diff options
| author |  Geoffrey Irving <geoffreyi@google.com> | 2016-04-14 11:52:13 -0800 |
|---|---|---|
| committer |  TensorFlower Gardener <gardener@tensorflow.org> | 2016-04-14 13:01:53 -0700 |
| commit | a58e5caa9a3e890a2137791e8e4a0870828fa882 (patch) | |
| tree | 8da87df94353c87422800ab8b42839a4a9d9582c /tensorflow/python/training/coordinator.py | |
| parent | 0d088fbaaa724a3b5d19870c1ae334671af3269e (diff) | |
Extend singleton case in _AggregatedGrads to IndexedSlices
If we are aggregating one gradient, we can just pass it through. Previously
the code did this only for tf.Tensor, not tf.IndexedSlices.
Change: 119880763
Diffstat (limited to 'tensorflow/python/training/coordinator.py')
0 files changed, 0 insertions, 0 deletions