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

Making the failure an OpError rather than InternalError because changing the compilation option to --define=with_mlir_support=true changes the place in the placer logic it fails.

Browse Source 
PiperOrigin-RevId: 275542262
Change-Id: If6172534903b9bff5337ed4b9887af03508dfcf9
This commit is contained in:
Rohan Jain 2019-10-18 14:03:56 -07:00 committed by TensorFlower Gardener
parent b00f7e2367
commit 8b935680a5
1 changed files with 1 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tensorflow/python/data/experimental/kernel_tests/replicate_cluster_test.py
View File

@ -108,9 +108,7 @@ class ReplicateClusterTest(test_base.DatasetTestBase, parameterized.TestCase):
it1 = dataset_ops.make_initializable_iterator(dataset1)
# We don't support stateful ops across processes in functions as of now.
with session.Session(self._target) as sess:
with self.assertRaisesRegexp(
errors.InternalError,
"tf.function with resource inputs residing on remote devices."):
with self.assertRaises(errors.OpError):
sess.run(it1.initializer)
@combinations.generate(