This repository contains implementations of standard autoencoder (AE) and variational autoencoder (VAE) architectures along with training procedures and evaluations.

Implemented a standard autoencoder architecture comprising encoder and decoder components.

Developed a variational autoencoder architecture, incorporating a reparameterization trick for the latent space.

Trained both models on the MNIST dataset, experimenting with various hyperparameters to optimize performance.

Defined the Generator and Discriminator networks for the GAN model.

Configured the training process, including setting up data loaders, initializing optimizers, and enabling GPU training.

Evaluated the GAN model by monitoring loss and other relevant metrics for both the Generator and Discriminator.

Generated new data samples using the trained Generator and compared their quality with the original dataset.