generative-models
generative-models copied to clipboard
shayneobrien
Annotated, understandable, and visually interpretable PyTorch implementations of: VAE, BIRVAE, NSGAN, MMGAN, WGAN, WGANGP, LSGAN, DRAGAN, BEGAN, RaGAN, InfoGAN, fGAN, FisherGAN
hello :) In the W-GAN model you mention in the comments the use of RMSProp optimizer, but you use in the code the Adam optimizer.
replace these lines - plt.scatter(m1s, m2s, c=labels) plt.legend([str(i) for i in set(labels)]) with the lines below - scatter = plt.scatter(m1s, m2s, c=labels) plt.legend(handles=scatter.legend_elements()[0], labels=set(labels))
In the notebook '04-wasserstein-gan-gradient-penalty.ipynb', the name of the model, 'WGANGP' and the trainer 'WGANGPTrainer' are incorrect concerning the corresponding src file (w_gp_gan.py). It should be changed into 'WGPGAN' and 'WGPGANTrainer'.
Metadata
Owner
Metadata
Annotated, understandable, and visually interpretable PyTorch implementations of: VAE, BIRVAE, NSGAN, MMGAN, WGAN, WGANGP, LSGAN, DRAGAN, BEGAN, RaGAN, InfoGAN, fGAN, FisherGAN