diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-10-16 19:21:40 +0100 |
|---|---|---|
| committer | Michele Di Giorgio <michele.digiorgio@arm.com> | 2019-11-06 15:53:24 +0000 |
| commit | dbdea0d1c025b18d4d82c278c87454427918f5b4 (patch) | |
| tree | 68bc25452f5d5b41006fb507c41516446cf8e457 /src/core/NEON/kernels/NEWeightsReshapeKernel.cpp | |
| parent | 75d47330e7ca0325cf5d83711452f6aeb085998f (diff) | |
| download | ComputeLibrary-dbdea0d1c025b18d4d82c278c87454427918f5b4.tar.gz | |
COMPMID-2308: NEConvolutionLayer: support QUANT8_SYMM_PER_CHANNEL filters
Change-Id: Ic1bf5f0d21ccd525f84213a360f7e199d7f50577
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2177
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/NEWeightsReshapeKernel.cpp')
| -rw-r--r-- | src/core/NEON/kernels/NEWeightsReshapeKernel.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/core/NEON/kernels/NEWeightsReshapeKernel.cpp b/src/core/NEON/kernels/NEWeightsReshapeKernel.cpp index 624833adfb..649316442e 100644 --- a/src/core/NEON/kernels/NEWeightsReshapeKernel.cpp +++ b/src/core/NEON/kernels/NEWeightsReshapeKernel.cpp @@ -49,7 +49,7 @@ TensorShape get_output_shape(const ITensorInfo *input, bool has_bias) Status validate_arguments(const ITensorInfo *input, const ITensorInfo *biases, const ITensorInfo *output) { //Note: ARM_COMPUTE_RETURN_ERROR_ON_CPU_F16_UNSUPPORTED(input) is not needed here as this kernel doesn't use NEON FP16 instructions. - ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::F16, DataType::F32); + ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::QSYMM8_PER_CHANNEL, DataType::F16, DataType::F32); ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(output); if(biases != nullptr) |