Python callable which returns a
tf.distribution.Distribution-like instance from a single input
representing the distribution's required mean, i.e.,
mean = linear_model_to_mean_fn(matmul(model_matrix, weights)).
Python callable which returns the
distribution's required mean as computed from the predicted linear
response, matmul(model_matrix, weights).
Python bool indicating that taken together,
distribution_fn and linear_model_to_mean_fn imply that the
distribution's variance is equivalent to d/dr
Python str used as TF namescope for ops created by member
functions. Default value: None (i.e., the subclass name).
Returns True when variance(r) == grad_mean(r) for all r.
Returns TF namescope prefixed to ops created by member functions.