Output format control in Prompt Engineering / GenAI - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output format of model predictions in JSON

Given a model that predicts labels for images, what is the output format of the following code snippet?

Prompt Engineering / GenAI

predictions = model.predict(images)
output = [{"image_id": i, "label": label} for i, label in enumerate(predictions)]
print(output)

A [{'label': 'cat'}, {'label': 'dog'}]

B [{'0': 'cat'}, {'1': 'dog'}]

C [('image_id', 0, 'label', 'cat'), ('image_id', 1, 'label', 'dog')]

D [{'image_id': 0, 'label': 'cat'}, {'image_id': 1, 'label': 'dog'}]

💡 Hint

Look at how the list comprehension builds dictionaries with keys 'image_id' and 'label'.

The code creates a list of dictionaries, each with keys 'image_id' and 'label', pairing the index with the predicted label.

❓ Metrics

intermediate

Interpreting accuracy output format

After training a classification model, the accuracy is printed as follows. What is the type and format of the accuracy output?

Prompt Engineering / GenAI

accuracy = model.evaluate(test_data)
print(f"Accuracy: {accuracy}")

A A dictionary with keys 'accuracy' and 'loss'

B A float number between 0 and 1 representing accuracy

C A list of accuracy values for each class

D A string describing accuracy percentage like '85%'

💡 Hint

Model evaluation usually returns numeric metrics as floats.

The evaluate method returns a float representing the overall accuracy between 0 and 1.

❓ Model Choice

advanced

Choosing output format for multi-label classification

You build a multi-label classifier that predicts multiple labels per input. Which output format best represents the model predictions?

A A string with comma-separated labels, e.g. ['cat,dog', 'bird']

B A single integer label per input, e.g. [0, 1]

C A list of label indices for each input, e.g. [[0,2], [1,3]]

D A dictionary with keys as labels and boolean values, e.g. [{'cat': True, 'dog': False}]

💡 Hint

Multi-label means multiple labels per input, so output must allow multiple labels.

A list of label indices per input clearly shows multiple labels predicted for each example.

🔧 Debug

advanced

Identifying output format error in prediction code

What error occurs when running this code that formats model output incorrectly?

Prompt Engineering / GenAI

preds = model.predict(data)
formatted = {i: preds[i] for i in preds}
print(formatted)

A KeyError because preds has no keys

B No error, prints correct dictionary

C TypeError because preds is a list and cannot be used as dict keys

D SyntaxError due to missing colon

💡 Hint

Check the type of preds and how it is used in the dict comprehension.

preds is a list, so iterating over it yields values, not keys, causing KeyError when indexing.

🧠 Conceptual

expert

Best practice for output format in AI model serving

When serving an AI model via an API, which output format is best for easy parsing and extensibility?

A JSON with keys for predictions, confidence scores, and metadata

B Plain text with predictions separated by commas

C Binary format with raw model output tensors

D CSV string with predictions only

💡 Hint

Consider formats that are human-readable and easy to extend with new fields.

JSON is widely supported, human-readable, and allows adding extra info like confidence and metadata.