tf.compat.v1.debugging

Assert tensor shapes and dimension size relationships between tensors.

tf.compat.v1.debugging.assert_shapes(
    shapes, data=None, summarize=None, message=None, name=None
)

This Op checks that a collection of tensors shape relationships satisfies given constraints.

Example:

n = 10
q = 3
d = 7
x = tf.zeros([n,q])
y = tf.ones([n,d])
param = tf.Variable([1.0, 2.0, 3.0])
scalar = 1.0
tf.debugging.assert_shapes([
 (x, ('N', 'Q')),
 (y, ('N', 'D')),
 (param, ('Q',)),
 (scalar, ()),
])

tf.debugging.assert_shapes([
  (x, ('N', 'D')),
  (y, ('N', 'D'))
])
Traceback (most recent call last):

ValueError: ...

Example of adding a dependency to an operation:

with tf.control_dependencies([tf.assert_shapes(shapes)]):
  output = tf.matmul(x, y, transpose_a=True)

If x, y, param or scalar does not have a shape that satisfies all specified constraints, message, as well as the first summarize entries of the first encountered violating tensor are printed, and InvalidArgumentError is raised.

Size entries in the specified shapes are checked against other entries by their hash, except:

a size entry is interpreted as an explicit size if it can be parsed as an integer primitive.
a size entry is interpreted as any size if it is None or '.'.

If the first entry of a shape is ... (type Ellipsis) or '*' that indicates a variable number of outer dimensions of unspecified size, i.e. the constraint applies to the inner-most dimensions only.

Scalar tensors and specified shapes of length zero (excluding the 'inner-most' prefix) are both treated as having a single dimension of size one.

Args
`shapes`	A list of (`Tensor`, `shape`) tuples, wherein `shape` is the expected shape of `Tensor`. See the example code above. The `shape` must be an iterable. Each element of the iterable can be either a concrete integer value or a string that abstractly represents the dimension. For example, `('N', 'Q')` specifies a 2D shape wherein the first and second dimensions of shape may or may not be equal. `('N', 'N', 'Q')` specifies a 3D shape wherein the first and second dimensions are equal. `(1, 'N')` specifies a 2D shape wherein the first dimension is exactly 1 and the second dimension can be any value. Note that the abstract dimension letters take effect across different tuple elements of the list. For example, `tf.debugging.assert_shapes([(x, ('N', 'A')), (y, ('N', 'B'))]` asserts that both `x` and `y` are rank-2 tensors and their first dimensions are equal (`N`). `shape` can also be a `tf.TensorShape`.
`data`	The tensors to print out if the condition is False. Defaults to error message and first few entries of the violating tensor.
`summarize`	Print this many entries of the tensor.
`message`	A string to prefix to the default message.
`name`	A name for this operation (optional). Defaults to "assert_shapes".

Returns
Op raising `InvalidArgumentError` unless all shape constraints are satisfied. If static checks determine all constraints are satisfied, a `no_op` is returned.

Raises
`ValueError`	If static checks determine any shape constraint is violated.

TensorFlow

tf.compat / v1 / v1.debugging.Overview

Example:

Args

Returns

Raises

TensorFlow

tf

tf.audio

tf.autograph

tf.bitwise

tf.compat

tf.config

tf.data

tf.debugging

tf.distribute

tf.dtypes

tf.errors

tf.estimator

tf.experimental

tf.feature_column

tf.graph_util

tf.image

tf.initializers

tf.io

tf.keras

tf.linalg

tf.lite

tf.lookup

tf.losses

tf.math

tf.metrics

tf.nest

tf.nn

tf.optimizers

tf.quantization

tf.queue

tf.ragged

tf.random

tf.raw_ops

tf.saved_model

tf.sets

tf.signal

tf.sparse

tf.strings

tf.summary

tf.sysconfig

tf.test

tf.tpu

tf.train

tf.version

tf.xla

tf.compat / v1 / v1.debugging.Overview

Example:

Args

Returns

Raises