Sube un modelo para explicar el contenedor tabular administrado

Sube un modelo para explicar el contenedor tabular administrado con el método upload_model.

Muestra de código

Python

Antes de probar este ejemplo, sigue las instrucciones de configuración para Python incluidas en la guía de inicio rápido de Vertex AI sobre cómo usar bibliotecas cliente. Para obtener más información, consulta la documentación de referencia de la API de Vertex AI Python.

Para autenticarte en Vertex AI, configura las credenciales predeterminadas de la aplicación. Si deseas obtener más información, consulta Configura la autenticación para un entorno de desarrollo local.

from google.cloud import aiplatform_v1beta1


def upload_model_explain_tabular_managed_container_sample(
    project: str,
    display_name: str,
    container_spec_image_uri: str,
    artifact_uri: str,
    input_tensor_name: str,
    output_tensor_name: str,
    feature_names: list,
    location: str = "us-central1",
    api_endpoint: str = "us-central1-aiplatform.googleapis.com",
    timeout: int = 300,
):
    # The AI Platform services require regional API endpoints.
    client_options = {"api_endpoint": api_endpoint}
    # Initialize client that will be used to create and send requests.
    # This client only needs to be created once, and can be reused for multiple requests.
    client = aiplatform_v1beta1.ModelServiceClient(client_options=client_options)

    # Container specification for deploying the model
    container_spec = {"image_uri": container_spec_image_uri, "command": [], "args": []}

    # The explainabilty method and corresponding parameters
    parameters = aiplatform_v1beta1.ExplanationParameters(
        {"xrai_attribution": {"step_count": 1}}
    )

    # The input tensor for feature attribution to the output
    # For single input model, y = f(x), this will be the serving input layer.
    input_metadata = aiplatform_v1beta1.ExplanationMetadata.InputMetadata(
        {
            "input_tensor_name": input_tensor_name,
            # Input is tabular data
            "modality": "numeric",
            # Assign feature names to the inputs for explanation
            "encoding": "BAG_OF_FEATURES",
            "index_feature_mapping": feature_names,
        }
    )

    # The output tensor to explain
    # For single output model, y = f(x), this will be the serving output layer.
    output_metadata = aiplatform_v1beta1.ExplanationMetadata.OutputMetadata(
        {"output_tensor_name": output_tensor_name}
    )

    # Assemble the explanation metadata
    metadata = aiplatform_v1beta1.ExplanationMetadata(
        inputs={"features": input_metadata}, outputs={"prediction": output_metadata}
    )

    # Assemble the explanation specification
    explanation_spec = aiplatform_v1beta1.ExplanationSpec(
        parameters=parameters, metadata=metadata
    )

    model = aiplatform_v1beta1.Model(
        display_name=display_name,
        # The Cloud Storage location of the custom model
        artifact_uri=artifact_uri,
        explanation_spec=explanation_spec,
        container_spec=container_spec,
    )
    parent = f"projects/{project}/locations/{location}"
    response = client.upload_model(parent=parent, model=model)
    print("Long running operation:", response.operation.name)
    upload_model_response = response.result(timeout=timeout)
    print("upload_model_response:", upload_model_response)

¿Qué sigue?

Para buscar y filtrar muestras de código para otros productos de Google Cloud, consulta el navegador de muestra de Google Cloud.