The transformer architecture's key mechanism, weighing relationships between all words at once, is called:
Q5easymcq
A model that handles more than one content type (e.g. text + image) is called:
Q6easymcq
The typical architecture behind image generation, starting from noise and iteratively refining it, is called a:
Q7easymcq
In the FM lifecycle, which stage comes immediately after 'Model selection'?
Q8easymcq
Most FMs (including Bedrock models) are billed based on:
Q9easymcq
MCP stands for:
Q10easymcq
Deliberately designing everything that goes into a model's context window (background, history, retrieved docs, tools) is called:
Q11hardmcq
Why does chunking matter specifically for RAG (previewed here, detailed in Lesson 7)?
Q12hardmulti
Which of the following are genuine agentic AI building blocks named in this lesson?
Select all that apply.
Q13hardmcq
Two documents about 'car maintenance' and 'automobile repair' end up close together in vector space despite sharing few exact words. This demonstrates:
Q14hardordering
Order the FM lifecycle stages correctly.
1.Fine-tuning
2.Deployment
3.Data selection
4.Evaluation
5.Pre-training
6.Model selection
Q15hardmatching
Match each term to its definition.
Token
Embedding
Transformer attention
Context engineering
Q16hardmcq
A team keeps sending very long prompts and documents on every call, driving up cost. What is the most directly relevant cost driver from this lesson?
Q17hardmcq
An agent needs to call an external weather API mid-conversation to answer a question. Which agentic building block does this require?
Q18hardmulti
Which of these correctly distinguish prompt engineering (Lesson 8) from context engineering (this lesson)?
Select all that apply.
Q19hardmcq
Which best describes 'memory management' as an agentic AI building block?
Q20hardmcq
Why is attention considered a major architectural advance over strictly sequential (word-by-word) processing?