Runs
Last updated
Last updated
The Runs menu execute a pipeline in the Service Portal Deka GPU. Each time a pipeline is executed, it is considered one run. On the Service Portal Cloudeka Deka MLOps, select the Runs menu.
In the Runs menu, there are several submenus along with their explanations as follows.
a. Active, serves to manage and monitor currently running or recently completed pipeline runs. This section provides essential functionalities to interact with ongoing experiments and evaluate their performance. The options available in the Active menu include.
Refresh
This option refreshes the list of active runs, ensuring that you have the most up-to-date information displayed.
Archive
This functionality allows you to archive runs that are no longer active, helping to keep your workspace organized by removing completed runs from the active list.
Clone Run
This option lets you duplicate an existing run, making it easier to rerun experiments with the same configuration without manually re-entering the details.
Compare Runs
This feature enables you to compare the results of different runs side by side, helping you analyze the performance and outcomes of various experiments effectively.
Create Runs
This feature allows you to initiate a new run of a pipeline. When you select "+ Create Run" you will be prompted to fill out necessary details, such as the pipeline to be used, run type (one-off or recurring), and any parameters required for the execution. This is the starting point for executing your machine learning workflows in Kubeflow Pipelines.
b. Archived, you will find only the Refresh option. The purpose of this option is to update the archived runs list, ensuring that any newly archived runs are visible. Since the archived runs are not actively running, there are no additional management options provided in this menu. The focus is on keeping the archived data current without cluttering the interface with unnecessary functionalities.