How to add custom metadata to a model in model registry via sagemaker pipeline step?

Question

based on sagemaker python sdk  , i can add custom/customer metadata via (sample code below). i am planning to use sagemaker model step and register the model in a model package group and create an end to end pipeline. is it possible to add these metadata at the time of model creation or registering model ?

```
boto3.client('sagemaker').update_model_package( ... 
CustomerMetadataProperties = {'key1': 'value1', 'key2': 'value2'})
```

Accepted Answer

Yes, it is possible. You can add custom metadata when you use the [register](https://sagemaker.readthedocs.io/en/stable/api/inference/model.html#sagemaker.model.Model.register) method of the Model object in the SageMaker SDK. The [Model step](https://docs.aws.amazon.com/sagemaker/latest/dg/build-and-manage-steps.html#step-type-model) from SageMaker Pipelines can then be used together with this register method.

Below is a snippet of code copied from the example provided on [this page](https://docs.aws.amazon.com/sagemaker/latest/dg/build-and-manage-steps.html#step-type-model), where I have added the `custom_metadata_properties` parameter with your example input. 
 
```
register_model_step_args = pipeline_model.register(
    content_types=["application/json"],
   response_types=["application/json"],
   inference_instances=["ml.t2.medium", "ml.m5.xlarge"],
   transform_instances=["ml.m5.xlarge"],
   model_package_group_name="sipgroup",
   customer_metadata_properties={"key1": "value1", "key2": "value2"}
)

step_model_registration = ModelStep(
   name="AbaloneRegisterModel",
   step_args=register_model_step_args,
)
```

How to add custom metadata to a model in model registry via sagemaker pipeline step?

相關內容