diff options
| author |  Benoit Steiner <bsteiner@google.com> | 2016-08-12 15:02:49 -0800 |
|---|---|---|
| committer |  TensorFlower Gardener <gardener@tensorflow.org> | 2016-08-12 16:16:31 -0700 |
| commit | a92da9e09fb476a9b267499e326919a89b826fb7 (patch) | |
| tree | 37c257b0d2d5e82b6f8caa8ad97e4f4de2b39a75 /tensorflow/contrib/quantization | |
| parent | 9ace75c7c88f30a488ce022be03b8f6dc1c9c990 (diff) | |
Merge changes from github.
Change: 130150683
Diffstat (limited to 'tensorflow/contrib/quantization')
| -rw-r--r-- | tensorflow/contrib/quantization/ops/array_ops.cc | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/tensorflow/contrib/quantization/ops/array_ops.cc b/tensorflow/contrib/quantization/ops/array_ops.cc index ea59a9ae3e..7dd64f82e0 100644 --- a/tensorflow/contrib/quantization/ops/array_ops.cc +++ b/tensorflow/contrib/quantization/ops/array_ops.cc @@ -146,7 +146,7 @@ If the mode is 'MIN_FIRST', then this approach is used: number_of_steps = 1 << (# of bits in T) range_adjust = number_of_steps / (number_of_steps - 1) range = (range_max - range_min) * range_adjust -range_scale = number_of_steps / range +range_scale = range / number_of_steps const double offset_input = static_cast<double>(input) - lowest_quantized; result = range_min + ((input - numeric_limits<T>::min()) * range_scale) ``` |