Examples and tutorials
This section introduces the examples in the kubeflow/examples repo.
Semantic code search
Use a Sequence to Sequence natural language processing model to perform a semantic code search. This tutorial runs in a Jupyter notebook and uses Google Cloud Platform (GCP).
Financial time series
Train and serve a model for financial time series analysis using TensorFlow on GCP.
GitHub issue summarization
Infer summaries of GitHub issues from the descriptions, using a Sequence to Sequence natural language processing model. You can run the tutorial in a Jupyter notebook or using TFJob. You use Seldon Core to serve the model.
MNIST image classification
Train and serve an image classification model using the MNIST dataset. You can choose to train the model locally, using GCP, or using Amazon S3. Serve the model using TensorFlow.
Object detection - cats and dogs
Train a distribued model for recognizing breeds of cats and dogs with the Tensorflow Object Detection API. Serve the model using TensorFlow.
Train a distributed PyTorch model on GCP and serve the model with Seldon Core.
Ames housing value prediction
Train an XGBoost model using the Kaggle Ames Housing Prices prediction on GCP. Use Selcon Core to serve the model locally, or GCP to serve it in the cloud.
Codelabs, workshops, and walkthroughs
Below is a list of recommended end-to-end tutorials, workshops, walkthroughs, and codelabs that are hosted outside the Kubeflow repositories.
Follow the tutorials to deploy Kubeflow and run a machine learning model.
Introduction to Kubeflow Codelab
Run MNIST with Kubeflow on Google Kubernetes Engine.
Introduction to Kubeflow Qwiklab
Run MNIST with resources provided by Qwiklabs.
Kubeflow End to End Codelab
Run GitHub Issue Summarization with Kubeflow on Google Kubernetes Engine.
Kubeflow End to End Qwiklab
Run GitHub Issue Summarization with resources provided by Qwiklabs.
The following blog posts provide detailed examples and use cases. Be aware that a blog post describes the interfaces at the time of publication of the post. Some interfaces are under rapid development and therefore may change frequently.
The Kubeflow blog
Visit the Kubeflow blog to keep up to date with news about the project and to learn how to use the latest features.
Using Kubeflow to train and serve a PyTorch model in Google Cloud Platform
January 23, 2019
This example demonstrates how you can use Kubeflow to train and serve a distributed Machine Learning model with PyTorch on a Google Kubernetes Engine cluster in Google Cloud Platform (GCP).
Getting started with Kubeflow Pipelines
November 16, 2018
This article describes how you can tackle ML workflow operations with Kubeflow Pipelines, and highlights some examples that you can try yourself. The examples revolve around a TensorFlow ‘taxi fare tip prediction’ model, with data pulled from a public BigQuery dataset of Chicago taxi trips.
How to create and deploy a Kubeflow machine learning pipeline
November 22 - December 4, 2018
A series of articles that walk you through the process of taking an existing real-world TensorFlow model and operationalizing the training, evaluation, deployment, and retraining of that model using Kubeflow Pipelines. Part 1 (creating and deploying a pipeline), and part 2 (using Jupyter notebooks).
Tutorials and overviews published in video format. Be aware that a video describes the interfaces at the time of publication of the video. Some interfaces are under rapid development and therefore may change frequently.
Machine Learning as Code: and Kubernetes with Kubeflow
December 15, 2018
Presenters: Jason “Jay” Smith and David Aronchick.
Artificial Intelligence at Cisco with Kubeflow
October 19, 2018
Presenter: Debo Dutta, Distinguished Engineer at Cisco.
CNCF (Cloud Native Computing Foundation) channel
A YouTube search for Kubeflow in the CNCF (Cloud Native Computing Foundation) channel.
Google Cloud Platform channel
A YouTube search for Kubeflow in the Google Cloud Platform channel.
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.