Module job_k8s 

Kubernetes-based distributed training backend.

Generates PyTorchJob YAML manifests and manages them via kubectl.

Functions

cancel_k8s_job

Cancel a Kubernetes job.

generate_pytorchjob_yaml 🔒

Generate a Kubeflow PyTorchJob YAML manifest.

run_k8s_job

Submit a distributed training job to Kubernetes via PyTorchJob CRD.