Visualizing embeddings (t-SNE) in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the t-SNE class from sklearn.manifold.

NLP

from sklearn.manifold import [1]

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

ATSNE

BtSNE

Ct_SNE

DTsne

Attempts:

3 left

💡 Hint

Common Mistakes

Using lowercase letters like 'tSNE' or 'Tsne' causes import errors.

Adding underscores like 't_SNE' is incorrect.

The correct class name is TSNE with all uppercase letters except the last two.

2fill in blank

medium

Complete the code to create a t-SNE object with 2 components for visualization.

NLP

tsne = TSNE(n_components=[1], random_state=42)

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

A3

B5

C1

D2

Attempts:

3 left

💡 Hint

Common Mistakes

Setting n_components to 3 or more makes visualization harder.

Using 1 component loses too much information.

For visualization, we reduce embeddings to 2 dimensions, so n_components=2.

3fill in blank

hard

Fix the error in the code to fit and transform the embeddings using t-SNE.

NLP

embeddings_2d = tsne.[1](embeddings)

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

Afit

Bfit_predict

Cfit_transform

Dtransform

Attempts:

3 left

💡 Hint

Common Mistakes

Using only fit does not return transformed data.

Using transform without fitting causes errors.

t-SNE requires fit_transform to both fit the model and reduce dimensions in one step.

4fill in blank

hard

Fill both blanks to create a scatter plot of the 2D embeddings with labels.

NLP

plt.scatter(embeddings_2d[:, [1]], embeddings_2d[:, [2]], c=labels, cmap='viridis')
plt.title('t-SNE visualization')
plt.show()

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

A0

B1

C2

D3

Attempts:

3 left

💡 Hint

Common Mistakes

Using indices 2 or 3 causes index errors.

Swapping x and y axes can confuse interpretation.

We plot the first dimension on x-axis (index 0) and second on y-axis (index 1).

5fill in blank

hard

Fill the blanks to create a dictionary of word embeddings filtered by length and visualize with t-SNE.

NLP

filtered_embeddings = {word: embedding for word, embedding in all_embeddings.items() if len(word) [1] 5}
tsne = TSNE(n_components=[2], random_state=42)
embeddings_2d = tsne.fit_transform(list(filtered_embeddings.values()))

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

A<

B>

C2

D3

Attempts:

3 left

💡 Hint

Common Mistakes

Using > 5 filters out short words instead of keeping them.

Setting n_components to 3 makes visualization 3D, which is harder to plot.

We filter words shorter than 5 (< 5), set t-SNE components to 2.