/Conditional-Diffusion-Model

Code for implemeting a conditional DDPM trained on CIFAR10

Primary LanguagePython

Conditional-Diffusion-Model

Code for implemeting a conditional DDPM trained on CIFAR10

Overview

This code modifies a basic denosing diffusion probailistic model (DDPM) to create a conditional DDPM trained on the CIFAR-10 dataset which is able to generate synthetic images from a combination of different input category labels

Project Structure

  1. cifar_ddpm.py - implements and trains conditional ddpm

  2. image_generator.py - produces generative images according to input conditions using trained model

Author

Louis Chapo-Saunders