wayground logo

Free Printable Worksheets

NEW

Font size

S
M
L
XL
Worksheets

RAG Quiz Answers

Total questions: 15

Worksheet time: 8mins

Name
Class
Date
1.

Which issue does RAG primarily help mitigate in large language models?

a)

Syntax errors in tokenization

b)

Hallucination and outdated knowledge

c)

Model overfitting on small datasets

d)

Slow training on GPUs

2.

In a RAG pipeline, which component locates relevant documents for a query?

a)

Generator component

b)

Retriever component

c)

Ranker component

d)

Summarizer component

3.

What kind of information is injected into the model to ground answers in RAG?

a)

User demographic data

b)

Factual domain information

c)

Random web samples

d)

Model hyperparameters

4.

In RAG, what are the model inputs typically enhanced with before generation?

a)

Retrieved text snippets

b)

Optimizer gradients

c)

Training labels

d)

System prompts only

5.

What does the acronym RAG stand for?

a)

Retrieve and Guess

b)

Retrieval and Generation

c)

Reasoning and Guidance

d)

Ranking and Grouping

6.

Embeddings used by retrievers are stored as what mathematical objects?

a)

Matrices of logits

b)

Vectors of numbers

c)

Trees of nodes

d)

Graphs of edges

7.

Which library is commonly used for efficient similarity search over embeddings?

a)

NumPy

b)

FAISS

c)

TensorFlow

d)

Pandas

8.

One benefit of RAG over purely parametric models is that external knowledge can be changed how?

a)

Data can be updated outside the model

b)

Weights are rewritten during inference

c)

Prompts auto-tune themselves

d)

Embeddings never require refresh

9.

In RAG, the generator’s role is best described as what?

a)

Selecting top-k documents

b)

Producing final response using context

c)

Computing embedding vectors

d)

Indexing documents for retrieval

10.

A typical retriever operation returns what from the index?

a)

All documents ever seen

b)

Top-1 nearest neighbor

c)

K most relevant documents

d)

Only exact keyword matches

11.

Which limitation of base LLMs motivates augmenting them with retrieval?

a)

No access to real-time data

b)

Inability to do arithmetic

c)

Poor sentiment detection

d)

Lack of tokenization rules

12.

RAG is especially helpful for which type of workload?

a)

Creative fiction writing tasks

b)

Domain-specific factual tasks

c)

General chit-chat conversations

d)

Adversarial prompt defense

13.

Effective retrieval depends most on which capability?

a)

Query understanding

b)

Grammar correction

c)

Topic generation

d)

Model pruning

14.

Similarity search in vector databases is based on what notion?

a)

Lexical overlap

b)

Semantic similarity

c)

Syntactic distance

d)

Statistical frequency

15.

Overall, adding retrieval to generation has what typical effect on answers?

a)

Increases narrative creativity

b)

Enhances factual accuracy

c)

Reduces token counts significantly

d)

Eliminates need for prompts