BERT tokenization (WordPiece) in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the BERT tokenizer from the transformers library.

NLP

from transformers import [1]

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

ATokenizer

BAutoTokenizer

CBertTokenizer

DBertModel

Attempts:

3 left

💡 Hint

Common Mistakes

Importing BertModel instead of BertTokenizer.

Using a generic Tokenizer class that does not exist.

Confusing AutoTokenizer with BertTokenizer.

The BertTokenizer class is used to tokenize text using BERT's WordPiece method.

2fill in blank

medium

Complete the code to load the pretrained BERT tokenizer for 'bert-base-uncased'.

NLP

tokenizer = BertTokenizer.[1]('bert-base-uncased')

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

Aload

Bfrom_pretrained

Cinit

Dtokenize

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'load' which is not a method of BertTokenizer.

Using 'init' which is a constructor, not for pretrained loading.

Using 'tokenize' which is for tokenizing text, not loading.

The from_pretrained method loads a pretrained tokenizer by name.

3fill in blank

hard

Fix the error in the code to tokenize the sentence using the tokenizer.

NLP

tokens = tokenizer.[1]('Hello, how are you?')

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

Atokenize

Bsplit

Cparse

Dencode

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'encode' which returns token IDs, not token strings.

Using 'split' which is a Python string method, not tokenizer method.

Using 'parse' which is not a tokenizer method.

The tokenize method splits the input text into WordPiece tokens.

4fill in blank

hard

Fill both blanks to create a dictionary of token ids and attention mask for the input text.

NLP

encoded_input = tokenizer('[1]', return_tensors='pt', padding=True, truncation=True)
input_ids = encoded_input['[2]']

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

AHello, how are you?

Binput_ids

Cattention_mask

Dtokens

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'tokens' instead of 'input_ids' as dictionary key.

Putting a variable name instead of a string in the first blank.

Using 'attention_mask' key when token IDs are needed.

The first blank is the input text string. The second blank is the key 'input_ids' to get token IDs from the encoded output.

5fill in blank

hard

Fill all three blanks to decode token ids back to the original text without special tokens.

NLP

decoded_text = tokenizer.[1](encoded_input['[2]'][0], skip_special_tokens=[3])

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

Adecode

Binput_ids

CTrue

DFalse

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'encode' instead of 'decode'.

Using 'attention_mask' key instead of 'input_ids'.

Setting skip_special_tokens to False, which keeps special tokens.

The decode method converts token IDs back to text. The key is 'input_ids'. Setting skip_special_tokens=True removes tokens like [CLS] and [SEP].