Contextual compression in Prompt Engineering / GenAI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the main goal of contextual compression in AI?

Contextual compression is used in AI to reduce the size of input data while preserving important information. What is the primary goal of this technique?

A To remove all redundant information and keep only the contextually relevant parts

B To increase the size of data for better model training

C To randomly select parts of data without considering their importance

D To convert data into a completely different format without preserving meaning

💡 Hint

Think about how compression should keep what matters most for understanding.

Contextual compression aims to keep only the parts of data that are important for the current context, removing redundant or irrelevant information to make processing efficient.

❓ Predict Output

intermediate

Output of simple contextual compression code snippet

Given the following Python code that simulates a simple contextual compression by filtering words based on a context list, what is the output?

Prompt Engineering / GenAI

text = 'The quick brown fox jumps over the lazy dog'
context_words = {'quick', 'fox', 'dog'}
compressed = ' '.join([word for word in text.split() if word in context_words])
print(compressed)

A "The quick brown fox jumps over the lazy dog"

B "quick brown fox dog"

C "The brown jumps over the lazy"

D "quick fox dog"

💡 Hint

Look at which words are kept based on the context_words set.

The code keeps only words that are in the context_words set. Those are 'quick', 'fox', and 'dog', joined by spaces.

❓ Model Choice

advanced

Choosing the best model for contextual compression in text data

You want to compress large text data by keeping only contextually important information for a chatbot. Which model type is best suited for this task?

A A transformer-based model with attention mechanisms

B A simple linear regression model

C A k-means clustering model

D A decision tree classifier

💡 Hint

Think about models that understand context and relationships in sequences.

Transformer models with attention can focus on important parts of text, making them ideal for contextual compression.

❓ Metrics

advanced

Evaluating contextual compression quality

Which metric best measures how well a contextual compression method preserves important information while reducing data size?

A Training loss of an unrelated classification model

B Reconstruction accuracy or similarity score between original and decompressed data

C Number of layers in the compression model

D Total time taken to run the compression algorithm

💡 Hint

Consider how to check if important information is still present after compression.

Reconstruction accuracy or similarity scores show how much important information is preserved after compression and decompression.

🔧 Debug

expert

Debugging a contextual compression function that fails to filter correctly

This Python function is supposed to compress text by keeping only words in a given context set. What error or issue does it have?

Prompt Engineering / GenAI

def compress_text(text, context_set):
    return ' '.join(word for word in text if word in context_set)

result = compress_text('hello world', {'hello'})
print(result)

A It raises a TypeError because join expects a list but gets a string

B It raises a SyntaxError due to missing parentheses

C It checks characters instead of words, so it filters letters not words

D It returns an empty string because context_set is empty

💡 Hint

Look at what 'for word in text' iterates over in Python strings.

Iterating over a string yields characters, not words. So the function filters letters, not words, causing incorrect compression.