179 questions
No questions match those filters.
← All questions
EvaluationEasy· Evaluation· LLM Judge· Human-in-the-loop
You don’t have ground-truth relevant documents for this...
This is one of the questions in the full AI/ML interview bank. Pro unlocks all 1789 questions; Premium includes the same bank plus the highest daily Practice limit.
See plansIn the absence of ground-truth data, you must generate a proxy for relevance. This is typically achieved by employing a human labeling panel or using an LLM to act as a judge. After establishing these labels, the standard classification metrics for retrieval remain applicable, treating the judge’s output as the ground truth for the purpose of calculating precision and recall.