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

Fix comment describing ignore_longer_outputs_than_inputs.

Browse Source 
PiperOrigin-RevId: 157400110
This commit is contained in:
A. Unique TensorFlower 2017-05-29 08:42:33 -07:00 committed by TensorFlower Gardener
parent 5f097217f4
commit e9d2fba8f1
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tensorflow/core/ops/ctc_ops.cc
View File

@ -77,8 +77,8 @@ ctc_merge_repeated: Scalar. If set to false, *during* CTC calculation
repeated non-blank labels will not be merged and are interpreted as repeated non-blank labels will not be merged and are interpreted as
individual labels. This is a simplified version of CTC. individual labels. This is a simplified version of CTC.
ignore_longer_outputs_than_inputs: Scalar. If set to true, during CTC ignore_longer_outputs_than_inputs: Scalar. If set to true, during CTC
calculation items have longer input sequences than output sequences calculation, items that have longer output sequences than input sequences
are ignored by returning zero-gradient for those items. are skipped: they don't contribute to the loss term and have zero-gradient.
loss: A vector (batch) containing log-probabilities. loss: A vector (batch) containing log-probabilities.
gradient: The gradient of `loss`. 3-D, shape: gradient: The gradient of `loss`. 3-D, shape:
`(max_time x batch_size x num_classes)`. `(max_time x batch_size x num_classes)`.