CS 7643 Deep Learning | Quiz 4 Practice | Fall 2026
GEORGIA TECH
CS 7643 - Deep Learning
Quiz 4 Practice
30 Original Questions, Answers & Detailed Rationales
Fall 2026 Alignment | Generative Models
FVBN / Autoregressive Models - VAEs - Diffusion - GANs
Scope note: Georgia Tech's current Fall 2026 schedule identifies Quiz 4 as an in-class quiz on Assignment 4
materials (Generative Models), scheduled for November 18, 2026. This document is independently written
practice material and is not an official or leaked Georgia Tech assessment.
Alignment date: September 7, 2026
Original Practice Material - September 2026
, CS 7643 Deep Learning | Quiz 4 Practice | Fall 2026
September 2026 Research Alignment
Current Fall 2026 evidence. The live Georgia Tech schedule says Quiz 4 is on Assignment 4 materials (Generative
Models). Before the quiz, the Fall sequence includes Generative Models: FVBN and VAE, Diffusion Models, and
GANs and Self-Supervised Learning. HW4 is due two days before Quiz 4.
Boundary evidence from the latest completed semester. Spring 2026 explicitly defined Quiz 4 as Assignment 4
Generative Models material, with the generative-model unit centered on variational autoencoders and diffusion. The
closest public Fall-format lecture sequence from Georgia Tech also includes FVBN/autoregressive models and
GANs. Therefore this practice set weights FVBN/autoregressive models, VAEs, diffusion, and GANs most heavily.
Important timing limitation. As of September 7, 2026, the Fall 2026 generative-model lectures have not yet
occurred. The current schedule establishes the assessment label and topic sequence, while detailed concept
alignment uses the most recent public Georgia Tech lecture materials from Fall 2025 and Spring 2026. Later
computer vision, 3D vision, and vision-language lectures are not treated as core Quiz 4 material because the
schedule specifically labels Quiz 4 as Generative Models.
Topic Priority Questions
Generative model taxonomy and
High Q1-Q2
density estimation
FVBN / autoregressive factorization Very High Q3-Q6
Variational autoencoders and ELBO Very High Q7-Q15
Diffusion models Very High Q16-Q24
GANs Very High Q25-Q29
Integrated model comparison High Q30
Practice format: Conceptual multiple choice, short answer, likelihood calculations, ELBO/KL calculations, diffusion
computations, and GAN objective reasoning. The emphasis is on concepts, computations, theory, and intuition,
matching Georgia Tech's published description of its in-class quizzes.
Original Practice Material - September 2026
GEORGIA TECH
CS 7643 - Deep Learning
Quiz 4 Practice
30 Original Questions, Answers & Detailed Rationales
Fall 2026 Alignment | Generative Models
FVBN / Autoregressive Models - VAEs - Diffusion - GANs
Scope note: Georgia Tech's current Fall 2026 schedule identifies Quiz 4 as an in-class quiz on Assignment 4
materials (Generative Models), scheduled for November 18, 2026. This document is independently written
practice material and is not an official or leaked Georgia Tech assessment.
Alignment date: September 7, 2026
Original Practice Material - September 2026
, CS 7643 Deep Learning | Quiz 4 Practice | Fall 2026
September 2026 Research Alignment
Current Fall 2026 evidence. The live Georgia Tech schedule says Quiz 4 is on Assignment 4 materials (Generative
Models). Before the quiz, the Fall sequence includes Generative Models: FVBN and VAE, Diffusion Models, and
GANs and Self-Supervised Learning. HW4 is due two days before Quiz 4.
Boundary evidence from the latest completed semester. Spring 2026 explicitly defined Quiz 4 as Assignment 4
Generative Models material, with the generative-model unit centered on variational autoencoders and diffusion. The
closest public Fall-format lecture sequence from Georgia Tech also includes FVBN/autoregressive models and
GANs. Therefore this practice set weights FVBN/autoregressive models, VAEs, diffusion, and GANs most heavily.
Important timing limitation. As of September 7, 2026, the Fall 2026 generative-model lectures have not yet
occurred. The current schedule establishes the assessment label and topic sequence, while detailed concept
alignment uses the most recent public Georgia Tech lecture materials from Fall 2025 and Spring 2026. Later
computer vision, 3D vision, and vision-language lectures are not treated as core Quiz 4 material because the
schedule specifically labels Quiz 4 as Generative Models.
Topic Priority Questions
Generative model taxonomy and
High Q1-Q2
density estimation
FVBN / autoregressive factorization Very High Q3-Q6
Variational autoencoders and ELBO Very High Q7-Q15
Diffusion models Very High Q16-Q24
GANs Very High Q25-Q29
Integrated model comparison High Q30
Practice format: Conceptual multiple choice, short answer, likelihood calculations, ELBO/KL calculations, diffusion
computations, and GAN objective reasoning. The emphasis is on concepts, computations, theory, and intuition,
matching Georgia Tech's published description of its in-class quizzes.
Original Practice Material - September 2026
