Create Pipelines
First, you need to click the “+ Upload Pipeline” button in the top right corner of the screen to start creating pipelines.

You will be directed to the “New Pipelines” window to fill out several sections before you click the “Create” button.
a. Upload pipeline or pipeline version: Choose between creating a new pipeline or creating a new pipeline version under an existing pipeline. If you select “Create a new pipeline,” the menu will ask you for the name and description of the pipeline. If you select “Create a new pipeline version under an existing pipeline,” it will overwrite the existing pipeline with the new one you are importing, so the menu will ask you for the name of the pipeline that will be overwritten. In both cases, the pipeline can be imported by uploading a .yaml, .zip, or .tar.gz file, or by providing an external link containing the .yaml, .zip, or .tar.gz file.
b. Select if the new pipeline will be private or shared: Choose between Private or Shared. Privately created pipelines will appear in the private section of the main menu, while shared pipelines will be shown in the shared menu.
c. Upload pipeline with the specified package, you need to provide a Pipeline Name, while the description is optional. Then, you can choose “Upload a file” or “Import by URL”.

Once you have finished setting up, you can click the “Create” button.

Next, you will be redirected back to the Pipelines menu with the list of pipelines available in this Service Portal.

Click on the name of the pipeline you just created, and you will be automatically directed to the details window of that pipeline. The graph visualization that appears should be the same as if you created a pipeline using the Python SDK. Initially, you will see a summary of the pipeline you just imported, and if you close it, you will get a detailed view of how your pipeline is visualized in the form of a graph. If you want to reopen the summary of the pipeline, you can click the “Show summary” button at the bottom left.

You can obtain the pipeline you just imported in the form of a YAML file by pressing the YAML button. In the top right corner, there are three options available for you to use if needed:
Create run This opens a menu to create individual runs of the pipeline. This is how you reproduce a pipeline result workflow. A few text boxes will appear for you to describe and assign your pipeline.
Upload version This opens a menu to upload and update your pipeline. The menu is the same one you see when you import a pipeline by selecting “Create a new pipeline version under an existing pipeline.”
Create experiment This opens a menu to create an experiment. The menu should look exactly like the one you see when you create new experiments with Experiments (KFP). After you describe the name and description of the experiment, it will show you a menu for customizing your runs, just like when you select “+ Create run”.

Last updated