Module 05 Knowledge Check
5 questions. Pass mark 4 out of 5. Answer every question before checking the answer key below, then retry after reading the feedback.
1. RAG evaluation must separate retrieval from generation because…
- A. It is conventional
- B. A wrong answer may mean the passage never arrived, or that it arrived and was misused
- C. Retrieval is slower
- D. They use different models
2. Recall@k measures…
- A. Answer quality
- B. Whether the answer-bearing document appears in the top k results
- C. Latency
- D. Citation accuracy
3. Citation accuracy asks…
- A. Whether a citation exists
- B. Whether the cited passage actually supports the specific claim
- C. Whether the link works
- D. How many citations there are
4. A rising refusal rate after a corpus update most likely indicates…
- A. A better model
- B. A retrieval or indexing regression
- C. Improved safety
- D. Lower cost
5. Testing with documents deliberately removed from the corpus checks…
- A. Speed
- B. Whether the system correctly says it does not know
- C. Cost
- D. Formatting
---
Answer Key and Explanations
Check these only after attempting every question.
1. B - A wrong answer may mean the passage never arrived, or that it arrived and was misused
Without the split you cannot tell whether to fix chunking or fix answering.
2. B - Whether the answer-bearing document appears in the top k results
It is the ceiling on grounded answering: if the passage is absent, nothing downstream can recover.
3. B - Whether the cited passage actually supports the specific claim
Present-but-unsupporting citations are the most misleading failure in RAG.
4. B - A retrieval or indexing regression
Refusal rate is a sensitive early indicator of retrieval breakage.
5. B - Whether the system correctly says it does not know
It is the cleanest way to measure invention under absence of evidence.
Practical Check
Apply this module to your own work: complete the module activity for *RAG Evaluation: Retrieval, Citation, and Answer Quality*, then write one sentence naming what your result shows and one naming what it does not.
Strong Answer Pattern
A strong answer names the task, the evidence used, the check performed, and the remaining limitation. It avoids "proved", "guaranteed", or "always" unless the evidence genuinely supports it.