Update getting-started.md #3950
Conversation
Example has a couple of bugs: * missing package * create_job argument deprecated Signed-off-by: Ronald Petty <[email protected]>
|
Hi @ronaldpetty. Thanks for your PR. I'm waiting for a kubeflow member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
There was a problem hiding this comment.
Thanks for the fix @ronaldpetty!
I left one comment.
/assign @kubeflow/wg-training-leads
| @@ -116,8 +117,7 @@ from kubeflow.training import TrainingClient | |||
| # Start PyTorchJob with 3 Workers and 1 GPU per Worker (e.g. multi-node, multi-worker job). | |||
| TrainingClient().create_job( | |||
| name="pytorch-ddp", | |||
| train_func=train_func, | |||
| num_procs_per_worker="auto", | |||
There was a problem hiding this comment.
We might need to keep it since this value indicates that we should use torchrun as an entrypoint: https://github.com/kubeflow/training-operator/blob/master/sdk/python/kubeflow/training/api/training_client.py#L479
There was a problem hiding this comment.
I have deleted my (cluster) environment, but will try to recreate. From what I saw "num_procs_per_worker" was removed (failed to run). I also loaded locally in a REPL and looked at the help and it was missing there as well.
There was a problem hiding this comment.
@ronaldpetty Can you check version of SDK that you are using ?
pip show kubeflow-training
Example has a couple of bugs: