Amazon SageMaker now offers new flexibility in configuring the model building pipelines
Amazon SageMaker Pipelines, the first purpose built continuous integration and continuous delivery (CI/CD) service for machine learning (ML), now allows customers to specify custom dependencies between the steps of the model building pipeline. Previously, specifying the output of a step as the input to another was the only option for specifying the dependency and the execution order between the two steps of the model building pipeline. Now, customers have the option of explicitly listing the steps that a given step execution needs to wait on.
Amazon SageMaker Pipelines, the first purpose built continuous integration and continuous delivery (CI/CD) service for machine learning (ML), now allows customers to specify custom dependencies between the steps of the model building pipeline. Previously, specifying the output of a step as the input to another was the only option for specifying the dependency and the execution order between the two steps of the model building pipeline. Now, customers have the option of explicitly listing the steps that a given step execution needs to wait on.
This new capability simplifies and provides customers additional flexibility in orchestrating the steps of the workflow to fit their model building requirements. This feature is available in all regions where SageMaker Pipelines is available. To learn more visit our documentation page.