Generative Models

Implementation of famous models for self-study

Autoencoders

• feedforward (MNIST)
• convolutional (MNIST)
• residual (MNIST)
• variational (MNIST, CIFAR)
• beta-variational (MNIST)
• convolutional with perceptual loss (MNIST)

GANs

• GAN (MNIST)
• DCGAN (MNIST)

References

[https://github.com/eriklindernoren/PyTorch-GAN] and [https://github.com/lyeoni/pytorch-mnist-GAN] were very helpful in understanding how the mathematics of GANs translated to code; as well as providing good initial hyperparameters and architectures