def assert_values_compatible_with_distribution_shape(
scoped_name: str, values: Any, dist_shape: tf.TensorShape) -> None:
"""Assert if a supplied values are compatible with a distribution's TensorShape.
A value is considered to have a consistent shape with the distribution if
two conditions are met.
1) It has a greater or equal number of dimensions when compared to the
distribution: len(values.shape) >= len(dist_shape)
2) The supplied values' shape is compatible with the distribution's shape:
dist_shape.is_compatible_with(values.shape[(len(values.shape) - len(dist_shape)):])
Parameters
----------
scoped_name: str
The variable's scoped name
values: Any
The supplied values
dist_shape: tf.TensorShape
The ``tf.TensorShape`` instance.
Returns
-------
None
Raises
------
EvaluationError
When the ``values`` shape is not compatible with the ``dist_shape``.
"""
value_shape = get_observed_tensor_shape(values)
if value_shape.rank < dist_shape.rank or not dist_shape.is_compatible_with(
value_shape[(len(value_shape) - len(dist_shape)):]):
raise EvaluationError(
EvaluationError.INCOMPATIBLE_VALUE_AND_DISTRIBUTION_SHAPE.format(
scoped_name, dist_shape, value_shape))
def _assert_compatible_shape(shape: tf.TensorShape, example_shape):
if not shape.is_compatible_with(example_shape):
raise ValueError(f"example shape {example_shape} is incompatible with "
f"feature shape {shape}")
