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/CreateWorkload.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/CreateWorkload.hpp')
-rw-r--r-- | src/armnn/test/CreateWorkload.hpp | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/src/armnn/test/CreateWorkload.hpp b/src/armnn/test/CreateWorkload.hpp index 581c621a16..b07e3b80a5 100644 --- a/src/armnn/test/CreateWorkload.hpp +++ b/src/armnn/test/CreateWorkload.hpp @@ -1149,7 +1149,7 @@ std::unique_ptr<DepthwiseConvolution2dFloat32Workload> CreateDepthwiseConvolutio DepthwiseConvolution2dLayer* const layer = graph.AddLayer<DepthwiseConvolution2dLayer>(layerDesc, "layer"); - layer->m_Weight = std::make_unique<ScopedTensorHandle>(TensorInfo({1, 2, 4, 4}, DataType)); // [ M, I, H, W ] + layer->m_Weight = std::make_unique<ScopedTensorHandle>(TensorInfo({1, 4, 4, 2}, DataType)); // [ 1, H, W, I*M ] layer->m_Weight->Allocate(); // Creates extra layers. @@ -1181,7 +1181,7 @@ std::unique_ptr<DepthwiseConvolution2dFloat32Workload> CreateDepthwiseConvolutio CHECK(queueDescriptor.m_Inputs.size() == 1); CHECK(queueDescriptor.m_Outputs.size() == 1); - CHECK((queueDescriptor.m_Weight->GetTensorInfo() == TensorInfo({1, 2, 4, 4}, DataType))); + CHECK((queueDescriptor.m_Weight->GetTensorInfo() == TensorInfo({1, 4, 4, 2}, DataType))); // Returns so we can do extra, backend-specific tests. return workload; |