V1XGBoostJob

polyaxon.polyflow.run.kubeflow.xgboost_job.V1XGBoostJob(kind='xgboost_job', clean_pod_policy=None, scheduling_policy=None, master=None, worker=None)

Kubeflow XGBoost-Job provides an interface to train distributed experiments with XGBoost.

  • Args:

YAML usage

run:
  kind: xgbjob
  cleanPodPolicy:
  schedulingPolicy:
  master:
  worker:

Python usage

from polyaxon.polyflow import V1KFReplica, V1XGBoost
from polyaxon.k8s import k8s_schemas
xgb_job = V1XGBoost(
    clean_pod_policy='All',
    master=V1KFReplica(...),
    worker=V1KFReplica(...),
)

Fields

kind

The kind signals to the CLI, client, and other tools that this component’s runtime is a xgbjob.

If you are using the python client to create the runtime, this field is not required and is set by default.

run:
  kind: xgbjob

cleanPodPolicy

Controls the deletion of pods when a job terminates. The policy can be one of the following values: [All, Running, None]

run:
  kind: xgbjob
  cleanPodPolicy: 'All'
 ...

schedulingPolicy

SchedulingPolicy encapsulates various scheduling policies of the distributed training job, for example minAvailable for gang-scheduling.

run:
  kind: xgbjob
  schedulingPolicy:
    ...
 ...

master

The master replica in the distributed XGBoostJob.

run:
  kind: xgbjob
  ps:
    replicas: 2
    container:
      ...
 ...

worker

The server replica in the distributed XGBoostJob.

run:
  kind: xgbjob
  worker:
    replicas: 2
    container:
      ...
 ...