Pipeline versioning and reproducibility in MLOps - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to save the pipeline version using MLflow.

MLOps

mlflow.set_experiment('my_experiment')
with mlflow.start_run():
    mlflow.log_param('pipeline_version', [1])

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

A'v1.0.0'

B1.0

Cversion

Dpipeline

Attempts:

3 left

💡 Hint

Common Mistakes

Using a number instead of a string for version.

Pipeline version should be a string like 'v1.0.0' to track reproducible runs.

2fill in blank

medium

Complete the code to load a specific pipeline version from the registry.

MLOps

from mlflow import pyfunc
model = pyfunc.load_model('models:/my_pipeline/[1]')

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

Aproduction

Bv1

C1

Dlatest

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'latest' which may not be stable.

Loading the 'production' stage ensures you get the stable pipeline version.

3fill in blank

hard

Fix the error in the code to ensure pipeline reproducibility by pinning dependencies.

MLOps

import subprocess
def save_requirements():
    with open('requirements.txt', 'w') as f:
        f.write([1])

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

Aos.system('pip freeze > requirements.txt')

Bpip freeze > requirements.txt

Csubprocess.check_output(['pip', 'freeze'], text=True)

Df.write('pip freeze')

Attempts:

3 left

💡 Hint

Common Mistakes

Writing the string 'pip freeze' instead of the command output.

Using subprocess.check_output captures the output of 'pip freeze' as a string and writes it to the file correctly.

4fill in blank

hard

Fill both blanks to create a reproducible pipeline run with a fixed random seed.

MLOps

import random
import numpy as np
random.[1](42)
np.random.[2](42)

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

Aseed

Brandom

Crandint

Dchoice

Attempts:

3 left

💡 Hint

Common Mistakes

Using methods that generate random numbers instead of setting the seed.

Setting the seed for both random and numpy.random ensures reproducible randomness.

5fill in blank

hard

Fill all three blanks to log pipeline metadata with MLflow for reproducibility.

MLOps

mlflow.log_param('[1]', 'v1.0.0')
mlflow.log_param('[2]', 'my_experiment')
mlflow.log_param('[3]', '2024-06-01')

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

Apipeline_version

Brun_id

Cstart_date

Dexperiment_name

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing up parameter names or using incorrect keys.

Logging 'pipeline_version', 'experiment_name', and 'start_date' helps track pipeline runs clearly.