how to update model version on model registry

Google Cloud Build with AI Custom ML & MLOps
1

Hi, currently I’m working on vertex ai pipelines. I’m confused on how to update my model version on model registry.

Versions:

kfp                                    2.0.0rc1
kfp-pipeline-spec                      0.2.2
kfp-server-api                         2.0.0rc1
google_cloud_pipeline_components       2.0.0b

Here’s the code snippet:

models = aiplatform.Model.list(filter=(“display_name={}“).format(MODEL_DISPLAY_NAME))
parent_model = models[0].resource_name
model_upload_op = ModelUploadOp(
project=PROJECT_ID,
display_name=MODEL_DISPLAY_NAME,
parent_model=,
unmanaged_container_model=unmanaged_model_importer.outputs[“artifact”],
).after(unmanaged_model_importer)

Error message:

InconsistentTypeException: Incompatible argument passed to the input 'parent_model' of component 'model-upload':
Argument type 'STRING' is incompatible with the input type 'google.VertexModel@0.0.1'

Is there any way to resolve this?
Thank you.

2

I also tried another method as below. No error, but it still uploaded to the model registry as a new model, not a new version.

models = aiplatform.Model.list(filter=(“display_name={}“).format(MODEL_DISPLAY_NAME))
model_id = models[0].resource_name
model_registry_importer = importer_node.importer(
    artifact_uri=f'https://us-central1-aiplatform.googleapis.com/v1/{model_id}',     
    artifact_class=artifact_types.VertexModel,
    metadata={"model_resource_name": model_id},
).after(unmanaged_model_importer)

model_upload_op = ModelUploadOp(
    project=PROJECT_ID,
    display_name=MODEL_DISPLAY_NAME,
    parent_model=model_registry_importer.outputs["artifact"],
    unmanaged_container_model=unmanaged_model_importer.outputs["artifact"],
).after(model_registry_importer)