Generative AI with Diffusion Models
NVIDIA Deep Learning Institute
Last updated June 16, 2026
Generative AI with Diffusion Models is an eight-hour, hands-on course from NVIDIA's Deep Learning Institute that takes developers under the hood of the models behind modern text-to-image tools. Working in PyTorch, you build a U-Net, learn how the denoising diffusion process turns random noise into coherent images, compare different diffusion approaches, and use CLIP to generate images from text prompts. A practical, code-first look at how image-generation systems actually work.
What you'll learn
- Building a U-Net to generate images from pure noise
- How the denoising diffusion process improves image quality
- Comparing different diffusion model approaches
- Generating images from text prompts with CLIP
Frequently asked questions about Generative AI with Diffusion Models
Who is Generative AI with Diffusion Models for?
Developers comfortable with PyTorch and deep-learning basics who want hands-on diffusion-model experience.
Is Generative AI with Diffusion Models free?
No — Generative AI with Diffusion Models is a paid course.
What are the prerequisites for Generative AI with Diffusion Models?
A basic understanding of deep learning concepts and familiarity with a framework such as PyTorch, TensorFlow, or Keras (the course uses PyTorch).
Do you need to code for Generative AI with Diffusion Models?
Yes — Generative AI with Diffusion Models involves hands-on coding.
Why we suggest this course
If you want to understand image generation by building it rather than just using it, this goes deep into the real mechanics in PyTorch. For developers with some deep-learning grounding who want hands-on diffusion experience.