diff options
Diffstat (limited to 'src/armnnSerializer/test/LstmSerializationTests.cpp')
| -rw-r--r-- | src/armnnSerializer/test/LstmSerializationTests.cpp | 72 |
1 files changed, 48 insertions, 24 deletions
diff --git a/src/armnnSerializer/test/LstmSerializationTests.cpp b/src/armnnSerializer/test/LstmSerializationTests.cpp index 3178bc990e..d8f8967bcd 100644 --- a/src/armnnSerializer/test/LstmSerializationTests.cpp +++ b/src/armnnSerializer/test/LstmSerializationTests.cpp @@ -1454,7 +1454,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo inputToInputWeightsInfo(inputToInputWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor inputToInputWeights(inputToInputWeightsInfo, inputToInputWeightsData); armnn::TensorShape inputToForgetWeightsShape = {4, 2}; @@ -1462,7 +1463,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo inputToForgetWeightsInfo(inputToForgetWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor inputToForgetWeights(inputToForgetWeightsInfo, inputToForgetWeightsData); armnn::TensorShape inputToCellWeightsShape = {4, 2}; @@ -1470,7 +1472,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo inputToCellWeightsInfo(inputToCellWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor inputToCellWeights(inputToCellWeightsInfo, inputToCellWeightsData); armnn::TensorShape inputToOutputWeightsShape = {4, 2}; @@ -1478,7 +1481,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo inputToOutputWeightsInfo(inputToOutputWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor inputToOutputWeights(inputToOutputWeightsInfo, inputToOutputWeightsData); // The shape of recurrent weight data is {outputSize, outputSize} = {4, 4} @@ -1487,7 +1491,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo recurrentToInputWeightsInfo(recurrentToInputWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor recurrentToInputWeights(recurrentToInputWeightsInfo, recurrentToInputWeightsData); armnn::TensorShape recurrentToForgetWeightsShape = {4, 4}; @@ -1495,7 +1500,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo recurrentToForgetWeightsInfo(recurrentToForgetWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor recurrentToForgetWeights(recurrentToForgetWeightsInfo, recurrentToForgetWeightsData); armnn::TensorShape recurrentToCellWeightsShape = {4, 4}; @@ -1503,7 +1509,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo recurrentToCellWeightsInfo(recurrentToCellWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor recurrentToCellWeights(recurrentToCellWeightsInfo, recurrentToCellWeightsData); armnn::TensorShape recurrentToOutputWeightsShape = {4, 4}; @@ -1511,7 +1518,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo recurrentToOutputWeightsInfo(recurrentToOutputWeightsShape, armnn::DataType::QAsymmU8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::ConstTensor recurrentToOutputWeights(recurrentToOutputWeightsInfo, recurrentToOutputWeightsData); // The shape of bias data is {outputSize} = {4} @@ -1520,7 +1528,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo inputGateBiasInfo(inputGateBiasShape, armnn::DataType::Signed32, biasScale, - biasOffset, true); + biasOffset, + true); armnn::ConstTensor inputGateBias(inputGateBiasInfo, inputGateBiasData); armnn::TensorShape forgetGateBiasShape = {4}; @@ -1528,7 +1537,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo forgetGateBiasInfo(forgetGateBiasShape, armnn::DataType::Signed32, biasScale, - biasOffset, true); + biasOffset, + true); armnn::ConstTensor forgetGateBias(forgetGateBiasInfo, forgetGateBiasData); armnn::TensorShape cellBiasShape = {4}; @@ -1536,7 +1546,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo cellBiasInfo(cellBiasShape, armnn::DataType::Signed32, biasScale, - biasOffset, true); + biasOffset, + true); armnn::ConstTensor cellBias(cellBiasInfo, cellBiasData); armnn::TensorShape outputGateBiasShape = {4}; @@ -1544,7 +1555,8 @@ TEST_CASE("SerializeDeserializeQuantizedLstm") armnn::TensorInfo outputGateBiasInfo(outputGateBiasShape, armnn::DataType::Signed32, biasScale, - biasOffset, true); + biasOffset, + true); armnn::ConstTensor outputGateBias(outputGateBiasInfo, outputGateBiasData); armnn::QuantizedLstmInputParams params; @@ -1655,12 +1667,14 @@ TEST_CASE("SerializeDeserializeQLstmBasic") armnn::TensorInfo inputWeightsInfo({numUnits, inputSize}, armnn::DataType::QSymmS8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::TensorInfo recurrentWeightsInfo({numUnits, outputSize}, armnn::DataType::QSymmS8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::TensorInfo biasInfo({numUnits}, armnn::DataType::Signed32, biasScale, biasOffset, true); @@ -1816,22 +1830,26 @@ TEST_CASE("SerializeDeserializeQLstmCifgLayerNorm") armnn::TensorInfo inputWeightsInfo({numUnits, inputSize}, armnn::DataType::QSymmS8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::TensorInfo recurrentWeightsInfo({numUnits, outputSize}, armnn::DataType::QSymmS8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::TensorInfo biasInfo({numUnits}, armnn::DataType::Signed32, biasScale, - biasOffset, true); + biasOffset, + true); armnn::TensorInfo layerNormWeightsInfo({numUnits}, armnn::DataType::QSymmS16, layerNormScale, - layerNormOffset, true); + layerNormOffset, + true); // Mandatory params std::vector<int8_t> inputToForgetWeightsData = GenerateRandomData<int8_t>(inputWeightsInfo.GetNumElements()); @@ -2003,32 +2021,38 @@ TEST_CASE("SerializeDeserializeQLstmAdvanced") armnn::TensorInfo inputWeightsInfo({numUnits, inputSize}, armnn::DataType::QSymmS8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::TensorInfo recurrentWeightsInfo({numUnits, outputSize}, armnn::DataType::QSymmS8, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::TensorInfo biasInfo({numUnits}, armnn::DataType::Signed32, biasScale, - biasOffset, true); + biasOffset, + true); armnn::TensorInfo peepholeWeightsInfo({numUnits}, armnn::DataType::QSymmS16, weightsScale, - weightsOffset, true); + weightsOffset, + true); armnn::TensorInfo layerNormWeightsInfo({numUnits}, armnn::DataType::QSymmS16, layerNormScale, - layerNormOffset, true); + layerNormOffset, + true); armnn::TensorInfo projectionWeightsInfo({outputSize, numUnits}, armnn::DataType::QSymmS8, weightsScale, - weightsOffset, true); + weightsOffset, + true); // Mandatory params std::vector<int8_t> inputToForgetWeightsData = GenerateRandomData<int8_t>(inputWeightsInfo.GetNumElements()); |