Samples and tutorials

Beta

This product is subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the Service Specific Terms. Pre-GA products are available "as is" and might have limited support. For more information, see the launch stage descriptions.

This page summarizes samples and tutorials that demonstrate how to use the assets that are available on AI Hub.

Kubeflow pipelines

The following samples and tutorials illustrate how to use Kubeflow pipelines.

• GitHub issue summarization is an advanced codelab focused on creating Kubeflow pipelines. This codelab demonstrates how to:

  • Set up a Kubeflow cluster using Google Kubernetes Engine
  • Build and run ML workflows using Kubeflow Pipelines
  • Run pipelines from within a Jupyter notebook

• Kubeflow Pipelines end-to-end on Google Kubernetes Engine is a tutorial that demonstrates how to:

  • Set up a Kubeflow cluster using Google Kubernetes Engine
  • Compile a sample pipeline by using the Kubeflow Pipelines SDK
  • Run the pipeline

• Getting started with Kubeflow Pipelines is a blog post that provides an introduction to building machine learning (ML) workflows with Kubeflow Pipelines. This blog post uses sample pipelines to demonstrate how to create, analyze, monitor, and make predictions with Kubeflow Pipelines.

The following videos provide overviews and demonstrations of Kubeflow Pipelines:

• Watch the AI Adventures introduction to Kubeflow for an overview of ML training and prediction on Kubernetes clusters.
• Watch understanding the earth, ML with Kubeflow Pipelines, for an overview of how Descartes Labs is using Kubeflow Pipelines to orchestrate ML workloads that process petabytes of satellite imagery.

Introduction to Kubeflow pipelines

This section provides background information to help you complete the samples and tutorials. The section is particularly useful if you're getting started with Kubeflow pipelines

Kubeflow pipelines are portable and scalable end-to-end machine learning (ML) workflows built using the Kubeflow Pipelines SDK. To learn how to create pipelines, see the Kubeflow pipelines author's guide.

Kubeflow Pipelines is a component of Kubeflow that provides a platform for building and deploying ML workflows. Kubeflow is an open source toolkit for running ML workloads on Kubernetes. Learn more about getting started with Kubeflow.

To set up Kubeflow Pipelines on Google Kubernetes Engine (GKE), choose one of the following options:

• Deploy Kubeflow using the command line. Choose this option if you want to deploy Kubeflow.
• Deploy Kubeflow Pipelines standalone to a cluster. Choose this option if you want to deploy only Kubeflow Pipelines to a GKE cluster.

TensorFlow modules

The following samples and tutorials illustrate how to use TensorFlow modules.

• Learn how to retrain a TensorFlow 2 SavedModel module to classify images of flowers.
• How to build a simple text classifier demonstrates how to use the text embedding module to train a simple sentiment classifier.
• Watch the TF Dev Summit 2019 overview of TensorFlow Hub to learn how you can reuse pieces of a TensorFlow graph to train a new model.

What's next

• Understand important concepts and terms by reading the introduction to AI Hub.