openeuler/kubeflow repository overview

⁠Quick reference

• The official Kubeflow docker image.

• Maintained by: openEuler CloudNative SIG⁠.

• Where to get help: openEuler CloudNative SIG⁠, openEuler⁠.

⁠Kubeflow | openEuler

Current Kubeflow images are built on the openEuler⁠. This repository is free to use and exempted from per-user rate limits.

Kubeflow is an open-source machine learning platform built on Kubernetes, providing end-to-end workflow management for AI/ML workloads. It is a CNCF graduated project that offers a comprehensive ecosystem including Notebook management, model training, hyperparameter tuning, model serving, and inference. This image includes the core notebook-controller and profile-controller components from the kubeflow/kubeflow repository.

Read more on Kubeflow Website⁠.

⁠Supported tags and respective dockerfile links

The tag of each kubeflow docker image is consist of the version of kubeflow and the version of basic image. The details are as follows

⁠Usage

In this usage, users can select the corresponding {Tag} based on their requirements.

• Pull the openeuler/kubeflow image from docker

  docker pull openeuler/kubeflow:{Tag}
  

  Copy

• Start the notebook-controller

  docker run -it --rm openeuler/kubeflow:{Tag} notebook-controller
  

  Copy

• Start the profile-controller

  docker run -it --rm openeuler/kubeflow:{Tag} profile-controller
  

  Copy

• Run with an interactive shell

  docker run -it --rm openeuler/kubeflow:{Tag} bash
  

  Copy

  The openeuler/kubeflow image is used to verify the integration between the upstream Kubeflow version and openEuler.

⁠Question and answering

If you have any questions or want to use some special features, please submit an issue or a pull request on openeuler-docker-images⁠.