using add_done_callback() and reponse.metadata() for automl

Question

I'm trying to use a standard example for automl. I would expect the create_model to launch a long running operation that will update the operation response once it's done and to then access the metadata (to get the model_id of the newly trained model) . but the script fails right away on metadata = response.metadata() with "TypeError: 'OperationMetadata' object is not callable".

I'm running this code inside a PythonOperator in Airflow/google composer if it makes any difference. I can see in the AutoML UI that the model starts training correctly.

My code is this, but it's basically the example usage that's in the docs.:

from google.cloud import automl

client = automl.AutoMlClient()

...
response = client.create_model(project_location, my_model)

def callback(operation_future):
   # Handle result.
   result = operation_future.result()

response.add_done_callback(callback)
metadata = response.metadata()

I'm using google-cloud-automl==0.9.0

I would recommend opening this as an issue at https://github.com/googleapis/google-cloud-python/issues. — Dustin Ingram, Dec 02 '19 at 17:35

score 1 · Accepted Answer · answered Jan 12 '20 at 23:52

I ran into the same problem. That example in the docs really led me astray. If you call the result() method on the response, it will actually wait until the model is finished training before it returns anything, so that's all you need to do. You can then get the model_id from that result.

import time
from google.cloud import automl

client = automl.AutoMlClient()

...
response = client.create_model(project_location, my_model)

model_id = response.result().name.split('/')[-1]

score 0 · Answer 2 · answered Dec 09 '19 at 21:35

0

Does https://googleapis.dev/python/automl/latest/gapic/v1beta1/tables.html help? It's a manually written Python SDk built on top of automl.AutoMlClient.

answered Dec 09 '19 at 21:35

Helin Wang

56
2

score 0 · Answer 3 · answered Dec 10 '19 at 12:30

I would suggest that instead of using the example you provided here basic usage that's in the docs.:.

It would be better to use the Google Cloud AutoML Operators How-to Guides Using Operators Google Cloud Operators.

Using the the auto_ml specific operators is the recommended method for the implementation of Auto_ML models within a Airflow/Composer environment.

Here is the complete API reference for the Python Client for Cloud AutoML API

using add_done_callback() and reponse.metadata() for automl

3 Answers3