diff options
author | Allen Lavoie <allenl@google.com> | 2018-06-22 11:13:43 -0700 |
---|---|---|
committer | TensorFlower Gardener <gardener@tensorflow.org> | 2018-06-22 11:16:40 -0700 |
commit | 51695446ec779bd069c5623bf696e4223ff2af48 (patch) | |
tree | be13cb094132252b1f0712ed651f76954b414b1e /tensorflow/contrib/cudnn_rnn | |
parent | 49934f7d1a289325480c618c658b7c4cdb8584c6 (diff) |
Split dependency tracking out from CheckpointableBase
Some unit test fiddling, but otherwise just moving code around.
My goal is to be able to use checkpointable data structures (or something like them) in Checkpointable's __setattr__ override. Checkpointable data structures depend on Layer, so Checkpointable and CheckpointableBase need to be in seprate files (so we can have the dependency chain CheckpointableBase->Layer->CheckpointableDataStructure->Checkpointable).
This will also require changes to python/keras/engine/__init__.py (which currently requires Network and Layer be imported together), but I'll do that in a separate change.
PiperOrigin-RevId: 201712549
Diffstat (limited to 'tensorflow/contrib/cudnn_rnn')
-rw-r--r-- | tensorflow/contrib/cudnn_rnn/python/ops/cudnn_rnn_ops.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/tensorflow/contrib/cudnn_rnn/python/ops/cudnn_rnn_ops.py b/tensorflow/contrib/cudnn_rnn/python/ops/cudnn_rnn_ops.py index 8822a7523f..748d7cd011 100644 --- a/tensorflow/contrib/cudnn_rnn/python/ops/cudnn_rnn_ops.py +++ b/tensorflow/contrib/cudnn_rnn/python/ops/cudnn_rnn_ops.py @@ -33,7 +33,7 @@ from tensorflow.python.ops import rnn_cell_impl from tensorflow.python.ops import state_ops from tensorflow.python.ops import variable_scope as vs from tensorflow.python.training import saver -from tensorflow.python.training.checkpointable import base as checkpointable_lib +from tensorflow.python.training.checkpointable import tracking as checkpointable_lib CUDNN_RNN_UNIDIRECTION = "unidirectional" CUDNN_RNN_BIDIRECTION = "bidirectional" |