tfp.layers.default_mean_field_normal_fn

View source on GitHub

Creates a function to build Normal distributions with trainable params.

This function produces a closure which produces tfd.Normal parameterized by a loc and scale each created using tf.get_variable.

is_singular Python bool if True, forces the special case limit of scale->0, i.e., a Deterministic distribution.
loc_initializer Initializer function for the loc parameters. The default is tf.random_normal_initializer(mean=0., stddev=0.1).
untransformed_scale_initializer Initializer function for the scale parameters. Default value: tf.random_normal_initializer(mean=-3., stddev=0.1). This implies the softplus transformed result is initialized near 0. It allows a Normal distribution with scale parameter set to this value to approximately act like a point mass.
loc_regularizer Regularizer function for the loc parameters.
untransformed_scale_regularizer Regularizer function for the scale parameters.
loc_constraint An optional projection function to be applied to the loc after being updated by an Optimizer. The function must take as input the unprojected variable and must return the projected variable (which must have the same shape). Constraints are not safe to use when doing asynchronous distributed training.
untransformed_scale_constraint An optional projection function to be applied to the scale parameters after being updated by an Optimizer (e.g. used to implement norm constraints or value constraints). The function must take as input the unprojected variable and must return the projected variable (which must have the same shape). Constraints are not safe to use when doing asynchronous distributed training.

make_normal_fn Python callable which creates a tfd.Normal using from args: dtype, shape, name, trainable, add_variable_fn.