Pytorch-Autoencoders-fashionMnist

Implementation of 3 flavours of autoencoders (vanilla AE, Convolutional-AE, Variational-AE) trained on fashionMnist dataset, courtsey of zalandoresearch.
A notebook is also provided which enables a simple and easy-to-use playfield to try out these different models.

Vanilla-Autoencoder Architecture

Simplified vanilla autoencoder architecture.

Convolutional-Autoencoder Architecture

Simplified convolutional-autoencoder architecture.

Variational-Autoencoder Architecture

Simplified variational-autoencoder architecture.

Training using a ubiquitous Learner

A common learner was devised to train all three models. This common Learner class enables training, testing, and progress visualization routine in a coherent manner.