experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Clarify assert docs

Browse Source 
PiperOrigin-RevId: 303125125
Change-Id: Ib69a093f0e9723cd275f1d6e55dddb1c44ca5765
This commit is contained in:
Alexandre Passos 2020-03-26 09:10:58 -07:00 committed by TensorFlower Gardener
parent f7132de3d4
commit f8c45a5b38
1 changed files with 11 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
tensorflow/python/ops/control_flow_ops.py
View File

@ -117,16 +117,6 @@ def Assert(condition, data, summarize=None, name=None):
If `condition` evaluates to false, print the list of tensors in `data`. If `condition` evaluates to false, print the list of tensors in `data`.
`summarize` determines how many entries of the tensors to print. `summarize` determines how many entries of the tensors to print.
NOTE: In graph mode, to ensure that Assert executes, one usually attaches
a dependency:
```python
# Ensure maximum element of x is smaller or equal to 1
assert_op = tf.Assert(tf.less_equal(tf.reduce_max(x), 1.), [x])
with tf.control_dependencies([assert_op]):
... code using x ...
```
Args: Args:
condition: The condition to evaluate. condition: The condition to evaluate.
data: The tensors to print out when condition is false. data: The tensors to print out when condition is false.
@ -141,8 +131,17 @@ def Assert(condition, data, summarize=None, name=None):
@end_compatibility @end_compatibility
Raises: Raises:
@compatibility(eager) @compatibility(TF1)
`tf.errors.InvalidArgumentError` if `condition` is not true When in TF V1 mode (that is, outside `tf.function`) Assert needs a control
dependency on the output to ensure the assertion executes:
```python
# Ensure maximum element of x is smaller or equal to 1
assert_op = tf.Assert(tf.less_equal(tf.reduce_max(x), 1.), [x])
with tf.control_dependencies([assert_op]):
... code using x ...
```
@end_compatibility @end_compatibility
""" """
if context.executing_eagerly(): if context.executing_eagerly():