#sampling-speed

1 question from Diffusion Models

Timed test

1 question · 1 min · score and explanations at the end

Q 1 / 1Score 0 / 0

Shuffle questions

Question 1: Single choice

easyAI-checkedDiffusion Models

Compared with GANs, what is the main practical drawback of diffusion models for image generation?

Sampling is slow: it takes many denoising steps through the network1
Their samples are less diverse, because they suffer from mode collapse2
Training is unstable, because a generator and a discriminator compete3
They cannot be conditioned on a class label or a text prompt4

Keys: 1–9 select · Enter check / next