GAN CIFAR Example

Small LSGAN-style executable path.

This trains:

• a generator z -> image toward the current CIFAR minibatch as a stable warm-up objective;
• a discriminator on real CIFAR images (1) and deterministic noise images (0).

The formal LSGAN objective decomposition lives in NN.Spec.Models.Gan and NN.MLTheory.Generative.Latent.GAN. A full alternating adversarial trainer can reuse the same generator/discriminator constructors and data path.

def NN.Examples.Models.Generative.Gan.exeName : String

def NN.Examples.Models.Generative.Gan.defaultLogJson : System.FilePath

def NN.Examples.Models.Generative.Gan.cfg : API.nn.models.VectorGenerativeConfig

@[reducible, inline]

abbrev NN.Examples.Models.Generative.Gan.Z : Shape

@[reducible, inline]

abbrev NN.Examples.Models.Generative.Gan.X : Shape

@[reducible, inline]

abbrev NN.Examples.Models.Generative.Gan.S : Shape

def NN.Examples.Models.Generative.Gan.mkGenerator : API.nn.M (API.nn.Sequential Z X)

def NN.Examples.Models.Generative.Gan.mkDiscriminator : API.nn.M (API.nn.Sequential X S)

def NN.Examples.Models.Generative.Gan.trainCurve (opts : Runtime.Autograd.Torch.Options) (xPath yPath : System.FilePath) (nRows seed steps : ℕ) : IO Runtime.Training.Curve

def NN.Examples.Models.Generative.Gan.main (args : List String) : IO UInt32