Skip to content

chore(trainer): Remove accelerator label from the runtimes #51

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jul 28, 2025
Merged

chore(trainer): Remove accelerator label from the runtimes #51

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 0 additions & 5 deletions python/kubeflow/trainer/api/trainer_client_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -450,9 +450,6 @@ def create_cluster_training_runtime(
metadata=models.IoK8sApimachineryPkgApisMetaV1ObjectMeta(
name=name,
namespace=namespace,
labels={
"trainer.kubeflow.org/accelerator": "gpu-tesla-v100-16gb",
},
),
spec=models.TrainerV1alpha1TrainingRuntimeSpec(
mlPolicy=models.TrainerV1alpha1MLPolicy(
Expand Down Expand Up @@ -514,7 +511,6 @@ def create_runtime_type(
trainer_type=types.TrainerType.CUSTOM_TRAINER,
framework=types.Framework.TORCH,
entrypoint=[constants.TORCH_ENTRYPOINT],
accelerator="gpu-tesla-v100-16gb",
accelerator_count=4,
),
)
Expand All @@ -541,7 +537,6 @@ def get_train_job_data_type(
trainer_type=types.TrainerType.CUSTOM_TRAINER,
framework=types.Framework.TORCH,
entrypoint=["torchrun"],
accelerator="gpu-tesla-v100-16gb",
accelerator_count=4,
),
),
Expand Down
4 changes: 0 additions & 4 deletions python/kubeflow/trainer/constants/constants.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,10 +72,6 @@
# single VM where distributed training code is executed.
NODE = "node"

# The label key to identify the accelerator type for model training (e.g. GPU-Tesla-V100-16GB).
# TODO: Potentially, we should take this from the Node selectors.
ACCELERATOR_LABEL = "trainer.kubeflow.org/accelerator"

# Unknown indicates that the value can't be identified.
UNKNOWN = "Unknown"

Expand Down
1 change: 0 additions & 1 deletion python/kubeflow/trainer/types/types.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,7 +167,6 @@ class Trainer:
trainer_type: TrainerType
framework: Framework
entrypoint: Optional[List[str]] = None
accelerator: str = constants.UNKNOWN
accelerator_count: Union[str, float, int] = constants.UNKNOWN


Expand Down
8 changes: 0 additions & 8 deletions python/kubeflow/trainer/utils/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -140,14 +140,6 @@ def get_runtime_trainer(
if isinstance(trainer.accelerator_count, (int, float)) and ml_policy.num_nodes:
trainer.accelerator_count *= ml_policy.num_nodes

# TODO (andreyvelich): Currently, we get the accelerator type from
# the runtime labels.
if (
runtime_metadata.labels
and constants.ACCELERATOR_LABEL in runtime_metadata.labels
):
trainer.accelerator = runtime_metadata.labels[constants.ACCELERATOR_LABEL]

return trainer


Expand Down