Practice Test 1 | Google Cloud Certified Professional Data Engineer | Dumps | Mock Test
The data scientists at your company have successfully built a deep neural network machine learning model to detect car plate numbers entering and exiting a parking lot of a high-rise condominium. The model was built using Tensorflow and the model was exported as SavedModel. As a data engineer, you are assigned to deploy their model. The company is using Google Cloud for its project.
Which approach is best for deploying the detection model?
A. Upload SavedModel object to Google Storage. Use Dataproc with Spark ML to use the model by accessing it using Google Storage Connector.
B. Deploy the model to Google Kubernetes Engine after wrapping SavedModel as docker image and uploading it to Google Container Registry.
C. Deploy the model to ML module in GCP after asking the data science team to convert the model to binary format using PyTorch.
D. Deploy the model exported as SavedModel directly to Vertex AI in GCP.
Answer: D
You can export your SavedModel directly to the Vertex AI. There is no need to take any other approach as it is more cumbersome.
Answer A is incorrect: This approach is too complicated and not necessary for this scenario.
Answer B is incorrect: There is no need to go through converting SavedModel to a docker image and use GKE. It’s not the recommended approach by Google anyway.
Answer C is incorrect because there is no need to convert the model to any other format. Cloud ML Engine supports Tensorflow and SavedModel format.
Source(s):
Google ML Engine: https://cloud.google.com/ml-engine/docs/tensorflow/prediction-overview
Deploying ML models to Google Cloud:
https://cloud.google.com/ml-engine/docs/tensorflow/deploying-models
Comments are closed, but trackbacks and pingbacks are open.