Data Science Model Deployments and Cloud Computing on GCP - Lab - Model Serving Using Endpoint with Python SDK

The video tutorial covers deploying an endpoint in Vertex AI, verifying its status, and running predictions. It explains optional deployment parameters like traffic splitting and replica counts. A function for batch predictions using Python is demonstrated. The tutorial concludes with a summary and an assignment for further practice.

What is the first step to verify if an endpoint is active in Vertex AI?

• Check the endpoint status in the console

• Run a prediction command

• Deploy a new model

• Restart the endpoint

What must be installed to run predictions on App Engine or Cloud Functions?

• Vertex AI SDK

• AI Platform Library

• Python 3.8

• Google Cloud SDK

How can you split traffic between two model versions?

• By using the traffic_split parameter

• By using a load balancer

• By deploying them on separate endpoints

• By setting different machine types

What is the purpose of setting minimum and maximum replica counts?

• To reduce deployment time

• To ensure model accuracy

• To manage resource allocation based on demand

• To control the number of predictions

What is the role of the prediction service client in the custom function?

• It monitors endpoint status

• It manages endpoint creation

• It handles prediction requests

• It deploys models

What is the main advantage of using a custom function for predictions?

• It reduces the cost of predictions

• It allows for more complex models

• It increases prediction speed

• It simplifies the prediction process

What is the expected outcome when running the custom function with a single list input?

• A deployment status

• A prediction result

• An error message

• A list of endpoints