diff options
| author |  Yifei Feng <yifeif@google.com> | 2018-05-24 19:12:26 -0700 |
|---|---|---|
| committer |  TensorFlower Gardener <gardener@tensorflow.org> | 2018-05-24 19:15:01 -0700 |
| commit | b59833c3fd91511b33255369016868e4ae6cda2e (patch) | |
| tree | ecbd70cfd3abb5d934f6eb4b7280a35e8589f5cf /tensorflow/core/util/tensor_format.h | |
| parent | 2b99d9cbc7166efedaff9eee11744348da30fc8a (diff) | |
Merge changes from github.
Revert #18413. Too many internal test failures due to the name scope change caused by this change.
Revert #18192. Cannot use re2::StringPiece internally. Need alternative for set call. Will pull and clean this up in a separate change.
PiperOrigin-RevId: 197991247
Diffstat (limited to 'tensorflow/core/util/tensor_format.h')
| -rw-r--r-- | tensorflow/core/util/tensor_format.h | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/tensorflow/core/util/tensor_format.h b/tensorflow/core/util/tensor_format.h index 58bc79aebf..d3d5602f92 100644 --- a/tensorflow/core/util/tensor_format.h +++ b/tensorflow/core/util/tensor_format.h @@ -75,7 +75,7 @@ enum FilterTensorFormat { FORMAT_OIHW = 1, // OIHW_VECT_I is the most performant tensor format for cudnn6's quantized - // int8 convolution and fused convolution. It is analagous to the NCHW_VECT_C + // int8 convolution and fused convolution. It is analogous to the NCHW_VECT_C // data format. It is laid out in the same order as OIHW, except that the size // of the Input Channels dimension is divided by 4, and a new dimension of // size 4 is appended, which packs 4 adjacent input channel weights into an |