Generative vs discriminative models in Prompt Engineering / GenAI - Interactive Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to define a generative model that learns the joint probability of data and labels.

Prompt Engineering / GenAI

model = GenerativeModel(p_x_y=[1])

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

AP(x, y)

BP(x|y)

CP(y|x)

DP(y)

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing conditional probability instead of joint probability.

A generative model learns the joint probability P(x, y) of data and labels.

2fill in blank

medium

Complete the code to define a discriminative model that learns the conditional probability of labels given data.

Prompt Engineering / GenAI

model = DiscriminativeModel(p_y_given_x=[1])

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

AP(x, y)

BP(x)

CP(x|y)

DP(y|x)

Attempts:

3 left

💡 Hint

Common Mistakes

Confusing joint probability with conditional probability.

A discriminative model learns the conditional probability P(y|x) of labels given data.

3fill in blank

hard

Fix the error in the code to train a discriminative model using maximum likelihood estimation.

Prompt Engineering / GenAI

loss = -sum(log([1](y|x)) for x, y in dataset)

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

AP(x|y)

BP(x, y)

CP(y|x)

DP(y)

Attempts:

3 left

💡 Hint

Common Mistakes

Using joint probability instead of conditional probability in loss.

Training a discriminative model uses the likelihood of labels given data, P(y|x).

4fill in blank

hard

Fill both blanks to complete the code that generates new data samples from a generative model.

Prompt Engineering / GenAI

samples = model.sample([1])
print('Generated samples:', [2])

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

Anum_samples

Bsamples

Cdata

Dlabels

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong variable names for sample count or output.

We specify how many samples to generate and then print the generated samples.

5fill in blank

hard

Fill all three blanks to complete the code that compares generative and discriminative models on accuracy.

Prompt Engineering / GenAI

gen_model = GenerativeModel()
disc_model = DiscriminativeModel()

accuracy_gen = gen_model.evaluate_accuracy([1])
accuracy_disc = disc_model.evaluate_accuracy([2])

print('Generative accuracy:', accuracy_gen)
print('Discriminative accuracy:', accuracy_disc)

better_model = [3] if accuracy_gen > accuracy_disc else disc_model

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

Atest_data

Cgen_model

Ddisc_model

Attempts:

3 left

💡 Hint

Common Mistakes

Using different data for evaluation or wrong model variable in comparison.

Both models are evaluated on the same test data. The better model is chosen by comparing accuracies.