|View source on GitHub|
Create a random variable for PixelCNN.
tfp.edward2.PixelCNN( *args, **kwargs )
See PixelCNN for more details.
Original Docstring for Distribution
Construct Pixel CNN++ distribution.
TensorShapeor tuple for the
[height, width, channels]dimensions of the image.
TensorShapeor tuple for the shape of the conditional input, or
Noneif there is no conditional input.
int, the number of layers (shown in Figure 2 of ) within each highest-level block of Figure 2 of .
int, the number of hightest-level blocks (separated by expansions/contractions of dimensions in Figure 2 of .)
int, the number of convolutional filters.
int, number of components in the logistic mixture distribution.
tuple, height and width in pixels of the receptive field of the convolutional layers above and to the left of a given pixel. The width (second element of the tuple) should be odd. Figure 1 (middle) of  shows a receptive field of (3, 5) (the row containing the current pixel is included in the height). The default of (3, 3) was used to produce the results in .
float, the dropout probability. Should be between 0 and 1.
string, the type of activation to use in the resnet blocks. May be 'concat_elu', 'elu', or 'relu'.
Truethen use weight normalization (works only in Eager mode).
Truethen use data-dependent initialization (has no effect if
int, the maximum value of the input data (255 for an 8-bit image).
int, the minimum value of the input data.
dtype: Data type of the
string, the name of the