Multilingual models in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a multilingual transformer model using Hugging Face Transformers.

NLP

from transformers import AutoModelForSequenceClassification
model = AutoModelForSequenceClassification.from_pretrained('[1]')

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

Agpt2

Bbert-base-uncased

Cxlm-roberta-base

Ddistilbert-base-uncased

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing a monolingual model like bert-base-uncased.

The xlm-roberta-base model is a popular multilingual transformer model that supports many languages.

2fill in blank

medium

Complete the code to tokenize input text for a multilingual model.

NLP

from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained('[1]')
inputs = tokenizer('Hello world', return_tensors='pt')

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

Abert-base-cased

Bxlm-roberta-base

Croberta-base

Dgpt2

Attempts:

3 left

💡 Hint

Common Mistakes

Using a tokenizer for a different model causes errors.

The tokenizer must match the multilingual model, so xlm-roberta-base is correct.

3fill in blank

hard

Fix the error in the code to correctly predict with a multilingual model.

NLP

outputs = model(**[1])
predictions = outputs.logits.argmax(dim=1)

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

Ainputs

Btokenizer

Ctext

Dlabels

Attempts:

3 left

💡 Hint

Common Mistakes

Passing raw text or tokenizer object instead of tokenized inputs.

The model expects tokenized inputs, so passing inputs is correct.

4fill in blank

hard

Fill both blanks to create a dictionary comprehension that maps languages to their ISO codes.

NLP

language_codes = {lang[1]: lang[2] for lang in ['eng', 'fra', 'spa', 'deu']}

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

A[:]

B[::]

C[:3]

D[3:]

Attempts:

3 left

💡 Hint

Common Mistakes

Using incorrect slice like [3:] which gets characters after index 3.

Using [:3] slices the first three characters, which are the ISO codes.

5fill in blank

hard

Fill all three blanks to create a dictionary of language names and their token counts from a multilingual tokenizer.

NLP

token_counts = [1]([2]: len(tokenizer.tokenize([2])) for [3] in ['Hello', 'Bonjour', 'Hola'])

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

Adict

Bword

Cw

Dtext

Attempts:

3 left

💡 Hint

Common Mistakes

Using different variable names inconsistently.

The dictionary is created with dict, and the loop variable is w used twice.