yingding / kf-examples

this repository contains examples of kubeflow pipeline

kf-examples

this repository contains examples of kubeflow pipeline for the talks at:

• Kubeflow Workshop and hands-on tutorials on 23 Feb. 2024
• GDG Cloud Munich October Gathering on 5 Oct. 2022

Notebooks

• exploring-iris-data-notebook: explores the iris dataset used later in the pipelines
• single-model-pipeline-example: demonstrates a pipeline trains a single ANN model with tensorflow and shows the performance metrics and confusion matrix visualization for the test dataset.
• multiple-models-pipeline-example: demonstrates a pipeline trains an ANN model with tensorflow, a SVM model with scikit-learn and shows the performance metrics and confusion matrix visualization for both trained models performing on the test dataset.
• patch-notebook-env-kfp-run: helps to setup the right environment while running a kfp pipeline with kfp python sdk from notebook the first time.

Issue

Should you encounter a 500 server issue while runing the demo pipelines with kfp python sdk for the first time from the Kubeflow notebook server.

Please double check whether the kfp-server-api 1.6.2 is outdated.

• For KF 1.4.1, kfp-server-api 1.7.1 are necessary.
• For KF 1.5.1, kfp-server-api 1.8.2 are necessary.
• For KF 1.6.1, KF 1.7.0 kfp-server-api 1.8.5 are necessary.

The major version number to the kfp-server backend shall match, so that a pipeline can be started.