kubeflow.fairing.deployers.tfjob package¶
Submodules¶
kubeflow.fairing.deployers.tfjob.tfjob module¶
-
class
kubeflow.fairing.deployers.tfjob.tfjob.TfJob(namespace=None, worker_count=1, ps_count=0, chief_count=1, runs=1, job_name='fairing-tfjob-', stream_log=True, labels=None, pod_spec_mutators=None, cleanup=False, annotations=None)¶ Bases:
kubeflow.fairing.deployers.job.job.JobHandle all the k8s’ template building to create tensorflow training job using Kubeflow TFOperator
-
create_resource()¶ create a tfjob training
-
generate_deployment_spec(pod_template_spec)¶ Returns a TFJob template
Parameters: pod_template_spec – template spec for pod
-
get_logs()¶ get logs
-
set_container_name(pod_template_spec)¶ - Sets the name of the main container to tensorflow.
- This is required for TfJobs
Parameters: pod_template_spec – spec for pod template
-