Welcome to FrEIA’s documentation!

Framework for Easily Invertible Architectures. Module to construct invertible networks with pytorch, based on a graph structure of operations.

Tutorial

• Quickstart guide
• Tutorial
  • Basic concepts
  • Invertible Operations
    • Coupling blocks
    • Defining custom invertible operations
  • Sequential API
  • Computation graph API
  • Tips & Tricks, FAQ
• Examples
  • Network architectures
    • Small fully-connected INNs
      • Simple INN in 2 dimensions
      • Conditional INN for MNIST
    • Convolutional INN with invertible downsampling
    • Bayes-flow
    • Invertible U-Net
  • Full training loops
    • Training: MNIST conditional normalizing flow
    • Training: CelebA normalizing flow

Package Documentation

• FrEIA.framework package
  • Module contents
    • ConditionNode
      • ConditionNode.__init__()
      • ConditionNode.build_module()
      • ConditionNode.consume_output()
      • ConditionNode.rev_input()
    • FeedForwardNode
      • FeedForwardNode.__init__()
      • FeedForwardNode.build_module()
      • FeedForwardNode.rev_input()
    • GraphINN
      • GraphINN.__init__()
      • GraphINN.get_module_by_name()
      • GraphINN.get_node_by_name()
      • GraphINN.log_jacobian_numerical()
      • GraphINN.node_list
    • InputNode
      • InputNode.__init__()
      • InputNode.build_module()
      • InputNode.rev_input()
    • Node
      • Node.build_module()
      • Node.rev_input()
    • OutputNode
      • OutputNode.__init__()
      • OutputNode.build_module()
      • OutputNode.consume_output()
      • OutputNode.rev_input()
    • ReversibleGraphNet
      • ReversibleGraphNet.__init__()
    • ReversibleSequential
      • ReversibleSequential.__init__()
    • SequenceINN
      • SequenceINN.__init__()
      • SequenceINN.append()
    • collect_nodes()
• FrEIA.modules package
  • Abstract template
    • InvertibleModule
      • InvertibleModule.__init__()
      • InvertibleModule.forward()
      • InvertibleModule.log_jacobian()
      • InvertibleModule.output_dims()
  • Coupling blocks
    • AllInOneBlock
      • AllInOneBlock.__init__()
    • NICECouplingBlock
      • NICECouplingBlock.__init__()
    • RNVPCouplingBlock
      • RNVPCouplingBlock.__init__()
    • GLOWCouplingBlock
      • GLOWCouplingBlock.__init__()
    • GINCouplingBlock
      • GINCouplingBlock.__init__()
    • AffineCouplingOneSided
      • AffineCouplingOneSided.__init__()
    • ConditionalAffineTransform
      • ConditionalAffineTransform.__init__()
  • Reshaping
    • IRevNetDownsampling
      • IRevNetDownsampling.__init__()
    • IRevNetUpsampling
      • IRevNetUpsampling.__init__()
    • HaarDownsampling
      • HaarDownsampling.__init__()
    • HaarUpsampling
      • HaarUpsampling.__init__()
    • Flatten
      • Flatten.__init__()
    • Reshape
      • Reshape.__init__()
  • Graph topology
    • Split
      • Split.__init__()
    • Concat
      • Concat.__init__()
  • Other learned transforms
    • ActNorm
      • ActNorm.__init__()
      • ActNorm.initialize()
      • ActNorm.load_state_dict()
      • ActNorm.scale
    • InvAutoAct
      • InvAutoAct.__init__()
    • InvAutoActTwoSided
      • InvAutoActTwoSided.__init__()
    • LearnedElementwiseScaling
      • LearnedElementwiseScaling.__init__()
    • OrthogonalTransform
      • OrthogonalTransform.__init__()
    • HouseholderPerm
      • HouseholderPerm.__init__()
  • Fixed (non-learned) transforms
    • PermuteRandom
      • PermuteRandom.__init__()
    • FixedLinearTransform
      • FixedLinearTransform.__init__()
    • Fixed1x1Conv
      • Fixed1x1Conv.__init__()
    • InvertibleSigmoid
      • InvertibleSigmoid.__init__()
  • Approximately- or semi-invertible transforms
    • InvAutoFC
      • InvAutoFC.__init__()
    • InvAutoConv2D
      • InvAutoConv2D.__init__()
    • IResNetLayer
      • IResNetLayer.__init__()
      • IResNetLayer.lipschitz_correction()
    • GaussianMixtureModel
      • GaussianMixtureModel.__init__()
      • GaussianMixtureModel.nll_loss()
      • GaussianMixtureModel.nll_upper_bound()
      • GaussianMixtureModel.normalize_weights()
      • GaussianMixtureModel.pick_mixture_component()

Indices and tables

• Index

• Module Index

• Search Page