feat(api): BREAKING CHANGE: Remove numProcPerNode from Torch API

Signed-off-by: Andrey Velichkevich <andrey.velichkevich@gmail.com>

Author: andreyvelich

api/openapi-spec/swagger.json

@@ -594,21 +594,6 @@ spec:
                             format: int32
                             type: integer
                         type: object
-                      numProcPerNode:
-                        anyOf:
-                        - type: integer
-                        - type: string
-                        default: auto
-                        description: |-
-                          numProcPerNode is the number of processes per node.
-                          This value is inserted into the `--nproc-per-node` argument of the `torchrun` CLI.
-                          Supported values: `auto`, `cpu`, `gpu`, or int value.
-                          Defaults to `auto`.
-                        x-kubernetes-int-or-string: true
-                        x-kubernetes-validations:
-                        - message: NumProcPerNode must be equal to auto, cpu, gpu,
-                            or int value
-                          rule: self > 0 || self in ['auto', 'cpu', 'gpu']
                     type: object
                 type: object
                 x-kubernetes-validations:

api/python_api/kubeflow_trainer_api/models/trainer_v1alpha1_torch_ml_policy_source.py

@@ -594,21 +594,6 @@ spec:
                             format: int32
                             type: integer
                         type: object
-                      numProcPerNode:
-                        anyOf:
-                        - type: integer
-                        - type: string
-                        default: auto
-                        description: |-
-                          numProcPerNode is the number of processes per node.
-                          This value is inserted into the `--nproc-per-node` argument of the `torchrun` CLI.
-                          Supported values: `auto`, `cpu`, `gpu`, or int value.
-                          Defaults to `auto`.
-                        x-kubernetes-int-or-string: true
-                        x-kubernetes-validations:
-                        - message: NumProcPerNode must be equal to auto, cpu, gpu,
-                            or int value
-                          rule: self > 0 || self in ['auto', 'cpu', 'gpu']
                     type: object
                 type: object
                 x-kubernetes-validations:

api/python_api/kubeflow_trainer_api/models/trainer_v1alpha1_trainer.py

@@ -4843,14 +4843,12 @@ spec:
                     format: int32
                     type: integer
                   numProcPerNode:
-                    anyOf:
-                    - type: integer
-                    - type: string
                     description: |-
                       numProcPerNode is the number of processes/workers/slots on every training node.
-                      For the Torch runtime: `auto`, `cpu`, `gpu`, or int value can be set.
-                      For the MPI runtime only int value can be set.
-                    x-kubernetes-int-or-string: true
+                      For the MPI runtime only int value can be set to represent number of slots per node.
+                      For the Torch runtime the value defaults to `auto` and can be overridden with an int.
+                    format: int32
+                    type: integer
                   resourcesPerNode:
                     description: resourcesPerNode defines the compute resources for
                       each training node.

charts/kubeflow-trainer/crds/trainer.kubeflow.org_clustertrainingruntimes.yaml

@@ -29,8 +29,7 @@ metadata:
 spec:
   mlPolicy:
     numNodes: 1
-    torch:
-      numProcPerNode: auto
+    torch: {}
   template:
     spec:
       replicatedJobs:

charts/kubeflow-trainer/crds/trainer.kubeflow.org_trainingruntimes.yaml

@@ -25,8 +25,7 @@ metadata:
 spec:
   mlPolicy:
     numNodes: 1
-    torch:
-      numProcPerNode: auto
+    torch: {}
   template:
     spec:
       replicatedJobs:

charts/kubeflow-trainer/crds/trainer.kubeflow.org_trainjobs.yaml

@@ -25,8 +25,7 @@ metadata:
 spec:
   mlPolicy:
     numNodes: 1
-    torch:
-      numProcPerNode: auto
+    torch: {}
   template:
     spec:
       volumeClaimPolicies:

charts/kubeflow-trainer/templates/runtimes/data-cache/torch-distributed-with-cache.yaml

@@ -25,8 +25,7 @@ metadata:
 spec:
   mlPolicy:
     numNodes: 1
-    torch:
-      numProcPerNode: auto
+    torch: {}
   template:
     spec:
       volumeClaimPolicies: