Operator Guides¶

Documentation for platform administrators deploying and managing Kubeflow Trainer in production.

This section contains guides for installing, configuring, and operating Kubeflow Trainer in Kubernetes clusters.

Installation & Migration¶

Installation

Install Kubeflow Trainer using kubectl or Helm

Migration from v1

Migrate from Kubeflow Training Operator v1 to Trainer v2

Training Runtimes

Configure TrainingRuntime and ClusterTrainingRuntime resources

ML Policies

Define ML-specific policies for training workloads

Job Templates

Customize job templates for different frameworks

Runtime Patches

Customize training runtime configuration with RuntimePatches

Extension Framework

Understand the plugin-based extension architecture

Job Scheduling

Integrate with Volcano, Kueue, Coscheduling, and KAI Scheduler