Additional tutorials in the form of Jupyter notebooks

The following Jupyter notebooks walk through various Vertex AI workflows:

AutoML

• Create, train, and deploy an AutoML text classification model

Custom training

• Custom training and batch prediction
• Custom training and online prediction

Vertex Explainable AI

The following notebooks use Vertex AI SDK for Python's high-level aiplatform library for most API calls:

• AutoML tabular binary classification model for batch prediction with explanation
• AutoML tabular classification model for online prediction with explanation
• Custom training image classification model for batch prediction with explanation
• Custom training image classification model for online prediction with explanation
• Custom training tabular regression model for batch prediction with explanation
• Custom training tabular regression model for online prediction with explanation

The following notebooks use the Vertex AI SDK for Python's lower-level aiplatform.gapic library. You might want to reference these notebooks if the high-level aiplatform library doesn't work for your use case:

• Custom training image classification model for batch prediction with explanation ("GAPIC")
• Custom training image classification model for online prediction with explanation ("GAPIC")
• Custom training tabular regression model for batch prediction with explanation ("GAPIC")
• Custom training tabular regression model for online prediction with explanation ("GAPIC")

Vertex AI Feature Store

• Manage sample ML features by using Vertex AI Feature Store

Model Monitoring

• Set up Model Monitoring to detect drift and training-prediction skew

Vertex ML Metadata

• Track parameters and metrics for custom training jobs
• Track parameters and metrics for locally trained models

Vertex AI Pipelines

• Create a pipeline with lightweight components based on Python functions
• Use a pipeline to create an AutoML image classification model
• Use a pipeline to create an AutoML tabular classification model
• Use a pipeline to create an AutoML tabular linear regression model
• Use a pipeline to create an AutoML text classification model
• Use a pipeline to create a custom-trained model
• Use the Kubeflow Pipelines SDK to build a pipeline with control structures
• Use the Kubeflow Pipelines SDK to visualize metrics and compare pipeline runs
• Use Vertex AI Pipelines with the Kubeflow Pipelines SDK

Vertex AI Vizier

• Use Vertex AI Vizier to optimize for multiple objectives simultaneously