DeepLearning.AIIntermediate
How Diffusion Models Work
Build generative diffusion models from scratch in PyTorch. Sampling, guidance, and U-Net architectures.
Duration1.5 hours
PriceFree
FormatInteractive
LevelIntermediate
What you'll learn
- The mathematical intuition behind noise addition and removal
- Build and train a U-Net architecture in PyTorch
- Implement classifier-free guidance for conditioned generation
- Accelerate sampling steps for faster image synthesis
About this course
A free DeepLearning.AI short course taught by Sharon Zhou. You will code the forward and reverse diffusion processes, train a U-Net on image generation, and implement classifier-free guidance. High signal, hands-on, and accessible via deeplearning.ai. We are not affiliated with DeepLearning.AI.
Prerequisites
Python and basic familiarity with PyTorch neural networks.