Trait std::GanModel

#include <std/ml/gan.h> source

trait GanModel : Parameterized + Generative {
    struct GanLosses train_step(const &FloatTensor real, ?&FloatTensor labels);
    struct GanLosses train(const &Dataset data, unsigned long epochs, unsigned long batchSize);
}

What a trainable GAN offers (Gan and Dcgan; CpuGanTrainer is generic over it): its two training entry points, its tensors and sampling.

Required methods

§struct GanLosses train_step(const &FloatTensor real, ?&FloatTensor labels)source

§struct GanLosses train(const &Dataset data, unsigned long epochs, unsigned long batchSize)source