Sentence-BERT for embeddings in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load the Sentence-BERT model for embeddings.

NLP

from sentence_transformers import SentenceTransformer
model = SentenceTransformer('[1]')

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

Abert-base-uncased

Broberta-base

Cgpt2

Dall-MiniLM-L6-v2

Attempts:

3 left

💡 Hint

Common Mistakes

Using a standard BERT model name instead of a Sentence-BERT model.

Using a language model like GPT-2 which is not for sentence embeddings.

The 'all-MiniLM-L6-v2' model is a popular Sentence-BERT model for generating sentence embeddings.

2fill in blank

medium

Complete the code to generate embeddings for a list of sentences.

NLP

sentences = ['I love machine learning.', 'Sentence embeddings are useful.']
embeddings = model.[1](sentences)

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

Aencode

Bpredict

Ctransform

Dfit

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'predict' which is for classification models.

Using 'fit' which is for training models.

The 'encode' method generates embeddings from sentences using the Sentence-BERT model.

3fill in blank

hard

Fix the error in the code to correctly compute cosine similarity between two embeddings.

NLP

from sklearn.metrics.pairwise import [1]
similarity = cosine_similarity([embeddings[0]], [embeddings[1]])[0][0]

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

Acosine_similarity

Beuclidean_distances

Cpairwise_distances

Dmanhattan_distances

Attempts:

3 left

💡 Hint

Common Mistakes

Using distance functions instead of similarity functions.

Using functions that return distances rather than similarity scores.

The 'cosine_similarity' function computes the cosine similarity between vectors, which is commonly used for embeddings.

4fill in blank

hard

Fill both blanks to create a dictionary comprehension mapping sentences to their embedding lengths.

NLP

lengths = {sentence: len(embedding) for sentence, embedding in zip(sentences, [1])}
filtered = {k: v for k, v in lengths.items() if v [2] 384}

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

Aembeddings

B>

C<

Dsentences

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'sentences' instead of 'embeddings' in the zip.

Using '<' instead of '>' for filtering.

We use 'embeddings' to get the vectors and filter those with length greater than 384.

5fill in blank

hard

Fill all three blanks to create a dictionary of sentences and their cosine similarity scores above 0.7 with the first sentence.

NLP

from sklearn.metrics.pairwise import [1]
scores = {sentence: [2]([embeddings[0]], [embedding])[0][0] for sentence, embedding in zip(sentences, embeddings)}
filtered_scores = {k: v for k, v in scores.items() if v [3] 0.7}

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

Acosine_similarity

Bcosine_distances

C>

D<

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'cosine_distances' which returns distances, not similarity.

Using '<' instead of '>' for filtering.

We import and use 'cosine_similarity' to compute scores and filter those greater than 0.7.