diff options
| author | Jan Eilers <jan.eilers@arm.com> | 2021-06-02 12:01:25 +0100 |
|---|---|---|
| committer | Jan Eilers <jan.eilers@arm.com> | 2021-06-16 11:31:42 +0000 |
| commit | 53ef79504b4c881c572735393c2eede5fa556c46 (patch) | |
| tree | f6e0cd27c4d03075fa154074c5b12d7c8c3149f7 /src/armnn/test/InferOutputTests.hpp | |
| parent | 77fe76bfa8cb798943821d1f3e432c228e1cdee3 (diff) | |
| download | armnn-53ef79504b4c881c572735393c2eede5fa556c46.tar.gz | |
IVGCVSW-5826 Change weights layout for depthwise to [1,H,W,I*M]
* This change is necessary because tflite uses a [1,H,W,I*M] format
and uses the I*M dimension for per axis quantization. Our previous
layout [M,I,H,W] can't handle the correlating quantization scales.
* Updates Onnx-, TfLiteParser and TfliteDelegate
* Updates the CpuRef, CpuAcc and GpuAcc backends
* Adjusts unit tests
* Adds test to ensure models with old layout can still be read and
executed
* Adds conversion function to previous layout [1,H,W,I*M] --> [M,I,H,W]
which can be used by backend developers
!android-nn-driver:5553
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: Ifef23368b8c3702cf315a5838d214f7dc13c0152
Diffstat (limited to 'src/armnn/test/InferOutputTests.hpp')
| -rw-r--r-- | src/armnn/test/InferOutputTests.hpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/armnn/test/InferOutputTests.hpp b/src/armnn/test/InferOutputTests.hpp index b8276de80c..6e2676ec8e 100644 --- a/src/armnn/test/InferOutputTests.hpp +++ b/src/armnn/test/InferOutputTests.hpp @@ -518,7 +518,7 @@ void DepthwiseConvolution2dInferOutputShapeTest() armnn::TensorShape inputShape(4, inputSize.data()); shapes.push_back(inputShape); - const std::vector<unsigned int> filterSize = { 1, 2, 3, 3}; + const std::vector<unsigned int> filterSize = { 1, 3, 3, 2 }; armnn::TensorShape filterShape(4, filterSize.data()); shapes.push_back(filterShape); |