timesead.models.layers.planar_nf

Classes

PlanarTransform

Implementation of the invertible transformation used in planar flow

PlanarFlow

Base class for all neural network modules.

Module Contents

class timesead.models.layers.planar_nf.PlanarTransform(dim: int, epsilon: float = 0.0001)

Bases: torch.nn.Module

Implementation of the invertible transformation used in planar flow

f(z) = z + u * h(dot(w.T, z) + b)

See Section 4.1 in https://arxiv.org/pdf/1505.05770.pdf.

Initialise weights and bias.

Parameters:

  • dim (int) – Dimensionality of the distribution to be estimated.

  • epsilon (float)

epsilon = 0.0001
w
b
u
forward(z: torch.Tensor) → Tuple[torch.Tensor, torch.Tensor]
Parameters:

z (torch.Tensor)

Return type:

Tuple[torch.Tensor, torch.Tensor]

get_u_hat() → None

Enforce w^T u >= -1. When using h(.) = tanh(.), this is a sufficient condition for invertibility of the transformation f(z). See Appendix A.1.

Return type:

None

class timesead.models.layers.planar_nf.PlanarFlow(dim: int, num_layers: int = 6)

Bases: torch.nn.Module

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Parameters:

Make a planar flow by stacking planar transformations in sequence.

Parameters:

  • dim (int) – Dimensionality of the distribution to be estimated.

  • num_layers (int) – Number of transformations in the flow.

layers
forward(z: torch.Tensor) → Tuple[torch.Tensor, torch.Tensor]
Parameters:

z (torch.Tensor)

Return type:

Tuple[torch.Tensor, torch.Tensor]