Why reproducibility builds trust in ML in MLOps - Test Your Understanding

Practice - 5 Tasks

1fill in blank

easy

Complete the code to save the model with reproducible results.

MLOps

model.save('model_[1].h5')

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

Av1

Btemp

Cfinal

Dtest

Attempts:

3 left

💡 Hint

Common Mistakes

Using generic names like 'temp' or 'test' which don't track versions.

Using a version like 'v1' helps keep track of reproducible model builds.

2fill in blank

medium

Complete the code to set a fixed random seed for reproducibility.

MLOps

import numpy as np
np.random.seed([1])

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

ANone

Brandom

C42

D-1

Attempts:

3 left

💡 Hint

Common Mistakes

Using None or 'random' which do not fix the seed.

Setting a fixed seed like 42 ensures the random numbers are the same every run.

3fill in blank

hard

Fix the error in the code to log parameters for reproducibility.

MLOps

mlflow.log_param('learning_rate', [1])

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

A0.01

B'0.01'

Clearning_rate

DNone

Attempts:

3 left

💡 Hint

Common Mistakes

Passing the learning rate as a string or variable name instead of a number.

Logging the parameter as a number (0.01) is correct for reproducibility tracking.

4fill in blank

hard

Fill both blanks to create a reproducible data split.

MLOps

train_data, test_data = train_test_split(data, test_size=[1], random_state=[2])

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

A0.2

B0.5

C42

DNone

Attempts:

3 left

💡 Hint

Common Mistakes

Using None for random_state or incorrect test sizes.

Using test_size=0.2 and random_state=42 ensures the split is reproducible.

5fill in blank

hard

Fill all three blanks to log model metrics reproducibly.

MLOps

mlflow.log_metric('[1]', [2], step=[3])

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

Aaccuracy

B0.95

C1

Dloss

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing metric names and values or missing the step parameter.

Logging 'accuracy' with value 0.95 at step 1 helps track reproducible metrics.