A general introduction of GAN for the lab seminar. Here are some generated samples:

This repository is organized as the following:

pytotch (version 1.0) implementation. WGAN-gp

Basic datasets. For images, we have cifar-10 and mnist. For the music, we offer the 'tab' dataset. We represents the music in so called 'piano-roll' fromat, which is a binary and sparse tensor.

paper backup

Slides for related works.

Implementation of the aforementioned topics in the slides. All are written in Tenseoflow (< 1.0).

• single image genration:
  SN-GAN, WGAN-GP

• conditional generation:
  concatenation, ACGAN