layers.rst

torchgan.layers

.. currentmodule:: torchgan.layers

This layers subpackage is a collection of popular building blocks for GAN architectures. Currently the following blocks are supported:

• Residual Blocks
  • ResidualBlock2d
  • ResidualBlockTranspose2d
• Densenet Blocks
  • BasicBlock2d
  • BottleneckBlock2d
  • TransitionBlock2d
  • TransitionBlockTranspose2d
  • DenseBlock2d
• Self Attention
  • SelfAttention2d
• Spectral Normalization
  • SpectralNorm2d
• Minibatch Discrimination
  • MinibatchDiscrimination1d
• Virtual Batch Normalization
  • VirtualBatchNorm

Residual Blocks

ResidualBlock2d

.. autoclass:: ResidualBlock2d
    :members:

ResidualBlockTranspose2d

.. autoclass:: ResidualBlockTranspose2d
    :members:

Densenet Blocks

BasicBlock2d

.. autoclass:: BasicBlock2d
    :members:

BottleneckBlock2d

.. autoclass:: BottleneckBlock2d
    :members:

TransitionBlock2d

.. autoclass:: TransitionBlock2d
    :members:

TransitionBlockTranspose2d

.. autoclass:: TransitionBlockTranspose2d
    :members:

DenseBlock2d

.. autoclass:: DenseBlock2d
    :members:

Self Attention

SelfAttention2d

.. autoclass:: SelfAttention2d
    :members:

Spectral Normalization

SpectralNorm2d

.. autoclass:: SpectralNorm2d
    :members:

Minibatch Discrimination

MinibatchDiscrimination1d

.. autoclass:: MinibatchDiscrimination1d
    :members:

Virtual Batch Normalization

VirtualBatchNorm

.. autoclass:: VirtualBatchNorm
    :members: