Abstractive summarization in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a pretrained abstractive summarization model using Hugging Face Transformers.

NLP

from transformers import pipeline
summarizer = pipeline('[1]')

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

Asummarization

Btext-classification

Ctranslation

Dquestion-answering

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'text-classification' instead of 'summarization'.

Confusing summarization with translation.

The pipeline task for abstractive summarization is 'summarization'.

2fill in blank

medium

Complete the code to generate a summary from the input text using the summarizer pipeline.

NLP

text = "Machine learning helps computers learn from data."
summary = summarizer([1], max_length=30, min_length=5, do_sample=False)

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

Atext

Btext.split()

Ctext.lower()

Dtext.upper()

Attempts:

3 left

💡 Hint

Common Mistakes

Passing a list of words instead of a string.

Passing a modified version of the text that changes meaning.

The summarizer expects a string input, so pass the variable 'text' directly.

3fill in blank

hard

Fix the error in the code to correctly decode the generated summary tokens.

NLP

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

model_name = 'facebook/bart-large-cnn'
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

inputs = tokenizer('Deep learning is a subset of machine learning.', return_tensors='pt')
summary_ids = model.generate(inputs['input_ids'], max_length=20, num_beams=4, early_stopping=True)
summary = tokenizer.[1](summary_ids[0], skip_special_tokens=True)

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

Aconvert_tokens_to_ids

Bdecode

Cencode

Dtokenize

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'encode' which converts text to tokens.

Using 'tokenize' which splits text into tokens but does not decode.

To convert token IDs back to text, use the 'decode' method.

4fill in blank

hard

Fill both blanks to create a dictionary comprehension that maps each word to its length only if the length is greater than 3.

NLP

words = ['data', 'AI', 'model', 'run']
lengths = { [1] : len([2]) for word in words if len(word) > 3 }

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

Aword

Bwords

Ddata

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'words' instead of 'word' inside the comprehension.

Using a fixed string like 'data' instead of the loop variable.

The key should be 'word' and the length is calculated on 'word'.

5fill in blank

hard

Fill all three blanks to create a dictionary comprehension that maps each word in 'texts' to its uppercase form only if the word length is less than 5.

NLP

texts = ['AI', 'learn', 'code', 'data']
result = { [1] : [2] for [3] in texts if len([3]) < 5 }

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

Aword.upper()

Bword

Dtext

Attempts:

3 left

💡 Hint

Common Mistakes

Using different variable names inconsistently.

Not applying the upper() method to the value.

The key is 'word', the value is 'word.upper()', and the loop variable is 'word'.