cddluv / pytorch-basic-gans

Simple Pytorch implementations of most used Generative Adversarial Network (GAN) varieties.

GANs

Simple Pytorch implementations of most used Generative Adversarial Network (GAN) varieties.

GPU or CPU

Support both GPU and CPU.

Dependencies

• Anaconda (Python 2.7)
• PyTorch 0.4.0

Table of Contents

• Vanilla GAN (GAN)
• Conditonal GAN (cGAN)
• Improved Conditonal GAN (Improved cGAN)
• Deep Convolutional GAN (DCGAN)
• Wasserstein GAN (WGAN)
• Improved Training of Wasserstein GAN (WGAN-GP)

Experiment Results

Vanilla GAN (GAN)

epoch 0	epoch 10	epoch 20	epoch 30	epoch 40
epoch 50	epoch 100	epoch 150	epoch 199	-
				-

---

Conditional GAN (cGAN)

epoch 0	epoch 10	epoch 20	epoch 30	epoch 40
epoch 50	epoch 100	epoch 150	epoch 199	-
				-

---

Improved Conditional GAN (Improved cGAN)

epoch 0	epoch 10	epoch 20	epoch 30	epoch 40
epoch 50	epoch 100	epoch 150	epoch 199	-
				-

Deep Convolutional GAN (DCGAN)

epoch 0	epoch 10	epoch 20	epoch 30	epoch 40
epoch 50	epoch 60	epoch 70	epoch 80	epoch 90

Wasserstein GAN (WGAN)

epoch 0	epoch 10	epoch 20	epoch 30	epoch 40
epoch 50	epoch 100	epoch 150	epoch 199	-
				-

Wasserstein GAN with Gradient Plenty (WGAN-GP)

epoch 0	epoch 10	epoch 20	epoch 30	epoch 40
epoch 50	epoch 100	epoch 150	epoch 199	-
				-

Acknowledgement

This project is going with the GAN Theory and Practice part of the Deep Learning Course: from Algorithm to Practice.

Contacts

If you have any question about the project, please feel free to contact with me.

E-mail: [email protected]