diff options
| author |  Alexandre Passos <apassos@google.com> | 2018-07-18 10:16:16 -0700 |
|---|---|---|
| committer |  TensorFlower Gardener <gardener@tensorflow.org> | 2018-07-18 10:22:51 -0700 |
| commit | 9cc29a75ce8131db67b48e92dac3c16a255b92ed (patch) | |
| tree | 73bf7a7483d8f7ae3872437609b6943218938ff4 /third_party/examples | |
| parent | 491b2d61156333c44e6bf06e2ac0a7ac02c4d310 (diff) | |
Allows constructing resource variables from tf.Variable.
Also adds arguments to control distributed aggregation to the tf.Variable constructor.
Removes tfe.Variable from examples as it's now unnecessary.
PiperOrigin-RevId: 205096552
Diffstat (limited to 'third_party/examples')
| -rw-r--r-- | third_party/examples/eager/spinn/spinn.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/third_party/examples/eager/spinn/spinn.py b/third_party/examples/eager/spinn/spinn.py index 67456a5bdf..c242ef3fdd 100644 --- a/third_party/examples/eager/spinn/spinn.py +++ b/third_party/examples/eager/spinn/spinn.py @@ -419,7 +419,7 @@ class SNLIClassifierTrainer(tfe.Checkpointable): # Create a custom learning rate Variable for the RMSProp optimizer, because # the learning rate needs to be manually decayed later (see # decay_learning_rate()). - self._learning_rate = tfe.Variable(lr, name="learning_rate") + self._learning_rate = tf.Variable(lr, name="learning_rate") self._optimizer = tf.train.RMSPropOptimizer(self._learning_rate, epsilon=1e-6) |