1. What is RAG ? Junior A. input/output filters that block unsafe, off-topic or policy-violating LLM content B. a parameterized prompt with slots for context and user input, versioned like code C. Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time D. fetching the k most similar chunks to the query to place in the prompt context
2. Which term means: "Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time"? Junior A. RAG B. Hallucination C. Top-k retrieval D. Guardrails
3. Which statement is correct? Junior A. RAG — a second-stage model that reorders retrieved candidates by relevance before they enter the prompt B. RAG — Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time C. RAG — fetching the k most similar chunks to the query to place in the prompt context D. RAG — a parameterized prompt with slots for context and user input, versioned like code
4. What is Embedding ? Junior A. a dense numeric vector representing text so that semantic similarity becomes vector distance B. fetching the k most similar chunks to the query to place in the prompt context C. Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time D. an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce
5. Which term means: "a dense numeric vector representing text so that semantic similarity becomes vector distance"? Junior A. Embedding B. Prompt template C. System prompt D. Vector database
6. Which statement is correct? Junior A. Embedding — splitting source documents into passages sized for embedding and retrieval B. Embedding — fetching the k most similar chunks to the query to place in the prompt context C. Embedding — a dense numeric vector representing text so that semantic similarity becomes vector distance D. Embedding — input/output filters that block unsafe, off-topic or policy-violating LLM content
7. What is Vector database ? Junior A. a second-stage model that reorders retrieved candidates by relevance before they enter the prompt B. the leading instruction that sets an LLM's role, constraints and tone for a session C. input/output filters that block unsafe, off-topic or policy-violating LLM content D. a store that indexes embeddings for fast approximate nearest-neighbor similarity search
8. Which term means: "a store that indexes embeddings for fast approximate nearest-neighbor similarity search"? Junior A. Top-k retrieval B. Vector database C. Prompt template D. Guardrails
9. Which statement is correct? Junior A. Vector database — the leading instruction that sets an LLM's role, constraints and tone for a session B. Vector database — a second-stage model that reorders retrieved candidates by relevance before they enter the prompt C. Vector database — a store that indexes embeddings for fast approximate nearest-neighbor similarity search D. Vector database — fetching the k most similar chunks to the query to place in the prompt context
10. What is Chunking ? Mid A. a parameterized prompt with slots for context and user input, versioned like code B. Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time C. fetching the k most similar chunks to the query to place in the prompt context D. splitting source documents into passages sized for embedding and retrieval
11. Which term means: "splitting source documents into passages sized for embedding and retrieval"? Mid A. Hallucination B. Guardrails C. System prompt D. Chunking
12. Which statement is correct? Mid A. Chunking — a dense numeric vector representing text so that semantic similarity becomes vector distance B. Chunking — splitting source documents into passages sized for embedding and retrieval C. Chunking — a second-stage model that reorders retrieved candidates by relevance before they enter the prompt D. Chunking — the leading instruction that sets an LLM's role, constraints and tone for a session
13. What is Top-k retrieval ? Mid A. fetching the k most similar chunks to the query to place in the prompt context B. a dense numeric vector representing text so that semantic similarity becomes vector distance C. a second-stage model that reorders retrieved candidates by relevance before they enter the prompt D. input/output filters that block unsafe, off-topic or policy-violating LLM content
14. Which term means: "fetching the k most similar chunks to the query to place in the prompt context"? Mid A. Top-k retrieval B. Reranking C. Prompt template D. Chunking
15. Which statement is correct? Mid A. Top-k retrieval — input/output filters that block unsafe, off-topic or policy-violating LLM content B. Top-k retrieval — an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce C. Top-k retrieval — fetching the k most similar chunks to the query to place in the prompt context D. Top-k retrieval — the leading instruction that sets an LLM's role, constraints and tone for a session
16. What is Prompt template ? Mid A. an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce B. a second-stage model that reorders retrieved candidates by relevance before they enter the prompt C. a store that indexes embeddings for fast approximate nearest-neighbor similarity search D. a parameterized prompt with slots for context and user input, versioned like code
17. Which term means: "a parameterized prompt with slots for context and user input, versioned like code"? Mid A. Prompt template B. Reranking C. Vector database D. Embedding
18. Which statement is correct? Mid A. Prompt template — fetching the k most similar chunks to the query to place in the prompt context B. Prompt template — input/output filters that block unsafe, off-topic or policy-violating LLM content C. Prompt template — a parameterized prompt with slots for context and user input, versioned like code D. Prompt template — a second-stage model that reorders retrieved candidates by relevance before they enter the prompt
19. What is System prompt ? Mid A. a dense numeric vector representing text so that semantic similarity becomes vector distance B. a second-stage model that reorders retrieved candidates by relevance before they enter the prompt C. the leading instruction that sets an LLM's role, constraints and tone for a session D. a parameterized prompt with slots for context and user input, versioned like code
20. Which term means: "the leading instruction that sets an LLM's role, constraints and tone for a session"? Mid A. Prompt template B. Hallucination C. System prompt D. Reranking
21. Which statement is correct? Mid A. System prompt — the leading instruction that sets an LLM's role, constraints and tone for a session B. System prompt — an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce C. System prompt — a store that indexes embeddings for fast approximate nearest-neighbor similarity search D. System prompt — fetching the k most similar chunks to the query to place in the prompt context
22. What is Hallucination ? Senior A. an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce B. Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time C. fetching the k most similar chunks to the query to place in the prompt context D. input/output filters that block unsafe, off-topic or policy-violating LLM content
23. Which term means: "an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce"? Senior A. Hallucination B. Top-k retrieval C. Embedding D. RAG
24. Which statement is correct? Senior A. Hallucination — a store that indexes embeddings for fast approximate nearest-neighbor similarity search B. Hallucination — input/output filters that block unsafe, off-topic or policy-violating LLM content C. Hallucination — Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time D. Hallucination — an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce
25. What is Reranking ? Senior A. a second-stage model that reorders retrieved candidates by relevance before they enter the prompt B. Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time C. an LLM producing fluent but factually wrong or unsupported output, which grounding aims to reduce D. fetching the k most similar chunks to the query to place in the prompt context
26. Which term means: "a second-stage model that reorders retrieved candidates by relevance before they enter the prompt"? Senior A. Reranking B. Vector database C. Chunking D. Guardrails
27. Which statement is correct? Senior A. Reranking — a store that indexes embeddings for fast approximate nearest-neighbor similarity search B. Reranking — a second-stage model that reorders retrieved candidates by relevance before they enter the prompt C. Reranking — a dense numeric vector representing text so that semantic similarity becomes vector distance D. Reranking — Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time
28. What is Guardrails ? Senior A. input/output filters that block unsafe, off-topic or policy-violating LLM content B. Retrieval-Augmented Generation: grounding an LLM's answer in documents fetched from a knowledge base at query time C. a store that indexes embeddings for fast approximate nearest-neighbor similarity search D. a parameterized prompt with slots for context and user input, versioned like code
29. Which term means: "input/output filters that block unsafe, off-topic or policy-violating LLM content"? Senior A. Reranking B. Hallucination C. Prompt template D. Guardrails
30. Which statement is correct? Senior A. Guardrails — the leading instruction that sets an LLM's role, constraints and tone for a session B. Guardrails — a store that indexes embeddings for fast approximate nearest-neighbor similarity search C. Guardrails — splitting source documents into passages sized for embedding and retrieval D. Guardrails — input/output filters that block unsafe, off-topic or policy-violating LLM content