Data Science Model Deployments and Cloud Computing on GCP - Lab - Custom Model Training Using SDK and Model Registries

The video tutorial demonstrates how to complete a manual training job using Vertex AI's web console and then replicate the process using Python's SDK. It covers setting up the Python SDK, Docker image, and service account, followed by executing a custom training job. Finally, it explains how to upload the trained model to Vertex AI's model registry, highlighting the importance of versioning and labeling models.

What is the purpose of using Python's SDK in the context of Vertex AI?

• To manually submit training jobs using the web console

• To automate the process of running custom training jobs

• To create Docker images for deployment

• To manage GCS bucket permissions

What is a necessary step when running the training job from a local computer?

• Setting up a new service account

• Building a new Docker image

• Exporting Google application credentials

• Creating a new GCS bucket

Which parameter is NOT required when defining a custom container training job?

• Display name

• Container URI

• Accelerator type

• Bucket name

What is the minimum replica count needed to run a training job?

• Two

• One

• Zero

• Three

What is the purpose of the model registry in Vertex AI?

• To export Google application credentials

• To maintain different versions of models

• To manage service accounts

• To store Docker images

What should be avoided in the version alias when uploading a model?

• Using spaces

• Using numbers

• Using uppercase letters

• Using special characters

What is an artifact URI in the context of model uploading?

• The location of the Docker image

• The location of the joblib file on GCS

• The URI of the service account

• The URI of the container registry