First interaction with GenAI APIs - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding GenAI API Response Structure

When you send a prompt to a GenAI API, what is the typical structure of the response you receive?

A A JSON object containing the generated text, usage details, and metadata about the request

B A binary file that needs special decoding to extract the generated text

C An XML file with multiple nested tags describing the model internals

D A plain text string with only the generated text and no additional information

💡 Hint

Think about what extra information might be useful besides just the generated text.

GenAI APIs usually return a JSON object that includes the generated text, usage statistics like token counts, and metadata such as model version. This helps users understand the output and manage usage.

❓ Predict Output

intermediate

Output of a GenAI API Call with a Simple Prompt

What will be the output of this Python code snippet that calls a GenAI API with a prompt?

Prompt Engineering / GenAI

response = {
    'choices': [{'message': {'content': 'Hello, how can I help you today?'}}],
    'usage': {'prompt_tokens': 5, 'completion_tokens': 7, 'total_tokens': 12}
}
print(response['choices'][0]['message']['content'])

A KeyError

B response

C {'message': {'content': 'Hello, how can I help you today?'}}

D Hello, how can I help you today?

💡 Hint

Look at how the dictionary keys are accessed to get the message content.

The code accesses the first choice's message content and prints it, so the output is the string inside 'content'.

❓ Model Choice

advanced

Choosing the Right GenAI Model for Summarization

You want to use a GenAI API to summarize long articles quickly and accurately. Which model type should you choose?

A A small model optimized for image recognition

B A general-purpose language model trained only on chat conversations

C A large language model fine-tuned specifically for summarization tasks

D A model designed for speech-to-text transcription

💡 Hint

Think about which model is trained for text summarization.

Models fine-tuned for summarization understand how to condense text while keeping key points, making them best for this task.

❓ Hyperparameter

advanced

Effect of Temperature Parameter in GenAI API Calls

What happens if you increase the temperature parameter in a GenAI API call?

A The output text becomes shorter and more concise

B The generated text becomes more random and creative

C The API call fails with an error

D The model runs faster but with less accuracy

💡 Hint

Temperature controls randomness in text generation.

Higher temperature values make the model pick less likely words, increasing creativity and randomness in the output.

❓ Metrics

expert

Interpreting Token Usage Metrics from GenAI API

After a GenAI API call, you receive usage metrics: prompt_tokens=50, completion_tokens=150, total_tokens=200. What does this tell you?

A The model used 50 tokens from your prompt and generated 150 tokens in response, totaling 200 tokens processed

B The model generated 50 tokens and used 150 tokens from the prompt, totaling 200 tokens

C The total tokens count is incorrect because 50 + 150 does not equal 200

D The model only processed 150 tokens in total

💡 Hint

Think about what prompt and completion tokens represent.

Prompt tokens count the input text tokens, completion tokens count the generated output tokens, and total tokens is their sum.