soumith / ganhacks

starter from "How to Train a GAN?" at NIPS2016

soumith/ganhacks Issues

Generator and discriminator schedule
Updated 6 months ago
GAN training example
Updated 7 months ago
Activation functions for training and testing phases
Updated a year ago1
MAE increasing when training!
Updated 2 years ago
Why Discriminator Loss 0 is a failure mode ?
Updated 2 years ago15
Generator ADAM X Discriminator SGD
Updated 2 years ago3
CycleGAN
Updated 2 years ago
Tricks to sharpen GAN's synthesis
Updated 2 years ago
Dose it need to transfer pretrained discriminator when transfer learning for GAN?
Updated 3 years ago1
Need help in understanding #1 and #16
Updated 4 years ago2
Reference for trick#16
Updated 4 years ago
Unexplainable GAN losses, need help
Updated 4 years ago3
Why "Construct different mini-batches for real and fake, i.e. each mini-batch needs to contain only all real images or all generated images"?
Updated 4 years ago3
> @KrnTneja : Thanks for your tricks. Could you provide any code to do it? I also meet the problem of loss D goes to zero
Closed 4 years ago1
Discriminator loss of WGAN-gp drop sharply to zero
Updated 4 years ago
Hacks relevancy in 2020
Updated 4 years ago1
balance loss maybe is good idea?
Updated 4 years ago
PixelShuffle, ConvTranspose2d + stride will get bad checkerboard
Updated 4 years ago2
Understand #10
Updated 4 years ago
Generator try to maximize, Discriminator try to minimize
Closed 4 years ago1
Paper / talk associated?
Closed 4 years ago
Validate a GAN
Updated 5 years ago5
for GAN ，why my D loss is increse，and G loss decrease to 0 at the begining
Updated 5 years ago10
Discriminator with accuracy 1, generator fools it perfectly. What's going on?
Updated 5 years ago5
Evaluate the Discriminator module over different set of images
Updated 5 years ago
is the training normal? how is the right train loss curve?
Updated 5 years ago
How to handle failure mode
Updated 5 years ago3
Question: Any interest in putting together a premium version for sale on SugarKubes?
Closed 5 years ago3
How would one go about training the network with one channel images?
Updated 5 years ago2
label smoothing one-sided?
Updated 5 years ago
Standard G & D Loss Values & Trends?
Updated 5 years ago5
Regarding sampling image in batches
Updated 5 years ago
not working in trick 13: adding gaussian noise to every layer of generator
Updated 6 years ago1
Using generator and discriminator loss to evaluate GANs.
Updated 6 years ago
Regarding 13: Add noise to inputs...
Updated 6 years ago1
G is always really smaller than D
Closed 6 years ago
There is no source that has empirically demonstrated a positive effect of replay buffer on GAN training
Closed 6 years ago1
Discriminator accuracy: 0.5 on average, but collapsed into 0 for negative samples and 1 for positive samples
Closed 6 years ago2
GAN - Multiple iterations for generator training
Closed 6 years ago
Label smoothing
Updated 6 years ago
Saturation of G_loss and D_loss
Updated 6 years ago
How hard it is to make a GAN experiment reproducible?
Updated 6 years ago
Alternating training vs. DCGAN?
Updated 6 years ago
Feasible loss simplification?
Updated 6 years ago2
Trick 11: How to schedule thresholds A and B?
Updated 6 years ago2
about generator loss
Updated 6 years ago
Replace BatchNorm with WeightNorm
Updated 6 years ago1
Update tips for WGANs
Updated 6 years ago3
measuring quality
Updated 6 years ago
Sampling from a sphere instead of a cube
Updated 6 years ago3