Extractive QA concept in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a pre-trained extractive QA model from Hugging Face.

NLP

from transformers import pipeline
qa_pipeline = pipeline('question-answering', model=[1])

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

A'distilbert-base-uncased-distilled-squad'

B'roberta-base'

C'gpt2'

D'bert-base-uncased'

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing a language model not fine-tuned for QA like 'gpt2'.

Using a base model without QA fine-tuning.

The 'distilbert-base-uncased-distilled-squad' model is a lightweight model fine-tuned for extractive question answering tasks.

2fill in blank

medium

Complete the code to provide context and question to the QA pipeline and get the answer.

NLP

result = qa_pipeline({'question': [1], 'context': context_text})

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

A'answer'

Bcontext_text

C'context'

D'What is extractive QA?'

Attempts:

3 left

💡 Hint

Common Mistakes

Passing the context as the question.

Using incorrect keys in the input dictionary.

The 'question' key must have the question string to ask the model.

3fill in blank

hard

Fix the error in accessing the answer text from the QA pipeline output.

NLP

answer_text = result[1]

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

A['answer']

B['text']

C['output']

D['result']

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'text' or 'result' keys which do not exist in the output.

Trying to access the answer as an attribute instead of a dictionary key.

The pipeline returns a dictionary with the key 'answer' containing the extracted answer string.

4fill in blank

hard

Fill both blanks to create a dictionary comprehension that maps each question to its answer using the QA pipeline.

NLP

answers = {q: qa_pipeline({'question': q, 'context': [1])[2] for q in questions}

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

Acontext_text

B['answer']

C['text']

Dcontext

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong variable names for context.

Accessing the answer with incorrect keys like ['text'].

The context is passed as 'context_text', and the answer is accessed with ['answer'] from the pipeline output.

5fill in blank

hard

Fill all three blanks to define a function that takes a question and context, runs the QA pipeline, and returns the answer text.

NLP

def answer_question([1], [2]):
    result = qa_pipeline({'question': [1], 'context': [2])
    return result[3]

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

Aquestion

Bcontext

C['answer']

Dresult

Attempts:

3 left

💡 Hint

Common Mistakes

Mismatching parameter names and dictionary keys.

Returning the whole result dictionary instead of just the answer.

The function parameters are 'question' and 'context'. The pipeline is called with these, and the answer is accessed with ['answer'].