This repository provides TensorFlow 2 implementations of four generative adversarial network variants: DCGAN (Deep Convolutional GAN), LSGAN (Least Squares GAN), WGAN-GP (Wasserstein GAN with gradient penalty), and DRAGAN (Deep Regret Analytic GAN). These models are trained on datasets like Fashion-MNIST, CelebA, and anime faces to generate synthetic images. The project serves as a reference implementation for researchers and developers looking to experiment with different GAN training objectives and architectures.