Generative AI & Diffusion Models
AI · Advanced
Definition
Generative models create new data; diffusion models learn to remove noise step by step to produce images.
Behind modern image, audio and video generators.
Diagram
Flowchart
Example: Noise schedule
x = 1.0
for beta in [0.1, 0.2]:
x *= (1 - beta) ** 0.5
print(round(x, 3))Output: 0.849
Key points
- GANs use generator vs discriminator
- Text conditions guide generation
Interview questions
GAN vs diffusion?
GANs generate in one step via adversarial training; diffusion denoises iteratively and trains more stably.