tf-operator

Kubeflow Training Operator

Overview

Before v1.2 release, tensorflow-operator can only support TFJob on Kubernetes. Starting from v1.3, Training Operator provides Kubernetes custom resources that makes it easy to run distributed or non-distributed TensorFlow/PyTorch/MXNet/XGBoost jobs on Kubernetes.

• For a complete reference of the custom resource definitions, please refer to the API Definition.
  • Tensorflow API Definition
  • PyTorch API Definition
  • MXNet API Definition
  • XGBoost API Definition
• For details on API design, please refer to the v1alpha2 design doc.
• For details of all-in-one operator design, please refer to the All-in-one Kubeflow Training Operator
• For details on its obersibility, please refer to the monitoring design doc.

Prerequisites

• Version >= 1.16 of Kubernetes
• Version >= 3.x of Kustomize
• Version >= 1.21.x of Kubectl

Installation

Master Branch

kubectl apply -k "github.com/kubeflow/tf-operator.git/manifests/overlays/standalone?ref=master"

Specific Release

kubectl apply -k "github.com/kubeflow/tf-operator.git/manifests/overlays/standalone?ref=v1.3.0"

Tensorflow Release Only

For users who prefer to use original tensorflow controllers, please checkout v1.2-branch, we will maintain the bug fix in this branch.

kubectl apply -k "github.com/kubeflow/tf-operator.git/manifests/overlays/standalone?ref=v1.2.0"

Quick Start

Please refer to the quick-start-v1.md and Kubeflow Training User Guide for more information.

API Documentation

Please refer to API Documentation.

• Tensorflow API Documentation
• PyTorch API Documentation
• MXNet API Documentation
• XGBoost API Documentation

Community

You can:

• Join our Slack channel.
• Check out who is using this operator.

This is a part of Kubeflow, so please see readme in kubeflow/kubeflow to get in touch with the community.

Contributing

Please refer to the DEVELOPMENT

Change Log

Please refer to CHANGELOG

Version Matrix

The following table lists the most recent few versions of the operator.