diff options
| author |  Justin Lebar <jlebar@google.com> | 2017-12-15 16:05:58 -0800 |
|---|---|---|
| committer |  TensorFlower Gardener <gardener@tensorflow.org> | 2017-12-15 16:09:46 -0800 |
| commit | deb50f80e5a87325adec5db826673e05acb4f5ab (patch) | |
| tree | 8eda7aaa1ee9b2a4e850e43790b1f7936b826b64 | |
| parent | f122ae338f958fa643ba05e3ff9660959bdde3a0 (diff) | |
[XLA] Update docs to indicate that BatchNormGrad returns {operand, scale, offset}, not {operand, offset, scale}.
PiperOrigin-RevId: 179251837
| -rw-r--r-- | tensorflow/docs_src/performance/xla/operation_semantics.md | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/tensorflow/docs_src/performance/xla/operation_semantics.md b/tensorflow/docs_src/performance/xla/operation_semantics.md index d6f05f81bf..71e5db5d9f 100644 --- a/tensorflow/docs_src/performance/xla/operation_semantics.md +++ b/tensorflow/docs_src/performance/xla/operation_semantics.md @@ -62,10 +62,11 @@ The output type is a tuple of three handles: |------------- | ----------------------- | ------------------------------------| |`grad_operand`| `ComputationDataHandle` | gradient with respect to input | : : : `operand` : -|`grad_offset` | `ComputationDataHandle` | gradient with respect to input | -: : : `offset` : |`grad_scale` | `ComputationDataHandle` | gradient with respect to input | : : : `scale` : +|`grad_offset` | `ComputationDataHandle` | gradient with respect to input | +: : : `offset` : + ## BatchNormInference |