Evaluation Judges and Ground Truth GENERATIVE-AI-ENGINEER-ASSOCIATE Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   Which of the following evaluation judges requires ground truth labels to compute its score?

   • AExact match
   • BReference-free LLM-as-judge (no reference answer provided)
   • CPerplexity on the model's own generated text
   • DSelf-evaluation (model critiques its own output)
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A team is evaluating a chatbot that answers questions about company policies. They have a set of questions with approved answers from the HR department. They want to evaluate the chatbot's responses for accuracy. Which evaluation judge should they use?

   • ASelf-evaluation by the chatbot
   • BReference-free LLM-as-judge
   • CLLM-as-judge with reference answers
   • DExact match
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   A team is evaluating a model that generates answers to trivia questions. They have a set of questions with correct answers. They want to evaluate the model's accuracy, but they are concerned that the model might give correct answers in different wording. Which evaluation judge should they use?

   • AReference-free LLM-as-judge
   • BLLM-as-judge with reference answers
   • CExact match
   • DROUGE
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   In the context of evaluating a generative AI model, what is meant by 'ground truth'?

   • AThe model's training data used to fine-tune its parameters
   • BThe set of evaluation metrics chosen for a specific task
   • CThe set of correct or expected outputs used as a reference to evaluate model performance
   • DThe model's own predictions on a test set
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   Which of the following best describes the role of ground truth in model evaluation?

   • AIt defines the architecture of the model being evaluated
   • BIt determines the hyperparameters used during training
   • CIt specifies the evaluation metrics to be used in the test
   • DIt provides the expected outputs that serve as a benchmark for comparing model predictions
   Reveal Solution Discussion Select an answer first