tfp.experimental.joint_distribution_layers.AffineLayer

Affine layer.

This represents a linear map: y = weights @ x + bias.

weights A floating point Tensor with shape [out_units, in_units].
bias A floating point Tensor with shape [out_units].

Methods

__call__

View source

Applies the layer to an input.

Args
x A floating point Tensor with shape [in_units].

Returns
y A floating point Tensor with shape [out_units].