Domain 2 of 5

Fundamentals of Generative AI

Tokens, embeddings and transformers, what a foundation model can and cannot be relied on to do, and the lifecycle that takes one from raw data to a deployed system with feedback coming back.

Concepts

~24%

Of the exam

Practice questions

Concepts in this domain

01Foundation models and how they workTokens, embeddings and transformers explained for somebody who has to make decisions about a model rather than build one.02What generative AI is good and bad atThe capabilities worth building on, the failure modes that are properties of the technology rather than bugs, and how to tell a suitable problem from an unsuitable one.03The foundation model lifecycleThe path from raw data to a deployed foundation model, and which stages you will ever be near when you are building on somebody else's.

Try a question from this domain

A finance team is forecasting the cost of a customer support assistant built on a hosted foundation model. They plan to budget per active user per month. What should they be told about how the cost actually behaves?

ACost is billed per user session, so the per user assumption is broadly correct.
BCost is fixed once the model is deployed, because inference does not consume metered resources.
CCost depends on the number of parameters in the model rather than on usage.
DCost is billed per token of input and output, so it scales with conversation length and document size rather than with user count.

16 questions on this domain.

One per page, with a worked explanation.

Start the set