Access to Resources in your Amazon VPC.

Show child fields
trial_component.source_detail.processing_job.network_config.vpc_config.security_group_ids[]
trial_component.source_detail.processing_job.network_config.vpc_config.subnets[]
trial_component.source_detail.processing_job.processing_end_time

The time that the processing job ended.

trial_component.source_detail.processing_job.processing_inputs[]
Show child fields
trial_component.source_detail.processing_job.processing_inputs[].app_managed

When True, input operations such as data download are managed natively by the processing job application. When False (default), input operations are managed by Amazon SageMaker.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition

Configuration for a Dataset Definition input.

Show child fields
trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition

Configuration for Athena Dataset Definition input.

Show child fields
trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.catalog

The name of the data catalog used in Athena query execution.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.database

The name of the database used in the Athena query execution.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt data generated from an Athena query execution.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.output_compression

The compression used for Athena query results.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.output_format

The data storage format for Athena query results.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.output_s3_uri

The location in Amazon S3 where Athena query results are stored.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.query_string

The SQL query statements, to be executed.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.athena_dataset_definition.work_group

The name of the workgroup in which the Athena query is being started.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.data_distribution_type

Whether the generated dataset is FullyReplicated or ShardedByS3Key (default).

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.input_mode

Whether to use File or Pipe input mode. In File (default) mode, Amazon SageMaker copies the data from the input source onto the local Amazon Elastic Block Store (Amazon EBS) volumes before starting your training algorithm. This is the most commonly used input mode. In Pipe mode, Amazon SageMaker streams input data from the source directly to your algorithm without using the EBS volume.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.local_path

The local path where you want Amazon SageMaker to download the Dataset Definition inputs to run a processing job. LocalPath is an absolute path to the input data. This is a required parameter when AppManaged is False (default).

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition

Configuration for Redshift Dataset Definition input.

Show child fields
trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.cluster_id

The Redshift cluster Identifier.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.cluster_role_arn

The IAM role attached to your Redshift cluster that Amazon SageMaker uses to generate datasets.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.database

The name of the Redshift database used in Redshift query execution.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.db_user

The database user name used in Redshift query execution.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt data from a Redshift execution.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.output_compression

The compression used for Redshift query results.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.output_format

The data storage format for Redshift query results.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.output_s3_uri

The location in Amazon S3 where the Redshift query results are stored.

trial_component.source_detail.processing_job.processing_inputs[].dataset_definition.redshift_dataset_definition.query_string

The SQL query statements to be executed.

trial_component.source_detail.processing_job.processing_inputs[].input_name

The name for the processing job input.

trial_component.source_detail.processing_job.processing_inputs[].s3_input

Configuration for downloading input data from Amazon S3 into the processing container.

Show child fields
trial_component.source_detail.processing_job.processing_inputs[].s3_input.local_path

The local path in your container where you want Amazon SageMaker to write input data to. LocalPath is an absolute path to the input data and must begin with /opt/ml/processing/. LocalPath is a required parameter when AppManaged is False (default).

trial_component.source_detail.processing_job.processing_inputs[].s3_input.s3_compression_type

Whether to GZIP-decompress the data in Amazon S3 as it is streamed into the processing container. Gzip can only be used when Pipe mode is specified as the S3InputMode. In Pipe mode, Amazon SageMaker streams input data from the source directly to your container without using the EBS volume.

trial_component.source_detail.processing_job.processing_inputs[].s3_input.s3_data_distribution_type

Whether to distribute the data from Amazon S3 to all processing instances with FullyReplicated, or whether the data from Amazon S3 is shared by Amazon S3 key, downloading one shard of data to each processing instance.

trial_component.source_detail.processing_job.processing_inputs[].s3_input.s3_data_type

Whether you use an S3Prefix or a ManifestFile for the data type. If you choose S3Prefix, S3Uri identifies a key name prefix. Amazon SageMaker uses all objects with the specified key name prefix for the processing job. If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want Amazon SageMaker to use for the processing job.

trial_component.source_detail.processing_job.processing_inputs[].s3_input.s3_input_mode

Whether to use File or Pipe input mode. In File mode, Amazon SageMaker copies the data from the input source onto the local ML storage volume before starting your processing container. This is the most commonly used input mode. In Pipe mode, Amazon SageMaker streams input data from the source directly to your processing container into named pipes without using the ML storage volume.

trial_component.source_detail.processing_job.processing_inputs[].s3_input.s3_uri

The URI of the Amazon S3 prefix Amazon SageMaker downloads data required to run a processing job.

trial_component.source_detail.processing_job.processing_job_arn

The ARN of the processing job.

trial_component.source_detail.processing_job.processing_job_name

The name of the processing job.

trial_component.source_detail.processing_job.processing_job_status

The status of the processing job.

trial_component.source_detail.processing_job.processing_output_config

Configuration for uploading output from the processing container.

Show child fields
trial_component.source_detail.processing_job.processing_output_config.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt the processing job output. KmsKeyId can be an ID of a KMS key, ARN of a KMS key, alias of a KMS key, or alias of a KMS key. The KmsKeyId is applied to all outputs.

trial_component.source_detail.processing_job.processing_output_config.outputs[]
Show child fields
trial_component.source_detail.processing_job.processing_output_config.outputs[].app_managed

When True, output operations such as data upload are managed natively by the processing job application. When False (default), output operations are managed by Amazon SageMaker.

trial_component.source_detail.processing_job.processing_output_config.outputs[].feature_store_output

Configuration for processing job outputs in Amazon SageMaker Feature Store. This processing output type is only supported when AppManaged is specified.

Show child fields
trial_component.source_detail.processing_job.processing_output_config.outputs[].feature_store_output.feature_group_name

The name of the Amazon SageMaker FeatureGroup to use as the destination for processing job output. Note that your processing script is responsible for putting records into your Feature Store.

trial_component.source_detail.processing_job.processing_output_config.outputs[].output_name

The name for the processing job output.

trial_component.source_detail.processing_job.processing_output_config.outputs[].s3_output

Configuration for processing job outputs in Amazon S3.

Show child fields
trial_component.source_detail.processing_job.processing_output_config.outputs[].s3_output.local_path

The local path of a directory where you want Amazon SageMaker to upload its contents to Amazon S3. LocalPath is an absolute path to a directory containing output files. This directory will be created by the platform and exist when your container's entrypoint is invoked.

trial_component.source_detail.processing_job.processing_output_config.outputs[].s3_output.s3_upload_mode

Whether to upload the results of the processing job continuously or after the job completes.

trial_component.source_detail.processing_job.processing_output_config.outputs[].s3_output.s3_uri

A URI that identifies the Amazon S3 bucket where you want Amazon SageMaker to save the results of a processing job.

trial_component.source_detail.processing_job.processing_resources

Identifies the resources, ML compute instances, and ML storage volumes to deploy for a processing job. In distributed training, you specify more than one instance.

Show child fields
trial_component.source_detail.processing_job.processing_resources.cluster_config

The configuration for the resources in a cluster used to run the processing job.

Show child fields
trial_component.source_detail.processing_job.processing_resources.cluster_config.instance_count

The number of ML compute instances to use in the processing job. For distributed processing jobs, specify a value greater than 1. The default value is 1.

trial_component.source_detail.processing_job.processing_resources.cluster_config.instance_type

The ML compute instance type for the processing job.

trial_component.source_detail.processing_job.processing_resources.cluster_config.volume_kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) that run the processing job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

trial_component.source_detail.processing_job.processing_resources.cluster_config.volume_size_in_gb

The size of the ML storage volume in gigabytes that you want to provision. You must specify sufficient ML storage for your scenario.

Certain Nitro-based instances include local storage with a fixed total size, dependent on the instance type. When using these instances for processing, Amazon SageMaker mounts the local instance storage instead of Amazon EBS gp2 storage. You can't request a VolumeSizeInGB greater than the total size of the local instance storage.

For a list of instance types that support local instance storage, including the total size per instance type, see Instance Store Volumes.

trial_component.source_detail.processing_job.processing_start_time

The time that the processing job started.

trial_component.source_detail.processing_job.role_arn

The ARN of the role used to create the processing job.

trial_component.source_detail.processing_job.stopping_condition

Configures conditions under which the processing job should be stopped, such as how long the processing job has been running. After the condition is met, the processing job is stopped.

Show child fields
trial_component.source_detail.processing_job.stopping_condition.max_runtime_in_seconds

Specifies the maximum runtime in seconds.

trial_component.source_detail.processing_job.tags[]
Show child fields
trial_component.source_detail.processing_job.tags[].key

The tag key. Tag keys must be unique per resource.

trial_component.source_detail.processing_job.tags[].value

The tag value.

trial_component.source_detail.processing_job.training_job_arn

The ARN of the training job associated with this processing job.

trial_component.source_detail.source_arn

The Amazon Resource Name (ARN) of the source.

trial_component.source_detail.training_job

Information about a training job that's the source of a trial component.

Show child fields
trial_component.source_detail.training_job.algorithm_specification

Information about the algorithm used for training, and algorithm metadata.

Show child fields
trial_component.source_detail.training_job.algorithm_specification.algorithm_name

The name of the algorithm resource to use for the training job. This must be an algorithm resource that you created or subscribe to on Amazon Web Services Marketplace.

You must specify either the algorithm name to the AlgorithmName parameter or the image URI of the algorithm container to the TrainingImage parameter.

Note that the AlgorithmName parameter is mutually exclusive with the TrainingImage parameter. If you specify a value for the AlgorithmName parameter, you can't specify a value for TrainingImage, and vice versa.

If you specify values for both parameters, the training job might break; if you don't specify any value for both parameters, the training job might raise a null error.

trial_component.source_detail.training_job.algorithm_specification.container_arguments[]
trial_component.source_detail.training_job.algorithm_specification.container_entrypoint[]
trial_component.source_detail.training_job.algorithm_specification.enable_sage_maker_metrics_time_series

To generate and save time-series metrics during training, set to true. The default is false and time-series metrics aren't generated except in the following cases:

trial_component.source_detail.training_job.algorithm_specification.metric_definitions[]
Show child fields
trial_component.source_detail.training_job.algorithm_specification.metric_definitions[].name

The name of the metric.

trial_component.source_detail.training_job.algorithm_specification.metric_definitions[].regex

A regular expression that searches the output of a training job and gets the value of the metric. For more information about using regular expressions to define metrics, see Defining metrics and environment variables.

trial_component.source_detail.training_job.algorithm_specification.training_image

The registry path of the Docker image that contains the training algorithm. For information about docker registry paths for SageMaker built-in algorithms, see Docker Registry Paths and Example Code in the Amazon SageMaker developer guide. SageMaker supports both registry/repository[:tag] and registry/repository[@digest] image path formats. For more information about using your custom training container, see Using Your Own Algorithms with Amazon SageMaker.

You must specify either the algorithm name to the AlgorithmName parameter or the image URI of the algorithm container to the TrainingImage parameter.

For more information, see the note in the AlgorithmName parameter description.

trial_component.source_detail.training_job.algorithm_specification.training_image_config

The configuration to use an image from a private Docker registry for a training job.

Show child fields
trial_component.source_detail.training_job.algorithm_specification.training_image_config.training_repository_access_mode

The method that your training job will use to gain access to the images in your private Docker registry. For access to an image in a private Docker registry, set to Vpc.

trial_component.source_detail.training_job.algorithm_specification.training_image_config.training_repository_auth_config

An object containing authentication information for a private Docker registry containing your training images.

Show child fields
trial_component.source_detail.training_job.algorithm_specification.training_image_config.training_repository_auth_config.training_repository_credentials_provider_arn

The Amazon Resource Name (ARN) of an Amazon Web Services Lambda function used to give SageMaker access credentials to your private Docker registry.

trial_component.source_detail.training_job.algorithm_specification.training_input_mode

The training input mode that the algorithm supports. For more information about input modes, see Algorithms.

Pipe mode

If an algorithm supports Pipe mode, Amazon SageMaker streams data directly from Amazon S3 to the container.

File mode

If an algorithm supports File mode, SageMaker downloads the training data from S3 to the provisioned ML storage volume, and mounts the directory to the Docker volume for the training container.

You must provision the ML storage volume with sufficient capacity to accommodate the data downloaded from S3. In addition to the training data, the ML storage volume also stores the output model. The algorithm container uses the ML storage volume to also store intermediate information, if any.

For distributed algorithms, training data is distributed uniformly. Your training duration is predictable if the input data objects sizes are approximately the same. SageMaker does not split the files any further for model training. If the object sizes are skewed, training won't be optimal as the data distribution is also skewed when one host in a training cluster is overloaded, thus becoming a bottleneck in training.

FastFile mode

If an algorithm supports FastFile mode, SageMaker streams data directly from S3 to the container with no code changes, and provides file system access to the data. Users can author their training script to interact with these files as if they were stored on disk.

FastFile mode works best when the data is read sequentially. Augmented manifest files aren't supported. The startup time is lower when there are fewer files in the S3 bucket provided.

trial_component.source_detail.training_job.auto_ml_job_arn

The Amazon Resource Name (ARN) of the job.

trial_component.source_detail.training_job.billable_time_in_seconds

The billable time in seconds.

trial_component.source_detail.training_job.checkpoint_config

Contains information about the output location for managed spot training checkpoint data.

Show child fields
trial_component.source_detail.training_job.checkpoint_config.local_path

(Optional) The local directory where checkpoints are written. The default directory is /opt/ml/checkpoints/.

trial_component.source_detail.training_job.checkpoint_config.s3_uri

Identifies the S3 path where you want SageMaker to store checkpoints. For example, s3://bucket-name/key-name-prefix.

trial_component.source_detail.training_job.creation_time

A timestamp that indicates when the training job was created.

trial_component.source_detail.training_job.debug_hook_config

Configuration information for the Amazon SageMaker Debugger hook parameters, metric and tensor collections, and storage paths. To learn more about how to configure the DebugHookConfig parameter, see Use the SageMaker and Debugger Configuration API Operations to Create, Update, and Debug Your Training Job.

Show child fields
trial_component.source_detail.training_job.debug_hook_config.collection_configurations[]
Show child fields
trial_component.source_detail.training_job.debug_hook_config.collection_configurations[].collection_name

The name of the tensor collection. The name must be unique relative to other rule configuration names.

trial_component.source_detail.training_job.debug_hook_config.collection_configurations[].collection_parameters

Parameter values for the tensor collection. The allowed parameters are "name", "include_regex", "reduction_config", "save_config", "tensor_names", and "save_histogram".

trial_component.source_detail.training_job.debug_hook_config.hook_parameters

Configuration information for the Amazon SageMaker Debugger hook parameters.

trial_component.source_detail.training_job.debug_hook_config.local_path

Path to local storage location for metrics and tensors. Defaults to /opt/ml/output/tensors/.

trial_component.source_detail.training_job.debug_hook_config.s3_output_path

Path to Amazon S3 storage location for metrics and tensors.

trial_component.source_detail.training_job.debug_rule_configurations[]
Show child fields
trial_component.source_detail.training_job.debug_rule_configurations[].instance_type

The instance type to deploy a custom rule for debugging a training job.

trial_component.source_detail.training_job.debug_rule_configurations[].local_path

Path to local storage location for output of rules. Defaults to /opt/ml/processing/output/rule/.

trial_component.source_detail.training_job.debug_rule_configurations[].rule_configuration_name

The name of the rule configuration. It must be unique relative to other rule configuration names.

trial_component.source_detail.training_job.debug_rule_configurations[].rule_evaluator_image

The Amazon Elastic Container (ECR) Image for the managed rule evaluation.

trial_component.source_detail.training_job.debug_rule_configurations[].rule_parameters

Runtime configuration for rule container.

trial_component.source_detail.training_job.debug_rule_configurations[].s3_output_path

Path to Amazon S3 storage location for rules.

trial_component.source_detail.training_job.debug_rule_configurations[].volume_size_in_gb

The size, in GB, of the ML storage volume attached to the processing instance.

trial_component.source_detail.training_job.debug_rule_evaluation_statuses[]
Show child fields
trial_component.source_detail.training_job.debug_rule_evaluation_statuses[].last_modified_time

Timestamp when the rule evaluation status was last modified.

trial_component.source_detail.training_job.debug_rule_evaluation_statuses[].rule_configuration_name

The name of the rule configuration.

trial_component.source_detail.training_job.debug_rule_evaluation_statuses[].rule_evaluation_job_arn

The Amazon Resource Name (ARN) of the rule evaluation job.

trial_component.source_detail.training_job.debug_rule_evaluation_statuses[].rule_evaluation_status

Status of the rule evaluation.

trial_component.source_detail.training_job.debug_rule_evaluation_statuses[].status_details

Details from the rule evaluation.

trial_component.source_detail.training_job.enable_inter_container_traffic_encryption

To encrypt all communications between ML compute instances in distributed training, choose True. Encryption provides greater security for distributed training, but training might take longer. How long it takes depends on the amount of communication between compute instances, especially if you use a deep learning algorithm in distributed training.

trial_component.source_detail.training_job.enable_managed_spot_training

When true, enables managed spot training using Amazon EC2 Spot instances to run training jobs instead of on-demand instances. For more information, see Managed Spot Training.

trial_component.source_detail.training_job.enable_network_isolation

If the TrainingJob was created with network isolation, the value is set to true. If network isolation is enabled, nodes can't communicate beyond the VPC they run in.

trial_component.source_detail.training_job.environment

The environment variables to set in the Docker container.

trial_component.source_detail.training_job.experiment_config

Associates a SageMaker job as a trial component with an experiment and trial. Specified when you call the following APIs:

Show child fields
trial_component.source_detail.training_job.experiment_config.experiment_name

The name of an existing experiment to associate with the trial component.

trial_component.source_detail.training_job.experiment_config.run_name

The name of the experiment run to associate with the trial component.

trial_component.source_detail.training_job.experiment_config.trial_component_display_name

The display name for the trial component. If this key isn't specified, the display name is the trial component name.

trial_component.source_detail.training_job.experiment_config.trial_name

The name of an existing trial to associate the trial component with. If not specified, a new trial is created.

trial_component.source_detail.training_job.failure_reason

If the training job failed, the reason it failed.

trial_component.source_detail.training_job.final_metric_data_list[]
Show child fields
trial_component.source_detail.training_job.final_metric_data_list[].metric_name

The name of the metric.

trial_component.source_detail.training_job.final_metric_data_list[].timestamp

The date and time that the algorithm emitted the metric.

trial_component.source_detail.training_job.final_metric_data_list[].value

The value of the metric.

trial_component.source_detail.training_job.hyper_parameters

Algorithm-specific parameters.

trial_component.source_detail.training_job.input_data_config[]
Show child fields
trial_component.source_detail.training_job.input_data_config[].channel_name

The name of the channel.

trial_component.source_detail.training_job.input_data_config[].compression_type

If training data is compressed, the compression type. The default value is None. CompressionType is used only in Pipe input mode. In File mode, leave this field unset or set it to None.

trial_component.source_detail.training_job.input_data_config[].content_type

The MIME type of the data.

trial_component.source_detail.training_job.input_data_config[].data_source

The location of the channel data.

Show child fields
trial_component.source_detail.training_job.input_data_config[].data_source.file_system_data_source

The file system that is associated with a channel.

Show child fields
trial_component.source_detail.training_job.input_data_config[].data_source.file_system_data_source.directory_path

The full path to the directory to associate with the channel.

trial_component.source_detail.training_job.input_data_config[].data_source.file_system_data_source.file_system_access_mode

The access mode of the mount of the directory associated with the channel. A directory can be mounted either in ro (read-only) or rw (read-write) mode.

trial_component.source_detail.training_job.input_data_config[].data_source.file_system_data_source.file_system_id

The file system id.

trial_component.source_detail.training_job.input_data_config[].data_source.file_system_data_source.file_system_type

The file system type.

trial_component.source_detail.training_job.input_data_config[].data_source.s3_data_source

The S3 location of the data source that is associated with a channel.

Show child fields
trial_component.source_detail.training_job.input_data_config[].data_source.s3_data_source.attribute_names[]
trial_component.source_detail.training_job.input_data_config[].data_source.s3_data_source.instance_group_names[]
trial_component.source_detail.training_job.input_data_config[].data_source.s3_data_source.s3_data_distribution_type

If you want SageMaker to replicate the entire dataset on each ML compute instance that is launched for model training, specify FullyReplicated.

If you want SageMaker to replicate a subset of data on each ML compute instance that is launched for model training, specify ShardedByS3Key. If there are n ML compute instances launched for a training job, each instance gets approximately 1/n of the number of S3 objects. In this case, model training on each machine uses only the subset of training data.

Don't choose more ML compute instances for training than available S3 objects. If you do, some nodes won't get any data and you will pay for nodes that aren't getting any training data. This applies in both File and Pipe modes. Keep this in mind when developing algorithms.

In distributed training, where you use multiple ML compute EC2 instances, you might choose ShardedByS3Key. If the algorithm requires copying training data to the ML storage volume (when TrainingInputMode is set to File), this copies 1/n of the number of objects.

trial_component.source_detail.training_job.input_data_config[].data_source.s3_data_source.s3_data_type

If you choose S3Prefix, S3Uri identifies a key name prefix. SageMaker uses all objects that match the specified key name prefix for model training.

If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want SageMaker to use for model training.

If you choose AugmentedManifestFile, S3Uri identifies an object that is an augmented manifest file in JSON lines format. This file contains the data you want to use for model training. AugmentedManifestFile can only be used if the Channel's input mode is Pipe.

trial_component.source_detail.training_job.input_data_config[].data_source.s3_data_source.s3_uri

Depending on the value specified for the S3DataType, identifies either a key name prefix or a manifest. For example:

  • A key name prefix might look like this: s3://bucketname/exampleprefix/

  • A manifest might look like this: s3://bucketname/example.manifest

    A manifest is an S3 object which is a JSON file consisting of an array of elements. The first element is a prefix which is followed by one or more suffixes. SageMaker appends the suffix elements to the prefix to get a full set of S3Uri. Note that the prefix must be a valid non-empty S3Uri that precludes users from specifying a manifest whose individual S3Uri is sourced from different S3 buckets.

    The following code example shows a valid manifest format:

    [ {"prefix": "s3://customer_bucket/some/prefix/"},

    "relative/path/to/custdata-1",

    "relative/path/custdata-2",

    ...

    "relative/path/custdata-N"

    ]

    This JSON is equivalent to the following S3Uri list:

    s3://customer_bucket/some/prefix/relative/path/to/custdata-1

    s3://customer_bucket/some/prefix/relative/path/custdata-2

    ...

    s3://customer_bucket/some/prefix/relative/path/custdata-N

    The complete set of S3Uri in this manifest is the input data for the channel for this data source. The object that each S3Uri points to must be readable by the IAM role that SageMaker uses to perform tasks on your behalf.

Your input bucket must be located in same Amazon Web Services region as your training job.

trial_component.source_detail.training_job.input_data_config[].input_mode

(Optional) The input mode to use for the data channel in a training job. If you don't set a value for InputMode, SageMaker uses the value set for TrainingInputMode. Use this parameter to override the TrainingInputMode setting in a AlgorithmSpecification request when you have a channel that needs a different input mode from the training job's general setting. To download the data from Amazon Simple Storage Service (Amazon S3) to the provisioned ML storage volume, and mount the directory to a Docker volume, use File input mode. To stream data directly from Amazon S3 to the container, choose Pipe input mode.

To use a model for incremental training, choose File input model.

trial_component.source_detail.training_job.input_data_config[].record_wrapper_type

Specify RecordIO as the value when input data is in raw format but the training algorithm requires the RecordIO format. In this case, SageMaker wraps each individual S3 object in a RecordIO record. If the input data is already in RecordIO format, you don't need to set this attribute. For more information, see Create a Dataset Using RecordIO.

In File mode, leave this field unset or set it to None.

trial_component.source_detail.training_job.input_data_config[].shuffle_config

A configuration for a shuffle option for input data in a channel. If you use S3Prefix for S3DataType, this shuffles the results of the S3 key prefix matches. If you use ManifestFile, the order of the S3 object references in the ManifestFile is shuffled. If you use AugmentedManifestFile, the order of the JSON lines in the AugmentedManifestFile is shuffled. The shuffling order is determined using the Seed value.

For Pipe input mode, shuffling is done at the start of every epoch. With large datasets this ensures that the order of the training data is different for each epoch, it helps reduce bias and possible overfitting. In a multi-node training job when ShuffleConfig is combined with S3DataDistributionType of ShardedByS3Key, the data is shuffled across nodes so that the content sent to a particular node on the first epoch might be sent to a different node on the second epoch.

Show child fields
trial_component.source_detail.training_job.input_data_config[].shuffle_config.seed

Determines the shuffling order in ShuffleConfig value.

trial_component.source_detail.training_job.labeling_job_arn

The Amazon Resource Name (ARN) of the labeling job.

trial_component.source_detail.training_job.last_modified_time

A timestamp that indicates when the status of the training job was last modified.

trial_component.source_detail.training_job.model_artifacts

Information about the Amazon S3 location that is configured for storing model artifacts.

Show child fields
trial_component.source_detail.training_job.model_artifacts.s3_model_artifacts

The path of the S3 object that contains the model artifacts. For example, s3://bucket-name/keynameprefix/model.tar.gz.

trial_component.source_detail.training_job.output_data_config

The S3 path where model artifacts that you configured when creating the job are stored. SageMaker creates subfolders for model artifacts.

Show child fields
trial_component.source_detail.training_job.output_data_config.compression_type

The model output compression type. Select None to output an uncompressed model, recommended for large model outputs. Defaults to gzip.

trial_component.source_detail.training_job.output_data_config.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption. The KmsKeyId can be any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

  • // KMS Key Alias

    "alias/ExampleAlias"

  • // Amazon Resource Name (ARN) of a KMS Key Alias

    "arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias"

If you use a KMS key ID or an alias of your KMS key, the SageMaker execution role must include permissions to call kms:Encrypt. If you don't provide a KMS key ID, SageMaker uses the default KMS key for Amazon S3 for your role's account. For more information, see KMS-Managed Encryption Keys in the Amazon Simple Storage Service Developer Guide. If the output data is stored in Amazon S3 Express One Zone, it is encrypted with server-side encryption with Amazon S3 managed keys (SSE-S3). KMS key is not supported for Amazon S3 Express One Zone

The KMS key policy must grant permission to the IAM role that you specify in your CreateTrainingJob, CreateTransformJob, or CreateHyperParameterTuningJob requests. For more information, see Using Key Policies in Amazon Web Services KMS in the Amazon Web Services Key Management Service Developer Guide.

trial_component.source_detail.training_job.output_data_config.s3_output_path

Identifies the S3 path where you want SageMaker to store the model artifacts. For example, s3://bucket-name/key-name-prefix.

trial_component.source_detail.training_job.profiler_config

Configuration information for Amazon SageMaker Debugger system monitoring, framework profiling, and storage paths.

Show child fields
trial_component.source_detail.training_job.profiler_config.disable_profiler

Configuration to turn off Amazon SageMaker Debugger's system monitoring and profiling functionality. To turn it off, set to True.

trial_component.source_detail.training_job.profiler_config.profiling_interval_in_milliseconds

A time interval for capturing system metrics in milliseconds. Available values are 100, 200, 500, 1000 (1 second), 5000 (5 seconds), and 60000 (1 minute) milliseconds. The default value is 500 milliseconds.

trial_component.source_detail.training_job.profiler_config.profiling_parameters

Configuration information for capturing framework metrics. Available key strings for different profiling options are DetailedProfilingConfig, PythonProfilingConfig, and DataLoaderProfilingConfig. The following codes are configuration structures for the ProfilingParameters parameter. To learn more about how to configure the ProfilingParameters parameter, see Use the SageMaker and Debugger Configuration API Operations to Create, Update, and Debug Your Training Job.

trial_component.source_detail.training_job.profiler_config.s3_output_path

Path to Amazon S3 storage location for system and framework metrics.

trial_component.source_detail.training_job.resource_config

Resources, including ML compute instances and ML storage volumes, that are configured for model training.

Show child fields
trial_component.source_detail.training_job.resource_config.instance_count

The number of ML compute instances to use. For distributed training, provide a value greater than 1.

trial_component.source_detail.training_job.resource_config.instance_groups[]
Show child fields
trial_component.source_detail.training_job.resource_config.instance_groups[].instance_count

Specifies the number of instances of the instance group.

trial_component.source_detail.training_job.resource_config.instance_groups[].instance_group_name

Specifies the name of the instance group.

trial_component.source_detail.training_job.resource_config.instance_groups[].instance_type

Specifies the instance type of the instance group.

trial_component.source_detail.training_job.resource_config.instance_type

The ML compute instance type.

SageMaker Training on Amazon Elastic Compute Cloud (EC2) P4de instances is in preview release starting December 9th, 2022.

Amazon EC2 P4de instances (currently in preview) are powered by 8 NVIDIA A100 GPUs with 80GB high-performance HBM2e GPU memory, which accelerate the speed of training ML models that need to be trained on large datasets of high-resolution data. In this preview release, Amazon SageMaker supports ML training jobs on P4de instances (ml.p4de.24xlarge) to reduce model training time. The ml.p4de.24xlarge instances are available in the following Amazon Web Services Regions.

  • US East (N. Virginia) (us-east-1)

  • US West (Oregon) (us-west-2)

To request quota limit increase and start using P4de instances, contact the SageMaker Training service team through your account team.

trial_component.source_detail.training_job.resource_config.keep_alive_period_in_seconds

The duration of time in seconds to retain configured resources in a warm pool for subsequent training jobs.

trial_component.source_detail.training_job.resource_config.volume_kms_key_id

The Amazon Web Services KMS key that SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) that run the training job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

The VolumeKmsKeyId can be in any of the following formats:

  • // KMS Key ID

    "1234abcd-12ab-34cd-56ef-1234567890ab"

  • // Amazon Resource Name (ARN) of a KMS Key

    "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"

trial_component.source_detail.training_job.resource_config.volume_size_in_gb

The size of the ML storage volume that you want to provision.

ML storage volumes store model artifacts and incremental states. Training algorithms might also use the ML storage volume for scratch space. If you want to store the training data in the ML storage volume, choose File as the TrainingInputMode in the algorithm specification.

When using an ML instance with NVMe SSD volumes, SageMaker doesn't provision Amazon EBS General Purpose SSD (gp2) storage. Available storage is fixed to the NVMe-type instance's storage capacity. SageMaker configures storage paths for training datasets, checkpoints, model artifacts, and outputs to use the entire capacity of the instance storage. For example, ML instance families with the NVMe-type instance storage include ml.p4d, ml.g4dn, and ml.g5.

When using an ML instance with the EBS-only storage option and without instance storage, you must define the size of EBS volume through VolumeSizeInGB in the ResourceConfig API. For example, ML instance families that use EBS volumes include ml.c5 and ml.p2.

To look up instance types and their instance storage types and volumes, see Amazon EC2 Instance Types.

To find the default local paths defined by the SageMaker training platform, see Amazon SageMaker Training Storage Folders for Training Datasets, Checkpoints, Model Artifacts, and Outputs.

trial_component.source_detail.training_job.retry_strategy

The number of times to retry the job when the job fails due to an InternalServerError.

Show child fields
trial_component.source_detail.training_job.retry_strategy.maximum_retry_attempts

The number of times to retry the job. When the job is retried, it's SecondaryStatus is changed to STARTING.

trial_component.source_detail.training_job.role_arn

The Amazon Web Services Identity and Access Management (IAM) role configured for the training job.

trial_component.source_detail.training_job.secondary_status

Provides detailed information about the state of the training job. For detailed information about the secondary status of the training job, see StatusMessage under SecondaryStatusTransition.

SageMaker provides primary statuses and secondary statuses that apply to each of them:

InProgress

  • Starting - Starting the training job.

  • Downloading - An optional stage for algorithms that support File training input mode. It indicates that data is being downloaded to the ML storage volumes.

  • Training - Training is in progress.

  • Uploading - Training is complete and the model artifacts are being uploaded to the S3 location.

Completed

  • Completed - The training job has completed.

Failed

  • Failed - The training job has failed. The reason for the failure is returned in the FailureReason field of DescribeTrainingJobResponse.

Stopped

  • MaxRuntimeExceeded - The job stopped because it exceeded the maximum allowed runtime.

  • Stopped - The training job has stopped.

Stopping

  • Stopping - Stopping the training job.

Valid values for SecondaryStatus are subject to change.

We no longer support the following secondary statuses:

  • LaunchingMLInstances

  • PreparingTrainingStack

  • DownloadingTrainingImage

trial_component.source_detail.training_job.secondary_status_transitions[]
Show child fields
trial_component.source_detail.training_job.secondary_status_transitions[].end_time

A timestamp that shows when the training job transitioned out of this secondary status state into another secondary status state or when the training job has ended.

trial_component.source_detail.training_job.secondary_status_transitions[].start_time

A timestamp that shows when the training job transitioned to the current secondary status state.

trial_component.source_detail.training_job.secondary_status_transitions[].status

Contains a secondary status information from a training job.

Status might be one of the following secondary statuses:

InProgress

  • Starting - Starting the training job.

  • Downloading - An optional stage for algorithms that support File training input mode. It indicates that data is being downloaded to the ML storage volumes.

  • Training - Training is in progress.

  • Uploading - Training is complete and the model artifacts are being uploaded to the S3 location.

Completed

  • Completed - The training job has completed.

Failed

  • Failed - The training job has failed. The reason for the failure is returned in the FailureReason field of DescribeTrainingJobResponse.

Stopped

  • MaxRuntimeExceeded - The job stopped because it exceeded the maximum allowed runtime.

  • Stopped - The training job has stopped.

Stopping

  • Stopping - Stopping the training job.

We no longer support the following secondary statuses:

  • LaunchingMLInstances

  • PreparingTrainingStack

  • DownloadingTrainingImage

trial_component.source_detail.training_job.secondary_status_transitions[].status_message

A detailed description of the progress within a secondary status.

SageMaker provides secondary statuses and status messages that apply to each of them:

Starting

  • Starting the training job.

  • Launching requested ML instances.

  • Insufficient capacity error from EC2 while launching instances, retrying!

  • Launched instance was unhealthy, replacing it!

  • Preparing the instances for training.

Training

  • Training image download completed. Training in progress.

Status messages are subject to change. Therefore, we recommend not including them in code that programmatically initiates actions. For examples, don't use status messages in if statements.

To have an overview of your training job's progress, view TrainingJobStatus and SecondaryStatus in DescribeTrainingJob, and StatusMessage together. For example, at the start of a training job, you might see the following:

  • TrainingJobStatus - InProgress

  • SecondaryStatus - Training

  • StatusMessage - Downloading the training image

trial_component.source_detail.training_job.stopping_condition

Specifies a limit to how long a model training job can run. It also specifies how long a managed Spot training job has to complete. When the job reaches the time limit, SageMaker ends the training job. Use this API to cap model training costs.

To stop a job, SageMaker sends the algorithm the SIGTERM signal, which delays job termination for 120 seconds. Algorithms can use this 120-second window to save the model artifacts, so the results of training are not lost.

Show child fields
trial_component.source_detail.training_job.stopping_condition.max_pending_time_in_seconds

The maximum length of time, in seconds, that a training or compilation job can be pending before it is stopped.

trial_component.source_detail.training_job.stopping_condition.max_runtime_in_seconds

The maximum length of time, in seconds, that a training or compilation job can run before it is stopped.

For compilation jobs, if the job does not complete during this time, a TimeOut error is generated. We recommend starting with 900 seconds and increasing as necessary based on your model.

For all other jobs, if the job does not complete during this time, SageMaker ends the job. When RetryStrategy is specified in the job request, MaxRuntimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt. The default value is 1 day. The maximum value is 28 days.

The maximum time that a TrainingJob can run in total, including any time spent publishing metrics or archiving and uploading models after it has been stopped, is 30 days.

trial_component.source_detail.training_job.stopping_condition.max_wait_time_in_seconds

The maximum length of time, in seconds, that a managed Spot training job has to complete. It is the amount of time spent waiting for Spot capacity plus the amount of time the job can run. It must be equal to or greater than MaxRuntimeInSeconds. If the job does not complete during this time, SageMaker ends the job.

When RetryStrategy is specified in the job request, MaxWaitTimeInSeconds specifies the maximum time for all of the attempts in total, not each individual attempt.

trial_component.source_detail.training_job.tags[]
Show child fields
trial_component.source_detail.training_job.tags[].key

The tag key. Tag keys must be unique per resource.

trial_component.source_detail.training_job.tags[].value

The tag value.

trial_component.source_detail.training_job.tensor_board_output_config

Configuration of storage locations for the Amazon SageMaker Debugger TensorBoard output data.

Show child fields
trial_component.source_detail.training_job.tensor_board_output_config.local_path

Path to local storage location for tensorBoard output. Defaults to /opt/ml/output/tensorboard.

trial_component.source_detail.training_job.tensor_board_output_config.s3_output_path

Path to Amazon S3 storage location for TensorBoard output.

trial_component.source_detail.training_job.training_end_time

Indicates the time when the training job ends on training instances. You are billed for the time interval between the value of TrainingStartTime and this time. For successful jobs and stopped jobs, this is the time after model artifacts are uploaded. For failed jobs, this is the time when SageMaker detects a job failure.

trial_component.source_detail.training_job.training_job_arn

The Amazon Resource Name (ARN) of the training job.

trial_component.source_detail.training_job.training_job_name

The name of the training job.

trial_component.source_detail.training_job.training_job_status

The status of the training job.

Training job statuses are:

  • InProgress - The training is in progress.

  • Completed - The training job has completed.

  • Failed - The training job has failed. To see the reason for the failure, see the FailureReason field in the response to a DescribeTrainingJobResponse call.

  • Stopping - The training job is stopping.

  • Stopped - The training job has stopped.

For more detailed information, see SecondaryStatus.

trial_component.source_detail.training_job.training_start_time

Indicates the time when the training job starts on training instances. You are billed for the time interval between this time and the value of TrainingEndTime. The start time in CloudWatch Logs might be later than this time. The difference is due to the time it takes to download the training data and to the size of the training container.

trial_component.source_detail.training_job.training_time_in_seconds

The training time in seconds.

trial_component.source_detail.training_job.tuning_job_arn

The Amazon Resource Name (ARN) of the associated hyperparameter tuning job if the training job was launched by a hyperparameter tuning job.

trial_component.source_detail.training_job.vpc_config

A VpcConfig object that specifies the VPC that this training job has access to. For more information, see Protect Training Jobs by Using an Amazon Virtual Private Cloud.

Show child fields
trial_component.source_detail.training_job.vpc_config.security_group_ids[]
trial_component.source_detail.training_job.vpc_config.subnets[]
trial_component.source_detail.transform_job

Information about a transform job that's the source of a trial component.

Show child fields
trial_component.source_detail.transform_job.auto_ml_job_arn

The Amazon Resource Name (ARN) of the AutoML job that created the transform job.

trial_component.source_detail.transform_job.batch_strategy

Specifies the number of records to include in a mini-batch for an HTTP inference request. A record is a single unit of input data that inference can be made on. For example, a single line in a CSV file is a record.

trial_component.source_detail.transform_job.creation_time

A timestamp that shows when the transform Job was created.

trial_component.source_detail.transform_job.data_capture_config

Configuration to control how SageMaker captures inference data for batch transform jobs.

Show child fields
trial_component.source_detail.transform_job.data_capture_config.destination_s3_uri

The Amazon S3 location being used to capture the data.

trial_component.source_detail.transform_job.data_capture_config.generate_inference_id

Flag that indicates whether to append inference id to the output.

trial_component.source_detail.transform_job.data_capture_config.kms_key_id

The Amazon Resource Name (ARN) of a Amazon Web Services Key Management Service key that SageMaker uses to encrypt data on the storage volume attached to the ML compute instance that hosts the batch transform job.

The KmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

trial_component.source_detail.transform_job.data_processing

The data structure used to specify the data to be used for inference in a batch transform job and to associate the data that is relevant to the prediction results in the output. The input filter provided allows you to exclude input data that is not needed for inference in a batch transform job. The output filter provided allows you to include input data relevant to interpreting the predictions in the output from the job. For more information, see Associate Prediction Results with their Corresponding Input Records.

Show child fields
trial_component.source_detail.transform_job.data_processing.input_filter

A JSONPath expression used to select a portion of the input data to pass to the algorithm. Use the InputFilter parameter to exclude fields, such as an ID column, from the input. If you want SageMaker to pass the entire input dataset to the algorithm, accept the default value $.

Examples: "$", "$[1:]", "$.features"

trial_component.source_detail.transform_job.data_processing.join_source

Specifies the source of the data to join with the transformed data. The valid values are None and Input. The default value is None, which specifies not to join the input with the transformed data. If you want the batch transform job to join the original input data with the transformed data, set JoinSource to Input. You can specify OutputFilter as an additional filter to select a portion of the joined dataset and store it in the output file.

For JSON or JSONLines objects, such as a JSON array, SageMaker adds the transformed data to the input JSON object in an attribute called SageMakerOutput. The joined result for JSON must be a key-value pair object. If the input is not a key-value pair object, SageMaker creates a new JSON file. In the new JSON file, and the input data is stored under the SageMakerInput key and the results are stored in SageMakerOutput.

For CSV data, SageMaker takes each row as a JSON array and joins the transformed data with the input by appending each transformed row to the end of the input. The joined data has the original input data followed by the transformed data and the output is a CSV file.

For information on how joining in applied, see Workflow for Associating Inferences with Input Records.

trial_component.source_detail.transform_job.data_processing.output_filter

A JSONPath expression used to select a portion of the joined dataset to save in the output file for a batch transform job. If you want SageMaker to store the entire input dataset in the output file, leave the default value, $. If you specify indexes that aren't within the dimension size of the joined dataset, you get an error.

Examples: "$", "$[0,5:]", "$['id','SageMakerOutput']"

trial_component.source_detail.transform_job.environment

The environment variables to set in the Docker container. We support up to 16 key and values entries in the map.

trial_component.source_detail.transform_job.experiment_config

Associates a SageMaker job as a trial component with an experiment and trial. Specified when you call the following APIs:

Show child fields
trial_component.source_detail.transform_job.experiment_config.experiment_name

The name of an existing experiment to associate with the trial component.

trial_component.source_detail.transform_job.experiment_config.run_name

The name of the experiment run to associate with the trial component.

trial_component.source_detail.transform_job.experiment_config.trial_component_display_name

The display name for the trial component. If this key isn't specified, the display name is the trial component name.

trial_component.source_detail.transform_job.experiment_config.trial_name

The name of an existing trial to associate the trial component with. If not specified, a new trial is created.

trial_component.source_detail.transform_job.failure_reason

If the transform job failed, the reason it failed.

trial_component.source_detail.transform_job.labeling_job_arn

The Amazon Resource Name (ARN) of the labeling job that created the transform job.

trial_component.source_detail.transform_job.max_concurrent_transforms

The maximum number of parallel requests that can be sent to each instance in a transform job. If MaxConcurrentTransforms is set to 0 or left unset, SageMaker checks the optional execution-parameters to determine the settings for your chosen algorithm. If the execution-parameters endpoint is not enabled, the default value is 1. For built-in algorithms, you don't need to set a value for MaxConcurrentTransforms.

trial_component.source_detail.transform_job.max_payload_in_mb

The maximum allowed size of the payload, in MB. A payload is the data portion of a record (without metadata). The value in MaxPayloadInMB must be greater than, or equal to, the size of a single record. To estimate the size of a record in MB, divide the size of your dataset by the number of records. To ensure that the records fit within the maximum payload size, we recommend using a slightly larger value. The default value is 6 MB. For cases where the payload might be arbitrarily large and is transmitted using HTTP chunked encoding, set the value to 0. This feature works only in supported algorithms. Currently, SageMaker built-in algorithms do not support HTTP chunked encoding.

trial_component.source_detail.transform_job.model_client_config

Configures the timeout and maximum number of retries for processing a transform job invocation.

Show child fields
trial_component.source_detail.transform_job.model_client_config.invocations_max_retries

The maximum number of retries when invocation requests are failing. The default value is 3.

trial_component.source_detail.transform_job.model_client_config.invocations_timeout_in_seconds

The timeout value in seconds for an invocation request. The default value is 600.

trial_component.source_detail.transform_job.model_name

The name of the model associated with the transform job.

trial_component.source_detail.transform_job.tags[]
Show child fields
trial_component.source_detail.transform_job.tags[].key

The tag key. Tag keys must be unique per resource.

trial_component.source_detail.transform_job.tags[].value

The tag value.

trial_component.source_detail.transform_job.transform_end_time

Indicates when the transform job has been completed, or has stopped or failed. You are billed for the time interval between this time and the value of TransformStartTime.

trial_component.source_detail.transform_job.transform_input

Describes the input source of a transform job and the way the transform job consumes it.

Show child fields
trial_component.source_detail.transform_job.transform_input.compression_type

If your transform data is compressed, specify the compression type. Amazon SageMaker automatically decompresses the data for the transform job accordingly. The default value is None.

trial_component.source_detail.transform_job.transform_input.content_type

The multipurpose internet mail extension (MIME) type of the data. Amazon SageMaker uses the MIME type with each http call to transfer data to the transform job.

trial_component.source_detail.transform_job.transform_input.data_source

Describes the location of the channel data, which is, the S3 location of the input data that the model can consume.

Show child fields
trial_component.source_detail.transform_job.transform_input.data_source.s3_data_source

The S3 location of the data source that is associated with a channel.

Show child fields
trial_component.source_detail.transform_job.transform_input.data_source.s3_data_source.s3_data_type

If you choose S3Prefix, S3Uri identifies a key name prefix. Amazon SageMaker uses all objects with the specified key name prefix for batch transform.

If you choose ManifestFile, S3Uri identifies an object that is a manifest file containing a list of object keys that you want Amazon SageMaker to use for batch transform.

The following values are compatible: ManifestFile, S3Prefix

The following value is not compatible: AugmentedManifestFile

trial_component.source_detail.transform_job.transform_input.data_source.s3_data_source.s3_uri

Depending on the value specified for the S3DataType, identifies either a key name prefix or a manifest. For example:

  • A key name prefix might look like this: s3://bucketname/exampleprefix/.

  • A manifest might look like this: s3://bucketname/example.manifest

    The manifest is an S3 object which is a JSON file with the following format:

    [ {"prefix": "s3://customer_bucket/some/prefix/"},

    "relative/path/to/custdata-1",

    "relative/path/custdata-2",

    ...

    "relative/path/custdata-N"

    ]

    The preceding JSON matches the following S3Uris:

    s3://customer_bucket/some/prefix/relative/path/to/custdata-1

    s3://customer_bucket/some/prefix/relative/path/custdata-2

    ...

    s3://customer_bucket/some/prefix/relative/path/custdata-N

    The complete set of S3Uris in this manifest constitutes the input data for the channel for this datasource. The object that each S3Uris points to must be readable by the IAM role that Amazon SageMaker uses to perform tasks on your behalf.

trial_component.source_detail.transform_job.transform_input.split_type

The method to use to split the transform job's data files into smaller batches. Splitting is necessary when the total size of each object is too large to fit in a single request. You can also use data splitting to improve performance by processing multiple concurrent mini-batches. The default value for SplitType is None, which indicates that input data files are not split, and request payloads contain the entire contents of an input object. Set the value of this parameter to Line to split records on a newline character boundary. SplitType also supports a number of record-oriented binary data formats. Currently, the supported record formats are:

  • RecordIO

  • TFRecord

When splitting is enabled, the size of a mini-batch depends on the values of the BatchStrategy and MaxPayloadInMB parameters. When the value of BatchStrategy is MultiRecord, Amazon SageMaker sends the maximum number of records in each request, up to the MaxPayloadInMB limit. If the value of BatchStrategy is SingleRecord, Amazon SageMaker sends individual records in each request.

Some data formats represent a record as a binary payload wrapped with extra padding bytes. When splitting is applied to a binary data format, padding is removed if the value of BatchStrategy is set to SingleRecord. Padding is not removed if the value of BatchStrategy is set to MultiRecord.

For more information about RecordIO, see Create a Dataset Using RecordIO in the MXNet documentation. For more information about TFRecord, see Consuming TFRecord data in the TensorFlow documentation.

trial_component.source_detail.transform_job.transform_job_arn

The Amazon Resource Name (ARN) of the transform job.

trial_component.source_detail.transform_job.transform_job_name

The name of the transform job.

trial_component.source_detail.transform_job.transform_job_status

The status of the transform job.

Transform job statuses are:

  • InProgress - The job is in progress.

  • Completed - The job has completed.

  • Failed - The transform job has failed. To see the reason for the failure, see the FailureReason field in the response to a DescribeTransformJob call.

  • Stopping - The transform job is stopping.

  • Stopped - The transform job has stopped.

trial_component.source_detail.transform_job.transform_output

Describes the results of a transform job.

Show child fields
trial_component.source_detail.transform_job.transform_output.accept

The MIME type used to specify the output data. Amazon SageMaker uses the MIME type with each http call to transfer data from the transform job.

trial_component.source_detail.transform_job.transform_output.assemble_with

Defines how to assemble the results of the transform job as a single S3 object. Choose a format that is most convenient to you. To concatenate the results in binary format, specify None. To add a newline character at the end of every transformed record, specify Line.

trial_component.source_detail.transform_job.transform_output.kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt the model artifacts at rest using Amazon S3 server-side encryption. The KmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

If you don't provide a KMS key ID, Amazon SageMaker uses the default KMS key for Amazon S3 for your role's account. For more information, see KMS-Managed Encryption Keys in the Amazon Simple Storage Service Developer Guide.

The KMS key policy must grant permission to the IAM role that you specify in your CreateModel request. For more information, see Using Key Policies in Amazon Web Services KMS in the Amazon Web Services Key Management Service Developer Guide.

trial_component.source_detail.transform_job.transform_output.s3_output_path

The Amazon S3 path where you want Amazon SageMaker to store the results of the transform job. For example, s3://bucket-name/key-name-prefix.

For every S3 object used as input for the transform job, batch transform stores the transformed data with an .out suffix in a corresponding subfolder in the location in the output prefix. For example, for the input data stored at s3://bucket-name/input-name-prefix/dataset01/data.csv, batch transform stores the transformed data at s3://bucket-name/output-name-prefix/input-name-prefix/data.csv.out. Batch transform doesn't upload partially processed objects. For an input S3 object that contains multiple records, it creates an .out file only if the transform job succeeds on the entire file. When the input contains multiple S3 objects, the batch transform job processes the listed S3 objects and uploads only the output for successfully processed objects. If any object fails in the transform job batch transform marks the job as failed to prompt investigation.

trial_component.source_detail.transform_job.transform_resources

Describes the resources, including ML instance types and ML instance count, to use for transform job.

Show child fields
trial_component.source_detail.transform_job.transform_resources.instance_count

The number of ML compute instances to use in the transform job. The default value is 1, and the maximum is 100. For distributed transform jobs, specify a value greater than 1.

trial_component.source_detail.transform_job.transform_resources.instance_type

The ML compute instance type for the transform job. If you are using built-in algorithms to transform moderately sized datasets, we recommend using ml.m4.xlarge or ml.m5.largeinstance types.

trial_component.source_detail.transform_job.transform_resources.volume_kms_key_id

The Amazon Web Services Key Management Service (Amazon Web Services KMS) key that Amazon SageMaker uses to encrypt model data on the storage volume attached to the ML compute instance(s) that run the batch transform job.

Certain Nitro-based instances include local storage, dependent on the instance type. Local storage volumes are encrypted using a hardware module on the instance. You can't request a VolumeKmsKeyId when using an instance type with local storage.

For a list of instance types that support local instance storage, see Instance Store Volumes.

For more information about local instance storage encryption, see SSD Instance Store Volumes.

The VolumeKmsKeyId can be any of the following formats:

  • Key ID: 1234abcd-12ab-34cd-56ef-1234567890ab

  • Key ARN: arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab

  • Alias name: alias/ExampleAlias

  • Alias name ARN: arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias

trial_component.source_detail.transform_job.transform_start_time

Indicates when the transform job starts on ML instances. You are billed for the time interval between this time and the value of TransformEndTime.

trial_component.start_time

When the component started.

trial_component.status

The status of the trial component.

Show child fields
trial_component.status.message

If the component failed, a message describing why.

trial_component.status.primary_status

The status of the trial component.

trial_component.tags[]
Show child fields
trial_component.tags[].key

The tag key. Tag keys must be unique per resource.

trial_component.tags[].value

The tag value.

trial_component.trial_component_arn

The Amazon Resource Name (ARN) of the trial component.

trial_component.trial_component_name

The name of the trial component.