Kubeflow Pipelines overview in MLOps - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the Kubeflow Pipelines SDK client.

MLOps

from kfp import [1]

Drag options to blanks, or click blank then click option'

Adsl

Bclient

Ccomponents

Dcompiler

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'dsl' instead of 'client' which is for pipeline definition, not interaction.

Using 'compiler' which is for compiling pipelines, not for client connection.

The client module is used to interact with Kubeflow Pipelines programmatically.

2fill in blank

medium

Complete the code to create a Kubeflow Pipelines client connected to the default endpoint.

MLOps

kfp_client = client.[1]()

Drag options to blanks, or click blank then click option'

AClient

Brun_pipeline

Cconnect

Dstart

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'connect' which is not a valid method in the client module.

Using 'run_pipeline' which is a method to run pipelines, not to create the client.

The Client() constructor creates a client object to interact with Kubeflow Pipelines.

3fill in blank

hard

Fix the error in the pipeline function decorator to define a Kubeflow pipeline.

MLOps

@dsl.[1]

Drag options to blanks, or click blank then click option'

Arun

Bcomponent

Cpipeline

Dtask

Attempts:

3 left

💡 Hint

Common Mistakes

Using '@dsl.component' which defines a component, not a pipeline.

Using '@dsl.run' which is not a valid decorator.

The @dsl.pipeline decorator defines a Kubeflow pipeline function.

4fill in blank

hard

Fill both blanks to compile a Kubeflow pipeline to a YAML file.

MLOps

compiler.[1](pipeline_func=my_pipeline, package_path='[2]')

Drag options to blanks, or click blank then click option'

Acompile

Brun

Cpipeline.yaml

Dstart

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'run' or 'start' instead of 'compile' for the first blank.

Using a wrong filename or missing the '.yaml' extension.

The compile function compiles the pipeline function into a YAML file named 'pipeline.yaml'.

5fill in blank

hard

Fill all three blanks to run a compiled pipeline on Kubeflow Pipelines.

MLOps

kfp_client.[1](experiment_name='[2]', job_name='[3]', pipeline_package_path='pipeline.yaml')

Drag options to blanks, or click blank then click option'

Acreate_run_from_pipeline_package

Bmy_experiment

Cmy_run

Drun_pipeline

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'run_pipeline' which is not a method of the client.

Confusing experiment name and job name values.

The method create_run_from_pipeline_package runs the pipeline YAML file under the experiment 'my_experiment' with the job name 'my_run'.