Text splitters in Prompt Engineering / GenAI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding the purpose of text splitters

Why do we use text splitters when preparing data for language models?

A To break long text into smaller chunks that fit model input limits

B To translate text into different languages automatically

C To remove all punctuation from the text

D To convert text into numerical vectors directly

💡 Hint

Think about the input size limits of language models.

Text splitters help break long documents into smaller pieces so they can be processed by models that have input size limits.

❓ Predict Output

intermediate

Output of a simple character-based text splitter

What is the output of this code that splits text into chunks of 5 characters?

Prompt Engineering / GenAI

text = 'HelloWorld!'
chunks = [text[i:i+5] for i in range(0, len(text), 5)]
print(chunks)

A ['Hello', 'World', '']

B ['Hello', 'World!']

C ['Hello', 'World', '!']

D ['HelloW', 'orld!']

💡 Hint

Look at how slicing works with step size 5.

The code slices the string every 5 characters, so it produces ['Hello', 'World', '!'].

❓ Model Choice

advanced

Choosing the best text splitter for semantic search

You want to split a long document into chunks that keep sentences intact for semantic search. Which splitter is best?

A Random splitter that splits text at random positions

B Character splitter that cuts every 100 characters

C Word splitter that splits text every 10 words regardless of sentence

D Sentence splitter that splits text by sentence boundaries

💡 Hint

Think about preserving meaning in chunks.

Sentence splitters keep sentences whole, preserving meaning better for semantic search than arbitrary character or word splits.

❓ Hyperparameter

advanced

Effect of chunk overlap in text splitting

What is the main effect of increasing chunk overlap when splitting text?

A It increases redundancy but helps preserve context across chunks

B It reduces the total number of chunks but loses context between chunks

C It speeds up processing by reducing chunk size drastically

D It removes stopwords automatically from each chunk

💡 Hint

Overlap means repeating some text in adjacent chunks.

Increasing overlap repeats some text in chunks, which helps keep context but adds redundancy.

🔧 Debug

expert

Identifying error in recursive text splitter code

What error does this recursive text splitter code raise?

Prompt Engineering / GenAI

def recursive_split(text, max_len):
    if len(text) <= max_len:
        return [text]
    else:
        split_point = text.rfind('.', 0, max_len)
        if split_point == -1:
            split_point = max_len
        return [text[:split_point]] + recursive_split(text[split_point:], max_len)

chunks = recursive_split('This is a sentence. This is another sentence.', 10)
print(chunks)

A ValueError because rfind returns -1

B RecursionError due to infinite recursion

C TypeError because of wrong data type

D No error, outputs correct chunks

💡 Hint

Check what happens when split_point equals max_len and slicing text[split_point:] repeatedly.

If no period is found, split_point equals max_len, so text[split_point:] does not shorten enough, causing infinite recursion and RecursionError.