How-Diffusion-Models-Work

Notes from How Diffusion Models Work by DeepLearning.ai

Contents

Intuition

Sampling

• With Extra Noise

explorer_pC0437cXSo.mp4

Training

Context Embedding

Faster Sampling

---

Notes

Taught By Sharon Zhou

Noted by Atul

• Example used throughout the course: Generate 16X16 size sprites for video games.

Intuition

• Goal : Given a lot of sprite images, generate even more sprite images

• What does the network learn?

  • Fine details
  • General outline
  • Everything in between

• Noising Process (bob as ink drop analogy)

• Denoising Process (what should the NN think?)

  • If its' Bob the sprite, keep it as it is
  • If its likely to be Bob, suggest more details to be filled
  • If its just an outline of a sprite, suggest general details for likely sprite(bob/fred/...)
  • If its nothing, suggest outline of a sprite

• Give the NN input noise, whose pixels are obtained from Normal distribution, and get a completely new sprite !

Sampling

• Assume you have a trained NN
• At each denoising step, it predicts noise, and subtracts it to get a better image
• NOTE: At each denoising step, some random noise is added again to prevent "mode collapse"

Neural Network

• UNet Architecture
  • Input and output of same size
  • First used for image segmentation

• Takes a noisy image, embeds into small space by downsampling, and upsamples to predict noise

• Can take more info. in form of embeddings

  • Time: related to timestep, and noise level added
  • Context: guides generation process

• Checkout forward() in sampling notebook

Training

Learns the distribution of what is "not noise"

• Sample training image, timestep t, and noise, randomly
  • Timestep helps control level of noise
  • randomisation ensures a stable model
• Add noise to image
• Input this into NN, which predicts the noise
• Compute loss between actual and predicted noise
• Backprop and learn

Control

• Embeddings are vectors , for instance, strings represented as number vectors
• Given as input to NN along with training image
• Get associated with a training example, and its properties
• Uses: Generate funky mixtures by combining embeddings
• Context formats
  • Text
  • Categories, one hot encoded (Eg. hero, non-hero, spells ...)

Fast Sampling : DDIM

• DDPM is slow!
  • Multiple timesteps, and markovian nature
• Skips steps, making the process deterministic
• Lower quality than DDPM

Summary

Other applications : Music, Inpainting, Textual Inversion