Sabari Narayana comments

Results 13 comments of


                                            Sabari Narayana

Hy

spam Issuues this can be closed

Duplicate hyperparameters waste compute and time

# Proposal: Prevent Duplicate Hyperparameter Suggestions **Issue**: [#2571](https://github.com/kubeflow/katib/issues/2571) **Summary**: Add duplicate detection in suggestion controller to prevent wasted compute on identical hyperparameter trials. ## Problem Katib allows duplicate suggestions to...

Duplicate hyperparameters waste compute and time

@andreyvelich Could you please review this proposal and let me know if you have any feedback? Once I get your input, I’ll start working on the issue.

Duplicate hyperparameters waste compute and time

Thanks @andreyvelich for the feedback! You're absolutely right - addressing the root cause makes more sense. I'll take over the migration from deprecated scikit-optimize to Optuna's GPSampler (#2280) first. **Plan:**...

Duplicate hyperparameters waste compute and time

/assign

Don't require the Trainjob to be suspended to update podTemplateOverrides

## TrainJob: Single API Request for Unsuspend + PodTemplateOverrides Update ## Related Issues - **Trainer:** [kubeflow/trainer#3043](https://github.com/kubeflow/trainer/issues/3043) - **Kueue:** [kubernetes-sigs/kueue#8296](https://github.com/kubernetes-sigs/kueue/issues/8296) --- ## Problem Statement When Kueue starts a TrainJob, it currently...

Don't require the Trainjob to be suspended to update podTemplateOverrides

Thanks for the detailed breakdown @kaisoz! This makes the path forward very clear. To summarize the fix needed in `pkg/runtime/framework/plugins/jobset/jobset.go` at `checkPodTemplateOverridesImmutability()`: **Current logic (line 167):** ```go suspended := ptr.Equal(newObj.Spec.Suspend,...