Updates constants referenced for kubeflow objects and set lower bound…

… on (#385) library to 1.8 Signed-off-by: Terry Kong <[email protected]>
NVIDIA · Jul 17, 2024 · eb9cc9e · eb9cc9e
1 parent b7cdbb0
commit eb9cc9e
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 4 deletions.
diff --git a/launcher_scripts/nemo_launcher/core/v2/step_k8s.py b/launcher_scripts/nemo_launcher/core/v2/step_k8s.py
@@ -166,7 +166,7 @@ def create_pytorchjob_resource(
         ),
     )
     pytorch_job = KubeflowOrgV1PyTorchJob(
-        api_version=f"{constants.KUBEFLOW_GROUP}/{constants.OPERATOR_VERSION}",
+        api_version=f"{constants.API_VERSION}",
         kind=constants.PYTORCHJOB_KIND,
         metadata=V1ObjectMeta(generate_name=generate_name, namespace=namespace),
         spec=KubeflowOrgV1PyTorchJobSpec(
@@ -290,7 +290,7 @@ def replica_template(n_replicas: int, container: V1Container):
     launcher = replica_template(n_replicas=1, container=launch_container,)
     worker = replica_template(n_replicas=n_workers, container=worker_container,)
     mpijob = KubeflowOrgV1MPIJob(
-        api_version=f"{constants.KUBEFLOW_GROUP}/{constants.OPERATOR_VERSION}",
+        api_version=f"{constants.API_VERSION}",
         kind=constants.MPIJOB_KIND,
         metadata=V1ObjectMeta(generate_name=generate_name, namespace=namespace),
         spec=KubeflowOrgV1MPIJobSpec(
@@ -334,7 +334,7 @@ def replica_template(n_replicas: int, container: V1Container):
 def delete_pytorchjob(name: str = "delete-pytorchjob"):
     manifest = dedent(
         f"""
-        apiVersion: {constants.KUBEFLOW_GROUP}/{constants.OPERATOR_VERSION}
+        apiVersion: {constants.API_VERSION}
         kind: {constants.PYTORCHJOB_KIND}
         metadata:
           name: {{{{inputs.parameters.metadata_name}}}}

diff --git a/requirements.txt b/requirements.txt
@@ -12,5 +12,5 @@ zstandard==0.15.2
 sqlitedict
 hera
 pydantic
-kubeflow-training
+kubeflow-training>=1.8
 kubernetes