ML Community Day is November 9! Join us for updates from TensorFlow, JAX, and more Learn more

tfp.experimental.vi.util.build_trainable_linear_operator_diag

Build a trainable LinearOperatorDiag instance.

tfp.experimental.vi.util.build_trainable_linear_operator_diag(
    shape, scale_initializer=0.01, diag_bijector=None, dtype=None, seed=None,
    name=None
)

shape Shape of the LinearOperator, equal to [b0, ..., bn, d], where b0...bn are batch dimensions and d is the length of the diagonal.
scale_initializer Variables are initialized with samples from Normal(0, scale_initializer).
diag_bijector Bijector to apply to the diagonal of the operator.
dtype tf.dtype of the LinearOperator.
seed PRNG seed; see tfp.random.sanitize_seed for details.
name str, name for tf.name_scope.

operator Trainable instance of tf.linalg.LinearOperatorDiag.