Class Dcgan
java.lang.Object
ai.beestream.jitnn.nn.models.Dcgan
DCGAN (Radford et al., 2016) for 28x28 single-channel images: a generator from a latent vector
through a 7x7 feature map and two stride-2 transposed convolutions to a tanh image, and a
discriminator of two stride-2 convolutions with LeakyReLU(0.2) and a linear logit. BatchNorm
everywhere but the generator's output and the discriminator's input layer.
-
Nested Class Summary
Nested ClassesModifier and TypeClassDescriptionstatic final classImages [B, 1, 28, 28] to logits [B, 1].static final classz [B, latent] to images [B, 1, 28, 28] in [-1, 1]. -
Method Summary