diff options
author | A. Unique TensorFlower <gardener@tensorflow.org> | 2018-01-19 12:08:47 -0800 |
---|---|---|
committer | TensorFlower Gardener <gardener@tensorflow.org> | 2018-01-19 12:13:53 -0800 |
commit | 83b751621439cc2b8a85450972414cf2f92a58cf (patch) | |
tree | 860f7f40a104388113121c79f7c6cb51cf7b1198 /tensorflow/contrib/lite/builtin_op_data.h | |
parent | 39ae44e9822ed76639bae3ccf800b36039d1da55 (diff) |
Add support for time_major shape format to the sequential RNN Op in TF Lite.
This option, if set, changes the shape format of the inputs and outputs to
[max_time, batch_size, depth]. If false, it uses [batch_size, max_time, depth].
By default, it is set to false.
PiperOrigin-RevId: 182569507
Diffstat (limited to 'tensorflow/contrib/lite/builtin_op_data.h')
-rw-r--r-- | tensorflow/contrib/lite/builtin_op_data.h | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/tensorflow/contrib/lite/builtin_op_data.h b/tensorflow/contrib/lite/builtin_op_data.h index 0c333f9e8c..3b43a1fd5d 100644 --- a/tensorflow/contrib/lite/builtin_op_data.h +++ b/tensorflow/contrib/lite/builtin_op_data.h @@ -83,6 +83,11 @@ typedef struct { TfLiteFusedActivation activation; } TfLiteRNNParams; +typedef struct { + bool time_major; + TfLiteFusedActivation activation; +} TfLiteSequenceRNNParams; + typedef struct { TfLiteFusedActivation activation; } TfLiteFullyConnectedParams; typedef enum { |