手动创建跟踪实体

导入跟踪实体和关联

1. 导入世系追踪实体。

   import sys
   !{sys.executable} -m pip install -q sagemaker
   
   from sagemaker import get_execution_role
   from sagemaker.session import Session
   from sagemaker.lineage import context, artifact, association, action
   
   import boto3
   boto_session = boto3.Session(region_name=region)
   sagemaker_client = boto_session.client("sagemaker")

2. 创建输入和输出项目。

   code_location_arn = artifact.Artifact.create(
       artifact_name='source-code-location',
       source_uri='s3://...',
       artifact_type='code-location'
   ).artifact_arn
   
   # Similar constructs for train_data_location_arn and test_data_location_arn
   
   model_location_arn = artifact.Artifact.create(
       artifact_name='model-location',
       source_uri='s3://...',
       artifact_type='model-location'
   ).artifact_arn

3. 训练模型并获取trial_component_arn这代表培训作业。

4. 将输入项目和输出工件与训练作业（试用组件）关联。

   input_artifacts = [code_location_arn, train_data_location_arn, test_data_location_arn]
   for artifact_arn in input_artifacts:
       try:
           association.Association.create(
               source_arn=artifact_arn,
               destination_arn=trial_component_arn,
               association_type='ContributedTo'
           )
       except:
           logging.info('association between {} and {} already exists', artifact_arn, trial_component_arn)
   
   output_artifacts = [model_location_arn]
   for artifact_arn in output_artifacts:
       try:
            association.Association.create(
               source_arn=trial_component_arn,
               destination_arn=artifact_arn,
               association_type='Produced'
           )
       except:
           logging.info('association between {} and {} already exists', artifact_arn, trial_component_arn)

5. 创建推理端点。

   predictor = mnist_estimator.deploy(initial_instance_count=1,
                                        instance_type='ml.m4.xlarge')

6. 创建终端节点上下文。

   from sagemaker.lineage import context
   
   endpoint = sagemaker_client.describe_endpoint(EndpointName=predictor.endpoint_name)
   endpoint_arn = endpoint['EndpointArn']
   
   endpoint_context_arn = context.Context.create(
       context_name=predictor.endpoint_name,
       context_type='Endpoint',
       source_uri=endpoint_arn
   ).context_arn

7. 将培训作业（试用组件）和终端上下文关联起来。

   association.Association.create(
       source_arn=trial_component_arn,
       destination_arn=endpoint_context_arn
   )

跟踪从端点到培训数据源的工作流

1. 导入跟踪实体。

   import sys
   !{sys.executable} -m pip install -q sagemaker
   
   from sagemaker import get_execution_role
   from sagemaker.session import Session
   from sagemaker.lineage import context, artifact, association, action
   
   import boto3
   boto_session = boto3.Session(region_name=region)
   sagemaker_client = boto_session.client("sagemaker")

2. 从终端节点 ARN 获取终端节点上下文。

   endpoint_context_arn = sagemaker_client.list_contexts(
       SourceUri=endpoint_arn)['ContextSummaries'][0]['ContextArn']

3. 从试用组件和终端上下文之间的关联中获取试用组件。

   trial_component_arn = sagemaker_client.list_associations(
       DestinationArn=endpoint_context_arn)['AssociationSummaries'][0]['SourceArn']

4. 从试用组件和终端上下文之间的关联中获取训练数据位置工件。

   train_data_location_artifact_arn = sagemaker_client.list_associations(
       DestinationArn=trial_component_arn, SourceType='Model')['AssociationSummaries'][0]['SourceArn']

5. 从训练数据位置工件中获取训练数据位置。

   train_data_location = sagemaker_client.describe_artifact(
       ArtifactArn=train_data_location_artifact_arn)['Source']['SourceUri']
       print(train_data_location)

   响应:

   s3://sagemaker-sample-data-us-east-2/mxnet/mnist/train