Databricks GenAI Engineer Agent Development Quiz

1. What is a key benefit of using Databricks Workspace for generative AI development?

Integrated notebooks, compute, and MLflow tracking in one platform

Lower cost than cloud-native alternatives

Built-in hardware acceleration for CPUs only

Automatic model quantization

2. Which Databricks feature helps track and compare different agent configurations and performance metrics?

Databricks SQL

MLflow Experiments

Databricks Repos

Databricks Clusters

3. What is the benefit of using function calling in LLM-based agents?

It allows the model to directly invoke external tools and APIs in a structured way

It reduces the size of the model

It eliminates the need for embeddings

It increases inference speed by 10x

4. In production agent systems, what role does Databricks Jobs play?

Scheduling and orchestrating agent workflows and batch inference

Tokenizing input text

Compressing model weights

Limiting API rate calls

5. Which approach allows an agent to reason step-by-step before generating a final answer?

Greedy decoding

Temperature sampling

Chain-of-thought reasoning

Token batching

6. What is the main challenge addressed by using a context window management strategy in long-running agents?

GPU memory overflow from accumulating conversation history

Network bandwidth limitations

Database connection timeouts

File system I/O bottlenecks

7. In Databricks MLflow, what is the primary purpose of model registry?

To track hyperparameter tuning only

To manage model versions, stage transitions, and production deployment

To encrypt model weights

To reduce inference latency

8. Which data structure is commonly used to store and retrieve embeddings efficiently in RAG systems?

Hash table

Vector database or index

B-tree

Linked list

9. What is the purpose of a system prompt in LLM-based agents?

To initialize GPU memory

To define the agent's role, behavior, and constraints

To compress model parameters

To schedule batch jobs

10. In the context of agent evaluation, what metric measures how often the agent produces factually correct responses?

Latency

Accuracy

Throughput

Precision

11. Which Databricks feature enables vector search and similarity-based retrieval for RAG applications?

Databricks Vector Search

Databricks SQL Analytics

Databricks Feature Store

Databricks Model Serving

12. Which parameter controls the randomness or creativity of an LLM's output?

Top-K

Temperature

Max tokens

Frequency penalty

13. What technique involves breaking down complex tasks into smaller, sequential steps for an LLM to solve?

Temperature scaling

Chain-of-thought prompting

Attention masking

Beam search

14. In agent design, what is the purpose of a tool or action module?

To tokenize input text

To enable the agent to interact with external APIs and systems

To compress model weights

To increase inference speed

15. Which Databricks service is used to deploy and serve generative AI models at scale?

Databricks Workflows

Databricks Model Serving

Databricks SQL

Databricks Jobs

16. What is the role of an embedding model in RAG systems?

Generates text responses

Converts text into numerical vectors for semantic search

Fine-tunes the base LLM

Manages API rate limiting

17. In the context of LLM fine-tuning, what does LoRA stand for?

Low-Rank Adaptation

Latency-Optimized Retrieval Architecture

Long-Range Attention

Linear Optimization Resource Allocation

18. Which of the following is a common framework for building autonomous agents with LLMs?

LangChain

Hugging Face Transformers

TensorFlow Lite

PyTorch Lightning

19. What is the primary advantage of using retrieval-augmented generation (RAG) in agent applications?

Reduces model inference latency

Grounds responses in external knowledge and reduces hallucinations

Eliminates the need for fine-tuning

Decreases model parameter count

20. In prompt engineering, what technique involves providing examples of desired input-output pairs to guide model behavior?

Zero-shot prompting

Few-shot prompting