# Components

- [GAN](/hypergan/components/gan.md): Create the domain object map and connect all components.
- [Aligned GAN](/hypergan/components/gan/aligned-gan.md)
- [Aligned Interpolated GAN](/hypergan/components/gan/aligned-interpolated-gan.md)
- [Standard GAN](/hypergan/components/gan/standard-gan.md)
- [Generator](/hypergan/components/generator.md): Manages the network used to create fake samples.
- [Configurable Generator](/hypergan/components/generator/configurable-generator.md)
- [DCGAN Generator](/hypergan/components/generator/dcgan-generator.md)
- [Resizable Generator](/hypergan/components/generator/resizable-generator.md)
- [Discriminator](/hypergan/components/discriminator.md): Manages the network used to discriminate real/fake samples.
- [DCGAN Discriminator](/hypergan/components/discriminator/dcgan-discriminator.md)
- [Configurable Discriminator](/hypergan/components/discriminator/configurable-discriminator.md)
- [Layers](/hypergan/components/layers.md)
- [add](/hypergan/components/layers/add.md)
- [cat](/hypergan/components/layers/cat.md)
- [channel\_attention](/hypergan/components/layers/channel_attention.md)
- [ez\_norm](/hypergan/components/layers/ez_norm.md)
- [layer](/hypergan/components/layers/layer.md)
- [mul](/hypergan/components/layers/mul.md)
- [multi\_head\_attention](/hypergan/components/layers/multi_head_attention.md)
- [operation](/hypergan/components/layers/operation.md)
- [pixel\_shuffle](/hypergan/components/layers/pixel_shuffle.md)
- [residual](/hypergan/components/layers/residual.md)
- [resizable\_stack](/hypergan/components/layers/resizable_stack.md)
- [segment\_softmax](/hypergan/components/layers/segment_softmax.md)
- [upsample](/hypergan/components/layers/upsample.md)
- [Loss](/hypergan/components/loss.md): Losses define how the error signals are interpreted for the two GAN players.
- [ALI Loss](/hypergan/components/loss/ali-loss.md): layer add for configurable component
- [F Divergence Loss](/hypergan/components/loss/f-divergence-loss.md)
- [Least Squares Loss](/hypergan/components/loss/least-squares-loss.md)
- [Logistic Loss](/hypergan/components/loss/logistic-loss.md)
- [QP Loss](/hypergan/components/loss/qp-loss.md): https://arxiv.org/abs/1811.07296
- [RAGAN Loss](/hypergan/components/loss/ragan-loss.md)
- [Realness Loss](/hypergan/components/loss/realness-loss.md)
- [Softmax Loss](/hypergan/components/loss/softmax-loss.md)
- [Standard Loss](/hypergan/components/loss/standard-loss.md)
- [Wasserstein Loss](/hypergan/components/loss/wasserstein-loss.md)
- [Latent](/hypergan/components/latent.md): Describes a prior distribution that feeds to a generator
- [Uniform Distribution](/hypergan/components/latent/uniform-distribution.md)
- [Trainer](/hypergan/components/trainer.md): Trains the GAN object.
- [Alternating Trainer](/hypergan/components/trainer/alternating-trainer.md)
- [Simultaneous Trainer](/hypergan/components/trainer/simultaneous-trainer.md)
- [Balanced Trainer](/hypergan/components/trainer/balanced-trainer.md)
- [Accumulate Gradient Trainer](/hypergan/components/trainer/accumulate-gradient-trainer.md)
- [Optimizer](/hypergan/components/optimizer.md): Optimizers move weights along gradients.  They are managed by the trainer.
- [Train Hook](/hypergan/components/trainhook.md): Train hooks provide training events and loss modification to trainers.
- [Adversarial Norm](/hypergan/components/trainhook/adversarial-norm.md): Custom research
- [Weight Constraint](/hypergan/components/trainhook/weight-constraint.md)
- [Stabilizing Training](/hypergan/components/trainhook/stabilizing-training.md): https://github.com/rothk/Stabilizing\_GANs
- [JARE](/hypergan/components/trainhook/jare.md): https://arxiv.org/abs/1806.09235
- [Learning Rate Dropout](/hypergan/components/trainhook/learning-rate-dropout.md): https://arxiv.org/abs/1912.00144
- [Gradient Penalty](/hypergan/components/trainhook/gradient-penalty.md): https://arxiv.org/pdf/1704.00028.pdf
- [Rolling Memory](/hypergan/components/trainhook/rolling-memory.md): (no paper)
- [Other GAN implementations](/hypergan/components/other-gan-implementations.md): Linked with awe