This repository offers a collection of generative model architectures and training infrastructure specifically for medical imaging applications. It includes diffusion models with multiple noise schedulers (DDPM, DDIM, PNDM), VAE variants (Autoencoder-KL, VQ-VAE), and GAN components. The library provides MONAI-compatible inferer classes for training, sampling synthetic images, and computing metrics like FID and MS-SSIM. It also includes tutorials for anomaly detection, inpainting, and super-resolution using these generative approaches.