WGAN-tensorflow
WGAN-tensorflow copied to clipboard
Zardinality
a tensorflow implementation of WGAN
`c_loss = tf.reduce_mean(fake_logit - true_logit)` maybe wrong. It should be `c_loss = tf.reduce_mean(fake_logit) - tf.reduce_mean(true_logit)`
when i use your lib,somentimes i will get a negative value.Why do i have a problem?
It seems that in the original paper the output of the discriminator(d_loss) is an estimate of EM distance, so should it be positive? The curve of d_loss shows it tends...
Hi, Thank you for sharing your code. No matter what model I use, I always get blurred Mnist images, should i Increase the number of iterations or do something else....
Hi, please share a WGAN code to convert synthetic images to real (with paired images) if you know. Any other vanila GAN would also be helpful.
我笔记本双显卡（集显+GTX980M），运行WGAN.ipynb时，采用默认的参数（deviece = '/gpu:0'），但运行时gpu并没有工作，只有cpu在满负荷跑，请问曹大大，会是哪里出了问题？跑其他深度学习代码GPU正常工作。 I run WGAN.ipynb on my laptop, and use the default config, (device = '/gpu:0')，but my GPU GTX980M doesn't work, only the CPU is running. Does anyone know...
Hi, I found both my d_loss and g_loss will grow extremely big with negative sign. According to the paper, discriminator should maximize the EM distance between fake and real data....
you did ly.conv2d_transpose(train, 128, 3, stride=2, activation_fn=tf.nn.relu, normalizer_fn=ly.batch_norm, padding='SAME', weights_initializer=tf.random_normal_initializer(0, 0.02)) I wonder if input variable is_training for function ly.batch_norm is always set to False. And if so does batch_norm...
mnist
Hello, I'm sorry to bother you, but I have some question, the mnist can be replaced with a pair of image?and can it be extracted to characteristics ?
