Why CI/CD differs for ML vs software in MLOps - Test Your Understanding

Practice - 5 Tasks

1fill in blank

easy

Complete the code to define the first step in ML CI/CD pipeline.

MLOps

def [1]():
    print("Start data validation")

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

Avalidate_data

Btrain_model

Cdeploy_model

Dtest_software

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing training or deployment as first step.

The first step in ML CI/CD is usually data validation to ensure quality input.

2fill in blank

medium

Complete the code to specify the artifact type unique to ML pipelines.

MLOps

artifact = '[1]'

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

Amodel_weights

Bbinary_executable

Cdocker_image

Dsource_code

Attempts:

3 left

💡 Hint

Common Mistakes

Confusing software binaries with ML model files.

ML pipelines produce model weights as artifacts, unlike software binaries.

3fill in blank

hard

Fix the error in the ML pipeline step that triggers retraining.

MLOps

if data_changed or [1]:
    retrain_model()

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

Atests_passed

Bmodel_deployed

Ccode_changed

Ddeployment_failed

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing deployment or test status instead of code changes.

Retraining triggers when data or code changes, not on deployment status.

4fill in blank

hard

Fill both blanks to define the ML pipeline stages for testing and deployment.

MLOps

def pipeline():
    if [1]():
        [2]()

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

Arun_model_tests

Bdeploy_model

Crun_software_tests

Dbuild_docker_image

Attempts:

3 left

💡 Hint

Common Mistakes

Using software test functions or building images instead of deploying.

ML pipelines test the model and then deploy it if tests pass.

5fill in blank

hard

Fill all three blanks to create a dictionary tracking ML model metrics after training.

MLOps

metrics = {
    '[1]': accuracy,
    '[2]': precision,
    '[3]': recall
}

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

Aaccuracy

Bprecision

Crecall

Dloss

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing metric names or using loss as a key incorrectly.

Common ML metrics include accuracy, precision, and recall to evaluate model quality.