Pipeline failed to deploy model: "service_account cannot be specified for deploying AutoML models"

[sangersteel]

I made a pipeline that almost mirrors step 6 of Intro to Vertex Pipelines which has managed to get past every step up until the model deployment side of things. The code snippet for my model deploy op is here:

And the associated error message in the logs for the deployment part of the pipeline was:

gcc_aip.ModelDeployOp(
    model=training_op.outputs["model"],
    endpoint=endpoint_op.outputs["endpoint"],
    dedicated_resources_min_replica_count=1,
    dedicated_resources_max_replica_count=1,
    dedicated_resources_machine_type="n1-standard-4",
    service_account = "vertex-notebooks@[redacted].iam.gserviceaccount.com"
    )

RuntimeError: Failed to create the resource. Error: {'code': 400, 'message': 'service_account cannot be specified for deploying AutoML Models.', 'status': 'FAILED_PRECONDITION'}

Does it have to do with a specific permission I need to give my service account? I don't know how to interpret this error.

[sf]

You are sure you can specify a service account here like this? Do you have any reference that uses service account here?

[1] https://cloud.google.com/blog/topics/developers-practitioners/use-vertex-pipelines-build-automl-...